Microsatellite instability (MSI) is the condition of genetic hypermutability (predisposition to mutation) that results from impaired DNA mismatch repair (MMR) in tumors. MSI is an effective marker for Lynch syndrome diagnosis and prognosis prediction in the treatment of certain cancer types. MSI is also an approved biomarker for predicting the efficacy of anti-PD-L1/PD-1 immunotherapeutic agents including Keytruda and Opdivo in solid tumors.
- Using a training set of paired tumor/normal samples with known MSI status (measured by PCR), we identified 19 out of 43 mononucleotide microsatellite loci covered by NovoPM whose repeat lengths can reliably reflect the MSI status of the tumor samples. These 19 loci are defined as the “MSI marker set”.
- For each tumor sample to be analyzed, the histogram of read counts for different repeat lengths of each locus in the MSI marker set are compared to that of the paired normal sample using the Mann-Whitney U test. A locus is considered unstable if the P-value is less than 0.05.
- The percentage of unstable loci in the MSI marker set is calculated as the NovoPM MSI score. The following thresholds are then established with another set of training samples:
- MSI score > 0.23: The tumor sample is MSI.
- 0.23 ≥ MSI score ≥ 0.17: MSI status of the tumor sample cannot be determined with this method. The use of a conventional method (e.g., PCR) is advised.
- MSI score < 0.17: The tumor sample is MSS.
NovoPM-MSI Validation with Novogene In-House Data
To analytically validate our NovoPM-MSI algorithm, we analyzed 113 colorectal cancer FFPE samples with both PCR (the current gold standard in clinical testing for MSI status) and NovoPMTM. The results are shown in Figure 1 and Table 1. Our NovoPM-MSI algorithm (“PPC” in Figure 1) has a sensitivity of 88.9% and a specificity of 96.2% with PCR. In addition, Table 2 shows that our NovoPM-MSI results have a strong correlation with the samples’ TMB status, which is consistent with the underlying biology of MSI and TMB.
Figure 1. NovoPM-MSI analytical validation with 113 colorectal cancer FFPE samples. X-axis shows the sample ID and Y-axis shows the MSI scores from NovoPM-MSI algorithm (PPC, green dot-line). Nine samples were tested to be MSI via PCR (“PCR_MSI Positive”). Eight of these nine samples were tested positive via NovoPM-MSI with the threshold of 0.2 (green dotted line). The sensitivity and specificity results are shown in Table 1. According to these data, we recommend that samples with a NovoPM-MSI score of 0.2±0.03 should be verified via PCR. The purple line shows MSI scores calculated from another algorithm (“DIF”) that did not produce acceptable sensitivity or specificity. This serves as an example of many other algorithms that were tested and then abandoned during the development of our final NovoPM-MSI algorithm.
Table 1. Concordance between PCR and NovoPM-MSI algorithm for 113 colorectal cancer FFPE samples.
Table 2. Correlation between MSI status and TMB status for 81 samples. TMB status (“low” or “high”) was determined with cut-offs calculated according to Zehir et al., Nat Med, 2017.