While a few studies have analyzed the characteristics of tumors with a high TMB in PC 18, there are still no studies focusing on PC tumors with a low TMB. Therefore, we focused on tumors with a particularly low TMB among PC cases, defined as TMB-ultra-low PC and attempted to characterize these samples using WES and GEP. In the present study, the median TMB in PC tumors was found to be 0.24, which is very low compared to other cancers; there were no TMB-high (TMB ≥ 5.0) tumors, 20% were classified as TMB-low (< 5.0, ≥ 1.0), and 80% were classified as TMB-ultra-low (< 1.0). The TMB-ultra-low PC had a significantly better prognosis than TBM-low PC. A multivariate analysis identified TMB-ultra-low PC as an independent favorable prognostic factor.
Previous studies investigating TMB in resected PC 18 reported that a TMB ≥ 10 mutations/Mb accounted for 7% of cases, 10 > TMB > 5 mutations/Mb accounted for 18%, and TMB ≤ 5 mutations/Mb accounted for 75%. This reported TMB is low compared to other cancers described thus far but is slightly higher than our present results, being generally consistent. In addition, those authors also revealed that the prognosis of patients could be stratified by the TMB value. However, their finding that tumors with a high TMB have a better prognosis and tumors with low TMB have a poorer prognosis than others differs from the trend in the present study, where TMB-ultra-low PC had a better prognosis than TMB-low PC. This discrepancy in the survival may be due to the cut-off value of the TMB. Previous studies placed cases with a TMB < 5, which accounts for a large portion of their PC cohort (75%), into a single category and did not focus on ultra-low-TMB cases (< 1.0) within that category or explore its characteristics. In the present study, we focused on ultra-low-TMB PC cases among PC, as this subset is representative of cancers with a particularly low TMB, and characterized the details for the first time.
Clinicopathologically, the TMB-ultra-low PC contained a higher rate of R-PC and lower rate of BR-PC. Furthermore, TMB-ultra-low PC included fewer cases of adenosquamous carcinoma, which is reported to have a poorer prognosis than adenocarcinoma 32. These clinicopathologic factors may have contributed to the particularly favorable prognosis of TMB-ultra-low PC. The mutational analysis showed that TMB-ultra-low PC had a significantly lower frequency of driver mutations and a significantly lower number of driver mutations than TMB-low PC. Since driver mutations have been reported to be a poor prognostic factor in PC 33,34, the possibility that differences in the frequency of driver mutations may be confounded by the results of the prognostic analysis should be considered. However, as shown in Table 3, in the present study, individual driver mutations were not significantly identified as prognostic factors in the prognostic factor analysis, although TMB was identified as an independent prognostic factor. We believe that this result supports TMB as a prognostic factor in resected PC and the favorable prognosis for TMB-ultra-low PC in contrast to the general recognition that low-TMB PC has a poor prognosis.
In our previous pan-cancer study 35, the association of TMB status and immune parameters, such as the expression of specific immune response genes, specific immune cell populations, and TCR repertoire profile were investigated. The analysis of immune cell populations in tumors demonstrated that the frequencies of exhausted CD8 + T-cells, activated effector CD8 + T-cells, and natural killer cells were significantly lower in TMB-ultra-low tumors. The T-cell receptor repertoire numbers and diversity evenness score (DE50) were higher in the TMB-ultra-low tumors. One possible mechanism underlying the relatively good prognosis of the TMB-ultra-low PC might be that the TMB-ultra-low PC features a balance between immunosuppression and immunostimulation. We plan to further investigate the precise mechanism responsible for these observations in the future.
Deficiency of TP53 activation has been recognized to be associated with enhanced CIN 27,36. Our gene expression signature analysis showed that TMB-ultra-low PC was associated with decreased TP53 inactivation and decreased CIN compared to TBM-low PC. Therefore, the TMB-ultra-low tumors may be less likely to accumulate somatic mutations and CNVs because TP53 activation is maintained. This means that TMB-ultra-low PC may also include tumors that have developed through other tumorigenic processes besides TP53 inactivation. A recent pan-cancer analysis revealed that chromothripsis(-like) events often occur in tumors with few driver mutations 37. Conversely, recent mutational analyses have shown that composite mutations led to tumorigenesis 38,39. Although the involvement of these mechanisms is speculated, to further understand the details of TMB-ultra-low PC, the integration of our analysis and whole-genome sequencing findings, including those of our germline mutation analysis, should be explored in the future.
Thus far, immunotherapy using ICIs has not been successful implemented in PC, mostly due to the immunosuppressive tumor microenvironment and the relatively low expression and/or low quality of tumor-specific neoantigens 40,41. The prediction of responses to ICIs using gene expression focuses on tumors harboring some mutations (≥ 10 mutations/Mb) 29–31, although most solid tumors were categorized as TMB-low. Our analysis revealed that there was no notable correlation between the TMB and T cell-inflamed gene expression profiling signatures, indicating treatment responsiveness to ICIs (P = 0.407). Although a high expression of T-cell inflammatory signatures in a population might be expected to indicate candidate ICI responsiveness despite a low number of mutations, the existing evidence concerning the correlation between the TMB and neoantigen expression and the efficacy of ICIs in PC is rather weak 42,43. Immunotherapy for PC with ICIs and their biomarkers should be explored further in the future.
PC is characterized by a marked desmoplastic reaction with dense fibrous stroma and low neoplastic cellularity 44. This low neoplastic cellularity can confound the analysis of the mutational and gene expression characteristics in actual tumor cells. Our previous pan-cancer study excluded samples with low tumor cellularity (< 0.3) from the mutation analysis. In the present PC study, the median estimated tumor content was 0.25. Adopting the threshold of 0.3 would exclude a large portion of samples from the analysis. This low tumor content is one concern as a limitation. A study involving low-cellularity PC samples highlighted the importance of deep sequencing of low-purity samples 25. We therefore compared the WES-calculated TMB with the TMB calculated from deep sequencing results using CCP to determine which was more appropriate for exploring the characteristics of TMB-ultra-low PC. The results indicated that the effect of a low tumor content on the TMB determined using WES was limited in PC. Notably, in the prognostic analysis, the TMB was identified as an independent prognostic factor, even though the tumor content was not identified as a prognostic factor even in univariate analysis. Based on these results, it is reasonable to assume that TMB-ultra-low PC has a better prognosis than TMB-low PC, although the tumor content is somewhat confounded by the TMB. Nevertheless, further investigation of PC-specific testing methods and appropriate cutoff values may be warranted.
In conclusion, the present study characterized PC tumors with a particularly low TMB, defined as TMB-ultra-low PC in the JCGA dataset. Detection rates of driver mutations and CNV were decreased in TMB-ultra-low PC. TMB-ultra-low PC was associated with decreased TP53 inactivation and decreased CIN. Furthermore, among tumors with low TMB, which have been considered a poor prognosis group based on the results of previous studies, we identified TMB-ultra-low PC as a subgroup of tumors with a relatively good prognosis. Our analysis focusing on TMB-ultra-low PC can provide insight into PC with low numbers of somatic alterations.