In sporadic colorectal tumours the BRAFV600E is associated with microsatellite instability (MSI-H) and inversely associated to KRAS mutations. Tumours from hereditary non-polyposis colorectal cancer (HNPCC) patients carrying germline mutations in hMSH2 or hMLH1 do not show BRAFV600E, however no consistent data exist regarding KRAS mutation frequency and spectrum in HNPCC tumours. We investigated KRAS in 158 HNPCC tumours from patients with germline hMLH1, hMSH2 or hMSH6 mutations, 166 MSI-H and 688 microsatellite stable (MSS) sporadic carcinomas. All tumours were characterized for MSI and 81 of 166 sporadic MSI-H colorectal cancer (CRCs) were analysed for hMLH1 promoter hypermethylation. KRAS mutations were observed in 40% of HNPCC tumours, and the mutation frequency varied upon the mismatch repair gene affected: 48% (29/61) in hMSH2, 32% (29/91) in hMLH1 and 83% (5/6) in hMSH6 (P = 0.01). KRAS mutation frequency was different between HNPCC, MSS and MSI-H CRCs (P = 0.002), and MSI-H with hMLH1 hypermethylation (P = 0.005). Furthermore, HNPCC CRCs had more G13D mutations than MSS (P < 0.0001), MSI-H (P = 0.02) or MSI-H tumours with hMLH1 hypermethylation (P = 0.03). HNPCC colorectal and sporadic MSI-H tumours without hMLH1 hypermethylation shared similar KRAS mutation frequency, in particular G13D. In conclusion, we show that depending on the genetic/epigenetic mechanism leading to MSI-H, the outcome in terms of oncogenic activation may be different, reinforcing the idea that HNPCC, sporadic MSI-H (depending on the hMLH1 status) and MSS CRCs, may target distinct kinases within the RAS/RAF/MAPK pathway.

Approximately 30% of all hereditary diffuse gastric cancer (HDGC) families carry CDH1 germline mutations. The other two thirds remain genetically unexplained and are probably caused by alterations in other genes. Using polymerase chain reaction (PCR)/single-strand conformation polymorphism (SSCP)/sequencing, we screened 32 Portuguese families with a history of gastric cancer and 23 patients with early onset gastric cancer for CDH1 germline mutations. In probands negative for CDH1 mutations, we screened genes involved in hereditary cancer syndromes in which gastric cancer may be one of the component tumours, namely p53 (Li-Fraumeni Syndrome) and hMLH1 and hMSH2 (HNPCC). We also screened in these patients for mutations in Caspase-10, a gene inactivated in sporadic gastric cancer, and SMAD4, a gene whose inactivation in mice is associated with signet-ring cell carcinoma of the stomach. One of the families fulfilling the HDGC criteria harboured a CDH1 germline mutation, and one of the families with incomplete criteria harboured a p53 germline mutation. No mutations were identified in hMLH1 and hMSH2, and only sequence variants were found in SMAD4 and Caspase-10. The present work reports for the first time CDH1 germline mutations in Portuguese gastric cancer families, and highlights the need for p53 mutation screening in families lacking CDH1 germline mutations, in a country with one of the highest incidences of gastric cancer in the world. No evidence was found for a role of germline mutations in SMAD4 and Caspase-10 in families lacking CDH1 mutations.

It is unclear whether the mutation spectra in WNT genes vary among distinct types of colorectal tumors. We have analyzed mutations in specific WNT genes in a cohort of 52 colorectal tumors and performed a meta-analysis of previous studies. Notably, significant differences were found among the mutation spectra. We have previously shown that in familial adenomatous polyposis, APC somatic mutations are selected to provide the "just-right" level of WNT signaling for tumor formation. Here, we found that APC mutations encompassing at least two beta-catenin down-regulating motifs (20 a.a. repeats) are significantly more frequent in microsatellite unstable (MSI-H) than in microsatellite stable (MSS) tumors where truncations retaining less than two repeats are more frequent (P = 0.0009). Moreover, in cases where both APC hits are detected, selection for mutations retaining a cumulative number of two 20 a.a. repeats became apparent in MSI-H tumors (P = 0.001). This type of mutations were also more frequent in proximal versus distal colonic tumors, regardless of MSI status (P = 0.0008). Among MSI-H tumors, CTNNB1 mutations were significantly more frequent in HNPCC than in sporadic lesions (28% versus 6%, P < 10-6) and were preferentially detected in the proximal colon, independently of MSI status (P = 0.017). In conclusion, the observed spectra of WNT gene mutations in colorectal tumors are likely the result from selection of specific levels of beta-catenin signaling, optimal for tumor formation in the context of specific anatomical locations and forms of genetic instability. We suggest that this may underlie the preferential location of MMR deficient tumors in the proximal colon.

We report the analysis of altogether 1050 suspected hereditary breast/ovarian cancer (HBOC) families, 524 fully screened for BRCA1/BRCA2 mutations and 526 tested only for the most common mutations. Of the 119 families with pathogenic mutations, 40 (33.6%) had the BRCA2 c.156_157insAlu rearrangement and 15 (12.6%) the BRCA1 c.3331_3334del mutation, the former being specific of Portuguese ancestry and the latter showing a founder effect in Portugal. Interestingly, the two most common mutations were found in a significant proportion of the HBOC families with an a priori BRCAPRO mutation probability <10%. We recommend that all suspected HBOC families from Portugal or with Portuguese ancestry, even those fulfilling moderately stringent clinical-criteria for genetic testing, should be specifically analyzed for the two most common BRCA1/BRCA2 founder mutations, and we here present a simple method for this first tier test. Screening of the entire coding regions of BRCA1 and BRCA2 should subsequently be offered to those families with a mutation probability ≥10% if none of those founder mutations are found.

BackgroundKRAS is an EGFR effector in the RAS/RAF/ERK cascade that is mutated in about 40% of metastatic colorectal cancer (mCRC). Activating mutations in codons 12 and 13 of the KRAS gene are the only established negative predictors of response to anti-EGFR therapy and patients whose tumors harbor such mutations are not candidates for therapy. However, 40 to 60% of wild-type cases do not respond to anti-EGFR therapy, suggesting the involvement of other genes that act downstream of EGFR in the RAS-RAF-MAPK and PI3K-AKT pathways or activating KRAS mutations at other locations of the gene.MethodsDNA was obtained from a consecutive series of 201 mCRC cases (FFPE tissue), wild-type for KRAS exon 2 (codons 12 and 13). Mutational analysis of KRAS (exons 3 and 4), BRAF (exons 11 and 15), and PIK3CA (exons 9 and 20) was performed by high resolution melting (HRM) and positive cases were then sequenced.ResultsOne mutation was present in 23.4% (47/201) of the cases and 3.0% additional cases (6/201) had two concomitant mutations. A total of 53 cases showed 59 mutations, with the following distribution: 44.1% (26/59) in KRAS (13 in exon 3 and 13 in exon 4), 18.6% (11/59) in BRAF (two in exon 11 and nine in exon 15) and 37.3% (22/59) in PIK3CA (16 in exon 9 and six in exon 20). In total, 26.4% (53/201) of the cases had at least one mutation and the remaining 73.6% (148/201) were wild-type for all regions studied. Five of the mutations we report, four in KRAS and one in BRAF, have not previously been described in CRC. BRAF and PIK3CA mutations were more frequent in the colon than in the sigmoid or rectum: 20.8% vs. 1.6% vs. 0.0% (P=0.000) for BRAF and 23.4% vs. 12.1% vs. 5.4% (P=0.011) for PIK3CA mutations.ConclusionsAbout one fourth of mCRC cases wild-type for KRAS codons 12 and 13 present other mutations either in KRAS, BRAF, or PIK3CA, many of which may explain the lack of response to anti-EGFR therapy observed in a significant proportion of these patients.