Artificial intelligence for detection of microsatellite instability in colorectal cancer-a multicentric analysis of a pre-screening tool for clinical application.

BACKGROUND: Microsatellite instability (MSI)/mismatch repair deficiency (dMMR) is a key genetic feature which should be tested in every patient with colorectal cancer (CRC) according to medical guidelines. Artificial intelligence (AI) methods can detect MSI/dMMR directly in routine pathology slides, but the test performance has not been systematically investigated with predefined test thresholds. METHOD: We trained and validated AI-based MSI/dMMR detectors and evaluated predefined performance metrics using nine patient cohorts of 8343 patients across different countries and ethnicities. RESULTS: Classifiers achieved clinical-grade performance, yielding an area under the receiver operating curve (AUROC) of up to 0.96 without using any manual annotations. Subsequently, we show that the AI system can be applied as a rule-out test: by using cohort-specific thresholds, on average 52.73% of tumors in each surgical cohort [total number of MSI/dMMR = 1020, microsatellite stable (MSS)/ proficient mismatch repair (pMMR) = 7323 patients] could be identified as MSS/pMMR with a fixed sensitivity at 95%. In an additional cohort of N = 1530 (MSI/dMMR = 211, MSS/pMMR = 1319) endoscopy biopsy samples, the system achieved an AUROC of 0.89, and the cohort-specific threshold ruled out 44.12% of tumors with a fixed sensitivity at 95%. As a more robust alternative to cohort-specific thresholds, we showed that with a fixed threshold of 0.25 for all the cohorts, we can rule-out 25.51% in surgical specimens and 6.10% in biopsies. INTERPRETATION: When applied in a clinical setting, this means that the AI system can rule out MSI/dMMR in a quarter (with global thresholds) or half of all CRC patients (with local fine-tuning), thereby reducing cost and turnaround time for molecular profiling.

What can the molecular pathologist offer for optimal decision making?

As a consequence of new innovative therapies and therapeutic combinations, the treatment of advanced colorectal cancer is becoming increasingly complex. Validated molecular biomarkers could contribute to patient management, but until recently, none has been routinely used. With the recognition that activating mutations of the KRAS oncogene can predict resistance to anti-epidermal growth factor receptor agents, the clinical value of biomarkers in advanced colorectal cancer has been brought to the fore. Prognostic and predictive biomarkers that reflect the molecular and therapeutic complexities of advanced colorectal cancer may provide valuable information regarding likely clinical outcome and therapeutic response and thus may improve patient management and therapeutic agent selection. Such biomarkers are discussed herein.

Molecular pathology--the future?

The diagnosis and management of complex disease requires multidisciplinary clinical collaboration. Histopathological assessment of tissues is an important part of this approach and relies on the morphological appraisal of stained tissues by light microscopy. Although considered the 'gold standard', morphology alone is becoming increasingly insufficient in disease where molecular characterisation has altered clinical practice. As the ability to define the molecular alterations associated with disease expands, the requirement to analyse such alterations for diagnostic and therapeutic purposes follows suit. Despite the widespread use of small scale molecular approaches such as immunohistochemistry and fluorescent in situ hybridisation, the demand for detailed molecular classification of disease states increases unabated. In this review, we summarise the clinical applications of various molecular techniques already used within the modern histopathology department. Thereafter, we consider novel high throughput, array based technologies and their possible impact on future histopathological practice.