MTClin: Integrating MammaTyper® to predict neoadjuvant therapy response in HER2+ breast cancer

Aim: Neoadjuvant therapy (NAT) is the standard of care for most HER2-positive early breast cancer (BC). However, 25 to 50% of patients fail to achieve a pathological complete response (pCR). This study evaluates the CE/IVD MammaTyper® kit (Cerca Biotech) as a predictor of response in this setting. Methods: A total of 161 HER2-positive (IHC score 3+) invasive BC patients treated with trastuzumab-based NAT were enrolled. After excluding seven cases (insufficient RNA amount or dual anti-HER2 blockade), 154 FFPE preoperatory core-biopsies were tested: 79 hormone-positive (HR+) and 75 hormone-negative (HR-). Ninety-one achieved a pCR, 63 attained a pathological partial response (pPR). MammaTyper®, an in vitro diagnostic RT-qPCR test, assessed ERBB2, ESR1, PGR, MKI67 mRNA levels. A Python-based Decision Tree Algorithm was used to predict pCR using ΔΔCq values of the four genes alongside tumor size and nodal status (MTClin). The model was further applied to a cT1N0 subgroup and to predict the recurrence risk at 3 and 5 years. Hyperparameters were tuned using GridSearchCV with a 5-fold cross-validation. Performance metrics included sensitivity, specificity, PPV and NPV. Results: Two decision trees were selected for optimal accuracy. In HR+ tumors, the model yielded 90.2 % sensitivity, 86.8 % specificity, 88.1 % PPV and 89.2 % NPV. In HR- tumors, sensitivity was 96 %, specificity 88 %, PPV 94.1 %, and NPV 91.7 %. Sensitivity and NPV were highest in the cT1N0 HR- subgroup. Recurrence prediction yielded AUCs of 0.98 (3 years) and 0.96 (5 years). Conclusions: MTClin may predict pCR and recurrence in HER2-positive BC, supporting tailored escalation or de-escalation treatment.