Evaluating tumor geography on routine H&E slides identifies high-risk recurrence patterns that volume-based assessments miss, according to a Mayo Clinic study.
Mayo Clinic researchers have found that artificial intelligence (AI)-enabled spatial analysis can identify patterns in routine pathology slides to help clinicians determine which patients with pancreatic cancer are at greater risk of recurrence after treatment and surgery, according to a study published in Clinical Cancer Research.
The findings indicate that examining how residual cancer is organized—rather than simply measuring how much tumor tissue remains—could refine recurrence risk estimates, particularly for patients whose tumors show limited response to chemotherapy prior to surgical resection, according to the study. Investigators noted that patients exhibiting a more fragmented, intermixed pattern of cancer and surrounding stroma experienced earlier disease recurrence, whereas the volume of residual cancer alone failed to reliably distinguish risk levels.
“Current pathology assessments largely tell us how much tumor is left after treatment. We wanted to know whether the geography of that remaining cancer could reveal additional biology about recurrence risk,” says Ryan Carr, MD, PhD, a Mayo Clinic oncologist and senior author of the study, in a release.
Analyzing the Spatial Architecture of Tumors
To evaluate these patterns, Dr Carr and his research team deployed an AI tool to differentiate cancer and stromal regions on standard pathology slides, quantifying how these zones formed patches, boundaries, and mixed areas, according to the release.
The study evaluated surgical tissue specimens from 203 patients with pancreatic ductal adenocarcinoma who received preoperative therapy but demonstrated only a limited pathologic response. By pairing an AI-driven digital pathology platform with analytical techniques adapted from landscape ecology, the team assessed standard hematoxylin and eosin (H&E) slides for tissue shape, fragmentation, and tumor-stroma intermixing, according to the researchers.
According to the study, two distinct spatial signatures predicted disease-free survival independent of cancer stage, lymph node status, and other established clinical or pathologic risk variables. Patients classified into the high-risk category under one spatial model showed a 71% higher adjusted risk of recurrence, while a second model identified high-risk patients as having more than double the adjusted risk.
Because this methodology utilizes standard H&E slides already prepared during routine clinical care, it could provide actionable prognostic data without requiring supplementary tissue testing, the authors noted.
“What is exciting is that this information is already present in the tissue,” says Carr in a release. “AI-enabled analysis gives us a way to measure features that are difficult to capture by eye and potentially add another layer of precision to how we assess risk after surgery.”
Immune Microenvironment Insights and Future Applications
The study also demonstrated that high-risk spatial configurations correlated with altered immune cell distribution, according to the release. In these high-risk patterns, fewer immune cells penetrated into the tumor core, accumulating instead along the tumor margins. Researchers emphasized that this spatial distribution highlights the critical role of the tumor microenvironment in driving treatment resistance and disease progression.
“Our long-term goal is to better identify which patients remain at greatest risk and ultimately use that knowledge to guide more individualized surveillance, adjuvant therapy, and clinical trial design,” says Carr in a release.
The study authors emphasized that while these findings are promising, prospective clinical studies are necessary to validate the spatial analysis platform before it can be integrated into routine clinical decision-making.
