In this study, we have reported the prevalence of somatic mutations in KRAS
PIK3CA, and BRAF genes among patients with advanced colorectal carcinoma from Sardinia, whose population shows genetic peculiarity due to geographical isolation and strong genetic drift . Prevalence of candidate gene mutations may vary among distinct populations due to concurrence of different environmental factors and genetic backgrounds. Furthermore, patients’ geographical origin within the same population may also account for different mutation rates in pathogenetic cancer genes, as already demonstrated for different types of cancer in Sardinian population by our group [15–17]. In summary, we observed a moderate rate of KRAS mutations (145/478; 30.3%) or PIK3CA mutations (67/384; 17.4%) and a very low rate of BRAF mutations (1/384; 0.3%) in a cohort of primary colorectal carcinomas.
Prevalence of KRAS mutations in our series is consistent with data from literature, indicating that such an alteration can be found in 30-40% of colorectal carcinomas . Controversial data have been instead published about the PIK3CA mutation rates, ranging from 7%  to 30%  of CRC cases presenting a mutated PIK3CA. In the present study, all detected PIK3CA variants have been previously demonstrated to be oncogenic in CRC cellular models  and commonly associated with colorectal carcinoma (17%; see Table 3); the mutation rate was therefore comparable with that described in majority of previous reports.
The most surprising finding was the nearly lack of BRAF mutations in our series. In recent meta-analyses, the BRAF V600E mutation - which represents the most common mutation in BRAF gene (more than 90% of cases) - was detected in about 9% of primary colorectal carcinomas [29, 30]. One could speculate that the very low frequency of BRAF mutation detected in our series may be somehow due to patients’ origin or, in other words, to the peculiarity of the genetically-isolated Sardinian population. On this regard, it cannot be excluded that different pathogenetic mechanisms of transformation could occur in different populations. Microsatellite instability (MSI), a recognized marker of a tendency for replication errors in human cancers, has been widely indicated as a factor associated with higher frequency of mutations in BRAF gene among colorectal carcinomas [29, 31, 32]. Although such an analysis was not conducted in the present study, Sardinian CRC population has been previously demonstrated to present an incidence of MSI similar to that observed in other CRC populations from Western countries [33, 34]. Therefore, this factor could not explain the striking discrepancy on the BRAF mutation prevalence in our series.
Considering the two prevalent alterations, KRAS and PIK3CA mutations were more or less equally distributed among the different patients’ subsets, and no statistically significant correlation with sex, onset age (though patients with older age at diagnosis and a higher tumour grade are more likely to present with a PIK3CA mutation), disease stage, primary CRC location, or tumour grading was observed (see Table 1).
As schematically represented in Figure 2, we made comparisons between prevalence of KRAS and PIK3CA mutations within different geographical areas of the island. In a population sharing a quite similar lifestyle and diet habit across the island (moreover, smokers were homogeneously distributed among patients of different origin - though such an information from medical records was available in only about 70% of cases from our series), the observation that a higher frequency of KRAS mutations was found in CRC cases from North Sardinia (43% vs. 21%) whereas a higher prevalence of PIK3CA mutations was found in patients from South Sardinia (24% vs. 10%) strongly suggest that different “genetic background” may also induce discrepant penetrance and distribution of somatic mutations in candidate cancer genes. As for similar data reported by our group in breast cancer and melanoma, the geographical distribution of the genetic variants in the island seems to be related to the specific large areas of Sardinia, which reflect its ancient history: the North area, delimited by the mountain chain crossing Sardinia and linguistically different from the rest of the island; and the Middle-South area, land of the ancient Sardinian population and domain of pastoral culture. Nevertheless, our findings clearly indicate that mutation frequency for any candidate cancer gene needs to be accurately evaluated in each geographical area within every single population. Due to this unexpected heterogeneity in distribution of somatic mutations in such few main genes, we already started the collection of tumour DNA samples from Sardinian colorectal cancer patients in order to perform a whole-exome sequencing and define a more comprehensive pattern of mutations in this population.
Finally, although performed on a limited number of CRC cases, our preliminary data seemed to indicate no correlation between PIK3CA mutations and response to the anti-EGFR treatment in patients with wild-type KRAS (objective responses were considered at first evaluation only) (see Table 5). A second study focused on evaluating such clinical aspects in a larger subset of patients is ongoing.
KRAS mutations are considered as an early event in the sequential accumulation of molecular alterations underlying the progression from colorectal adenoma to malignant carcinoma, resulting in an important tumour growth advantage. During the recent past years, a targeted therapy with monoclonal antibodies (cetuximab and panitumumab), blocking the EGFR-driven cell proliferation signals, has been introduced into the therapy of metastatic colorectal cancer [9, 35]. No significant response to therapy with anti-EGFR antibodies have been observed in colorectal cancer patients exhibiting KRAS mutations . Moreover, majority of colorectal carcinomas exhibiting wild-type KRAS do not respond to such therapies either . These phenomena are expected from the molecular point of view, since EGFR tyrosine kinase transmits proliferation signals via RAS-GTPase on the cell membrane inner surface, which in turn can bind effector proteins such as RAF or PIK3CA kinases . Therefore, the occurrence of functional integrity of the RAS-driven pathways - BRAF-MEK-ERK and PIK3CA-AKT - is necessary in order to really interfere with tumour cell growth through inhibition of EGFR target. In other words, the assessment of mutational status of BRAF and PIK3CA genes into the KRAS wild-type population may indeed improve the selection of patients presenting such a functional integrity of the RAS-driven pathways (though we are aware that additional alterations in downstream effectors may intervene). In contrast to KRAS, the heterogeneity of BRAF and PIK3CA mutations has not been adequately investigated in colorectal cancer thus far.
In the present study, because some somatic mutations (in BRAF and PIK3CA genes or in KRAS and PIK3CA genes) occurred concomitantly in a given patient, which is in line with literature, a total of about 45% of all patients showed at least one mutation in any of these three genes (see Table 4). Therefore, our data suggest that including mutation analyses for BRAF and PIK3CA in addition to KRAS into a standard diagnostic setting of colorectal cancer would allow the identification of an additional fraction (in our case, about 15%) of patients who cannot be considered as “true wild-type” for such main proliferation-controlling genes. However, whether or not these additional patients might benefit from EGFR-specific antibody therapy has to be verified in prospective clinical studies.