145P - Identification of next generation sequencing (NGS)-based genomic signature predicting resistance to immunotherapy (IO) in patients (pts) with metastatic non-small cell lung cancer (mNSCLC): A single-center cohort study

Background

IO ± chemotherapy (CT) is first-line therapy for mNSCLC without actionable mutations. We aimed to identify a genomic signature potentially predicting resistance to IO using NGS.

Methods

Pts with mNSCLC who received IO or CT/IO were eligible. NGS was performed by TSO500HT® assay (DNA [522 genes], RNA [55 genes], Tumor Mutational Burden, Microsatellite Instability) or Oncomine Focus Assay® (DNA, 35 genes) plus Archer’s FusionPlex Lung Panel® (RNA, 17 genes), according to quality/quantity cellularity. A least absolute shrinkage and selection operator (LASSO) Cox regression was conducted to choose genes. The correlation between LASSO-selected genes and progression-free survival (PFS) was evaluated by univariable Cox regression. A risk score was derived using regression coefficients of significant genes in the multivariable analysis. Pts were divided into two groups based on the median risk score (0 vs >0). The risk score was evaluated in a multivariable model including clinical features.

Results

56 pts were retrospectively included. 32 (57%) received CT/IO and 23 (43%) IO. Median follow-up was 10.4 months. Among 346 genes, 15 resulted potentially predictive of PFS. At univariable analysis, 6 genes were significantly correlated to PFS: AKT3 (HR 40.39, 95% IC 5.58-292.30, p<0.001), ARID5B (HR 4.98, 95% IC 1.11-22.24, p=0.04), CD276 (HR 14.33, 95% IC 2.60-78.99, p=0.002), PAX8 (HR 8.33, 95% IC 1.78-38.95, p=0.007), PDGFRB (HR 4.88, 95% IC 1.42-16.75, p=0.01) and RICTOR (HR 3.82, 95% IC 1.08-13.57, p=0.04). In the multivariable analysis PAX8 was excluded and a genomic signature with 5 genes was considered. PFS was significantly different in the two risk groups: HR 13.56, 95% IC 5.30-34.67, p<0.001, with a median PFS of 1.64 (95% IC 0.76-NA) vs 13.25 months. (95% IC 10.88-NA). The genomic signature remained prognostic in a multivariable model including age, smoke, histology and regimen (HR 15.71; 95% IC 5.19-47.52, p<0.001).

Conclusions

A genomic signature including AKT3, ARID5B, CD276, PDGFRB and RICTOR alterations was significantly associated with shorter PFS in pts receiving IO. Validation in external cohort is ongoing.

Editorial acknowledgement

Clinical trial identification

Legal entity responsible for the study

Medical Oncology Unit, Fondazione Policlinico Agosto Gemelli.

Funding

Has not received any funding.

Disclosure

All authors have declared no conflicts of interest.