Immune checkpoint inhibitors (ICKi) have been approved for patients with metastatic colorectal cancer (mCRC) displaying MSI/dMMR (microsatellite instability, defective mismatch repair). We aimed to evaluate the accuracy of standard immunohistochemistry and PCR methods for the detection of MSI/dMMR in mCRC in routine clinical practice.
The study was performed on a multicenter retrospective cohort of mCRCs previously determined as MSI and/or dMMR by local assessment and on a prospective, single center cohort of patients included in ICKi trials based on positive MSI and/or dMMR status previously determined by the originating institutes. We re-assessed dMMR and MSI status in our specialized diagnostic center using immunohistochemistry (antibodies directed against MLH1, MSH2, MSH6 and PMS2), and pentaplex PCR (BAT-25, BAT-26, NR-21, NR-24 and NR-27). The positive predictive value (PPV) of local assessment was the primary objective of the study. Detection rate (i.e. conclusive result) and sensitivity of immunohistochemistry and PCR by central review were evaluated.
Nine false-positives (9.8%) were found in the retrospective cohort (N = 92). These were initially diagnosed as MSI and/or dMMR by the originating institute but were reclassified as MMR proficient /microsatellite stable in our laboratory (PPV=90.2%; 95%CI, 82.2-95.0). The PPV in the prospective cohort (N = 39) was 92.3% (95%CI, 79.0-98.1), with the 3 false-positive patients experiencing progressive disease with ICKi treatment. Amongst the 119 true-positive mCRCs, the detection rate and sensitivity were respectively 100% and 95.8% for immunohistochemistry, while for pentaplex PCR these were 81.5% and 95.9%. Only the combination of immunohistochemistry and pentaplex PCR methods resulted in 100% detection rate and 100% sensitivity.
Local assessment of MSI/dMMR status in mCRC resulted in misdiagnosis of 9.1% of cases as false positive and subsequently incorrect treatment with ICKi. We recommend new guidelines that mandate dual testing of mCRC samples in experienced diagnostic centers using both PCR and immunohistochemistry.
Clinical trial identification
Legal entity responsible for the study
ARCAD Foundation, Institut National du Cancer, Ligue Contre Le Cancer.
T. André: Consultancy: Bristol-Myers Squibb, Fees: Merck. All other authors have declared no conflicts of interest.