Genetically regulated expression underlies cellular sensitivity to chemotherapy in diverse populations.

Most cancer chemotherapeutic agents are ineffective in a subset of patients; thus, it is important to consider the role of genetic variation in drug response. Lymphoblastoid cell lines (LCLs) in 1000 Genomes Project populations of diverse ancestries are a useful model for determining how genetic factors impact the variation in cytotoxicity. In our study, LCLs from three 1000 Genomes Project populations of diverse ancestries were previously treated with increasing concentrations of eight chemotherapeutic drugs, and cell growth inhibition was measured at each dose with half-maximal inhibitory concentration (IC50) or area under the dose-response curve (AUC) as our phenotype for each drug. We conducted both genome-wide association studies (GWAS) and transcriptome-wide association studies (TWAS) within and across ancestral populations. We identified four unique loci in GWAS and three genes in TWAS to be significantly associated with the chemotherapy-induced cytotoxicity within and across ancestral populations. In the etoposide TWAS, increased STARD5 predicted expression associated with decreased etoposide IC50 (P = 8.5 × 10-8). Functional studies in A549, a lung cancer cell line, revealed that knockdown of STARD5 expression resulted in the decreased sensitivity to etoposide following exposure for 72 (P = 0.033) and 96 h (P = 0.0001). By identifying loci and genes associated with cytotoxicity across ancestral populations, we strive to understand the genetic factors impacting the effectiveness of chemotherapy drugs and to contribute to the development of future cancer treatment.

Ototoxicity After Cisplatin-Based Chemotherapy: Factors Associated With Discrepancies Between Patient-Reported Outcomes and Audiometric Assessments.

OBJECTIVES: To provide new information on factors associated with discrepancies between patient-reported and audiometrically defined hearing loss (HL) in adult-onset cancer survivors after cisplatin-based chemotherapy (CBCT) and to comprehensively investigate risk factors associated with audiometrically defined HL. DESIGN: A total of 1410 testicular cancer survivors (TCS) ≥6 months post-CBCT underwent comprehensive audiometric assessments (0.25 to 12 kHz) and completed questionnaires. HL severity was defined using American Speech-Language-Hearing Association criteria. Multivariable multinomial regression identified factors associated with discrepancies between patient-reported and audiometrically defined HL and multivariable ordinal regression evaluated factors associated with the latter. RESULTS: Overall, 34.8% of TCS self-reported HL. Among TCS without tinnitus, those with audiometrically defined HL at only extended high frequencies (EHFs) (10 to 12 kHz) (17.8%) or at both EHFs and standard frequencies (0.25 to 8 kHz) (23.4%) were significantly more likely to self-report HL than those with no audiometrically defined HL (8.1%) [odds ratio (OR) = 2.48; 95% confidence interval (CI), 1.31 to 4.68; and OR = 3.49; 95% CI, 1.89 to 6.44, respectively]. Older age (OR = 1.09; 95% CI, 1.07 to 1.11, p < 0.0001), absence of prior noise exposure (OR = 1.40; 95% CI, 1.06 to 1.84, p = 0.02), mixed/conductive HL (OR = 2.01; 95% CI, 1.34 to 3.02, p = 0.0007), no hearing aid use (OR = 5.64; 95% CI, 1.84 to 17.32, p = 0.003), and lower education (OR = 2.12; 95% CI, 1.23 to 3.67, p = 0.007 for high school or less education versus postgraduate education) were associated with greater underestimation of audiometrically defined HL severity, while tinnitus was associated with greater overestimation (OR = 4.65; 95% CI, 2.64 to 8.20 for a little tinnitus, OR = 5.87; 95% CI, 2.65 to 13.04 for quite a bit tinnitus, and OR = 10.57; 95% CI, 4.91 to 22.79 for very much tinnitus p < 0.0001). Older age (OR = 1.13; 95% CI, 1.12 to 1.15, p < 0.0001), cumulative cisplatin dose (>300 mg/m2, OR = 1.47; 95% CI, 1.21 to 1.80, p = 0.0001), and hypertension (OR = 1.80; 95% CI, 1.28 to 2.52, p = 0.0007) were associated with greater American Speech-Language-Hearing Association-defined HL severity, whereas postgraduate education (OR = 0.58; 95% CI, 0.40 to 0.85, p = 0.005) was associated with less severe HL. CONCLUSIONS: Discrepancies between patient-reported and audiometrically defined HL after CBCT are due to several factors. For survivors who self-report HL but have normal audiometric findings at standard frequencies, referral to an audiologist for additional testing and inclusion of EHFs in audiometric assessments should be considered.

Clinical and genetic risk factors for radiation-associated ototoxicity: A report from the Childhood Cancer Survivor Study and the St. Jude Lifetime Cohort.

BACKGROUND: Cranial radiation therapy (CRT) is associated with ototoxicity, which manifests as hearing loss and tinnitus. The authors sought to identify clinical determinants and genetic risk factors for ototoxicity among adult survivors of pediatric cancer treated with CRT. METHODS: Logistic regression evaluated associations of tinnitus (n = 1991) and hearing loss (n = 2198) with nongenetic risk factors and comorbidities among CRT-treated survivors in the Childhood Cancer Survivor Study. Genome-wide association studies (GWASs) of CRT-related tinnitus and hearing loss were also performed. RESULTS: Males were more likely to report CRT-related tinnitus (9.4% vs 5.4%; P = 5.1 × 10-4 ) and hearing loss (14.0% vs 10.7%; P = .02) than females. Survivors with tinnitus or hearing loss were more likely to experience persistent dizziness or vertigo (tinnitus: P < 2 × 10-16 ; hearing loss: P = 6.4 × 10-9 ), take antidepressants (tinnitus: P = .02; hearing loss: P = .01), and report poorer overall health (tinnitus: P = 1.5 × 10-6 ; hearing loss: P = 1.7 × 10-6 ) in comparison with controls. GWAS of CRT-related tinnitus revealed a genome-wide significant signal in chromosome 1 led by rs203248 (P = 1.5 × 10-9 ), whereas GWAS of CRT-related hearing loss identified rs332013 (P = 5.8 × 10-7 ) in chromosome 8 and rs67522722 (P = 7.8 × 10-7 ) in chromosome 6 as nearly genome-wide significant. A replication analysis identified rs67522722, intronic to ATXN1, as being significantly associated with CRT-related hearing loss (P = .03) and de novo hearing loss (P = 3.6 × 10-4 ). CONCLUSIONS: CRT-associated ototoxicity was associated with sex, several neuro-otological symptoms, increased antidepressant use, and poorer self-reported health. GWAS of CRT-related hearing loss identified rs67522722, which was supported in an independent cohort of survivors. LAY SUMMARY: Hearing loss and subjective tinnitus (the perception of noise or ringing in the ear) are long-term side effects of cancer treatment and are common in children treated with radiation to the brain. These toxicities can affect childhood development and potentially contribute to serious learning and behavioral difficulties. This study's data indicate that males are at greater risk for hearing loss and tinnitus than females after radiation therapy to the brain. Those who develop these toxicities are more likely to use antidepressants and report poorer overall health. Health care providers can improve the management of survivors by informing patients and/or their parents of these risks.

Clinical evaluation of germline polymorphisms associated with capecitabine toxicity in breast cancer: TBCRC-015.

PURPOSE: Capecitabine is important in breast cancer treatment but causes diarrhea and hand-foot syndrome (HFS), affecting adherence and quality of life. We sought to identify pharmacogenomic predictors of capecitabine toxicity using a novel monitoring tool. METHODS: Patients with metastatic breast cancer were prospectively treated with capecitabine (2000 mg/m2/day, 14 days on/7 off). Patients completed in-person toxicity questionnaires (day 1/cycle) and automated phone-in assessments (days 8, 15). Correlation of genotypes with early and overall toxicity was the primary endpoint. RESULTS: Two hundred and fifty-nine patients were enrolled (14 institutions). Diarrhea and HFS occurred in 52% (17% grade 3) and 69% (9% grade 3), respectively. Only 29% of patients completed four cycles without dose reduction/interruption. In 39%, the highest toxicity grade was captured via phone. Three single nucleotide polymorphisms (SNPs) associated with diarrhea-DPYD*5 (odds ratio [OR] 4.9; P = 0.0005), a MTHFR missense SNP (OR 3.3; P = 0.02), and a SNP upstream of MTRR (OR 3.0; P = 0.03). GWAS elucidated a novel HFS SNP (OR 3.0; P = 0.0007) near TNFSF4 (OX40L), a gene implicated in autoimmunity including autoimmune skin diseases never before implicated in HFS. Genotype-gene expression analyses of skin tissues identified rs11158568 (associated with HFS via GWAS) with expression of CHURC1, a transcriptional activator controlling fibroblast growth factor (beta = - 0.74; P = 1.46 × 10-23), representing a previously unidentified mechanism for HFS. CONCLUSIONS: This is the first cancer pharmacogenomic study to use phone-in self-reporting, permitting augmented toxicity characterization. Three germline toxicity SNPs were replicated, and several novel SNPs/genes having strong functional relevance were discovered. If further validated, these markers could permit personalized capecitabine dosing.

Integration of genetic and functional genomics data to uncover chemotherapeutic induced cytotoxicity.

Identifying genetic variants associated with chemotherapeutic induced toxicity is an important step towards personalized treatment of cancer patients. However, annotating and interpreting the associated genetic variants remains challenging because each associated variant is a surrogate for many other variants in the same region. The issue is further complicated when investigating patterns of associated variants with multiple drugs. In this study, we used biological knowledge to annotate and compare genetic variants associated with cellular sensitivity to mechanistically distinct chemotherapeutic drugs, including platinating agents (cisplatin, carboplatin), capecitabine, cytarabine, and paclitaxel. The most significantly associated SNPs from genome wide association studies of cellular sensitivity to each drug in lymphoblastoid cell lines derived from populations of European (CEU) and African (YRI) descent were analyzed for their enrichment in biological pathways and processes. We annotated genetic variants using higher-level biological annotations in efforts to group variants into more interpretable biological modules. Using the higher-level annotations, we observed distinct biological modules associated with cell line populations as well as classes of chemotherapeutic drugs. We also integrated genetic variants and gene expression variables to build predictive models for chemotherapeutic drug cytotoxicity and prioritized the network models based on the enrichment of DNA regulatory data. Several biological annotations, often encompassing different SNPs, were replicated in independent datasets. By using biological knowledge and DNA regulatory information, we propose a novel approach for jointly analyzing genetic variants associated with multiple chemotherapeutic drugs.

Adverse Health Outcomes in Relationship to Hypogonadism After Chemotherapy: A Multicenter Study of Testicular Cancer Survivors.

BACKGROUND: This study examined the prevalence of hypogonadism, its clinical and genetic risk factors, and its relationship to adverse health outcomes (AHOs) in North American testicular cancer survivors (TCS) after modern platinum-based chemotherapy. PATIENTS AND METHODS: Eligible TCS were <55 years of age at diagnosis and treated with first-line platinum-based chemotherapy. Participants underwent physical examinations and completed questionnaires regarding 15 AHOs and health behaviors. Hypogonadism was defined as serum testosterone levels ≤3.0 ng/mL or use of testosterone replacement therapy. We investigated the role of 2 single nucleotide polymorphisms (rs6258 and rs12150660) in the sex hormone-binding globulin (SHBG) locus implicated in increased hypogonadism risk in the general population. RESULTS: Of 491 TCS (median age at assessment, 38.2 years; range, 18.7-68.4 years), 38.5% had hypogonadism. Multivariable binary logistic regression analysis identified hypogonadism risk factors, including age at clinical evaluation (odds ratio [OR], 1.42 per 10-year increase; P= .006) and body mass index of 25 to <30 kg/m2 (OR, 2.08; P= .011) or ≥30 kg/m2 (OR, 2.36; P= .005) compared with <25 kg/m2. TCS with ≥2 risk alleles for the SHBG SNPs had a marginally significant increased hypogonadism risk (OR, 1.45; P= .09). Vigorous-intensity physical activity appeared protective (OR, 0.66; P= .07). Type of cisplatin-based chemotherapy regimen and socioeconomic factors did not correlate with hypogonadism. Compared with TCS without hypogonadism, those with hypogonadism were more likely to report ≥2 AHOs (65% vs 51%; P= .003), to take medications for hypercholesterolemia (20.1% vs 6.0%; P<.001) or hypertension (18.5% vs 10.6%; P= .013), and to report erectile dysfunction (19.6% vs 11.9%; P= .018) or peripheral neuropathy (30.7% vs 22.5%; P= .041). A marginally significant trend for increased use of prescription medications for either diabetes (5.8% vs 2.6%; P= .07) or anxiety/depression (14.8% vs 9.3%; P= .06) was observed. CONCLUSIONS: At a relatively young median age, more than one-third of TCS have hypogonadism, which is significantly associated with increased cardiovascular disease risk factors, and erectile dysfunction. Providers should screen TCS for hypogonadism and treat symptomatic patients.

Cancer pharmacogenomics: strategies and challenges.

Genetic variation influences the response of an individual to drug treatments. Understanding this variation has the potential to make therapy safer and more effective by determining selection and dosing of drugs for an individual patient. In the context of cancer, tumours may have specific disease-defining mutations, but a patient's germline genetic variation will also affect drug response (both efficacy and toxicity), and here we focus on how to study this variation. Advances in sequencing technologies, statistical genetics analysis methods and clinical trial designs have shown promise for the discovery of variants associated with drug response. We discuss the application of germline genetics analysis methods to cancer pharmacogenomics with a focus on the special considerations for study design.

Racial disparities in omission of oncotype DX but no racial disparities in chemotherapy receipt following completed oncotype DX test results.

PURPOSE: To examine racial/ethnic disparities in Oncotype DX (ODX) testing among patients with node-negative, estrogen receptor (ER)-positive, human epidermal growth factor receptor 2 (HER2)-negative breast cancers and possible racial/ethnic disparities in chemotherapy receipt following ODX testing within Recurrence Score (RS) category (Not Done, Low, Intermediate, High), as well as chemotherapy receipt time trends within RS categories. METHODS: A retrospective cohort list of 125,288 women who were potentially indicated for ODX testing from 2010 to 2014 was obtained using the National Cancer Database. We fit multivariate logistic regression predicting chemotherapy receipt, adjusting for clinical factors, patient demographic factors, and hospital-level factors, separately by RS category, and calculated odds ratios (OR) and 95% confidence intervals (CI), as well as time trends. RESULTS: Overall, ODX testing was completed for 46.1% of Non-Hispanic (NH) Whites, 43.9% of NH Blacks, and 41.7% of Hispanics. Among patients who did not receive ODX testing, NH Black and Hispanic women both experienced statistically significant increases in chemotherapy receipt relative to NH White women (NH Black OR 1.23; 95% CI 1.11-1.37; Hispanic OR 1.23; 95% CI 1.07-1.42). However, among patients with ODX results, no statistically significant racial/ethnic differences in chemotherapy receipt were observed within strata of RS category. Trend analyses demonstrated increasing adherence to national guidelines for ODX testing. CONCLUSIONS: We identified racial disparities in omission of ODX testing but no differences in chemotherapy receipt if ODX test results were obtained, suggesting increasing access to ODX testing may improve racial equality in efficacious use of adjuvant chemotherapy for ER-positive HER2-negative breast cancer.

Identification of Novel Protein Expression Changes Following Cisplatin Treatment and Application to Combination Therapy.

Determining the effect of chemotherapeutic treatment on changes in protein expression can provide important targets for overcoming resistance. Due to challenges in simultaneously measuring large numbers of proteins, a paucity of data exists on global changes. To overcome these challenges, we utilized microwestern arrays that allowed us to measure the abundance and modification state of hundreds of cell signaling and transcription factor proteins in cells following drug exposure. HapMap lymphoblastoid cell lines (LCLs) were exposed to cisplatin, a chemotherapeutic agent commonly used to treat testicular, head and neck, non-small cell lung, and gynecological cancers. We evaluated the expression of 259 proteins following 2, 6, and 12 h of cisplatin treatment in two LCLs with discordant sensitivity to cisplatin. Of these 259 proteins, 66 displayed significantly different protein expression changes (p < 0.05). Fifteen of these proteins were evaluated in a second pair of LCLs with discordant sensitivities to cisplatin; six demonstrated significant differences in expression. We then evaluated a subset of 63 proteins in a second set of LCLs with discordant sensitivity, and 40% of those that were significant in the first pair were also significant in the second part with concordant directionality (p < 0.05). We functionally validated one of the top proteins identified, PDK1, and demonstrated a synergistic relationship between cisplatin and a PDK1 inhibitor in multiple lung cancer lines. This study highlights the potential for identifying novel targets through an understanding of cellular changes in protein expression and modification following drug treatments.

Identification and validation of genetic variants that influence transcription factor and cell signaling protein levels.

Many genetic variants associated with human disease have been found to be associated with alterations in mRNA expression. Although it is commonly assumed that mRNA expression changes will lead to consequent changes in protein levels, methodological challenges have limited our ability to test the degree to which this assumption holds true. Here, we further developed the micro-western array approach and globally examined relationships between human genetic variation and cellular protein levels. We collected more than 250,000 protein level measurements comprising 441 transcription factor and signaling protein isoforms across 68 Yoruba (YRI) HapMap lymphoblastoid cell lines (LCLs) and identified 12 cis and 160 trans protein level QTLs (pQTLs) at a false discovery rate (FDR) of 20%. Whereas up to two thirds of cis mRNA expression QTLs (eQTLs) were also pQTLs, many pQTLs were not associated with mRNA expression. Notably, we replicated and functionally validated a trans pQTL relationship between the KARS lysyl-tRNA synthetase locus and levels of the DIDO1 protein. This study demonstrates proof of concept in applying an antibody-based microarray approach to iteratively measure the levels of human proteins and relate these levels to human genome variation and other genomic data sets. Our results suggest that protein-based mechanisms might functionally buffer genetic alterations that influence mRNA expression levels and that pQTLs might contribute phenotypic diversity to a human population independently of influences on mRNA expression.

Protein quantitative trait loci identify novel candidates modulating cellular response to chemotherapy.

Annotating and interpreting the results of genome-wide association studies (GWAS) remains challenging. Assigning function to genetic variants as expression quantitative trait loci is an expanding and useful approach, but focuses exclusively on mRNA rather than protein levels. Many variants remain without annotation. To address this problem, we measured the steady state abundance of 441 human signaling and transcription factor proteins from 68 Yoruba HapMap lymphoblastoid cell lines to identify novel relationships between inter-individual protein levels, genetic variants, and sensitivity to chemotherapeutic agents. Proteins were measured using micro-western and reverse phase protein arrays from three independent cell line thaws to permit mixed effect modeling of protein biological replicates. We observed enrichment of protein quantitative trait loci (pQTLs) for cellular sensitivity to two commonly used chemotherapeutics: cisplatin and paclitaxel. We functionally validated the target protein of a genome-wide significant trans-pQTL for its relevance in paclitaxel-induced apoptosis. GWAS overlap results of drug-induced apoptosis and cytotoxicity for paclitaxel and cisplatin revealed unique SNPs associated with the pharmacologic traits (at p<0.001). Interestingly, GWAS SNPs from various regions of the genome implicated the same target protein (p<0.0001) that correlated with drug induced cytotoxicity or apoptosis (p ≤ 0.05). Two genes were functionally validated for association with drug response using siRNA: SMC1A with cisplatin response and ZNF569 with paclitaxel response. This work allows pharmacogenomic discovery to progress from the transcriptome to the proteome and offers potential for identification of new therapeutic targets. This approach, linking targeted proteomic data to variation in pharmacologic response, can be generalized to other studies evaluating genotype-phenotype relationships and provide insight into chemotherapeutic mechanisms.

Mouse models of human AML accurately predict chemotherapy response.

The genetic heterogeneity of cancer influences the trajectory of tumor progression and may underlie clinical variation in therapy response. To model such heterogeneity, we produced genetically and pathologically accurate mouse models of common forms of human acute myeloid leukemia (AML) and developed methods to mimic standard induction chemotherapy and efficiently monitor therapy response. We see that murine AMLs harboring two common human AML genotypes show remarkably diverse responses to conventional therapy that mirror clinical experience. Specifically, murine leukemias expressing the AML1/ETO fusion oncoprotein, associated with a favorable prognosis in patients, show a dramatic response to induction chemotherapy owing to robust activation of the p53 tumor suppressor network. Conversely, murine leukemias expressing MLL fusion proteins, associated with a dismal prognosis in patients, are drug-resistant due to an attenuated p53 response. Our studies highlight the importance of genetic information in guiding the treatment of human AML, functionally establish the p53 network as a central determinant of chemotherapy response in AML, and demonstrate that genetically engineered mouse models of human cancer can accurately predict therapy response in patients.

Genetic Variants Contributing to Colistin Cytotoxicity: Identification of TGIF1 and HOXD10 Using a Population Genomics Approach.

Colistin sulfate (polymixin E) is an antibiotic prescribed with increasing frequency for severe Gram-negative bacterial infections. As nephrotoxicity is a common side effect, the discovery of pharmacogenomic markers associated with toxicity would benefit the utility of this drug. Our objective was to identify genetic markers of colistin cytotoxicity that were also associated with expression of key proteins using an unbiased, whole genome approach and further evaluate the functional significance in renal cell lines. To this end, we employed International HapMap lymphoblastoid cell lines (LCLs) of Yoruban ancestry with known genetic information to perform a genome-wide association study (GWAS) with cellular sensitivity to colistin. Further association studies revealed that single nucleotide polymorphisms (SNPs) associated with gene expression and protein expression were significantly enriched in SNPs associated with cytotoxicity (p ≤ 0.001 for gene and p = 0.015 for protein expression). The most highly associated SNP, chr18:3417240 (p = 6.49 × 10-8), was nominally a cis-expression quantitative trait locus (eQTL) of the gene TGIF1 (transforming growth factor ß (TGFß)-induced factor-1; p = 0.021) and was associated with expression of the protein HOXD10 (homeobox protein D10; p = 7.17 × 10-5). To demonstrate functional relevance in a murine colistin nephrotoxicity model, HOXD10 immunohistochemistry revealed upregulated protein expression independent of mRNA expression in response to colistin administration. Knockdown of TGIF1 resulted in decreased protein expression of HOXD10 and increased resistance to colistin cytotoxicity. Furthermore, knockdown of HOXD10 in renal cells also resulted in increased resistance to colistin cytotoxicity, supporting the physiological relevance of the initial genomic associations.

Linking the genetic architecture of cytosine modifications with human complex traits.

Interindividual variation in cytosine modifications could contribute to heterogeneity in disease risks and other complex traits. We assessed the genetic architecture of cytosine modifications at 283,540 CpG sites in lymphoblastoid cell lines (LCLs) derived from independent samples of European and African descent. Our study suggests that cytosine modification variation was primarily controlled in local by single major modification quantitative trait locus (mQTL) and additional minor loci. Local genetic epistasis was detectable for a small proportion of CpG sites, which were enriched by more than 9-fold for CpG sites mapped to population-specific mQTL. Genetically dependent CpG sites whose modification levels negatively (repressive sites) or positively (facilitative sites) correlated with gene expression levels significantly co-localized with transcription factor binding, with the repressive sites predominantly associated with active promoters whereas the facilitative sites rarely at active promoters. Genetically independent repressive or facilitative sites preferentially modulated gene expression variation by influencing local chromatin accessibility, with the facilitative sites primarily antagonizing H3K27me3 and H3K9me3 deposition. In comparison with expression quantitative trait loci (eQTL), mQTL detected from LCLs were enriched in associations for a broader range of disease categories including chronic inflammatory, autoimmune and psychiatric disorders, suggesting that cytosine modification variation, while possesses a degree of cell linage specificity, is more stably inherited over development than gene expression variation. About 11% of unique single-nucleotide polymorphisms reported in the Genome-Wide Association Study Catalog were annotated, 78% as mQTL and 31% as eQTL in LCLs, which covered 37% of the investigated diseases/traits and provided insights to the biological mechanisms.

Chemotherapy-induced peripheral neuropathy: Current status and progress.

As there are increasing numbers of cancer survivors, more attention is being paid to the long term unwanted effects patients may experience as a result of their treatment and the impact these side effects can have on their quality of life. Chemotherapy-induced peripheral neuropathy (CIPN) is one of the most common long-term toxicities from chemotherapy. In this review we will briefly review the clinical presentation, evaluation and management of chemotherapy-induced peripheral neuropathy, with a focus on CIPN related to platinum and taxane agents. We will then discuss current clinical models of peripheral neuropathy and ongoing research to better understand CIPN and develop potential treatment options.

Poly-omic prediction of complex traits: OmicKriging.

High-confidence prediction of complex traits such as disease risk or drug response is an ultimate goal of personalized medicine. Although genome-wide association studies have discovered thousands of well-replicated polymorphisms associated with a broad spectrum of complex traits, the combined predictive power of these associations for any given trait is generally too low to be of clinical relevance. We propose a novel systems approach to complex trait prediction, which leverages and integrates similarity in genetic, transcriptomic, or other omics-level data. We translate the omic similarity into phenotypic similarity using a method called Kriging, commonly used in geostatistics and machine learning. Our method called OmicKriging emphasizes the use of a wide variety of systems-level data, such as those increasingly made available by comprehensive surveys of the genome, transcriptome, and epigenome, for complex trait prediction. Furthermore, our OmicKriging framework allows easy integration of prior information on the function of subsets of omics-level data from heterogeneous sources without the sometimes heavy computational burden of Bayesian approaches. Using seven disease datasets from the Wellcome Trust Case Control Consortium (WTCCC), we show that OmicKriging allows simple integration of sparse and highly polygenic components yielding comparable performance at a fraction of the computing time of a recently published Bayesian sparse linear mixed model method. Using a cellular growth phenotype, we show that integrating mRNA and microRNA expression data substantially increases performance over either dataset alone. Using clinical statin response, we show improved prediction over existing methods. We provide an R package to implement OmicKriging (http://www.scandb.org/newinterface/tools/OmicKriging.html).

Genetic and epigenetic variants contributing to clofarabine cytotoxicity.

2-chloro-2-fluoro-deoxy-9-D-arabinofuranosyladenine (Clofarabine), a purine nucleoside analog, is used in the treatment of hematologic malignancies and as induction therapy for stem cell transplantation. The discovery of pharmacogenomic markers associated with chemotherapeutic efficacy and toxicity would greatly benefit the utility of this drug. Our objective was to identify genetic and epigenetic variants associated with clofarabine toxicity using an unbiased, whole genome approach. To this end, we employed International HapMap lymphoblastoid cell lines (190 LCLs) of European (CEU) or African (YRI) ancestry with known genetic information to evaluate cellular sensitivity to clofarabine. We measured modified cytosine levels to ascertain the contribution of genetic and epigenetic factors influencing clofarabine-mediated cytotoxicity. Association studies revealed 182 single nucleotide polymorphisms (SNPs) and 143 modified cytosines associated with cytotoxicity in both populations at the threshold P ≤ 0.0001. Correlation between cytotoxicity and baseline gene expression revealed 234 genes at P ≤ 3.98 × 10(-6). Six genes were implicated as: (i) their expression was directly correlated to cytotoxicity, (ii) they had a targeting SNP associated with cytotoxicity, and (iii) they had local modified cytosines associated with gene expression and cytotoxicity. We identified a set of three SNPs and three CpG sites targeting these six genes explaining 43.1% of the observed variation in phenotype. siRNA knockdown of the top three genes (SETBP1, BAG3, KLHL6) in LCLs revealed altered susceptibility to clofarabine, confirming relevance. As clofarabine's toxicity profile includes acute kidney injury, we examined the effect of siRNA knockdown in HEK293 cells. siSETBP1 led to a significant change in HEK293 cell susceptibility to clofarabine.

Relating human genetic variation to variation in drug responses.

Although sequencing a single human genome was a monumental effort a decade ago, more than 1000 genomes have now been sequenced. The task ahead lies in transforming this information into personalized treatment strategies that are tailored to the unique genetics of each individual. One important aspect of personalized medicine is patient-to-patient variation in drug response. Pharmacogenomics addresses this issue by seeking to identify genetic contributors to human variation in drug efficacy and toxicity. Here, we present a summary of the current status of this field, which has evolved from studies of single candidate genes to comprehensive genome-wide analyses. Additionally, we discuss the major challenges in translating this knowledge into a systems-level understanding of drug physiology, with the ultimate goal of developing more effective personalized clinical treatment strategies.

Comprehensive genetic analysis of cytarabine sensitivity in a cell-based model identifies polymorphisms associated with outcome in AML patients.

A whole-genome approach was used to investigate the genetic determinants of cytarabine-induced cytotoxicity. We performed a meta-analysis of genome-wide association studies involving 523 lymphoblastoid cell lines (LCLs) from individuals of European, African, Asian, and African American ancestry. Several of the highest-ranked single-nucleotide polymorphisms (SNPs) were within the mutated in colorectal cancers (MCC) gene. MCC expression was induced by cytarabine treatment from 1.7- to 26.6-fold in LCLs. A total of 33 SNPs ranked at the top of the meta-analysis (P < 10(-5)) were successfully tested in a clinical trial of patients randomized to receive low-dose or high-dose cytarabine plus daunorubicin and etoposide; of these, 18 showed association (P < .05) with either cytarabine 50% inhibitory concentration in leukemia cells or clinical response parameters (minimal residual disease, overall survival (OS), and treatment-related mortality). This count (n = 18) was significantly greater than expected by chance (P = .016). For rs1203633, LCLs with AA genotype were more sensitive to cytarabine-induced cytotoxicity (P = 1.31 × 10(-6)) and AA (vs GA or GG) genotype was associated with poorer OS (P = .015), likely as a result of greater treatment-related mortality (P = .0037) in patients with acute myeloid leukemia (AML). This multicenter AML02 study trial was registered at www.clinicaltrials.gov as #NCT00136084.

Hydroxylation and N-dechloroethylation of Ifosfamide and deuterated Ifosfamide by the human cytochrome p450s and their commonly occurring polymorphisms.

The hydroxylation and N-dechloroethylation of deuterated ifosfamide (d4IFO) and ifosfamide (IFO) by several human P450s have been determined and compared. d4IFO was synthesized with deuterium at the alpha and alpha' carbons to decrease the rate of N-dechloroethylation and thereby enhance hydroxylation of the drug at the 4' position. The purpose was to decrease the toxic and increase the efficacious metabolites of IFO. For all of the P450s tested, hydroxylation of d4IFO was improved and dechloroethylation was reduced as compared with nondeuterated IFO. Although the differences were not statistically significant, the trend favoring the 4'-hydroxylation pathway was noteworthy. CYP3A5 and CYP2C19 were the most efficient enzymes for catalyzing IFO hydroxylation. The importance of these enzymes in IFO metabolism has not been reported previously and warrants further investigation. The catalytic ability of the common polymorphisms of CYP2B6 and CYP2C9 for both reactions were tested with IFO and d4IFO. It was determined that the commonly expressed polymorphisms CYP2B6*4 and CYP2B6*6 had reduced catalytic ability for IFO compared with CYP2B6*1, whereas CYP2B6*7 and CYP2B6*9 had enhanced catalytic ability. As with the wild-type enzymes, d4IFO was more readily hydroxylated by the polymorphic variants than IFO, and d4IFO was not dechloroethylated by any of the polymorphic forms. We also assessed the use of specific inhibitors of P450 to favor hydroxylation in human liver microsomes. We were unable to separate the pathways with these experiments, suggesting that multiple P450s are responsible for catalyzing both metabolic pathways for IFO, which is not observed with the closely related drug cyclophosphamide.