Ipilimumab with or without nivolumab in PD-1 or PD-L1 blockade refractory metastatic melanoma: a randomized phase 2 trial

Abstract

In this randomized phase 2 trial, blockade of cytotoxic T-lymphocyte protein 4 (CTLA-4) with continuation of programmed death protein 1 (PD-1) blockade in patients with metastatic melanoma who had received front-line anti-PD-1 or therapy against programmed cell death 1 ligand 1 and whose tumors progressed was tested in comparison with CTLA-4 blockade alone. Ninety-two eligible patients were randomly assigned in a 3:1 ratio to receive the combination of ipilimumab and nivolumab, or ipilimumab alone. The primary endpoint was progression-free survival. Secondary endpoints included the difference in CD8 T cell infiltrate among responding and nonresponding tumors, objective response rate, overall survival and toxicity. The combination of nivolumab and ipilimumab resulted in a statistically significant improvement in progression-free survival over ipilimumab (hazard ratio = 0.63, 90% confidence interval (CI) = 0.41-0.97, one-sided P = 0.04). Objective response rates were 28% (90% CI = 19-38%) and 9% (90% CI = 2-25%), respectively (one-sided P = 0.05). Grade 3 or higher treatment-related adverse events occurred in 57% and 35% of patients, respectively, which is consistent with the known toxicity profile of these regimens. The change in intratumoral CD8 T cell density observed in the present analysis did not reach statistical significance to support the formal hypothesis tested as a secondary endpoint. In conclusion, primary resistance to PD-1 blockade therapy can be reversed in some patients with the combination of CTLA-4 and PD-1 blockade. Clinicaltrials.gov identifier: NCT03033576 .

Extended Data Figure 1.

Kaplan-Meier plot of overall survival.

Extended Data Figure 2.

CONSORT diagram of tumour biopsies assessed for CD8-positive cell quantitation.

Extended Data Figure 3.. Semi-automated workflow for CD8-positive cell quantitation in tumour biopsies.

A) Representative images for manual annotation of tumour and tumour periphery regions. B) Screenshot of cell segmentation, identifying CD8-positive cells (red) and CD8-negative cells (blue), for CD8 cell quantitation.

Extended Data Figure 4.. Model for the mechanism of anti-CTLA-4-mediated sensitization of tumours to anti-PD-1.

Releasing the CTLA-4 checkpoint at the lymph node to promote T-cell trafficking (A to B) and expansion of CD8 T cells by concurrent anti-PD-1 therapy at the tumour site (B to C).

Extended Date Figure 5.

Model for the mechanism of reversal of tumour resistance to anti-PD-1 with the addition of anti-CTLA-4.

Figure 1.. CONSORT diagram.

The diagram includes patient enrolment, randomisation and follow up. All eligible patients who were randomised were included in the intention-to-treat population. The safety population included all patients who received at least one dose of treatment that they were randomly assigned.

Figure 2.. Analysis of progression-free survival (PFS).

A) Kaplan-Meier estimates of PFS as assessed by local investigators. The 6-month PFS estimates were 34% (90% CI: 25%−43%) and 13% (4%−27%) for the combination therapy versus ipilimumab alone groups, respectively. B) Forest plot for PFS according to subgroups. The hazard ratio (HR) and 90% confidence interval (CI) from a Cox regression model are reported and represented by the solid black squares and error bars; no adjustment was made for multiple comparisons.

Figure 3.. Depth of response and response duration.

A and B) Waterfall plots of overall response in the nivolumab and ipilimumab group (A) and the ipilimumab group (B). The plots show the change in RECIST target lesions. Lines indicate threshold for objective response (≥30% decrease) or disease progression (>20% increase). Asterisks and hatched bars denote patients whose best response was progression due to new lesions (n=23 with nivolumab and ipilimumab, and n=14 with ipilimumab) or clear worsening of non-measurable disease (n=2 with nivolumab and ipilimumab) without assessable RECIST changes, symptomatic deterioration (n=1 with nivolumab and ipilimumab, and n=1 with ipilimumab), or death due to disease (n=2 with nivolumab and ipilimumab, and n=1 with ipilimumab). Two patients on the combination group with changes in RECIST measurements greater than 100% over baseline are denoted on the far left with hatched bars capped at 100%. Two patients on the combination group did not have complete follow-up disease assessment data and were not included. C) Swimmer’s plot of the course of patients with an objective response to therapy. Bars represent progression-free survival (PFS) from time of registration. Patients with arrows are alive and progression-free.

Figure 4.. CD8 cell quantitation in biopsy specimens.

A) Density of CD8-positive cells detected within the annotated tumour. Paired baseline-on-treatment biopsies are indicated by points connected by lines. Dotted lines indicate patients with observed features of pathological response present in the on-treatment biopsy. The number of patients in each group is indicated by the number along the x-axis (Ipilimumab, N=2 partial response [PR], N=3 stable disease [SD], N=13 progressive disease [PD]; nivolumab and ipilimumab, N=8 complete response [CR], N=9 PR, N=9 SD, N=27 PD). Box plots indicate the median (middle line), 25^th and 75^th percentiles (box) and 5^th and 95^th percentiles (whiskers). CD8-positive cell density was compared across biopsies using Wilcoxon rank-sum for two-group comparisons and Wilcoxon signed-rank tests for paired comparisons. B) Representative images of biopsies. Biopsy of a patient with response to nivolumab and ipilimumab with increased CD8 density (left panel); biopsy from a patient with response to nivolumab and ipilimumab and detection of features of pathological regression in the on-treatment biopsy with decrease in CD8 density in that area (middle); and biopsy from a patient with progressive disease following nivolumab and ipilimumab with no change in CD8 cell density (right). Scale bars in each panel are 100um.