Prevalence and Phenotypes of APC and MUTYH Mutations in Patients With Multiple Colorectal Adenomas FREE

The presence of multiple colorectal adenomas may be attributable to familial adenomatous polyposis (FAP), an autosomal dominant polyposis syndrome resulting from germline mutations in the adenomatous polyposis coli (APC) gene (NCBI Entrez Gene 324).¹ Individuals with APC mutations may present with “classic” polyposis (≥100 adenomas) and develop thousands of adenomas in the second or third decade of the disease. Approximately 10% of individuals with APC mutations may have milder disease, with 20 to 99 adenomas at an older age of onset. Multiple colorectal adenomas may also arise secondary to mutations in the mutY homolog (MUTYH) gene (NCBI Entrez Gene 4595).^2- 3 Individuals with MUTYH- associated polyposis (MAP) are at an increased risk of colorectal cancer that may develop in the presence of few polyps.⁴

Although it is established that the clinical presentation of FAP and MAP may overlap, 2 important issues warrant further study. First, the relative contribution of biallelic MUTYH mutations to APC mutations in individuals with multiple adenomas is unknown. Current estimates have been derived from highly selected clinic-based patients with multiple adenomas and no APC mutation.^5- 8 Studies evaluating the prevalence of both APC and MUTYH mutations in attenuated polyposis have been small, and their findings have not been validated.^9- 10 Second, guidelines for when genetic evaluation should be performed in individuals with multiple colorectal adenomas vary, and data to support such guidelines are limited.^11- 14

We evaluated the frequency of APC and MUTYH mutations by the number of colorectal adenomas among individuals who had undergone clinical genetic testing. We also studied the relationship between the number of adenomas and age at diagnosis of adenoma and colorectal cancer and the prevalence of pathogenic APC or MUTYH mutations to inform future guidelines for genetic testing in individuals with multiple adenomas.

This cross-sectional study was conducted among 8903 individuals whose clinicians (physicians, physician assistants, or nurse practitioners) submitted blood samples for genetic testing for APC and MUTYH mutations to a commercial laboratory (Myriad Genetic Laboratories Inc) between 2004 and 2011 as part of clinical care because of the patient's personal or family history of colorectal cancer, colorectal polyps, or both. Clinicians, genetic counselors, or other members of the clinical staff completed a prespecified test order form that included age at testing, ancestry (Western/Northern European, Central/East European, Ashkenazi, Latin American/Caribbean, African, Asian, Near East/Middle Eastern, Native American, other), cancer history (colorectal cancer, endometrial cancer, other), age at cancer diagnosis, age at colorectal adenoma diagnosis and adenoma count (1, 2-5, 6-9, 10-19, 20-99, 100-999, and ≥1000), and family history of cancer (relative, cancer site, age at diagnosis) and colorectal adenomas in first-, second- and third-degree relatives. From the original 8903 individuals, we excluded 227 individuals for whom personal as well as family histories were missing, for a final study cohort of 8676.

The study was investigator initiated and approved by the Dana-Farber Cancer Institute institutional review board.

Clinical genetic testing consisted of full gene sequencing and large rearrangement analysis of the APC gene. Full gene sequence determination was performed in the forward and reverse direction of approximately 8532 base pairs comprising 15 exons and 420 adjacent noncoding intronic base pairs. For large rearrangement analyses, all exons of APC were examined for evidence of deletions and duplications by standard Southern blot methods. All individuals also underwent DNA sequence analysis of specific portions of MUTYH exons 7 and 13 designed to detect the 2 most common MUTYH mutations (Y179C, G396D). Full MUTYH gene sequencing was performed if 1 of the 2 most common mutations was identified. Individuals with deleterious or suspected deleterious mutations were defined as mutation-positive. Suspected deleterious mutations included genetic variants for which the available evidence indicated likelihood, but not proof, that the mutation is deleterious. Genetic testing techniques did not change during the study period (2004-2011).

The primary outcome was the prevalence of pathogenic APC or pathogenic biallelic MUTYH mutations. Covariates of interest included the number and age at diagnoses of adenomas, the presence of and age at colorectal cancer diagnosis, and the presence of colorectal cancer in a first-degree relative. In individuals diagnosed with the same cancer more than once, the age at diagnosis was defined as the youngest age at diagnosis. Age was categorized a priori into 4 categories (<30, 30-39, 40-49, and ≥50 years). For individuals with adenomas identified more than once, a cumulative adenoma count was computed. Adenoma count was analyzed as an ordinal variable (<10, 10-19, 20-99, 100-999, and ≥1000 adenomas).

Bivariable analyses were used to assess the association between mutation status and covariates of interest. χ² tests were performed for categorical variables and t tests for continuous data. Results were reported as odds ratios with 95% confidence intervals. P < .05 (2-sided) was considered statistically significant.

Multiple imputation was used to obtain estimates of missing data for adenoma count (398/7225 [6%]), age at adenoma diagnosis (1912/7225 [26%]), and age at colorectal cancer diagnosis (67/2306 [3%]).¹⁵ The coefficients of 5 rounds of imputation (performed in R using the AregImpute function) were combined to obtain the final estimates for missing data. Multivariable logistic regression analysis was performed on the imputed data set to assess the independent associations of the presence of a pathogenic mutation (APC or biallelic MUTYH) and covariates of interest. Multinomial logistic regression analyses were used to examine the differences in phenotypic characteristics between individuals with a pathogenic APC mutation and biallelic MUTYH mutations and to derive the probability of these mutations based on clinical characteristics.

Statistical analyses were performed using SAS version 9.2 (SAS Institute Inc) and R version 2.11.0 (R Foundation for Statistical Computing).

Of the 8676 individuals included in the study, 4324 (50%) were male and 6323 (73%) were of European ancestry (Table 1). Of the included individuals, 1508 (17%) had a pathogenic APC mutation, 422 (5%) had biallelic pathogenic MUTYH mutations, 168 (2%) had a monoallelic pathogenic MUTYH mutation, and 6578 (76%) had a nonpathogenic APC or MUTYH alteration or no alteration in either gene.

Overall, 7225 individuals (83%) were reported to have a history of adenomas, with a median age of 47 years at adenoma diagnosis, and 517 (6%) were reported to have extraintestinal manifestations associated with a familial polyposis syndrome. Of the remaining 1451 individuals (17%) without a history of adenomas, 527 (36%) had a personal history of colorectal cancer and 184 (13%) had a history of either a cancer that was not colorectal cancer or an extraintestinal manifestation associated with familial polyposis. A personal history of colorectal cancer was reported in 2306 individuals (27%); of these, 1779 (77%) had a history of both colorectal cancer and adenomas. Approximately one-third of the study population reported having a first-degree relative with a history of colorectal cancer.

Of the 7225 individuals with a reported history of colorectal adenomas, 1457 (21%) had a classic polyposis phenotype (≥100 adenomas [1338 with 100-999 adenomas and 119 with ≥1000 adenomas]) and 3253 (45%) had an attenuated phenotype (20-99 adenomas) (Table 2).

Of the 119 individuals with 1000 or more adenomas, 95 (80% [95% CI, 71%-87%]) had a pathogenic APC mutation and 2 (2% [95% CI, 0.2%-6%]) had biallelic MUTYH mutations. In contrast, among 1338 individuals with 100 to 999 adenomas, 756 (56% [95% CI, 54%-59%]) had an APC mutation and 94 (7% [95% CI, 6%-8%]) had biallelic MUTYH mutations. The presence of a first-degree relative with colorectal cancer did not significantly influence APC or MUTYH mutation prevalence in individuals with 1000 or more adenomas.

Of the 3253 individuals with 20 to 99 polyps, 326 (10% [95% CI, 9%-11%]) had a pathogenic APC mutation and 233 (7% [95% CI, 6%-8%]) had biallelic MUTYH mutations. In these patients with an attenuated FAP phenotype, having a first-degree relative with colorectal cancer was associated with a higher APC mutation prevalence than if no such history existed (15% [95% CI, 13%-17%] and 8% [95% CI, 7%-9%], respectively).

Of the 970 individuals with 10 to 19 adenomas, APC and biallelic MUTYH mutations were present in 50 (5% [95% CI, 4%-7%]) and 37 (4% [95% CI, 3%-5%]), respectively. The majority of mutation carriers did not report a family history of colorectal cancer.

Overall, the prevalence of APC and MUTYH mutations varied with adenoma count, with APC mutation rate progressively increasing with increasing polyp burden and MUTYH mutation rates remaining relatively constant across different categories (Figure).

We performed bivariable and multivariable logistic regression analyses to evaluate the association of a pathogenic mutation in either gene with clinical characteristics (Table 3). In the multivariable logistic regression analysis, controlling for a family history of colorectal cancer in a first-degree relative, individuals with 10 to 19 adenomas were significantly more likely to have pathogenic APC mutations or biallelic MUTYH mutations than those with fewer than 10 adenomas (odds ratio [OR], 2.7 [95% CI, 1.9-3.7]). The odds of a mutation increased with adenoma count (20-99: OR, 6.4 [95% CI, 4.9-8.4]; 100-999: OR, 30.7 [95% CI, 23.4-40.3]; ≥1000: OR, 77.5 [95% CI, 45.3-132.4]). Colorectal adenomas prior to age 50 years were associated with an increased likelihood of pathogenic APC or biallelic MUTYH mutations, which increased progressively with earlier age at diagnosis (40-49 years: OR, 2.4 [95% CI, 2.0-2.8]; 30-39 years: OR, 4.2 [95% CI, 3.5-5.2]; <30 years: OR, 8.7 [95% CI, 7.1-10.6]).

To examine the differences between the phenotypic characteristics of individuals with a pathogenic APC mutation and biallelic MUTYH mutations, we performed multinomial logistic regression analysis (logistic regression for a categorical dependent variable with ≥2 categories [APC, biallelic MUTYH, nonpathogenic APC, or MUTYH alteration/no APC or MUTYH alteration/monoallelic MUTYH)] (Table 3). The odds of carrying a pathogenic APC mutation were significantly increased in patients with more than 10 adenomas (10-19: OR, 2.4 [95% CI, 1.6-3.6]; 20-99: OR, 6.0 [95% CI, 4.3-8.2]; 100-999: OR, 40.1 [95% CI, 29.2-55.1]; ≥1000: OR, 124.0 [95% CI, 69.7-220.7]). Age at adenoma diagnosis was also associated with an APC mutation (<30 years: OR, 15.4 [95% CI, 12.2-19.5]; 30-39 years: OR, 6.1 [95% CI, 4.8-7.8]; 40-49 years: OR, 2.7 [95% CI, 2.2-3.4]). Individuals with 10 to 19 adenomas were significantly more likely to have biallelic MUTYH mutations than no mutation or a monoallelic MUTYH mutation. The odds of biallelic MUTYH mutations increased with increasing number of adenomas (10-19: OR, 2.9 [95% CI, 1.7-5.1]; 20-99: OR, 6.6 [95% CI, 4.1-10.6]; 100-999: OR, 12.5 [95% CI, 7.6-20.6]).

The multinomial logistic regression model (eSupplement) was also used to derive the predicted probability of pathogenic APC and MUTYH mutations based on phenotypic characteristics and family history of colorectal cancer. The C statistic was 0.81 (95% CI, 0.73-0.89) for APC and 0.59 (95% CI, 0.49-0.68) for MUTYH when the model included the number of adenomas alone, 0.88 (95% CI, 0.82-0.95) for APC and 0.59 (95% CI, 0.49-0.69) for MUTYH when the model included the number and age at adenoma diagnoses, 0.89 (95% CI, 0.82-0.95) for APC and 0.65 (95% CI, 0.55-0.74) for MUTYH when the presence of colorectal cancer and age at colorectal cancer diagnosis were added to the model, and 0.89 (95% CI, 0.82-0.95) for APC and 0.66 (95% CI, 0.56-0.75) for MUTYH when the presence of a first-degree relative with colorectal cancer was also included in the model.

To illustrate how the prediction probabilities derived from these models may be used in a clinical setting and the differences in APC and MUTYH mutation probability based on clinical characteristics, 20 clinical scenarios with their respective predicted mutation probabilities are reported in Table 4. For example, for an individual with multiple adenomas diagnosed at age 20 years and no history of colorectal cancer in a first-degree relative, the probabilities of APC and biallelic MUTYH mutations range from 97% (95% CI, 93.4%-100.0%) for APC and 0.5% (95% CI, 0.0%-1.9%) for MUTYH with 1000 or more adenomas to 89% (95% CI, 83.0%-95.2%) for APC and 3% (95% CI, 0.0%-6.9%) for MUTYH with 100 to 999 adenomas, to 59% (95% CI, 49.3%-68.6%) for APC and 8% (95% CI, 2.7%-13.4%) for MUTYH with 20 to 99 adenomas, and 38% (95% CI, 28.6%-47.7%) for APC and 6% (95% CI, 1.4%-10.7%) for MUTYH with 10 to 19 adenomas.

We evaluated the relative frequencies of mutations in the APC and MUTYH genes in a large number of individuals who had undergone genetic testing. Our results help further inform the understanding of the genetic epidemiology of the classic hereditary colorectal cancer syndrome FAP and shed some light on the important differences in disease patterns between carriers of APC mutations vs those with biallelic MUTYH mutations.

The clinical syndrome of FAP was first reported in 1847. In 1975, Bussey¹⁶ described the clinical characteristics of patients with hundreds to thousands of colorectal polyps. In 1991, the APC gene was cloned and found to be mutated in patients with FAP.^1,17- 18MUTYH -associated polyposis was described in 2002 when Al-Tassan et al² noted biallelic germline mutations in the base excision repair gene MUTYH in a family with recessive inheritance of multiple colorectal adenomas and colorectal cancer.

Previous studies (predating the discovery of MAP) have reported widely varying prevalence of pathogenic APC mutations among individuals with a classic polyposis phenotype (52%-82%), likely attributable to varying mutation analysis techniques and patient selection.^19- 24 However, these studies primarily involved small cohorts that were geographically and ethnically homogeneous. After the discovery of MUTYH, APC mutation–negative probands with classic FAP were screened for MUTYH mutations. These relatively small studies reported MUTYH mutation prevalence rates ranging from 7.5% to 20% in individuals with classic polyposis.^5,7

The results of our study, in which all individuals were tested for both APC and MUTYH mutations, indicate significant heterogeneity in mutation prevalence, even among individuals with a classic polyposis phenotype. Among individuals with 1000 or more adenomas, 80% (95% CI, 71%-87%) had a pathogenic APC mutation, and MUTYH played a minor role (2% [95% CI, 0.2%-6%]). The distribution and prevalence of mutations was markedly different, however, in individuals with 100 to 999 adenomas (still considered classic polyposis)—only 56% (95% CI, 54%-59%) were APC carriers, and a higher proportion (7% [95% CI, 6%-8%]) had biallelic MUTYH mutations. No pathogenic APC or MUTYH mutations were detected in 18% (95% CI, 12%-27%) of individuals with 1000 or more adenomas and in 35% (95% CI, 33%-38%) with 100 to 999 adenomas, a finding potentially attributable in part to genes that have not been identified.

In contrast, in the 3253 individuals with attenuated polyposis, prevalence rates of pathogenic APC and MUTYH mutations were similar (10% [95% CI, 9%-11%] and 7% [95% CI, 6%-8%], respectively). This MUTYH prevalence rate is lower than those reported in prior reports from smaller cohorts of patients with attenuated polyposis, in which estimates have ranged from 22% to 29%.^{5- 8,10,25- 28}

We did not evaluate the genotype-phenotype correlation among individuals with APC mutations as has been previously reported, because this study aimed to highlight the clinical characteristics associated with a pathogenic mutation in either of the 2 familial polyposis genes (APC or MUTYH) and the differences in these characteristics between mutation carriers. Ten or more adenomas and young-onset (<50 years) adenomas were associated with a mutation in either gene (APC or MUTYH). There was an incremental increase in the odds of a mutation, with an increasing number of adenomas and earlier age at adenoma diagnosis. Individuals with 10 or more adenomas and young-onset adenomas were significantly more likely to have an APC mutation. The presence of 10 or more adenomas was associated with a pathogenic MUTYH mutation, but in contrast to individuals with an APC mutation, the odds of a mutation did not incrementally increase with earlier age at diagnosis and were highest at ages between 30 and 49 years.

The study population is both a weakness and strength. This was not a population-based study, and participants had undergone testing based on a personal or family history suggestive of a polyposis syndrome by clinicians (physicians, physician assistants, nurse practitioners) who may have had variable expertise in genetic evaluation; therefore, the prevalence estimates, particularly in the groups with fewer numbers of individuals, must be interpreted with caution because of potential ascertainment and referral bias.²⁹ Nonetheless, this cohort is representative of individuals for whom genetic testing for APC and MUTYH genes should be considered and reflects the characteristics of the population at risk. We did not verify the pathology of polyps or the clinical data provided on the test order form. Although data were provided by clinicians whose specific specialty or training was not reported on the form, other studies using similar methods of data collection for cohorts tested for familial colorectal cancer syndromes have been externally validated, suggesting that the data are likely to be accurate and are likely not to vary between the groups being compared.^30- 31 We also used multiple imputation techniques for missing data so as to minimize selection bias, which has been demonstrated to be particularly important in genetic association studies, in which missing data may be distributed differentially and may generate spurious associations. However, results obtained from using both complete case data and imputed data were similar.

The test order form did not elicit a history of hyperplastic polyps, which have been reported in small cohorts with MAP.³² However, only a small percentage of patients with MAP present with hyperplastic polyposis, and adenomatous polyps and colorectal cancer remain the most common clinical presentation. Targeted sequence analysis was performed to detect the 2 most common MUTYH mutations (Y179C and G396D), and full MUTYH gene sequencing was performed in a small percentage of individuals. Full MUTYH sequencing may have led to an increase in some prevalence estimates; however, it is known that Y179C and G396D mutations account for the majority of mutant alleles in individuals of North American and European ancestry, who comprised the majority of our study participants.^7,33- 35 The use of MUTYH gene rearrangement analysis and allele-specific APC analysis, which have recently been reported but are not widely available commercially, may also result in a small improvement in the yield of testing.³⁶

Through evaluation of the phenotypic differences between mutation carriers in this large study, a pattern has emerged. Overall, in individuals with multiple adenomas, the APC mutation rate progressively increases with increasing polyp burden, whereas the MUTYH mutation rate remains relatively constant across different categories. Furthermore, the prevalence of APC mutations varies significantly among individuals with classic polyposis (≥1000 adenomas: 80% [95% CI, 71%-87%]; 100-999 adenomas: 56% [95% CI, 54%-59%]). In contrast, biallelic MUTYH mutations are rare in individuals with 1000 or more adenomas, and their prevalence is relatively constant among individuals with fewer than 1000 adenomas.

Our evaluation of individuals who underwent genetic testing because of a personal or family history suggestive of a familial polyposis syndrome suggests that genetic evaluation for APC and MUTYH mutations may be considered in individuals with 10 or more adenomas. However, our results are derived from a selected cohort of high-risk individuals and need to be validated in larger populations of unselected patients.

The mutation probabilities reported here may assist clinicians in their decision to recommend genetic evaluation and counsel patients undergoing genetic testing. However, it remains important to also consider the limitations of genetic testing at present, because one-third of patients with a classic FAP phenotype are found to not carry a mutation in either the APC or MUTYH gene. Such individuals should undergo periodic reevaluation as other susceptibility genes are identified.