Pfefferle Adam D, Spike Benjamin T, Wahl Geoff M, Perou Charles M
Department of Pathology and Laboratory Medicine, University of North Carolina at Chapel Hill, Chapel Hill, NC, 27599, USA,
Breast Cancer Res Treat. 2015 Jan;149(2):425-37. doi: 10.1007/s10549-014-3262-6. Epub 2015 Jan 10.
Mammary gland morphology and physiology are supported by an underlying cellular differentiation hierarchy. Molecular features associated with particular cell types along this hierarchy may contribute to the biological and clinical heterogeneity observed in human breast carcinomas. Investigating the normal cellular developmental phenotypes in breast tumors may provide new prognostic paradigms, identify new targetable pathways, and explain breast cancer subtype etiology. We used transcriptomic profiles coming from fluorescence-activated cell sorted (FACS) normal mammary epithelial cell types from several independent human and murine studies. Using a meta-analysis approach, we derived consensus gene signatures for both species and used these to relate tumors to normal mammary epithelial cell phenotypes. We then compiled a dataset of breast cancer patients treated with neoadjuvant anthracycline and taxane chemotherapy regimens to determine if normal cellular traits predict the likelihood of a pathological complete response (pCR) in a multivariate logistic regression analysis with clinical markers and genomic features such as cell proliferation. Most human and murine tumor subtypes shared some, but not all, features with a specific FACS-purified normal cell type; thus for most tumors a potential distinct cell type of 'origin' could be assigned. We found that both human luminal progenitor and mouse fetal mammary stem cell features predicted pCR sensitivity across all breast cancer patients even after controlling for intrinsic subtype, proliferation, and clinical variables. This work identifies new clinically relevant gene signatures and highlights the value of a developmental biology perspective for uncovering relationships between tumor subtypes and their potential normal cellular counterparts.
乳腺的形态和生理由潜在的细胞分化层级所支撑。沿着这一层级与特定细胞类型相关的分子特征，可能导致在人类乳腺癌中观察到的生物学和临床异质性。研究乳腺肿瘤中的正常细胞发育表型，可能会提供新的预后范式，识别新的可靶向通路，并解释乳腺癌亚型的病因。我们使用了来自多项独立的人类和小鼠研究中经荧光激活细胞分选（FACS）的正常乳腺上皮细胞类型的转录组图谱。通过荟萃分析方法，我们得出了两个物种的共识基因特征，并利用这些特征将肿瘤与正常乳腺上皮细胞表型联系起来。然后，我们汇编了一组接受新辅助蒽环类和紫杉烷化疗方案治疗的乳腺癌患者数据集，以确定在包含临床标志物和细胞增殖等基因组特征的多变量逻辑回归分析中，正常细胞特征是否能预测病理完全缓解（pCR）的可能性。大多数人类和小鼠肿瘤亚型与特定的FACS纯化正常细胞类型共享一些但并非全部特征；因此，对于大多数肿瘤，可以确定一种潜在的不同“起源”细胞类型。我们发现，即使在控制了内在亚型、增殖和临床变量之后，人类管腔祖细胞和小鼠胎儿乳腺干细胞特征仍能预测所有乳腺癌患者的pCR敏感性。这项工作识别了新的临床相关基因特征，并突出了发育生物学视角在揭示肿瘤亚型与其潜在正常细胞对应物之间关系方面的价值。
