ABSTRACT Tumors deficient in homologous recombination display high levels of copy number variants
(CNVs) including amplifications, deletions, and breakpoints in their genomes. This homologous recombination
deficiency (HRD) phenotype has been associated with pathogenic BRCA mutations in multiple cancers and
can be exploited clinically through a synthetic lethal interaction with agents, such as inhibitors of poly ADP-
ribose polymerase (PARP) and chemotherapy regimens such as FOLFIRINOX, that increase the level of DNA
damage in tumor genomes. Strikingly, subsets of tumors that are wild type for BRCA1 and BRCA2 have shown
similar clinical responses to these agents. Preliminary studies suggest that responder tumors have elevated
numbers of genomic aberrations even in the absence of variants in the BRCA genes or in other mediators of
homologous recombination mediated DNA repair (e.g. PALB2, ATM, and BRIP1). Thus the nature and extent
of these lesions in tumor genomes, regardless of genotype, provides a biomarker for patients who will
respond to DNA damage and repair targeting therapies.
It is estimated that in 2019, 56,770 Americans will be diagnosed with pancreatic ductal adenocarcinoma
(PDA) and 45,750 will die from the disease, making PDA the third most common cause of cancer death.
Recent clinical trials have made modest improvements in overall survival. Studies suggest that similar to
other tumors, elevated numbers of chromosomal aberrations define a HRD signature in a subset of PDAs.
A fundamental hypothesis is that this genomic signature predicts those PDA patients likely to respond to
PARP inhibitors and to DNA damaging agents. However PDA biopsies are difficult to characterize due to
complex genomes and heterogeneous cellularity, as cancer cells represent on average only 25% of the cells
within the tumor. Furthermore, biopsies frequently contain multiple neoplastic populations that cannot be
distinguished by morphology based methods. To address these clinical challenges we validated DNA content
based flow sorting of PDA tissues. Our published methods yield highly purified (>95%) samples suitable for
whole genome analyses from a variety of clinical samples. These include both fresh frozen and formalin fixed
paraffin embedded (FFPE) tissues with low tumor cell content (<10-20%) and high amounts (>90%) of necrosis
and debris. We have identified PDAs with extensive numbers of interstitial aberrations (IAs) in their genomes
similar to those observed in HRD-positive BRCAmut tumors. Our results suggest that elevated numbers of IAs
correlate with clinical response in PDA. In this study we will establish analytical and processing procedures for
our HRD-IA assay then establish a score that distinguishes HRD+ BRCAmut tumors. We will then exploit
samples treated with FOLFIRINOX to validate the application of our HRD-IA assay to identify those PDA
patients, BRCAmut and BRCAwt, who respond to DNA damage targeting agents. The final phase of this work will
validate our sorting and CNV based HRD-IA assay for CLIA application and clinical use.