Search

11565 Background: Predicting response to checkpoint inhibitors (CPIs) using biological knowledge-based decision processes with machine learning (ML) has a great potential to predict rapid progression in patients treated with checkpoint inhibitors (CPIs) (hyperprogressive disease (HPD)) as well as responders. ML models risk overfitting data and do not always evaluate the underlying biology, thus performing well in the initial training cohort but lack generalizability when extended to other cohorts. Biology-based decision may not perform as well initially due to limited understanding and a simplified rule set, but often perform equally well when extended to larger similar cohorts of patients. Methods: A custom NGS cancer immune gene expression assay compared 87 patients treated with CPIs classified as CR, PR, or SD versus 12 HPD. A ML-based polynomial regression model based on 54 immune-related genes combined with mutational burden was optimized for prediction of response. A biological 4-gene decision tree model was constructed independently based on ML. A second biological decision tree incorporated the weighted average relative rank of the expression of multiple genes in 4 different immune functions including immune cell infiltration, regulation, activation, and cytokine signaling. Bayesian model average (BMA) incorporated all three models’ results into the final prediction. Results: For87 patients classified as CR, PR, or SD the PPV >96% for responders and a NPV >90% for non-responders was achieved with the regression model, however with response indeterminate for 24% of the population. While the two biological decision tree models’ PPV were in the 70% range, they accurately revealed the critical genes’ roles in immune response with strong literature support. BMA process integrated these three models resulted in a PPV >96% and a NPV >90% and eliminated the indeterminate group. For HPD a unique biology related to priming of short term memory T-cells was identified. Conclusion: Prediction of response to CPIs is best attained by combining ML with biological knowledge. Decision tree models using a large panel of immune related genes in the context of archival samples from patients treated with CPIs can be used to better understand the biology of responders versus non-responders and provides new insights into HPD.

e14590 Background: The benefit of CPI is limited to a subgroup of patients. Significant toxicity and financial burden make it imperative to identify biomarkers predictive of clinical benefit. We aimed to analyze if immune related adverse events (irAE), laboratory parameters (besides PD-L1 status), prior chemotherapy and concomitant medication use could predict response to CPI. Methods: 216 consecutive patients treated at RPCI with single/multiple CPI from 2011 to 2016 were identified. Demographics, irAE (CTCAE v4.03), concomitant medication use, response (RECIST v1.1) and laboratory parameters were recorded. Progression free (PFS) and overall survival (OS) were calculated. Clinical benefit (responder) was defined as objective response or stable disease with PFS ≥ 6 months. Statistical analysis was performed using SAS v9.4. Results: 185/216 patients (86%) had clinical outcomes with first CPI available (55.7% melanoma, 28.1% lung, 10.8% kidney, 3.8% bladder, 1.6% others), 14% died before disease assessment. Median age was 61 years; 64% were males. 31% were responders (R); 69% non-responders (NR). Median PFS and OS for R were not reached and for NR were 4.7/7 months (p < 0.001). 26% R developed rash vs. 12% NR (p = 0.02). irAE diarrhea was associated with better OS (p = 0.03); rash with better PFS (p = 0.007) and OS (p = 0.05). Concomitant beta-blocker/NSAID use was noted in 35%/28% R vs 20%/12% NR (p = 0.04; p = 0.01). Higher median absolute eosinophil (AEC) and absolute lymphocyte count (ALC) were noted in R (400 and 2000/µl in R vs. 300 and 1600/µl in NR; p = 0.008 and 0.024). NR were noted to have a higher median lactate dehydrogenase (LDH) of 565 vs. 503 (p < 0.001). AEC > 100/µl and ALC > 1000/ µl were associated with better OS (P = 0.03; P = 0.001). Absolute neutrophil count (ANC) > 7500/µl and leucocyte count > 8750/ µl were associated with inferior PFS (p = 0.005; P = 0.02) and OS (p = 0.002; p = 0.03). Pre-CPI chemotherapy did not correlate with response to CPI. Conclusions: Among irAE, diarrhea and rash predicted improved outcomes. Eosinophilia, lymphocytosis, neutrophilia, leukocytosis and LDH could serve as potential prognostic biomarkers to CPI. Prospective studies are warranted to validate these findings.