MSIsensor

Overview

MSIsensor is a computational tool for detecting microsatellite instability (MSI) from paired tumor-normal next-generation sequencing data. It scans the sequenced genome for microsatellite loci (using the msisensor scan sub-command) and computes a per-locus distribution of repeat-unit lengths in tumor vs. normal reads; an overall MSI score (percentage of unstable loci) is computed across all evaluated sites. A threshold of ≥10 is commonly used to classify samples as MSI-H (high instability). MSIsensor is widely deployed in clinical NGS pipelines as the reference standard for MSI calling, including within the MSK-IMPACT clinical sequencing program at Memorial Sloan Kettering Cancer Center.

Used by

  • Used as the reference comparator in the development and validation of mimsi, assessed across a held-out test set (n=317), a prospective validation cohort (n=5,037 with MMR IHC), and a global MSK-IMPACT comparison cohort (n=45,112 samples) PMID:39746944.
  • Used in the MSK-IMPACT clinical pipeline for MSI inference in 2,336 PDAC patients; cut-off ≥10 for MSI-H classification PMID:39753968.
  • Used alongside MiMSI and IDYLLA MSI to assess microsatellite instability in 244 GBC samples; identified 6 tumors (3%) as MSI-High PMID:36228155
  • Applied to WGS data of 28 metastatic NENs to predict MSI status; classified PN4 (homozygous MSH6/MLH1 loss, TMB ~11 mut/Mb) as MSI-low — illustrating that dMMR does not always predict MSI-high in non-colorectal tumors PMID:24326773.
  • MSIsensor used to cross-check signature-based MSI calls in the MSK-IMPACT cohort; identified 102 MSI patients across 11 tumor types, 45% previously untested for MMR deficiency PMID:28481359
  • MSIsensor used to determine MSI status from targeted sequencing data in 295 metastatic EGC patients (MSI-H threshold ≥10); identified 9/295 (3%) MSI-H tumors PMID:29122777
  • MSIsensor score ≥10 used to classify 99 MSI-H/hypermutated (8.7%) and 1,027 MSS (90.6%) cases among 1,134 COADREAD tumors; 98.6% concordance with MMR IHC PMID:29316426
  • MSIsensor used to assess microsatellite instability in SUMMIT basket-trial patients with ERBB2/ERBB3-mutant solid tumors profiled by MSK-IMPACT; results contributed to TMB and co-mutation analysis PMID:29420467

Notes

  • Operates on paired tumor-normal BAM files; evaluates loci generated by msisensor scan v0.2.
  • Threshold ≥10 for MSI-H in the MSK-IMPACT clinical deployment PMID:39753968.
  • Sensitivity 0.67 and auROC 0.907 on a held-out test set enriched for difficult (low-purity, low-coverage) cases — substantially lower than mimsi (sensitivity 0.895, auROC 0.971) PMID:39746944.
  • Indeterminate rate of 3.8% (n=1,724 of 45,112 MSK-IMPACT samples) vs. 0.47% for MiMSI PMID:39746944.
  • Fails to report in 118 of 5,037 prospective cohort cases (2.3%) due to low coverage or purity PMID:39746944.
  • 96% concordance with MiMSI for definitive MSS/MSI-H calls PMID:39746944.

Sources

  • PMID:39746944 — Ziegler et al. benchmarked MSIsensor against MiMSI across multiple cohorts; MSIsensor sensitivity 0.67 vs. MiMSI 0.895 on held-out test set (317 samples), with particular underperformance in low-purity cases (<30% purity: MSIsensor 72.8% vs MiMSI 85.1%, McNemar’s P=8.244×10⁻⁷); indeterminate rate 3.8% across 45,112 MSK-IMPACT samples PMID:39746944.
  • PMID:39753968 — MSIsensor (cutoff ≥10) used for MSI inference in the 2,336-patient MSK PDAC cohort sequenced with MSK-IMPACT; MSI was one of the OncoKB level 1 biomarkers tracked in the actionability landscape analysis PMID:39753968.

This page was processed by crosslinker on 2026-05-06. - PMID:36228155

This page was processed by crosslinker on 2026-05-06. - PMID:24326773

This page was processed by wiki-cli on 2026-05-09. - PMID:28481359

This page was processed by entity-page-writer on 2026-05-15. - PMID:29122777

This page was processed by entity-page-writer on 2026-05-15. - PMID:29316426

This page was processed by wiki-cli on 2026-05-15. - PMID:29420467

This page was processed by wiki-cli on 2026-05-15.