Prediction of response to immune checkpoint blockade in patients with metastatic colorectal cancer with microsatellite instability

•ChallengES tumor mutation burden (TMB) and MSIsensor for predicting response to immunotherapy in MSI mCRC.•Demonstration that MSI misdiagnoses lead to a false ability of TMB and MSIsensor to predict ICI resistance in MSI mCRC.•Identification and validation of DNA and RNA signatures predictive of response to ICI in these patients with mCRC MSI.•Evidence for a fibrosis signature associated with functional ICI resistance in MSI mCRC. BackgroundMismatch repair-deficient (dMMR) tumors displaying microsatellite instability (MSI) represent a paradigm for the success of immune checkpoint inhibitor (ICI)-based immunotherapy, particularly in patients with metastatic colorectal cancer (mCRC). However, a proportion of patients with dMMR/MSI mCRC exhibit resistance to ICI. Identification of tools predicting MSI mCRC patient response to ICI is required for the design of future strategies further improving this therapy.Patients and methodsWe combined high-throughput DNA and RNA sequencing of tumors from 116 patients with MSI mCRC treated with anti-programmed cell death protein 1 ± anti-cytotoxic T-lymphocyte-associated protein 4 of the NIPICOL phase II trial (C1, NCT03350126, discovery set) and the ImmunoMSI prospective cohort (C2, validation set). The DNA/RNA predictors whose status was significantly associated with ICI status of response in C1 were subsequently validated in C2. Primary endpoint was progression-free survival by immune RECIST (iRECIST) (iPFS).ResultsAnalyses showed no impact of previously suggested DNA/RNA indicators of resistance to ICI, e.g. MSIsensor score, tumor mutational burden, or specific cellular and molecular tumoral contingents. By contrast, iPFS under ICI was shown in C1 and C2 to depend both on a multiplex MSI signature involving the mutations of 19 microsatellites hazard ratio cohort C2 (HRC2) = 3.63; 95% confidence interval (CI) 1.65-7.99; P = 1.4 × 10–3] and the expression of a set of 182 RNA markers with a non-epithelial transforming growth factor beta (TGFB)-related desmoplastic orientation (HRC2 = 1.75; 95% CI 1.03-2.98; P = 0.035). Both DNA and RNA signatures were independently predictive of iPFS.ConclusionsiPFS in patients with MSI mCRC can be predicted by simply analyzing the mutational status of DNA microsatellite-containing genes in epithelial tumor cells together with non-epithelial TGFB-related desmoplastic RNA markers. Mismatch repair-deficient (dMMR) tumors displaying microsatellite instability (MSI) represent a paradigm for the success of immune checkpoint inhibitor (ICI)-based immunotherapy, particularly in patients with metastatic colorectal cancer (mCRC). However, a proportion of patients with dMMR/MSI mCRC exhibit resistance to ICI. Identification of tools predicting MSI mCRC patient response to ICI is required for the design of future strategies further improving this therapy. We combined high-throughput DNA and RNA sequencing of tumors from 116 patients with MSI mCRC treated with anti-programmed cell death protein 1 ± anti-cytotoxic T-lymphocyte-associated protein 4 of the NIPICOL phase II trial (C1, NCT03350126, discovery set) and the ImmunoMSI prospective cohort (C2, validation set). The DNA/RNA predictors whose status was significantly associated with ICI status of response in C1 were subsequently validated in C2. Primary endpoint was progression-free survival by immune RECIST (iRECIST) (iPFS). Analyses showed no impact of previously suggested DNA/RNA indicators of resistance to ICI, e.g. MSIsensor score, tumor mutational burden, or specific cellular and molecular tumoral contingents. By contrast, iPFS under ICI was shown in C1 and C2 to depend both on a multiplex MSI signature involving the mutations of 19 microsatellites hazard ratio cohort C2 (HRC2) = 3.63; 95% confidence interval (CI) 1.65-7.99; P = 1.4 × 10–3] and the expression of a set of 182 RNA markers with a non-epithelial transforming growth factor beta (TGFB)-related desmoplastic orientation (HRC2 = 1.75; 95% CI 1.03-2.98; P = 0.035). Both DNA and RNA signatures were independently predictive of iPFS. iPFS in patients with MSI mCRC can be predicted by simply analyzing the mutational status of DNA microsatellite-containing genes in epithelial tumor cells together with non-epithelial TGFB-related desmoplastic RNA markers.