Inflamed immune phenotype as a novel predictive marker of immune checkpoint inhibitors for non-small cell lung cancer
Martina Carullo, Carlotta Antoniotti, Changho Ahn, Chiyoon Oum, Valentina Angerilli, Francesca Bergamo, Carolina Sciortino, Alessandra Boccaccino, Maria Alessandra Calegari, Jessica Gasparello, Stefano Tamberi, Rossana Intini, Mario Scartozzi, Camilla Damonte, Chiara Citterio, Lisa Salvatore, Chiara Boccaccio, Matteo Fassan, Chan-Young Ock, Chiara Cremolin
ESMO, 2025
Abstract
Background T-cell infiltration in tumors and the surrounding stroma has been suggested as a prognostic biomarker in non-small cell lung cancer (NSCLC) patients treated with immune checkpoint inhibitors (ICIs). This study aims to assess the predictive role of an artificial intelligence (AI)-based spatial tumor-infiltrating lymphocytes (TIL) density assessment and immune phenotype (IP) classification in advanced NSCLC.
Methods This multicenter prospective observational study included patients with stage II to IV or recurrent lung cancer (LC-SCRUM-IBIS). For spatial TIL analysis, we used Luit SCOPE IO, an AI-powered WSI analyzer of TIL densities in the tumor microenvironment that can define four immune phenotypes (IPs): inflamed, immune-excluded, immune-desert, and non-inflamed.
Results Between February 2017 and May 2018, 1,017 lung cancer patients were enrolled. Of these patients, 482 were NSCLC patients who had submitted adequate tissue samples for TIL analysis and were treated with ICIs. Among patients, the median age was 64 years, 27% were female, 79% were ever smokers, stage II/III/IV/recurrence; 9/91/330/52, Non-Sq/Sq; 357/125, and inflamed/immune-excluded/immune-desert/non-inflamed; 74/79/242/87. Median (m) progression-free survival (PFS) and overall survival (OS) were 9.6/31.9 months with inflamed IP, 3.4/13.5 months with immune-excluded IP, 3.7/14.7 months with immune-desert IP, and 2.9/14.6 months with non-inflamed IP, respectively. Also, patients with inflamed IP showed longer mPFS (9.6 vs. 3.4 months, HR 0.56, 95% CI: 0.43 – 0.75, P < 0.001) and mOS (31.9 vs. 14.6 months, HR 0.51, 95% CI: 0.37 – 0.69, p < 0.001) compared to those with other IPs. However, in patients who received chemotherapy (n = 295), there was no difference in mPFS (7.9 vs. 8.1 months, HR 0.83, 95% CI: 0.53 – 1.30, p = 0.4) and mOS (43.4 vs. 23.6 months, HR 0.65, 95% CI: 0.39 – 1.10, p = 0.1) for inflamed IP compared to other IPs.
Conclusions Inflamed IP demonstrated favorable PFS and OS in NSCLC patients treated with ICIs, comparable to previously reported results in NSCLC patients with high PD-L1 expression. AI-powered spatial TIL analysis may offer a novel predictive marker of ICIs for NSCLC.