Acquired Resistance to Immune Checkpoint Inhibitors Across Tumor Types: A Reconstructed Individual Patient Data Analysis.

BACKGROUND

Some patients with cancer who initially benefit from checkpoint inhibitors (CPIs) achieve lasting remission or cure. Acquired resistance to immunotherapy (ARI) refers to disease progression occurring after an initial period of clinical benefit from immune checkpoint blockade. Despite its clinical relevance, the incidence of ARI across tumor types remains largely unknown. We aimed to estimate the overall incidence of ARI across tumor types and examine its association with tumor response.

METHODS

ARI rates were estimated using a pooled analysis of digitized progression-free survival (PFS) curves from EMA- and FDA-approved phase III trials of CPIs in solid tumors. Only tumors with an overall response rate (ORR) to CPIs ≥20% were considered. PFS ≥6 months was used as a proxy for response to CPIs, based on the Society for Immunotherapy of Cancer (SITC) definition of ARI. Cancer mortality data from GLOBOCAN 2022 were used to estimate the annual incidence of advanced disease across tumor types.

RESULTS

Reconstructed pseudo-individual patient data from 15,199 patients across 11 tumor types and 41 phase III trials were included. The estimated pan-cancer weighted rate of ARI was 73.5% (95% CI, 71.6-75.4) at 3 years, ranging from 25.4% in deficient mismatch repair colorectal cancer to 83.1% in squamous esophageal carcinoma; approximately 1,001,815 patients per year may develop ARI globally. RECIST 1.1 complete response (CR), partial response (PR), and stable disease (SD) rates correlated with ARI (CR and PR: ρ = -0.63 and -0.66; P=.044 and .031, respectively; SD: ρ = +0.68; P=.025), and these associations remained consistent after adjusting for tumor type.

CONCLUSIONS

A considerable proportion of patients who respond to CPIs develop ARI. The variability in ARI incidence across tumor types is explained by the rates of CR, PR, and SD, indicating that the strength of the antitumor response, rather than tumor histology, informs ARI rates.