Several groups have reported the genetic-phenotypic correlation with SLs via analysis of many SL patients, but the patients’ germline genetic background is not homogenous, and these reports do not conclusively indicate whether SLs arise from “pure” somatic genetic and/or epigenetic alterations or not. Therefore, this study aimed to analyse and compare the somatic genetic profiles among multiple SLs within a patient with SPS. Our study demonstrates that (a) favourable DNA samples (≥ 4.0 DIN) can be obtained from FFPE tissues ≤ 2 years old to detect appropriate somatic DNA profiles using NGS, (b) pure somatic SL DNA profiles within a SPS patient were compatible with previous SL reports using patients with heterogeneous germline genetic backgrounds, and (c) pure DNA profiles of TA are quite different from that of other SLs within a patient with SPS. To our knowledge, this is the first study to demonstrate a pure somatic genetic profile compared among SLs within a patient.
Although FFPE tissue is generally prepared for histopathological diagnosis, it can also be used for molecular diagnosis, such as genomic analysis using NGS, as it can be clearly observed from the histopathological picture that the lesion is a tumour within the FFPE sample. This applies to our study in that SLs are histopathologically varied, and it is important to determine the location of SL. Therefore, we used FFPE samples, detected the SL location, and extracted DNA only at this location. Further, it is important to analyse NGS with high-quality DNA samples, and evaluate the quality of pre-analytical and analytical processes, especially when using DNA from FFPE due to the tendency of the sample to degrade. In our study, we evaluated the DNA derived from FFPE by DIN and found that DNA was degraded in FFPE tissues stored for more than two years, which is consistent with recent reports by other groups showing that the high degradation of DNA extracted from FFPE is unsuitable for sequence analysis. However, NGS analysis in our study was successful even with aged FFPE samples, probably because the amplicon for analysis was designed to be as small as 200 bp, even if the nucleic acid was degraded. [25–27] Thus, it is important to consider the best sample from which to obtain DNA to match the progress of genomic analysis techniques, such as long-read sequencing. [28]
Many groups have reported the influence of pathological genetic variations of BRAF, such as c.1799T > A and p.Val600Glu, on the progression of HPs, SSLs, TSAs, and KRAS pathogenic variants for HPs and TSAs, but these analyses were performed among patients with heterogeneous germline backgrounds. [10–13] To detect a pure somatic genetic variation profile, we compared the genetic profiles of dome-serrated lesions within identical patients. In Case #1, a known pathogenic variant of BRAF (c.1799T > A, p.Val600Glu) was detected in one SSL (#1–4) and one TSA (#1–5). Previous reports have demonstrated that the BRAF variant was found in almost all SSLs, where there are two distinct pathways for progression to TSAs, that is, the KRAS-driven pathway and the BRAF-driven pathway, and it is compatible that we detected the BRAF variant commonly in two different SLs in patient #1. [12 13 20 21] When we focus on the two lesions, it is interesting that genetic profiles, other than that of the BRAF variant, appear quite different (BLM, AXIN2, CDC27, and MLH1 in #1–4, and RET, ERBB2, STK11, and TCERG1 in #1–5, as seen in Table 4). Therefore, the two SLs must be initiated by the common BRAF pathogenic variant, followed by progression via the accumulation of different genetic profiles, but further accumulated findings should be considered. In Case #2, all SSLs displayed pathogenic variants of BRAF (c.1799T > A, p.Val600Glu), as expected from previous reports. [14] In addition, it is interesting that we confirmed the adenoma-carcinoma pathway by detecting somatic APC deletion (c.4249_4265delATTATAAGCCCCAGTGA) as a driver variant (VAF: 55.4%) in TA (#2–2). Notably, the somatic genetic profile of the TA is quite different from other SSLs within Case #2 (#2 − 1, #2–3, #2–4), which indicates that the serrated pathway in the SSL adenoma-carcinoma sequence in TAs does not have common driver variants at the initiation stage, and that the accumulated genetic variant profile is distinct between the two pathways.
RNF43 has been reported as one of the key genes when pathogenic germline or somatic variants are detected in SLs. [23 29 30] Giannakis et al. demonstrated that somatic mutations in RNF43 occur in 18.9–17.6% of CRC cases, and the majority of RNF43 somatic mutations were truncating events. Despite this, it is still unclear from their reports whether CRC arises from SLs. [29] Taken together, it is possible that the somatic RNF43 splice-site variant detected in our study in SSLs of Case #1 (#1–3) is pathogenic in the serrated polyposis-cancer sequence, although additional questions remain as limitations, such as the existence of two hits for the lesion by genetic or epigenetic alteration.
As for epigenetic features in SLs, it has been reported that silencing of MLH1 plays an important role in the progression of SLs, especially with the BRAF pathogenic variant, [4 16] but in our IHC study, no deficiency of MLH1 protein could be seen among SLs in two patients with SPS. Apparently, this result does not agree with a previous report, but it is not clear whether MLH1 was silenced to completely suppress the expression of MLH1 protein. Moreover, it must be noted that previous clinical reports have demonstrated that deficient-MMR has not been identified in HPs, TSAs, or SSLs, but has been reported in SSL with dysplasia (SSLD) only. [21 31] Additionally, SSLD is the only pre-cancerous colorectal lesion in which MLH1 is methylated. [32] Regarding the occurrence of deficient MMR, patients with pre-cancerous lesions, especially with SSLD, require careful surveillance after resection.
Among the SLs and non-SLs (TA), we examined SBS type and checked which substitutional types would be correspond to the MS pattern previously reported [33]. Interestingly, we found that HP, TSA, and TA have similar substitutional pattern in that these three lesions have mainly observed C > T characterized as aging, and SBS of SSL is distinct from these three lesions. To confirm our results from multigene panel testing, more huge number of SBS should be analyzed using whole exome sequencing or whole genome sequencing.
The present study has several limitations: i) IHC for DNA MMR was performed only for pre-cancerous lesions and not cancerous lesions, because cancer was not found in the two analysed SPS patients; ii) the analysed number of patients with SPS was small; and iii) the methylation profile was not evaluated. These findings require further investigation in future studies. However, to the best of our knowledge, the present study is the first to focus on pure somatic genetic variant profiles of multiple SLs compared within a SPS patient, and it may represent the first step in understanding the mechanism behind the differentiation into different histological types of serrated lesions.
In conclusion, we compared pure somatic genetic profiles among SLs within patients with SPS as a homogenous germline background using NGS. A known common pathogenic BRAF variant could be seen in SSLs and TSAs, but another genetic profile of each lesion appeared quite different. In addition, the genetic profile of TA is consistent with the profile of the adenoma-carcinoma sequence pathway and distinct from that of other SLs within the same patient with SPS. More research is required to understand the molecular mechanisms underlying the progression to SL for patient care and colorectal cancer prevention.