222P - Artificial intelligence (AI) based prognostication from baseline computed tomography (CT) scans in a phase III advanced non-small cell lung cancer (aNSCLC) trial

Background

Baseline (BL) disease burden, measured via sum of longest diameters (SLD) of target lesions, is a known prognostic factor and contributes to evaluating treatment response in phase 1-3 solid tumor clinical trials. However, SLD may not comprehensively capture overall disease burden and thus may have limited prognostic utility. AI tools like imaging-based prognostication (IPRO), a fully automated deep learning pipeline independently trained on real-world CT scans and survival outcomes of people with aNSCLC, may enable more accurate prognostication and assessment of treatment effect.

Methods

We retrospectively examine how SLD and IPRO predict OS from 453 pre-treatment CT scans acquired in NExUS (NCT00449033), a phase 3 randomized controlled trial evaluating chemotherapy in combination with either sorafenib or placebo for first-line treatment of subjects with aNSCLC. The two study arms did not have a difference in OS and were combined for this study. For both SLD and IPRO, we measure the concordance index (c-index) and Kaplan-Meier curves stratified by covariate tertiles to estimate median overall survival (mOS). Hazard ratios (HRs) from proportional hazards models based on the standardized continuous values estimate the prognostic strength.

Results

The mOS was 12.0 months (95% CI: 10.3 – 13.2), 295 patients (64.8%) were male, and 401 (88%) were diagnosed as stage IV. At baseline, IPRO and SLD yielded a c-index of 0.64 and 0.60, respectively. The HR per 1SD change for IPRO (1.56) was 13% higher than for SLD (1.38). Table: 222P

Summary of mOS and HR for OS by SLD and IPRO

Conclusions

IPRO is more strongly correlated with OS compared to SLD at BL, and future work may prove IPRO to have greater clinical utility than traditional radiologic lesion measurements.

Clinical trial identification

NCT00449033.

Editorial acknowledgement

Legal entity responsible for the study

The authors.

Funding

Bayer.

Disclosure

O.F. Khan: Financial Interests, Personal, Invited Speaker: Pfizer, AstraZeneca, Gilead, Knight Therapeutics, Novartis; Financial Interests, Personal, Other, Research Support: Altis Labs, Inc. J. Montalt-Tordera: Financial Interests, Institutional, Full or part-time Employment: Bayer. J. Riskas, V. Sivan, O. Samorodova, J. Hennessy, A. Mitchell: Financial Interests, Personal, Full or part-time Employment: Altis Labs. S.A. Haider, S.S. Mohammadi, E. Di Tomaso, T. Banerji, C. Glaus: Financial Interests, Personal, Full or part-time Employment: Bayer. F. Baldauf-Lenschen: Financial Interests, Personal and Institutional, Leadership Role: Altis Labs.