Copy number polymorphisms and anticancer pharmacogenomics

Abstract

Background: Recent studies have investigated the contribution of copy number variants (CNVs) to disease susceptibility in a multitude of complex disorders, including systemic lupus erythematosus, Crohn's disease, and various neurodevelopmental disorders. Relatively few CNV studies, however, have been conducted on pharmacologic phenotypes even though these structural variants are likely to play an important role. We developed a genome-wide method to identify CNVs that contribute to heterogeneity in drug response, focusing on drugs that are widely used in anticancer treatment regimens.

Results: We conducted a comprehensive genome-wide study of CNVs from population-scale array-based and sequencing-based surveys by analyzing their effect on cellular sensitivity to platinating agents and topoisomerase II inhibitors. We identified extensive CNV regions associated with cellular sensitivity to functionally diverse chemotherapeutics, supporting the hypothesis that variation in copy number contributes to variation in drug response. Interestingly, although single nucleotide polymorphisms (SNPs) tag some of the CNVs associated with drug sensitivity, several of the most significant CNV-drug associations are independent of SNPs; consequently, they represent genetic variations that have not been previously interrogated by SNP studies of pharmacologic phenotypes.

Conclusions: Our findings demonstrate that pharmacogenomic studies may greatly benefit from the study of CNVs as expression quantitative trait loci, thus contributing broadly to our understanding of the complex traits genetics of CNVs. We also extend our PACdb resource, a database that makes available to the scientific community relationships between genetic variation, gene expression, and sensitivity to various drugs in cell-based models.

Figure 1

Three transcripts cis-regulated by CNVs identified by whole genome sequencing data predict cellular sensitivity to functionally diverse chemotherapeutics. Through ENCODE/Gencode annotation, 376 genes were recently identified [12] to be cis-regulated by CNV deletions. At P < 0.05, we found 21 gene expression correlations with carboplatin, 15 with cisplatin, 23 with daunorubicin, and 21 with etoposide. Three transcripts (MOXD1 on chromosome 6, RCC1 on chromosome 1, SULF2I on chromosome 20) were significant after multiple testing adjustment (p_adj < 0.05) and were the only CNV deletions associated with all four agents at the nominal P < 0.05 threshold. The plots show level of expression (x axis) and IC₅₀ (y axis).

Figure 2

A schematic of the approach to the discovery of drug susceptibility-associated CNVs. We added to PACdb the CNV associations with cellular sensitivity to the chemotherapeutic agents. Note that some drug-associated CNVs (dCNVs) are poorly interrogated by SNPs and are thus independent of drug-associated SNPs (dSNPs). SCAN includes the results of eQTL mapping of CNVs to transcriptional expression.

Figure 3

A CNV associated with cellular sensitivity to both carboplatin and cisplatin. We identified an amplification CNV, CNVR7826_full on chromosome 20, that is associated with both carboplatin sensitivity (a) (P = 0.0056) and cisplatin sensitivity (b) (P = 0.0023). The x-axis is the copy number (3 or 4) for the CNV; the y-axis is the log₂-transformed IC₅₀.

Figure 4

The relationship between CCND1 expression and etoposide IC₅₀ in the discovery set and in an independent replication set of 52 CEPH LCLs. (a) Discovery set; (b) independent replication set of 52 CEPH LCLs. Panel (b) describes the expression- IC₅₀ relationship using real-time expression; the relationship is consistent with that in the discovery set of CEU samples.