Heterogeneity in etiology and prognosis of breast cancer by intrinsic subtypes

Abstract

Background: Breast cancer is heterogeneous in clinical behavior by intrinsic subtypes. However, it is unclear whether the intrinsic subtypes reflect underlying etiological classification. Furthermore, the etiological heterogeneity of breast cancer might differently contribute to tumor progression by intrinsic subtypes. This study aims to evaluate heterogeneity in etiology and prognosis of intrinsic subtypes in breast cancer. Materials and methods: From the Seoul Breast Cancer Study (SEBCS), a total of 3,433 breast cancer patients and 3,870 healthy controls were included in the analysis. The intrinsic subtype of breast cancer was defined based on hormone receptor (HR), composed of estrogen receptor (ER) and progesterone receptor (PR), and human epidermal growth factor receptor 2 (HER2) status. As explanatory variables, the genetic risk factors previously identified by genome-wide association study (GWAS) and established or probable non-genetic risk factors of breast cancer were considered. The distribution of intrinsic subtypes was compared among age groups, menopausal status, and tumor-node-metastasis (TNM) stages in the study subjects and among races by a systematic review and meta-analysis of previous studies. To evaluate the etiological heterogeneity of intrinsic subtypes, associations of genetic and non-genetic factors on breast cancer risk between age frequency matched cases and controls were tested by a multivariate logistic regression model. To evaluate that the etiological heterogeneity of intrinsic subtypes affect prognosis, associations of identified genetic and non-genetic risk factors on breast cancer survival were tested by a multivariate Cox proportional hazard model. Furthermore, a two-stage GWAS was conducted to identify novel genetic markers for breast cancer survival in 2,023 patients for discovery and additionally independent 1,616 patients for replication. Results: The distribution of subtypes was as follows

1,617 HR+ HER2- (56.1%), 398 HR+ HER2+ (13.8%), 395 HR- HER2+ (13.7%), 473 HR- HER2- (16.4%) with differences between age groups and TNM stages (P<0.01). The difference in distribution of subtypes was observed among Asian (HR+ HER2-: 57.7%, HR+ HER2+: 13.3%, HR- HER2+: 11.8%, HR- HER2-: 16.3%), Caucasian (HR+ HER2-: 62.4%, HR+ HER2+ 16.3%, HR- HER2+: 7.8%, HR- HER2-: 13.1%), and African population (HR+ HER2-: 49.5%, HR+ HER2+: 16.2%, HR- HER2+: 12.7%, HR- HER2-: 20.9%) by meta-analysis. In the case-control analyses, earlier menarche and lifetime endogenous estrogen exposure consistently increased the risk of all subtypes. The higher body mass index (BMI) in premenopausal women showed the association with the risk of HR- HER2- tumors (odds ratio (OR)=1.08, 95% confidence interval (CI)=1.02-1.13), but not the other subtypes. Having more number of children also showed the expected association with the risk of HR- HER2+ (OR=1.15, 95% CI=1.02-1.29) and HR- HER2- tumors (OR=1.13, 95% CI=1.01-1.26). Among identified genetic factors by previous GWAS, rs2046210 in 6q25 (ESR1, C6orf97) was related to the risk of all subtypes with no heterogeneity. However, rs10069690 in 5p15 (TERT, CLPTM1L) was associated with the risk of HR+ HER2+ (OR=0.71, 95% CI=0.52-0.99, Ptrend=4.10×10-2) and HR- HER2+ tumors (OR=0.63, 95% CI=0.45-0.89, Ptrend=8.53×10-3) and rs9485372 in 6q25 (ESR1, C6orf97) with risk of HR+ HER2- (OR=0.82, 95% CI=0.73-0.91, Ptrend=3.76×10-4) and HR- HER2+ tumors (OR=0.69, 95% CI=0.56-0.84, Ptrend=2.51×10-4) with statistically significant heterogeneity across subtypes (Prs10069690<0.01 and Prs9485372=0.02). In survival analyses, earlier menarche and longer endogenous estrogen exposure rather tended to show better survival, specifically in HR+ HER2- and HR+ HER2+ tumors. The higher BMI related to HR- HER2- tumors and more number of children related to HR- HER2+ and HR- HER2- tumors were not associated with survival. The prognostic relevance of identified single nucleotide polymorphisms (SNPs) which were associated with specific subtype was also not observed. In the two-stage GWAS stratified by intrinsic subtypes, rs166870 at 15q25 (MTHFS) and rs100825036 at 10q21 (PCDH15) were associated with survival of HR+ HER2- and HR- HER2- tumors, respectively (Prs166870=2.88×10-7 and Prs10825036=3.54×10-7 in the combined set). Conclusions: There were heterogeneity in distribution, risk factors, and prognostic factors of different intrinsic subtypes. Although the prognostic relevance of etiologically heterogeneous factors was not observed, novel genetic prognostic factors, rs166870 and rs10825036, were identified in HR+ HER2- and HR- HER2- tumors.