Problems and solutions in the evaluation of hormone receptors in breast cancer.

An article by Badve et al in this issue of the Journal of Clinical Oncology describes an interesting study comparing two methods of assessing estrogen-receptor (ER) and progesterone-receptor (PR) expression in invasive breast cancer (IBC). One method measured protein by immunohistochemistry (IHC) and the other measured RNA using quantitative reverse-transcription polymerase chain reaction (RT-PCR). All assays utilized formalin-fixed, paraffin-embedded tissue samples from a case-control study cohort derived from the Eastern Cooperative Oncology Group clinical trial E2197, which randomly assigned patients with high-risk IBC to adjuvant doxorubicin/cyclophosphamide versus doxorubicin/docetaxel, with the addition of tamoxifen in receptor-positive cases. The cases (n 179) represented all patients who relapsed (5 years follow-up) matched to an excess of controls (n 597) who did not relapse, with available paraffin blocks. The IHC assays were performed twice in different laboratories. One was in the laboratories of institutions using local (and almost certainly diverse) staining protocols and definitions of positivity, most likely evaluating single 3to 5m histologic sections from intact paraffin blocks of tumor. The other was in an expert central laboratory using a standardized staining protocol on single 3to 5m sections from tissue microarrays (TMAs) containing two 1.0-mm cores from each tumor, with positive defined as an Allred Score of more than 2 (corresponding to 1% to 10% weakly positive cells). The RT-PCR assays were conducted with RNA extracted from multiple (three to six) 10m sections of intact paraffin blocks at Genomic Health Inc (GHI; Redwood City, CA) using the Oncotype DX assay, which evaluates 21 genes to determine a Recurrence Score (RS). The genes include ER and PR, with definitions of positive (6.5 and 5.5 units, respectively) determined in previous studies. Overall, there was a fairly high concordance of positive and negative results across all assays, ranging from 90% to 93% for ER, and 84% to 88% for PR. Considering a positive result by any assay to be truly positive, the discordant cases favored RT-PCR over central IHC for assessing ER (7.2% of all cases; 91% of RT-PCR positive), central IHC over RT-PCR for PR (10% of all cases; 81% of IHC positive), and discordances were evenly balanced comparing both receptors combined (ie, ER and/or PR positive receptor positive; 6.6% of all cases). All ER and PR assays were strongly associated (P .0001) with relapse in all patients combined, but only marginally in the subset of patients with ER-positive tumors (P .0144 by RT-PCR and P .0913 by IHC). However, the RS was a highly significant (P .0001) predictor of relapse in tamoxifen-treated ER-positive patients, consistent with previous studies. In current medical practice, ER and PR status are critical in determining the use and estimating the benefit of adjuvant hormonal therapy, and there is a widely (and justifiably) held perception that IHC, the present standard methodology for measuring ER and PR status, is unreliable and inaccurate in a substantial proportion of cases. The importance of the Badve et al study stems from its investigation of alternative methods for evaluating receptors that may be more reliable and accurate. In the limited scope of this study, it was successful by demonstrating that RT-PCR is at least equivalent to IHC in its ability to identify receptor-positive cases (considering ER and PR combined), marginally superior in predicting outcome in ER-positive patients, and superior in technical precision, which are all encouraging results. There are many causes for the problems associated with measuring ER and PR by IHC. Some involve preanalytic issues unrelated to IHC itself, such as delayed fixation of tissue, which allows proteins to degrade. Others are analytic in nature, such as the use of diverse reagents with unequal sensitivities, or procedures which inadequately expose proteins masked during tissue fixation. Even postanalytic events may contribute, in the sense that tumors with low levels of receptors (eg, 1% to 10% positive cells) may respond to hormonal therapy, and some laboratories use arbitrary definitions of positive that are too high (eg, 10% positive cells). With newer reagents and protocols, IHC assays for hormone receptors can even be too sensitive, resulting in bimodal results (negative or very high) that are inaccurate because breast cancers show a broad range of receptor expression. RT-PCR does not suffer this shortcoming and is able to more accurately convey this continuum. The central ER IHC results in the Badve et al study were essentially bimodal, suggesting that the assay was saturated, but also that insufficient sensitivity of the assay per se was not the reason that most of the discordant results were positive by RT-PCR but negative by IHC, although some of preanalytic issues alluded to could have resulted in false-negative IHC results. For example, perhaps some of the cases were poorly fixed and RT-PCR is better able to accommodate degraded RNA than IHC to accommodate degraded protein, although this has not been proven. Some of the discordances may have resulted from sampling error caused by the much smaller amount of tissue analyzed by IHC compared with RT-PCR in this study, which may have been inadequate to fully account for the heterogeneity of receptor expression in the tumors. Although some studies suggest that IHC performed on TMAs can closely approximate the results obtained by evaluating much larger JOURNAL OF CLINICAL ONCOLOGY E D I T O R I A L VOLUME 26 NUMBER 15 MAY 2