Background: Immune checkpoint inhibitors (ICIs) are now a therapeutic standard in advanced non-small cell lung cancer (NSCLC), but strong predictive markers for ICIs efficacy are still lacking. We evaluated machine learning models built on simple clinical and biological data to individually predict response to ICIs. Methods: Patients with metastatic NSCLC who received ICI in second line or later were included. We collected clinical and hematological data and studied the association of this data with disease control rate (DCR), progression free survival (PFS) and overall survival (OS). Multiple machine learning (ML) algorithms were assessed for their ability to predict response. Results: Overall, 298 patients were enrolled. The overall response rate and DCR were 15.3% and 53%, respectively. Median PFS and OS were 3.3 and 11.4 months, respectively. In multivariable analysis, DCR was significantly associated with performance status (PS) and hemoglobin level (OR 0.58, p < 0.0001; OR 1.8, p < 0.001). These variables were also associated with PFS and OS and ranked top in random forest-based feature importance. Neutrophil-to-lymphocyte ratio was also associated with DCR, PFS and OS. The best ML algorithm was a random forest. It could predict DCR with satisfactory efficacy based on these three variables. Ten-fold cross-validated performances were: accuracy 0.68 ± 0.04, sensitivity 0.58 ± 0.08; specificity 0.78 ± 0.06; positive predictive value 0.70 ± 0.08; negative predictive value 0.68 ± 0.06; AUC 0.74 ± 0.03. Conclusion: Combination of simple clinical and biological data could accurately predict disease control rate at the individual level.

Background: Resistance to PD1/L1 immune checkpoint inhibitors (ICIs) in advanced NSCLC patients is observed in about 80% of individuals with no robust predictive biomarker yet. The PIONeeR trial (NCT03493581) aims to predict such resistances through a comprehensive multiparametric biomarkers analysis. Methodology: Among the &gt;300 advanced NSCLC patients (pts) recruited in PIONeeR, we focused on the first 137 ≥2nd line ECOG PS0-1 pts treated with single-agent nivolumab, pembrolizumab or atezolizumab. Tumor tissue was collected at baseline and pts were re-biopsied at 6 weeks, and blood-sampled every cycle throughout the 24 weeks post C1D1. Response to PD1/L1 ICIs was assessed by RECIST 1.1 every 6 weeks. Immune contexture was characterized in tumor & blood of each pt through FACS for circulating immune cell subtypes quantification and endothelial activation, blood soluble factors dosage, dual- & multiplex IHC/digital pathology to quantify immune cells infiltrating the tumor, WES for TMB & ICI plasma dosage, leading to 331 measured biomarkers in addition to routine clinical parameters. Multivariable (MV) logistic regression was used to examine the association of each biomarker (controlled by sex, age, smoking status, histological type & PDL1+ Tumor Cells) with the risk of Early Progression (EP), i.e. within 3.5 months of treatment. Multivariable Cox regression analysis was conducted for association with PFS and OS. Results: Overall, the 137 pts were mainly male (64%), smokers (92%) and &lt;70yrs (68%). Tumors were mainly non-squamous (79%) with &gt;1% PDL1+ TC in 36% of the cases, and 21% of pts were still on treatment at data cut-off. Archived samples were available for 80% of pts at inclusion and re-biopsy was available in 52.9% of these cases. The median follow up was 19.8 months, 22.5% of pts did not progress at data cut-off while 62% presented EP. Tumor Cytotoxic T-cells density, especially PD1+ were lower in EP (MV OR=0.45, p=0.022); conversely, higher proportions of circulating cytotoxic T-cells and activated T-cells (HLA-DR+) were observed in EP (MV OR=3.8, p&lt;0.001). Among other biomarkers, Tregs (MV OR=0.44, p=0.018), NK cell subsets (MV OR≤0.44, p&lt;0.05), albumin (MV OR=0.4, p&lt;0.01) and PDL1 TC % (MV OR=0.27, p&lt;0.01) were decreased whereas alkaline phosphatase was increased (OR=3, p=0.018). &gt;65% inter-pt variability was observed in plasma exposures for all ICIs, with 8-10% of pts displaying trough levels below the target engagement threshold. Data will be presented through unsupervised clustering algorithms & multi-modal supervised learning methods. Changes after 6 weeks of treatment will be analyzed to further investigate drugs mechanisms of action. Conclusion: The PIONeeR trial provides with the 1st comprehensive biomarkers’ analysis to establish predictive models of resistance in advanced NSCLC pts treated with PD1/L1 ICIs and highlights how tumor and circulating biomarkers are complementary. Citation Format: Laurent Greillier, Florence Monville, Vanina Leca, Frédéric Vely, Stephane Garcia, Joseph Ciccolini, Florence Sabatier, Gilbert Ferrani, Nawel Boudai, Lamia Ghezali, Marcellin Landri, Clémence Marin, Mourad Hamimed, Laurent Arnaud, Melanie Karlsen, Kevin Atsou, Sivan Bokobza, Pauline Fleury, Arnaud Boyer, Clarisse Audigier-Valette, Stéphanie Martinez, Hervé Pegliasco, Patrice Ray, Lionel Falchero, Antoine Serre, Nicolas Cloarec, Louisiane Lebas, Stephane Hominal, Patricia Barre, Sarah Zahi, Ahmed Frikha, Pierre Bory, Maryannick Le Ray, Lilian Laborde, Virginie Martin, Richard Malkoun, Marie Roumieux, Julien Mazieres, Maurice Perol, Eric Vivier, Sebastien Benzekry, Jacques Fieschi, Fabrice Barlesi. Comprehensive biomarkers analysis to explain resistances to PD1-L1 ICIs: The precision immuno-oncology for advanced non-small cell lung cancer (PIONeeR) trial [abstract]. In: Proceedings of the American Association for Cancer Research Annual Meeting 2022; 2022 Apr 8-13. Philadelphia (PA): AACR; Cancer Res 2022;82(12_Suppl):Abstract nr LB120.

BackgroundImmune checkpoint inhibitors (ICIs) are now a therapeutic standard in advanced non-small cell lung cancer (NSCLC), but strong predictive markers for ICIs efficacy are still lacking. We evaluated machine learning models built on simple clinical and biological data to individually predict response to ICIs.MethodsPatients with metastatic NSCLC who received ICI in second line or later were included. We collected clinical and hematological data and studied the association of this data with disease control rate (DCR), progression free survival (PFS) and overall survival (OS). Multiple machine learning (ML) algorithms were assessed for their ability to predict response.ResultsOverall, 298 patients were enrolled. The overall response rate and DCR were 15.3 % and 53%, respectively. Median PFS and OS were 3.3 and 11.4 months, respectively. In multivariable analysis, DCR was significantly associated with performance status (PS) and hemoglobin level (OR 0.58, p<0.0001; OR 1.8, p<0.001). These variables were also associated with PFS and OS and ranked top in random forest-based feature importance. Neutrophils-to-lymphocytes ratio was also associated with DCR, PFS and OS. The best ML algorithm was a random forest. It could predict DCR with satisfactory efficacy based on these three variables. Ten-fold cross-validated performances were: accuracy 0.68 ± 0.04, sensitivity 0.58 ± 0.08; specificity 0.78 ± 0.06; positive predictive value 0.70 ± 0.08; negative predictive value 0.68 ± 0.06; AUC 0.74 ± 0.03.ConclusionCombination of simple clinical and biological data could accurately predict disease control rate at the individual level.Highlights-Machine learning applied to a large set of NSCLC patients could predict efficacy of immunotherapy with a 69% accuracy using simple routine data-Hemoglobin levels and performance status were the strongest predictors and significantly associated with DCR, PFS and OS-Neutrophils-to-lymphocyte ratio was also associated with outcome-Benchmark of 8 machine learning models