Reimbursement Review

Pembrolizumab (Keytruda)

Sponsor: Merck

Therapeutic area: Unresectable or metastatic microsatellite instability-high or mismatch repair deficient solid tumours

This multi-part report includes:

Clinical Review

Pharmacoeconomic Review

Ethics Review

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

dMMR = mismatch repair deficient; MSI-H = microsatellite instability-high; NOC = Notice of Compliance.

Introduction

Mismatch repair (MMR) deficiency results in an inactivation of the DNA repair system and can be caused by mutations in genes encoding proteins that are responsible for detecting and correcting errors in mismatched base pairs.^1-3 Microsatellites are repetitive stretches of 1 to 6 base pairs that can result from a defective MMR system or inactivation of the MMR system.^2,3 MMR deficiency can be observed by comparing the variation in length of the microsatellite in the tumour tissue versus the same patient’s healthy tissue.⁴ This is often termed microsatellite instability (MSI) and large variation in microsatellite length is referred to as microsatellite instability-high (MSI-H).^2,4 Cancers that are MSI-H or MMR deficient (dMMR) tend to have a high tumour mutational load and are more responsive to PD-1–based immunotherapy.⁴ Immune checkpoint inhibitors, such as pembrolizumab, prevent tumour cells from evading the immune system; the cancer cells are then recognized by T cells and can trigger an antitumour immune response.⁴

MSI-H or dMMR mutations have been detected in more than 30 cancers.^5-8 It was reported in a meta-analysis from 2022 that there is an estimated pooled prevalence of 2.7% for MSI-H status and 2.9% for dMMR status across different solid tumour types in adults.⁶ Endometrial, colorectal, small intestine, and gastric cancers showed a higher prevalence of MSI-H (8.5% to 21.9%) compared with cervical, esophageal, bladder or urothelial, lung, and skin cancers, which showed a lower prevalence of MSI-H (less than 5%).⁶ According to the clinical experts consulted for this review, patients with metastatic solid tumours that have confirmed MSI-H or dMMR status have inferior outcomes when treated with conventional therapies and tend to show improved outcomes with immune checkpoint inhibitors. As a group, these patients have a worse prognosis compared with those whose tumours are not MSI-H or dMMR, although prognosis varies based on tumour type.⁹

There is high variability in first-line treatments and the standard of care (SOC) in solid tumours expressing MSI-H or dMMR mutations. In some instances, immunotherapy has become the SOC. In patients with MSI-H or dMMR unresectable or metastatic colorectal cancer (CRC), pembrolizumab received a recommendation to reimburse as a first-line therapy and is the currently funded SOC.¹⁰ Also, pembrolizumab received a recommendation to reimburse as a second-line or later therapy in adult patients with MSI-H or dMMR unresectable or metastatic endometrial cancer whose tumours have progressed following prior therapy and who have no satisfactory alternative treatment options, and it is currently funded in most jurisdictions.¹¹ Regardless of tumour type, patients with unresectable or metastatic MSI-H or dMMR cancers that have progressed after prior standard systemic therapy have a very poor prognosis, and the SOC for these patients is typically chemotherapy-based regimens, which provide limited clinical benefit and are associated with significant cumulative toxicity.

Pembrolizumab is a high-affinity antibody against PD-1 that reactivates tumour-specific cytotoxic T lymphocytes in the tumour microenvironment.¹² Pembrolizumab is available as a solution for infusion (100 mg per 4 mL vial) that is administered as an IV infusion over 30 minutes.^12,13 The recommended dosage in adult patients is either 200 mg every 3 weeks or 400 mg every 6 weeks until unacceptable toxicity or disease progression, or up to 24 months or 35 doses of 200 mg or 18 doses of 400 mg, whichever is longer, in patients without disease progression.^12,13 The recommended dosage in pediatric patients is 2 mg/kg (up to a maximum of 200 mg) every 3 weeks until disease progression or unacceptable toxicity, or up to 24 months or 35 doses, whichever is longer, in patients without disease progression.^12,13 In Canada, pembrolizumab has been issued market authorization to treat numerous types of cancers and the drug has been reviewed by Canada’s Drug Agency (CDA-AMC) for several other indications.

The objective of this report is to review and critically appraise the evidence submitted by the sponsor on the beneficial and harmful effects of pembrolizumab, 100 mg per 4 mL vial solution for infusion, in the treatment of adult and pediatric patients with unresectable or metastatic MSI-H or dMMR solid tumours, as determined by a validated test, that have progressed following prior treatment and who have no satisfactory alternative treatment options.

Testing Procedure Considerations

MSI-H and dMMR status can be assessed by next-generation sequencing (NGS) and immunohistochemistry (IHC). According to the clinical experts, testing for MSI-H and dMMR status is currently part of the SOC for the most common unresectable or metastatic solid tumour types in adult patients and is not part of the SOC for any pediatric patients. Potential increases in testing requirements to determine eligibility for pembrolizumab are not anticipated to substantially impact health care resources or pose an additional burden to patients and caregivers in either adult or pediatric populations.

Perspectives of Patients, Clinicians, and Drug Programs

The information in this section is a summary of input provided by the patient and clinician groups that responded to the CDA-AMC call for input and by the clinical experts consulted for the purpose of this review.

Patient Input

Patient group input for this review was submitted by Colorectal Cancer Canada and the Colorectal Cancer Resource & Action Network (CCRAN) (the latter working in collaboration with Canadian Cancer Survivor Network, Craig’s Cause Pancreatic Cancer Society, Canadian Breast Cancer Network, and Ovarian Cancer Canada). The 2 patient groups collected information through patient and caregiver interviews and online surveys. All contributions were from adult patients and/or their caregivers and do not include the perspectives of pediatric patients.

According to the respondents, symptoms of the disease as well as the side effects of the SOC treatments they received had the greatest impact on their physical, mental, and social health. The impact of diagnosis, disease recurrence, and treatment was felt by patients’ families, friends, and communities.

The most important outcomes noted by patients and caregivers included preventing disease progression and disease recurrence, prolonging survival, having a durable response, and improving health-related quality of life (HRQoL) and symptoms.

Patients who received pembrolizumab noted many benefits, including symptom relief, quick response to treatment, minimal adverse effects, being easier to receive than conventional treatments, providing better HRQoL, and avoiding long-term toxicities associated with conventional chemotherapies.

The groups emphasized the importance of wide and equitable access to biomarker testing, treatments for those with appropriate biomarker status (i.e., MSI-H or dMMR) that go beyond tumour type, and drugs to treat rare and pediatric cancers.

Clinician Input

Input From the Clinical Experts Consulted by CDA-AMC

The clinical experts highlighted the heterogeneity and aggressive phenotype of MSI-H and dMMR solid tumours. For adult patients with MSI-H and/or dMMR metastatic cancer, there is a need for effective therapeutic options if SOC or salvage chemotherapy fails. Meanwhile, for pediatric patients, there is a general need for new treatments because the current treatment options in both the front-line and relapsed setting are limited for this population.

The clinical experts highlighted that pembrolizumab is already indicated for use as SOC in non–MSI-H and non-dMMR tumours (e.g., kidney, urothelial, gastric, lung, or biliary tract) as well as in some cancers with MSI-H and/or dMMR mutations (e.g., CRC or endometrial); thus, pembrolizumab may not be considered in later lines of therapy, depending on prior response. For the pediatric population, 1 clinical expert highlighted that, in general, pembrolizumab and immune checkpoint inhibitors are used less frequently and, thus, experience with these is limited. In adult and pediatric patients, the clinical experts highlighted that, for the indication under review, pembrolizumab is intended for use after first-line therapy specific to the solid tumour site, although the ideal sequencing of treatments is unknown. The clinical experts noted that pembrolizumab would not be used in combination with existing treatments for any indication in this setting, based on currently available data.

The experts considered the patients most likely to benefit from treatment with pembrolizumab to be those with confirmed MSI-H and/or dMMR positivity as, at this time, there are no other validated biomarkers or clinical factors that can better predict treatment benefit for these patients than MSI-H and/or dMMR status. The clinical experts highlighted that underdiagnosis due to lack of testing is a risk, particularly in rarer cancers that are not reflexively tested using NGS or IHC and in pediatric patients, among whom testing is even more limited. One expert highlighted that up to 50% of patients will not benefit from treatment despite biomarker positivity. Additionally, the experts highlighted that, particularly for rarer cancers, treatment with SOC does not result in long-term benefits. The clinical experts also noted that the patients least suitable for pembrolizumab are those who are negative for the MSI-H and/or dMMR biomarker, those who have previously received and/or whose cancer did not respond to an immune checkpoint inhibitor during a previous line of therapy, and those with severe active autoimmune disease. In the pediatric setting, 1 clinical expert highlighted that the population of patients with high-grade gliomas, for which MSI and MMR mutations are known and the testing for which is more routine, would benefit from pembrolizumab. Conversely, the prevalence of MSI-H and/or dMMR mutations appears much lower in pediatric patients; thus, testing for MSI-H and/or dMMR mutations is uncommon (except for high-grade gliomas) in pediatric patients because clinicians do not often see pediatric patients with some of the more common adult cancers that harbour MSI-H or dMMR mutations (e.g., CRC, endometrial, or pancreatic).

In both the adult and pediatric populations, the primary aim of treatment is disease control and prolongation of life, with low toxicity. The clinical experts noted that complete responses are uncommon for these patients; thus, durability and stability are also important. Per the clinical experts, treatment tolerance is generally evaluated at each appointment, with radiologic response assessments typically occurring every 2 to 3 treatment cycles or approximately every 9 to 12 weeks, depending on local protocols.

There is extensive experience with pembrolizumab in oncology. The clinical experts highlighted that treatment with pembrolizumab is generally discontinued in the event of intolerable adverse events (AEs) and clear evidence of disease progression. One concern highlighted by the clinical experts is the potential risk of premature discontinuation due to pseudoprogression; therefore, treating physicians must carefully weigh the need to identify true disease progression against the risk of prematurely halting a therapy that may still offer significant clinical benefit to patients.

As noted, immune checkpoint inhibitors have become the SOC in the treatment of many adult cancers, and there is growing experience in their use across various health care settings, including academic and community settings; however, in the pediatric setting, this is likely limited to specialist tertiary hospitals. Specialists, such as medical oncologists (as well as pediatric oncologists in the case of pediatric patients), are needed to select appropriate patients for treatment and oversee the management of therapy.

Clinician Group Input

Five clinician groups submitted input for this CDA-AMC review: the Society of Gynecologic Oncology of Canada and 4 Ontario Health (Cancer Care Ontario) Drug Advisory Committees (for genitourinary cancer, breast cancer, gynecology cancer, and central nervous system [CNS] cancer).

The input from the clinician groups aligned with that of the clinical experts consulted for this review with regard to treatment goals, the unmet needs of this patient population, assessing treatment response, the drug’s place in therapy, deciding when to discontinue treatment, the types of specialists who should manage these patients, and when patients should be treated with pembrolizumab. The clinician groups noted that access to biomarker testing varies based on tumour type and jurisdiction.

Drug Program Input

The drug programs identified implementation issues relating to relevant comparators, considerations for the initiation of therapy, considerations for the prescribing of therapy, generalizability, the funding algorithm, care provision issues, and system and economic issues. Refer to Table 5 for more details.

Clinical Evidence

Systematic Review

Description of Studies

Three studies were included in this review: KEYNOTE-158, KEYNOTE-164, and KEYNOTE-051. All were single-arm trials that evaluated pembrolizumab in different solid tumours. The KEYNOTE-158 trial enrolled adult patients with advanced MSI-H or dMMR solid tumours (non-CRC) who had disease progression following prior therapy. The KEYNOTE-164 trial enrolled adult patients with advanced MSI-H or dMMR CRC that progressed following prior therapy into 2 cohorts. Cohort A included patients who had previously been treated with 2 or more lines of SOC therapies, which had to have included fluoropyrimidine, oxaliplatin, and irinotecan. Cohort B included patients previously treated with at least 1 line of prior systemic SOC therapy. The KEYNOTE-051 trial enrolled pediatric patients with MSI-H or dMMR cancers whose disease had progressed following prior therapy. Patients who had previously received anti–PD-1, anti–PD-L1, or anti–PD-L2 drugs were excluded from all trials. The primary end point for the KEYNOTE-158 and KEYNOTE-164 trials was objective response rate (ORR), and the secondary end points were duration of response (DOR), progression-free survival (PFS), overall survival (OS), safety, and tolerability. For the KEYNOTE-051 trial, the primary end points were safety, tolerability, and ORR, and the secondary end points were DOR, PFS, and OS.

The KEYNOTE-158 trial enrolled 373 patients with a variety of MSI-H and/or dMMR solid tumours, primarily endometrial (25%), gastric (14%), and small intestine and ovarian (7% each), among other rarer cancers (26%), who had disease progression after prior treatments. The mean age was 59.2 years (standard deviation [SD] = 13.1 years), and most patients (56%) had 2 or more prior lines of therapy. The KEYNOTE-164 trial enrolled 124 patients (61 in cohort A and 63 in cohort B) with MSI-H or dMMR CRC who had received 1 or more prior lines of therapy, while the majority had received 2 or more lines (76%). The mean age of included patients was 56.1 years (SD = 14.9 years), and there were slightly more males (56%) than females (44%). The KEYNOTE-051 trial included 7 pediatric patients with a mean age of 11 years (SD = 4.3 years), mostly with brain tumours (n = 6; 86%). In total, 57% of patients had received 1 prior therapy, while 29% had received 2 or more lines of therapy.

Efficacy Results

Response Rate

Adult: In the KEYNOTE-158 trial, the median duration of follow-up was 17 months (range, 0.2 to 71.4 months). The ORR per the Response Evaluation Criteria in Solid Tumours Version 1.1 (RECIST 1.1) was 33.8% (95% confidence interval [CI], 29.0% to 38.8%). The best objective response included a complete response (CR) in 40 patients (10.7%) and a partial response (PR) in 86 patients (23.1%). The tumour-specific ORR in the KEYNOTE-158 trial showed varied response rates, ranging from 4.8% (brain) to 59.3% (small intestine) across different cancer types. Tumours with higher ORRs compared with the pooled result included endometrial, gastric, small intestine, cholangiocarcinoma, urothelial, and salivary cancers. Tumours with lower ORRs included ovarian, pancreatic, brain, sarcoma, breast, cervical neuroendocrine, prostate, adrenocortical, mesothelioma, thyroid, small cell lung, and renal cancers.

For the KEYNOTE-164 trial, the median duration of follow-up was 31.4 months (range, 0.2 to 65.2 months) for cohort A and 52.7 months (range, 0.1 to 56.6 months) for cohort B. The pooled ORR (cohorts A and B) was 33.9% (95% CI, 25.6% to 42.9%); 9.7% had a CR and 24.2% had a PR. The ORR in cohort A was 32.8% (95% CI, 21.3% to 46.0%). The objective response included a CR in 3 patients (4.9%) and a PR in 17 patients (27.9%). The ORR in cohort B was 34.9% (95% CI, 23.3% to 48.0%). The objective response included a CR in 9 patients (14.3%) and a PR in 13 patients (20.6%).

Pediatric: In the KEYNOTE-051 trial, the median duration of follow-up was 5.2 months (range, 0.3 to 28.2 months), and the ORR was 0% (95% CI, 0.0% to 41.0%). No patients experienced a CR or PR, 1 patient (14.3%) had stable disease, and 5 patients (71.4%) experienced progressive disease. Tumour-specific ORR results were not analyzed in the KEYNOTE-051 trial because all but 1 patient had brain cancer.

Duration of Response

Adult: Among the 126 patients in the KEYNOTE-158 trial with a response, the median DOR was 63.2 months (range, 1.9 to 63.9 months). Both ends of the range represent patients who were ongoing in the study without the event at the time of the data cut-off. The tumour-specific median DORs were as follows (for tumours represented by at least 10 patients): endometrial cancer (63.2 months), gastric cancer (not reached), and small intestine cancer (not reached). The DOR event-free probability at 6, 12, and 24 months was 95.2%, 88.5%, and 72.3%, respectively.

In the KEYNOTE-164 trial, among the 42 patients with a response in the overall population, the median DOR was not reached (range, 4.4 to 58.5 months among patients with a response who were ongoing in the study). The DOR event-free probability at 6, 12, and 24 months was 97.6%, 95.1%, and 92.2%, respectively. Similarly, the median DOR in cohort A (n = 20) was not reached (range, 6.2 to 58.5 months among patients with a response who were ongoing), and not reached in cohort B (n = 22) (range, 4.4 to 52.4 months among patients with a response who were ongoing).

Pediatric: DOR was not measured in the KEYNOTE-051 trial because there were no patients with a response.

Overall Survival

Adult: A total of 230 of 373 patients (61.7%) in the KEYNOTE-158 trial died, with a median OS of 19.8 months (95% CI, 14.5 to 25.8 months). The OS event-free probability at 6, 12, and 24 months was 72.1%, 58.6%, and 46.5%, respectively.

In the KEYNOTE-164 trial, 63 of 124 patients (50.8%) died, with a median OS of 36.1 months (95% CI, 24.0 months to not reached). The OS event-free probability at 12 and 24 months was 74.2% and 59.1%, respectively. In cohort A, 38 of 61 patients died (62.3%), with a median OS of 31.4 months (95% CI, 21.4 to 58.0 months). In cohort B, 31 of 63 patients died (49.2%), with a median OS of 47.0 months (95% CI, 19.2 months to not reached).

Pediatric: In the KEYNOTE-051 trial, 5 of 7 patients (71.4%) died, representing a median OS of 7.7 months (95% CI, 1.9 months to not reached). The OS event-free probability at 6, 12, and 18 months was 50.0%, 33.3%, and 33.3%, respectively.

Progression-Free Survival

Adult: A total of 275 of 373 patients (73.7%) in the KEYNOTE-158 trial experienced a PFS event, with a median PFS of 4.0 months (95% CI, 2.4 to 4.3 months). The PFS event-free probability at 6, 12, and 24 months was 43.2%, 35.1%, and 28.8%, respectively.

In the KEYNOTE-164 trial, 83 of 124 patients (66.9%) experienced a PFS event, with a median PFS of 4.0 months (95% CI, 2.1 to 7.4 months). The PFS event-free probability at 6, 12, and 24 months was 45.8%, 37.5%, and 33.8%, respectively. In cohort A, 44 of 61 patients (72.1%) had a PFS event, with a median PFS of 2.3 months (95% CI, 2.1 to 8.1 months). In cohort B, 40 of 63 patients (63.5%) had a PFS event, with a median PFS of 4.1 months (95% CI, 2.1 to 18.9 months).

Pediatric: In the KEYNOTE-051 trial, 6 of 7 patients (85.7%) experienced a PFS event, with a median PFS of 1.7 months (95% CI, 0.4 months to not reached). The PFS event-free probability at 6 months was 16.7% and at 12 months it was not estimable.

Health-Related Quality of Life

Adult: In the KEYNOTE-158 trial, the European Organisation for Research and Treatment of Cancer Quality of Life Questionnaire Core 30 (EORTC QLQ-C30) global health status (GHS) and quality of life (QoL) score (ranging from 0 for worst to 100 for best) had a baseline mean of 64.40 points (SD = 20.12 points) across 364 patients. By week 9, the mean GHS and QoL score was 67.48 points (SD = 22.48 points) in 265 patients, and the least squares (LS) mean change from baseline was 3.08 points (95% CI, 0.32 to 5.84 points). A total of 57 patients (21.5%) experienced a deterioration in HRQoL at week 9. Deterioration was defined as a negative change of 10 points or more for each EORTC QLQ-C30 scale or subscale.

HRQoL was not assessed in the KEYNOTE-164 trial.

Pediatric: HRQoL was not assessed in the KEYNOTE-051 trial.

Harms Results

Adult

In the KEYNOTE-158 trial, 96% of patients experienced at least 1 AE, with the most common being diarrhea (25%), fatigue (24%), nausea (21%), asthenia (20%), vomiting (19%), and pruritus (19%). At least 1 serious adverse event (SAE) was reported in 133 out of 373 patients (36%). The SAEs reported in at least 2% of patients included sepsis (2%) and pneumonia (2%). A total of 49 of 373 patients (13%) discontinued pembrolizumab in the KEYNOTE-158 trial, and 20 patients (5.4%) died due to AEs.

In the KEYNOTE-164 trial, 100% of patients experienced AEs, with the most frequent being fatigue (34%), diarrhea (32%), nausea (31%), abdominal pain (27%), and vomiting (26%). At least 1 SAE was reported in 56 of 124 patients (45%). SAEs reported in at least 2% of patients included dyspnea (4%), sepsis (4%), abdominal pain (3%), a small intestine obstruction (3%), urinary tract infection (3%), and ileus (2%). In total, 10 of 124 patients (8%) discontinued treatment due to AEs, and 4 of 124 patients (3%) had an AE that resulted in death.

Pediatric

In the KEYNOTE-051 trial, all patients experienced at least 1 AE, with the most common being anemia (71.4%), headache (71.4%), vomiting (57.1%), decreased lymphocyte count (42.9%), pyrexia (42.9%), and rash (42.9%). At least 1 SAE was reported in 6 of 7 patients (85.7%). The most common SAE was vomiting (28.6%). A total of 2 of 7 patients (28.6%) discontinued treatment due to AEs. There were no deaths due to AEs.

Critical Appraisal

KEYNOTE-158, KEYNOTE-164, and KEYNOTE-051 are all single-arm trials, which raised several important considerations due to the lack of any relevant comparison group. Single-arm trials are inherently limited in their ability to provide causal inferences, making it difficult to distinguish between treatment effects and natural disease progression. Only ORR and DOR can be definitively attributed to the antitumour activity of pembrolizumab because a response is unlikely to occur spontaneously. This absence of a direct comparator creates challenges in interpreting time-to-event end points (PFS, OS) and HRQoL. The open-label design may also introduce a risk of performance bias, although there is no clear evidence that this was a concern in the trials. There is a risk of bias in the measurement of the outcomes, particularly HRQoL and subjective AEs, because knowledge of the intervention can impact patient expectations and perceptions about the benefits and harms of treatment. This risk of bias in the outcome assessment was mitigated for the response outcomes (PFS, ORR, DOR) which, in the KEYNOTE-158 and KEYNOTE-164 trials, were assessed by an independent review committee. OS is an objective outcome unlikely to be affected by such bias. The assessment of HRQoL in the KEYNOTE-158 trial was at high risk of bias due to missing outcome data. Despite high adherence among available patients, close to 30% of outcome data were missing at week 9. The results of the KEYNOTE-158 trial, which included adult patients with varying tumour types, showed considerable heterogeneity in response across tumour types. Many tumour types were represented by the very small number of patients (< 20). The interpretation of the potential differences in response across small samples of different cancer types is therefore challenging because these may represent either actual differences in treatment effects or natural statistical variation. The Health Canada reviewers’ report also acknowledged the variation in therapeutic benefit across tumour types, citing small sample sizes and that the numbers of patients recruited for each histology reflects the natural prevalence of the MSI-H and/or dMMR biomarker, and that enrolling a larger number of patients may not have been feasible, given the broad indication. The Health Canada reviewers’ report also noted that the use of a pooled ORR and DOR in the KEYNOTE-158 trial is reasonable for pursuing a tissue-agnostic indication.¹⁴ There was a small degree of heterogeneity across cohorts A and B in the KEYNOTE-164 trial. In the KEYNOTE-051 trial, only 7 patients were included, which limits the reliability of assessing the efficacy and safety of pembrolizumab.

In the KEYNOTE-158, KEYNOTE-164, and KEYNOTE-051 trials, patients who had previously received anti–PD-1, anti–PD-L1, or anti–PD-L2 drugs were excluded. The clinical experts consulted by CDA-AMC indicated that only a small proportion of patients in Canada would meet this eligibility criterion because most MSI-H and/or dMMR solid tumours are treated with anti–PD-1 or anti–PD-L1 drugs as SOC in earlier lines of therapy. This limits the generalizability of the trial results to the smaller population of patients with MSI-H or dMMR cancers in Canada who would not have received anti–PD-1 or anti–PD-L1 drugs in an earlier line. For cancers such as colorectal, endometrial, and non–small cell lung cancer, pembrolizumab is already used in earlier treatment stages. Additionally, other immune checkpoint inhibitors are widely used in solid tumours like gastric, mesothelioma, breast, small cell lung, and biliary tract cancers, regardless of MSI-H or dMMR status. The KEYNOTE-051 trial included only 7 patients, primarily with brain cancers, which also limits the generalizability of the findings. This small cohort is unlikely to adequately represent the broader patient population in terms of age, sex, performance status, disease stage, and other factors. Furthermore, HRQoL was not assessed in 2 trials (KEYNOTE-164 and KEYNOTE-051); thus, the generalizability of any HRQoL results to other indications or populations remains uncertain.

GRADE Summary of Findings and Certainty of the Evidence

For the pivotal studies identified in the sponsor’s systematic review, Grading of Recommendations, Assessment, Development and Evaluation (GRADE) was used to assess the certainty of the evidence for the outcomes considered most relevant to inform CDA-AMC expert committee deliberations, and a final certainty rating was determined as outlined by the GRADE Working Group.^15,16 Although GRADE guidance is not available for noncomparative studies, the review team assessed pivotal single-arm trials for study limitations (which refer to internal validity or risk of bias), inconsistency across studies (or populations), indirectness, and publication bias to present these important considerations. Because the lack of a comparator arm does not allow for a conclusion to be drawn on the effect of the intervention versus any comparator, the certainty of evidence for single-arm trials starts at very low certainty with no opportunity for rating up.

Table 2 presents a GRADE summary of the findings for pembrolizumab that were assessed in the KEYNOTE-158, KEYNOTE-164, and KEYNOTE-051 trials. The selection of outcomes for GRADE assessment was based on the sponsor’s summary of clinical evidence, consultation with clinical experts, and the input received from patient and clinician groups and public drug plans. The following list of outcomes was finalized in consultation with expert committee members:

• response (ORR [CR and PR], and DOR)

• survival (OS and PFS)

• HRQoL (EORTC QLQ-C30)

• notable harms (adverse events of special interest [AEOSIs]), i.e., immune-mediated AEs and infusion-related reactions.

Table 2: GRADE Summary of Findings for Pembrolizumab for Patients With Unresectable or Metastatic MSI-H and/or dMMR Solid Tumours

AEOSI = adverse event of special interest (immune-mediated event or infusion-related reaction); CI = confidence interval; CR = complete response; CRC = colorectal cancer; dMMR = mismatch repair deficient; DOR = duration of response; EORTC QLQ-C30 = European Organisation for Research and Treatment of Cancer Quality of Life Questionnaire Core 30; GRADE = Grading of Recommendations Assessment, Development and Evaluation; MSI-H = microsatellite instability-high; NA = no assessment; NR = not reported; ORR = objective response rate; OS = overall survival; PFS = progression-free survival; PR = partial response; QoL = quality of life; RECIST 1.1 = Response Evaluation Criteria in Solid Tumours Version 1.1; SAE = serious adverse event; vs. = versus.

Note: In column 3 (effect), “+” indicates the patient was ongoing in the study without the event at the time of analysis.

^aIn the absence of a comparator group, conclusions about efficacy relative to any comparator cannot be drawn and the certainty of evidence is started at “very low” and cannot be rated up.

^bRated down 1 level for serious study limitations because results are based on an interim analysis. There is a risk of overestimating treatment effects.

^cRated down 1 level due to inconsistency for ORR, CR, and DOR. The KEYNOTE-158 trial included a mixed solid tumour population. There was heterogeneity in the effects across different solid tumours for these outcomes; there was no subgroup information for OS, PFS, or harms available to assess these outcomes.

^dHRQoL data were not reported for the KEYNOTE-164 trial (adults with CRC) and the KEYNOTE-051 trial (children with mixed solid tumours). Rated down 2 levels for study limitations due to the risk of bias in the assessment of the outcome (open-label trial with subjective assessment) and due to missing outcome data.

Source: KEYNOTE-158,¹⁷ KEYNOTE-164,¹⁸ and KEYNOTE-051¹⁹ Clinical Study Reports.

Long-Term Extension Studies

No long-term extension studies were provided by the sponsor.

Indirect Comparisons

To support the economic evaluation for tumour-agnostic submissions, the sponsor submitted several indirect treatment comparisons (ITCs), each focusing on comparing pembrolizumab with other relevant treatments in 4 different solid tumour types: colorectal, endometrial, small intestine, and gastric cancers. In the CRC ITC, pembrolizumab was compared with pooled chemotherapy and anti–vascular endothelial growth factor (VEGF) plus chemotherapy using a naive indirect comparison, and trifluridine and tipiracil hydrochloride (TAS-102) through an unanchored matching-adjusted indirect comparison (MAIC). In the endometrial cancer ITC, an unanchored MAIC was conducted to compare pembrolizumab with the treatment of physician’s choice (TPC) (doxorubicin or paclitaxel). For the small intestine cancer ITC, a naive indirect comparison was used to compare pembrolizumab with nanoparticle albumin-bound (nab)-paclitaxel. The gastric cancer ITC included a naive indirect comparison to compare pembrolizumab with leucovorin calcium (folinic acid) + fluorouracil + irinotecan hydrochloride (FOLFIRI); ramucirumab plus paclitaxel; ramucirumab monotherapy; paclitaxel; and irinotecan.

Efficacy Results

Colorectal Cancer

The colorectal ITC included 2 naive ITCs comparing pembrolizumab with chemotherapy and chemotherapy plus anti-VEGF, and 1 MAIC comparing pembrolizumab with TAS-102. Only PFS and OS results were reported. In both of the naive ITCs, the pembrolizumab group had a lower median age and fewer patients with an Eastern Cooperative Oncology Group Performance Status (ECOG PS) of 0 compared with the comparator group. In the unanchored MAIC, matching factors included age, sex, and ECOG PS.

The naive ITC for OS estimated a hazard ratio (HR) of 0.30 (95% CI, 0.24 to 0.39) when compared with chemotherapy alone, and 0.37 (95% CI, 0.29 to 0.48) when compared with chemotherapy in combination with an anti-VEGF, favouring pembrolizumab. The naive ITC for PFS estimated an HR of 0.43 (95% CI, 0.34 to 0.54) when compared with chemotherapy alone, and 0.53 (95% CI, 0.42 to 0.67) when compared with chemotherapy in combination with an anti-VEGF, favouring pembrolizumab.

In the unanchored MAIC comparing pembrolizumab with TAS-102, the estimated HR for OS was 0.21 (95% CI, 0.15 to 0.30) and, for PFS, was 0.32 (95% CI, 0.23 to 0.45), both favouring pembrolizumab.

Endometrial Cancer

For the endometrial cancer ITC, an unanchored MAIC was conducted to compare pembrolizumab with TPC (doxorubicin or paclitaxel). ORR, PFS, and OS results were reported. Matching factors included age, race, ECOG PS, number of prior lines of therapy, and histology status.

In the unanchored MAIC, the estimated response ratio was 4.27 (95% CI, 2.11 to 8.64) for ORR, favouring pembrolizumab. The estimated HR was 0.31 (95% CI, 0.19 to 0.53) for PFS and 0.24 (95% CI, 0.13 to 0.45) for OS, both favouring pembrolizumab compared with TPC. In each case, sensitivity analyses yielded results that aligned with the main analysis in both the direction and magnitude of effect.

Small Intestine Cancer

For the small intestine cancer ITC, a naive ITC was conducted to compare pembrolizumab with nab-paclitaxel. Only PFS and OS results were presented. The pembrolizumab group had a similar median age and a greater number of patients with an ECOG PS of 0 compared with the comparator group.

The naive ITC estimated an HR of 0.18 (95% CI, 0.07 to 0.45) for OS and 0.22 (95% CI, 0.09 to 0.52) for PFS, both favouring pembrolizumab.

Gastric Cancer

For the gastric cancer ITC, a naive ITC was conducted to compare pembrolizumab with FOLFIRI, ramucirumab plus paclitaxel, ramucirumab monotherapy, paclitaxel, and irinotecan. Only OS and PFS results were presented. The pembrolizumab group had a higher median age and a greater number of patients with an ECOG PS of 0 compared with all comparator groups.

For OS, the naive ITC estimated an HR of 0.43 (95% CI, 0.26 to 0.69) versus FOLFIRI, 0.35 (95% CI, 0.22 to 0.53) versus ramucirumab, 0.44 (95% CI, 0.29 to 0.66) versus ramucirumab plus paclitaxel, and 0.38 (95% CI, 0.26 to 0.56) versus irinotecan, all favouring pembrolizumab. The evidence was insufficient to demonstrate a difference between pembrolizumab and paclitaxel.

For PFS, the naive ITC estimated an HR of 0.43 (95% CI, 0.28 to 0.67) versus FOLFIRI, 0.37 (95% CI, 0.24 to 0.58) versus ramucirumab, 0.45 (95% CI, 0.31 to 0.65) versus ramucirumab plus paclitaxel, and 0.33 (95% CI, 0.23 to 0.47) versus irinotecan, all favouring pembrolizumab. The evidence was insufficient to demonstrate a difference between pembrolizumab and paclitaxel.

Harms Results

No harms were evaluated in the submitted ITCs.

Critical Appraisal

The sponsor conducted systematic literature reviews (SLRs) to identify studies for inclusion in the 4 ITCs. Although the SLRs appeared comprehensive, an a priori protocol was not provided, which prevented the evaluation of the risk of selective reporting based on the magnitude, direction, or statistical significance of the effects. The included comparator studies were deemed to be primarily at a low or unclear risk of bias. Additionally, the literature searches for colorectal, endometrial, and small intestine cancer studies were conducted between July and August 2023; it is not clear whether any new relevant studies have become available since then. Violations of the proportional hazards assumption in most comparisons further undermine the validity of the estimated HRs.

The majority of the ITCs, except for the endometrial cancer ITC, included comparator studies that did not evaluate MSI-H or dMMR status because this was infeasible. According to the clinical experts consulted by the review team, patients with dMMR or MSI-H status typically have a worse prognosis compared with those whose tumours are not MSI-H or dMMR, although prognosis varies based on tumour type. Without the dMMR or MSI-H status of patients in the comparator arms, the impact cannot be quantified. Naive indirect comparisons were used due to substantial reductions in effective sample size (ESS) when attempting to match baseline characteristics; however, such methods are highly susceptible to confounding bias from differences in patient characteristics and study methodologies. Unanchored MAICs were employed in other cases. These matched on a small number of factors. Although these factors were relevant according to the clinical experts consulted by the review team, it is unlikely that they represent all known and unknown prognostic and effect-modifying variables. As a result, there is a high risk of residual confounding. Small ESSs after matching and differences in patient characteristics, such as ECOG PS scores, reduce the reliability of the results.

The colorectal, endometrial, and gastric cancer comparisons may no longer be highly relevant due to changes in the current SOC because patients are likely to have already received immunotherapy in the first-line setting. This shift means that comparisons may not accurately reflect current patient populations or outcomes. Additionally, the analyses did not assess harms or HRQoL outcomes.

Studies Addressing Gaps in the Evidence From the Systematic Review

Description of Studies

The International Replication Repair Deficiency Consortium (IRRDC) data came from an observational, registry-based study of pediatric patients (N = 18, including 1 patient who was aged 24 years with a total of 20 tumours) with confirmed or suspected DNA replication repair deficiency.²⁰ A patient’s cancer type was categorized as either CNS tumours or non-CNS tumours. Patients were treated with pembrolizumab between May 2015 and March 2019. Objective tumour response was the outcome of interest. The data cut-off date was March 2022.

The whole-exome sequencing (WES) data were obtained from across the clinical development program and data were evaluated from 7 trials of pembrolizumab monotherapy.²⁰ The patients (N = 21) consisted of adults (median age was 65 years) with advanced solid tumours (gastric or gastroesophageal junction cancer, prostate cancer, cholangiocarcinoma, head and neck squamous cell carcinoma, triple-negative breast cancer) who had previously received at least 1 systemic treatment. End points of interest included ORR, DOR, PFS, and OS.

Efficacy Results

From the IRRDC dataset (pediatric patients), 17 tumours had measurable disease at baseline (3 tumours were not measurable). Based on the 17 tumours, 4 patients (23.5%) experienced objective treatment response, and 9 patients (52.9%) experienced stable disease. Four tumours (23.5%) continued to progress. Furthermore, 11 of 20 tumours (55.0%) had not progressed by 6 months and 15 of 18 patients (83.3%) were alive at 12 months.

In the WES dataset (adults), the ORR was 52.4% (95% CI, 29.8% to 74.3%). Median PFS was 17.8 months (95% CI, lower limit of 4.3 months and upper limit not reached) while the PFS rates were 56.7% and 45.9% at 12 and 36 months (there were no CIs), respectively. Median DOR and median OS were not achieved. The OS rates were 66.7% and 61.5% at 12 and 24 months (there were no CIs), respectively.

Harms Results

The collection of safety data was not a specific intent of the IRRDC study (pediatric patients); therefore, harms data are limited.²¹ According to the summary of clinical safety, no new safety signals were identified, and pancreatitis was the only harm that occurred in more than 1 patient (N = 2). The following harms were reported in 1 patient each: diarrhea, pneumonia, gastritis, dry skin, transient hypothyroidism, tolerable intermittent elevations in liver enzymes, skin rash, severe headaches, and seizure.

Safety data were not reported for the WES dataset (adults).

Critical Appraisal

Although the datasets provide more efficacy information on pediatric and adult populations with MSI-H cancers, they are small, noncomparative, and do not address the lack of direct or indirect evidence for pembrolizumab in this setting. Moreover, no protocols were available for review and the methods were not well described; thus, there is a risk of selective reporting. It was noted that for the IRRDC study, the treatment of patients was at the discretion of the clinical team and changes to dosages may not have been standardized across patients, making it challenging to interpret the results. Outcome measures were reviewed centrally by a blinded, independent committee, which can lower the risk of bias in the outcome measurement. The tumour types in these datasets generally cover the same types as those found in the pivotal trials (i.e., the datasets provide limited information additional to what is already available). Based on the available information, more than half of the patients had an ECOG PS of 1, which may not be representative of patients in poorer health who could receive pembrolizumab for MSI-H or dMMR tumours in clinical practice in Canada. There were few data on pediatric patients contributing to the pivotal trial evidence, and the IRRDC dataset modestly increases the amount of information available for younger patients. There was limited reporting of harms (reported only for the IRRDC dataset); assessments of HRQoL were not reported. Despite the use of real-world registry data, which can improve the generalizability of the results, the internal and external validity issues minimize the utility and applicability of the findings to clinical practice.

Conclusions

Patients and clinicians emphasized a high unmet need for new, effective treatments for patients with unresectable or metastatic MSI-H and/or dMMR solid tumours if SOC or salvage chemotherapy fails. Evidence for this review consisted of 3 pivotal trials (KEYNOTE-158, KEYNOTE-164, and KEYNOTE-051), multiple ITCs, and 2 noncomparative datasets in pediatric and adult patients with MSI-H cancer (IRRDC and WES datasets). The evidence about the efficacy of pembrolizumab in adults from 2 pivotal trials, KEYNOTE-158 and KEYNOTE-164, was very uncertain due to the single-arm design. The clinical experts consulted by the review team indicated that the response to treatment (ORR) was clinically meaningful and durable (DOR) compared with what is typically observed with SOC treatments. However, responses were heterogeneous across specific cancer types, many of which were represented by a small number of patients. Despite an inability to draw causal conclusions regarding time-to-event end points (OS and PFS), the clinical experts stated that the results were promising based on the natural history of the disease and experience in clinical practice. The results for HRQoL were inconclusive due to the open-label design and missing data. Results for the WES dataset were supportive of the pivotal trials but were affected by similar limitations.

In the pediatric population, due to the small number of patients (N = 7) enrolled in the KEYNOTE-051 trial and the single-arm design, it is difficult to ascertain the benefit of pembrolizumab in this group of patients. No patients in the KEYNOTE-051 trial had a PR or CR. Supportive evidence from the single-arm IRRDC registry study provides preliminary evidence from a small sample of patients to suggest that some pediatric patients with MSI-H or dMMR malignancies may respond to pembrolizumab. No new safety signals were identified in any of the pembrolizumab trials.

The indirect comparative evidence from the ITCs (naive indirect comparisons and unanchored MAICs) was limited to 4 cancer types in adults: colorectal, endometrial, small intestine, and gastric. The results of the ITCs suggested that PFS and OS were improved compared with SOC, which aligned with the expectations of the clinical experts consulted by the review team. However, the ITCs had significant limitations that impacted the internal validity of the findings and precluded definitive conclusions about the comparative efficacy of pembrolizumab versus SOC treatments. Additionally, immune checkpoint inhibitors are now used in the first line for colorectal, endometrial, and gastric cancers, reducing the relevance of these comparisons. No comparative safety data were provided; thus, the relative safety of pembrolizumab compared with other treatments is unknown. However, the clinical experts consulted for this review emphasized the breadth of experience with using pembrolizumab to treat other cancers, highlighting that pembrolizumab is better tolerated and has fewer and less severe side effects than conventional chemotherapy.

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 pembrolizumab, 100 mg per 4 mL vial solution for infusion, in the treatment of adult and pediatric patients with unresectable or metastatic MSI-H or dMMR solid tumours, as determined by a validated test, whose disease has progressed following prior treatment and who have no satisfactory alternative treatment options.

Disease Background

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

Error repair of mismatched base pairs during DNA replication is an essential function for the maintenance of genomic integrity.³ The MMR deficiency results in an inactivation of the DNA repair system and can be caused by sporadic or inherited mutations in the gene (e.g., MLH1, MSH2, MSH6, and PMS2) encoding proteins that are responsible for detecting and correcting errors in mismatched base pairs or by methylation of the MLH1 promoter.^1-3

Microsatellites are repetitive stretches of 1 to 6 base pairs that can result from a defective MMR system or inactivation of the MMR system.^2,3 The dMMR status can be observed by comparing the variation in length of the microsatellite in a patient’s tumour tissue versus their healthy tissue.⁴ This is often termed MSI, and a large variation in microsatellite length is referred to as MSI-H.^2,4

Mutations in the DNA result in abnormal proteins that can be neoantigens (as opposed to self-antigens).⁴ Cancers that are MSI-H or dMMR tend to have a high tumour mutational load and are more responsive to PD-1–based immunotherapy.⁴ Immune checkpoint inhibitors, such as pembrolizumab, prevent tumour cells from evading the immune system so they can be recognized by T cells and can trigger an antitumour immune response.⁴

MSI-H or dMMR mutations have been detected in more than 30 cancers.^5-8 It was reported in a meta-analysis from 2022 that there is an estimated pooled prevalence rate of 2.7% for MSI-H status and 2.9% for dMMR status across different solid tumour types in adults.⁶ Endometrial, colorectal, small intestine, and gastric cancers showed a higher prevalence of MSI-H (8.5% to 21.9%), compared with cervical, esophageal, bladder or urothelial, lung, and skin cancers, which showed a lower prevalence of MSI-H (less than 5%).⁶ The same meta-analysis noted that, when data were available by disease stage, dMMR and MSI statuses were identified more often in early-stage disease.⁶ The meta-analysis did not report on solid tumours in pediatric patients due to the limited evidence available.

According to the clinical experts consulted for this review, patients with metastatic solid tumours that have confirmed MSI-H or dMMR status have inferior outcomes (e.g., ORR, PFS, and DOR) when treated with conventional therapies, and tend to show improved outcomes with immune checkpoint inhibitors. They also noted that metastatic MSI-H or dMMR tumours can be characterized by faster disease progression during or after chemotherapy treatment, resulting in shorter survival times compared with patients with the same cancer stage and intact MMR. As a group, the former patients have a worse prognosis compared with those whose tumours are not MSI-H or dMMR, although prognosis varies based on tumour type.⁹

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 the CDA-AMC review team.

There is high variability in first-line treatments and in the SOC in solid tumours expressing MSI-H or dMMR mutations. In some instances, immunotherapy has become the SOC, with the clinical experts consulted for this review highlighting that pembrolizumab or other immune checkpoint inhibitors are available in earlier lines of therapy for most tumour types, regardless of MSI-H or dMMR status. In patients with MSI-H or dMMR unresectable or metastatic CRC, pembrolizumab received a recommendation to reimburse as a first-line therapy and is the currently funded SOC.¹⁰ Also, pembrolizumab received a recommendation to reimburse as a second-line or later therapy in adult patients with MSI-H or dMMR unresectable or metastatic endometrial cancer whose tumours have progressed following prior therapy and who have no satisfactory alternative treatment options, and it is currently funded in most jurisdictions.¹¹ Regardless of tumour type, patients with unresectable or metastatic MSI-H or dMMR cancers that have progressed after prior standard systemic therapy and who have no satisfactory options have a very poor prognosis, and the SOC for these patients is typically chemotherapy-based regimens, which provide limited clinical benefit and are associated with significant cumulative toxicity.

The proposed place in therapy for pembrolizumab in this submission is in accordance with the Health Canada–approved indication, i.e., as monotherapy for the treatment of adult and pediatric patients with unresectable or metastatic MSI-H or dMMR solid tumours, as determined by a validated test, that have progressed following prior treatment and who have no satisfactory alternative treatment options.²² The clinical experts consulted for this review noted that when pembrolizumab or other immune checkpoint inhibitors are available as SOC, it is unlikely that patients would receive pembrolizumab in later lines of therapy.

Drug Under Review

Pembrolizumab is a high-affinity antibody against PD-1 that exerts a dual-ligand blockade of the PD-1 pathway, including PD-L1 and PD-L2, on antigen-presenting and tumour cells.¹² By inhibiting the PD-1 receptor from binding to its ligands, pembrolizumab reactivates tumour-specific cytotoxic T lymphocytes in the tumour microenvironment.¹²

Pembrolizumab is available as a solution for infusion (100 mg per 4 mL vial) that is administered as an IV infusion over 30 minutes.^12,13 The recommended dosage in adult patients is either 200 mg every 3 weeks or 400 mg every 6 weeks until unacceptable toxicity or disease progression, or up to 24 months or 35 doses of 200 mg or 18 doses of 400 mg, whichever is longer, in patients without disease progression.^12,13 The recommended dosage in pediatric patients is 2 mg/kg (up to a maximum of 200 mg) every 3 weeks until disease progression or unacceptable toxicity, or up to 24 months or 35 doses, whichever is longer, in patients without disease progression.^12,13

In Canada, pembrolizumab has been issued market authorization to treat numerous types of cancers and has been reviewed by CDA-AMC for several indications. The Health Canada–approved indication and reimbursement request for this review is as monotherapy for the treatment of adult and pediatric patients with unresectable or metastatic MSI-H or dMMR solid tumours, as determined by a validated test, that have progressed following prior treatment and who have no satisfactory alternative treatment options.¹² Pembrolizumab received a Notice of Compliance on August 29, 2024.

Key characteristics of pembrolizumab and other treatments available for unresectable or metastatic MSI-H or dMMR solid tumours are summarized in Table 3.

Table 3: Key Characteristics of Pembrolizumab and Other Treatments for MSI-H or dMMR Tumours

b.i.d. = twice daily; CRC = colorectal cancer; dMMR = mismatch repair deficient; EGFR = epidermal growth factor receptor; FOLFIRI = leucovorin calcium (folinic acid) + fluorouracil + irinotecan hydrochloride; FOLFOX = leucovorin calcium (folinic acid) + fluorouracil + oxaliplatin; MSI-H = microsatellite instability-high; nab = nanoparticle albumin-bound; PD = pharmacodynamics; PK = pharmacokinetics; q.2.w. = every 2 weeks; q.3.w. = every 3 weeks; RNA = ribonucleic acid; SJS = Stevens-Johnson syndrome; TEN = toxic epidermal necrolysis; VEGF = vascular endothelial growth factor.

^aHealth Canada–approved indication.

Source: Product monographs and Cancer Care Ontario.^4,12,24-34

Testing Procedure Considerations

Different testing methods are available to detect MSI-H and/or dMMR status. The expression of the 4 MMR proteins is assessed by IHC,³⁵ whereas MSI-H status can be detected by NGS³⁶ or polymerase chain reaction. NGS can also help define tumour mutational burden, which can function as a surrogate marker for MSI-H or dMMR status.^37,38 According to the clinical experts consulted for this review, both IHC and NGS should be used concurrently to determine MSI-H and/or dMMR status. Molecular profiling using NGS and/or IHC has become part of SOC for many tumour types because targeted drugs tailored to be effective in the presence of certain alterations have likewise entered the SOC.³⁹

Both IHC and NGS testing can be conducted using archival tumour tissue collected during diagnostic biopsy or surgical resection. If archival tumour tissue is not available, NGS testing can be conducted in certain tumour types using circulating tumour DNA obtained through a peripheral blood sample.⁴⁰ A clinical expert noted that, in clinical practice, around 20% to 30% of patients do not have enough archival tissue available and may require testing of circulating tumour DNA; sometimes, a fresh biopsy is required.

The review team considered the potential impacts of MSI-H and/or dMMR testing conducted to ascertain eligibility for pembrolizumab as monotherapy for the treatment of adult and pediatric patients with unresectable or metastatic MSI-H or dMMR solid tumours that have progressed following prior treatment and who have no satisfactory alternative treatment options. Considerations included impacts to health systems, patients (including families and caregivers), and costs, and the team determined that these impacts are not anticipated to be substantial compared with the current SOC. Ethical considerations related to the increase in testing requirements are detailed in the Ethics Review report. Key considerations and relevant information were sourced from materials submitted by the sponsor, input from the clinical experts (including the clinical expert panel consulted by the review team), and sources from the literature. These were validated by the review team when possible and are summarized in Table 4.

Table 4: Considerations for MSI and MMR Testing for Establishing Treatment Eligibility With Pembrolizumab in Unresectable or Metastatic MSI-H or dMMR Solid Tumours

CDA-AMC = Canada’s Drug Agency; dMMR = mismatch repair deficient; ICI = immune checkpoint inhibitor; IHC = immunohistochemistry; MMR = mismatch repair; MSI = microsatellite instability; MSI-H = microsatellite instability-high; NGS = next-generation sequencing; PARP = poly(adenosine diphosphate ribose) polymerase; SOC = standard of care.

^aBudget impact analysis submitted by the sponsor.

^bCDA-AMC has not evaluated or critically appraised this evidence to determine its validity or reliability.

Perspectives of Patients, Clinicians, and Drug Programs

The full patient and clinician group submissions received by CDA-AMC are available in the consolidated patient and clinician group input document for this review on the project website.

Patient Group Input

This section was prepared by the CDA-AMC review team based on the input provided by 2 patient groups.

There were 2 patient group input submissions for this review, which were provided by Colorectal Cancer Canada and CCRAN (the latter working in collaboration with Canadian Cancer Survivor Network, Craig’s Cause Pancreatic Cancer Society, Canadian Breast Cancer Network, and Ovarian Cancer Canada). Colorectal Cancer Canada gathered information from an interview with 1 patient with CRC in Canada in 2024 for the current review, and through an online survey conducted in 2023 for a previous review that included responses from 15 patients with CRC and 1 caregiver. The CCRAN submission collected information from 6 patient and caregiver interviews across 4 pathologies. All contributions were from adult patients and/or their caregivers and do not include the perspectives of pediatric patients.

Patients described specific symptoms related to their cancer before beginning any treatment (e.g., weight changes, pain, fatigue, itching, difficulty sleeping and/or eating, vomiting, bloating, and jaundice). Patients stated that cancer symptoms had the greatest impact on their ability to work, exercise, and participate in social activities. After diagnosis, most patients received SOC chemotherapy for their specific type of cancer (e.g., FOLFOX [leucovorin calcium (folinic acid) + fluorouracil + oxaliplatin], FOLFIRI, capecitabine) and experienced disease progression. The most common intolerable side effects included neuropathy, nausea, and fatigue. Other side effects that impacted their HRQoL including insomnia, brain fog, digestive concerns, and hair loss. The negative effects that treatments had on their physical health resulted in detriments to their mental, social, and emotional health and prevented patients from participating in normal activities, volunteering, and working. This was further compounded for patients who lived in rural communities because travel and time away from their homes and loved ones for treatment was burdensome. The patient groups highlighted that the impact of diagnosis, disease, recurrence, and treatment went beyond the patient and was felt by their families and caregivers, friends, and communities.

The patients with experience with pembrolizumab gained access to the drug through various channels, including clinical trials, private payment, special access or compassionate access programs, or provincial funding. Treatment duration ranged from 9 to 50 cycles. The patients interviewed for the CCRAN submission had a terminal diagnosis when they started pembrolizumab and, at the time of the interview, most had no evidence of disease. While on pembrolizumab, patients indicated experiencing symptom relief and a quick and measurable response to treatment based on imaging and laboratory results. They also noted minimal side effects (fatigue, joint pain, inflammation, hypothyroidism, skin rash and dryness, and asymptomatic elevated lipase or liver enzymes), which patients considered to be an acceptable trade-off. Of those contributing to the CCRAN input, 2 of 6 patients had to withhold therapy due to elevated enzyme levels but remained asymptomatic. Patients reported various benefits of pembrolizumab, including that it was easier to receive than the prior treatments, required fewer concomitant medications (e.g., antiemetics, steroids), had a shorter infusion time, provided a better HRQoL, and left them feeling that they had avoided the long-term toxicities associated with conventional chemotherapies and the risks associated with invasive procedures.

Input from both patient groups emphasized the need for new, safe, and effective treatment options that are accessible to treat refractory, metastatic disease. According to the respondents, there were general concerns around a notable delay in diagnosis, variable access to tumour biomarker testing, and the need for patients to relocate for treatment and be away from their families and friends for extended periods of time. Furthermore, patients advocated for wide and equitable access to biomarker testing, treatments for tumours with appropriate biomarker status (i.e., MSI-H or dMMR) that go beyond tumour type and, especially, for the availability of drugs to treat rare and pediatric cancers. The most important outcomes noted by patients and caregivers included preventing disease progression and disease recurrence, prolonging survival, having a durable response, and improving HRQoL and symptoms such as pain.

Clinician Input

Input From Clinical Experts Consulted by CDA-AMC

All CDA-AMC 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). As part of the review of pembrolizumab, a panel of 3 clinical experts from across Canada was convened to characterize unmet therapeutic needs, assist in identifying and communicating situations when there are gaps in the evidence that could be addressed through the collection of additional data, promote the early identification of potential implementation challenges, gain further insight into the clinical management of patients living with a condition, and explore the potential place in therapy of the drug (e.g., potential reimbursement conditions). A summary of this panel discussion follows.

Unmet Needs

MSI-H and dMMR solid tumours are a diverse and heterogenous group of cancers that typically have a more aggressive phenotype and a reduced sensitivity to SOC chemotherapeutic agents. For adult patients with MSI-H and/or dMMR metastatic cancer, there is a need for effective therapeutic options if SOC or salvage chemotherapy fails. Meanwhile, for pediatric patients, there is a general need for new treatments because the current treatment options in both the front-line and relapsed setting are limited for this population (for example, pediatric high-grade glioma). The clinical experts highlighted that pembrolizumab has the potential to address this unmet need by providing an efficacious treatment option for this niche population.

Place in Therapy

The clinical experts highlighted the variability in treatment options and sequencing, depending on tumour site and diagnosis. In adult and pediatric patients, the clinical experts highlighted that, for the indication under review, pembrolizumab is intended for use after first-line therapy specific to the solid tumour site, although the ideal sequencing of treatments is unknown. The clinical experts highlighted that pembrolizumab is already indicated for use as SOC in non–MSI-H and/or non-dMMR tumours (e.g., kidney, urothelial, gastric, lung, biliary tract) and in some cancers with MSI-H and/or dMMR mutations (e.g., CRC, endometrial); thus, pembrolizumab may not be considered in later lines of therapy, depending on prior response.

For the pediatric population, 1 clinical expert highlighted that, in general, pembrolizumab and immune checkpoint inhibitors are used less frequently and, thus, experience with these is limited (except in specific subpopulations such as Hodgkin lymphoma). However, it was noted that therapies that are proven safe and effective in the pediatric population would be used as indicated, and access to such therapies is important, given the lack of available treatments, particularly in the relapsed setting.

The clinical experts noted that pembrolizumab would not be used in combination with existing treatments for any indication in this setting based on currently available data.

Patient Population

MSI-H and/or dMMR mutations are present across a broad range of cancer types, occurring in both common and rarer cancers. As previously noted, hypermutated tumours are generally more aggressive and do not often respond to SOC treatments. Per the clinical experts, this represents a population of patients currently in need of new treatments because the options are limited and of limited effectiveness. The clinical experts highlighted that underdiagnosis due to lack of testing is a risk, particularly in rarer cancers that are not reflexively tested using NGS or IHC, or in pediatric patients, among whom testing is even more limited.

The experts considered that the patients most likely to benefit from treatment with pembrolizumab would be those with confirmed MSI-H and/or dMMR positivity because, at this time, there are no other validated biomarkers or clinical factors that can better predict treatment benefit for these patients than MSI-H and/or dMMR status. One expert highlighted that up to 50% of patients will not benefit from treatment despite biomarker positivity. Additionally, the experts highlighted that, particularly for rarer cancers, treatment with SOC does not result in long-term benefits.

The clinical experts also highlighted that the patients least suitable for pembrolizumab are those who are negative for the MSI-H and/or dMMR biomarker, those who have previously received and/or did not respond to an immune checkpoint inhibitor during a previous line of therapy, and those with severe active autoimmune disease. However, the experts noted that in the case of certain, well-controlled autoimmune diseases, pembrolizumab may be used, but this would be decided in discussion with patients (and caregivers).

In the pediatric setting, 1 clinical expert highlighted that the population of patients with high-grade gliomas would benefit from pembrolizumab because such cancers are known to potentially have MSI and MMR mutations and, thus, testing for these biomarkers is more routine. Conversely, the prevalence of MSI-H and dMMR mutations appears much lower in pediatric patients; thus, testing for these mutations is uncommon (except for high-grade gliomas) because clinicians do not often see pediatric patients with some of the more common adult cancers that harbour MSI-H and/or dMMR mutations (e.g., CRC, endometrial, pancreatic). It was also noted that classic MSI testing based on a limited number of loci may not always identify patients, such as those with constitutional dMMR mutations, who nevertheless have significant MSI (albeit at other loci).

Assessing the Response to Treatment

Pembrolizumab has been available as a treatment in multiple cancers for many years, and the clinical experts noted that it is generally well tolerated. The primary aim of treatment in the spectrum of MSI-H and/or dMMR cancers may vary but, generally, disease control and prolongation of life with reduced toxicity are the most important goals of treatment in both the adult and pediatric populations.

The clinical experts noted that current clinical trials aim to address important outcomes that are assessed in clinical practice. These outcomes generally include objective response, disease stability, DOR, and HRQoL. The clinical experts noted that CRs are uncommon for these patients; thus, durability and stability are also important. Standard imaging tests (CT, MRI, PET, and so forth) are generally used to measure treatment response. Per the clinical experts, treatment tolerance is usually evaluated at each appointment, with radiologic response assessments typically occurring every 2 to 3 treatment cycles or approximately every 9 to 12 weeks, depending on local protocols.

Discontinuing Treatment

The clinical experts referred to several situations in which treatment with pembrolizumab may be discontinued, including intolerable AEs and clear evidence of disease progression (e.g., clinical [worsening of symptoms] or radiographic evidence). As previously noted, there is extensive experience with pembrolizumab in oncology. The clinical experts highlighted that immune-related AEs with pembrolizumab are generally mild and manageable with treatment interruption, symptomatic measures, or medical therapy (e.g., topical or oral steroids). However, in the case of severe immune-related AEs, pembrolizumab should be discontinued. One expert highlighted that in cases of oligometastatic progression, when progression is limited to 1 or 2 lesions, local therapies such as stereotactic body radiation therapy may be employed to treat the progressing lesions while continuing immune checkpoint therapy. Ultimately, the decision to discontinue treatment must carefully weigh the need to identify true disease progression against the risk of prematurely halting a therapy that may still offer significant clinical benefit to patients.

One additional concern highlighted by the clinical experts is the potential risk of premature discontinuation due to pseudoprogression. Pseudoprogression has been shown to occur with the use of immune checkpoint inhibitors due to immune infiltration that presents as disease progression on imaging. However, patients may continue to improve clinically or demonstrate a late response; thus, premature discontinuation is a risk that must be considered.

Prescribing Considerations

As noted, immune checkpoint inhibitors have become the SOC in the treatment of many adult cancers, and there is growing experience in their use across various health care settings, including academic and community settings; however, in the pediatric setting, this is likely limited to specialist tertiary hospitals.

Specialists, such as medical oncologists (as well as pediatric oncologists in the case of pediatric patients), are needed to select appropriate patients for treatment and oversee the management of therapy. This includes monitoring side effects, managing toxicities, and determining when to continue or discontinue treatment. The clinical experts noted that while specialists play a key role in these decisions, the actual delivery and routine monitoring of the treatment is currently handled by other health care professionals, such as general practitioners, especially in community hospitals and rural facilities, with proper guidance.

Clinician Group Input

This section was prepared by the CDA-AMC review team based on the input provided by 5 clinician groups.

Five clinician groups submitted input for this CDA-AMC review: the Society of Gynecologic Oncology of Canada and 4 Ontario Health (Cancer Care Ontario) Drug Advisory Committees (for genitourinary cancer, breast cancer, gynecology cancer, and CNS cancer). The Society of Gynecologic Oncology of Canada consists of health care professionals involved in the treatment and prevention of gynecologic cancer. Information for the Society of Gynecologic Oncology of Canada’s submission was based on completed and published clinical trials as well as the expert opinion of the physician members of the organization’s board of directors. The 4 Ontario Health Drug Advisory Committees, which provide timely evidence-based clinical and health system guidance on drug-related issues, gathered information for the submissions via email, videoconference, and published literature.

According to the 5 clinician groups, the most important goals of treatment include prolonging life, delaying disease progression, reducing the severity of cancer-related symptoms, improving HRQoL, and minimizing treatment-related toxicities and adverse effects. For patients with recurrent cancer, there is a lack of safe and effective treatment options and therapy is often palliative in nature. Although treatments for recurrent disease are available, prognoses remain poor because many treatments have not been shown to improve OS or disease-related symptoms and they are often toxic.

According to the clinician groups, patients whose disease has progressed on prior therapy and who have confirmed MSI-H or dMMR tumours could receive pembrolizumab monotherapy if they have not previously received the drug in the metastatic or unresectable setting. Based on input from the Society of Gynecologic Oncology of Canada, patients can be identified by IHC assessment of primary tissue for dMMR status, highlighting that testing is well validated in this setting and can be requested on an as-needed basis for patients with advanced or recurrent disease, and that misdiagnosis is rare. However, the breast cancer Drug Advisory Committee stated that MSI-H or dMMR testing is not standard for breast cancer in Ontario. The clinician groups highlighted that the patients who would not receive pembrolizumab are those with poor performance status, at risk of significant toxicity, or who have contraindications to immunotherapy.

Treatment response is assessed based on the patient’s symptoms, a clinical examination, a tumour assessment by diagnostic imaging (CT or MRI), or tumour markers as determined by the treating physician. Symptoms and a clinical examination are completed before each treatment cycle, while diagnostic imaging is typically performed every 2 to 3 cycles (every 6 to 9 weeks). According to some clinician groups, a clinically meaningful response includes prolonged survival, maintaining radiographic disease control (i.e., tumour response or stabilization) with good treatment tolerance, and stable or improving symptoms (e.g., symptoms related to pain, vomiting, or functional status).

In line with the clinical experts consulted for this review, the CNS Cancer Drug Advisory Committee cautioned against premature discontinuation based on pseudoprogression. Patients with clinically stable disease would continue treatment; however, repeat imaging that confirms ongoing disease progression would warrant treatment discontinuation. For immune-related toxicities, treatment should be withheld and managed as per standard guidelines and, if the event has resolved, treatment could be reinitiated at the physician’s discretion; however, grade 4 toxicity would require discontinuation. Contraindication to immunotherapy, significant drug intolerance, and patient choice are other reasons for stopping the drug.

The clinician groups and the clinical experts consulted for this review agreed that patients should be managed at cancer care centres (e.g., hospital outpatient settings, tertiary cancer care centres, community satellite clinics) by oncologists with expertise in treating advanced disease, using systemic therapies, and managing immune-related AEs.

Drug Program Input

The drug programs provide input on each drug being reviewed through CDA-AMC 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 CDA-AMC are summarized in Table 5.

Table 5: Summary of Drug Plan Input and Clinical Expert Response

CDA-AMC = Canada’s Drug Agency; CNS = central nervous system; CRC = colorectal cancer; dMMR = mismatch repair deficient; ECOG PS = Eastern Cooperative Oncology Group Performance Status; EGFRi = epidermal growth factor receptor inhibitor; IHC = immunohistochemistry; MSI = microsatellite instability; MSI-H = microsatellite instability-high; NGS = next-generation sequencing; PCR = polymerase chain reaction; pERC = pan-Canadian Oncology Drug Review Expert Review Committee; SOC = standard of care; VEGF = vascular endothelial growth factor.

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 pembrolizumab 200 mg IV or pembrolizumab 2 mg/kg IV in the treatment of patients with unresectable or metastatic MSI-H or dMMR solid tumours, as determined by a validated test, that have progressed following prior treatment and who have no satisfactory alternative treatment options. The focus will be placed on comparing pembrolizumab with relevant comparators and identifying gaps in the current evidence.

A summary of the clinical evidence included by the sponsor in the review of pembrolizumab is presented in 4 sections, with a critical appraisal of the evidence included at the end of each section. The first section, the systematic review, includes pivotal studies and randomized controlled trials (RCTs) that were selected according to the sponsor’s systematic review protocol. The CDA-AMC 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. The third section includes indirect evidence from the sponsor. The fourth section includes additional studies that were considered by the sponsor to address important gaps in the systematic review evidence.

Included Studies

Clinical evidence from the following is included in the CDA-AMC review and appraised in this document:

• 3 pivotal studies identified in the systematic review

• ITCs for 4 tumour types (CRC, endometrial cancer, small intestine cancer, and gastric cancer)

• 2 additional studies addressing gaps in the evidence.

Systematic Review

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

Description of Studies

Three clinical trials with evidence relevant to the targeted reimbursement request were included in this review: KEYNOTE-158,⁴⁸ KEYNOTE-164,⁴⁹ and KEYNOTE-051.¹⁹

KEYNOTE-158 Trial

KEYNOTE-158 is an ongoing, single-arm, open-label, phase II basket trial that evaluated pembrolizumab monotherapy in adult patients with advanced (unresectable and/or metastatic) MSI-H solid tumours that have progressed on SOC therapy. Patients received pembrolizumab 200 mg intravenously every 3 weeks until progressive disease, unacceptable toxicity, or an intercurrent illness that prevents further administration of treatment. The primary end point was ORR and the secondary end points were DOR, PFS, OS, safety, and tolerability. Patients were enrolled across 60 centres in 18 countries (included 16 patients from 2 sites in Quebec and Ontario, Canada). A total of 373 patients with MSI-H and/or dMMR cancers were assessed in the KEYNOTE-158 trial.⁴⁸ The database cut-off for the data presented herein was January 12, 2022.

KEYNOTE-164 Trial

KEYNOTE-164 is a single-arm, open-label, phase II trial that evaluated pembrolizumab in adult patients with locally advanced unresectable or metastatic (stage IV) dMMR or MSI-H CRC whose tumours have progressed on SOC therapy. The KEYNOTE-164 trial included patients in 2 cohorts, cohort A and cohort B, all of whom had received prior standard treatments. For cohort A, patients had to have been treated with at least 2 lines of SOC therapies that must have included fluoropyrimidine, oxaliplatin, and irinotecan. For cohort B, patients were required to have had at least 1 line of systemic SOC, including fluoropyrimidine plus oxaliplatin or fluoropyrimidine plus irinotecan, with or without an anti-VEGF or anti-epidermal growth factor receptor (EGFR) monoclonal antibody. All patients received pembrolizumab 200 mg intravenously every 3 weeks until progressive disease, unacceptable toxicity, or an intercurrent illness that prevents further administration of treatment. The primary end point was ORR and the secondary end points were DOR, PFS, OS, safety, and tolerability. Patients were enrolled across 34 centres in 10 countries (included 5 patients from 1 site in Quebec, Canada). A total of 124 patients were assessed in the KEYNOTE-164 trial.⁴⁹ The database cut-off for the data presented herein was February 19, 2021.

KEYNOTE-051 Trial

The KEYNOTE-051 trial is an ongoing, single-arm, open-label, phase I and II basket trial that evaluated pembrolizumab in pediatric patients with MSI-H or dMMR cancers that has progressed on SOC therapy.⁵⁰ Patients received pembrolizumab 2 mg/kg intravenously every 3 weeks for 35 cycles (approximately 24 months). The primary end points were safety, tolerability, and ORR, and the secondary end points were DOR, PFS, and OS. A total of 7 patients were included in the MSI-H cancer cohort in the KEYNOTE-051 trial.¹⁹ Patients were enrolled from 58 centres across 12 countries, including 2 sites in Canada. However, it is unclear from which sites the 7 patients were evaluated. The database cut-off for the data presented herein was January 18, 2022.

Table 6: Details of Sponsor Studies Included in the Systematic Review

AE = adverse event; CNS = central nervous system; CRC = colorectal cancer; dMMR = mismatch repair deficient; DOR = duration of response; ECOG PS = Eastern Cooperative Oncology Group Performance Status; EGFR = epidermal growth factor receptor; EORTC QLQ-C30 = European Organisation for Research and Treatment of Cancer Quality of Life Questionnaire Core 30; HRQoL = health-related quality of life; INRC = International Neuroblastoma Response Criteria; IRC = independent review committee; mAb = monoclonal antibody; MSI = microsatellite instability; MSI-H = microsatellite instability-high; ORR = objective response rate; OS = overall survival; PFS = progression-free survival; q.3.w. = every 3 weeks; RECIST 1.1 = Response Evaluation Criteria in Solid Tumours Version 1.1; SOC = standard of care; VEGF = vascular endothelial growth factor.

Sources: Marabelle et al. (2020),⁴⁸ Maio et al. (2022),^51-53 Le et al. (2020),⁴⁹ Le et al. (2021),⁵⁴ Le et al. (2023),⁵⁵ Geoerger et al. (2020),⁵⁶ sponsor’s clinical summary for Keytruda for MSI-H,¹³ clinicaltrials.gov,³³ and Clinical Study Reports for the KEYNOTE-158,¹⁷ KEYNOTE-051,¹⁹ and KEYNOTE-164¹⁸ trials.

Populations

Inclusion and Exclusion Criteria

The eligibility criteria for the 3 KEYNOTE trials included in this review are summarized in Table 6.

The KEYNOTE-158 trial enrolled adult patients with advanced MSI-H or dMMR non-CRC cancers who had disease progression following prior standard first-line therapy. Patients were enrolled into cohorts A through J, depending on their cancer type (anal, biliary, neuroendocrine, endometrial, cervical, vulvar, small cell lung, mesothelioma, thyroid, or salivary gland); MSI-H status was retrospectively tested in these patients. Cohort K prospectively included patients with any solid tumour (except CRC) that was MSI-H. Only patients with MSI-H, non-CRC solid tumours (cohorts A through K) were of interest to this review. Patients had to have adequate organ function and an ECOG PS score of 0 or 1, and had to provide an archival or newly obtained tissue sample from a tumour lesion that had not been previously irradiated. Patients who had previously received anti–PD-1, anti–PD-L1, or anti–PD-L2 drugs were excluded.

The KEYNOTE-164 trial enrolled adult patients with advanced MSI-H or dMMR CRC into either cohort A, if they had been previously treated with at least 2 lines of SOC therapies that must have included fluoropyrimidine, oxaliplatin, and irinotecan, or cohort B, if they had been previously treated with at least 1 line of systemic SOC therapy (fluoropyrimidine and oxaliplatin or fluoropyrimidine plus irinotecan, with or without anti-VEGF or anti-EGFR monoclonal antibody). Patients had to have adequate organ function and an ECOG PS score of 0 or 1, and had to provide an evaluable tissue sample for biomarker analysis taken from a tumour lesion that had not been previously irradiated. Patients who had previously received anti–PD-1, anti–PD-L1, or anti–PD-L2 drugs were excluded.

The KEYNOTE-051 trial enrolled pediatric patients aged 6 months to 18 years with MSI-H or dMMR cancers who had disease progression following prior therapy. Patients had to have adequate organ function and an ECOG PS score of 0 or 1, and had to provide an archival or newly obtained tissue sample of a tumour lesion that had not been previously irradiated. Patients who had previously received anti–PD-1, anti–PD-L1, or anti–PD-L2 drugs were excluded.

Interventions

For the KEYNOTE-158 and KEYNOTE-164 trials, patients received pembrolizumab 200 mg intravenously every 3 weeks until progressive disease, unacceptable toxicity, or an intercurrent illness that prevents further administration of treatment. For the KEYNOTE-051 trial, patients received pembrolizumab 2 mg/kg intravenously every 3 weeks for 35 cycles (approximately 24 months).^52,55,56 Patients in the KEYNOTE-158, KEYNOTE-164, and KEYNOTE-051 trials were prohibited from receiving certain therapies during the screening and treatment phase, including any re-treatment following relapse. The restricted therapies included antineoplastic systemic chemotherapy or biological therapy, any immunotherapy or chemotherapy not specified in the protocol, investigational drugs apart from pembrolizumab, and radiation therapy. Additionally, live vaccines were not allowed within 30 days before the first dose of trial treatment or during the trial. Systemic glucocorticoids were allowed only in specific circumstances, such as managing AEs for symptoms of immunologic etiology, with the sponsor’s approval.^52,55,56

At the discretion of the investigator, patients with stable disease or better who discontinued pembrolizumab treatment after 24 months for reasons other than disease progression or intolerability, or participants who experienced a CR and stopped pembrolizumab treatment after 8 cycles and who had received at least 2 cycles administered after the date of the initial CR, could be eligible for up to 1 year of pembrolizumab re-treatment (second course) upon experiencing disease progression.^52,55,56

Outcomes

A list of efficacy end points assessed in this Clinical Review report is provided in Table 7, followed by descriptions of the outcome measures. Summarized end points are based on 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 by CDA-AMC and the input from the patient and clinician groups and public drug plans. Using the same considerations, the CDA-AMC review team selected end points that were considered to be most relevant to inform the deliberations of the CDA-AMC expert committee, and finalized this list of end points in consultation with members of the committee. All summarized efficacy end points were assessed using GRADE. Select notable harms outcomes that were considered important for informing the deliberations of the expert committee were also assessed using GRADE. The patient and clinician groups, as well as the clinical experts consulted, indicated that durable response, prevention of progression, prolongation of life, and improved HRQoL with a low number of harms were important goals of treatment. This was reflected in the included end points: ORR, DOR, PFS, OS, HRQoL, SAEs, and AEOSIs (immune-mediated and infusion-related reactions).

Table 7: Outcomes Summarized From the Included Studies

AEOSI = adverse event of special interest (immune-mediated event or infusion-related reaction); DOR = duration of response; EORTC QLQ-C30 = European Organisation for Research and Treatment of Cancer Quality of Life Questionnaire Core 30; NR = not reported; ORR = objective response rate; OS = overall survival; PFS = progression-free survival; RECIST 1.1 = Response Evaluation Criteria in Solid Tumours Version 1.1; SAE = serious adverse event.

Sources: Marabelle et al. (2020),⁴⁸ Maio et al. (2022),^51-53 sponsor’s clinical summary for Keytruda for MSI-H,¹³ clinicaltrials.gov,³³ and Clinical Study Reports for the KEYNOTE-158,¹⁷ KEYNOTE-051,¹⁹ and KEYNOTE-164¹⁸ trials.

Efficacy Outcomes

Overall Survival

OS was defined as the time from either treatment allocation (KEYNOTE-158 trial) or treatment initiation (KEYNOTE-164 and KEYNOTE-051 trials) to death from any cause. In cases where a patient did not die but started a new anticancer therapy, the patient was censored at the time of the new treatment initiation. Missing OS observations were censored at the last recorded assessment. Patients without efficacy evaluation data or without survival data were censored at day 1 in the OS analyses.

Progression-Free Survival

PFS was defined as the time from day 1 of study treatment to the first documented disease progression per RECIST 1.1 or death due to any cause, whichever occurred first. In cases where a patient did not die or experience disease progression but started a new anticancer therapy, the patient’s data were censored at the time of treatment initiation. Missing PFS observations were censored at the last recorded assessment. Patients without efficacy evaluation data or survival data were censored at day 1 in the PFS analyses. The PFS end point was assessed by independent central radiological review (KEYNOTE-164 and KEYNOTE-158 trials) or by the investigator (KEYNOTE-051 trial).

Objective Response Rate

ORR was defined as the proportion of patients who have a confirmed CR or PR per RECIST 1.1 assessed by an independent central radiological review (KEYNOTE-158 and KEYNOTE-164 trials) or by the investigator (KEYNOTE-051 trial). Patients with no assessment (unknown or missing) were considered to have not experienced a response and were included in the denominator.

Duration of Response

DOR was assessed among patients with a CR or PR and defined as the time from first documented response to subsequent disease progression (per RECIST 1.1) or death from any cause, whichever occurred first. In the KEYNOTE-158 and KEYNOTE-164 trials, patients who were ongoing in the study, had started a subsequent treatment, or had missed at least 2 disease assessments were censored at the last adequate disease assessment. Patients who died or progressed after 1 missed assessment were considered to have had an event. Patients who were alive, had not progressed, had not initiated new anticancer treatment, and were not lost to follow-up were considered ongoing responders at the time of analysis.

Health-Related Quality of Life

HRQoL was only assessed in the KEYNOTE-158 trial using the EORTC QLQ-C30. The EORTC QLQ-C30 contains 30 items and measures, 5 functional dimensions (physical, role, emotional, cognitive, and social), 3 symptom items (fatigue, nausea or vomiting, and pain), 6 single items (dyspnea, sleep disturbance, appetite loss, constipation, diarrhea, and financial impact), and a global health and QoL scale. A summary of the questionnaire, its measurement properties, and estimated minimal important difference (MID) is in Table 8. Prespecified exploratory HRQoL end points included changes from baseline to week 9 in the EORTC QLQ-C30 GHS and QoL.¹³

A mean difference of 10 points or more has been estimated as a MID when interpreting data collected with the EORTC QLQ-C30 questionnaire.⁵⁷ However, a more recent review has estimated an MID for within-group improvement in GHS ranging from 4 points (brain cancer) to 13 points (ovarian cancer), and for deterioration ranging from −5 points (head or neck, lung cancers) to −8 points (breast cancer).^58,59

Harms Outcomes

AE reporting was identical across the KEYNOTE-158, KEYNOTE-164, and KEYNOTE-051 studies. AEs, irrespective of causality, were reported from the time of treatment initiation through 30 days after the last dose of study treatment or before the initiation of a new anticancer treatment, whichever occurred first (intensity was assessed by the investigator according to the National Cancer Institute Common Terminology Criteria for Adverse Events Version 4.03).

SAEs reported from the time of treatment initiation until 90 days after the last dose of study treatment or 30 days after discontinuation of study treatment if the patient initiated a new anticancer treatment, whichever occurred first (intensity was assessed by the investigator according to the National Cancer Institute Common Terminology Criteria for Adverse Events Version 4.03). SAEs included AEs that result in death or are life-threatening, require inpatient hospitalization or prolong existing hospitalization, or result in persistent or significant disability and/or incapacity. SAEs also included congenital anomalies and/or births, deaths, or other important medical events.

Notable harms of interest to this review included immune-mediated events and infusion-related reactions associated with pembrolizumab. A predefined list of preferred terms was developed by the sponsor to consistently characterize the nature and frequency of each AEOSI regardless of causality and as reported by investigators. These preferred terms are considered medically equivalent to immune-mediated events and infusion-related reactions.

Table 8: Summary of Outcome Measures and Their Measurement Properties

dMMR = mismatch repair deficient; ECOG PS = Eastern Cooperative Oncology Group Performance Status; EORTC QLQ-C30 = European Organisation for Research and Treatment of Cancer Quality of Life Questionnaire Core 30; MID = minimal important difference; MSI-H = microsatellite instability-high.

Statistical Analysis

Sample Size and Power Calculation

Sample size considerations were included for all KEYNOTE trials. For KEYNOTE-158, the trial planned to enrol a minimum of 200 patients and up to approximately 1,350 patients over a period of around 90 months. Because both the sample size and the underlying response rate were unknown, no power calculations were included in this study.

In the KEYNOTE-164 trial, approximately 60 patients were planned for each of cohorts A and B. For cohort A, with 60 patients, the study had 93% power to reject the null hypothesis of an ORR of 15%, with a 1-sided type I error rate of 2.5%, assuming a true ORR of 35%. Historically, the response rate had been less than 5%. For cohort B, with a historical response rate of about 20% and a sample size of 60 patients, if at least 19 responders were observed, the lower bound of the 95% CI for ORR would have exceeded 20%.

In the KEYNOTE-051 trial, power calculations were for the “all participants as treated” (APaT) population and had 25 patients per indication planned for enrolment. With up to 25 patients per indication, the study had 84% power to show that the ORR induced by pembrolizumab exceeded 10% at a 1-sided alpha level of 8%, assuming a true ORR of 35%. The null hypothesis of 10% assumed that the population consisted of patients with incurable solid tumours that had not responded to multiple lines of standard therapy.

Multiplicity

The KEYNOTE-158 trial allowed for multiple interim analyses to allow for hypothesis generation. Analyses were performed without multiplicity control.

In the KEYNOTE-164 trial, for cohort A, the overall type I error rate for ORR analyses was controlled at 2.5% (1-sided) using a group sequential approach for the interim analysis and final analysis. For cohort B, no multiplicity adjustment was applied.

In the KEYNOTE-051 trial, interim analyses were performed to monitor the ORR of the enrolled patients in the APaT set. Within a tumour type, after at least 10 patients had had at least 1 postbaseline response assessment, a sequential monitoring procedure was used to evaluate for efficacy and futility simultaneously, based on the number of patients with a confirmed or unconfirmed response. The design assumed a 1-sided type I error rate of 8%.

Statistical and Analytical Plans

The efficacy analyses included in this submission were based on the APaT populations (or relevant subsets of those populations) and included all participants with MSI-H and/or dMMR tumours in cohorts A through K who received at least 1 dose of pembrolizumab in the KEYNOTE-158 trial (N = 373), all participants in cohorts A and B who received at least 1 dose of pembrolizumab in the KEYNOTE-164 trial (N = 124), and all participants with MSI-H and/or dMMR tumours who received at least 1 dose of pembrolizumab in the KEYNOTE-051 trial (N = 7).

The data cut-off dates for the analyses were January 12, 2022, for the KEYNOTE-158 trial (interim analysis); February 19, 2021, for the KEYNOTE-164 trial (final analysis); and January 18, 2022, for the KEYNOTE-051 trial (interim analysis). The analysis methods were similar across all trials. The point estimate and 95% CI for the ORR, based on RECIST 1.1, were provided using an exact binomial distribution (Clopper and Pearson) method. Participants without response data (i.e., data were nonevaluable, missing, or unknown) were counted as nonresponders but included in the denominator. In the KEYNOTE-164 trial, a P value was to be provided for testing that showed the ORR was greater than 15% (cohort A); results for cohort B were descriptive. For the KEYNOTE-158 and KEYNOTE-051 trials, the results were provided descriptively. Time-to-event outcomes were presented as median values with 95% CIs. DOR, PFS, and OS were summarized by Kaplan-Meier (KM) methods.

In the KEYNOTE-158 trial, ORR and DOR were provided for the overall population as well as by tumour type. Results for the KEYNOTE-164 trial were provided for cohorts A and B and pooled within the sponsor’s submission. Results by tumour type were not provided for the KEYNOTE-051 trial because all but 1 patient in that trial had brain cancer. Sensitivity analyses were not described.

In the KEYNOTE-158 trial, an adherence summary was provided, defined as the proportion of patients who completed at least 1 item out of all eligible patients who were expected to complete questionnaires at a given time point. This calculation excluded patients missing due to the study design, such as death or study discontinuation. LS mean (95% CI) changes from baseline to week 9 in the EORTC QLQ-C30 GHS and QoL scores were presented using a repeated measures model, with missing data assumed to be missing at random. The proportion of participants showing improvement, deterioration, or stability in EORTC QLQ-C30 scores from baseline to week 9 was summarized with no imputation for missing data. Postbaseline scores were classified as “improved,” “stable,” or “deteriorated” according to a change of 10 points or more for each EORTC QLQ-C30 scale or subscale.

Analysis Populations

Analysis sets included in the KEYNOTE-158, KEYNOTE-164, KEYNOTE-051 trials are summarized in Table 9.

Table 9: Analysis Populations for the KEYNOTE-158, KEYNOTE-164, and KEYNOTE-051 Trials

APaT = all participants as treated; dMMR = mismatch repair deficient; FAS = full analysis set; HRQoL = health-related quality of life; MSI-H = microsatellite instability-high.

Sources: KEYNOTE-158,¹⁷ KEYNOTE-051,¹⁹ and KEYNOTE-164¹⁸ Clinical Study Reports.

Results

Patient Disposition

In all 3 trials, all allocated patients received treatment: 373 (100%) in the KEYNOTE-158 trial, 124 (100%) in the KEYNOTE-164 trial, and 7 (100%) in the KEYNOTE-051 trial (Table 10). Data reported in the sponsor’s summary of clinical evidence for the KEYNOTE-164 trial pooled cohorts A and B. The proportion of patients with unsuccessful screening were not reported for the KEYNOTE-158 or KEYNOTE-051 trial. For the KEYNOTE-164 trial, 148 patients were screened, with 24 (16.2%) being screened out due to eligibility criteria not being met.

Study discontinuation rates were 62.5% in the KEYNOTE-158 trial, 55.6% in the KEYNOTE-164 trial, and 71.4% in the KEYNOTE-051 trial. The primary reason for discontinuation was death, occurring in 59.0% of patients in the KEYNOTE-158 trial, 46.0% in the KEYNOTE-164 trial, and 71.4% in the KEYNOTE-051 trial. Other reasons included withdrawal by the patient and AEs, with AEs causing discontinuation in 3.2% of patients in the KEYNOTE-164 trial, but no patients withdrew due to AEs in the KEYNOTE-158 or KEYNOTE-051 trials.

Table 10: Summary of Patient Disposition in the Systematic Review for Sponsor Studies

APaT = all patients as treated; FAS = full analysis set; MSI-H = microsatellite instability-high; NA = not applicable; NR = not reported.

Sources: KEYNOTE-158,¹⁷ KEYNOTE-051,¹⁹ KEYNOTE-164¹⁸ Clinical Study Reports.

Baseline Characteristics

The summary of baseline characteristics in the 3 included trials is presented in Table 11. The baseline characteristics outlined are limited to those that are most relevant to this review or were felt to affect the outcomes or interpretation of the study results.

The KEYNOTE-158 trial enrolled 373 adult patients with MSI-H and/or dMMR tumours. The majority of the patients were female (61%), with a mean age of 59.2 (SD = 13.1) years, and 36% of the patients were aged 65 years or older. The most common cancer types in this cohort were colorectal (25%), endometrial (14%), and gastric (14%); there was a smaller representation (4% to 7%) of other cancers, including small intestine, ovarian, biliary, pancreatic, brain, sarcomas, and other rare cancers. Regarding metastatic status, 91% of the patients were classified as having M1 disease, indicating the presence of metastases. Regarding ECOG PS, 46% of the patients had an ECOG PS of 0, and 54% had an ECOG PS of 1. Most patients had undergone prior therapy, with 42% receiving 1 prior line of treatment and 56% receiving 2 or more lines.

The KEYNOTE-164 trial included 124 patients with MSI-H and/or dMMR CRC. The sex distribution for patients was 56% male and 44% female. The mean age was 56.1 years (SD = 14.9 years), and 33% were aged 65 years or older. Regarding ECOG PS, 41% of the patients had a status of 0 and 59% had a status of 1. In terms of metastatic status, 97% of patients were classified as M1, indicating metastatic disease. The majority of the patients (76%) had received 2 or more prior lines of treatment.

The KEYNOTE-051 trial included a cohort of 7 pediatric patients with MSI-H and/or dMMR tumours. The majority of the patients (71%) were female, and the age of patients ranged from 3 to 16 years, with a mean age of 11 years (SD = 4.3 years). This cohort mainly consisted of patients with brain tumours (86%), with 1 patient having another type of cancer (14%). Most patients (71%) were classified as having M1 disease, indicating their cancer was metastatic, and 57% of the patients had undergone 2 or more prior lines of therapy.

Table 11: Summary of Baseline Characteristics — Participants With MSI-H and/or dMMR Cancers in the KEYNOTE-158, KEYNOTE-164, and KEYNOTE-051 Trials (APaT)

APaT = all participants as treated; dMMR = mismatch repair deficient; ECOG PS = Eastern Cooperative Oncology Group Performance Status; MSI-H = microsatellite instability-high; M0 = cancer has not spread to other parts of the body; M1 = cancer has spread to other parts of the body; MX = metastasis cannot be measured; NR = not reported; SD = standard deviation.

Note: The data cut-off dates were January 12, 2022, for the KEYNOTE-158 trial (interim analysis), February 19, 2021, for the KEYNOTE-164 trial (final analysis), and January 18, 2022, for the KEYNOTE-051 trial (interim analysis).

Sources: Clinical Study Report for KEYNOTE-158¹⁷ and sponsor’s clinical summary.¹³

Exposure to Study Treatments

Patient exposure for the included studies is presented in Table 12.

In the KEYNOTE-158 trial, the median duration of treatment with pembrolizumab was 6.18 months (range, 0.03 to 29.67 months) and the mean duration of treatment was 10.19 months (SD = 9.42 months). Patients received a median treatment of 9 cycles (range, 1 to 35 cycles), with a mean of 15.06 cycles (SD = 13.14 cycles). Among all patients, 85.3% were treated for at least 1 month, 65.4% for 3 months or more, 50.4% for 6 months or more, and 37.0% for 12 months or more. Additionally, 28.2% of patients were treated for 18 months or more, and 25.7% were treated for 21 months or more.

In the KEYNOTE-164 trial, the median duration of treatment with pembrolizumab was 273 days (approximately 9.0 months) in cohort A (range, 1 to 1,405 days) and 226 days (approximately 7.4 months) in cohort B (range, 1 to 1,157 days). The mean duration of treatment was 427.41 days (SD = 403.01 days) for cohort A and 382.62 days (SD = 350.65 days) for cohort B. Patients received a median treatment of 14 cycles in cohort A (range, 1 to 52 cycles) and 11 cycles in cohort B (range, 1 to 48 cycles), with a mean of 18.90 cycles (SD = 15.10 cycles) in cohort A and 17.52 cycles (SD = 14.75 cycles) in cohort B. All patients (100%) were treated for more than 0 months, with 89.5% treated for at least 1 month, 73.4% for 3 months or more, 54.0% for 6 months or more, and 43.5% for 12 months or more. Additionally, 16.9% of patients were treated for 21 months or more, and 7.3% were treated for 24 months or more.

In the KEYNOTE-051 trial, which included 7 patients with MSI-H tumours, the median duration of treatment with pembrolizumab was 43 days (approximately 1.4 months; range, 1 to 453 days). The mean duration of treatment was 114.9 days (SD = 165.4 days). Patients received a median treatment of 3 cycles (range, 1 to 21 cycles), with a mean of 6.1 cycles (SD = 7.4 cycles). All patients (100%) were treated for more than 0 months, with 71.4% treated for at least 1 month. However, only 28.6% of patients were treated for 3 months or more, and the same proportion (28.6%) was treated for 6 months or more. One patient (14.3%) was treated for 12 months or more.

Table 12: Summary of Patient Exposure in the Systematic Review for Sponsor Studies

MSI-H = microsatellite instability-high; NR = not reported; SD = standard deviation.

Sources: KEYNOTE-158,¹⁷ KEYNOTE-051,¹⁹ and KEYNOTE-164 Clinical Study Reports.¹⁸ Details included in the table are from the sponsor’s summary of clinical evidence.

Concomitant Treatments

In the KEYNOTE-158 trial, a total of 5 patients (1.3%) received concomitant antineoplastic agents, 8 patients (2.1%) were receiving endocrine treatment, 3 patients (0.8%) were receiving immunostimulants, and 6 patients (1.6%) were receiving immunosuppressants.

In cohort A in the KEYNOTE-164 trial, a total of 25 patients (41.0%) received concomitant antineoplastic agents and 3 patients (4.9%) received endocrine therapy. In cohort B, a total of 24 patients (38.1%) received concomitant antineoplastic agents and 1 patient (1.6%) received an immunostimulant.

In the KEYNOTE-051 trial, all patients received concomitant treatments. The most frequently reported concomitant medications were acetaminophen and levetiracetam, in 4 participants each.

Efficacy

The summary of the efficacy outcomes for the sponsor’s studies is presented in Table 13 and Table 14. In the KEYNOTE-158 trial, the median duration of follow-up was 17 months (range, 0.2 to 71.4 months). For the KEYNOTE-164 trial, the median duration of follow-up was 31.4 months (range, 0.2 to 65.2 months) for cohort A and 52.7 months (range, 0.1 to 56.6 months) for cohort B. In the KEYNOTE-051 trial, the median length of follow-up was 5.2 months (range, 0.3 to 28.2 months).

Objective Response Rate

Adult

The ORR per RECIST 1.1 in the KEYNOTE-158 trial (varied tumours) was 33.8% (95% CI, 29.0% to 38.8%). The best objective response included 40 patients (10.7%) with a CR and 86 patients (23.1%) with a PR. The tumour-specific ORR in the KEYNOTE-158 trial showed varied response rates, ranging from 4.8% (brain) to 59.3% (small intestine) across different cancer types (Table 43). Tumours with higher ORRs compared with the pooled result included endometrial, gastric, small intestine, cholangiocarcinoma, urothelial, and salivary cancers. Tumours with lower ORRs included ovarian, pancreatic, brain, sarcoma, breast, cervical neuroendocrine, prostate, adrenocortical, mesothelioma, thyroid, small cell lung, and renal cancers.

The ORR in cohort A was 32.8% (95% CI, 21.3% to 46.0%). The objective response included a CR in 3 patients (4.9%) and a PR in 17 patients (27.9%), with 11 patients (18.0%) experiencing stable disease and 28 patients (45.9%) experiencing progressive disease. The ORR in cohort B was 34.9% (95% CI, 23.3% to 48.0%). The objective response included a CR in 9 patients (14.3%) and a PR in 13 patients (20.6%), with 13 patients (20.6%) experiencing stable disease and 25 patients (39.7%) experiencing progressive disease.

Pediatric

In the KEYNOTE-051 trial, the ORR was 0% (95% CI, 0.0% to 41.0%). No patients experienced a CR or PR; 1 patient (14.3%) had stable disease. Tumour-specific ORR results were not reported in the KEYNOTE-051 trial.

Table 13: Key Efficacy Outcomes in the KEYNOTE-158, KEYNOTE-164, and KEYNOTE-051 Trials, APaT Populations

APaT = all participants as treated; CI = confidence interval; CR = complete response; CRC = colorectal cancer; dMMR = mismatch repair deficient; DOR = duration of response; MSI-H = microsatellite instability-high; NA = not applicable; NE = not evaluable; NR = not reported; ORR = objective response rate; OS = overall survival; PFS = progression-free survival; PR = partial response; RECIST 1.1 = Response Evaluation Criteria in Solid Tumours Version 1.1; TTR = time to response.

Note: “+” indicates the TTR event (e.g., response, progression, survival) was ongoing at the time of the analysis.

^aOnly tumours affecting ≥ 20 participants are reported.

^bFrom product-limit (Kaplan-Meier) method for censored data.

^cPatients with no postbaseline assessment at data cut-off, including missing, discontinued, or death before the first postbaseline scan. One of the 7 patients in the MSI-H cohort of the KEYNOTE-051 trial received the first dose of pembrolizumab just 9 days before the data cut-off date and therefore, no postbaseline imaging scans were available for efficacy analyses.

Sources: Sponsor’s clinical summary.¹³

Duration of Response

Adult

In the KEYNOTE-158 trial (varied tumours), among 126 patients with a response, the median DOR was 63.2 months (range, 1.9 to 63.9 months). Both ends of the range represent patients who were ongoing in the study without the event at the time of the data cut-off. Tumour-specific median DOR was as follows: endometrial cancer (63.2 months), gastric cancer (not reached), small intestine cancer (not reached), ovarian cancer (not reached), biliary tract cancer (30.6 months), pancreatic cancer (not reached), and brain cancer (18.9 months). Among patients who experienced a response, the event-free probabilities at 6, 12, 24, 36, 48, and 60 months after the first response were 95.2%, 88.5%, 72.3%, 69.4%, 66.0%, and 66.0%, respectively.

In the KEYNOTE-164 trial (all patients with CRC), among 42 patients with a response, the median DOR was not reached (range, 4.4 to 58.5 months for patients ongoing with response). Among patients who experienced a response, the event-free probabilities at 6, 12, 18, 24, and 36 months after the first response were 97.6%, 95.1%, 92.2%, 92.2%, and 92.2% respectively. The median DORs in cohorts A (n = 20) and B (n = 22) were not reached (range, 6.2 to 58.5 months for patients ongoing with response) and not reached (range, 4.4 to 52.4 months for patients ongoing with response), respectively.

Pediatric

DOR was not assessed in the KEYNOTE-051 trial because no patients had a response.

Overall Survival

Adult

At the data cut-off date, 230 out of 373 patients (61.7%) in the KEYNOTE-158 trial (varied tumours) had died. The median OS was 19.8 months (95% CI, 14.5 to 25.8 months). The OS event-free probabilities at 6, 12, 24, 36, 48, and 60 months were 72.1%, 58.6%, 46.5%, 39.2%, 37.1%, 34.8%, and not reached, respectively. The OS for individual tumour types was not reported.

In KEYNOTE-164 (all patients had CRC), 63 of 124 patients (50.8%) had died, with a median OS of 36.1 months (95% CI, 24.0 months to not estimable). The OS event-free probabilities at 12, 24, 36, and 48 months were 74.2%, 59.1%, 50.5%, 44.3%, respectively. In cohort A, 38 of 61 patients died (62.3%), with a median OS of 31.4 months (95% CI, 21.4 to 58.0 months). In cohort B, 31 of 63 patients died (49.2%), with a median OS of 47.0 months (95% CI, 19.2 months to not reached).

Pediatric

In the KEYNOTE-051 trial, 5 of 7 patients (71.4%) had died, with a median OS of 7.7 months (95% CI, 1.9 months to not estimable). The OS event-free probabilities at 6, 12, and 18 months were 50.0%, 33.3%, 33.3%, respectively. OS for individual tumour types was not reported.

Progression-Free Survival

Adult

At the cut-off date, 275 of 373 patients (73.7%) in the KEYNOTE-158 trial (varied tumours) had experienced a PFS event, with a median PFS of 4.0 months (95% CI, 2.4 to 4.3 months). The PFS event-free probabilities at 6, 12, 24, 36, 48, 60, and 72 months were 43.2%, 35.1%, 28.8%, 25.0%, 24.0%, 22.0%, and not reached, respectively. PFS for individual tumour types was not reported.

In KEYNOTE-164 (all patients had CRC), 83 of 124 patients (66.9%) had experienced a PFS event, with a median PFS of 4.0 months (95% CI, 2.1 to 7.4 months). The PFS event-free probabilities at 6, 12, 24, and 36 months were 45.8%, 37.5%, 33.8%, and 31.5%, respectively. In cohort A, 44 of 61 patients (72.1%) had a PFS event, with a median PFS of 2.3 months (95% CI, 2.1 to 8.1 months). In cohort B, 40 of 63 patients (63.5%) had a PFS event, with a median PFS of 4.1 months (95% CI, 2.1 to 18.9 months).

Pediatric

In the KEYNOTE-051 trial, 6 of 7 patients (85.7%) experienced a PFS event, with a median PFS of 1.7 months (95% CI, 0.4 months to not reached). The PFS event-free probability at 6 months was 16.7% and for 12 months it was not reached. PFS for individual tumour types was not reported.

Health-Related Quality of Life

Adult

In the KEYNOTE-158 trial (varied tumours) at baseline, adherence rates were 93.4% for the EORTC QLQ-C30. By week 9, adherence rates were 90%. After baseline, adherence rates ranged from 71.5% to 90.0% for the EORTC QLQ-C30 through week 111. Completion rates declined over time due to participant discontinuation as a result of disease progression, AEs, death, or physician decision.

The mean baseline EORTC QLQ-C30 GHS and QoL score was 64.40 points (SD = 20.12 points) across 364 patients. By week 9, the mean score was 67.48 points (SD = 22.48 points) in 265 patients, with an LS mean change from baseline of 3.08 points (95% CI, 0.32 to 5.84 points). A total of 57 patients (21.5%) experienced a deterioration in HRQoL at week 9, according to the sponsor’s 10-point threshold for any subscale. Based on the observed cases (n = 265), 32.8% of patients reported an improvement (≥ 10-point increase), 45.7% reported stability (< 10-point change), and 21.5% reported deterioration (≥ 10-point decrease) in GHS and QoL at week 9.

HRQoL was not assessed in the KEYNOTE-164 trial.

Pediatric

HRQoL was not assessed in the KEYNOTE-051 trial.

Table 14: Key HRQoL Outcomes (KEYNOTE-158 Trial, APaT Population)

APaT = all participants as treated; CI = confidence interval; EORTC QLQ-C30 = European Organisation for Research and Treatment of Cancer Quality of Life Questionnaire Core 30; GHS = global health status; HRQoL = health-related quality of life; LS = least squares; QoL = quality of life; SD = standard deviation.

Note: Data cut-off date of January 12, 2022.

Source: KEYNOTE-158 Clinical Study Report.¹⁷

Harms

The summary of the harms outcomes for the included studies is presented in Table 15.

Adverse Events

Adult

At the January 12, 2022, cut-off date, 358 of 373 patients (96%) had experienced at least 1 AE in the KEYNOTE-158 trial. The most common AEs included diarrhea (25%), fatigue (24%), nausea (21%), vomiting (19%), and abdominal pain (14%). Of 373 patients, 197 (53%) had experienced at least 1 grade 3 to 5 AE. The grade 3 to 5 AEs that were reported in at least 5% of patients included anemia (6%) and abdominal pain (1%).

At the September 9, 2019, cut-off date, all 124 patients (100%) had experienced at least 1 AE. The most common AEs included fatigue (34%), diarrhea (32%), nausea (31%), abdominal pain (27%), and vomiting (26%). Of 124 patients, 68 (55%) had experienced at least 1 grade 3 to 5 AE. The grade 3 to 5 AEs that were reported in at least 5% of patients in the KEYNOTE-164 trial were anemia (7%) and abdominal pain (6%).

Pediatric

At the January 18, 2022, cut-off date, 7 of 7 patients (100%) had experienced at least 1 AE in the KEYNOTE-051 trial. The most common AEs included headache (71.4%), anemia (71.4%), vomiting (57.1%), pyrexia (42.9%), and rash (42.9%). A total of 5  of 7 patients (71.4%) had experienced at least 1 grade 3 to 5 AE. The grade 3 to 5 AEs that were reported in at least 5% of patients in the KEYNOTE-051 trial included anemia (14.3%) and abdominal pain (28.6%).

Serious Adverse Events

Adult

At the January 12, 2022, cut-off date, 133 of 373 patients (36%) had reported at least 1 SAE. The SAEs that were reported in at least 2% of patients in the KEYNOTE-158 trial included sepsis (2%) and pneumonia (2%).

At the September 9, 2019, cut-off date, 56 of 124 patients (45%) had reported at least 1 SAE. The SAEs that were reported in at least 2% of patients in the KEYNOTE-164 trial included dyspnea (4%), sepsis (4%), abdominal pain (3%), a small intestine obstruction (3%), urinary tract infection (3%), and ileus (2%).

Pediatric

At the January 18, 2022, cut-off date, 6 of 7 patients (85.7%) had reported at least 1 SAE. The SAE that was reported in at least 10% of patients in the KEYNOTE-051 trial was vomiting (28.6%).

Withdrawals Due to Adverse Events

Adult

At the January 12, 2022, cut-off date, 49 patients (13%) had discontinued pembrolizumab due to AEs in the KEYNOTE-158 trial.

At the September 9, 2019, cut-off date, 10 patients (8%) had discontinued treatment due to AEs in the KEYNOTE-164 trial.

Pediatric

At the January 18, 2022, cut-off date, 2 patients (28.6%) in the KEYNOTE-051 trial had discontinued pembrolizumab due to AEs.

Mortality

Adult

At the January 12, 2022, cut-off date, 220 patients had died in the KEYNOTE-158 trial; 20 of these patients died due to AEs.

At the September 9, 2019, cut-off date, 32 out of 61 patients (52.5%) had died in cohort A in the KEYNOTE-164 trial. In cohort B, 25 out of 63 patients (39.7%) had died. A total of 2 patients died in each cohort due to AEs.

Pediatric

At the January 18, 2022, cut-off date, 5 patients had died in the KEYNOTE-051 trial; none of these deaths were due to AEs.

Notable Harms

Notable harms of interest to this review were immune-mediated events and infusion-related reactions. The incidence of notable harms in the included trials is summarized in Table 15.

Adult

At the January 12, 2022, cut-off date, 83 out of 373 patients (22%) in the KEYNOTE-158 trial had experienced at least 1 AEOSI (immune-mediated event or infusion-related reaction associated with pembrolizumab). The most common AEOSIs were hypothyroidism (10%), hyperthyroidism (4%), pneumonitis (3%), and colitis (3%).

At the September 9, 2019, cut-off date, 37 out of 124 patients (30%) in the KEYNOTE-164 trial had experienced at least 1 AEOSI. The most common AEOSIs (immune-mediated events and infusion-related reactions associated with pembrolizumab) were hypothyroidism (15%), hyperthyroidism (8%), pneumonitis (2%), and colitis (2%).

Pediatric

At the January 18, 2022, cut-off date, 2 patients (28.6%) in the KEYNOTE-051 trial had experienced at least 1 AEOSI (immune-mediated event or infusion-related reaction associated with pembrolizumab). The AEOSIs reported were hypothyroidism (14.3%) and drug hypersensitivity (14.3%).

Table 15: Key Harms Data (APaT Population) for Sponsor Studies

APaT = all participants as treated; AE = adverse event; AEOSI = adverse event of special interest (immune-mediated event or infusion-related reaction associated with pembrolizumab); SAE = serious adverse event.

Note: The data cut-off dates were January 12, 2022, for the KEYNOTE-158 trial and February 19, 2021, for the KEYNOTE-164 trial.

Sources: KEYNOTE-158,¹⁷ KEYNOTE-164,¹⁸ and KEYNOTE-051¹⁹ Clinical Study Reports.

Critical Appraisal

Internal Validity

KEYNOTE-158, KEYNOTE-164, and KEYNOTE-051 are all single-arm trials, which raised several important considerations due to the lack of any comparison group. A single-arm study design is usually used to provide preliminary evidence of the efficacy of a treatment and to collect safety data and not intended to be confirmatory for efficacy.⁶⁶ Formal statistical hypothesis tests were not presented. According to the FDA, ORR is a direct measure of antitumour activity that may be attributed to the study drug and is evaluable with a single-arm design.⁶⁷ However, single-arm trials are inherently limited in their ability to unambiguously support causal conclusions versus any comparator because it is not possible to distinguish between treatment effects, placebo effects, and natural history. Although the inclusion and exclusion criteria for the trials were stated, the selection procedures were not described; therefore, the potential for selection bias cannot be excluded.

An open-label design can introduce a risk of performance bias (i.e., the care or concomitant treatments received may differ based on knowledge of the intervention), but there was no evidence that this occurred. Similarly, the open-label design could introduce a risk of bias in the measurement of the outcomes, particularly HRQoL and subjective AEs. A relevant tool (EORTC QLQ-C30) with evidence of validity, reliability, and responsiveness was used in the assessment of HRQoL. However, the assessment of subjective end points can be biased because knowledge of the intervention can impact patient expectations and perceptions about the benefits and harms of treatment (i.e., there is a possibility for both to be overestimated). This risk of bias in the outcome assessment was mitigated for the response outcomes (PFS, ORR) in the KEYNOTE-158 and KEYNOTE-164 trials because these were outcomes assessed by an independent review committee. The measurement of these outcomes was done by the investigator in the KEYNOTE-051 trial, which might introduce some bias; however, this likely did not occur for the ORR because no patients responded. OS is an objective outcome that is not likely to be affected by such bias. However, it should be noted that OS results would be reflective of the influence of both pembrolizumab and subsequent treatments.

The risk of bias due to missing outcome data was considered to be low across all trials for the response and time-to-event outcomes because the rate of withdrawal from the studies was relatively low. In the assessment of ORR, patients with missing data were considered to be nonresponders, which would be a conservative assumption in a single-arm design. The assessment of HRQoL in the KEYNOTE-158 trial was at high risk of bias because, despite high adherence among available patients, close to 30% of the outcome data were missing at week 9. The analysis of the change from baseline employed a mixed model for repeated measures in which missing data are implicitly imputed under the assumption that they are missing at random. However, the reasons for the missing data suggest that this assumption is not valid, and no sensitivity analyses were provided to judge the robustness of the results.

The results of the KEYNOTE-158 and KEYNOTE-164 trials, which included adult patients with varying tumour types, show considerable heterogeneity in response across tumour types. Several tumour types were represented by a very small number of enrolled patients, which reduces the reliability of the results (i.e., the effects may be unstable and not reproducible in a larger sample). The interpretation of the potential differences in response across small samples of different cancer types is challenging because these may represent either actual differences in treatment effects or natural statistical variation. However, in its review, Health Canada acknowledged that the numbers of patients recruited for each histology reflects the natural prevalence of the MSI-H and/or dMMR biomarker, and that enrolling a larger number of patients may not have been feasible, given the broad indication.¹⁴ In the KEYNOTE-051 trial, only 7 pediatric patients were included (below the target of 25), which limits the reliability of assessing the efficacy and safety of pembrolizumab in this population. Additionally, rare or delayed AEs may not be adequately captured, leading to gaps in the harms profile, although there is extensive experience with pembrolizumab in many cancer indications. Quantification of the extent of uncertainty was challenged by a lack of CIs presented across several outcomes in this trial.

External Validity

In the KEYNOTE-158, KEYNOTE-164, and KEYNOTE-051 trials, patients who had previously received anti–PD-1, anti–PD-L1, or anti–PD-L2 drugs were excluded. The clinical experts consulted by CDA-AMC indicated that only a small proportion of the patient population in Canada would meet this eligibility criterion because most MSI-H and/or dMMR solid tumours are now treated with anti–PD-1 or anti–PD-L1 drugs that are the SOC in earlier lines of therapy. For cancers such as colorectal, endometrial, non–small cell lung cancer, kidney cancer, and urothelial carcinoma, pembrolizumab is already used in earlier treatment stages. Additionally, other immune checkpoint inhibitors are widely used in solid tumours like those found in gastric, mesothelioma, breast, small cell lung, and biliary tract cancers. Therefore, most of the available evidence in adults (KEYNOTE-158 and KEYNOTE-164 trials) is for tumours that are now treated with immune checkpoint inhibitors in earlier lines of therapy. The KEYNOTE-051 trial included only 7 pediatric patients, primarily with brain cancers, which also limits the generalizability of the findings. This small cohort may not adequately represent the broader patient population in terms of age, sex, performance status, disease stage, and other factors. As a result, it becomes difficult to confidently apply the outcomes of this trial to a broader population. Furthermore, HRQoL was not assessed in 2 trials (KEYNOTE-164 and KEYNOTE-051); thus, the generalizability of any HRQoL results to the CRC cohort and pediatric patients remains uncertain.

GRADE Summary of Findings and Certainty of the Evidence

Methods for Assessing the Certainty of the Evidence

For the pivotal studies identified in the sponsor’s systematic review, GRADE was used to assess the certainty of the evidence for outcomes considered most relevant to inform the deliberations of the CDA-AMC expert committee, and a final certainty rating was determined as outlined by the GRADE Working Group:^15,16

• 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.”

Although GRADE guidance is not available for noncomparative studies, to present these important considerations, the CDA-AMC review team assessed pivotal single-arm trials for study limitations (which refer to internal validity or risk of bias), inconsistency across studies (or populations), indirectness, and publication bias. Because the lack of a comparator arm does not allow for a conclusion to be drawn on the effect of the intervention versus any comparator, the certainty of evidence for single-arm trials started at very low certainty with no opportunity for rating up.

Although pooled evidence is available for the KEYNOTE-158 and KEYNOTE-164 trials, there are clinical characteristics that differ between these trials, such as cancer type, which may have an impact on important clinical outcomes. Therefore, GRADE was assessed separately in 3 patient groups based on these differences, as follows: adults with mixed solid tumours, adults with CRC, and children with mixed solid tumours.

Results of the GRADE Assessments

Table 2 presents the GRADE summary of findings for pembrolizumab for the treatment of unresectable or metastatic MSI-H or dMMR solid tumours assessed in adult and pediatric patients from the KEYNOTE-158, KEYNOTE-164, and KEYNOTE-051 trials.

Long-Term Extension Studies

No long-term evidence was submitted by the sponsor 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 CDA-AMC review team.

Objectives for the Summary of Indirect Evidence

No comparative evidence was available in the pivotal trials for patients with unresectable or metastatic MSI-H or dMMR solid tumours that progressed following prior treatment. This section provides a summary and appraisal of the indirect evidence submitted by the sponsor to address these knowledge gaps. These data were used to inform the pharmacoeconomic model.

Description of Indirect Comparisons

A total of 4 ITC reports were submitted, each focusing on a different solid tumour type: colorectal, endometrial, small intestine, and gastric cancers. ITCs were conducted only in tumour types for which there was a sufficient number of pembrolizumab-treated patients to conduct the analysis. All of the submitted ITCs were for the adult population only.

In the CRC ITC, pembrolizumab was compared with pooled chemotherapy, anti-VEGF plus chemotherapy, and TAS-102, representing the SOC interventions. Naive indirect comparisons were used for comparisons between pembrolizumab and pooled chemotherapy or anti-VEGF plus chemotherapy, while an unanchored MAIC was used to compare pembrolizumab with TAS-102. In the endometrial cancer ITC, an unanchored MAIC was conducted to compare pembrolizumab with TPC (doxorubicin or paclitaxel). For the small intestine cancer ITC, a naive indirect comparison was used to compare pembrolizumab with nab-paclitaxel. The gastric cancer ITC included a naive indirect comparison to compare pembrolizumab with FOLFIRI, ramucirumab plus paclitaxel, ramucirumab monotherapy, paclitaxel, and irinotecan.

Study Selection Methods

A systematic review was conducted to identify clinical trials evaluating treatments in adult patients with advanced or metastatic colorectal carcinoma who have received at least 1 prior line of therapy, irrespective of MSI-H or dMMR status. CDA-AMC did not receive any protocols outlining the systematic review and ITC for the different tumour types. The dates of the searches for the SLRs were in August 2023 (CRC ITC), May 2023 (small intestine cancer ITC), and June 2023 (gastric cancer ITC). Details of the study selection criteria can be found in Table 16. Two reviewers, working independently, reviewed all abstracts and proceedings identified by the search according to the selection criteria, with the exception of outcome criteria, which were only applied during the screening of full-text publications. The full texts of the studies identified as eligible during abstract screening were then screened by the same 2 reviewers. The studies selected at this stage were then included for data extraction. A third reviewer was included to reach consensus for any discrepancies.

For the endometrial cancer ITC, the KEYNOTE-775 trial was identified and selected based on a previous reimbursement submission reviewed by CDA-AMC.¹¹ KEYNOTE-775 was a phase III RCT that evaluated the combination of pembrolizumab and lenvatinib versus TPC chemotherapy (doxorubicin or paclitaxel) for patients with advanced endometrial carcinoma following at least 1 prior platinum-based regimen in any setting. The study included subgroup analyses of ORR, PFS, and OS by MMR status (i.e., proficient MMR versus dMMR). For consistency, the ITC presented in this section and the comparator study selection methods are similar to those methods found in the material recently reviewed by CDA-AMC for the reimbursement recommendation for pembrolizumab monotherapy for the treatment of adult patients with unresectable or metastatic MSI-H or dMMR endometrial cancer whose tumours have progressed following prior therapy and who have no satisfactory alternative treatment options.¹¹ This indication is already reimbursed by a majority of jurisdictions in Canada. The comparator in the previous submission was deemed to be still clinically relevant and no updated ITC was conducted. The ITC methods and results are summarized in Table 16.

ITC Design: CRC

Objectives

To compare the treatment efficacy of pembrolizumab as part of the KEYNOTE-164 trial versus SOC interventions in patients with MSI-H colorectal carcinomas and at least 1 prior line of therapy.

ITC Analysis Methods

A dataset with the individual patient data (IPD) from the KEYNOTE-164 trial and the pseudo-IPD from the comparator studies was used. The comparator studies were chosen based on SLRs of treatments commonly used for patients with MSI-H colorectal carcinoma and at least 1 prior line of therapy. The comparator group was pooled from multiple studies, although the pooling method was not specified. These individual comparators included pooled chemotherapy, anti-VEGF plus chemotherapy, and TAS-102, representing SOC interventions. To allow a treatment effect estimation on OS and PFS within the context of time-to-event analyses, a software tool was used to replicate the data from the published KM curves.⁷¹ Pseudo-IPD were derived for comparator interventions by using the number of participants at risk over time alongside the digitized KM curves, per the methods developed by Guyot et al.⁷⁵

The baseline characteristics of the KEYNOTE-164 participants were reweighted to align with those of participants from the selected comparator studies to assess the potential reduction in ESS. For chemotherapy and anti-VEGF comparisons, where most studies included participants with only 1 prior line of therapy, KEYNOTE-164 participants with more than 1 prior line of therapy were excluded from the analyses, reducing the sample size from 124 to 30. Further reweighting for age, sex, and ECOG PS resulted in a significant reduction in ESS and, as a result, only unadjusted ITCs could be conducted for these comparators.

Table 16: Study Selection Criteria and Methods for Indirect Comparisons

5-FU = 5-fluorouracil; AE = adverse event; ASCO = American Society of Clinical Oncology; CAPOX = capecitabine and oxaliplatin; CENTRAL = Cochrane Central Register of Controlled Trials; CNS = central nervous system; CRC = colorectal cancer; dMMR = mismatch repair deficient; DOR = duration of response; ECOG PS = Eastern Cooperative Oncology Group Performance Status; ESMO = European Society for Medical Oncology; FIGO = International Federation of Gynecology and Obstetrics; FOLFIRI = leucovorin calcium (folinic acid) + fluorouracil + irinotecan hydrochloride; FOLFOX = leucovorin calcium (folinic acid) + fluorouracil + oxaliplatin; FOLFOXIRI = folinic acid, 5-fluorouracil, oxaliplatin, and irinotecan; HRQoL = health-related quality of life; ICU = intensive care unit; ITC = indirect treatment comparison; MSI-H = microsatellite instability-high; nab = nanoparticle albumin-bound; ORR = objective response rate; OS = overall survival; PFS = progression-free survival; PICOS = population, intervention, comparator, outcomes, and study design; RCT = randomized controlled trial; SAE = serious adverse event; SLR = systematic literature review; TAS-102 = trifluridine and tipiracil hydrochloride; WDAE = withdrawal due to adverse event.

Sources: Colorectal cancer SLR report,⁷⁰ Merck Canada colorectal cancer ITC report,⁷¹ small intestine adenocarcinoma SLR report,⁷² Merck Canada small intestine cancer ITC report,⁷³ gastric cancer SLR report,⁷⁴ Merck Canada gastric cancer ITC report.⁷³

Unadjusted ITCs, without adjustment for confounders or effect modifiers, were performed using Cox proportional hazard models, applying both pseudo-IPD from comparator arms and the KEYNOTE-164 data. The model included treatment as the sole covariate. The results summarized for each time-to-event end point included HRs with 95% CIs, P values, median survival times with corresponding CIs, and the number and percentage of events per treatment arm (pembrolizumab versus pooled comparators). In cases with 0 events in 1 treatment group, the 2-sided Wald test was replaced by a 2-sided score test. Log-cumulative hazard plots and Schoenfeld residual plots were generated for diagnostic purposes. When multiple suitable comparator arms were identified, pseudo-IPD data were merged, and baseline characteristics were summarized using weighted averages.

An unanchored MAIC approach was used for the comparison of pembrolizumab to TAS-102 (Table 17). To account for differences in baseline characteristics between participants in KEYNOTE-164 and the aggregated data from comparator studies (Table 44), IPD from the KEYNOTE-164 trial were reweighted to align with the mean baseline characteristics reported in the external studies, following the methods described by Signorovitch et al. (2012)⁷⁶ and Signorovitch et al. (2010).⁷⁷ Before reweighting, any participants in the KEYNOTE-164 trial who did not meet the eligibility criteria of the comparator trials were excluded. Potential effect modifiers and prognostic factors identified as relevant matching variables, based on input from the clinical experts, included age, sex, and ECOG PS.

For all comparisons in CRC, the proportional hazards assumption did not hold for both OS and PFS outcomes. Time-varying HR analyses were not conducted due to the limited sample size of the colorectal carcinoma subpopulation in the KEYNOTE-164 trial, which could have caused parametrization issues. Diagnostic plots were provided to aid in interpreting this assumption.⁷¹

Table 17: ITC Analysis Methods — Colorectal Cancer

ECOG PS = Eastern Cooperative Oncology Group Performance Status; ITC = indirect treatment comparison; KM = Kaplan-Meier; MAIC = matching-adjusted indirect comparison; OS = overall survival; PFS = progression-free survival; TAS-102 = trifluridine and tipiracil hydrochloride; VEGF = vascular endothelial growth factor.

Source: Merck Canada colorectal cancer ITC report.⁷¹

Results of ITC: CRC

Summary of Included Studies

Baseline characteristics for selected comparator studies are provided in Table 18, Table 19, and Table 20. Table 18 and Table 19 provide a summary of the key patient characteristics in the pembrolizumab arm (KEYNOTE-164 study) and the pooled chemotherapy arm of the ITC and the anti-VEGF plus chemotherapy arm used in the naive indirect comparisons. For these 2 comparators, patients had received at least 1 prior line of systemic therapy. For the chemotherapy with or without anti-VEGF comparators, because the majority of comparator studies included only participants with 1 prior line of therapy, patients in the KEYNOTE-164 trial with more than 1 prior line of therapy were excluded from the naive indirect comparison. The results of the risk-of-bias assessment or quality of the included studies was not provided.

The characteristics of the patients in the pembrolizumab arm versus the TAS-102 arm after reweighting in the unanchored MAIC are described in Table 20; these patients had all received at least 2 prior lines of systemic treatment. For the TAS-102 comparison, there was a limitation in comparing the baseline characteristics of the comparator studies. It was assumed, for matching purposes within the MAIC framework, that all participants had received at least 2 prior lines of therapy. However, there may have been a small proportion of participants in the comparator studies who had fewer than 2 prior lines of therapy, which could impact the accuracy of the comparison.

The homogeneity assessment of included studies is provided in Table 21; few characteristics were available for comparison.

Table 18: Key Baseline Characteristics for Participants With MSI-H CRC and at Least 1 Line of Prior Therapy — Pembrolizumab Versus Pooled Chemotherapy (APaT Population)

APaT = all participants as treated; CRC = colorectal cancer; ECOG PS = Eastern Cooperative Oncology Group Performance Status; MSI-H = microsatellite instability-high; NR = not reported; Q1 = quartile 1; Q3 = quartile 3; SD = standard deviation.

^aNumber of participants: Included the APaT population of patients with MSI-H CRC and at least 1 line of prior therapy. Database cut-off date: February 19, 2021.

^bNumber of participants was based on Aparicio et al. (2022), Giantonio et al. (2007), Graeven et al. (2007), Li et al. (2018), Masi et al. (2015), Moore et al. (2016), Passardi et al. (2017), Peeters et al. (2013), Pietrantonio et al. (2020), Randolph Hecht et al. (2017), Rothenberg et al. (2008), Tabernero et al. (2015), Van Cutsem et al. (2012), Van Cutsem et al. (2011), and Yasui et al. (2015); modified APaT population.

Source: Merck Canada CRC indirect treatment comparison report.⁷¹

Table 19: Key Baseline Characteristics for Participants With MSI-H CRC and at Least 1 Line of Prior Therapy — Pembrolizumab Versus Anti-VEGF and Chemotherapy (APaT Population)

APaT = all participants as treated; CRC = colorectal cancer; ECOG PS = Eastern Cooperative Oncology Group Performance Status; MSI-H = microsatellite instability-high; NR = not reported; Q1 = quartile 1; Q3 = quartile 3; SD = standard deviation; VEGF = vascular endothelial growth factor.

^aNumber of participants: Included the APaT population of patients with MSI-H CRC and at least 1 line of prior therapy. Database cut-off date: February 19, 2021.

^bNumber of participants: Based on Bendell et al. (2013), Bendell et al. (2013), Cunningham et al. (2013), Giantonio et al. (2007), Iwamoto et al. (2015), Iwamoto et al. (2015), Li et al. (2018), Li et al. (2021), Masi et al. (2015), O’Neil et al. (2014), Passardi et al. (2017), Shi et al. (2017), Van Cutsem et al. (2012), modified APaT population.

Source: Merck Canada CRC indirect treatment comparison report.⁷¹

Table 20: Key Baseline Characteristics Before and After Weighting for Participants With MSI-H CRC and at Least 2 Lines of Prior Therapy — Pembrolizumab Versus TAS-102 (APaT Population)

APaT = all participants as treated; CRC = colorectal cancer; ECOG PS = Eastern Cooperative Oncology Group Performance Status; MSI-H = microsatellite instability-high; NR = not reported; Q1 = quartile 1; Q3 = quartile 3; SD = standard deviation; TAS-102 = trifluridine and tipiracil hydrochloride.

^aNumber of participants: Included the APaT population of patients with MSI-H CRC and at least 2 lines of prior therapy. Database cut-off date: February 19, 2021.

^bNumber of participants: Based on Mayer et al. (2015), Pfeiffer et al. (2023), Prager et al. (2023), Xu et al. (2018), and Yoshino et al. (2012); modified APaT population.

Source: Merck Canada CRC indirect treatment comparison report.⁷¹

Table 21: Assessment of Homogeneity — Colorectal Cancer

dMMR = mismatch repair deficient; ECOG PS = Eastern Cooperative Oncology Group Performance Status; MSI-H = microsatellite instability-high; OS = overall survival; PFS = progression-free survival; TAS-102 = trifluridine and tipiracil hydrochloride; VEGF = vascular endothelial growth factor.

Source: Merck Canada colorectal cancer indirect treatment comparison report.⁷¹

Results

In the naive indirect comparisons, pembrolizumab was favoured for OS versus chemotherapy alone or in combination with an anti-VEGF drug. The unadjusted OS analysis estimated an HR of 0.30 (95% CI, 0.24 to 0.39) when compared with chemotherapy alone, and an HR of 0.37 (95% CI, 0.29 to 0.48) when compared with chemotherapy in combination with an anti-VEGF drug. The unadjusted PFS analysis estimated an HR of 0.43 (95% CI, 0.34 to 0.54) when compared with chemotherapy alone, and an HR of 0.53 (95% CI, 0.42 to 0.67) when compared with chemotherapy in combination with an anti-VEGF drug, favouring pembrolizumab.

In the unanchored MAIC comparing pembrolizumab with TAS-102, pembrolizumab was favoured for OS and PFS. The unanchored OS and PFS analysis after matching estimated HRs of 0.21 (95% CI, 0.15 to 0.30) and 0.32 (95% CI, 0.23 to 0.45), respectively (Table 22, Table 23, Table 24, Table 25).

Table 22: Unadjusted Indirect OS Comparison of Pembrolizumab Versus Selected Comparators for Participants With MSI-H CRC and at Least 1 Prior Therapy (APaT Population)

APaT = all participants as treated; CI = confidence interval; CRC = colorectal cancer; HR = hazard ratio; MSI-H = microsatellite instability-high; NA = not applicable; OS = overall survival; VEGF = vascular endothelial growth factor.

^aFrom product-limit (Kaplan-Meier) method.

^bBased on Cox regression model with treatment as a covariate.

^cTwo-sided P value using Wald test (score test in case of 0 events in 1 treatment group).

Source: Merck Canada CRC indirect treatment comparison report.⁷¹

Table 23: Unanchored OS MAIC of Pembrolizumab Versus TAS-102 for Participants With MSI-H CRC and at Least 2 Prior Therapies (APaT Population)

APaT = all participants as treated; CI = confidence interval; CRC = colorectal cancer; HR = hazard ratio; MAIC = matching-adjusted indirect comparison; MSI-H = microsatellite instability-high; NA = not applicable; OS = overall survival; TAS-102 = trifluridine and tipiracil hydrochloride.

Note: Matching covariates included age (median), sex, and Eastern Cooperative Oncology Group Performance Status.

^aDatabase cut-off date: February 19, 2021.

^bNumber of participants: APaT population with MSI-H CRC and at least 2 lines of prior therapy.

^cNumber of participants: APaT population based on Mayer et al. (2015), Pfeiffer et al. (2023), Prager et al. (2023), Xu et al. (2018), Yoshino et al. (2012).

^dFrom product-limit (Kaplan-Meier) method.

^eBased on Cox regression model with treatment as a covariate.

^fTwo-sided P value using Wald test (score test in case of 0 events in 1 treatment group).

Source: Merck Canada CRC indirect treatment comparison report.⁷¹

Table 24: Unadjusted Indirect PFS Comparison of Pembrolizumab Versus Selected Comparators for Participants With MSI-H CRC and at Least 1 Prior Therapy (APaT Population)

APaT = all participants as treated; CI = confidence interval; CRC = colorectal cancer; HR = hazard ratio; MSI-H = microsatellite instability-high; NA = not applicable; PFS = progression-free survival; VEGF = vascular endothelial growth factor.

^aFrom product-limit (Kaplan-Meier) method.

^bBased on Cox regression model with treatment as a covariate.

^cTwo-sided P value using Wald test (score test in case of 0 events in 1 treatment group).

Source: Merck Canada CRC indirect treatment comparison report.⁷¹

Table 25: Unanchored PFS MAIC of Pembrolizumab Versus TAS-102 for Participants With MSI-H CRC and at Least 2 Prior Therapies (APaT Population)

APaT = all participants as treated; CI = confidence interval; CRC = colorectal cancer; HR = hazard ratio; MAIC = matching-adjusted indirect comparison; MSI-H = microsatellite instability-high; NA = not applicable; PFS = progression-free survival; TAS-102 = trifluridine and tipiracil hydrochloride.

Note: Matching covariates included age (median), sex, and Eastern Cooperative Oncology Group Performance Status.

^aDatabase cut-off date: February 19, 2021.

^bNumber of participants: APaT population of patients with MSI-H CRC and at least 2 lines of prior therapy.

^cNumber of participants: Based on Mayer et al. (2015), Pfeiffer et al. (2023), Prager et al. (2023), Xu et al. (2018), Yoshino et al. (2012), modified APaT population.

^dFrom product-limit (Kaplan-Meier) method.

^eBased on Cox regression model with treatment as a covariate.

^fTwo-sided P value using Wald test (score test in case of 0 events in 1 treatment group).

Source: Merck Canada CRC indirect treatment comparison report.⁷¹

ITC Design: Endometrial Cancer

Objectives

The objective was to compare pembrolizumab in participants with MSI-H endometrial carcinoma that were part of the KEYNOTE-158 study (particularly in cohort K and some in cohort D; data cut-off date of January 12, 2022) against the comparator arm (doxorubicin or paclitaxel, denoted as TPC) in the dMMR subpopulation of the KEYNOTE-775 trial using information from Makker et al. (2021).⁷⁸

ITC Analysis Methods

The endometrial cancer ITC used unanchored MAICs to assess the relative efficacy of pembrolizumab monotherapy in participants with previously treated MSI-H or dMMR advanced or metastatic endometrial carcinoma, compared with SOC (Table 26). The following baseline characteristics, identified as potential effect modifiers or key prognostic factors based on clinical expertise, were selected as matching variables:

• age (median)

• race (Asian, Black, white, other)

• ECOG PS (0 versus 1)

• number of prior lines of therapy (1 versus ≥ 2)

• histology status (endometrioid carcinoma, others).

For time-to-event end points (such as OS and PFS), an unstratified Cox proportional hazards model was applied, with treatment as the only covariate and using the Efron method for handling ties. This model used a universal weight of 1 for the comparator arm (pseudo-IPD for TPC from the KEYNOTE-775 trial) and weights derived from the MAIC matching step for the pembrolizumab arm (IPD from the KEYNOTE-158 trial). HRs for pembrolizumab versus TPC and the corresponding 95% CIs were estimated both before and after weighting, using the sandwich (empirical) covariance matrix. A 2-sided Wald test was applied (all treatment groups had ≥ 1 event).

For the end point (i.e., ORR), a weighted response ratio with no stratification factor was used to estimate the treatment difference between pembrolizumab and TPC, with a universal weight 1 for the comparator arm (pseudo-IPD for TPC from the KEYNOTE-775 trial) and the weights from the MAIC matching step for the pembrolizumab arm (IPD from the KEYNOTE-158 trial).

Table 26: ITC Analysis Methods — Endometrial Cancer

ECOG PS = Eastern Cooperative Oncology Group Performance Status; ITC = indirect treatment comparison; KM = Kaplan-Meier; MAIC = matching-adjusted indirect comparison; ORR = objective response rate; OS = overall survival; PFS = progression-free survival.

Source: Merck Canada endometrial cancer ITC report.⁸⁰

Results of ITC: Endometrial Cancer

Table 27 provides a summary of the key patient characteristics in the pembrolizumab arm (KEYNOTE-158 study) and the TPC arm of the ITC. The baseline homogeneity assessment of included studies is provided in Table 28.

Summary of Included Studies

Table 27: MAIC of Pembrolizumab Versus Selected Comparators for Participants With MSI-H Endometrial Carcinoma and at Least 1 Line of Prior Therapy

APaT = all participants as treated; ECOG PS = Eastern Cooperative Oncology Group Performance Status; ESS = effective sample size; MAIC = matching-adjusted indirect comparison; MSI-H = microsatellite instability-high; TPC = treatment of physician’s choice (doxorubicin or paclitaxel).

^aNumber of participants and comparators based on Makker et al. (2021).

^bDatabase cut-off date: January 12, 2022.

^cNumber of participants: Included the APaT population of patients in the KEYNOTE-158 trial with MSI-H endometrial carcinoma (cohorts D and K) and at least 1 line of prior therapy.

^dESS computed as the square of the summed weights divided by the sum of the squared weights. Weighted according to matched baseline characteristics of selected comparators.

Source: Merck Canada endometrial cancer indirect treatment comparison report.⁸⁰

Table 28: Assessment of Homogeneity — Endometrial Cancer

ECOG PS = Eastern Cooperative Oncology Group Performance Status; ITC = indirect treatment comparison; KM = Kaplan-Meier; ORR = objective response rate; OS = overall survival; PFS = progression-free survival; TPC = treatment of physician’s choice (doxorubicin or paclitaxel).

Source: Merck Canada endometrial cancer ITC report.⁸⁰

Table 29: Unanchored MAIC Analysis of OS (Pembrolizumab Versus TPC) for Patients With MSI-H Endometrial Carcinoma and at Least 1 Line of Prior Therapy

CI = confidence interval; ECOG PS = Eastern Cooperative Oncology Group Performance Status; HR = hazard ratio; MAIC = matching-adjusted indirect comparison; MSI-H = microsatellite instability-high; NA = not applicable; OS = overall survival; TPC = treatment of physician’s choice (doxorubicin or paclitaxel).

^aDatabase cut-off date: January 12, 2022.

^bMatching was done on the following covariates: age (median), ECOG PS, race, prior lines of therapy, and histology status.

^cNumber of participants: For pembrolizumab, this included all participants with MSI-H endometrial carcinoma (cohorts D and K) and at least 1 line of prior therapy in the as-treated population of the KEYNOTE-158 trial. For TPC, the data were based on Makker et al. (2021).

^dSample size after matching computed as the sum of the weights.

^eFrom product-limit (Kaplan-Meier) method for censored data.

^fBased on Cox regression model with treatment as a covariate.

^gTwo-sided P value using Wald test (score test in case of 0 events in 1 treatment group).

Source: Merck Canada endometrial cancer indirect treatment comparison report.⁸⁰

Results

OS results for the unanchored MAIC between pembrolizumab from the KEYNOTE-158 trial and the TPC data from the KEYNOTE-775 trial are presented in Table 29. After weighting, the HR was 0.24 (95% CI, 0.13 to 0.45) favouring pembrolizumab compared with TPC. The sensitivity analysis for OS yielded results that aligned with the main analysis in both the direction and magnitude of effect.

Table 30: Unanchored MAIC Analysis of PFS (Pembrolizumab Versus TPC) for Patients With MSI-H Endometrial Carcinoma and at Least 1 Line of Prior Therapy

CI = confidence interval; ECOG PS = Eastern Cooperative Oncology Group Performance Status; HR = hazard ratio; MAIC = matching-adjusted indirect comparison; MSI-H = microsatellite instability-high; NA = not applicable; PFS = progression-free survival; TPC = treatment of physician’s choice (doxorubicin or paclitaxel).

^aDatabase cut-off date: January 12, 2022.

^bMatching was done on the following covariates: age (median), ECOG PS, race, prior lines of therapy, and histology status.

^cNumber of participants: For pembrolizumab, this included all participants in the as-treated population in the KEYNOTE-158 trial with MSI-H endometrial carcinoma (cohorts D and K) and at least 1 line of prior therapy. For TPC, the number of participants was based on Makker et al. (2021).

^dThe sample size after matching was computed as the sum of the weights.

^eFrom product-limit (Kaplan-Meier) method for censored data.

^fBased on Cox regression model with treatment as a covariate.

^gTwo-sided P value using Wald test (score test in case of 0 events in 1 treatment group).

Source: Merck Canada endometrial cancer indirect treatment comparison report.⁸⁰

Table 31: Unanchored MAIC Analysis of ORR (Pembrolizumab Versus TPC) for Patients With MSI-H Endometrial Carcinoma and at Least 1 Line of Prior Therapy

CI = confidence interval; ECOG PS = Eastern Cooperative Oncology Group Performance Status; ESS = effective sample size; MAIC = matching-adjusted indirect comparison; MSI-H = microsatellite instability-high; NA = not applicable; ORR = objective response rate; TPC = treatment of physician’s choice (doxorubicin or paclitaxel).

^aNumber of participants: This included all participants in the as-treated population of the KEYNOTE-158 trial with MSI-H endometrial carcinoma (cohorts D and K) and at least 1 line of prior therapy. Database cut-off date: January 12, 2022.

^bNumber of participants: Based on Makker et al. (2021).

^cMatching was done on the following covariates: age (median), ECOG PS, race, prior lines of therapy, and prior adjuvant or neoadjuvant therapy.

^dESS after matching computed as the sum of the weights.

^eBased on a robust sandwich estimator using PROC GENMOD in the safety analysis set.

^fTwo-sided P value based on Wald test.

Source: Merck Canada endometrial cancer indirect treatment comparison report.⁸⁰

For PFS, the results are detailed in Table 30. Following weighting, the HR was 0.31 (95% CI, 0.19 to 0.53), favouring pembrolizumab compared with TPC. The sensitivity analysis for PFS yielded results that aligned with the main analysis in both the direction and magnitude of effect.

The ORR results are outlined in Table 31. After weighting, the response ratio was 4.27 (95% CI, 2.11 to 8.64), favouring pembrolizumab compared with TPC. The sensitivity analysis for ORR yielded results that aligned with the main analysis in both the direction and magnitude of effect.

ITC Design: Small Intestine Cancer

Objectives

To compare pembrolizumab in participants with MSI-H small intestine cancer who were part of the KEYNOTE-158 trial versus selected interventions.

ITC Analysis Methods

A naive indirect comparison was performed based on Cox proportional hazard models using pseudo-IPD for the nab-paclitaxel arm and IPD from the KEYNOTE-158 trial for the pembrolizumab arm (Table 32). The model included treatment as a single covariate. The following results were summarized for each time-to-event end point analyzed: HRs with corresponding 95% CIs and P values, median survival time with corresponding 95% CIs, and number and percentage of events by treatment arm (pembrolizumab versus nab-paclitaxel). Log-cumulative hazard plots as well as Schoenfeld residual plots were produced for diagnostic purposes. KM curves for OS and PFS from Overman et al.⁸¹ were used to derive pseudo-IPD for the nab-paclitaxel arm of the ITC. Digital software was used to replicate the data from the published KM curves. Pseudo-IPD were derived using the number of participants at risk over time alongside the digitized KM curves, per the methods developed by Guyot et al.⁷⁵ Due to the reduction in the ESS after matching, the baseline characteristics of participants with MSI-H small intestine cancer in the KEYNOTE-158 trial versus the aggregated baseline characteristics of the selected comparator study, the impact on ESS was deemed too substantial to perform an MAIC. As a result, only naive unadjusted ITCs were conducted.

Table 32: ITC Analysis Methods — Small Intestine Cancer

IPD = individual patient data; ITC = indirect treatment comparison; OS = overall survival; PFS = progression-free survival.

Source: Merck Canada small intestine cancer ITC report.⁸²

Results of ITC: Small Intestine Cancer

Table 33 provides a summary of the key patient characteristics in the pembrolizumab arm (KEYNOTE-158 study) and the nab-paclitaxel arm of the ITC. In these 2 treatment arms, all patients had received at least 1 prior line of systemic therapy; the median number of lines of prior therapy was 1 in the pembrolizumab arm versus 2 in the nab-paclitaxel arm. The homogeneity assessment of included studies is provided in Table 34.

Table 33: Key Baseline Characteristics for Participants With MSI-H Small Intestine Carcinoma and at Least 1 Line of Prior Therapy

ECOG PS = Eastern Cooperative Oncology Group Performance Status; MSI-H = microsatellite instability-high; nab = nanoparticle albumin-bound; NR = not reported; Q1 = first quartile; Q3 = third quartile; SD = standard deviation.

^aNumber of participants: For the pembrolizumab efficacy analysis, this included all participants with MSI-H small intestine carcinoma and at least 1 line of prior therapy in the as-treated population in the KEYNOTE-158 trial. Database cut-off date: January 12, 2022.

^bNumber of participants: mAPaT population based on Overman (2018).

Source: Merck Canada small intestine cancer indirect treatment comparison report.⁸²

Results

The naive indirect comparisons resulted in an HR of 0.18 (95% CI, 0.07 to 0.45) for OS (Table 35) and an HR of 0.22 (95% CI, 0.09 to 0.52) for PFS (Table 36), both favouring pembrolizumab.

Table 34: Assessment of Homogeneity — Small Intestine Cancer

dMMR = mismatch repair deficient; ECOG PS = Eastern Cooperative Oncology Group Performance Status; KM = Kaplan-Meier; MSI-H = microsatellite instability-high; nab = nanoparticle albumin-bound; OS = overall survival; PFS = progression-free survival.

Source: Merck Canada small intestine cancer indirect treatment comparison report.⁸²

Table 35: Unadjusted OS Analysis — Unadjusted ITC of Pembrolizumab Versus Nab-Paclitaxel for Patients With MSI-H Small Intestine Carcinoma and at Least 1 Line of Prior Therapy

APaT = all participants as treated; CI = confidence interval; HR = hazard ratio; ITC = indirect treatment comparison; KM = Kaplan-Meier; MSI-H = microsatellite instability-high; nab = nanoparticle albumin-bound; NA = not applicable; OS = overall survival; TPC = treatment of physician’s choice (doxorubicin or paclitaxel).

^aNumber of participants: For the pembrolizumab efficacy analysis, this included all participants with MSI-H small intestine carcinoma and at least 1 line of prior therapy in the as-treated population in the KEYNOTE-158 trial; database cut-off date: January 12, 2022. For TPC, the data were based on the Overman (2018) modified APaT population.

^bFrom product-limit (Kaplan-Meier) method.

^cBased on Cox regression model with treatment as a covariate.

^dTwo-sided P value using Wald test (score test in case of 0 events in 1 treatment group).

Source: Merck Canada small intestine cancer ITC report.⁸²

Table 36: Unadjusted PFS Analysis — Unadjusted ITC of Pembrolizumab Versus Nab-Paclitaxel for Patients With MSI-H Small Intestine Carcinoma and at Least 1 Line of Prior Therapy

APaT = all participants as treated; CI = confidence interval; HR = hazard ratio; ITC = indirect treatment comparison; KM = Kaplan-Meier; MSI-H = microsatellite instability-high; NA = not applicable; nab = nanoparticle albumin-bound; PFS = progression-free survival; TPC = treatment of physician’s choice (doxorubicin or paclitaxel).

^aNumber of participants: For the pembrolizumab efficacy analysis, this included all patients with MSI-H small intestine carcinoma and at least 1 line of prior therapy in the as-treated population of the KEYNOTE-158 trial; database cut-off date: January 12, 2022. For TPC, the data were based on the Overman (2018) modified APaT population.

^bFrom product-limit (Kaplan-Meier) method.

^cBased on Cox regression model with treatment as a covariate.

^dTwo-sided P value using Wald test (score test in case of 0 events in 1 treatment group).

Source: Merck Canada small Iintestine cancer ITC report.⁸²

ITC Design: Gastric Cancer

Objectives

To compare the efficacy in patients with MSI-H gastric carcinomas in the KEYNOTE-158 trial who received pembrolizumab versus selected interventions.

ITC Analysis Methods

Naive indirect comparisons were performed based on Cox proportional hazard models using pseudo-IPD for the comparator arms and IPD from the KEYNOTE-158 trial for the pembrolizumab arm (Table 37). In some cases, the comparator arm included pooled data across multiple studies, but it was not clear how this pooling was undertaken. The model included treatment as a single covariate. The following results were summarized for each time-to-event end point analyzed: HRs with corresponding 95% CIs and P values, median survival time with the corresponding 95% CI, and number and percentage of events by treatment arm (pembrolizumab versus selected comparators). If there were no events in 1 of the treatment groups, the 2-sided Wald test was replaced with a 2-sided score test. Log-cumulative hazard plots as well as Schoenfeld residual plots were produced for diagnostic purposes. A software tool was used to replicate the data from the published KM curves. Pseudo-IPD were derived using the number of participants at risk over time alongside the digitized KM curves, per the methods developed by Guyot et al.⁷⁵ Due to the reduction in the ESS after matching the baseline characteristics of participants with MSI-H gastric cancer in the KEYNOTE-158 trial to the aggregated baseline characteristics of the selected comparator studies, the impact on ESS was deemed too substantial to perform an MAIC. As a result, only unadjusted ITCs were conducted.

Table 37: ITC Analysis Methods — Gastric Cancer

ITC = indirect treatment comparison; KM = Kaplan-Meier; OS = overall survival; PFS = progression-free survival.

Source: Merck Canada gastric cancer ITC report.⁷³

Results of ITC: Gastric Cancer

Summary of Included Studies

The selected comparator studies are listed in Table 45 of Appendix 1. The baseline characteristics of the patients in each treatment group of interest in each study is reported in Table 38. The comparators trials did not assess MSI-H or dMMR status. The comparator interventions include FOLFIRI, ramucirumab plus paclitaxel, ramucirumab, paclitaxel, and irinotecan. The homogeneity assessment of the included studies is provided in Table 39.

Table 38: Baseline Characteristics of the ITC Treatment Arms

ECOG PS = Eastern Cooperative Oncology Group Performance Status; FOLFIRI = leucovorin calcium (folinic acid) + fluorouracil + irinotecan hydrochloride; ITC = indirect treatment comparison; mAPaT = modified all participants as treated; MSI-H = microsatellite instability-high; NR = not reported; Q1 = quartile 1; Q3 = quartile 3; SD = standard deviation.

^aNumber of participants: The efficacy analysis included all participants with MSI-H gastric carcinoma and at least 1 line of prior therapy in the KEYNOTE-158 trial in the as-treated population; database cut-off date: January 12, 2022.

^bNumber of participants: Based on comparator-specific pooled data for the mAPaT population from the sources reported in Table 45.

Source: Merck Canada gastric cancer ITC report.⁷³

Table 39: Assessment of Homogeneity — Gastric Cancer

dMMR = mismatch repair deficient; ECOG PS = Eastern Cooperative Oncology Group Performance Status; FOLFIRI = leucovorin calcium (folinic acid) + fluorouracil + irinotecan hydrochloride; KM = Kaplan-Meier; MAIC = matching-adjusted indirect comparison; MSI-H = microsatellite instability-high; OS = overall survival; PFS = progression-free survival.

Source: Merck Canada gastric cancer indirect treatment comparison report.⁷³

Results

The naive indirect comparisons demonstrated that pembrolizumab was favoured for OS (Table 40) and PFS (Table 41) compared with FOLFIRI, ramucirumab, ramucirumab and paclitaxel, and irinotecan in patients with gastric cancer. The evidence was insufficient to demonstrate a difference between pembrolizumab and paclitaxel.

Table 40: Unadjusted OS Analysis — ITC of Pembrolizumab Versus Comparators for Patients With MSI-H Gastric Carcinoma and at Least 1 Line of Prior Therapy

CI = confidence interval; FOLFIRI = leucovorin calcium (folinic acid) + fluorouracil + irinotecan hydrochloride; HR = hazard ratio; ITC = indirect treatment comparison; MSI-H = microsatellite instability-high; mAPaT = modified all participants as treated; NA = not applicable; OS = overall survival.

^aNumber of participants: For the pembrolizumab efficacy analysis, this included all participants with MSI-H gastric carcinoma and at least 1 line of prior therapy in the as-treated population in the KEYNOTE-158 trial; database cut-off date: January 12, 2022. For comparators, the analysis was based on comparator-specific pooled data for the mAPaT population from the sources reported in Table 45.

^bFrom product-limit (Kaplan-Meier) method.

^cBased on Cox regression model with treatment as covariate.

^dTwo-sided P value using Wald test (score test in case of 0 events in 1 treatment group).

Source: Merck Canada gastric cancer ITC report.⁷³

Table 41: Unadjusted PFS Analysis — ITC of Pembrolizumab Versus Comparators for Patients With MSI-H Gastric Carcinoma and at Least 1 Line of Prior Therapy

APaT = all participants as treated; CI = confidence interval; FOLFIRI = leucovorin calcium (folinic acid) + fluorouracil + irinotecan hydrochloride; ITC = indirect treatment comparison; MSI-H = microsatellite instability-high; NA = not applicable; PFS = progression-free survival.

^aNumber of participants: For the pembrolizumab efficacy analysis, this included all patients with MSI-H gastric carcinoma and at least 1 line of prior therapy in the as-treated population in the KEYNOTE-158 trial; database cut-off date: January 12, 2022. For comparators, the analysis was based on comparator-specific pooled data for the modified APaT population from the sources reported in Table 45.

^bFrom product-limit (Kaplan-Meier) method.

^cBased on Cox regression model with treatment as covariate.

^dTwo-sided P value using Wald test (score test in case of 0 events in 1 treatment group).

Source: Merck Canada gastric cancer ITC report.⁷³

Critical Appraisal of ITCs

The sponsor-submitted SLRs that were used to identify studies for inclusion in all of the submitted ITCs, except the ITC for endometrial cancer, appeared comprehensive. However, an a priori protocol and subsequent analyses were not provided, which prevented any evaluation of the risk of selective-results reporting (i.e., based on the magnitude, direction, or statistical significance of the effects). The included comparator studies were deemed to be primarily at a low or unclear risk of bias. Additionally, the SLR search for colorectal and small intestine cancer was conducted in July and August 2023; it is not clear whether new relevant studies have become available since then. In some cases, data were pooled across a number of studies to produce comparator arms for the analyses. The rationale and methods for pooling were not provided; thus, it should be considered that the pooling may have introduced bias and/or obscured heterogeneity in the results across the comparator studies. The sponsor noted that the proportional hazards assumption may have been violated in most comparisons (excluding the small intestine cancer ITC and the gastric cancer ITC with irinotecan and paclitaxel as comparators), which would undermine the validity of the estimated HRs.

All ITCs, except for the endometrial cancer ITC, included comparator studies that did not evaluate MSI-H or dMMR status. According to the clinical experts consulted by the review team, patients with dMMR or MSI-H tumours typically have a worse prognosis compared with those with cancer that is not MSI-H or dMMR, although prognosis varies based on tumour type. Without data on the dMMR or MSI-H status of patients in the comparator arms, the impact is not clear. Most of the ITCs conducted were unadjusted (naive), except for the endometrial cancer ITC and the pembrolizumab versus TAS-102 comparison for the CRC ITC. Unadjusted indirect comparisons were undertaken due to substantial reductions in ESS when attempting to match baseline characteristics across study arms. However, naive indirect comparisons are not considered a valid method to produce comparative effect estimates for decision-making because they are at high risk of bias due to confounding (which may be substantial). Such confounding may be due to both differences in patient characteristics and the methodologies across studies (e.g., different length of follow-up). As a result, it is not possible to discern whether the effects produced by these analyses are the result of the treatments or the differences in known or unknown prognostic and/or effect-modifying variables.

The other indirect comparisons used unanchored MAIC methodology, in which a small number of potential effect-modifying or prognostic variables were selected for matching by the sponsor based on clinical expert input. In the CRC MAIC between pembrolizumab and TAS-102, age, sex, and ECOG PS were identified as matching factors. For the endometrial cancer MAIC, adjusting factors included age, race, ECOG PS, number of prior lines of therapy, and histology. The clinical experts consulted by the review team considered these factors to be relevant. While the baseline characteristics included in the matching and reweighting are important, a key requirement for the validity of an unanchored MAIC is that all prognostic factors and effect modifiers have been identified and included in the weighting process. As this is a strong assumption that is likely not possible to meet, it has been proposed by the National Institute for Health and Care Excellence (NICE) that the extent of residual confounding be assessed statistically.⁸³ In the absence of such analyses for the MAICs presented, the magnitude of residual bias in the relative treatment effect estimates produced by the unanchored MAICs is unknown and could be substantial.⁸³ Additionally, the MAIC approach cannot adjust for methodological differences that may exist across the treatment arms (e.g., length of follow-up). In the endometrial cancer MAIC, there was a notable reduction in ESS following matching, which suggests that the results may be driven by a small subset of the original KEYNOTE-158 population. Some comparisons included a small sample size (e.g., the small intestine ITC), which would reduce the reliability of the effect estimates.

Several external validity considerations were identified. The colorectal, endometrial, and gastric cancer comparisons may no longer be highly relevant due to changes in the current SOC because patients would have already received immune checkpoint inhibitors in the first-line setting. This shift means that the comparisons may not accurately reflect current patient populations or outcomes. Furthermore, no harms or HRQoL outcomes were assessed in the analyses; therefore, it is not possible to inform on the comparative effects of these outcomes, despite them being important to patients. Another limitation is that the matching in an MAIC produces a population that is more similar to the comparator trials, which could differ from the population considered for the indication, potentially affecting the generalizability of the results.

Studies Addressing Gaps in the Systematic Review Evidence

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

The sponsor submitted 2 studies, Le et al. (2015) and Bellone et al. (2021), that did not meet the eligibility criteria for inclusion in the systematic review:^50,84

• The study by Le et al. (2015)⁸⁴ did not evaluate the Health Canada–approved dosage for pembrolizumab and was therefore not eligible for inclusion in this review.

• The study by Bellone et al. (2021)⁵⁰ assessed pembrolizumab in patients with dMMR and/or MSI-H endometrial cancer that progressed after standard therapy. Because there was pivotal evidence for these patients from the KEYNOTE-158 trial (summarized in the systematic review), CDA-AMC did not consider the study to provide additional insight. The clinical experts consulted for this review highlighted a gap in the evidence of pembrolizumab rechallenging in patients who previously received pembrolizumab and whose disease responded to immune checkpoint inhibitors and then progressed later. However, because patients with prior exposure to immune checkpoint inhibitors were excluded from the study, the potential gap for patients who were previously treated with these drugs and subsequently received pembrolizumab remains unaddressed; therefore, the study by Bellone et al. (2021) was not included.

A summary of the evidence gaps and studies that may address these gaps is included in Table 42.

Table 42: Summary of Gaps in the Systematic Review Evidence

IRRDC = International Replication Repair Deficiency Consortium; MSI-H = microsatellite instability-high; ORR = objective response rate; WES = whole-exome sequencing.

Source: Details included in the table are from the sponsor’s summary of clinical evidence.¹³

Description of Studies

The IRRDC data came from an observational, registry-based study of pediatric patients (including 1 patient who was aged 24 years) with confirmed or suspected DNA replication repair deficiency.²⁰ Data were collected for the treatment administered and from medical records (demographics, cancer diagnosis, date of initiation and completion of anti–PD-1 therapy, choice of anti–PD-1 therapy, and survival outcomes), ongoing radiological monitoring, response assessment, and management of adverse effects. Patients were treated with pembrolizumab between May 2015 and March 2019. Pathology, radiology, molecular, and biomarker analyses were centrally assessed. Imaging data were collected to assess objective tumour response and the radiologist was blinded to the clinically determined response. Tumour measurements were documented according to the Response Assessment in Neuro-Oncology criteria or by RECIST 1.1. The data cut-off date was March 2022.

The WES data were obtained from across the drug’s clinical development program and data were evaluated from 7 trials of pembrolizumab monotherapy.²⁰ Patients included adults who had advanced solid tumours and previously received systemic treatment. End points included ORR, DOR, and PFS based on a blinded independent central review (RECIST 1.1) and OS.

Populations

The IRRDC dataset consisted of 18 patients who had 20 tumours. The median age was 14 years (range, 4 to 24 years) and the proportion of females and males was split evenly. The type of cancer was categorized as either CNS tumours (10 patients with glioblastoma, 2 patients with astrocytoma, and 1 patient with high-grade glioma) or non-CNS tumours (7 patients with colorectal adenocarcinoma). The 2 patients with 2 types of cancer had both glioblastoma and colorectal adenocarcinoma. All patients received pembrolizumab 2 mg/kg (up to 200 mg) IV every 3 weeks. Discontinuations and follow-up duration were not reported for the IRRDC dataset.

The WES dataset consisted of 21 patients with MSI-H tumours from the KEYNOTE-028, KEYNOTE-012, KEYNOTE-059, KEYNOTE-061, KEYNOTE-055, KEYNOTE-199, and KEYNOTE-086 trials. The median age was 65 years (range, 51 to 89 years) and 57.1% were female, while 42.9% were male. Overall, 11 patients (52.4%) had an ECOG PS score of 1 and scores were not reported for other patients. Cancer types included gastric or gastroesophageal junction cancer (76.2%), prostate cancer (9.5%), cholangiocarcinoma (4.8%), head and neck squamous cell carcinoma (4.8%), and triple-negative breast cancer (4.8%). All patients had previously received at least 1 prior systemic therapy and 95.2% had metastatic disease. All patients received pembrolizumab. Of the 21 patients, 6 (28.1%) completed the study and 12 (57.1%) discontinued pembrolizumab. The main reason for treatment discontinuation was progressive disease, while the main reason for study discontinuation was death. The median follow-up was 18.2 months (range, 1.0 to 55.4 months).

Efficacy

From the IRRDC dataset (pediatric patients), 17 tumours were measurable at baseline (3 tumours were not measurable). Based on the 17 tumours, 4 patients (23.5%) experienced an objective treatment response and 9 patients (52.9%) experienced stable disease. Four tumours (23.5%) continued to progress. Furthermore, 11 of 20 tumours (55.0%) had not progressed by 6 months and 15 of 18 patients (83.3%) were alive at 12 months.

In the WES dataset (adults), the ORR was 52.4% (95% CI, 29.8% to 74.3%). Median PFS was 17.8 months (95% CI, 4.3 months to upper limit not reached) while the PFS rates were 56.7% and 45.9% at 12 and 36 months (there were no CIs), respectively. Median DOR and median OS were not achieved. The OS rates were 66.7% and 61.5% at 12 and 24 months (there were no CIs), respectively.

Harms

The collection of safety data was not a specific intent of the IRRDC study (pediatric patients); therefore, harms data are limited.²¹ According to the summary of clinical safety, no new safety signals were identified, and pancreatitis was the only harm that occurred in more than 1 patient (n = 2). The following harms were reported in 1 patient each: diarrhea, pneumonia, gastritis, dry skin, transient hypothyroidism, tolerable intermittent elevations in liver enzymes, skin rash, severe headaches, and seizure.

Safety data were not reported for the WES dataset (adults).

Critical Appraisal

Internal Validity

Although the datasets provide more efficacy data in pediatric and adult populations with MSI-H cancers, they are small, noncomparative, and do not address the lack of direct or indirect evidence for pembrolizumab in this setting. Moreover, no protocols were available for review and the methods were not well described; thus, there is a risk of selective reporting. It was noted that, for the IRRDC study, the treatment of patients was at the discretion of the clinical team and changes to dosages may not have been standardized across patients, making it challenging to interpret the results. Outcome measures were reviewed centrally by a blinded, independent committee, which can lower the risk of bias in the outcome measurement. While registries allow for access to additional data that clinical trials may not have, concerns with the use of this type of data include the lack of randomization, uncertain data quality (e.g., accuracy and consistency of collecting and maintaining the data), and missing or incomplete data that can bias results, particularly when compared with the rigour of standardized RCT follow-up.^85,86

External Validity

Tumour types in these datasets generally covered the same types as those in the pivotal trials (i.e., the datasets provide limited information additional to what is already available). Based on what has been reported for the WES dataset, more than half of the patients had an ECOG PS of 1 indicating that these patients were relatively healthy, which may not be representative of patients in poorer health who may receive pembrolizumab for MSI-H or dMMR tumours in clinical practice in Canada. The IRRDC dataset only modestly increases the amount of information available for younger patients with MSI-H cancers treated with pembrolizumab because there were few pediatric patients contributing to the pivotal trial evidence. There was limited reporting of harms and only for the IRRDC dataset; assessments of HRQoL were not reported. Despite the use of real-world registry data, which could improve the generalizability of the results, the internal and external validity issues minimize the utility and applicability of the findings to clinical practice.

Discussion

Summary of Available Evidence

The systematic review included 3 clinical trials: KEYNOTE-158, KEYNOTE-164, and KEYNOTE-051. Both KEYNOTE-158 and KEYNOTE-164 were phase II trials, while KEYNOTE-051 was a combined phase I and II trial. All were single-arm trials that evaluated pembrolizumab in different solid tumour cancers. The KEYNOTE-158 trial enrolled 373 adult patients with advanced MSI-H or dMMR non-CRC who experienced disease progression after prior treatments. The KEYNOTE-164 trial enrolled 124 adult patients with advanced MSI-H or dMMR CRC that had progressed following 1 or 2 prior lines of therapy (i.e., after treatment with fluoropyrimidine and either oxaliplatin or irinotecan, with or without an anti-VEGF or anti-EGFR-based therapy). The KEYNOTE-051 trial is an ongoing, open-label, multicentre study aimed at establishing the pediatric dosing regimen for pembrolizumab and evaluating the tolerability of pembrolizumab in 7 pediatric patients with MSI-H or dMMR cancers.

In the KEYNOTE-158 trial, the mean age was 59.2 years (SD = 13.1 years), and 36% of the patients were aged 65 years or older. The majority of the participants were female (61%). The majority of tumours were endometrial (25%), followed by gastric (14%). Other rarer cancers represented 26% of the cancer types. Regarding ECOG PS, 46% of the patients had a status of 0, and 54% had a status of 1. The majority of patients (56%) had received 2 or more prior lines of therapy. In the KEYNOTE-164 trial, the mean age of patients was 56.1 years (SD = 14.9 years), and 33% were aged 65 years or older. There were slightly more males (56%) than females (44%). In total, 41% of patients had an ECOG PS of 0, and 59% had an ECOG PS of 1. Most patients (76%) had received 2 or more prior lines of therapy. In the KEYNOTE-051 trial, the mean age of the 7 pediatric patients was 11 years (SD = 4.3 years) and ranged from 3 to 16 years. ECOG PS was not reported for this cohort, but 57% of the patients had a Karnofsky score of 100. In total 57% of patients had received only 1 prior therapy, while 29% had received 2 or more lines of therapy.

The sponsor submitted several ITCs, each focusing on comparing pembrolizumab with other relevant treatments in different solid tumour types: colorectal, endometrial, small intestine, and gastric cancers. The CRC ITCs compared pembrolizumab with pooled chemotherapy, anti-VEGF plus chemotherapy, and TAS-102. Naive indirect comparisons were used for comparisons between pembrolizumab and pooled chemotherapy or anti-VEGF plus chemotherapy, while an unanchored MAIC was used to compare pembrolizumab with TAS-102. For endometrial cancer, an unanchored MAIC was conducted to compare pembrolizumab with TPC (doxorubicin or paclitaxel). For small intestine cancer, a naive indirect comparison was used to compare pembrolizumab with nab-paclitaxel. For gastric cancer, naive indirect comparisons were used to compare pembrolizumab with multiple comparators, which included FOLFIRI, ramucirumab plus paclitaxel, ramucirumab monotherapy, paclitaxel, and irinotecan.

The sponsor submitted evidence from 4 studies to address gaps in the pivotal evidence; however, 2 were excluded for not providing sufficient additional insight to support using pembrolizumab for the treatment of unresectable or metastatic MSI-H or dMMR solid tumours. Two datasets were included, the IRRDC (N = 18) dataset and the WES pooled dataset (N = 21) that included pediatric and adult patients, respectively, with MSI-H status tumours who received pembrolizumab.

Interpretation of Results

Efficacy

The 3 pivotal sponsor trials (KEYNOTE-158, KEYNOTE-164, and KEYNOTE-051) included different patient populations. The KEYNOTE-158 trial included adult patients with multiple solid tumour types, including endometrial, gastric, small intestine, ovarian, biliary, pancreatic, brain, sarcomas, and other rare cancers; KEYNOTE-164 included adult patients with CRC; and KEYNOTE-051 included pediatric patients with solid tumours that consisted mainly of brain tumours. All sponsor-provided studies were single-arm trials; therefore, the findings need to be interpreted in consideration of the single-arm design, which typically cannot support causal conclusions about the effect of a drug versus any comparator. However, ORR is a surrogate end point that can be evaluated as a direct measure of antitumour activity in a single-arm design by considering the magnitude of the effect and the proportion of CRs.⁶⁷ It is not possible to robustly evaluate evidence for time-to-event end points (i.e., PFS, OS) in a single-arm design; therefore, the interpretation of the findings for these end points from the single-arm trials is informed primarily by expert opinion.

There are limited treatment options available for patients with solid tumours when SOC options were unsuccessful, particularly for adults with rare cancers and pediatric patients. The clinical experts consulted by CDA-AMC stated that the efficacy observed in the adult trials represents a notable improvement compared with SOC, based on their clinical experience. The experts noted that they would not expect SOC treatments to result in a CR or PR, rather, the goal at an advanced disease stage is stabilization and is typically what is observed in terms of response. However, the clinical experts indicated that only a small proportion of the patient population in Canada would meet the eligibility criteria to receive pembrolizumab for the submitted indication because most solid tumours are now treated with anti–PD-1 or anti–PD-L1 drugs in earlier lines of therapy (e.g., pembrolizumab is funded as SOC in the first line for MSI-H and dMMR CRC, and in the second line in MSI-H and dMMR endometrial cancer among other cancers for which immune checkpoint inhibitors are available as first-line or second-line therapy). This limits the number of patients in Canada who will be eligible to receive pembrolizumab in later lines of therapy and limits the generalizability of the studies because the majority of the patients in the included studies are not representative of the eligible population in Canada. Furthermore, there are no data to support pembrolizumab after the failure of immune checkpoint inhibitors because patients with prior exposure were excluded from the studies.

Because the KEYNOTE trials that were conducted in adults enrolled a small number of patients with tumours relevant to this indication, the generalizability of the trial results to the broader population in Canada is limited. It is important to note that the KEYNOTE-158 trial encompassed multiple solid tumour types, with variability in ORR and DOR observed across these different tumours. Despite this variability and the limited evidence for several tumour types, the tumour-agnostic indication was based on the biological rationale that MSI-H and dMMR alterations drive similar mechanisms of immune invasion, regardless of tumour type.¹⁴

In the pediatric study (KEYNOTE-051), no patients experienced an objective response (CR or PR). The small sample size of the KEYNOTE-051 trial, which included only 7 patients, also limits the generalizability of its findings and restricts the ability to draw robust conclusions about pembrolizumab’s efficacy and safety across the broader pediatric population with MSI-H or dMMR tumours that have progressed on prior therapy. The IRRDC study included a small sample of children and provided preliminary evidence to suggest that some MSI-H and dMMR tumours are responsive to pembrolizumab in this population. The clinical experts noted that the prevalence of MSI-H or dMMR in the pediatric population appears much lower because clinicians do not often see patients with some of the more common adult cancers that harbour MSI-H or dMMR mutations (e.g., CRC, endometrial, pancreatic). Given the limited evidence regarding MSI-H malignancies in children, the biological plausibility of pembrolizumab's efficacy in this population is mainly theoretical.¹⁴

Several ITCs were conducted to compare pembrolizumab with various comparators across different cancer types. Due to limitations in the available data, suboptimal methods were employed, including naive comparisons and unanchored MAICs that matched on few variables. Naive comparisons are generally not considered valid for deriving comparative effect estimates because they do not adjust for potential confounders. Unanchored MAICs require assumptions that are difficult or impossible to fully satisfy, such as the absence of unmeasured confounding. The analyses typically suggested that pembrolizumab may be favoured over the comparators in terms of OS, PFS, and ORR. The clinical experts believed it was plausible that pembrolizumab would outperform the comparators based on the single-arm trial results. However, these findings must be interpreted with caution due to significant methodological limitations. These concerns included the absence of a predefined protocol, violation of the proportional hazards assumption, lack of quality appraisals, a high risk of bias due to confounding, and small sample sizes for some comparisons. As a result, robust conclusions could not be drawn from the indirect comparisons.

Harms

Generally, no new safety signals were identified in the sponsor-provided KEYNOTE trials. The open-label design of the trials may introduce bias in reporting subjective harms, potentially affecting the reliability of the safety data. Additionally, there is limited safety information available for pediatric patients, which restricts the understanding of pembrolizumab's safety profile in this population. Because the sponsor did not provide ITCs for safety outcomes, the relative safety of pembrolizumab compared with other treatments remains unknown.

Conclusion

Patients and clinicians emphasized a high unmet need for new, effective treatments for patients with unresectable or metastatic MSI-H and/or dMMR solid tumours if SOC or salvage chemotherapy fails. Evidence for this review consisted of 3 pivotal trials (KEYNOTE-158, KEYNOTE-164, and KEYNOTE-051), multiple ITCs, and 2 noncomparative datasets in pediatric and adult patients with MSI-H cancer (IRRDC and WES). The evidence on the efficacy of pembrolizumab in adults from 2 pivotal trials, KEYNOTE-158 and KEYNOTE-164, was very uncertain due to the single-arm design. The clinical experts consulted by the review team indicated that the response to treatment (ORR) was clinically meaningful and durable (DOR) compared with what is typically observed with SOC treatments. However, responses were heterogeneous across specific cancer types, many of which were represented by a small number of patients. Despite an inability to draw causal conclusions regarding time-to-event end points (OS and PFS), the clinical experts stated that the results were promising based on the natural history of the disease and experience in clinical practice. The results for HRQoL were inconclusive due to the open-label design and missing data. The results from the WES dataset were supportive of the pivotal trials but were affected by similar limitations.

In the pediatric population, due to the small number of patients (N = 7) enrolled in the KEYNOTE-051 trial and the single-arm design, it is difficult to ascertain the benefit of pembrolizumab in this group of patients. No patients in the KEYNOTE-051 trial had a PR or CR. Supportive evidence from the single-arm IRRDC registry study provides preliminary evidence from a small sample of patients to suggest that some pediatric patients with MSI-H or dMMR malignancies may respond to pembrolizumab. No new safety signals were identified in any of the pembrolizumab trials.

The indirect comparative evidence from the ITCs (naive indirect comparisons and unanchored MAICs) was limited to 4 cancer types in adults: colorectal, endometrial, small intestine, and gastric. Results of the ITCs suggested improved PFS and OS compared with SOC, which aligned with the expectations of the clinical experts consulted by the review team. However, the ITCs had significant limitations that impacted the internal validity of the findings and precluded drawing definitive conclusions about the comparative efficacy of pembrolizumab versus SOC treatments. Additionally, immune checkpoint inhibitors are now used in the first line for colorectal, endometrial, and gastric cancers, reducing the relevance of these comparisons. No comparative safety data were provided; thus, the relative safety of pembrolizumab compared with other treatments is unknown. However, the clinical experts consulted for this review emphasized the breadth of experience with using pembrolizumab in other cancers, highlighting that pembrolizumab is better tolerated and has fewer and less severe side effects than conventional chemotherapy.

References

1.Marcus L, Lemery SJ, Keegan P, Pazdur R. FDA Approval Summary: Pembrolizumab for the Treatment of Microsatellite Instability-High Solid Tumors. Clin Cancer Res. 2019;25(13):3753-3758. PubMed

2.Mishima S, Naito Y, Akagi K, et al. Japanese Society of Medical Oncology/Japan Society of Clinical Oncology/Japanese Society of Pediatric Hematology/Oncology-led clinical recommendations on the diagnosis and use of immunotherapy in patients with DNA mismatch repair deficient (dMMR) tumors, third edition. Int J Clin Oncol. 2023;28(10):1237-1258.

3.Olave MC, Graham RP. Mismatch repair deficiency: The what, how and why it is important. Genes Chromosomes Cancer. 2022;61(6):314-321. PubMed

4.Overman MJ, Morse M. Tissue-agnostic cancer therapy: DNA mismatch repair deficiency, tumor mutational burden, and response to immune checkpoint blockade in solid tumors. In: Post TW, ed. UpToDate. Waltham (MA): UpToDate; 2024: https://www.uptodate.com. Accessed 2024 Jul 17.

5.Akagi K, Oki E, Taniguchi H, et al. Real-world data on microsatellite instability status in various unresectable or metastatic solid tumors. Cancer Sci. 2021;112(3):1105-1113. PubMed

6.Kang YJ, O'Haire S, Franchini F, et al. A scoping review and meta-analysis on the prevalence of pan-tumour biomarkers (dMMR, MSI, high TMB) in different solid tumours. Sci Rep. 2022;12(1):20495. PubMed

7.Lorenzi M, Amonkar M, Zhang J, Mehta S, Liaw K-L. Epidemiology of Microsatellite Instability High (MSI-H) and Deficient Mismatch Repair (dMMR) in Solid Tumors: A Structured Literature Review. J Oncol. 2020;2020:1-17.

8.Programme de gestion thérapeutique des médicaments. Inhibiteurs du point de contrôle immunitaire (IPCI) pour le traitement des tumeurs avec instabilité microsatellitaire élevée/déficience du système de réparation des erreurs d’apariements (MSI-H / dMMR). Rapport d’évaluation [sponsor supplied reference]. 2021.

9.Li K, Luo H, Huang L, Luo H, Zhu X. Microsatellite instability: a review of what the oncologist should know. Cancer Cell Int. 2020;20(1):16. PubMed

10.CADTH Reimbursement Recommendation: Pembrolizumab (Keytruda). Can J Health Technol. 2021;1(7). https://www.cadth.ca/sites/default/files/attachments/2021-07/PC0235%20Keytruda%20-%20CADTH%20Final%20Rec%20revised_1.pdf. Accessed 2023 Nov 24.

11.CADTH Reimbursement Recommendation: Pembrolizumab (Keytruda). Can J Health Technol. 2023;3(2). https://www.cadth.ca/sites/default/files/DRR/2023/PC0280%20Keytruda%20-%20Final%20CADTH%20Recommendation.pdf. Accessed 2023 Nov 24.

12.Keytruda (pembrolizumab): solution for infusion 100 mg/4 mL vial [product monograph DRAFT]. Kirkland (QC): Merck Canada Inc.; 2023 Aug 18.

13.Keytruda Clinical Evidence [internal sponsor's report]. In: Drug Reimbursement Review sponsor submission: Keytruda (pembrolizumab), solution for infusion 100 mg/4 mL vial Kirkland (QC): Merck Canada Inc.; 2024.

14.Health Canada reviewer's report: Keytruda (pembrolizumab) [internal sponsor's report]. In: Drug Reimbursement Review sponsor submission: pembrolizumab, 200 mg Q3W IV solution for infusion. Merck Canada Inc.

15.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

16.Santesso N, Glenton C, Dahm P, et al. GRADE guidelines 26: informative statements to communicate the findings of systematic reviews of interventions. J Clin Epidemiol. 2020;119:126-135. PubMed

17.Merck Sharp & Dohme. Data on file. Pembrolizumab (MK-3475) - MSI-H Cancer (KEYNOTE-158). Clinical Study Report P158V11MK3475. 2022.

18.Merck Sharp & Dohme. Data on file. Pembrolizumab (MK-3475) - KEYNOTE-164. Clinical Study Report P164V04MK3475 (end of study synoptic report). 2022.

19.Merck Sharp & Dohme. Data on file. Pembrolizumab (MK-3475) - Pediatric Advanced Solid Tumors or Lymphomas (KEYNOTE-051). Clinical Study Report P051V03MK3475. 2022.

20.Common Technical Document section 2.7.3: summary of clinical efficacy [internal sponsor's report]. In: Drug Reimbursement Review sponsor submission: Keytruda (pembrolizumab), solution for infusion 100 mg/4 mL vial. Kirkland (QC): Merck Canada Inc.; 2023.

21.Common Technical Document section 2.7.4: summary of clinical safety [internal sponsor's report]. In: Drug Reimbursement Review sponsor submission: Keytruda (pembrolizumab), solution for infusion 100 mg/4 mL vial. Kirkland (QC): Merck Canada Inc.; 2023.

22.KEYTRUDA (pembrolizumab) draft version [product monograph]. Merck Canada, Inc.; 2023 Oct 11.

23.Tian Z, Yao W. Albumin-Bound Paclitaxel: Worthy of Further Study in Sarcomas. Front Oncol. 2022;12:815900. PubMed

24.Lonsurf (trifluridine and tipiracil): tablet, 15 mg trifluridine/6.14 mg tipiracil (as tipiracil hydrochloride), 20 mg trifluridine/8.19 mg tipiracil (as tipiracil hydrochloride) [product monograph]. Oakville (ON): Taiho Pharma Canada, Inc.; 2020 Oct 29: https://pdf.hres.ca/dpd_pm/00058926.PDF. Accessed 2024 Aug 15.

25.PACLitaxel powder for injectable suspension nanoparticle, albumin-bound paclitaxel: 100 mg paclitaxel/vial, lyophilized powder [product monograph]. Toronto (ON): Apotex Inc.; 2021 Nov 16: https://pdf.hres.ca/dpd_pm/00071231.PDF. Accessed 2024 Aug 15.

26.Cyramza (ramucirumab): intravenous injection 10 mg ramucirumab/mL [product monograph]. Toronto (ON): Eli Lilly Canada Inc.; 2022 Jun 9: https://pdf.hres.ca/dpd_pm/00067328.PDF. Accessed 2024 Aug 15.

27.Leucovorin calcium injection: solution, 10 mg/mL, intravenous and intramuscular [product monograph]. Toronto (ON): Teva Canada Limited 2023 Nov 30: https://pdf.hres.ca/dpd_pm/00073616.PDF. Accessed 2024 Aug 15.

28.Irinotecan hydrochloride injection: solution, 20 mg/mL, intravenous [product monograph]. Toronto (ON): Teva Canada Limited 2023 Mar 28: https://pdf.hres.ca/dpd_pm/00070343.PDF. Accessed 2024 Aug 15.

29.Fluorouracil solution: 50 mg / mL, intravenous [product monograph]. Boucherville (QC): Sandoz Canada Inc.; 2024 May 31: https://pdf.hres.ca/dpd_pm/00075846.PDF. Accessed 2024 Aug 15.

30.Oxaliplatin injection: 5 mg/mL, sterile concentrate 50 mg / 10 mL, 100 mg / 20 mL, 200 mg / 40 mL for intravenous infusion [product monograph]. Boucherville (QC): Sandoz Canada Inc.; 2024 Jul 30: https://pdf.hres.ca/dpd_pm/00076515.PDF. Accessed 2024 Aug 15.