CADTH Reimbursement Review

Durvalumab (Imfinzi)

Sponsor: AstraZeneca Canada Inc.

Therapeutic area: Biliary tract cancer

This multi-part report includes:

Clinical Review

Pharmacoeconomic Review

Stakeholder Input

Clinical Review

Executive Summary

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

Table 1: Submitted for Review

BTC = biliary tract cancer; NOC = Notice of Compliance; q.3.w. = every 3 weeks; q.4.w. = every 4 weeks.

Introduction

Biliary tract cancer (BTC) refers to a heterogeneous group of gastrointestinal (GI) adenocarcinomas in the liver, gallbladder, and bile ducts. There are 4 subtypes of BTC: intrahepatic cholangiocarcinoma (IHCC), extrahepatic cholangiocarcinoma (EHCC), gallbladder cancer (GBC), and ampulla of Vater (AoV) cancer.^1-3 Although BTCs comprise less than 1% of all cancers, they account for 10% to 15% of primary liver cancers,^4,5 which are and 12th and 18th most common cancer in males and females in Canada, respectively, in 2021.⁶ The most common subtype of BTC is GBC.^7,8 IHCC makes up approximately 10% to 20% of BTC cases, whereas EHCC makes up in 30% to 40% of BTC cases.^5,9 The incidence of BTCs varies globally, depending on various risk factors,⁴ with an incidence of cholangiocarcinoma and GBC of 0.3 to 3.5 per 100,000 and 1.6 to 2.0 per 100,000, respectively, in Europe, the US, and Australasia^2,5 although incidence rates are said to be increasing.^1,10 Although there are few estimates for BTC in Canada, 1 study estimated the average national incidence rate of GBC and extrahepatic BTCs to be 30.92 cases per 1,000,000 individuals per year (approximately 3 per 100,000),¹¹ which increased between 1994 and 2012.¹² In Canada and the US, it has been estimated that there are approximately 400 and 5,000 new cases of cholangiocarcinoma diagnosed each year, respectively.^4,13 although these numbers are nearly 20 years old and may not be reflective of current incidence rates.

Symptoms of BTCs are often nonspecific and include nausea, emesis, anorexia, weight loss, abdominal pain, and jaundice. As such, the majority of patients present with locally advanced or metastatic BTC and up to 90% of BTC cases are inoperable at the time of diagnosis.^14,15 Symptoms often reflect the location of the cancer; patients with IHCC often present with nonspecific symptoms that include fever, weight loss, and/or abdominal pain, and patients with EHCC present with jaundice due to biliary obstruction.^16,17

For patients with locally advanced or metastatic BTC, platinum-based chemotherapy, most commonly the combination of gemcitabine (1,000 mg/m²) plus cisplatin (25 mg/m²), has remained, for more than 10 years, the preferred first-line standard of care (SOC) regimen for patients with advanced BTC.⁹ There is currently no standard second-line treatment option for patients with locally advanced or metastatic BTC who experience disease progression after first-line treatment. Palliative treatment options include 5-fluorouracil (5-FU) or capecitabine monotherapy, a 5-FU, leucovorin, and irinotecan (FOLFIRI) combination, or a 5-FU, folinic acid (leucovorin), and oxaliplatin (FOLFOX) combination. In addition, pemigatinib has Health Canada market authorization with conditions for the treatment of adults with previously treated, unresectable locally advanced or metastatic cholangiocarcinoma that has a fibroblast growth factor receptor 2 fusion or other rearrangement,¹⁸ although it recently received a do not reimburse CADTH recommendation and is not funded in Canada.¹⁹

Durvalumab (Imfinzi) is a fully humanized immunoglobulin G1 kappa monoclonal antibody that selectively blocks the interaction of programmed cell death 1 ligand 1 (PD-L1) with programmed cell death 1 protein (PD-1) and CD80. Durvalumab has received a Notice of Compliance (NOC) from Health Canada for the treatment of patients with locally advanced or metastatic BTC in combination with gemcitabine-based chemotherapy. Durvalumab is also indicated for the treatment of locally advanced, unresectable, stage III, non–small cell lung cancer (NSCLC) and for extensive-stage small cell lung cancer (ES-SCLC). In addition, durvalumab has received marketing authorization with conditions for the treatment of locally advanced or metastatic urothelial carcinoma, pending the results of clinical trials.²⁰ Durvalumab has been previously reviewed by CADTH for the treatment of patients with locally advanced, unresectable NSCLC who have received curative-intent, platinum-based chemoradiation therapy, for a maximum of 12 months, as well as for the first-line treatment of adults with ES-SCLC in combination with etoposide and either carboplatin or cisplatin.^21,22

The objective of the current report is to review the beneficial and harmful effects of durvalumab (50 mg/mL) 1,500 mg in combination with chemotherapy for the treatment of locally advanced or metastatic BTC.

Stakeholder Perspectives

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

Patient Input

Patient group input for the review of durvalumab was provided in a submission by the Canadian Cancer Survivor Network (CCSN), with participation from the Canadian Liver Foundation, Canadian Organization for Rare Disorders, Cholangiocarcinoma Foundation, Colorectal Cancer Resource and Action Network (CCRAN), Gastrointestinal Society, and Regroupement québécois des maladies orphelines. The CCSN is a national network of patients, families, survivors, friends, community partners, funders, and sponsors who have come together to promote the best SOC for patients with cancer.

The CCSN and participating patient groups conducted an online survey between July 18 and August 2, 2022, to collect quantitative data on durvalumab. Of the 58 individuals who responded to the survey, 12 had experience with durvalumab, 25 did not have experience with durvalumab, and 17 identified as caregivers. Of the 58 survey respondents, 21 were living in Canada, 35 were living in the US, 1 was living in the UK, and 1 was living in Spain. Additionally, the CCRAN conducted 7 interviews between July 7 and July 22, 2022, with 4 patients and 3 caregivers in Canada and the US who had experience with durvalumab.

In the CCSN survey, 29 respondents disclosed their disease stage, including 18 with late-stage or metastatic, 4 with middle-stage, 3 with early-stage, and 4 with unknown disease stage. Respondents to the CCSN survey reported symptoms related to BTC that had an impact on their quality of life (QoL) and day-to-day life, which included abdominal pain, loss of appetite and weight loss, nausea and vomiting, itching, dark urine, fever, jaundice, and light coloured, greasy stools. Caregivers indicated that caring for a loved 1 with BTC had an impact on their lifestyle, and reported feeling emotionally drained, experiencing challenges in managing medications and medical appointments, and being unable to plan ahead. The most commonly used treatments reported by respondents to the CCSN survey included gemcitabine plus cisplatin, immunotherapy, radiation, surgical therapy, targeted therapy, and FOLFOX. Most CCSN respondents reported tiredness, difficulty sleeping, hair loss, nausea and vomiting, muscle weakness, numbness and tingling of the arms and legs, and diarrhea as adverse effects associated with treatment. Although most respondents indicated no issues with accessing treatments, they did cite travel costs, limited availability in the community, and financial hardship due to cost as challenges.

Most survey respondents with durvalumab experience indicated that there was little to no difference in symptom management, side effects, ease of use, or disease progression between durvalumab and other therapies they had received for BTC. Respondents reported fatigue, constipation, decreases in white blood cell and platelet counts, and anemia as side effects of durvalumab. In the CCRAN interviews, 4 patients and 3 caregivers described their experience with durvalumab for BTC. Patients accessed durvalumab through compassionate access programs, clinical trials, private insurance coverage, and by paying out of pocket. Most CCRAN interviewees described few to no side effects associated with durvalumab and reported that their cancer had regressed with treatment. Further, CCRAN respondents indicated that durvalumab was easier to use, with a shorter duration of infusion, compared to other treatment options. There was a consensus among patients that durvalumab should be made available to eligible patients.

Survey respondents reported that new treatments should maintain QoL, prolong life, provide a cure, reduce side effects from current treatments, delay the onset of symptoms, and be easy to use. When asked to describe how much of an improvement would be needed from a new treatment to make it better than current treatments, the consensus was that prolonged life with similar or reduced side effects was most important, and that QoL and ease of access should be maintained as much as possible.

Clinician Input

Input From Clinical Experts Consulted by CADTH

First-line therapy for advanced BTC has remained gemcitabine and cisplatin since publication of the ABC-02 study in 2010; however, median overall survival (OS) is less than 12 months, so the experts considered prolonged survival an important unmet need for patients with locally advanced or metastatic BTC. The clinical experts highlighted that surgery is currently the only curative treatment for BTC, but because the disease is often detected at an advanced stage, it is usually inoperable at diagnosis. Aside from the current SOC of gemcitabine and cisplatin, patients with a relatively poor performance status often receive gemcitabine alone. The experts noted that patients commonly progress after first-line treatment and there is currently no standard second-line treatment available. There are no predictive biomarkers for locally advanced or metastatic BTC. Therefore, according to the clinical experts, patients most suitable for durvalumab plus gemcitabine and cisplatin are those with preserved organ function and good performance status, regardless of the presence of cancer-related symptoms. Conversely, patients least suitable for durvalumab plus gemcitabine and cisplatin are those with a contraindication to immunotherapy. The experts also felt that it would not be appropriate to recommend other treatments before initiating treatment with durvalumab.

The experts noted that, in clinical practice, patients would be assessed every 3 weeks during routine follow-up for changes in symptoms such as fatigue and pain, and clinical and functional status, and response to treatment would be assessed every 9 to 12 weeks with imaging. The experts also noted that tumour marker assessments of carbohydrate antigen 19 to 9 (CA19 to 9) are often evaluated and followed for those with adequate biliary decompression and elevated CA 19 to 9 levels. Per the experts, durvalumab would be discontinued at clinical or radiologic progression, a confirmed worsening of symptoms, or unacceptable immune-related toxicity. The clinical experts stated that durvalumab should only be prescribed by medical oncologists and administered by qualified nurses under the supervision of a medical oncologist in a systemic treatment unit.

Clinician Group Input

Two clinician groups provided input for the review of durvalumab: the Canadian Gastrointestinal Oncology Evidence Network (CGOEN), represented by 7 clinicians, and the Ontario Health Cancer Care Ontario (OH-CCO) Gastrointestinal Cancer Drug Advisory Committee (DAC), represented by 5 clinicians. The CGOEN is a virtual network of Canadian GI oncology clinicians who contribute knowledge about GI cancer and its treatments. The OH-CCO DAC provides evidence-based clinical and health system guidance on drug-related issues.

Both clinician groups noted that current treatment goals for patients with unresectable, metastatic BTC include extending patients’ lives, delaying disease progression, and maintaining QoL. The CGEON indicated that cisplatin plus gemcitabine is the only currently available treatment option for patients with unresectable BTC, although the OH-CCO DAC also indicated that carboplatin and gemcitabine may be used in the first-line setting. The CGEON clinicians noted that the majority of BTC patients do not reach second-line treatment as the disease progresses rapidly, and that second-line treatment with 5FU and oxaliplatin provides a modest survival benefit but is poorly tolerated. The clinicians highlighted the fact that none of the molecularly targeted drugs for BTC are funded in Canada. Thus, both clinician groups emphasized that the limited number of treatment options and the moderate survival benefit provided by gemcitabine and cisplatin treatment constitute a significant unmet need for more effective treatments in this setting.

Given the lack of available options, both clinician groups indicated that there was no rationale for patients to try other treatments before initiating durvalumab plus gemcitabine and cisplatin, and that the addition of durvalumab to the current SOC would not affect the sequencing of subsequent therapy. The CGEON and OH-CCO DAC inputs indicated that patients with unresectable BTC who align with the clinical trial criteria would be most suited for treatment with durvalumab. Patients least suitable for treatment were identified as those with contraindications to immunotherapy, inadequate liver or renal function, or an Eastern Cooperative Oncology Group Performance Status (ECOG PS) of 3 or more. The CGEON and OH-CCO DAC both indicated that in clinical practice, clinical condition and/or radiologic progression are used to determine response to treatment. The CGEON indicated that a clinically meaningful response to treatment would be maintenance or improvement in QoL and prolongation of survival, and that disease progression or intolerance would be factors for discontinuing treatment. The groups agreed that durvalumab could be administered in any setting where standard chemotherapy is delivered, under the supervision of a specialist. The OH-CCO DAC agreed with the weight-based dosing method with a cap, which is consistent with other disease-site regimens, and noted that flat dosing can result in overtreatment.

Drug Program Input

The drug programs identified the following jurisdictional implementation issues: relevant comparators, considerations for initiation of therapy, considerations for continuation or renewal of therapy, considerations for discontinuation of therapy, considerations for prescribing of therapy, generalizability, and system and economic issues. Refer to Table 3 for more details.

Clinical Evidence

Pivotal Studies and Protocol-Selected Studies

Description of Study

TOPAZ-1 was a double-blind, placebo-controlled, international, randomized, phase III study to evaluate the efficacy and safety of adding durvalumab to the established chemotherapy regimen of gemcitabine and cisplatin in patients with previously untreated, unresectable, locally advanced, or metastatic BTC. Patients were randomized in a 1:1 ratio to either durvalumab 1,500 mg (n = 341) or placebo (n = 344) delivered by IV infusion (on day 1 and every 3 weeks [q.3.w.] thereafter) in combination with cisplatin 25 mg/m² and gemcitabine 1,000 mg/m² (each administered on days 1 and 8, and q.3.w. thereafter) for up to 8 cycles, followed by 1,500 mg durvalumab or placebo delivered by IV infusion every 4 weeks (q.4.w.) until clinical progression (or RECIST 1.1-defined radiological progressive disease [PD]), unacceptable toxicity, withdrawal of consent, or another discontinuation criterion. The primary end point of the TOPAZ-1 trial was OS, with secondary end points of progression-free survival (PFS), objective response rate (ORR), duration of response (DOR), disease control rate (DCR), and health-related quality of life (HRQoL), and treatment tolerability.^23,24

Baseline characteristics of the TOPAZ-1 trial were well balanced between treatment groups; however, according to the clinical experts, the trial may have enrolled a healthier group of patients with a lower ECOG PS than would be seen in the population in Canada. In the TOPAZ-1 trial, patients were mostly Asian (56.4%), with an even balance of males (50.4%) and females (49.6%) and a median age of 64 years, and most patients had initially unresectable disease (█████████). Most patients had IHCC (█████████), followed by GBC (█████████) and EHCC (█████████). There were 3 planned data cut-offs (DCOs); 2 interim analyses and 1 final analysis, although the second interim analysis (IA-2) was considered the final analysis. As of the most recent DCO (February 25, 2022), the median follow-up in the TOPAZ-1 trial was ██████ ███████ ███████████ ███ months and █████████████ ███████████ ███ months for the durvalumab plus gemcitabine and cisplatin and placebo plus gemcitabine and cisplatin groups for OS.^23,24

Efficacy Results

Results of the efficacy analyses of the TOPAZ-1 trial are summarized in Table 2. As of the final efficacy analysis of the TOPAZ-1 trial (IA-2, DCO of August 11, 2021), the median OS was 12.8 months (95% confidence interval [CI], 11.1 to 14.0 months) in the durvalumab plus gemcitabine and cisplatin group, and 11.5 months (95% CI, 10.1 to 12.5 months) in the placebo plus gemcitabine and cisplatin group. Durvalumab plus gemcitabine and cisplatin was associated with a prolonged OS compared to placebo plus gemcitabine and cisplatin (hazard ratio [HR], 0.80; 95% CI, 0.66 to 0.97; P = 0.021).²³ With an additional 6.5 months of follow-up (DCO of February 25, 2022), the median OS was 12.9 months (95% CI, 11.6 to 14.1 months) for durvalumab plus gemcitabine and cisplatin, and 11.3 months (95% CI, 10.1 to 12.5 months) for placebo plus gemcitabine and cisplatin, and durvalumab plus gemcitabine and cisplatin remained favoured over placebo plus gemcitabine and cisplatin (HR, 0.76; 95% CI, 0.64 to 0.91). Results for OS rate were consistent at IA-2 and the 6.5-month follow-up, with landmark OS rates at 12 months, 18 months, and 24 months of 54.3% versus 47.1%, 34.8% versus 24.1%, and 23.6% versus 11.5%, respectively.²⁴

At IA-2, the key secondary end point of PFS was in line with the primary end point. Durvalumab plus gemcitabine and cisplatin was associated with a prolonged PFS compared to placebo plus gemcitabine and cisplatin (HR, 0.75; 95% CI, 0.63 to 0.89; P = 0.001), with a median PFS of 7.2 months (95% CI, 6.7 to 7.4 months) for durvalumab plus gemcitabine and cisplatin, and 5.7 months (95% CI, 5.6 to 6.7 months) for placebo plus gemcitabine and cisplatin.²³ Results for PFS were not available at the 6.5-month additional follow-up DCO.

ORR was a secondary end point of the TOPAZ-1 study but was a primary outcome at the first interim analysis (IA-1). █████ █ ██ █████ █████ ███████ █████████ █████████ ████ █ ████ █████ ██ ████ ██ ██████ ██████████ █████ █ ██ ████ ██████ ███████ █████████ █████████ ██ ████████████ ████ ███ █████████. At IA-2, the ORR was 26.7% for durvalumab plus gemcitabine and cisplatin and 18.7% for placebo plus gemcitabine and cisplatin (odds ratio [OR], 1.60; 95% CI, 1.11 to 2.31; P = 0.011). The statistical test for this outcome was not adjusted for multiplicity, so there is an increased risk of type I error. Only 7 (2.1%) patients in the durvalumab plus gemcitabine and cisplatin group and 2 (0.6%) patients in the placebo plus gemcitabine and cisplatin group achieved a complete response (CR). No results for ORR were available at the 6.5-month update. The median DOR was 6.4 months ████ ███████████ ██████████████████ ███ in the durvalumab plus gemcitabine and cisplatin group and 6.2 months ██████ ███████ ███████████ ████████ ████ in the placebo plus gemcitabine and cisplatin group.²³

Secondary end points for HRQoL consisted of the time to deterioration and improvement rates for the European Organisation for Research and Treatment of Cancer Quality of Life Questionnaire Core 30 (EORTC QLQ-C30) and the EORTC 21-Item Cholangiocarcinoma and Gallbladder Cancer Quality of Life Questionnaire (EORTC QLQ-BIL21). For global health status (GHS)/QoL, the median time to deterioration was 7.4 months (95% CI, 5.6 to 8.9 months) for durvalumab plus gemcitabine and cisplatin compared to 6.7 months (95% CI, 5.6 to 7.9 months) for placebo plus gemcitabine and cisplatin. The median time to deterioration in functional groups ranged from 5.6 months to 10.1 months for durvalumab plus gemcitabine and cisplatin and 6.5 months to 10.0 months for placebo plus gemcitabine and cisplatin. The median time to deterioration in multiple and single symptom items ranged from 3.0 months for fatigue to 18.2 months for diarrhea in the durvalumab plus gemcitabine and cisplatin group and 3.5 months for fatigue to 11.0 months for diarrhea in the placebo plus gemcitabine and cisplatin group. █████ █ ██ ██████████ █ ██████ █████████ █████████ ████ ██████████ ██ ████ ██ ██████ ██████████ ███████████ █ ██ ███ ██ █████ ██ █████ █████████ █████████ ████ ██████████ ██ ████ ██ ██████ ██████████ ███████████ █ ██ ██████ ████ ████ ███ █████████ █████████ ████ ████ ██████ ██ ████ ██ ██████ ██████████ ███████████ █ ██ ██ ████ ████ ██ █████ █████████ █████████ ████ ██████████ ██ ████ ██ ██████ ██████████ █████ ███ ███ █ ██ █████ ███ ██ ████ ███ █████████ █████ ████ ████ ██████████ ██ ████ ██ ██████ ██████████ ███ ███ █████ █ ██ █ █ ███ █████ ███████ █████████ █████████ ████ ██████████ ██ ████ ██ ██████ ██████████ █ ████ ██████ █ ██ ██ ██ ██ ████ ███████ █████████ ██ ███████ ████ ██████████ ██ ████ ██ ██████ ██████████ ██ ██ ███ ████ █ ██ █ ███ ██████ ███████ █████████ █████████ ████ ██████████ ██ ████ ██ ██████ ██ ██ ██████ ██ █████████ █ ██ █ ████ █████ ███████ █████████ █████████ ████ ████████ ██ ██ ████ ██ ██████ ██ ████████ █████ █ ██ ██ ████████ ███████ █████████ █████████ █████

The median time to symptom deterioration on the EORTC QLQ-BIL21 ranged from 3.5 months to 11.7 months for durvalumab plus gemcitabine and cisplatin, and 3.7 months to 14.2 months for placebo plus gemcitabine and cisplatin. The proportion of patients experiencing improvement in symptom domains ranged from ███ ███████ ████████ █████████ for durvalumab plus gemcitabine and cisplatin, and from ████ ██████████████ █████ ██ ██ for placebo plus gemcitabine and cisplatin.²³

Harms Results

As of the final analysis (IA-2), the overall incidence of treatment-emergent adverse events (TEAEs) in the TOPAZ-1 study was comparable between durvalumab plus gemcitabine and cisplatin and placebo plus gemcitabine and cisplatin (336 [99.4%] versus 338 [98.8%]). The most frequent TEAEs for both durvalumab plus gemcitabine and cisplatin and placebo plus gemcitabine and cisplatin included anemia (163 [48.2%] versus ███ ██████ ██████ ███) and nausea █ ███ ██ ███████████ ██████████ ███ ████████ ███████, with differences of at least 5% only observed between the durvalumab plus gemcitabine and cisplatin and placebo plus gemcitabine and cisplatin groups for nausea. Grade 3 or 4 adverse events (AEs) were reported at a similar frequency in the 2 treatment groups, with 250 patients (74.0%) and 257 patients (75.1%) in the durvalumab plus gemcitabine and cisplatin group and the placebo plus gemcitabine and cisplatin group, respectively. The most common grade 3 or 4 AEs included anemia ███████ █████ ███████████ ████, decreased neutrophil count (21.0% versus 25.7%), and neutropenia (20.1% versus 21.1%). The incidence of serious adverse events (SAEs) was similar in the durvalumab plus gemcitabine group and the cisplatin and placebo plus gemcitabine and cisplatin group █████ █████████ ████ versus 151 [44.2%]), with the most common SAE being cholangitis in both groups (██ ███████ ████ █████ █████████ [5.0%]). The proportion of patients that discontinued treatment due to TEAEs was █ ████████ in the durvalumab plus gemcitabine and cisplatin group, and 15.2% in the placebo plus gemcitabine and cisplatin group, driven mainly by █████ █ ██ ██████████ ███████ ████ █████ ███ ██████ ██ ███ in the placebo group. Deaths due to AEs were reported in 13 (3.8%) patients in the durvalumab plus gemcitabine and cisplatin group, and 14 (4.1%) patients in the placebo plus gemcitabine and cisplatin group, with most deaths in the durvalumab group due to █████ █ ██ ██████████ ███████ █████████ █████████ ████ ██ ███ █████ ██ ████ ██ ██████ ██████████ █████ █ ██ ██████████ ███████ █████████ █████████ ██ █ ███ ███ █ ███████ █████████ █████ █████████ ████████ in the placebo group.^23,24

The incidence of notable harms, including immune-mediated AEs (imAEs), infusion-related reactions (IRRs), infections, and GI events was generally more frequent in the durvalumab plus gemcitabine and cisplatin group than in the placebo plus gemcitabine and cisplatin group. At the final analysis, imAEs were identified for 43 (12.7%) patients in the durvalumab plus gemcitabine and cisplatin group and 16 (4.7%) patients in the placebo plus gemcitabine and cisplatin group; of these, 8 (2.4%) and 5 (1.5%) patients, respectively, experienced grade 3 or 4 imAEs. IRRs were reported in ███ ████ ███ ████ ███████████ ████████ ███ patients in the durvalumab plus gemcitabine and cisplatin and placebo plus gemcitabine and cisplatin groups, respectively. Infections and infestations occurred in ███████ ████ ███████ patients in the durvalumab plus gemcitabine and cisplatin group, and ████ █████████ ███ ██ patients in the placebo plus gemcitabine and cisplatin group. GI disorders occurred in █ ██ ██████ ███ ████ ██████ ███ ██████████ █████ █████ patients in the durvalumab plus gemcitabine and cisplatin and placebo plus gemcitabine and cisplatin groups, respectively. Immune-mediated GI events occurred in 3 ███ ██████ ██████ ███ patients in the durvalumab plus gemcitabine and cisplatin group and in 1 (0.3%) patient in placebo plus gemcitabine and cisplatin group.^23,24

Table 2: Summary of Key Results From TOPAZ-1

AE = adverse event; CI = confidence interval; Cis = cisplatin; CR = complete response; D = durvalumab; DCO = data cut-off; FAS = full analysis set; Gem = gemcitabine; GI = gastrointestinal; HR = hazard ratio; IA-2 = interim analysis 2; imAE = immune-mediated adverse event; IRR = infusion-related reaction; NR = not reported; OR = odds ratio; OS = overall survival; ORR = objective response rate; PFS = progression-free survival; PR = partial response; SAE = serious adverse event; WDAE = withdrawal due to adverse event.

^aP value for ORR was not controlled for multiple testing.

Sources: TOPAZ-1 CSR (IA-2 and 6.5-month update).^23,24

Critical Appraisal

TOPAZ-1 was an international, double-blind, phase III randomized controlled trial (RCT). Appropriate methods of randomization, treatment allocation, and stratification were employed using Interactive Web Response Systems (IWRS) and Interactive Voice Response Systems (IVRS). The methods for randomization were considered by the CADTH review team to result in a low risk of bias. Patients did not differ with regard to baseline demographic or disease characteristics, indicating that randomization was successful. There was potential for unblinding in the TOPAZ-1 study in the event of differences in rates of disease progression and known treatment-related toxicities associated with durvalumab, such as imAEs. Overall, withdrawals and discontinuations due to AEs were similar across treatment groups and, therefore, likely did not result in unblinding. Methods to maintain blinding were also appropriate, although the number of patients who were unblinded for various reasons was not reported. Two interim analyses and 1 final analysis were planned for the TOPAZ-1 trial. As of IA-2, the study met its primary end point ██ ██████ ██████ ████ ████ █████ and was considered the final analysis by the Independent Data Monitoring Committee. Trials that stop early for benefit may overestimate the relative benefits of treatment,^25,26 although the presence and extent of overestimation is uncertain. The 6.5-month follow-up was performed at 76.9% maturity for OS, which demonstrated consistent benefit compared to IA-2, although comparisons between DCOs were not formally conducted. Acceptable methods to account for multiplicity were conducted in the TOPAZ-1 trial for the primary efficacy end point of OS (via family-wise error rate) and the selected key secondary end point, PFS, using a hierarchical testing procedure. Other secondary outcomes, including ORR, DOR, EORTC QLQ-C30 and EORTC QLQ-BIL21, and exploratory outcomes (such as Patient Global Impression of Severity [PGI-S], Patient Reported Outcomes version of the Common Terminology Criteria for Adverse Events [PRO-CTCAE], and the 5-Level EQ-5D [EQ-5D-5L]), were not controlled for multiplicity, so need to be interpreted with consideration of the increased risk of type I error. For all time points for all HRQoL and patient-reported outcome (PRO) end points, there was a notable amount of missing data; thus, there was a high risk of attrition bias for these outcomes.The inclusion and exclusion criteria for TOPAZ-1 trial were generally as expected for patients with locally advanced or metastatic BTC, although histologic diagnosis is not always confirmed, the clinical experts consulted by CADTH explained. It was considered that patients included in the study may be a less sick population for various reasons, including ECOG PS, where the clinical experts noted that patients with an ECOG PS of 2 would also be eligible to receive durvalumab plus gemcitabine and cisplatin, provided they have adequate organ function. The clinical experts consulted by CADTH highlighted some characteristics that were not reflective of clinical practice in Canada, primarily that there was an overrepresentation of Asian patients in the TOPAZ-1 trial (56.4%), which is not representative of what would be seen in the population in Canada. This was a reflection of where the trial was conducted, as no Canadian sites were included, and most patients were enrolled from Asian countries (n = 374 [54.6%]). Regarding BTC subtypes, the experts felt that the proportion of patients with IHCC was higher than seen in the real world, and the proportion with EHCC was lower than seen in the real world, which may have a subtle effect on prognosis. The clinical experts consulted by CADTH noted that the proportion of patients receiving subsequent anticancer therapy was higher than expected, although that likely reflected the location where patients were enrolled in the trial. It was hypothesized by the clinical experts that the elevated proportion of patients receiving subsequent anticancer therapy was due to the baseline ECOG PS, where patients likely remained fit enough for subsequent lines of therapy. This was not considered reflective of clinical practice in Canada by the experts, so may not be generalizable to the population in Canada. Outcomes of the TOPAZ-1 trial were clinically relevant and important to both clinicians and patients. Measures of HRQoL were prespecified in the protocol and suggested no detriment to QoL with durvalumab plus gemcitabine and cisplatin; however, no conclusions for HRQoL or treatment tolerability can be drawn, owing to the methodological limitations associated with these analyses.

Indirect Comparisons

No indirect evidence was included in the sponsor’s submission to CADTH. One systematic literature review (SLR) and network meta-analysis (NMA) was identified in the literature search that matched the inclusion and exclusion criteria of this review. Feng et al. (2022)²⁷ conducted an NMA comparing the OS and PFS of different first-line treatment regimens for patients with advanced BTC using methods of random-effects. A total of 19 different treatment regimens were included in the NMA, although only comparisons involving treatments identified in the CADTH review protocol (Table 4) were of interest to this review. Data provided for comparisons of interest were derived from the TOPAZ-1 trial. There was insufficient evidence to draw conclusions for the outcome of OS (HR, 0.27; 95% CI, 0.06 to 1.29), whereas for PFS, the results produced by the NMA for comparisons of interest were in favour of durvalumab plus gemcitabine and cisplatin (HR, 0.22; 95% CI, 0.08 to 0.62).²⁷ There was notable heterogeneity in the populations from the studies included in the NMA, including differences in age, ECOG PS, and primary tumour location, with different proportions of patients with IHCC, EHCC and GBC, and some trials including patients with AoV cancer. This heterogeneity likely resulted in the notable imprecision, given the wide 95% CIs, and uncertainty in the comparative efficacy estimates.

Other Relevant Evidence

No long-term extension studies or other relevant studies were included in the sponsor’s submission to CADTH.

Conclusions

Patients and clinicians highlighted the need for new effective treatments that prolong life, maintain QoL, and reduce side effects compared to current treatments. In the randomized, phase III TOPAZ-1 trial, durvalumab plus gemcitabine and cisplatin demonstrated a statistically significant, albeit modest, improvement in median OS compared with placebo plus gemcitabine and cisplatin, which is the current SOC in Canada for patients with locally advanced or metastatic BTC. However, the clinical importance of the modest survival gain afforded by treatment with durvalumab compared with SOC is subjective and may vary from patient to patient, although the OS rates may also suggest that patients achieve longer OS benefit. Analyses of secondary outcomes, including PFS, and ORR supported the efficacy of adding durvalumab to the SOC. Because the findings for survival and response outcomes were based on an interim analysis, the benefits of durvalumab relative to SOC may be overestimated, although the presence and extent of any overestimation is uncertain. No definitive conclusions could be drawn about the impact of treatment with durvalumab compared with SOC on HRQoL or time to deterioration, owing to limitations of the analyses, including a lack of statistical testing and risk of attrition bias. The safety of durvalumab is well known, considering its use in other indications; however, aside from the AEs specific to durvalumab, the frequency of AEs and SAEs was similar in the 2 treatment groups. Overall, although there are limitations associated with stopping the trial at the second interim analysis, and also surrounding the statistical analysis for secondary end points, the TOPAZ-1 study demonstrated improvement in OS with the addition of durvalumab to gemcitabine and cisplatin. However, given the modest gain in OS observed in the trial, the clinical benefit is likely to be incremental, and there remains uncertainty about the magnitude of clinical benefit in the locally advanced or metastatic BTC population in Canada.

Introduction

Disease Background

BTC refers to a heterogeneous group of GI adenocarcinomas in the liver, gallbladder, and bile ducts. There are 4 subtypes of BTC: IHCC, which occurs in the small bile ducts in the periphery of the liver; EHCC, which occurs in the main ducts of the hilum or distal region of the bile duct; GBC; and AoV.^1-3

Although BTCs comprise less than 1% of all cancers, they account for 10% to 15% of primary liver cancers,^4,5 which are the fifth most common cancers worldwide,^28,29 and the 12th and 18th most common cancers in males and females in Canada, respectively, in 2021.⁶ The other 85% to 90% of primary liver cancers are hepatocellular carcinomas.^28,29 The most common subtype of BTC is GBC.^7,8 IHCC makes up approximately 10% to 20% of BTC cases, whereas EHCC makes up 30% to 40% of BTC cases.^5,9 The incidence of BTC varies globally, depending on various risk factors,⁴ with an incidence of cholangiocarcinoma and GBC of 0.3 to 3.5 per 100,000 and 1.6 to 2.0 per 100,000, respectively,^2,5 in Europe, the US, and Australasia, although incidence rates are said to be increasing.^1,10 Rates of BTC tend to be higher in Asian cultures, with China, Thailand, South Korea, Japan, and Taiwan having some of the highest rates of incidence and mortality;³⁰ the highest incidence, of 90 per 100,000, is seen in Northeast Thailand.³¹ Although there are few estimates of BTC in Canada, 1 study estimated the average national incidence rate of GBC and extrahepatic BTC at 30.92 cases per 1,000,000 individuals per year (approximately 3 per 100,000),¹¹ which increased between 1994 and 2012.¹² It has been estimated that there are approximately 400 and 5,000 new cases of cholangiocarcinoma diagnosed each year in Canada and the US,, respectively,^4,13 although these numbers are nearly 20 years old and may not reflect current incidence rates.

Symptoms of BTCs are often nonspecific and include nausea, emesis, anorexia, weight loss, abdominal pain, and jaundice. As such, the majority of patients present with locally advanced or metastatic BTC, and up to 90% of BTC cases are inoperable at the time of diagnosis.^14,15 Symptoms often reflect the location of the cancer; patients with IHCC often present with nonspecific symptoms, including fever, weight loss, and/or abdominal pain, whereas patients with EHCC present with jaundice due to biliary obstruction.^16,17 Generally, a diagnosis of BTC is made with MRI, and magnetic resonance cholangiopancreatography, ultrasound, or CT scan are used to identify the disease, locate the level of biliary obstruction, and identify any regional lymphadenopathy or metastatic disease. In general, a pathologic diagnosis with endoscopic retrograde cholangiopancreatography, percutaneous transhepatic cholangiography, or magnetic resonance cholangiopancreatography should be obtained before the initiation of nonsurgical treatments.^5,9 In cases of advanced disease, serum CA19 to 9 has some added value as a tumour marker, but it is not considered diagnostic.³²

Risk factors for BTC vary by subtype and by region. Well-established risk factors in the Western world include primary sclerosing cholangitis, chronic inflammation or infection, age, hepatitis C infection, cirrhosis, and exposure to various toxins, such as dioxins and vinyl chlorides. In regions where liver flukes are common, patients with liver fluke infections are at an increased risk of BTC.^1,4,5,9 The incidence and risk of GBC is closely associated with cholelithiasis (gall stones), gallbladder polyps, primary sclerosing cholangitis, chronic infection (e.g., Salmonella typhi), congenital malformations, obesity, and other risk factors for gallbladder disease.^1,5,33 Globally, GBC is more common in females, whereas cholangiocarcinoma, particularly EHCC, is more common in males.^1,5,34

BTCs are typically staged using the American Joint Committee on Cancer tumour, node, metastases staging system. The prognosis for patients with BTCs is poor, with estimated 5-year survival rates in the US of 30%, 24%, and 2% for local, regional, and distant metastatic EHCC, respectively, and of 15%, 6%, and 2%, respectively, for IHCC. Five-year survival rates for GBC are similar, at 8% and 7% for patients with stages IIIA and IIIB disease, respectively, and at 4% and 2% for patients with stages IVA and IVB disease, respectively.¹⁴

Standards of Therapy

In patients with resectable, early-stage disease, surgery is the only potentially curative option. However, few patients are diagnosed with resectable disease, and nearly 50% relapse after surgery.^14,15,35

For patients with locally advanced or metastatic disease, treatment is often palliative. Platinum-based chemotherapy, most commonly the combination of gemcitabine (1,000 mg/m²) plus cisplatin (25 mg/m²), has remained, for more than 10 years, the preferred first-line SOC regimen for suitable patients with advanced BTC.⁹ The combination of gemcitabine plus cisplatin was established after results of the ABC-02 study in patients with locally advanced or metastatic BTC demonstrated better OS (11.7 months versus 8.1 months; HR, 0.64; 95% CI, 0.52 to 0.80) and better PFS (8.0 months versus 5.0 months; HR, 0.63; 95% CI, 0.51 to 0.77) than gemcitabine monotherapy.³⁶

Although not the focus of this review, there is currently no standard second-line treatment option for patients with locally advanced or metastatic BTC who experience disease progression after first-line treatment. Palliative treatment options include 5-FU monotherapy, FOLFIRI, or FOLFOX. Pemigatinib also has Health Canada market authorization with conditions for the treatment of adults with previously treated, unresectable locally advanced or metastatic cholangiocarcinoma that has a fibroblast growth factor receptor 2 fusion or other rearrangement,¹⁸ although it recently received a do not reimburse CADTH recommendation and is not publicly funded in Canada.¹⁹

Drug

Durvalumab is a fully immunoglobulin G1 kappa monoclonal antibody that selectively blocks the interaction of PD-L1 with PD-1 and CD80. Expression of PD-L1 helps tumours evade detection and elimination by the immune system. When PD-L1 interacts with PD-1 and CD80, the PD-L1/ PD-1 pathway decreases cytotoxic T-cell activity, proliferation, and cytokine production. By selectively blocking the interaction of PD-L1 with PD-1 and CD80, durvalumab inhibits the immunosuppressive effects of the PD-L1/PD-1 pathway, enhancing antitumour immune responses.²⁰

Durvalumab is provided as a single-use vial of 120 mg durvalumab/2.4 mL (nominal concentration of 50 mg/mL) and 500 mg durvalumab/10 mL (nominal concentration of 50 mg/mL) and is administered as an IV infusion over 60 minutes. The recommended dose of durvalumab is 1,500 mg in combination with chemotherapy every 3 weeks, followed by 1,500 mg every 4 weeks as monotherapy until disease progression or unacceptable toxicity. Patients with a body weight of 30 kg or less must receive weight-based dosing of durvalumab at 20 mg/kg in combination with chemotherapy dosed every 3 weeks, followed by monotherapy at 20 mg/kg every 4 weeks until weight increases to greater than 30 kg. When given on the same day, durvalumab should be administered before chemotherapy.²⁰

Durvalumab has received an NOC from Health Canada for the treatment of patients with locally advanced or metastatic BTC in combination with gemcitabine-based chemotherapy.²⁰ Durvalumab is also indicated for the treatment of locally advanced, unresectable, stage III NSCLC and for ES-SCLC. In addition, durvalumab has received marketing authorization with conditions for the treatment of locally advanced or metastatic urothelial carcinoma, pending the results of clinical trials.²⁰ Durvalumab has been previously reviewed by CADTH for the treatment of patients with locally advanced, unresectable NSCLC following curative-intent, platinum-based chemoradiation therapy, for up to a maximum of 12 months, as well as for first-line treatment of adult patients with ES-SCLC in combination with etoposide and either carboplatin or cisplatin.^21,22

Stakeholder Perspectives

Patient Group Input

The information in this section is a summary of input provided by the patient groups who responded to CADTH’s call for patient input and from clinical experts consulted by CADTH for the purpose of this review.

Patient group input for the review of durvalumab was provided as a submission by the CCSN, with participation from the Canadian Liver Foundation, Canadian Organization for Rare Disorders, Cholangiocarcinoma Foundation, CCRAN, Gastrointestinal Society, and Regroupement québécois des maladies orphelines. The CCSN is a national network of patients, families, survivors, friends, community partners, funders, and sponsors who have come together to promote the best SOC for patients with cancer.

The CCSN and participating patient groups conducted an online survey between July 18 and August 2, 2022, to collect quantitative data on durvalumab. Of the 58 individuals who responded to the survey, 12 had experience with durvalumab, 25 did not have experience with durvalumab, and 17 identified as caregivers. Of the 58 survey respondents, 21 were living in Canada, 35 were living in the US, 1 was living in the UK, and 1 was living in Spain. Additionally, the CCRAN conducted 7 interviews between July 7 and July 22, 2022, with 4 patients and 3 caregivers in Canada and the US who had experience with durvalumab.

In the CCSN survey, 29 respondents disclosed their disease stage, including 18 with late-stage or metastatic, 4 with middle-stage, 3 with early-stage, and 4 with unknown disease stage. Respondents to the CCSN survey reported BTC symptoms that had an impact on their QoL and day-to-day life, which included abdominal pain, loss of appetite and weight loss, nausea and vomiting, itching, dark urine, fever, jaundice, and light coloured, greasy stools. Caregivers indicated that caring for a loved 1 with BTC had an impact on their lifestyle, and reported feeling emotionally drained, experiencing challenges in managing medications and medical appointments, and being unable to plan ahead. The most commonly used treatments reported by respondents to the CCSN survey included gemcitabine plus cisplatin, immunotherapy, radiation, surgical therapy, targeted therapy, and FOLFOX. Most CCSN respondents reported tiredness, difficulty sleeping, hair loss, nausea and vomiting, muscle weakness, numbness and tingling of the arms and legs, and diarrhea as adverse effects associated with treatment. Although most respondents indicated no issues with accessing treatments, they did cite travel costs, limited availability in the community, and financial hardship due to cost as challenges.

Most survey respondents with durvalumab experience indicated that there was little to no difference in symptom management, side effects, ease of use, or disease progression between durvalumab and other therapies they had received for BTC. Respondents reported fatigue, constipation, decreases in white blood cell and platelet counts, and anemia as side effects of durvalumab. In the CCRAN interviews, 4 patients and 3 caregivers described their experience with durvalumab for BTC. Patients accessed durvalumab through compassionate access programs, clinical trials, private insurance coverage, and by paying out of pocket. Most CCRAN interviewees described few to no side effects associated with durvalumab and reported that their cancer had regressed with treatment. Further, CCRAN respondents indicated that durvalumab was easier to use, with a shorter duration of infusion compared to other treatment options. There was a consensus among patients that durvalumab should be made available to eligible patients.

Survey respondents reported that new treatments should maintain QoL, prolong life, provide a cure, reduce side effects from current treatments, delay the onset of symptoms, and be easy to use. When asked to describe how much of an improvement would be needed from a new treatment to make it better than current treatments, the consensus was that prolonged life with similar or reduced side effects was most important, and that QoL and ease of access should be maintained as much as possible.

Clinician Input

Input From Clinical Experts Consulted by CADTH

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

Unmet Needs

First-line therapy for advanced BTC has not changed in more than a decade and remains cytotoxic chemotherapy in the form of gemcitabine and cisplatin. The median OS with current SOC is just less than 12 months. As such, the experts considered prolonged survival an important unmet need for patients with locally advanced or metastatic BTC because less than 15% of patients remain alive 24 months after diagnosis. Additionally, improvement in disease-related symptoms was considered important.

Place in Therapy

According to the clinical experts consulted by CADTH, surgery is currently the only curative treatment for BTC; however, the disease is often detected at an advanced stage and is usually inoperable at diagnosis. The standard first-line treatment for locally advanced or metastatic BTC has been the combination of gemcitabine and cisplatin since publication of the ABC-02 study in 2010, which demonstrated better OS and PFS than gemcitabine alone.³⁶ For patients with a relatively poor performance status, gemcitabine alone is often used. One expert noted that for patients with biliary obstruction, interventions such as ERCP or percutaneous transhepatic cholangiography often have to be considered to relieve the obstruction before treatment.

The experts emphasized that there is currently no standard second-line treatment for locally advanced or metastatic BTC, and patients commonly progress after first-line treatment. Several clinical trials have aimed to improve survival outcomes with the addition of more targeted therapies, the clinical experts noted, but the results of these studies have largely been negative. However, a recent RCT comparing FOLFOX to active symptom control showed survival benefits, the experts reported.³⁷

Given that the mechanism of action of durvalumab differs from that of cytotoxic chemotherapy, the experts explained that its complementary use with the current gemcitabine and cisplatin SOC (up to 8 cycles) is expected to shift the first-line treatment paradigm and is expected to provide incremental improvement in outcomes.

Patient Population

There is currently no companion diagnostic required, or there are no predictive biomarkers for locally advanced or metastatic BTC. Microsatellite instability (MSI) is rare in this population, the experts noted, although many Canadian centres have reflex testing for MSI. And not all patients will have had a pathologic diagnosis, they added, as some patients with extrahepatic disease are not amenable to a biopsy.

According to the clinical experts and given that the TOPAZ study²³ did not specify a biomarker, such as PD-L1 expression, patients who are candidates for gemcitabine and cisplatin and who have preserved organ function and a good performance status are most suitable for durvalumab plus gemcitabine and cisplatin, regardless of the presence of cancer-related symptoms.

Conversely, the experts added, durvalumab plus gemcitabine and cisplatin would only be suitable for patients who do not have a contraindication to immunotherapy. The experts also felt that it would not be appropriate to recommend other treatments before initiating treatment with durvalumab.

Assessing Response to Treatment

The clinical experts agreed that patients would be assessed clinically during routine follow-up (i.e., q.3.w., given the cycle length) for changes in symptoms, such as fatigue and pain, and for clinical and functional status. Response to treatment would be assessed every 9 to 12 weeks with cross sectional imaging, most often a CT scan, although MRI or ultrasound may be used to evaluate tumour response or stability.

The experts also emphasized that tumour marker assessments of CA 19 to 9 are often evaluated and followed in patients with adequate biliary decompression and elevated CA 19 to 9 levels.

Discontinuing Treatment

The clinical experts stated that durvalumab would be discontinued at objective, clinical, or radiologic disease progression, confirmed worsening of symptoms, or unacceptable immune-related toxicity.

Prescribing Conditions

The clinical experts stated that durvalumab should only be prescribed by medical oncologists. Durvalumab should be administered in a systemic chemotherapy unit by qualified nurses under the supervision of a medical oncologist. The experts also noted that, at this time, all centres would likely be familiar with immune checkpoint inhibitors and many would have had experience with durvalumab administration.

Clinician Group Input

Two clinician groups provided input for the review of durvalumab: the CGOEN, represented by 7 clinicians; and the OH-CCO DAC, represented by 5 clinicians. The CGOEN is a virtual network of Canadian GI oncology clinicians who contribute knowledge about GI cancer and its treatments. The OH-CCO DAC provides evidence-based clinical and health system guidance on drug-related issues.

Both clinician groups noted that current treatment goals for patients with unresectable, metastatic BTC include extending lives, delaying disease progression, and maintaining QoL. The CGEON indicated that cisplatin plus gemcitabine is the only currently available treatment option for patients with unresectable BTC, although the OH-CCO DAC also indicated that carboplatin and gemcitabine may be used in the first-line setting. The CGEON clinicians noted that the majority of BTC patients do not reach second-line treatment as the disease progresses rapidly, and that second-line treatment with 5FU and oxaliplatin provides a modest survival benefit but is poorly tolerated. The clinicians noted that no molecularly targeted drugs for BTC are funded in Canada. Thus, both clinician groups emphasized that the limited number of treatment options and the moderate survival benefit provided by gemcitabine and cisplatin treatment constitute a significant unmet need for more effective treatments in this setting.

Given the lack of available options, both clinician groups indicated that there was no rationale for patients to try other treatments before initiating durvalumab plus gemcitabine and cisplatin, and that the addition of durvalumab to the current SOC would not affect the sequencing of subsequent therapy. The CGEON and OH-CCO DAC inputs indicated that patients with unresectable BTC who align with the clinical trial criteria would be most suited for treatment with durvalumab. Patients least suitable for treatment were identified as those with contraindications to immunotherapy, inadequate liver or renal function, or an ECOG PS of 3 or more. The CGEON and OH-CCO DAC both indicated that, in clinical practice, clinical condition and/or radiologic progression are used to determine response to treatment. The CGEON indicated that a clinically meaningful response to treatment would be maintenance or improvement in QoL and prolongation of survival, whereas disease progression or intolerance would be factors for discontinuing treatment. The groups agreed that durvalumab could be administered in any setting where standard chemotherapy is delivered, under the supervision of a specialist. The OH-CCO DAC agreed with the weight-based dosing method with a cap, which is consistent with other disease-site regimens, and noted that flat dosing can result in overtreatment.

Drug Program Input

The drug programs provide input on each drug being reviewed through CADTH’s reimbursement review processes by identifying issues that may affect their ability to implement a recommendation. The implementation questions and corresponding responses from the clinical experts consulted by CADTH are summarized in Table 3.

Table 3: Summary of Drug Plan Input and Clinical Expert Response

AoV = ampulla of Vater; BTC = biliary tract cancer; ECOG PS = Eastern Cooperative Oncology Group Performance Status; OS = overall survival; PAG = Provincial Advisory Group; q.3.w. = every 3 weeks; q.4.w. = every 4 weeks; SOC = standard of care.

Clinical Evidence

The clinical evidence included in the review of durvalumab is presented in 3 sections. The first section, the Systematic Review, includes pivotal studies provided in the sponsor’s submission to CADTH and Health Canada, as well as studies that were selected according to an a priori protocol. The second section includes indirect evidence selected from the literature that met the selection criteria specified in the review. No indirect evidence was provided by the sponsor. The third section includes sponsor-submitted long-term extension studies and additional relevant studies that were considered to address important gaps in the evidence included in the Systematic Review.

Systematic Review (Pivotal and Protocol-Selected Studies)

Objectives

To perform a systematic review of the beneficial and harmful effects of durvalumab (50 mg/mL) 1,500 mg in combination with chemotherapy for the treatment of locally advanced or metastatic BTC.

Methods

Studies selected for inclusion in the Systematic Review included pivotal studies provided in the sponsor’s submission to CADTH and Health Canada, as well as those meeting the selection criteria presented in Table 4. Outcomes included in the CADTH review protocol reflect outcomes considered to be important to patients, clinicians, and drug plans.

Of note, the systematic review protocol presented below was established before the granting of an NOC from Health Canada.

Table 4: Inclusion Criteria for the Systematic Review

AE = adverse events; BTC = biliary tract cancer; DCR = disease control rate; DOR = duration of response; ECOG PS = Eastern Cooperative Oncology Group Performance Status; EHCC = extrahepatic cholangiocarcinoma; GBC = gall bladder cancer; GI = gastrointestinal; HRQoL = health-related quality of life; IHCC = intrahepatic cholangiocarcinoma; ORR = objective response rate; OS = overall survival; PD-L1 = programmed cell death 1 ligand 1; PFS = progression-free survival; PGI-S = Patient Global Impression of Severity; q.3.w. = every 3 weeks; q.4.w. = every 4 weeks; RCT = randomized controlled trial; SAE = serious adverse event; TTD = time to treatment discontinuation; WDAE = withdrawal due to adverse event.

The literature search for clinical studies was performed by an information specialist using a peer-reviewed search strategy according to the PRESS Peer Review of Electronic Search Strategies checklist.³⁸

Published literature was identified by searching the following bibliographic databases: MEDLINE All (1946–) via Ovid and Embase (1974–) via Ovid. All Ovid searches were run simultaneously as a multifile search. Duplicates were removed using Ovid deduplication for multifile searches, followed by manual deduplication in Endnote. The search strategy comprised both controlled vocabulary, such as the National Library of Medicine’s MeSH (Medical Subject Headings), and keywords. The main search concepts were Imfinzi or durvalumab and biliary cancer. The clinical trials registries searched were the US National Institutes of Health’s clinicaltrials.gov, WHO’s International Clinical Trials Registry Platform (ICTRP) search portal, Health Canada’s Clinical Trials Database, and the European Union Clinical Trials Register.

No filters were applied to limit the retrieval by study type. Retrieval was not limited by publication date or by language. Conference abstracts were excluded from the search results. Refer to Appendix 1 for the detailed search strategies.

The initial search was completed on August 10, 2022. Regular alerts updated the search until the meeting of the CADTH pan-Canadian Oncology Drug Review Expert Committee (pERC) on December 7, 2022.

Grey literature (literature that is not commercially published) was identified by searching relevant websites from the Grey Matters: A Practical Tool For Searching Health-Related Grey Literature checklist.³⁹ Included in this search were the websites of regulatory agencies (FDA and European Medicines Agency). Google was used to search for additional internet-based materials. Refer to Appendix 1 for more information on the grey literature search strategy.

Two CADTH clinical reviewers independently selected studies for inclusion in the review based on titles and abstracts, according to the predetermined protocol. Full-text articles of all citations considered potentially relevant by at least 1 reviewer were acquired. Reviewers independently made the final selection of studies to be included in the review, and differences were resolved through discussion.

Findings From the Literature

A total of 1 study was identified from the literature for inclusion in the Systematic Review (Figure 1). That study is summarized in Table 5.

Figure 1: Flow Diagram for Inclusion and Exclusion of Studies

Table 5: Details of Included Studies

ALT = alanine aminotransferase; ANC = absolute neutrophil count; AST = aspartate aminotransferase; CHF = congestive heart failure; CrCl = creatinine clearance; CSP = clinical study protocol; ctDNA = circulating tumour DNA; CTLA-4 = cytotoxic T-lymphocyte-associated antigen-4; DCR = disease control rate; DOR = duration of response; ECOG PS = Eastern Cooperative Oncology Group Performance Status; EORTC QLQ-BIL21 = European Organisation for Research and Treatment of Cancer 21-Item Cholangiocarcinoma and Gallbladder Cancer Quality of Life Questionnaire; EORTC QLQ-C30 = European Organisation for Research and Treatment of Cancer Quality of Life Questionnaire Core 30; GI = gastrointestinal; HBc = hepatitis B core; HBV = hepatitis B virus; HCV = hepatitis C virus; HRQoL = health-related quality of life; IBD = inflammatory bowel disease; ILD = interstitial lung disease; IP = investigational product; MSI = microsatellite instability; NCI-CTCAE = National Cancer Institute Common Terminology Criteria for Adverse Events; ORR = objective response rate; OS = overall survival; PD = progressive disease; PD-1 = programmed cell death 1 protein; PD-L1 = programmed cell death 1 ligand 1; PD-L2 = programmed cell death 1 ligand 2; PFS = progression-free survival; PGI-S = Patient Global Impression of Severity; PRO-CTCAE = Patient Reported Outcomes version of the Common Terminology Criteria for Adverse Events; q.3.w. = every 3 weeks; q.4.w. = every 4 weeks; RCT = randomized controlled trial; RECIST = Response Evaluation Criteria in Solid Tumours; SLE = systemic lupus erythematosus; TMB = tumour mutation burden; TL = target lesions; ULN = upper limit of normal; WHO = WHO.

Source: TOPAZ-1 CSR (IA-2).²³

Description of Studies

One study, TOPAZ-1, was included in this review. The TOPAZ-1 trial was a double-blind, placebo-controlled, international, randomized, phase III study designed to evaluate the efficacy and safety of adding durvalumab to the established chemotherapy regimen of gemcitabine and cisplatin in patients with previously untreated, unresectable, locally advanced or metastatic BTC. The primary objective of the TOPAZ-1 trial was to confirm superiority of the combination of durvalumab plus gemcitabine and cisplatin over gemcitabine and cisplatin alone in terms of OS in patients with first-line advanced BTC.²³

A total of 685 patients were centrally randomized 1:1 ratio by interactive response technology to either cisplatin 25 mg/m² and gemcitabine 1,000 mg/m² (each administered on days 1 and 8 q.3.w.) in combination with durvalumab 1,500 mg (n = 341) or placebo (n = 344) delivered by IV infusion (on day 1 q.3.w.); starting on cycle 1, for up to 8 cycles. After completion of the chemotherapy treatment period, patients received 1,500 mg durvalumab or placebo delivered by IV infusion every 4 weeks until clinical progression (or RECIST 1.1-defined radiological PD), unacceptable toxicity, withdrawal of consent, or another discontinuation criterion. Randomization was stratified by disease status (initially unresectable versus recurrent) and primary tumour site (IHCC versus EHCC versus GBC). The TOPAZ-1 trial enrolled patients from 17 countries,²³ but no Canadian investigative sites were included.

The TOPAZ-1 trial planned to include 3 DCOs consisting of 2 interim analyses and 1 final analysis. The DCO for IA-1 was December 18, 2020. The DCO for IA-2 was August 11, 2021 (database lock: September 13, 2021). The TOPAZ-1 trial met its primary end point at IA-2, so this was considered the final formal analysis.²³ A 6.5-month update of IA-2 was conducted based on a DCO of February 25, 2022 (database lock: March 22, 2022).²⁴

Populations

Inclusion and Exclusion Criteria

The inclusion and exclusion criteria for the TOPAZ-1 trial are summarized in Table 5. Briefly, eligible patients were required to have histologically confirmed, unresectable advanced or metastatic adenocarcinoma of the biliary tract, including IHCC, EHCC, or GBC, as well as a WHO/ECOG PS of 0 or 1. Additionally, if unresectable or metastatic at diagnosis, patients were required to have untreated disease; however, patients who had received prior adjuvant therapy were permitted, provided that it was completed more than 6 months ahead of this study. Given the different genetic profile ampullary cancer, such patients were excluded to minimize diversity. Patients with active or prior autoimmune disorders were excluded because of the mechanism of action of durvalumab and the known AEs associated with immunotherapy.²³

Baseline Characteristics

Baseline characteristics of patients in the TOPAZ-1 trial are summarized in Table 6. Baseline characteristics were well balanced between the durvalumab plus gemcitabine and cisplatin group and the placebo plus gemcitabine and cisplatin group. Patients were mostly Asian in the durvalumab plus gemcitabine and cisplatin group and the placebo plus gemcitabine and cisplatin group (54.3% and 58.4%, respectively), with an even split between male (49.6% and 51.2%, respectively) and female (50.4% and 48.8%, respectively) patients. The median age was 64 years (range, 20 to 85 years), with most patients younger than 65 years (53.1% and 53.5%, respectively, in the durvalumab plus gemcitabine and cisplatin group and the placebo plus gemcitabine and cisplatin group). The majority of patients had metastatic disease at baseline (88.9% and 83.1%, respectively, in the durvalumab plus gemcitabine and cisplatin group and the placebo plus gemcitabine and cisplatin group). Across groups, the proportion of patients with a WHO/ECOG PS of 0 (50.7% and 49.3%, respectively, in the durvalumab plus gemcitabine and cisplatin group and the placebo plus gemcitabine and cisplatin group) or 1 (47.4% and 52.6%, respectively) was well balanced. Patients were stratified by disease status (initially unresectable [80.4% and 80.2%, respectively, in the durvalumab plus gemcitabine and cisplatin group and the placebo plus gemcitabine and cisplatin group] versus recurrent [19.6% versus 19.8%, respectively]) and by primary tumour location (IHCC [56.9% and 57.3% respectively, in the durvalumab plus gemcitabine and cisplatin group and the placebo plus gemcitabine and cisplatin group] versus EHCC [18.2% and 18.3%, respectively] versus GBC [24.9% and 24.4%, respectively]).²³

Table 6: Summary of Baseline Characteristics (IA-2; DCO of August 11, 2021)

BMI = body mass index; Cis = cisplatin; D = durvalumab; DCO = data cut-off; ECOG PS = Eastern Cooperative Oncology Group Performance Status; EHCC = extrahepatic cholangiocarcinoma; FAS = full analysis set; GBC = gallbladder cancer; Gem = gemcitabine; IA-2 = second interim analysis; IHCC = intrahepatic cholangiocarcinoma; IVRS = Interactive Voice Response System; MSI = microsatellite instability; PD-L1 = programmed cell death 1 ligand 1; TIP = tumour and/or immune cell positivity; WHO = WHO.

^aOverall, 5 of 333 (1.5%) patients with an MSI result were MSI high. MSI status missing includes MSI-unknown and not tested.

Source: TOPAZ-1 CSR (IA-2).²³

Interventions

The dosing schedule of durvalumab or placebo plus gemcitabine and cisplatin is displayed in Figure 2. In the durvalumab plus gemcitabine and cisplatin arm, patients received 25 mg/m² cisplatin and 1,000 mg/m² gemcitabine administered on days 1 and 8 every 3 weeks plus 1,500 mg durvalumab starting on cycle 1, for a maximum of 8 cycles, delivered by IV infusion every 3 weeks. Patients randomized to the comparator arm received 25 mg/m² cisplatin and 1,000 mg/m² gemcitabine administered on days 1 and 8 every 3 weeks plus placebo delivered by IV infusion every 3 weeks starting on cycle 1, for a maximum of 8 cycles. During day 1 visits of each cycle, administration of durvalumab or placebo occurred first, followed by chemotherapy. Given the potential disadvantages, including cumulative toxicities, continuing chemotherapy after 8 cycles was not allowed in the TOPAZ-1 trial. After completing the gemcitabine and cisplatin treatment period (first 8 cycles), patients received 1,500 mg durvalumab or placebo as monotherapy delivered by IV infusion every 4 weeks until clinical progression (or RECIST 1.1-defined radiological PD), unacceptable toxicity, withdrawal of consent, or any other discontinuation criterion was met.²³

Patients, investigators, and study centre staff were blinded to treatment allocation; IV solutions of durvalumab and placebo were blinded using a translucent-coloured or opaque sleeve, fastened with tamper-evident tape to maintain the double-blind conditions. An unblinded pharmacist prepared the durvalumab and placebo for patients as specified by the randomization scheme and IVRS. Investigators were unblinded to treatment allocation only in cases of medical emergencies or, at progression of disease, patients may have been unblinded at the request of the investigator to inform decisions regarding subsequent anticancer treatment. After the final analysis, open-label durvalumab was provided up to the time that patients discontinued treatment.²³

Figure 2: Durvalumab or Placebo Plus Gemcitabine and Cisplatin Dosing Schedule

d/c = discontinuation; PD = progressive disease; q3w = every 3 weeks; q4w = every 4 weeks; RECIST 1.1 = Response Evaluation Criteria in Solid Tumours Version 1.1.

Source: TOPAZ-1 CSR (IA-2).²³

If a patient’s weight fell to 30 kg or below, the patient received weight-based dosing equivalent to 20 mg/kg of durvalumab (or placebo) every 3 weeks (i.e., with chemotherapy) or every 4 weeks (i.e., during monotherapy) after consultation between the investigator and the study physician, until the patient’s weight increased to more than 30 kg, at which point the patient was to start receiving the fixed dosing of durvalumab 1,500 mg (or placebo) every 3 weeks or every 4 weeks.²³

Dose Modifications, Reductions, or Interruptions

Local standard clinical practice regarding dose modifications were followed for the combination of gemcitabine and cisplatin. Gemcitabine and cisplatin may have been delayed and subsequently resumed, per local standard clinical practice. If dosing was delayed for reasons other than causally related toxicity, dosing occurred as soon as possible. Dose reductions were permitted for gemcitabine and cisplatin. For patients who discontinued gemcitabine and cisplatin before completion of cycle 8 due to toxicity, treatment could continue with durvalumab or placebo monotherapy (q.4.w.) at the investigator’s discretion once toxicity resolved to grade 2 or less.²³

Dose reductions were not permitted for durvalumab (or placebo). Guidelines for the management of imAEs, IRRs, and nonimmune-mediated reactions for durvalumab were provided in the dosing modification and toxicity management guidelines. During the first 8 cycles of durvalumab or placebo in combination with gemcitabine and cisplatin, if durvalumab or placebo was discontinued due to toxicity related to either drug, the patient was discontinued from study treatment, completed the treatment discontinuation visit, and entered the follow-up period. Subsequent anticancer treatment was to be held until the toxicity resolved to grade 2 or less. At the discretion of the investigator, gemcitabine and cisplatin could be continued as the postdiscontinuation anticancer treatment.^23,41

Treatment Discontinuation and Withdrawal From Study

Patients were discontinued from study treatment for the following reasons^23,41:

• Withdrawal of consent from further treatment. Patients were free to discontinue treatment or withdraw from the study at any time. Patients who discontinued treatment were to continue to participate in the study for safety and survival follow-up, unless consent was specifically withdrawn for all further participation in any study procedures and assessments.

• An AE that, in the opinion of the investigator or the sponsor, contraindicated further dosing.

• Any AE that met the criteria for discontinuation as defined in the dosing modification and toxicity management guidelines or as defined in the local prescribing information for gemcitabine and cisplatin.

• Pregnancy or intent to become pregnant.

• Noncompliance with the study protocol that, in the opinion of the investigator or sponsor, warranted withdrawal from treatment (e.g., refusal to adhere to scheduled visits).

• Initiation of alternative anticancer therapy, including another investigational drug.

• Clinical progression or radiological progression and investigator determination that the patient is no longer benefiting from treatment.

Patients who permanently discontinued treatment for reasons other than RECIST 1.1-defined radiological PD were to continue to have RECIST scans performed every 6 weeks (± 1 week) for the first 24 weeks (relative to the date of randomization) and then every 8 weeks (± 1 week) thereafter (relative to the date of randomization) until PD plus at least 1 additional follow-up scan or death (whichever came first).^23,41

Treatment Through Progression

During the treatment period, patients who were clinically stable at an initial RECIST 1.1-defined radiological PD could continue to receive study treatment at the discretion of the investigator (with input from the patient) if they were deemed to be receiving clinical benefit. For these patients, imaging for tumour assessments were to be continued at a maximum scan interval of 8 weeks for the duration of treatment unless there was unacceptable toxicity, withdrawal of consent, or another discontinuation criterion was met.^23,41

Patients with RECIST 1.1-defined radiological PD who continued to receive their assigned treatment at the discretion of the investigator (with input from the patient) could receive treatment until it no longer provided clinical benefit, and imaging for tumour assessments were continued at a maximum scan interval of 8 weeks for the duration of treatment unless there was unacceptable toxicity, withdrawal of consent, or another discontinuation criterion is met. For patients who discontinued treatment due to radiological PD, 1 additional follow-up scan was performed at the next scheduled imaging visit and no less than 4 weeks after the prior assessment of PD. All patients were followed for survival until the end of the study.^23,41

Patients with rapid tumour progression or with symptomatic progression that required urgent medical intervention (e.g., central nervous system metastasis, respiratory failure due to tumour compression, or spinal cord compression) were not eligible to continue durvalumab (or placebo). For all patients who were treated through progression, the investigator was to ensure that the patients did not have any significant, unacceptable, or irreversible toxicities that indicated that continued treatment would not further benefit them.^23,41

Crossover from the placebo plus gemcitabine and cisplatin arm to the durvalumab plus gemcitabine and cisplatin arm was not permitted.^23,41

Concomitant and Rescue Therapy

Concomitant medications or treatments included those that were deemed necessary to provide adequate prophylactic or supportive care (e.g., acetaminophen or diphenhydramine); best supportive care (including antibiotics, nutritional support, correction of metabolic disorders, optimal symptom control, and pain management) and inactivated viruses (i.e., the influenza vaccine) were permitted for use, when necessary, for all patients.^23,41

Restricted, prohibited, and permitted concomitant medications included:^23,41

• any investigational anticancer therapy, monoclonal antibodies against CTLA-4, PD-1, or PD-L1 other than those under investigation in this study were not to be given while the patient was on study treatment.

• any concurrent chemotherapy, radiotherapy, immunotherapy, or biologic or hormonal therapy for cancer treatment other than those under investigation in this study. Concurrent use of hormones for noncancer-related conditions (e.g., insulin for diabetes and hormone replacement therapy) was acceptable. Local treatment of isolated lesions, excluding target lesions, for palliative intent was acceptable (e.g., radiotherapy for bone metastasis).

• live attenuated vaccines, which should not be administered until 30 days after the last dose of investigational product.

• immunosuppressive medications, including but not limited to systemic corticosteroids at doses exceeding 10 mg/day of prednisone or equivalent, methotrexate, azathioprine, and tumour necrosis factor-alpha blockers, which should not be given concomitantly or used for premedication before the intraosseous infusions.

• epidermal growth factor receptor tyrosine kinase inhibitors, which should not be given concomitantly and should be used with caution in the 90 days after the last dose of durvalumab, owing to the potential for pneumonitis and transaminase increases.

• herbal and natural remedies, which may have immune-modulating effects and should not be given concomitantly.

Rescue medications (steroids and other immunosuppressants) were available to manage imAEs experienced by patients receiving durvalumab. The permitted immunosuppressants were infliximab (e.g., for colitis) and mycophenolate (e.g., for hepatitis). Blinded and unblinded access and notifications were to be controlled using the IVRS or IWRS.²³

Outcomes

A list of efficacy end points identified in the CADTH review protocol that were assessed in the clinical trials included in this review is provided in Table 7 and is summarized in the subsequent text. A detailed discussion and critical appraisal of the outcome measures is provided in Appendix 3.

Table 7: Summary of Outcomes of Interest Identified in the CADTH Review Protocol

BOR = best overall response; DCR = disease control rate; DOR = duration of response; EORTC QLQ-BIL21 = European Organisation for Research and Treatment of Cancer 21-Item Cholangiocarcinoma and Gallbladder Cancer Quality of Life Questionnaire; EORTC QLQ-C30 = European Organisation for Research and Treatment of Cancer Quality of Life Questionnaire Core 30; EQ-5D-5L = 5-Level ED-EQ; HRQoL = health-related quality of life; ORR = objective response rate; OS = overall survival; PFS = progression-free survival; PGI-S = Patient Global Impression of Severity; PRO-CTCAE = PRO-CTCAE = Patient Reported Outcomes version of the Common Terminology Criteria for Adverse Events.

Efficacy Outcomes

In line with the primary objective of the TOPAZ-1 study, the primary end point was OS. PFS was a key secondary end point of the TOPAZ-1 trial. Other secondary end points in the TOPAZ-1 trial consisted of ORR, DOR, and DCR, safety, and HRQoL as measured by EORTC QLQ-C30 and EORTC QLQ-BIL21. Exploratory outcomes included symptom severity as measured by the PGI-S, treatment tolerability as measured by the PRO-CTCAE, and HRQoL as measured by the EQ-5D-5L.²³

The analysis of secondary end points of PFS, ORR, DOR, and DCR were conducted by investigator RECIST 1.1 assessments. For IA-1 to support early registration, a blinded independent central review (BICR) was performed for the end points of ORR and DOR for the first 200 randomized patients for at least 32 weeks. After IA-1, all images were collected and stored for potential BICR if considered necessary by the sponsor.²³

Overall Survival

The primary outcome of the TOPAZ-1 trial, OS, was defined as the time from the date of randomization until death from any cause. Any patient not known to have died at the time of analysis was censored on the last recorded date on which the patient was known to be alive.²³

Progression-Free Survival

PFS was a key secondary end point of the TOPAZ-1 trial and was defined as the time from the date of randomization until the date of RECIST 1.1-defined radiological PD or death, regardless of whether the patient withdrew from therapy or received another anticancer therapy before progression. Patients who had not progressed or died at the time of analysis were censored at the time of the most recent date of assessment after their last evaluable RECIST 1.1 assessment. However, if the patient progressed or died after 2 or more missed visits, the patient was censored at the time of the most recent evaluable RECIST 1.1 assessment before the 2 missed visits.²³

Objective Response Rate

The secondary end point of ORR, defined as the percentage of patients with confirmed CR or partial response (PR), per investigator, according to RECIST 1.1 criteria. The ORR was analyzed using all scans, regardless of whether they were scheduled or not. Data obtained up to progression, or the most recent evaluable assessment in the absence of progression, were included in the assessment of ORR. Patients who discontinued treatment without progression, received a subsequent anticancer therapy, and then responded were not considered responders in the ORR. The ORR was also defined using the BICR data (at IA-1 only) to establish a visit response of CR or PR, with the denominator being a subset of all randomized patients with measurable disease at baseline, per BICR.²³

Duration of Response

The DOR, per investigator assessment, was a secondary end point of the TOPAZ-1 trial and was defined as the time from the date of first documented response to the date of documented progression or death in the absence of disease progression. For patients with a measurable disease at baseline, DOR was summarized by treatment group. At IA-1, DOR was also defined using BICR data.²³

Disease Control Rate

The DCR, per investigator assessment, was defined as the rate of best objective response of no evidence of disease, CR, PR, or stable disease, according to RECIST 1.1. The DCR was summarized by treatment for the overall treatment duration at 24 weeks, 32 weeks, and 48 weeks. For IA-1, the DCR was also assessed by BICR.²³

HRQoL and Patient-Reported Outcomes

PRO assessments were completed using electronic case report forms at the study site. Questionnaires were completed before treatment dosing and before any other study procedures in the following order: EORTC QLQ-C30, EORTC QLQ-BIL21, PGI-S, EQ-5D-5L, and PRO-CTCAE.²³

EORTC QLQ-C30 and EORTC QLQ-BIL21

The EORTC QLQ-C30 assesses the impact of treatment on disease-related symptoms, impact of the disease, and HRQoL. It includes 30 items and assesses HRQoL and health status through 5 functional scales (physical, role, cognitive, emotional, and social), 3 symptom scales (fatigue, pain, and nausea and vomiting), and 1 GHS/QoL scale. It also includes 6 single-item symptom and impact measures: dyspnea, insomnia, appetite loss, constipation, diarrhea, and financial difficulties. A high score on a functional or GHS/QoL scale indicates a high level of functioning or global HRQoL, whereas a high score on a symptom scale or item indicates a high level of symptom burden.^23,42

The EORTC QLQ-BIL21 was developed to measure QoL specifically for patients with BTC. It consists of 21 questions: 3 single-item assessments related to treatment side effects, difficulties with drainage bags or tubes, and concerns regarding weight loss; and 18 items grouped into 5 symptom scales (pain [4 items], tiredness [3 items], jaundice [3 items], anxiety [4 items], and eating [4 items]). The EORTC QLQ-BIL21 has demonstrated content validity in BTC patients. It was developed based on a literature review, in-depth patient interviews, and health care provider interviews specific to BTC populations.⁴³ Its measurement properties have also been tested in BTC populations and include evidence of acceptable internal consistency reliability, test-retest reliability (reproducibility), and construct validity.^23,44

A clinically meaningful change or difference on the EORTC QLQ-C30 and EORTC QLQ-BIL21, based on patients with breast or lung cancer, was defined in the TOPAZ-1 trial as an absolute change in score from baseline of at least 10 points;^23,45 however, no minimally important difference (MID) for the EORTC QLQ-C30 or EORTC QLQ-BIL21 for patients with BTC was identified in the literature by the CADTH review team (Appendix 3).

Five-Level EQ-5D

The EQ-5D-5L is a generic questionnaire that provides a simple descriptive profile of health and a single index value for health status. The EQ-5D-5L questionnaire comprises 5 questions that cover 5 dimensions of health: mobility, self-care, usual activities, pain/discomfort, and anxiety/depression. Respondents also assess their health today using the EQ-visual analogue scale (VAS), which ranges from 0 (worst imaginable health) to 100 (best imaginable health).²³ No evidence of the validity, reliability, or responsiveness of the EQ-5D-5L, or of an MID among patients with BTC was identified in the literature by the CADTH review team (Appendix 3).

Symptom Severity

Patient Global Impression of Severity

The PGI-S item is included to assess how patients perceive the overall severity of their cancer symptoms over the previous 7 days. It is a single-item questionnaire, in which patients choose response options on a 6-point scale that range from no symptoms to very severe.²³ No evidence of the validity, reliability, or responsiveness of this single item, or of an MID among patients with BTC was identified in the literature by the CADTH review team (Appendix 3).

Treatment Tolerability

PRO-CTCAE and EORTC QLQ-BIL21 Item 49

Patient-reported treatment tolerability was explored using the PRO-CTCAE and EORTC QLQ-BIL21 item 49. The PRO-CTCAE is a tool developed by the National Cancer Institute to address tolerability from the patients’ perspective. The PRO-CTCAE is an item-bank of symptoms. Six different symptomatic AEs were selected for the PRO-CTCAE, based on available safety reports of durvalumab and gemcitabine and cisplatin. Generally, symptomatic AEs that were not included in the EORTC were prioritized for the PRO-CTCAE to reduce patient burden, per published recommendations.^23,46 No evidence of the validity, reliability, or responsiveness of the PRO-CTCAE, nor of an MID among patients with BTC was identified in the literature by the CADTH review team (Appendix 3). The psychometric properties of the PRO-CTCAE were evaluated in 940 adults with various cancer types.⁴⁷

Item 49 of the EORTC QLQ-BIL21 assesses a patient’s global impression of treatment tolerability with the question, “To what extent have you been troubled with side effects from your treatment?” Item 49 was included to complement the findings of the PRO-CTCAE.^23,41

Harms Outcomes

Safety and tolerability were assessed in terms of AEs, SAEs, deaths, physical examinations, laboratory findings, WHO/ECOG PS, vital signs, electrocardiograms, and exposure, which were collected for all patients. AEs were defined as an undesirable medical condition (other than the progression of the malignancy under evaluation) or the deterioration of a preexisting medical condition after or during exposure to a pharmaceutical product, whether or not it was considered causally related to the product. The severity of AEs was defined according to the National Cancer Institute Common Terminology Criteria for Adverse Events version 5.0.^23,48

Statistical Analysis

Sample Size and Power Calculation

Approximately 672 patients were to be randomized in a 1:1 ratio to either durvalumab plus gemcitabine and cisplatin or placebo plus gemcitabine and cisplatin in the global cohort. Approximately 130 patients from China were to be randomized. Once the global cohort was closed, sites in China continued recruitment until 130 patients were randomized.^23,41

The TOPAZ-1 trial was powered to demonstrate the superiority in OS of durvalumab plus gemcitabine and cisplatin over placebo plus gemcitabine and cisplatin in patients previously untreated for unresectable, locally advanced or metastatic BTC.^23,41

The hypothesis of improved OS was to be tested when:^23,41

• approximately 397 OS events had occurred across both groups (59% maturity) (IA-2), and

• approximately 496 OS events had occurred across both groups (74% maturity) (final analysis).

The primary analysis of OS was based on a log-rank test for the interim analysis and a Fleming-Harrington (FH) (0,1) test for the final analysis. The log-rank test was also performed at the final analysis as a sensitivity analysis.^23,41

Sample size calculations were performed under the assumption that the true average HR for OS is 0.745, corresponding to an approximate 4-month improvement in median OS over the control group of 11.7 months. Using a 2-sided log-rank test for the final analysis, approximately 496 OS events were estimated to provide 90% power to demonstrate statistical significance at the 2-sided level of 4.20%. The 4.9% (2-sided) alpha allocation for the OS analysis was controlled at IA-2 and final analysis using the Lan-DeMets spending function that approximates the O’Brien-Fleming approach, in which the significance level applied depends on the information fraction. The smallest treatment difference that could be statistically significant at the final analysis was an HR of 0.833 using log-rank test, or a 2.3-month improvement in median OS. With a planned 21-month nonlinear recruitment period, with accrual weight equal to 1.5 and a minimum follow-up period of 19 months assumed, it was anticipated that the final analysis will be performed approximately 40 months after the first patient is randomized. With a log-rank test at IA-2 and a FH(0,1) test at the final analysis, the overall power was at least 86%, based on an assumed average HR of 0.745 under the assumption of proportional hazards or up to a 6-month delayed effect (i.e., delayed separation of the OS curves by up to 6 months).^23,41

Interim and Final Analyses

Two interim analyses and 1 final analysis were planned for the evaluation of efficacy.^23,41

IA-1: The objective of IA-1 was to assess clinical activity with ORR and DOR to support early registration of durvalumab plus gemcitabine and cisplatin using both investigator assessments and BICR assessments according to RECIST 1.1. The minimum efficacy criterion for IA-1 was that the lower bound of the 2-sided exact 95% CI for ORR in the durvalumab plus gemcitabine and cisplatin arm was higher than the ORR point estimate in the placebo plus gemcitabine and cisplatin arm. The planned DCO for IA-1 was to occur when at least 200 patients (i.e., approximately 100 per arm) completed at least 32 weeks of follow-up. No formal comparison or statistical testing between arms was performed, although a nominal significance level of 0.001 was allocated to IA-1. Descriptive summaries of ORR, including a 2-sided exact Clopper-Pearson 95% CI, were presented by treatment group. Further descriptive summaries of ORR were produced by subgroups based on confirmed BICR assessments.^23,41

IA-1 was conducted in March 2021 and included 369 randomized patients who had had an opportunity to be followed up for at least 32 weeks. At the time of IA-1, the Independent Data Monitoring Committee determined that an acceptable safety profile was observed and the prespecified efficacy criteria for ORR had been met. The study blinding was maintained for all study personnel until full unblinding, at the time of IA-2.^23,41

IA-2: IA-2 tested for early superiority of durvalumab plus gemcitabine and cisplatin relative to placebo plus gemcitabine and cisplatin. This analysis was performed when approximately 80% of the final number of OS events was expected to be reached (approximately 397 of 496 OS events). The 2-sided alpha level allocated to OS was 0.049 and was controlled at the interim and final analyses, accounting for the correlation structure between the test statistics at IA-2 and the final analysis. The significance level for the primary confirmatory OS analysis at the final analysis was determined based on the actual alpha spending at IA-2 and the correlation structure between the IA-2 log-rank statistic and the final analysis FH(0,1) statistic. As a secondary evaluation, PFS was also analyzed at the time of OS for IA-2 and the final analysis, but only if superiority is confirmed with OS. The alpha level for PFS was controlled at the interim and final analysis using the Lan-DeMets spending function that approximates a Pocock approach. The significance levels for the PFS analyses using the log-rank test were calculated by specifying the information fraction for each analysis. The information fraction was calculated as the number of PFS events at the analysis time point divided by the total number of events at the final analysis time point.^23,41

Results from IA-2, a 6.5-month update for additional follow-up on OS, and safety outcomes are included in this report.

Statistical and Analytical Plans

Primary Efficacy End Point Analyses

For the primary end point of OS, the stratified log-rank test (adjusted for disease status and primary tumour location as entered into the IVRS or IWRS at randomization) was used for statistical inference, the stratified Cox proportional hazards model was used for quantification of the treatment effect under the assumption of proportional hazards with Efron ties, and the CI was calculated using the profile likelihood approach. Kaplan-Meier plots of OS were presented by treatment, and median OS and estimated OS rates at 12, 18, and 24 months were presented.²³

Because of the mechanism of action of immunotherapy, delayed separation of survival curves may be observed between groups in clinical trials, violating the proportional hazards assumption. The standard log-rank test, although optimal under proportional hazards, suffers substantial power loss in handling survival data with delayed separation of survival curves. At IA-2, the proportional hazards assumption was assessed by examining plots of complementary log-log (event times) versus log (time) and, if concerning, by fitting a time-dependent covariate (adding a treatment-by-time or treatment-by-ln[time] interaction term) to assess the extent to which this represents random variation. As a lack of proportionality was evident, the variation in treatment effect was described by presenting piecewise HR (Cox modelling) calculated over distinct time periods.²³

Sensitivity Analyses: No adjustment to the significance level for testing of the sensitivity analyses was made. Sensitivity analysis was conducted by examining censoring patterns to rule out attrition bias using a Kaplan-Meier plot of time to censoring, in which the censoring indicator of OS was reversed. The number of patients prematurely censored (i.e., the last known alive date is prior to DCO) was summarized by treatment group. Additionally, median durations of follow-up were provided for censored patients alive at DCO (i.e., time from randomization to date of censoring [i.e., date last known to be alive]) for each group.²³

Additional sensitivity analysis using Cox proportional hazards modelling, which contained treatment, the stratification factors, and the covariates of age, sex, race, ECOG PS, and BTC stage (locally advanced versus metastatic), were used to assess the effect of covariates on the HR estimate interactions between treatments, and stratification factors were also tested to rule out any qualitative interaction using the approach of Gail and Simon (1985).^23,49

Subgroup Analyses: Comparisons of efficacy between durvalumab plus gemcitabine and cisplatin and placebo plus gemcitabine and cisplatin were performed for the following prespecified subgroups of interest in the full analysis set (FAS):²³

• sex (male versus female)

• age at randomization (< 65 years versus ≥ 65 years)

• PD-L1 expression

• disease status (initially unresectable versus recurrent)

• primary tumour location (IHCC versus EHCC versus GBC)

• race (Asian versus non-Asian)

• region (Asia versus rest of the world)

• WHO/ECOG PS (0 versus 1 at screening)

• locally advanced versus metastatic BTC.

For these subgroup analyses, any patient with missing values was excluded from that particular subgroup. No adjustment to the significance level for testing of the subgroups was conducted, and all analyses were considered supportive of the analysis of OS and PFS. For each subgroup, the HR and 95% CI were calculated from a Cox proportional hazards model that only contained a term for treatment.²³

To assess the consistency of the treatment effect between subgroups for plausible subgroups, the presence of quantitative interactions between treatment and stratification factors was checked using an overall global interaction test (at the 2-sided 10% significance level). Where statistically significant, an attempt to determine the cause and type of interaction was made using stepwise backward selection on the saturated model until a final model was reached in which all included interactions were significant and all excluded interactions were nonsignificant. All main effects were included in the model, regardless of whether the corresponding interaction term was still present to identify factors that independently alter the treatment effect and prevent identification of multiple correlated interactions. Quantitative interactions identified using this procedure were then tested to rule out any qualitative interaction using the approach of Gail and Simon (1985).^23,49

Secondary Efficacy End Point Analyses

The analysis of the secondary efficacy end points of PFS, ORR, DOR, and DCR were based on investigator assessments using RECIST 1.1.²³

Progression-Free Survival: The analysis of the key secondary end point of PFS was based on RECIST 1.1, per investigator assessments, using a methodology similar to that described for OS, including estimated PFS rates at 6, 9, 12, and 24 months, using the Kaplan-Meier curves. The assumption of proportional hazards was examined, and other sensitivity analyses were performed analogous to OS. An additional sensitivity analysis of PFS was performed to assess possible evaluation-time bias that could be introduced if scans were not performed at the protocol-scheduled time points. The midpoint between the time of progression and the previous evaluable RECIST 1.1 assessment was analyzed using a stratified log-rank test, per the main PFS analysis. For patients whose death was treated as a PFS event, the date of death was used to derive the PFS time used in the analysis.²³

Attrition bias was assessed by repeating the PFS analysis, except that the actual PFS event times, rather than the censored times, of patients who progressed or died in the absence of progression immediately after 2 or more nonevaluable tumour assessments were included. In addition, patients receiving subsequent therapy before progression or death were censored at their last evaluable assessment before receiving the subsequent therapy; this was further supported by a Kaplan-Meier plot of the time to censoring, in which the censoring indicator of the PFS analysis was reversed. Cox proportional hazards modelling (conducted analogous to OS) was used to assess the effect of covariates on the HR estimate. Subgroup analyses comparing PFS (per RECIST 1.1 using investigator assessments) were conducted analogous to OS.²³

Objective Response Rate: ORR was the primary end point for IA-1. The ORR, using the BICR data (at IA-1 only) to define a visit response of CR or PR, was analyzed for the FAS at 32 weeks. A comparison of confirmed best overall response (BOR) for BICR versus investigator assessment was provided.²³

At IA-2, the ORR was compared between treatment groups using a stratified Cochran-Mantel Haenszel test as the primary test. That test used the same stratification factors as for OS, and results were presented in terms of OR and P value.²³

As a sensitivity analysis, ORR was analyzed using logistic regression models (adjusted for the same stratification factors as for OS) and presented in terms of OR (OR greater than 1 favouring treatment with durvalumab plus gemcitabine and cisplatin) together with its associated profile likelihood 95% CI and P value. The ORR analysis was performed in the subset of patients in the FAS who had measurable disease at baseline, with a sensitivity analysis based on the complete FAS. Additionally, for each treatment group, BOR was summarized by n (%) for each category (CR, PR, stable disease, PD, no evidence of disease, and not evaluable).²³

Subgroup analyses for ORR were performed using the subgroups specified for OS.²³

Duration of Response: Kaplan-Meier plots of DOR (per investigator) were presented, along with median DOR. In addition, the number of censored responders (split by lost to follow-up and still in response), number (%) of patients with a DOR of at least 3 months, at least 6 months, at least 9 months, and at least 12 months, and the time to onset of response from randomization were summarized.²³

Further descriptive summaries of DOR were produced by subgroups (disease status, primary tumour location, and region). Swimmer plots that show the profile of each responder were also produced. Subgroup analyses were performed based on geographic region, primary tumour location, and disease status.²³

Health-Related Quality of Life: The PRO analyses for EORTC QLQ-C30 and EORTC QLQ-BIL21 included time to deterioration (using a stratified log-rank test that provided a P value and a stratified Cox proportional hazard model that provided HR and 95% CI), adjusted mean change from baseline score (using a mixed-effect model for repeated measures that provided an estimate of the treatment difference, 95% CI, and P value), and BOR (using a logistic regression model that provided OR, 95% CI, and P value). Visit responses (improvement, no change, and deterioration) and visit-specific and overall scores (across all visits) were summarized descriptively. Deterioration and improvement were defined as an increase or decrease of at least 10 points, respectively, on symptom scales and items. Changes of less than 10 points were considered to be no change.²³

For the EQ-5D-5L, descriptive statistics (e.g., n, mean, median, standard deviation, minimum, and maximum) were reported for the EQ-5D index score and the EQ-VAS score, as were the change from baseline for the EQ-5D index score and the EQ-VAS score.²³

Symptom Severity (PGI-S): Data for the PGI-S were summarized descriptively as the number of patients and corresponding percentages for each category in the questionnaire at each visit by treatment group.²³

Treatment Tolerability (PRO-CTCAE, QLQ-BIL21 Item 49): Data for the PRO-CTCAE and EORTC QLQ-BIL21 were summarized descriptively.²³

Safety Analyses

Safety data were not formally analyzed but were summarized descriptively according to the treatment received.²³

Multiple Testing

The objective of IA-1 was to evaluate the efficacy of durvalumab plus gemcitabine and cisplatin in terms of clinical activity based on all patients in the global cohort randomized at least 32 weeks before the IA-1 DCO (December 18, 2020). A small alpha expenditure of 0.001 was allocated to IA-1 for ORR.^23,41

Multiplicity adjustment was incorporated in IA-2 and the final analysis of the primary end point (OS) and the selected key secondary end point (PFS). For the planned OS analysis, control of the family-wise error rate at the remaining 4.9% level (2-sided) was achieved with a combined approach of alpha allocation to the OS analyses (IA-2 and the FA) using the O’Brien-Fleming alpha spending function and a hierarchical testing procedure; that is, PFS was to be tested only if OS met statistical significance at IA-2 or the final analysis.^23,41

The IA-2 OS analysis was conducted when 424 of the 496 expected OS events at the final analysis (61.9% overall maturity for OS) had occurred, using the Lan-DeMets spending function, approximating O’Brien-Fleming boundaries, to ensure strong control of the type I error (with a 2-sided interim P value of < 0.0300) for the treatment comparison. As OS met statistical significance at IA-2, significance levels for PFS at IA-2 for the log-rank test were derived based on the Lan-DeMets alpha spending function, approximating Pocock boundaries, which strongly controls the type I error at the 0.049 level (2-sided).^23,41

Analysis Populations

The following analysis populations of interest were defined in the TOPAZ-1 trial:²³

• FAS, which consisted of all randomized patients and was analyzed on an intention-to-treat basis. The FAS was used for all efficacy analyses.

• Safety analysis set, which consisted of all patients who received at least 1 dose of the study treatment. Patients were analyzed according to the treatment they received.

• PRO analysis set, which was defined as all patients from the FAS except those with no questionnaire translation available or those who did not complete the questionnaire because of other physical or language reasons (included with statistical analysis plan (SAP) amendment on June 7, 2021).

Protocol Amendments, Deviations, and Changes to Planned Analyses

Protocol Amendments

The original clinical study protocol was finalized on December 19, 2018. There were subsequently 6 global protocol amendments, for a total of 7 versions. Changes to the study protocol after the initiation of patient recruitment are summarized in Table 8. It was unclear how many patients were enrolled at each protocol amendment; therefore, the impact of potential biases due to protocol amendments remains unknown. According to the investigators, none of the protocol amendments affected the conduct or integrity of the study, nor did it affect the patients’ safety.^23,41

Table 8: Important Amendments to the Original Study Protocol After the Start of Patient Recruitment

ALT = alanine aminotransferase; AST = aspartate aminotransferase; CMH = Cochran-Mantel Haenszel; CSR = Clinical Study Report; ctDNA = circulating tumour DNA; DCO = data cut-off; FA = final analysis; FH = Fleming-Harrington; HR = hazard ratio; IA-1 = first interim analysis; IA-2 = second interim analysis; MSI = microsatellite instability; ORR = objective response rate; OS = overall survival; PD = progressive disease; RECIST 1.1 = Response Evaluation Criteria in Solid Tumours Version 1.1; TMB = tumour mutation burden.

Sources: TOPAZ-1 Protocol and CSR (IA-2).^23,41

Protocol Deviations

The number of patients with important protocol deviations in each treatment group are summarized in Table 9. █████ █ ██ ██████████ ███████ █████████ █████████ ████ ██████████ ██ ████ ██ ██████ ██████████ ████████ ███ █ ██ ███ ███████ ███████ █████████ █████████ ████ ██████████ ██ ████ ██ ██████ ████████ ██ ███ ███ █████ █ ██ █████ █████ ███████ ██████ ███ █████████ ████ ██████████ ██ ████ ██ ██████ ██████ ████ ████ ███ ██ ██ █ ██ ██████████ ███████ █████████ █████████ ████ ██████████ ██ ████ ██ ██████ ██████████ ██ ███ ██ ████ █ ██ ██████ ████ ███████ █████████ █████████ ████ ██████████ ██ ████ ██ ██████ ████ ██████ ██ █ ████ ████ █ ██ ██████████ ███████ █████████ █████████ ████ ██████████ ██ ████ ██ ██████ █████ █████ ██ █ ███ █████ ██████ █████ ███████ █████████ ████████

Table 9: Important Protocol Deviations (FAS)

Cis = cisplatin; D = durvalumab; FAS = full analysis set; Gem = gemcitabine; PBO = placebo; RECIST 1.1 = Response Evaluation Criteria in Solid Tumours Version 1.1.

█████ █ ██ ██████████ ███████ █████████ █████████ ████ ██████████ ██ ████ ██ ██████ ██████████ █████ ██ ████ █ ██ ██████████ ███████ █████████ █████████ ████ ██████████ ██ ████ ██ ██████ ██████████ ███ ██ ██████ █ ██ ██████████ ███████ █████████ █████████ ████ ██████████ ██ ████ ██ ██████ ██████████ ██ ███ █ ██ ██████████ ███████ █████████ █████████ ████

Source: TOPAZ-1 CSR (IA-2).²³

Changes to Planned Analyses

Changes to the SAP between August 2020 to October 2021 (SAP Amendments version 2 to version 6) were implemented to align with the clinical study protocol. Deviations from the planned analysis in the protocol included the following:²³

• all PRO analyses were performed on the PRO analysis set and not the FAS, as stated in protocol

• no subgroup analysis was performed for DCR

• no sensitivity analysis was carried out for ORR or DOR for patients with 32 weeks of follow-up as assessed by BICR, as BICR data were only collected until IA-1.

After unblinding of the study data, the following posthoc analyses were conducted as supplemental analyses.^23,41

• For OS:

  • The plot of kernel-smoothed estimates of the hazard functions versus time to show the departure from proportional hazards.

  • OS by confirmed ORR (RECIST 1.1, per investigator).

  • A forest plot of HR by additional PD-L1 tumour and/or immune cell positivity cut-offs (5% and 10%) was produced.

• Forest plot of HR for ORR by subgroup to visually present this prespecified subgroup analysis.

• A waterfall plot for DOR.

• Demographic and disease characteristics (including stratification factors) presented by region (Asia versus rest of the world).

• Time to treatment discontinuation for patients with any causally related AE leading to discontinuation of any study treatment.

• The potential for hepatitis reactivation was evaluated by shifts in viral load for patients with hepatitis B or C.

• Additional by-patient data to provide clarity around certain cases.

Results

Patient Disposition

TOPAZ-1 was a randomized, phase III clinical trial. Table 10 summarizes the disposition of enrolled patients at IA-2 (DCO of August 11, 2021) and as of the 6.5-month update (DCO of February 25, 2022). A total of 914 patients were screened for eligibility and 685 were randomized to receive durvalumab plus gemcitabine and cisplatin (n = 341) or placebo plus gemcitabine and cisplatin (n = 344).²³

With 6.5 months of additional follow-up, the proportion of patients who discontinued treatment was slightly higher than at IA-2. At the 6.5-month update (DCO of February 25, 2022), ██████████████ patients discontinued treatment in the durvalumab plus gemcitabine and cisplatin group, as did ██████████████ patients in the placebo plus gemcitabine and cisplatin group. █████ █ ██ ██████████ ███████ ██ ███████ █████████ ████ ██████████ ██ ████ ██ ██████ ██ ████████ ████ ███████ █ ██ █ █████████ ███████ █████████ █████████ ████ ██████████ ██ ████ ██ ██████ ██ ████ ████ ██████ █████ █ ██ ███ ███████ ███████ █████████ ████ █████ ████ ██████████ ██ ████ ██ ██████ ██████████ █████ █ ██ ██████████ ███████ █████████ █████████ ███████ ██████ ███████████²⁴

Table 10: Patient Disposition

Cis = cisplatin; D = durvalumab; DCO = data cut-off; FAS = full analysis set; Gem = gemcitabine; IA-2 = second interim analysis; N/A = not applicable; PBO = placebo; RECIST = Response Evaluation Criteria in Solid Tumours; SAS = safety analysis set.

Sources: TOPAZ-1 CSR (IA-2 and 6.5-month update).^23,24

Exposure to Study Treatments

Exposure to study treatments at IA-2 and the 6.5-month update is summarized in Table 11. At the 6.5-month update, the median duration of treatment was ████████ █████████████ ██████████████ █████ █████ with durvalumab plus gemcitabine and cisplatin and ████████████ ██████ ████ ███ ██████████ ██████████ with placebo plus gemcitabine and cisplatin. The difference in duration of treatment was due to the monotherapy period for durvalumab after 8 cycles of chemotherapy.²⁴

Table 11: Duration of Durvalumab or Placebo Exposure (Safety Analysis Set)

Cis = cisplatin; D = durvalumab; DCO = data cut-off; Gem = gemcitabine; PBO = placebo.

^aTotal treatment duration = (min [last dose date where dose > 0 plus {20 if last dose in period 1 (combination) or 27 if last dose in period 2 (maintenance)}, date of death, date of DCO] – first dose date plus 1)/(365.25/12).

^bActual treatment duration = total treatment duration minus the total duration of delays.

^cTotal treatment duration = (min [last dose date period 1 where dose > 0 plus 20, date of death, date of DCO] – first dose date plus 1)/(365.25/12).

^dActual treatment duration = total treatment duration minus the total duration of delays.

^eTotal treatment duration = (min [last dose date period 2 where dose > 0 plus 27, date of death, date of DCO] – first dose date plus 1)/(365.25/12).

Sources: TOPAZ-1 CSR (IA-2 and 6.5-month update).^23,24

The median number of treatment cycles for each treatment is summarized in Table 12. The median number of cycles for durvalumab was 10 (range, 1 to 29) and the median number of cycles for placebo was 8 (range, 1 to 26). A total of 60.1% and 59.2% of patients in the durvalumab plus gemcitabine and cisplatin group received at least 8 cycles of gemcitabine and cisplatin, respectively. Conversely, 52.0% and 50.0% of patients in the placebo plus gemcitabine and cisplatin group received at least 8 cycles of gemcitabine and cisplatin, respectively. For durvalumab, 30.8% of patients received 12 or more treatment cycles, whereas 19.6% of patients received 12 or more treatment cycles of placebo.²³

The median dose intensity for both treatment groups were identical. The median dose intensity for durvalumab and placebo was 100%, whereas the median dose intensity for both gemcitabine and cisplatin was 93.8%.²³

Table 12: Number of Treatment Cycles Received and Relative Dose Intensity (IA-2, Safety Analysis Set)

IA-2 = second interim analysis; N/A = not available; SD = standard deviation.

Note: Relative dose intensity is the percentage of the actual dose intensity delivered relative to the intended dose intensity through treatment discontinuation.

Source: TOPAZ-1 CSR (IA-2).²³

Dose Modifications

Results for dose modifications, including dose delays, infusion interruptions, and dose reductions, as of IA-2 are summarized in Table 13. No results for dose modifications were available at the 6.5-month update █████ █ ██ ██████████ ███████ ██ ██ █████ █████████ ████ ██████████ ██ ████ ██ ██████ ██████████ ███████████ █ ██ ██████████ ███████ ██ ██ █████ █████████ ████ ██████████ ██ ████ ██ ██████ ██████████ ███████████ █ ██ ██████████ ███████ █ █ ███████ █████████ ████ ██████████ ██ ████ ██ ██████ ██████████ █████ █ ██ ██████████ ███████ █ ███ █████ █████████ ██████████████. Dose reductions were not permitted for durvalumab or placebo. █████ █ ██ ███ ███ ████ ███████ █████████ █████████ ████ ██████████ ██ ████ ██ ██████ ██████████ ███████████ █ ██ ███ ██ ████ █ ███████ █████████ █████████ ████ ██████████ ██ ████ ██ ██████ ██████████ █████ ██████ █ ██ ██ █ ███ ████ ███████ █████████ █████ ████ ████ ██████████ ██ ████ ██ ██████ ██████████ ███████████ █ ██ ██ ███ █████ ███████ █████████ █████████ ████ ██████████ ██ ████ ██ ██████ ██████████ ██ ███ ███ ███ █ ██ ██████ ████ ███████ █████████ █████████ ████ ██████████ ██ ████ ██ ██████ ██████████ ██ ██ ███████ █ ██ ████ ██████ ███████ █████████ █████████ ████ ██████████ ██ ████ ██ ██████ ██████████ ███ ██ ██████ █ ██ █████ █████ ███████ █████████ █████████ ████ ██████████ ██ ████ ██ ██████ ██████████ ██ ████ █████ █ ██ ██████████ ███████ █████████ █████████ ████ ██████████ ██ ████ ██ ██████ ██████████ ███ ██ ██████ █ ██ ██████████ ███████ █████████ █████████ ████ ██████████ ██ ████ ██ ██████ ██ ███ █████ ███████████ █ ██ ██████████ ███████ █████████ █████████ ████ ██████████ ██ ████ ██ ██████ ██ █ ██ ███ ██ █████ █ ██ ██████████ ███████ ████ █████ █████████ ███ ███████ ███████ ███████ ███████ █ █ ███ ████ █████

Table 13: TOPAZ-1 Dose Delays, Interruptions, Reductions (IA-2, Safety Analysis Set)

AE = adverse event; NA = not applicable; NR = not reported.

Source: TOPAZ-1 CSR (IA-2).²³

Subsequent Anticancer Therapy

Results for subsequent anticancer therapy received at IA-2 and as of the 6.5-month follow-up (DCO of February 25, 2022) are summarized in Table 14. █████ █ ██ ██████████ ███████ █████████ █████████ ████ ██████████ ██ ████ ██ ██ ████ ██████████ ███████████ █ ██ ██████████ ███████ █████████ █████████ ████ ██████████ ██ ████ ██ ███ ███ █████ █████ ██████ █████ █ ██ ██████████ ███████ █████████ █████████ ████ ██████████ ██ ████ ██ ██████ ███████ ███ ████████ ██████ █████ ███ ██████ ████ █████████████²⁴

Although not considered subsequent anticancer therapy, █████ █ ██ ██████████ ███████ █████████ █████████ ████ █ ██ ███████ ██ ████ ██ ██████ ██ ████████ ███████████ █ ██ ██████████ ███ ████ █████████ █████████ █ ███ ██ █████ ███ ██ ████ ██ ██████ ██████████ ███████████ █ ██ ████ ██████ ███████ █████████ █████████ █ ███ ██ ████████ ██ ████ ██ ██████ ██████████ ███████████ █ ██ ██████████ ███████ █████████ █ ███ █████ █ ███ █ █████████ ██ ████ ██ ██████ ████ ██████ ███████████ █ ██ ███ ███████ ███████ █████████ ███ ██ ████ ████ █████ █████ ██ ████ ██ ██████ ██████████ █████████████████████████████████████████ patients in the durvalumab plus gemcitabine and cisplatin and placebo plus gemcitabine and cisplatin groups, respectively.²³

Table 14: Subsequent Anticancer Therapy (FAS)

Cis = cisplatin; D = durvalumab; DCO = data cut-off; FAS = dull analysis set; Gem = gemcitabine; IA-2 = second interim analysis; PBO = placebo.

Note: Therapies post-discontinuation of study treatment were displayed. A patient could have more than 1 cancer therapy.

Sources: TOPAZ-1 CSR (IA-2 and 6.5-month update).^23,24

Efficacy

Only efficacy outcomes and analyses of subgroups identified in the review protocol are reported here. Refer to Appendix 3 for detailed efficacy data.

Overall Survival

OS was the primary end point of IA-2 of the TOPAZ-1 study. Results for OS at IA-2 and 6.5-month additional follow-up are summarized in Table 15 and by Kaplan-Meier plot in Figure 3.

At the preplanned IA-2 (DCO of August 11, 2021), a total of 424 death events had occurred (198 [58.1%] in the durvalumab plus gemcitabine and cisplatin group, and 226 [65.7%] in the placebo plus gemcitabine and cisplatin group), representing ████ ███ ███ ███ ███████ ██████ ██████████. The median duration of follow-up in censored patients was ███ ██████ ██████ ███████ ██ █████ ███████ in the durvalumab plus gemcitabine and cisplatin group, and ██ █████ ██████ ███ █████ ███ █████ █ █████ ███ █████ █████ in the placebo plus gemcitabine and cisplatin group. The HR for the durvalumab plus gemcitabine and cisplatin versus placebo plus gemcitabine and cisplatin was 0.80 (95% CI, 0.66 to 0.97; P = 0.021) in favour of durvalumab plus gemcitabine and cisplatin, with a median OS of 12.8 months (95% CI, 11.1 to 14.0 months) for durvalumab plus gemcitabine and cisplatin and 11.5 months (95% CI, 10.1 to 12.5 months) for placebo plus gemcitabine and cisplatin.

Table 15: Overall Survival (FAS)

CI = confidence interval; Cis = cisplatin; D = durvalumab; DCO = data cut-off; FAS = full analysis set; Gem = gemcitabine; HR = hazard ratio; IA-2 = second interim analysis 2; OS = overall survival.

^aIncludes patients known to be alive at DCO.

^bIncludes patients with unknown survival status, patients who were lost to follow-up, or patients with “other” recorded on the case report form.

^cAt IA-2, the investigator made a transcription error that was clarified at the 6.5-month update analysis. At IA-2, the last known date that patient was known to be alive was erroneously reported as June, 29 2021. The last known date that the patient was known to be alive is July 24, 2020.

^dCalculated using the Kaplan-Meier technique. CI for median OS derived using the Brookmeyer-Crowley method.

^eThe analysis was performed using a stratified Cox proportional hazards model (ties = Efron), adjusted for disease status and primary tumour location. The CI was calculated using a profile likelihood approach. HRs < 1 favour durvalumab, and is associated with a longer OS than placebo.

^fThe primary end point of OS was controlled for multiple testing using the Lan-DeMets spending function, approximating O’Brien-Fleming boundaries. The 2-sided significance level was < 0.0300.

Sources: TOPAZ-1 CSR (IA-2 and 6.5-month update).^23,24

At the 6.5-month update (DCO of February 25, 2022), a total of ███████ death events had occurred (248 [72.7%] in the durvalumab plus gemcitabine and cisplatin group, and 279 [81.1%] in the placebo plus gemcitabine and cisplatin group), representing 76.9% overall maturitsy for OS. █████ █ ██ ██████████ ███████ █████████ █████████ ████ ██████████ ██ ████ ██ ██████ ██████████ ███████████ █ ██ ██████████ ███████ █████████ █████████ ████ █ ███ █ █████ ██ ████ ██ ██████ ██████████ █████ █ ██ ██████████ ███████ █████████ █████████ ██ ██ ██ █ █ ████ █████ █████████████ ██████████. With 6.5 months of additional follow-up, the HR for the durvalumab plus gemcitabine and cisplatin versus placebo plus gemcitabine and cisplatin was 0.76 (95% CI, 0.64 to 0.91) in favour of durvalumab plus gemcitabine and cisplatin, with a median OS of 12.9 months (95% CI, 11.6 to 14.1 months) for durvalumab plus gemcitabine and cisplatin and 11.3 months (95% CI, 10.1 to 12.5 months) for placebo plus gemcitabine and cisplatin.

Results for the OS rate were consistent at IA-2 and at the 6.5-month additional follow-up. The OS rate at 18 months was 34.8% (95% CI, 29.6% to 40.0%) for durvalumab plus gemcitabine and cisplatin and 24.1% (95% CI, 19.6% to 28.9%) for placebo plus gemcitabine and cisplatin, and at 24 months was 23.6% (95% CI, 18.7% to 28.9%) and 11.5% (95% CI, 7.6% to 16.2%), respectively.^23,24

Posthoc Analyses of the Primary End Point

A posthoc analysis was conducted confirming the nonproportionality of the hazards assumption, given that delayed separation of Kaplan-Meier curves is common in clinical trials of immunotherapy in combination with chemotherapy. Results of the posthoc analysis were consistent with the primary analysis, confirming nonproportional hazards, and the piecewise HR was █████ █ ██ ███ ███████ ███████ █████████ █████████ ████ ██████████ ██ ████ ██ ██████ ██████████ ███████████ █ ██ █ ██████ ███ ███████ █████████ █████████ ████ ██████████ ██ ████ ██ ██████ ████ ██████ █████████ ████ █████ ██ ██████^23,24 An additional posthoc analysis analyzing OS by response status was conducted, given that the OS benefit for patients receiving durvalumab plus gemcitabine and cisplatin was driven not only by responders but also by patients with a BOR of stable disease. Results for this posthoc analysis are summarized in Figure 4 of Appendix 2. █████ █ ██ ██████████ ███████ █████ ████ █████████ ████ ██████████ ██ ████ ██ ██████ ██████████ ███████████ █ ██ ████████ ██ ████ █ ██ ██████ ███ ███ ██████ ████ ██████████ ██ ████ ██ ██████ ███ ███████ ███████████ █ ██ ████████ ██ █ ███ ███ █████████ █████████ ████ ██████████ ██ ████ ██ ██████ ██████████ ███████████ █ ██ ███ ███████ ██ █████ █████████ ███ ██████ ████ ████ ██████ ██ ████ ██ ██████ ██████████ █████ ██████ █ ██ ██ █████ ███ ███████ █████████ █████████ ████ ██████████ ██ ████ ██ ██████ ██████████ ███████████ █ ██ ██████ ████ ███████ █████████ █████████ ████ ██████████ ██ ████ ██ ██████ ██████████ █████ █ ██ ██████ ████ ███████ █████████ █████████ ███████████ █████████

Sensitivity Analyses

Results for the prespecified sensitivity analysis examining censoring patterns to identify potential attrition bias by reversing the censoring indicator of OS suggested that the HR for OS was ██████ █████████ ███████ ███████ ████████ █████████ ██. The sponsor considered this supportive of the primary analysis.²³

An additional sensitivity analysis of OS HR estimates adjusted for the potential effect of covariates using Cox proportional hazards modelling was similar to that obtained without such adjustment █████ █ ██ ███████ ███ ███████ █████████ █████████ ██ ███████ █████.²³

Results of the sensitivity analyses at the 6.5-month follow-up were consistent with the results at IA-2.²⁴

Subgroup Analyses

Results for prespecified subgroups in the subgroups of interest to this review at IA-2 and after 6.5 months of additional follow-up are summarized in Table 16. The full subgroup analysis from the TOPAZ-1 trial is summarized in Figure 4 of Appendix 2. At IA-2, the OS was generally consistent across subgroups. Results of the subgroup analyses of interest after 6.5 months of additional follow-up were generally consistent with the IA-2 subgroups, as well as with the primary analysis.²⁴

Progression-Free Survival

PFS was a key secondary end point of the TOPAZ-1 study, and results are summarized in Table 17 and by Kaplan-Meier plot for IA-2 in Figure 5. No results for PFS were available at the 6.5-month update. At the time of the IA-2 DCO (August 11, 2021), a total of ███ █████████ PFS events had occurred; ██ ██████ ████ in the durvalumab plus gemcitabine and cisplatin group, and ████ █████ █ ██ in the placebo plus gemcitabine and cisplatin group ██████ █████████ ████ ██████████ ██ █████. The HR for durvalumab plus gemcitabine and cisplatin versus placebo gemcitabine and cisplatin was 0.75 (95% CI, 0.63 to 0.89; P = 0.001) in favour of durvalumab. The median PFS was 7.2 months (95% CI, 6.7 to 7.4 months) in the durvalumab plus gemcitabine and cisplatin group, and 5.7 months (95% CI, 5.6 to 6.7 months) in the placebo plus gemcitabine and cisplatin group. The PFS rate at 6, 9, and 12 months was █████ █ ██ █████ █████ ███████ █████████ █████████ ███████████ █████████ ███ ███ for the durvalumab plus gemcitabine and cisplatin group, and █████ █ ██ ██████████ ███████ ████ █████ █████████ ██ ██ ████ ████████████ ███████ ███████ ████ in the placebo plus gemcitabine and cisplatin group, respectively.²³

Table 16: OS by Subgroup of Interest (FAS)

CI = confidence interval; Cis = cisplatin; D = durvalumab; DCO = data cut-off; ECOG PS = Eastern Cooperative Oncology Group Performance Status; EHCC = extrahepatic cholangiocarcinoma; FAS = full analysis set; GBC = gallbladder cancer; Gem = gemcitabine; HR = hazard ratio; IA-2 = second interim analysis; IHCC = intrahepatic cholangiocarcinoma; PBO = placebo; PD-L1 = programmed cell death 1 ligand 1; TIP = tumour and/or immune cell positivity; WHO = WHO.

Note: HRs < 1 favour durvalumab plus gemcitabine plus cisplatin.

Sources: TOPAZ-1 CSR (IA-2 and 6.5-month update).^23,24

Table 17: PFS by Investigator Assessment According to RECIST 1.1 (FAS)

CI = confidence interval; Cis = cisplatin; D = durvalumab; DCO = data cut-off; FAS = full analysis set; Gem = gemcitabine; HR = hazard ratio; IA-2 = second interim analysis; PBO = placebo; PFS = progression-free survival; RECIST = Response Evaluation Criteria in Solid Tumours.

^aOnly includes progression events that occurred within 2 missed visits of the last evaluable assessment.

^bTarget lesions, nontarget lesions, and new lesions are not necessarily mutually exclusive categories.

^cRECIST 1.1 progression event occurred 2 or more consecutive missed visits after last evaluable RECIST 1.1 assessment.

^dDeath occurred after 2 or more consecutive missed visits after last evaluable RECIST 1.1 assessment.

^eCalculated using the Kaplan-Meier technique.

^fThe HR and its CI were estimated using a stratified Cox proportional hazards model (ties = Efron) adjusted for disease status and primary tumour location.

^gThe P value is based on a stratified log-rank test and tested at the 0.0481 significance level. HRs < 1 favour durvalumab plus gemcitabine plus cisplatin, and are associated with a longer PFS than placebo plus gemcitabine plus cisplatin. PFS was included in the multiple-testing procedure and was formally tested after a statistically significant result for the primary outcome, OS.

Source: TOPAZ-1 CSR (IA-2).²³

Sensitivity Analyses

Sensitivity analyses of the key secondary end point of PFS to evaluate the potential for attrition bias and the impact of covariates on the treatment effect were consistent with the primary analysis. █████ █ ██ ████████ ██ ███████ █████████ █████ ████ █ ███ █████ █████ ██ ████ ██ ██████ ████ ██████ ███████████ █ ██ ██████████ ███████ ███ ██████ ██████ ███ ████ ███ ██ █████ ██ ████ ██ ██████ ████ ██████ █████ █ ██ ██████████ ███████ █████████ ███████ ██ ████.

██ ███ █ ██ ██████████ ███████ █████████ █████████ ████ ██████████ ██ ████ ██ ██████ ██████████ ██ ███ ██████ █ ██ ████████ ██ ███████ █████████ █████████ ████ ██████████ ██ ████ ██ ██████ ██████████ ███ ███ █████ █ ██ █████ █████ ███████ ██ ████ ███ █████████ ████ ███████ ███ ██ ████ ██ ██████ ██ ███ █████ ██████

Subgroup Analyses

Results of subgroup analyses of PFS from the TOPAZ-1 trial are summarized in Figure 6. For subgroups of interest to this review, results appeared consistent with the primary analysis, although no HRs were presented.²³

Clinical Response

Objective Response Rate

ORR was the primary outcome at IA-1 and was a secondary outcome at IA-2. Results for ORR at both IA-1 and IA-2 are summarized in Table 18. No results for ORR were available at the 6.5-month update. Results at IA-1 were assessed by the investigator and by BICR. █████ █ ██ ██████████ ███████ █████████ █████████ ████ ██████ ████ ██ ████ ██ ██ ██ ██ ██████ ████ ███████ ████ █ ██ ████ ██████ ███████ █████████ █████████ ████ ██████ ████ ██ ████ ██ ██ ████ ███ ██ █████ █████ █ ██ ██████████ ███████ █████████ █████████ ██████ ████████████ ██ ██████ ████ ███ ███████.

Table 18: ORR — Stratified CMH Test (FAS Patients With Measurable Disease at Baseline)

CI = confidence interval; Cis = cisplatin; CMH = Cochran-Mantel Haenszel; D = durvalumab; DCO = data cut-off; DCR = disease control rate; FAS 32w = full analysis set at 32 weeks; Gem = gemcitabine; IA-2 = interim analysis 2; OR = odds ratio; ORR = objective response rate; PBO = placebo.

^aResponses included confirmed complete or partial response, per investigator, according to RECIST 1.1. They do not include patients who discontinued randomized treatment without progression or received a subsequent anticancer therapy and then responded.

^bORs > 1 favour durvalumab plus gemcitabine plus cisplatin.

^cNominal 2-sided P value was not adjusted for multiple testing.

Source: TOPAZ-1 CSR (IA-2).²³

At IA-2, per the investigator, 91 (26.7%) patients in the durvalumab plus gemcitabine and cisplatin group and 64 (18.7%) in the placebo plus gemcitabine and cisplatin group achieved an ORR. The OR of achieving a response with durvalumab plus gemcitabine and cisplatin versus placebo plus gemcitabine and cisplatin was 1.60 (95% CI, 1.11 to 2.31; P = 0.011) in favour of durvalumab plus gemcitabine and cisplatin. The median time to response was 1.6 months for durvalumab plus gemcitabine and cisplatin and 2.7 months for placebo plus gemcitabine and cisplatin.²³

DCR (Table 18) was a secondary outcome of the TOPAZ-1 trial. At IA-2, the DCR in the FAS was 85.3% in the durvalumab plus gemcitabine and cisplatin group and 82.6% in the placebo plus gemcitabine and cisplatin group.²³

Sensitivity Analysis: █████ █ ██ ██████████ ███████ ████ █████ █████████ ████ ██████████ ██ ████ ██ ███ ███ ████ ██████ ███████████ █ ██ ██████████ ███████ █████████ █████████ ████ ████ ██████ ██ ████ ██ ██████ ██ ████ ████ █████ █ ██ ██████████ ███████ █████████ █████████ ██████████████ ██████ ████████.

Subgroup Analysis: A post-hoc subgroup analysis of ORR at IA-2 was conducted. Results of the subgroup analyses are summarized in Figure 7. █████ █ ██ ██████████ ███████ █████████ █████████ ████ ██████████ ██ ████ ██ ██ █ ███ ██ ██ ██████ ███████████ █ ██ ██████████ ███████ █████████ █████████ ████ ██████████ ██ ████ ██ ██ ████ █████ █████ █████ █ ██ ██████████ ███████ █████████ █████████ █████████ █████ ██████████████.

Duration of Response

DOR in responders was a secondary outcome of the TOPAZ-1 trial. Results for DOR at IA-2 are summarized in Table 19. No results for DOR were available at the 6.5-month update. More confirmed responses were observed in the durvalumab plus gemcitabine and cisplatin group (91 [26.7%] patients) than in the placebo plus gemcitabine and cisplatin group (64 [18.7%] patients). The median DOR was 6.4 months (95% CI, 5.9 to 8.1 months) with durvalumab plus gemcitabine and cisplatin and 6.2 months (95% CI, 4.4 to 7.3 months) with placebo plus gemcitabine and cisplatin. Most patients in the durvalumab plus gemcitabine and cisplatin group and the placebo plus gemcitabine and cisplatin group subsequently progressed or died (65.9% versus 79.7%).²³

The percent of patients remaining in response at 9 and 12 months was 32.6% and 26.1%, respectively, for durvalumab plus gemcitabine and cisplatin and 25.3% and 15.0%, respectively, for placebo plus gemcitabine and cisplatin.²³

Table 19: Duration of Objective Response per Investigator (FAS; Patients With Objective Response and Measurable Disease at Baseline)

CI = confidence interval; Cis = cisplatin; D = durvalumab; DCO = data cut-off; DOR = duration of response; FAS = full analysis set; Gem = gemcitabine; IA-2 = second interim analysis.

^aOne patient received surgical resection for BTC and remained on study treatment after surgery. The site continued scans (each resulting in RECIST not evaluable). Per the SAP, DOR was censored at patient’s last evaluable RECIST 1.1 assessment.

^bFor 2 patients, the ongoing responses were apparently maintained with subsequent anticancer therapy. Both patients discontinued treatment due to subjective disease progression. Per PFS censoring methodology, DOR was censored at patient’s last evaluable RECIST 1.1 assessment for both cases.

^cOne patient responded to treatment but then withdrew consent for study participation, without RECIST progressive disease, and later died. Due to withdrawal of consent, follow-up scans were not performed. In the absence of at least 2 or more missed visits, the patient was censored at the time of the latest evaluable RECIST 1.1 assessment.

^dDOR is the time from the first documentation of CR or PR until the date of progression, death, or the last evaluable RECIST assessment for patients who do not progress. The DOR was calculated following the PFS methodology.

^eCalculated using the Kaplan-Meier technique.

Source: TOPAZ-1 CSR (IA-2).²³

Health-Related Quality of Life

The PRO analysis set was used for all HRQoL and PRO measures; however, the sample size varied by measure. HRQoL measured by the EORTC QLQ-C30 and EORTC QLQ-BIL21 were secondary end points in the TOPAZ-1 trial. The PGI-S, PRO-CTCAE, and EQ-5D measures were exploratory end points of the TOPAZ-1 trial.²³

EORTC QLQ-C30

The PRO analysis set for the EORTC QLQ-C30 included 318 patients in the durvalumab plus gemcitabine and cisplatin group, and 328 patients in the placebo plus gemcitabine and cisplatin group. PRO analyses for EORTC QLQ-C30 at IA-2 included time to deterioration and improvement rates (Table 20). For GHS/QoL, 132 (49.1%) and 135 (48.8%) patients in the durvalumab plus gemcitabine and cisplatin and the placebo plus gemcitabine and cisplatin groups, respectively, experienced deterioration. The HR for durvalumab plus gemcitabine and cisplatin versus placebo plus gemcitabine and cisplatin for GHS/QoL was 0.87 (95% CI, 0.69 to 1.12). For GHS/QoL, the median time to deterioration was 7.4 months (95% CI, 5.6 to 8.9 months) for durvalumab plus gemcitabine and cisplatin and 6.7 months (95% CI, 5.6 to 7.9 months) for placebo plus gemcitabine and cisplatin. Results for the proportion of patients experiencing deterioration in functional domains ranged from ████████████ in the durvalumab plus gemcitabine and cisplatin group and ████████████████████████ in the placebo plus gemcitabine and cisplatin group. The median time to deterioration in functional domains ranged from █ ████ ██████ █████ ████████ for durvalumab plus gemcitabine and cisplatin and █████ █████ █████ ███████ ██ for placebo plus gemcitabine and cisplatin. In the multiple and single symptom domains, the proportion of patients experiencing deterioration ranged from 26.5% (diarrhea) to 52.4% (fatigue) in the durvalumab plus gemcitabine and cisplatin group, and from 26.8% (diarrhea) to 50.0% (fatigue) in the placebo plus gemcitabine and cisplatin group. The median time to deterioration in multiple and single symptom items ranged from ████ ███████ █████ █████ ██ █ █ █████████ ████████ ██████ for diarrhea with durvalumab plus gemcitabine and cisplatin and from █████ █ ███ ████ ███████ █ █ for fatigue to ██ ███ ██ █████ ██████ ██████ for diarrhea with placebo plus gemcitabine and cisplatin.²³

Improvements in GHS/QoL or functional domains were defined by the investigators as 2 consecutive assessments at least 14 days apart that show an increase from baseline score greater than or equal to 10 points. For GHS/QoL at IA-2, █████ █████ ███ ██ ███████ ██ patients in the durvalumab plus gemcitabine and cisplatin group and █████ ██ ██ ███ patients in the placebo plus gemcitabine and cisplatin group showed improvement. Across functional domains, ██████ █████████ █████ ████ of patients in the durvalumab plus gemcitabine and cisplatin group and ██ █████ ███ ██ ██████████ ██ of patients in the placebo plus gemcitabine and cisplatin group showed improvement. Improvement in multiple and single symptom domains was defined as 2 consecutive assessments at least 14 days apart that show a decrease from baseline score greater than or equal to 10 points. In total, █ ██████ ████ ██ ████ ███████ of patients in the durvalumab plus gemcitabine and cisplatin group and ████ ██████████ ███████████ ███████ ████ of patients in the placebo plus gemcitabine and cisplatin group showed improvement in symptoms at IA-2.²³

Table 20: Time to Deterioration and Improvement Based on BOR for EORTC QLQ-C30 (IA-2, PRO Analysis Set)

BOR = best overall response; CI = confidence interval Cis = cisplatin; D = durvalumab; Gem = gemcitabine; EORTC QLQ-C30 = European Organisation for Research and Treatment of Cancer Quality of Life Questionnaire Core 30; GHS = global health status; HR = hazard ratio; IA-2 = second interim analysis; N/A = not applicable; NC = not calculable; NR = not reported; OR = odds ratio; PBO = placebo; PRO = patient-reported outcome; QoL = quality of life; TTD = time to deterioration.

^aHRs are presented for time to deterioration. HRs < 1 favour durvalumab plus gemcitabine plus cisplatin. ORs are presented for improvement rates based on BOR. ORs > 1 favour durvalumab plus gemcitabine plus cisplatin.

Source: TOPAZ-1 CSR (IA-2).²³

█████ █ ██ ██████████ ███████ █████████ █████████ ████ ██████████ ██ ████ ██ ██████ ██████ ████ █ █ ██ ██ █████ █ ██ ██████████ ███████ █████████ █████████ ████ ██████████ ██ ████ ██ ██████ ██████████ ██ ███ ██ ████ █ ██ ██████████ ███████ █████████ ████ █████ ████ ██████████ ██ ████ ██ ██████ █████████ █ ███ ███ █████ █ ██ ██████████ ███████ █████████ █████████ ████ ██████████ ██ ████ ██ ██████ ███ ███ ████ █ ████ ██████ █ ██ ████ ██████ ███████ █████████ ███ ██████ ████ ██████████ ██ ████ ██ ██████ █ ██ ███████ ██ ████ █████████████ ███████

EORTC QLQ-BIL21

The PRO analysis set for the EORTC QLQ-BIL21 included 305 patients in the durvalumab plus gemcitabine and cisplatin group and 322 patients in the placebo plus gemcitabine and cisplatin group. PRO analyses for EORTC QLQ-BIL21 at IA-2 included time to deterioration and improvement rates (Table 21). Clinically meaningful symptom deterioration was defined by the investigators as an increase from baseline of greater than or equal to 10 points for symptom scales or items. The proportion of patients experiencing deterioration in symptoms ranged from 20.6% (jaundice) to 48.5% (tiredness) with durvalumab plus gemcitabine and cisplatin and from 19.9% (jaundice) to 47.9% (tiredness) with placebo plus gemcitabine and cisplatin. The median time to symptom deterioration ranged from █████ █ ██ ██████████ ███████ █████████ █████████ ██████████████ for durvalumab plus gemcitabine and cisplatin and from █████ █ ██ ██████████ ███████ █████████ █████████ ██████████████ with placebo plus gemcitabine and cisplatin.²³

Improvement rates for EORTC QLQ-BIL21 at IA-2 are also summarized in Table 21. Improvement in symptoms was defined by the investigators as 2 consecutive assessments at least 14 days apart that show a decrease from baseline score greater than or equal to 10 points. █████ █ ██ ██████████ ███████ █████████ █████████ ████ ██████████ ██ ████ ██ ██████ █████ █████ ███████████ █ ██ ██████████ ███████ █████████ █████████ ████ ██████████ ██ ████ ██ ██████ ██ ██ ██████ █████ █ ██ ██████████ ███████ █████████ █████████ ████.

Table 21: Time to Symptom Deterioration and Improvement Rates for the EORTC QLQ-BIL21 (IA-2, PRO Analysis Set)

BOR = best overall response; CI = confidence interval; Cis = cisplatin; D = durvalumab; EORTC QLQ-BIL21 = European Organisation for Research and Treatment of Cancer 21-Item Cholangiocarcinoma and Gallbladder Cancer Quality of Life Questionnaire; Gem = gemcitabine; HR = hazard ratio; IA −2 = second interim analysis; N/A = not applicable; NC = not calculable; OR = odds ratio; PBO = placebo; PRO = patient-reported outcome; TTD = time to deterioration.

^aHRs are presented for time to deterioration. HRs < 1 favour durvalumab plus gemcitabine plus cisplatin. ORs are presented for improvement rates based on BOR. ORs > 1 favour durvalumab plus gemcitabine plus cisplatin.

Source: TOPAZ-1 CSR (IA-2).²³

EQ-5D-5L

The impact of treatment and disease state on health state utility, assessed with the EQ-5D-5L, was an exploratory outcome of the TOPAZ-1 trial. ███████████ ███ █████ ███████████ █ █████ ██ ██████ ███ ███ ████ ██████ ████ █████████ █████ █████ ██████ ██████████████ ███████ ███████████████ ██ ████████████ ████ ███████ ██ ██ ████ ██████ ██████████████ ███████████ █████████ ████ ██ ███████████ ██████ ██████████ ████ ██ ██████████ ████ ██████ ████ ████████████ █████ ███ ███████ ███████ ██████████████ █████ ██████████ ███ ███ ████████ ████ █████████ ██████ ██████████ ████████ ███████████████ ███████ ████████ ██ █████████ ████████ █████████ ██████████████ ███████████████ ██████. Results for change from baseline in EQ-5D-5L index score are summarized in Figure 8. ███ ███████████████ ████████████ ██████████████████ █████ ██████████ ███████ █████ ████████████████████ ██████████ ███████████ ████ ███████ ██ █ ███ ███████ █████ ████ ████ ████████ ██████ ██████ ███████████ █ ██ ██████ ███ █████ █ ██ ██████████ ███████ █████████ ████ █████ ████ ██████████ ██ ████ ██ ██████ ██████████ ██████████ ███████████ ██████ ██ █████████████. Results for change from baseline in EQ-VAS are summarized in Figure 9. ███ ███ ████ █████ █████ ██████████ █████ ███████████ ███████ ███████████████ ██ ███ ██ █████ ████ ████████████ ██████ █████ ██ ██ ████ ██████ ████████ ███████████ ███████████ ██████████████ ███ ██████ ████████ ██████████ ██ ███ ██ ███████ ████ ████████ ███ ████████ ██ ██ █████ ███ ███████ ████ ████ ████ ███ █████████████.

Symptom Severity

PGI-S of Cancer Symptoms

PGI-S was an exploratory outcome of the TOPAZ-1 trial. ██████████████ █████████ █████ █ ██ ████████ ██ ███████ ██████ ███ █████████ ████ ██████████ ██ ████ ██ ██████ ██████████ ███████████ █ ██ ██████████ ███████ ███ ██████ █████████ ████ ██████████ ██ ████ ██ ██████ ██████████ ██████ ████ ██████ ████████ ██ ██ ███████ ██████ ██████ █████ █ ██ ██████████ ███████ █████████ █████████ ████ ██████████ ██ ████ ██ ██████ ██████████ ███████████ █ ██ ██████████ ███████ █████████ █████████ ████ ██████ ████ ██ ████ ██ ██████ ██████ ████ ███ ████ ███████████████ ███████████████████ ██ ████████ ██████████████████ ███████████ ████ ██████████████ ███████ █████████████████████ ████ ██████████ ███████ ███████ ████ ███ ████ █████ ██████████████ █████ █████ █ ██ ██████████ ███████ █████ ████ █████████ ████ ████ ██████ ██ ████ ██ ██████ ██████████ ███████████████████ █████ █ ██ ██████████ ███████ █████████ █████████ ████ ██████████ ██ ████ ██ ██████ ████ ██████ ███████████ █ ██ ██████████ ███████ █████ ████ ████ █████ ████ ██████████ ██ ████ ██ ██████ ██████████ ████ ███████ █ ██ ██████████ ███████ █████████ ███ ██████ ████ ██████████ ██ ████ ██ ██████ ████ ██████ ███████████ █ ██ ██████████ ███████ █████████ ███ ██████ ████ ██████████ ██ ████ ██ ██████ ██████████ ██████████████.

Treatment Tolerability

PRO-CTCAE

Treatment tolerability, assessed with the PRO-CTCAE, was an exploratory outcome of the TOPAZ-1 trial. █████ █ ██ ████ ██ ████ ███████ █████████ █████████ ████ ██████████ ██ ████ ██ ██████ ██████████ ██ █████████ █ ██ ██ █████ ███ ███████ █████████ █████████ ████ ██████████ ██ ████ ██ ██████ ██████████ ███████████ █ ██ █████ █████ ███████ █████████ █████████ ████ ██████████ ██ ████ ██ ██████ ██ ████████ ██ ██ █ ███ ███ █ ██ █████ █████ ███████ █████████ █ ████████ ████ ████ ██████ ██ ████ ██ ██████ ██████ ████ ██ ██ ███████ █ ██ ███████ ███ ███████ ███ ██████ █████████ ████ ██████████ ██ ████ ██ ██████ ███ ████ ███ ██ ██ ██ █████ █ ██ ██████████ ███████ █████ ████ ██████ ███ ████ █████ █████ ██ ████ ██ ██████ ██████████ ██ ██ ███ ████ █ ██ ███████ ███ ███████ █████████ █████ ████ ████ ██████████ ██ ████ ██ ██████ █ ███████ ██ ██ ███ ██████ █ ██ ██████████ ███████ █████████ █████████ ████ ██████████ ██ ████ ██ ██████ █████ █████ ███████████ █ ██ ██████ ████ ███████ ███ ██████ █████████ ████ ████ ██████ ██ ████ ██ █████ █ ███ ██ █████ █████ ██████ █ ██ ██████████ ███████ █████████ █ ████████ ███ █ ██████████ ██ ████ ██ ██████ ██ ████████ ███████████ █ ██ ██████████ ███████ █████████ █████████ ████ ██████████ ██ ████ █ █ ██████ ████ ██████ ████ ███████ ██ █████ ██████ ██████ ███ ████ ████ █████ ████████ ██████████ █████ █████████ ██████ ██████.

Harms

Only harms identified in the review protocol are reported here. Refer to Table 22 and Table 23 for detailed harms data from IA-2 and the 6.5-month update.

Adverse Events

At both IA-2 and the 6.5-month update, TEAEs were reported in 99.4% and 98.8% of patients in the durvalumab plus gemcitabine and cisplatin and placebo plus gemcitabine and cisplatin groups, respectively. The most frequently reported TEAEs in the durvalumab plus gemcitabine and cisplatin group included anemia (163 [48.2%]), nausea (137 [40.5%]), constipation (109 [32.2%]), and neutropenia (107 [31.7%]). The most frequently reported TEAEs in the placebo plus gemcitabine and cisplatin group were anemia (154 [45.0%]), nausea (119 [34.8%]), decreased neutrophil count (106 [31.0%]), and neutropenia (102 [29.8%]). Nausea was the only TEAE that had a greater than 5% difference between durvalumab plus gemcitabine and cisplatin and placebo plus gemcitabine and cisplatin (40.5% versus 34.8%). AEs with known association to durvalumab were reported at a higher rate with durvalumab plus gemcitabine and cisplatin than with placebo plus gemcitabine and cisplatin, including rash (11.8% versus 7.9%), pruritis (12.7% versus 8.5%), and pyrexia (20.4% versus 17.0%).²³

As of the 6.5-month update, grade 3 or 4 AEs were reported at a similar frequency in the 2 treatment groups, with a total of 250 (74.0%) and 257 (75.1%) in the durvalumab plus gemcitabine and cisplatin and placebo plus gemcitabine and cisplatin groups, respectively. █████ █ ██ ██████████ ███████ █████████ █████████ ████ ██████████ ██ ████ ██ ██████ █ ██ ████ ███ ███████████ █ ██ ██████████ ███████ █████████ █████████ ████ ██████████ ██ ████ ██ ██████ █████ █████ ██████.

█████ █ ██ ██████████ ███████ █████████ █████████ ████ ██████████ ██ ████ ██ ██████ ██████████ ███ ██ ██ ████ █ ██ ██████████ ███████ █████████ █████████ ████ ██████████ ██ ████ ██ ██████ ██████████ ███ ██ ██████ █ ██ ██████████ ███████ █████████ █████████ ████ ██████ ████ ██ ████ ██ ██████ ██████████ ██ █ ████ ████ █ ██ ██████████ ███████ █████████ █████████ ████ ██████████ ██ ████ ██ ██████ ██ █ █ ██████ ██ █████ ████ █ ██ ██████████ ███████ █████████ █████████ ████ ██████████ ██ ████ ██ ██ ████ █ ███ ██████ ██████

Serious Adverse Events

SAEs at the 6.5-month follow-up were consistent with those in IA-2. SAEs were well balanced across treatment groups and were reported for ████████ ████████ ███████ patients in the durvalumab plus gemcitabine and cisplatin group and ███ █████ ███████ ████████ patients in the placebo plus gemcitabine and cisplatin group.^23,24

█████ █ ██ ██████████ ███████ █████████ █████████ ████ ██████████ ██ ████ ██ ██████ ██████████ ██ ██ ██ █████ █ ██ ██████████ ███████ █████████ █████████ ████ ██████████ ██ ████ ██ ██████ ███ ███ ████ ███ ████████ █ ██ ██████████ ███████ █████████ █████████ ████ ██████████ ██ ████ ██ ██████ ██████████ █████ █ ██ ██████████ ███████ █████████ █████████ ████████ █████████████████████

Withdrawals Due to Adverse Events

At IA-2 of the TOPAZ-1 study, 44 (13.0%) patients in the durvalumab plus gemcitabine and cisplatin group and 52 (15.2%) patients in the placebo plus gemcitabine and cisplatin group discontinued treatment because of AEs. █████ █ ██ ██████████ ███████ █████████ █████████ ████ ██████████ ██ ████ ██ ██████ ██████████ ███████████ █ ██ ██████████ ██ ██ ██ █ █████████ █████████ ████ ██████████ ██ ████ ██ ██████ ██████████ ███████████ █ ██ ██████████ █ ██ ████ ███ ██████ █████████ ████ ██████████ ██ ████ ██ ██████ ██████████ ███████████ █ ██ ██████████ ██ █████ █████████ █████████ ████ ██████████ ██ ████ ██ ██████ ██████████ ██████.

At the 6.5-month update, 1 patient in the durvalumab plus gemcitabine and cisplatin group was removed from the dataset, as the withdrawal due to asthenia was considered to be related to disease progression by the investigator, for a total of 43 (12.7%) patients who discontinued due to AEs. █████ █ ██ ██████████ ███████ █████████ █████████ ████ ██████████ ██ ████ ██ ██████ █████ █████ ███████████ █ ██ ██████████ ███████ █████████ █████████ ████ ██████████ ██ ████ ██ ███ ███ ██████████ ███████████ █ ██ ██████████ ███████ █████████ █████████ ████ █ ██ ███ ████ ██ ████ ██ ██████ ███ ███████ ███████████ █ ██ ██████████ ███████ █████████ █████████ ████ ██ ███ █████ ██ ████ ██ ██████ ██████████ ███████████ █ ██ ██████████ ███████ █████████ █████████ ████ ███ ███████ ██ ████ ██ ██████ ██████████ ███████ ████████ ██████████████.

Mortality

At IA-2, of the 341 patients who were randomized to receive durvalumab plus gemcitabine and cisplatin, 12 (3.6%) patients experienced fatal AEs. There were 14 (4.1%) patients who experienced fatal AEs in the placebo plus gemcitabine and cisplatin group due █████ █ ██ ██████████ ███████ █████████ █████████ ████ ██████████ ██ ████ ██ ██████ ██ ████ ████ ███████████ █ ██ ██████████ ███████ █████████ █████████ ████ ██████████ ██ ████ ██ ██████ ████ ██ ██ ██ ████ ███████ █ ██ ██████████ ███████ █████████ █████████ ████ ██████████ ██ ████ ██ ██████ ███ ██████ █ ████ ███████ █ ██ ██████████ ███████ █████████ █████████ ████ ██████████ ██ ████ ██ ██████ ██████████ ████████ ███████████ ██████████.

At the 6.5-month update, 1 additional AE leading to death was reported in the durvalumab plus gemcitabine and cisplatin group, for a total of 13 (3.8%); AE-related deaths in the placebo plus gemcitabine and cisplatin remained at 14 (4.1%). As of the 6.5-month update, █████ █ ██ ██████████ ███████ █████████ █████████ ████ ██████████ ██ ████ ██ ██████ ██████████ ████ ███████ █ ██ ██████████ ███████ █████████ █████████ ████ ██████████ ██ ████ ██ ██████ ███ █ ██ █ ███ ███████████ █ ██ ██████████ ███████ █████████ █████████ ████ ██████████ ██ ████ ██ ██████ ██ ██ ██████ ███████████ ██████ ████████████.

Table 22: Summary of Harms (Safety Analysis Set)