Abstract CT246: An ensemble multi-omics framework for chemo de-escaltion NSCLC patients with PD-L1-low/negative disease in the PEOPLE phase II trial (NCT03447678)

Abstract Advanced non-small cell lung cancer (NSCLC) patients with PD-L1-low/negative expression face uncertainty regarding optimal treatment selection, as some derive meaningful benefit from pembrolizumab monotherapy while others require chemotherapy escalation. This study leverages multi-omic profiling from the PEOPLE Phase II trial (NCT03447678) to discover immune biomarkers that identify patients likely to benefit from immunotherapy alone, supporting precision-guided chemo de-escalation strategies for fragile patients.Among 87 patients treated with front-line pembrolizumab, circulating immune profiling (CIP), tumor gene expression panels (GEP), and gut metagenomics were collected and log₂-normalized. Compared to previously published work on the PEOPLE dataset, a novel population of atypical monocytes (CD14+CD33+CD56+CD16− HLA-DR) was identified thought an unsupervised analysis and included in the analysis. Immune profiling data, including this population, alongside transcriptomic and microbial data, was integrated using an ensemble analytical framework comprising three complementary methods: unsupervised multi-omic factorization (MOFA2), supervised multi-block classification (DIABLO in mixOmics), and patient-level similarity network fusion (SNF) for molecular subtype discovery. Remarkable concordance emerged across all three methods. Long survivors consistently demonstrated enrichment for NK cells (CD56dim CD16+), myeloid dendritic cells (CD1C+), T cell activation markers (CD3D, CD6, CD69), B-cell signatures (CD79A, BLK), cytotoxic markers (GZMK), and health-promoting gut bacteria (Phocaeicola massiliensis with short-chain fatty acid production). Conversely, short survivors exhibited elevated atypical monocytes (CD14+CD33+CD56+), pmn-MDSCs, neutrophils/granulocytes, IL1A expression, and dysbiotic microbial profiles (Lactobacilli, R. timonensis, S. mutans). SNF successfully identified a discrete poor-prognosis cluster characterized by this myeloid-driven inflammatory signature. This ensemble multi-omic approach reveals biologically meaningful patient stratification that transcends PD-L1 status alone, identifying a subpopulation with superior immunotherapy response despite low PD-L1 expression. The novel CD14+CD33+CD56+ atypical monocyte population represents a promising immunosuppressive biomarker warranting independent validation. These findings support precision-guided chemo de-escalation in PD-L1-low/negative advanced NSCLC, with prospective validation and predictive modeling as immediate next steps for clinical deployment. Citation Format: Luca Mauro Invernizzi, Ghazal Farhikhteh, Arianna Rigamonti, Sabrina Pervin Abu, Donato Cosi, Cecilia Silvestri, Luca Agnelli, Tiziana Triulzi, Marta Brambilla, Michela Francesconi, Francesco Sgambelluri, Mario Occhipinti, Maria Chiara Garassino, Roberta Mortarini, Filippo De Braud, Claudia Proto, Francesco Trovò, Monica Ganzinelli, Giuseppe Lo Russo, Arsela Prelaj. An ensemble multi-omics framework for chemo de-escaltion NSCLC patients with PD-L1-low/negative disease in the PEOPLE phase II trial (NCT03447678) [abstract]. In: Proceedings of the American Association for Cancer Research Annual Meeting 2026; Part 2 (Late-Breaking, Clinical Trial, and Invited Abstracts); 2026 Apr 17-22; San Diego, CA. Philadelphia (PA): AACR; Cancer Res 2026;86(8_Suppl):Abstract nr CT246.