Deep learning identifies pathobiological features within H&E images associated with genomic alterations and progression on anti-PD(L)1 in HUDSON, an AstraZeneca-sponsored Phase II clinical trial

• Novel biomarkers are needed to better predict which patients will
respond to immunotherapy.

• Machine learning (ML) models have the potential to quantitatively
characterize the tumor and tumor microenvironment (TME). However,
most ML models require training to be done on a subset of samples from
a novel dataset before deployment on the remainder of the dataset. This
training-test approach requires a larger number of samples than is
generally available for small clinical trials. There is a need for pre-trained
ML models that can be applied to small datasets.

• PathAI previously trained ML models on NSCLC samples from
commercial and clinical datasets to identify and quantify cellular
composition, tissue architecture, and blood vessel features in the TME.

• Here we deployed PathAI’s ML models to H&E images from HUDSON
(NCT03334617), an AstraZeneca Phase II platform clinical trial, to
identify morphologic features associated with genomic alterations and
time to progression on anti-PD(L)1 therapies.