Quality of life with first-line pembrolizumab for PD-L1-positive advanced gastric/gastroesophageal junction adenocarcinoma: results from the randomised phase III KEYNOTE-062 study.

BACKGROUND: In the randomised phase III KEYNOTE-062 study, pembrolizumab was non-inferior to chemotherapy for overall survival in patients with programmed death-ligand 1 (PD-L1)-positive [combined positive score (CPS) ≥1] advanced gastric/gastroesophageal junction (GEJ) cancer. We present findings of prespecified health-related quality-of-life (HRQOL) analyses for pembrolizumab versus chemotherapy in this population.
MATERIALS AND METHODS: HRQOL, a secondary endpoint, was measured in patients who received ≥1 dose of study treatment and completed ≥1 HRQOL questionnaire [European Organisation for the Research and Treatment of Cancer (EORTC) 30-question quality-of-life (QLQ-C30), EORTC 22-question quality-of-life gastric-cancer-specific module (QLQ-STO22)]. Least squares mean (LSM) change (baseline to week 18) in global health status/quality of life (GHS/QOL; EORTC QLQ-C30) and time to deterioration (TTD) in GHS/QOL, nausea/vomiting and appetite loss scores (EORTC QLQ-C30) and abdominal pain/discomfort scores (EORTC QLQ-STO22) were evaluated.
RESULTS: The HRQOL population comprised 495 patients with CPS ≥1 (pembrolizumab, 252; chemotherapy, 243). Compliance rates at week 18 were similar for pembrolizumab and chemotherapy (EORTC QLQ-C30, 87.9% and 81.9%; EORTC QLQ-STO22, 87.9% and 81.3%, respectively). There was no between-arm difference in LSM score change in GHS/QOL [-0.16; 95% confidence interval (CI) -5.01 to 4.69; P = 0.948]. The LSM score change for most subscales showed comparable worsening in both arms. TTD for GHS/QOL [hazard ratio (HR), 0.96; 95% CI, 0.67-1.38; P = 0.826], appetite loss (HR, 0.83; 95% CI, 0.58-1.20; P = 0.314) and pain (HR, 1.22; 95% CI, 0.78-1.91; P = 0.381) were similar between arms. Longer TTD was observed for pembrolizumab versus chemotherapy for nausea/vomiting (HR, 0.61; 95% CI, 0.44-0.85; P = 0.003).
CONCLUSIONS: HRQOL was maintained with first-line treatment with pembrolizumab in patients with PD-L1-positive advanced gastric/GEJ cancer and was similar between pembrolizumab and chemotherapy in this population.

Introduction

Gastric cancer is among the most commonly diagnosed cancers globally, with an incidence of >1 million new cases (sixth most common among cancer types) and almost 800 000 deaths (third most leading cause of cancer deaths) annually. The 5-year survival rate for patients with advanced stage gastric/gastroesophageal junction (GEJ) cancer is 5%-10%. Advanced gastric/GEJ cancer is also marked by worsening overall health-related quality of life (HRQOL) as measured by validated questionnaires. Physical, social and emotional functioning and the disease-related symptom profile (weight loss, abdominal pain, vomiting, gastric obstruction, bleeding) are more burdensome in the advanced stage of gastric/GEJ cancer than in earlier stages. In addition, chemotherapy is known to cause drug-related symptoms (abdominal pain, fatigue, nausea/vomiting, diarrhoea) in patients with advanced gastric/GEJ cancer. For most patients with advanced gastric/GEJ cancer, the standard of care first-line treatment is doublet chemotherapy,3, 4, 5, 6 which offers a modest but statistically significant benefit of about 1 month in overall survival (OS) compared with single-agent chemotherapy. Chemotherapy regimens used as first-line therapy, regardless of type, have been found to maintain HRQOL over time.

The programmed death 1 (PD-1) inhibitor pembrolizumab has demonstrated antitumour activity and a manageable toxicity profile in patients with programmed death-ligand 1 (PD-L1)–positive [combined positive score (CPS) ≥1] advanced gastric/GEJ adenocarcinoma., In cohort 1 of the phase II KEYNOTE-059 trial (NCT02335411), pembrolizumab monotherapy was given to patients with gastric/GEJ adenocarcinoma in the third-line or later setting. Among patients whose tumours express CPS ≥1, the objective response rate (ORR) was 16% and the median duration of response (DOR) was 16 months. Pembrolizumab also demonstrated antitumour activity in patients with previously untreated gastric/GEJ adenocarcinoma enrolled in cohorts 2 (pembrolizumab plus chemotherapy) and 3 (pembrolizumab monotherapy) of the KEYNOTE-059 study; patients in cohort 3 were required to have CPS ≥1 tumours. The ORR was 60% in cohort 2 and 26% in cohort 3, and safety was tolerable with both regimens. The KEYNOTE-062 trial (NCT02494583) was a randomised, active-controlled, phase III study of pembrolizumab as monotherapy or in combination with chemotherapy compared with placebo plus chemotherapy as first-line treatment of patients with advanced gastric or GEJ cancer. At the final analysis, pembrolizumab monotherapy was non-inferior to chemotherapy for OS among patients with CPS ≥1 tumours [median OS, 10.6 versus 11.1 months; hazard ratio (HR), 0.91; 99.2% confidence interval (CI), 0.69-1.18 (prespecified non-inferiority margin = 1.2)]. Although not formally tested, there was a clinically meaningful improvement in OS in patients with CPS ≥10 tumours (median OS, 17.4 versus 10.8 months; HR, 0.69; 95% CI, 0.49-0.97). Pembrolizumab monotherapy offered an improved safety profile with a lower incidence of any-grade (54% versus 92%) and grade 3-4 (16% versus 68%) treatment-related adverse events (AEs) compared with chemotherapy.

Here we present findings from the prespecified secondary and exploratory HRQOL analyses in patients with CPS ≥1 gastric/GEJ cancer who received pembrolizumab monotherapy versus chemotherapy in the KEYNOTE-062 trial.

Materials and methods

Study design and patients

KEYNOTE-062 (ClinicalTrials.gov: NCT02494583) was a global, randomised, phase III clinical trial of first-line pembrolizumab monotherapy or pembrolizumab plus chemotherapy versus chemotherapy in patients with advanced gastric/GEJ cancer (Supplementary Figure S1, available at https://doi.org/10.1016/j.esmoop.2021.100189). In the primary analysis, pembrolizumab versus chemotherapy met the criteria for non-inferiority, whereas pembrolizumab-chemotherapy versus chemotherapy did not meet the criteria for superiority. Therefore, we focused on the HRQOL outcomes for pembrolizumab monotherapy compared with chemotherapy in this HRQOL analysis. Details of this comparison in the primary study have been reported elsewhere and are briefly summarised here. Eligible patients had locally advanced unresectable or metastatic gastric/GEJ adenocarcinoma that was human epidermal growth factor receptor 2 (HER2)/neu-negative and PD-L1–positive (CPS ≥1) and Eastern Cooperative Oncology Group performance status 0 or 1. PD-L1 positivity was tested using PD-L1 IHC 22C3 pharmDx (Agilent Technologies, Carpinteria, CA, USA) and was measured using CPS [defined as the number of PD-L1–positive cells (tumour cells, lymphocytes, macrophages) as a proportion of the total number of viable tumour cells × 100]. Patients were randomly assigned 1 : 1 : 1 to pembrolizumab 200 mg, pembrolizumab-chemotherapy [cisplatin 80 mg/m2/day on day 1 plus 5-fluorouracil (FU) 800 mg/m2/day on days 1 to 5 or capecitabine 1000 mg/m2 twice daily] or placebo-chemotherapy every 3 weeks for a maximum of 35 cycles (∼2 years). The chemotherapy regimen was decided by the investigator before randomisation.

The study protocol and all amendments were approved by the appropriate ethics committee at each centre. The study was conducted in accordance with the protocol, its amendments and the standards of Good Clinical Practice. All patients provided written informed consent.

HRQOL outcomes and assessments

HRQOL outcomes reported here were prespecified secondary and exploratory endpoints from KEYNOTE-062. Secondary endpoints included mean change from baseline to week 18 in the global health status/quality-of-life (GHS/QOL), functioning and symptom scores of the European Organisation for the Research and Treatment of Cancer (EORTC) core 30 quality-of-life questionnaire (QLQ-C30) and the EORTC QLQ 22-question quality-of-life gastric cancer-specific module (QLQ-STO22) as well as time to deterioration (TTD) in the GHS/QOL, nausea/vomiting and appetite loss scores of the EORTC QLQ-C30 and the abdominal pain/discomfort scores of the EORTC QLQ-STO22. Week 18 was selected based on a blinded review of compliance/completion rates where rates were 60%-80%. Health status from the EuroQol five-dimension, three-level (EQ-5D-3L) questionnaire was an exploratory endpoint.

HRQOL questionnaires were administered by qualified site personnel and completed electronically by the patient, before study drug administration or AE/disease status evaluation and in the following order: EQ-5D-3L, EORTC QLQ-C30, EORTC QLQ-STO22. The EQ-5D-3L is a standardised instrument that has been translated into 170 languages and provides data for use in economic models and analyses of health utility or quality-adjusted life-years and addresses five health state dimensions: mobility, self-care, usual activities, pain/discomfort and anxiety/depression. The EORTC QLQ-C30 is a widely used and well-validated cancer-specific HRQOL instrument that has been translated to and validated in 81 languages and comprises a GHS/QOL scale, five functional dimensions (physical, role, emotional, cognitive and social), three symptom scales (fatigue, nausea/vomiting and pain) and six single-item measures (dyspnoea, sleep disturbance, appetite loss, constipation, diarrhoea and financial difficulties). The EORTC QLQ-STO22, designed for use in clinical trials in addition to the EORTC QLQ-C30 to assess disease-specific treatment measures and translated into 60 languages, consists of 22 items: symptoms of dysphagia (three items), pain (four items), reflux symptoms (three items), eating restrictions (five items), anxiety (three items), dry mouth, taste, body image and hair loss. Questionnaires were administered at baseline; at weeks 3, 6, 9 and 12; every 6 weeks thereafter up to 1 year or end of treatment (whichever came first); and at the 30-day post-treatment discontinuation follow-up visit.

As previously defined for the EORTC QLQ-C30, a mean change of 5 to 10 points in EORTC QLQ-C30 scores represents a small change (‘little’ change as reported by patients), 10 to 20 points represents a moderate change and >20 points represents a large change (‘very much’ as reported by patients); a decline from baseline of ≥10 points on the functional or global health and QOL scales was considered a clinically meaningful deterioration. Therefore, deterioration was defined as a ≥10-point decline from baseline when measuring TTD.

Statistical analysis

The HRQOL analysis population comprised all patients who received ≥1 dose of study treatment and completed ≥1 HRQOL questionnaire. Compliance and completion rates were summarised by treatment arm and visit and were reported for all three HRQOL questionnaires. Compliance rate was defined as the proportion of patients who completed ≥1 HRQOL questionnaire among those expected to complete the questionnaires at each visit (excluding patients missing by design because they discontinued study treatment). Completion rate was defined as the proportion of patients who completed ≥1 HRQOL questionnaire among the total HRQOL analysis population at each visit. Change in least squares mean (LSM) score from baseline to week 18 was assessed using a constrained longitudinal data analysis model based on the missing-at-random assumption. Descriptive analyses of mean score and mean score changes from baseline [±standard error (SE)] in the GHS/QOL and subscale scores of the EORTC QLQ-C30 and nausea/vomiting subscale scores for the EORTC QLQ-STO22 were summarised through week 48 to further depict trends. The Kaplan–Meier method was used to estimate the TTD survival curve for the GHS/QOL score and subscales of the EORTC QLQ-C30 and EORTC QLQ-STO22. The TTD analysis was censored at death. Cox proportional hazards model with treatment as a covariate was used to assess the magnitude of treatment differences. The median TTD HR and 95% CI were reported.

Analyses were conducted using the final analysis data cut-off date of 26 March 2019 [median follow-up, 11.3 months (range, 0.2-41.2 months)].

Results

Patients

A total of 763 patients were randomly assigned in KEYNOTE-062 (pembrolizumab monotherapy, n = 256; pembrolizumab plus chemotherapy, n = 257; chemotherapy, n = 250). The HRQOL population in the current analysis of pembrolizumab monotherapy versus chemotherapy comprised 495 patients who received treatment and completed ≥1 HRQOL questionnaire by the final analysis (26 March 2019): 252 in the pembrolizumab arm and 243 in the chemotherapy arm.

HRQOL compliance and completion

Among patients in the HRQOL analysis population, compliance rates at week 18 were similar in the pembrolizumab arm and the chemotherapy arm for EORTC QLQ-C30 (87.9% and 81.9%), EORTC QLQ-STO22 (87.9% and 81.3%) and EQ-5D-3L (87.9% and 82.5%) questionnaires, respectively. Completion rates of all three questionnaires decreased from baseline because of treatment discontinuation attributed to disease progression, death or AEs (Supplementary Tables S1-S3, available at https://doi.org/10.1016/j.esmoop.2021.100189).

Change in HRQOL from baseline to week 18

EORTC QLQ-C30 and EORTC QLQ-STO22

Baseline GHS/QOL scores of the EORTC QLQ-C30 were well balanced between the pembrolizumab and chemotherapy arms (Table 1).

Table 1

Change from baseline in EORTC QLQ-C30 GHS/QOL symptom scores and EORTC QLQ-STO22 symptom subscale scores at week 18

Treatment	Baseline score, mean (SD)	Week 18 score, mean (SD)	Change from baseline at week 18, LSM (95% CI)a	Difference in LSM (95% CI)
EORTC QLQ-C30 GHS/QOL
Pembrolizumab	n = 23962.7 (21.8)	n = 10266.4 (20.2)	n = 251−1.9 (−5.8 to 2.0)	−0.16 (−5.0 to 4.7)P = 0.948
Chemotherapy	n = 23462.4 (21.1)	n = 14063.6 (20.2)	n = 243−1.8 (−5.2 to 1.7)
EORTC QLQ-C30 nausea/vomiting
Pembrolizumab	n = 23916.1 (21.9)	n = 10215.9 (23.3)	n = 2515.8 (1.3-10.3)	−2.1 (−7.8 to 3.7)P = 0.477
Chemotherapy	n = 23416.8 (21.9)	n = 14021.2 (25.2)	n = 2437.9 (4.0-11.8)
EORTC QLQ-C30 appetite loss
Pembrolizumab	n = 23933.9 (33.9)	n = 10221.9 (29.1)	n = 251−3.5 (−9.4 to 2.3)	−4.6 (−11.9 to 2.7)P = 0.217
Chemotherapy	n = 23436.6 (33.3)	n = 14032.9 (30.7)	n = 2431.0 (−4.1 to 6.2)
EORTC QLQ-STO22 pain
Pembrolizumab	n = 23930.4 (22.1)	n = 10222.0 (20.2)	n = 251−1.1 (−4.7 to 2.4)	2.4 (−2.2 to 6.9)P = 0.308
Chemotherapy	n = 23328.8 (21.3)	n = 13922.3 (20.4)	n = 243−3.5 (−6.6 to −0.4)

CI, confidence interval; GHS, global health status; LSM, least squares mean; EORTC QLQ-C30, European Organisation for the Research and Treatment of Cancer core 30 quality-of-life questionnaire; EORTC QLQ-STO22, European Organisation for the Research and Treatment of Cancer 22-question quality-of-life gastric cancer-specific module; QOL, quality of life; SD, standard deviation.

At week 18, mean GHS/QOL scores were 66.4 [standard deviation (SD) ±20.2] with pembrolizumab and 63.6 (SD ±20.2) with chemotherapy. No clinically meaningful between-arm differences in LSM score were observed (LSM difference, −0.16; 95% CI, −5.0 to 4.7; nominal P = 0.948). Results were similar for the EORTC QLQ-C30 symptom subscales of nausea/vomiting and appetite loss and the EORTC QLQ-STO22 symptom subscale of pain (Table 1). The LSM score change from baseline to week 18 for most of the EORTC QLQ-C30 and EORTC QLQ-STO22 function and symptom subscales showed comparable worsening in both treatment arms (Figure 1). The largest between-arm difference in LSM score was for the taste subscale of the EORTC QLQ-STO22, whereas patients in the chemotherapy arm reported greater worsening.

Figure 1

LSM (95% CI) change from baseline to week 18 in (A) EORTC QLQ-C30 GHS/QOL and functional subscale scores, (B) EORTC QLQ-C30 symptom subscale scores and (C) EORTC QLQ-STO22 symptom subscale scores.

CI, confidence interval; GHS, global health status; LSM, least squares mean; EORTC QLQ-C30, European Organisation for the Research and Treatment of Cancer core 30 quality-of-life questionnaire; EORTC QLQ-STO22, European Organisation for the Research and Treatment of Cancer 22-question quality-of-life gastric cancer-specific module; QOL, quality of life.

Descriptive analyses of mean change from baseline revealed that GHS/QOL scores remained relatively stable for both arms (Figure 2A). A slight improvement was observed in the pembrolizumab arm between weeks 18 and 42. Descriptive analyses of mean score change from baseline through week 48 of follow-up demonstrated a general trend in improvement for the EORTC QLQ-C30 nausea/vomiting and appetite loss subscales and the EORTC QLQ-STO22 pain subscale (Figure 2B-D). However, these changes from baseline were not clinically meaningful.

Figure 2

Mean (±SE) change from baseline by study visit in (A) EORTC QLQ-C30 GHS/QOL scores, (B) EORTC QLQ-C30 symptom subscale nausea/vomiting scores, (C) EORTC QLQ-C30 subscale appetite loss scores and (D) EORTC QLQ-STO22 symptom pain subscale scores. Parts C and D are continued on next page.

GHS, global health status; EORTC QLQ-C30, European Organisation for the Research and Treatment of Cancer core 30 quality-of-life questionnaire; EORTC QLQ-STO22, European Organisation for the Research and Treatment of Cancer 22-question quality-of-life gastric cancer-specific module; QOL, quality of life; SE, standard error.

EQ-5D-3L visual analogue scale

The EQ-5D-3L visual analogue scale (VAS) score decreased from baseline to week 18 in the pembrolizumab arm (LSM score change, −5.98; 95% CI, −9.28 to −2.69) and the chemotherapy arm (LSM score change, −4.78; 95% CI, −7.70 to −1.87), indicating worsening. There was no difference in LSM between arms (LSM difference, −1.20; 95% CI, −5.41 to 3.00; nominal P = 0.574) (Supplementary Figure S2, available at https://doi.org/10.1016/j.esmoop.2021.100189).

Time to deterioration at week 18

The proportion of patients with deterioration at week 18 (defined as a ≥10-point worsening from baseline) was similar between the pembrolizumab and chemotherapy arms for the GHS/QOL score, the EORTC QLQ-C30 appetite loss subscale and the EORTC QLQ-STO22 pain subscale. In the EORTC QLQ-C30 nausea/vomiting subscale, substantially fewer patients experienced deterioration at week 18 with pembrolizumab (23.4%) than with chemotherapy (39.7%) (Figure 3).

Figure 3

Proportion of patients with deteriorated EORTC QLQ-C30 and EORTC QLQ-STO22 scores at week 18.

GHS, global health status; EORTC QLQ-C30, European Organisation for the Research and Treatment of Cancer core 30 quality-of-life questionnaire; EORTC QLQ-STO22, European Organisation for the Research and Treatment of Cancer 22-question quality-of-life gastric cancer-specific module; QOL, quality of life.

TTD in GHS/QOL (HR, 0.96; 95% CI, 0.67-1.38; nominal P = 0.826), EORTC QLQ-C30 appetite loss subscale (HR, 0.83; 95% CI, 0.58-1.20; nominal P = 0.314), and EORTC QLQ-STO22 pain subscale (HR, 1.22; 95% CI, 0.78-1.91; nominal P = 0.381) were similar between arms (Figure 4A, C, D). Longer TTD was observed for pembrolizumab than for chemotherapy for the EORTC QLQ-C30 nausea/vomiting subscale (HR, 0.61; 95% CI, 0.44-0.85; nominal P = 0.003) (Figure 4B).

Figure 4

TTD in (A) EORTC QLQ-C30 GHS/QOL, (B) EORTC QLQ-C30 nausea/vomiting subscale, (C) EORTC QLQ-C30 appetite loss subscale and (D) EORTC QLQ-STO22 pain subscale. Parts C and D are continued on next page.

CI, confidence interval; GHS, global health status; NR, not reached; EORTC QLQ-C30, European Organisation for the Research and Treatment of Cancer core 30 quality-of-life questionnaire; EORTC QLQ-STO22, European Organisation for the Research and Treatment of Cancer 22-question quality-of-life gastric cancer-specific module; QOL, quality of life; TTD, time to deterioration; -, the median was not reached.

a From product-limit (Kaplan–Meier) method for censored data.

b Based on stratified Cox regression model with treatment as a covariate.

c Two-sided P value based on stratified log-rank test.

Discussion

This analysis showed that HRQOL was similar in patients with PD-L1−positive (CPS ≥1) advanced gastric/GEJ tumours who received first-line pembrolizumab and those who received chemotherapy. Compliance rates of the three HRQOL questionnaires were high (≥81%) and generally equivalent between treatment arms at week 18, the primary analysis time point for HRQOL. General HRQOL as measured by EORTC QLQ-C30 GHS/QOL scores were comparable between treatment arms from baseline to week 18. The EQ-5D-3L VAS, which measures general HRQOL regardless of disease status, was also equivalent between arms from baseline to week 18.

All five functional subscales and nearly all symptom subscales of the EORTC QLQ-C30 showed comparable scores between arms. The same was true of the EORTC QLQ-STO22 symptom subscales with the exception of taste, which showed a worsening trend for patients treated with chemotherapy than for patients treated with pembrolizumab. TTD for the EORTC QLQ-C30 nausea/vomiting subscale was longer for pembrolizumab-treated patients than for chemotherapy-treated patients. The primary analysis of the KEYNOTE-062 trial demonstrated a favourable safety profile for pembrolizumab monotherapy compared with chemotherapy. The incidence of treatment-related AEs associated with symptom and single-item subscales (fatigue, nausea, vomiting, decreased appetite, constipation, diarrhoea) was lower for pembrolizumab-treated patients than for chemotherapy-treated patients. Not all symptoms that occurred more frequently among chemotherapy-treated patients resulted in an observable change in HRQOL; no clinically meaningful differences in LSM score from baseline to week 18 were observed in appetite loss, constipation and diarrhoea scales or in TTD for appetite loss. Taken together, pembrolizumab offers favourable safety compared with chemotherapy and maintains HRQOL.

Intensification of chemotherapy for gastric/GEJ cancer can provide modest gains in OS, and first-line chemotherapy regimens in gastric/GEJ cancer tend to maintain HRQOL as measured by GHS/QOL. Evidence from immunotherapy trials has demonstrated that monotherapy with PD-1 inhibitors can provide efficacy benefits and tolerable safety while maintaining or improving HRQOL in patients with solid tumours., In a systematic review of HRQOL outcomes, there was a consistent prolongation of the time to symptom deterioration and better symptom control at different follow-up points with nivolumab, pembrolizumab and atezolizumab in patients with lung cancer, melanoma, head and neck cancer and urothelial cancer. Findings from the present analysis of KEYNOTE-062 demonstrate the ability of pembrolizumab monotherapy to maintain HRQOL in the first-line setting in patients with gastric/GEJ cancer.

One limitation of these HRQOL analyses is the partially blinded design of the study. Patients were not fully blinded to pembrolizumab monotherapy because only one type of study treatment was administered in that arm of the trial. Conversely, administration of pembrolizumab or placebo was blinded in the combination chemotherapy arms. Future data releases will elucidate the HRQOL profile of double-blinded pembrolizumab or placebo compared with chemotherapy.

Conclusion

In this study in patients who had advanced gastric/GEJ adenocarcinoma with PD-L1 CPS ≥1 tumours, HRQOL was similar between the pembrolizumab and the chemotherapy arms in the first-line setting. Longer TTD was observed for pembrolizumab than for chemotherapy for the nausea/vomiting subscale in EORTC QLQ-C30, which aligned with the increased incidence of nausea and vomiting treatment-related AEs from chemotherapy in the primary safety analysis. Overall, HRQOL was similar in patients with PD-L1-positive advanced gastric/GEJ adenocarcinoma, whether they received pembrolizumab or chemotherapy as first-line treatment.