HER2 Signatures in Breast Cancer: Ready to Go to Print?

Abstract

Breast cancer is a heterogeneous disease that is currently being classified into at least four main classes: basal-like, which is primarily progesterone receptor (PR) negative, estrogen receptor (ER) negative, and HER2 negative (ie, triple negative); luminal A (ER positive and low grade); luminal B (ER positive and high grade); and HER2 positive. Diagnosis of these subtypes is based on gene expression profiles and on traditional clinical and pathologic parameters, such as the status of ER, PR, and HER2, as well as histologic grade and the presence of specific proliferation markers. However, the ability of these criteria to predict disease progression or metastasis is imperfect, especially in tumors with intermediate histologic grade or intermediate ER expression. As such, clinicians are increasingly looking further into gene expression profiling to aid in the molecular diagnosis and treatment of the tumor. Recently, a number of studies linking gene expression profiles to clinical outcome in breast cancer have been demonstrated to outperform all of the known clinicopathologic parameters. This has lead to the clinical development of some of these signatures. For example, Mammaprint (Agendia, Huntington Beach, CA) has received US Food and Drug Administration approval to aid in the prognosis of patients with node-negative disease. Using DNA microarrays, this assay measures the expression of 70 genes and calculates a prognostic score that classifies the patients into a good or poor prognosis. Similarly, Oncotype DX (Genomic Health, Redwood City, CA), which has received endorsement from the American Society of Clinical Oncology for classifying patients with ER-positive, node-negative disease treated with tamoxifen, analyzes the expression of 21 known cancer-related genes using quantitative reversetranscriptase polymerase chain reaction. The Oncotype DX system categorizes these measurements into a quantitative recurrence score to classify patients into low-risk, intermediate-risk, or high-risk groups. Although many gene expression signatures provide a link between diagnosis, treatment, and clinical outcome in a substantial subset of breast cancers, present genomic signatures add only modest prognostic information for patients overexpressing HER2, who are usually classified as high risk in these signatures. The HER2 gene is amplified or overexpressed in 15% to 20% of invasive breast carcinomas and results in increased metastatic potential and poor clinical outcome. However, HER2-positive breast cancer is hardly a homogeneous disease, because a proportion of patients do not experience relapse or have a protracted clinical course. Likewise, although the humanized monoclonal antibody trastuzumab (Herceptin, Roche Biomedical Laboratories, Basel, Switzerland) improves survival in early and advanced HER2-positive breast cancer, the beneficial effect of trastuzumab is not uniform, because primary or acquired resistance to trastuzumab is not uncommon. Hence the variability of clinical outcome and response to antiHER2 therapies provides evidence that HER2-positive breast tumors may be heterogeneous in nature. Taking these observations as a backdrop, in this issue of the Journal of Clinical Oncology, Staaf et al sought to correlate the genetic heterogeneity of HER2-positive tumors with clinical outcome using molecular profiling. In a retrospective study using 58 HER2-amplified tumors, unsupervised gene expression analysis revealed three separate subtypes independent of stage, histologic grade, and ER status. Importantly, one of these subtypes (cluster 2) had a significantly worse clinical outcome, with overall survival 12% in the poor-prognosis group compared with 50% to 55% in the good prognosis groups over a 10-year follow-up period. Staaf et al then performed two-class Sam analysis of gene expression data comparing the poor-prognosis subtype with the two good-prognosis subtypes to identify a 158-gene prognostic predictor which they termed HER2derived prognostic predictor (HDPP). HDPP classification significantly improved stratification of good and poor prognosis in both overall survival and distant metastasis-free survival compared with unsupervised analysis in their HER2 tumor set. More importantly, the authors demonstrated strong prognostic value of the HDPP gene signature in HER2-overexpressing tumors from multiple breast cancer data sets that were generated from different microarray platforms. It is worth noting that Staaf et al observed that HDPP showed superior prognostic stratification of HER2-positive tumors in the Netherlands Cancer Institute data set as compared with both the Mammaprint and Oncotype DX systems. The prognostic value of the HDPP signature was not restricted to HER2-positive tumors. They demonstrated that along with HER2 molecular subtypes, HDPP performed well in basal-like tumors and showed association for survival in triple-negative tumors. In contrast, HDPP showed no prognostic value in luminal A, luminal B, or normal-like subtypes. After analyzing the findings from Staaf et al, the question that we need to address is whether we are ready to use the HDPP gene signature in the management of HER2-positive breast cancer? The answer is clearly no, at least for now. As with all previous published JOURNAL OF CLINICAL ONCOLOGY E D I T O R I A L S VOLUME 28 NUMBER 11 APRIL 1