NovoPM™ Tumor Mutation Burden (TMB) Analysis Overview
Tumor mutation burden (TMB) represents the total number of mutations per coding area of a tumor genome calculated through the genomic sequencing of tumor tissue samples. The value of TMB has been found to correlate with the efficacy of anti-PD-L1/PD-1 immunotherapies in some tumor types.
- TMB is calculated based on the coding DNA sequence (CDS) regions included in the NovoPM panel (approximately 1.5Mb).
- Somatic mutations are identified by VarScan2.
- The following mutations are excluded from the calculation of TMB:
- Germline mutations
- Low frequency mutations (threshold=2%)
- Known driver mutations (EGFR, MET, BRAF, PIK3CA, NF1, KRAS, and NOTCH family)
- Synonymous mutations
- Repeat regions
NovoPM-TMB Validation with Public Data
The NovoPM TMB algorithm was first validated with a set of public whole exome sequencing (WES) data generated from 34 NSCLC samples (Rizvi et al., Science, 2015). TMB was calculated from (a) the entire set of WES data using our bioinformatics pipeline of TMB, or (b) the genes included in NovoPM using the bioinformatics pipeline of NovoPM TMB. The results of these two methods showed strong linear correlation (Figure 1; R2=0.9175). Similar analyses were then conducted with the lung adenocarcinoma (LUAD) WES dataset and lung squamous cell carcinoma (LUSC) WES dataset from The Cancer Genome Atlas (TCGA). Both showed highly linear correlations between mutation counts calculated from whole exome sequencing and from NovoPM (Figure 2 and Figure 3; R2=0.91 and 0.83, respectively).
Figure 1. Correlation of mutation counts between WES and NovoPM using public whole exome data (Shan et al., WCLC 2017 – poster presentation by Novogene)
Figure 2. Correlation of mutation counts between WES and NovoPM using lung adenocarcinoma dataset from TCGA (Shan et al., WCLC 2017 – poster presentation by Novogene)
Figure 3. Correlation of mutation counts between WES and NovoPM using lung squamous cell carcinoma dataset from TCGA (Shan et al., WCLC 2017 – poster presentation by Novogene)
NovoPM-TMB Validation with Novogene In-House Data
In addition to the in silico validation with public data as described above, we also validated the NovoPM TMB algorithm with in-house WES data from 115 lung cancer samples and the results showed strong linear correlation (Figure 4; R2=0.944).
To validate both the wet-lab procedure and the bioinformatics pipeline of NovoPM TMB, we carried out parallel WES and NovoPM panel sequencing of 15 lung cancer and colorectal cancer samples. The sequencing data were then analyzed in a similar fashion and the results showed strong linear correlation (Figure 5; R2=0.882).
To test the reproducibility of the NovoPM TMB assay, triplicates of 5 in-house lung cancer FFPE samples were analyzed with the NovoPM wet-lab procedure and bioinformatics pipeline. The results demonstrated that the assay is highly reproducible (Figure 6).
We also found that our NovoPM TMB results showed a statically significant correlation with MMR status or POLE mutation status in 262 Chinese lung cancer samples (Table 1), which further proves the validity of this assay.
Figure 4. Correlation of TMB between WES and NovoPM using lung cancer WES samples sequenced in Novogene
Figure 5. Correlation of TMB using lung cancer and colorectal cancer samples which were sequenced by both WES and NovoPM at Novogene
Figure 6. Five FFPE samples TMB precision experiments show good reproducibility (Shan et al., WCLC 2017 – poster presentation by Novogene)
*Cut-offs calculated according to Zehir et al., Nat Med, 2017
Table 1. Correlation between TMB and MMR status or POLE mutation status of 262 Chinese lung cancer samples