Can a gastrointestinal pathologist identify microsatellite instability in colorectal cancer with reproducibility and a high degree of specificity?

Abstract

Clinical features usually initiate evaluation for Lynch Syndrome (LS) but some colorectal cancer (CRC) histopathology findings are compatible with high microsatellite instability (MSI-H) that also occurs in LS. This led to the suggestion that pathologists request MSI analysis, which is an expensive addition to routine histology. We aimed to see if a Gastrointestinal Pathologist could identify MSI-H features with reproducibility and high (95%) specificity (MSI-H 95%). Histopathology of all CRCs received during 2005 and 4 MSI-H controls were scored using 2 published methods, "MsScore" and "PathScore". MSI analysis was performed on CRCs scored by either method as probable MSI-H 95% and results compared. To examine reproducibility of histopathology, 100 coded slides, including 25 scored MSI-H 95% and 75 scored low, were re-examined to now identify those needing MSI analysis. Costs were evaluated for identifying MSI-H with or without scoring. All 227 CRCs were scored for possible MSI-H 95%; 24 had high scores and MSI analysis. DNA analysis proved 14 MSI-H, PathScore identified 13 (95%), MsPath identified 9 (64%), histopathology alone identified 7 (50%). Reproducibility for identifying histopathology characteristics of MSI-H at re-examination, without scoring, was "moderate agreement" (Kappa statistic = 0.4615). Costs for identifying MSI-H by PathScore were the lowest, $436/identification. Conclusions; PathScore identified the most proven MSI-H CRCs at lowest cost and even an experienced gastrointestinal pathologist has difficulties identify MSI-H without scoring. So, scoring can be facilitated by a computerized evaluation form for routine CRC histology, prompting score computation and recommendation for MSI analysis with high specificity.

Artificial Intelligence for Histology-Based Detection of Microsatellite Instability and Prediction of Response to Immunotherapy in Colorectal Cancer

“…Over half of the tumors analyzed had less than 5% chance of harboring MSI, presenting the potential for significant cost savings [26]. In another cohort, the model by Greenson et al detected 93% of tumors with MSI and outperformed MsPath [40].…”

Clinical actionability of triaging DNA mismatch repair deficient colorectal cancer from biopsy samples using deep learning

“… 12 However, pathologists still find it challenging to accurately identify MMR status solely based on visual inspections of tissue morphology. 13 Recent advances in artificial intelligent image analysis techniques, especially deep learning approaches, have shown promising performance in various histopathological analytic tasks, including diagnosis, prognosis estimation, and gene mutation prediction. 14 , 15 , 16 , 17 There are growing evidences supporting the possible use of deep learning (DL) for H&E stained image-based MMR status detection in CRC, with an area-under-ROC curves (AUROC) between 0·77 and 0·96.…”