Type of publication:

Conference abstract

Author(s):

Gandara D.R.; Carbone D.P.; Dicker A.P.; Christopoulos P.; Puzanov I.; Jain P.; Farrugia D.; Brown S.; Moskovitz M.; Bar J.; Hassani A.; *Chatterjee A.; Abu-Amna M.; Polychronis A.; Brewster A.; Lou Y.; VanderWalde N.A.; Gottfried M.; Lahav C.; Lowenthal G.; Sela I.; Harel M.; Elon Y.; Schneider M.A.

Citation:

Journal of Thoracic Oncology. Conference: The 2024 World Conference on Lung Cancer. San Diego United States. 19(10 Supplement) (pp S354-S355), 2024. Date of Publication: October 2024.

Abstract:

Introduction: Small cell lung cancer (SCLC) is an aggressive disease with limited treatment options. Immune checkpoint inhibitor (ICI) therapy with concurrent chemotherapy is the preferred first-line treatment for patients with extensive-stage SCLC. However, the addition of ICIs to chemotherapy only modestly improves clinical outcomes while posing a risk of ICI-related toxicities. Thus, identifying patients likely to benefit from ICIs is critical for optimizing treatment decisions. Here, we describe a test derived from a novel computational model that analyzes pretreatment plasma proteomic profiles to predict clinical outcomes in patients with SCLC receiving ICI-based therapies. Method(s): An observational study collected pretreatment plasma samples from 79 patients with extensive-stage SCLC treated with ICI-based therapy (NCT04056247). Proteomic profiling of plasma samples was performed using aptamer-based technology, measuring approximately 7000 proteins per sample. A machine learning model was developed to predict the clinical benefit (CB) from ICI-based therapy, where CB was defined as 6-month progression-free survival. Given the limited cohort size, CB prediction was achieved by integrating two computational models. Model 1, based on 146 plasma proteomic biomarkers, was developed from the SCLC dataset using cross-validation. Model 2, based on a 4-protein signature, was developed from a previously reported NSCLC dataset (Christopoulos et al. JCO prec. onc. 2024). The hybrid model stratified patients into two groups (i.e., 'positive' or 'negative') based on a pre-defined CB probability threshold. Bioinformatic analysis of the SCLC-specific proteomic biomarkers was performed to gain insight into the potential mechanisms driving ICI therapeutic benefit and resistance in SCLC. Result(s): The model displayed a robust predictive capability, as demonstrated by the area under the curve (AUC) of the receiver operating characteristic (ROC) plot of 0.63 (p-value = 0.02) and a high correlation between the predicted CB (i.e., model output) and the observed CB rate (R2 = 0.93). Furthermore, overall survival (OS) was significantly longer in patients stratified to the positive group compared to those in the negative group. Median OS was 14 months versus 8.8 months in positive versus negative patient groups (Hazard ratio = 0.47, 95% Confidence interval: 0.25-0.90, p-value = 0.02). Bioinformatic analysis of model proteins revealed significant enrichment of lung tumor-associated proteins, poor prognostic factors in lung cancer, extracellular matrix-related proteins, intermediate filaments, and replicative immortality (Fisher exact test; FDR<0.1). Multiple model proteins are also known to be involved in fibroblast growth factor signaling and glutathione metabolism. Given their association with different treatment resistance mechanisms, such proteins represent potential targets for intervention. Conclusion(s): We describe preliminary results from a novel pretreatment plasma proteomics-based predictive model that can potentially inform treatment decisions for patients with SCLC. Bioinformatic analysis demonstrates that the model is based on a composite of biologically and clinically relevant biomarkers. The potential clinical utility of this model is being investigated in a large prospective clinical trial.