-
Views
-
Cite
Cite
Ning Qing Liu, Marcel Smid, John W.M. Martens, John A. Foekens, Arzu Umar, Response, JNCI: Journal of the National Cancer Institute, Volume 106, Issue 9, September 2014, dju254, https://doi.org/10.1093/jnci/dju254
Close - Share Icon Share
Extract
We are writing in response to the comment of Bakinde and colleagues, who correctly state that triple-negative breast cancer (TNBC) is a very heterogeneous disease. Indeed, it has been shown in a large meta-study of publically available gene expression data (n = 3247, of which 587 are TNBC cases) that TNBC actually consists of seven different subtypes, including two basal-like (BL1 and BL2), an immunomodulatory (IM), a mesenchymal (M), a mesenchymal stem-like (MSL), a luminal androgen receptor (LAR), and an unstable subtype ( 1 ). In their comment, Bakinde and colleagues express their concerns about whether our described prognostic signature for TNBC ( 2 ) may have a different outcome in African American (AA) patients and may thus be less applicable because TNBC occurs more frequently in this group of patients.
Unfortunately, we do not know the exact ethnic composition of our studied patient population, but in general it is fair to assume that the majority of patients had a Caucasian white (CW) background, because all samples were gathered from five European centers. We did investigate whether the TNBC subtypes described by Lehmann et al. ( 1 ) were present in our cohort based on gene expression data (n = 66). Indeed, we identified all seven subtypes in our small cohort ( Figure 1A ) with comparable frequency distribution, as in the Lehmann dataset, except for the LAR and MSL subtypes, which seem to be underrepresented in our own ( 2 ) cohort ( Figure 1B ). Although Lehmann et al. also do not report on the ethnic background of their study cohort, it is clear that a substantial proportion (n = 1350) of the analyzed gene expression data comes from different centers from within the United States. One could assume that these datasets include AA patients, suggesting that there must be reasonable overlap in molecular characteristics between TNBC tumors of AA and CW women. In fact, it has been recently described that, although the frequency is higher, TNBC in AA women is not a unique disease compared with CW women, because the tumors are similar at the molecular level and 10-year survival is very similar between the two populations ( 3 ). Similarly, it has been reported for several Asian populations that the frequency of TNBC is higher than observed in Western countries ( 4–6 ). This explains why the 70-gene prognostic signature (MammaPrint TM ), originally developed using tumors from predominantly CW women, predicts substantially more ‘high-risk’ patients in a Korean cohort compared with European cohorts ( 7 ). These findings emphasize the importance of studying breast cancer subtypes separately from each other.
