Assessment of PIK3CA mutations in human epidermal growth factor receptor 2-positive breast cancer: clinical validity but not utility.

Abstract

The use of anti–human epidermal growth factor receptor 2 (HER2) targeted agents in addition to cytotoxic chemotherapy has had a substantial and positive impact on the survival of patients with HER2-overexpressing or -amplified breast cancer. However, treatment is not effective for every patient. Therefore, over the last decade or more there has been a quest to identify biomarkers that predict benefit, or more importantly, lack of benefit, from anti-HER2 therapies. Unfortunately, this quest has been unsuccessful, and so we continue to treat all patients with HER2-positive breast cancer with costly, inconvenient, and occasionally toxic therapies to ensure that we include the roughly one third who will benefit with improved diseasefree and overall survival. Indeed, the most recent update of the American Society of Clinical Oncology/College of American Pathologists HER2 guidelines were admittedly directed toward sensitivity over specificity of the assays, given that these treatments have been so effective and are reasonably well tolerated. In the article that accompanies this editorial, Loibl et al from the German Breast Group have evaluated the association between presence of PIK3CA mutations and response to anti-HER2 therapy by analyzing specimens collected from patients who participated in three previously conducted, prospective neoadjuvant clinical trials. Their data indicate that the presence of PIK3CA mutations is associated with a poorer pathologic complete response (pCR) in patients who received trastuzumab, lapatinib, or a combination of the two. Although intriguing, the exact clinical application of these data requires a thoughtful discussion of their meaning. Substantial attention has been paid to the development and testing of new therapeutic agents, but less rigor has been applied when identifying and validating tumor biomarkers. However, recent efforts to generate the same sense of value for tumor biomarkers that we have for anticancer treatments have gained traction. It is important to understand that a tumor biomarker is an indication of cancer biology and behavior, but that there may be one or more tests for that biomarker, and that they may not provide identical, or even similar, results. In this regard, a tumor biomarker test should not be incorporated into routine clinical care until it has cleared a number of hurdles, best articulated by the Evaluation of Genomic Applications in Practice and Prevention Working Group of the Centers for Disease Control and Prevention. First, a tumor biomarker must be demonstrated to have analytic validity. In other words, the assay must accurately, reliably, and reproducibly measure what it is intended to measure. Second, the tumor biomarker test must have clinical (or biologic) validity, which implies that it separates one population into two or more groups with distinctly different outcomes. However, for a tumor biomarker test to be incorporated into standard of care, it must also be shown to have clinical utility. Establishment of clinical utility requires a high level of evidence from either prospective or prospective–retrospective studies demonstrating that application of the marker to direct therapy improves patient outcomes with sufficient magnitude to justify testing all eligible patients. How should these high levels of evidence be generated? Ideally, as with new therapeutics, one would perform a prospective trial in which the clinical utility of the tumor biomarker test is the primary objective. Several authors have published various trial designs to address this issue. However, unlike new therapeutics, it might be possible to generate high levels of evidence for a tumor biomarker test using archived specimens. Simon et al have proposed criteria for such studies. These criteria are rigorous, requiring that the archived specimens be collected from patients who participated in previously conducted clinical trials that addressed a use to which the tumor biomarker test might be applied, and that a prospectively written correlative science protocol be prepared with a set study design and analytic plan. These criteria raise several points. It does seem that the assay for PIK3CA mutations has high analytic validity. However, we must ask: what is the intended use of the tumor biomarker assay? In the accompanying article, Loibl et al investigated the apparent prediction of anti-HER2 therapy benefit or lack of benefit. However, their clinical trials were not designed to address this question. In other words, patients were not randomly assigned to receive or not receive antiHER2 therapy, nor were they given single-modality anti-HER2 therapy, without chemotherapy, to determine if response was a result of the anti-HER2 therapy or the chemotherapy. Just as importantly, we must ask: what is the study end point? Although response is a convenient surrogate end point to determine JOURNAL OF CLINICAL ONCOLOGY E D I T O R I A L VOLUME 32 NUMBER 29 OCTOBER 1