Abstract Numerous oncology drugs have pharmacogenomics warnings on their FDA labels, yet pharmacogenomics screening is not routinely applied in clinical practice. Pharmacogenomics testing is used to reduce the chance of drug-induced toxicities, improve patient outcomes, and reduce treatment costs by tailoring therapies to the patient's genotype. Here we present the NantOmics pharmacogenomics test, which uses whole genome (DNA) and whole exome sequencing data from FFPE tumors and matched normal samples to predict and advise physicians of possible drug-induced toxicities. Using FDA labels and Clinical Pharmacogenomics Implementation Consortium (CPIC) guidelines, we developed a clinical pharmacogenomics panel including 31 markers in 11 genes linked to toxicities from 14 cancer therapies, as well as a research panel of 13 gene-drug pairs reported in primary literature. The test screens for both germline and somatic variants to predict how an oncology patient may respond to routinely-used chemotherapies. Several genes in our panel, including CYP2D6 and other CYP family members, prove challenging to properly genotype due to high levels of polymorphism, multiple similar haplotypes, and structural variations. We have established a kmer-based approach to accurately determine haplotypes for these genes, as well as coverage-based methods to infer hybridizations and copy number variations. For pharmacogenes with ambiguous genotypes comprising multiple heterozygous SNPs, we utilize any allele fraction imbalance observed at these loci in the tumor DNA and/or RNA sequencing data to disambiguate the correct genotype. The test was run on 1,172 patient samples, 95.8% of which contained a variant in at least one gene from our panel. Furthermore, 7.5% of patients had genomic variants associated with severe or life-threatening drug toxicities. For all alleles in our clinical panel, we observed similar allele frequencies to those reported in the ExAC database. Our test was validated on several cell lines from the CDC GeT-RM, and was able to accurately determine the genotypes for all genes present in our test panels in each of the cell lines tested. To further validate our test, we ran it on a cohort of patients previously genotyped by an independent CLIA-validated PCR-based panel, as well as on a set of synthetic data. In all validation studies, we were able to demonstrate that the test detects each variant in our panel, and correctly determines patient genotype in all studied cases. Given the high percentage of patients with potentially treatment-altering genomic variants, these results underscore the need for more routine pharmacogenomics screening in the oncological setting. Citation Format: Camille Schwartz, John Little, Charles Vaske, Stephen Benz, Patrick Soon-Shiong, Shahrooz Rabizadeh, J. Zachary Sanborn. The NantOmics Pharmacogenomics Test: An integrative panomic approach to pharmacogenomics screening [abstract]. In: Proceedings of the American Association for Cancer Research Annual Meeting 2018; 2018 Apr 14-18; Chicago, IL. Philadelphia (PA): AACR; Cancer Res 2018;78(13 Suppl):Abstract nr 3888.
Read full abstract