Incorporating biomarkers to improve statistical power of immunotherapeutic neoadjuvant clinical trials in patients with triple-negative breast cancer.

Neoadjuvant (preoperative) approach to breast cancer treatment has become widely accepted. Traditionally the primary objective of neoadjuvant treatment is to improve subsequent surgical intervention and the effectiveness are often evaluated by its ability achieving complete pathological response (pCR), the eradication of the malignant disease in the breast and axillary lymph nodes. More recently, neoadjuvant treatment has also become recognized as an in vivo, preoperative 'window of opportunity' to explore the efficacy of novel agents such as immunotherapies where a wealth of tumor biomarkers are also routinely collected to quantify antitumor immunity. However, one challenge to combine the traditional pCR and efficacy biomarkers is that these tumor biomarkers are only partially available and cannot be measured in patients who have achieved pCR. In this article, a stepwise hypothesis testing procedure is proposed to combine a continuous tumor biomarker with the conventional binary endpoint in a two-arm randomized phase II superiority trial to improve statistical power. The operating characteristics of proposed procedure is illustrated with a real-world example and the performance is also evaluated numerically.