Effectiveness of Immune Checkpoint Inhibitors in Patients With Advanced Esophageal Squamous Cell Carcinoma: A Meta-analysis Including Low PD-L1 Subgroups

Abstract

Importance: Immune checkpoint inhibitors (ICIs) have improved survival outcomes of patients with advanced esophageal squamous cell carcinoma in both first- and second-line settings. However, the benefit of ICIs in patients with low programmed death ligand 1 (PD-L1) expression remains unclear.

Objective: To derive survival data for patient subgroups with low PD-L1 expression from clinical trials comparing ICIs with chemotherapy in esophageal squamous cell carcinoma and to perform a pooled analysis.

Data sources: Kaplan-Meier curves from the randomized clinical trials were extracted after a systematic search of Scopus, Embase, PubMed, and Web of Science from inception until October 1, 2021.

Study selection: Randomized clinical trials that investigated the effectiveness of anti-PD-1-based regimens for advanced esophageal squamous cell carcinoma and that reported overall survival (OS), progression-free survival, or duration of response were included in this meta-analysis.

Data extraction and synthesis: Kaplan-Meier curves of all-comer populations, subgroups with high PD-L1, and those with low PD-L1 (when available) were extracted from published articles. A graphic reconstructive algorithm was used to calculate time-to-event outcomes from these curves. In studies with unreported curves for subgroups with low PD-L1 expression, KMSubtraction was used to impute survival data. KMSubtraction is a workflow to derive unreported subgroup survival data with from subgroups. An individual patient data pooled analysis including previously reported and newly imputed subgroups was conducted for trials with the same treatment line and PD-L1 scoring system. Data analysis was conducted from January 1, 2022, to June 30, 2022.

Main outcomes and measures: Primary outcomes included Kaplan-Meier curves and hazard ratios (HRs) for OS for subgroups with low PD-L1 expression. Secondary outcomes included progression-free survival and duration of response.

Results: The randomized clinical trials CheckMate-648, ESCORT-1st, KEYNOTE-590, ORIENT-15, KEYNOTE-181, ESCORT, RATIONALE-302, ATTRACTION-3, and ORIENT-2 were included, totaling 4752 patients. In the pooled analysis of first-line trials that evaluated a tumor proportion score (CheckMate-648 and ESCORT-1st), no significant benefit in OS was observed with immunochemotherapy compared with chemotherapy in the subgroup of patients who had a tumor proportion score lower than 1% (HR, 0.91; 95% CI, 0.74-1.12; P = .38) compared with chemotherapy. In the pooled analysis of first-line trials that evaluated combined positive score (KEYNOTE-590 and ORIENT-15), there was a significant but modest OS benefit for immunochemotherapy compared with chemotherapy in the subgroup with a combined positive score lower than 10 (HR, 0.77; 95% CI, 0.62-0.94; P = .01).

Conclusions and relevance: Findings suggest a lack of survival benefit of ICI-based regimens in the first-line setting compared with chemotherapy alone in the subgroup with a tumor proportion score lower than 1%.

Figure 1.. Kaplan-Meier Plots for Overall Survival in Low Programmed Death Ligand 1 (PD-L1) Subgroups Derived With KMSubtraction (Studies With the Tumor Proportion Score [TPS] Available)

A, In the first-line CheckMate-648 study, there was no significant difference in overall survival for the subgroup with a TPS lower than 1% for both ipilimumab plus nivolumab dual immunotherapy (hazard ratio [HR], 0.98; 95% CI, 0.75-1.28; P = .89) and nivolumab-based immunochemotherapy (HR, 0.99; 95% CI, 0.76-1.30; P = .97) compared with chemotherapy alone. B, In the second-line ESCORT study, there was no statistically significant difference in overall survival for immunotherapy compared with chemotherapy alone (HR, 0.78; 95% CI, 0.59-1.02; P = .07) for the subgroup with a TPS lower than 1%.

Figure 2.. Kaplan-Meier Plots for Overall Survival in Low Programmed Death Ligand 1 (PD-L1) Subgroups Derived With KMSubtraction (Studies With the Combined Positive Score [CPS] Available)

A, In the global multinational studies on pembrolizumab, the first-line KEYNOTE-590 trial subgroup with a CPS lower than 10 demonstrated no significant difference in overall survival for immunochemotherapy compared with chemotherapy alone (hazard ratio [HR], 0.91; 95% CI, 0.69-1.20; P = .50). B, In the second-line KEYNOTE-181 trial, for the subgroup with a CPS lower than 10, there was no significant difference in overall survival for immunotherapy compared with chemotherapy alone (HR, 0.90; 95% CI, 0.68-1.18; P = .44). C, In the sintilimab-based ORIENT-15 trial, for the subgroup with a CPS lower than 10, there was a significant difference in overall survival for immunochemotherapy compared with chemotherapy (HR, 0.62; 95% CI, 0.45-0.85; P = .003).

Figure 3.. Kaplan-Meier Plots for Progression-Free Survival and Duration of Response in Low Programmed Death Ligand 1 (PD-L1) Subgroups Derived With KMSubtraction

A, In the subgroup of CheckMate-648 with a tumor proportion score (TPS) lower than 1%, no significant difference in progression-free survival was observed between immunochemotherapy and chemotherapy (hazard ratio [HR], 0.98; 95% CI, 0.75-1.28; P = .88); there was a significantly inferior progression-free survival for ipilimumab plus nivolumab dual immunotherapy compared with chemotherapy alone (HR, 1.47; 95% CI, 1.14-1.90; P = .003). B, In the subgroup of ORIENT-15 with a combined positive score (CPS) lower than 10, a significant difference in progression-free survival was observed between immunochemotherapy and chemotherapy (HR, 0.52; 95% CI, 0.39-0.70; P < .001). C, In the subgroup of CheckMate-648 with a TPS lower than 1%, the median duration of response was 6.9 months (95% CI, 5.8-15.1 months; HR, 0.91; 95% CI, 0.58-1.44; P = .70) for nivolumab-based immunochemotherapy, 10.3 months (95% CI, 5.8-16.7 months; HR, 0.87; 95% CI, 0.51-1.49; P = .61) for nivolumab plus ipilimumab, and 7.2 months (95% CI, 5.8-10.1 months) for chemotherapy. D, In the subgroup of ORIENT-15 with a CPS lower than 10, the median duration of response was 7.2 months (95% CI, 5.8-15.2 months; HR, 0.64; 95% CI, 0.43-0.96; P = .03) for sintilimab-based immunochemotherapy and 6.9 months (95% CI, 5.7-8.4 months) for chemotherapy.

Figure 4.. One-Stage Pooled Analysis of First-line Studies With the Tumor Proportion Score (TPS) Available (CheckMate-648 and ESCORT-1st)

A, There was a significant difference in overall survival for the overall population comparing immunochemotherapy (hazard ratio [HR], 0.73; 95% CI, 0.63-0.84; P < .001) with chemotherapy alone. B, There was a significant difference in overall survival for the population with a TPS of 1% or higher comparing immunochemotherapy (HR, 0.57; 95% CI, 0.47-0.70; P < .001) with chemotherapy alone. C, There was no significant difference in overall survival for the subgroup with a TPS lower than 1% comparing immunochemotherapy (HR, 0.91; 95% CI, 0.74-1.12; P = .38) with chemotherapy alone.

Figure 5.. One-Stage Pooled Analysis of First-line Combined Positive Score (CPS) Studies (ORIENT-15 and KEYNOTE-590)

A, There was a significant difference in overall survival for the overall group comparing immunochemotherapy (hazard ratio [HR], 0.67; 95% CI, 0.58-0.78; P < .001) with chemotherapy alone. B, There was a significant difference in overall survival for the group with a CPS of 10 or higher comparing immunochemotherapy (HR, 0.60; 95% CI, 0.49-0.73; P < .001) with chemotherapy alone. C, There was a significant difference in overall survival for the subgroup with a CPS lower than 10 comparing immunochemotherapy with chemotherapy alone (HR, 0.77; 95% CI, 0.62-0.94; P = .01).