Background: The treatment landscape of triple negative breast cancer (TNBC) has rapidly changed in recent years, with immuno-oncology (IO) therapies, such as immune checkpoint inhibitors, leading to improvement in patient outcomes. However, the number of patients who benefit is less than 15%, and the therapy itself can prompt severe immune-related adverse events. Breast cancer tumor volume measurements derived from magnetic resonance imaging (MRI) have proven to be a strong predictor of outcome (survival) in patients who either achieve pathological complete response (pCR) or have residual cancer burden (RCB) (ACRIN6657/CALGB150007 I-SPY1). Using an integrative computational approach, we developed both the TumorIO biomarker and the TumorIO Score to predict response to IO therapy in breast cancer patients. Here, we assess the volumetric response in IO-treated patients and its relationship with the TumorIO Score.
Methods: We assessed the relationship between breast tumor volumetric response to IO therapy and the TumorIO Score, a previously developed metric that relates probability of IO responsiveness to spatially-resolved tumor biology. We grouped patients into TumorIO Score high vs. low populations, and compared final volumes assessed by MRI and percent response to therapy between groups.
Results: We analyzed the volumetric response of IO-treated patients from the I-SPY2 trial and from an independent cohort (total n = 67). Over 96% of the IO-treated patients in the I-SPY2 trial (n = 55) had a greater than 90% volumetric response, while in the independent cohort (n = 12) only 50% of patients had greater than 90% volumetric response. Across both cohorts, 95.8% of patients with a high TumorIO Score achieved > 90% reduction in tumor volume. Additionally, we identified three patients in our cohort that appeared not to respond to IO therapy, who were later classified as progressor, pseudo-progressor, and non-responder patients. The TumorIO biomarker was able to differentiate these three patients, as it associates a unique distinctive biological phenomenon with either response or lack of response.
Conclusions: The TumorIO Score provides a novel technology to address several challenges in neoadjuvant IO therapy, namely: predicting which patients are likely to have pCR in response to IO therapy, as well as identifying patients likely to have a substantial volumetric response to IO. Additionally, this method may facilitate subcategorizing patients likely to show radiological pseudo-progression prior to IO administration, although a larger cohort is needed to validate these findings.
View publication here.
