Reimbursement Review

Ivosidenib (Tibsovo)

Sponsor: Servir Canada Inc.

Therapeutic area: Acute myeloid leukemia

This multi-part report includes:

Clinical Review

Pharmacoeconomic Review

Testing Procedure Assessment

Clinical Review

Abbreviations

Executive Summary

An overview of the submission details for the drug under review is provided in Table 1.

Table 1: Background Information of Application Submitted for Review

AML = acute myeloid leukemia; NOC = Notice of Compliance.

Introduction

Acute myeloid leukemia (AML) is a heterogeneous hematologic malignancy characterized by the clonal expansion of myeloid blasts in the bone marrow, peripheral blood, and/or other tissues.^1,2 Typical symptoms of AML include fatigue, pale skin, dyspnea, infection, dizziness, headache, and coldness in hands and feet.^3-5 Furthermore, leukopenia and neutropenia increase the risk of infections and fever, while thrombocytopenia increases the likelihood of bruising, bleeding, frequent or severe nosebleeds, bleeding gums, and heavy menstrual bleeding.⁵ Other symptoms include weight loss, night sweats, and loss of appetite.^6,7 AML is 1 of the most aggressive forms of leukemia.⁵ The Cancer Quality Council of Ontario has reported age-standardized 1-year (2017 to 2018) and 5-year survival rates (2014 to 2018) of 42.1% and 19.9%, respectively.⁸

The prevalence of AML ranges from 0.6 to 11.0 per 100,000 persons for all age categories, genders, and ethnicities globally.^9,10 The national age-standardized incidence rate for AML was reported to be 3.8 per 100,000 persons by Statistics Canada in 2018.¹¹ Approximately 1,600 patients in Canada were diagnosed with AML in 2022.¹² It is estimated that 6% to 10% of all people with AML carry an IDH1 mutation, with an estimated incidence ranging from 0.24 to 0.40 per 100,000 persons.^13-20 The incidence of IDH1-mutated AML is low, and it is considered to be a rare disease.²¹ Approximately 40% to 50% of people with newly diagnosed AML are ineligible for standard induction chemotherapy regimens because of older age, insufficient Karnofsky performance status or Eastern Cooperative Oncology Group (ECOG) performance status, and/or comorbid conditions.^12,22-25

The treatment goals for patients with AML who are not eligible to receive intensive induction chemotherapy are to prolong life, alleviate symptoms, reduce dependency on blood transfusion, reduce infections, and improve patients’ quality of life (QoL). Treatment options for patients with newly diagnosed AML who carry a mutation in the IDH1 enzyme and are ineligible for standard intensive chemotherapy (because of insufficient performance status, a comorbid medical condition, or age) are limited. In Canada, active treatment options that are currently publicly funded for patients with AML who are ineligible for standard intensive chemotherapy, although not specific to patients carrying an IDH1 mutation, include:^1,26-29

• venetoclax combined with azacitidine

• monotherapy with azacitidine or low-dose cytarabine (LDAC) if the patients are not considered candidates for combination therapy.

Ivosidenib is an inhibitor of the mutant IDH1 enzyme. Mutant IDH1 converts alpha-ketoglutarate to 2-hydroxyglutarate, which blocks cellular differentiation and promotes tumorigenesis in both hematologic and nonhematologic malignancies.³⁰ On July 19, 2024, ivosidenib in combination with azacitidine was approved by Health Canada for the treatment of adult patients with newly diagnosed AML with an IIDH1 R132 mutation who are not eligible to receive intensive induction chemotherapy. The sponsor’s reimbursement request is aligned with the Health Canada–approved indication. IIDH1 R132 mutation must be confirmed before the combination regimen is initiated.³⁰

The recommended dose for ivosidenib is 500 mg (2 × 250 mg tablets) taken orally once daily. Ivosidenib should be started on cycle 1 day 1 and administered once daily during the 28-day cycle. It should be started in combination with azacitidine at 75 mg/m² of body surface area, intravenously or subcutaneously, once daily on days 1 to 7 of each 28-day cycle. It is recommended that patients be treated for a minimum of 6 cycles. Treatment should be continued until disease progression or until treatment is no longer tolerated by the patient.³⁰

The objective of this report is to review and critically appraise the evidence submitted by the sponsor on the beneficial and harmful effects of ivosidenib (250 mg film-coated tablets) in combination with azacitidine for the treatment of adult patients with newly diagnosed IDH1-mutated AML who are not eligible for intensive induction chemotherapy.

Perspectives of Patients, Clinical Input, and Drug Input

The information in this section is a summary of the input provided by the patient and clinician groups who responded to our call for input and from the clinical experts consulted by the review team for this submission.

Patient Input

Two patient groups, the Leukemia & Lymphoma Society of Canada (LLSC) and Heal Canada, provided input to the review of ivosidenib. The LLSC is a national organization with charitable status dedicated to finding a cure for blood cancers and improving the QoL of people affected by blood cancers and their families by funding life-enhancing research and providing educational resources, services, and support. Heal Canada is a registered not-for-profit organization that aims to empower patients, improve health care outcomes, and advocate for equitable access to quality health care across Canada. Data were gathered through online surveys or emails with people diagnosed with AML and their caregivers. Eighty-three respondents participated in the survey from the LLSC, and 7 of those respondents identified as having the IDH1 mutation. The LLSC also conducted two 1-on-1 interviews with patients currently living with AML. Heal Canada launched an online survey to assess different characteristics of patients living with blood cancer. Of the 22 respondents, 5 had been diagnosed with AML. Information was also gathered from semistructured interviews with 2 patients and 2 caregivers. No patients or caregivers from Heal Canada had experience with ivosidenib; the LLSC interviewed 1 patient with previous experience with ivosidenib.

Most respondents reported that the mental, physical, and financial effects of AML have significant negative impact on the lives of patients and caregivers. The respondents described the challenges linked to the currently available treatments, such as intolerable side effects, lack of treatment response, and the limited options available to patients. Both respondent groups indicated that important patient outcomes included improved health-related QoL (HRQoL) (related to better control of anemia without transfusion or with fewer transfusions, as well as a lower infection rate), improved disease control, and prolonged survival. The patient who had experience with ivosidenib was initially treated with induction chemotherapy after a diagnosis of IDH1-mutated AML. After relapse on chemotherapy, the patient started ivosidenib and reported great response and minimal side effects from the treatment.

Clinician Input

Input From Clinical Experts Consulted by the Review Team for This Submission

The clinical experts identified the following unmet needs associated with the available treatments for patients with AML who are ineligible for intensive induction chemotherapy: first, not all patients respond to available therapies, and the outcomes for patients with AML (with or without IIDH1 R132 mutation) who are not eligible for intensive chemotherapy are poor; second, patients who respond to available therapy eventually relapse and succumb to their disease. Therefore, the clinical experts indicated that for patients in the target population, the most important treatment goals are to prolong remission and survival, reduce transfusion requirement, reduce the risk of infection and bleeding, and improve HRQoL.

The clinical experts indicated that ivosidenib would be reserved as first-line therapy for patients with AML who carry the IIDH1 R132 mutation and who are not eligible for intensive chemotherapy because of their age, comorbidities, or preference. Ivosidenib in combination with azacitidine could potentially replace the currently available combination therapy for these patients.

The clinical experts stated that only patients with a diagnosis of de novo AML with IIDH1 R132 mutation who are not eligible for intensive induction chemotherapy would be eligible to receive treatment with ivosidenib.

According to the experts, important outcomes for patients with AML are survival and improvements in HRQoL, response rates (in particular, complete remission [CR]), transfusion requirements, infection rates, and safety. The experts also noted that in clinical practice, patients’ responses to treatment are typically assessed every 28 days, corresponding to the length of the treatment cycles for azacitidine.

The experts noted that treatment with a combination of ivosidenib and azacitidine will be discontinued if disease progression is detected, if patients experience intolerable adverse events (AEs), and/or based on patient preference.

The clinical experts noted that, in general, patients should be treated by a hematologist and/or a hematologist or oncologist with experience in AML management. Treatment with ivosidenib can be administered in both inpatient and outpatient settings.

Clinician Group Input

Two clinician groups provided input for the review of ivosidenib in combination with azacitidine: the LLSC Clinician Network and the Ontario Health (Cancer Care Ontario) (OH-CCO) Hematology Cancer Drug Advisory Committee.

In general, the clinician group input was consistent with the input provided by the clinical experts consulted by the review team. The treatment goals for this patient population would be to prolong life, improve QoL, reduce transfusion requirement, and experience remission. The clinician groups noted that the current publicly funded treatment options for patients with AML who are not eligible for intensive chemotherapy include venetoclax plus azacitidine, single-drug azacitidine, LDAC, and best supportive care. The OH-CCO Drug Advisory Committee also mentioned venetoclax plus LDAC as an available therapy. However, not all patients respond to these therapies. In addition, both clinician groups suggested that treatment with venetoclax plus azacitidine is associated with increased risk of neutropenic fever and infections compared to azacitidine alone. According to the clinicians, infections may result in hospitalizations, which might last days to weeks depending on severity. The clinicians from the LLSC Clinician Network added that no tumour lysis syndrome monitoring is required with ivosidenib plus azacitidine. The clinician groups noted that specific inhibitors may offer a chance for increased treatment response and suggested ivosidenib plus azacitidine be considered as first-line therapy and become the new standard of care for adult patients with newly diagnosed IDH1-mutated AML who are not eligible for intensive induction chemotherapy or stem cell or bone marrow transplant. Both clinician groups indicated that remission rate, stabilization, and improvement in the frequency and severity of symptoms — such as improvement in blood counts, fewer transfusions, leukemia-free survival, and overall survival (OS), using usual leukemia response timelines — are the outcomes used to determine whether a patient is responding to ivosidenib plus azacitidine. Reasons for treatment discontinuation identified by the clinician groups included disease progression, intolerable side effects, and patient preference. Both clinician groups noted that ivosidenib plus azacitidine can be given in the inpatient and outpatient settings, or even in community centres that have experience treating acute leukemias.

Both the LLSC Clinician Network and the OH-CCO Drug Advisory Committee noted that timely results of testing for IDH1 mutation are required to identify patients who would benefit from and be eligible for this treatment.

Drug Program Input

Input was obtained from the drug programs that participate in our reimbursement review process. Refer to Table 4 for further information. The following were identified as key factors that could potentially impact the implementation of our recommendation for ivosidenib in combination with azacitidine:

• considerations for initiation of therapy

• considerations for discontinuation of therapy

• considerations for prescribing of therapy

• generalizability

• care provision issues.

Clinical Evidence

Systematic Review

Description of Studies

One international, phase III, multicentre, double-blind randomized controlled trial (RCT), the AGILE trial (N = 146), evaluated the efficacy and safety of ivosidenib plus azacitidine compared to placebo plus azacitidine in adult patients with newly diagnosed AML with an IIDH1 R132 mutation who were not eligible to receive intensive induction chemotherapy. Patients were recruited from 89 study sites across 20 countries. Eligible patients were randomized 1:1 to receive either ivosidenib (500 mg orally once daily) plus azacitidine (75 mg/m²/day, subcutaneous or IV) for 7 days, in 28-day cycles, or placebo in combination with azacitidine. The primary efficacy end point in the AGILE study was event-free survival (EFS). Key secondary end points were CR rates, OS, CR and CR with partial hematologic recovery (CRh), and objective response rate (ORR). Additional secondary end points in this study included HRQoL (measured by the European Organisation for Research and Treatment of Cancer Quality of Life Questionnaire Core 30 [EORTC QLQ-C30]), transfusion requirement, and harms. The majority of patients (73.3% per investigator [76% per Interactive Web Response System]) had de novo AML at initial diagnosis. There were more male patients in the ivosidenib plus azacitidine group (58.3%) than in the placebo plus azacitidine group (51.4%). According to the WHO classification of AML, fewer patients in the ivosidenib plus azacitidine group (22.2%) had AML with recurrent genetic abnormalities than in the placebo plus azacitidine group (32.4%); more patients in the ivosidenib plus azacitidine group (38.9%) had AML with myelodysplasia-related changes than in the placebo plus azacitidine group (35.1%). IIDH1 R132C was the most common polymorphism (65.8% of patients). In total, 63.9% of patients in the ivosidenib plus azacitidine group and 67.6% of patients in the placebo plus azacitidine group had an ECOG performance status score of 0 to 1. Cytogenetic risk status, as assessed by the investigators based on the 2017 National Comprehensive Cancer Network guidelines, was intermediate (63.0%: 66.7% in the ivosidenib plus azacitidine group versus 59.5% in the placebo plus azacitidine group) or poor (24.7%: 22.2% in the ivosidenib plus azacitidine group versus 27.0% in the placebo plus azacitidine group) for most patients at baseline. The median bone marrow blast at baseline was 52.5% (range, 17% to 100%).

Two data cut-offs (DCOs) were available for the AGILE trial. The first DCO (March 18, 2021) represents an unplanned early interim analysis by the Independent Data Monitoring Committee (IDMC), which occurred before the protocol-specified number of events for the planned analysis. Because of a notable difference in the number of deaths, which favoured ivosidenib, the IDMC recommended that trial recruitment should end early, treatment assignment should be unblinded, and crossover to ivosidenib should be allowed. The stopping boundaries were therefore adjusted, and this interim analysis became the final analysis. A later DCO (June 30, 2022) was available for OS, transfusion requirement, and harms.

Efficacy Results

The AGILE study met its primary end point. As of the DCO of March 18, 2021, the between-group difference in the EFS rate was 19.7% (95% confidence interval [CI], ███ ██ ████) at 6 months and 25.3% (95% CI, ███ ██ ████) at 12 months, favouring ivosidenib. Improvement in EFS was largely driven by the proportion of patients with treatment failure, assigned an event time of the date of randomization: 42 patients (58.3%) in the ivosidenib plus azacitidine group versus 59 patients (79.7%) in the placebo plus azacitidine group had treatment failure. The median EFS in the ivosidenib plus azacitidine group was 0.03 months (95% CI, 0.03 months to 11.01 months) and 0.03 months (95% CI, not estimable [NE] to NE) in the placebo plus azacitidine group. The median did not appear different between groups because the majority of events were treatment failures, which were assigned the date of randomization. The corresponding hazard ratio (HR) was 0.33 (95% CI, 0.16 to 0.69; P = 0.0011). Predefined sensitivity analyses supported the robustness of the primary analysis and suggested an EFS benefit associated with ivosidenib in the short-term.

Treatment with ivosidenib plus azacitidine was associated with prolonged OS and met the prespecified efficacy boundary for a statistically significant OS benefit at the DCO of March 18, 2021. At the updated DCO of June 30, 2022, 37 patients (50.7%) in the ivosidenib plus azacitidine group and 58 (77.3%) in the placebo plus azacitidine group had died. The median OS was 29.3 months (95% CI, 13.2 months to NE) in the ivosidenib plus azacitidine group and 7.9 months (95% CI, 4.1 months to 11.3 months) in the placebo plus azacitidine group (P < 0.0001). The corresponding HR was 0.42 (95% CI, 0.27 to 0.65). The between-group differences in the Kaplan-Meier–estimated OS rate were 24.6% (95% CI, ███ ██ ████) at 12 months and 35.7% (95% CI, ████ ██ ████) at 24 months.

The results of subgroup analyses for OS and EFS (prespecified for EFS) based on various patient baseline characteristics were consistent with those in the overall population.

As of March 18, 2021, the CR rate was 47.2% (95% CI, 35.3% to 59.3%) in the ivosidenib plus azacitidine group and 14.9% (95% CI, 7.7% to 25.0%) in the placebo plus azacitidine group. However, these estimates were affected by high risk of bias due to missing data.

As of the DCO of June 30, 2022, a higher proportion of patients in the ivosidenib plus azacitidine group (██ ████████ ███████) did not require red blood cell (RBC) and/or platelet transfusion than in the placebo plus azacitidine group (██ ████████ ███████). This measurement was from a nonrandomized subset of the population. According to the clinical experts, improved CR rates and a reduced transfusion requirement are considered clinically meaningful changes, and better CR rates and, in their opinion, reduced transfusion can subsequently be translated to improved HRQoL and, potentially, prolonged survival.

Harms Results

Overall, the safety results from the 2 DCOs were consistent.

As of the DCO of March 18, 2021, the proportion of patients who experienced at least 1 AE was 98.6% (70 patients) in the ivosidenib plus azacitidine group and 100% (73 patients) in the placebo plus azacitidine group. Patients treated with ivosidenib plus azacitidine were more likely (5% or more) to report the following AEs than patients treated with placebo plus azacitidine: vomiting (29 patients [40.8%] versus 19 patients [26.0%]), neutropenia (20 [28.2%] versus 12 [16.4%]), thrombocytopenia (20 [28.2%] versus 15 [20.5%]), prolonged electrocardiogram QT interval (14 [19.7%] versus 5 [6.8%]), insomnia (13 [18.3%] versus 9 [12.3%]), differentiation syndrome (10 [14.1%] versus 6 [8.2%]), pain in extremity (10 [14.1%] versus 3 [4.1%]), hematoma (9 [12.7%] versus 1 [1.4%]), arthralgia (8 [11.3%] versus 3 [4.1%]), headache (8 [11.3%] versus 2 [2.7%]), leukocytosis (8 [11.3%] versus 1 [1.4%]), and leukopenia (6 [8.5%] versus 2 [2.7%]).

Grade 3 and higher AEs were reported in 66 patients (93.0%) in the ivosidenib plus azacitidine group and 69 patients (94.5%) in the placebo plus azacitidine group. In both groups, commonly reported grade 3 and higher AEs were anemia (25.4% of patients in the ivosidenib plus azacitidine group versus 26.0% in the placebo plus azacitidine group), febrile neutropenia (28.2% versus 34.2%), neutropenia (26.8% versus 16.4%), thrombocytopenia (23.9% versus 20.5%), and pneumonia (22.5% versus 28.8%).

The proportion of patients who experienced serious adverse events (SAEs) was 69.0% (46 patients) in the ivosidenib plus azacitidine group and 82.2% (60 patients) in the placebo plus azacitidine group. Commonly reported SAEs in the 2 treatment groups were febrile neutropenia (23.9% of patients in the ivosidenib plus azacitidine group versus 27.4% in the placebo plus azacitidine group) and pneumonia (19.7% versus 21.9%).

The overall incidences of treatment-emergent adverse events (TEAEs) that led to combination treatment discontinuation were similar between the treatment groups: 19 patients (26.8%) in the ivosidenib plus azacitidine group and 19 patients (26.0%) in the placebo plus azacitidine group.

Differentiation syndrome and infection were identified by the clinical experts as notable harms for treatment with ivosidenib. As of June 30, 2022, differentiation syndrome was reported in 10 patients (13.9%) in the ivosidenib plus azacitidine group and 6 patients (8.1%) in the placebo plus azacitidine group. Infection was reported in 25 patients (34.7%) in the ivosidenib plus azacitidine group and 38 patients (51.4%) in the placebo plus azacitidine group.

Critical Appraisal

In the AGILE study, there were some imbalances in baseline patient characteristics between the 2 treatment groups, for example sex, WHO classification of AML, and cytogenetic risk status as assessed by the investigator. These imbalances are likely to be the result of the small sample size, within which prognostic balance is not likely to be assured; as such, there is some risk that the observed effects are overestimated or underestimated. In addition, the postbaseline transfusion requirement outcome was measured among approximately half the population who required transfusions at baseline. Randomization is not necessarily upheld in this population. However, the results of transfusion requirement in patients who were dependent on transfusion at baseline did not differ significantly from those in the overall population. Therefore, the potential for bias is unlikely to have an important impact on the study findings specific to this outcome.

The study originally had no planned interim analyses. Observations of a notable difference in the number of deaths (favouring ivosidenib) by the IDMC prompted an unplanned interim analysis before the protocol-defined number of events. To control for multiplicity, new stopping boundaries were calculated based on the observed information fraction that were not outlined in the original statistical analysis plan. Because the results are from an unplanned interim analysis (which became the final analysis), even though the new stopping boundaries are appropriate, there is a risk of overestimation of the true effects of the study drug.

HRQoL was assessed using the EORTC QLQ C-30, although this is not an AML-specific instrument. Even though a minimally important difference (MID) for EORTC QLQ C-30 score for patients with AML was not identified from the literature, a range of potential between-group MIDs (3 to 11 points for improvement and –5 to –13 points for deterioration on the global QoL scale) were established based on clinical trials of 9 cancer types and may provide some guidance when determining the clinical relevance of the findings for HRQoL in the AGILE study. The completion rate of the EORTC QLQ C-30 was low. The completion rates were ██████ █████ ███ ████ at 6 months, 12 months, and 18 months of the study. The evidence for HRQoL was considered to be very uncertain because of large amounts of missing data and imprecision; the CIs included the potential for little-to-no clinically meaningful difference between groups. The missing data imputation approach used may not adequately address the issue. Therefore, there is a high risk of bias because of the large amount of missing HRQoL outcome data in this study; the direction of bias cannot be predicted.

EFS was the primary efficacy outcome in this study. It is a composite end point, and the sample size of the AGILE study was small. In the AGILE study, almost all events occurred at baseline (i.e., 1 component of the end point). As such, there were few patients left at risk postbaseline; as a result, the EFS could not robustly characterize the long-term efficacy of the study drug.³¹ The correlations between EFS and OS were modest in the published research that provided trial-level information. However, 1 major limitation of these studies was that they were not specific to the population nor the drug class of interest, and therefore the ability to generalize the study findings was not clear.^32-34

According to feedback from the clinical experts, the eligibility criteria and baseline characteristics of the patients randomized in the AGILE study generally reflected a patient population in Canadian clinical practice that would receive combination therapy of ivosidenib plus azacitidine. The clinical experts noted that the results from the AGILE study could be generalized to patients with IDH1-mutated AML in Canada who would be treated with ivosidenib plus azacitidine. The clinical experts suggested that some flexibility should be applied in using ivosidenib plus azacitidine in patients with slightly worse ECOG performance status than in the trial. Patients’ IDH1 mutation status should be confirmed before the treatment. The experts indicated that the outcome measures in the AGILE study were generally appropriate and clinically relevant for clinical trials of AML.

In the AGILE study, ivosidenib in combination with azacitidine was compared with azacitidine monotherapy. The clinical experts consulted for this review indicated that azacitidine alone is not the most appropriate comparator for the study drug combination in the study population. Instead, venetoclax plus azacitidine is currently the most commonly used combination therapy in the target patient population. In practice, monotherapy with azacitidine would typically be used for patients who cannot tolerate treatment with the combination of venetoclax and azacitidine. There is a lack of direct evidence within the AGILE study with which to examine the efficacy and safety of the study drug compared with the other combination regimens.

GRADE Summary of Findings and Certainty of the Evidence

For pivotal studies and RCTs identified in the sponsor’s systematic review, the Grading of Recommendations, Assessment, Development, and Evaluations (GRADE) approach was used to assess the certainty of the evidence for the outcomes considered most relevant to inform the expert committee deliberations, and a final certainty rating was determined, as outlined by the GRADE Working Group.^35,36

Following the GRADE approach, evidence from RCTs started as high-certainty evidence and could be rated down for concerns related to study limitations (which refer to internal validity or risk of bias), indirectness, imprecision of effects, and publication bias.

When possible, certainty was rated in the context of the presence of an important (nontrivial) treatment effect; if this was not possible, certainty was rated in the context of the presence of any treatment effect (i.e., the clinical importance is unclear). In all cases, the target of the certainty of evidence assessment was based on the point estimate and where it was located relative to the threshold for a clinically important effect (when a threshold was available) or to the null. The threshold for a clinically important effect for OS and EFS in the study population was not obtained. Therefore, the target of the certainty of evidence assessment was the presence or absence of any (non-null) effect for survival rates. The threshold for a clinically important effect for the EORTC QLQ-C30 score was set according to the presence or absence of an important effect based on thresholds identified in the literature.³⁷ In addition, the target of the certainty of evidence assessment was the presence or absence of any non-null effect for CR, CR plus CR with incomplete hematologic recovery (CRi), and transfusion requirements. For some harm events (e.g., differentiation syndrome), because of the unavailability of the absolute difference in effects, the certainty of evidence was summarized narratively.

Table 2 presents the GRADE summary of findings for ivosidenib plus azacitidine versus placebo plus azacitidine.

The selection of outcomes for the GRADE assessment was based on the sponsor’s summary of clinical evidence, consultation with clinical experts, and input received from patient and clinician groups and public drug plans. The following list of outcomes was finalized in consultation with members of the expert committee:

• OS

• EFS

• CR

• CR plus CRi

• change from baseline in EORTC QLQ C-30 scores

• transfusion requirements

• any SAEs

• risk of AEs of special interest (differentiation syndrome, infection).

Table 2: Summary of Findings for Ivosidenib Plus Azacitidine Versus Placebo Plus Azacitidine for Patients With IDH1-Mutated AML

AML = acute myeloid leukemia; CI = confidence interval; CR = complete remission; CRi = complete remission with incomplete hematologic recovery; DCO = data cut-off; EFS = event-free survival; EORTC QLQ-C30 = European Organisation for Research and Treatment of Cancer Quality of Life Questionnaire Core 30; FAS = full analysis set; LS = least squares; NA = not applicable; NR = not reported; OR = odds ratio; OS = overall survival; RCT = randomized controlled trial; SAE = serious adverse event.

Notes: Study limitations (which refer to internal validity or risk of bias), indirectness, imprecision of effects, and publication bias were considered when assessing the certainty of the evidence. All serious concerns in these domains that led to the rating down of the level of certainty are documented in the table footnotes; the between-group differences of the efficacy and harm outcomes in this table were requested from the sponsor.

^aThe outcome of OS at the DCO of June 30, 2022, was not multiplicity adjusted; however, significance was met at an earlier multiplicity-adjusted analysis at the DCO of March 18, 2021.

^bRated down 1 level for serious imprecision. No threshold of clinical importance could be estimated, but it was considered that the effect estimate and entire CI were consistent with important benefit. The sample size and number of events are small, resulting in potential for overestimation of the true effect.

^cRated down 1 level for serious imprecision. A threshold of clinical importance could not be estimated, but it was judged that the lower bound of the 95% CI includes the potential for little-to-no important difference.

^dRated down 2 levels for very serious imprecision. The sample size is small for this composite end point; after the large majority of events assigned to the date of randomization due to treatment failure (which was the first component of the composite EFS end point), too few patients remained event-free to robustly assess the long-term effects on EFS.

^e,fRated down 1 level for serious imprecision (results were from interim analysis of study with small sample size and low number of events). Rated down 1 level for risk of bias due to what appears to be a large amount of missing outcome data due to no postbaseline assessment.

^gDid not rate down for risk of bias. Although only a subset of the population was represented, in which randomization may not be upheld, results appeared similar when compared to analysis of the full population. Rated down 2 levels for very serious imprecision. Using the null as the threshold, the point estimate suggests benefit while the lower bound of the CI suggests harm.

^h,iRated down 2 levels for very serious study limitations because of risk of bias due to missing outcomes data (data were available for 9% to 33% of the study population). Rated down 1 level for serious imprecision. The between-group difference of EORTC QLQ-C30 subscales exceed the identified minimally important difference for the global health states subscale in this instrument. However, the 95% CI included the possibility of little-to-no difference. Statistical testing for this outcome was not adjusted for multiplicity in the study and should be considered as supportive evidence.

^jRated down 1 level for serious imprecision. No threshold of clinical importance could be established; therefore, the null was used. The point estimate suggests benefit, but the 95% CI included the possibility of little-to-no difference.

^kRated down 2 levels for very serious imprecision. No threshold of clinical importance could be established; therefore, the null was used. The point estimate suggests harm, but the 95% CI includes the possibility of little-to-no difference or benefit.

^lRated down 1 level for serious imprecision. No threshold of clinical importance could be established; therefore, the null was used. The point estimate suggests benefit, but the 95% CI includes the potential for little-to-no difference.

Source: AGILE Clinical Study Report.^38,39 Details included in the table are from the sponsor’s summary of clinical evidence.

Long-Term Extension Studies

No relevant long-term extension studies were submitted by the sponsor.

Indirect Comparisons

Description of Studies

One report of 4 indirect treatment comparisons (ITCs) — 1 network meta-analysis [NMA] and 3 matching-adjusted indirect comparisons [MAICs] — was submitted by the sponsor to compare the treatment benefits and harms of ivosidenib plus azacitidine with other active therapies for the treatment of IDH1-mutated AML. A feasibility assessment was conducted to determine the feasibility of conducting indirect comparisons in the study population for the outcome of interest and to assess the heterogeneities across the included studies. The efficacy of ivosidenib versus comparators (venetoclax plus azacitidine, azacitidine, LDAC, decitabine, venetoclax plus LDAC, and glasdegib plus LDAC) on OS, EFS, CR rates, and transfusion requirement were evaluated, based on evidence from 6 RCTs.

Efficacy Results

For this submission, venetoclax plus azacitidine was identified as the most relevant comparator. As per the clinical experts consulted for this review, it is currently the most commonly used therapy in the patient population of interest. Evidence comparing ivosidenib plus azacitidine to venetoclax plus azacitidine was only available through a sponsor-submitted ITC report. The rarity of the population of interest limits the size and number of clinical studies completed with potential comparators and adds to the practical challenges when indirectly comparing treatment options. Based on the results of the NMA and MAICs, the evidence is insufficient to conclude whether ivosidenib plus azacitidine differs from venetoclax plus azacitidine in terms of OS, EFS, CR rates, or transfusion requirement in patients with untreated AML. The limitations associated with the ITCs included limited evidence from 6 RCTs, heterogeneity in the included trials, and imprecision of study results from the wide credible intervals (CrIs) or CIs for these outcomes.

Harms Results

Harm outcomes were not assessed in the ITCs.

Critical Appraisal

There was no a priori protocol for the ITCs; therefore, it cannot be known whether the analyses presented were selected from multiple analyses of the data. Although appropriate methods were used to reduce the risk of bias and error in data extraction, it was unknown if the risk of bias in the included trials was assessed by 2 independent reviewers. In addition, risk of bias was assessed at the level of the trial, rather than at the level of the reported results (i.e., per outcome), which ignores that risk of bias can vary by reported result within a trial. Some of the studies included within the NMA had some potential for risk of bias.

Six RCTs were included in the NMA. Heterogeneities were identified in the analysis populations, which included IDH1 mutation status, gender, type of AML diagnosis, cytogenic risk, performance status, median bone marrow blast, differences in placebo effect across placebo-controlled studies, and differences in the definition of EFS. For the time-to-event comparisons (e.g., EFS), lengths of follow-up were different, and with longer follow-up it may be expected that the HR would be attenuated, even when the proportional hazards (PH) assumption is not formally violated. The bias would likely favour the study drug. These differences would undermine the validity of the NMA, which relies on the transitivity assumption being upheld. The use of fixed-effect models was chosen based on the deviance information criterion. However, the use of fixed rather than random effects models means that the CrIs are unlikely to adequately express the uncertainty arising from the heterogeneity. The limited number of included studies did not allow for meta-regression or other techniques to adjust for differences in effect modifiers across studies within the NMA. The rarity of the population of interest limits the size and number of clinical studies completed with potential comparators and adds to the practical challenges when indirectly comparing treatment options.

In the NMA, given the lack of closed loops in the networks, consistency in the ITC analyses could not be tested, which increases the level of uncertainty. When comparing ivosidenib plus azacitidine with other combination regimens, the 95% CrIs for the point estimates were wide for some efficacy outcomes and spanned the null; therefore, confidence in the relative effect estimates for efficacy was limited because of the imprecision indicated by the wide CrIs for these outcomes, which precludes any conclusions as to which treatment may be favoured.

In the MAICs, the following potential effect modifier or prognostic factors were identified through the literature and a deliberating process by the sponsor: age, gender, ECOG performance status, type of AML, cytogenetic risk of AML, bone marrow blasts, and IDH1 mutation. The clinical experts consulted for this review agreed that these are relevant effect modifiers and prognostic variables. However, it is unclear if the identification of potential effect modifiers through the literature would be sufficient to identify all relevant treatment effect modifiers. The populations in the AGILE study and the other comparator studies were weighted and matched. Within the unanchored MAIC there was no reported estimate of the potential residual bias due to unadjusted confounders; as a result, the magnitude of residual confounding remains uncertain.

Before adjustment, the median OS and EFS for the placebo plus azacitidine groups were substantially different, suggesting reduced comparability of the populations. The main differences for the 2 studies used (AGILE and VIALE-A) is that in the AGILE study, the patients were younger and had a better ECOG performance status and a lower proportion of the patients had high-risk cytogenic status. The effective sample size (ESS) for the anchored MAICs was reduced by approximately one-third, suggesting that the results are heavily influenced by a subset of the sample population in the trial who may not be representative of the full sample population. The reduction in the ESS and the sample size in general resulted in wide CIs. Furthermore, there is uncertainty about comparing the population with IDH1 mutation to the intention-to-treat (ITT) population in the VIALE-A study. It was not possible to adjust for this factor.

The study population for this review includes patients with AML with IDH1 mutation who are ineligible for intensive chemotherapy. However, most of the selected trials were not specifically for IDH1-mutated AML. No other studies included only patients with IDH1 mutation, and it is not clear in the other included trials whether there were separate results for this particular subgroup. The prognostic significance of IDH1 status in AML, or whether IDH1 status may be a treatment effect modifier, remains uncertain. According to the clinical experts consulted for this review, the effect modifiers identified in patients with AML by the sponsor are also considered effect modifiers in patients with IDH1-mutated AML.

In this ITC report, several efficacy outcomes were analyzed, such as OS, EFS, and CR rates (not evaluated in the MAICs). However, other efficacy end points of interest to patients and clinicians (e.g., HRQoL), as well as harms, were not investigated. Therefore, the relative treatment effect of ivosidenib plus azacitidine versus relevant comparators on patients’ HRQoL and on harms remains unknown.

Studies Addressing Gaps in the Evidence From the Systematic Review

No relevant studies addressing gaps in the evidence from the systematic review were submitted by the sponsor.

Conclusions

Adult patients with newly diagnosed AML with an IIDH1 R132 mutation who are not eligible to receive intensive induction chemotherapy have a poor prognosis. Patients and clinicians highlighted the need for new treatments that prolong life, improve remission, reduce transfusion requirements, and maintain HRQoL. Evidence from a randomized, double-blind, phase III RCT (the AGILE study) showed that treatment with ivosidenib plus azacitidine likely results in a clinically important increase in the probability of OS at 12 months and 24 months compared to placebo plus azacitidine in the target population. Evidence from the trial also showed that ivosidenib plus azacitidine likely results in a clinically important increase in the probability of EFS at 6 months. EFS was a composite end point driven by treatment failure events; postbaseline, too few patients remained event-free to robustly characterize other components of the end point (i.e., relapse and death). The rates of CR, as well as CR plus CRi, and the need for transfusions may be improved with treatment with ivosidenib plus azacitidine compared with placebo plus azacitidine. Evidence on HRQoL was very uncertain because of the limitations of the analyses, including risk of bias due to missing data and imprecision. In terms of harms, evidence from the AGILE study suggested that treatment with ivosidenib plus azacitidine may result in an increase in differentiation syndrome but likely results in fewer infections and SAEs than treatment with placebo plus azacitidine.

There is a lack of direct comparative evidence between ivosidenib plus azacitidine and other relevant treatments for patients with AML who are not eligible for intensive chemotherapy, such as venetoclax plus azacitidine, which is currently the most commonly used treatment in the target patient population. Indirect evidence from a sponsor-submitted NMA of 6 trials and 3 MAICs comparing patients from the AGILE study to patients treated with venetoclax plus azacitidine in the VIALE-A study was insufficient to conclude whether treatment with ivosidenib plus azacitidine differs from treatment with venetoclax plus azacitidine in terms of OS, EFS, CR rates, and transfusion requirements. There was substantial uncertainty in the treatment effect estimates (indicated by wide CrIs) from the ITCs because of limited efficacy data and important heterogeneity across studies. No comparisons of HRQoL or harms, which are important to patients and clinicians, were conducted.

Introduction

The objective of this report is to review and critically appraise the evidence submitted by the sponsor on the beneficial and harmful effects of ivosidenib (tablets, 250 mg, oral use) in combination with azacitidine for the treatment of adult patients with newly diagnosed AML with an IIDH1 R132 mutation who are not eligible to receive intensive induction chemotherapy.

Disease Background

Contents within this section have been informed by materials submitted by the sponsor and clinical expert input. The following has been summarized and validated by our review team.

AML is a heterogeneous hematologic malignancy characterized by the clonal expansion of myeloid blasts in the bone marrow, peripheral blood, and/or other tissues.^1,2 Although the cause of AML is not known, several factors are associated with an increased risk of this disease, such as increasing age, male sex, genetic factors, environmental factors and lifestyle, drugs, chemical exposure, and antecedent blood disorders.⁴⁰ Commonly reported signs of AML are anemia, leukopenia, neutropenia, and thrombocytopenia, which result from the dysfunctional clonal expansion of myeloid progenitor cells. Typical symptoms of AML include fatigue, pale skin, dyspnea, infection, dizziness, headache, and coldness in hands and feet.^3-5 Furthermore, leukopenia and neutropenia increase the risk of infections and fever, while thrombocytopenia increases the likelihood of bruising, bleeding, frequent or severe nosebleeds, bleeding gums, and heavy menstrual bleeding. Other symptoms include weight loss, night sweats, and loss of appetite.^6,7 Occasionally, patients experience hepatomegaly, splenomegaly, or a soft tissue mass due to myeloid sarcoma.⁵

AML is 1 of the most aggressive forms of leukemia.⁵ Poorer prognosis is associated with increased age,^41,42 secondary AML (AML after prior diagnosis of myelodysplasia, myeloproliferative neoplasm, or aplastic anemia, as opposed to de novo AML, in which patients have no clinical history of prior myelodysplastic syndrome, myeloproliferative disorder, or exposure to potentially leukemogenic therapies or agents),⁴³ and certain molecular subtypes.² The Cancer Quality Council of Ontario has reported age-standardized 1-year (2017 to 2018) and 5-year survival rates (2014 to 2018) of 42.1% and 19.9%, respectively.⁸ Furthermore, AML mortality is strongly related to age, with the highest mortality rates in older people.⁴¹ Five-year net survival (based on the combined results from 2010 to 2012) reported by Statistics Canada was 62% for people aged 15 to 44 years, 44% for people aged 45 to 54 years, 24% for people aged 55 to 64 years, 10% for people aged 65 to 74 years, and 3% for people aged 75 years and older.⁴²

The prevalence of AML ranges from 0.6 to 11.0 per 100,000 persons for all age categories, genders, and ethnicities globally.^9,10 The national age-standardized incidence rate for AML was reported to be 3.8 per 100,000 persons by Statistics Canada in 2018.¹¹ CCO and the Cancer Quality Council of Ontario have reported relatively higher age-standardized incidence rates of 4.4 and 4.6 per 100,000 persons in Ontario, in 2016 and 2018, respectively.^44,45 Approximately 1,600 people in Canada were diagnosed with AML in 2022.¹² It is estimated that 6% to 10% of all people with AML carry an IDH1 mutation (with an estimated incidence ranging from 0.24 to 0.40 per 100,000 persons).^13-20 The incidence of IDH1-mutated AML is low, and it is considered to be a rare disease.²¹

Diagnosis of AML is based on morphology, immunophenotyping, cytogenetics and molecular cytogenetics, molecular testing, demographics and medical history, detailed family history, patient bleeding history, and performance status.^1,46-48 Approximately 40% to 50% of people with newly diagnosed AML are ineligible for standard induction chemotherapy regimens because of older age, poor Karnofsky performance status or ECOG performance status, and/or comorbid conditions.^12,22-25 Multiple international guidelines, such as those of the National Comprehensive Cancer Network, European LeukemiaNet, the American Society of Hematology and the College of American Pathologists, and the European Society for Medical Oncology, recommend testing for IDH1 mutations to identify patients who may benefit from IDH1-targeted treatments.^1,47-49

Standards of Therapy

Contents within this section have been informed by materials submitted by the sponsor and clinical expert input. The following has been summarized and validated by our review team.

The majority of patients with AML are aged 60 years or older. While results of treatment have improved steadily in younger adults over the past 20 years, there have been limited changes in outcomes among older adults. When treated with chemotherapy alone, this age group has an estimated 2-year survival probability of approximately 20% and 10% at 4 years and 5 years, respectively. The reasons for the unsatisfactory outcome in older adults likely relate to the increased frequency of unfavourable cytogenetics among older patients with AML, a greater frequency of antecedent myelodysplasia, as well as reduced ability to tolerate intensive chemotherapy. High-dose chemotherapy is not beneficial to older adults with AML. There has been an intense interest in the introduction of new treatment modalities.⁵⁰ The patient and clinician groups that provided input for this submission and the clinical experts consulted for this review indicated that the unmet therapeutic need for patients with AML stems from the poor outcomes (e.g., disease progression or relapse after previous remission, transfusion dependency, intolerable side effects, short life expectancy) in this patient group despite the currently available treatments and from the limited treatment options available if the patients’ current therapies fail. According to the clinical experts, the treatment goals for patients with AML who are not eligible to receive intensive induction chemotherapy are to prolong life, extend time in remission, alleviate symptoms, reduce dependency on blood transfusion, reduce infections, and improve QoL.

Treatment options for patients with newly diagnosed AML who carry a mutation in the IDH1 enzyme and are ineligible for the standard intensive chemotherapy (because of poor performance status, a comorbid medical condition, or age) are limited. In Canada, treatments that are currently publicly funded for patients with AML who are ineligible for standard intensive chemotherapy, but not specific to those carrying an IDH1 mutation, include:^1,26-29

• venetoclax combined with azacitidine (currently the mainstay and most frequently used treatment in the target patient population)

• monotherapy with azacitidine or LDAC if the patients are not considered candidates for combination or targeted therapy.

Before the introduction of venetoclax combination therapies, single-agent azacitidine or LDAC were recommended for patients with AML who were not eligible for intensive induction chemotherapy.⁵¹ Azacitidine and LDAC are widely available and reimbursed across Canada. Venetoclax (a small-molecule inhibitor of BLC-2, a protein that inhibits cells from programmed cell death²⁶) plus azacitidine or LDAC, or glasdegib (an inhibitor of the Hedgehog signal transduction pathway²⁷) plus LDAC, have been approved by Health Canada for the treatment of newly diagnosed AML in adult patients aged 75 years or older or who are otherwise not eligible to receive intensive induction chemotherapy. Glasdegib received a negative reimbursement recommendation in 2020 and, according to the clinical experts consulted for this review, is not routinely used in Canadian clinical practice. Venetoclax plus LDAC received a negative reimbursement recommendation in 2021, and according to the clinical experts consulted for this review, this regimen is not funded in jurisdictions in Canada, although some patients may have access to this treatment via a compassionate program. The clinical experts indicated that venetoclax plus azacitidine is currently the most commonly used treatment for the target patient population in Canada. Venetoclax plus azacitidine was recommended for reimbursement for patients with newly diagnosed AML aged 75 years or older or who have comorbidities that preclude the use of intensive induction chemotherapy. With the currently approved and reimbursed treatment options in Canada, the median OS in patients with AML, regardless of IDH mutation status, is approximately 5 months with LDAC,^52-54 10 months with azacitidine,^24,55 and 15 months with venetoclax plus azacitidine.⁵⁵ In patients with AML with IDH1 mutation who are not eligible for intensive induction chemotherapy, the quality of evidence for treatment with venetoclax plus azacitidine is low, limited to post hoc subgroup analyses with a small number of patients.^55-57

Drug Under Review

Ivosidenib has a non-cytotoxic mechanism of action. It is an inhibitor of the mutant IDH1 enzyme. Mutant IDH1 converts alpha-ketoglutarate to 2-hydroxyglutarate, which blocks cellular differentiation and promotes tumorigenesis in both hematologic and nonhematologic malignancies. The mechanism of action of ivosidenib beyond its ability to reduce 2-hydroxyglutarate and restore cellular differentiation is not fully understood.³⁰

On July 19, 2024, ivosidenib in combination with azacitidine was approved by Health Canada for the treatment of adult patients with newly diagnosed AML with an IIDH1 R132 mutation who are not eligible to receive intensive induction chemotherapy. The sponsor’s reimbursement request is aligned with the Health Canada–approved indication. The IIDH1 R132 mutation must be confirmed before the combination regimen is initiated.³⁰

Ivosidenib is provided as 250 mg film-coated tablets. The recommended dose is 500 mg ivosidenib (2 × 250 mg tablets) taken orally once daily. Ivosidenib should be started on cycle 1 day 1 and administered once daily during the 28-day cycle. It should be started in combination with azacitidine at 75 mg/m² of body surface area, intravenously or subcutaneously, once daily on days 1 to 7 of each 28-day cycle. The first treatment cycle of azacitidine should be given at 100% of the dose. It is recommended that patients be treated for a minimum of 6 cycles. Treatment should be continued until disease progression or until treatment is no longer tolerated by the patient.³⁰ Patients with AML and treated with ivosidenib have reported differentiation syndrome, which can be life-threatening or fatal if not treated.³⁰

Key characteristics of ivosidenib plus azacitidine are summarized in Table 3, with other treatments available for untreated or newly diagnosed AML.

Table 3: Key Characteristics of Ivosidenib, Venetoclax, Azacitidine, and Cytarabine

AML = acute myeloid leukemia; ARDS = adult respiratory distress syndrome; CNS = central nervous system; GI = gastrointestinal; LDAC = low-dose cytarabine; RNA = ribonucleic acid; SC = subcutaneous; TLS = tumour lysis syndrome.

^aHealth Canada–approved indication.

Sources: Product monographs for ivosidenib,³⁰ venetoclax,²⁶ and cytarabine.²⁹

Perspectives of Patient, Clinical Input and Drug Program Input

Patient Group Input

This section was prepared by the review team based on the input provided by patient groups. The full original patient inputs received by us have been included in the Patient, Clinical Input, and Drug Program Input section of this report.

We received 2 patient group submissions, from the LLSC and from Heal Canada. The LLSC is a national organization with charitable status dedicated to finding a cure for blood cancers and improving the QoL of people affected by blood cancers and their families by funding life-enhancing research and providing educational resources, services, and support. Heal Canada is a registered not-for-profit organization that aims to empower patients, improve health care outcomes, and advocate for equitable access to quality health care across Canada.

Data for the LLSC input were gathered using 1 online survey, distributed through various social media channels and directly by email in March 2024. The survey was developed and distributed by the LLSC, in English only. Eighty-three respondents proceeded with the survey, of which 7 respondents identified as having the IDH1 mutation. The LLSC also conducted 2 1-on-1 interviews with patients currently living with AML.

Heal Canada launched an online survey to assess different characteristics of patients living with blood cancer on February 27, 2024. Of the 22 respondents, 5 had been diagnosed with AML. Information was also gathered from semistructured interviews with 2 patients and 2 caregivers.

Most respondents in both patient groups reported that the mental, physical, and financial effects of AML have significant impact on the lives of patients and caregivers. According to Heal Canada, the predominant symptoms of AML are extreme fatigue, weakness, and tiredness, which make it difficult to accomplish basic daily tasks such as showering, washing dishes, cleaning the house, and shopping. People with this condition tend to be heavily dependent on their caregivers. The LLSC revealed that both patients and caregivers are forced to change how, if, and when they can interact with the people close to them, which has both mental and physical impacts for those affected. The caregiver burden is significant, especially for older patients and those living alone before being diagnosed with AML.

In terms of the currently available treatments, the LLSC highlighted that doses of the prescribed medications have to be decreased, or treatment has to be discontinued, when there are intolerable side effects or no response to treatment. However, if the available treatments fail and stem cell or bone marrow transplant is not an option, the only alternative is often best supportive care until death. Heal Canada provided some details about patients who mentioned receiving azacitidine or best supportive care (e.g., blood transfusion) and noted that both treatments necessitate frequent blood transfusion, which remains the most critical burden for patients with AML. Heal Canada also indicated that current treatment options have limited efficacy and significant harms and that patients may not receive active treatment but rather best supportive care. Patients expressed that they often feel trapped, with no real options to treat their cancer and improve their QoL.

Both patient groups indicated that important patient outcomes included improved HRQoL (related to better control of anemia without transfusion or with fewer transfusions, as well as a lower infection rate), improved disease control, and prolonged survival.

No patients or caregivers from Heal Canada had experience with ivosidenib, while the LLSC interviewed 1 patient with previous experience with ivosidenib. The patient was initially diagnosed with IDH1-mutated AML in June 2021 and started induction chemotherapy treatments immediately. After relapse on induction chemotherapy, the patient started ivosidenib with great response and minimal side effects, and she had been in remission since then. She was aware of the option of getting a transplant, but she was also frustrated about not having a donor.

Heal Canada reported that the turnaround time of companion testing is different across the country, and the LLSC commented that treatment with ivosidenib may be delayed in some treatment facilities if laboratory results are not made available within a short window of time. The LLSC noted that testing for IDH1 mutation is part of the next-generation sequencing panel, which is conducted on all patients with AML, and does not require an additional blood test.

Clinician Input

Input From Clinical Experts Consulted for this Review

All our review teams include at least 1 clinical specialist with expertise in the diagnosis and management of the condition for which the drug is indicated. Clinical experts are a critical part of the review team and are involved in all phases of the review process (e.g., providing guidance on the development of the review protocol, assisting in the critical appraisal of clinical evidence, interpreting the clinical relevance of the results, and providing guidance on the potential place in therapy). The following input was provided by 2 clinical specialists with expertise in the diagnosis and management of AML.

Unmet Needs

The clinical experts identified the following unmet needs associated with the currently available treatments for patients with AML who are ineligible for intensive induction chemotherapy: first, not all patients respond to available therapies, and effective treatments for this patient population are lacking, making the outcomes for patients with AML (with or without IIDH1 R132 mutation) who are not eligible for intensive chemotherapy extremely poor; second, patients who respond to available therapy eventually relapse and succumb to their disease. Therefore, the clinical experts indicated that for patients in the target population, the most important treatment goals are to prolong remission and survival, reduce transfusion requirements, reduce the risk of infection and bleeding, and improve HRQoL.

Place in Therapy

The clinical experts indicated that based on its unique mechanism of action (inhibition of the mutated IIDH1 R132) and the available clinical evidence, ivosidenib would be reserved as first-line therapy for patients with AML who carry the IIDH1 R132 mutation and who are not eligible for intensive chemotherapy because of their age or comorbidities. Ivosidenib in combination with azacitidine could potentially replace the currently available combination therapy for these patients.

Patient Population

The clinical experts stated that only patients with a diagnosis of de novo AML with IIDH1 R132 mutation who are not eligible for induction chemotherapy would be eligible to receive treatment with ivosidenib. The experts also noted that testing for the IDH1 mutation is routinely performed in many specialized leukemia centres across Canada, although not in all jurisdictions (e.g., not in Manitoba). However, delays of days to weeks in receiving the test results have been reported, which makes it challenging for the clinician to select the appropriate treatment for patients with newly diagnosed AML; they can either initiate treatment with the currently used therapies before a patient’s IDH1 status is verified or wait until the patient’s IDH1 mutation status can be obtained. In addition, the clinical experts suggested that some flexibility should be applied in using ivosidenib plus azacitidine in patients with slightly lower ECOG performance status than in the trial.

Assessing the Response to Treatment

The experts noted that important outcomes for patients with AML are survival, HRQoL, response rates (in particular CR), and safety. Other outcomes of interest to the clinicians include transfusion requirements and infection rates. The experts also noted that in clinical practice, patients’ response to treatment are typically assessed every 28 days, corresponding to the length of treatment cycles for azacitidine.

Discontinuing Treatment

According to the clinical experts consulted for this review, treatment with a combination of ivosidenib and azacitidine will be discontinued if there is evidence of disease progression, as demonstrated by either an increased number of blasts in the bone marrow according to the standards of the International Working Group or, if a bone marrow aspiration is not performed, worsening of blood counts and/or an increased number of circulating blasts. Other reasons for treatment discontinuation include intolerable AEs related to the treatment and patient preference.

Prescribing Considerations

The clinical experts noted that patients should be treated by a hematologist and/or hematologist or oncologist with experience in AML management. Treatment with ivosidenib can be administered in both inpatient and outpatient settings.

Clinician Group Input

This section was prepared by the review team based on the input provided by clinician groups. The full original clinician group input(s) received by the team have been included in the Patient, Clinical Input and Drug Program Input section of this report.

Two clinician groups provided input for the review of ivosidenib in combination with azacitidine: the LLSC Clinician Network and the OH-CCO Hematology Cancer Drug Advisory Committee.

In general, the input from the 2 clinician groups was consistent with the input provided by the clinical experts consulted by the review team. The treatment goals for this patient population would be to prolong life, improve QoL, reduce transfusion requirements, and experience remission. The clinician groups noted that the current publicly funded treatment options for patients with AML who are not eligible for intensive chemotherapy include venetoclax plus azacitidine, single-agent azacitidine, LDAC, and best supportive care. The OH-CCO Drug Advisory Committee also mentioned venetoclax plus LDAC as an available therapy. However, not all patients respond to these therapies. In addition, both clinician groups suggested that treatment with azacitidine plus venetoclax is associated with increased risk of neutropenic fever and infections compared to azacitidine alone. According to the clinicians, infections may result in hospitalizations, which might last days to weeks depending on severity. The clinicians from LLSC Clinician Network added that no tumour lysis syndrome monitoring is required with ivosidenib plus azacitidine. The clinician groups noted that specific inhibitors may offer a chance for increased treatment response and suggested ivosidenib plus azacitidine be considered as first-line therapy and become the new standard of care for adult patients with newly diagnosed IDH1-mutated AML who are not eligible for intensive induction chemotherapy or stem cell or bone marrow transplant. Both clinician groups indicated that remission rate and stabilization and improvement in the frequency and severity of symptoms — such as improvement in blood counts, fewer transfusions, leukemia-free survival, and OS, using usual leukemia response timelines — are the outcomes used to determine whether a patient is responding to ivosidenib plus azacitidine. Reasons for treatment discontinuation identified by the clinician groups included disease progression, intolerable side effects, and patient preference. Both clinician groups noted that ivosidenib plus azacitidine can be given in the inpatient and outpatient settings, or even in community centres that have experience treating acute leukemias.

Both the LLSC Clinician Network and the OH-CCO Drug Advisory Committee noted that timely results of testing for IDH1 mutation are required to identify patients who would benefit from and be eligible for this treatment.

Drug Program Input

The drug programs provide input on each drug being reviewed through our reimbursement review processes by identifying issues that may impact their ability to implement a recommendation. The implementation questions and corresponding responses from the clinical experts consulted by the review team are summarized in Table 4.

Table 4: Summary of Drug Plan Input and Clinical Expert Response

5 to 2-2 = azacitidine was administered on days 1 through 5 and days 8 and 9 of each 28-day cycle; AML = acute myeloid leukemia; CR = complete remission; ECG = electrocardiogram; ECOG = Eastern Cooperative Oncology Group; HMA = hypomethylating agent; KPS = Karnofsky performance status; LDAC = low-dose cytarabine; MDS = myelodysplastic syndrome; PCR = polymerase chain reaction; pERC = CADTH pan-Canadian Oncology Review Expert Review Committee; SC = subcutaneous.

Clinical Evidence

The objective of this Clinical Review Report is to review and critically appraise the clinical evidence submitted by the sponsor on the beneficial and harmful effects of ivosidenib (250 mg per tablet, oral use) in combination with azacitidine for the treatment of adult patients with newly diagnosed AML with an IIDH1 R132 mutation who are not eligible to receive intensive induction chemotherapy. The focus will be placed on comparing ivosidenib in combination with azacitidine to relevant comparators and identifying gaps in the current evidence.

A summary of the clinical evidence included by the sponsor in the review of ivosidenib in combination with azacitidine is presented in 4 sections, with critical appraisal of the evidence included at the end of each section. The first section, the systematic review, includes the pivotal studies and RCTs that were selected according to the sponsor’s systematic review protocol. Our assessment of the certainty of the evidence in this first section using the GRADE approach follows the critical appraisal of the evidence. The second section would include sponsor-submitted long-term extension studies; however, none were submitted by the sponsor. The third section includes indirect comparisons from the sponsor. The fourth section would include additional studies that were considered by the sponsor to address important gaps in the systematic review evidence; however, no studies addressing gaps were submitted by the sponsor.

Included Studies

Clinical evidence from the following are included in the current review and appraised in this document:

• One pivotal study (AGILE)^38,39 identified in the systematic review

• Four ITCs:⁵⁸ 1 NMA and 3 MAICs

Systematic Review

Contents within this section have been informed by materials submitted by the sponsor. The following has been summarized and validated by the review team.

Description of Study

The AGILE study (also known as AG120-C-009) is an ongoing, phase III, multicentre, double-blind RCT assessing the efficacy and safety of ivosidenib in combination with azacitidine compared with placebo in combination with azacitidine in patients with newly diagnosed IDH1-mutated AML who were ineligible for intensive induction chemotherapy. The primary objective of the study was to compare EFS between ivosidenib and placebo (each combined with azacitidine), as described in Table 5. Key secondary objectives of this study included comparing the remission rates and OS in patients treated with ivosidenib plus azacitidine versus with placebo plus azacitidine. HRQoL, transfusion requirements, and harms were also assessed in the AGILE study.

All recruited patients underwent screening procedures within the 4 weeks before randomization to determine eligibility. A screening bone marrow aspirate (or peripheral blood sample if bone marrow aspirate was not available) was required for confirmation of IDH1 mutation at a central laboratory. Patients eligible for study treatment were randomized 1:1 to receive oral ivosidenib or placebo, both administered in combination with subcutaneous or IV azacitidine. The randomization schedule was generated by an independent statistical group, and the randomization assignment was implemented by interactive response technologies. The patients, investigators, sponsor, and clinical research unit staff who dealt directly with patients were blinded to the treatment assignment. Ivosidenib and matched placebo were packaged and labelled identically so that the study pharmacist remained blinded to treatment assignment. Randomization was stratified by de novo status (de novo AML and secondary AML) and geographic region (US and Canada; Western Europe, Israel, and Australia; Japan; and rest of the world). As of the DCO of March 18, 2021, 89 sites globally had enrolled patients, including 2 sites in Canada. A total of 295 patients were screened, of which 146 underwent randomization (72 to the ivosidenib plus azacitidine group and 74 to the placebo plus azacitidine group). As of the second DCO on June 30, 2022, 2 more patients were included in the AGILE study, 1 in each treatment group. Updated analyses based on the data available on the second DCO were performed for OS, transfusion requirements, and harms.

The IDMC reviewed the safety data as of March 18, 2021, based on the 146 patients enrolled in the AGILE study. A greater number of deaths were observed in the placebo plus azacitidine group than in the ivosidenib plus azacitidine group. This prompted another unblinded analysis for efficacy, which included OS, EFS, and clinical response, and led to the IDMC recommendation to halt recruitment to the study on May 12, 2021. Because of a notable difference in the number of deaths, which favoured ivosidenib, the IDMC recommended that trial recruitment should end early, treatment assignment should be unblinded, and crossover to ivosidenib should be allowed. Patients who were already receiving ivosidenib plus azacitidine could continue to receive treatment on the same assessment schedule. Before crossover to ivosidenib, the investigators evaluated the patients to determine their safety eligibility based on the inclusion and exclusion criteria of the study.

The characteristics of the AGILE study are summarized in Table 5.

Table 5: Details of Studies Included in the Systematic Review

AE = adverse event; AESI = adverse event of special interest; AML = acute myeloid leukemia; ANC = absolute neutrophil count; CNS = central nervous system; CR = complete remission; CRh = complete remission with partial hematologic recovery; CRi = complete remission with incomplete hematologic recovery; CRp = complete remission with incomplete platelet recovery; DOCR = duration of complete remission; DOCRh = duration of CR plus CRh; DOCRi = duration of CR plus CRi (including CRp); DOR = duration of response; ECG = electrocardiogram; ECHO = echocardiogram; ECOG PS = Eastern Cooperative Oncology Group performance status; EFS = event-free survival; EORTC QLQ-C30 = European Organisation for Research and Treatment of Cancer Quality of Life Questionnaire Core 30; HMA = hypomethylating agent; HRQoL = health-related quality of life; IDMC = Independent Data Monitoring Committee; LVEF = left ventricular ejection fraction; MDS = myelodysplastic syndrome; MLFS = morphologic leukemia-free state; MUGA = multigated acquisition; ORR = objective response rate; OS = overall survival; PR = partial remission; RBC = red blood cell; RCT = randomized controlled trial; SAE = serious adverse event; SC = subcutaneous; TTCR = time to complete remission; TTCRh = time to CR plus CRh; TTCRi = time to CR plus CRi (including CRp); TTR = time to first response; ULN = upper limit of normal.

Source: AGILE Clinical Study Report.³⁸ Details in the table are from the sponsor’s summary of clinical evidence.

Populations

Inclusion and Exclusion Criteria

Eligible patients in the AGILE study were aged 18 years or older and had a centrally confirmed diagnosis of previously untreated AML with IIDH1 R132 mutation confirmed using an appropriate diagnostic test. Patients were required to be ineligible for intensive induction chemotherapy and to have an ECOG performance status score of 0 to 2 (on a 5-point scale, in which higher scores indicate greater disability). Additional eligibility criteria included no previous treatment with an IDH1 inhibitor or hypomethylating agent for myelodysplastic syndrome, as well as adequate hepatic and renal function. Patients were excluded if they were candidates for intensive induction chemotherapy for their AML, had received any prior treatment for AML (except for non-oncolytic treatments, such as hydroxyurea or leukapheresis) or prior hypomethylating agent for myelodysplastic syndrome, had received prior IDH1 inhibitor therapy, had severe cardiac disorder or pulmonary disorder, or had an active uncontrolled systemic fungal, bacterial, or viral infection without improvement despite appropriate antibiotics, antiviral therapy, and/or other treatment. Full inclusion and exclusion criteria, including the definition of ineligibility for intensive induction chemotherapy, in this study are provided in Table 5.

Interventions

Eligible patients in the AGILE study were randomly assigned to receive oral ivosidenib or placebo, both administered in combination with azacitidine. Ivosidenib 500 mg or matched placebo was administered orally once daily during weeks 1 to 4 in continuous 28-day cycles. Crossover between treatment arms was not permitted until the study was unblinded after the primary end point (EFS) reached statistical significance at an unplanned early interim analysis. Patients who were initially randomized to placebo plus azacitidine who met key safety eligibility criteria were given the opportunity to receive ivosidenib plus azacitidine following unblinding. Patients who were already receiving ivosidenib plus azacitidine could continue to receive this treatment on the same assessment schedule. Azacitidine at a dose of 75 mg/m²/day was administered subcutaneously or intravenously for 1 week every 4 weeks. A full 7 days of azacitidine was required, but as per institutional practice, a schedule of 5 days of daily dosing, followed by no dose received on the weekend, and 2 daily doses given again at the start of the next week, was allowed; the same schedule was to be used for each patient throughout the duration of treatment, when possible. Patients were to be treated for a minimum of 6 cycles of combination therapy.

Treatment was to be discontinued if any of the following occurred: death, disease relapse, disease progression, treatment failure (defined as patients with a response less than CR after receiving treatment for at least 24 weeks), clinical progression (within 24 weeks) not confirmed by International Working Group assessment, patient lost to follow-up, development of unacceptable AEs, confirmed pregnancy, withdrawal by patient, protocol violation, or end of study.

During the study, treatment with strong CYP3A4 inducers, sensitive CYP3A4 substrate medications with narrow therapeutic windows, and anticancer therapy (with the exception of hydroxyurea) were not allowed when patients were receiving the study treatment. Patients could receive analgesics, antiemetics, anti-infectives, antipyretics, antimicrobial prophylaxis, and blood products, as necessary. Information on subsequent therapies was collected during the EFS and OS follow-up periods.

Outcomes

A list of efficacy end points assessed in this Clinical Review Report is provided in Table 6, followed by descriptions of the outcome measures. Summarized end points are based on the outcomes included in the sponsor’s summary of clinical evidence as well as any outcomes identified as important to this review according to the clinical experts consulted for this review and input from patient and clinician groups and from public drug plans. Using the same considerations, the review team selected the end points considered most relevant to inform the expert committee deliberations and finalized this list of end points in consultation with members of the expert committee. OS, EFS, CR, HRQoL (measured with the EORTC QLQ C-30), and transfusion requirement were assessed using GRADE. SAEs and select notable harms outcomes considered important for informing the expert committee deliberations were also assessed using GRADE. Results for the ORR are presented in Appendix 1 but are not appraised for certainty using GRADE because the main treatment goals in the study population were to prolong survival and improve HRQoL, as indicated by the clinical experts consulted for this review and noted by the input, and because there is no evidence that ORR is a validated surrogate for OS.

Table 6: Outcomes Summarized From the AGILE Study

AE = adverse event; CR = complete remission; CRi = complete remission with incomplete hematologic recovery; DCO = data cut-off; EFS = event-free survival; EORTC QLQ-C30 = European Organisation of Research and Treatment of Cancer Quality of Life Questionnaire Core 30; GRADE = Grading of Recommendations, Assessment, Development, and Evaluations; OS = overall survival; SAE = serious adverse event; SoF = Summary of Findings; WDAE = withdrawal due to adverse event.

^aStatistical testing for these end points was adjusted for multiple comparisons (e.g., hierarchal testing). CR plus CR with partial hematologic recovery and objective response rate were included in the sponsor’s hierarchy tests; however, the certainty of the results of these outcomes was not assessed using GRADE in this Clinical Review Report.

Source: AGILE Clinical Study Report.³⁸ Details included in the table are from the sponsor’s summary of clinical evidence.

Overall Survival

OS was defined as the time from date of randomization to the date of death due to any cause. Once the study was unblinded, survival follow-up continued. All patients who were alive after an EFS event were to be contacted every 8 weeks for survival follow-up until death, withdrawal by patient, loss to follow-up, or the sponsor ending the study. Results of OS on both DCO dates are available.

In the AGILE study, reasons for censoring for OS included withdrawal of consent, loss to follow-up, and being alive at the DCO.

Event-Free Survival

EFS was a composite end point, defined as the time from randomization until treatment failure, relapse from remission, or death from any cause, whichever occurred first. Treatment failure was defined as a patient not experiencing CR by week 24 (i.e., being on treatment > 24 weeks without CR or treatment discontinuation ≤ 24 weeks without CR); these patients were considered to have an event at day 1 of randomization.

All patients who discontinued study treatment without experiencing death, disease relapse, or treatment failure, or without withdrawing consent were followed every day 1 (± 7 days) of weeks 9, 17, 25, 33, 41, and 53, and every 24 weeks thereafter, for EFS until they experienced treatment failure, relapse, or death; until they withdrew from the study; until 173 EFS events had occurred; or until deemed necessary by the IDMC.

During the AGILE study, the initial primary end point was changed from OS to EFS. The sponsor’s justification was that the sample size estimation showed that this change allowed for a smaller sample (200 instead of 398), a more feasible trial size in this rare patient population. Furthermore, EFS was considered by the sponsor to more accurately describe the contribution of a novel therapy to clinical benefit by removing the potentially confounding effects of posttrial therapies and by capturing treatment failure as an event. A previous phase Ib/II study (AG-221-AML-005) provided encouraging preliminary safety and efficacy data (primary efficacy end point: overall response rate) comparing an IDH1 inhibitor plus azacitidine with azacitidine alone.⁶¹ These considerations supported the amendment of the protocol to include a primary end point of EFS as a measure of clinical benefit for the treatment of patients with AML who are ineligible for intensive induction chemotherapy. This change was recorded in protocol amendment 5, on January 9, 2020, which was before unblinding of the data. OS was kept as a key secondary end point in the AGILE study and included in a fixed-sequence testing procedure to control the overall type I error rate.²¹

Reasons for censoring for EFS included the following: CR by 24 weeks, starting subsequent anticancer therapy; CR by 24 weeks, relapse or death documented after 2 or more missing disease assessments; CR by 24 weeks, lost to follow-up; CR by 24 weeks, withdrawal by patient; and CR by 24 weeks, ongoing in study without relapse or death.

Treatment Response

Multiple outcome measures related to response and remission were included in the AGILE study:

• Rate of CR, where CR is defined as bone marrow blasts less than 5% and no Auer rods, absence of extramedullary disease, absolute neutrophil count greater than or equal to 11,000/μL, platelet count greater than or equal to 100,000/μL, and independence from RBC transfusions. This outcome was assessed until the date of relapse. Only assessments performed on or before the start date of subsequent anticancer therapies were considered in the determination of this response end point. CR rate was 1 of the key secondary efficacy end points in the AGILE study.

• Rate of CR plus CRi (including CR with incomplete platelet recovery), where CRi (including CR with incomplete platelet recovery) was defined as all CR criteria except for residual neutropenia where the absolute neutrophil count was less than 1,000/μL or thrombocytopenia where the platelet count was less than 100,000/μL without platelet transfusion for at least 1 week before disease assessment.

Health-Related Quality of Life

In the AGILE study, HRQoL was measured using the EORTC QLQ C-30 and the EQ-5D questionnaires. Results of the disease-specific instrument EORTC QLQ-C30⁶² are included in this Clinical Review Report. EORTC QLQ-C30 score is a self-reported measure of HRQoL for patients with cancer who are receiving cancer treatment. The EORTC QLQ-C30 contains 30 items in total, and each item is evaluated on a 4-point or 7-point Likert scale. These 30 items can be categorized into 1 global health status/QoL scale, 5 functional scales, 3 symptom scales, and 6 single-item scales. Each scale is scored from 0 to 100, with a higher score representing more of the concept (e.g., more functioning or more symptoms). Each of the multi-item scales includes a different set of items. No item occurs in more than 1 scale (Table 7):

• one global health status/QoL scale (2 items)

• five functional scales: physical functioning (5 items), role functioning (2 items), emotional functioning (4 items), cognitive functioning (2 items), social functioning (2 items)

• three symptom scales: fatigue (3 items), nausea and vomiting (2 items), pain (2 items)

• six single-item scales relating to dyspnea, insomnia, appetite loss, constipation, diarrhea, and financial difficulties.

Table 7: Summary of Outcome Measures and Their Measurement Properties

AML = acute myeloid leukemia; EORTC QLQ-C30 = European Organisation for Research and Treatment of Cancer Quality of Life Questionnaire Core 30; HRQoL = health-related quality of life; ICC = intraclass correlation coefficient; MID = minimally important difference; QoL = quality of life.

^aPatients who were receiving IV chemotherapy at the time were excluded from the test-retest assessment.

Transfusion Requirements

The summary of on-treatment transfusion data included the number of patients with any on-treatment transfusions, the number of patients with each type of on-treatment transfusion (whole blood, packed RBCs, platelet, plasma, and other), the total number of units per patient with each type of on-treatment transfusion, and the reasons for which each type of transfusion was administered. Results of transfusion requirements on both DCO dates are available.

Hospitalization

Hospitalization due to AEs in the safety analysis population was evaluated in the AGILE study. No further information was available for the assessment on hospitalization. This outcome is presented in the Harms section in this report.

Harms

The harms of treatment with ivosidenib plus azacitidine were assessed by review of AEs, SAEs, discontinuations of the study intervention due to AEs, and AEs of special interest.

• AE: any untoward medical occurrence associated with the use of a drug in humans, whether or not considered drug related.

• SAE: an AE is considered serious if it results in death, a life-threatening condition, inpatient hospitalization or prolongation of existing hospitalization, a persistent or significant incapacity or substantial disruption of the ability to conduct normal life functions, congenital anomaly or congenital anomaly in a neonate or infant born to a parent exposed to study treatment, or an important medical event as assessed by the investigator or sponsor.

• Differentiation syndrome and infections were identified by the clinical experts consulted for this review as important notable harms for treatment with ivosidenib plus azacitidine. Results of harms on both DCO dates are available.

Statistical Analysis

Sample Size and Power Calculation

Assumptions for the placebo plus azacitidine group in the AGILE study were based on results from study AZA-AML-001, a phase III RCT comparing azacitidine with conventional care regimens in patients with newly diagnosed AML. Based on results from previous clinical trials, the CR rate at 24 weeks was assumed to be 20% for the placebo plus azacitidine group. For patients who experienced CR by 24 weeks, the median EFS was assumed to be 14.6 months.

Assumptions for the ivosidenib plus azacitidine treatment group in the AGILE study were based on results from study AG-221-AML-005, a phase Ib/II study comparing enasidenib (an IDH2 inhibitor) plus azacitidine with azacitidine alone in patients with newly diagnosed AML and IDH2 mutation. The CR rate by 24 weeks was assumed to be 40%. For patients who experienced CR by 24 weeks, a target HR of 0.76 for EFS was assumed (equivalent to a median EFS among patients who experienced response to treatment of 14.6 months in the placebo plus azacitidine group versus 19.2 months in the ivosidenib plus azacitidine group, assuming an exponential distribution).

Based on simulation results, the average overall HR over 10,000 simulations for the entire population was 0.641. Given that the assumption of PH was not met based on the EFS definition, the overall HR was less meaningful in this context. Therefore, the overall HR for the entire population was not part of the study design assumptions. Under these assumptions, 173 EFS events were required to provide 80% power at a 1-sided alpha of 0.025 level of significance to reject the null hypothesis using a stratified log-rank test. Assuming a recruitment period of approximately 44 months, with an accrual rate of 3 patients per month during the first 10 months and 5 patients per month thereafter, along with an assumed 5% overall dropout rate, approximately 200 patients with previously untreated IDH1-mutated AML were planned to be randomized to the 2 treatment groups in a 1:1 ratio. Given these assumptions, it was estimated that the analysis of the primary end point for EFS would occur approximately 52 months after the first patient was randomized. However, the planned sample size was not achieved. Instead, study enrolment ended early on the advice of the IDMC.

Statistical Testing

There were no planned interim analyses for efficacy in this study. Following the recommendation of the IDMC, enrolment in the study was stopped before the planned number of patients had been enrolled. It was decided that the IDMC DCO of March 18, 2021, would be used for the study’s primary analysis, which was an unplanned analysis before reaching the protocol prespecified 173 EFS events for the primary end point of EFS.

In the AGILE study, EFS was not initially the primary efficacy end point. The sponsor amended the protocol, and EFS became the primary end point as of January 9, 2020 (refer to the Outcomes section for more details). EFS was tested using the log-rank test stratified by the randomization stratification factors. The basis for a claim of efficacy would be the statistical significance of EFS in favour of the ivosidenib plus azacitidine group when the 1-sided P value was less than 0.025. This was later adjusted to a P value of 0.0046 at the unplanned interim analysis. Kaplan-Meier estimates of EFS were presented by treatment group, together with a summary of associated statistics. In addition, the EFS rates at 1 day and at 3, 6, 9, 12, 18, 24, and 36 months were estimated with corresponding 2-sided 95% CIs. The HR was estimated using a Cox PH model stratified by the randomization strata. Given that the PH assumption was not met, based on the EFS definition, the overall HR may not be meaningful. In the analyses for EFS, if the PH assumption was violated when large departures from the PH assumption were observed, the log-rank test would be underpowered to detect differences in the survival distributions for the treatment groups, and a test of the difference in the restricted mean survival time between the treatment group and the control group may be more appropriate to determine the superiority of the treatment group compared to the control group with respect to the time-to-event end point. The associated 95% CI for the difference in restricted mean survival time and 1-sided P value were generated.

As EFS is a composite end point, the estimates for each component were summarized, including CR rate by 24 weeks, and EFS among patients who experienced CR by 24 weeks.

Key secondary end points in the AGILE study were CR, OS, CR plus CRh, and ORR. A description of the statistical analyses for each efficacy outcome reported in the AGILE study is provided in Table 8.

The HR of OS was estimated using a Cox PH model stratified by the randomization strata. Kaplan-Meier estimates (product-limit estimates) were presented by treatment group, together with a summary of associated statistics, including the median OS time with 2-sided 95% CIs. In particular, the OS rate at 3, 6, 9, 12, 18, 24, and 36 months was estimated, with corresponding 2-sided 95% CIs.

Multiplicity Control

To control the overall type I error rate, the fixed-sequence testing procedure was planned to adjust for multiple statistical testing of the primary and key secondary efficacy end points. These end points were intended to be tested in the following order:

• EFS

• CR rate

• OS

• CR plus CRh rate

• ORR.

Table 8: Statistical Analysis of Efficacy End Points in the AGILE Study

AML = acute myeloid leukemia; CI = confidence interval; CMH = Cochran-Mantel-Haenszel; CR = complete remission; CRi = complete remission with incomplete hematologic recovery; DCO = data cut-off; eCRF = electronic case report form; EFS = event-free survival; EORTC QLQ-C30 = European Organisation for Research and Treatment of Cancer Quality of Life Questionnaire Core 30; EOT = end of treatment; FAS = full analysis set; HR = hazard ratio; IRT = interactive response technologies; ITT = intention to treat; OS = overall survival; PH = proportional hazards; RMST = restricted mean survival time.

^aAccording to the sponsor-provided additional information, no imputation was performed to handle missing data in the analyses for CR, CR plus CRi, and transfusion requirement. Instead, best response was reported for all patients in line with the ITT analysis. A patient’s best response may have been reported as “not assessed” if they did not have any postbaseline disease assessment, or as “not evaluable.”

Source: AGILE Clinical Study Report.³⁸ Details included in the table are from the sponsor’s summary of clinical evidence.

No control of the alpha level was made for the other analyses. To account for the unplanned interim analysis by the IDMC and early termination of the study, an individual set of group-sequential boundaries was applied separately to each of the efficacy end points. The stopping boundaries were calculated using the Lan and De Mets alpha spending function for 1-sided tests, accounting for the amount of information actually available (i.e., information fraction) at the time of the analysis.

For time-to-event end points, such as EFS and OS, the information fraction is the ratio of the number of actual events at the time of the analysis to the number of planned events as specified in the protocol. For binary end points, such as CR, the information fraction is the ratio of the number of patients who are randomized at the time of the analysis to the planned sample size as specified in the protocol. For EFS, CR, and OS, the 1-sided P value boundaries are 0.0046 (based on a 62.4% information fraction), 0.0087 (based on a 73.0% information fraction), and 0.0017 (based on a 51.0% information fraction), respectively.

The updated results of EFS, CR, and some other efficacy end points at the second DCO (June 30, 2022) were not available, because after unblinding the study and analyzing the primary efficacy end point, the invasive procedures and the necessary visits to the clinics for treatment response assessment (e.g., CR) were not warranted beyond those performed as per standard of care.

Sensitivity Analyses

Sensitivity analyses for EFS were performed to evaluate the robustness of the EFS end point. Details of these sensitivity analyses are presented in Table 8.

Subgroup Analyses

Prespecified subgroup analyses for EFS are presented in Table 9. Treatment groups were compared for EFS using a 2-sided unstratified log-rank test for each category, and the unstratified HR and its corresponding 95% CI was computed for each category and depicted in a Forest plot. If there was a small number of patients within a category (< 5% of the patients in the full analysis set [FAS]), the categories were pooled (if 3 or more categories are prespecified for the subgroup) or the subgroup would not be analyzed (if there are only 2 prespecified categories in the subgroup). Efficacy analyses in subgroups were purely exploratory and intended to evaluate the consistency of treatment effect.

Table 9: Subgroup Analyses Performed for EFS in the AGILE Study

AML = acute myeloid leukemia; ECOG PS = Eastern Cooperative Oncology Group performance status; eCRF = electronic case report form; EFS = event-free survival; IRT = interactive response technologies; WBC = white blood cell.

^aFor bone marrow blasts, bone marrow aspirate was used as the primary source. If a bone marrow aspirate assessment was not available, a bone marrow biopsy assessment was used.

Source: AGILE Clinical Study Report.³⁸ Details included in the table are from the sponsor’s summary of clinical evidence.

Analysis Populations

Analysis populations of the AGILE study are summarized in Table 10.

Table 10: Analysis Populations of the AGILE Study

AML = acute myeloid leukemia; CR = complete remission; ECOG PS = Eastern Cooperative Oncology Group performance status; FAS = full analysis set; HMA = hypomethylating agent; ITT = intention to treat; PPS = per-protocol set; SAS = safety analysis set.

^aKey secondary end points were CR rate, overall survival, CR plus CR with partial hematologic recovery rate, and objective response rate.

Source: AGILE Clinical Study Report.³⁸ Details included in the table are from the sponsor’s summary of clinical evidence.

Results

Patient Disposition

A summary of patient disposition in the AGILE study is provided in Table 11.

As of the DCO of March 18, 2021, 144 patients (98.6%) were randomized and received at least 1 dose of the study treatment: 71 patients in the ivosidenib plus azacitidine group and 73 patients in the placebo plus azacitidine group. One patient in each treatment group died before receiving study treatment. Twenty-six patients (36.1%) were ongoing with ivosidenib plus azacitidine, and 12 patients (16.2%) were ongoing with placebo plus azacitidine as of this cut-off date.

A total of 106 patients discontinued ivosidenib or placebo: 45 (62.5%) in the ivosidenib plus azacitidine group and 61 (82.4%) in the placebo plus azacitidine group. The main reasons for treatment discontinuation were AEs (27.4%) and progressive disease (17.1%). Other reasons included patient withdrawal (10.3%), clinical progression (6.2%), lack of treatment benefit (6.2%), other (4.8%), and death (1 patient in the placebo plus azacitidine group), with similar proportions of patients in both groups within each category. For patients who discontinued azacitidine, the distribution of discontinuation rates due to these reasons were also similar across both groups.

Eighty-five patients (58.2%) discontinued the study: 34 in the ivosidenib plus azacitidine group and 51 in the placebo plus azacitidine group. Among them, 74 patients (50.7%) died (including 1 patient per group who died due to COVID-19), 10 patients (6.8%) withdrew, and 1 patient (0.7%) was lost to follow-up (placebo plus azacitidine).

Table 11: Summary of Patient Disposition From the AGILE Study (FAS, DCO March 18, 2021)

AE = adverse event; DCO = data cut-off; FAS = full analysis set; IWG = International Working Group; PPS = per-protocol set; SAS = safety analysis set.

^aAs of June 30, 2022, 2 more patients were enrolled in the AGILE study, 1 in each treatment group.

Source: AGILE Clinical Study Report.³⁸ Details included in the table are from the sponsor’s summary of clinical evidence.

Baseline Characteristics

The baseline characteristics outlined in Table 12 are those that are most relevant to this review or were felt to affect the outcomes or interpretation of the study results.

The overall patient population was composed of a similar proportion of male and female patients (80 [54.8%] and 66 [45.2%], respectively; there were more male patients in the ivosidenib plus azacitidine group [58.3%] than in the placebo plus azacitidine group [51.4%]). The patients were primarily older than 75 years (82 patients [56.2%]). Race was unreported for most patients (84 [57.5%]).

The majority of patients (73.3% per investigator [76% per Interactive Web Response System]) had the subtype de novo (primary) AML at initial diagnosis, whereby AML arises as a new condition (the remaining patients in the trial had secondary AML, whereby the disease appears alongside a history of hematologic disorders).⁶⁵ Based on the WHO classification of AML, fewer patients in the ivosidenib plus azacitidine group (22.2%) had AML with recurrent genetic abnormalities than in the placebo plus azacitidine group (32.4%); more patients in the ivosidenib plus azacitidine group (38.9%) had AML with myelodysplasia-related changes than in the placebo plus azacitidine group (35.1%). IIDH1 R132C was the most common polymorphism (65.8% of patients). In total, 63.9% of patients in the ivosidenib plus azacitidine group and 67.6% of patients in the placebo plus azacitidine group had an ECOG performance status score of 0 to 1. Cytogenetic risk status, as assessed by the investigators based on the 2017 National Comprehensive Cancer Network guidelines, was intermediate (63.0% of patients) or poor (24.7% of patients) for most patients at baseline. The baseline median bone marrow blast proportion was 52.5% (range, 17% to 100%).

A summary of baseline characteristics in the AGILE study is provided in Table 12.

Table 12: Summary of Baseline Characteristics From the AGILE Study (FAS)

AML = acute myeloid leukemia; ECOG PS = Eastern Cooperative Oncology Group performance status; FAS = full analysis set; MDS = myelodysplastic syndrome; SD = standard deviation.

^aIDH1 mutation for these patients was confirmed with local testing.

^bFor bone marrow blasts, bone marrow aspirate was used as the primary source. If a bone marrow aspirate assessment was not available, a bone marrow biopsy assessment was used.

Source: AGILE Clinical Study Report.³⁸ Details included in the table are from the sponsor’s summary of clinical evidence.

Exposure to Study Treatments

A summary of patient exposure in the AGILE study is provided in Table 13.

Overall, the median duration of exposure to ivosidenib plus azacitidine was 5.79 cycles (Q1: 1.25; Q3: 15.25), and the median duration of exposure to placebo plus azacitidine was 2.32 cycles (Q1: 1.25; Q3: 5.82).

Table 13: Patient Exposure in the AGILE Study (SAS, DCO March 18, 2021)

DCO = data cut-off; IQR = interquartile range; NR = not reported; Q1 = first quartile; Q3 = third quartile; SAS = safety analysis set; SD = standard deviation.

Note: Duration of exposure (days) = (date of last dose – date of first dose + 1).

Source: AGILE Clinical Study Report.³⁸ Details included in the table are from the sponsor’s summary of clinical evidence.

Concomitant Medications and Co-Interventions

███ ████████ ██ ████ █████████ ██████ ███████ received concomitant medications. Overall, the most frequently used medications in both groups were similar. In the ivosidenib plus azacitidine group (N = 71), the most frequently used medications (≥ 20 patients) included ███████████ ██ ██ ███████ █████████ ███████████ ██ ██ ███████ █████████ ████████████ ██████ ██ ██████████ ██████ ██ ██ ███████ █████████ ██████████ ██ ██ ███████ █████████ ███ ████████████████ ██ ██ ███████ ████████. In the placebo plus azacitidine group (N = 73), the most frequently used medications (≥ 20 patients) included ███████████ ██ ██ ███████ █████████ ██████████ ██ ██ ███████ █████████ ███████████ ██ ██ ███████ █████████ ████████████ ██ ██ ███████ █████████ ████████████ ██████ ██ ██████████ ██████ ██ ██ ███████ █████████ ██████████████ ██ ██ ███████ █████████ █████████ ██ ██ ███████ █████████ █████████ ██ ██ ███████ █████████ █████████ ████████ ██ ██ ███████ █████████ ███ ██████████████ █████████████ ██ ██ ███████ ████████. The most frequently used concomitant medications in the AGILE study are summarized in Table 14.

Table 14: Most Frequently Used Concomitant Medications in the AGILE Study (SAS, DCO March 18, 2021)

DCO = data cut-off; SAS = safety analysis set.

Source: AGILE Clinical Study Report.³⁸ Details included in the table are from the sponsor’s summary of clinical evidence.

Subsequent Treatments

A summary of subsequent anticancer therapies in the AGILE study is provided in Table 15. Fourteen patients (19.4%) in the ivosidenib plus azacitidine group (N for FAS = 72) and 16 patients (21.6%) in the placebo plus azacitidine group (N for FAS = 74) had at least 1 subsequent anticancer therapy. As of the DCO of March 18, 2021, 4 patients (5.6%) in the ivosidenib plus azacitidine group (N for safety analysis set [SAS] = 71) and 1 patient (1.4%) in the placebo plus azacitidine group (N for SAS = 73) had had an allogeneic hematopoietic stem cell transplant. As of the DCO of June 30, 2022, a total of 7 patients (4.7%) had had an allogeneic hematopoietic stem cell transplant: 5 (6.8%) in the ivosidenib plus azacitidine group and 2 (2.7%) in the placebo plus azacitidine group. Five patients initially treated with placebo plus azacitidine crossed over to the ivosidenib plus azacitidine group after March 18, 2021.³⁹

Table 15: Subsequent Anticancer Treatments From the AGILE Study (FAS, DCO March 18, 2021)

DCO = data cut-off; FAS = full analysis set.

Notes: Patients with multiple medications within a preferred term are counted only once in that preferred term. Patients with multiple medications within an Anatomic Therapeutic Chemical classification are counted only once in that classification.

Source: AGILE Clinical Study Report.³⁸ Details included in the table are from the sponsor’s summary of clinical evidence.

Efficacy

Event-Free Survival

EFS was the primary efficacy end point in the AGILE study. As of the DCO of March 18, 2021, the median EFS was 0.03 months (95% CI, 0.03 months to 11.01 months) in the ivosidenib plus azacitidine group and 0.03 months (95% CI, NE to NE) in the placebo plus azacitidine group. The median did not appear different between groups because of the majority of events being treatment failure, which were assigned the date of randomization. The corresponding HR was 0.33 (95% CI, 0.16 to 0.69; P = 0.0011). The primary end point met the defined boundary for statistical significance. Forty-two patients (58.3%) in the ivosidenib plus azacitidine group experienced treatment failure, as did 59 patients (79.7%) in the placebo plus azacitidine group, and were considered to have had an EFS event at day 1. Most of the treatment failure events were due to treatment discontinuation without CR: 36.1% in the ivosidenib plus azacitidine group and 64.9% in the placebo plus azacitidine group. Three patients (4.2%) in the ivosidenib plus azacitidine group and 2 patients (2.7%) in the placebo plus azacitidine group had relapsed AML. One patient (1.4%) in each group died. Data were censored for 26 patients (36.1%) in the ivosidenib plus azacitidine group and 12 patients (16.2%) in the placebo plus azacitidine group. The between-group difference in EFS rates for ivosidenib plus azacitidine versus placebo plus azacitidine were 19.7% (95% CI, ███ ██ █████) at 6 months and 25.3% (95% CI, ███ ██ █████) at 12 months (Table 16). The between-group differences in EFS rates at later time points were not reported in the study. In addition, updated EFS results at the second DCO (June 30, 2022) were not available, because after the study was unblinded and the primary efficacy end point analyzed, invasive procedures and the associated visits to the clinics were not warranted beyond those performed as part of standard of care.

Table 16: Summary of Event-Free Survival in the AGILE Study (FAS, DCO March 18, 2021)

CI = confidence interval; CR = complete remission; DCO = data cut-off; EFS = event-free survival; FAS = full analysis set; HR = hazard ratio; NE = not estimable; NR = not reported.

^aEstimated using the product-limit (Kaplan-Meier) method. The CIs are calculated using the Brookmeyer and Crowley method with log-log transformation.

^bHR is estimated using a Cox proportional hazards model stratified by the randomization stratification factors (acute myeloid leukemia status and geographic region), with placebo plus azacitidine as the denominator.

^cP value is calculated from the 1-sided log-rank test stratified by the randomization stratification factors (acute myeloid leukemia status and geographic region); the stopping boundary was 1-sided P = 0.0046.

^dEFS rate is the estimated probability that a patient will remain event-free up to the specified time point. EFS rates are obtained from the Kaplan-Meier survival estimates. CIs are calculated using the Greenwood formula and log-log transformation.

Source: AGILE Clinical Study Report.³⁸ Details included in the table are from the sponsor’s summary of clinical evidence.

The Kaplan-Meier EFS curves are shown in Figure 1.

The restricted mean survival time for EFS calculated up to 18.2 months was 7.1 months in the ivosidenib plus azacitidine group and 3.1 months in the placebo plus azacitidine group. The between-group difference was 4.0 months (95% CI, 1.5 to 6.5; P = 0.0009).

In general, the results of the sensitivity analyses were consistent with those of the primary analysis for EFS in terms of median EFS, HR, and proportion of censored patients, except for sensitivity analysis 5, where EFS was tested using the log-rank test stratified by randomization stratification factors and based on the FAS. In this sensitivity analysis, patients who did not experience CR by week 24 were not considered to have had an EFS event at day 1 of randomization; the event time was either 24 weeks or the end of treatment, whichever was earlier. In this analysis, the median EFS was ███ ██████ ████ ██ ███ ██ ████ in the ivosidenib plus azacitidine group versus ███ ██████ ████ ██ ███ ██ ████ in the placebo plus azacitidine group. The HR was similar to the main analysis ████ ████ ███ ████ ██ ██████ ████████████. In another sensitivity analysis, where EFS with treatment failure was defined as a lack of CR, CRi, or morphologic clearance of leukemic cells from the marrow after at least 24 weeks of treatment, the median EFS was 22.9 months (95% CI, 7.5 months to NE) with ivosidenib plus azacitidine and 4.1 months (95% CI, 2.7 months to 6.8 months) with placebo plus azacitidine (HR = 0.39; 95% CI, 0.24 to 0.64; P < 0.001).

Results of the subgroup analyses showed that the improvement in EFS with ivosidenib plus azacitidine was generally consistent across the prespecified subgroups (Figure 2).

Figure 1: Kaplan-Meier Plot of EFS in the AGILE Study (FAS, DCO March 18, 2021)

AG-120 = ivosidenib; CI = confidence interval; DCO = data cut-off; EFS = event-free survival; FAS = full analysis set; NE = not estimable.

Source: AGILE Clinical Study Report.³⁸

Figure 2: Forest Plot of EFS by Subgroups in the AGILE Study (FAS, DCO March 18, 2021)

AG-120 = ivosidenib; AML = acute myeloid leukemia; AZA = azacitidine; CI = confidence interval; DCO = data cut-off; ECOG PS = Eastern Cooperative Oncology Group performance status; eCRF = electronic case report form; EFS = event-free survival; FAS = full analysis set; IRT = interactive response technologies; NE = not estimable; ROW = rest of world; WBC = white blood cell.

Notes: The hazard ratio is calculated from the unstratified Cox regression model, with placebo plus azacitidine as the denominator, with 2-sided 95% CI. “Greater than or equal to 20% of baseline blasts” was reported for 1 patient within the ivosidenib plus azacitidine arm. This patient is not included in the subgroup analyses for baseline percent bone marrow blasts.

Source: AGILE Clinical Study Report.³⁸

Overall Survival

At the DCO of March 18, 2021, the median follow-up time was approximately 15 months for both treatment groups. Twenty-eight patients (38.9%) in the ivosidenib plus azacitidine group and 46 patients (62.2%) in the placebo plus azacitidine group had died (HR = 0.44; 95% CI, 0.27 to 0.73; P = 0.0005). OS met the predefined boundary for statistical significance. The median OS was 24.0 months (95% CI, 11.3 months to 34.1 months) in the ivosidenib plus azacitidine group and 7.9 months (95% CI, 4.1 months to 11.3 months) in the placebo plus azacitidine group (Figure 3). The OS rates were 45.4% (at 24 months) to 84.2% (at 3 months) in the ivosidenib plus azacitidine group and 20.5% (at 24 months) to 66.6% (at 3 months) in the placebo plus azacitidine group. Between-group differences in the OS rate at these time points were not available as of March 18, 2021. Treatment effect was maintained in the analysis of the per-protocol set (data not shown in this report).

After the DCO of June 30, 2022, the median follow-up time for OS was similar between the treatment groups in the FAS: ████ ██████ in the ivosidenib plus azacitidine group and ████ ██████ in the placebo plus azacitidine group. Two more patients were enrolled in the AGILE study. Five patients originally in the placebo group crossed over to the ivosidenib group. As of June 30, 2022, 95 OS events had occurred: 37 in the ivosidenib plus azacitidine group and 58 in the placebo plus azacitidine group (HR = 0.42; 95% CI, 0.27 to 0.65; P < 0.0001). The median OS was 29.3 months (95% CI, 13.2 months to NE) in the ivosidenib plus azacitidine group and 7.9 months (95% CI, 4.1 months to 11.3 months) in the placebo plus azacitidine group. The between-group difference in OS rate for ivosidenib plus azacitidine versus placebo plus azacitidine was 24.6% (95% CI, ███ ██ ████) at 12 months and 35.7% (95% CI, ████ ██ ████) at 24 months, respectively (Table 17). OS at the second DCO was not multiplicity adjusted (but significance was met at the earlier test).

Table 17: Summary of OS in the AGILE Study (FAS, DCO March 18, 2021, and June 30, 2022)

CI = confidence interval; DCO = data cut-off; FAS = full analysis set; HR = hazard ratio; NE = not estimable; OS = overall survival.

^aEstimated using the product-limit (Kaplan-Meier) method. The CIs are calculated using the Brookmeyer and Crowley method with log-log transformation.

^bHR is estimated using a Cox proportional hazards model stratified by the randomization stratification factors (acute myeloid leukemia status and geographic region), with placebo plus azacitidine as the denominator.

^cP value is calculated from the 1-sided log-rank test stratified by the randomization stratification factors (acute myeloid leukemia status and geographic region); the stopping boundary was 1-sided P = 0.0017.

^dOS rate is the estimated probability that a patient will remain alive to the specified time point. OS rates are obtained from the Kaplan-Meier survival estimates. CIs are calculated using Greenwood’s formula and log-log transformation.

^eOS at DCO of June 30, 2022, was not multiplicity adjusted.

Source: AGILE Clinical Study Report.^38,39 Details included in the table are from the sponsor’s summary of clinical evidence.

Figure 3: Kaplan-Meier Plot of OS in the AGILE Study (FAS, DCO March 18, 2021)

AG-120 = ivosidenib; CI = confidence interval; DCO = data cut-off; FAS = full analysis set; OS = overall survival.

Source: AGILE Clinical Study Report.³⁸

Treatment Response

CR Rate

At the DCO of March 18, 2021, the CR rate was 47.2% in the ivosidenib plus azacitidine group and 14.9% in the placebo plus azacitidine group (odds ratio [OR] = 4.76; 95% CI, 2.15 to 10.50; 1-sided P < 0.0001). The between-group difference was 31% (95% CI, ██ ██ ███). For comparisons of the CR rate between ivosidenib plus azacitidine and placebo plus azacitidine, ████ ██████ ████ were NE and █████ ██████ █████ were not assessed because of a lack of postbaseline assessments.

CR Plus CRi Rate

At the DCO of March 18, 2021, the CR plus CRi rate was 54.2% in the ivosidenib plus azacitidine group and 16.2% in the placebo plus azacitidine group (OR = 5.90; 95% CI, 2.69 to 12.97; 1-sided P < 0.0001). The between-group difference in the CR plus CRi rate was 37% (95% CI, ██ ██ ███).

Detailed results of treatment response rates in the AGILE study are presented in Table 18.

HRQoL (Measured With EORTC QLQ C-30)

In the EORTC QLQ C-30, higher scores in the global health status subscale indicate better HRQoL.

At baseline, the mean scores for EORTC QLQ-C30 subscales were similar between the treatment groups (data not shown).

At 6, 12, and 18 months, the proportion of patients who were available to complete the HRQoL assessment was ██████ █████ ███ ████ of the FAS, respectively. The dropout rate was higher in the placebo plus azacitidine group than in the ivosidenib plus azacitidine group, which could be partially explained by there being more deaths in the former group.

Table 18: Summary of Treatment Response Rates in the AGILE Study (FAS, DCO March 18, 2021)

CI = confidence interval; CMH = Cochran-Mantel-Haenszel; CR = complete remission; CRi = complete remission with incomplete hematologic recovery; CRp = complete remission with incomplete platelet recovery; DCO = data cut-off; FAS = full analysis set; OR = odds ratio.

^aThe CI of the percentage is calculated with the Clopper and Pearson (exact binomial) method.

^bThe CMH estimate for the OR is calculated with placebo plus azacitidine as the control (denominator).

^cIf the primary analysis of EFS is significant, a stratified CMH test will be used to compare CR between the 2 treatment arms. The 1-sided P value is calculated from the CMH test stratified by the randomization stratification factors (acute myeloid leukemia status and geographic region); the stopping boundary was 1-sided P = 0.0087. CR plus CRi was not adjusted for multiplicity in the AGILE study.

Source: AGILE Clinical Study Report.³⁸ Details included in the table are from the sponsor’s summary of clinical evidence.

Table 19: Summary of Change From Baseline in Global Health Status/QoL Subscale of EORTC QLQ-C30 (FAS, DCO March 18, 2021)

CI = confidence interval; DCO = data cut-off; EORTC QLQ-C30 = European Organisation for Research and Treatment of Cancer Quality of Life Questionnaire Core 30; FAS = full analysis set; QoL = quality of life.

Notes: A 2-sided P value is reported. The least squares mean and 95% CI are estimated from the mixed effect model on the change from baseline across visits for all scales, with baseline score, treatment arm, time, randomization stratification factors (acute myeloid leukemia status and geographic region) and an interaction between treatment arm and time as fixed effects, and patient as a random effect. The unstructured covariance structure is used to define covariance between random effects. Unscheduled visits are excluded from the analysis.

^aThis outcome was not multiplicity adjusted.

Source: AGILE Clinical Study Report.³⁸ Details included in the table are from the sponsor’s summary of clinical evidence.

At 6, 12, and 18 months of treatment, patients in the ivosidenib plus azacitidine group reported increased global health status/QoL subscale scores from baseline; however, this trend was not observed in the placebo plus azacitidine group. The point estimates for the least squares mean between-group difference in this score exceeded the MIDs identified from the literature (a 3-point to 11-point increase indicates clinically meaningful improvement in this subscale). The MIDs identified from the literature were not specific to patients with AML.

Transfusion Requirement

At baseline, the proportion of patients who were transfusion dependent was 54.2% in the ivosidenib plus azacitidine group and 54.1% in the placebo plus azacitidine group. Among patients who were transfusion dependent at baseline, a higher proportion of patients receiving ivosidenib plus azacitidine (██ ███████ ████████) experienced RBC and/or platelet transfusion independence than those receiving placebo plus azacitidine (██ ███████ ████████) (OR ████ ███ ██ ███ ██ ████ ████████) at the DCO of March 18, 2021.

Similarly, at the DCO of June 30, 2022, a higher proportion of patients in the ivosidenib plus azacitidine group (██ ███████ ████████) experienced RBC and/or platelet transfusion independence than in the placebo plus azacitidine group (██ ███████ ████████) (OR ████ ███ ██ ███ ██ ██████████).

Table 20: Transfusion Requirements in the AGILE Study (FAS, DCO of March 18, 2021, and June 30, 2022)

CI = confidence interval; DCO = data cut-off; FAS = full analysis set; OR = odds ratio; RBC = red blood cell; RD = risk difference.

Note: The outcome of transfusion requirements was not adjusted for multiplicity.

^aDenominators are the number of patients who required transfusion at baseline.

Source: AGILE Clinical Study Report.^38,39 Details included in the table are from the sponsor’s summary of clinical evidence.

Harms

Summaries of safety data were presented by treatment group for the safety analysis set. In general, the safety results were consistent between the 2 DCO dates.

Adverse Events

As of the DCO of March 18, 2021, the proportion of patients who experienced at least 1 AE was 98.6% (70 patients) in the ivosidenib plus azacitidine group and 100% (73 patients) in the placebo plus azacitidine group. Patients treated with ivosidenib plus azacitidine were more likely (5% or more) to report the following TEAEs than patients treated with placebo plus azacitidine: vomiting (29 patients [40.8%] in the ivosidenib plus azacitidine group versus 19 patients [26.0%] in the placebo plus azacitidine group), neutropenia (20 [28.2%] versus 12 [16.4%]), thrombocytopenia (20 [28.2%] versus 15 [20.5%]), prolonged electrocardiogram QT interval (14 [19.7%] versus 5 [6.8%]), insomnia (13 [18.3%] versus 9 [12.3%]), differentiation syndrome (10 [14.1%] versus 6 [8.2%]), pain in extremity (10 [14.1%] versus 3 [4.1%]), hematoma (9 [12.7%] versus 1 [1.4%]), arthralgia (8 [11.3%] versus 3 [4.1%]), headache (8 [11.3%] versus 2 [2.7%]), leukocytosis (8 [11.3%] versus 1 [1.4%]), and leukopenia (6 [8.5%] versus 2 [2.7%]). In general, the severity of AEs was mild.

The following TEAEs were more likely (5% or more) to be reported in the placebo plus azacitidine group than in the ivosidenib plus azacitidine group: constipation (38 patients [52.1%] in the placebo plus azacitidine group versus 19 patients [26.8%] in the ivosidenib plus azacitidine group), pyrexia (29 [39.7%] versus 24 [33.8%]), febrile neutropenia (25 [34.2%] versus 20 [28.2%]), asthenia (24 [32.9%] versus 11 [15.5%]), pneumonia (23 [31.5%] versus 17 [23.9%]), hypokalemia (21 [28.8%] versus 11 [15.5%]), decreased appetite (19 [26.0%] versus 11 [15.5%]), edema peripheral (16 [21.9%] versus 8 [11.3%]), weight decrease (12 [16.4%] versus 4 [5.6%]), cough (11 [15.1%] versus 6 [8.5%]), and sepsis (6 [8.2%] versus 2 [2.8%]).

Grade 3 and higher AEs were reported in 66 patients (93.0%) in the ivosidenib plus azacitidine group and 69 patients (94.5%) in the placebo plus azacitidine group. In both groups, the commonly reported grade 3 and higher AEs were as follows (shown as ivosidenib plus azacitidine versus placebo plus azacitidine): anemia (25.4% versus 26.0%), febrile neutropenia (28.2% versus 34.2%), neutropenia (26.8% versus 16.4%), thrombocytopenia (23.9% versus 20.5%), and pneumonia (22.5% versus 28.8%).

As of the DCO of June 30, 2022, the number of patients who reported at least 1 AE was ██ ███████ in the ivosidenib plus azacitidine group and ██ ██████ in the placebo plus azacitidine group. Grade 3 and higher AEs were reported in ██ ███████ patients in the ivosidenib plus azacitidine group and ██ ███████ patients in the placebo plus azacitidine group. In both groups, the commonly reported grade 3 and higher AEs were as follows (shown as ivosidenib plus azacitidine versus placebo plus azacitidine): anemia (█████ ███ █████), febrile neutropenia (█████ ███ █████), neutropenia (█████ ███ █████), thrombocytopenia (█████ ███ █████), and pneumonia (█████ ███ █████).

Serious AEs

The proportion of patients who experienced SAEs was 69.0% (49 patients) in the ivosidenib plus azacitidine group and 82.2% (60 patients) in the placebo plus azacitidine group.

Commonly reported SAEs in the 2 treatment groups were febrile neutropenia (23.9% of patients in the ivosidenib plus azacitidine group versus 27.4% in the placebo plus azacitidine group) and pneumonia (19.7% versus 21.9%).

Results relating to SAEs at the DCO of June 30, 2022, were similar to those at the DCO of March 18, 2021.

Withdrawal Due to AEs

The overall incidences of TEAEs that led to combination treatment discontinuation were similar between the treatment groups: 19 patients (26.8%) in the ivosidenib plus azacitidine group and 19 patients (26.0%) in the placebo plus azacitidine group. Of the AEs leading to treatment discontinuation, only pulmonary embolism occurred in more than 1 patient (it occurred in 2 patients [2.8%] in the ivosidenib plus azacitidine group). The overall incidence of TEAEs that led to discontinuation of only ivosidenib or placebo was similar between the 2 groups (██████ patients in the ivosidenib plus azacitidine group versus ██████ in the placebo plus azacitidine group). TEAEs that led to discontinuation of only azacitidine occurred in █████ patients in the ivosidenib plus azacitidine group versus ██████ in the placebo plus azacitidine group and included febrile neutropenia (██████ patient per treatment group) and thrombocytopenia (██████ patient in the ivosidenib plus azacitidine group).

Mortality

Ten patients (14.1%) in the ivosidenib plus azacitidine group and 21 patients (28.8%) in the placebo plus azacitidine group died because of AEs during the study as of the DCO of March 18, 2021.

As of the DCO of June 30, 2022, ██ ████████ ███████ in the ivosidenib plus azacitidine group and ██ ████████ ███████ in the placebo plus azacitidine group died because of AEs during the study.

Notable Harms

Among the AEs of special interest reported by the sponsor, patients treated with ivosidenib plus azacitidine reported more cases of prolonged electrocardiogram QT intervals, leukocytosis, and differentiation syndrome (differentiation syndrome was identified as 1 of the most important AEs by the clinical experts consulted by the review team).

As of the DCO of June 30, 2022, differentiation syndrome was reported in 10 patients (13.9%) in the ivosidenib plus azacitidine group and 6 patients (8.1%) in the placebo plus azacitidine group. The majority of differentiation syndrome events occurring in patients in the ivosidenib plus azacitidine group were grade 2 (██████ █████████████), with only ██████ patients experiencing grade 3 differentiation syndrome. In the placebo plus azacitidine group, ██████ patients experienced grade 2 differentiation syndrome, ██████ patients experienced grade 3 differentiation syndrome, and ██████ patient experienced grade 4 differentiation syndrome.

Key harms at the DCOs of March 18, 2021, and June 30, 2022, are summarized in Table 21.

Table 21: Summary of Harms Results From the AGILE Study (SAS, DCO March 18, 2021, and June 30, 2022)

AE = adverse event; AESI = adverse event of special interest; DCO = data cut-off; SAS = safety analysis set; TEAE = treatment-emergent adverse event; WDAE = withdrawal due to adverse event.

Source: AGILE Clinical Study Report.^38,39 Details included in the table are from the sponsor’s summary of clinical evidence.

Hospitalizations due to AEs

At the DCO of March 18, 2021, the rates of hospitalization for TEAEs were similar for both treatment groups. The days of hospitalization per person-year of drug exposure were ███ for the ivosidenib plus azacitidine group and ███ for the placebo plus azacitidine group (Table 22).

Hospitalization due to other reasons was not assessed in the AGILE study.

Table 22: Hospitalizations for Adverse Events in the AGILE Study (SAS, DCO March 18, 2021)

DCO = data cut-off; SAS = safety analysis set.

Source: AGILE Clinical Study Report.³⁸ Details included in the table are from the sponsor’s summary of clinical evidence.

Critical Appraisal

Internal Validity

In the AGILE study, appropriate methods of randomization and allocation concealment were employed. The randomization schedule was prepared by an independent statistical group and stratified by de novo status and geographic region. The allocation was implemented using interactive response technology. Several baseline patient characteristics were balanced between the 2 treatment groups, for example demographic characteristics, disease characteristics, and prior anticancer therapy. The use of concomitant therapies and subsequent anticancer therapies was also generally balanced across the groups and consistent with the clinical practice in Canada. There were some imbalances in baseline patient characteristics between the 2 treatment groups, for example gender, WHO classification of AML, and cytogenetic risk status as assessed by the investigator. These imbalances are likely to be the result of the small sample size, within which prognostic balance is not likely to be assured; as such, there is some risk that the observed effects are overestimated or underestimated. In addition, the postbaseline transfusion independence outcome was measured among approximately half the population who required transfusions at baseline. Randomization is not necessarily upheld in this population. However, the results of transfusion requirement in patients who were dependent on transfusion at baseline did not differ significantly from those in the overall population. Therefore, the potential for bias is unlikely to have an important impact on the study findings specific to this outcome.

The study originally had no planned interim analyses. Observations of a notable difference in the number of deaths (favouring ivosidenib) by the IDMC prompted an unplanned interim analysis before the protocol-defined number of events. To control for multiplicity, new stopping boundaries were calculated based on the observed information fraction that were not outlined in the original statistical analysis plan. Because the results are from an unplanned interim analysis (which became the final analysis), even though the new stopping boundaries are appropriate, there is a risk of overestimation of the true effects of the study drug. Some of the important clinical outcomes were analyzed without multiplicity adjustment, for example HRQoL assessment using the EORTC QLQ-C30. As such, there would be an increased risk of false-positive conclusions (i.e., erroneously rejecting the null hypothesis); however, the reported results for these patient-reported outcomes were not statistically significant at later time points.

The patients, investigators, sponsor, and clinical research unit staff who deal directly with patients were blinded to treatment allocation until the final analysis for the primary end point unless emergency unblinding was required. Following the early interim analysis, the AGILE study was unblinded, and patients who received placebo plus azacitidine could switch to ivosidenib plus azacitidine.

HRQoL was assessed using a cancer-specific instrument. Even though the EORTC QLQ C-30 is not an AML-specific instrument and an MID for patients with AML was not identified from the literature, a range of potential between-group MIDs (3 to 11 points for improvement and –5 to –13 points for deterioration on the global QoL scale) were established based on clinical trials of 9 cancer types and may provide some guidance when determining the clinical relevance of the findings for HRQoL in the AGILE study. Even though no threshold of clinical importance could be estimated in patients with AML, the review team leaned on these MID ranges identified in other cancer types when assessing the GRADE imprecision domain for the EORTC QLQ C-30 results in the AGILE study. The completion rate of the EORTC QLQ C-30 was low. The completion rates were ██████ █████ ███ ████ at 6 months, 12 months, and 18 months of the study. Missing data were implicitly imputed within the mixed model, with the assumption of “missing at random.” However, there were no sensitivity analyses, and it is unlikely that the missing at random assumption is plausible. As a result, there is a high risk of bias because of the large amount of missing outcome data.

In the analysis of EFS, patients were censored if CR was documented by 24 weeks and 1 of the following occurred: the patients started subsequent anticancer therapy, relapse or death was documented after 2 or more missing disease assessments, the patient was lost to follow-up, withdrawal by patient, or the CR was ongoing without patient relapse or death. For patients who experienced CR by 24 weeks, no 1 was lost to follow-up. In the analysis of OS, patients were censored if they were alive or lost to follow-up or if they withdrew consent. The proportion of patients who were lost to follow-up was very low. Therefore, the effect of missing data on survival outcomes was not considered significant. For other binary end points, such as CR and CR plus CRi, there appeared to be a large amount of data missing, labelled as “not assessed”: █████ in the ivosidenib plus azacitidine group versus █████ in the placebo plus azacitidine group were not assessed because of a lack of postbaseline assessments for CR assessment. Subsequently, a high risk of bias may be introduced with unclear direction; no reason for the missing data was reported.

EFS was the primary efficacy outcome in this study. This is a composite end point, which was defined as the time from randomization until treatment failure (i.e., patient does not experience CR by week 24), relapse from remission, or death from any cause, whichever occurred first. In the AGILE study, almost all events occurred at baseline (i.e., 1 component of the end point). As such, there were few patients left at risk postbaseline; as a result, the EFS could not robustly characterize the long-term efficacy of the study drug.³¹ The correlations between EFS and OS were modest in the published research that provided trial-level information. However, 1 major limitation of these studies was that they were not specific to the population nor the drug class of interest, and therefore the ability to generalize the study findings was not clear.^32-34

Predefined sensitivity analyses were conducted to evaluate the robustness of the primary EFS results. Overall, the results of the sensitivity analyses were generally aligned with the primary analysis of EFS, which supported the robustness of the results. Prespecified subgroup analyses generally supported consistency in the overall direction of effect of ivosidenib across subgroups; some subgroups were small, resulting in wide CIs.

There was low risk of selective reporting bias; all presented end points had been specified in the statistical analysis plan; however, as mentioned previously, the interim analysis was unplanned.

As described in the Outcomes section, OS was the primary efficacy end point at the beginning of this trial, but it was replaced with EFS. The sponsor’s justification was that this change allowed for a smaller sample size and therefore a more feasible trial in this rare patient population. Furthermore, EFS was considered by the sponsor to more accurately describe the contribution of a novel therapy to clinical benefit by removing the potentially confounding effects of posttrial therapies and by capturing treatment failure as an event. Meanwhile, previous research provided encouraging preliminary safety and efficacy data when comparing an IDH1 inhibitor plus azacitidine with azacitidine alone. All these factors supported the amendment of the protocol of the AGILE study to use EFS as a measure of clinical benefit for the treatment of patients with AML who are ineligible for intensive induction chemotherapy. Moreover, this change was done before unblinding of the data and therefore was not likely to bias the study results.

External Validity

According to feedback from the clinical experts consulted for this review, the eligibility criteria and baseline characteristics of the patients randomized in the AGILE study generally reflected a patient population in Canadian clinical practice that would receive combination therapy of ivosidenib plus azacitidine. The clinical experts noted that the results from the AGILE study could be generalized to patients with IDH1-mutated AML in Canada who would be treated with ivosidenib plus azacitidine. The clinical experts also indicated that in clinical practice, ECOG performance status criteria would not always be used; in addition, some flexibility should be applied in terms of using ivosidenib plus azacitidine in patients with slightly worse ECOG performance status than in the trial. The potential benefits and risks of this treatment for individual patients need to be assessed. Patients’ IDH1 mutation status should be confirmed before the treatment. The experts indicated that the outcome measures in the AGILE study were generally appropriate and clinically relevant for clinical trials of AML.

In the AGILE study, ivosidenib in combination with azacitidine was compared with azacitidine monotherapy. The clinical experts consulted for this review indicated that azacitidine alone is not the most appropriate comparator for the study drug combination in the study population. Instead, venetoclax plus azacitidine is currently the most commonly used combination therapy in the target patient population.

In practice, monotherapy with azacitidine would typically be used for patients with increased frailty that would make treatment with the combination of venetoclax and azacitidine unreasonable. There is a lack of direct evidence within the AGILE study with which to examine the relative efficacy and safety of the study drug compared with other combination regimens.

GRADE Summary of Findings and Certainty of the Evidence

Methods for Assessing the Certainty of the Evidence

For pivotal studies and RCTs identified in the sponsor’s systematic review, GRADE was used to assess the certainty of the evidence for the outcomes considered most relevant to inform the expert committee deliberations, and a final certainty rating was determined, as outlined by the GRADE Working Group.^35,36

• High certainty: We are very confident that the true effect lies close to that of the estimate of the effect.

• Moderate certainty: We are moderately confident in the effect estimate — The true effect is likely to be close to the estimate of the effect, but there is a possibility that it is substantially different. We use the word “likely” for evidence of moderate certainty (e.g., “X intervention likely results in Y outcome”).

• Low certainty: Our confidence in the effect estimate is limited — The true effect may be substantially different from the estimate of the effect. We use the word may for evidence of low certainty (e.g., “X intervention may result in Y outcome”).

• Very low certainty: We have very little confidence in the effect estimate — The true effect is likely to be substantially different from the estimate of effect. We describe evidence of very low certainty as “very uncertain.”

Following the GRADE approach, evidence from RCTs started as high-certainty evidence and could be rated down for concerns related to study limitations (which refer to internal validity or risk of bias), inconsistency across studies, indirectness, imprecision of effects, and publication bias.

When possible, certainty was rated in the context of the presence of an important (nontrivial) treatment effect; if this was not possible, certainty was rated in the context of the presence of any treatment effect (i.e., the clinical importance is unclear). In all cases, the target of the certainty of evidence assessment was based on the point estimate and where it was located relative to the threshold for a clinically important effect (when a threshold was available) or to the null. The threshold for a clinically important effect for OS and EFS in the study population was not obtained. Therefore, the target of the certainty of evidence assessment was the presence of absence of any (non-null) effect for survival rates. The threshold for a clinically important effect for the EORTC QLQ-C30 score was set according to the presence or absence of an important effect based on thresholds identified in the literature.³⁷ In addition, the target of the certainty of evidence assessment was the presence or absence of any non-null effect for CR, CR plus CRi, and transfusion requirements. For some harm events (e.g., differentiation syndrome), because of the unavailability of the absolute difference in effects, the certainty of evidence was summarized narratively.

Results of GRADE Assessments

Table 2 presents the GRADE summary of findings for ivosidenib plus azacitidine versus placebo plus azacitidine.

Long-Term Extension Studies

There were no relevant long-term extension studies submitted for this review.

Indirect Evidence

Contents within this section have been informed by materials submitted by the sponsor. The following has been summarized and validated by the review team.

Objectives for the Summary of Indirect Evidence

Aside from the comparison to placebo plus azacitidine, there was no direct evidence comparing ivosidenib plus azacitidine against other relevant comparators for the treatment of newly diagnosed or untreated patients with IDH1-mutated AML; therefore, a review of indirect evidence was undertaken and submitted by the sponsor.⁵⁸ A research protocol is not available for this study; however, detailed selection criteria and methods of analyses were provided in the ITC report submitted by the sponsor.

Description of Indirect Comparisons

Objectives

The objective of the submitted ITC report was to derive estimates of the relative efficacy of ivosidenib plus azacitidine versus existing therapies for patients with treatment-naive or newly diagnosed AML with IDH1 mutation who are ineligible for intensive chemotherapy, by means of either an NMA or an MAIC.

Study Selection Methods

A summary of the study selection criteria and methodology used to conduct the systematic review contributing to the ITCs is given in Table 23. Overall, clinical trials of adult patients with newly diagnosed or untreated AML who were ineligible for intensive chemotherapy were included. Eligible patients could include those aged 75 years or older; had severe heart, pulmonary, liver, or renal disorders; or had a greater than 20% blast count. Treatments of interest included ivosidenib plus azacitidine, venetoclax plus azacitidine, azacitidine monotherapy, LDAC, venetoclax plus LDAC, and glasdegib plus LDAC. The outcomes considered in this ITC report included OS, EFS, treatment response rates, transfusion independence, and transfusion burden, which were deemed most important to capture and convey the treatment benefit to payers and clinicians and to provide outputs amenable for economic modelling. Multiple databases, conference abstracts, and trial registries were searched to identify relevant evidence. The search was last updated on January 31, 2023. Study selection was carried out by 2 reviewers independently. Data was extracted using a standardized data extraction form; however, it was unclear if this was completed by the 2 independent reviewers. Risk of bias in the included RCTs was assessed at the study level using the Cochrane Risk of Bias tool v2.0; the number of reviewers who contributed was not reported.

Table 23: Study Selection Criteria and Methods for Systematic Review Contributing to the ITCs Submitted by the Sponsor

AML = acute myeloid leukemia; CR = complete remission; CRh = complete remission with partial hematologic recovery; CRi = complete remission with incomplete hematologic recovery; DOR = duration of response; ECOG PS = Eastern Cooperative Oncology Group performance status; EFS = event-free survival; FEV₁ = forced expiratory volume in 1 second; ICTRP = International Clinical Trials Registry Platform; ITC = indirect treatment comparison; LDAC = low-dose cytarabine; LVEF = left ventricular ejection fraction; OS = overall survival; PICOS = population, intervention, comparison, outcomes and study; RCT = randomized controlled trial; SLR = systematic literature review; ULN = upper limit of normal

^aVenetoclax plus LDAC and glasdegib plus LDAC combinations are not recommended for reimbursement in Canada; however, they were included in the analyses for consistency with the SLR strategy and the ITC PICOS criteria.

Source: Sponsor-submitted ITC report.⁵⁸ Details included in the table are from the sponsor’s summary of clinical evidence.

Feasibility Appraisal

Before the analysis, a comprehensive feasibility assessment was conducted to verify whether an ITC could be made. This assessment looked at the ability to pool across studies within each treatment group and the presence and extent of between-study heterogeneity. A rationale for conducting an NMA and various MAICs was not provided in this ITC analysis.

ITC Analysis Methods

Network Meta-Analyses

The AGILE study was the only study comparing treatments exclusively in patients with IDH1 mutation. The relative efficacy of ivosidenib plus azacitidine versus existing therapies was therefore estimated for adults with previously untreated AML ineligible for intensive chemotherapy irrespective of mutation status, using an NMA for the outcomes of OS, EFS, CR rates, CR plus CRi rates, CR plus CRh rates, and transfusion independence. In addition to the NMA, 3 MAICs were conducted in comparison with venetoclax plus azacitidine to account for population imbalances: an anchored MAIC of OS for ivosidenib plus azacitidine versus venetoclax plus azacitidine in the ITT population of the VIALE-A study; an unanchored MAIC of OS for ivosidenib plus azacitidine versus the venetoclax plus azacitidine group from the IDH1-mutated subgroup in the VIALE-A study; an anchored MAIC of EFS for ivosidenib plus azacitidine versus venetoclax plus azacitidine in the ITT population of the VIALE-A study.

A summary of the NMA methods is presented in Table 24.

Table 24: Network Meta-Analysis Methods

BGR = Brooks-Gelman-Rubin; CR = complete remission; CRi = complete remission with incomplete hematologic recovery; CrI = credible interval; DIC = deviance information criterion; EFS = event-free survival; HR = hazard ratio; OR = odds ratio; OS = overall survival.

Source: Sponsor-submitted indirect treatment comparison.⁵⁸ Details included in the table are from the sponsor’s summary of clinical evidence.

Analysis Framework

Analyses were run in a Bayesian framework.⁶⁶

Assessment of Convergence

Under a Bayesian approach, posterior densities for the unknown parameters are estimated using Markov chain Monte Carlo simulations for each model. The proposed analyses were based on a burn-in of (at least) 20,000 iterations and a further sample of (at least) 40,000 iterations or until convergence was achieved. Convergence was assessed by checking the Brooks-Gelman-Rubin diagnostic in OpenBUGS. In addition, the Monte Carlo error was captured, which reflects both the number of simulations and the degree of autocorrelation. This should be no more than 5% of the posterior standard deviation of the parameters of interest. Finally, visual inspection of trace-density plots was carried out.

As suggested by the National Institute for Health and Care Excellence (NICE) Decision Support Unit (DSU) report,⁶⁶ a normal distribution with zero mean and variance equal to 104 was used for treatment effects and a uniform distribution with range zero to 2 was used for the between-trial standard deviation. Vague (flat/uninformative) priors were used for all calculations.

Inconsistency Assessment

As none of the evidence networks for OS, EFS, CR, CR + CRi, or transfusion requirement have closed loops, a consistency assessment as per NICE DSU Technical Support Document 4 was not possible.⁶⁷

Model Selection

The conducted analyses consisted of binary outcomes (CR, CR plus CRi, transfusion independence) and time-to-event outcomes (hazard rates for OS and EFS). A binomial model with a logit link function was employed for binary outcomes and a normal model with an identity link function was employed for time-to-event outcomes, based on NICE guidance.⁶⁶

Both fixed effects and random effects models were considered for each analysis, and results from each model were run. However, only 1 model was chosen from which to draw inferences. The deviance information criterion was used to choose the appropriate model for the data.⁶⁶

Following the feasibility assessment, meta-regression was not carried out to adjust for differences in study-level effect modifiers because of a lack of data.

Presentation of Results

For categorical outcomes, ORs were used to reflect the relative treatment effects between interventions; for time-to-event outcomes, HRs were used. Forest plots are presented using the posterior median of OR or HR for each pairwise treatment comparison. The 2.5th and 97th percentiles to capture the 95% CrI of OR or HR are also provided. For time-to-event outcomes (OS and EFS), a median HR less than 1 indicates favourable results for ivosidenib plus azacitidine. For categorical outcomes, a median OR greater than 1 indicates favourable results for ivosidenib plus azacitidine.

Matching-Adjusted Indirect Comparisons: The VIALE-A study is a phase III RCT comparing venetoclax plus azacitidine with placebo plus azacitidine in patients with treatment-naive AML.⁵⁵ A study by Pollyea et al.⁵⁶ pooled data from the VIALE-A study and a prior phase Ib study (single-arm study of patients with treatment-naive AML, investigating the safety and pharmacokinetics of venetoclax combined with decitabine or azacitidine). None of these studies was specific to patients with IDH1-mutated AML. Since an anchored MAIC was not feasible to adjust for within-study imbalances in potential effect modifiers in these studies, an unanchored MAIC for OS was conducted for the comparison with the IDH1 mutation subgroup, where the baseline characteristics of patients in the ivosidenib plus azacitidine group in the AGILE study were matched to the baseline characteristics of patients in the venetoclax plus azacitidine group in Pollyea et al.⁵⁶ The baseline characteristics in Pollyea et al. reflect the population with IDH1/2 mutations, as the baseline characteristics for patients with IDH1 mutations were not reported in the VIALE-A study or in Pollyea et al. For the comparisons of the OS and EFS outcomes with the ITT population from the VIALE-A study, an anchored MAIC was feasible and therefore conducted.

Identification of Effect Modifiers and Prognostic Variables: The final list of treatment effect modifiers to be adjusted for in the MAIC analyses was determined through a deliberative process that considered statistical analyses conducted using the AGILE study individual patient data, and a review of the effect modifiers identified in a previous published MAIC,⁶⁸ and a simulated treatment comparison⁶⁹ of therapies in the indication of interest. Input from the clinical experts consulted by the sponsor was also sought to validate the covariate selection process.

Quantitative Analysis

Analyses were performed using the AGILE study individual patient data for a wide set of variables identified through the literature as potential prognostic variables and/or effect modifiers for first-line treatment of AML.^68,69 For the end points of interest (i.e., OS and EFS), the following variables that were commonly reported in both studies were considered and assessed:

• Median age

• Gender

• ECOG PS

• Type of AML

• Intermediate cytogenetic risk

• Poor cytogenetic risk

• Bone marrow blasts

• IDH1 mutation based on central testing

Multivariable regression models were fitted for the OS and EFS end points, including all potential prognostic variables as covariates. For OS and EFS, a Cox PH model was fitted. Variable selection was then performed based on the statistical significance of each prognostic variable in the model. Specifically, statistical testing to assess prognostic variable status consisted of likelihood ratio tests between a Cox model, with the variable of interest as a covariate, and a null intercept-only model. Given the relatively small sample size of the ivosidenib plus azacitidine group in the AGILE study (i.e., 72 patients), the classical significance threshold (i.e., 5%) was relaxed, and more conservative significance levels of 10%, 15%, and 20% were used. The most appropriate threshold was discussed and agreed with clinical experts consulted by the sponsor. This method violates the recommendations of the NICE DSU that the list of variables be identified before the analysis.

In addition to the likelihood ratio test, a stepwise selection process was used for the covariate selection, where the model with the lowest Akaike Information Criterion value was considered the best fitting. For the selection of the effect modifiers, the same process was followed as that described for prognostic variables, with the only difference being that the interaction between covariates and the treatment was explored.

Covariates Considered in Previous Studies

In previously published MAICs for AML,^68,69 the following covariates were adjusted for: age, AML type, bone marrow blast count, cytogenetic risk, ECOG PS, gender, neutrophil count, platelet count, poor cytogenetic risk category, and response status.

Estimation of MAIC Weights: To enable an adjusted comparison between ivosidenib plus azacitidine and the available comparative evidence sources, individual patients in the AGILE study were assigned statistical weights that adjust for their overrepresentation or underrepresentation relative to the average prognostic variables observed in the comparative evidence source. As a result, after weighting, the average baseline characteristics (mean and variance or proportion of patients within a category) were balanced for the patients in the AGILE study and the comparator populations.

Weights were derived using a form of propensity score weighting. In the absence of patient-level data for the comparative evidence source, a method of moments and the quasi-Newton optimization Broyden-Fletcher-Goldfarb-Shanno algorithm was used to allow a propensity score logistic regression model to be estimated and ensure that the weights balanced the mean covariate values.

Following the estimation of the weights, the distribution of the rescaled weights was visually examined to determine whether specific patient(s) or groups of patients (based on covariate values) are overrepresented or underrepresented in the analysis.

The robustness of the analyses was also evaluated by approximating the ESS.

Missing Data: During the matching process, the estimation of patient-specific weights required that matching covariates were available for all patients enrolled in the AGILE study. Missing data (if any) for patients in the AGILE study were identified once the covariates to adjust for had been selected. Specifically, 1 patient in the ivosidenib plus azacitidine group and 1 patient in the azacitidine group did not report the percentage of bone marrow blasts; thus, in scenarios where this covariate was used in the matching process, these patients were removed from the initial sample size.

Statistical Analyses Incorporating MAIC Weights: After the matching procedure was conducted and the weights were derived, efficacy outcomes were compared between balanced treatment groups using analyses that incorporate the derived weights. For the OS and EFS end points, a reweighted relative treatment effect (and standard error) for ivosidenib plus azacitidine versus the relevant comparator treatments was estimated using the reweighted absolute effect of ivosidenib plus azacitidine and the reported absolute effect of the relevant comparator treatment. The same statistical approach was followed in the anchored and unanchored cases, with the only difference being that the relative treatment effect of ivosidenib plus azacitidine versus venetoclax plus azacitidine was established via the common comparison against azacitidine in the anchored case.

Model Fitting and Model Selection: For survival outcomes (OS and EFS), the assumption of PH was assessed by visually inspecting the log-cumulative hazard plots for nonlinearities and by inspecting the Schoenfeld residuals. HRs were obtained by fitting a weighted Cox PH model whenever the PH assumption was met. When the PH assumption did not hold, survival models were fitted to the original and weighted AGILE study data as well as to the digitized comparator data. Alternative survival parametric models, including exponential, Weibull, log-logistic, log-normal, Gompertz, and generalized gamma distributions, were fitted to the weighted AGILE study and the digitized comparator data. Model selection included visual comparison as well as calculation of the Akaike Information Criterion and the Bayesian Information Criterion, where lower values indicated better fit to the data.

Results of the Feasibility Assessment

Ten studies were identified through the systematic literature review for the ITCs and were included in the feasibility assessment.

Key considerations in the feasibility assessment included the availability of the outcomes of interest, study design, characteristics of patient populations, posology of evaluated interventions, definitions, and methods for ascertainment of outcomes. Several limitations for the ITCs were identified by the feasibility assessment:

• None of the comparator studies were conducted in the target population (specifically, in relation to IDH1 mutation).

• In studies reporting mutation subgroup data, IDH1 is based on post hoc analyses with small patient numbers, and IDH1 is not a stratification factor for randomization in those studies.

• Population baseline characteristics for the IDH1 subgroup are not available for venetoclax plus azacitidine; the IDH1/2 baseline characteristics in Pollyea et al. are unbalanced between treatment arms.

• Notable differences in placebo arm rates are observed across placebo-controlled studies (i.e., the AGILE study and the IDH1 mutation subgroup from the VIALE-A study as reported in Pollyea et al.), which suggest differences in populations across the studies and raise concerns about outcome homogeneity.

The feasibility assessment identified heterogeneity in the analysis populations arising from a lack of published subgroup data for patients with IDH1 mutation, heterogeneity in other patient demographic and disease characteristics (gender, type of AML, cytogenic risk, ECOG performance status, and median bone marrow blast), differences in placebo arm rates across placebo-controlled studies, and differences in the definition of EFS.

Table 25: Assessment of Homogeneity of Studies Included in the ITCs

AML = acute myeloid leukemia; CR = complete remission; CRi = complete remission with incomplete hematologic recovery; ECOG PS = Eastern Cooperative Oncology Group performance status; EFS = event-free survival; ITC = indirect treatment comparison; LDAC = low-dose cytarabine; OS = overall survival; RBC = red blood cell; RCT = randomized controlled trial.

Source: Sponsor-submitted ITC report.⁵⁸ Details included in the table are from the sponsor’s summary of clinical evidence.

Results of the NMA

Following the feasibility assessment, 6 studies were selected to contribute to the evidence networks for 6 outcomes of interest and the following treatments:

• ivosidenib plus azacitidine

• azacitidine

• LDAC

• decitabine

• venetoclax plus azacitidine

• venetoclax plus LDAC

• glasdegib plus LDAC.

Results specific to patients with IDH1 mutation are reported only for venetoclax plus azacitidine in the VIALE-A study by DiNardo et al. and a pooled analysis by Pollyea et al. (pooled data from the VIALE-A study and a single-arm phase Ib study) but are based on post hoc subgroup analyses with small sample sizes (specifically, fewer than 20 patients with IDH1 mutation were enrolled in the azacitidine group in the VIALE-A study, which does not meet the sample size inclusion criterion in the feasibility assessment).

Studies were excluded from the NMA if there were serious quality concerns (results of risk of bias assessment are presented in Appendix 1), if outcomes of interest were not reported, or if there were highly uncertain study results. During the development of the NMA, new data cuts from the AGILE study (June 30, 2022; median follow-up: 28.6 months) and the VIALE-A study (December 1, 2021; median follow-up: 43.2 months) were available for the analysis of OS. The newest data cut from the VIALE-A study was ultimately published early in 2024,⁷¹ and data from the new data cut were used in updated data analyses of OS.

Of the 6 studies included in the NMA, 3 were considered to be at low risk of bias (DiNardo et al. [2020]; Dombret et al. [2015]; Wei et al. [2021]), concerns of risk of bias applied to 2 (Heuser et al. [2021]; Kantarjian et al. [2012]), and the AGILE study was not assessed.

The diagrams for the network of evidence for OS, EFS, CR, CR plus CRi, and transfusion independence are presented in Figures 4 to 8.

Figure 4: Network of Evidence for Overall Survival

LDAC = low-dose cytarabine.

Source: Sponsor-submitted indirect treatment comparison.⁵⁸

Results

Network Meta-Analyses

Both fixed-effect and random-effect models were fitted to the data, and the fixed-effect model was preferred across all analyses in this NMA.

For the comparison of ivosidenib plus azacitidine to venetoclax plus azacitidine and venetoclax plus LDAC, associated CrIs were wide, suggesting uncertainty about which regimen could be favoured. Results suggest that ivosidenib plus azacitidine was favoured over LDAC monotherapy for all outcomes and that azacitidine monotherapy was favoured for all outcomes except for transfusion independence (the CrI was wide).

Detailed results from the NMA are presented in Table 26.

Figure 5: Network of Evidence for Event-Free Survival

LDAC = low-dose cytarabine.

Source: Sponsor-submitted indirect treatment comparison.⁵⁸

Figure 6: Network of Evidence for Complete Remission

LDAC = low-dose cytarabine.

Source: Sponsor-submitted indirect treatment comparison.⁵⁸

Figure 7: Network of Evidence for CR Plus CR With Incomplete Hematologic Recovery

CR = complete remission; LDAC = low-dose cytarabine.

Source: Sponsor-submitted indirect treatment comparison.⁵⁸

Figure 8: Network of Evidence for Transfusion Requirements

LDAC = low-dose cytarabine.

Source: Sponsor-submitted indirect treatment comparison.⁵⁸

Table 26: Summary of Efficacy Outcome Measures in the Sponsor-Submitted ITCs, Results From NMA, Fixed-Effect Models

CR = complete remission; CRi = complete remission with incomplete hematologic recovery; Crl = credible interval; EFS = event-free survival; HR = hazard ratio; ITC = indirect treatment comparison; ITT = intention to treat; LDAC = low-dose cytarabine; NMA = network meta-analysis; OR = odds ratio; OS = overall survival.

Notes: Venetoclax plus LDAC is not recommended for reimbursement in Canada. Bolded values indicate statistically significant differences between the 2 treatments. An HR less than 1 indicates “favours ivosidenib” for OS and EFS; an OR less than 1 indicates “favours comparator” for CR, CR plus CRi, and transfusion independence.

Source: Sponsor-submitted indirect treatment comparison.⁵⁸

Matching-Adjusted Indirect Comparisons

During the MAIC analyses, each covariate selection approach identified different sets of covariates for inclusion in the MAIC analysis per outcome of interest. Three scenarios have been proposed to determine the covariates to adjust for in the matching process:

• unanchored MAIC for OS: matching the AGILE study to the IDH1 subgroup in the VIALE-A study, with matching based on the IDH1/2 population described in the Pollyea et al. study

• anchored MAIC for OS: matching the AGILE study to the ITT population from the VIALE-A study

• anchored MAIC for EFS: matching the AGILE study to the ITT population from the VIALE-A study.

The rationale for matching against the IDH1 population in the VIALE-A study is that, despite the limitations of the data, this population reflects the target population for decision-making and is hence presented even if the results should be interpreted with great caution. The rationale for matching against the ITT population in the VIALE-A study, which includes a broader population of patients than in the AGILE study, is to explore whether IDH1 mutation status is an effect modifier for venetoclax plus azacitidine and to overcome the main limitation of the IDH1 mutation data as they are based on a post hoc subgroup with small sample size, where randomization is broken with explained and unexpected effect modification.

The baseline characteristics in the AGILE study before and after matching to the IDH1 population or ITT population are presented in Appendix 1, Tables 30 to 33. There may not be adequate information to determine if the matching is adequate when only the baseline characteristics on which the sponsor matched were provided, rather than all baseline characteristics (which could have become unbalanced during the matching of the other variables). In addition, for the ITT population, it was not possible to match on IDH1 status, because all patients in the AGILE study have an IDH1 mutation.

In the unanchored MAIC for OS in the IDH1 population, the following were matched for the base-case analysis: age, sex, ECOG performance status, AML type, cytogenetic risk, and bone marrow blasts; age and bone marrow blasts were matched for scenario analysis 1; and age, bone marrow blasts, and ECOG performance status were matched for scenario analysis 2.

In the anchored MAIC for OS in the ITT population, the following were matched for the base-case analysis: age, sex, ECOG performance status, AML type, cytogenetic risk, and bone marrow blasts; bone marrow blasts were matched for scenario analysis 1.

In the anchored MAIC for EFS in the ITT population, the following were matched for the base-case analysis: age, sex, ECOG performance status, AML type, cytogenetic risk, and bone marrow blasts; sex and ECOG performance status were matched for scenario analysis 1; and sex, cytogenetic risk, and ECOG performance status were matched for scenario analysis 2.

The results from the unanchored MAIC for OS in the IDH1 population showed that after matching (base case), the median OS was ████ ██████ ████ ██ ████ ██ █████ with ivosidenib plus azacitidine, compared to ████ ██████ ████ ██ ███ ██ ███ ████████ with venetoclax plus azacitidine.

The results from the anchored MAIC for OS in the ITT population showed that after matching, the median OS was ████ ██████ ████ ██ ███ ██ █████ with ivosidenib plus azacitidine, compared to ████ ██████ ████ ██ ███ ██ ███ ████████ with venetoclax plus azacitidine.

The results from the anchored MAIC for EFS in the ITT population showed that after matching, the median EFS was ████ ██████ ████ ██ ███ ██ ███ ████████ with ivosidenib plus azacitidine, compared to ███ ██████ ████ ██ ███ ██ █████ with venetoclax plus azacitidine.

Detailed results from the MAIC are presented in Table 27.

Table 27: Median Overall Survival and Event-Free Survival Before and After Matchings

BC = base case; CI = confidence interval; EFS = event-free survival; ESS = effective sample size; HR = hazard ratio; ITT = intention to treat; MAIC = matching-adjusted indirect comparison; NA = not applicable; NR = not reached; OS = overall survival; SA = scenario analysis.

Source: Sponsor-submitted indirect treatment comparison.⁵⁸

Critical Appraisal of ITCs

There was no a priori protocol for the ITCs; therefore, it cannot be known whether the analyses presented were selected from multiple analyses of the data (e.g., based on the magnitude and direction of observed effects). In this ITC report, studies were identified by searching multiple databases based on prespecified inclusion and exclusion criteria. Studies were selected by 2 independent reviewers; thus, the error and bias in the study selection process were minimized. Appropriate methods were used to reduce the risk of bias and error in data extraction. It was unknown if the risk of bias in the included trials was assessed by the 2 independent reviewers. In addition, risk of bias was assessed at the level of the trial, rather than at the level of the reported results (i.e., per outcome), which ignores that risk of bias can vary by reported result within a trial. Some of the studies included within the NMA had some potential for risk of bias. Risk of bias in the AGILE study was not assessed.

One of the major concerns for the ITCs is that the included trials could have been highly heterogeneous in terms of study design and patient characteristics at baseline. Six RCTs were included in the NMA. Heterogeneities were identified in the analysis populations, which included IDH1 mutation status, gender, type of AML diagnosis, cytogenic risk, performance status, median bone marrow blast, differences in placebo arm rates across placebo-controlled studies, and differences in the definition of EFS. For the time-to-event comparisons (e.g., EFS), lengths of follow-up were different, and with longer follow-up it may be expected that the HR would be attenuated, even without formal violation of the PH assumption. The bias would likely favour the study drug. These differences would undermine the validity of the NMA, which relies on the transitivity assumption (i.e., that the trials are similar for all important effect modifiers) being upheld. The use of fixed-effect models was chosen based on the deviance information criterion. However, the use of fixed-effect models (assuming no between-study heterogeneity) rather than random effects models means that the CrIs are unlikely to adequately express the uncertainty arising from the heterogeneity. The limited number of included studies did not allow for meta-regression or other techniques to adjust for differences in effect modifiers across studies within the NMA. The rarity of the population of interest limits the size and number of clinical studies completed with potential comparators and adds to the practical challenges when indirectly comparing treatment options.

In the NMA, given the lack of closed loops in the networks, consistency in the ITC analyses could not be tested. All comparisons are therefore informed only by indirect evidence, which increases the level of uncertainty. Efficacy data were sparse in this NMA for the comparison of ivosidenib versus placebo in combination with azacitidine. The 95% CrIs for the point estimates were wide for some efficacy outcomes and spanned the null when comparing with other combination regimens; therefore, confidence in the effect estimates for efficacy of the study drugs was limited because of the imprecision indicated by the wide CrIs for these outcomes, which precludes any conclusions as to which treatment may be favoured.

In the MAICs, the following potential effect modifier or prognostic factors were identified through the literature and a deliberating process by the sponsor: age, gender, ECOG performance status, type of AML, cytogenetic risk of AML, bone marrow blasts, and IDH1 mutation. The clinical experts consulted for this review agreed that these are relevant effect modifiers and prognostic variables. However, it is unclear if the identification of potential effect modifiers through the literature would be sufficient to identify all relevant treatment effect modifiers. The populations in the AGILE study and the other comparator studies were weighted and matched. Within the unanchored MAIC there was no reported estimate of the potential residual bias due to unadjusted confounders; as a result, the magnitude of residual confounding remains uncertain.

Before adjustment, the median OS and EFS for the placebo plus azacitidine groups were substantially different, suggesting reduced comparability of the populations. The main differences for the 2 studies used (AGILE and VIALE-A) is that in the AGILE study, the patients were younger and had a better ECOG performance status and a lower proportion of the patients had high-risk cytogenic status. The unanchored MAIC matched the characteristics of the patients with IDH1/2 mutation from the VIALE-A study because the characteristics for IDH1 were unavailable. In the anchored MAICs, the ESS reduced by approximately one-third after the weighting process, suggesting that the results are heavily influenced by a subset of the sample population in the trial who may not be representative of the full sample population. The reduction in the ESS and the sample size in general resulted in wide CIs. Furthermore, there is uncertainty about comparing the population with IDH1 mutation to the ITT population in the VIALE-A study. It was not possible to adjust for this factor.

The study population for this review is patients with AML with IDH1 mutation who are ineligible for intensive chemotherapy. However, most of the selected trials were not specifically for IDH1-mutated AML. No other studies included only patients with IDH1 mutation, and it is not clear in the other included trials whether there were separate results for this particular subgroup. The prognostic significance of IDH1 status in AML, or whether this IDH1 status may be a treatment effect modifier, remains uncertain. According to the clinical experts consulted for this review, the aforementioned patient characteristics (e.g., de novo AML status, region, age, baseline ECOG performance status, sex, race, baseline cytogenetic risk status, WHO classification of AML, baseline white blood cell count, and baseline percent bone marrow blasts) were considered treatment effect modifiers in patients with AML and IDH1-mutated AML.

In this ITC report, several efficacy outcomes were analyzed, such as OS, EFS, and CR (not evaluated in the MAICs). However, other efficacy end points of interest to patients and clinicians (e.g., HRQoL), as well as harms, were not investigated. Therefore, the relative treatment effect of ivosidenib plus azacitidine versus relevant comparators on patients’ HRQoL and on harms remains unknown.

Studies Addressing Gaps in the Systematic Review Evidence

There were no relevant studies addressing the gaps in the systematic review evidence submitted for this review.

Discussion

Summary of Available Evidence

The evidence included in the systematic review consisted of 1 pivotal phase III, double-blind RCT, the AGILE study (N = 146). The purpose of this study was to evaluate the efficacy and safety of the combination of ivosidenib plus azacitidine versus placebo plus azacitidine in adult patients with newly diagnosed AML with an IIDH1 R132 mutation who are not eligible to receive intensive induction chemotherapy. Patients were randomized either to ivosidenib 500 mg orally once daily plus azacitidine 75 mg/m²/day, subcutaneous or IV, for 7 days, in 28-day cycles, or to placebo in combination with azacitidine. The primary efficacy end point in the AGILE study was EFS. Other relevant outcomes in this study included OS, remission rates, HRQoL measured by the EORTC QLQ-C30, transfusion requirement, and harms. The majority (73.3% per investigator [76% per Interactive Web Response System]) of patients had de novo AML at initial diagnosis. Based on the WHO classification of AML, fewer patients in the ivosidenib plus azacitidine group (22.2%) had AML with recurrent genetic abnormalities than in the placebo plus azacitidine group (32.4%), and more patients in the ivosidenib plus azacitidine group (38.9%) had AML with myelodysplasia-related changes than in the placebo plus azacitidine group (35.1%). IIDH1 R132C was the most common polymorphism (65.8% of patients). In total, 63.8% of patients in the ivosidenib plus azacitidine group and 67.6% of patients in the placebo plus azacitidine group had an ECOG performance status score of 0 to 1. Cytogenetic risk status, as assessed by the investigators based on the 2017 National Comprehensive Cancer Network guidelines, was intermediate (63.0%: 66.7% in ivosidenib plus azacitidine group versus 59.5% in placebo plus azacitidine group) or poor (24.7%: 22.2% in ivosidenib plus azacitidine group versus 27.0% in placebo plus azacitidine group) for most patients at baseline. The median bone marrow blast proportion at baseline was 52.5% (range, 17% to 100%).

Two DCOs were available for the AGILE study. The first DCO (March 18, 2021) represents an unplanned early interim analysis by the IDMC, which occurred before the protocol-specified number of events for the planned analysis. Because of a notable difference in the number of deaths, which favoured ivosidenib, the IDMC recommended that trial recruitment should end early, treatment assignment should be unblinded, and crossover to ivosidenib should be allowed. The stopping boundaries were therefore adjusted, and this became the final analysis. A later DCO (June 30, 2022) was available for OS, transfusion independence, and harms. The results of the interim analysis of efficacy end points are at risk of overestimating the true effects of ivosidenib plus azacitidine.

One ITC report (comprising 1 NMA and 3 MAICs) was submitted by the sponsor to compare the treatment efficacy and safety of ivosidenib plus azacitidine with other active therapies (e.g., venetoclax plus azacitidine, azacitidine monotherapy, LDAC monotherapy, and venetoclax plus LDAC) for the treatment of IDH1-mutated AML. The comparative efficacy of ivosidenib versus venetoclax, in combination with azacitidine, was evaluated based on evidence from 6 RCTs.

Interpretation of Results

Efficacy

According to the patient groups and the clinical experts consulted for this review and the clinician groups that submitted input for this review, important unmet needs for patients with IDH1-mutated AML who are not eligible for intensive chemotherapy, include therapies that offer durable remission, can prolong life, can reduce transfusion dependency, and would improve patients’ HRQoL. The AGILE study met its primary end point at an unplanned interim analysis by the IDMC that occurred at the DCO of March 18, 2021. The results suggested that treatment with ivosidenib plus azacitidine is likely to be associated with a clinically important improvement in EFS rates, compared with treatment with placebo plus azacitidine: the between-group difference in EFS rate was 19.7% (95% CI, ███ ██ ████) at 6 months, favouring ivosidenib plus azacitidine. The between-group differences were affected by imprecision, where the lower bound of the CI included effects that might not be considered clinically important. Improvement in EFS was largely driven by the proportion of patients who experienced treatment failure, assigned an event time of the date of randomization: 42 patients (58.3%) in the ivosidenib plus azacitidine group versus 59 patients (79.7%) in the placebo plus azacitidine group experienced treatment failure. EFS is a composite end point, and the sample size of the AGILE study was small; following the large number of events of treatment failure, too few patients remained event-free to robustly characterize the long-term treatment effect of ivosidenib plus azacitidine on EFS.³¹

Treatment with ivosidenib plus azacitidine was associated with prolonged OS. At the DCO of March 18, 2021, OS met the stopping boundary, leading to the unplanned interim analysis for a statistically significant OS benefit for ivosidenib plus azacitidine. At the updated DCO of June 30, 2022, 37 patients (50.7%) in the ivosidenib plus azacitidine group and 58 (77.3%) in the placebo plus azacitidine group had died. The median OS was 29.3 months (95% CI, 13.2 months to NE) in the ivosidenib plus azacitidine group and 7.9 months (95% CI, 4.1 to 11.3 months) in the placebo plus azacitidine group (P < 0.0001). The corresponding HR was 0.42 (95% CI, 0.27 to 0.65). In addition, the OS rates at various time points showed that ivosidenib plus azacitidine likely results in a clinically relevant increase in the probability of OS at 1 year and 2 years, compared with placebo plus azacitidine. The between-group differences in the Kaplan-Meier–estimated OS rate were 24.6% (95% CI, ███ ██ ████) at 12 months and 35.7% (95% CI, ████ ██ ████) at 24 months. There was some potential for overestimation of the true effect due to small sample size. The results of the prespecified subgroup analyses for OS and EFS based on various patient baseline characteristics were consistent with those in the overall population.

Treatment with ivosidenib plus azacitidine may be associated with higher CR rates than treatment with placebo plus azacitidine. As of the DCO of March 18, 2021, the CR rate was 47.2% (95% CI, 35.3% to 59.3%) in the ivosidenib plus azacitidine group and 14.9% (95% CI, 7.7% to 25.0%) in the placebo plus azacitidine group. However, these estimates were affected by high risk of bias due to missing data. As of the DCO of June 30, 2022, a higher proportion of patients in the ivosidenib plus azacitidine group (██ ███████ ████████) were RBC and/or platelet transfusion independent than in the placebo plus azacitidine group (██ ███████ ████████). This was measured in a nonrandomized subset of the population. According to the clinical experts, improved CR rates and reduced transfusion dependence are considered clinically meaningful changes, and better CR rates and, in their opinion, reduced transfusion dependence can subsequently be translated to improved HRQoL and potentially prolonged survival.

HRQoL measured by the EORTC QLQ-C30 was a secondary outcome in the AGILE study. The evidence for HRQoL was considered to be very uncertain because of large amounts of missing data and imprecision; the CIs included the potential for little-to-no clinically meaningful difference between groups.

For this submission, venetoclax plus azacitidine was identified as the most relevant comparator for the indication under review. Comparative evidence of ivosidenib plus azacitidine versus venetoclax plus azacitidine was available through a sponsor-submitted ITC analysis. The rarity of the population of interest limits the size and number of clinical studies completed with potential comparators and adds to the practical challenges when indirectly comparing treatment options. Based on the results of the NMA and the MAICs, the evidence is insufficient to conclude whether ivosidenib plus azacitidine differs from venetoclax plus azacitidine in terms of OS, EFS, CR rates, or transfusion requirement in patients with untreated AML because of the limitations associated with the ITC report, such as limited evidence from the 6 RCTs, heterogeneity existing in the included trials, and imprecision of study results from the wide CrIs or CIs for these outcomes. There was no evidence to compare impacts on HRQoL for ivosidenib plus azacitidine versus any comparators outside the AGILE trial. For the comparisons between ivosidenib plus azacitidine and azacitidine or LDAC monotherapies, the results were in the same direction as results observed in the AGILE study and were as expected by clinical experts based on their experience with combination and monotherapies in clinical practice.

Harms

Overall, the safety results from the 2 DCOs are consistent. As of the DCO of March 18, 2021, in the AGILE trial the proportion of patients who experienced at least 1 AE was 98.6% (70 patients) in the ivosidenib plus azacitidine group and 100% (73 patients) in the placebo plus azacitidine group. Patients treated with ivosidenib plus azacitidine were more likely (5% or more) to report the following AEs than patients treated with placebo plus azacitidine: vomiting (29 patients [40.8%] versus 19 patients [26.0%]), neutropenia (20 [28.2%] versus 12 [16.4%]), thrombocytopenia (20 [28.2%] versus 15 [20.5%]), prolonged electrocardiogram QT interval (14 [19.7%] versus 5 [6.8%]), insomnia (13 [18.3%] versus 9 [12.3%]), differentiation syndrome (10 [14.1%] versus 6 [8.2%]), pain in extremity (10 [14.1%] versus 3 [4.1%]), hematoma (9 [12.7%] versus 1 [1.4%]), arthralgia (8 [11.3%] versus 3 [4.1%]), headache (8 [11.3%] versus 2 [2.7%]), leukocytosis (8 [11.3%] versus 1 [1.4%]), and leukopenia (6 [8.5%] versus 2 [2.7%]).

Evidence from the AGILE study showed that ivosidenib plus azacitidine is likely to be associated with a reduction in SAEs compared to placebo plus azacitidine. The proportion of patients who experienced SAEs was 69.0% (46 patients) in the ivosidenib plus azacitidine group and 82.2% (60 patients) in the placebo plus azacitidine group. Commonly reported SAEs in the 2 treatment groups were febrile neutropenia (23.9% of patients in the ivosidenib plus azacitidine group versus 27.4% in the placebo plus azacitidine group) and pneumonia (19.7% versus 21.9%). The clinical experts noted that the increased incidence of febrile neutropenia and pneumonia in the placebo plus azacitidine group may be related to disease progression in this treatment group, rather than being an adverse effect from the study drug.

The overall incidences of TEAEs that led to combination treatment discontinuation were similar between the treatment groups: 19 patients (26.8%) in the ivosidenib plus azacitidine group and 19 patients (26.0%) in the placebo plus azacitidine group.

Ivosidenib plus azacitidine may increase the rate of differentiation syndrome compared to placebo plus azacitidine; however, this was informed by few events. As of June 30, 2022, differentiation syndrome was reported in 10 patients (13.9%) in the ivosidenib plus azacitidine group and 6 patients (8.1%) in the placebo plus azacitidine group. Infection was likely to be reduced with ivosidenib plus azacitidine: infection was reported in 25 patients (34.7%) in the ivosidenib plus azacitidine group and 38 patients (51.4%) in the placebo plus azacitidine group.

There was no direct or indirect evidence comparing the harms of ivosidenib plus azacitidine to any other relevant comparators, including venetoclax plus azacitidine.

Conclusion

Adult patients with newly diagnosed AML with an IIDH1 R132 mutation who are not eligible to receive intensive induction chemotherapy have a poor prognosis. Patients and clinicians highlighted the need for new treatments that prolong life, improve remission, reduce transfusion requirements, and maintain HRQoL, compared to the current treatments. Evidence from a double-blind, phase III RCT (the AGILE study) showed that treatment with ivosidenib plus azacitidine likely results in a clinically important increase in the probability of OS at 12 months and 24 months compared to placebo plus azacitidine in the target population. Evidence from the trial also showed that ivosidenib plus azacitidine likely results in a clinically important increase in the probability of EFS at 6 months. EFS was a composite end point driven by treatment failure events: postbaseline, too few patients remained at risk to robustly characterize other components of the end point (i.e., relapse and death). The rates of CR, as well as CR plus CRi, and the need for transfusions may be improved with treatment with ivosidenib plus azacitidine compared with placebo plus azacitidine. Evidence on HRQoL was very uncertain because of the limitations of the analyses, including risk of bias due to missing data and imprecision. In terms of harms, evidence from the AGILE study suggested that treatment with ivosidenib plus azacitidine may result in an increase in differentiation syndrome but likely results in a reduction in the proportion of patients who experience SAEs and infections compared with treatment with placebo plus azacitidine.

There is a lack of direct comparative evidence between ivosidenib plus azacitidine and other relevant active treatments for patients with AML who are not eligible for intensive chemotherapy, such as venetoclax plus azacitidine, which is currently the most commonly used treatment in the target patient population. Indirect evidence from a sponsor-submitted NMA of 6 trials and 3 MAICs comparing patients from the AGILE study to patients treated with venetoclax plus azacitidine in the VIALE-A study was insufficient to conclude whether treatment with ivosidenib plus azacitidine differs from treatment with venetoclax plus azacitidine in terms of OS, EFS, CR rates, and transfusion dependence. There was substantial uncertainty in the treatment effect estimates (indicated by wide CrIs) from the ITCs because of limited efficacy data and important heterogeneity across studies. No comparisons of HRQoL or harms were conducted.

References

1.National Comprehensive Cancer Network. Clinical Practice Guidelines in Oncology (NCCN Guidelines®). Acute Myeloid Leukemia Version 1.2024. 2024; https://www.nccn.org/guidelines/guidelines-detail?category=1&id=1411. Accessed 2024 Mar 7.

2.Chennamadhavuni A, Lyengar V, Mukkamalla SKR, Shimanovsky A. Leukemia [accessed by sponsor]. StatPearls. Treasure Island (FL): StatPearls Publishing LLC; 2024: https://www.ncbi.nlm.nih.gov/pubmed/32809325.

3.Vakiti A, Reynolds SB, Mewawalla P. Acute Myeloid Leukemia. StatPearls. Treasure Island (FL): StatPearls Publishing Copyright © 2023, StatPearls Publishing LLC.; 2024.

4.Pleyer C, Afzal A, Shomali W, et al. Impact of red blood cell and platelet transfusions in acute myeloid leukemia (AML) patients undergoing remission induction chemotherapy. Blood. 2015;126(23):2127-2127.

5.Shiffer CA, Wang ES. Overview of the complications of acute myeloid leukemia. In: Post TW, ed. UpToDate. Waltham (MA): UpToDate: https://www.uptodate.com. Accessed 2023 Oct 27.

6.National Organization for Rare Disorders. Acute Myeloid Leukemia. 2018; https://rarediseases.org/rare-diseases/acute-myeloid-leukemia/. Accessed 2023 Oct 27.

7.American Cancer Society. What Is Acute Myeloid Leukemia (AML)? [accessed by sponsor]. https://www.cancer.org/cancer/types/acute-myeloid-leukemia/about/what-is-aml.html. Accessed 2023 Oct 27.

8.Cancer Quality Council of Ontario. Cancer Survival. 2021; https://www.csqi.on.ca/en/2020/indicators/cancer-survival. Accessed 2023 Oct 27.

9.Lubeck DP, Danese M, Jennifer D, Miller K, Richhariya A, Garfin PM. Systematic literature review of the global incidence and prevalence of myelodysplastic syndrome and acute myeloid leukemia. Blood. 2016;128(22):5930-5930.

10.Turbeville S, Francis KM, Behm I, et al. Prevalence and incidence of acute myeloid leukemia may be higher than currently accepted estimates among the ≥65 year-old population in the United States. Blood. 2014;124(21):958-958.

11.Statistics Canada. Number of new cases and age-standardized rates of primary cancer, by cancer type and sex. 2023; https://www150.statcan.gc.ca/t1/tbl1/en/cv.action?pid=1310074701. Accessed 2023 Oct 2.

12.CADTH Reimbursement Review: Venetoclax (Venclexta). Can J Health Technol. 2023;1(10). https://www.cadth.ca/sites/default/files/DRR/2021/PC0238-combined-report-FINAL.pdf. Accessed 2023 Oct 26.

13.Patel JP, Gonen M, Figueroa ME, et al. Prognostic relevance of integrated genetic profiling in acute myeloid leukemia. N Engl J Med. 2012;366(12):1079-1089. PubMed

14.Abbas S, Lugthart S, Kavelaars FG, et al. Acquired mutations in the genes encoding IDH1 and IDH2 both are recurrent aberrations in acute myeloid leukemia: prevalence and prognostic value. Blood. 2010;116(12):2122-2126. PubMed

15.Chotirat S, Thongnoppakhun W, Promsuwicha O, Boonthimat C, Auewarakul CU. Molecular alterations of isocitrate dehydrogenase 1 and 2 (IDH1 and IDH2) metabolic genes and additional genetic mutations in newly diagnosed acute myeloid leukemia patients. J Hematol Oncol. 2012;5:5. PubMed

16.Chou WC, Hou HA, Chen CY, et al. Distinct clinical and biologic characteristics in adult acute myeloid leukemia bearing the isocitrate dehydrogenase 1 mutation. Blood. 2010;115(14):2749-2754. PubMed

17.Marcucci G, Maharry K, Wu YZ, et al. IDH1 and IDH2 gene mutations identify novel molecular subsets within de novo cytogenetically normal acute myeloid leukemia: a Cancer and Leukemia Group B study. J Clin Oncol. 2010;28(14):2348-2355. PubMed

18.Mardis ER, Ding L, Dooling DJ, et al. Recurring mutations found by sequencing an acute myeloid leukemia genome. N Engl J Med. 2009;361(11):1058-1066. PubMed

19.Paschka P, Schlenk RF, Gaidzik VI, et al. IDH1 and IDH2 mutations are frequent genetic alterations in acute myeloid leukemia and confer adverse prognosis in cytogenetically normal acute myeloid leukemia with NPM1 mutation without FLT3 internal tandem duplication. J Clin Oncol. 2010;28(22):3636-3643. PubMed

20.Bullinger L, Dohner K, Dohner H. Genomics of Acute Myeloid Leukemia Diagnosis and Pathways. J Clin Oncol. 2017;35(9):934-946. PubMed

21.Montesinos P, Recher C, Vives S, et al. Ivosidenib and Azacitidine in IDH1-Mutated Acute Myeloid Leukemia. N Engl J Med. 2022;386(16):1519-1531. PubMed

22.Sadashiv SK, Hilton C, Khan C, et al. Efficacy and tolerability of treatment with azacitidine for 5 days in elderly patients with acute myeloid leukemia. Cancer Med. 2014;3(6):1570-1578. PubMed

23.Cashen AF, Schiller GJ, O’Donnell MR, DiPersio JF. Multicenter, phase II study of decitabine for the first-line treatment of older patients with acute myeloid leukemia. J Clin Oncol. 2010;28(4):556-561. PubMed

24.Dombret H, Seymour JF, Butrym A, et al. International phase 3 study of azacitidine vs conventional care regimens in older patients with newly diagnosed AML with >30% blasts. Blood. 2015;126(3):291-299. PubMed

25.Estey EH. Acute myeloid leukemia: 2012 update on diagnosis, risk stratification, and management. Am J Hematol. 2012;87(1):89-99. PubMed

26.AbbVie Corporation. VENCLEXTA® (venetoclax) Product Monograph. April 5, 2022 [sponsor provided reference].

27.Pfizer Canada ULC. DAURISMO® (glasdegib) Product Monograph. October 31, 2022 [sponsor provided reference].

28.Celgene Inc, a Bristol Myers Squibb company. VIDAZA® (azacitidine) Product Monograph. February 13, 2024 [sponsor provided reference].

29.Pfizer Canada I. Cytarabine Injection (100 mg/mL) Product Monograph. November 16, 2017 [sponsor provided reference]. 2017.

30.Tibsovo (ivosidenib): 250 mg tablets, oral [product monograph]. Laval (QC): Servier Canada Inc.; 2024 Jul 19.

31.Woods A, Norsworthy KJ, Wang X, et al. FDA Approval Summary: Ivosidenib in Combination with Azacitidine for Treatment of Patients with Newly Diagnosed Acute Myeloid Leukemia with an IDH1 Mutation. Clin Cancer Res. 2024;30(7):1226-1231. PubMed

32.Othus M, van Putten W, Lowenberg B, et al. Relationship between event-free survival and overall survival in acute myeloid leukemia: a report from SWOG, HOVON/SAKK, and MRC/NCRI. Haematologica. 2016;101(7):e284-286. PubMed

33.Assouline S, Wiesinger A, Spooner C, Jovanovic J, Schlueter M. Validity of event-free survival as a surrogate endpoint in haematological malignancy: Review of the literature and health technology assessments. Crit Rev Oncol Hematol. 2022;175:103711. PubMed

34.Norsworthy KJ, Gao X, Ko CW, et al. Response Rate, Event-Free Survival, and Overall Survival in Newly Diagnosed Acute Myeloid Leukemia: US Food and Drug Administration Trial-Level and Patient-Level Analyses. J Clin Oncol. 2022;40(8):847-854. PubMed

35.Balshem H, Helfand M, Schunemann HJ, et al. GRADE guidelines: 3. Rating the quality of evidence. J Clin Epidemiol. 2011;64(4):401-406. PubMed

36.Santesso N, Glenton C, Dahm P, et al. GRADE guidelines 26: informative statements to communicate the findings of systematic reviews of interventions. Journal of clinical epidemiology. 2020;119:126-135. PubMed

37.Musoro JZ, Coens C, Sprangers MAG, et al. Minimally important differences for interpreting EORTC QLQ-C30 change scores over time: A synthesis across 21 clinical trials involving nine different cancer types. Eur J Cancer. 2023;188:171-182. PubMed

38.Clinical Study Report AG120-C-009: A Phase 3, Multicenter, Double-Blind, Randomized, Placebo-Controlled Study of AG-120 in Combination with Azacitidine in Subjects ≥18 Years of Age with Previously Untreated Acute Myeloid Leukemia with an IDH1 Mutation [internal sponsor’s report]. Suresnes (FR): Institut de Recherches Internationales Servier; 2021 Dec 8.

39.Clinical Study Report AG120-C-009 (addendum): A Phase 3, Multicenter, Double-Blind, Randomized, Placebo-Controlled Study of AG-120 in Combination with Azacitidine in Subjects ≥18 Years of Age with Previously Untreated Acute Myeloid Leukemia with an IDH1 Mutation. Suresnes (FR): Institut de Recherches Internationales Servier; 2023 Aug 21.

40.American Cancer Society. Risk Factors for Acute Myeloid Leukemia (AML) [accessed by sponsor]. https://www.cancer.org/cancer/types/acute-myeloid-leukemia/causes-risks-prevention/risk-factors.html. Accessed 2023 Oct 27.

41.Klepin HD, Rao AV, Pardee TS. Acute myeloid leukemia and myelodysplastic syndromes in older adults. J Clin Oncol. 2014;32(24):2541-2552. PubMed

42.Statistics Canada. Age-specific five-year net survival estimates for primary sites of cancer, by sex, three years combined. 2020; https://www150.statcan.gc.ca/t1/tbl1/en/tv.action?pid=1310015801. Accessed 2023 Oct 2.

43.Cheson BD, Bennett JM, Kopecky KJ, et al. Revised recommendations of the International Working Group for Diagnosis, Standardization of Response Criteria, Treatment Outcomes, and Reporting Standards for Therapeutic Trials in Acute Myeloid Leukemia. J Clin Oncol. 2003;21(24):4642-4649. PubMed

44.Cancer Care Ontario. Ontario Cancer Statistics 2020 Ch 5: Cancer Incidence Rates and Trends. 2020; https://www.cancercareontario.ca/en/statistical-reports/ontario-cancer-statistics-2020/ch-5-cancer-incidence-rates-trends. Accessed 2023 Oct 27.

45.Cancer Quality Council of Ontario. Cancer Incidence. 2021; https://www.csqi.on.ca/en/2020/indicators/cancer-incidence. Accessed 2023 Oct 27.

46.Dohner H, Estey E, Grimwade D, et al. Diagnosis and management of AML in adults: 2017 ELN recommendations from an international expert panel. Blood. 2017;129(4):424-447. PubMed

47.Dohner H, Wei AH, Appelbaum FR, et al. Diagnosis and management of AML in adults: 2022 recommendations from an international expert panel on behalf of the ELN. Blood. 2022;140(12):1345-1377. PubMed

48.Arber DA, Borowitz MJ, Cessna M, et al. Initial Diagnostic Workup of Acute Leukemia: Guideline From the College of American Pathologists and the American Society of Hematology. Arch Pathol Lab Med. 2017;141(10):1342-1393. PubMed

49.Heuser M, Ofran Y, Boissel N, et al. Acute myeloid leukaemia in adult patients: ESMO Clinical Practice Guidelines for diagnosis, treatment and follow-up. Ann Oncol. 2020;31(6):697-712. PubMed

50.Lowenberg B, Griffin JD, Tallman MS. Acute myeloid leukemia and acute promyelocytic leukemia. Hematology Am Soc Hematol Educ Program. 2003;2003(1):82-101. PubMed

51.Brandwein JM, Zhu N, Kumar R, et al. Treatment of older patients with acute myeloid leukemia (AML): revised Canadian consensus guidelines. Am J Blood Res. 2017;7(4):30-40. PubMed

52.Kantarjian HM, Thomas XG, Dmoszynska A, et al. Multicenter, randomized, open-label, phase III trial of decitabine versus patient choice, with physician advice, of either supportive care or low-dose cytarabine for the treatment of older patients with newly diagnosed acute myeloid leukemia. J Clin Oncol. 2012;30(21):2670-2677. PubMed

53.Cortes JE, Heidel FH, Hellmann A, et al. Randomized comparison of low dose cytarabine with or without glasdegib in patients with newly diagnosed acute myeloid leukemia or high-risk myelodysplastic syndrome. Leukemia. 2019;33(2):379-389. PubMed

54.Wei AH, Montesinos P, Ivanov V, et al. Venetoclax plus LDAC for newly diagnosed AML ineligible for intensive chemotherapy: a phase 3 randomized placebo-controlled trial. Blood. 2020;135(24):2137-2145. PubMed

55.DiNardo CD, Jonas BA, Pullarkat V, et al. Azacitidine and Venetoclax in Previously Untreated Acute Myeloid Leukemia. N Engl J Med. 2020;383(7):617-629. PubMed

56.Pollyea DA, DiNardo CD, Arellano ML, et al. Impact of Venetoclax and Azacitidine in Treatment-Naive Patients with Acute Myeloid Leukemia and IDH1/2 Mutations. Clin Cancer Res. 2022;28(13):2753-2761. PubMed

57.Committee for Medicinal Products for Human Use. Assessment report: Venclyxto (venetoclax). (European public assessment report). Amsterdam (NL): European Medicines Agency; 2021: https://www.ema.europa.eu/en/documents/variation-report/venclyxto-h-c-4106-ii-0030-epar-assessment-report-variation_en.pdf. Accessed 2023 Nov 8.

58.Tibsovo in first-line acute myeloid leukemia patients with IDH1 mutation ineligible for intensive chemotherapy - indirect treatment comparison report (version 8.0) [internal sponsor’s report]. Kirkland (QC): IQVIA Inc.; 2023 Apr 18.

59.Dohner H, Montesinos P, Polo SV, et al. AML-295 AGILE: A Global, Randomized, Double-Blind, Phase 3 Study of Ivosidenib + Azacitidine Versus Placebo + Azacitidine in Patients With Newly Diagnosed Acute Myeloid Leukemia With an IDH1 Mutation. Clin Lymphoma Myeloma Leuk. 2022;22:S234.

60.Institut de Recherches Internationales Servier. NCT03173248: Study of AG-120 (Ivosidenib) vs. Placebo in Combination With Azacitidine in Participants With Previously Untreated Acute Myeloid Leukemia With an IDH1 Mutation (AGILE). ClinicalTrials.gov. Bethesda (MD): U.S. National Library of Medicine; 2024: https://classic.clinicaltrials.gov/ct2/show/NCT03173248. Accessed 2024 May 25.

61.DiNardo CD, Schuh AC, Stein EM, et al. Enasidenib plus azacitidine versus azacitidine alone in patients with newly diagnosed, mutant-IDH2 acute myeloid leukaemia (AG221-AML-005): a single-arm, phase 1b and randomised, phase 2 trial. Lancet Oncol. 2021;22(11):1597-1608. PubMed

62.Fayers PM, Aaronson NK, Bjordal K. The EORTC QLC-C30 scoring manual (3rd Edition). Brussels (BE): EORTC; 2001: https://www.eortc.org/app/uploads/sites/2/2018/02/SCmanual.pdf. Accessed 2024 May 25.

63.Giesinger JM, Kieffer JM, Fayers PM, et al. Replication and validation of higher order models demonstrated that a summary score for the EORTC QLQ-C30 is robust. J Clin Epidemiol. 2016;69:79-88. PubMed

64.Kaupp-Roberts SD, Yadegarfar G, Friend E, et al. Validation of the EORTC QLQ-BIL21 questionnaire for measuring quality of life in patients with cholangiocarcinoma and cancer of the gallbladder. Br J Cancer. 2016;115(9):1032-1038. PubMed

65.Reid J. Secondary vs de novo acute myeloid leukemia. 2022; https://aml-hub.com/medical-information/secondary-vs-de-novo-acute-myeloid-leukemia. Accessed 2024 Aug 15.

66.Dias S, Welton NJ, Sutton AJ, Ades AE. NICE DSU Technical Support Document 2: A Generalised Linear Modelling Framework for Pairwise and Network Meta-Analysis of Randomised Controlled Trials. Report by the Decision Support Unit. August 2011 (last updated April 2014) [accessed by sponsor]. 2011; https://www.ncbi.nlm.nih.gov/books/NBK310366/pdf/Bookshelf_NBK310366.pdf.

67.Dias S, Welton NJ, Sutton AJ, Caldwell DM, Lu G, Ades AE. NICE DSU Technical Support Document 4: Inconsistency in Networks of Evidence Based on Randomised Controlled Trials [accessed by sponsor]. NICE Decision Support Unit Technical Support Documents. London (GB): National Institute for Health and Care Excellence (NICE); 2014: https://www.ncbi.nlm.nih.gov/pubmed/27466656.

68.Chen C, Huls GA, Cameron H, et al. Matching Adjusted Indirect Comparison (MAIC) of Oral Azacitidine (Oral-AZA) Versus Injectable AZA in Patients (pts) with Acute Myeloid Leukemia (AML) in First Remission after Intensive Induction Chemotherapy. Blood. 2021;138(Supplement 1):3015-3015.

69.Tremblay G, Daniele P, Bell T, Chan G, Brown A, Cappelleri JC. Comparative effectiveness of glasdegib versus venetoclax combined with low-dose cytarabine in acute myeloid leukemia. J Comp Eff Res. 2021;10(7):603-612. PubMed

70.Phillippo D, Ades T, Dias S, Palmer S, Abrams KR, Welton NJ. NICE DSU Technical Support Document 18: Methods for population-adjusted indirect comparisons in submissions to NICE [accessed by sponsor]. Vol 18: NICE Decision Support Unit; 2016.

71.Pratz KW, Jonas BA, Pullarkat V, et al. Long-term follow-up of VIALE-A: Venetoclax and azacitidine in chemotherapy-ineligible untreated acute myeloid leukemia. Am J Hematol. 2024;99(4):615-624. PubMed

Appendix 1: Detailed Outcome Data

Please note that this appendix has not been copy-edited.

Table 28: Summary of ORR in the AGILE Study (FAS, DCO March 18, 2021)

CI = confidence interval; DCO = data cut-off; FAS = full analysis set; NR = not reported; OR = odds ratio; ORR = objective response rate.

Note: ORR was one of the key secondary outcomes in the AGILE study. It was defined as the rate of CR, CR with incomplete hematologic recovery (CRi) [including CR with incomplete platelet recovery (CRp)], partial remission (PR), and morphologic leukemia-free state (MLFS).

^aCI of percentage is calculated with the Clopper and Pearson (exact Binomial) method.

^bCochran-Mantel-Haenszel (CMH) estimate for OR is calculated with placebo plus azacitidine as the control (denominator).

^cIf the primary analysis of EFS, CR, OS and CR plus CRh are significant, a stratified Cochran-Mantel-Haenszel (CMH) test will be used to compare ORR between the 2 treatment arms. 1-sided P value is calculated from CMH test stratified by the randomization stratification factors (AML status and geographic region).

Source: AGILE Clinical Study Report.³⁸ Details included in the table are from the sponsor’s summary of clinical evidence.

Table 29: RoB-2 Tool for Assessing Risk of Bias in Randomized Trials

NA = not available.

^aStudy was recommended by Servier.

Source: Sponsor-submitted indirect treatment comparison.⁵⁸

Table 30: Baseline Characteristics in the AGILE Study Before and After Matching to IDH1 Population for OS (Unanchored MAIC)

AML = acute myeloid leukemia; BC = base case; ECOG = Eastern Cooperative Oncology Group; IPD = individual patient data; SA = scenario analysis.

Source: Sponsor-submitted indirect treatment comparison.⁵⁸

Table 31: Baseline Characteristics in the AGILE Study Before and After Matching to ITT Population for OS (Anchored MAIC)

AML = acute myeloid leukemia; BC = base case; ECOG = Eastern Cooperative Oncology Group; IPD = individual patient data; ITT = intention to treat; MAIC = matching-adjusted indirect comparison; OS = overall survival; SA = scenario analysis.

Source: Sponsor-submitted indirect treatment comparison.⁵⁸

Table 32: Baseline Characteristics in the AGILE Study Before and After Matching to ITT Population for OS (Anchored MAIC) With New Data Cut for the AGILE Study and the VIALE-A Study

AML = acute myeloid leukemia; BC = base case; ECOG = Eastern Cooperative Oncology Group; IPD = individual patient data; ITT = intention to treat; MAIC = matching-adjusted indirect comparison; OS = overall survival; SA = scenario analysis.

Source: Sponsor-submitted indirect treatment comparison.⁵⁸

Table 33: Baseline Characteristics in the AGILE Study Before and After Matching to ITT Population for EFS (Anchored MAIC)