Introduction

Incorporating cancer genetic risk assessment into gastroenterology practice provides an excellent opportunity to optimize screening and surveillance strategies for colorectal cancer (CRC). A genetic predisposition to CRC has been recognized for decades, and there are several well-defined hereditary CRC syndromes. Historically, syndrome-specific testing was the standard for detection of highly penetrant familial CRC syndromes [1, 2]. The clinical introduction of non-Sanger-based next-generation sequencing (NGS) panels in early 2012 has greatly enhanced the identification of families with a hereditary predisposition to cancer. Despite increased availability of NGS panels and improved management and outcomes for hereditary CRC syndromes, genetic counseling and testing are often not performed for this indication [3]. This is in contrast to hereditary breast cancer, which has greater public awareness, more providers with genetic expertise, and a better defined systematic approach in identifying affected patients [4].

Among patients diagnosed with CRC, approximately 30% report a family history of CRC [5]. However, only 3–6% of all CRC patients carry identifiable highly penetrant gene mutations associated with hereditary CRC syndromes [6]. As new cancer predisposition genes are discovered, this gap continues to lessen [7]. With decreased cost and increased data made available by NGS, it has become apparent that cancer genes not previously associated with CRC are being detected, expanding the phenotypic spectrum of hereditary CRC. This has led to increased complexity in characterization of these cancer syndromes. The authors of this review will discuss guidelines for CRC genetic risk assessment, available screening tools, referral of high-risk patients, novel CRC-associated genes, and new developments in this field.

Hereditary CRC syndromes

Highly penetrant hereditary CRC syndromes are a well-characterized group of diseases with established inheritance and diagnostic criteria but with variable phenotypes, cancer risk, and management. Some, such as classic familial attenuated polyposis (FAP) in which patients typically present in early adolescence with hundreds to thousands of colorectal polyps, are easily identified. These patients have a 100% reported incidence of CRC if left untreated. Other well-established hereditary CRC syndromes include Lynch syndrome, attenuated FAP, MUTYH-associated polyposis (MAP), Peutz-Jeghers syndrome (PJS), juvenile polyposis syndrome (JPS), Cowden/PTEN hamartoma tumor syndrome, and serrated polyposis syndrome (SPS) [8,9,10].

For the practicing gastroenterologist, characterization and management of these patients are well-described in National Comprehensive Cancer Network (NCCN) guidelines and will not be reviewed here in depth [11••]. Management recommendations, such as early endoscopic surveillance, prophylactic risk-reducing surgery, and chemoprevention have been proven to decrease mortality and prevent cancer in these patients.

Universal tumor testing for lynch syndrome

Lynch syndrome (previously called hereditary nonpolyposis colorectal cancer [HNPCC]) represents the most common hereditary CRC syndrome. It is inherited in an autosomal dominant fashion and accounts for 3% of all CRC cases with a lifetime risk of CRC between 20 and 80% depending on the gene mutation [12]. The clinical criteria for identifying Lynch syndrome, including Amsterdam II criteria and revised Bethesda Guidelines, are cumbersome and lead to lower genetic screening rates in these patients [13]. Furthermore, these clinical criteria fail to recognize between 30 and 50% of Lynch syndrome patients [14, 15]. To improve detection of Lynch syndrome, universal CRC tumor screening for a mismatch repair deficiency (MMR-d) is now recommended. MMR-d is present in approximately 90% of Lynch syndrome-associated CRC tumors [16]. If screening via immunohistochemical (IHC) and/or microsatellite instability (MSI) testing raises concern for Lynch syndrome, referral for genetic counseling and testing is recommended [17]. IHC testing for the four mismatch repair (MMR) proteins (MLH1, MSH2, MSH6, and PMS2) can be carried out in a pathology laboratory whereas DNA testing for MSI must be done in a molecular laboratory. Since results of both screening tests correlate highly with each other, IHC testing for MMR proteins has become more prevalent within universal screening programs. In order to reduce the number of patients with sporadic CRC tumors that must be referred for genetic counseling and testing, additional testing of tumors with absent MLH1 expression is required [8]. The presence of a somatic BRAF mutation in the colon tumor and/or MLH1 hypermethylation significantly lowers the likelihood of Lynch syndrome as these findings typically represent sporadic mechanisms for absent MLH1 expression. As a result, this reflex testing has been incorporated by many universal tumor testing programs.

Utilizing universal CRC tumor screening, 10% of all CRC patients will require further genetic testing for Lynch syndrome. This has been consistently demonstrated across different ethnic groups [18]. Compared to other screening strategies, universal CRC tumor testing has been shown to be an effective screening tool for Lynch syndrome, although it is not the most cost-effective [19]. Guidelines recommend universal tumor testing as well as the use of clinical criteria, but many clinical settings still lack a standardized approach.

Penn State Health embarked on a public health initiative to increase the identification of Lynch syndrome in our patient population. Beginning in May of 2014, all biopsy/surgical specimens containing invasive colorectal adenocarcinoma were screened for Lynch syndrome using IHC staining for the 4 MMR proteins. Patients whose specimens demonstrated absent protein expression for 1 or more of the MMR proteins were contacted by a member of the Cancer Genetics Program and offered an appointment for genetic counseling and testing. As of October 2018, 544 colorectal cancer specimens were screened for Lynch syndrome. Of the 51 (9.4%) specimens that demonstrated abnormal protein expression, 35 (68.6%) patients met with a genetic counselor/geneticist and elected to pursue genetic testing. Of those 35 patients who pursued testing, 9 (25.7%) patients were confirmed to have Lynch syndrome (4 MLH1+, 3 MSH2+, 1 MSH6+, and 1 PMS2+).

A recent study found that up-front tumor sequencing was simpler and had superior sensitivity than current approaches to Lynch syndrome screening, while simultaneously providing critical information for treatment selection [20]. In this study, the tumor sequencing identified KRAS, NRAS, or BRAF mutations that could affect therapy for stage IV CRC, thus helping to avoid another test. In addition, the tumor sequencing had the benefit of identifying patients with germline DPYD mutations which are associated with toxicity to fluorouracil chemotherapy, and thus could be useful for treatment selection.

Risk assessment screening tools

Identifying patients with hereditary CRC syndromes starts with a detailed personal and family history of cancer and premalignant GI conditions [11••]. Once this has been obtained, utilization of the NCCN guidelines can help clinicians assess the need for gene testing (Table 1).

Table 1 NCCN guidelines indication for genetic risk assessment of hereditary CRC syndromes [11]
Full size table

In today’s healthcare environment, it is impractical for gastroenterologists to obtain a comprehensive, 3-generation family history that a genetic counselor would obtain. ACG clinical guidelines recommend a more practical approach to obtaining a CRC and polyp history of first- and second-degree relatives including age of diagnosis(es) [8]. While most gastroenterologists routinely seek personal and family history to risk-stratify patients for changes in screening age or surveillance intervals, genetic cancer risk assessment is often missed. This can have important implications for the patient and family members [21]. This issue is compounded by open-access colonoscopy in which gastroenterologists meet patient just minutes prior to the procedure. The ability of clinicians to determine the cumulative number of neoplastic polyps that a patient has had over time is limited. In attempts to overcome these problems, several different risk assessment tools have been developed to help identify hereditary CRC syndromes.

Clinical prediction algorithms

Clinical prediction algorithms are one type of risk assessment tool. These utilize comprehensive online questionnaires that allow for any individual to calculate the probability that a person carries a Lynch syndrome-associated germline mutation. One example is PREMM5, which calculates an individual’s risk of carrying a mutation in any of the known 4 MMR genes responsible for Lynch syndrome (MLH1, MSH2, MSH6, and PMS2) and/or EPCAM which is upstream of MSH2. PREMM5 incorporates age, gender, and personal and family history of cancer into a clinical prediction algorithm. A recent study found that the previously defined ≥ 5% threshold of these calculators may be inferior to the ≥ 2.5% threshold in the PREMM5 model. This lower threshold increases the number of identified mutation carriers while preserving a high negative predictive value of 99% [22•]. Nonetheless, NCCN guidelines recommend clinical judgment when determining threshold [11••]. Simulation models showed the PREMM5 screening strategy to be cost-effective and decreased the incidence of CRC by 43.9% in patients 25–35 years of age [23]. An earlier version of this prediction model (PREMM1,2,6) was assessed in the community setting utilizing a self-administered electronic tablet version. In addition to detecting 6 new cases of Lynch syndrome in over 3000 patients screened during a 6-month time period, there was high endoscopist and patient satisfaction [24]. This also allows for real-time feedback and incorporation into the patient’s medical record.

Screening questionnaires

Screening questionnaires are a practical type of risk assessment tool. Kastrinos et al. developed and validated a popular 3-question survey that can be incorporated into the pre-procedural evaluation for patients undergoing colonoscopy [25].

Kastrinos 3-question CRC risk assessment tool

  • Do you have a first-degree relative (mother, father, brother, sister, child) with any of the following conditions diagnosed before age 50:

  • Colon or rectal cancer?

  • Cancer of the uterus, ovary, stomach, small intestine, urinary tract (kidney, ureter, bladder), bile ducts, pancreas, or brain?

  • Do you have 3 or more relatives (this includes parents, brothers, sisters, children, grandparents, aunts, uncles, cousins) with a history of colon or rectal cancer?

  • Have you had any of the following conditions diagnosed before age 50:

  • Colon or rectal cancer?

  • Colon or rectal polyps?

Using a statistical analysis method called recursive partitioning analysis, 9 clinical risk factors were ranked based on ability to capture the most high-risk individuals when asked sequentially. This ultimately resulted in the 3-question survey above. The sensitivity was found to be 77%, which was also validated prospectively. In individuals with known Lynch syndrome and MMR gene defects, the survey was able to identify 95% of high-risk patients. A criticism of Kastrinos’s 3-question survey has been its lack of specificity with 20% of patients deemed high-risk and thus potentially warranting cancer genetics referral. A recent survey proposed by Guivatchian et al. included Kastrinos’s 3 question tool plus 1 question incorporating Lynch syndrome-associated cancers and 1 question on the lifetime cumulative polyp number [26••]. The Guivatchian questionnaire found 10% of patients to be high-risk, and of these patients, 10% were found to have heritable genetic mutations. Another study using mailed questionnaires on family history, sent to patients prior to outpatient colonoscopies, increased the number of patients referred to genetic counseling compared to standard of care (3.7% vs. 1.6%) [27]. In-office questionnaires can be completed by nurses with high patient satisfaction [28]. While feasible, the real-world efficacy of identifying germline mutations through questionnaires has been variable with one study identifying 2 germline mutations out of 6031 screened (3.3/10,000) [29].

A similar expanded 5-question version of the Kastrinos risk assessment tool was created and studied at Penn State Health. This expanded questionnaire incorporated 2 additional Lynch syndrome-associated cancer questions [30]. Patients 40 years of age and older in 10 different outpatient primary care sites were surveyed, and 2438 surveys were completed (23% response rate). Of these, 15.7% self-identified as high-risk. Only 31% of high-risk patients ages 40–49 were up-to-date on CRC screening compared to over 80% of average-risk and high-risk adults ≥ 50 years of age. In a subsequent study of Penn State Health employees over 40 years of age using a similar electronic questionnaire, 33.4% (878 individuals) self-identified as high-risk. In sub-group analysis, only 45.8% of high-risk individuals ages 40–49 reported up-to-date CRC screening [31]. Taken together, these two studies document a disparity in screening high-risk younger individuals for hereditary CRC syndromes.

There has also been the development of consumer-oriented CRC risk assessment tools to help patients collect their family history, such as the Office of the Surgeon General’s “My Family Health Portrait” (MFHP) [32]. MFHP has been externally validated with reasonable sensitivity and specificity compared to genetic counselor pedigree review although it is not used by most healthcare systems [33].

Limitations and barriers

There are several limitations and barriers to these various screening strategies. With all of these screening strategies, a great number of patients do not present for genetic counseling, decline testing, or do not follow through with testing [26••, 29, 34]. A limitation specific to PREMM5 is that it was not designed for screening for hereditary CRC syndromes other than Lynch syndrome. Furthermore, several of the screening strategies listed above are designed to be used in endoscopy centers where a patient may not present until 50 years of age. Other barriers to implementation of screening include physician motivation, time constraint, and knowledge. Prior to genetic testing, informed consent should include a discussion on the significance of a “positive” result including its potential psychological implications, the potential for genetic discrimination, and cost [8, 34, 35]. An increase in testing of a proband’s relatives has been noted with many genetics labs offering free testing within 90 days of the index patient’s pathogenic variant finding. While for Lynch syndrome, this cascade testing is recommended for all first-degree relatives 18 years of age or older; this frequently does not occur [36, 37]. There remains no single best way to screen for high-risk individuals for hereditary CRC syndromes, but at a minimum, gastroenterologists should strive to implement syndrome-specific NCCN guideline recommendations (Table 1) [19].

Once high-risk individuals are identified, it is recommended to refer these patients for cancer genetic counseling and testing. In-house referrals are optimal if genetic counselors are on staff, but alternative options exist. In today’s market of web-based services, a genetic counseling professional can be identified by visiting the National Society of Genetic Counselor’s website (www.nsgc.org), and searching for the nearest cancer genetics counselor [38]. Some academic centers have established comprehensive cancer genetics programs that have affiliations with other hospitals and care centers. This allows for broadened involvement in educational and patient care opportunities via videoconference and telemedicine, respectively [4]. If these options are not available, there is a growing number of telephone and web-based genetic counseling services. For clinicians looking for up-to-date and reliable information, a free online resource, GeneReviews®, is also available.

Early onset colorectal cancer

Recent studies looking at the prevalence of germline mutations in patients with CRC diagnosed less than 50 years of age using NGS panels further support the referral of all such cases for genetic counseling and testing. In one study looking at 450 patients with CRC diagnosed under 50 years of age, 72 (16%) were found to have an underlying germline mutation [39]. Overall, 8.4% of the patients were found to have Lynch syndrome while 8% were found to have other hereditary cancer syndromes involving mutations in high-penetrance CRC genes (5 APC; 1 APC/PMS2; 2 biallelic MUTYH; 1 SMAD4), low-penetrance CRC genes (3 APC c.3920 T > A, p.I1307K; 7 monoallelic MUTYH), and high- or moderate-penetrance genes not traditionally associated with CRC (3 ATM; 1 ATM/CHEK2; 2 BRCA1; 4 BRCA2; 1 CDKN2A; 2 PALB2). Importantly, 24 of 72 mutation positive patients (33.3%) did not meet established genetic testing criteria for the gene(s) in which they had a mutation.

Novel CRC predisposition genes

Multi-gene panel testing of cancer predisposition genes through a multitude of commercially available options has allowed for an increased number of genes to be analyzed at a markedly decreased cost. This can be as inexpensive as $250 for 84 genes with a turn-around time of 1–2 weeks [40]. Over the last several years, there has been the discovery of several new genes that may predispose individuals to CRC (Table 2).

Table 2 Novel colon cancer predisposition genes [41,42,43]
Full size table

These novel genes add to the expanding spectrum and overlap of CRC syndromes (Fig. 1). They have been characterized similarly to other well-described highly penetrant hereditary CRC syndromes based on inheritance patterns, clinical and histopathologic findings, and cellular mechanisms. For example, the polyp phenotype of polymerase proofreading-associated polyposis (PPAP) is similar to that of attenuated FAP. POLE is one of the gene mutations associated with PPAP. Its presence in sporadic tumors has been associated with a favorable prognosis, similar to microsatellite instability (MSI) highlighting the overlap of these genes [44]. Another novel syndrome, hereditary mixed polyposis syndrome (HMPS) associated with GREM1 gene mutation, presents with varied polyp histology and adds to the spectrum of hereditary CRC syndromes [45]. NTHL1- and MUTYH-associated polyposis (NAP and MAP, respectively) are both recessive polyposis syndromes which predispose to other cancer risks as well. The spectrum of benign and malignant tumors, though, in individuals with biallelic NTHL1 mutations appears to be broader [46].

Fig. 1
figure 1

Phenotypic classification of nonpolyposis and polyposis CRC syndromes, mode of inheritance, causal genes, and affected molecular pathways. Note: germline AXIN2 autosomal dominant mutations (Wnt pathway) may cause oligodontia-colorectal cancer syndrome characterized by severe permanent tooth agenesis and the presence of CRC or precancerous colonic or gastric lesions of variable types (adenomas, hyperplastic polyps). Due to the still undefined CRC and polyposis phenotype, it has not been included in the figure. BER, base excision repair; CMMRD, constitutional mismatch repair deficiency; HMPS, hereditary mixed polyposis syndrome; MAP, MUTYH-associated polyposis; MMR, DNA mismatch repair; PPAP, polymerase proofreading-associated polyposis; SPS, serrated polyposis syndrome. Used with permission from John C. Wiley and Sons [41].

Full size image

The use of NGS technology has resulted in the discovery of novel cancer genes which are now being associated with previously described clinical diagnoses. The diagnosis of serrated polyposis syndrome (SPS) remains dependent on World Health Organization (WHO) clinical criteria, but NGS technology has identified an associated gene, RNF43, albeit at a low mutation frequency [47]. The diagnosis of Familial Colorectal Cancer Type X (FCCTX) requires that families meet Amsterdam criteria, have mismatch repair-proficient (MMR-p) tumors, and lack germline mutations [41]. These families carry a twofold increase in CRC and lack the extracolonic tumors seen in Lynch syndrome. More recently, limited data has shown an association with FCCTX and the RPS20 gene [42•].

Many of these newly described genes are only documented in a limited number of individuals and/or families, which make clinical correlations and optimal management strategies challenging. One recent study concluded that patients with either CHEK2 or APC p.I1307K should start screening for CRC at age 45 and screening for those with monoallelic MUTYH should begin at age 50 [43]. The numerous other hereditary cancer genes not traditionally associated with CRC continue to be investigated with regard to their possible role in CRC. These include ATM, BRCA1, BRCA2, CDKN2A, PALB2, and TP53. The absolute risk of CRC associated with these genes is debated. It is unclear if the data is skewed by background population mutations or pleiotropism (gene manifests itself in a variety of clinical phenotypes) [48, 49]. On the other hand, various gene mutations previously believed to predispose to CRC, such as FANCM, FAN1, BUB1, BUB3, LRP6, and PTPN12, have recently been shown to lack this association [42•]. Other novel candidate genes that appear to play a role as moderate- or low-risk genes have been described including AXIN2, GALNT12, and biallelic MSH3 (Table 2) [50,51,52]. One problem with NGS panels is the absence in consistency of included genes, which points to the current lack of consensus inclusion criteria. In one study of 10 different commercially available NGS panels, only 6 of the well-characterized genes (APC, MLH1, MSH2, MSH6, PMS2, and MUTYH) were consistently included in all 10 panels [53].

Variants of uncertain significance

Results of gene testing return as one of three general possibilities: pathogenic variant, variant of uncertain significance (VUS), or benign variant. VUSs ultimately result in proteins with unclear function (absent, identical, increased, decreased, or altered). With this, cancer risk is unclear, and therefore, VUS carrier results lie on a spectrum from benign to pathogenic. Approximately 20–30% of patients undergoing CRC gene testing will receive a VUS result [54, 55]. Given that a VUS cannot be used to stratify patients into high or low risk, the result cannot be used to guide the medical management of patients and/or their families [38, 55]. If a VUS is found, patients are counseled to follow guidelines based on personal and family history. Unfortunately, though, many physicians misinterpret VUSs as pathogenic variants [56].

New developments

When multi-gene panel testing does not identify a CRC-predisposing mutation, CRC screening is based on personal and family history alone. Through the technology of NGS, genome-wide studies have been able to find single nucleotide polymorphisms (SNPs) that are associated with an increased risk of CRC. A recent study that utilized a 45 SNP panel found that individuals with two first-degree relatives with CRC who were also in the highest risk SNP quintile should undergo CRC screening 16 years earlier compared to average risk adults [57•]. The authors acknowledge the lack of evaluation of cost-effectiveness, resource feasibility, and insurance implications. Another study that used a 63 SNP panel and evaluated lifestyle and environmental factors determined the risk of CRC better than family history alone [58]. Although not yet mainstream, polygenetic risk scores, utilizing a validated set of SNPs, in combination with NGS panels including both high and moderate cancer risk susceptibility genes, will likely become increasingly utilized in clinical care to risk stratify and guide medical management.

The use of RNA analysis to re-categorize or clarify VUSs is also under investigation. Using this technique, an exonic duplication in the MSH2 gene was found to result in abnormal transcription leading to Lynch syndrome [59•]. RNA analysis is now being offered by some labs in conjunction with DNA analysis to better classify VUSs involving splice site junctions and duplications that may be in tandem, as well as to detect deep intronic mutations. A recent study using a methylation tolerance (MT) assay evaluated cellular response to cytotoxic effects of methylating agents to determine the effect on VUSs in MMR genes [60]. This novel technique may be used to reclassify VUSs found in up to 30% of patients.

Lastly, recent studies have shown that not only Lynch-associated tumors, but tumors with high microsatellite instability (MSI-h)/mismatch repair deficiency (MMR-d) are particularly susceptible to immune-based therapies (i.e., pembrolizumab) with patients showing durable responses in treatment-refractory advanced metastatic disease. This finding is due to the accumulation of frameshift mutations at hotspot repeat sequences due to MSI and in turn leads to the development of immunogenic neopeptides/neoantigens which are recognized by CD8 + tumor infiltrating lymphocytes. Interestingly, patients with Lynch syndrome who are cancer-free have been found to harbor circulating cytotoxic T cells targeted against MSI-induced frameshift neoantigens which suggests the possibility that immune-based therapies (i.e., immune checkpoint inhibitors, vaccination) may be used to prevent cancer in patients with Lynch syndrome. Clinical trials, such as NCT03631641, are beginning to study the application of such novel concepts and give hope to all patients with a hereditary predisposition to cancer.

Conclusion

The widespread use of NGS technology has expanded the spectrum of hereditary CRC genes which can be analyzed at significantly less cost and, as a result, has significantly increased access to testing. It is imperative that gastroenterologists utilize various screening strategies to identify high-risk individuals who may benefit from genetic counseling and testing. Future research should focus on optimizing practical strategies for risk assessment, further characterizing CRC predisposition genes, clarifying VUSs, and standardizing management strategies for these patients.