Limits...
Multivariate genetic determinants of EEG oscillations in schizophrenia and psychotic bipolar disorder from the BSNIP study.

Narayanan B, Soh P, Calhoun VD, Ruaño G, Kocherla M, Windemuth A, Clementz BA, Tamminga CA, Sweeney JA, Keshavan MS, Pearlson GD - Transl Psychiatry (2015)

Bottom Line: We assessed eight data-driven EEG frequency activity derived from group-independent component analysis (ICA) in conjunction with a reduced subset of 10,422 SNPs through novel multivariate association using parallel ICA (para-ICA).Para-ICA extracted five frequency and nine SNP components, of which theta and delta activities were significantly correlated with two different gene components, comprising genes participating extensively in brain development, neurogenesis and synaptogenesis.The gene clusters were enriched for biological properties affecting neural circuitry and involved in brain function and/or development.

View Article: PubMed Central - PubMed

Affiliation: Olin Neuropsychiatry Research Center, Hartford Hospital, Institute of Living, Hartford, CT, USA.

ABSTRACT
Schizophrenia (SZ) and psychotic bipolar disorder (PBP) are disabling psychiatric illnesses with complex and unclear etiologies. Electroencephalogram (EEG) oscillatory abnormalities in SZ and PBP probands are heritable and expressed in their relatives, but the neurobiology and genetic factors mediating these abnormalities in the psychosis dimension of either disorder are less explored. We examined the polygenic architecture of eyes-open resting state EEG frequency activity (intrinsic frequency) from 64 channels in 105 SZ, 145 PBP probands and 56 healthy controls (HCs) from the multisite BSNIP (Bipolar-Schizophrenia Network on Intermediate Phenotypes) study. One million single-nucleotide polymorphisms (SNPs) were derived from DNA. We assessed eight data-driven EEG frequency activity derived from group-independent component analysis (ICA) in conjunction with a reduced subset of 10,422 SNPs through novel multivariate association using parallel ICA (para-ICA). Genes contributing to the association were examined collectively using pathway analysis tools. Para-ICA extracted five frequency and nine SNP components, of which theta and delta activities were significantly correlated with two different gene components, comprising genes participating extensively in brain development, neurogenesis and synaptogenesis. Delta and theta abnormality was present in both SZ and PBP, while theta differed between the two disorders. Theta abnormalities were also mediated by gene clusters involved in glutamic acid pathways, cadherin and synaptic contact-based cell adhesion processes. Our data suggest plausible multifactorial genetic networks, including novel and several previously identified (DISC1) candidate risk genes, mediating low frequency delta and theta abnormalities in psychoses. The gene clusters were enriched for biological properties affecting neural circuitry and involved in brain function and/or development.

No MeSH data available.


Related in: MedlinePlus

Q–Q plot of theoretical and empirical P-values from logistic regression for (a) schizophrenia (SZ) vs healthy controls (HCs) and (b) psychotic bipolar disorder (PBP) vs HCs. Logistic regression was applied to individual markers in case–control fashion for both SZ and PBP groups.
© Copyright Policy - open-access
Related In: Results  -  Collection

License
getmorefigures.php?uid=PMC4490286&req=5

fig1: Q–Q plot of theoretical and empirical P-values from logistic regression for (a) schizophrenia (SZ) vs healthy controls (HCs) and (b) psychotic bipolar disorder (PBP) vs HCs. Logistic regression was applied to individual markers in case–control fashion for both SZ and PBP groups.

Mentions: Raw SNP data in categorical format (homozygous (AA, BB) and heterozygous (AB or BA)) were numerically coded for the number of minor alleles (AA=0, AB=1 and BB=2, assuming B is a minor allele) based on additive model. Supplementary Figure S1 illustrates the processing pipeline for quality control of SNPs. The first stage of SNP data processing removed individuals with poor genotyping quality by inspecting for discordant sex, exceeding missing rate (>3%), abnormal heterozygosity (>3 s.d. from mean) and unusual relatedness between individuals (identity by descent >0.1875). Exclusion criteria for individual SNPs42 were as follows: minor allele frequency <5% call rate <98% P<0.00001 for deviation from Hardy–Weinberg expectation (in unrelated unaffected individuals); linkage disequilibrium >0.8 in block sizes of 10 kb; significantly differing genotype call rates between cases (SZ and PBP probands) and controls (P<0.00001). There were 575 687 autosomal SNPs after quality control analysis. SNP data were adjusted for hidden population stratification (PCA-based eigenstrat) to minimize false positives by identifying and correcting those PCA components (top three in the current sample)43 for which the loading coefficients (LCs) were significantly associated with the self-reported ethnicity. No significant difference in LCs was detected between cases and controls. The q–q plot (see Figure 1) shows no substantial inflation in the SNP data. To gain statistical power, the SNP quality control and univariate analyses were carried out on all 620 subjects with genotype data, but only data specific to subjects (n=306) from the current sample were selected for multivariate association analysis.


Multivariate genetic determinants of EEG oscillations in schizophrenia and psychotic bipolar disorder from the BSNIP study.

Narayanan B, Soh P, Calhoun VD, Ruaño G, Kocherla M, Windemuth A, Clementz BA, Tamminga CA, Sweeney JA, Keshavan MS, Pearlson GD - Transl Psychiatry (2015)

Q–Q plot of theoretical and empirical P-values from logistic regression for (a) schizophrenia (SZ) vs healthy controls (HCs) and (b) psychotic bipolar disorder (PBP) vs HCs. Logistic regression was applied to individual markers in case–control fashion for both SZ and PBP groups.
© Copyright Policy - open-access
Related In: Results  -  Collection

License
Show All Figures
getmorefigures.php?uid=PMC4490286&req=5

fig1: Q–Q plot of theoretical and empirical P-values from logistic regression for (a) schizophrenia (SZ) vs healthy controls (HCs) and (b) psychotic bipolar disorder (PBP) vs HCs. Logistic regression was applied to individual markers in case–control fashion for both SZ and PBP groups.
Mentions: Raw SNP data in categorical format (homozygous (AA, BB) and heterozygous (AB or BA)) were numerically coded for the number of minor alleles (AA=0, AB=1 and BB=2, assuming B is a minor allele) based on additive model. Supplementary Figure S1 illustrates the processing pipeline for quality control of SNPs. The first stage of SNP data processing removed individuals with poor genotyping quality by inspecting for discordant sex, exceeding missing rate (>3%), abnormal heterozygosity (>3 s.d. from mean) and unusual relatedness between individuals (identity by descent >0.1875). Exclusion criteria for individual SNPs42 were as follows: minor allele frequency <5% call rate <98% P<0.00001 for deviation from Hardy–Weinberg expectation (in unrelated unaffected individuals); linkage disequilibrium >0.8 in block sizes of 10 kb; significantly differing genotype call rates between cases (SZ and PBP probands) and controls (P<0.00001). There were 575 687 autosomal SNPs after quality control analysis. SNP data were adjusted for hidden population stratification (PCA-based eigenstrat) to minimize false positives by identifying and correcting those PCA components (top three in the current sample)43 for which the loading coefficients (LCs) were significantly associated with the self-reported ethnicity. No significant difference in LCs was detected between cases and controls. The q–q plot (see Figure 1) shows no substantial inflation in the SNP data. To gain statistical power, the SNP quality control and univariate analyses were carried out on all 620 subjects with genotype data, but only data specific to subjects (n=306) from the current sample were selected for multivariate association analysis.

Bottom Line: We assessed eight data-driven EEG frequency activity derived from group-independent component analysis (ICA) in conjunction with a reduced subset of 10,422 SNPs through novel multivariate association using parallel ICA (para-ICA).Para-ICA extracted five frequency and nine SNP components, of which theta and delta activities were significantly correlated with two different gene components, comprising genes participating extensively in brain development, neurogenesis and synaptogenesis.The gene clusters were enriched for biological properties affecting neural circuitry and involved in brain function and/or development.

View Article: PubMed Central - PubMed

Affiliation: Olin Neuropsychiatry Research Center, Hartford Hospital, Institute of Living, Hartford, CT, USA.

ABSTRACT
Schizophrenia (SZ) and psychotic bipolar disorder (PBP) are disabling psychiatric illnesses with complex and unclear etiologies. Electroencephalogram (EEG) oscillatory abnormalities in SZ and PBP probands are heritable and expressed in their relatives, but the neurobiology and genetic factors mediating these abnormalities in the psychosis dimension of either disorder are less explored. We examined the polygenic architecture of eyes-open resting state EEG frequency activity (intrinsic frequency) from 64 channels in 105 SZ, 145 PBP probands and 56 healthy controls (HCs) from the multisite BSNIP (Bipolar-Schizophrenia Network on Intermediate Phenotypes) study. One million single-nucleotide polymorphisms (SNPs) were derived from DNA. We assessed eight data-driven EEG frequency activity derived from group-independent component analysis (ICA) in conjunction with a reduced subset of 10,422 SNPs through novel multivariate association using parallel ICA (para-ICA). Genes contributing to the association were examined collectively using pathway analysis tools. Para-ICA extracted five frequency and nine SNP components, of which theta and delta activities were significantly correlated with two different gene components, comprising genes participating extensively in brain development, neurogenesis and synaptogenesis. Delta and theta abnormality was present in both SZ and PBP, while theta differed between the two disorders. Theta abnormalities were also mediated by gene clusters involved in glutamic acid pathways, cadherin and synaptic contact-based cell adhesion processes. Our data suggest plausible multifactorial genetic networks, including novel and several previously identified (DISC1) candidate risk genes, mediating low frequency delta and theta abnormalities in psychoses. The gene clusters were enriched for biological properties affecting neural circuitry and involved in brain function and/or development.

No MeSH data available.


Related in: MedlinePlus