Transforming Growth Factor β Receptor Type I and Transforming Growth Factor β1 Polymorphisms Are Not Associated with Postmenopausal Breast Cancer

A polymorphic allele in transforming growth factor β receptor 1 (TGFβR1) is hypothesized to increase risk of cancer (1). The allele, designated as TGFβR1*6A, results from the deletion of three alanines within a nine-alanine stretch in exon 1, and in vitro studies have shown that TGFβR1*6A is an impaired mediator of TGF-β antiproliferative signals compared with wild-type (TGFβR1*9A; refs. 2, 3). Most previous studies (2, 4, 5), including a meta-analysis (6), have found an association between TGFβR1*6A and increased risk of breast cancer, but the majority of these studies were hospital based (2, 5, 6). One study (5) further suggested that TGFβR1*6A interacts with the TGFβ1 T29C polymorphism. The TGFβ1 29C allele leads to significantly higher serum levels of TGF-β1 (7, 8), is hypothesized to reduce breast cancer risk (8), and has been examined in many previous breast cancer studies (5, 8-15). We used a nested case-control study within the Cancer Prevention Study II Nutrition Cohort (16) to examine whether the TGFβR1*6A allele influenced risk of postmenopausal breast cancer either alone or in combination with TGFβ1 T29C.

This analysis includes 502 postmenopausal breast cancer cases and 505 controls drawn from a subgroup of the cohort who donated blood samples (n = 21,965 women). Controls were matched on age, race/ethnicity, and date of blood collection; cases and controls had no previous history of cancer (other than nonmelanoma skin cancer). The cohort and case control study are detailed elsewhere (16, 17).

Genotyping was done at Applied Biosystems (Foster City, CA) using DNA purified from buffy coat. The TGFβ1 T29C assay was done using TaqMan as previously described (12). The TGFβR1 assay has also previously been described (1). Following PCR amplification, TGFβR1 alleles were visualized using an ABI Prism sequence analyzer. A 115-bp peak corresponded to the *9A allele and a 107-bp peak corresponded to the *6A allele. Laboratory personnel were blinded to case-control status and 10% blind duplicates were included in the assays. The concordance rate was 100% for TGFβ1 and 98% for TGFβR1. Overall success rate for the genotyping assays was >96% for both loci and there were no deviations from Hardy-Weinberg equilibrium.

Unconditional logistic regression was used to examine the association between TGFβ1 and TGFβR1 and breast cancer while controlling for matching factors. We examined whether the association with the variants of interest differed by stage at diagnosis using general summary stage to classify cases as in situ, localized, or regional/distant metastasis. We evaluated whether known breast cancer risk factors were confounders in the logistic models but their inclusion in the models did not appreciably change our effect estimates, and thus we did not include them in the final models.

To examine possible interactions between *6A (TGFβR1) and T29C (TGFβ1) alleles, individuals were classified into low, intermediate, or high TGF-β signalers based on previously published criteria shown in Table 1 (5).

Cases were mostly Caucasian (99%) with median age of 68 years (range, 46-83 years) at diagnosis. Table 1 shows the genotype frequencies, odds ratios (OR), and 95% confidence intervals (95% CI) for the association between breast cancer and the TGFβR1*6A and TGFβ1 T29C overall and by stage. Because the TGFβR1*6A allele is uncommon (allele frequency among controls, 10.7%), we combined women homozygous and heterozygous for the *6A allele and compared them with *9A homozygotes. We found no association with breast cancer and TGFβR1*6A (OR, 0.95; 95% CI, 0.69-1.30, for 6A carriers versus 9A/9A) or with TGFβ1 C/C compared with T/T (OR, 0.89; 95% CI, 0.61-1.31), and no differences in analyses stratified by stage.

Because the TGFβ1 T allele is associated with lower circulating levels of TGF-β1 than the C allele, and the TGFβR1*6A allele is a compromised form of wild-type TGFβR1, a previous study combined these variants to classify individuals as high, intermediate, or low signalers (5). We found no statistically significant differences across these groups (P_trend = 0.70).

We found no evidence of association of breast cancer with TGFβR1*6A nor with TGFβ1 T29C allele, either alone or in combination. Our study had ∼80% power to detect an OR of >1.5 for allele frequencies of 11% as observed for TGFβR1*6A among our controls (α = 0.05, β = 0.20). Previous studies have reported ORs in the range of 1.5 to 1.6 (2, 4, 5); thus, we had adequate power to observe an association of similar magnitude in our population.

A meta-analysis of TGFβR1*6A that included 1,420 cases of breast cancer reported a summary OR of 1.38 (95% CI, 1.14-1.67) among *6A carriers compared with *9A/*9A (6). However, there are important methodologic limitations in previous studies included in the meta-analysis. None were population based and most used external controls and did not account for differences in age and ethnicity between case and control groups (2, 4-6, 10).

TGFβ1 and breast cancer has been studied more extensively than TGFβR1, and the results have been mixed (5, 8-15). Only one previous study (5) has looked at these two polymorphisms jointly and found increased risk of breast cancer among the group that they defined as low signalers compared with high signalers (OR, 1.69; 95% CI, 1.08-2.66). By their definition, high signalers are those with TGFβ1 (C/C) and TGFβR1 (9A/9A). We did not observe an association between breast cancer and low signaling in our study (OR, 1.12; 95% CI, 0.69-1.81).

In summary, although both the variants examined in this study have been shown to affect TGF-β signaling, we found no evidence that they are associated with postmenopausal breast cancer, either alone or in combination.