Abstract
Breast cancer is the most diagnosed cancer and the second leading cause of cancer death in women in
the USA. Immune checkpoint therapy (ICI) is revolutionizing the therapy of a number of malignancies, including
triple negative breast cancer (TNBC). Therapeutic blockade of immune checkpoints, such as by anti-PD-1,
removes the tumor-initiated suppression of the immune system and unleashes prolonged anti-tumor immunity.
Despite the encouraging success, many patients develop severe and sometimes life-threating adverse effects,
or many also fail to benefit from immunotherapy. Existing immune therapy response predictive markers have
only modest positive predictive values and there are no clinically useful markers of toxicity. Our proposed study
will focus on comprehensive evaluation of serological auto-antibodies (AAb) as potential predictors of immune
related adverse event (irAE) and benefit from therapy. Serum based biomarkers exploit easy sample
accessibility and directly measure immune responses. Our platform can be rapidly adopted for clinical use.
Our study leverages collaboration between experts on immunoproteomics and biomarker discovery and
physician scientists specializing in breast cancer. Our study will analyze more than 3000 longitudinal serum
samples collected from 1,195 patients enrolled in a randomized phase III trial (SWOG S1418) that tests the
efficacy of one year of single agent adjuvant pembrolizumab (anti-PD-1 antibody) therapy compared to
observation in high risk TNBC. Serum samples were collected at baseline and at 13 and 55 weeks after
starting therapy, all samples are annotated with detailed toxicity and outcome information that were collected
during this FDA registration trial. We will employ two complementary high-throughput antibody profiling
technologies including Nucleic Acid Programmable Protein Array (NAPPA) and Multiple In Solution Antibody
Assay (MISPA) that we developed. NAPPA allows the detection of antibodies against tens of thousands freshly
produced full length proteins in hundreds of samples while MISPA enables more precise quantification of
hundreds of antibodies in thousands of samples and could be scaled for clinical use. Our approach involves (i)
a discovery step using the NAPPA arrays that will interrogate serum samples for AAbs against 18,000 human
proteins in the discovery cohort, (ii) a verification step of identified outcome-predictive AAb using the MISPA
platform using the discovery samples to select the final set of candidate markers, (iii) and blinded validation of
candidate AAb using MISPA on the held out validation set. Identification of biomarkers that can predict
clinically-relevant immune-related adverse events or recurrence, and hence inefficacy of therapy, would help
better estimate the risk benefit ratio of these very expensive and potentially toxic therapies. Predictors of
adverse events could also influence patient monitoring and allow earlier therapeutic intervention. The lessons
learned from this proposed study on TNBC will likely impact a broad spectrum of other cancers since adverse
events are not cancer type dependent.
