Background:
Immuno-oncology (IO) therapies, such as immune checkpoint inhibitor antibodies, have emerged as promising treatments for early-stage breast cancer (ESBC). Although immunotherapy is standard-of-care for ESBC, the number of benefiting patients remains small, and the therapy itself can prompt severe immune-related adverse events. To date, efforts to identify biomarkers of IO response, such as those targeting the PD-1/PD-L1 signaling axis, have failed to robustly demarcate predictive features that translate across cohorts. Here, we applied an integrative computational approach to develop the IOScope biomarker, which predicts IO response in ESBC. Our method overcomes many of the limitations of invasive, costly, and time-consuming IO response analyses derived from transcriptomics data, and advances beyond the inability of single-site biopsies to account for tumor heterogeneity.
Methods:
We analyzed single-cell and whole-tissue RNA-seq data from non-IO treated ESBC patients in the I-SPY trials, as well as 23 manually-curated gene expression signatures representing the “hallmarks of cancer”, in order to associate PD-L1 levels with local tumor biology. We then correlated PD-L1 expression to biophysical features derived from patient DCE-MRIs to generate the IOScope biomarker. We quantified IOScope within patient tumors using integrative computational modeling and developed a corresponding IOScope Score. The IOScope biomarker and Score were then applied to data from IO-treated patients in the I-SPY2 clinical trial to assess their performance in predicting pCR in response to IO therapy. We additionally validated IOScope in a virtual clinical trial, comparing pCR rates to empirical data.
Results:
We validated the IOScope biomarker and the IOScope Score in a small, independent cohort of IO-treated patients (n = 17) and correctly predicted pCR response in 15/17 individuals (88.2% accuracy, comprising 10/12 in TNBC and 5/5 in HR+/HER2-). We further applied the IOScope Score in a virtual clinical trial (n = 292) simulating IO administration in IO-naïve cohort (Virtual Tumor Bank, SimBioSys, Inc.). Using this approach, we predicted an increase in pCR rates of 28.5% (TNBC) and 8.9% (HR+/HER2-) with the addition of IO therapy to a standard chemotherapy background. These predicted increases in pCR rate compare favorably to published rates increases produced by IO in both cancer subtypes.
Conclusion:
The IOScope biomarker and IOScope Score represent a next generation approach using integrative computational and biophysical analysis to assess cancer responsiveness to immunotherapy. This biomarker, predicated on MRI-derived tumor microenvironmental features, allows for immuno-oncology profiling of tumors in a rapid, non-invasive manner and may confer high clinical decision impact to further enable personalized oncologic care.
View publication here.
