Biology of Precancers and Opportunities for Cancer Interception: Lesson from Colorectal Cancer Susceptibility Syndromes

Cancer is thought to be a multistep evolutionary process arising from a single cell, acquiring genomic alterations, which provide a fitness advantage (1, 2). Syndromic constitutional pathogenic variants (PV) predispose individuals to a high risk of developing cancer, providing an opportunity to identify and explore precancerous lesions in these patients. Of the cancer susceptibility syndromes affecting the gastrointestinal tract, the two best defined syndromes are Lynch syndrome (LS) and familial adenomatous polyposis (FAP). Studying these groups on regular cancer surveillance provides insight into the tumor biology, potential avenues for prevention, and their effectiveness. In this article, we will review some of the lessons that can be learned from these conditions which may afford opportunities for cancer prevention.

FAP is a rare dominantly inherited syndrome, the hallmark of which is the development of up to hundreds or thousands of colorectal adenomas and almost inevitable development of colorectal cancer without intervention. Constitutional PV's in the tumor suppressor gene adenomatous polyposis coli (APC) result in constitutive activation of the Wnt signaling pathway through deregulation of β-catenin, causing downstream effects on proliferation and differentiation within colonic crypts. Somatic APC variants are also found in more than 80% of sporadic colorectal tumors. Given its well defined phenotype FAP has been considered a good model for the study of colorectal cancer and chemoprevention. Due to the ubiquitous expression of APC, it is not surprising that FAP is a multisystem disorder and there are other precancerous lesions to consider, for which intervention may be required.

LS is an autosomal dominant inherited condition which increases the lifetime risk of colorectal cancer, as well as other predominantly epithelial cancers, and affects 1 in 400 adults (3, 4). It is defined by the presence of a constitutional PV in one of the DNA replication mismatch repair (MMR) genes. LS-related cancer diagnoses are common in younger people below national screening ages, with a significant impact on quality-adjusted life years gained through early diagnosis and prevention. Therefore, interventions are applied to try and mitigate these risks, which include regular colonoscopy, chemoprophylaxis and preventative gynecologic surgery (5). Thorough screening programs throughout patients’ lifetimes allow for multiple time-point studies of the evolution of LS-associated cancers from pre-dysplasia (normal biopsies) through to early and late dysplasia (adenocarcinoma), illuminating the otherwise unseen precancerous stage of lesion development in sporadic cancers. Recent cancer prevention and early detection innovations have moved towards the integration of molecular knowledge and risk stratification profiles, to allow for a more accurate characterization of at-risk individuals (6).

Mismatch repair deficiency (dMMR) drives mutagenesis and genomic instability in LS-associated cancer, along with approximately 15% of sporadic colorectal cancer. dMMR is not itself an initiator of oncogenesis alone, but instead accelerates mutation accumulation and therefore the chances of acquiring a driver mutation, a phenomenon termed hypermutator phenotype.

Compelling data have been published relating to premalignant molecular events in the context of LS colorectal cancer, including the identification of dMMR but morphologically nonneoplastic colonic crypts (7, 8). These crypts exhibit MMR gene variants and are likely precursors to malignancy, but also undergo a high rate of regression under immune surveillance. Other evidence that dMMR is an early event in LS carcinogenesis includes data which indicate that 77% of adenomas are dMMR, and the heterogeneity of somatic driver mutations in LS cancers other than dMMR (9). Indeed, Ahadova and colleagues described how known colorectal cancer mutations such as KRAS and APC mutations commonly occur after the onset of dMMR in LS (Fig. 1). Together this suggests dMMR clones evoke high negative selective pressures on them by the immune system; a mechanism that could potentially be harnessed and heightened by preventative medicine.

The time required for cancer development in LS has been estimated to be much shorter than in sporadic colorectal cancer: the time interval between a clear colonoscopy and later carcinoma formation in LS is around 2 to 3 years (10) whereas estimations of sporadic cancer evolution suggest it could take decades to evolve (11), with incidences of sporadic cancer within 3 years of a clear colonoscopy being quite rare (12). This acceleration of cancer onset is attributed to bypassing the need for two loss of function events in each of the MMR alleles, leading to an increase in mutational rate and therefore likelihood of oncogenic mutations (13). The shortened timeframe of cancer evolution in LS allows for a timelier investigation of the processes needed for cells to overcome for malignancy. The information from LS studies may also be translated to understanding the longer sporadic colorectal cancer development.

The identification of ‘submucosal’ colorectal cancer is a phenomenon rarely observed outside the context of LS. It may account for only a small number of colorectal cancer in the context of LS, but nevertheless suggests that some cancers could arise via a non-adenoma precursor route (Fig. 2; ref. 14). Colonoscopy may have limited efficacy in the prevention of such colorectal cancers, and require augmentation in its approach, or goals more focused on early diagnosis than prevention. This pathway of progression seems to be more common in MLH1 and MSH2 variant carriers (15) and less common in PMS2-associated colorectal cancer, an observation that may reflect the minimal risk of cancers in PMS2 carriers under surveillance. It has been shown that CTNNB1 and TP53 mutations occur more commonly in tumors lacking evidence of non-flat morphology polypoid growth (9).

Unexpectedly high rates of post-colonoscopy colorectal cancer have been reported, with cumulative incidences up to 45% in MLH1 carriers, for example, in comparison of the 9.3% rates reported in population-based screening programs. It is important to note that none of these studies considered the potential impact of quality of colonoscopy on this potential failure of colonoscopy surveillance. These data have led to the generation of the hypothesis that post-colonoscopy colorectal cancer in LS arises from an “invisible” subepithelial lesion that progresses directly to colorectal cancer. However, the true proportion of submucosal versus mucosal tumors is unknown (16).

Tumor mutational signatures may indicate a genetic predisposition that underlies the cancer, and facilitate personalized cancer therapy and prevention, by integrating the somatic mutation landscape within a single tumor to identify somatic molecular patterns associated with distinct oncogenic pathways (17). In addition, appreciation of early somatic events in hereditary colorectal cancer precursors including adenomas may contribute towards prediction models for progression of patients undergoing longitudinal endoscopic surveillance (18).

Mismatch variants accumulate in dMMR cancers resulting in hypermutated tumors with somatic mutations, including insertion/deletion mutations within repetitive sequences which are susceptible to replication errors, including in exons (coding microsatellites, cMS). These indels promote translational frameshifts, which generate truncated frameshift peptide (FSP)-encoding neoproteins (19). Several studies have identified a large spectrum of genes affected by such frameshift mutations, demonstrating that indel mutations affecting key tumor suppressors, such as the TGFBR2, are enriched in dMMR cancers. Specific dMMR-related FSP neoantigens can encompass neo-epitopes completely unknown to the host's immune system. This mechanism is commonly considered to be responsible for the high immunogenicity of dMMR cancers, demonstrated by several studies showing dense local immune cell infiltration and reactivity to FSPs. Nonneoplastic dMMR colonic crypts are seen in healthy LS carriers, suggesting that the healthy colon of patients with LS is itself a key source of immunogenic FSPs that serve to auto-vaccinate such patients and suppress dMMR-induced carcinogenesis (7).

Vaccination with recurrent FSPs that are shared by multiple dMMR tumors of different patients is a promising approach to boost immune surveillance of dMMR precancerous cell clones, and potentially immune-mediated regression of subclinical dMMR tumors for effective immunoprevention. A phase IIa clinical trial in patients with a history of dMMR colorectal cancer has demonstrated the safety and immunologic efficacy of a trivalent recurrent FSP-based vaccine (19). Whether recurrent FSP vaccination can reduce LS or sporadic dMMR tumor burden or prolong patient survival is unknown however.

A more precise understanding of the mechanisms by which LS-associated carcinogenesis escapes immune surveillance may facilitate the translation of such discoveries into immune-based cancer prevention, for example by elucidating the role of HLA genotypes. Early data from mouse models suggest this is a potentially fruitful mechanism in combination with NSAIDs, with early studies in human subjects indicating the safety of LS vaccines (19, 20).

In sporadic colorectal cancer, correlations have been made between gut microorganisms, intestinal barrier function and inflammation. There are few data to indicate a ‘microbiome signature’ in hereditary colorectal cancer syndromes (21). There appears to be an interaction between MMR and TGFBR2 inactivation in inflammation-associated colon tumorigenesis (22). However, the dominant characteristic thus far of the microbiome in LS relates to previous colorectal resection (21).

In FAP, biofilms including Escherichia coli and Bacteroides fragilis have been observed penetrating the mucous barrier, indicating that the intestinal barrier function is compromised (23). There are data suggesting impaired cellular immunity and tumor surveillance in FAP. This implies that mucosal immune dysfunction may contribute to carcinogenesis in this predominantly genetically driven condition (24).

Microbiome signatures from average risk screening populations may augment existing population screening tools, including the fecal immunochemical test. This is being explored as part of an existing multicenter trial of people with LS in the UK (25). Other potential biomarkers are being evaluated in patients with LS, with the advantage that the benefits in such populations may be more tangible in a shorter timeframe, before translation of other predominantly high-risk populations. These include organ-specific biomarkers (e.g., urinary biomarkers for endometrial or urothelial cancer and fecal or plasma genomic markers for colorectal cancer), as well as non–organ-specific biomarkers including cell free DNA from plasma (26–29).

Although much focus is on the cancer risk in the large bowel in both FAP and LS, both carry an increased risk of cancer relating to the upper gastrointestinal (UGI) tract.

In FAP, there are emerging data regarding gastric adenomas and indeed an increased risk of gastric cancer in FAP (30–33), but these entities in FAP are yet to be well defined in terms of life-time risk, pathways to cancer and speed of cancer development; furthermore the differences in gastric cancer risk and pathways to cancer between the Western World and Asia make interpreting data in FAP even more challenging. It is also clear that there is an increased risk for gastric and duodenal cancer in patients with LS. However, the current data are difficult to interpret due to ascertainment bias and the true lifetime risks are not well established. Upper gastrointestinal tract surveillance in LS is variably recommended (5, 34) and this variability may relate to few data regarding the incidence, natural history or intervention outcomes for premalignant lesions. Therefore, it is difficult to extrapolate any of these data to sporadic disease.

Sporadic adenomas in the duodenum are rare, with a prevalence as low as 0.03% in those referred for diagnostic upper gastrointestinal endoscopy (35). They are usually sited at the level of, or just distal to, the ampulla of Vater. In FAP, duodenal adenomas are common; indeed, an inherited polyposis syndrome may account for 60% or more of those patients in whom a duodenal adenoma is found (36). The distribution of adenomas in FAP mirrors that of sporadic duodenal adenomas, with a propensity to develop in the peri-ampullary area (37).

In FAP, the lifetime risks of developing one or more adenomas approaches 100% (38) and of developing duodenal cancer is 5% to 10%therefore it is an area which seems ripe for study to better understand the sporadic counterpart. A staging system has been adopted (Table 1). Prospective data from a number of studies have shown that there is slow progression of all stages of duodenal disease which infrequently develop cancer (38–40). Understanding this natural history is useful in terms of understanding if and when to intervene in those with sporadic duodenal adenomas, or for studying the effect of any intervention.

Bile is a likely aetiological factor in duodenal adenoma development. The distribution of adenomas within the duodenum mimics exposure to bile in both sporadic and FAP duodenal adenomas. Bile from FAP and non-FAP patients has similar mutagenicity, however, the composition of bile relating to bile acids does appear to differ (41, 42). DNA adduct formation, as a measure DNA carcinogen exposure, is increased in FAP versus controls (43–45). Furthermore, the dietary administration of unconjugated bile, in a mouse model of FAP, increases duodenal tumor burden (46). Further studies to better define the etiology of duodenal adenoma in FAP, may well have direct benefits in terms of understanding and managing sporadic disease and directing new therapies.

Exploring chemoprevention in inherited conditions with a high cancer risk may be helpful as a model for sporadic cancer prevention. One of the difficulties in such studies is establishing what is an appropriate primary endpoint to measure. A reduction in cancer risk is the most important endpoint. In LS, with its lack of polyposis phenotype, then cancer has been used appropriately as the endpoint to be measured. However, in FAP, where there are evidence-based interventions to reduce cancer risk in the large bowel (namely prophylactic surgery) and where cancer risk in the duodenum remains uncommon, a number of surrogate endpoints have been used, however there remains much debate as to whether these endpoints in FAP are clinically meaningful and appropriate.

Aspirin and nonsteroidal anti-inflammatory drugs have been the most widely studied chemoprevention drugs in LS, FAP, and indeed, sporadic colorectal cancer. In CAPP2, patients with LS were assigned to receive 600-mg aspirin daily or placebo. The primary endpoint was development of colorectal cancer. The initial report was of significant benefit (47), however longer-term follow up outcomes demonstrate a reduction in colorectal cancer risk in those receiving aspirin [HR, 0.65; 95% confidence interval (CI), 0.43–0.97; P = 0.035] for aspirin versus placebo. These results reflects the growing data on the benefit of aspirin in prevention of sporadic colorectal cancer (48). It is interesting however to note that the timeframe before the benefit is shorter in LS than sporadic colorectal cancer, where benefit may be seen after 10 years (49). This difference likely relates to the rates of carcinogenesis between LS and sporadic colorectal cancer (49). Studies of aspirin in FAP have yielded conflicting results, likely due to the methodologies used (Table 2)

Sulindac and eflornithine was studied against either single agent alone, initially no difference in outcomes for the lower gastrointestinal tract was reported (50), but a post hoc analysis of participants with at least a partial intact lower gastrointestinal tract showed a statistically significant reduction in disease progression and need for surgery (once censored for polypectomy >10 mm; ref. 51). The combination of sulindac with erlotinib (52) and single agent weekly erlotinib (53) lead to a significant reduction in polyp burden after 6 months of treatment compared with placebo adverse events were common, limiting their potential as treatment in an otherwise well population, in these combination therapy studies.

Other agents have been trialled in FAP (Table 2), all of which have similar methodologic flaws; being short term and lacking a hard endpoint of cancer and using surrogate endpoints, the validity of which have to be questioned. This is especially important given there are a number of reports of cancer arising whilst on chemoprevention (54–57). Currently there are no data to support any agent preventing colorectal cancer in FAP, along with prevention of surgery data remaining weak with inadequate follow up.

Studies have addressed chemoprevention in the setting of duodenal polyposis in FAP. Sulindac has been studied in several randomized and non-randomized settings (58–61) and in combination with eflornithine (50). Given the possible role of bile and that bile-induced DNA adduct formation is pH-sensitive, ranitidine and the bile salt sequestrant ursodeoxycholic acid have also been investigated in FAP (62, 63). All failed to demonstrate regression of duodenal disease. Furthermore, all studies except the combination study were small and short-term.

Combined treatment with sulindac and erlotinib (52) and a phase II study with weekly erlotinib (53) have shown statistical benefit in terms of polyp burden but adverse events were unacceptably high. Celecoxib treatment (800 mg/day) leads to a nonsignificant decrease in polyp burden, although subgroup analysis showed a significant decrease in polyp burden in those with Spigelman stage III or IV disease (64).

In summary, no agent has been shown to reduce long-term duodenal cancer risk and none have been shown to reduce the need for major resectional surgery, which are the most clinically meaningful endpoints in FAP. With a better understanding of the etiology of duodenal adenomas, more targeted agents could be considered as a chemoprevention strategy to study.

In LS there is less data regarding chemoprevention in the upper gastrointestinal tract. In the CAPP2 study, a reduction in risk for extra-colonic LS-related cancers, including gastric and duodenal cancers, was not observed. However, a protective effect of resistant starch against non–colorectal cancer LS cancers [incidence rate ratios (IRR), 0.52; 95% CI, 0.32–0.84; P = 0.0075] was noted (65).

It is clear that for chemoprevention studies to be appropriately designed, there needs to be an understanding of tumor biology to dictate timeframes required for follow up. Robust clinically meaningful endpoints are required; cancer as an endpoint however may not be possible, as the size of study to show benefit and the duration may not be feasible for new agents with a role in cancer prevention in the setting a non-accelerated cancer pathway. These problems may limit the utility of FAP as a model to study sporadic colorectal cancer chemoprevention.

Hereditary syndromes provide an opportunity to study early precancer evolution surveillance and interventions to prevent cancer. Understanding genotype and phenotype in colorectal cancer susceptibility syndromes allows for customized clinical trials in preventive medicine to be highly effective in their outcomes. The increasing knowledge of precancer lesions and their biology may aid novel strategies for individualized prevention. It is essential to consider the adverse effects of such interventions when weighed against quality of life, particularly considering the age at which medication may begin. Exciting advancements are being made in the field of cancer vaccination against LS colorectal cancer, however until trials show their validity the recommended surveillance protocols must be upheld. Extrapolating information from these hereditary colorectal cancer studies can also help pinpoint how certain sporadic cancers evolve and offer opportunities for cancer prevention.

No disclosures were reported.