Comparative Analysis of Durable Responses on Immune Checkpoint Inhibitors Versus Other Systemic Therapies

A Pooled Analysis of Phase III Trials

Elvire Pons-Tostivint, MD; Aurélien Latouche, PhD; Pauline Vaflard, MD; Francesco Ricci, MD, PhD; Delphine Loirat, MD, PhD; Ségolène Hescot, MD, PhD; Marie-Paule Sablin, MD; Roman Rouzier, MD, PhD; Maud Kamal, PhD; Claire Morel, MSc; Charlotte Lecerf, MSc; Vincent Servois, MD; Xavier Paoletti, PhD; Christophe Le Tourneau, MD, PhD

Disclosures

JCO Precis Oncol. 2019;2019(3) 

In This Article

Results

Characteristics of the Trials

Our search retrieved 353 studies (Figure 2). Thirty-one of these studies met our selection criteria. Twelve of these were excluded for the following reasons: absence of a monotherapy ICI arm (n = 3), adjuvant setting (n = 3), duplicate trials (n = 5), and insufficient follow-up (n = 1). The 19 remaining phase III trials involved 11,640 patients treated in 42 different treatment arms (Table 1).

Figure 2.

Flowchart of study selection.

Tumor types included melanoma (seven trials),[8–12,23,24] NSCLC (five trials),[1–3,19] urothelial carcinoma (two trials),[6,20] prostate cancer (two trials),[21,22] head and neck squamous cell carcinoma (one trial),[4] gastric cancer (one trial),[5] and renal cancer (one trial).[7] A total of 7,769 patients were randomly assigned to an ICI treatment in 26 arms. One of these 26 arms combined two ICIs. The remaining 3,871 patients were treated with chemotherapy (11 arms), a molecularly targeted agent (one arm), a vaccine (one arm), and a placebo (three arms). Twelve (63%) of the 19 trials were conducted in patients having received at least one prior line of treatment in the recurrent and/or metastatic setting, whereas patients were treated in the first-line recurrent and/or metastatic setting in five trials (26%) and without restriction in terms of prior treatments in the two remaining trials (11%). PD-L1 expression was used for patient selection in two trials (11%). Crossover from the control arm to the ICI arm was allowed in five trials (26%).

Proportion of Durable Responses

The mean of the median follow-ups was 15.7 months (range, 5 to 38 months). The mean median PFS was 3.8 months (95% CI, 3.0 to 4.7 months) in the 26 ICI arms versus 3.5 months (95% CI, 2.9 to 4.0) in the 16 other arms (P = .98, Mann-Whitney U test). The mean proportion of patients who had a PFS that exceeded three times the median PFS was 25% (95% CI, 22% to 28%) in the ICI arms versus 11% (95% CI, 9% to 14%) in the other arms (P < .001, Mann-Whitney U test; Figure 3A; Table 2).

Figure 3.

Proportion of patients who had (A) a progression-free survival (PFS) that exceeded three times the median PFS and (B) an overall survival (OS) that exceeded two times the median OS according to drug classes. Each dot represents a treatment arm. Horizontal lines represent the means, and vertical lines represent the standard deviations. ICI, immune checkpoint inhibitor.

The mean median OS was 14.6 months (95% CI, 11 to 18 months) in the ICI arms versus 11.0 months (95% CI, 7.6 to 14 months) in the other arms (P = .03, Mann-Whitney U test). The mean proportion of patients who experienced an OS that exceeded two times the median OS was 30% (95% CI, 28% to 32%) in the ICI arms versus 23% (95% CI, 21% to 26%) in the other arms (P < .001, Mann-Whitney U test; Figure 3B; Table 2).

Predictors of Durable Responses

Because durable responses were well reported in patients with melanoma treated with ICIs,[15] we compared the proportion of durable responses in patients with and without melanoma. Estimations also were made in the subgroup of patients with NSCLC, given the high number of trials in this latter tumor type. Melanoma trials involved 4,649 patients treated in 13 ICI arms and four non-ICI arms (Table 1). Nonmelanoma trials involved 6,991 patients treated in 13 ICI arms and 12 non-ICI arms (Table 1). The mean proportion of patients in the ICI arms who had a PFS that exceeded three times the median PFS was 28% (95% CI, 22% to 34%) in the melanoma trials versus 23% (95% CI, 20% to 26%) in the nonmelanoma trials (P not significant, Mann-Whitney U test; Figure 4A; Table 2). The mean proportion of patients in the ICI arms who experienced an OS that exceeded two times the median OS was 30% (95% CI, 26% to 33%) in the melanoma trials versus 30% (95% CI, 26% to 33%) in the nonmelanoma trials (P not significant; Mann-Whitney U test). Results were similar when the subgroup of nonmelanoma trials was restricted to NSCLC trials (Table 2).

Figure 4.

Proportion of patients who had a progression-free survival (PFS) that exceeded three times the median PFS according to (A) tumor type (melanoma v nonmelanoma), (B) clinical setting (first v second line or greater of treatment), and (C) drug classes (anti–programmed cell death 1 [PD-1]/programmed death-ligand 1 [PD-L1] v anti–cytotoxic T-cell lymphocyte-4 [CTLA-4] agents). Each dot represents a treatment arm. Horizontal lines represent the means, and vertical lines represent the standard deviations.

We then evaluated whether the number of previous lines of treatment in the recurrent and/or metastatic setting affected the results. First-line setting trials involved 3,039 patients treated in seven ICI arms and four non-ICI arms (Table 1). Trials beyond the first-line recurrent and/or metastatic setting involved 6,991 patients treated in 14 ICI arms and 12 non-ICI arms (Table 1). The mean proportion of patients in the ICI arms who had a PFS that exceeded three times the median PFS was 30% (95% CI, 20% to 40%) in the first-line trials versus 23% (95% CI, 19% to 26%) in the beyond-the-first-line trials (P not significant, Mann-Whitney U test; Figure 4B; Table 2). The mean proportion of patients who had an OS that exceeded two times the median OS in the first-line setting could not be calculated because of a too-short follow-up.

We finally evaluated whether the proportion of durable responses depended on the target of the ICIs (PD-1/PD-L1 v CTLA-4). We excluded from this analysis the combination arm with nivolumab and ipilimumab. Trials with an anti–PD-1/PD-L1 agent involved 4,470 patients treated in 16 arms, whereas trials with an anti–CTLA-4 agent involved 2,658 patients treated in eight arms (Table 1). The mean proportion of patients treated with an anti–PD-1/PD-L1 agent who had a PFS that exceeded three times the median PFS was 28% (95%CI, 24% to 31%) versus 18%(95% CI, 15% to 21%) for patients treated with an anti–CTLA-4 agent (P < .001, Mann-Whitney U test; Figure 4C; Appendix Table A1). The mean proportion of patients who had an OS that exceeded two times the median OS was 31%(95%CI, 28%to 34%) in patients treated with an anti–PD-1/PD-L1 agent compared with 29% (95% CI, 26% to 32%) in patients treated with an anti–CTLA-4 agent (P not significant, Mann-Whitney U test; Appendix Table A1).

In multivariable analysis, the effect of treatment with anti–PD-1/PD-L1 agents (P < .001) and the effect of first-line treatment (P = .02) were statistically associated with a higher mean proportion of durable responses (Appendix Table A2).

Comments

3090D553-9492-4563-8681-AD288FA52ACE

processing....