Pathology

Histologic normalization in microscopic colitis links to less steroid use

In a retrospective cohort of 380 microscopic colitis patients with repeat colonoscopy, 164 (43.2%) demonstrated histologic normalization at first re-evaluation. Patients who normalized had substantially lower adjusted odds of corticosteroid use after the repeat procedure (OR 0.26, 95% CI 0.13–0.53). The association remained significant among symptomatic patients (OR 0.22; 95% CI 0.08–0.59) and was weaker in asymptomatic patients. There were no differences in hospitalizations or surgeries between normalized and persistent-histology groups. 1

PathoROB shows foundation models are variably vulnerable to non-biological features

This benchmark evaluated 20 pathology foundation models for susceptibility to non-biological variation using representation and output-level metrics. All models showed robustness deficits, and non-robust representations produced downstream diagnostic errors in patch, slide, retrieval, and clustering tasks. Strategies such as selecting more robust models, improving vision-language alignment, and post-hoc robustification reduced but did not eliminate risk. The authors recommend standardized robustness evaluation prior to clinical FM deployment. 2

Neoadjuvant therapy increases pathology workload for breast excisions

In a single-institution retrospective cohort of 241 breast resections after neoadjuvant therapy (NAT), NAT cases required more tissue blocks (median 20.0 vs 13.0) than controls. NAT specimens more often needed gross resampling (23.7% vs 7.1%) and additional pathologist review (16.2% vs 4.6%). Median pathology report turnaround time was longer for NAT cases (17.0 days vs 15.0 days). Pretreatment clinical stage correlated with macroscopic identifiability of the tumour bed. 3