BackgroundPreoperative therapy with chemotherapy and the HER2-targeted monoclonal antibody trastuzumab is valuable for patients with large or locally advanced HER2-positive (HER2+) breast cancers but traditional methods of measuring HER2 expression do not accurately stratify patients for likelihood of response. Quantitative immunofluorescent approaches have the potential to provide a mathematically continuous measure of HER2. Here we seek to determine whether quantitative measurement of HER2 or phospho-HER2 correlates with likelihood of response to trastuzumab- containing neoadjuvant therapy.MethodsWe evaluated core biopsy samples from 27 HER2+ breast cancer patients enrolled in a preoperative clinical trial using trastuzumab, nab-paclitaxel and carboplatin combination therapy (BrUOG BR-211B (NCT00617942)). Tumor core biopsies were taken before initiation of treatment and 9–13 days after patients received "run-in" doses of either single agent trastuzumab or nab-paclitaxel. The AQUA method of quantitative immunofluorescence was used for analysis of in situ protein expression. Patients then received 18 weeks of treatment, followed by surgery to assess pathologic response to the neoadjuvant regimen.ResultsA HER2 score of 2111 by AQUA analysis has been shown to be equivalent to HER2 3+ by immunohistochemical staining in previous studies. Of 20 evaluable patients, 10 cases who achieved a pathologic complete response (pathCR) with neoadjuvant treatment had a mean HER2 level of 10251 compared with 4766 in the patients without pathCR (p = 0.0021). Measurement of phospho-HER2 showed no difference in pathCR vs non-pathCR groups. In 9 patients who had HER2 levels repeated after a single treatment with trastuzumab there was no evidence of a reduction in the HER2 or phospho-HER2 levels following that exposure.ConclusionsHigh levels of HER2 are associated with achievement of a pathCR in the preoperative setting, while levels of Phospho-HER2 were not predictive of response. This data suggests that accurate measurement of HER2 may help determine the likelihood of response in the pre-surgical setting. Further validation in larger cohorts is required, but this pilot data shows the feasibility of this approach.