Prevalence and Characteristics of MUTYH-Associated Polyposis in Patients with Multiple Adenomatous and Serrated Polyps

See related article by Borras et al., p. 1061

MUTYH-associated polyposis (MAP; OMIM #608456) is an autosomal recessive disease that usually appears in patients with an attenuated polyposis phenotype (1, 2). This syndrome is associated with biallelic mutations in the MUTYH gene. Patients with MAP exhibit a mean of 50 polyps (3), and this disease is responsible for 7% of attenuated adenomatous polyposis and 6.6% of classic polyposis cases (4). MAP cases with colorectal cancer not showing a polyposis phenotype have also been described previously (5). Colorectal cancer risk in patients with biallelic mutations is about 80% by the age of 70 years without treatment, and colorectal cancer is diagnosed simultaneously with the diagnosis of polyposis in approximately 50% of patients (6). Patients with MAP can also present duodenal adenomatous polyps and polyps in the fundus (6, 7). Extraintestinal neoplasias of breast, gastric, thyroid, testis, and hematologic origin have been described in MAP syndrome (8, 9).

Two common mutations, c.536A>G; p.Y179C and c.1187G>A; p.G396D, are reportedly responsible for approximately 80% of MAP cases in Caucasian populations (2, 10), although large deletions (11) and other low-frequency mutations have also been identified in this gene (12). The International Society for Gastrointestinal Hereditary Tumors (InSiGHT) database presently includes 300 unique DNA variants of the MUTYH gene, some of which are probable founder mutations in different populations. However, about 20% to 33% of the tested index patients of southern European populations do not carry one of these most common mutations (13).

It has been recently reported that the MAP phenotype differs from other previously described polyposis syndromes in that it may involve the coexistence of both adenomatous and serrated polyps (14). However, to date, MAP diagnosis has been based on the presence of adenomatous polyps (4, 15, 16). This differential phenotype makes it unclear which patients should be tested for MUTYH, and whether patients with serrated polyps or those with multiple adenomatous and serrated polyps should also be investigated for this disease. Moreover, different authors have proposed different diagnostic strategies, and no consensus has been reached (4, 5, 17, 18).

The present study aimed to determine the prevalence of MUTYH mutations in a population of patients with multiple colonic polyps, including both adenomatous and serrated polyps, and to explore the diagnostic yield of different strategies for germline MUTYH testing in these patients. Moreover, we investigated the usefulness of somatic molecular markers in the polyps for MAP diagnosis. Our results could lead to better characterization of the syndrome and enhanced strategies for the genetic diagnosis of the disease, consequently improving colorectal cancer prevention and the management and surveillance of these patients.

The study included 405 patients from the EPIPOLIP study, a multicenter nationwide project that investigated causes of multiple colonic polyps and incorporated patients from 24 Spanish hospitals. Patients were retrospectively recruited during the years 2009 to 2010 (19). Patients diagnosed with at least 10 polyps of any histology were included in the present study. Patients previously diagnosed with familial adenomatous polyposis, Lynch syndrome, or inflammatory bowel disease, and those who had only hyperplastic rectosigmoid polyps were excluded. Informed written consent was obtained from all participants. The study was approved by the ethical committees from the participating hospitals.

Demographic data about age, sex, and personal and familial history of polyposis, colorectal cancer, or other neoplasia were collected. Endoscopy reports and the corresponding histopathology reports were also reviewed to collect information about the number, size, morphology, distribution, and histology of colonic polyps. Polyps were classified as adenomatous or serrated. Polyps were considered proximal if located in the transverse or ascending colon or cecum, and distal if located in the descending or sigmoid colon or rectum. A central review of all specimens was performed by three experienced pathologists (A. Payá, C. Egoavil, and C. Alenda). On the basis of the polyp histology, cases were classified as adenomatous polyposis, multiple adenomatous, and serrated polyps, or only multiple serrated polyps (20). Serrated polyps included hyperplastic polyps, traditional serrated adenomas, sessile serrated polyps, and mixed hyperplastic/adenomatous polyps.

DNA was extracted from peripheral blood samples of the 405 patients using the QIAamp DNA Kit (QIAGEN) following the manufacturer's instructions.

A total of 605 polyps were collected from 56 patients. From each polyp, we prepared nine 5-μm paraffin-embedded sections containing only the representative biopsy. Genomic DNA was extracted with the QIAamp DNA Investigator Kit (QIAGEN) following the manufacturer's instructions. All samples were dissected macroscopically by a pathologist (A. Payá, C. Alenda, and C. Egoavil) to ensure that they contained predominantly neoplastic tissue.

All cases were analyzed by PCR and sequencing of the MUTYH exons 7 and 13, where the two most frequent mutations are located (c.536A>G; p.Y179C and c.1187G>A; p.G396D). All patients were also examined for the frameshift change c.1227_1228dup (Glu410GlyfX43), which was previously reported as a common mutation in different Mediterranean populations (21, 22). Patients who were heterozygous for any of these three mutations were sequenced for the whole coding region and intron–exon boundaries to determine the existence in trans, of a second mutational event. In addition, 216 patients who showed none of the three analyzed mutations were also investigated for germline mutations along the whole coding sequence of the MUTYH gene. This group was a representative sample of the initial population, with no significant differences in age, gender, familial history, diagnosis of colorectal cancer, or in the number, size, type, and location of polyps (Fig. 1).

Cases with more than 10 adenomas were also tested for APC mutations (n = 183). Direct amplicon sequencing was performed using BigDye v3.1 terminators and a 3500 Genetic Analyzer (Applied Biosystems). Sequencing results were analyzed using Sequencing Analysis v. 5.1 and Variant Reporter v. 1.1 software (Applied Biosystems).

The 183 cases with more than 10 adenomas were tested for large insertion/deletion of the APC gene. Cases with biallelic or monoallelic pathogenic mutations were analyzed for large rearrangements of the MUTYH gene. These analyses were performed by multiplex ligation-dependent probe amplification (MLPA) using MLPA Kits P043 APC, and P378 MUTYH (MRC-HOLLAND). MLPA assays were performed following the manufacturer's instructions. PCR fragments were separated and quantified by capillary electrophoresis on a 3500 Genetic Analyzer (Applied Biosystems). Fragments were analyzed using GeneMapper v. 4.0 analysis software (Applied Biosystems).

Genetic analysis results were interpreted on the basis of the American College of Medical Genetics and Genomics (Bethesda, MD) recommendations for Standards for Interpretation of Sequence Variations and the InSiGHT database. Bioinformatics tools were used to perform in silico analyses for estimating the functional effects at the RNA and protein levels of the MUTYH base substitution variants with unknown clinical significance. Polymorphism Phenotyping v2 (PolyPhen-2) and SNPs3D can predict the possible impact of an amino acid substitution on the structure and function of a human protein using straightforward physical and comparative considerations. A PolyPhen-2 variant prediction score of less than 5% was considered benign, and a score of more than 95% was considered to be probably damaging. SNPs3D predictions produced support vector machine (SVM) values; values below −0.5 were associated with deleterious variants, whereas those more than +0.5 were associated with neutral mutations. Splicing was studied using SplicePort, which predicts losses or gains of donor and acceptor splice sites. When these bioinformatics tools made at least two concordant predictions of pathogenicity for a given variants of uncertain significance (VUS), it was classified as clinically significant.

Cases with polyp DNA available were tested for somatic KRAS mutation (n = 56). We analyzed a mean of 10 polyps per patient. Direct sequencing was performed to identify KRAS mutation at exon 1, including codons 12 and 13. Both mutations were assessed by direct amplicon sequencing with BigDye v1.1 terminators and a 3500 Genetic Analyzer (Applied Biosystems).

The V600E BRAF mutation was detected by real-time PCR (ABI PRISM 7500, Applied Biosystems) using specific TaqMan probes and the allelic discrimination software (Applied Biosystems), as previously described by Benlloch and colleagues (23).

Data were tested for statistical significance using SPSS software (SPSS 19.0). Parametric continuous variables are reported as mean ± SD, whereas nonparametric continuous variables are reported as median (Q2–Q3 interquartile range). Categorical variables are reported as frequency or percentage. The χ² test, followed by the Fisher exact test where appropriate, was used to identify correlations between categorical parameters. The Student t test was used for quantitative data, but the Kruskal–Wallis test was used for comparisons between more than two groups. A P value of less than 0.05 was considered significant.

A total of 405 unrelated patients with at least 10 polyps from the EPIPOLIP study were included in the present investigation. The mean age at diagnosis was 57.5 years (SD, 12.55; range, 13–84). There were 286 males (70.6%). Personal history of colorectal cancer was reported for 145 patients (35.8%), and 162 patients (40%) had a familial history of colorectal cancer or colonic polyps in first-degree relatives. The median number of polyps was 20 (25–75 interquartile range, 13–36). A majority of polyps were <10 mm, with patients having a median of 15 polyps <10 mm (25–75 interquartile range, 11–33) and a median of two polyps ≥10 mm (25–75, interquartile range, 0–4). Patients had a median of eight proximal polyps (25–75 interquartile range, 4–14) and 10 distal polyps (25–75 interquartile range, 5–20). Polyp histology revealed that 135 cases were adenomatous polyposis (33.3%), 219 involved multiple adenomatous and serrated polyps (54.1%), and 51 (12.6%) only showed multiple serrated polyps, 40 of which met the World Health Organization (WHO) criteria for serrated polyposis (9.9%; ref. 20).

Of the 405 total patients, 183 (45.2%) had more than 10 adenomas. These patients were tested for APC mutation, and 6 cases (3.3%) showed a germline heterozygous pathogenic mutation in the APC gene. Each case with APC mutation showed exclusively adenomatous polyps.

All of the 405 patients were studied for MUTYH germline mutations. Eighteen (4.4%) had biallelic mutations of the Y179C and G396D variants (Fig. 1, group 1; and Table 1), of which 4 (22.2%) had a homozygous p.Y179C mutation, 6 (33.3%) had a homozygous p.G396D mutation, and 8 (44.5%) were compound heterozygous. No patient showed biallelic c.1227_1228dup mutation. Two patients were carriers of compound heterozygous Y179C and c.1227_1228dup mutations (Fig. 1, group 2; and Table 1).

Homozygotes carriers for each of the two common mutations did not exhibit any differences in age at diagnosis, familial history, colorectal cancer predisposition, or in the number, location, or size of polyps. However, serrated polyps were more common in p.G396D homozygotes carriers (83.3%) than in p.Y179D homozygotes carriers (0%; χ²; P = 0.05). Four (50%) compound heterozygous cases showed only adenomatous polyps, while the remaining four showed both adenomas and serrated polyps. The mean ages at diagnosis were 47 (homozygous p.Y179C), 57 (homozygous p.G396D), and 57 years (heterozygous p.Y179C/p.G396D), indicating that homozygous mutations in Y179C were associated with early onset of MAP syndrome (χ², P = 0.05). Nineteen monoallelic carriers of any of the three most common mutations were found, including seven monoallelic carriers for the Y179C mutation, 10 for the G396D mutation, and two for the c.1227_1228dup mutation.

We performed whole-gene analysis for the 19 patients who were heterozygous for any of the three common mutations. Nine of these patients (47.4%) showed a second MUTYH mutation, including 3 that had been previously described as pathogenic (the nonsense pathogenic mutation Gln338X, the known pathogenic changes c.389-1G>A and c.721C>T; refs. 24, 25). Of the new six VUS, three were classified as pathogenic based on the predictions of in silico studies (Table 2). Therefore, another 6 patients were genetically diagnosed with MAP (Fig. 1, group 2; and Table 1).

Whole-gene analysis was also performed in a representative sample of 216 patients who did not show any of the three recurrent mutations. Only one of these patients (0.5%) had a biallelic pathogenic MUTYH mutation. This patient (Table 1, # 27) was diagnosed at 40 years of age, and had one brother who was also diagnosed with polyposis at 40 years of age; no genotyping data were accessible for this relative.

Another patient exhibited two concurrent VUS, both considered to be likely benign. Whole-gene analysis also identified 7 patients as heterozygous, but only one VUS identified in these patients was classified as probably pathogenic based on the in silico results. No large rearrangements were identified in these patients (Tables 1 and 2).

Overall, a total of 27 patients were diagnosed with MAP due to biallelic pathogenic mutations (Tables 1 and 3), indicating an estimated 6.7% prevalence of MAP among APC mutation-negative cases with multiple polyps in the colon (27 of 405). Of these 27 patients, 16 (59.3%) had colorectal cancer and serrated polyps were found in 11 (40.8%). One patient fulfilled the WHO criteria for serrated polyposis (Table 1, patient #6; ref. 20), which supposes a 2.5% of patients with serrated polyposis (1 of 40 patients with serrated polyposis included in the study). Four patients (14.8%) showed other extracolonic cancers (2 had endometrial cancer, 1 ovarian cancer, and 1 testicular germ cell tumor). In all MAP cases, family history was consistent with recessive inheritance with one or more affected siblings—except for patients 3 and 23 (Table 1), who showed inheritance consistent with vertical transmission.

Comparing MAP cases with the patients with multiple colonic polyps who did not show MUTYH mutation (Table 3), we found that patients with MAP had an earlier age of polyposis onset (52.7 vs. 60.2; P = 0.005) and more polyps (P < 0.001), as expected. Patients with MAP also more commonly had a personal history of colorectal cancer (59.3% vs. 35%; P = 0.014) and high-grade dysplasia adenomas (P = 0.02). Disregarding those patients who were diagnosed with colorectal cancer in the first colonoscopy, we observed that the incidence of new cases of colorectal cancer in patients with MAP was also higher than in the non-MUTYH patients (P = 0.03).

Serrated polyps were more frequently located in the right colon in patients diagnosed with MAP (P = 0.008). Larger polyps (>1 cm) were more frequently found in cases without MUTYH mutation (P = 0.03). No significant differences were found regarding the presence of family history of polyps and colorectal cancer or extracolonic cancer (Table 3).

Comparing patients with biallelic mutations in the two common MUTYH hotspots (group 1) with those carrying other pathogenic mutations (group 2) revealed no significant differences in age at diagnosis; total number, size, and location of polyps; presence of serrated polyps; familial history; extracolonic cancer; or colorectal cancer predisposition (Table 4). However, there was a trend for patients of group 2 to have an earlier age at diagnosis (48.5 vs. 54.8 years).

Patients with MAP who carried the G396D mutation showed older age at diagnosis, and this mutation was significantly associated with the presence of serrated lesions (P = 0.008). Carriers of the Y179C mutation more frequently showed proximal polyps (Table 4).

A total of 605 polyps from 56 patients included in this study were analyzed for the KRAS Gly12Cys and BRAF V600E somatic mutations. According to histology, 50.2% of the polyps were classified as tubular adenomas, 9.5% as tubulovillous adenomas, and 40.3% as serrated polyps. Among patients with MAP, 10.3% of polyps were serrated, whereas 58.7% were serrated in MUTYH-negative patients

Polyps from 13 patients with MAP were analyzed for the KRAS Gly12Cys somatic mutation. Eleven of these patients (84.6%) showed this mutation in at least 10% of their polyps (mean, 29%; range, 10%–50%; Tables 1 and 2). One of the five carriers of heterozygous MUTYH mutations showed the KRAS Gly12Cys mutation in 50% of polyps. This mutation was not observed in any of the 373 polyps from 36 MUTYH-negative patients, or in the 2 patients who carried benign VUS. Comparing MAP cases with those without MUTYH mutation revealed a strong association between germinal MUTYH mutation and KRAS Gly12Cys somatic mutation in polyps (P < 0.001; Table 3).

KRAS Gly12Cys somatic mutation was found more frequently in serrated polyps and tubulovillous adenomas (61.5% and 63.6%, respectively) than in tubular adenomas (14%; P < 0.001). Gly12Cys mutation was more frequent in larger polyps (>5 mm): 38.1% versus 19.3% (P = 0.02). No differences were found relating to polyp location.

On the other hand, BRAF V600E mutation was not detected in any of the 178 analyzed polyps from patients with MAP, and was detected in 31.2% (116 of 373) of polyps from MUTYH-negative patients. BRAF mutation was predominantly found in serrated polyps (Table 3).

We found that only 18 of 27 (66.6%) MAP cases could be explained by biallelic mutations in the two common MUTYH hotspots. Analysis of the two hotspots and the third common variant in our population resulted in a sensitivity of 74.1% (20 of 27 cases). Whole-gene analysis of cases heterozygous for any of the three common variants achieved the best performance, diagnosing 96.3% of cases. Table 5 shows the performance characteristics of these strategies.

Considering the characteristics of our patients with MAP, investigation of the MUTYH gene in cases with only adenomatous polyps would achieve a sensitivity of only 59.3%. If we had performed the MUTYH study in only patients with more than 30 adenomas, 10 patients with MAP (37%) would not have been identified. However, by including patients with multiple adenomatous and serrated polyps and more than 10 adenomas, only 1 patient would remain undiagnosed (Table 5). Regarding age at diagnosis, limiting MUTYH analysis to patients younger than 60 years would reduce the sensitivity to 70.4% (Table 5).

The present report describes several notable results. First, almost half of the identified MAP cases showed both adenomatous and serrated polyps, especially when G396D mutation was involved. There were even cases showing less than 10 adenomas, raising the question of whether there is an adenoma cutoff for MAP diagnosis in cases with both adenomatous and serrated polyps. Second, we found that the MAP diagnostic strategy should start with analysis of the most common mutations, taking into account locally prevalent mutations, and it should include whole-gene analysis only in cases that present a heterozygous mutation in a gene hotspot. Third, in our population, the majority of MUTYH mutation carriers showed the somatic KRAS Gly12Cys mutation in their polyps, but BRAF V600E mutation was not found in serrated polyps of patients with MAP.

Biallelic germline mutations in MUTYH accounted for 6.7% of the analyzed cases that had an attenuated polyposis phenotype. This mutation detection rate was lower than found in previous reports (2, 22, 26, 27), but similar to that described in more recent studies of cases with multiple colorectal adenomas (4). The frequencies of the Y179C and G396D alleles were consistent with those in previous studies (10, 22, 28), although the present population showed a higher heterogeneity of MUTYH mutations—including 10 new variants, five of which were considered pathogenic. Our study confirms previous reports indicating the presence of both adenomatous and serrated polyps in almost half of the MAP cases. These results support the notion that MAP should be particularly suspected in patients with multiple adenomas coexisting with serrated polyps throughout the colon, especially when familial history suggests a recessive hereditary pattern, and in patients who are young or who have a high number of polyps. We found one MAP patient fulfilling the WHO criteria for serrated polyposis (20), which represents a 2.5% of patients with serrated polyposis included in the study, supporting the need for studies of MUTYH in patients with this disease (14).

Massive parallel sequencing methodologies are undoubtedly going to change the analysis strategy in cancer genetics; however, until next-generation sequencing becomes broadly available in diagnostic laboratories, Sanger sequencing remains essential. Our present findings have important implications for the design of mutation detection strategies. We recommend a sequential analysis strategy using the direct sequencing of exons 7 and 13 of MUTYH for MAP diagnosis, starting with examination of the most locally prevalent mutations. In our study, more than 95% of diagnosed MAP cases had at least one of the three most common mutations, which were all located in these exons. Correct MAP diagnosis requires identification of the locally prevalent mutations that exist in a large proportion of cases with MUTYH germline mutation by analyzing the whole gene in a number of cases. Our results support the need for complete MUTYH gene analysis only in patients who are carriers of prevalent mutations (Y179C, G396D, and Glu410GlyfX43), even if recessive inheritance is not confirmed. We found a very low diagnostic yield from whole-gene analysis in patients with multiple adenomatous and serrated polyps. No MUTYH rearrangements were identified in this study. Not having analyzed the whole gene in every case was a limitation of our study, which precludes the possibility of having more accurate values for the sensitivity and specificity of the different strategies. However, given the extremely low rate of biallelic pathogenic mutation found in the 216 studied cases, we decided not to complete the whole-gene testing in all cases.

It has been suggested that there is a genotype–phenotype association for MUTYH mutations, with worse pathogenicity and earlier age of onset in carriers of the Y179C allele (29). Our results offer new insight into this association. We report that the G396D variant was strongly associated with the presence of serrated polyps, even in heterozygosis. We also found 2 patients who developed endometrial cancer; one had a sister with MUTYH biallelic mutation who also had both endometrial and colorectal cancer. Endometrial cancer has been proposed as an extracolonic neoplasia in MAP syndrome (30), and the present findings might support this partnership. However, the number of MAP cases that we included precludes our ability to provide solid evidence about the association between MAP and extracolonic neoplasms.

In our population, germline mutations in APC or MUTYH accounted for only a small portion of cases with multiple colonic polyps. Additional related mutations have been reported in genes involved in the base excision repair system or the Wnt pathway (31–35). Other mechanisms, such as mosaicism and deep intronic APC gene mutations, are also reportedly associated with attenuated polyposis, and could explain a small proportion of cases (36, 37). The existence of other yet unknown genes involved in cases showing multiple colonic polyps with incomplete penetrance, as well as complex interactions between environmental factors and modifier genes might further account for the unexplained cases.

One proposed molecular hallmark of carcinomas caused by MUTYH deficiency is the presence of the c.34G>T (p.Gly12Cys) KRAS mutation in MAP colorectal cancers. Previous reports have indicated that this mutation occurs at a high frequency in tumors of patients with MAP, and it has been suggested that this analysis could be implemented as a prescreening test to help select persons for MUTYH genetic testing (38, 39). Our results confirmed the existence of this KRAS Gly12Cys somatic mutation in the polyps of patients with MAP, supporting its possible role as a MAP diagnostic tool, especially in atypical cases with young onset colorectal cancer or recessive colorectal cancer familial history. This molecular marker is more frequently found in serrated and tubulovillous adenomas and, interestingly, it shows a high specificity, having only been found in the polyps of patients with MAP and not in other cases with multiple colonic polyps. This marker could also be a useful tool for classifying VUS at MUTYH; analyzing the presence of the somatic mutation Gly12Cys in polyps or colorectal cancer could support the pathogenicity of the VUS. This technique may be limited by the need for testing multiple polyps, as well as cost or technical problems in obtaining polyp tissue. Further research should be performed to study possible uses of this promising molecular marker of MAP. Moreover, in this study, we found no polyp showing the somatic BRAF V600E mutation in patients with MAP. We believe this to be a very striking fact, given the high proportion of BRAF mutation that we found in the serrated polyps of MUTYH-negative patients. This finding suggests the possibility of using BRAF mutation in serrated polyps as a negative molecular marker for MAP.

In summary, our present results support the existence of different polyp phenotypes in MAP. Genetic testing should be offered, not only in cases with attenuated adenomatous polyposis as has been classically described, but also and especially in cases showing multiple adenomatous and serrated polyps. The determination of clinical and molecular hallmarks of MAP will enable more reliable correct identification of possible carriers of pathogenic MUTYH mutations.

No potential conflicts of interest were disclosed.

Conception and design: C. Guarinos, J. Cubiella, F. Rodríguez-Moranta, A. Castillejo, R. Jover

Development of methodology: C. Guarinos, C. Egoavil, R. Salas, F. Rodríguez-Moranta, A. Castillejo, C. Alenda, J.-L. Soto

Acquisition of data (provided animals, acquired and managed patients, provided facilities, etc.): C. Egoavil, M. Rodríguez-Soler, L. Pérez-Carbonell, J. Cubiella, F. Rodríguez-Moranta, L. de-Castro, L. Bujanda, A. Serradesanferm, D. Nicolás-Pérez, M. Herráiz, F. Fernández-Bañares, A. Herreros-de-Tejada, E. Aguirre, J. Balmaña, M.-L. Rincón, A. Pizarro, F. Polo-Ortiz, A. Castillejo, C. Alenda, A. Payá, R. Jover

Analysis and interpretation of data (e.g., statistical analysis, biostatistics, computational analysis): C. Guarinos, M. Rodríguez-Soler, R. Salas, J. Cubiella, A. Castillejo, C. Alenda, J.-L. Soto, R. Jover

Writing, review, and/or revision of the manuscript: C. Guarinos, C. Egoavil, J. Cubiella, L. de-Castro, M. Herráiz, F. Fernández-Bañares, E. Aguirre, J. Balmaña, A. Castillejo, J.-L. Soto, R. Jover

Administrative, technical, or material support (i.e., reporting or organizing data, constructing databases): C. Guarinos, C. Egoavil, R. Salas, A. Herreros-de-Tejada, E. Aguirre, R. Jover

Study supervision: J. Cubiella, A. Castillejo, J.-L. Soto, R. Jover

Performance of the genetic testing of patients: M. Juárez

This work was supported by the Instituto de Salud Carlos III (PI08/0726, INT-09/208, PI11/2630, and INT-12/078). C. Guarinos received a predoctoral grant from Conselleria d'Educació de la Generalitat Valenciana (VALi+d. EXP ACIF/2010/018). M. Rodríguez-Soler received a grant from Fundación de la Comunidad Valenciana para la Investigación en el Hospital General Universitario de Alicante and Instituto de Salud Carlos III (Rio-Hortega grant CM11/00066) and L. Pérez-Carbonell received grants from Instituto de Salud Carlos III (FI07-00303) and Fundación Alfonso Martín Escudero.

The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.