Table 1.
Data set | rs10158645 MAF |
CSF KYNA (nM)
|
CSF HVA (nM)
|
Psychosis
|
||||||||
---|---|---|---|---|---|---|---|---|---|---|---|---|
N | r | P | IEa | N | r | P | IEb | N | OR (95% CI) | P | ||
II | 0.15 | 76 | 0.58 | 3 × 10−8 | 65 (36–101)c | 72 | 0.54 | 7 × 10−7 | 0.71 (0.15–1.8)c | 76 | 4.0 (1.4–12) | 0.011 |
III | 0.15 | 30 | 0.48d | 7 × 10−3 | ||||||||
IV | 0.18 | 565 | 1.5 (1.1–2.0) 6.5 (1.5–28)e |
0.020 0.013e |
||||||||
II+IV | 0.17 | 641 | 1.6 (1.2–2.2) 7.3 (1.7–32)e |
2 × 10−3 8 × 10−3 |
Abbreviations: BD, bipolar disorder; CI, confidence interval; CSF, cerebrospinal fluid; HVA, homovanillic acid; IE, indirect effect; KYNA, kynurenic acid; MAF, minor allele frequency; OR, odds ratio. All reported P-values are two sided.
IE = indirect effect for mediation. Non-parametric bootstrap method, the PROCESS script for SPSS (see http://afhayes.com/spss-sas-and-mplus-macros-and-code.html). Causal variable = allele frequency in rs10158645, mediator = CSF KYNA concentration and outcome variable = CSF HVA concentration. Direct effect = 43 (−0.53 to 86).
Causal variable = CSF KYNA concentration, mediator = CSF HVA concentration and outcome variable = psychosis (0/1). Direct effect = 0.15 (−0.61to 0.90).
95% Bootstrap CI (percentile bootstrap = 20 000).
Spearman’s coefficient.
In sample I, only one patient was homozygous for the minor G allele in rs1018645 and the analysis was performed under an additive model. In sample II, 21 patients were homozygous for the minor G allele (18 with psychosis) and additional analyses under a recessive model was performed in sample II and the combined sample (I+II).