Polygenic inheritance of paclitaxel-induced sensory peripheral neuropathy driven by axon outgrowth gene sets in CALGB 40101 (Alliance).

Peripheral neuropathy is a common dose-limiting toxicity for patients treated with paclitaxel. For most individuals, there are no known risk factors that predispose patients to the adverse event, and pathogenesis for paclitaxel-induced peripheral neuropathy is unknown. Determining whether there is a heritable component to paclitaxel-induced peripheral neuropathy would be valuable in guiding clinical decisions and may provide insight into treatment of and mechanisms for the toxicity. Using genotype and patient information from the paclitaxel arm of CALGB 40101 (Alliance), a phase III clinical trial evaluating adjuvant therapies for breast cancer in women, we estimated the variance in maximum grade and dose at first instance of sensory peripheral neuropathy. Our results suggest that paclitaxel-induced neuropathy has a heritable component, driven in part by genes involved in axon outgrowth. Disruption of axon outgrowth may be one of the mechanisms by which paclitaxel treatment results in sensory peripheral neuropathy in susceptible patients.

Hypospadias and variants in genes related to sex hormone biosynthesis and metabolism.

We examined whether variants in genes related to sex hormone biosynthesis and metabolism were associated with hypospadias in humans. We examined 332 relatively common tag single-nucleotide polymorphisms (tagSNPs) in 20 genes. Analyses included 633 cases (84 mild, 322 moderate, 212 severe and 15 undetermined severity) and 855 population-based non-malformed male controls born in California from 1990 to 2003. We used logistic regression models to estimate odds ratios (OR) and 95% confidence intervals (CI) for each SNP. Several of the 332 studied SNPs had p < 0.01: one in CYP3A4, four in HSD17B3, one in HSD3B1, two in STARD3, 10 in SRD5A2 and seven in STS. In addition, haplotype analyses gave several associations with p < 0.01. For HSD17B3, 14-SNP and 5-SNP blocks had ORs of 1.5 (95% CI 1.1, 2.0, p < 0.001) and 2.8 (95% CI 1.6, 4.8, p < 0.001) respectively. For SRD5A2, 9-SNP, 3-SNP and 8-SNP blocks had ORs of 1.7 (95% CI 1.3, 2.2, p < 0.001), 1.4 (95% CI 1.1, 1.8, p = 0.008) and 1.5 (95% CI 1.2, 1.9, p = 0.002) respectively. Our study indicates that several genes that contribute to sex hormone biosynthesis and metabolism are associated with hypospadias risk.

Association of CYP2C9*2 with bosentan-induced liver injury.

Bosentan (Tracleer) is an endothelin receptor antagonist prescribed for the treatment of pulmonary arterial hypertension (PAH). Its use is limited by drug-induced liver injury (DILI). To identify genetic markers of DILI, association analyses were performed on 56 Caucasian PAH patients receiving bosentan. Twelve functional polymorphisms in five genes (ABCB11, ABCC2, CYP2C9, SLCO1B1, and SLCO1B3) implicated in bosentan pharmacokinetics were tested for associations with alanine aminotransferase (ALT), aspartate aminotransferase (AST), and DILI. After adjusting for body mass index, CYP2C9*2 was the only polymorphism associated with ALT, AST, and DILI (ß = 2.16, P = 0.024; ß = 1.92, P = 0.016; odds ratio 95% CI = 2.29-∞, P = 0.003, respectively). Bosentan metabolism by CYP2C9*2 in vitro was significantly reduced compared with CYP2C9*1 and was comparable to that by CYP2C9*3. These results suggest that CYP2C9*2 is a potential genetic marker for prediction of bosentan-induced liver injury and warrants investigation for the optimization of bosentan treatment.

Variations in the heme oxygenase-1 microsatellite polymorphism are associated with plasma CD14 and viral load in HIV-infected African-Americans.

Heme oxygenase-1 (HO-1) is an anti-inflammatory enzyme that maintains homeostasis during cellular stress. Given previous findings that shorter length variants of a HO-1 promoter region GT(n) microsatellite polymorphism are associated with increased HO-1 expression in cell lines, we hypothesized that shorter variants would also be associated with increased levels of HO-1 expression, less inflammation and lower levels of inflammation-associated viral replication in human immunodeficiency virus (HIV)-infected subjects. Healthy donors (n = 20) with shorter GT(n) repeats had higher HO-1 mRNA transcript in peripheral blood mononuclear cells stimulated with lipopolysaccharide (r = -0.38, P = 0.05). The presence of fewer GT(n) repeats in subjects with untreated HIV disease was associated with higher HO-1 mRNA levels in peripheral blood (r = -0.41, P = 0.02); similar observations were made in CD14(+) monocytes from antiretroviral-treated subjects (r = -0.36, P = 0.04). In African-Americans, but not Caucasians, greater GT(n) repeats were correlated with higher soluble CD14 levels during highly active antiretroviral therapy (r = 0.38, P = 0.007) as well as higher mean viral load off-therapy (r = 0.24, P = 0.04). These data demonstrate that the HO-1 GT(n) microsatellite polymorphism is associated with higher levels of HO-1 expression and that this pathway may have important effects on the association between inflammation and HIV replication.

A common 5'-UTR variant in MATE2-K is associated with poor response to metformin.

Multidrug and toxin extrusion 2 (MATE2-K (SLC47A2)), a polyspecific organic cation exporter, facilitates the renal elimination of the antidiabetes drug metformin. In this study, we characterized genetic variants of MATE2-K, determined their association with metformin response, and elucidated their impact by means of a comparative protein structure model. Four nonsynonymous variants and four variants in the MATE2-K basal promoter region were identified from ethnically diverse populations. Two nonsynonymous variants-c.485C>T and c.1177G>A-were shown to be associated with significantly lower metformin uptake and reduction in protein expression levels. MATE2-K basal promoter haplotypes containing the most common variant, g.-130G>A (>26% allele frequency), were associated with a significant increase in luciferase activities and reduced binding to the transcriptional repressor myeloid zinc finger 1 (MZF-1). Patients with diabetes who were homozygous for g.-130A had a significantly poorer response to metformin treatment, assessed as relative change in glycated hemoglobin (HbA1c) (-0.027 (-0.076, 0.033)), as compared with carriers of the reference allele, g.-130G (-0.15 (-0.17, -0.13)) (P=0.002). Our study showed that MATE2-K plays a role in the antidiabetes response to metformin.

COX2 genetic variation, NSAIDs, and advanced prostate cancer risk.

Collective evidence suggests that cyclooxygenase 2 (COX2) plays a role in prostate cancer risk. Cyclooxygenase 2 is the major enzyme that converts arachidonic acid to prostaglandins, which are potent mediators of inflammation. Nonsteroidal anti-inflammatory drugs (NSAIDs) inhibit the enzymatic activity of COX2 and long-term use of NSAIDs appears to modestly lower the risk of prostate cancer. We investigated whether common genetic variation in COX2 influences the risk of advanced prostate cancer. Nine single-nucleotide polymorphisms (SNPs) in COX2 were genotyped among 1012 men in our case-control study of advanced prostate cancer. Gene-environment interactions between COX2 polymorphisms and NSAID use were also evaluated. Information on NSAID use was obtained by questionnaire. Three SNPs demonstrated nominally statistically significant associations with prostate cancer risk, with the most compelling polymorphism (rs2745557) associated with a lower risk of disease (odds ratio (OR) GC vs GG=0.64; 95% confidence interval (CI): 0.49-0.84; P=0.002). We estimated through permutation analysis that a similarly strong result would occur by chance 2.7% of the time. Nonsteroidal anti-inflammatory drug use was associated with a lower risk of disease in comparison to no use (OR=0.67; 95% CI: 0.52-0.87). No significant statistical interaction between NSAID use and rs2745557 was observed (P=0.12). Our findings suggest that variation in COX2 is associated with prostate cancer risk.

Model-free linkage analysis with covariates confirms linkage of prostate cancer to chromosomes 1 and 4.

As with many complex genetic diseases, genome scans for prostate cancer have given conflicting results, often failing to provide replication of previous findings. One factor contributing to the lack of consistency across studies is locus heterogeneity, which can weaken or even eliminate evidence for linkage that is present only in a subset of families. Currently, most analyses either fail to account for locus heterogeneity or attempt to account for it only by partitioning data sets into smaller and smaller portions. In the present study, we model locus heterogeneity among affected sib pairs with prostate cancer by including covariates in the linkage analysis that serve as surrogate measures of between-family linkage differences. The model is a modification of the Olson conditional logistic model for affected relative pairs. By including Gleason score, age at onset, male-to-male transmission, and/or number of affected first-degree family members as covariates, we detected linkage near three locations that were previously identified by linkage (1q24-25 [HPC1; LOD score 3.25, P=.00012], 1q42.2-43 [PCAP; LOD score 2.84, P=.0030], and 4q [LOD score 2.80, P=.00038]), near the androgen-receptor locus on Xq12-13 (AR; LOD score 3.06, P=.00053), and at five new locations (LOD score > 2.5). Without covariates, only a few weak-to-moderate linkage signals were found, none of which replicate findings of previous genome scans. We conclude that covariate-based linkage analysis greatly improves the likelihood that linked regions will be found by incorporation of information about heterogeneity within the sample.

Predicting quantitative trait levels by modeling SNP interaction.

Predicting phenotype from genotype is difficult when the phenotype is affected by a gene with numerous weakly penetrant alleles that differ only in the pattern of their single nucleotide polymorphisms (SNPs). While it is probable that SNP interactions affect phenotype, to our knowledge no one has determined the most effective way of evaluating whether SNPs interact and of modeling the interaction. Therefore, to explore this issue, we investigate here three methods of modeling SNP interaction using data from Genetic Analysis Workshop 12. Since major gene 5 (MG5) has sequence information and explains 37% of the variation in quantitative trait 5 (Q5), we focus on using SNPs within MG5 to predict Q5 among individuals who married into the pedigree. As a preliminary screening step, we reduced the number of SNPs from 269 to 34 based on their association with Q5. In our first models we assumed that SNPs affected Q5 in a simple additive manner. These models explained 34% and 15% of the variation in Q5 in women and men, respectively. Our second model was a linear model, which used individual SNPs and simple interaction terms as predictors. These models explained 36% and 16% of the variation in Q5 levels for women and men, respectively. Our last model was a "hit"-based model which was motivated by the hypothesis that disequilibrium between SNPs may reflect the fact that SNPs affect phenotype by acting in concert with other SNPs within their "disequilibrium set." Thus, the number of hits within the disequilibrium sets were used as predictors. These models explained 35% and 19% of the variation in Q5 for women and men, respectively. Our results suggest that phenotype can be predicted from complex patterns of weakly penetrant SNPs using relatively simple models. We concluded that SNP interaction either was not included in the simulation model, or had only a weak impact on Q5 levels.

Hierarchical modeling of the relation between sequence variants and a quantitative trait: addressing multiple comparison and population stratification issues.

When analyzing the relation between genetic sequence information and disease traits, false-positive associations can arise due to multiple comparisons and population stratification. In an attempt to address these issues, we incorporate into a conventional analytic model higher-level--or "prior"--models that use additional information to improve estimates while allowing for differing population structures. We apply this hierarchical model to simulated data from the Genetic Analysis Workshop 12. We focus on the effects of common candidate gene sequence variants on quantitative risk factor 5 (Q5) levels. In particular, we compare the regression coefficients (and 95% confidence intervals) obtained from conventional (one-stage) analyses versus the corresponding results from the hierarchical analyses. When examining either the marry-ins or all subjects in the general and isolate populations, the conventional model detected numerous sites in candidate genes 1-5 and 7 that had statistically significant regression coefficients (alpha level = 0.05). In contrast, our hierarchical model primarily only detected associations for variants in candidate gene 2, which is the casual gene for Q5.

Impact of preadjusting a quantitative phenotype prior to sib-pair linkage analysis when gene x environment interaction exists.

The investigation of potential gene x environment (G x E) interactions is an important facet in the study of complex diseases. When G x E interaction exists, linkage analyses of the interacting gene must treat the environmental factor appropriately. Specifically, the common approach of regressing out an environmental factor prior to linkage analysis may be inappropriate if that factor has an interaction with the gene. This is explored here in the Genetic Analysis Workshop 12 simulated data set using the G x E interaction between major gene four (MG4) and environmental factor two (E2). The analysis shows that preadjusting the quantitative trait three (Q3) phenotype for the main effects of several environmental variables, including one (E2) that interacts with MG4, affects the results of a Haseman-Elston linkage analysis. In particular, the agreement in detecting linkage between preadjusting versus not preadjusting was only 78% and 66% using alpha levels of 0.05 and 0.10, respectively. For both approaches, incorporating an interaction term in the regression models enabled linkage to be detected where the evidence was either minimal or not present in an identical-by-descent main effects-only model. Furthermore, preadjustment for E2 did not appear to account for the major discrepancies between the approaches.

Replication linkage study for prostate cancer susceptibility genes.

BACKGROUND: Since the publication of the first genome screen for prostate cancer (CaP) 5 years ago, over a dozen linkage studies have appeared. Most attention has been directed to chromosome 1, where two separate regions have been identified as harboring a prostate cancer susceptibility locus: HPC1 in the 1q24-25 interval and PCaP in the 1q42.2-43 interval. Linkage analysis of chromosome 16 has also provided evidence of harboring two loci predisposing to CaP. METHODS: We report on a replication linkage study of chromosomes 1 and 16 in 45 new and 4 expanded multiplex CaP families. Multipoint Z-scores were obtained for 30 highly polymorphic short-sequence tandem repeat markers spanning chromosome 1, and 22 markers spanning chromosome 16. RESULTS: The replication sample gave no evidence for a CaP susceptibility locus in the 1q24-25 interval and equivocal evidence for such a locus at 1q42.2-43. With respect to chromosome 16, positive Z-scores were obtained over a contiguous interval covering the entire p arm and the proximal half of the q arm. CONCLUSIONS: The linkage analysis of our replication sample does not support the existence of HPC1, and the evidence for the existence of PCaP remains equivocal. Evidence of a susceptibility locus on 16p remains strong, but the evidence for a susceptibility locus on 16q is weakened.

Multilevel modeling in epidemiology with GLIMMIX.

Previous work has shown that multilevel modeling can be a valuable technique for epidemiologic analysis. The complexity of using this approach, however, continues to restrict its general application. A critical factor is the lack of flexible and appropriate software for multilevel modeling. SAS provides a macro, GLIMMIX, that can be used for multilevel modeling, but that is not sufficient for a complete epidemiologic analysis. We here provide additional code to obtain epidemiologic output from GLIMMIX, illustrated with new data on diet and breast cancer from the European Community Multicenter Study on Antioxidants, Myocardial Infarction, and Breast Cancer (EURAMIC). Our results give epidemiologists an easily used tool for fitting multilevel models.

Testing drug response in the presence of genetic information: sampling issues for clinical trials.

Progress towards construction of a dense map of di-allelic markers across the human genome has generated considerable enthusiasm for pharmacogenomic applications. To date, however, nearly all of the effort on single nucleotide polymorphism (SNP) projects has been focused on marker identification and screening, not on how the SNP genotype data actually can be used in clinical trials to advance medical practice. Here, we explore how different properties of SNPs impact the size, scope and design of clinical trials using a simple trial design. We evaluate the clinical trial sampling requirements under different allele frequencies, gene action, gene effect size and number of markers in a genome screen. Power and sample size calculations suggest that allele frequency and type of gene action can have a dramatic impact on trial sample sizes, in that under some conditions the required sample sizes are too large to be applicable in a costly clinical trial setting. In other situations, however, pharmacogenomic clinical trials can yield significant sampling/cost savings over traditional trials. These properties are discussed with regard to the general usage of genetic information in clinical trial settings.

CYP3A activity in African American and European American men: population differences and functional effect of the CYP3A4*1B5'-promoter region polymorphism.

OBJECTIVE: Cytochrome P4503A (CYP3A) activity exhibits considerable interindividual variability. Possible differences in CYP3A activity were investigated in European American and African American men with the use of midazolam as an in vivo probe. METHODS: Midazolam was simultaneously administered intravenously (1 mg, [15N3]-labeled) and orally (2 mg, unlabeled in capsule form) to 15 young healthy European American men and a similar group of men of African American descent. Plasma concentration-time curves were measured. The subjects were subsequently genotyped with respect to the CYP3A4*B1 polymorphism (A-290G) in the 5'-promoter (nifedipine-specific element) region. RESULTS: The oral bioavailability of midazolam was about equally determined by intestinal and hepatic extraction with CYP3A activity at the former site exhibiting greater variability. Oral bioavailability was related to intestinal metabolism (r = 0.98), whereas hepatic CYP3A activity contributed little to the interindividual variability (r = 0.03). A lower systemic clearance (265+/-54 versus 310+/-56 mL/min; P = .04), but not oral clearance, was observed in African Americans. With one exception, the African Americans possessed a variant CYP3A4*1B allele (4 heterozygotes A/G and 10 homozygote G/G), whereas all of the European Americans were wild-type homozygotes (A/A). Hepatic CYP3A activity and the systemic clearance of midazolam were about 30% lower in G/G homozygotes than in A/A homozygotes (252+/-53 versus 310+/-54 mL/min; P = .02), and a gene-dose effect was present (P = .01). There was no genotype/phenotype relationship with respect to the oral clearance of midazolam. CONCLUSION: Comparison of CYP3A activity between populations is complicated by frequency distribution differences in the regulatory CYP3A4*1B polymorphism and lower hepatic CYP3A activity associated with the variant allele. However, this reduction is modest; therefore no major and clinically important difference in CYP3A activity is present between Americans of African or European descent.

Identification and fine mapping of a region showing a high frequency of allelic imbalance on chromosome 16q23.2 that corresponds to a prostate cancer susceptibility locus.

Linkage to a prostate cancer susceptibility locus was recently reported on chromosome 16q23. We now report a region exhibiting a high frequency of allelic imbalance (AI) corresponding to this locus in tumors from 51 men diagnosed with prostate cancer using the same linked markers. The highest frequency of AI was found at markers D16S3096 (45%) and D16S516 (53%) that map to chromosome 16q23.2. In addition, 19 of the 51 (37%) prostate tumors showed interstitial AI involving one or both of these markers. This result strongly suggests that a candidate prostate cancer tumor suppressor gene maps between markers D16S3096 and D16S516. We estimate that the distance between these markers is approximately 118 kb using a Stanford radiation hybrid panel. We observed a positive association with family history (P = 0.048) when comparing those men showing interstitial AI at markers D16S3096 and/or D16S516 with those without any imbalance at these two markers. Taken together, these data suggest that we have precisely localized a region of chromosome 16q23.2 that may harbor a prostate cancer tumor suppressor gene implicated in the development of non-familial and possibly familial forms of prostate cancer.

Genomewide scan for prostate cancer-aggressiveness loci.

The aggressiveness of prostate cancer (PCa) varies widely: some tumors progress to invasive, potentially life-threatening disease, whereas others stay latent for the remainder of an individual's lifetime. The mechanisms resulting in this variability are not yet understood, but they are likely to involve both genetic and environmental influences. To investigate genetic factors, we conducted a genomewide linkage analysis of 513 brothers with PCa, using the Gleason score, which reflects tumor histology, as a quantitative measure of PCa aggressiveness. To our knowledge, this is the first time that a measure of PCa aggressiveness has been directly investigated as a quantitative trait in a genomewide scan. We employed a generalized multipoint Haseman-Elston linkage-analysis approach that regresses the mean-corrected cross product between the brothers' Gleason scores on the estimated proportion of alleles shared by brothers identical by descent at each marker location. Our results suggest that candidate regions on chromosomes 5q, 7q, and 19q give evidence for linkage to PCa-aggressiveness genes. In particular, the strongest signals detected in these regions were at the following markers (with corresponding P values): for chromosome 5q31-33, between markers D5S1480 and D5S820 (P=.0002); for chromosome 7q32, between markers D7S3061 and D7S1804 (P=.0007); and, for chromosome 19q12, at D19S433 (P=.0004). This indicates that one or more of these candidate regions may contain genes that influence the progression of PCa from latent to invasive disease. Identification of such genes would be extremely valuable for elucidation of the mechanism underlying PCa progression and for determination of treatment in men in whom this disease has been diagnosed.

A genome screen of multiplex sibships with prostate cancer.

Analysis of a genome screen of 504 brothers with prostate cancer (CaP) who were from 230 multiplex sibships identified five regions with nominally positive linkage signals, on chromosomes 2q, 12p, 15q, 16p, and 16q. The strongest signal in these data is found on chromosome 16q, between markers D16S515 and D16S3040, a region suspected to contain a tumor-suppressor gene. On the basis of findings from previous genome screens of families with CaP, three preplanned subanalyses were carried out, in the hope of increasing the subgroup homogeneity. Subgroups were formed by dividing the sibships into a group with a positive family history (FH+) that met criteria for "hereditary" CaP (n=111) versus those which did not meet the criteria (n=119) and by dividing the families into those with a mean onset age below the median (n=115) versus those with a mean onset age above the median (n=115). A separate subanalysis was carried out for families with a history of breast cancer (CaB+ [n=53]). Analyses of these subgroups revealed a number of potentially important differences in regions that were nonsignificant when all the families were analyzed together. In particular, the subgroup without a positive family history (FH-) had a signal in a region that is proximal to the putative site of the HPC1 locus on chromosome 1, whereas the late-age-at-onset group had a signal on 4q. The CaB+ subgroup revealed a strong linkage signal at 1p35.1.

Clinical trials in the genomic era: effects of protective genotypes on sample size and duration of trial.

It is well known that individuals can vary widely in their disease susceptibilities. One potential source of this variation is the genetic makeup of individuals, which can confer either protection or susceptibility to disease. Here we examine the effects of protective genotypes on the sample sizes and time required to detect differences between clinical trial arms. We show that including individuals with protective genotypes in a clinical trial can increase required sample sizes and trial duration. One can deal with this issue by pregenotyping subjects and selectively enrolling them based on their genotype. Thus we also calculate the number of individuals that must be recruited and pregenotyped to fulfill sample size requirements. The benefits of genotypically screening study subjects will depend on numerous factors, including ease of patient recruitment, cost of genotyping, long-term costs of study (or long-term cost per subject), and the strength of the protective effect. We present several examples that show the potential value of incorporating information about protective genotypes into a clinical trial.

On the relative sample size required for multiple comparisons.

Multiple comparisons are commonly made in epidemiologic and genetic research. How to appropriately adjust for multiple comparisons remains a controversial issue. This note demonstrates, however, that large increases in the number of comparisons has a limited effect on the sample size required to maintain an experimentwise alpha-level. In particular, the relative sample size required increases only linearly with the logarithm of the number of comparisons made.