We read with great interest the article by Hosey et al. ( 1 ) , which proposed BRCA1mediated transcriptional activation of ESR1 in breast cancer cell lines, thus suggesting a direct link between the BRCA1 gene and estrogen response that involves OCT-1 . We assessed whether these in vitro results could be clinically confirmed in an independent cohort of primary breast cancer specimens that included 23 sporadic and 27 hereditary breast tumors in which full DNA sequencing and multiplex

more »

... ndent probe amplification of the BRCA1 gene detected no mutations. We assessed estrogen receptor (ER) status as previously described ( 2 ) and determined the mRNA expression levels of the ESR1 , BRCA1 , and OCT-1 genes by real-time quantitative polymerase chain reaction (TaqMan gene expression assay identification numbers Hs00174860_m1, Hs00173233_m1, and Hs00231250_m1, respectively; Applied Biosystems, Foster City, CA), using the housekeeping gene 18S ribosomal RNA (4319413E) for normalization. BRCA1 mRNA levels in the 34 ERpositive (ER ϩ ) samples overlapped with those of the 16 ER-negative (ER Ϫ ) samples (range of log 2 -transformed BRCA1 expression: Ϫ 19.8 to Ϫ 15.95 in ER Ϫ and Ϫ 19.6 to Ϫ 14.7 in ER ϩ ), resulting in a weak association between BRCA1 expression and ER status (ratio of mean BRCA1 expression of ER ϩ to ER Ϫ samples = 1.61, 95% confidence interval [CI] = 0.99 to 2.56, P = .054 [two-sided t test with unequal variances using the log 2 -transformed data]). There was no association between OCT-1 expression and ER status (mean expression ratio = 0.83, 95% CI = 0.58 to 1.19, P = .30). These results were further confi rmed in publicly available datasets ( 3 ): neither BRCA1 nor OCT-1 expression was associated with ER status in sporadic breast can-cers ( P < .05 for two of 13 datasets for BRCA1 and for three of 11 datasets for OCT-1 ). On average, ER ϩ samples that had higher BRCA1 expression also had higher ESR1 expression ( P = .004, Wald test), consistent with the proposal of Hosey et al. ( 1 ). In a series of 19 BRCA1-mutant samples from our institution for which ER status and ESR1 and BRCA1 mRNA expression levels were available, we found four ER ϩ samples; interestingly, in those four samples, BRCA1 expression was positively associated with ESR1 expression. To see whether the existence of ER ϩ , BRCA1 -mutant cases could be explained by the presence of specifi c deleterious BRCA1 mutations that disrupt protein function but do not abrogate BRCA1mediated ESR1 transcription, we analyzed the type and location of the BRCA1 mutation in a series of BRCA1 -mutant patients with primary breast cancer who had no personal history of ovarian cancer (including 24 ER ϩ and 65 ER Ϫ BRCA1 -mutant patients). We identifi ed 41 different mutations of the BRCA1 gene -20 in the ER ϩ patients and 31 in the ER Ϫ patients -that were distributed along the entire coding sequence, with no evidence of preferential grouping of mutations according to the ER status. Twelve of these mutations were present in both the ER ϩ and ER Ϫ patients, including the 5382insC mutation [the mutation that was present in the HCC1937 cell line that Hosey et al. used for the in vitro experiments ( 1 )], which we detected in three ER ϩ patients and 12 ER Ϫ patients. Finally, we observed that patients who were members of the same family (and thus carried the same BRCA1 mutation) had tumors that differed with respect to ER status, which indicates that the type of BRCA1 mutation alone does not determine the ER status of breast tumors. Our results suggest that ESR1 expression levels might be infl uenced by genes other than BRCA1 and OCT-1 and/or by mechanisms (4 -6) other than those proposed by Hosey et al. ( 1 ).