AI-Driven Digital Pathology Diagnostic for Colorectal Cancer Biomarker to Optimize Patient Access to Immunotherapy

Colorectal cancer (CRC), with nearly two million new cases and one million deaths globally in 2020, stands as the third most prevalent cancer worldwide and the second leading cause of cancer-related deaths. A critical genomic biomarker for CRC is Microsatellite Instability (MSI), accounting for approximately 15% of all CRC cases. Recent clinical research underscores the prognostic and therapeutic significance of the MSI phenotype, particularly following the approval of immune checkpoint inhibitor (ICI) therapies. Patients with MSI-positive tumors are often good candidates for ICI therapy, while those with microsatellite stable (MSS) tumors typically are not. Consequently, global consensus guidelines now advocate for MSI testing to guide optimal treatment strategies. Pre-screening instruments that can eliminate the necessity for universal patient testing are emerging as a means to streamline this process and alleviate the burden on laboratory personnel and resources.

A team of scientists from from Owkin, Inc. (Paris, France), in collaboration with French pathology labs has performed a blind validation of MSIntuit CRC, a groundbreaking AI-powered digital pathology diagnostic created by Owkin. Designed as a pre-screening tool, MSIntuit CRC seeks to refine the precision in diagnosing and treating CRC. The tool, leveraging patient-derived data, is crafted to offer reciprocal benefits to patients. The study's findings indicate that MSIntuit CRC can successfully exclude nearly half of the MSS patients while accurately identifying over 96% of MSI patients, in line with the performance of established gold standard methods (92-95%). Such innovations are set to transform the screening process, enabling quicker and more extensive patient screening.

The robustness of the study was evidenced by the blind validation conducted on 600 consecutive CRC cases over two years from nine distinct pathology labs, minimizing the selection bias risk. Additionally, the validation maintained consistent performance across two different pathology slide scanners, with sensitivities of 96% and 98% respectively. The study's focus on sensitivity and specificity as indicators of performance and other methodological strategies underscores the AI model's reliability, ensuring the diagnostic's suitability for clinical use and its applicability across various laboratories and regions.

“With the increasing number of biomarkers to be routinely tested in clinical practice, the need for tools that can both ease bottleneck and resource pressures while ramping up biomarker testing is paramount,” said Meriem Sefta, Chief Diagnostic Officer at Owkin. “Our solution represents the first step towards the development of an AI diagnostic that can identify actionable biomarkers from a single H&E slide used in clinical routine, pushing us closer to realizing a precision medicine future.“

Related Links:
Owkin, Inc.