27P - Analytical validation of a small amplicon NGS panel for MSI detection

Background

With the development of immune checkpoint inhibitors (ICI), microsatellite instability (MSI) has become an important biomarker. While large multigene hybrid capture next-generation sequencing (NGS) panels are highly accurate in detecting MSI, the sensitivity of small amplicon-based panels is unknown.

Methods

Next-generation sequencing (NGS) was performed on FFPE samples from patients with any stage colorectal cancer (CRC) using the Solo Atlas Pro amplicon panel covering 38 genes and 39 mononucleotide short tandem repeats for MSI analysis. 5-loci polymerase chain reaction (PCR)-based MSI analysis was used as a reference method. MSI by NGS was estimated based on the distribution of k-mers. Statistical analysis was performed using Cohen’s kappa (к), Mann Whitney and Fisher’s exact test.

Results

A total of 160 samples were analyzed using NGS and PCR. Median coverage was 2352x (range, 238-28447х), MAPD – 0.368 (range, 0.236-0.731). Based on PCR results, 42 (46.25%) samples were MSI, 118 (73.75%) – MSS. Based on NGS results, 40 (25%) samples were MSI, 120 (75%) were MSS. NGS and PCR were concordant in 98.75% (158/160) of samples (к=0.97). Sensitivity of NGS was 95.24% (95% CI, 83.84%-99.42%), specificity – 100.00% (96.92%-100.00%), accuracy – 98.75% (95.56%-99.85%). All NGS false positives cases had suboptimal FFPE quality, suggested by absence of any somatic mutations identified. MSI samples had a significantly higher amount of somatic mutations than MSS samples (median, 3 vs 1, p<0.0001). In silico mixing of the sequencing data of 11 MSI samples demonstrated that the minimal sample fraction with the detectable MSI was 0.5%; at 2.5% MSI was detectable in all samples.

Conclusions

The Atlas Pro amplicon panel demonstrates high concordance with PCR and is highly sensitive in detecting MSI in CRC samples.

Editorial acknowledgement

Clinical trial identification

Legal entity responsible for the study

The authors.

Funding

Russian Science Foundation (Grant №22-75-10154).

Disclosure

All authors have declared no conflicts of interest.