Nomogram for prediction of pathological complete response in human epidermal growth factor 2 (HER22) receptor positive breast cancer patients treated with neoadjuvant systemic therapy

Objective: Our primary objective was to make a nomogram to predict the pathologic complete response (pCR) rate in primary HER2-positive breast cancer patients treated with neoadjuvant systemic therapy. Our hypothesis is high HER2 FISH ratio and low ER expression level is associated with high pCR rate in HER2 positive breast cancer patients who treated with NST with trastuzumab. ^ Methods: This study is a retrospective single-center study approved by IRB at The University of Texas MD Anderson Cancer Center and the University of Texas Health Science Center at Houston School of Public Health. Patients with histologically confirmed Stage I-III human epidermal growth factor 2 (HER2) positive invasive breast carcinoma who received neoadjuvant systemic chemotherapy were included in this study. Variables collected from patient charts and database include demographics (age, race, menopausal status, and body mass index), date of death/recurrence/last follow-up, estrogen receptor (ER)/progesterone receptor (PR) status as continuous variable from 0 to 100%, HER2 status (FHER2/CEP FISH ratio), histology (histological subtype, nuclear grade) from diagnostic biopsy specimen, clinical stage, neoadjuvant chemotherapy regimens, and response to neoadjuvant systemic therapy (pCR/non-pCR). ^ Descriptive statistics were used to summarize all variables collected in this study. The relationship between the pCR and covariates was estimated by using a univariable logistic regression model followed by a multivariable logistic regression model to make a prediction model. Backward stepwise method was used for variable selection. Upon construction of a model, the discriminatory abilities of the models was assessed using Harrel Concordance-index (c-index) using both training and validation samples. ^ Results: A total of 733 patients diagnosed with primary HER2 positive invasive breast cancer at MD Anderson Cancer Center from January 1, 1999 to June 30, 2013 were included and were divided into two cohorts with a 2:1 ratio; training set (n=489) and validation set (n=244). Those patients are well distributed and there is no significant difference between training and validation sets. Univariable analysis showed low estrogen receptor (ER) level (ER; Odds ratio 0.99, 95%CI 0.98-0.99 p\u3c0.01) are strongly related to high rate of pCR. There was a trend that inflammatory breast cancer (IBC) yielded low odds of pCR (OR 0.54 95%CI 0.27-1.306 p=0.07) and that high HER2/CEP ratio provided higher probability of pCR (OR 1.06 95%CI 1.0-1.12 p=0.05). By using backward stepwise methods including all the variables, ER, HER2 FISH ratio, IBC status, and neoadjuvant chemotherapy regimen remained significant. By using those covariates, a nomogram to predict pCR was constructed. The prediction model had a concordance index of 0.71 (95%CI, 0.67-0.76) in training set and 0.65 (95%CI, 0.58-0.72) in validation set. Calibration plot showed a good accuracy. ^ Conclusion: We have shown that high HER2 FISH ratio and low ER expression level are associated with high pCR rate in HER2 positive breast cancer patients who treated with NST. Our findings provided a well validated nomogram considering ER expression level, HER2 FISH ratio, IBC or non-IBC status, and regimen to predict pathological compete response after neoadjuvant systemic therapy in Stage I-III newly diagnosed invasive breast cancer. By using this model, we can potentially stratify the HER2 positive patients treated neoadjuvant systemic therapy who would be most likely or least likely to achieve pCR. For patients for whom the probability of pCR is low, we may need to investigate additional treatment option. External validation in prospective manner is warranted to test the clinical applicability.