A genome-wide association study of breast cancer has revealed single nucleotide polymorphisms (SNPs) in five genes with novel association to susceptibility: TNRC9, FGFR2, MAP3K1, H19 and LSP1 [1]. The results were confirmed for FGFR2 and TNRC9 in two independent studies [2, 3]. However, these studies revealed little of the mechanisms underlying these associations. Pooling of such a large amount of cases, as performed in these studies, inevitably leads to concealment of the various histological and clinico-pathological subtypes. This suggests that the observed genes are either of universal importance for breast cancer development, are associated with a subgroup that dominates the overall pool or are associated with any subgroup but with an association sufficiently strong to dominate the overall result. Breast cancer patients can be divided into five distinct molecular subtypes based on their expression profiles [4]. The existence of these five subtypes, luminal A, luminal B, basal-like, ErbB2+, and normal-like, have been confirmed in independent datasets [5] and they are associated with different clinical outcomes [6]. If the probability to develop a given subtype of breast cancer is genetically determined, we might expect to find that the newly discovered susceptibility genes [1] are differentially expressed in the various tumour subtypes, and that their transcription is regulated in cis by SNPs within them. With this in mind, we retrieved the mRNA expression data of TNRC9, FGFR2, MAP3K1, H19 and LSP1 from 112 breast tumours representing all five subtypes [7]. Significantly different mRNA levels between the subtypes were found for all the five genes by ANOVA analysis (Table 1). For instance, TNRC9 was up-regulated in luminal A, luminal B and ErbB2+subtypes and down-regulated in the basal-like subtype (p = 4.5 × 10-7). FGFR2 was up-regulated in luminal A and basal-like subtypes and down-regulated in luminal B and ErbB2+ subtypes (p = 3.1 × 10-5), while MAP3K1 was up-regulated in luminal A and the normal-like subtypes and down-regulated in luminal B, ErbB2+ and basal-like subtypes (p = 5.2 × 10-5). Furthermore, we could calculate the association between SNPs residing within these genes and their tumour expression levels since genotype data on these patients have been generated using an Illumina 109K SNP array. The three genes whose expression levels were most significantly associated with tumour subtype (TNRC9, FGFR2 and MAP3K1) all harboured SNPs within them displaying a significant association with gene expression level (Table 1). One of these SNPs, rs9940048 in TNRC9, displayed a significantly different genotype distribution between the subtypes, with breast cancer patients homozygous for the low frequency allele over-represented in the basal-like subtype (p = 0.003), in concordance with the observation that the basal-like tumours had the lowest levels of TNRC9 mRNA.

Table 1 P values after ANOVA analyses
Full size table

Conclusion

Our results suggest that SNPs in the recently discovered susceptibility genes may exert their effect through the expression of their genes in tumours, giving rise to the various breast cancer subtypes. Thus, stratification of patients by their molecular subtypes may give much more power to classic case control studies, and genes of no or borderline significance may appear to be high-penetrant for certain subtypes and, therefore, be identifiable.