CADTH Reimbursement Review

Sotorasib (Lumakras)

Sponsor: Amgen Canada Inc.

Therapeutic area: KRAS G12C-mutated, advanced non–small cell lung cancer

This multi-part report includes:

Clinical Review

Pharmacoeconomic Review

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

NOC = Notice of Compliance; NOC/c = Notice of Compliance with conditions; NSCLC = non–small cell lung cancer.

Introduction

Lung cancer is 1 of the most diagnosed cancers, with non–small cell lung cancer (NSCLC) accounting for approximately 80% to 88% of all lung cancer diagnoses in Canada.^1,2 In 2020, the incidence of NSCLC in Canada was estimated to be 60.5 per 100,000 persons.³ In 2021, an estimated 29,600 new cases of lung cancer and 21,150 deaths due to lung cancer were projected.⁴ Survival from lung cancer across all stages and histologies is poor, with a 5-year net survival rate of 22%.⁵ NSCLC often remains asymptomatic until the disease is in its advanced stages.^6,7 When patients with NSCLC present with symptoms, these are usually nonspecific and difficult to attribute to lung cancer.⁷ Compared to other advanced cancers, advanced NSCLC is associated with a higher prevalence and intensity of symptoms, such as pain, dyspnea, cough, decreased appetite, weight loss, and depression, as well as lower health-related quality of life (HRQoL).⁸ Because early diagnosis of NSCLC is challenging,⁹ approximately two-thirds of patients have advanced or metastatic disease at diagnosis, at which point curative treatments are not possible.^4,10 The median overall survival (OS) of patients with metastatic NSCLC (stages IVA and IVB) is poor, ranging from 8 months to 11 months; the 5-year OS ranges from 4% to 6%. The 5-year net survival rate for stage IV NSCLC is 5.2%.¹¹

NSCLC often holds oncogenic driver mutations that lead to uncontrolled cell growth and proliferations.¹² Of these, mutations within the RAS family account for more than 30% of all mutated oncogenes in NSCLC, causing approximately 1 million deaths worldwide annually.¹³ Within the RAS family, KRAS is the isoform most frequently altered in NSCLC. Approximately 1 in 4 patients with NSCLC harbour KRAS mutations.¹⁴ Patients with KRAS G12C-mutated NSCLC have a lower proportion of response to cytotoxic chemotherapy and decreased survival compared to the overall population of patients with NSCLC.¹⁵ The KRAS G12C subtype represents almost half of all KRAS mutations in NSCLC and is identified in approximately 13% of patients with NSCLC.^16,17 Based on an estimation in the Health Canada Reviewers Report, the incidence of patients in Canada living with NSCLC and harbouring the KRAS G12C mutation is approximately 7.9 per 100,000 persons.¹⁸

Sotorasib is a highly selective inhibitor of KRAS G12C that suppresses the rapid growth of cancer cells. Sotorasib forms an irreversible, covalent bond with the unique cysteine of KRAS G12C, locking the protein in an inactive state that prevents downstream signalling without affecting wild-type KRAS. Sotorasib blocks KRAS signalling, inhibits cell growth, promotes apoptosis, and is associated with antitumour inflammatory responses and immunity in KRAS G12C tumour models. On September 10, 2021, sotorasib received a Notice of Compliance with conditions (NOC/c) from Health Canada for the treatment of adult patients with KRAS G12C-mutated, locally advanced (not amenable to curative therapy) or metastatic NSCLC who have received at least 1 prior systemic therapy. The Health Canada market authorization with conditions is pending the results of trials to verify the clinical benefit of sotorasib. Sotorasib underwent advance consideration under NOC/c, as well as Project Orbis, at Health Canada. Health Canada recommends that sotorasib be administered at a dose of 960 mg (8 × 120 mg tablets) orally once daily until disease progression or unacceptable toxicity. Up to 2 dose reductions are permitted in the case of adverse events (AEs).

The objective of this review is to evaluate the beneficial and harmful effects of sotorasib 960 mg (8 × 120 mg oral tablets) for the treatment of KRAS G12C-mutated, locally advanced (not amenable to curative therapy) or metastatic NSCLC in adult patients who have received at least 1 prior systemic therapy.

Stakeholder Perspectives

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 Input

The Lung Health Foundation (LHF) (formerly the Ontario Lung Association), Lung Cancer Canada (LCC) and the Canadian Cancer Survivor Network (CCSN) submitted joint input. A total of 5 people (4 patients and 1 caregiver) responded to telephone survey interviews conducted from August 2022 to September 2022. Of the 4 patients with lung cancer, 3 had experience with sotorasib; 1 patient with KRAS G12C-mutated lung cancer had no experience with sotorasib. All respondents were female. All respondents but 1 were from Canada (Nova Scotia, British Columbia, Quebec, and Ontario); the remaining respondent was from the US.

From the submitted input, 1 patient said they felt “rock bottom” when their lung cancer progressed on chemotherapy and radiation, leading them to be on oxygen, to acquire a debilitating cough as well as shortness of breath from eating or talking, and to require assistance to shower. Another patient detailed the mental and emotional side effects associated with multiple courses of immunotherapy and chemotherapy, such as depression, anxiety, panic attacks, and severe mood swings, which remained debilitating even during remission. The sole caregiver interviewed stated that it is mentally, physically, and financially challenging to care for a patient with lung cancer with comorbidities (e.g., preparing meals, making arrangements for transportation to medical appointments, managing daily responsibilities) and that they felt burned out without many sources of support. Three patients who had experience with sotorasib reported significant tumour reduction (i.e., ranging from a 50% to 65% reduction to no evidence of disease within 5 weeks to 1.5 years). They said they experienced mild side effects (e.g., fatigue, aches, and pains when walking for extended periods of time, a minor rash, diarrhea, shortness of breath, and increased liver enzymes) while on sotorasib. However, they said these side effects did not have much impact on their daily activities and/or quality of life. According to their input, these patients felt hopeful that they could plan for the future.

In terms of important key outcomes, 1 patient said they were most interested in obtaining a cure while maintaining a good quality of life. Respondents also expressed their hope for a treatment option with an oral route of administration that would be accessible from home, delay the onset of symptoms, prolong their life, and improve functionality and mobility, with fewer side effects. The caregiver expressed the importance of treatment that can be accessed from home, limiting the need to travel to infusion clinics. Lastly, it was pointed out in the survey that wait times for lung surgeries across Canada are unacceptable; patients and caregivers would like to treat the lung cancer in the early stages. The caregiver surveyed said that the wait time for the surgery was the most difficult aspect (or the most challenging adverse effect) of current treatment.

Clinician Input

Input From Clinical Experts Consulted by CADTH

According to the clinical experts, the treatment options for patients with KRAS G12C-mutated NSCLC who have progressed on standard therapy are limited. There is an unmet need in this patient population for an efficacious treatment associated with fewer AEs. According to the clinical experts consulted for this review, sotorasib would change the line of therapies for patients with KRAS G12C-mutated NSCLC. The clinical experts anticipated that sotorasib would be used following immunotherapy and platinum-doublet chemotherapy. Accordingly, docetaxel would then move to being either third-line or fourth-line therapy. The clinical experts noted that sotorasib would not be combined with other drugs at this time. They suggested that sotorasib be limited to patients with KRAS G12C-mutated NSCLC. Moreover, they noted that treatment with sotorasib would not be suitable for patients with an Eastern Cooperative Oncology Group Performance Status (ECOG PS) of 3 or 4, with severe organ dysfunction, or with untreated, symptomatic brain metastasis. The clinical experts did note that patients with untreated, asymptomatic brain lesions may be suitable for sotorasib; ideally, these patients should have their cases discussed at a multidisciplinary tumour board round at a centre with expertise in stereotactic radiosurgery. Based on input from the clinical experts, patients should undergo clinical and toxicity assessment per cycle (typically every 3 to 4 weeks) and imaging every 3 months to assess response to treatment in clinical practice. Based on input from the clinical experts, improved progression-free survival (PFS), improved OS, and maintenance or improvement in quality of life are considered meaningful responses to treatment in this population. The clinical experts suggested that treatment with sotorasib be discontinued under the following 3 scenarios: patient decision to stop treatment with sotorasib; unacceptable toxicity due to sotorasib; and disease progression without clinical benefits. The clinical experts agreed that patients with documented disease progression could continue sotorasib if they were deriving clinical benefit. They also noted that sotorasib may be prescribed by a medical oncologist in an outpatient oncology clinical setting.

Clinician Group Input

Clinician group input was provided by 2 groups: the Ontario Health – Cancer Care Ontario (OH-CCO) Lung Cancer Drug Advisory Committee (LC DAC), which provided input from 3 clinicians who had joint discussion through email; and LCC, which provided input based on a review of the literature and proceedings from recent conferences compiled by 26 clinicians. Both groups identified the following 3 goals of therapy: reducing tumour burden; improving symptoms; and prolonging life while upholding patients’ values and desires. Both groups also noted that all patients eventually progress on currently available treatment; thus, durability of response is also an important treatment goal. In addition, LCC added the need for treatment that is associated with reduced toxicity and resource utilization. Indeed, the OH-CCO LC DAC endorsed the advent of an oral anticancer treatment without the side effects and life impacts associated with chemotherapy to slow disease progress and improve length and quality of life. According to input from the OH-CCO LC DAC, sotorasib would be placed as a second- or third-line monotherapy for patients who have progressed on, or are unable to tolerate, platinum-based chemotherapy and immunotherapy (where appropriate) and possibly docetaxel. The OH-CCO LC DAC clinician group stated that sotorasib would not affect immunotherapy or platinum-doublet chemotherapy use; however, sotorasib may be preferred to docetaxel if it can demonstrate a meaningful improvement in survival or quality of life, given that docetaxel has a proven survival benefit compared to best supportive care (BSC) and other chemotherapy regimens (ifosfamide), albeit with an unfavourable side effect profile.

The input from LCC placed sotorasib as a second-line standard of care for patients with advanced KRAS G12C NSCLC and suggested that it be used as a single drug after at least 1 line of prior systemic treatment has not worked. According to both clinician groups, patients with KRAS G12C-mutated, advanced NSCLC (i.e., stage IV or recurrent) who had received prior therapy would be best suited for sotorasib. LCC added that it is uncertain whether patients with an ECOG PS of 3 or 4 would benefit from treatment with sotorasib. Both clinician groups noted that patients eligible for treatment with sotorasib should be identified by a validated molecular diagnostic test, preferably next-generation sequencing (NGS). Both clinician groups noted that improvement in symptoms, stability of disease, tumour shrinkage, radiographic reduction of disease site from baseline, and prolonged survival are indicative of a clinically meaningful response to treatment. The OH-CCO LC DAC added that the definition of a clinically meaningful improvement in frequency or severity of symptoms depends on patients and varies across physicians. Based on input from LCC, response to treatment should be determined by the treating physician based on CT imaging and assessed every 2 to 3 months, similarly to other oral tyrosine kinase inhibitors used in Canada.

Based on input from the clinician groups, discontinuing treatment with sotorasib should be considered in the event of loss of clinical benefit (as indicated by unequivocal disease progression, such as symptoms, polyprogression, and so on) or intolerable side effects. LCC noted that treatment with sotorasib may continue in patients with oligoprogression who are amenable to local therapy (radiation or surgery); who are newly diagnosed or have experienced progression of brain metastases and have been treated with brain radiation or surgery; and who have asymptomatic disease without overt progression on imaging associated with increased symptom burden. The OH-CCO LC DAC emphasized that a combination of clinical judgment, radiological interpretation, and clinically standardized scales, such as the Edmonton Symptom Assessment System, should be used to determine if a patient is benefiting from therapy. The OH-CCO-LC also noted that the Response Evaluation Criteria in Solid Tumours Version 1.1 (RECIST 1.1) and its derivatives are designed for clinical trials, not clinical practice, and should not be used to determine discontinuation of treatment. Both clinician groups agreed that outpatient clinics would be an appropriate treatment setting for sotorasib in addition to the hospital setting. Both groups stated that a medical oncologist, general practitioner in oncology under medical oncology supervision, or respirologist experienced in treating patients with lung cancers should be involved in diagnosing, treating, and monitoring patients on sotorasib.

Drug Program Input

Input was obtained from the drug programs that participate in the CADTH reimbursement review process. The following were identified as key factors that could potentially affect the implementation of a CADTH recommendation for sotorasib:

• consideration for prescribing of therapy

• care provision

• systemic and economic issues.

Clinical Evidence

Pivotal Studies and Protocol-Selected Studies

Description of Studies

Two sponsor-conducted studies were included in this systematic review: the CodeBreaK 100 study (the pivotal trial) and the CodeBreaK 200 study.^19-21

The CodeBreaK 100 study is an ongoing, multicentre, nonrandomized, open-label, single-group, phase I and II study. Phase I of the CodeBreaK 100 study was a first-in-human dose exploration (part 1) and dose expansion (part 2) study aimed at evaluating the safety, tolerability, pharmacokinetics (PKs), and pharmacodynamics of sotorasib in adult patients with KRAS G12C-mutated advanced NSCLC, colorectal cancer, and other solid tumours.¹⁹ Phase II of the CodeBreaK 100 study was designed to evaluate the efficacy and safety of sotorasib as monotherapy in adult patients with KRAS G12C-mutated advanced NSCLC, colorectal cancer, and other solid tumours.²⁰ CADTH’s review of the CodeBreaK 100 study focused on the phase II efficacy and safety results in adult patients with KRAS G12C-mutated advanced NSCLC. Phase II of the CodeBreaK 100 study enrolled a total of 224 patients with KRAS G12C-mutated advanced solid tumours across 59 sites. Of the patients enrolled, 126 (56.3%) had NSCLC. Patients self-administered sotorasib 960 mg (8 × 120 mg tablets) orally once daily and continued treatment without interruption until disease progression, treatment intolerance, withdrawal of consent, or death. Tumour response was assessed through contrast-enhanced CT or MRI and assessed per RECIST 1.1 by blinded independent central review (BICR). Patients underwent a safety follow-up visit 30 days (± 7 days) after the last dose of sotorasib before any new anticancer treatment was started. Following safety follow-up visits, patients were followed long-term for health condition, disease status, and subsequent anticancer treatment every 12 weeks (± 2 weeks) for up to 3 years after the last patient was enrolled or until withdrawal of consent, loss to follow-up, or death, whichever occurred first. The study team was blinded to the efficacy data. The primary efficacy end point for phase II of the CodeBreaK 100 study was objective response rate (ORR), which was a composite of CR and partial response (PR). Response was assessed by BICR. Secondary efficacy end points included duration of response (DOR), disease control rate (DCR), time to response (TTR), PFS, OS, and the Non–Small Cell Lung Cancer Symptom Assessment Questionnaire (NSCLC-SAQ). Exploratory outcomes included changes in patient-reported, cancer-specific symptoms and HRQoL measures, including the European Organisation for Research and Treatment Quality of Life Questionnaire Core 30 (EORTC QLQ-C30) and its Lung Cancer Module (EORTC QLQ-LC13); the EQ-5D-5L and EQ visual analogue scale (EQ VAS); item GP5 (“I am bothered by side effects of treatment”) from the Functional Assessment of Cancer Therapy – General (FACT-G); the patient global impression of change (PGIC) and patient global impression of severity (PGIS) questionnaires; and the Patient-Reported Outcomes version of the Common Terminology Criteria for Adverse Events (PRO-CTCAE). The data cut-off date for the primary analysis of phase II was September 1, 2020, which was updated at the data cut-off date of March 15, 2022. The updated data cut-off date for OS at 2 years was September 9, 2022.

Patients enrolled in the CodeBreaK 100 study had a mean age of 62.9 years (standard deviation [SD] = 9.3 years). Most patients were aged 18 years to 64 years (53.2%) and white (81.7%); 81% were former smokers. Regarding disease characteristics, 61.9% and 98% of patients had stage IV disease at initial diagnosis and screening, respectively. At the time of screening, metastatic disease was identified in 96.8% of enrolled patients, with the most common site of metastasis found in the bone (48.4%). The most common histology type among enrolled patients was nonsquamous adenocarcinoma (95.2%). A total of 42.9%, 34.9%, and 22.2% of patients had received 1, 2, or 3 prior lines of anticancer therapy, respectively. The most common types of prior anticancer therapy were immunotherapy checkpoint inhibitors (92.1%) and chemotherapy (91.3%). A total of 81% of patients had received both platinum-based chemotherapy and anti–programmed cell death 1 (PD-1) or anti–programmed cell death-ligand 1 (PD-L1) immunotherapy.

The CodeBreaK 200 study is an ongoing, multicentre (148 sites across 22 countries), randomized (1:1), open-label, parallel-group, phase III study evaluating the efficacy of oral sotorasib (960 mg daily) (n = 171) versus IV docetaxel (75 mg/m² every 3 weeks) (n = 174) in adult patients with KRAS G12C-mutated NSCLC who progressed after prior platinum-based chemotherapy and a checkpoint inhibitor. Randomization was stratified by the number of prior lines of therapy in advanced disease (1 versus 2 versus > 2), race (Asian versus non-Asian), and history of central nervous system (CNS) involvement (present or absent). Patients continued treatment without interruption until disease progression, treatment intolerance, withdrawal of consent, or death. Tumour response was assessed through contrast-enhanced CT or MRI and assessed per RECIST 1.1 by BICR every 6 weeks from cycle 1, day 1 until week 49, and then at 9-week intervals thereafter. A cycle was defined as 21 days in length (± 3 days), unless a delay was medically necessary. Patients randomized to the docetaxel treatment group who experienced disease progression, confirmed by radiological assessment after study initiation, were permitted to cross over to the sotorasib treatment group (n = 46). The primary end point of the CodeBreaK 200 study was PFS. Key secondary end points included OS, ORR, and HRQoL, as assessed by the EORTC QLQ-C30 and EORTC QLQ-LC13. Other secondary end points included DOR, TTR, and DCR. The data cut-off date of the primary analysis was August 2, 2022.

Patients randomized in the CodeBreaK 200 study had a mean age of 63.5 years (SD = 9.5 years). Most patients were aged 18 years to 64 years (53.9%), white (82.9%), and current or former smokers (96.2%). With regard to disease characteristics, metastatic disease was identified in 95.1% of randomized patients. The most common histology type among the randomized patients was nonsquamous (96.8%). A total of 42.9%, 40.9%, and 16.2% of patients received 1, 2, and 3 or more complete prior lines of therapy, respectively; 34.2% were on maintenance therapy. Between the 2 treatment groups, a greater proportion of patients were male in the sotorasib treatment group (63.7%) relative to the docetaxel treatment group (54.6%). Other imbalances in baseline characteristics between the sotorasib and docetaxel treatment groups were noted for the following: PD-L1 protein expression greater than or equal to 1% and less than 50% (sotorasib = 26.9%; docetaxel = 40.2%); ECOG PS score of 0 at cycle 1, day 1 (sotorasib = 38.6%; docetaxel = 33.9%); primary refractory to last prior line of therapy (sotorasib = 39.2%; docetaxel = 32.8%); and initial response with subsequent growth (sotorasib = 20.5%; docetaxel = 27.0%).

Baseline patient characteristics were generally similar between the CodeBreaK 100 study and the CodeBreaK 200 study.

Efficacy Results

A summary of key results for the CodeBreaK 100 study and the CodeBreaK 200 study is presented in Table 2. In the CodeBreaK 200 study, the Maurer-Bretz²² multiple testing procedure was used to control the study-level overall type I error rate below 1-sided 0.025 levels, starting with PFS. If all 3 hypotheses of PFS, ORR, and OS were sequentially rejected, then the end points of change from baseline over 12 weeks for the symptoms of dyspnea, cough, and pain, as measured by the EORTC QLQ-C30 and EORTC QLQ-LC13, would be tested using Holm’s procedure. The key secondary HRQoL outcomes assessed in the CodeBreaK 200 study were not statistically tested because the hierarchical testing was stopped at the non–statistically significant OS end point.

Overall Survival

In CodeBreaK 100, the proportion of observed deaths at the time of the primary data cut-off (September 1, 2020) was 38.1%. The median OS was 12.0 months (95% confidence interval [CI], 9.5 months to not evaluable [NE]). The probabilities of survival based on Kaplan-Meier (KM) estimates at 3 months, 6 months, 9 months, and 12 months were 89.5% (95% CI, 82.7% to 93.8%), 75.5% (95% CI, 66.8% to 82.2%), 63.4% (95% CI, 53.8% to 71.5%), and 51.6% (95% CI, 36.7% to 64.5%), respectively. Results for OS at the updated analysis with the data cut-off date of March 15, 2021, were generally consistent with the results from the primary data cut-off date, with median OS of 12.5 months (95% CI, 10.0 months to NE). The results for OS at 2 years, with the data cut-off date of September 9, 2022, were consistent with the results from the previous data cut-off dates, with median OS of 12.48 months (95% CI, 9.99 months to 19.29 months). The probabilities of survival based on KM estimates at 18 months and 24 months were 42.08% (95% CI, 32.97% to 50.90%) and 31.56% (95% CI, 23.15% to 40.29%), respectively.

In the CodeBreaK 200 study, the proportions of observed deaths at the time of data cut-off (August 2, 2022) were 63.7% and 54.0% in the sotorasib and docetaxel treatment arms, respectively. The CodeBreaK 200 study was not powered to detect a difference in OS. The median OS rates were 10.64 months (95% CI, 8.94 months to 13.96 months) in the sotorasib group and 11.3 months (95% CI, 9.00 months to 14.85 months) in the docetaxel group. The stratified hazard ratio (HR) for OS was 1.01 (95% CI, 0.77 to 1.33; P = 0.53) following treatment with sotorasib versus docetaxel. In total, 59 patients randomized to docetaxel crossed over to receive treatment with sotorasib; crossover occurred in 46 patients following disease progression (per protocol), while 13 patients received sotorasib as subsequent treatment upon disease progression. The results of the sensitivity analyses exploring the crossover effect on OS in the CodeBreaK 200 study were consistent with the main analysis. In the analyses based on patients who were per-protocol crossovers, the HRs for survival were 1.01 (95% CI, 0.66 to 1.49) in the rank-preserving structural failure time (RPSFT) analysis; 0.99 (95% CI, 0.73 to 1.34) in the inverse-probability-of-censoring weighting (IPCW) adjusted analysis; and 0.885 (95% CI, 0.17 to 1.33) in the 2-stage approach adjusted analysis. Sensitivity analysis of OS exploring the crossover effect among all 59 crossover patients using the 2-stage approach resulted in an HR for survival of 0.82 (95% CI, 0.14 to 1.33). KM plots of sensitivity analysis of OS exploring crossover effects are presented in Figure 20 for 46 patients who were per-protocol crossovers and in Figure 21 for all 59 patients who received sotorasib upon progression.

Progression-Free Survival

In the CodeBreaK 100 study, 56.9% of patients experienced progression or death due to any cause at the time of the September 2020 data cut-off. The median PFS was 6.7 months (95% CI, 4.9 months to 8.1 months). The probabilities of PFS at 3 months, 6 months, and 9 months were 67.5% (95% CI, 58.2% to 75.2%), 51.5% (95% CI, 41.9% to 60.4%), and 36.2% (95% CI, 26.7% to 45.8%), respectively. The sensitivity analysis for PFS using the investigator’s assessment was consistent with the results obtained by BICR. At the time of the March 2021 data cut-off, 70.2% of patients had experienced progression or death due to any cause. The median PFS rate was consistent with the earlier data cut-off (6.8 months; 95% CI, 5.1 months to 8.2 months).

Among patients in the CodeBreaK 200 study, disease progression or death on or before the data cut-off date was observed in 71.3% of patients in the sotorasib treatment group and 58.0% of patients in the docetaxel group. The median PFS was 5.62 months (95% CI, 4.27 months to 7.75 months) in the sotorasib group and 4.47 months (95% CI, 3.02 months to 5.68 months) in the docetaxel group at the time of data cut-off. The stratified HR for disease progression or death was 0.66 (95% CI, 0.51 to 0.86; P = 0.002) in favour of sotorasib versus docetaxel. The results of the sensitivity analyses for PFS using the investigator’s assessment were consistent with the results obtained by BICR.

Patient-Reported Outcomes

European Organisation for Research and Treatment of Cancer Quality of Life Questionnaire Core 30: In the CodeBreaK 100 study, 98 patients (78%) completed the EORTC QLQ-C30 at baseline. The number of patients available to complete the measure diminished consistently with each cycle. At the end of treatment, 38 patients (43.2%) completed the assessment. Over time, scores for both global health status and physical functioning appeared to remain relatively stable. Change from baseline in global health status ranged from –5.33 (SD = 15.57) (cycle 11, n = 25) to + 1.37 (SD = 19.44) (cycle 5, n = 61). Change from baseline in physical functioning ranged from –8.57 (SD = 20.33) (cycle 13, n = 14) to + 6.67 (SD = 22.05) (cycle 7, n = 33).

In the CodeBreaK 200 study, the baseline EORTC QLQ-C30 — a key secondary outcome — was completed by 168 patients (98.2%) and 158 patients (90.8%) in the sotorasib and docetaxel treatment groups, respectively. Compliance rates for the questionnaire remained consistently high, above 90%. The number of patients available to complete the measure diminished consistently with each cycle. At cycle 5, day 1 (week 12), the EORTC QLQ-C30 was completed by 69 patients (39.7%) in the docetaxel treatment group and by 106 patients (62.0%) in the sotorasib treatment group. The differences between groups in least squares change from baseline to week 12 were 6.93 (95% CI, 3.66 to 10.19) for global health status and 8.78 (95% CI, 5.39 to 12.17) for physical functioning, favouring treatment with sotorasib (n = 106) relative to docetaxel (n = 69).

EQ-5D-5L and EQ VAS: In the CodeBreaK 100 study, results for the EQ-5D-5L remained relatively stable during the treatment period. At baseline (n = 86), most patients (68% to 94%) reported that they had no problems or slight problems with the dimensions of health assessed by the EQ-5D-5L. At the end of the treatment phase (n = 28; 22.2%), 42.9% and 21.4% of patients reported that they had no problems or slight problems, respectively. EQ VAS scores remained relatively stable during the treatment period. The mean EQ VAS score at baseline was 70.2 (SD = 17.5; n = 86), indicating that patients generally rated their health favourably (with a score of 100 indicating best health imaginable). Change in mean score from baseline over the treatment period ranged from –24.0 (SD = 32.8) in cycle 15 (n = 4) to 2.7 (SD = 18.6) in cycle 3 (n = 69); and 2.7 (SD = 16.6; cycle 7, n = 31). By the end of the treatment phase, the mean EQ VAS score was 61.5 (SD = 19.5; n = 28), which was associated with a mean change from baseline of –10.6 (SD = 19.3). The number of patients available to complete both the EQ-5D-5L and the EQ VAS dropped by more than 50% after cycle 6 in the CodeBreaK 100 study.

In the CodeBreaK 200 study, a baseline EQ-5D-5L was completed by 160 patients (93.6%) in the sotorasib treatment group and 138 patients (79.3%) in the docetaxel treatment group. The number of patients available to complete the measure diminished consistently with each cycle. At cycle 5, day 1, the EQ-5D-5L was completed by 105 patients (61.4%) in the sotorasib treatment group and 67 patients (38.5%) in the docetaxel treatment group. At baseline, most patients in both the sotorasib treatment group (64.4% to 93.2%) and docetaxel treatment group (69.6% to 90.6%) reported that they had no problems or slight problems with the dimensions of health assessed by the EQ-5D-5L. At cycle 5, day 1, 79.0% to 96.2% of patients in the sotorasib treatment group and 61.2% to 85.1% of patients in the docetaxel treatment group reported that they had no problems or slight problems. Between-group differences for the change from baseline on the EQ-5D-5L scales were not reported. Baseline EQ VAS — a secondary outcome — was completed by 166 patients (97.1%) and 154 patients (88.5%) in the sotorasib and docetaxel treatment groups, respectively. Compliance rates for this measure were consistently greater than 80%. The number of patients available to complete the measure diminished consistently with each cycle. At cycle 5, day 1, the EQ VAS was completed by 106 patients (62.0%) in the sotorasib group and 69 patients (39.7%) in the docetaxel group. At baseline, mean EQ VAS scores were 67.6 (SD = 19.9) and 68.3 (SD = 20.3) in the sotorasib and docetaxel treatment groups, respectively. At cycle 5, day 1, the mean EQ VAS score for patients in the sotorasib treatment group was 73.2 (SD = 18.6), which was associated with a mean change from baseline of 2.2 (SD = 15.5) for sotorasib. For patients in the docetaxel treatment group (n = 67), the mean EQ VAS score at cycle 5, day 1 was 67.7 (SD = 20.8), which was associated with a mean change from baseline of –5.8 (SD = 18.2). Between-group differences for change from baseline on the EQ VAS were not reported.

EORTC QLQ-C30 Symptom Scales: In the CodeBreaK 100 study, baseline EORTC QLQ-C30 symptom scales were completed by 94 patients (76.4%). The number of patients available to complete the measure diminished consistently with each cycle. After cycle 6, the number of patients available to complete assessments had dropped by more than 50%. From baseline (n = 94) to the end of the treatment phase of the study (n = 32), mean scores were sustained for diarrhea (no change) and dyspnea (mean change from baseline = 3.13 [SD = 39.1]). The symptom scores for fatigue, nausea and/or vomiting, and pain increased during the study period.

In the CodeBreaK 200 study, the dyspnea subscale was considered the most important, and was the only symptom-specific scale of the EORTC QLQ-C30 reported. At cycle 2, day 1, a total of 148 patients (86.5%) in the sotorasib treatment group and 112 patients (64.4%) in the docetaxel treatment group completed the dyspnea subscale. At cycle 5, day 1, the proportions of patients available to complete the dyspnea subscale of the EORTC QLQ-C30 were 61.4% and 38.5% in the sotorasib and docetaxel treatment groups, respectively. The difference between groups in the least squares change from baseline for dyspnea at week 12 was –10.09 (95% CI, –13.39 to–6.78), favouring treatment with sotorasib (n = 105) relative to docetaxel (n = 67).

EORTC QLQ-LC13 Symptom Scales: The 3 main symptoms of lung cancer are dyspnea, cough, and chest pain; thus, these were the focus of the assessment. In the CodeBreaK 100 study, the least squares mean scores were maintained or decreased over time for the EORTC QLQ-LC13 subscales of dyspnea, cough, and chest pain from baseline (n = 86) to the end of the treatment phase at cycle 17 (n = 28). The least squares mean changes from baseline ranged from –11.2 (95% CI, –16.2 to –6.1) in cycle 2 (n = 77) to –7.1 (95% CI, –13.2 to –1.1) in cycle 6 (n = 45) for cough; from –4.9 (95% CI, –10.3 to 0.4) in cycle 7 (n = 31) to –0.44 (95% CI, –5.12 to 4.24) in cycle 6 (n = 45) for chest pain; and from –3.4 (95% CI, –7.8 to 1.0) in cycle 4 (n = 58) to –0.14 (95% CI, –5.7 to 5.41) in cycle 9 (n = 25) for dyspnea. After cycle 6, the number of patients available to complete assessments had dropped by more than 50%.

In the CodeBreaK 200 study, the dyspnea, cough, and chest pain subscales of the EORTC QLQ-LC13 were completed at baseline by 166 patients (97.1%) and 154 patients (88.5%) in the sotorasib and docetaxel treatment groups, respectively. Compliance rates were consistently high, mostly above 90%. The numbers of patients who completed the EOTRC QLQ-LC13 at cycle 5, day 1 (week 12) in the sotorasib and docetaxel treatment groups were 106 (62.0%) and 69 (39.7%), respectively. The total number of patients available to provide assessments dropped by more than 50% at cycle 6, day 1. The odds of improved symptoms at week 12 favoured sotorasib relative to docetaxel for cough (odds ratio = 3.21; 95% CI, 1.55 to 6.65) and for dyspnea (odds ratio = 3.58; 95% CI, 1.98 to 6.46). The odds of improved chest pain at week 12 favoured sotorasib compared to docetaxel (odds ratio = 1.56; 95% CI, 0.82 to 2.96).

Objective Response Rate

In CodeBreaK 100, 1.6% of patients were documented as having CR on sotorasib, while 35.8% had PR to treatment at the time of the September 2020 data cut-off. The ORR (CR plus PR) was 37.4% (95% CI, 28.84% to 46.58%). The study achieved the predetermined threshold for a positive outcome (lower limit of the 95% CI for ORR > 23%). The mean time to objective response was 1.95 months (SD = 1.23 months). At the time of the March 2021 data cut-off, 3.2% of patients were documented as having CR, while 33.9% had PR. The ORR (CR plus PR) was consistent with the earlier data cut-off (37.1%; 95% CI, 28.60% to 46.23%). The mean time to objective response as observed at the updated analysis was 2.11 months (SD = 1.71 months).

In the CodeBreaK 200 study, the proportions of patients documented as having an objective response (CR or PR) to treatment at the time of data cut-off in the sotorasib and docetaxel groups were 28.1% (95% CI, 21.5 to 35.4) and 13.2% (95% CI, 8.6 to 19.2), respectively, resulting in a difference in proportion of 14.8% (95% CI, 6.4 to 23.1; P < 0.001) in favour of sotorasib. The odds of objective response were higher among patients randomized to the sotorasib group relative to those randomized to docetaxel (odds ratio = 2.6; 95% CI, 1.48 to 4.56; P < 0.001). The mean time to objective response as of the August 2, 2022, data cut-off date was 2.43 months (SD = 1.80 months) among patients in the sotorasib treatment group and 3.24 months (SD = 2.08 months) in the docetaxel treatment group. Of note, there was an imbalance across groups in the proportion of patients for whom response assessments were not done (5.8% and 24.1% of patients in the sotorasib and docetaxel groups, respectively). The sponsor clarified that the majority of patients who did not have ORR measured did not receive docetaxel because of withdrawal of consent or other reasons.

Duration of Response

In the CodeBreaK 100 study, a DOR of at least 3 months and of at least 6 months was documented in 76.1% and 50.0% of patients, respectively, at the primary data cut-off (September 1, 2020). The median DOR was 8.4 months (95% CI, 6.9 months to 8.4 months). The probabilities of response based on KM estimates at 3 months and 6 months were 89.9% (95% CI, 75.3% to 96.1%) and 76.2% (95% CI, 59.1 to 86.9), respectively. Results for DOR at the updated analysis with the data cut-off date of March 15, 2021, were consistent overall with the results from the primary data cut-off, with DORs of at least 3 months and 6 months documented in 82.6% and 60.9% of patients, respectively. The median DOR was 11.1 months (95% CI, 6.9 months to NE), and the probabilities of objective response based on KM estimates at 3 months and 6 months were 90.5% (95% CI, 76.7 to 96.3) and 70.8% (95% CI, 54.3 to 82.2), respectively. At 9 months and 12 months, the probabilities of objective response were 57.3% (95% CI, 40.4 to 71.0) and 48.2% (95% CI, 31.5 to 63.0), respectively.

In the CodeBreaK 200 study, the proportions of patients with DORs of at least 3 months and 6 months were not available to the CADTH review team. The median DORs based on BICR of disease response per RECIST v1.1 were 8.64 months (95% CI, 7.06 months to 17.97 months) and 6.80 months (4.27 months to 8.28 months) in the sotorasib and docetaxel treatment groups, respectively. No comparative effect estimate was provided.

Harms Results

Adverse Events

At the September 1, 2020, data cut-off date in the CodeBreaK 100 study, a total of 125 patients (99.2%) in the NSCLC cohort had AEs. Overall, 75 patients (59.5%) had grade 3 or worse AEs, and 21 patients (16.1%) had grade 4 or worse AEs. The 3 most common AEs reported were diarrhea (49.2%), nausea (29.4%), and fatigue (24.6%). AEs reported at the subsequent updated analysis with the data cut-off date of March 15, 2021, were consistent overall with those from the primary data analysis. At the time of the updated analysis, 125 patients (99.2%) reported at least 1 AE; among these, 77 patients (61.1%) had grade 3 or worse AEs and 24 patients (19.0%) had grade 4 or worse AEs. The 3 most common AEs reported at the time of the updated analysis were diarrhea (50.8%), nausea (31.0%), and fatigue (25.4%).

In the CodeBreaK 200 study, 98% of patients in both the sotorasib and docetaxel treatment groups reported at least 1 AE. The most common AEs reported in the sotorasib treatment group were diarrhea (41.4%), nausea (26.0%), and decreased appetite (23.1%). The most common AEs reported in the docetaxel treatment group were fatigue (29.8%), diarrhea (25.8%), and nausea (25.8%). Alopecia occurred in 3 patients (1.8%) in the sotorasib group and 35 patients (23.2%) in the docetaxel group. Grade 3 or worse AEs were reported by 121 patients (71.6%) in the sotorasib treatment group and 91 patients (60.3%) in the docetaxel treatment group. Grade 4 or worse AEs were reported in 48 patients (28.4%) and 35 patients (23.2%) in the sotorasib and docetaxel treatment groups, respectively. The most common grade 3 or worse AEs in the sotorasib treatment group were diarrhea (13.6%), NSCLC (10.1%), increased alanine aminotransferase (ALT) (8.3%), and increased aspartate aminotransferase (AST) (5.9%). In the docetaxel treatment group, the most common grade 3 or worse AEs were neutropenia (8.6%), anemia (6.6%), fatigue (6.0%), pneumonia (6.0%), and febrile neutropenia (5.3%).

Serious Adverse Events

In the CodeBreaK 100 study, a total of 63 patients (50%) in the NSCLC cohort had at least 1 serious adverse event (SAE) at the time of primary data analysis (September 1, 2020). The most frequently reported SAEs were pneumonia (7.1%), NSCLC (6.3%), and pleural effusion (4.8%). Of those reporting SAEs, 60 patients (47.6%) had a grade 3 or worse AE and 19 patients (15.1%) had a grade 4 or worse AE. SAEs at the time of the subsequent updated analysis with the March 15, 2021, cut-off date were documented in 69 patients (54.8%). The grades of SAEs at the time of the updated analysis were not available to the CADTH review team. The most frequently reported SAEs at the time of the updated analysis were pneumonia (7.1%), NSCLC (6.3%), and pleural effusion (4.0%).

In the CodeBreaK 200 study, SAEs were reported by 91 patients (53.8%) in the sotorasib treatment group and by 67 patients (44.4%) in the docetaxel treatment group. The most frequently reported SAE in the sotorasib treatment group was NSCLC (10.7%). The most frequently reported SAE in the docetaxel treatment group was pneumonia (6.6%).

Withdrawals Due to AEs

A total of 11 patients (8.7%) in the NSCLC cohort of the CodeBreaK 100 study discontinued treatment due to AEs at the time of primary data analysis (September 1, 2020) and subsequent updated analysis (March 15, 2022). The most common reasons for treatment discontinuation were drug-induced liver injury (2.4%), increased ALT (1.6%), increased AST (1.6%), and pneumonitis (1.6%) at the time of both the primary and updated analyses.

In the CodeBreaK 200 study, 28 patients (16.6%) and 24 patients (15.9%) in the sotorasib and docetaxel treatment groups, respectively, discontinued treatment due to AEs. Reasons for discontinuing treatment in the sotorasib treatment group included increased ALT (3.6%), NSCLC (3.0%), increased blood bilirubin (2.4%), increased AST (1.2%), asthenia (1.2%), and increased blood alkaline phosphatase (ALP) (1.2%). In the docetaxel treatment group, reasons for discontinuing treatment included fatigue (2.0%), NSCLC (1.3%), anemia (1.3%), febrile neutropenia (1.3%), pneumonitis (0.7%), and asthenia (0.7%).

Dose Modification Due to AEs

In the CodeBreaK 100 study, a total of 48 patients (38.1%) and 22 patients (17.5%) required dose modification of sotorasib due to AEs at the time of the primary data analysis (September 1, 2020) and the subsequent updated analysis (March 15, 2021), respectively. Reasons for dose modification were consistent at both data analysis points. The most common reasons for dose modification due to AE were diarrhea (8.7%), increased AST (8.7%), increased ALT (8.7%), and increased blood ALP (4.0%).

In the CodeBreaK 200 study, dose modifications due to AEs were required by 26 patients (15.4%) and 43 patients (28.5%) in the sotorasib and docetaxel treatment groups, respectively. The most frequently reported reason for dose modification in the sotorasib treatment group was diarrhea (8.3%), followed by increased ALT (3.6%), increased AST (1.8%), and increased blood ALP (0.6%). In the docetaxel treatment group, the most frequently reported reasons for dose modification were neutropenia (4.6%) and fatigue (4.0%), followed by asthenia (3.3%) and diarrhea (2.0%).

Dose Interruption

In the CodeBreaK 100 study, dose interruptions due to AEs were documented in 46 patients (36.5%) at the time of the March 15, 2021, data cut-off date.

In the CodeBreaK 200 study, dose interruptions due to AEs were reported in 85 patients (50.9%) in the sotorasib treatment group and in 42 patients (27.8%) in the docetaxel treatment group. The most frequently reported reasons for dose interruption in the sotorasib treatment group were diarrhea (15.4%), increased ALT (5.9%), increased AST (5.3%), nausea (4.7%), decreased appetite (3.0%), and fatigue (1.2%). In the docetaxel treatment group, dose interruptions were due to pneumonia (4.6%), fatigue (3.3%), and nausea (0.7%).

Mortality

At the time of the primary data analysis at the data cut-off of September 1, 2020, in the CodeBreaK 100 study, 18 patients (14.3%) had died due to an AE. At the time of the subsequent, updated analysis on March 15, 2021, 20 patients (15.9%) had died due to an AE. None of the deaths were considered attributable to treatment-related AEs.

In the CodeBreaK 200 study, fatal AEs were recorded in 37 patients (21.9%) in the sotorasib treatment group and in 18 patients (11.9%) in the docetaxel treatment group. Seventeen patients (10.1%) in the sotorasib treatment group and 5 patients (3.3%) in the docetaxel group experienced fatal AEs related to NSCLC disease progression.

Notable Harms

At the time of the CodeBreaK 100 primary data analysis on September 1, 2020, a total of 40 patients (31.7%) in the NSCLC cohort reported hepatotoxicity, with the most common documented as increased AST (21.4%), increased ALT (20.6%), and increased blood ALP (13.5%). Renal toxicity was reported in 21 patients (16.7%) in the NSCLC cohort, and included hyponatremia (5.6%) and hypoalbuminemia (3.2%), hyperkaliaemia (2.4%), hypocalcemia (2.4%), hypophosphatemia (1.6%), and increased blood creatine (0.8%). There were no documented cases of interstitial lung disease. Pneumonitis was reported in 3 patients (2.4%) in the NSCLC cohort. Peripheral neuropathy was reported in 1 patient (0.8%) in the NSCLC cohort. Lastly, gastrointestinal (GI) toxicity was reported in 93 patients (73.8%) in the NSCLC cohort, and included diarrhea (49.2%), nausea (29.4%), and vomiting (18.3%). Reported notable harms at the subsequent updated analysis with the data cut-off date of March 15, 2021, were consistent overall with those from the primary data analysis. At the updated data analysis on March 15, 2021, a total of 40 patients (31.7%) reported hepatoxicity, with the most common documented as increased AST (21.4%), increased ALT (20.6%), and increased blood ALP (13.5%). Renal toxicity was reported in 23 patients (18.3%) and included hyponatremia (7.1%), hypoalbuminemia (3.2%), increased blood creatine (0.8%), hyperkalemia (2.4%), hypocalcemia (3.2%), and hypophosphatemia (2.4%). There were no documented cases of interstitial lung disease at the time of the updated analysis. Pneumonitis was reported in 3 patients (2.4%) in the NSCLC cohort. Peripheral neuropathy was reported in 1 patient (0.8%) in the NSCLC cohort. The overall number of patients with GI toxicity was not reported at the time of the updated analysis. The following notable AEs related to GI toxicity were reported: diarrhea (50.8%), nausea (31.0%), vomiting (18.3%), constipation (19.0%), and abdominal pain (8.7%).

In the CodeBreaK 200 study, hepatoxicity AEs, including increased AST (10.7%), increased ALT (10.7%), increased blood ALP (7.7%), and increased gamma-glutamyltransferase (4.0%), were reported in the sotorasib treatment group, while in the docetaxel treatment group, 2% or less of patients reported hepatoxicity due to these reasons. Renal toxicity related to hyponatremia was reported in 8 patients (4.7%) in the sotorasib treatment group and in 4 patients (2.6%) in the docetaxel treatment group, while hypoalbuminemia was reported in 4 patients (2.4%) in the sotorasib treatment group and 8 patients (5.3%) in the docetaxel treatment group. Interstitial lung disease was reported in 1 patient in the sotorasib treatment group, and pneumonitis was reported in 3 patients (1.8%) in the sotorasib treatment group and 3 patients (2.0%) in the docetaxel treatment group. Peripheral neuropathy was documented in 1 patient (0.6%) in the sotorasib treatment group and in 16 patients (10.6%) in the docetaxel group. Notable GI toxicities reported in the sotorasib treatment group included diarrhea (41.4%), nausea (26.0%), vomiting (13.0%), constipation (13.0%), and abdominal pain (11.8%). Notable GI toxicities reported in the docetaxel treatment group included diarrhea (25.8%), nausea (24.5%), constipation (19.2%), and vomiting (9.9%).

Table 2: Summary of Key Results From CodeBreaK 100 and CodeBreaK 200 Studies

AE = adverse event; ALP = alkaline phosphatase; ALT = alanine aminotransferase; AST = aspartate aminotransferase; CI = confidence interval; CNS = central nervous system; CR = complete response; DOR = duration of response; GI = gastrointestinal; HR = hazard ratio; KM = Kaplan-Meier; NA = not applicable; NE = not evaluable; NR = not reported; NSCLC = non–small cell lung cancer; ORR = objective response rate; OS = overall survival; PFS = progression-free survival; PR = partial response; RECIST 1.1 = Response Evaluation Criteria in Solid Tumours Version 1.1; SAE = serious adverse event; SD = standard deviation; vs. = versus.

Notes: Randomization stratification factors are number of prior lines of therapy in advanced disease (1 vs. 2 vs. > 2), race (Asian vs. non-Asian), and history of CNS involvement (yes vs. no).

For patients who continued treatment postprogression or patients who crossed over from docetaxel to sotorasib, the first date of progression is used for PFS analysis; the patient’s response post–first progression or postcrossover is not used in the primary analyses to evaluate objective response or DOR.

The median OS at 2 years in the CodeBreaK 100 study with data cut-off September 9, 2022, was 12.48 (95% CI, 9.99 to 19.29). The probabilities of survival based on KM estimates at 18 months and 24 months were 42.08% (95% CI, 32.97 to 50.90%) and 31.56% (95% CI, 23.15 to 40.29%), respectively.

Survival status may include publicly available records (where permitted) searched by investigator after the patient ended the study.

^a95% CIs are based on estimated variance for log-log transformation of the KM survival estimate.

^bPFS and OS rates and 95% CIs were estimated using the method by Kalbfleisch and Prentice with log-log transformation.

^cStratification factors: number of lines of therapy in advanced disease (1 vs. 2 vs. > 2), race (Asian vs. non-Asian), history of CNS involvement (yes vs. no).

^dHR and 95% CI were estimated using a stratified Cox proportional hazard model. An HR of less than 1.0 represents a lower average death rate and longer OS for sotorasib relative to docetaxel. The P value was calculated using a stratified log-rank test.

^eThe P value was adjusted for multiple comparison using the Maurer-Bertz multiple testing procedure. The PFS and ORR hypothesis was rejected before testing OS.

^fThe P value was calculated using a stratified log-rank test.

^gThe exact 95% CI was calculated using the Clopper-Pearson method.

^hCalculated using the stratified Cochran-Mantel-Haenszel chi-square test.

^ITime to response was calculated among confirmed responders’ best overall response of PR or CR.

^jDOR was calculated among confirmed responders.

^k95% CIs were based on estimated variance for log-log transformation of the KM survival estimate.

^lFrequency greater than 0% in any treatment group.

^mIn the CodeBreaK 200 study, hepatotoxicity of interest was reported using Standardised MedDRA Query for hepatic disorders (narrow). Hepatotoxicity of interest was reported in 7 patients (4.6%) in the docetaxel group and 41 patients (24.3%) in the sotorasib group. Renal toxicity of interest was reported using Standardized MedDRA Query (acute renal failure, chronic kidney disease). Renal toxicity of interest did not occur in the docetaxel group, but was reported for the sotorasib group in 10 patients (5.9%), including acute kidney injury and renal failure in 3 patients each.

Sources: Clinical Study Report for phase II of the CodeBreaK 100 study;²⁰ Clinical Study Report for the CodeBreaK 200 study.²¹

Critical Appraisal

The single-group, open-label, nonrandomized design of the CodeBreaK 100 study makes interpretation of the efficacy and safety of sotorasib challenging. The lack of comparison with an active comparator, BSC, and/or placebo precludes the ability to draw causal inferences or assess the relative therapeutic benefits or safety of sotorasib. Interpretation of time‐to‐event end points, such as OS and PFS, is limited in single‐group studies; because all patients with KRAS G12C-mutated advanced NSCLC received the same treatment, the extent to which the observed survival is due to the natural history of the tumour, or the intervention remains unclear. The CodeBreaK 100 trial included no formal statistical significance or hypothesis testing and point estimates with 95% CIs were reported to estimate the magnitude of treatment effect. To mitigate the limitations associated with open-label studies, tumour response and disease progression outcomes were blinded to the study investigator and assessed by BICR. However, as noted by the FDA assessment of sotorasib, the analyses of PFS and OS are uninterpretable in the single-group study.²³ The results for patient-reported outcomes (PROs) were inconclusive, given the noncomparative, open-label design of the trial and the substantial decline in the number of patients available to provide assessments over time. Due to the previously mentioned limitations, the ability to draw firm conclusions on the magnitude of clinical benefit and safety of sotorasib was limited. The results were found to be generalizable, overall, to the clinical setting by the clinical experts consulted by CADTH for the purpose of this review; however, the patients in the trial were judged to be younger and healthier (i.e., with lower ECOG PS scores) than those typically seen in practice. In addition, patients with active brain metastases, who may be considered for treatment in practice, were excluded.

In the CodeBreaK 200 study, patients were randomized centrally using interactive response technology, a method that is typically adequate for concealing allocation until treatment assignment. While the randomization stratification factors appear to be appropriate, several between-group imbalances at baseline suggest that prognostic balance may not have been achieved between groups. Of concern was the proportion of patients in the docetaxel treatment group who were randomized but not dosed (13.2%), namely due to refusal of treatment, and discontinued the study by way of withdrawal of consent (22.4%). As a result, unequal censoring between the sotorasib and docetaxel treatment groups was observed (e.g., a 10% difference in censoring between the groups for OS and a 13% difference in censoring for PFS), introducing uncertainty about treatment effects. In addition, a sizable proportion of patients in the docetaxel treatment group (24%) had an outcome of “not measured” in the ORR analysis. The sponsor clarified that the majority of patients who did not have ORR measured did not receive docetaxel because of withdrawal of consent or other reasons. The extent and direction of bias associated with the ORR analysis are uncertain. The risk of attrition bias cannot be ruled out due to the disproportionate loss of patients by way of self-withdrawal. There is added uncertainty for the PFS rates measured over the study period due to the limited number of patients at risk by month 12. To minimize the risk of bias in the measurement of the outcomes associated with the open-label design, patients’ responses to treatment were blinded to the study investigator, and tumour response was confirmed by radiologic evidence and done by BICR, as per RECIST 1.1. The sensitivity analyses of PFS and ORR demonstrated consistency between the BICR and the investigator’s assessment of tumour response, suggesting that the risk for bias in response outcomes from the open-label design is likely not substantial. Objective outcomes, like OS, are not prone to bias due to knowledge of the intervention.

The CodeBreaK 200 study assessed HRQoL and symptom burden — outcomes deemed important by patients and clinicians — as key secondary outcomes. However, the key secondary HRQoL outcomes were not statistically tested because the hierarchical testing was stopped at the non–statistically significant OS end point. Of note, due to protocol amendments that allowed for crossover and subsequent reduction in sample size, the CodeBreaK 200 study was not powered to detect a difference in OS. Furthermore, the open-label nature of the CodeBreaK 200 study increases the risk of bias in the measurement of the subjective HRQoL^24,25 Missing baseline data and the low completion rates in both the sotorasib and docetaxel treatment groups add further uncertainty to the HRQoL and other symptom burden outcomes. Focusing on the docetaxel treatment group, after cycle 3, fewer than 50% of patients were available to provide assessment for the EORTC QLQ-C30, the subscales of the EORTC QLQ-LC13, and the EQ-5D-5L. Consequently, there would be no reliable assessment for the docetaxel treatment group for HRQoL and symptom burden measures. The HRQoL and symptom burden outcomes are at high risk of attrition bias because long-term survivors tend to be healthier patients. The analysis of efficacy results followed a defined statistical analysis plan, employed appropriate censoring criteria, and accounted for patient crossover from the docetaxel treatment group to the sotorasib treatment group. The sponsor conducted additional sensitivity analyses of OS that adjusted for the effect of crossover. The results of these analyses were consistent with the main analysis in that the CIs were too wide to draw any conclusions of certainty with respect to whether sotorasib or docetaxel was favoured for OS.

Indirect Comparisons

At the time of initial submission to CADTH in August 2022, data from the CodeBreaK 200 study were not available. In the absence of direct comparative data between sotorasib and docetaxel from the CodeBreaK 200 trial, the sponsor submitted a matching-adjusted indirect comparison (MAIC), which was used to inform the pharmacoeconomic model. The unanchored MAIC estimated the comparative OS and PFS between sotorasib and docetaxel in patients with locally advanced or metastatic NSCLC (stages IIB to IV) and confirmed KRAS mutation in the absence of direct comparative evidence from a randomized trial. The MAIC was based on individual patient data from the single-group, open-label index trial, the CodeBreaK 100 study, and from aggregate-level data from the docetaxel (75 mg/m²) plus placebo group of the double-blind, randomized SELECT-1 trial.²⁶ In November 2022, this submission was temporarily suspended because the sponsor informed CADTH that the economic evaluation would be revised to include data from the CodeBreaK 200 study. In May 2023, the temporary suspension was lifted upon receipt of the revised economic model from the sponsor. After receiving data from the CodeBreaK 200 study, which provided direct comparative evidence between sotorasib and docetaxel, the CADTH review team determined that the MAIC no longer addressed any gap in the evidence; therefore, the submitted MAIC was not included in the clinical report.

Other Relevant Evidence

This section summarizes the partial results of the global expanded access program (EAP). Although this study did not meet the systematic review inclusion criteria, it provides supportive evidence for patients with an ECOG PS of 2, which was a patient group excluded from the CodeBreaK 100 study and the CodeBreaK 200 study. Furthermore, the clinical expert consulted by CADTH and the input received from clinician groups for this submission also expressed a desire to treat patients with ECOG PS 2. The input received from the clinician group, LCC, for this submission indicated that benefit in patients with ECOG PS 3 and 4 remains debatable. The CADTH review team summarized the study designs and data of the global EAP to provide supplemental evidence for decision-making.

The global EAP provided for compassionate use of sotorasib before local regulatory approval. Patients eligible for the EAP were aged 18 years to 99 years with an ECOG PS equal to or less than 2, and had pathologically documented, locally advanced and unresectable or metastatic, KRAS G12C-mutated NSCLC confirmed through molecular testing. Patients had to demonstrate that they had exhausted other standard of care options for locally advanced and unresectable or metastatic NSCLC, including platinum-based combination chemotherapy and PD-1 and/or PD-L1 immunotherapy (unless medically contraindicated). Patients were excluded from participating if they were participating in any ongoing clinical study of sotorasib; if they had mixed small cell lung cancer or mixed NSCLC histology, active brain metastases, active hepatitis B or hepatitis C virus, or current active malignancy other than NSCLC; and if they had previously enrolled in a prior sotorasib study. The primary end points included safety (such as treatment-emergent adverse events [TEAEs], AEs of interest, and SAEs). Key secondary end points included OS and treatment duration, with real-world PFS as the ad hoc end point. The sponsor-submitted abstract, poster, and presentation summarized the data from 2 global protocols under the EAP (Amgen studies 20190236 [Study 436] and 20190442 [Study 442]) that evaluated the safety and efficacy of sotorasib outside the registrational trial setting in patients with advanced KRAS G12C-mutated NSCLC in multiple countries (US, Spain, Argentina, Brazil, Israel, Saudi Arabia, and Taiwan) across 49 centres.^27,28 Real-world PFS was estimated for Study 436 and defined as the time from the start of treatment to end of protocol due to disease progression or death, any death before new anticancer therapy, or end of commercial sotorasib, whichever occurred first.

A total of 147 patients received sotorasib 960 mg daily under EAP Study 436. The median number of prior lines of anticancer therapy reported by patients before receiving sotorasib was 2 (up to 8). At baseline, a total of 25.2% of patients had an ECOG PS of 2, and 32.7% of patients had brain metastases.

Efficacy Results

Real-world PFS was estimated for the 92 patients enrolled in Study 436 of the global EAP study.²⁷ The median real-world PFS as of the data cut-off date of June 24, 2022, was 6.7 months (95% CI, 4.6 months to 8.3 months) with 60 events (65.2%). OS was estimated for 147 patients. The median OS as of the data cut-off date of November 8, 2022, was 9.5 months (95% CI, 8.6 months to 12.0 months). The median real-world PFS was comparable to the PFS efficacy results reported in the CodeBreaK 100 study and the CodeBreaK 200 study.

Harms Results

The global EAP study²⁷ presented AEs as TEAEs; thus, these were not reported by the CADTH review team.

Critical Appraisal

The CADTH review team was unable to complete a robust critical appraisal of the internal and external validity of the study results. A key limitation to the global EAP study is its noncomparative design and lack of statistical testing, precluding causal conclusions. The study was open-label, and the method for ascertaining PFS or OS was not reported; therefore, the potential for and extent of any bias in the measurement of this outcome is unknown. Although the results were comparable to the PFS efficacy results reported in the CodeBreaK 100 and CodeBreaK 200 trials, the magnitude of the treatment effect for the real-world PFS should be interpreted with uncertainty in light of these limitations. The study did not report on outcomes other than PFS and OS that are important to patients, clinicians, and drug plans (e.g., HRQoL, symptoms, harms).

Conclusions

One nonrandomized, single-group, open-label, phase I and II study (the CodeBreaK 100 study) and 1 randomized, comparative, open-label, phase III study (the CodeBreaK 200 study) provided evidence for the efficacy and safety of sotorasib in adult patients with KRAS G12C-mutated, advanced NSCLC. In the CodeBreaK 100 study, the results for ORR and DOR were considered clinically meaningful in the target population by the clinical experts consulted by CADTH. There was uncertainty around the magnitude of the clinical benefit, given the limitations in the evidence from the noncomparative phase I and II clinical trial. The nonrandomized design of the CodeBreaK 100 study made interpreting the PFS and OS events attributable to sotorasib challenging because it is unclear whether the observations were due to natural history or the effect of treatment with sotorasib. In the CodeBreaK 200 study, treatment with sotorasib demonstrated a beneficial impact on PFS and ORR compared to docetaxel. Although the difference in median PFS of approximately 1 month was modest, the clinical experts consulted by CADTH felt that the PFS rates for sotorasib at 12 months appeared encouraging and clinically meaningful. The clinical experts agreed that the impact of sotorasib on objective response was clinically meaningful. There was uncertainty around the treatment effect observed in the CodeBreaK 200 study because a larger proportion of patients in the docetaxel treatment group versus the sotorasib group withdrew from the trial before receiving treatment and because of unequal censoring between the sotorasib and docetaxel treatment groups. The CIs associated with the treatment effect of sotorasib relative to docetaxel on OS in the CodeBreaK 200 study were too wide to determine if either treatment was favoured. The results for HRQoL and symptom severity were descriptive in nature and remained inconclusive due to diminishing numbers of patients available to complete the assessments over time. The clinical experts consulted by CADTH noted that the TEAEs observed with sotorasib could be adequately managed in clinical practice and that sotorasib appeared to have a favourable toxicity profile compared to docetaxel.

Introduction

Disease Background

Lung cancer is 1 of the most diagnosed cancers, and NSCLC accounts for 80% to 88% of all lung cancer diagnoses in Canada.^1,2 In 2020, the incidence of NSCLC in Canada was estimated to be 60.5 per 100,000 persons.³ In 2021, an estimated 29,600 new cases of lung cancer and 21,150 deaths due to lung cancer were projected.⁴ Survival from lung cancer across all stages and histologies is poor, with a 5-year net survival rate of 22%.⁵ NSCLC often remains asymptomatic until the disease is in its advanced stages.^6,7 Because early diagnosis is challenging,⁹ approximately two-thirds of patient have advanced disease at diagnosis, when curative treatments are not possible.^4,10 Indeed, 20% of patients are diagnosed with locally advanced disease (stage III), and 50% are diagnosed with metastatic disease. Median OS is poor in patients with metastatic NSCLC (stages IVA and IVB), ranging from 8 months to 11 months; 5-year OS ranges from 4% to 6%. The 5-year net survival rate for stage IV NSCLC is 5.2%.¹¹ Among patients with metastatic disease, the most common sites of metastases are brain, liver, and bone. Among those patients with advanced NSCLC who are eligible for treatment, approximately 4% may die each week while waiting for therapy initiation.²⁹

NSCLC is classified according to histologic subtypes, with the most common being adenocarcinoma and squamous cell carcinoma. Squamous cell cancer is a subtype of NSCLC that usually starts in the cells that line the bronchi in the centre of the lungs, whereas the adenocarcinoma subtypes typically start in the glandular cells on the outer part of the lung. In addition, NSCLC often holds an oncogenic driver mutation that leads to uncontrolled cell growth and proliferation. Gene mutations discovered in NSCLC include anaplastic lymphoma kinase, epidermal growth factor receptor, ROS proto-oncogene 1, and RAS.¹² Mutations within the RAS family account for more than 30% of all mutated oncogenes in NSCLC, causing approximately 1 million deaths worldwide annually.¹³ Approximately 1 in 4 patients with NSCLC harbour KRAS mutations.¹⁴ KRAS is the isoform most frequently altered in NSCLC. It is most commonly found in patients who smoke or have formerly smoked, in Western populations (versus Asian populations), and in patients with nonsquamous histology. Patients with KRAS G12C-mutated NSCLC have a lower proportion of response to cytotoxic chemotherapy and decreased survival compared to the overall population of patients with NSCLC.¹⁵ The KRAS G12C subtype represents almost half of all KRAS mutations in NSCLC and is identified in approximately 13% of patients with nonsquamous NSCLC.^16,17 Based on an estimation in the Health Canada Reviewers Report, the incidence of patients living in Canada with NSCLC harbouring the KRAS G12C mutation was approximately 7.9 per 100,000 persons.¹⁸ The KRAS G12C mutation does not usually occur in the presence of other known driver mutations in NSCLC.³⁰

When patients with NSCLC present with symptoms, these are usually nonspecific and difficult to attribute to lung cancer.⁷ The most common symptoms include fatigue, cough, chest, or shoulder pain, hemoptysis, weight loss, dyspnea, hoarseness, bone pain, and fever.⁷ Advanced NSCLC is associated with a higher prevalence and intensity of symptoms such as pain, dyspnea, cough, decreased appetite, weight loss, and depression, as well as lower HRQoL compared to other advanced cancers.⁸ When lung cancer is suspected, diagnostic procedures include imaging with CT, PET, and/or MRI scans, bronchoscopy with or without endobronchial ultrasound bronchoscopy, and tissue biopsy.³¹ Pathologic testing of biomarkers on lung biopsy specimens assists in determining treatment options and risk stratification.

Standards of Therapy

According to the clinical experts consulted by CADTH for this review, the goals of treatment in the noncurative setting are preserving or improving quality of life and extending quantity of life.

The current treatment algorithm for locally advanced (not amenable to curative treatment) or metastatic NSCLC without a driver mutation is first-line immunotherapy with or without platinum-doublet chemotherapy. Based on input from the clinical experts consulted by CADTH for this review, immunotherapy-based regimens are superior to chemotherapy alone. The clinical experts agreed that chemotherapy by itself would be considered as a treatment option only for patients who are ineligible for immunotherapy; therefore, sequencing chemotherapy followed by an immunotherapy would be extremely rare. The clinical experts noted that for patients who have progressed on immunotherapy and a platinum-doublet chemotherapy, standard therapy options are docetaxel or BSC. However, docetaxel has low response rates and is quite toxic. The clinical experts highlighted that while KRAS G12C-mutated lung adenocarcinomas do benefit from immunotherapy, patients who progress on immunotherapy have limited treatment options and limited life expectancy.

The clinical experts anticipated that sotorasib will be used to replace docetaxel (i.e., clinicians would first exhaust platinum-based chemotherapy and immunotherapy options before moving on to sotorasib). The clinical experts noted that there is insufficient evidence at this time to inform alternative sequencing (e.g., immunotherapy in the first line, sotorasib in the second line, followed by platinum-based chemotherapy or docetaxel).

Drug

Sotorasib is a highly selective inhibitor of KRAS G12C that suppresses the rapid proliferation of cancer cells. Sotorasib forms an irreversible, covalent bond with the unique cysteine of KRAS G12C, locking the protein in an inactive state that prevents downstream signalling without affecting wild-type KRAS. Sotorasib blocks KRAS signalling, inhibits cell growth, promotes apoptosis, and is associated with antitumour inflammatory responses and immunity in KRAS G12C tumour models.

On September 10, 2021, sotorasib received an NOC/c from Health Canada for the treatment of adult patients with KRAS G12C-mutated, locally advanced (not amenable to curative therapy) or metastatic NSCLC who have received at least 1 prior systemic therapy. The Health Canada market authorization with conditions is pending the results of trials to verify sotorasib’s clinical benefit. Sotorasib underwent advance consideration under NOC/c, as well as Project Orbis, at Health Canada. The sponsor’s reimbursement request is aligned with the Health Canada indication. Health Canada recommends that sotorasib be administered at a dose of 960 mg (8 × 120 mg tablets) orally once daily until disease progression or unacceptable toxicity. A maximum of 2 dose reductions are permitted in the case of AEs.

The product monograph for sotorasib documents the following common side effects: diarrhea, musculoskeletal pain, nausea, fatigue, hepatotoxicity, and cough. The product monograph also contains warmings for serious side effects, including hepatotoxicity and interstitial lung disease/pneumonitis.

Submission History

At the time of the initial submission to CADTH in August 2022, data from the CodeBreaK 200 study were not available. In the absence of a head-to-head comparison of sotorasib and docetaxel, the sponsor submitted an indirect treatment comparison in the form of an unanchored MAIC, which was used to inform the pharmacoeconomic model. The unanchored MAIC estimated the comparative OS and PFS between sotorasib and docetaxel in patients with locally advanced or metastatic NSCLC (stages IIB to IV) and confirmed KRAS mutation in the absence of direct comparative evidence from a randomized trial. The MAIC was based on individual patient data from the single-group, open-label index trial, the CodeBreaK 100 study, and aggregate-level data from the docetaxel 75 mg/m² plus placebo group of the double-blind, randomized SELECT-1 trial.²⁶ In November 2022, this submission was temporarily suspended because the sponsor informed CADTH that the economic evaluation would be revised to include data from the CodeBreaK 200 study. In May 2023, the temporary suspension was lifted upon receipt of the revised economic model from the sponsor. After receiving data from the CodeBreaK 200 study, which provided direct evidence comparing sotorasib and docetaxel, the CADTH review team determined that the MAIC no longer addressed any gap in the evidence, given that the CodeBreaK 200 study was a phase III, head-to-head, randomized controlled trial (RCT) comparing the efficacy and safety of sotorasib and docetaxel. Therefore, the MAIC was not included in the clinical report.

Stakeholder Perspectives

Patient Group Input

This section was prepared by CADTH staff based on the input provided by patient groups. The full original patient input received by CADTH is included in the stakeholder section at the end of this report.

The LHF (formerly the Ontario Lung Association), LCC, and the CCSN submitted joint input. The LHF is a registered charity that provides programs and services to patients and health care providers, invests in lung research, and advocates for improved policies in lung health. LCC, a member of the Global Lung Cancer Coalition, is a registered national charitable organization that serves as a resource for education, patient support, research, and advocacy, focusing exclusively on lung cancer. The CCSN is a national network of patients, families, friends, community partners, and sponsors who promote the best standard of care, patient support, survivorship, and quality of end-of-life care. A total of 5 people (4 patients and 1 caregiver) responded to a telephone survey interview from August to September 2022. Of the 4 patients with lung cancer, 3 had experience with sotorasib; 1 patient with KRAS G12C-positive lung cancer did not. All respondents were female. All but 1 were from Canada (Nova Scotia, British Columbia, Quebec, and Ontario); the remaining respondent was from the US.

From the submitted input, 1 patient said they felt “rock bottom” when their lung cancer progressed on chemotherapy and radiation, leading them to be on oxygen, to acquire a debilitating cough as well as shortness of breath from eating or talking, and to require assistance to shower. Another patient detailed the mental and emotional side effects associated with multiple courses of immunotherapy and chemotherapy, such as depression, anxiety, panic attacks, and severe mood swings, which remained debilitating even while in remission. The sole caregiver interviewed stated that it is mentally, physically, and financially challenging to take care of a patient with lung cancer with comorbidities (e.g., preparing meals, making arrangements for transportation to medical appointments, managing daily responsibilities) and that they feel burned out without many sources of support.

Three patients who had experience with sotorasib reported significant tumour reduction (i.e., ranging from a 50% to 65% reduction to no evidence of disease within 5 weeks to 1.5 years). They said they experienced mild side effects while on sotorasib, such as fatigue, aches, and pains when walking for extended periods of time, a small or minor rash, diarrhea, shortness of breath, and increased liver enzymes. However, they said these side effects did not have much impact on their daily activities or quality of life. According to the input, these patients felt hopeful again that they could plan for the future.

In terms of important key outcomes, 1 patient said they were most interested in obtaining a cure while maintaining a good quality of life. Respondents also expressed their hope for a treatment option with an oral route of administration that would be accessible from home, delay the onset of symptoms, prolong their life, and improve functionality and mobility, with fewer side effects. The caregiver expressed the importance of treatment that can be accessed from home, limiting the need to travel to infusion clinics. Lastly, it was pointed out in the survey that wait times for lung surgeries across Canada are unacceptable; patients and caregivers would like to treat the lung cancer in the early stages. The caregiver surveyed said that the wait time for the surgery was the most difficult aspect (or the most challenging adverse effect) of current treatment.

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 NSCLC.

Unmet Needs

According to the clinical experts, treatment options for patients with KRAS G12C-mutated NSCLC who have progressed on standard therapy are limited. The clinical experts noted that KRAS was once believed to be undruggable. A new class of drugs, including sotorasib, has provided targeted options for patients with KRAS G12C-mutated NSCLC. However, given that no publicly funded, targeted drugs are currently accessible, therapy options are limited to docetaxel, which is associated with high levels of toxicity. Therefore, there is an unmet need for an efficacious treatment associated with fewer AEs. The clinical experts were careful to note that lung cancer is a stigmatized disease due to its association with smoking. The availability of a targeted treatment in a population of patients who smoke or have smoked is unprecedented. The clinical experts also expressed a need for a targeted therapy that can be administered orally versus an IV drug that requires patients to travel to cancer centres. Finally, in this patient population with high rates of intracranial metastases, 1 clinical expert expressed the need for a treatment that would also be able to treat brain metastases.

Place in Therapy

According to the clinical experts consulted for this review, sotorasib would change the sequence of therapies for patients with KRAS G12C-mutated NSCLC. The clinical experts expect that sotorasib would be used following immunotherapy and platinum-doublet chemotherapy. The clinical experts noted that sotorasib is anticipated to displace docetaxel, but not platinum-based chemotherapy (i.e., clinicians would exhaust platinum-based chemotherapy and immunotherapy options before moving on to sotorasib). Accordingly, docetaxel would move to being a third- or fourth-line therapy. Based on feedback from the clinical experts consulted by CADTH for this review, there is insufficient evidence at this time to inform alternative sequencing (e.g., immunotherapy in first-line therapy and sotorasib in second-line therapy, followed by platinum-based chemotherapy or docetaxel), given that patients enrolled in the CodeBreaK 200 study were required to have progressed on a PD-1 or PD-L1 inhibitor and platinum-based chemotherapy; accordingly, there is no phase III evidence to support the use of sotorasib as second-line treatment after first-line, single-drug PD-1 or PD-L1 inhibitors. The clinical experts noted that sotorasib would not be combined with other drugs at this time.

Patient Population

Patients with KRAS G12C-mutated NSCLC are identified through NGS. According to the clinical experts, NGS is standard of care for adenocarcinomas of the lung, and any patient with nonsquamous histology has reflexive molecular testing performed to evaluate for any mutations.

The clinical experts suggested that sotorasib be limited to patients with KRAS G12C-mutated NSCLC. They stated that sotorasib should not be extended to other KRAS G12C-mutated NSCLC because the biology of the G12C mutation is distinct from those of other KRAS subtypes. Moreover, treatment with sotorasib is not suitable for patients with an ECOG PS of 3 or 4, with severe organ dysfunction, or with untreated, symptomatic brain metastasis. The clinical experts did note that patients with untreated, asymptomatic brain lesions may be suitable for treatment with sotorasib; ideally, these patients should have their cases discussed at a multidisciplinary tumour board round at a centre with expertise in stereotactic radiosurgery.

Assessing Response to Treatment

Based on input from the clinical experts, patients should undergo clinical and toxicity assessments per cycle (typically every 3 weeks to 4 weeks) and imaging approximately every 3 months to assess their response to treatment in clinical practice. Based on input from the clinical experts, improved PFS, improved OS, and maintenance or improvement in quality of life are considered meaningful responses to treatment in this population.

Discontinuing Treatment

Based on input from the clinical experts, treatment with sotorasib should be discontinued under the following 3 scenarios: patient decision to stop treatment with sotorasib; unacceptable toxicity due to sotorasib; and disease progression without clinical benefits. The clinical experts agreed that patients with documented disease progression could continue sotorasib if they were deriving clinical benefit.

Prescribing Conditions

The clinical experts suggested that sotorasib be prescribed by a medical oncologist in an outpatient oncology clinical setting.

Clinician Group Input

This section was prepared by CADTH staff based on the input provided by clinician groups. The full original clinician group input received by CADTH has been included in the stakeholder section at the end of this report.

Input was received from 2 clinician groups: the OH-CCO LC DAC and LCC. The OH-CCO LC DAC provides clinical and health system guidance on drug-related issues in support of CCO’s mandate, including the provincial drug reimbursement programs and the systemic treatment program. Input from the OH-CCO was provided by 3 clinicians who had joint discussions through email. LCC is a national charity that increases awareness on lung cancer, advocates for and educates patients with lung cancer and their families, and supports research on and access to the best care for lung cancer. Input from LCC was compiled by 26 clinicians and was based on a review of the literature and proceedings from recent conferences.

Both clinician groups identified the following goals of therapy: reducing tumour burden; improving symptoms; and prolonging life while upholding patients’ values and desires. Both clinician groups also noted that all patients eventually progress on currently available treatment; thus, durability of response is an important treatment goal. In addition, LCC added the need for treatment that is associated with reduced toxicity and resource utilization. Indeed, OH-CCO LC DAC endorsed the advent of an oral anticancer treatment without the side effects and life impacts associated with chemotherapy to slow disease progress and ideally improve length and quality of life. According to input from the OH-CCO LC DAC, sotorasib would be placed as second- or third-line monotherapy for patients who have progressed on or are unable to tolerate platinum-based chemotherapy and immunotherapy (where appropriate) and possibly docetaxel. The OH-CCO LC DAC clinician group stated that sotorasib would not affect the use of immunotherapy or platinum-doublet chemotherapy; however, sotorasib may be preferred to docetaxel if it can demonstrate a meaningful improvement in survival or quality of life, given that docetaxel has a proven survival benefit compared to BSC and other chemotherapy regimens (ifosfamide), albeit with an unfavourable side effect profile. Input from LCC places sotorasib as second-line standard of care for patients with advanced KRAS G12C-mutated NSCLC and suggests that sotorasib be used as a single drug after at least 1 line of prior systemic treatment has failed. Based on input from both clinician groups, patients with a KRAS G12C-mutated, advanced NSCLC (stage IV or recurrent) who received prior therapy would be best suited for sotorasib. The LCC added that it is uncertain whether patients with an ECOG PS of 3 or 4 would benefit from treatment with sotorasib. Both clinician groups noted that patients eligible for treatment with sotorasib should be identified by a validated molecular diagnostic test, preferably NGS. Both clinician groups noted that improvement in symptoms, stable disease, tumour shrinkage, radiographic reduction of disease site from baseline, and prolonged survival are indicative of a clinically meaningful response to treatment. The OH-CCO LC DAC added that the definition of a clinically meaningful improvement in frequency or severity of symptoms varies from 1 patient to another and across physicians. Based on input from LCC, response to treatment should be determined by the treating physician based on CT imaging and assessed every 2 months to 3 months, as is done in Canada when using other oral tyrosine kinase inhibitors. Based on input from the clinician groups, discontinuing treatment with sotorasib should be considered in the event of loss of clinical benefit (as indicated by unequivocal disease progression, including symptoms, polyprogression, and so on) or intolerable side effects. The LCC noted that treatment with sotorasib may continue in patients with oligoprogression who are amenable to local therapy (radiation or surgery); who are newly diagnosed or have experienced progression of brain metastases and have been treated with brain radiation or surgery; and who have asymptomatic disease (without overt progression on imaging associated with increased symptom burden). The OH-CCO LC DAC emphasized that a combination of clinical judgment, radiological interpretation, and clinically standardized scales, such as the Edmonton Symptom Assessment System, should be used to determine if a patient is benefiting from therapy. The OH-CCO-LC also noted that RECIST 1.1 and its derivatives are designed for clinical trials, not clinical practice, and should not be used to determine discontinuation of treatment. Both clinician groups agreed that outpatient clinics would be an appropriate treatment setting for sotorasib in addition to the hospital setting. Both groups stated that a medical oncologist, general practitioner in oncology under medical oncology supervision, or respirologist experienced in treating patients with lung cancers should be involved in diagnosing, treating, and monitoring patients on sotorasib.

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

ALT = alanine aminotransferase; AST = aspartate aminotransferase; CYP3A4 = cytochrome P450 3A4 enzyme; ILD = interstitial lung disease; pERC = CADTH pan-Canadian Oncology Drug Review Expert Review Committee; P-gp = permeability glycoprotein; PPI = proton pump inhibitor.

Clinical Evidence

The clinical evidence included in the review of sotorasib 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 those studies that were selected according to an a priori protocol. The second section would normally include indirect evidence from the sponsor and indirect evidence selected from the literature that met the selection criteria specified in the review; however, for this review, no relevant indirect evidence was identified. The third section includes an additional relevant study that was considered to address an important gap in the evidence included in the systematic review.

Systematic Review (Pivotal and Protocol-Selected Studies)

Objective

To perform a systematic review of the beneficial and harmful effects of sotorasib 960 mg (8 × 120 mg oral tablets) for the treatment of KRAS G12C-mutated, locally advanced (not amenable to curative therapy) or metastatic NSCLC in adult patients who have received at least 1 prior systemic therapy.

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 those considered important to patients, clinicians, and drug plans.

Table 4: Inclusion Criteria for the Systematic Review

AE = adverse event; DOR = duration of response; ECOG = Eastern Cooperative Oncology Group; GI = gastrointestinal; HRQoL = health-related quality of life; NSCLC = non–small cell lung cancer; ORR = objective response rate; OS = overall survival; PFS = progression-free survival; RCT = randomized controlled trial; SAE = serious adverse event; 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.³²

Published literature was identified by searching the following bibliographic databases: MEDLINE through Ovid and Embase through 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. Search concepts were developed based on the elements of the Patients, Interventions, Comparisons, and Outcomes (PICO) framework and research questions. The main search concept was Lumakras or sotorasib. The following clinical trial registries were searched: the US National Institutes of Health’s clinicaltrials.gov, the WHO International Clinical Trials Registry Platform 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 September 19, 2022. Regular alerts updated the search until the meeting of the CADTH pan-Canadian Oncology Drug Review Expert Committee on September 12 and 13, 2023.

Grey literature (literature that is not commercially published) was identified by searching relevant websites from CADTH’s Grey Matters: A Practical Tool For Searching Health-Related Grey Literature checklist.³³ Included in this search were the websites of regulatory agencies (i.e., the 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.

These searches were supplemented through contacts with appropriate experts. In addition, the sponsor was contacted for information regarding unpublished studies.

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 2 studies were identified from the literature for inclusion in the systematic review (Figure 1). The included studies are summarized in Table 5. A list of excluded studies is presented in Appendix 2.

Table 5: Details of Included Studies

AE = adverse event; ALK = anaplastic lymphoma kinase; ALT = alanine aminotransferase; ANC = absolute neutrophil count; AST = aspartate aminotransferase; AUC = area under the curve; BPI-SF = Brief Pain Inventory – Short Form; CHF = congestive heart failure; Cmax = maximum plasma concentration; CNS = central nervous system; CR = complete response; CRC = colorectal cancer; DCR = disease control rate; DOR = duration of response; ECOG PS = Eastern Cooperative Oncology Group Performance Status; EGFR = epidermal growth factor receptor; EORTC QLQ-C30 = European Organisation for Research and Treatment of Cancer Quality of Life Questionnaire Core 30; EORTC QLQ-LC13 = European Organisation for Research and Treatment of Cancer Quality of Life Questionnaire Lung Cancer Module; EQ VAS = EQ visual analogue scale; FACT-G = Functional Assessment of Cancer Therapy – General; FPE = first patient enrolled; GI = gastrointestinal; HRQoL = health-related quality of life; INR = international normalized ratio; IP = investigational product; LPE = last patient enrolled; MI = myocardial infarction; MDRD = Modification of Diet in Renal Disease; NR = not reported; NSCLC = non–small cell lung cancer; NSCLC-SAQ = Non–Small Cell Lung Cancer Symptom Assessment Questionnaire; OS = overall survival; PD-1 = programmed cell death 1; PD-L1 = programmed cell death 1 ligand 1; PFS = progression-free survival; PFS 2 = progression-free survival on next line of treatment; PGIC = patient global impression of change; PGIS = patient global impression of severity; PK = pharmacokinetic; PR = partial response; PRO-CTCAE = patient-reported outcome version of the Common Terminology Criteria for Adverse Events; PTT = partial thromboplastin time; QTc = correct heart rate interval; RCT = randomized controlled trial; RECIST 1.1 = Response Evaluation Criteria in Solid Tumours Version 1.1; ROS1 = ROS proto-oncogene 1 receptor tyrosine kinase; TPS = Tumour Proportion Score; TTR = time to response; ULN = upper limit of normal.

^aThree additional reports were included: the Clinical Study Report for phase II of the CodeBreaK 100 study;²⁰ Clinical Study Report for the CodeBreaK 200 study,^21,36 and Health Canada Reviewers Report.¹⁸

Only inclusion and exclusion criteria applied to the indication population are reported. The list of exclusion criteria for the CodeBreaK 200 study represents key criteria and is not exhaustive.

Sources: Clinical Study Report for phase II of the CodeBreaK 100 study;²⁰ Clinical Study Report for the CodeBreaK 200 study.²¹

Description of Studies

Two sponsor-conducted studies were included in this systematic review: the CodeBreaK 100 study and the CodeBreaK 200 study.

CodeBreaK 100 Study

The CodeBreaK 100 study was the pivotal study submitted by the sponsor. It is an ongoing, phase I and II multicentre, nonrandomized, open-label, single-group study.^19,20 Phase I was a first-in-human dose exploration (part 1) and dose expansion (part 2) study aimed at evaluating the safety, tolerability, PKs, and pharmacodynamics of sotorasib in adult patients with KRAS G12C-mutated advanced NSCLC, colorectal cancer, and other solid tumours.¹⁹ Phase II was designed to evaluate the efficacy and safety of sotorasib as monotherapy in adult patients with KRAS G12C-mutated advanced NSCLC, colorectal cancer, and other solid tumours.²⁰ A schematic of the CodeBreaK 100 study design is illustrated in Figure 2. CADTH’s review of the CodeBreaK 100 study focuses on the phase II efficacy and safety results in adult patients with KRAS G12C-mutated, advanced NSCLC.

Figure 1: Flow Diagram for Inclusion and Exclusion of Studies

A total of 224 patients with KRAS G12C-mutated, advanced solid tumours were enrolled at 59 sites across 11 countries, including 8 patients (3.6%) across 2 sites in Canada. At enrolment, each patient received a unique study identification number, which was provided on all study documentation and used to identify the patient throughout the clinical study. The study identification number was assigned by an interactive response technology. Of the patients enrolled, 126 patients (56.3%) had NSCLC. Sotorasib was dispensed at the study centre by a qualified staff member. On study visit days, patients were required to take sotorasib at the study centre. At all other times, patients self-administered sotorasib at home. Patients self-administered sotorasib 960 mg (8 × 120 mg tablets) orally once daily and continued treatment without interruption until disease progression, treatment intolerance, withdrawal of consent, or death. Tumour response was assessed using contrast-enhanced CT or MRI and per RECIST 1.1 through BICR using an independent radiological central laboratory. Patients underwent safety follow-up visits 30 days (± 7 days) after the last dose of sotorasib or before any new anticancer treatment was started. Following safety follow-up visits, patients were followed long-term for health condition, disease status, and subsequent anticancer treatment every 12 weeks ( ±  2 weeks) for up to 3 years after the last patient was enrolled or until withdrawal of consent, loss to follow-up, or patient death, whichever occurred first. The study team was blinded to the efficacy data.

The primary efficacy end point for phase II of the CodeBreaK 100 study was ORR, which was a composite of CR and PR. Secondary efficacy end points included DOR, DCR, TTR, PFS, and OS. Exploratory outcomes included changes in cancer-specific symptoms and patient-reported HRQoL.

An independent data review team conducted interim safety reviews after 30, 50, 70, and 100 patients had been enrolled and treated with sotorasib for at least 21 days. The data cut-off date for the primary analysis was determined on the assumption that most responders would experience a response in the first 1.5 months to 3 months from the start of treatment. The data cut-off date for phase II was September 1, 2020. Interim and updated analyses were conducted on December 1, 2020, and March 15, 2021, respectively. The results of the primary and updated analysis of March 15, 2021, are presented in this review because the results from the December 1, 2020, analysis were consistent with both the primary and updated analyses. The updated data cut-off date for OS at 2 years was September 9, 2022, and is also presented.

Figure 2: Schematic of CodeBreaK 100 Study

AMG 510 = sotorasib; BID = twice daily; DLRT = dose level review team; IP = investigational product; NSCLC = non–small cell lung cancer; QD = once daily; RP2D = recommended phase II dose.

Note: The DLRT was responsible for reviewing data in the dose expansion phase to confirm the RP2D and determine the benefit to risk ratio of proceeding to phase II of the study, according to the futility and efficacy thresholds described in the statistical analysis section.

Source: Clinical Study Report for phase II of the CodeBreaK 100 study.²⁰

CodeBreaK 200 Study

The CodeBreaK 200 study is an ongoing, multicentre, randomized, open-label, parallel-group, phase III study evaluating the efficacy and safety of oral sotorasib versus IV docetaxel in adult patients with KRAS G12C-mutated NSCLC who progressed after prior platinum-based chemotherapy and a checkpoint inhibitor.²¹ A schematic of the CodeBreaK 200 study design is illustrated in Figure 3.

Figure 3: Schematic of CodeBreaK 200 Study

AMG 510 = sotorasib; CNS = central nervous system; EOS = end of study; Q3W = every 3 weeks; QD = once daily; vs. = versus.

^a Treatment with investigational product continued until independent central confirmation of progression, intolerance of treatment, initiation of another anticancer

therapy, withdrawal of consent, or death. Patients who consented to treatment beyond progression or consented to cross over from docetaxel to sotorasib continued to

receive the investigational product after independent central confirmation of progression at the time of first progressive disease.

^b Upon permanent discontinuation from the study treatment for any reason, a safety follow-up visit was performed 30 days (± 7 days) after the end of the last dosing

interval of the investigational product.

Source: Clinical Study Report for the CodeBreaK 200 study.²¹

A total of 345 patients with KRAS G12C-mutated, advanced solid tumours were enrolled at 148 sites across 22 countries, including 4 patients (1.16%) at 3 sites in Canada. Patients were randomized 1:1 to receive oral sotorasib 960 mg once daily (n = 171) or IV docetaxel 75 mg/m² every 3 weeks (n = 174). Randomization was stratified by number of prior lines of therapy in advanced disease (1 versus 2 versus > 2), race (Asian versus non-Asian), and history of CNS involvement (present or absent). Each of the following were considered as 1 line of prior therapy: chemoradiation for locally advanced and unresectable disease, if progression occurred within 6 months of end of treatment; chemoradiation followed by planned systematic therapy, including checkpoint inhibitor, or vice versa without documented intervening progression, if progression occurred within 6 months of end of treatment; adjuvant therapy, if progression occurred within 6 months of end of treatment; and each new systematic anticancer regimen for progressive locally advanced and unresectable or metastatic disease. Maintenance therapy and chemotherapy regimen adjustments for intolerability were not considered new lines of therapy. At enrolment, each patient received a unique study identification number that was provided on all study documentation and used to identify the patient throughout the clinical study. The study identification number was assigned by an interactive response technology. The first day a patient received study medication was noted as cycle 1, day 1. Each treatment cycle was 21 (± 3) days. Patients continued treatment without interruption until disease progression, treatment intolerance, withdrawal of consent, or death. Tumour response was assessed using contrast-enhanced CT or MRI and per RECIST 1.1 using an independent radiological central laboratory. Patients randomized to the docetaxel treatment group who experienced radiologic disease progression according to RECIST 1.1 were given the opportunity to cross over into the sotorasib treatment group. Patients randomized to the sotorasib treatment group could continue study treatment following radiologic progression if they continued to demonstrate clinical benefit. Patients who consented to treatment beyond progression or to cross over from docetaxel to sotorasib continued to receive the investigational product after independent central confirmation of progression at the time of first progressive disease. Patients who discontinued treatment before RECIST 1.1 disease progression were followed with tumour assessment until independent central confirmation of progression, withdrawal of consent, or start of another anticancer therapy, then followed for subsequent anticancer therapy and survival.

The primary end point for the CodeBreaK 200 study was PFS as assessed by BICR per RECIST 1.1. Key secondary end points included OS, ORR, and cancer-specific HRQoL measures, while secondary outcomes included DCR, disease- and treatment-related symptoms, and other patient-reported HRQoL measures.

There were 2 planned PFS efficacy analyses. These were conducted when approximately 160 (interim) and 230 (primary) PFS events occurred. An interim analysis of OS was planned for superiority of sotorasib over docetaxel, either at the time when PFS achieved statistical significance or after 175 OS events were observed, whichever occurred later. The primary OS analysis was to occur when at least 198 OS events had been observed. The primary analysis of ORR was to be performed when PFS was claimed to be statistically significant and the last randomized patient had had the opportunity to have at least 12 weeks of follow-up. Patients will be followed for 5 years after the last patient is enrolled, or until withdrawal of consent, loss to follow-up, or death, whichever occurs first. The data cut-off date for the CodeBreaK 200 study was August 2, 2022, at which point the results from the planned primary analysis of PFS were analyzed.

Protocol Amendments

The protocol for the CodeBreaK 100 study was amended 6 times. The phase II portion of the CodeBreaK 100 study was added with amendment 2. The protocol-defined statistical analyses were amended twice. In amendment 3 (May 22, 2019), the sample size was adjusted based on a new benchmark rate (i.e., 23%) to exclude from the lower limit of the 95% CI for observed ORR; and the futility interim analysis was updated to allow for continuous monitoring using the Bayesian predictive probability method. In amendments 3 and 4, PROs were added. In amendment 5, the planned sample size was increased to 250 patients from 200 patients. In amendment 6, the frequency of tumour assessments was increased, and language was added to allow treatment with sotorasib to continue beyond disease progression in patients who continue to have clinical benefit, in the investigator’s opinion.

The protocol for the CodeBreaK 200 study was amended 4 times. In amendment 3 (February 15, 2021), after initiation of the study and per regulatory guidance, the initial planned enrolment of 650 patients (powered for the secondary end point of OS) was reduced to 330 patients (powered for the primary end point of PFS), and crossover from docetaxel to sotorasib was permitted following disease progression. The PFS interim analysis, planned at approximately 70% information fraction (by which time about 160 PFS events had been observed from both groups), was also incorporated at this time. Other amendments included administrative changes and clarifications.

Populations

Inclusion and Exclusion Criteria

The key inclusion and exclusion criteria for phase II of the CodeBreaK 100 study and for the CodeBreaK 200 study are summarized in Table 5. Briefly, patients eligible for inclusion in both studies were aged 18 years or older with advanced KRAS G12C-mutated solid tumours identified by molecular testing. For patients with NSCLC, KRAS G12C mutation was confirmed by central testing through therascreen KRAS Rapid Gast Quenching polymerase chain reaction from Qiagen before enrolment. For inclusion, patients must have had at least 1 prior line of anticancer therapy, progressed on the prior line of therapy, have measurable disease per RECIST 1.1 criteria, have an ECOG PS score of 1 or less, have a corrected heart rate interval of 470 msec or less, have the ability to take oral medications, and have adequate hematological, renal, hepatic, and coagulation laboratory assessments. Patients with active brain metastases from nonbrain tumours were excluded from the study. Of note, patients in the CodeBreaK 200 study must have progressed on immunotherapy and platinum-based chemotherapy, whereas patients in the CodeBreaK 100 study must have progressed after receiving anti-PD-1 or anti-PD-L1 immunotherapy, unless contraindicated, and/or platinum-based combination chemotherapy and targeted therapy, if actionable oncogenic driver mutations were identified (i.e., epidermal growth factor receptor, anaplastic lymphoma kinase, or ROS proto-oncogene 1).

Baseline Characteristics

Baseline demographics, disease characteristics, and prior treatments of patients enrolled in the CodeBreaK 100 study and the CodeBreaK 200 study are summarized in Table 6.

Patients enrolled in the CodeBreaK 100 study had a mean age of 62.9 years (SD = 9.3 years). Most patients were aged 18 years to 64 years (53.2%), white (81.7%), and current or former smokers (92.9%). In regard to disease characteristics, most patients were assessed as having stage IV disease at initial diagnosis (61.9%) and screening (96.0%). At the time of screening, metastatic disease was identified in 96.8% of enrolled patients, with the most common site of metastasis found in the bone (48.4%). The most common histology type among enrolled patients was nonsquamous adenocarcinoma (95.2%). A total of 42.9%, 34.9%, and 22.2% of patients received 1, 2, or 3 prior lines of anticancer therapy, respectively. The most common types of anticancer therapy were immunotherapy checkpoint inhibitors (92.1%) and chemotherapy (91.3%). A total of 81.0% of patients received both platinum-based chemotherapy and anti-PD-1 or anti-PD-L1 immunotherapy.

Patients enrolled in the CodeBreaK 200 study had a mean age of approximately 63 years (SD range, 9.1 to 9.9 years). Most patients were aged 18 years to 64 years (53.9%), white (82.9%), and current or former smokers (96.2%). In regard to disease characteristics, metastatic disease was identified in 95.1% of randomized patients. The most common histology type among the randomized patients was nonsquamous (96.8%). A total of 42.9%, 40.9%, and 16.2% of patients received 1, 2, or 3 or more complete prior lines of therapies, respectively; 34.2% were on maintenance therapy. Overall, 25.7% and 20.1% of patients had received prior treatment with platinum-based chemotherapy in the sotorasib and docetaxel treatment groups, respectively, while 7.6% and 5.7%, respectively, had received non–platinum-containing chemotherapy. Prior treatment with immunotherapy with platinum and alone was received by 37.4% and 25.1% of patients in the sotorasib group, respectively, and by 39.7% and 30.5% of patients in the docetaxel group, respectively. Liver, bone, and brain metastases were documented in 17.5%, 47.4%, and 33.9% of patients in the sotorasib treatment group, respectively, and in 20.1%, 39.7%, and 34.5% of patients in the docetaxel treatment group, respectively. Between the 2 treatment groups, a greater proportion of patients were male in the sotorasib treatment group (63.7%) relative to the docetaxel treatment group (54.6%). Other imbalances in baseline characteristics between the sotorasib and docetaxel treatment groups were noted: PD-L1 protein expression greater than or equal to 1% and less than 50% (sotorasib = 26.9%; docetaxel = 40.2%); ECOG PS of 0 at cycle 1, day 1 (sotorasib = 38.6%; docetaxel = 33.9%); primary refractory to last prior line of therapy (sotorasib = 39.2%; docetaxel = 32.8%); and initial response with subsequent growth (sotorasib = 20.5%; docetaxel = 27.0%).

Baseline patient characteristics were generally similar between the CodeBreaK 100 study and the CodeBreaK 200 study.

Interventions

Intervention

In both the CodeBreaK 100 study and the CodeBreaK 200 study, patients received 960 mg sotorasib (8 × 120 mg tablet) orally once daily in a fasted state (i.e., no food or liquid other than water 2 hours before to 1 hour after dosing). In the CodeBreaK 200 study, a cycle was defined as 21 days in length, unless a delay was medically necessary. Sotorasib was dispensed in clinic by a qualified staff member. Patients were instructed to take sotorasib at approximately the same time each day (± 2 hours) without interruption. Patients were instructed to skip a dose if 6 hours had passed from the scheduled time of dosing. A dose could be replaced if the patient had vomited within 15 minutes of dosing, and if all the tablets administered were accounted for and intact by visual inspection.

In the event of toxicity or AEs, up to 2 dose reductions (when sotorasib was given as monotherapy) were permitted. Patients were required to permanently discontinue sotorasib if they experienced an AE that required dose reductions to less than 240 mg of sotorasib as monotherapy.

Treatment with sotorasib was discontinued in the event of a toxicity that warranted stopping of treatment, in the opinion of the investigator. Patients were permanently discontinued from treatment with sotorasib in the event that:

• AST or ALT was 3 times the upper limit of normal (ULN) and the international normalized ratio was 1.5 times the ULN in the presence of no important alternative causes for evaluated AST or ALT values.

• AST or ALT was 3 times the ULN, and total bilirubin was 2 times the ULN, in the presence of no important alternative causes of elevated AST or ALT and/or total bilirubin values.

In the CodeBreaK 100 study and the CodeBreaK 200 study, treatment with sotorasib following disease progression was permitted in patients who continued to have clinical benefits from the therapy, based on investigator’s judgment, providing the following protocol-defined criteria were met:³⁷

• The patient provided a separate informed consent beyond disease progression.

• There was no threat to vital organs or anatomic sites requiring urgent alternative medical intervention.

• No other treatment discontinuation criteria were met.

• There were no significant, unacceptable, or irreversible toxicities related to any dose of sotorasib or any treatment-related AEs (based on the Common Terminology Criteria for Adverse Events, grade 4) at the current dose at the time of progression.

• There were no current signs or symptoms of clinical disease progression that decreased the ECOG PS to greater than 2.

• The patient had disease progression in terms of small metastases or oligometastases, or the patient had disease progression and was a candidate for stereotactic body radiotherapy, palliative radiation, or surgery.

• The patient was willing to undergo a biopsy of a progressing lesion. (If tumour biopsy was not clinically feasible or advisable, the patient could continue only upon agreement by the investigator and Amgen Inc. medical monitor.)

• The Amgen Inc. medical monitor approved the continuation.

• Palliative radiation and/or surgery for new, progressive, or symptomatic lesions was permitted, but sotorasib was required to be withheld during these treatments.

• If the time between the patient’s initial progression scan and the restart of sotorasib was more than 4 weeks, the scan had to be repeated to serve as the new baseline.

Continued growth of tumour(s) on subsequent scans resulted in permanent discontinuation from sotorasib.

Table 6: Summary of Baseline Characteristics for Patients Enrolled in CodeBreaK 100 and CodeBreaK 200 Studies

CNS = central nervous system; CR = complete response; ECOG PS = Eastern Cooperative Oncology Group Performance Status; EGFR = epidermal growth factor receptor; KEAP1 = Kelch-like ECH-associated protein 1; NR = not reported; PD-1 = programmed cell death 1; PD-L1 = programmed cell death 1 ligand 1; SD = standard deviation; VEGF = vascular endothelial growth factor.

Note: The data cut-off dates were September 1, 2020, and March 15, 2021, for the CodeBreaK 100 study and August 2, 2022, for the CodeBreaK 200 study.

^aBaseline ECOG PS is measured at screening visit predose. Patients may satisfy the ECOG PS enrolment eligibility during the screening period, but subsequently have a baseline ECOG PS of 2 before the first dose.

^bEach patient may have multiple prior therapies. Types of prior anticancer therapies were adjudicated and included therapies given in any treatment setting.

^cPlatinum-based chemotherapy and anti-PD-1 or anti-PD-L1 could be in combination or across different lines.

^dNumber of prior lines and best response on prior lines of therapy include therapies in metastatic disease and adjuvant therapy immediately before metastasis with progression occurring by or within 6 months of treatment ending.

Sources: Clinical Study Report for phase II of the CodeBreaK 100 study²⁰ and Clinical Study Report for the CodeBreaK 200 study.²¹

Comparator

No comparator intervention was used in the CodeBreaK 100 study.

The comparator intervention in the CodeBreaK 200 study was docetaxel. Docetaxel was administered by IV at a dose of 75 mg/m² every 3 weeks following the failure of prior platinum-based chemotherapy and immune checkpoint inhibitors. Docetaxel was infused in clinic by a qualified staff member over 1 hour. Additional administration time was allowed, as per local guidelines. Dose reduction of docetaxel was permitted, based on investigator opinion, in the event of select toxicities, including: grade 3 or higher related thrombocytopenia associated with bleeding; febrile neutropenia; neutrophils greater than 500 cells/mm³ for more than 1 week; grade 4 anemia; absolute neutrophil count of less than 1,500 cells/mm³; severe or cumulative cutaneous reactions; other grade 3 or 4 nonhematological toxicities; grade 3 or higher peripheral neuropathy; bilirubin greater than the ULN or — if AST and/or ALT is greater than 1.5 times the ULN, concomitant with ALP — greater than 2.5 times the ULN; grade 3 or greater AST or ALT elevation related to docetaxel for more than 2 weeks; concurrent AST or ALT greater than 3 times the ULN and total bilirubin greater than 2 times the ULN if related to docetaxel; and severe hypersensitivity. Patients could continue docetaxel after radiologic progression if the investigator believed they would continue to derive clinical benefit from treatment and if the criteria outlined earlier for continuing treatment with sotorasib after disease progression were met.

Crossover From Docetaxel to Sotorasib

Patients randomized to the docetaxel treatment group in the CodeBreaK 200 study who discontinued docetaxel due to disease progression were allowed to cross over to the sotorasib group at the discretion of the principal investigators. Treatment with sotorasib after crossover could not be initiated any earlier than 14 days after the last dose of docetaxel, regardless of time of disease progression. Patients who entered the long-term follow-up phase of the CodeBreaK 200 study were not permitted to cross over into the sotorasib treatment group if they received any systematic anticancer therapy other than docetaxel.

The CodeBreaK 200 study employed 2 crossover conditions. Crossover condition 1 required patients who permanently discontinued chemotherapy due to an AE, or for any reason other than progressive disease, to have had radiologic progression and undergo independent central confirmation of progression at the time of first progressive disease before crossover. Crossover after progressive disease was considered based on the following conditions:

• Disease progression limited to CNS only must be treated before crossover.

• Patients with symptomatic brain metastasis could not cross over.

• There was no threat to vital organs or critical anatomic sites (e.g., CNS metastasis, respiratory failure due to tumour compression, spinal cord compression) requiring urgent alternative medical intervention.

• Brain imaging (CT or MRI with contrast) was completed before initiating crossover therapy with sotorasib. If a brain scan was performed more than 28 days from the date of the first crossover dose of sotorasib, then the brain scan was repeated.

• All inclusion criteria requirements were met on administration day 1 of crossover treatment with sotorasib.

• No treatment discontinuation criteria were met.

• No significant, unacceptable, or irreversible toxicities related to any dose of docetaxel or treatment-related AEs at the current dose at the time of progression were reported (with the exception of alopecia and nail discoloration). Additional nonsignificant toxicities were allowed, but had to be approved by the medical monitor before crossover.

• No current signs or symptoms of clinical disease progression that have decreased the ECOG PS to greater than 2 were observed.

• Palliative radiation and/or surgery for new, progressive, or symptomatic lesions was permitted before crossover, provided that treatment with sotorasib was withheld until after completion of these treatments and until 7 days after completion of radiation.

• No other systemic anticancer directed therapeutic drugs were required during crossover.

Crossover condition 2 required consideration of immediate crossover for all patients who were randomized to receive docetaxel, should an early efficacy of the study be noted by the data monitoring committee.

Concomitant Treatment

In both the CodeBreaK 100 study and the CodeBreaK 200 study, patients were permitted to receive any concomitant medications or treatment deemed necessary by the investigator to provide adequate supportive care. Medications and supplements that were known to be strong inducers of cytochrome P450 3A4 enzyme (e.g., St. John’s wort), or that were known cytochrome P450 3A4 enzyme–sensitive substrates with a narrow therapeutic window, were not provided for 14 days before enrolment or during the study period unless approved by the principal investigator and the sponsor’s medical monitor. In the CodeBreaK 200 study, coadministration of sotorasib with proton pump inhibitor and H2 receptor antagonists was to be avoided; when this was not possible, then sotorasib was to be administered 4 hours before or 10 hours after a local antacid.

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. These end points are further summarized here. A detailed discussion and critical appraisal of the outcome measures is provided in Appendix 5.

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

AE = adverse event; BPI-SF = Brief Pain Inventory – Short Form; CR = complete response; DOR = duration of response; EORTC QLQ-C30 = European Organisation for Research and Treatment of Cancer Quality of Life Questionnaire Core 30; EORTC QLQ-LC13 = European Organisation for Research and Treatment of Cancer Quality of Life Questionnaire Lung Cancer Module; EQ VAS = EQ visual analogue scale FACT-G = Functional Assessment of Cancer Therapy – General; HRQoL = health-related quality of life; NA = not assessed; NSCLC-SAQ = Non–Small Cell Lung Cancer Symptom Assessment Questionnaire; OS = overall survival; PFS = progression-free survival; PGIC = patient global impression of change; PGIS = patient global impression of severity; PR = partial response; PRO-CTCAE = patient-reported outcome version of the Common Terminology Criteria for Adverse Events.

Sources: Clinical Study Report for phase II of the CodeBreaK 100 study;²⁰ Protocol and Clinical Study Report for the CodeBreaK 200 study.^21,36

Overall Survival

OS was designated a secondary outcome measure in phase II of the CodeBreaK 100 study and a key secondary outcome in the CodeBreaK 200 trial. The outcome was defined as the time from the start of treatment (the CodeBreaK 100 study) or time of randomization (the CodeBreaK 200 study) until the event of death due to any cause. Patients were censored at the date on which they were last known to be alive. If the last known date alive was after the date that triggered the analysis (i.e., the data cut-off date), patients were censored at the analysis trigger date. In the CodeBreaK 200 study, OS was estimated regardless of the subsequent anticancer therapy (for patients who started new anticancer therapies during the trial) and regardless of crossover from the docetaxel to sotorasib group.

Progression-Free Survival

PFS was a secondary outcome measure in phase II of the CodeBreaK 100 study and was designated as the primary end point in the CodeBreaK 200 study. PFS was defined as the interval from the start of treatment (the CodeBreaK 100 study) or time of randomization (the CodeBreaK 200 study) to disease progression or death due to any cause, whichever came first. The censoring strategies used in the CodeBreaK 100 study and the CodeBreaK 200 study are described in Appendix 3. Patients who did not progress or die were censored at their last date of evaluable disease assessment (including those who started a new anticancer therapy before a PFS event). Progression was based on independent radiologic assessment by BICR of disease response per RECIST 1.1. Among patients who continued treatment postprogression or crossed over from docetaxel to sotorasib in the CodeBreaK 200 study, the first date of progression was used for analysis, and patients’ response to post–first progression or postcrossover were not used in the analysis of this end point.

Health-Related Quality of Life

In the CodeBreaK 100 study, the following HRQoL outcome measures were assessed as exploratory outcomes: EORTC QLQ-C30; EORTC QLQ-LC13; and the EQ-5D-5L and EQ VAS.

In the CodeBreaK 200 study, the EORTC QLQ-C30 subscales of physical functioning and global health status and the EORTC QLQ-LC13 subscales of dyspnea, cough, and chest pain were assessed as key secondary outcomes, while the EQ VAS portion of the EQ-5D-5L was assessed as a secondary outcome.

The EORTC QLQ-C30 is a self-reported, multidimensional, cancer-specific, self-administered questionnaire for evaluating HRQoL. The questionnaire consists of 30 questions, including 5 function scales (physical, role, cognitive, emotional, and social), 1 global health status/global quality of life scale, 3 symptom scales (fatigue, pain, nausea, and vomiting), and 6 single items that assess additional symptoms (dyspnea, appetite loss, sleep disturbance, constipation, and diarrhea) and financial impact.³⁸ Scales and single items of the questionnaire range in score from 0 to 100, with higher scores on the functional and global health status/quality of life scales indicating higher levels of functioning and health status/quality of life, respectively. Higher scores on the symptom scales or items represent a greater presence of symptoms.³⁸ The construct and convergent validity of the EORTC QLQ-C30 has been deemed good in a cohort of patients with NSCLC who are not amenable to curative or life-prolonging treatments.²⁵ All items included in the scale except cognitive function and pain showed acceptable reliability.^25,39 The CADTH review team did not identify an estimated minimal important difference (MID) for the EORTC QLQ-C30 among patients with NSCLC. Estimations of within-patient meaningful improvement and worsening on key scores that were used by the sponsor are summarized in Table 8.

Table 8: Meaningful Improvement and Worsening on Key Scales of the EORTC QLQ-C30

EORTC QLQ-C30 = European Organisation for Research and Treatment of Cancer Quality of Life Questionnaire Core 30; QoL = quality of life.

Sources: Cocks et al. (2012);⁴⁰ Maringwa et al. (2011).⁴¹

The EORTC QLQ-LC13 is a responsive questionnaire specific to lung cancer that is supplementary to the EORTC QLQ-C30 and consists of 13 items addressing symptoms associated with lung cancer and its standard treatment. Symptoms mentioned in the EORTC QLQ-C30 include cough (1 question), hemoptysis (1 question), dyspnea (3 questions), and pain (3 questions); pain medication is also included (1 question). Treatment-related side effects mentioned include sore mouth or tongue (1 question), dysphagia (1 question), neuropathy (1 question), and alopecia (1 question). All items are scored on a 4-point categorical scale ranging from 1 (not at all) to 4 (very much), except for 1 pain-related item that has a dichotomous response category of yes or no. Scale and items scores are transformed on a scale ranging from 0 to 100, with higher scores representing increased symptom burden.⁴² The construct validity has been established between pain score and disease type and between ECOG PS and scores for dyspnea, coughing, and pain; correlation between spirometry results and dyspnea scores was found to be weak.⁴² The internal consistency estimated for the dyspnea scale has been confirmed to be acceptable (i.e., Cronbach alpha > 0.7), while the internal consistency for pain was found to be unacceptable when used alone, without the EORTC QLQ-C30 core questionnaire pain items.³⁹ The CADTH review team did not identify an estimated MID for the EORTC QLQ-LC13 among patients with NSCLC.

The EQ-5D-5L is a generic, self-reported HRQoL outcome measure that may be applied to a variety of health conditions and treatments.⁴³ Its first 2 components assesses 5 domains: mobility, self-care, usual activities, pain or discomfort, and anxiety or depression.⁴³ Each domain has 5 levels: no problem; slight problems; moderate problems; severe problems; and extreme problems. The descriptive system classifies respondents (aged ≥ 12 years) based on the following 5 dimensions: mobility, self-care, usual activities, pain or discomfort, and anxiety or depression. The EQ-5D-5L has 5 possible levels for each domain, and respondents are asked to choose the level that reflects their health state for each of the 5 domains.⁴³ The second component of the EQ-5D-5L is a 20 cm EQ-VAS that has end points labelled 0 and 100, with respective anchors of “worst imaginable health state” and “best imaginable health state.” Respondents are asked to rate their health by drawing a line from an anchor box to the point on the EQ-VAS that best represents their health on that day. The EQ-5D-5L index score is generated by applying a multiattribute utility function to the descriptive system.⁴⁴ Different utility functions are available that reflect the preferences of specific populations (e.g., in the US or UK). Scores less than 0 represent health states that are valued by society as being worse than dead, while scores of 0 and 100 are assigned to the health states “dead” and “perfect health,” respectively. The measurement properties have not been assessed in patients with NSCLC. The CADTH review team did not identify an estimated MID for the EQ-5D-5L or EQ VAS among patients with NSCLC.

Symptom Burden

The following symptom burden scales were assessed as exploratory outcomes in both the CodeBreaK 100 study and the CodeBreaK 200 study:

• FACT-G

• PGIS and PGIC

• PRO-CTCAE.

The CodeBreaK 100 study also assessed the NSCLC-SAQ and the symptom-specific scales of the EORTC QLQ-C30 and the EORTC QLQ LC13 as exploratory outcomes. The Brief Pain Inventory – Short Form (BPI-SF) was assessed as an exploratory outcome in the CodeBreaK 200 study.

The NSCLC-SAQ is a self-reported, symptom-based PRO instrument that consists of 7 scale items assessing 1 of 5 symptom concepts of NSCLC: cough, pain, dyspnea, fatigue, and appetite. The recall period is 1 week. Each of the 7 items is scored on a 5-point verbal rating scale from 0 to 4 (from “not at all” to “very severe” for 3 items assessing intensity and from “never” to “always” for 4 items assessing frequency). The total NSCLC-SAQ score is the sum of all 5 domain scores, ranging from 0 to 20. Higher scores are indicative of more severe symptomatology. Of note, if any domain score is missing, the total score is not calculated.^45,46 The NSCLC-SAQ has been found to be valid: content validity was ensured during the developmental phase; convergent validity has been ensured using the National Comprehensive Cancer Network Functional Assessment of Cancer Therapy Lung Symptom Index 17-Item Version, EORTC QLQ-C30, EORTC QLQ-LC13, and the EQ-5D-5L (Spearman rank-order r = 0.3 to 0.5); and known-group validity has been ensured with other questionnaires [P < 0.0001 to 0.001]). It has also been found to be reliable (i.e., Cronbach alpha > 0.7 for internal consistency and interclass correlation coefficient > 0.7 for reliability, based on test-retest 1 week to 2 weeks apart) and responsive (to improvement over 18 weeks; P < 0.001).^45,47,48 Improvement in symptoms is indicated by a decrease in total score of 3 to 5 points, which was identified using categories of correlated anchor variables (PGIS for lung cancer).⁴⁸

FACT-G is a cancer-specific, preference-based HRQoL instrument that can be used across a variety of tumour types. FACT-G item GP5 asks patients to respond to the statement, “I am bothered by side effects of treatment” on a 5-point Likert scale from 0 (not at all) to 4 (very much).⁴⁹ No validation information was identified by the CADTH review team. The MID for item GP5 is unknown.

The PGIS and PGIC are collected for the symptoms of cough, dyspnea, and chest pain to provide context for the interpretation of key lung cancer symptoms. Patients are asked to rate their experiences of any cough, chest pain, and shortness of breath in the preceding week as being none, mild, moderate, or severe. They are then asked to compare their current experience of cough, chest pain, and shortness of breath to their experiences of these at the beginning of the study and report whether they were now much better, a little better, about the same, a little worse, or much worse. The CADTH review team did not identify validation information or an estimated MID for the PGIS and PGIC among patients with NSCLC.

The PRO-CTCAE is a PRO measure designed to evaluate symptomatic toxicity in patients enrolled in clinical trials of cancer drugs.⁵⁰ Based on the phase I dose exploration portion of the CodeBreaK 100 study, the following questions were selected for assessment in the CodeBreaK 100 study and the CodeBreaK 200 study: mouth sores (severity, interference), mouth cracking (severity), itchy skin (severity), numbness (severity, interference), pain (frequency, severity, interference), aching muscles (frequency, severity, interference), and aching joints (frequency, severity, and interference). Patients are asked to assess the frequency, severity, and/or interference of adverse effects. The recall period for the PRO-CTACE is the preceding 7 days. The CADTH review team did not identify validation information or an estimated MID for the PRO-CTCAE among patients with NSCLC.

The BPI-SF is a self-administered assessment tool used to measure the severity of pain and its impact of functioning.⁵¹ The BPI-SF consists of 9 questions targeting the occurrence of pain; areas of pain; rating of pain at its worst in the last 24 hours; rating of pain at its least in the last 24 hours; specifying the average pain level; specifying the current pain level; specifying the treatments or medications currently being taken; specifying the percentage of pain relief obtained from medications in the preceding 24 hours; and specifying how much the pain has interfered in several areas of life in the preceding 24 hours (i.e., general activity, mood, walking ability, normal work, relationships, sleep, and enjoyment of life). The BPI-SF was administered to patients in clinics to interpret changes in patient-reported symptoms of chest pain. The CADTH review team did not identify validation information or an estimated MID for the BPI-SF among patients with NSCLC.

Objective Response

Objective response was designated as the primary outcome measure in phase II of the CodeBreaK 100 study and was a key secondary end point in the CodeBreaK 200 study; it was reported as a rate. The ORR was defined as the proportion of patients with a best overall response of confirmed CR or confirmed PR as measured by CT or MRI scans and assessed by RECIST 1.1. Response was assessed by BICR.

Among patients who continued treatment postprogression or who crossed over from docetaxel to sotorasib in the CodeBreaK 200 study, the first date of progression was used for analysis; the patient’s response to post–first progression or postcrossover were not used in the analysis of this end point.

Duration of Response

The DOR was a secondary outcome in phase II of the CodeBreaK 100 study and in the CodeBreaK 200 study. DOR was defined as the time from first CR or PR to disease progression or death due to any cause. DOR was calculated only for those patients who had a confirmed best overall response of PR or CR.

Among patients who continued treatment postprogression or who crossed over from docetaxel to sotorasib in the CodeBreaK 200 study, the first dates of progression were used for analysis, and patients’ responses to post–first progression postcrossover were not used in the analysis of this end point.

Statistical Analysis

Sample Size Determination

In the CodeBreaK 100 study, sample size determination was based on the phase III REVEL clinical trial of ramucirumab plus docetaxel as second-line treatment, which demonstrated an ORR of 23% (95% CI, 20% to 26%),^52,53 establishing the selected benchmark ORR to exclude as 23%. A sample size of 105 patients with NSCLC was determined to provide 90% probability that the lower limit of the ORR 95% CI exceeded the tumour-specific benchmark ORR of 23%.

In the CodeBreaK 200 study, sample size determination was based on the assumption that the true treatment effect HR for PFS was 0.65 for sotorasib versus docetaxel. With 230 PFS events at the time of the primary PFS analysis, there would be approximately 90% power to show a statistically significant difference between groups at a 0.025 1-sided significant level. It was estimated that, with 330 patients enrolled, approximately 19 months would be required to reach 230 PFS events, and 13 months would be required to reach 70% of the target PFS events (interim PFS analysis). This estimation was based on a median PFS of 5 months and 7.7 months for the docetaxel and sotorasib groups, respectively,⁵⁴ with a 10% dropout rate.

Planned Analyses

Primary Analyses

In the CodeBreaK 100 study, the primary analysis for ORR was set to occur approximately 8.5 months after 105 patients with NSCLC were enrolled in phase II. The timing of the primary analysis was based on the assumption that response to treatment would be seen at the first or second scan. Patients with unconfirmed PR or CR were counted as responders for this analysis if they were still enrolled in the study at the time of data cut-off and had the potential for future confirmative scans.

In the CodeBreaK 200 study, the primary analysis for PFS was expected to occur at approximately 19 months, when 230 PFS events were observed. The primary analysis for OS was expected to occur when at least 198 OS events (approximately 60% maturity) were observed (approximately 3 months after the PFS primary analysis). The estimation was based on a median OS of 9 months for the docetaxel group and 12 months for the sotorasib group, for an HR of 0.75. Multiplicity was adjusted as necessary based on the O’Brien-Fleming–type alpha spending function. The primary analysis of ORR was performed when PFS was claimed to be statistically significant and the last randomized patient had had the opportunity to have at least 12 weeks of follow-up.

Final Analyses

The final analyses for both the CodeBreaK 100 study and the CodeBreaK 200 study are set to occur when the ends of the studies (last patient, last visit) have been reached. The purpose of the final analyses is to summarize the efficacy and safety after all patients have completed long-term follow-up. Patients will be followed for 3 years and 5 years in the CodeBreaK 100 study and the CodeBreaK 200 study, respectively. The final analysis will be based on BICR of disease response per RECIST 1.1.

Interim Analyses and Early Stopping

In the CodeBreaK 100 study, safety was assessed after approximately 30, 50, 70, and 100 patients had been treated with sotorasib for 21 days. There were no formal stopping rules for safety. Interim futility analyses were conducted in a continuous manner using Bayesian predictive probability.⁵⁵ For patients with NSCLC, interim futility analysis began after approximately 25 patients had received at least 1 dose of sotorasib and had at least 7 weeks of response data starting from day 1. After the initial interim analysis, subsequent interim analyses were conducted after each set of 10 patients became evaluable for a tumour response. For the purposes of the futility analysis, patients with unconfirmed PR or CR were counted as responders if they were still in enrolled in the study at the time of data cut-off and if future confirmation scans were a possibility. The go criterion was considered to be met if the probability of the true ORR exceeded the benchmark set at probability (ORR > 0.23) greater than or equal to 80%. The futility analysis was based on site-assessed disease response per RECIST 1.1.

In the CodeBreaK 200 study, an interim analysis of PFS was planned when approximately 70% (160) of the target PFS events had been observed from both groups, or when the enrolment was completed and the last randomized patient had at least 6 weeks of follow-up, whichever occurred last. The monitoring boundary for early stopping for efficacy was based on an O’Brien-Fleming–type alpha spending function for multiplicity adjustment. If PFS achieved statistical significance at the interim analysis, an administrative interim summary for OS was set to be performed. An administrative interim summary of OS was performed at the PFS interim analysis, with approximately 107 OS events (approximately 32% maturity) observed. A nominal alpha of 0.01% was spent on this interim summary.

Multiplicity

In the CodeBreaK 200 study, the Maurer-Bretz²² multiple testing procedure among PFS, OS, and ORR was used to control the study-level, overall type I error rate below 1-sided 0.025 levels. The testing procedure is illustrated in Figure 4. Starting with PFS, if the null hypothesis of PFS (i.e., that PFS survival distribution of the sotorasib treatment group is the same as in the docetaxel group) was rejected, then ORR would be tested using a 1-sided alpha divided by 5 (0.005) level. If the ORR null hypothesis (i.e., that the odds ratio of the ORR between the sotorasib group and the docetaxel group is 1) failed to be rejected, then OS would be tested using a 1-sided 0.001 level at the time of PFS interim analysis, otherwise using 4 alpha divided by 5 (0.002) level. With the rejection of OS null hypothesis (i.e., that OS survival distribution of the sotorasib group is the same as for docetaxel group), ORR could be retested using the 1-sided full alpha (0.025) level.

If all 3 hypotheses of PFS, OS, and ORR are rejected, then the next 3 end points of change from baseline over 12 weeks for the symptoms of dyspnea, cough, and pain, as measured by the EORTC QLQ-C30 and EORTC QLQ-LC13, would be tested using Holm’s procedure (Figure 5). Hypotheses (i.e., that differences in the mean change from baseline over time to 12 weeks equal 0) would be rejected sequentially based on the smallest P value.

If the 6 hypotheses listed are rejected, the change from baseline over time to week 12 in physical functioning (i.e., EORTC QLQ-C30) and global health status (i.e., EORTC QLQ-C30) are then tested using Holm’s procedure (Figure 6). The hypotheses (i.e., that differences in mean change from baseline over time to 12 weeks equal 0) would be rejected sequentially based on the smallest P value.

Figure 4: Maurer-Bretz Multiple Testing Procedure for PFS, OS, and ORR in CodeBreaK 200 Study

ORR = objective response rate; OS = overall survival; PFS = progression-free survival.

Source: Protocol for the CodeBreaK 200 study.³⁶

Figure 5: Holm’s Multiple Testing Procedure for Change Over Time in Dyspnea, Cough, and Pain in CodeBreaK 200 Study

dysp. = dyspnea.

Source: Protocol for the CodeBreaK 200 study.³⁶

Figure 6: Holm’s Multiple Testing Procedure for Change Over Time in Physical Functioning and Global Health Status in CodeBreaK 200 Study

ghs = global health status; phys. = physical functioning.

Source: Protocol for the CodeBreaK 200 study.³⁶

Statistical Methods

The statistical analyses of the efficacy end points in the CodeBreaK 100 study and the CodeBreaK 200 study are summarized in Table 9.

Overall Survival

The distribution of OS, including median, was estimated using the KM method (curves, quartiles) in both the CodeBreaK 100 study and the CodeBreaK 200 study. Also reported were the OS rates at selected time points. In the CodeBreaK 100 study, the 95% CI for the median and other percentiles was constructed using the Klein and Moeschberger method with log-log transformation.⁵⁶ The 95% CI for the OS rate was estimated using the method by Kalbfleisch and Prentice with log-log transformation.⁵⁷ In the CodeBreaK 200 study, OS was estimated using the KM method with the HR and 95% CI estimated using a Cox proportional hazards model stratified by the randomization stratification factors.

Additional analyses of OS that adjusted for the potential confounding effect of patients randomized to docetaxel who later cross over to receive sotorasib were conducted and included the RPSFT model,⁵⁸ IPCW,⁵⁹ and 2-stage approach.⁶⁰ The RPSFT model provides a randomization-based estimate of the treatment effect, assumes a multiplicative effect of treatment on time to event, and allows reconstruction of the hypothetical docetaxel time to event, assuming that all patients initially randomized to docetaxel would not cross over to sotorasib. The IPCW method censors patients who cross over at the time of crossover (i.e., weight = 0 during the interval after crossover; therefore, dropped from the model), then up-weights similar patients who did not cross over; as a result, it adjusts for bias of informative censoring. The 2-stage approach adjusts for the crossover effect by estimating the relative treatment effect of OS that would have been observed in the docetaxel group in the absence of crossover, based on the adjusted survival time fitted. Both the 2-stage approach and IPCW method assume no unmeasured confounders.

Progression-Free Survival

In the CodeBreaK 100 study, PFS was analyzed using the same method as described previously for OS.

In the CodeBreaK 200 study, the primary efficacy outcome of PFS was estimated using the KM method. The HR and its 95% CI were estimated using a Cox proportional hazards model stratified by the randomization factors.

HRQoL and PRO Outcomes

Among the HRQoL outcomes assessed in the CodeBreaK 100 study, continuous variables (i.e., EORTC QLQ-C30, EORTC QLQ-LC13, NSCLC-SAQ, and EQ-5D-5L) were summarized by sample size, mean, and SD, median, 25th and 75th percentiles, and minimum and maximum. Categorical variables (i.e., PRO-CTCAE, GP5, EQ-5D-5L non-VAS items, PGIS, and PGIC) were summarized by sample size and percentage in each category. The percentages were based on the nonmissing categories.

In both the CodeBreaK 100 study and the CodeBreaK 200 study, changes from baseline to week 12 on the EORTC QLQ-C30 (dyspnea, physical functioning, and global health status) and EORTC-LC13 (dyspnea, cough, and chest pain) were analyzed using a restricted-maximum, likelihood-based, mixed-effects model for repeated measures under the assumption of missing at random. The model included time and baseline score as fixed effects and used an unstructured covariance structure. A stratified analysis of change from baseline was explored by adding the following baseline covariates: number of prior lines of therapy (1 versus > 1); race (Asian versus non-Asian); and brain metastasis (yes or no). Covariates with less than 5% of the whole population in 1 category were excluded.

The presentations for all other HRQoL and PRO measures were descriptive in nature.

Objective Response Rate

In the CodeBreaK 100 study, the primary efficacy outcome of objective response was summarized as a percentage, with a Clopper-Pearson exact 95% CI. Patients without a postbaseline tumour assessment were considered nonresponders.

In the CodeBreaK 200 study, the outcome of objective response was summarized as a percentage, with a Clopper-Pearson exact 95% CI. The inferential comparison for ORR was made using the Cochran-Mantel-Haenszel chi-square test controlling for the randomization stratification factors. An estimate of the common odds ratio was provided as a measure of the relative treatment effect.

Duration of Response

DOR was calculated only for those patients who had an overall response of PR or better. The distribution of DOR, including median and quartiles, was characterized using KM methods. The 95% CI for median and quartiles for DOR were constructed using the same methods described for OS. The rates for selected duration were reported and estimated using the methods described previously. Patients were censored following the same censoring strategy applied to PFS. KM curves were constructed only if at least 10 patients had PR or CR.

Table 9: Statistical Analysis of Efficacy End Points

BPI-SF = Brief Pain Inventory – Short Form; CI = confidence interval; CNS = central nervous system; CR = complete response; DOR = duration of response; EORTC QLQ-C30 = European Organisation for Research and Treatment of Cancer Quality of Life Questionnaire Core 30; EORTC QLQ-LC13 = European Organisation for Research and Treatment of Cancer Quality of Life Questionnaire Lung Cancer Module; EQ VAS = EQ visual analogue scale; FACT-G = Functional Assessment of Cancer Therapy – General; GEE = generalized estimating equation; HR = hazard ratio; IPCW = inverse-probability-of-censoring weighting; MMRM = mixed-effects model for repeated measures; NR = not reported; NSCLC-SAQ = Non–Small Cell Lung Cancer Symptom Assessment Questionnaire; OR = odds ratio; ORR = objective response rate; OS = overall survival; PFS = progression-free survival; PGIC = patient global impression of change; PGIS = patient global impression of severity; PR = partial response; PRO = patient-reported outcome; PRO-CTCAE = patient-reported outcome version of the Common Terminology Criteria for Adverse Events; RECIST 1.1 = Response Evaluation Criteria in Solid Tumours Version 1.1; RBEE = randomization-based estimate of treatment effect; RPSFT = rank-preserving structural failure time; vs. = versus.

Sources: Clinical Study Report for phase II of the CodeBreaK 100 study;²⁰ Clinical Study Report for the CodeBreaK 100 study PRO data;⁶¹ Protocol for the CodeBreaK 200 study;³⁶ Clinical Study Report for the CodeBreaK 200 study.²¹

Subgroup Analysis

For each of the primary end points and selected secondary end points, the following predefined subgroup analyses relevant to the CADTH protocol were performed in both the CodeBreaK 100 study and the CodeBreaK 200 study:

• ECOG PS (0, 1)

• tumour histology (squamous, nonsquamous)

• number of prior lines of anticancer therapy (1, 2, > 2)

• prior anti-PD1 or anti-PD-L1 therapy (yes, no)

• prior platinum-based chemotherapy (yes, no)

• prior platinum-based chemotherapy and prior anti–PD1 or anti-PD-L1 therapy (yes, no).

No formal hypothesis testing in the subgroups was performed.

Analysis Populations

Four populations were defined for the purpose of data analysis related to the CodeBreaK 100 study, and 5 populations were defined for the CodeBreaK 200 study. Definitions of the populations are summarized in Table 10.

Table 10: Summary of Analysis Populations

BPI-SF = Brief Pain Inventory – Short Form; BICR = blinded independent central review; DOR = duration of response; EORTC QLQ-C30 = European Organisation for Research and Treatment of Cancer Quality of Life Questionnaire Core 30; EORTC QLQ-LC13 = European Organisation for Research and Treatment of Cancer Quality of Life Questionnaire Lung Cancer Module; FACT-G = Functional Assessment of Cancer Therapy – General; HRQoL = health-related quality of life; ITT = intention to treat; ORR = objective response rate; OS = overall survival; PFS = progression-free survival; PGIC = patient global impression of change; PGIS = patient global impression of severity; PP = per protocol; PRO = patient-reported outcome; PRO-CTCAE = patient-reported outcome version of the Common Terminology Criteria for Adverse Events; RECIST 1.1 = Response Evaluation Criteria in Solid Tumours Version 1.1; TTR = time to response.

^aPatients who stopped disease assessment before 7 weeks were included if the data cut-off date was at least 7 weeks after their first dose date.

Sources: Clinical Study Report for phase II of the CodeBreaK 100 study;²⁰ Protocol for the CodeBreaK 200 study;³⁶ Clinical Study Report for the CodeBreaK 200 study.²¹

Results

Patient Disposition

Patient disposition is summarized in Table 11.

Among the 126 patients who were enrolled and received treatment in the CodeBreaK 100 study, 70.6% discontinued treatment. The most common reason for treatment discontinuation was disease progression (55.6%), followed by AEs (8.7%). Of the 57 patients who discontinued from the study, death was cited as the main reason in 45.2%. A total of 123 patients were included in both the full analysis set and the ORR analysis set. By the time of the updated data cut-off date of March 15, 2021, 81.7% of patients had discontinued treatment. The most common reason for treatment discontinuation was disease progression (65.9%), followed by AEs (8.7%). Of the 75 patients who had discontinued from the study by the time of the updated analysis, death was the main reason for study discontinuation (50.0%).

In the CodeBreaK 200 study, approximately 44% of the 616 patients screened were not randomized into the study. The most common reason for this was the absence of documentation of KRAS G12C mutation confirmed by central testing before enrolment (23.6%), followed by having active brain metastases (22.5%), other reasons (9.6%), not having an ECOG PS of equal to or less than 1 (8.9%), and not having adequate hematological laboratory assessments (5.2%). In total, 171 patients and 174 patients were randomized to receive sotorasib 960 mg orally once daily or docetaxel infusion 75 mg/m² every 3 weeks, respectively. Of those randomized, 1.2% of patients in the sotorasib treatment group and 13.2% of those in the docetaxel group were not dosed with an investigational product. Of the 23 patients who were not dosed in the docetaxel treatment group, 16 were not dosed on request of the patient. Of those who received at least 1 dose of the investigational product, 147 patients (86.5%) and 143 patients (82.2%) discontinued treatment in the sotorasib and docetaxel treatment groups, respectively. The most common reason for treatment discontinuation was disease progression (experienced by 60.2% of patients in the sotorasib treatment group and 54.6% of patients in the docetaxel treatment group). Of the 247 patients who discontinued from the study, 121 patients (70.8%) and 126 patients (72.4%) were from the sotorasib and docetaxel treatment groups, respectively. Death was cited as the main reason for study discontinuation in both the sotorasib treatment group (60.8%) and the docetaxel treatment group (48.9%). A greater proportion of patients in the docetaxel treatment group (22.4%) withdrew consent from study participation relative to the sotorasib treatment group (7.0%). At the time of the August 2, 2022, data cut-off date, the full analysis set included 171 and 174 patients in the sotorasib and docetaxel treatment groups, respectively.

During the CodeBreaK 200 study, 46 patients (26.4%) who had been randomized to the docetaxel treatment group crossed over to receive treatment with sotorasib following confirmed progressive disease; an additional 13 patients (7.5%) received sotorasib as a new anticancer therapy. The sponsor clarified that the 13 patients who crossed over outside of the protocol did not differ structurally from the per-protocol crossover patients. All 13 patients received sotorasib upon radiographic progression; however, for most patients, crossover was not available at the clinical site. For 1 patient, the investigator opted not to request crossover from the medical monitor and sought sotorasib administration outside of the clinical trial protocol. At the time of the August 2, 2022, data cut-off date, 12 crossover patients (26.1%) were continuing treatment with sotorasib, while 34 crossover patients (73.9%) had discontinued treatment with sotorasib. The most common reason for discontinuing treatment with sotorasib was disease progression (in 26 crossover patients [56.5%]). Radiographic and clinical disease progression was documented in 22 crossover patients (47.8%) and 4 crossover patients (8.7%), respectively. As of the data cut-off date, 25 crossover patients (54.3%) were continuing in the study, and 21 crossover patients (45.7%) had discontinued the study due to death (n = 17; 37%) or withdrawal of consent (n = 4; 8.7%).

Table 11: Patient Disposition

FAS = full analysis set; NA = not applicable; NR = not reported; NSCLC = non–small cell lung cancer; ORR = objective response rate; PI = principal investigator; PP = per protocol.

^aOther included patient refusal to participate in the study, PI decision, and withdrawal of consent.

^bOther included investigator decision, no clinical benefit, withdrawal of consent, ineligibility to resume sotorasib, loss of clinical benefit.

Sources: Clinical Study Report for phase II of the CodeBreaK 100 study;²⁰ Clinical Study Report for the CodeBreaK 200 study.²¹

Protocol Deviations

Protocol deviations during the CodeBreaK 100 study and the CodeBreaK 200 study are summarized in Table 12.

In the CodeBreaK 100 study, important protocol deviations were documented for 40.5% of patients. The most-cited deviations involved missing data (31.7%), comprising 22.2% related to missing key safety or laboratory samples and 9.5% related to missing screening assessments. In the CodeBreaK 200 study, important protocol deviations were documented in 67.3% and 49.4% of patients in the sotorasib and docetaxel treatment groups, respectively. The most-cited protocol deviations in the CodeBreaK 200 study were off-schedule procedures, which occurred in 36.8% and 27.6% of patients in the sotorasib and docetaxel treatment groups, respectively.

Table 12: Protocol Deviations in CodeBreaK 100 and CodeBreaK 200 Studies

CNS = central nervous system; GCP = good clinical practice; IRT = interactive response technology; NR = not reported; NSCLC = non–small cell lung cancer; PK = pharmacokinetic; QTc = correct heart rate interval; TA = received the wrong treatment or incorrect dose; TC = other treatment adherence.

Notes: Deviation categories are not mutually exclusive. Multiple deviations within the same category are counted once per patient.

Data cut-off dates were September 1, 2020, for the CodeBreaK 100 study and August 2, 2022, for the CodeBreaK 200 study.

^aEach occurrence is counted, including multiple events of the same important protocol deviation for a single patient.

Sources: Clinical Study Report for phase II of the CodeBreaK 100 study;²⁰ Clinical Study Report for the CodeBreaK 200 study.²¹

Exposure to Study Treatments

Exposure to Investigational Product

Patients’ exposure to investigational products in the CodeBreaK 100 study and the CodeBreaK 200 study is summarized in Table 13.

In the CodeBreaK 100 study, the median treatment duration with sotorasib was 24.07 weeks (range, 1.0 week to 52.1 weeks) at a median dose intensity of 100% (range, 15.1% to 100%). The median number of doses per patient was 168 (range, 7 doses to 360 doses). At the time of the updated March 15, 2021, data cut-off date, the median treatment duration with sotorasib was 24.07 weeks (range, 1.0 week to 50.00 weeks) at a median dose intensity of 100% (range, 90.31% to 100%). The median number of doses per patient at the time of the updated data cut-off was 168 (range, 68.0 doses to 338.0 doses).

In the CodeBreaK 200 study, the median treatment duration with sotorasib was 19.86 weeks (range, 0.4 weeks to 101.3 weeks), administered over a median of 7.0 cycles (range, 1 cycle to 34 cycles) at a median relative dose intensity of 100% (range, 23.7% to 100%). Among patients randomized to receive sotorasib, 28 patients (16.4%) continued treatment with sotorasib after initial disease progression. In the docetaxel group, the median duration of treatment was 12 weeks (range, 3.0 weeks to 101.0 weeks) over a median of 4.0 cycles (range, 1 cycle to 33 cycles) at a median relative dose intensity of 94.8% (range, 48.9% to 105.6%).

Of the 46 patients randomized to docetaxel in the CodeBreaK 200 study who crossed over to the sotorasib treatment group, the median time from randomization to first administration of sotorasib after crossover was 5.83 months (range, 1.87 months to 13.63 months). The median treatment duration with sotorasib was 21.0 weeks (range, 2.6 weeks to 62.0 weeks) over 7.0 cycles (range, 1 cycle to 21 cycles) at a relative dose intensity of 97.07% (range, 36.8% to 100%).

Table 13: Exposure to Investigational Products in CodeBreaK 100 and CodeBreaK 200 Studies (Safety Analysis Set)

IP = investigational product; NR = not reported; NSCLC = non–small cell lung cancer; SD = standard deviation.

^aThe average dose delivered is the cumulative dose divided by the number of days on treatment.

^bThe relative dose intensity is equal to the actual dose intensity divided by the planned dose intensity multiplied by 100, where actual (planned) dose intensity is the actual (planned) cumulative dose (in mg) divided by the actual (planned) duration of IP administration.

Sources: Clinical Study Report for phase II of the CodeBreaK 100 study;²⁰ Clinical Study Report for the CodeBreaK 200 study.²¹

Concomitant Medications

The use of concomitant medications in the CodeBreaK 100 study and the CodeBreaK 200 study is summarized in Table 14.

All patients in the CodeBreaK 100 study reported treatment with a concomitant medication. The most used concomitant medication was paracetamol (44%), followed by ibuprofen (23.8%), pantoprazole (23.0%), and folic acid (22.2%).

In the CodeBreaK 200 study, the use of concomitant medications was reported in 100% and 98.7% of patients in the sotorasib and docetaxel treatment groups, respectively. The most common concomitant medication used in the sotorasib treatment group was paracetamol (45.0%), followed by viral vaccine (21.9%), loperamide hydrochloride (20.7%), and pantoprazole (20.1%). In the docetaxel treatment group, the most common concomitant medication used was paracetamol (42.4%), followed by viral vaccines (21.2%) and ondansetron (22.5%).

Table 14: Use of Concomitant Medication in CodeBreaK 100 and CodeBreaK 200 Studies (Safety Analysis Set)

NR = not reported; NSCLC = non–small cell lung cancer.

Note: The data cut-off dates were September 1, 2020, for the CodeBreaK 100 study and August 2, 2022, for the CodeBreaK 200 study.

Sources: Clinical Study Report for phase II of the CodeBreaK 100 study;²⁰ Clinical Study Report for the CodeBreaK 200 study.²¹

New and Subsequent Therapy

A summary of new anticancer therapy used in the CodeBreaK 200 study is presented in Table 15. The per-protocol definition of the requirements to be eligible to receive a new anticancer therapy was not provided. The administration of new anticancer therapy before any per-protocol crossover was documented in 62 patients (36.3%) and 28 patients (16.1%) in the sotorasib and docetaxel treatment groups, respectively.

Table 15: Summary of New Anticancer Therapy in CodeBreaK 200 Study (Full Analysis Set)

NA = not applicable; SD = standard deviation.

Note: Data cut-off date August 2, 2022.

^aSummarizes crossover patients per protocol flagged based on whether the patients randomized to docetaxel received at least 1 dose of sotorasib.

^bSummarizes patients who received any new anticancer therapy. Patients who received both new anticancer therapy and crossover therapy were included only if anticancer therapy was received before crossover.

Source: Clinical Study Report for the CodeBreaK 200 study.²¹

A summary of subsequent anticancer therapy is presented in Table 16. The need for any subsequent anticancer therapy, including crossover therapy, was documented in 36.3% and 42.0% of patients in the sotorasib and docetaxel treatment groups, respectively. Among patients in the sotorasib group who required subsequent anticancer therapy, chemotherapy was the most frequently administered therapy (21.1%). In the docetaxel treatment group, KRAS G12C inhibitor was the most frequently administered subsequent anticancer therapy (33.95%).

Table 16: Summary of Subsequent Anticancer Therapy in CodeBreaK 200 Study (Full Analysis Set)

Note: The data cut-off date was August 2, 2022.

Source: Clinical Study Report for the CodeBreaK 200 study.²¹

Efficacy

Only those efficacy outcomes and analyses of subgroups identified in the review protocol are reported. Refer to Appendix 4 for detailed efficacy data.

Overall Survival

OS data for the CodeBreaK 100 study and the CodeBreaK 200 study are summarized in Table 17. A KM plot of median OS for patients with NSCLC in the CodeBreaK 100 study is illustrated in Figure 7 for the data cut-off date of March 15, 2021, and in Appendix 4 for the data cut-off date of September 1, 2020. The results for OS at 2 years in the CodeBreaK 100 study are summarized in Appendix 4. The KM plot of OS for patients in the CodeBreaK 200 study is illustrated in Figure 8.

In the CodeBreaK 100 study, the proportion of observed deaths at the time of primary data cut-off (September 1, 2020) was 38.1%. The median OS was 12.0 months (95% CI, 9.5 months to NE). The probabilities of survival based on KM estimates at 3 months, 6 months, 9 months, and 12 months were 89.5% (95% CI, 82.7% to 93.8%), 75.5% (95% CI, 66.8% to 82.2%), 63.4% (95% CI, 53.8% to 71.5%), and 51.6% (95% CI, 36.7% to 64.5%), respectively. Results for OS at the updated analysis with data cut-off date of March 15, 2021, were generally consistent with the results from the primary data cut-off date, with median OS of 12.5 months (95% CI, 10.0 months to NE). The results for OS at 2 years, with a data cut-off date of September 9, 2022, were consistent with the results from the previous data cut-off dates, with a median OS of 12.48 months (95% CI, 9.99 months to 19.29 months). The probabilities of survival based on KM estimates at 18 months and 24 months were 42.08% (95% CI, 32.97% to 50.90%) and 31.56% (95% CI, 23.15% to 40.29%), respectively.

In the CodeBreaK 200 study, the proportions of observed deaths at the time of data cut-off (August 2, 2022) were 63.7% and 54.0% in the sotorasib and docetaxel treatment groups, respectively. The median OS was 10.64 months (95% CI, 8.94 months to 13.96 months) in the sotorasib group and 11.3 months (95% CI, 9.00 months to 14.85 months) in the docetaxel group. The stratified HR for OS was 1.01 (95% CI, 0.77 to 1.33; P = 0.53) following treatment with sotorasib versus docetaxel. At the time of the analysis, 36.3% and 46.0% of patients in the sotorasib and docetaxel groups, respectively, were censored. The imbalance was accounted for mainly by more patients withdrawing consent in the docetaxel group (18.4%) than in the sotorasib group (4.7%).

Sensitivity Analysis

In total, 59 patients randomized to docetaxel crossed over to receive treatment with sotorasib; per-protocol crossover occurred in 46 patients following disease progression, while 13 patients received sotorasib as subsequent treatment following disease progression. The sensitivity analyses of OS exploring the crossover effect in the CodeBreaK 200 study led to results that were consistent with the main analysis. In the analyses based on patients who were per-protocol crossovers, the HRs for survival were 1.01 (95% CI, 0.66 to 1.49) in the RPSFT analysis, 0.99 (95% CI, 0.73 to 1.34) in the IPCW adjusted analysis, and 0.885 (95% CI, 0.17 to 1.33) in the 2-stage approach adjusted analysis. A sensitivity analysis of OS exploring the crossover effect among all 59 crossover patients using the 2-stage approach resulted in an HR for survival of 0.82 (95% CI, 0.14 to 1.33). KM plots of sensitivity analysis of OS exploring crossover effect are presented in Figure 20 for 46 patients who were per-protocol crossovers and in Figure 21 for all 59 patients who received sotorasib upon progression.

Progression-Free Survival

PFS data for the CodeBreaK 100 study and the CodeBreaK 200 study are summarized in Table 18. KM plots for median PFS for patients with NSCLC in the CodeBreaK 100 study and the CodeBreaK 200 study are illustrated in Figure 9 and Figure 10, respectively.

Table 17: OS by RECIST 1.1 Based on BICR in CodeBreaK 100 Study (Safety Analysis Set) and CodeBreaK 200 Study (Full Analysis Set)

BICR = blinded independent central review; CI = confidence interval; CNS = central nervous system; HR = hazard ratio; KM = Kaplan-Meier; NE = not evaluable; NR = not reported; NSCLC = non–small cell lung cancer; ORR = objective response rate; OS = overall survival; PFS = progression-free survival; RECIST 1.1 = Response Evaluation Criteria in Solid Tumours Version 1.1; vs. = versus.

Note: Survival status may include publicly available records (where permitted) searched by the investigator after patients ended the study.

^a95% CIs are based on estimated variance for log-log transformation of the KM survival estimate.

^bOS rates and 95% CIs were estimated using the method by Kalbfleisch and Prentice with log-log transformation.

^cThe median OS at 2 years in the CodeBreaK 100 study with data cut-off September 9, 2002, was 12.48 months (95% CI, 9.99 months to 19.29 months). The probabilities of survival based on KM estimates at 18 months and 24 months were 42.08% (95% CI, 32.97% to 50.90%) and 31.56% (95% CI, 23.15% to 40.29%), respectively.

^dStratification factors: number of lines of therapy in advanced disease (1 vs. 2 vs. > 2), race (Asian vs. non-Asian), history of CNS involvement (yes vs. no).

^eThe HR and 95% CI were estimated using a stratified Cox proportional hazard model. An HR of less than 1.0 represents a lower average death rate and longer OS for sotorasib relative to docetaxel.

^fThe P value was calculated using a stratified log-rank test.

^gThe P value was adjusted for multiple comparison using the Maurer-Bertz multiple testing procedure. The PFS and ORR hypothesis was rejected before testing OS.

Sources: Clinical Study Report for phase II of the CodeBreaK 100 study;²⁰ Clinical Study Report for the CodeBreaK 200 study.²¹

In the CodeBreaK 100 study, 56.9% of patients had experienced progression or death due to any cause by the time of the September 2020 data cut-off. The median PFS was 6.7 months (95% CI, 4.9 months to 8.1 months). The probabilities of PFS at 3 months, 6 months, and 9 months were 67.5% (95% CI, 58.2% to 75.2%), 51.5% (95% CI, 41.9% to 60.4%), and 36.2% (95% CI, 26.7% to 45.8%), respectively. The results of the sensitivity analysis for PFS using the investigator’s assessment were consistent with the results obtained by BICR. At the time of the March 2021 data cut-off, 70.2% of patients had experienced progression or death due to any cause. Median PFS was consistent with the earlier data cut-off (i.e., 6.8 months; 95% CI, 5.1 months to 8.2 months).

Figure 7: KM Plot of OS of NSCLC in CodeBreaK 100 Study at the Data Cut-Off Date of March 15, 2021 (Safety Analysis Set)

CI = confidence interval; KM = Kaplan-Meier; NE = not evaluable; NSCLC = non–small cell lung cancer; OS = overall survival.

Note: Death is considered an event.

Source: Clinical Study Report for phase II of the CodeBreaK 100 study.²⁰

Figure 8: KM Plot of OS in CodeBreaK 200 Study at the Data Cut-Off Date of August 2, 2020 (Full Analysis Set)

AMG 510 = sotorasib; CI = confidence interval; HR = hazard ratio; KM = Kaplan-Meier; OS = overall survival.

Source: Clinical Study Report for the CodeBreaK 200 study.²¹

Among patients in the CodeBreaK 200 study, disease progression or death on or before the data cut-off date was observed in 71.3% of patients in the sotorasib treatment group and 58.0% of patients in the docetaxel group. The median PFS was 5.62 months (95% CI, 4.27 months to 7.75 months) in the sotorasib group and 4.47 months (95% CI, 3.02 months to 5.68 months) in the docetaxel group at the time of data cut-off. The stratified HR for disease progression or death was 0.66 (95%, 0.51 to 0.86; P = 0.002) in favour of sotorasib versus docetaxel. At the time of the analysis, 28.7% and 42.0% of patients in the sotorasib and docetaxel groups, respectively, were censored. The imbalance was mainly accounted for by more patients without a baseline disease assessment in the docetaxel group (11.5%) compared with the sotorasib group (0.6%).

The results of the sensitivity analysis for PFS using the investigator’s assessment were consistent with the results obtained by BICR.

Table 18: PFS by RECIST 1.1 Based on BICR in CodeBreaK 100 and CodeBreaK 200 Studies (Full Analysis Set)

BICR = blinded independent central review; CI = confidence interval; CNS = central nervous system; HR = hazard ratio; KM = Kaplan-Meier; NR = not reported; NSCLC = non–small cell lung cancer; PFS = progression-free survival; RECIST = Response Evaluation Criteria in Solid Tumours; vs. = versus.

Note: In the CodeBreaK 200 study, for patients who continued treatment postprogression or patients who crossed over from docetaxel to sotorasib, the first date of progression is used for PFS analysis.

^aPatients who did not have any postbaseline tumour assessment per the investigator assessment were not reviewed by the BICR. These patients were considered as having no baseline in the BICR database and summarized accordingly in the table.

^b95% CIs are based on estimated variance for log-log transformation of the KM survival estimate.

^cEstimated using the method by Kalbfleisch and Prentice with log-log transformation.

^dStratification factors: number of lines of therapy in advanced disease (1 vs. 2 vs. > 2), race (Asian vs. non-Asian), history of CNS involvement (yes vs. no).

^eHR and 95% were estimated using a Cox proportional hazard model. An HR of less than 1.0 indicates a lower average event rate and a longer PFS for sotorasib relative to docetaxel.

^fThe P value was calculated using a stratified log-rank test.

^gThe P value was adjusted for multiple comparison using the Maurer-Bertz multiple testing procedure.

Sources: Clinical Study Report for phase II of the CodeBreaK 100 study;²⁰ Clinical Study Report for the CodeBreaK 200 study.²¹

Figure 9: KM Plot of PFS of NSCLC in CodeBreaK 100 Study at Cut-Off Date of March 15, 2021 (Full Analysis Set)

CI = confidence interval; KM = Kaplan-Meier; NSCLC = non–small cell lung cancer; PFS = progression-free survival.

Note: An event is radiological progression or death, whichever occurs first.

Source: Clinical Study Report for phase II of the CodeBreaK 100 study.²⁰

Subgroup Analysis

Subgroup analyses of PFS among patients in the CodeBreaK 200 study at the time of the August 2, 2022, data cut-off are presented in Table 19. The findings for PFS for sotorasib versus docetaxel were generally consistent across the subgroups and consistent with the overall effect.

Figure 10: KM Plot of PFS of Sotorasib Versus Docetaxel in CodeBreaK 200 Study at Data Cut-Off Date of August 2, 2022 (Full Analysis Set)

AMG 510 = sotorasib; CI = confidence interval; HR = hazard ratio; KM = Kaplan-Meier; PFS = progression-free survival.

Note: An event is radiological progression or death, whichever occurs first.

Source: Clinical Study Report for the CodeBreaK 200 study.²¹

Table 19: Subgroup Analysis of PFS by BICR in CodeBreaK 200 Study

CI = confidence interval; CNS = central nervous system; ECOG = Eastern Cooperative Oncology Group; NR = not reported; PFS = progression-free survival; vs. = versus.

Note: Randomization stratification factors are number of prior lines of therapy in advanced disease (1 vs. 2 vs. > 2), race (Asian vs. non-Asian), and history of CNS involvement (yes vs. no). For subgroup analyses, a stratification factor is not used for the stratified analysis if the stratification factor is the subgroup of interest. If the number of patients in a subgroup level is less than 10 within any treatment group, then only the numbers of events and/or patients are displayed for that subgroup, not the inferential statistics. Data cut-off date August 2, 2022.

^aEvents are disease progression and death.

^bMedians were estimated using the Kaplan-Meier method; 95% CIs were estimated using the method by Klein and Moeschberger with log-log transformation.

^cHazard ratios and 95% CIs were estimated using a stratified Cox proportional hazards model. A hazard ratio of less than 1.0 indicates a lower average death rate and a longer OS for sotorasib relative to docetaxel.

Source: Clinical Study Report for the CodeBreaK 200 study.²¹

Health-Related Quality of Life

In the CodeBreaK 200 study, hierarchical testing was stopped at the non–statistically significant OS end point. Accordingly, the reporting of key secondary HRQoL outcomes in the CodeBreaK 200 study is descriptive. All HRQoL and PRO data presented from the CodeBreaK 100 study and the CodeBreaK 200 study should be considered descriptive in nature.

European Organisation for Research and Treatment of Cancer Quality of Life Questionnaire Core 30

In the CodeBreaK 100 study, 98 patients (78%) completed the EORTC QLQ-C30 at baseline. Compliance rates for the questionnaire were consistently high, above 90%. The number of patients available to complete the measure diminished over time. At the end of treatment, 38 patients (43.2%) completed the assessment. Least squares mean changes from baseline in EORTC QLQ-C30 global health status and physical function — the health scales deemed most important by patients — over time are illustrated in Figure 11 and Figure 12, respectively. Over time, scores for both global health status and physical functioning appeared to remain relatively stable. Changes from baseline in global health status ranged from –5.33 (SD = 15.57) in cycle 11 (n = 25) to + 1.37 (SD = 19.44) in cycle 5 (n = 61). Changes from baseline in physical functioning ranged from –8.57 (SD = 20.33) in cycle 13 (n = 14) to + 6.67 (SD = 22.05) in cycle 7 (n = 33).

In the CodeBreaK 200 study, the EORTC QLQ-C30 — a key secondary outcome — was completed at baseline by 168 patients (98.2%) and 158 patients (90.8%) in the sotorasib and docetaxel treatment groups, respectively. The number of patients available to complete the measure diminished with each cycle. At cycle 5, day 1 (week 12), the EORTC QLQ-C30 was completed by 69 patients (39.7%) in the docetaxel treatment group and 106 patients (62.0%) in the sotorasib treatment group. The least squares mean changes from baseline at week 12 for the EORTC QLQ-C30 global health status and physical functioning scales are illustrated in Figure 13. The differences between groups in least squares change from baseline to week 12 were 6.93 (95% CI, 3.66 to 10.19) for global health status and 8.78 (95% CI, 5.39 to 12.17) for physical functioning, favouring treatment with sotorasib (n = 106) relative to docetaxel (n = 69).

EQ-5D-5L and EQ VAS

A summary of the EQ-5D-5L results for the CodeBreaK 100 study is presented in Table 20. Patients’ results on the EQ-5D-5L remained relatively stable during the treatment period. At baseline (n = 86; 68.3%), most patients (68% to 94%) reported that they had no problems or slight problems with the dimensions of health assessed by the EQ-5D-5L. At the end of the treatment phase (n = 28; 22.2%), 42.9% and 21.4% of patients reported no problems or slight problems, respectively. The EQ VAS scores remained relatively stable during the treatment period. The mean EQ VAS score at baseline was 70.2 (SD = 17.5; n = 86), indicating that patients generally rated their health favourably (with a score of 100 indicating best health imaginable). Change in mean score from baseline over the treatment period ranged from –24.0 (SD = 32.8) in cycle 15 (n = 4) to 2.7 (SD = 18.6) in cycle 3 (n = 69) and 2.7 (SD = 16.6; cycle 7, n = 31). By the end of the treatment phase (n = 28), the mean EQ VAS score was 61.5 (SD = 19.5), which was associated with a mean change from baseline of –10.6 (SD = 19.3). For both the EQ-5D-5L and the EQ VAS, the number of patients available to complete assessments dropped by more than 50% after cycle 6 in the CodeBreaK 100 study.

In the CodeBreaK 200 study, a baseline EQ-5D-5L was completed by 160 patients (93.6%) in the sotorasib treatment group and 138 patients (79.3%) in the docetaxel treatment group. The number of patients available to complete the measure diminished consistently with each cycle. At cycle 5, day 1, the EQ-5D-5L was completed by 105 patients (61.4%) in the sotorasib treatment group and 67 patients (38.5%) in the docetaxel treatment group. A summary of the EQ-5D-5L scales in the CodeBreaK 200 study is presented in Table 21. At baseline, most patients in both the sotorasib (64.4% to 93.2%) and docetaxel (69.6% to 90.6%) treatment groups reported that they had no problems or slight problems with the dimensions of health assessed by the EQ-5D-5L. At cycle 5, day 1, 79.0% to 96.2% of patients in the sotorasib treatment group and 61.2% to 85.1% of patients in the docetaxel treatment group reported that they had no problems or slight problems. Between-group differences for change from baseline on the EQ-5D-5L scales were not reported. Baseline EQ VAS — a secondary outcome — was completed by 166 patients (97.1%) and 154 patients (88.5%) in the sotorasib and docetaxel treatments, respectively. Compliance rates for this measure were consistently greater than 80%. The number of patients available to complete the measure diminished consistently with each cycle. At cycle 5, day 1, the EQ VAS was completed by 106 patients (62.0%) in the sotorasib group and 69 patients (39.7%) in the docetaxel group. Changes in EQ VAS scores following treatment with sotorasib and docetaxel in the CodeBreaK 200 study are illustrated in Figure 14. At baseline, mean EQ VAS scores were 67.6 (SD = 19.9) and 68.3 (SD = 20.3) in the sotorasib and docetaxel treatment groups, respectively. At cycle 5, day 1, the mean EQ VAS score for patients in the sotorasib treatment group was 73.2 (SD = 18.6), which was associated with a mean change from baseline of 2.2 (SD = 15.5) for sotorasib. For patients in the docetaxel treatment group (n = 67), the mean EQ VAS score at cycle 5, day 1 was 67.7 (SD = 20.8), which was associated with a mean change from baseline of –5.8 (SD = 18.2). Between-group differences for change from baseline on the EQ VAS were not reported.

Figure 11: Least Squares Mean Change From Baseline in EORTC QLQ-C30 Global Health Status Over Time in CodeBreaK 100 Study (PRO Analysis Set)

EORTC QLQ-C30 = European Organisation for Research and Treatment of Cancer Quality of Life Questionnaire Core 30; NSCLC = non–small cell lung cancer; PRO = patient-reported outcome.

Note: The data cut-off date was September 1, 2020.

Source: Clinical Study Report for CodeBreaK 100 study PRO data.⁶¹

Figure 12: Least Squares Mean Change From Baseline in EORTC QLQ-C30 Physical Functioning Over Time in CodeBreaK 100 Study (PRO Analysis Set)

EORTC QLQ-C30 = European Organisation for Research and Treatment of Cancer Quality of Life Questionnaire Core 30; NSCLC = non–small cell lung cancer; PRO = patient-reported outcome.

Note: The data cut-off date was September 1, 2020.

Source: Clinical Study Report for CodeBreaK 100 study PRO data.⁶¹

Figure 13: Least Squares Mean Change From Baseline for EORTC QLQ-C30 Scales in CodeBreaK 200 Study — MMRM (EORTC QLQ-C30 Analysis Set)

AMG 510 = sotorasib; CI = confidence interval; CNS = central nervous system; EORTC QLQ-C30 = European Organisation for Research and Treatment of Cancer Quality of Life Questionnaire Core 30; LS = least squares; MMRM = mixed-effects model for repeated measures; QOL = quality of life; vs. = versus.

Notes: The data cut-off date was August 2, 2022.

MMRM is based on change from baseline up to cycle 5, day 1 (week 12) as the dependent variable; intercept, time, baseline score, treatment, treatment-by-time interaction, and randomization stratification factors as fixed effects; and patient intercept and slope of time for the change from baseline as random effects. Overall represents all postbaseline visits, including visits beyond cycle 5, day 1, excluding follow-up visits. Randomization stratification factors: number of prior lines of therapy in advanced disease (1 vs. 2 vs. > 2), race (Asian vs. non-Asian), and history of CNS involvement (yes vs. No).

Source: Clinical Study Report for the CodeBreaK 200 study.²¹

Table 20: Summary of EQ-5D-5L in CodeBreaK 100 Study (PRO Analysis Set)

NSCLC = non–small cell lung cancer; PRO = patient-reported outcome.

Note: The data cut-off date was September 1, 2020.

^aEnd of treatment was defined as the last assessment for the protocol-specified treatment phase of the study for an individual patient.

Source: Clinical Study Report of PRO data for the CodeBreaK 100 study.⁶¹

Table 21: Summary of EQ-5D-5L Dimensions in CodeBreaK 200 Study (EQ-5D-5L Analysis Set)