Type I error rates of rare single nucleotide variants are inflated in tests of association with non-normally distributed traits using simple linear regression methods.

In this study, the effects of (a) the minor allele frequency of the single nucleotide variant (SNV), (b) the degree of departure from normality of the trait, and (c) the position of the SNVs on type I error rates were investigated in the Genetic Analysis Workshop (GAW) 19 whole exome sequence data. To test the distribution of the type I error rate, 5 simulated traits were considered: standard normal and gamma distributed traits; 2 transformed versions of the gamma trait (log10 and rank-based inverse normal transformations); and trait Q1 provided by GAW 19. Each trait was tested with 313,340 SNVs. Tests of association were performed with simple linear regression and average type I error rates were determined for minor allele frequency classes. Rare SNVs (minor allele frequency < 0.05) showed inflated type I error rates for non-normally distributed traits that increased as the minor allele frequency decreased. The inflation of average type I error rates increased as the significance threshold decreased. Normally distributed traits did not show inflated type I error rates with respect to the minor allele frequency for rare SNVs. There was no consistent effect of transformation on the uniformity of the distribution of the location of SNVs with a type I error.

The FKBP5-gene in depression and treatment response--an association study in the Sequenced Treatment Alternatives to Relieve Depression (STAR*D) Cohort.

BACKGROUND: In a recent study of several antidepressant drugs in hospitalized, non-Hispanic White patients, Binder et al. reported association of markers located within the FKBP5 gene with treatment response after 2 and 5 weeks. Individuals homozygous for the TT-genotype at one of the markers (rs1360780) reported more depressive episodes and responded better to antidepressant treatment. There was no association between markers in FKBP5 and disease. The present study aimed at studying the associated FKBP5 markers in the ethnically diverse Sequenced Treatment Alternatives to Relieve Depression (STAR*D) sample of non-hospitalized patients treated with citalopram. METHODS: We used clinical data and DNA samples from 1809 outpatients with non-psychotic major depressive disorder (DSM-IV criteria), who received up to 14 weeks of citalopram. A subset of 1523 patients of White non-Hispanic or Black race was matched with 739 control subjects for a case-control analysis. The markers rs1360780 and rs4713916 were genotyped on the Illumina platform. TaqMan-assay was used for marker rs3800373. RESULTS: In the case-control analysis, marker rs1360780 was significantly associated with disease status in the White non-Hispanic sample after correction for multiple testing. A significant association was also found between rs4713916 and remission. Markers rs1360780 and rs4713916 were in strong linkage disequilibrium in the White non-Hispanic but not in the Black population. There was no significant difference in the number of previous episodes of depression between genotypes at any of the three markers. CONCLUSIONS: These results indicate that FKBP5 is an important target for further studies of depression and treatment response.

Association of GRIK4 with outcome of antidepressant treatment in the STAR*D cohort.

OBJECTIVE: An initial pharmacogenetic study of the Sequenced Treatment Alternatives to Relieve Depression (STAR*D) clinical trial reported an association between genetic variation in the HTR2A gene and outcome of citalopram treatment. By design, the study analyzed only those markers that showed reproducible association in the first wave of genotypes (comprising 1,297 patients) in the complete cohort of patients. The purpose of the present study was to utilize a second wave of genotype results, for a more powerful analysis, in the complete cohort of patients with available deoxyribonucleic acid (DNA) samples. METHOD: The authors tested the association between treatment response and 768 markers that were genotyped in the full set of 1,816 eligible patients from the STAR*D cohort. In order to control for multiple testing, the subjects were divided into two study groups: discovery and replication. RESULTS: In addition to the previously identified marker in the HTR2A gene, a new marker (rs1954787) in the GRIK4 gene, which codes for the kainic acid-type glutamate receptor KA1, was observed. The effect size of the GRIK4 marker alone was modest, but homozygote carriers of the treatment-response-associated marker alleles of both the GRIK4 and HTR2A genes were 23% less likely to experience nonresponse to treatment relative to participants who did not carry any of these marker alleles. CONCLUSIONS: The findings demonstrate that genetic variation in a kainic acid-type glutamate receptor is reproducibly associated with response to the antidepressant citalopram. This finding suggests that the glutamate system plays an important role in modulating response to selective serotonin reuptake inhibitors (SSRIs).

Association between a functional serotonin transporter promoter polymorphism and citalopram treatment in adult outpatients with major depression.

CONTEXT: The HTTLPR, a functional polymorphism of the serotonin transporter gene solute carrier family 6 (neurotransmitter transporter, serotonin), member 4 (SLC6A4), promoter, affects transcription and may be involved in antidepressant drug treatment outcome, although response rates with antidepressants can be lower in patients who experience adverse effects. OBJECTIVE: To test the hypothesis that HTTLPR is associated with treatment outcome to citalopram. DESIGN: A clinical effectiveness trial, Sequenced Treatment Alternatives to Relieve Depression, collected DNA samples from outpatients with nonpsychotic major depressive disorder who received citalopram in the first treatment step. The triallelic HTTLPR locus was genotyped in 1775 samples to discriminate between long (L) and short (S) alleles, followed by the A > G substitution. The low-expression S and L(G) alleles were grouped together compared with the high-expression L(A) allele. SETTING: Eighteen primary care and 23 psychiatric care sites across the United States. PARTICIPANTS: Ages 18 to 75 years, meeting criteria for single or recurrent nonpsychotic major depression. MAIN OUTCOME MEASURES: Categorical response, remission, tolerance, and adverse effect burden. RESULTS: Expression-based grouping produced a significant finding of association between the L(A) allele and adverse effect burden in the entire sample (P = .004 [genotype frequency]; P < .001 [allele frequency]). To control for bias from population stratification, a white American subsample was analyzed. A lesser adverse effect burden was associated with L(A)L(A) genotype frequency (P = .03) or L(A) allele frequency (P = .007). These findings in white patients did not hold when the L allele was undifferentiated. No association was observed between treatment outcome phenotypes and HTTLPR. Development of diarrhea and the presence of the low-expression S or L(G) alleles were the strongest risk factors associated with adverse effect burden. CONCLUSIONS: The HTTLPR polymorphism is associated with citalopram adverse effects. Because the L(A) allele confers increased SLC6A4 transcription, increased serotonin transporter levels in brain and other tissues may lead to fewer adverse effects for antidepressant medications that target the transporter.

Covariate-based linkage analysis: application of a propensity score as the single covariate consistently improves power to detect linkage.

Successful identification of genetic risk loci for complex diseases has relied on the ability to minimize disease and genetic heterogeneity to increase the power to detect linkage. One means to account for disease heterogeneity is by incorporating covariate data. However, the inclusion of each covariate will add one degree of freedom to the allele sharing based linkage test, which may in fact decrease power. We explore the application of a propensity score, which is typically used in causal inference to combine multiple covariates into a single variable, as a means of allowing for multiple covariates with the addition of only one degree of freedom. In this study, binary trait data, simulated under various models involving genetic and environmental effects, were analyzed using a nonparametric linkage statistic implemented in LODPAL. Power and type I error rates were evaluated. Results suggest that the use of the propensity score to combine multiple covariates as a single covariate consistently improves the power compared to an analysis including no covariates, each covariate individually, or all covariates simultaneously. Type I error rates were inflated for analyses with covariates and increased with increasing number of covariates, but reduced to nominal rates with sample sizes of 1000 families. Therefore, we recommend using the propensity score as a single covariate in the linkage analysis of a trait suspected to be influenced by multiple covariates because of its potential to increase the power to detect linkage, while controlling for the increase in the type I error.

Allele frequency misspecification: effect on power and Type I error of model-dependent linkage analysis of quantitative traits under random ascertainment.

BACKGROUND: Studies of model-based linkage analysis show that trait or marker model misspecification leads to decreasing power or increasing Type I error rate. An increase in Type I error rate is seen when marker related parameters (e.g., allele frequencies) are misspecified and ascertainment is through the trait, but lod-score methods are expected to be robust when ascertainment is random (as is often the case in linkage studies of quantitative traits). In previous studies, the power of lod-score linkage analysis using the "correct" generating model for the trait was found to increase when the marker allele frequencies were misspecified and parental data were missing. An investigation of Type I error rates, conducted in the absence of parental genotype data and with misspecification of marker allele frequencies, showed that an inflation in Type I error rate was the cause of at least part of this apparent increased power. To investigate whether the observed inflation in Type I error rate in model-based LOD score linkage was due to sampling variation, the trait model was estimated from each sample using REGCHUNT, an automated segregation analysis program used to fit models by maximum likelihood using many different sets of initial parameter estimates. RESULTS: The Type I error rates observed using the trait models generated by REGCHUNT were usually closer to the nominal levels than those obtained when assuming the generating trait model. CONCLUSION: This suggests that the observed inflation of Type I error upon misspecification of marker allele frequencies is at least partially due to sampling variation. Thus, with missing parental genotype data, lod-score linkage is not as robust to misspecification of marker allele frequencies as has been commonly thought.

Variation in the gene encoding the serotonin 2A receptor is associated with outcome of antidepressant treatment.

Depressive disorders account for a large and increasing global burden of disease. Although the condition of many patients improves with medication, only a minority experience full remission, and patients whose condition responds to one medication may not have a response to others. Individual variation in antidepressant treatment outcome is, at present, unpredictable but may have a partial genetic basis. We searched for genetic predictors of treatment outcome in 1,953 patients with major depressive disorder who were treated with the antidepressant citalopram in the Sequenced Treatment Alternatives for Depression (STAR*D) study and were prospectively assessed. In a split-sample design, a selection of 68 candidate genes was genotyped, with 768 single-nucleotide-polymorphism markers chosen to detect common genetic variation. We detected significant and reproducible association between treatment outcome and a marker in HTR2A (P range 1 x 10(-6) to 3.7 x 10(-5) in the total sample). Other markers in HTR2A also showed evidence of association with treatment outcome in the total sample. HTR2A encodes the serotonin 2A receptor, which is downregulated by citalopram. Participants who were homozygous for the A allele had an 18% reduction in absolute risk of having no response to treatment, compared with those homozygous for the other allele. The A allele was over six times more frequent in white than in black participants, and treatment was less effective among black participants. The A allele may contribute to racial differences in outcomes of antidepressant treatment. Taken together with prior neurobiological findings, these new genetic data make a compelling case for a key role of HTR2A in the mechanism of antidepressant action.

Critical values and variation in type I error along chromosomes in the COGA dataset using the applied pseudo-trait method.

BACKGROUND: By analyzing a "pseudo-trait," a trait not linked or associated with any of the markers tested, the distribution of the test statistic under the null hypothesis can provide the critical value for the appropriate percentile of the distribution. In addition, the anecdotal observation that p-values tend to be more significant near the telomeres was investigated. RESULTS: The applied pseudo-trait (APT) method was applied to the Affymetrix and Illumina SNPs in the Collaborative Study on the Genetics of Alcoholism dataset to determine appropriate critical values for regression of offspring on mid-parent (ROMP) and Haseman-Elston association and linkage analyses, investigating the occurrence of type I errors in different chromosomal locations, and the extent to which the critical values obtained depend on the type of pseudo-trait used. CONCLUSION: On average, the 5 percentile critical values obtained for this study were less than the expected 0.05. The distribution of p-values does not seem to depend on chromosomal position for ROMP association analysis methods, but does in some cases for Haseman-Elston linkage analysis. Results vary with different pseudo-traits.

Importance sampling method of correction for multiple testing in affected sib-pair linkage analysis.

Using the Genetic Analysis Workshop 13 simulated data set, we compared the technique of importance sampling to several other methods designed to adjust p-values for multiple testing: the Bonferroni correction, the method proposed by Feingold et al., and naïve Monte Carlo simulation. We performed affected sib-pair linkage analysis for each of the 100 replicates for each of five binary traits and adjusted the derived p-values using each of the correction methods. The type I error rates for each correction method and the ability of each of the methods to detect loci known to influence trait values were compared. All of the methods considered were conservative with respect to type I error, especially the Bonferroni method. The ability of these methods to detect trait loci was also low. However, this may be partially due to a limitation inherent in our binary trait definitions.