199P - Novel machine learning (ML) algorithm to predict immunotherapy response in small cell (SCLC) and non-small cell (NSCLC) lung cancer

Background

Immune checkpoint inhibitors (ICI) are used to manage patients with SCLC and NSCLC lung cancer. Yet, response rates are often low and identifying patients that will benefit from ICIs can be challenging. Accurate and accessible tools that predict ICI responses could enable a precision medicine approach that improves patient outcomes. This study aimed to use ML to predict response to ICI therapies in patients with lung cancer based on clinically available data.

Methods

334 eligible records were reprocessed using one hot encoding. 161 patients had complete datasets available. Differences in data distribution were handled using the Synthetic Minority Oversampling Technique. Six ML algorithms were trained, including Linear regression, Support Vector Classifier, XGBoost Classifier, Random Forest, Decision Tree and Gaussian Naïve Bayes Classifier. The algorithms used 80% of the training data, were tested on 20% of validation data and used the Grid Search Cross-Validation technique for hyperparameter optimization.

Results

Of the 161 patients, 9% had SCLC and 80% had NSCLC. Patients receiving Pembrolizumab, Nivolumab and Atezolizumab comprised 62%, 11% and 25% respectively. XGBoost Classifier predicted response with the most accuracy, 64%. The artificial intelligence (AI) algorithm predicted and stratified ICI response better than PD-L1 levels alone (Table). The model showed good performance status, female gender and adenocarcinoma sub-type predicted response to ICI. Conversely, M1, N2 staging, male gender, squamous cell carcinoma sub-type and receiving Atezolizumab were predictive of disease progression. Table: 199P

Conclusions

Multiple novel ML models, developed using clinically available data, showed that ICI type, histopathology sub-type and TMN staging impact ICI response in lung cancer. Future studies will seek to include more SCLC cases and compare the prediction accuracy among the three ICIs.

Clinical trial identification

Editorial acknowledgement

Jane Webb for data access from electronic medical records Denny Wong for data cleaning and pre-processing.

Legal entity responsible for the study

The authors.

Funding

Has not received any funding.

Disclosure

V. Balaji: Financial Interests, Personal, Full or part-time Employment: Curenetics. V. Tailor: Financial Interests, Personal, Full or part-time Employment: CurenX. G. Powell: Financial Interests, Personal, Full or part-time Employment: Curenetics. R. Shah: Financial Interests, Personal, Advisory Role: Boehringer Ingelheim, AstraZeneca , Roche, Bristol Myers Squibb, MSD, Pfizer, Lilly, Novartis, Takeda, Bayer, BeiGene, Guardant Health, Sanofi, EQRx. O.M. Adeleke: Financial Interests, Personal, Stocks or ownership: Curenetics. All other authors have declared no conflicts of interest.