Colorectal carcinoma (CRC) develops through a process termed the adenoma-carcinoma sequence, which refers to a stepwise process of mutational activation of oncogenes (Vogelstein et al., 1998; Kinzler and Vogelstein, 1996). Adenomatous polyposis coli (APC) mutations are most frequently observed in the early stages of colorectal carcinogenesis. Additional genetic mutations, including CTNNB1, FBXW7, KRAS, BRAF, PIC3CA, p53, and SMAD4, accumulate during CRC development (Jones et al., 2008). The Cancer Genome Atlas (TCGA, 2012) has increased our understanding of the pathogenesis of CRC, in which certain driver mutations are identified under malignant conditions.

Familial adenomatous polyposis (FAP) is an inherited disease characterized by the development of more than 100 adenomatous polyps in the large intestine. Germline variants in APC are inherited from parents or sometimes occur spontaneously. APC, which is located on chromosome 5q, encodes 2843 amino acids. APC is a component of the Wnt signaling pathway and plays a role in intercellular adhesion (Miyaki et al., 2008). Germline pathogenic variants are distributed widely from exons 1 to 15, and genotype-phenotype correlations are observed depending on the location of the variants (Miyoshi et al., 1992). The typical form of FAP with profuse polyposis is associated with variations in the mutation cluster region (MCR) between codons 1,286 and 1,513 (Kohler et al., 2008). Somatic APC mutations in colorectal cancers are typically nonsense variants and insertion and/or deletion mutations that result in amino acid truncation. APC mutations within the MCR produce truncated APC proteins that lack all axin-binding sites and all but 1 or 2 of the 20 amino acid repeats (20AARs) (Kohler et al., 2008). Patients with germline mutations after codon 1,399, which retain 2 to three intact 20AARs, acquire truncating mutations in the MCR. Conversely, patients with germline mutations to the MCR tend to have somatic mutations in the MCR and retain 1 to 2 20AARs (Albuquerque et al., 2002). This suggests that second hits in APC are selected to produce a ‘just-right’ level of β-catenin signaling optimal for colorectal tumor development. The combined hits resulting in partial loss of β-catenin regulation are suggested as a loose fit model (Crabtree et al., 2003). Studies of adenomas from patients with attenuated FAP have further revealed third hits in APC targeting the germline mutant allele to achieve an optimal genotype. Since these extensive studies concerning the two-hit hypothesis for colorectal tumorigenesis were conducted in the late 20th century, Sanger sequencing was used to identify germline variants in patients with polyposis. Next-generation sequencing (NGS) has been conducted for large scale regional analyses (Takao et al., 2021; Torrezan et al., 2013). However, few studies have investigated the molecular alterations in early precancerous colorectal tumors of FAP patients, compared with sporadic tumors using NGS.

We explored molecular alterations using target sequencing in the early phase of adenoma-carcinoma sequences and analyzed the APC gene alterations in these early lesions by comparing colorectal neoplasms in FAP patients with sporadic colonic neoplasms.

We investigated early genetic events associated with colorectal carcinogenesis using NGS. Our genetic analysis indicated a high frequency of APC somatic mutations in adenomas of both FAP patients and sporadic cases. APC mutations were observed in 75% of the colorectal adenomas, and the frequency of the mutation was sustained through the adenoma-carcinoma sequences (Figure 4). Thus, APC mutations are most likely associated with the initiation of colorectal adenoma. Peterson et al. has developed a mathematical model of colorectal cancer initiation and demonstrated that the order of the driver genes was APC loss as the first event followed by KRAS gain in large majority of cases (Paterson et al., 2020). The frequency of mutation of colorectal cancer was compared to the genetic alterations deposited in the database (Supplementary Figure S4). APC mutations are the most frequently observed mutations, and KRAS mutations are considered major cancer-related genetic mutations. Early carcinogenesis in our FAP patients was similar to that in sporadic cases when these genetic alterations were compared in our study. The late phase of colorectal carcinogenesis may be regulated by MAPK signaling, such as KRAS and BRAF, but not by WNT signaling. Another carcinogenic process (e.g., for mucinous carcinoma) is regulated by a p53 mutation without an APC mutation. We further showed that KRAS mutations were commonly identified in HG adenoma and early carcinoma of FAP patients. A previous study on FAP tumors suggested that KRAS mutations were not necessary in APC-associated FAP adenomas, as codon 12 mutations are rarely (10%) observed (Obrador-Hevia et al., 2010; Bettington et al., 2013). A later study detected KRAS mutations in 23% of FAP tumor samples and somatic APC mutations in 69% of the patients (Li et al., 2020). These discrepancies may have depended on the dysplasia grade in these studies, as the previous study included many small adenomas. Among our FAP patients, the frequency of KRAS mutations in LG adenomas was 24%, which corresponds to the rate reported in the previous study. The comparison between FAP-associated and sporadic neoplasia clearly demonstrated that the MAPK pathway is associated with carcinogenesis in some FAP patients. Colorectal tumors with KRAS mutations were frequent in some FAP patient, and the positive rates differed among the patients. Subsequently, we propose intra-patient homogeneity and inter-patient heterogeneity in genetic mutations of colorectal tumors in FAP.

There is a clinical question of whether APC mutations contribute to the progression of advanced CRC. The frequency of APC mutations in carcinomas, which has been investigated in a large number of studies in TCGA, is 51% in hypermutated tumors and 81% in non-hypermutated tumors. Truncated APC protein is thus expected to be critical for the initiation of benign colorectal adenomas, but not for the promotion of malignant lesions. APC mutations were recently reported to be associated with a good prognosis (Mondeca et al., 2020), whereas KRAS and BRAF mutations were reported to be associated with later carcinogenesis and poor prognosis in patients with CRC (Walther et al., 2009; Pai et al., 2012). In our FAP cohort, the frequency of KRAS mutation in patients with HG adenomas or carcinomas was higher than that in patients with sporadic tumors. The disease severity of colorectal tumors in FAP and sporadic tumors may vary depending on the metabolic process or proliferative activity (Li et al., 2020). Carcinogenicity differs among patients because KRAS mutations are frequently observed in polyps of some patients and not in others. Remarkably, polyps in the FAP patients with whole APC deletion (FAP13 and FAP14) frequently possessed KRAS mutations (5/6 and 4/5, in Figure 3). The reason for the intra-patient homogeneity in oncogene mutations should be investigated further. APC loss establishes the adenomatous phenotype, while KRAS activation enhances proliferation, reduces apoptosis, and facilitates invasion. The adenoma-carcinoma sequence may be accelerated in some patients through genetic and environmental changes including bacterial colonization.

The germline APC variants were not located within a particular region in the 15 FAP patients, with variant located at the 5′ terminal (S89* in FAP11) and two at the 3′-terminal (Q1928Rfs* in FAP1, F1933Lfs* in FAP5). Whereas APC somatic mutations in the MCR were frequently observed in polyps of our FAP patients. In contrast, APC mutations in sporadic adenomas were more widely distributed in the 5′- half of the gene. Mutations in this region have been reported to maintain 1 or 2 of the 20AARs, controlling the ability to bind β-catenin, which translocates into the nucleus and activates transcriptional factors in the Wnt signaling pathway (Aitchison et al., 2020). The ‘just-right’ signaling model explains these observations. In our analysis of attenuated FAP patients, in whom germline pathogenic variants were in the 5′ and 3′ terminals, the somatic mutations were located heterogeneously, retaining 0-3 20AARs. The loose-fit hypothesis was clearly adopted. The number of preserved 20AARs is quite different in sporadic colorectal cancer (Christie et al., 2013). In contrast, FAP patients with whole APC deletion (FAP13 and 14) had somatic mutations in MCR retaining 2 20AARs. The correlation between germline variant and somatic mutations is associated with genotype-phenotype correlation.

One of the important discoveries in this study was the three-hit APC alteration, which was detected in nine polyps from seven FAP patients. They retain the first hit in the germline, and the second and/or third hits occurs in the somatic region (Sieber et al., 2006). Judging from the locations of the second and third hits, we could not identify any rules. The three-hit hypothesis that mutations and polyploidy cause multiple changes in a single cancer cell was proposed in 2009 (Segditsas et al., 2009). Subsequently, many types of genetic alterations were detected by NGS analysis because it is more sensitive than Sanger sequencing. Ultrasensitive genotyping of APC mutations using nanofluidic PCRs clearly demonstrates intra-crypt heterogeneity (Gausachs et al., 2017; Garcia et al., 1999). Therefore, the second and third hits were considered to occur at separate crypt levels. The heterogeneity of tumor cells in a single polyp may produce multiple crypts with a few hits in FAP carcinogenesis.

The present study has some technical limitations. First, the genetic alterations analyzed in this study were limited. The whole-exome sequence of the mutational signature was investigated in a recent study (Aitchison et al., 2020) Because FFPE specimens were used for genetic analysis, short-read sequencing was performed. Long-read sequencing may provide additional information regarding the zygosity of the three-hits. Second, epigenetic alterations have not yet been investigated. The most common possible epigenetic change (i.e., gene methylation) frequently reduces proteins (Sieber et al., 2006). Transcriptomic profiling of carcinogenesis requires fresh tissues from patients.

Through NGS analysis of colorectal neoplasms, we identified the genetic profile of the colonic adenoma-carcinoma sequence and the high prevalence of APC somatic mutations in colorectal tumors from both FAP patients and sporadic cases. A combined analysis of germline and somatic alterations revealed ‘three-hit’ alterations in the APC gene among FAP patients, suggesting that the heterogeneity of colorectal carcinogenesis contribute to these genetic changes. The observed correlation between germline pathogenic variants and somatic mutations resembles a genotype-phenotype relationship. Further structural analyses are needed to elucidate the mechanisms of tumor development in colorectal polyps of FAP patients.