MUTYH-associated polyposis (MAP)

3.1. Mutation spectrum 

To date, 105 unique sequence variants have been reported in the LOVD database for the MUTYH gene. These are predominantly missense mutations, but to a lesser extent also small deletions, duplications and one insertion. For all reported mutations see: (www.lovd.nl/mutyh). In the LOVD, mutations are annotated according to the longest possible (hypothetical) coding sequence, NM_001128425.1, as recommended by the Human Genome Variation Society (HGVS) in collaboration with The National Center for Biotechnology Information (NCBI). This sequence differs from the most frequently reported transcript NM_001048171.1 by insertion of 42 nucleotides after c.157 (www.lovd.nl/mutyh). For example: the p.Y165C annotation becomes p.Y179C and p.G382D becomes p.G396D.

In Caucasian populations, a biallelic status for the hot spot mutations p.Y179C and/or p.G396D is reported in up to 70% of MAP patients. Furthermore, 90% of the western MAP population carries at least one of these mutations [19]. Since these mutations were never found in Korean, Japanese or Jewish persons of European origin [20], [21], [22], [23], [24], existence of founder mutations and ethnic differentiation is assumed for MUTYH mutations. Other common mutations are c.1147delC (p.A385fs mutation) in western countries [19], p.P405L in Dutch [25], c.1437_1439del in Italian (p.Glu480del) [26], p.E480X in British Indian, p.Y104X in Pakistani [27], c.1227_1228dup in Spanish and Portuguese [28] and p.A359V in Japanese and Korean patients [20], [22], [29].

3.2. Functionality assays 

Functionality assays have been performed for a number of MUTYH variants. Studies on other species (like E. coli) show that the equivalent location for the human p.Y179 in the HhH region plays an important role in 8-oxoG recognition and adenine mispair specificity, intercalation into the DNA duplex and stability of the protein-DNA complex. The equivalent location for the human p.G396 in the C-terminal domain is also involved in 8-oxoG recognition and adenine mispair specificity, and is thought to be important for conformational flexibility of the MUTYH protein [11], [12], [13], [14], [30]. Results of functionality assays on these variants are not consistent, depending on the setting some studies found a complete absence of activity for both variants [31] while others found only partially reduced activity [32], [33]. Furthermore the p.G396D variant is significantly less catalytically compromised than the p.Y179C in some studies [3], [33], [34]. It has been suggested that the effective glycosylase activity of MUTYH is more reduced in vivo because the search process is more demanding, as the size of the chromosomal DNA is larger than the oligonucleotide substrates used in vitro and there is more competition for binding with other DNA binding proteins [32], [35], [36]. Differences in outcomes might also be explained by lack of consideration and correction of the active enzyme fraction in an in vitro setting. Recently, it was shown that both variants (and other variants) – when corrected for active enzyme fraction – showed the same threefold reduction in glycosylase activity compared to the wild-type protein in a vitro experiment [37]. In contrast, in a cellular context using rifampicine resistance assays, different results for the assessed variants were found. While the ability to suppress mutations was reduced only twofold for the p.G396D variant, no mutation suppression was observed upon expression of the p.Y179C variant. These results suggest that the intrinsic activity and active fraction of certain variants may be further compromised in a cellular context because of other factors such as protein stability and folding [37]. Also posttranslational modification, such as phosphorylation, might further reduce glycosylase activity for some variants in a human cell [16], [37].

Severely hampered or completely absent DNA binding and repair activity have subsequently been shown for the c.1147delC [33], [34], p.R185W [38], p.R241W [39], p.V246F [39], p.R245L [40], p.R245H, p.P295L, p.P405L [37], p.Y104X, p.Q391X, p.E480X [33], p.A473D [41] and p.G286E variants [22]. The p.R274Q is partially active [33]. The c.934-2A>C (IVS10-2A>C) variant causes the production of an aberrant mRNA transcript encoding a truncated MUTYH protein and immunofluorescence analysis shows the absence of this variant in the nucleus [42]. The mouse MUTYH variant of p.A373V did not show any impaired DNA glycosylase activity [22].

A number of variants are frequently found in polyposis patient and cancer cases as well as controls, the p.Q338H (previously p.Q324H) is found in 40–45%, the c.504+35G>A (IVS6+35G>A) in 20–25% and the p.V22M variant in 10–15%. The V22M variant showed the same activity as wild-type MUTYH protein in vitro in one study [33]. The pathogenic significance of p.Q338H is disputed. One study also reported no difference with respect to repair activity for the p.Q338H SNP compared to wild type [43]. In contrast, others found that this SNP was partially impaired with regard to adenine removal [33]. Recently it was reported that another variant at the same position, the Q324R variant, showed a reduced rate of adenine excision in the glycosylase assays but the enzyme activity was sufficient to mediate repair and prevent mutations in a cellular context [37].

In conclusion, the MUTYH BER protein has an important role in DNA repair following oxidative damage. Several different mutations, mainly missense mutations, have been found. The two most common mutations in the Western population are p.Y179C and p.G396D, with probable different effects on MUTYH glycosylase function.

4.1. Clinical management of MAP patients 

In 2008, a consensus meeting recommended colonoscopic surveillance of MAP patients every 2 years for patients from 18–20 years old [79]. When the number of polyps is greater than is allowed for by endoscopic removal, subtotal colectomy is indicated. Because of a high risk of metachronous tumors, total colectomy and ileorectal anastomosis or IPAA should be taken into consideration when operating upon a MAP patient. Since hyperplastic polyps and sessile serrated adenomas are a common finding in patients with MAP, and can eventually evolve into more malignant lesions, screening should include the detection and removal of these polyps [80], [81].

Beginning surveillance when patients are 18–20 years old seems sufficient. So far, only two cases have been reported with age of onset <20 years: a 14-year-old boy with hundreds of adenomas throughout the colon and an intramucosal carcinoma (now considered as high-grade dysplasia) and a 13-year-old boy with more than a hundred colon adenomas and gastric cancer at age 17 [57]. In these cases, additional genetic or environmental factors might explain the young age of onset.

Upper gastrointestinal tract screening is advised to begin from the age of 25–30 years [79]. The recommended intervals between screening depend on the severity of disease according to the Spigelman classification [79].

4.2. Geno-phenotype correlations 

In accordance with a number of functionality assays that show a greater reduction in MUTYH glycosylase activity for p.Y179C as compared to p.G396D (see Section 3.2), we recently showed that the phenotype for MAP patients with biallelic p.G396D mutations was less severe than for p.Y179C homozygotes [19]. Patients with a homozygous p.G396D mutation or compound heterozygous p.G396D/p.Y179C mutations presented with MAP later and had a significantly lower hazard of developing CRC than patients with a homozygous p.Y179C mutation (P<0.001). The mean ages of CRC diagnosis were 58 years (homozygous p.G396D), 52 years (c. heterozygous p.G396D/p.Y179C) and 46 years (homozygous Y179C; p=0.001, linear regression).

Based on the range of age at diagnosis of CRC (37–70 years), it may be appropriate to initiate surveillance later for p.G396D homozygotes and p.G396D/p.Y179C compound heterozygotes as compared to p.Y179C homozygotes. Unexpectedly, the p.G396D mutation in combination with mutations other than p.Y179C did not show a milder phenotype. Severity similar to that of the p.Y179C homozygotes was observed. A later age of CRC diagnosis in p.G382D homozygotes compared to p.Y165C homozygotes was also found by others [78]. Ideally, long-term prospective studies of MAP should be undertaken to extend the evidence for refinement of clinical guidelines.

4.3. Survival 

We found that MAP CRC patients have better overall survival than sporadic CRC cases (Nielsen et al., submitted). Lynch syndrome patients with CRC have also been shown to have better survival than sporadic CRC patients [82], [83]. Interestingly, MAP carcinomas show some histological similarities to Lynch syndrome carcinomas (see Section 6 of this review for more details). A higher immune response might be the underlying cause of improved survival, induced by accumulation of peptide fragments derived from secondary mutated proteins that are presented at the cell surface as part of the major histocompatibility complex. Because of the activated immune system, a strong selective pressure can be expected to favor the outgrowth of tumor cell clones with an immune evasive phenotype. Recently, it was indeed shown that loss of HLA class I receptor expression (involved in presenting mutated proteins to the immune system) is a frequent event in MAP carcinomas, similar to MMR-deficient colorectal tumors [84], [85], supporting the existence of an anti-tumor immune response [86].

4.4. Eligibility for MUTYH mutation screening 

MUTYH screening should be directed to patients with between 10 and a few hundred polyps (adenomas and/or hyperplastic polyps), especially in the context of a family history that is compatible with recessive inheritance, although a vertical transmission of CRC does not rule out the possibility of biallelic MUTYH mutations [25], [87]. Biallelic MUTYH mutations are found in about 28% of (APC germ-line mutation-negative) patients with 10–100 polyps and in 14% of patients with 100–1000 polyps (Table 1). One study showed that the combination of >15 synchronous adenomas and developing CRC younger than 50 years of age yielded a sensitivity of 75% and a specificity of 93.8% for detecting biallelic MUTYH mutations [58]. In the absence of a dominant family history of polyposis, it is reasonable to carry out MUTYH mutation analysis before or concomitantly with APC analysis in individuals with fewer than 1000 adenomas.

In population-based CRC studies where patients were collected on the basis of the diagnosis of CRC and not on polyps, biallelic MUTYH mutations were found in 0.3–2.0% (mean 0.4%) and 0.8–6.2% (mean 1.4%) of patients under 49–55 years when stratified for age (Table 1, Table 2). In eight population-based studies, 28 out of 79 (35%) proven biallelic MUTYH mutation carriers had no polyps aside from their CRC while seventeen (22%) had a limited number of adenomas (i.e., <10) [48], [49], [58], [59], [60], [61], [66], [67], [78]. Therefore, biallelic MUTYH mutations are not invariably associated with a polyposis phenotype and therefore, the practicing pathologist and clinician should also consider biallelic MUTYH mutations in CRC patients with zero or less than 10 polyps. However, no biallelic MUTYH mutations are found in patients with less than 10–15 polyps and no CRC so far (Table 1). Although, it should be considered that in MAP patients younger than 40 years the polyp development can still be slow, and MUTYH DNA analysis might be considered in these cases even if the polyp count is less than 15. A sudden rapid acceleration of adenoma development at age 52 years was recently found in a patient with a single CRC at age 44 years and previously only slow adenoma development [88]. A pre-screening assay for c.34G>T in KRAS in tumor material can also be helpful in these atypical MAP patients (see Section 6 of this review).

Table 2. Percentage of MUTYH biallelics in CRC patients according to age.

	Age	CRC with or without polyps
	<50 years	1.1% (29/2605)
		Range 0.8–6.2%
	>50 years	0.3% (28/11150)
		Range 0.0–0.6%*

Cumulating data from eight studies [49], [54], [58], [59], [63], [69], [72], [78].

Despite the relatively large proportion of population-specific mutations in MAP, it is recommended to screen the entire MUTYH gene. About 17% of 185 tested index patients in Germany, the Netherlands and the UK do not carry one of the two most common missense mutations (p.Y179C and p.G396D) [19]. In patients of southern European descent, this proportion is even higher (20–33%) [26], [28], [45], [74]. In non-western countries, the mutation spectrum is completely different [22], [27].

When biallelic mutations have been found in a patient, testing should be offered to siblings because they have a 25% chance of having inherited biallelic MUTYH mutations. Since MUTYH mutations are present in 1–2% of the population, children of mono- or biallelic MUTYH mutation carriers have a 0.5–1.0% chance of inheriting two MUTYH mutations because the other parent might (also) be a MUTYH heterozygote. A cost effectiveness analysis of Dutch MAP patients estimated that MUTYH testing in spouses of MAP patients, and to a lesser extent in spouses of MUTYH heterozygotes, was acceptable in terms of cost per quality-adjusted life year [89]. Therefore, genetic testing of spouses and children should be discussed with and offered to counselees.

5.1. Small bowel lesions 

Duodenal polyposis occurred in 17% of 276 MAP patients and the lifetime risk for duodenal cancer was 4% (relative risk (SIR) 129; 95% CI 16–66). Duodenal cancer was also reported in four other patients [45], [77], [94]. One case report described a MAP patient with three synchronous jejunal carcinomas at age 39 years [68]. Except for two carcinoid tumors in the small bowel, no other small bowel carcinomas have been reported [93].

5.2. Gastric cancer 

Among 150 endoscopically surveyed patients, 17 (11%) had gastric lesions and a higher risk of gastric cancer was observed, although this trend was not significant (3 out 150 surveyed patients, SIR 4.2, 95% CI 0.9–12) [93]. One case described in the literature had gastric cancer at age 17, suggesting the presence of other causative factors [57]. One additional MAP patient with gastric cancer has been reported [22]. Several Japanese groups studying a possible association of MUTYH mutations and gastric cancer show somewhat conflicting results. No biallelic MUTYH mutations among 20 familial cases of gastric cancer were found in Japanese patients. However, the authors found a biallelic novel splice site variant (IVS10-2A) in an additional group of 148 consecutive gastric cancer cases. This variant was not present in controls [42]. In 2006 a higher frequency of a specific variant of the MUTYH haplotype (p.P18L and p.G25D on the same allele) in suspected hereditary gastric cancer cases compared with controls was found (OR 4.4, p=0.021) [95]. No clear pathogenic MUTYH mutations in the normal tissue of 23 selected gastric cancer patients was found [96].

For now, mutated forms of MUTYH do not seem to be a strong causative factor in gastric cancer.

5.3. Endometrial cancer 

A non-significant trend towards endometrial cancer in 118 female MAP patients (two patients, SIR 4.6 (95% CI: 0.6–16.5)) was found. The ages at diagnosis were 47 and 54 years [93]. Four endometrial carcinomas in female MAP patients between 46 and 58 years of age have been reported by others [77], [97], [98]. The tumors in two patients were examined for KRAS mutations. The MAP-specific c.34G>T mutation was found in one patient, underscoring a causative role of defective MUTYH [97]. One of these patients was identified in a large series of consecutive endometrial cancers (n=225) [98]. Another study found no biallelic MUTYH patients among 213 endometrial cancer patients [99].

5.4. Breast cancer 

An association of breast cancer and MUTYH is supported by the fact that BRCA1 and BRCA2 are involved in 8-oxoG repair [100] and breast tumors are found in combined MUTYH/APC deficient mice [91]. The incidence of breast cancer in females in the European cohort with MAP was significantly increased (SIR 3.0; 95%CI 1.5–5.3), but only if the number of cancers rather than the number of affected females was considered. In contrast to hereditary breast and ovarian cancer, breast cancer in MAP patients was a late-onset manifestation occurring between the fifth and eighth decades of life. Breast cancer was also diagnosed at age 56 in one male MAP patient who tested negative for BRCA1 and BRCA2 germ-line mutations (SIR 53.5; 95% CI 1.4–298) [93]. In contrast, two large breast cancer case–control studies (including more than 5000 cases in total) found no association between breast cancer and mono- or biallelic MUTYH mutations [101], [102].

5.5. Other cancers 

A low to moderate but significant increase in the incidence of ovarian and bladder carcinomas, as well as skin cancer (SIR 5.7, SIR 7.2 and SIR 2.8, respectively), was found in the 276 European MAP patients. Skin cancers included melanomas, squamous epithelial carcinomas and basal cell cancers. The risk of bladder cancer in MAP was similar to the risk of urinary tract cancers in Lynch syndrome. However, in contrast to Lynch syndrome, no cancers were located in the upper part of the urinary tract (renal pelvis, ureter) and the MAP-associated cancers included one squamous cell carcinoma in addition to urothelial cancers [93]. Ovarian cancer was observed three times. In one cancer, the histology was known and involved a mucinous cystadenocarcinoma. Furthermore, two cases of thyroid carcinoma, each with a different histology, were observed in this cohort [93]. One additional case has been reported elsewhere [103]. The two cancers had different histological types, which does not support a clear association of thyroid cancer with MAP, in contrast to the established association with FAP. Furthermore, single MAP patients with scapular chondrosarcoma [94], a high-grade astrocytoma [94], and a Schwannoma [45] have been reported.

No overrepresentation of mono- or biallelic MUTYH mutations was found in patients with hepatocellular carcinoma [104], cholangiocellular carcinoma [104], (childhood) leukaemia [105], prostate cancer [106], [107] sporadic squamous oral/oropharyngeal carcinoma [108] or sporadic pancreatic cancer [109] as compared to healthy controls. In 276 patients with lung cancer, no overrepresentation of MUTYH mutations (mono- or biallelic) was found [110]. However, in a recent study, biallelic MUTYH p.Q338H (previously p.Q324H) showed a significant association with lung cancer risk in smokers (adjusted OR 3.82, p=0.022) [111].

5.6. Sebaceous gland tumors 

Sebaceous gland tumors (adenomas, epitheliomas and carcinomas), which are a well-known characteristic of the Muir–Torre variant of Lynch syndrome and are very rare in the general population, have been reported in four MAP patients [44], [103], [112]. In the European cohort, sebaceous gland tumors occurred in five patients (1.8% of all included MAP patients) [93]. Furthermore, sebaceous gland hyperplasia (SGH), which is not part of the Muir–Torre syndrome clinical definition, has been reported in four MAP patients. In all of these cases, a somatic BRAF p.V600E mutation was detected [51]. The authors suggest that detection of BRAF mutations in sebaceous hyperplasia can help identify MAP cases [51]. Notably, the mutation found concerned a T>A transversion, which is not associated with MUTYH deficiency.

5.7. FAP manifestations 

Some manifestations also seen in familial adenomatous polyposis (FAP) have been reported in a small number of MAP patients; lipomas in 3% (8/276) [93]; congenital hypertrophy of the retinal pigment epithelium (CHRPE) in 5.5% (3/62 analyzed patients [93] and 5 more cases in the literature) [26], [54], [62]; osteomas in two cases [26]; jaw-bone cyst in 11 cases [93]; epidermoid cyst in 3 cases [93]; desmoid tumor in 1 case [68]; and pilomatrixomas (pilomatricomas) in three siblings [113]. Overall, the incidence of FAP-related manifestations in MAP patients seems to be low.

In conclusion, several extracolonic tumors have been reported in MAP, but significantly higher risks were only observed for duodenal carcinoma, ovarian, bladder and skin cancer. Although extracolonic tumors are less prevalent than in Lynch syndrome patients, the extracolonic tumor spectrum is somewhat similar to that found in Lynch syndrome patients. The spectrum of cancers and their advanced age at onset in MAP do not suggest specific surveillance recommendations other than frequent upper gastrointestinal tract screening.

6.1. MAP colorectal carcinomas 

In further studies, APC mutations were found in 14–83% of MAP CRCs, all G>T transversions with a predilection for G bases in AGAA or TGAA motifs. A specific KRAS mutation (the c.34G>T in codon 12) is present in 64% of CRCs [54], [114], [115] In contrast, KRAS mutations are found in 29% of sporadic CRCs and 22% of sporadic MSI-high carcinomas. Furthermore, the c.34G>T transversion represents just 8% of KRAS mutations in sporadic CRCs and 0% of KRAS mutations in MSI-high CRCs [116]. P53 and SMAD4 mutations were found in MAP CRCs less frequently (21–60 and 0–26%, respectively). These were not predominantly G>T changes [114], [115].

In contrast to sporadic colon cancer, where physical loss of genetic material is the main characteristic, MAP CRCs are often near-diploid (52–92%) [114], [117] and commonly contain chromosomal regions of copy-neutral loss of heterozygosity (LOH) (71%) [117]. In contrast, another study found aneuploidy in 80% of MAP-adenomas [118], although another study did not [119]. One possible explanation for the difference is that different technical platforms were used [117].

Several studies investigated microsatellite instability (MSI) in MAP cancers. Two small studies found MSI-high phenotypes in 33 and 16% of patients (1/3 and 1/6 MAP CRCs, respectively) [54], [75] and suggested an MSI pathway (by inactivation of MLH1) in some MAP CRCs. Others found the MSI-high phenotype in a minority of cancers (0–18%, mean: 4%, 3/77) [67], [78], [81], [115]. One study found that CRCs of mono- and biallelic MUTYH carriers are more often MSI-low than sporadic CRCs (23.5% vs. 9%). However, only one carcinoma (6%, 1/17) was characterized as MSI-high [67].

Although the MSI pathway does not appear a common pathway in MAP tumors, we recently found that MAP CRCs show some histological similarities to cancers characterized by MSI (Lynch syndrome or sporadic MSI-high). A preferential proximal location was found in 69% (40/58), (meta) synchronous tumors in 23% (10/44), a high rate of mucinous histotype in 21% (9/42), and high frequency of tumor infiltrating lymphocytes (TILs) in 74% (marked in 17%) [115] (see Fig. 2 for examples of MAP carcinomas). A frequent proximal location, metachronous CRC and high numbers of TILs in comparison to sporadic CRCs were also shown by others [67], [114].

View Large Image Download to PowerPoint

Fig. 2. (a) mucinous MAP CRC, 5× and (b) tumor infiltrating lymphocytes in a MAP CRC, 40×.

Previously, MUTYH nuclear immunohistochemistry (IHC) staining was shown to discriminate between MUTYH-proficient and MUTYH-deficient carcinomas [120]. However, others were not able to replicate these results and concluded that MUTYH IHC is not a feasible technique to discriminate MUTYH-associated CRCs from sporadic CRCs [81], [121].

In conclusion, a KRAS hotspot mutation (c.34G>T in codon 12) is present in 64% of MAP CRCs and, less frequently, G>T transversions are found in the APC gene. Since G>T transversions are found predominantly in genes mutated in early tumor development (KRAS and APC), but not in P53 and SMAD4 (implicated in tumor progression), this finding might indicate a predominant MUTYH effect in early carcinogenesis. Furthermore, MAP CRCs have some specific molecular and histological features that differentiate them from sporadic CRCs, and show overlap with MSI-high and Lynch syndrome CRCs, such as a preferential proximal location, mucinous histotype and increased presence of TILs. These features should direct the pathologist towards a MAP etiology of CRC as an alternative cause of deficient mismatch repair, especially when diagnosed at a young age and in combination with polyps and/or a recessive inheritance pattern. KRAS analysis can be implemented as a pre-screening test that helps select patients with CRC who are eligible for MUTYH mutation screening [66].

6.2. Adenomas, hyperplastic polyps and sessile serrated adenoma 

MAP patients can present with conventional adenomas as well as serrated adenomas, hyperplastic polyps and mixed (hyperplastic and adenomatous) polyps [62], [73], [80], [81], [103], [122]. In 47% (8/17) of MAP patients 1 or more hyperplastic polyps and/or sessile serrated adenoma (SSAs) were found. Three of these patients fulfilled the criteria for the hyperplastic polyposis syndrome [80].

Mutations in KRAS are present in 18–23% of MAP adenomas [80], [123]. Similar to MUTYH-associated carcinomas, the hotspot mutation is a c.34G>T transversion in codon 12. This mutation was found in 100% of MAP adenomas with a KRAS mutation compared to only 13% of KRAS mutations in sporadic adenomas [123]. Another research group also found exclusively G:C to T:A transversions in MAP adenomas, although the proportion of c.34G>T transversions was not mentioned [80]. The frequency of KRAS mutation increases with the degree of dysplasia, villous content and size [123]. APC mutations are found in 41% of MAP adenomas and comprise exclusively G:C to T:A transversions [80]. No G:C to T:A transversions in APC were found in sporadic adenomas in this study [80].

In MUTYH-associated HPs and SSAs, no APC mutations were found, but KRAS mutations in codon 12 were present in 70% of patients (51/73); 94% of mutations comprised G:C to T:A transversions. In the control group, KRAS mutations were present in only 17% (7/41) of patients and no APC mutations were present. The authors concluded that HPs and SSAs are a common finding in patients with MAP and are causally related to biallelic MUTYH mutations. They subsequently suggested the existence of two distinct pathways, one leading to conventional adenomas with APC and/or KRAS mutations and one separate non-APC route leading to HPs and SSAs with KRAS mutations [80]. Another study found MUTYH biallelic mutations in 1 out of 38 patients with hyperplastic polyposis syndrome [73], suggesting that this gene should be considered in individuals with both adenomatous and hyperplastic polyposis. Further investigation is necessary to evaluate the prevalence of MUTYH mutations in patients with hyperplastic polyposis syndrome.

6.3. Somatic MUTYH mutations 

Somatic inactivation of MUTYH in CRC is reported by one group (2 out of 48 sporadic CRC cases) [124] but not by others [47], [63], [64]. It does not seem to be a frequent occurrence in CRC. One study found somatic MUTYH mutations in both alleles of 2 out of 95 sporadic gastric cancer cases [125]. A recent study found that the nuclear expression of MUTYH was attenuated in the mucosa of ulcerative colitis (UC)-associated neoplasia and UC without neoplasms compared to mucosa unaffected by UC. No pathological MUTYH mutations were detected in any of the UC-associated neoplasia cases [126].

7.1. Case–control studies 

Seventeen case–control studies addressing this question have been performed to date [47], [48], [49], [58], [59], [60], [61], [63], [65], [67], [71], [75], [76], [78], [127], [128], [129]. The majority of studies found a small overrepresentation of MUTYH heterozygotes in CRC cases with odds ratios between 0.78 and 4.7, although independent statistical significance was reached in only three [58], [59], [71]. Among CRC patients under age 50 or patients with familial CRC and/or less than 10 polyps no [49], [60] or only a slight overrepresentation of heterozygotes is seen [60], [69], [72].

To achieve more statistical power, several meta-analyses have been performed, with ORs or RRs between 1.11 and 1.27 [61], [127], [130]. Of these, only one reached (borderline) statistical significance (OR 1.27; 95% CI 1.01–1.61) [130]. There are several possible biases in these studies. One possible weak point in a number of these studies is that genetic screening was limited to the two common mutations (p.Y179D and p.G396D) or a subset of mutations in most studies. In a recent study, the entire MUTYH gene was screened and a significant odds ratio of 1.48 (95% CI.1.02–2.16) was found [67]. If screening had been limited to the two common mutations, the authors state that they would have failed to detect a significant increase (including only p.G396D and p.Y179C; RR 1.45; 95% CI 0.95–2.3). The authors also speculated that the less common variants may be more penetrant than the two common mutations. The mutational spectrum of MAP patients might not be completely representative of the unselected CRC patient spectrum [67]. Therefore, restricted genotyping for the most common mutations in MAP patients might lead to an underestimation of the risk for CRC in MUTYH heterozygotes. By contrast, the OR might also be overestimated when no or limited additional screening was performed because p.Y179C or p.G396D heterozygotes will be compound heterozygotes [61], [127]. For example, in two studies additional rare MUTYH variants were found in p.Y179C and p.G396D heterozygotes [48], [59].

Stratification by mutations or age might be needed to find a significant effect of MUTYH heterozygotic mutations (with correction for multiple testing). Two studies found a lower OR in p.G396D carriers (1.1 and 1.3) than in p.Y179C carriers (1.24 and 2.1), although these numbers did not reach significance [48], [58]. A large case–control study showed a significant relative risk after stratification for age: a risk ratio of 1.68 (95% CI of 1.07–2.95) for CRC in MUTYH heterozygous carriers aged >55 years [59]. This result was not confirmed by Webb et al. [127], who stratified 10-year age bands in a large case–control study and found no evidence for a relationship between cancer risk and age.

One small study suggested an association of the p.Q338H polymorphism with colorectal cancer susceptibility in patients with no smoking history (adjusted OR 4.08, p=0.022) [131]. Recently a significantly increased risk for rectal carcinoma (but not for colon) was found for p.Q338H homozygotes (OR:1,52, 95% CI:1.06–2.17) in a Swedish cohort (1785 cases and 1722 controls) [132]. Previously, the same research group reported in a smaller cohort (412 CRC cases and 188 controls) no evidence of an increased colorectal cancer risk for p.Q338H homo- or heterozygotes [76] This research group also noted that three (until that time unreported) variants at codon 437 (previously codon 423) were found to be overrepresented in CRC cases compared to healthy controls (5/447 vs. 0/478, p=0.02, Chi-square test). The pathological relevance of variants at this codon needs to be further studied. Another study demonstrated an increased CRC risk for the c.IVS1+11T MUTYH variant (OR 1.43, p=0.042) and a specific haplotype containing, amongst others, the c.IVS1+11T and p.Q338H variants (OR 1.43, p=0.046) [133].

7.2. Family studies 

Another strategy to examine whether MUTYH heterozygotes have an increased CRC risk is to study family members of MAP patients. One research group found a twofold significant increase in the incidence of colorectal cancer in 347-obligate MUTYH-heterozygous parents compared to the general population (standardized incidence ratio (SIR), 2.12). Their colorectal cancer mortality was not significantly increased (standardized mortality ratio (SMR), 1.02), nor was overall cancer risk (SIR 0.92), cancer mortality (SMR 1.12) or overall mortality (SMR 0.94) [87]. Another research group used a kin-cohort design to estimate CRC risk among 300 first-degree relatives of 39 CRC cases with either monoallelic or biallelic MUTYH mutations. An approximately threefold increased risk was found among monoallelic carriers, which was significant in persons aged >55 years and non-significant in persons aged <55 years [134]. A similar kin-cohort approach was used by two other large research groups. A somewhat – though non-significant – higher CRC prevalence (ratios of around 1.7) was found for first-degree relatives of MUTYH mutation carriers compared to CRC controls [78], [127].

7.3. Somatic inactivation of the wild-type MUTYH allele? 

A possible mechanism underlying CRC in MUTYH heterozygotes might be somatic inactivation of the wild-type MUTYH allele through loss of heterozygosity (LOH) at chromosome 1 (corresponding to the chromosomal location of MUTYH). LOH of this region was found in 8 (47%) of 17 colorectal tumors from monoallelic MUTYH gene mutation carriers but in only 2 (20%) of 10 sporadic colorectal tumors [48]. This argument was disputed by others because LOH at chromosome 1p in tumors is highly variable and subject to confounding [135]. In seven adenomas from a heterozygous p.G396D mutation carrier two somatic APC changes were found; neither change was a G:C to T:A transversion, thus providing no evidence of defective MUTYH activity [62]. Another study showed a significant positive correlation between the presence of monoallelic germ-line MUTYH variants and G:C to T:A transversions in KRAS or APC and 1p LOH. They suggest that single germ-line MUTYH variants may influence genetic pathways in CRC [65].

In conclusion, the risk of CRC among MUTYH heterozygotes in the population seems to be only marginally increased and present data do not justify colonoscopic screening in heterozygous MUTYH carriers. For heterozygous relatives of MAP patients available data suggest that such individuals have a two or at most threefold increase in their risk of CRC at an age similar to that in the general population and thus are expected to benefit from population screening measures or could be offered average moderate-risk colorectal screening based on their family history. Furthermore, heterozygotes do not seem to have an increased overall risk of cancer.

8.1. Association MUTYH and MSH6 

On a molecular level, MUTYH has been shown to be physically associated with the MSH2/MSH6 complex via the MSH6 subunit. The MSH6 binding site is mapped to a conserved region in the MUTYH gene and the binding and glycosylase activities of MUTYH are enhanced by the MSH2/MSH6 complex [17]. Moreover, the MSH2/MSH6 complex can also bind DNA duplexes containing oxoG:A mispairs [140], [141] and overexpression of the MSH2/MSH6 complex significantly decreased the rate of G:C to T:A transversions in MUTYH mutant E. coli cells [142]. Based on this interaction, it was hypothesized that both genes act cooperatively to confer an increased CRC risk. Indeed, patients with a mild MSH6 missense mutation had an MUTYH monoallelic mutation more frequently than controls, suggesting a contribution to adenomas and/or CRC development [143]. In addition, mutations in MSH6 were also more frequently found in heterozygote MUTYH CRC patients than in control CRC patients (11.5% vs. 0%) [144]. However, a different study did not find overrepresentation of MUTYH heterozygotes among MSH6 missense carriers [145].

On the other hand, abrogation of both MSH6 DNA mismatch repair and base repair might be mutually exclusive, as we previously showed in a patient with compound heterozygous MUTYH mutations and a pathogenic MSH6 germ-line mutation. This patient had an extremely mild clinical phenotype with only a few adenomas at age 56 years. No second hit of MSH6 was apparent in any of the adenomas, due to retained MSH6 nuclear expression and a lack of microsatellite instability [146].

For now it seems that biallelic or monoallelic germ-line mutations of MUTYH are unlikely to play a significant role in Lynch syndrome, HNPCC and HNPCC-like patients. It remains to be determined whether interaction between MUTYH and MSH6 contributes to the cause of polyps and/or CRC. Previously, the p.G396D and p.Y179C MUTYH mutations were not found in APC-positive FAP patients, suggesting that mutations in MUTYH do not play a modifying role in APC-driven FAP either [147].