Cause, Epidemiology, and Histology of Polyps and Pathways to Colorectal Cancer

INTRODUCTION

Gilbertson first suggested in the 1960s that colorectal cancer (CRC) may arise from intermediary lesions in the colon.¹ Subsequent studies demonstrated foci of adenocarcinoma in adenomatous polyps, termed “adenomas,” which suggested these polyps might develop into cancers.² In the late 1980s, Fearon and Vogelstein described CRC as a genetic disease, with the polyp-to-carcinoma progression as a sequence of specific genetic mutations.³ Some of these mutations may be preexisting in the blood (ie, the germline), whereas others accumulate over time in the colonic epithelium due to environmental or other factors (i.e., somatic mutations) and drive the tumorigenesis pathways from a precursor lesion to CRC. Given the length of time it takes to progress through the polyp-to-carcinoma sequence, it was suggested that CRC could be prevented by detecting and removing these precancerous colon polyps.³

Since then, several studies have demonstrated that CRC screening and removal of precancerous colon adenomas leads to a reduction in CRC incidence and mortality by approximately 50%.^4–12 Ongoing research into the cause and epidemiology of these precancerous colon lesions continues to inform CRC prevention by helping to refine optimal screening strategies, create new screening modalities, and identify new therapeutic targets.

CAUSE

In the mid-1970s, pathology studies of polyps and CRCs suggested that colorectal adenocarcinoma may progress from adenomatous polyps. Areas of adenomatous tissue are sometimes found in cancers, whereas conversely, foci of cancer are often observed in larger adenomas (i.e., greater than 1 cm).² A natural history study of unresected colonic polyps greater than 1 cm in size in patients who declined surgical resection reported a 24% risk of invasive adenocarcinoma at the site of the index polyp and a 35% risk of carcinoma at any colonic site after 20 years.¹³ In addition, familial adenomatous polyposis (FAP), an inherited disease, is caused by germline mutations in the adenomatous polyposis coli (APC) tumor suppressor gene and is associated with numerous colonic adenomas. Patients with FAP almost universally develop CRC by age 40 years if they are not managed with total proctocolectomy.¹⁴ Inactivating mutations in the APC tumor suppressor gene are also commonly found in sporadic nonhereditary lesions, such as small adenomas and larger carcinomas, whereas oncogenic KRAS and loss-of-function TP53 mutations are mostly limited to adenomas greater than 1 cm in size and carcinomas.³ Based on these studies and others, in 1990 Fearon and Vogelstein proposed a multistep genetic model of colorectal carcinogenesis in which inactivation of the APC tumor suppressor gene occurs first in normal colonic mucosa, followed by activating mutations in the KRAS oncogene and subsequent additional mutations. Recent efforts to classify CRCs by gene expression profiles have resulted in 4 consensus molecular subtypes (CMS1–4), each associated with distinct molecular features and clinical outcomes (Fig. 1).¹⁵

Fig. 1.

Pathways of colorectal carcinogenesis. Activation of the Wnt pathway (primarily via APC mutation) or a mutation in BRAF can initiate colorectal tumorigenesis. BRAF mutations promote tumorigenesis via the serrated neoplasia pathway, leading to microsatellite instability (MSI) with hypermutation or microsatellite stable (MSS) without hypermutation (indicated in the figure). Colorectal tumor classifications include chromosomal instability (CIN), MSI, and the serrated pathway (see CMS). EMT, epithelial to mesenchymal transition; H, high; L, low; neg, negative. (From Nguyen LH, Goel A, Chung DC. Pathways of Colorectal Carcinogenesis. Gastroenterology. 2020 Jan;158(2):291–302.)

Chromosomal Instability Pathway

Somatic mutations accumulate with aging in the normal intestine and colon at a rate of approximately 40 novel mutations per year.¹⁶ Colonic stem cell division is correlated with cancer risk in epidemiologic analyses, which suggests that mutagenesis in normal colonic stem cells underlies cancer initiation.¹⁷ However, the rate of spontaneous mutagenesis is believed to be too low to account for the multiple mutations required for CRC development (proposed to be 3 mutations).¹⁸ Rather, up to 70% of CRCs exhibit chromosomal changes such as somatic copy number alterations caused by aneuploidy, deletions, insertions, amplifications, or loss of heterozygosity. These changes are caused by defective chromosomal segregation that results in karyotypic variability between cells.

Karyotypic instability typically results in mutations in the tumor suppressor genes APC and TP53 and in the oncogene KRAS. Mutations in APC result in activation of the Wnt signaling pathway and are typically the earliest event in the initiation of colon adenomas in the chromosomal instability pathway. CMS2 CRC are characterized by activation of the Wnt signaling pathway.¹⁵ Studies in animal models and patient-derived colon cultures or “organoids” have demonstrated that loss of APC transforms intestinal or colonic stem cells (located at the base of crypts) into adenomas by activating the Wnt signaling pathway and that additional mutations in oncogenic KRAS and TP53 then accelerate progression to cancer.^19–22 Indeed, mutations in KRAS often arise after inactivating APC mutations and are found in approximately 40% of tumors.³ Oncogenic KRAS mutations result in constitutive activation of the ectodermal growth factor receptor (EGFR) signaling pathway. Inhibitors of EGFR are used to treat metastatic CRC but are ineffective in KRAS-mutant cancer.²³ CMS3 cancers exhibit metabolic dysregulation such as sugar, amino acid, and fatty acid metabolism.¹⁵ Other commonly mutated genes in the chromosomal instability pathway that occur late in the adenoma-carcinoma sequence include PIK3CA (found in 10%–20% of colorectal tumors) and TP53 (found in greater than 60% of colorectal cancers).^24,25 Mutations in genes that encode proteins in the transforming growth factor pathway such as SMAD2 and SMAD4 explain only a small portion of the 70% of advanced cancers with loss of heterozygosity in chromosome arm 18q.^25,26 Recent deep sequencing studies of mutations in CRC have confirmed previously identified cancer-associated genes and have revealed many uncharacterized somatic mutations that could be either “driver” mutations with an essential role in disease pathogenesis or “passenger” mutations that are incidental (Fig. 2).^25,27,28

Fig. 2.

Mutational landscape in microsatellite stable (MSS) colorectal cancer. Seshagiri and colleagues performed somatic exome sequencing of 57 MSS colorectal cancers. Genes were evaluated for significance using Q score criteria. Each circle represents a gene, and the size of the circle is proportional to the mutation count for that gene. The genes are represented in order of increasing number of expected mutations from left to right on the x axis. Genes with a statistically significant Q score are labeled. (From Seshagiri S, Stawiski EW, Durinck S, Modrusan Z, Storm EE, Conboy CB, Chaudhuri S, Guan Y, Janakiraman V, Jaiswal BS, Guillory J, Ha C, Dijkgraaf GJ, Stinson J, Gnad F, Huntley MA, Degenhardt JD, Haverty PM, Bourgon R, Wang W, Koeppen H, Gentleman R, Starr TK, Zhang Z, Largaespada DA, Wu TD, de Sauvage FJ. Recurrent R-spondin fusions in colon cancer. Nature. 2012 Aug 30;488(7413):660–4.)

Microsatellite Instability Pathway

The microsatellite instability (MSI) pathway is observed in approximately 15% of CRCs. Our current understanding of the MSI pathway of CRC is based on the discovery, first made in yeast by Paul Modrich, that DNA microsatellite sequence fidelity after DNA replication is maintained by a DNA mismatch repair (MMR) system.²⁹ The MMR genes are MSH2, MLH1, MSH6, MSH3, PMS2, and EPCAM; the MSH2 protein forms heterodimers with MSH6 as well as MSH3, MLH1 pairs with PMS2, and EPCAM encodes a protein that regulates MSH2. MSI (i.e., MSI-high) refers to the presence of frameshifted microsatellite sequences from genomic DNA due to defective MMR of DNA replication mistakes at microsatellites.³⁰ Defective MMR due to a germline mutation in an MMR gene causes accelerated colorectal tumorigenesis in an autosomal dominant inherited condition known as Lynch syndrome. Lynch syndrome is the most common hereditary CRC syndrome and accounts for approximately 3% of CRC. These patients often present 10 to 30 years younger than patients with sporadic CRC.³¹ These tumors are classified as CMS1 (i.e., hypermethylation of DNA, MSI-high, and immune cell infiltration).²⁹ Colorectal tumors in patients with Lynch syndrome that arise from germline defects in MMR genes have MSI but typically do not have BRAF mutations. In contrast, the most common cause of the sporadic MSI phenotype is epigenetic silencing of MLH1 because of promoter methylation. These MSI pathway cancers often manifest with BRAF mutations and a lower frequency of APC and TP53 mutations.³² In addition, the MSI pathway is characterized by a high level of methylation at regulator regions in the genome, known as the cytosine/guanine (CpG) island methylator phenotype (CIMP).

Serrated Pathway

Recent advances in endoscopic technology have led to the detection of flat or sessile polyps that pathologically have a serrated or saw-toothed appearance. Serrated neoplasia may account for 15% of CRCs. Hyperplastic polyps comprise approximately two-thirds of serrated lesions but rarely progress to cancer.³³ Hyperplastic polyps feature narrow crypt bases with serration confined to the upper region. On the other hand, sessile serrated lesions (SSLs) are distinguished from hyperplastic polyps by abnormal proliferation, branching or dilated T-/L-shaped crypts, and are often challenging to diagnose by pathologists. Traditional serrated adenomas are serrated lesions with villiform projections and are less common than sessile serrated lesions.³⁴

A distinguishing feature of the sessile serrated pathway is an activating mutation in the BRAF oncogene, which is rare in convention adenomas from the chromosomal instability pathway; BRAF and KRAS mutations are mutually exclusive.³⁵ BRAF-mutant serrated lesions can progress into cancer either through the MSI pathway, via mutations in an MMR gene or MLH1 methylation causing an MSI-high phenotype, or with additional mutations in TP53 and other cancer-associated genes that result in a traditional serrated adenoma and then an microsatellite stable (MSS) cancer.³⁶ Serrated lesions that progress on the MSI-high pathway are often classified as CMS1. Serrated lesions that progress along the MSS pathway typically develop into CMS4 cancers, with an immunosuppressive microenvironment that facilitates tumor invasion, immune evasion, and poor survival.¹⁵ The Wnt signaling pathway is often activated in SSLs, via inactivating mutations in RNF43 (a gene encoding an E3 ubiquitin ligase that negatively regulates Wnt signaling) rather than missense mutations in APC.³⁷ Serrated pathway cancers (both MSI-high and MSS) also feature high levels of cytosine/guanine (CpG) island methylation or CIMP; hypermethylation of these promoter-rich regions leads to inactivation of nearby tumor suppressor genes and subsequent tumorigenesis.³⁸ BRAF-mutant cancers that arise from the sessile serrated pathway may respond to treatment with dual BRAF/MEK inhibition.^39,40

EPIDEMIOLOGY

Prevalence Rates in the United States and Internationally

Although CRC is a preventable disease, it remains the second leading cause of cancer death in the United States and globally, with a lifetime risk of approximately 4% to 5%.^41–43 Current models that inform the United States Preventative Services Task Force (USPSTF) guidelines on CRC screening estimate that, in the absence of screening, between 7.7% and 8.5% of 40-year-olds would develop CRC in their lifetimes, and 3.2% to 3.4% would die from this disease.⁴⁴ Incidence rates are highest in developed countries, but increasing westernization of diet and lifestyle in developing countries is expected to significantly increase the global burden of disease by 2030.^42,45 Fortunately, CRC incidence and mortality is decreasing in the United States (Fig. 3), most European countries, and much of the Asia-Pacific.^45–48 CRC incidence has decreased in the United States by greater than 35% since the widespread use of screening in the 1990s.⁴⁹ Indeed, detecting precancerous lesions is now an important goal of CRC screening,^44,50,51 as removal of polyps during colonoscopy or sigmoidoscopy has been shown to effectively prevent CRC in observational and randomized studies.^{5,9,10,12,52–56} Other factors also explain this decline in CRC incidence, such as decreased tobacco use, improved diet, more widespread use of aspirin for cardioprotection, more effective surgical techniques, and advances in chemotherapy.⁵⁷ Therefore, understanding the epidemiology of precancerous lesions in the colon is critical to developing effective CRC prevention programs.

Fig. 3.

Incidence rates of colorectal cancer and prevalence rates of adenomas/SSLs by age. (A) CRC incidence rates by age and compared across time periods (screening era of 2014–2018 vs an unscreened era in 1975). (B) Prevalence rates of adenomas on screening colonoscopy compared with SSLs on index colonoscopy by age. Figure was produced from data found in References ^49,70,76,142

Models that inform adenoma prevalence data are based on global autopsy studies before the dissemination of screening programs. These models estimate the prevalence of adenomas from approximately 20% at age 45 years to greater than 50% to 60% by age 85 years.^58–68 Contemporary estimates of adenoma prevalence are based on studies of colonoscopy, which generally report adenomas in 20% to 53% of individuals 50 years or older, including up to 9.7% with advanced adenomas (defined as an adenoma sized ≥10 mm or with villous features or with high-grade dysplasia).^{4,11,12,54,55,57,69–75} Meta-analyses of colonoscopy studies (age ≥50 years) suggest an overall adenoma prevalence rate of approximately 24% and advanced adenoma prevalence rate of about 4.5%.^76,77 However, rates of adenomas are also estimated to be higher in men and increase with age (see Fig. 3).^70,77 Recent evidence has also suggested a trend toward a greater proportion of adenomas in the proximal colon with increasing age, although it is unclear if this is a true increase or related to recent improvements in methods of detection for proximal polyps.^70,77–79

Between a quarter to one-third of all CRCs develop from the sessile serrated pathway.^80–82 Serrated lesions include both SSLs and hyperplastic polyps.^57,81,82 Although it remains inconclusive whether hyperplastic polyps and SSLs arise independently,⁸³ and whether hyperplastic polyps have malignant potential (particularly if >5 mm and/or proximally located),^81,82,84,85 current evidence indicates that SSLs may develop into cancer. .^81,82 Rates of SSL detection vary between 1% and 18%; the true prevalence remains uncertain.^33,86–93 The high variability in prevalence estimates is likely due to differences in detection, resection, and classification practices across studies. For example, sessile and flat lesions are approximately 5 times more likely to be missed than pedunculated polyps, and when detected, there is variability among pathologists in distinguishing hyperplastic from SSLs.^94–96 Importantly, almost half of patients with SSLs are found to have adenomas on the same examination.^89–93 Yet, in contrast to adenomas, prevalence rates of SSLs generally do not substantially vary by age (see Fig. 3) or sex (despite conflicting evidence suggesting either sex may have higher risk^88,97). On the other hand, hyperplastic polyps have prevalence rates on colonoscopy studies generally between 10% and 15% but can be up to 30% in certain populations.^{11,33,90,97–102} Hyperplastic polyps are typically less than or equal to 5 mm and occur in the distal colon, whereas SSLs are larger and have a predilection for the proximal colon.⁸¹ Finally, traditional serrated adenomas are villous polyps, often erythematous and (semi-)pedunculated, and located in the distal colon (with an average size of about 15 mm). These polyps have prevalence rates of less than 1% at screening colonoscopy.^{86,89,91,101–104}

Table 1 summarizes the epidemiology of different colon polyps. Most importantly, detection of colon polyps is critical to preventing CRC. In landmark studies by Corley and colleagues and Kaminski and colleagues, increased adenoma detection rates (ADR) in the highest quintile on screening colonoscopy were associated with a linear decrease in incidence of interval CRC by 48% to 82%.^71,74 These findings led to the development of ADR as an important quality metric of screening colonoscopy, with recommendations for ADR of greater than or equal to 25% for the general screening population. Similar efforts are ongoing to establish benchmark detection rates for SSLs.^80,89,90 Given an increasing focus on quality metrics emphasizing adenoma detection, together with advances in colonoscopy technology and technique, prevalence rates of adenoma and SSL are expected to increase.^71,74,105 However, it remains unclear whether increased detection of diminutive and nonadvanced polyps will translate to further decreases in CRC incidence and mortality or whether the benefits will be outweighed by the cost and risks of increased utilization of follow-up colonoscopy.^106–109 Therefore, significant efforts are also ongoing to improve CRC screening uptake among the unscreened population and address important disparities and barriers that remain to optimal primary prevention, especially given that most CRC diagnoses are in unscreened populations.^110–113

Table 1.

Summary of colon polyp epidemiology

Features	Adenomas	Sessile Serrated Lesions	Hyperplastic Polyps
Prevalence	Up to 60%	1%–18%a	Up to 30%
Age≥50 y	20%–60%	No data	No data
Age<50 y	20%	No data	No data
Cause	Chromosomal instability or microsatellite instability pathways	Sessile serrated pathway	Sessile serrated pathway
Location	Evenly distributedb (34% rectum, 32% distal colon, and 34% proximal colon)	Up to 90% proximal colon	Up to 80% distal colon
Natural history	If size <10 mm, typically with low risk of progression to CRC. Higher risk of progression if size ≥10 mm or with villous or high-grade dysplasia, with ~1%–5% annual progression to CRC.	Possibly slower growth rate compared with adenomas, with annual progression to CRC of <1%. SSLs ≥10 mm with similar risk of CRC compared with large adenomas.	Unlikely neoplastic, although size >5–10 mm and/or proximally location may increase risk.

^aTrue prevalence unknown.

^bIn individuals younger than 50 years, location is shifted to rectum (42%), distal colon (31%), and proximal colon (27%).

Natural History

Cancer prevention by detecting and removing a precancerous lesion is a feature unique to CRC and only a few other malignancies (e.g., cervical cancer). IThe mean overall time from adenoma incidence to cancer development ranges from 10.6 to 25.8 years across all age groups, and the time from preclinical cancer development to (symptomatic) cancer detection is estimated at approximately 1.6 to 4.0 years.^{44,58,114,115} However, understanding the natural history of colon polyps, or the “window” of the precancerous lesion’s existence, has important implications for CRC screening and surveillance programs. For example, slower progression rates would lead to increased effectiveness of screening and decreased utility of surveillance, whereas increased progression rates would have the opposite impact.

Growth rates are believed to be highest in adenomas with size greater than or equal to 10 mm or with villous or high-grade dysplasia. These “advanced adenomas” are felt to be nearing CRC in progression, as studies estimate approximately 1% to 5% annual progression rates of advanced adenomas to CRC, with rates in the lower range for women overall, and rates typically on the higher end in older age groups.¹¹⁶ A longitudinal barium enema study of polyps greater than or equal to 10 mm in size estimated the cumulative risk of invasive CRC at 1, 5, and 10 years of 1%, 2.5%, and 8%, respectively.¹³ i Longitudinal barium enema studies also found that most of polyps less than 10 mm in size grew so slow that “they could not have achieved a significant size during the longest human life span.”^117,118 However, although some argue that polyps greater than or equal to 5 mm may be of little or no clinical import, given that less than 1% contain advanced features, studies suggest that up to 10% of small (6–9 mm) polyps actually harbor advanced histology, including 1% with CRC^{106,108,119–122}; this would argue the importance of colonoscopy with polypectomy in patients with small polyps, though this is an area of ongoing investigation.

Finally, despite increasing recognition of the contribution of the sessile serrated pathway to a relevant proportion of CRCs, flat SSLs were largely unobservable or unrecognized in the past and are not included in the current models that inform national guidelines. Yet, a recent study that incorporated the SSL pathway into modeling estimates found that predictions of screening effectiveness were similar to models that assume all CRCs arise only from adenomas. This finding is likely related to the fact that many patients with SSLs also have adenomas (and thus undergo complete resection of both lesions) and improved recognition of these lesions in practice is now effectively mitigating risk from both lesions.¹²³ This study also estimated that annual rates of progression from SSLs to CRC were less than 1% across age groups and sex (much less than rates for advanced adenomas).¹¹⁶ A recent CT colography study confirmed the suggestion of slower growth rates for SSLs: when comparing SSLs (mean size 9.3 mm) with adenomas (mean size 6.3 mm), only 22% of SSLs achieved a 20% growth increase over mean 5.3 years compared with 41% of adenomas.¹²⁴ Stratifying the malignant potential of various SSLs is currently uncertain and an area of ongoing research.

YOUNG-ONSET NEOPLASIA

Incidence and mortality of young-onset (YO) CRC, defined as CRC occurring in individuals younger than 50 years, has been increasing worldwide over the past 3 decades.^125–127 In 1992 the incidence of CRC among young adults was 8.6 per 100,000, which has increased by 45% to 12.5 per 100,000 in 2015.¹²⁸ Furthermore, the CRC mortality rate among young adults has increased by 1.3% annually from 2008 through 2017, compared with a decrease of 3% annually in adults aged 65 years and older.¹²⁹ Patients with YO CRC have a higher prevalence (53%–72%) of advanced-stage (III–IV) disease at the time of diagnosis compared with average-onset (AO) CRC patients (41%–63%).^130–132 This disparity has been suggested to be related to either delays in diagnosis^125,130 versus more aggressive tumor biology.^132–136 However, a large retrospective analysis that excluded patients with syndromic or genetic predispositions for CRC found no difference in the tumor histopathology or mutational burden of sporadic YO CRC compared with sporadic AO CRC.¹³⁵ Furthermore, although CRC in individuals older than 50 years is generally equally distributed throughout the colon, YO CRC occurs most frequently in the rectum, then distal colon, and least commonly in the proximal colon (42%, 31%, and 27%, respectively).^125,137,138 In fact, patients with YO CRC commonly present with rectal bleeding or other symptoms associated with left-sided disease.¹³⁵ However, colonoscopy is still recommended in adults younger than 50 years with other “alarm” symptoms, such as unexplained hematochezia and/or iron deficiency anemia, given the significant association with YO CRC (5-year cumulative incidence for YO CRC of 0.45% if these signs or symptoms are present).¹³⁹

In 2018, the American Cancer Society issued a qualified recommendation to start CRC screening at the age of 45 years for average-risk persons.¹⁴⁰ Although this statement was based on modeling data, emerging observational data support the suggestion that individuals aged 40 to 49 years may benefit from screening to detect and remove precancerous adenomas. A recent meta-analysis by Kolb and colleagues of average-risk patients undergoing colonoscopy younger than 50 years found that the rates of any colorectal neoplasm and advanced neoplasm were 14% and 2%, respectively.⁷⁶ Furthermore, this study showed that the rate of advanced neoplasia was statistically similar in the 45 to 49 years age group compared with the 50 to 59 years population (3.6% vs 4.2%, respectively). Another systemic review studying the yield of screening colonoscopy in both average risk and symptomatic individuals younger than 50 years (18–49 years) found that the prevalence of YO adenoma was 9%, with increasing age being the most consistent risk factor.¹⁴¹ And although there is a paucity of data related to SSL prevalence in those younger than 50 years, Anderson and colleagues found a similar prevalence rate of SSL in those aged between 40 and 49 years and older than 50 years.¹⁴² Taken together, these studies have informed recent CRC screening guidelines, which suggest that expanding screening to individuals aged 45 years may yield a similar impact on the reduction of CRC incidence as has been observed in those aged 50 years and older.^44,143 Nevertheless, the true impact of screening and surveillance on YO CRC is still unknown and more studies are required to address this question.

SUMMARY

Studies of the cause and epidemiology of precancerous colon polyps have significantly affected clinical practice and public policy in the last few decades. This research has informed public health initiatives for education and risk mitigation, screening recommendations across various populations, new (and often less-invasive) screening modalities, and potential chemotherapeutic options. However, despite the increasing detection of precancerous colon polyps, the lifetime risk for CRC remains approximately 4% to 5%.⁴⁹ In the future, personalized risk assessment tools based on clinical and genetic epidemiologic insights could help better target screening and surveillance resources to truly high-risk individuals, while reducing exposure to the costs and risks inherent to large-scale screening (particularly with invasive procedures) among those at low risk of CRC.

KEY POINTS.

• Colorectal cancer (CRC) is a genetic disease that develops from precursor colon lesions/polyps that progress through various tumorigenesis pathways.

• Clinical and genetic epidemiologic studies of colon polyps inform CRC screening programs, which aim to detect and remove precancerous colon polyps to prevent CRC.

• The incidence of CRC has increased markedly in younger individuals, but the epidemiology of young onset neoplasia remains unclear.

CLINICS CARE POINTS.

• Understanding the epidemiology of colon polyps will help providers have informed discussions with patients regarding personalized colorectal cancer risk assessments.

• Insights in the natural history of colon polyps, including the long latency period of tumorigenesis pathways, will allow providers to make recommendations for appropriate colorectal cancer risk reduction strategies, including follow-up intervals for screening.

• Knowledge about the increasing burden of young-onset colorectal cancer will help providers advise patients on the timing of CRC screening and follow-up studies. Furthermore, providers must have a low threshold to investigate any ‘alarm” symptoms in patients, particularly those younger than 50 years.