Artificial intelligence applied to routine pathology slides may help identify which patients with rare cancers are responding to pembrolizumab within weeks of starting treatment, according to a new study that analyzed changes in the tumor microenvironment before and during therapy.
Researchers evaluated more than 500 biopsy specimens collected from 84 patients enrolled in a phase II pembrolizumab trial covering 10 rare tumor types. A deep learning system analyzed standard hematoxylin and eosin (H&E)-stained whole-slide images to measure two tissue features: the density of immune cells infiltrating the tumor and the proportion of tissue occupied by tumor cells. Rather than relying only on pretreatment findings, the investigators compared baseline biopsies with samples collected about two to three weeks after treatment began.
The findings, published in Journal for Immunotherapy of Cancer, suggest that dynamic tissue changes may be more clinically informative than a single baseline assessment. Patients whose follow-up biopsies showed increasing immune-cell infiltration and decreasing tumor content experienced better outcomes than those without these changes. Baseline immune-cell density alone appeared useful only in tumor types that already had relatively high immune infiltration, highlighting the value of serial tissue assessment rather than one-time biomarker measurement.
The study underscores the growing role of AI-assisted digital pathology as a quantitative companion to conventional slide review. Because the analysis was performed on routine H&E sections, the approach could be easier to implement than more specialized immune profiling methods while providing reproducible measurements of tissue changes over time. The investigators also reported that AI-derived measurements corresponded with findings from multiplex immunofluorescence, supporting the biological relevance of the image-based analysis.
The authors suggest that early repeat biopsies could eventually help guide treatment decisions by identifying patients who are benefiting from immunotherapy before conventional imaging provides clear answers. Such information could support earlier treatment modification for nonresponders while avoiding unnecessary exposure to ineffective therapy.
Lead researcher Aung Naing, from the University of Texas MD Anderson Cancer Center, said, "While this AI-powered approach needs validation, this is an exciting step forward because it shows that meaningful insights can be extracted from routine pathology samples across a diverse group of rare cancers."
Newsletters
Receive the latest pathologist news, personalities, education, and career development – weekly to your inbox.
