The TGFβ pathway plays an important role in tumour development and progression . Recently, there has been interest in the contribution of genetic variation in this pathway to cancer susceptibility. Several polymorphic variants in genes involved in this pathway have been studied . The TGFBR1_Int7G24A variant has been previously associated with cancer risk, but the role of this variant has yet to be discovered. Zhang et al  reported that homozygous but not heterozygous carriers of this variant have an increased risk of developing non-small cell lung cancer. Chen et al  reported the association of this variant with the incidence of renal cell carcinomas (OR: 2.2; 95% CI: 1.22-3.96) and transitional cell carcinoma of the bladder (OR: 2.45; 95% CI: 1.8-3.16). In a meta-analysis of these three studies , the pooled OR was 1.76 (95% CI: 1.33-2.34; P < 0.0001). Another case-control study  found an association with breast cancer (OR: 2.42; 95% CI: 1.55-3.79), suggesting that the Int7G24A variant could also represent a risk factor for invasive and metastatic breast cancers (OR: 2.61; 95% CI: 1.65-4.11). Skoglund-Lundin et al.  did not detect a significant association between CRC risk and the Int7G24A variant when they analysed data from cases with a familial history of colorectal cancer. The expectedly low-penetrance effect of this putative susceptibility allele would be very difficult to detect in cases with high-penetrance alleles. Moreover, the authors did not stratify their data according to sex.
An exclusion criterion for the present study was the diagnosis of any familial cancer syndrome predisposing to CRC, such as hereditary non-polyposis colorectal cancer, familial adenomatous polyposis, Cowden disease and Peutz-Jeghers syndrome. We were interested to evaluate the effect of this SNP for sporadic CRC susceptibility and tumour progression. The potential effect of the SNP on familial syndromes as modifier of the penetrance, or the phenotype expression should be approached independently in further specific studies.
Control individuals were selected from those patients treated at an emergency department with diverse non-cancerous pathologies and with no personal history of cancer. The probability that controls suffered from diseases putatively caused by the same risk factors as the patients with CRC is extremely low.
It is well known that one of the main limitations and sources of confounding the results in association studies comes from population stratification and inappropriate sample sizes. A priori, our study with 504 sporadic cases and 504 controls allowed us to detect an OR of 1.5 for G/A heterozygous individuals (considering a frequency of 0.3) and an OR of 2.0 for A/A homozygous individuals (considering a frequency of 0.025) with 80% power (two-sided test, alpha level 5%). Estimated genotype frequencies were obtained as preliminary results in a subset of our control population and were similar to the frequencies registered at the SNP database for European populations http://www.ncbi.nlm.nih.gov/projects/SNP/. The results presented here show an increase of CRC susceptibility for those individuals carrying the Int7G24A variant allele with a dosage effect (additive model of inheritance). Adjusting the OR by age, we only observed an association of Int7G24A variant carriers with CRC risk in subjects aged less than 70 years (Table 1). Thus, the genetic risk effect is more evident in younger individuals. Ageing implies a huge amount of acquired predisposing carcinogenic factors that could overshadow the mild effect of the susceptibility alleles.
When we analysed the data adjusting by gender, we found that the Int7G24A variant was strongly associated with CRC risk in men but not women. We do not know why. Evidence is accumulating supporting gender-related differences in the development of CRC, but to our knowledge, few reports have been published regarding sex-specific relationship between SNPs and CRC. Slattery et al  reported that the oestrogen receptor beta CA repeat polymorphism was associated with an increase relative risk of colon cancer in women but not in men. On the contrary, increasing number of CAG repeats in the androgen receptor was directly associated with colon cancer among men, but not in women. Both oestrogen and androgen receptors are present in colorectal tissue and it has been suggested that these might be important in regulating the CRC risk associated with these hormones. Activation of the TGFβ signal transduction system is also subject to hormonal regulation . This striking connection might suggest the involvement of sex steroid hormones acting through the TGFβ pathway in the aetiology of CRC.
The apolipoprotein E (ApoE) epsilon 2/3 polymorphism is another example of gender-specific modulation of CRC risk and prognosis where men have a highly significant association but there is no association in women . Bae et al  found a gender-specific association between polymorphisms of the gene for vascular endothelial growth factor (VEGF 936 C>T) and CRC in the Korean population. T allele-bearing genotypes significantly increased the risk for CRC in women but not in men. Moreover, there are gender-related survival differences in association with EGFR (gene for epidermal growth factor receptor) polymorphisms in patients with metastatic colon cancers . Again, oestrogens and androgens may be behind all these associations because of their functional links with ApoE , VEGF  and EGFR .
It is possible that the positive associations shown here were not based on real effects of this polymorphism, but rather reflect unknown differences in population ancestry between the case and control groups . We consider the probability of false-positive inference attributable to population stratification to be small, because the cases and control individuals were recruited from an ethnically homogeneous population with no indication of a significant amount of recent genetic admixture.
Given our results and those in the literature, it seems reasonable to speculate that there might be a gender-dependent allelic architecture for CRC risk associated with sex steroids. This would introduce a new level of complexity in the 'common disease-common variant' hypothesis . Our association studies between this polymorphism and clinicopathological factors did not show any statistically significant results. The lack of difference in genotype distribution between male and female patients with CRC might be attributed to the contribution and compensation of other gender-related susceptibility polymorphisms, where the risks affect women preferentially. To date, how this intronic SNP affects oncogenesis remains to be elucidated. It has been proposed that this SNP creates an alternative splice site within intron 7. In vitro analysis with breast and ovarian cell lines from carriers of the Int7G24A variant have shown the retention of the seventh intron in the mRNA up to the site of the Int7G24A variant . We wanted to study if this alternative splicing also occurred in the normal colorectal tissue from patients with CRC. However, we concluded that factors other than alternative splicing might account for the association of this polymorphism with cancer, although we cannot rule out the possibility that Int7G24A variant is a marker representing a TGFBR1 haplotype. In this regard, it is interesting to note the likely 'bystander susceptibility effect' for the TGFBR1*6A allele because of linkage disequilibrium with the unknown causative mutation of the allele-specific expression (ASE) in the TGFBR1 gene . In our series, no linkage disequilibrium was found between the TGFBR1*6A and Int7G24A variant (unpublished results). No information is available regarding the association between variant and ASE. Further studies are required to characterize the molecular mechanisms by which variant is involved in gender-specific susceptibility to CRC.