PathAI And Genentech Present On AI-Powered Pathologic Response Assessment In NSCLC At The American Society Of Clinical Oncology Virtual Scientific Program 2021
Application of PathAI’s machine learning (ML) models to data from the Genentech-sponsored LCMC3 trial showed that model calculations of pathologic response (PathR) in non-small cell lung cancer (NSCLC) were robust and suggest that these models have the potential to support pathologists assessing PathR in clinical trials and in practice.
PathAI, a global provider of AI-powered technology applied to pathology, today announced that new data highlighting the application of its ML-based PathR algorithm to clinical trials as an aid to pathologists will be presented in the American Society of Clinical Oncology (ASCO) Virtual Scientific Program 2021, held from June 4-8, 2021. These results will be shared in full in the oral presentation, Artificial intelligence (AI)–powered pathologic response (PathR) assessment of resection specimens after neoadjuvant atezolizumab in patients with non-small cell lung cancer: Results from the LCMC3 study (Abstract #106) in the Clinical Science Symposium session, Artificial Intelligence: Optimizing Cancer Care Using Imaging and Pathology.
Recommended AI News: Qualys Announces Passing of Former CEO and Industry Visionary Philippe Courtot
The study describes ML-quantification of PathR, a clinically significant histologic endpoint that is currently calculated manually from tumor resection slides by pathologists. PathR, specifically MPR (10% or less viable tumor remaining), can be used as an efficacy endpoint in clinical trials investigating neoadjuvant therapies in patients with resectable NSCLC. MPR is studied as a potential surrogate efficacy outcome measure for disease free survival (DFS) or overall survival (OS).
“Through our partnership with PathAI, we are better understanding response and histopathologic changes in the tumor of lung cancer patients that received Tecentriq prior to surgery, and together developed a tool that could potentially support and simplify pathologists’ day to day work.” said Nai-Shun Yao, M.D., vice president and chief medical partner, Oncology at Genentech.
ML-quantification of tumor histology features can provide a reproducible assessment of percent viable tumor that was shown to be at least as accurate as manual assessment. These ML-based tools have the potential to impact how PathR is assessed in clinical trials, potentially providing a rapid and more scalable solution to MPR determination. PathAI models were trained to identify and quantify cells and tissues within NSCLC tumor samples collected following neoadjuvant atezolizumab therapy, as part of the LCMC3 clinical trial sponsored by Genentech, a member of the Roche Group, and from those measurements, determine PathR (digital PathR). ML-model assessed MPR strongly agreed with manual MPR (AUROC = 0.975) and showed comparable DFS and OS for both digital PathR and manual PathR, however it should be noted that DFS and OS data in LCMC3 are still immature. Digitally assessed MPR was significantly associated with better DFS and OS, whereas manually assessed MPR showed a non-significant trend toward better DFS and OS. These results suggest that ML-based quantification of PathR after neoadjuvant immunotherapy may serve as surrogate for survival outcome measures, but further data maturation and validation is needed.
This investigation, together with another also presented at ASCO 2021 (poster #3061), demonstrate PathAI’s multi-faceted approach toward integrating scalable AI-powered tools that generate clinically relevant, accurate, and reproducible pathology scores into oncology clinical trial workflows.