Mutational spectrum of the APC and MUTYH genes and genotype–phenotype correlations in Brazilian FAP, AFAP, and MAP patients

Patients with multiple colorectal adenomas are screened for germline mutations in two distinct genes, APC and MUTYH. According to the polyp number and age of onset, the phenotype of APC-mutated patients can be classified as classical familial adenomatous polyposis (FAP: more than 100 polyps, early onset) or attenuated FAP (AFAP: fewer than 100 polyps with later onset)[1‐3]. MUTYH biallelic mutation carriers usually present 10 to 100 polyps and are categorized as having MUTYH-associated polyposis (MAP)[4].

FAP/AFAP (OMIM #175100) is a dominantly inherited colorectal cancer (CRC) predisposing syndrome[1, 2] caused by mutations in the tumor suppressor gene adenomatous polyposis coli (APC). The encoded APC protein controls β-catenin turnover in the Wnt pathway[5, 6]. Besides colonic polyposis and colorectal cancer, individuals with FAP can present a number of benign extracolonic features, including multiple osteomas, epidermoid cysts, desmoid tumors, and congenital hypertrophy of the retinal pigment epithelium[2]. Over 1100 different pathogenic APC mutations have been reported to date in the Leiden Open Variation Database (http://​www.​lovd.​nl/​2.​0/​), the majority of them being nonsense mutations or small insertions or deletions that lead to a truncated protein. Mutations causing AFAP have been reported to occur mainly in three regions of APC: at the 5^′ end (the first five exons), in the alternatively spliced region of exon 9, or at the 3^′ end (after codon 1580)[7‐9].

MAP (OMIM #608456) is a recessively inherited syndrome caused by biallelic mutations in the mutY homolog (MUTYH) gene that maps to chromosome 1p34.1[4]. MUTYH encodes a DNA glycosylase that plays a key role in the base excision repair pathway by removing mispaired bases caused by the oxidation product 8-oxoG[4]. Nearly 300 different sequence variants have been identified in this gene (LOVD Mutation Database), including about 80 pathogenic mutations distributed throughout the gene at positions corresponding to different functional domains of the encoded protein[10]. In contrast to APC pathogenic variants, which mostly result in a truncated or absent protein, most MUTYH pathogenic variants are missense substitutions and only a minority are splice site or truncating mutations[11].

With regard to clinical features, most MUTYH- mutated patients present 10 to 100 colorectal adenomas, usually with later onset compared with FAP patients[4, 12, 13]. MUTYH mutation carriers represent approximately 7.5% of patients with more than 100 adenomas without an APC mutation, 40% of all patients with 10–100 polyps, and 0.3–1.7% of patients with fewer than 10 polyps and early-onset CRC with no family history[12, 14‐16]. Furthermore, it has also been reported MAP patients having no polyps at the time of CRC diagnosis[17, 18].

Because of the observed overlap between the clinical phenotype of FAP/AFAP and MAP syndromes, the identification of the causative mutation has important implications for family management, allowing effective clinical surveillance and accurate genetic counseling. Moreover, the spectrum of mutations and the genotype–phenotype correlations may have clinical impact and can be population-specific. Therefore, it is important to obtain genetic and clinical data from FAP/AFAP and MAP families in different populations.

The aim of this study was to conduct a comprehensive molecular analysis to determine the spectrum of point mutations and large genomic rearrangements (LGR) in the APC and MUTYH genes in a series of 23 Brazilian polyposis patients. This paper summarizes the mutation screening data and outlines the most cost-effective approach to detect APC and MUTYH mutations in Brazilian polyposis patients. In addition, we discuss the genotype–phenotype associations found in these families in the context of previously described data from the literature.

Twenty-three Brazilian families with a clinical diagnosis of classical or attenuated polyposis were included in this study. The majority of the index cases (15) presented an intermediate or severe FAP phenotype (> 100 polyps) and 13 of them harbored an APC pathogenic mutation, while one patient was mutation-negative and one had a monoallelic MUTYH mutation. The remaining eight patients presented an attenuated polyposis burden (< 100 polyps), among whom five carried biallelic mutations in the MUTYH gene, one carried a novel APC duplication of exons 1–3, one presented a novel APC missense variant, and one was mutation-negative. Seven novel germline mutations (six pathogenic and one variant of unknown significance) were detected in this cohort, and two of them have been recently published by our group[21, 23]. The APC and MUTYH mutation spectrum, including information about previous reports of the detected mutations, is summarized in Table1.

This is the first report of a comprehensive mutational analysis and genotype–phenotype correlation in Brazilian polyposis families. Through direct sequencing of the APC and MUTYH genes, MLPA, aCGH, and duplex qPCR, we were able to identify pathogenic mutations in 20 of 23 index cases — a detection rate of 87%. Of the remaining three patients, two were mutation-negative and one harbored a novel APC missense variant (p.Val1789Leu). Because of the lack of affected relatives for co-segregation analysis and the inconclusive results given by in silico analysis and control population screening, the clinical significance of this alteration is yet to be determined. Interestingly, a parallel study from our institution, performed in high risk cancer patients, revealed that this patient also presents a rare germline microdeletion of the PIP gene possibly associated with an increased cancer risk (Silva 2013, unpublished observations), suggesting that these two alterations may be acting in synergy.

The detection rate in polyposis patients from other populations varies markedly, ranging from 39 to 90%; the variation reflects different selection criteria for testing and diverse sensitivity of screening strategies[4, 27, 37‐40]. Our data reinforce the need to apply a combination of mutation-screening methods to detect the disease-causing mutation in polyposis patients efficiently. Most APC mutations previously described were identified with conventional methods, for instance denaturing high performance liquid chromatography or the protein truncation test, which can have a relatively low detection rate. Nowadays, the gold standard detection method for polyposis patients is direct DNA sequencing of all APC and MUTYH coding exons (including intron–exon boundaries), accompanied by screening for LGR, as was performed here.

The majority of mutations identified in our cohort were distinct, except for two families who shared the codon 1309 hotspot APC mutation and three families who presented the p.Tyr179Cys hotspot mutation in one of the MUTYH alleles. The absence of the commonly reported APC mutation at codon 1061, and the relatively low frequency of the hotspot mutations APC 1309, MUTYH p.Tyr179Cys, and p.Gly396Asp are consistent with the fact that Brazilian patients represent an admixed population, probably lacking a founder APC or MUTYH mutation.

The p.Tyr179Cys and p.Gly396Asp MUTYH mutations were identified in three families and one family, respectively, and corresponded to 44% of all mutated alleles identified in this gene (4/9). These are also the most prevalent mutations in populations of European origin, probably because of a founder effect, and account for approximately 80% of all reported mutant alleles[12]. A recent report described a screen for these two variants in 30 Brazilian patients with clinical phenotypes of MAP and FAP; 5/30 patients were identified as carrying one of these two hotspot mutations, and four of them were in a biallelic state[41]. However, because the entire coding sequence of MUTYH was not evaluated in all patients in this study, we cannot perform comparisons with the frequency found in our patients.

In our series, we could not identify a causative mutation in two index cases—one with attenuated polyposis and one with an intermediate polyposis phenotype. A possible explanation for the polyposis phenotype in these patients is the presence of unusual mutations in the APC or MUTYH genes, such as intron or promoter point mutations, epimutations or genetic mosaicism. In this sense, a recent study demonstrated that up to 8% of APC/MUTYH-negative polyposis patients presented a deep intronic APC variant that led to an aberrant transcript[42]. A second possibility is the existence of other susceptibility genes, and with the current possibility of screening all coding genes by next generation exome sequencing, it can be anticipated that novel polyposis-predisposing genes will be identified.

Considering that this study is the first comprehensive analysis of APC and MUTYH mutations in Brazilian polyposis patients, we attempted to determine the most cost-effective approach to detect the causative mutation in this population. In patients presenting fewer than 100 polyps (N = 8), 62% carried a biallelic mutation in the MUTYH gene. Among patients with more than 100 polyps (15 cases), three cases presented a mutation in APC exon 8 and eight cases (53%) exhibited a mutation between codons 1017 and 1650 of APC exon 15. Interestingly, this initial region of exon 15 comprises only 16% of the coding sequence of APC, yet presented a high mutation rate in our cohort. Therefore, based on our results, an optimized scheme for the molecular diagnosis of APC and MUTYH mutations in the Brazilian population might be obtained as follows: for patients presenting > 100 polyps, codons 1017–1650 of APC exon 15 should be sequenced first, followed by exon 8, and then the remaining APC exons; for patients presenting fewer than 100 polyps, MUTYH should be firstly screened. Because none of the studied MUTYH patients presented only the hotspot mutations p.Tyr179Cys or p.Gly396Asp, the whole gene should be sequenced, instead of undertaking an initial search for these variants, as recommended for other populations[18].

Genotype–phenotype correlations in polyposis syndromes are of great clinical interest, because they can contribute to better genetic counseling and simplify mutation screening. In several studies, an association between the location of the mutation and the clinical manifestations has been observed[7, 20, 27, 34‐36]. However, since several contradictions have also been reported[22], it remains unclear whether the genetic information should guide clinical decision-making, such as the extent of the prophylactic colectomy or the protocol for clinical surveillance[7, 43‐47].

Regarding the age at which clinical surveillance should begin, the established guidelines suggest that classical FAP patients should start endoscopic surveillance from the early teens, while AFAP and MAP families could start surveillance at age 18–20[46]. Similar to the literature, our results demonstrated that severe FAP patients had an earlier age of onset (on average 10 years younger than AFAP or MAP patients). However, the most premature case in our series was a 7-year-old patient from a family with an APC mutation at codon 1017 — a region usually associated with an intermediate phenotype. A particularly aggressive phenotypic expression of this mutation was observed in this family; several relatives presented a high number of polyps (> 1000), an early age of onset, and desmoid tumors. This case demonstrates the importance of considering the family clinical history when planning the surveillance of other family members.

One of the most clinically important discrepancies observed in our study concerns desmoid tumors, which, even if histologically benign, can lead to life-threatening complications through their size and impingement on vital structures. Indeed, desmoid tumors represent the second leading cause of death in FAP patients[46, 48] and were identified at high frequency in our cohort, occurring in 57% (8/14) of APC-mutated families. Although described as usually associated with mutations after codon 1444[22], only two out of eight families affected by desmoid tumors in our study harbored mutations after this codon. In concordance with our findings, previous studies with large cohorts have also failed to confirm this association[49], or described different boundaries for the increased risk of these tumors, such as codon 1310[50] or 1395[35]. Furthermore, besides the location of APC mutations, several other risk factors are suggested to be related with desmoid tumor development, such as surgical trauma[51], pregnancy[52] and especially positive family history for desmoid tumors[49]. In this regard, the last appears to be the most important risk factor for our population, since five of eight families with desmoid tumors presented more than one relative affected by this tumor.

The differences observed in certain phenotypic features between our series and those of others may be because of our relatively small number of FAP and MAP families, may reflect some selection or data collection bias, or may be related to phenotypic peculiarities in this specific population. In this sense, the majority of FAP genotype-phenotype studies were performed in Europeans cohorts[34‐40, 43, 47, 49] and the self-declared ethnic origin of most families from our study was also European (Portuguese, Italian and Spanish, mainly) - except from two Japanese and one Arabian families. However, it is important to highlight that Brazilians represent an extremely admixed population, with most individuals presenting some degree of African and Amerindian ancestry[53].

Furthermore, intra and interfamilial variations in the FAP phenotype are also well documented in other populations, and it is likely that modifying genes and environmental factors, as well as functional polymorphisms of the normal APC allele, play a crucial role in determining the clinical course of disease[54, 55]. In this regard, the different genetic background and/or environmental factors of our population could be responsible for the phenotypic differences observed in our study. For instance, in our series, desmoid tumors were much more prevalent in APC-mutated patients (57%) than in others previous studies (10-15%)[22, 49, 54], indicating that perhaps our set of patients represents a distinct group regarding this extracolonic feature.

Finally, the lack of a clear phenotypic expression of the mutations identified in our study complicates clinical predictions based on knowledge of the mutation site, and as a result, we can make no specific surveillance and management recommendations for the Brazilian population. In order to accomplish that, larger studies need to be carried out. Instead, we recommend that clinical decisions regarding an individual patient should not be based strictly on the genotype, but mainly on the colonic phenotype and family clinical history.

In this comprehensive investigation of the APC and MUTYH mutation spectrum in Brazilian polyposis patients, we identified a high frequency of germline mutations, allowing the identification of several novel pathogenic variants and the proposal of a cost-effective screening approach for this population. Notably, a significant number of APC mutation-positive families were not consistent with predicted genotype–phenotype correlations, and this should be taken into consideration for genetic counseling and patient management of our population.