Information extraction from free text for aiding transdiagnostic psychiatry: constructing NLP pipelines tailored to clinicians' needs.

BACKGROUND: Developing predictive models for precision psychiatry is challenging because of unavailability of the necessary data: extracting useful information from existing electronic health record (EHR) data is not straightforward, and available clinical trial datasets are often not representative for heterogeneous patient groups. The aim of this study was constructing a natural language processing (NLP) pipeline that extracts variables for building predictive models from EHRs. We specifically tailor the pipeline for extracting information on outcomes of psychiatry treatment trajectories, applicable throughout the entire spectrum of mental health disorders ("transdiagnostic"). METHODS: A qualitative study into beliefs of clinical staff on measuring treatment outcomes was conducted to construct a candidate list of variables to extract from the EHR. To investigate if the proposed variables are suitable for measuring treatment effects, resulting themes were compared to transdiagnostic outcome measures currently used in psychiatry research and compared to the HDRS (as a gold standard) through systematic review, resulting in an ideal set of variables. To extract these from EHR data, a semi-rule based NLP pipeline was constructed and tailored to the candidate variables using Prodigy. Classification accuracy and F1-scores were calculated and pipeline output was compared to HDRS scores using clinical notes from patients admitted in 2019 and 2020. RESULTS: Analysis of 34 questionnaires answered by clinical staff resulted in four themes defining treatment outcomes: symptom reduction, general well-being, social functioning and personalization. Systematic review revealed 242 different transdiagnostic outcome measures, with the 36-item Short-Form Survey for quality of life (SF36) being used most consistently, showing substantial overlap with the themes from the qualitative study. Comparing SF36 to HDRS scores in 26 studies revealed moderate to good correlations (0.62-0.79) and good positive predictive values (0.75-0.88). The NLP pipeline developed with notes from 22,170 patients reached an accuracy of 95 to 99 percent (F1 scores: 0.38 - 0.86) on detecting these themes, evaluated on data from 361 patients. CONCLUSIONS: The NLP pipeline developed in this study extracts outcome measures from the EHR that cater specifically to the needs of clinical staff and align with outcome measures used to detect treatment effects in clinical trials.

Genome wide association identifies common variants at the SERPINA6/SERPINA1 locus influencing plasma cortisol and corticosteroid binding globulin.

Variation in plasma levels of cortisol, an essential hormone in the stress response, is associated in population-based studies with cardio-metabolic, inflammatory and neuro-cognitive traits and diseases. Heritability of plasma cortisol is estimated at 30-60% but no common genetic contribution has been identified. The CORtisol NETwork (CORNET) consortium undertook genome wide association meta-analysis for plasma cortisol in 12,597 Caucasian participants, replicated in 2,795 participants. The results indicate that <1% of variance in plasma cortisol is accounted for by genetic variation in a single region of chromosome 14. This locus spans SERPINA6, encoding corticosteroid binding globulin (CBG, the major cortisol-binding protein in plasma), and SERPINA1, encoding α1-antitrypsin (which inhibits cleavage of the reactive centre loop that releases cortisol from CBG). Three partially independent signals were identified within the region, represented by common SNPs; detailed biochemical investigation in a nested sub-cohort showed all these SNPs were associated with variation in total cortisol binding activity in plasma, but some variants influenced total CBG concentrations while the top hit (rs12589136) influenced the immunoreactivity of the reactive centre loop of CBG. Exome chip and 1000 Genomes imputation analysis of this locus in the CROATIA-Korcula cohort identified missense mutations in SERPINA6 and SERPINA1 that did not account for the effects of common variants. These findings reveal a novel common genetic source of variation in binding of cortisol by CBG, and reinforce the key role of CBG in determining plasma cortisol levels. In turn this genetic variation may contribute to cortisol-associated degenerative diseases.

Medical comorbidities in children and adolescents with autism spectrum disorders and attention deficit hyperactivity disorders: a systematic review.

Somatic disorders occur more often in adult psychiatric patients than in the general adult population. However, in child and adolescent psychiatry this association is unclear, mainly due to a lack of integration of existing data. To address this issue, we here present a systematic review on medical comorbidity in the two major developmental disorders autism spectrum disorder (ASD) and attention deficit hyperactivity disorder (ADHD) and formulate clinical recommendations. The literature was searched using the PubMed and PsycINFO databases (2000-1 May 2016) with the keywords "[((child and adolescent) AND (Autism OR Attention Deficit Hyperactivity Disorder* OR ADHD)) AND ("Cardiovascular Diseases" [Mesh] OR "Endocrine System Diseases" [Mesh] OR "Immune System Diseases" [Mesh] OR "Neurobehavioral Manifestations" [Mesh] OR "Gastrointestinal Disorders" [Mesh] OR Somatic OR Autoimmune disease OR Nervous system disease OR Infection OR Infectious disease)]. Two raters independently assessed the quality of the eligible studies. The initial search identified 5278 articles. Based on inclusion and exclusion criteria 104 papers were selected and subsequently subjected to a quality control. This quality was assessed according to a standardized and validated set of criteria and yielded 29 studies for inclusion. This thorough literature search provides an overview of relevant articles on medical comorbidity in ADHD and/or ASD, and shows that medical disorders in these children and adolescents appear to be widespread. Those who work with children with ASD and/or ADHD should be well aware of this and actively promote routine medical assessment. Additionally, case-control studies and population-based studies are needed to provide reliable prevalence estimates. Future studies should furthermore focus on a broader evaluation of medical disorders in children and adolescents with ADHD and/or ASD to improve treatment algorithm in this vulnerable group.

A genome-wide approach to children's aggressive behavior: The EAGLE consortium.

Individual differences in aggressive behavior emerge in early childhood and predict persisting behavioral problems and disorders. Studies of antisocial and severe aggression in adulthood indicate substantial underlying biology. However, little attention has been given to genome-wide approaches of aggressive behavior in children. We analyzed data from nine population-based studies and assessed aggressive behavior using well-validated parent-reported questionnaires. This is the largest sample exploring children's aggressive behavior to date (N = 18,988), with measures in two developmental stages (N = 15,668 early childhood and N = 16,311 middle childhood/early adolescence). First, we estimated the additive genetic variance of children's aggressive behavior based on genome-wide SNP information, using genome-wide complex trait analysis (GCTA). Second, genetic associations within each study were assessed using a quasi-Poisson regression approach, capturing the highly right-skewed distribution of aggressive behavior. Third, we performed meta-analyses of genome-wide associations for both the total age-mixed sample and the two developmental stages. Finally, we performed a gene-based test using the summary statistics of the total sample. GCTA quantified variance tagged by common SNPs (10-54%). The meta-analysis of the total sample identified one region in chromosome 2 (2p12) at near genome-wide significance (top SNP rs11126630, P = 5.30 × 10(-8) ). The separate meta-analyses of the two developmental stages revealed suggestive evidence of association at the same locus. The gene-based analysis indicated association of variation within AVPR1A with aggressive behavior. We conclude that common variants at 2p12 show suggestive evidence for association with childhood aggression. Replication of these initial findings is needed, and further studies should clarify its biological meaning. © 2015 Wiley Periodicals, Inc.

The role of maternal stress during pregnancy, maternal discipline, and child COMT Val158Met genotype in the development of compliance.

Maternal discipline is an important predictor of child committed compliance. Maternal stress can affect both parenting and child development. In a large population-based cohort study (N = 613) we examined whether maternal discipline mediated the association between maternal stress during pregnancy and child compliance, and whether COMT or DRD4 polymorphisms moderated the association between maternal discipline and child compliance. Family-related and general stress were measured through maternal self-report and genetic material was collected through cord blood sampling at birth. Mother-child dyads were observed at 36 months in disciplinary tasks in which the child was not allowed to touch attractive toys. Maternal discipline and child compliance were observed in two different tasks and independently coded. The association between family stress during pregnancy and child committed compliance was mediated by maternal positive discipline. Children with more COMT Met alleles seemed more susceptible to maternal positive discipline than children with more COMT Val alleles.

Maternal smoking during pregnancy and child emotional problems: the relevance of maternal and child 5-HTTLPR genotype.

Serotonin is involved in the development of neural circuits modulating emotional behavior. The short allele (s) of a polymorphism (5-HTTLPR) of the serotonin transporter gene is a risk factor for psychopathology in the presence of environmental stressors. Maternal smoking is associated with growth restriction of the human fetal brain and adverse effects of nicotine on the developing serotonin system have been documented. We hypothesized that maternal smoking interacts with both child and mother 5-HTTLPR genotype as a risk factor for later child emotional problems. In a sample of n = 1,529 mother-child dyads, smoking habits were assessed by questionnaires during pregnancy. Child emotional problems were measured by the Child Behavior Checklist at the child's age of 3 years. Maternal smoking during pregnancy significantly increased the risk for emotional problems in children carrying the s-allele; ß = 0.24, P = 0.03 (mother-report), and ß = 0.46, P = 0.001 (father-report). In children heterozygous at 5-HTTLPR and exposed to maternal prenatal smoking (n = 79) risk of emotional problems increased with each additional s-allele the mother carried. The associations between 5-HTTLPR and child emotional problems were not moderated by paternal prenatal smoking. These findings imply that the vulnerability for emotional problems in s-allele carriers may already originate in fetal life.

Prenatal and postnatal psychological symptoms of parents and family functioning: the impact on child emotional and behavioural problems.

Although relations of various parental psychological problems and family functioning with child development are well documented, it remains unclear whether specific prenatal or specific postnatal risk factors are independently associated with child emotional and behavioural problems, or whether observed associations can be explained by general parental psychopathology. Using a stepwise approach, we examined the effects of prenatal and postnatal parental depressive symptoms, prenatal and postnatal hostility of the parents, as well as prenatal family functioning on the risk of child emotional and behavioural problems. This study was embedded in Generation R: a population-based cohort from foetal life onwards. Mothers and fathers of 2,698 children provided information about depressive symptoms, symptoms of hostility and family functioning during pregnancy and 3 years after birth. Mother and father each reported on child behaviour when the child was 3 years old. Parental depressive symptoms increased the risk of child emotional and behavioural problems, but this increase was explained by postnatal parental hostile behaviour. Postnatal symptoms of hostility of mothers (OR = 1.34, p value <0.001) and postnatal symptoms of hostility of fathers (OR = 1.30, p value <0.001) each contributed independently to the risk of child emotional and behavioural problems. Postnatal parental hostility is associated with an increased risk of child emotional and behavioural problems, independent of parental depressive symptoms. These findings suggest that prevention and intervention strategies should focus on psychological symptoms of both mothers and fathers, in particular on hostile behaviour, in families with young children.

The longitudinal association of the diurnal cortisol rhythm with internalizing and externalizing problems in pre-schoolers. The Generation R Study.

BACKGROUND: Studies investigating the association between diurnal cortisol rhythm and behavioural problems in young children have yielded inconsistent results. We tested the hypothesis that variations in diurnal cortisol rhythm in pre-schoolers are already related to problem behaviour early in life with a cross-sectional and longitudinal design. METHODS: This study was embedded in Generation R, a population-based cohort from foetal life onwards. Parents collected saliva samples from their infant at 5 moments during 1 day. In 322 infants aged 12-20 months, we determined the diurnal cortisol rhythm by calculating the area under the curve (AUC), the cortisol awakening response (CAR), and the diurnal slope. Problem behaviour was assessed at ages 1.5 and 3 years with the Child Behavior Checklist/1.5-5 years. RESULTS: No cross-sectional associations between the cortisol composite measures and problem behaviour were found at 1.5 years. However, cortisol predicted change in internalizing problems as assessed from 1.5 to 3 years, but not change in externalizing problems. Children with higher AUC levels, flatter slopes and a more positive CAR at baseline were more likely to score higher on the Internalizing Problems scale (ß per nmol/L AUC: 0.08, 95% CI: 0.00; 0.17, p=0.04; ß per nmol/L/h slope: 0.57, 95% CI: 0.17; 0.98, p=0.006; ß per nmol/L CAR: 0.04, 95% CI: 0.01; 0.08, p=0.02) at follow-up. CONCLUSIONS: Variations in diurnal cortisol rhythm are associated with change in internalizing problems in pre-schoolers. The results suggest that variations in diurnal cortisol patterns early in life precede internalizing problems.

Common variation near ROBO2 is associated with expressive vocabulary in infancy.

Twin studies suggest that expressive vocabulary at ~24 months is modestly heritable. However, the genes influencing this early linguistic phenotype are unknown. Here we conduct a genome-wide screen and follow-up study of expressive vocabulary in toddlers of European descent from up to four studies of the EArly Genetics and Lifecourse Epidemiology consortium, analysing an early (15-18 months, 'one-word stage', N(Total) = 8,889) and a later (24-30 months, 'two-word stage', N(Total)=10,819) phase of language acquisition. For the early phase, one single-nucleotide polymorphism (rs7642482) at 3p12.3 near ROBO2, encoding a conserved axon-binding receptor, reaches the genome-wide significance level (P=1.3 × 10(-8)) in the combined sample. This association links language-related common genetic variation in the general population to a potential autism susceptibility locus and a linkage region for dyslexia, speech-sound disorder and reading. The contribution of common genetic influences is, although modest, supported by genome-wide complex trait analysis (meta-GCTA h(2)(15-18-months) = 0.13, meta-GCTA h(2)(24-30-months) = 0.14) and in concordance with additional twin analysis (5,733 pairs of European descent, h(2)(24-months) = 0.20).

A genome-wide association meta-analysis of preschool internalizing problems.

OBJECTIVE: Preschool internalizing problems (INT) are highly heritable and moderately genetically stable from childhood into adulthood. Gene-finding studies are scarce. In this study, the influence of genome-wide measured single nucleotide polymorphisms (SNPs) was investigated in 3 cohorts (total N = 4,596 children) in which INT was assessed with the same instrument, the Child Behavior Checklist (CBCL). METHOD: First, genome-wide association (GWA) results were used for density estimation and genome-wide complex trait analysis (GCTA) to calculate the variance explained by all SNPs. Next, a fixed-effect inverse variance meta-analysis of the 3 GWA analyses was carried out. Finally, the overlap in results with prior GWA studies of childhood and adulthood psychiatric disorders and treatment responses was tested by examining whether SNPs associated with these traits jointly showed a significant signal for INT. RESULTS: Genome-wide SNPs explained 13% to 43% of the total variance. This indicates that the genetic architecture of INT mirrors the polygenic model underlying adult psychiatric traits. The meta-analysis did not yield a genome-wide significant signal but was suggestive for the PCSK2 gene located on chromosome 20p12.1. SNPs associated with other psychiatric disorders appeared to be enriched for signals with INT (λ = 1.26, p < .03). CONCLUSION: Our study provides evidence that INT is influenced by many common genetic variants, each with a very small effect, and that, even as early as age 3, genetic variants influencing INT overlap with variants that play a role in childhood and adulthood psychiatric disorders.

FTO at rs9939609, food responsiveness, emotional control and symptoms of ADHD in preschool children.

The FTO minor allele at rs9939609 has been associated with body mass index (BMI: weight (kg)/height (m)(2)) in children from 5 years onwards, food intake, and eating behaviour. The high expression of FTO in the brain suggests that this gene may also be associated with behavioural phenotypes, such as impulsivity and control. We examined the effect of the FTO minor allele (A) at rs9939609 on eating behaviour, impulsivity and control in young children, thus before the BMI effect becomes apparent. This study was embedded in the Generation R Study, a population-based cohort from fetal life onwards. 1,718 children of European descent were genotyped for FTO at rs9939609. With logistic regression assuming an additive genetic model, we examined the association between the FTO minor allele and eating behaviour, impulsivity and control in preschool children. There was no relation between FTO at rs9939609 and child BMI at this age. The A allele at rs9939609 was associated with increased food responsiveness (OR 1.21, p = 0.03). Also, children with the A allele were less likely to have symptoms of ADHD (OR 0.74, p = 0.01) and showed more emotional control (OR 0.64, p = 0.01) compared to children without the A allele. Our findings suggest that before the association between FTO and BMI becomes apparent, the FTO minor allele at rs9939609 leads to increased food responsiveness, a decreased risk for symptoms of ADHD and better emotional control. Future studies are needed to investigate whether these findings represent one single mechanism or reflect pleiotropic effects of FTO.

Variation in the glucocorticoid receptor gene at rs41423247 moderates the effect of prenatal maternal psychological symptoms on child cortisol reactivity and behavior.

Prenatal maternal psychopathology affects child development, but some children seem more vulnerable than others. Genetic variance in hypothalamic-pituitary-adrenal axis genes may influence the effect of prenatal maternal psychological symptoms on child emotional and behavioral problems. This hypothesis was tested in the Generation R Study, a population-based cohort from fetal life onward. In total, 1727 children of Northern European descent and their mothers participated in this study and were genotyped for variants in the glucocorticoid receptor (GR) gene (rs6189/rs6190, rs10052957, rs41423247, rs6195, and rs6198) and the FK506-binding protein 5 (FKBP5) gene (rs1360780). Prenatal maternal psychological symptoms were assessed at 20 weeks pregnancy and child behavior was assessed by both parents at 3 years. In a subsample of 331 children, data about cortisol reactivity were available. Based on power calculations, only those genetic variants with sufficient minor allele frequencies (rs41423247, rs10052957, and rs1360780) were included in the interaction analyses. We found that variation in GR at rs41423247 moderates the effect of prenatal maternal psychological symptoms on child emotional and behavioral problems (beta 0.41, SE 0.16, p=0.009). This prenatal interaction effect was independent of mother's genotype and maternal postnatal psychopathology, and not found for prenatal psychological symptoms of the father. Moreover, the interaction between rs41423247 and prenatal psychological symptoms was also associated with decreased child cortisol reactivity (beta -2.30, p-value 0.05). These findings emphasize the potential effect of prenatal gene-environment interaction, and give insight in possible mechanisms accounting for children's individual vulnerability to develop emotional and behavioral problems.

The Generation R Study: a review of design, findings to date, and a study of the 5-HTTLPR by environmental interaction from fetal life onward.

OBJECTIVE: First, we give an overview of child psychiatric research in the Generation R Study, a population-based cohort from fetal life forward. Second, we examine within Generation R whether the functional polymorphism (5-HTTLPR) in the promoter of the serotonin transporter gene interacts with prenatal maternal chronic difficulties, prenatal maternal anxiety or postnatal maternal anxiety to influence child emotional development. METHOD: A total of 2,136 northern European children were genotyped for 5-HTTLPR and rs25531. Mothers reported chronic difficulties and anxiety symptoms at 20 weeks' pregnancy and when the child was 3 years old. Child emotion recognition was observed at 3 years, and child emotional problems were assessed with the CBCL/1½-5 at 5 years. RESULTS: There were consistent main effects of maternal difficulties and anxiety on child emotional problems, but no main effect of 5-HTTLPR. Moreover, children with the s allele were at increased risk for emotional problems if their mothers reported prenatal anxiety symptoms (ß = 2.02, p < .001) or postnatal anxiety symptoms (ß = 1.64, p < 0.001). Also, in children of mothers with prenatal anxiety symptoms, the s allele was associated with less accurate emotion-matching (ß = -0.11, p = .004). CONCLUSIONS: This population-based study shows that vulnerability due to 5-HTTLPR is not specific for certain adverse exposures or severe events, but suggests that the small effects of gene-environment interaction on emotional development become manifest early in life.

Genetics of cortisol secretion and depressive symptoms: a candidate gene and genome wide association approach.

BACKGROUND: Depressive patients often have altered cortisol secretion, but few studies have investigated genetic variants in relation to both cortisol secretion and depression. To identify genes related to both these conditions, we: (1) tested the association of single nucleotide polymorphisms (SNPs) in hypothalamic-pituitary-adrenal-axis (HPA-axis) candidate genes with a summary measure of total cortisol secretion during the day (cortisol(AUC)), (2) performed a genome wide association study (GWAS) of cortisol(AUC), and (3) tested the association of identified cortisol-related SNPs with depressive symptoms. METHODS: We analyzed data on candidate SNPs for the HPA-axis, genome-wide scans, cortisol secretion (n=1711) and depressive symptoms (the Centre for Epidemiology Studies Depression Scale, CES-D) (n=2928) in elderly persons of the Rotterdam Study. We used data from the Whitehall II study (n=2836) to replicate the GWAS findings. RESULTS: Of the 1456 SNPs in 33 candidate genes, minor alleles of 4 SNPs (rs9470080, rs9394309, rs7748266 and rs1360780) in the FKBP5 gene were associated with a decreased cortisol(AUC) (p<1×10(-4) after correction for multiple testing using permutations). These SNPs were also associated with an increased risk of depressive symptoms (rs9470080: OR 1.19 (95%CI 1.0; 1.4)). The GWAS for cortisol yielded 2 SNPs with p-values of 1×10(-06) (rs8062512, rs2252459), but these associations could not be replicated. CONCLUSIONS: These results suggest that variation in the FKBP5 gene is associated with both cortisol(AUC) and the likelihood of depressive symptoms.

Serotonin transporter polymorphism moderates effects of prenatal maternal anxiety on infant negative emotionality.

BACKGROUND: Consistent with the fetal programming hypothesis, effects of maternal prenatal anxiety have been found to predict various measures of infant temperament in the early postnatal period. In recent years, a polymorphism in the serotonin transporter gene (5-HTTLPR) emerged as a moderator of diverse environmental influences on different outcomes, with individuals carrying the short allele being generally more vulnerable to adversity. METHODS: We tested whether the association between self-reported maternal anxiety at 20 weeks gestation (Brief Symptom Inventory) and mother-rated infant negative emotionality at 6 months after birth (Infant Behavior Questionnaire-Revised) would be moderated by the 5-HTTLPR in a large Dutch cohort sample (n = 1513). We hypothesized that infants carrying the 5-HTTLPR short allele would be more susceptible and therefore more affected by both low and high prenatal maternal anxiety vis-à-vis negative emotionality than other genotypes. RESULTS: Findings of a significant gene × environment interaction (B = .65, p = .01) were supportive of a vulnerability model, with infants carrying the short allele being more negatively emotional when mothers reported anxiety during pregnancy, whereas there was no difference between genotypes on negative emotionality when maternal anxiety was low. CONCLUSIONS: The association between maternal anxiety during pregnancy and negative emotionality in early infancy was significant in infants carrying one or more copies of the short allele but not in those homozygous for the long allele. The 5-HTTLPR short allele might increase vulnerability to adverse environmental influences as early as the fetal period.

FKBP5 and resistant attachment predict cortisol reactivity in infants: gene-environment interaction.

Quality of the parent-infant attachment relationship influences physiological stress regulation. Genetic factors also contribute to the stress regulatory HPA-axis. Quality of attachment as an index of the rearing environment (measured with the Strange Situation Procedure, SSP), and HPA-axis related SNPs (BclI, rs41423247; TthIIII, rs10052957; GR-9ß, rs6198; N363S, rs6195; ER22/23EK, rs6189 and 6190; and FKBP5, rs1360780) were hypothesized to be related to cortisol reactivity in the stressful SSP. In this large population based sample, FKBP5 rs1360780, but not GR haplotype, was related to cortisol reactivity. Moreover, we found a significant interaction effect for insecure-resistant attachment and FKBP5 rs1360780, indicating a double-risk for heightened cortisol reactivity levels in infants with one or two T-alleles of the FKBP5 SNP and an insecure-resistant attachment relationship with their mother. Findings are discussed from the perspective of gene-environment interaction.