Estimating effects of proband characteristics on familial risk: II. The association between age at onset and familial risk in the Maryland schizophrenia sample

Family Genetic Risk Scores and the Genetic Architecture of Major Affective and Psychotic Disorders in a Swedish National Sample

Importance

Family and genetic approaches have traditionally been used to evaluate our diagnostic concepts. Using a novel method, the family genetic risk score (FGRS), can we validate the genetic architecture of major affective and psychotic disorders in a national Swedish sample?

Objective

To determine whether FGRSs, calculated for the entire Swedish population, can elucidate the genetic relationship between major affective and psychotic disorders and clarify the association of genetic risk with important clinical features of disease.

Design, Setting, and Participants

This cohort study included the native Swedish population born from January 1, 1950, through December 31, 1995, and followed up through December 31, 2017. Data were collected from Swedish population-based primary care, specialist, and hospital registers, including age at first registration for a psychiatric diagnosis and number of registrations for major depression, bipolar disorder, and schizophrenia. Data were analyzed from October 15, 2020, to February 2, 2021.

Exposures

FGRSs for major depression, bipolar disorder, and schizophrenia calculated from morbidity risks for disorders in first- through fifth-degree relatives, controlling for cohabitation.

Main Outcomes and Measures

Diagnoses of major depression, bipolar disorder, schizophrenia, schizoaffective disorder, and other nonaffective psychoses (ONAPs), age at registration, and number of registrations for major depression, bipolar disorder, and schizophrenia. Diagnostic conversion of major depression to bipolar disorder and ONAPs to schizophrenia was assessed by Cox proportional hazards regression models.

Results

The cohort included 4 129 002 individuals (51.4% male) with a mean (SD) age at follow-up of 45.5 (13.4) years. Mean FGRSs for major depression, bipolar disorder, and schizophrenia produced distinct patterns for major depression, bipolar disorder, schizophrenia, schizoaffective disorder, and ONAPs with large separations between disorders. In major depression, bipolar disorder, and schizophrenia, high FGRSs were associated with early age at onset and high rates of recurrence: a high mean FGRS for bipolar disorder was associated with early age at onset (younger than 25 years, 0.11; 95% CI, 0.11-0.12) and higher recurrence (8 or more registrations, 0.11; 95% CI, 0.11-0.12) in major depression. The schizophrenia FGRS was separately associated with psychotic and nonpsychotic forms of major depression (0.10; 95% CI, 0.06-0.14 vs 0.03; 95% CI, 0.02-0.03) and bipolar disorder (0.22; 95% CI, 0.16-0.28 vs 0.11; 95% CI, 0.09-0.12). The bipolar disorder and schizophrenia FGRSs were associated with conversion from major depression to bipolar disorder (eg, hazard ratio, 1.70 [95% CI, 1.63-1.78] for high vs low bipolar FGRS) and ONAP to schizophrenia (eg, hazard ratio, 1.38 [95% CI, 1.27-1.51] for high vs low schizophrenia FGRS).

Conclusions and Relevance

In this Swedish cohort study, the FGRSs for major depression, bipolar disorder, and schizophrenia for the Swedish population clearly separated major affective and psychotic disorders from each other in a larger and more representative patient sample than previously possible. These findings provide possible validation, from a genetic perspective, for these major diagnostic categories. These results replicated and extended prior observations on more limited samples of the association of FGRS with age at onset, recurrence, psychotic subtypes, and diagnostic conversions.

Family History of Psychiatric Disorders and Clinical Factors Associated With a Schizophrenia Diagnosis

Background

Schizophrenia (SCH) and bipolar disorder (BD) have both shared and unique genetic risk factors and clinical characteristics. The aim of the present study was to identify potential risk factors significantly associated with SCH, relative to a BD reference group.

Methods

Data were obtained from medical records of patients that entered a major Mexico City hospital during 2009–2010 presenting psychotic symptoms (n = 1132; 830 cases of SCH, 302 cases of BD; 714 men and 418 women). SCH and BD diagnoses were compared with respect to a number of family and clinical characteristics. Logistic and linear regression analyses were used to respectively identify factors selectively associated with the SCH diagnosis relative to the BD diagnosis and explore the relationship between PANSS scores and parental age at time of birth to the age of SCH onset.

Results

Patients with SCH showed greater functional impairment than those with BD. Family history of mental illness, premorbid schizoid-like personality, and obstetric trauma were significantly associated with the SCH diagnosis. The association of obstetric trauma with SCH was greatest in male patients with a family history of mental illness. In women, increased paternal and decreased maternal age at time of the patient’s birth were associated with an earlier age of SCH onset.

Conclusion

Male gender, showing premorbid schizoid-like personality, familial SCH, and obstetric trauma are risk factors that distinguish SCH from BD. Additionally, our results suggest that risk for SCH relative to BD may be importantly influenced by interactions between familial risk, gender, and obstetric trauma.

Functional characterization of two variants in the 3'-untranslated region (UTR) of transcription factor 4 gene and their association with schizophrenia in sib-pairs from multiplex families

Transcription factor 4 (TCF4) gene plays an important role in nervous system development and it always associated with the risk of schizophrenia. Since miRNAs regulate targetgenes by binding to 3'UTRs of target mRNAs, the functional variants located in 3'UTR of TCF4 are highly suggested to affect the gene expressions in schizophrenia. To test the hypothesis regarding the effects of the variants located in 3'UTR of TCF4, we conducted an in silico analysis to identify the functional variants and their predicted functions. In this study, we sequenced the 3'UTR of TCF4 in 13 multiplex schizophrenia families and 14 control families. We found two functional variants carried by three unrelated patients. We determined that the C allele of rs1272363 and the TC insert of rs373174214 might suppress post- transcriptional expression. Secondly, we cloned the region that flanked these two variants into a dual luciferase reporter system and compared the luciferase activities between the pmirGLO-TCF4 (control), pmirGLO-TCF4-rs373174214 and pmirGLO-TCF4-rs1273263. Both pmirGLO-TCF4-rs373174214 and pmirGLO-TCF4-rs1273263 caused lower reporter gene activities, as compared to the control. However, only the C allele of rs1272363 reduced the luciferase activity significantly (p = 0.0231). Our results suggested that rs1273263 is a potential regulator of TCF4 expression, and might be associated with schizophrenia.

Associations between risk factors for schizophrenia and concordance in four monozygotic twin samples

Concordance for schizophrenia is high in monozygotic twins but the extent to which concordance varies according to the presence of other schizophrenia risk factors is not well established. We aimed to investigate this in systematically ascertained twin samples. DSM-III-R/DSM-IV diagnoses were made from original data or published case histories from four systematically ascertained monozygotic twin samples. Probandwise concordance for schizophrenia was calculated according to the presence of psychotic disorder in first-degree relatives, birth order, gender, and age-at-onset. Logistic regression analysis was also performed to adjust for potential confounders. Psychotic disorder in parents and earlier age-at-onset were significantly associated with higher probandwise concordance for schizophrenia, including after adjustment for potential confounders. For example, when no parents had a psychotic disorder concordance was 34/88 (38.6%) versus 10/16 (62.5%) when one parent was affected; and for age-at-onset <23 years concordance was 25/46 (54.3%), declining to 13/44 (29.5%) for age-at-onset >30 years. These results are consistent with psychotic disorder in parents and age-at-onset being markers of the level of familial liability to schizophrenia and these factors may be useful in genetic counseling of monozygotic twins and in identifying and managing those at particularly high risk, if these findings are further replicated.

Family history of psychiatric disorders and age at first contact in schizophrenia: An epidemiological study

Background

The risk for schizophrenia has been associated with a family history of this and other psychiatric disorders. The relationship between age at first contact and family history of psychiatric illness is not certain.

Aims

To estimate the risks for schizophrenia associated with a range of psychiatric diagnoses in family members and to investigate the relationship between these risks and age at first contact for schizophrenia.

Method

A nested case–control study design was employed. Psychiatric admission data and socio-economic data were available for 7704 cases admitted between 1981 and 1998 in Denmark, 192 590 gender- and age-matched controls, and for the parents and siblings of all subjects.

Results

Controlling for socio-economic factors, risk for schizophrenia was associated with a family history of all psychiatric disorders except substance misuse and independently with a family history of suicide. The risk for schizophrenia associated with a family history of psychiatric disorders decreased as age at first contact increased.

Conclusions

Risk for schizophrenia is associated with a range of psychiatric disorders in family members and these risks are not constant across the risk period.

Detecting Familial Aggregation

In addition to characterizing the distribution of genetic features of populations (mutation and allele frequencies; measures of Hardy-Weinberg equilibrium), genetic epidemiology and statistical genetics aim to explore and define the role of genomic variation in risk of disease or variation in traits of interest. To facilitate this kind of exploration, genetic epidemiology and statistical genetics address a series of questions: 1. Does the disease tend to cluster in families more than expected by chance alone? 2. Does the disease appear to follow a particular genetic model of transmission in families? 3. Does variation at a particular genomic position tend to cosegregate with disease in families? 4. Do specific genetic variants tend to be carried more frequently by those with disease than by those without these variants in a given population (or across families)? The first question can be examined using studies of familial aggregation or correlation. An ancillary question: "how much of the susceptibility to disease (or variation in disease-related traits) might be accounted for by genetic factors?" is typically answered by estimating heritability, the proportion of variance in a trait or in risk to a disease attributable to genetics. The second question can be formally tested using pedigrees for which disease affection status or trait values are available through a modeling approach known as segregation analysis. The third question can be answered with data on genomic markers in pedigrees with affected members informative for linkage, where meiotic cross-over events are estimated or assessed. The fourth question is answerable using genotype data on genomic markers on unrelated affected and unaffected individuals and/or families with affected members and unaffected members. All of these questions can also be explored for quantitative (or continuously distributed) traits by examining variation in trait values between family members or between unrelated individuals. While each of these questions and the analytical approaches for answering them is explored extensively in subsequent chapters (heritability in Chapters 8 and 9 ; segregation in Chapter 12 ; linkage in Chapters 13 - 17 ; and association in Chapters 18 - 20 ), this chapter focuses on statistical methods to address questions of familial aggregation of qualitative phenotypes (e.g., disease status) or quantitative phenotypes.While studies exploring genotype-phenotype correlations are arguably the most important and common type of statistical genetic study performed, these studies are performed under the assumption that genetic contributors at least partially explain risk of a disease or a trait of interest. This may not always be the case, especially with diseases or traits known to be strongly influenced by environmental factors. For this reason, before any of the last three questions described above can be answered, it is important to ask first whether the disease clusters among family members more than unrelated persons, as this constitutes evidence of a possible heritable contribution to disease, justifying the pursuit of studies answering the other questions. In this chapter, the underlying principles of familial aggregation studies are addressed to provide an understanding and set of analytical tools to help answer the question if diseases or traits of interest are likely to be heritable and therefore justify subsequent statistical genetic studies to identify specific genetic causes.

Neuropsychiatric Features in Primary Mitochondrial Disease

Mitochondrial diseases are a clinically heterogeneous group of disorders that ultimately result from dysfunction of the mitochondrial respiratory chain. There is some evidence to suggest that mitochondrial dysfunction plays a role in neuropsychiatric illness; however, the data are inconclusive. This article summarizes the available literature published in the area of neuropsychiatric manifestations in both children and adults with primary mitochondrial disease, with a focus on autism spectrum disorder in children and mood disorders and schizophrenia in adults.

Sex differences in the genetic risk for schizophrenia: history of the evidence for sex-specific and sex-dependent effects

Although there is a long history to examinations of sex differences in the familial (and specifically, genetic) transmission of schizophrenia, there have been few investigators who have systematically and rigorously studied this issue. This is true even in light of population and clinical studies identifying significant sex differences in incidence, expression, neuroanatomic and functional brain abnormalities, and course of schizophrenia. This review highlights the history of work in this arena from studies of family transmission patterns, linkage and twin studies to the current molecular genetic strategies of large genome-wide association studies. Taken as a whole, the evidence supports the presence of genetic risks of which some are sex-specific (i.e., presence in one sex and not the other) or sex-dependent (i.e., quantitative differences in risk between the sexes). Thus, a concerted effort to systematically investigate these questions is warranted and, as we argue here, necessary in order to fully understand the etiology of schizophrenia.

Gender differences in schizophrenia and first-episode psychosis: a comprehensive literature review

Recent studies have begun to look at gender differences in schizophrenia and first-episode psychosis in an attempt to explain the heterogeneity of the illness. However, a number of uncertainties remain. This paper tries to summarize the most important findings in gender differences in schizophrenia and first-psychosis episodes. Several studies indicate that the incidence of schizophrenia is higher in men. Most of the studies found the age of onset to be earlier in men than in women. Findings on symptoms are less conclusive, with some authors suggesting that men suffer more negative symptoms while women have more affective symptoms. Premorbid functioning and social functioning seem to be better in females than males. However, cognitive functioning remains an issue, with lack of consensus on differences in neuropsychological profile between women and men. Substance abuse is more common in men than women with schizophrenia and first-episode psychosis. In terms of the disease course, women have better remission and lower relapse rates. Lastly, there is no evidence of specific gender differences in familial risk and obstetric complications. Overall, gender differences have been found in a number of variables, and further study in this area could help provide useful information with a view to improving our care of these patients.

Detecting familial aggregation

Beyond calculating parameter estimates to characterize the distribution of genetic features of populations (frequencies of mutations in various regions of the genome, allele frequencies, measures of Hardy-Weinberg disequilibrium), genetic epidemiology aims to identify correlations between genetic variants and phenotypic traits, with considerable emphasis placed on finding genetic variants that increase susceptibility to disease and disease-related traits. However, determining correlation alone does not suffice: genetic variants common in an isolated ethnic group with a high burden of a given disease may show relatively high correlation with disease but, as markers of ethnicity, these may not necessarily have any functional role in disease. To establish a causal relationship between genetic variants and disease (or disease-related traits), proper statistical analyses of human data must incorporate epidemiologic approaches to examining sets of families or unrelated individuals with information available on individuals' disease status or related traits.Through different analytical approaches, statistical analysis of human data can answer several important questions about the relationship between genes and disease: 1. Does the disease tend to cluster in families more than expected by chance alone? 2. Does the disease appear to follow a particular genetic model of transmission in families? 3. Do variants at a particular genetic marker tend to cosegregate with disease in families? 4. Do specific genetic markers tend to be carried more frequently by those with disease than by those without, in a given population (or across families)? The first question can be examined using studies of familial aggregation or correlation. An ancillary question: "how much of the susceptibility to disease (or variation in disease-related traits) might be accounted for by genetic factors?" is typically answered by estimating heritability, the proportion of disease susceptibility or trait variation attributable to genetics. The second question can be formally tested using pedigrees for which disease affection status or trait values are available through a modeling approach known as segregation analysis. The third question can be answered with data on pedigrees with affected members and genotype information at markers of interest, using linkage analysis. The fourth question is answerable using genotype information at markers on unrelated affected and unaffected individuals and/or families with affected and unaffected members. All of these questions can also be explored for quantitative (or continuously distributed) traits by examining variation in trait values between family members or between unrelated individuals. While each of these questions and the analytical approaches for answering them is explored extensively in subsequent chapters (heritability in Chapters 9 and 10, segregation in Chapter 12, linkage in Chapters 13-17, and association in Chapters 18-21 and 23), this chapter focuses on statistical methods to answer questions of familial aggregation.

Models of population-based analyses for data collected from large extended families

Large studies of extended families usually collect valuable phenotypic data that may have scientific value for purposes other than testing genetic hypotheses if the families were not selected in a biased manner. These purposes include assessing population-based associations of diseases with risk factors/covariates and estimating population characteristics such as disease prevalence and incidence. Relatedness among participants however, violates the traditional assumption of independent observations in these classic analyses. The commonly used adjustment method for relatedness in population-based analyses is to use marginal models, in which clusters (families) are assumed to be independent (unrelated) with a simple and identical covariance (family) structure such as those called independent, exchangeable and unstructured covariance structures. However, using these simple covariance structures may not be optimally appropriate for outcomes collected from large extended families, and may under- or over-estimate the variances of estimators and thus lead to uncertainty in inferences. Moreover, the assumption that families are unrelated with an identical family structure in a marginal model may not be satisfied for family studies with large extended families. The aim of this paper is to propose models incorporating marginal models approaches with a covariance structure for assessing population-based associations of diseases with their risk factors/covariates and estimating population characteristics for epidemiological studies while adjusting for the complicated relatedness among outcomes (continuous/categorical, normally/non-normally distributed) collected from large extended families. We also discuss theoretical issues of the proposed models and show that the proposed models and covariance structure are appropriate for and capable of achieving the aim.

Genetics of obsessive-compulsive disorder

Obsessive-compulsive disorder is a common debilitating condition affecting individuals from childhood through adult life. There is good evidence of genetic contribution to its etiology, but environmental risk factors also are likely to be involved. The condition probably has a complex pattern of inheritance. Molecular studies have identified several potentially relevant genes, but much additional research is needed to establish definitive causes of the condition.

Twin study of illness history variables in psychosis

BACKGROUND

Genetic factors are important in the aetiology of psychotic disorders, but it is unclear how far they influence aspects of illness history within psychoses.

AIMS

To investigate the extent, and type of familial aggregation for a range of illness history variables in psychosis.

METHODS

Two-hundred-and-twenty-four proband-wise twin pairs (106 monozygotic, 118 same-sex dizygotic), where probands had psychosis, were ascertained from the Maudsley Twin Register in London. We investigated the following illness history variables, rated using the OPCRIT checklist: age at onset; chronicity of course; mode of onset; psychotic/affective predominance; pre-morbid social adjustment; and the presence of a psychosocial precipitant. We used Mx statistical modelling software to analyse correlations of variables within pairs of monozygotic twins concordant for psychosis; and relationships between variables in probands and risk of psychosis in monozygotic and dizygotic co-twins.

RESULTS

There was a high monozygotic within-pair correlation for age at onset (intra-class correlation=0.9); moderate correlations for chronicity of course (polychoric correlation=0.4) and psychotic/affective predominance (polychoric correlation=0.5); and lower non-significant correlations for other variables. No variables consistently predicted risk of psychosis in co-twins.

CONCLUSIONS

Illness history variables in psychosis show a broad range of familial aggregation. It is likely that familial influences are predominantly modifying effects, independent of susceptibility factors for psychosis.

Heritability of schizophrenia and major affective disorder as a function of age, in the presence of strong cohort effects

It remains unclear whether age at onset for major psychiatric disorders is a useful marker of etiologic and genetic heterogeneity. The authors examined how heritability of schizophrenia and major affective disorders varied with age at onset. The sample was drawn from a large archival data set collected by Lionel Penrose, comprising 3,109 families with two or more members first hospitalized in Ontario between 1874 and 1944. The authors studied 1,295 sibships with schizophrenia (n = 487), major affective disorder (n = 378), both (n = 234) or neither (n = 196) of these disorders. Proportional hazards models were used to estimate how the hazard of hospitalization for each disorder (schizophrenia or major affective disorder) varied with proband age at onset, adjusted for changes in age at onset distribution between 1874 and 1944. A sibling's risk of hospitalization for the same illness significantly increased for each 10-year decrease in age at onset of the proband both for schizophrenia (hazard ratio = 1.21, 95 % confidence interval: 1.06, 1.39), and for affective disorder (hazard ratio = 1.29,95 % CI: 1.14, 1.45). Gender of proband was unrelated to sibling risk of the same illness, and tests of interaction effects between proband age at onset and gender on sibling risk were nonsignificant.

Multipoint linkage mapping using sibpairs: non-parametric estimation of trait effects with quantitative covariates

Multipoint linkage analysis using sibpair designs remains a common approach to help investigators to narrow chromosomal regions for traits (either qualitative or quantitative) of interest. Despite its popularity, the success of this approach depends heavily on how issues such as genetic heterogeneity, gene-gene, and gene-environment interactions are properly handled. If addressed properly, the likelihood of detecting genetic linkage and of efficiently estimating the location of the trait locus would be enhanced, sometimes drastically. Previously, we have proposed an approach to deal with these issues by modeling the genetic effect of the target trait locus as a function of covariates pertained to the sibpairs. Here the genetic effect is simply the probability that a sibpair shares the same allele at the trait locus from their parents. Such modeling helps to divide the sibpairs into more homogeneous subgroups, which in turn helps to enhance the chance to detect linkage. One limitation of this approach is the need to categorize the covariates so that a small and fixed number of genetic effect parameters are introduced. In this report, we take advantage of the fact that nowadays multiple markers are readily available for genotyping simultaneously. This suggests that one could estimate the dependence of the generic effect on the covariates nonparametrically. We present an iterative procedure to estimate (1) the genetic effect nonparametrically and (2) the location of the trait locus through estimating functions developed by Liang et al. ([2001a] Hum Hered 51:67-76). We apply this new method to the linkage study of schizophrenia to illustrate how the onset ages of each sibpair may help to address the issue of genetic heterogeneity. This analysis sheds new light on the dependence of the trait effect on onset ages from affected sibpairs, an observation not revealed previously. In addition, we have carried out some simulation work, which suggests that this method provides accurate inference for estimating the location of quantitative trait loci.

Marginal analysis for clustered failure time data

Clustered failure time data are commonly encountered in biomedical research where the study subjects from the same cluster (e.g., family) share the common genetic and/or environmental factors such that the failure times within the same cluster are correlated. Two approaches that are commonly used to account for the intra-cluster association are frailty models and marginal models. In this paper, we study the marginal proportional hazards model, where the structure of dependence between individuals within a cluster is unspecified. An estimation procedure is developed based on a pseudo-likelihood approach, and a risk set sampling method is proposed for the formulation of the pseudo-likelihood. The asymptotic properties of the proposed estimators are studied, and the related issues regarding the statistical efficiencies are discussed. The performances of the proposed estimator are demonstrated by the simulation studies. A data example from a child vitamin A supplementation trial in Nepal (Nepal Nutrition Intervention Project-Sarlahi, or NNIPS) is used to illustrate this methodology.

A review of the evidence from family, twin and adoption studies for a genetic contribution to adult psychiatric disorders

Family, twin and adoption studies have provided major evidence for the role of genetics in numerous psychiatric disorders including obsessive-compulsive disorder, panic disorder, major depressive disorder, bipolar disorder, schizophrenia and Alzheimer's disease. As the search for patterns of inheritance and candidate genes of these complex disorders continues, we review relevant findings from quantitative genetic studies and outline the main challenges for the field of psychiatric genetics to focus on in order to more definitively establish the underpinnings of genetic and environmental influences of adult psychopathology.

A genomewide linkage study of age at onset in schizophrenia

There is strong evidence for a genetic contribution to age at onset of schizophrenia, which probably involves both susceptibility loci for schizophrenia and modifying loci acting independent of disease risk. We sought evidence of linkage to loci that influence age at onset of schizophrenia in a sample of 94 affected sibling pairs with DSM-IV schizophrenia or schizoaffective disorder, and age at first psychiatric contact of 45 years or less. Individuals were genotyped for 229 microsatellite markers spaced at approximately 20 cM intervals throughout the genome. Loci contributing to age at onset were sought by a quantitative maximum-likelihood multipoint linkage analysis using MAPMAKER/SIBS. A nonparametric multipoint analysis was also performed. The genomewide significance of linkage results was assessed by simulation studies. There were six maximum-likelihood LOD score peaks of 1.5 or greater, the highest being on chromosome 17q (LOD = 2.54; genomewide P = 0.27). This fulfils Lander and Kruglyak's [1995: Nat Genet 11:241-247] criteria for suggestive linkage in that it would be expected to occur once or less (0.3 times) per genome scan. However, this finding should be treated with caution because the LOD score appeared to be almost solely accounted for by the pattern of ibd sharing at one marker (D17S787), with virtually no evidence of linkage over flanking markers. None of the linkage results achieved genomewide statistical significance, but the LOD score peak on chromosome 13q (LOD = 1.68) coincided with the region showing maximum evidence for linkage in the study by Blouin et al. [1998: Nat Genet 20:70-73] of categorical schizophrenia.

Long-term survivor mixture model with random effects: application to a multi-centre clinical trial of carcinoma

A mixture model incorporating long-term survivors has been adopted in the field of biostatistics where some individuals may never experience the failure event under study. The surviving fractions may be considered as cured. In most applications, the survival times are assumed to be independent. However, when the survival data are obtained from a multi-centre clinical trial, it is conceived that the environmental conditions and facilities shared within clinic affects the proportion cured as well as the failure risk for the uncured individuals. It necessitates a long-term survivor mixture model with random effects. In this paper, the long-term survivor mixture model is extended for the analysis of multivariate failure time data using the generalized linear mixed model (GLMM) approach. The proposed model is applied to analyse a numerical data set from a multi-centre clinical trial of carcinoma as an illustration. Some simulation experiments are performed to assess the applicability of the model based on the average biases of the estimates formed.

Statistical designs for familial aggregation

In the past two decades, it has become increasingly clear that genetic factors contribute to the aetiology of many common diseases including cancers, coronary disease, allergy and psychiatric disorders. While one goal of genetic epidemiological studies is to locate susceptibility genes for these complex diseases, it is important that strong evidence of familial aggregation be established at an early stage of research. In this paper, we discuss several study designs useful to address some issues such as (1) detecting familial aggregation, (2) testing for gene-environment interaction, (3) identifying homogeneous subgroups and (4) measuring magnitude and patterns of familial correlations. These designs include the conventional case-control design and the family case-control design. For each of these two study designs, we discuss analytical strategies as well as their strengths and weaknesses. Throughout, several examples from real studies are used for illustrative purposes.

Childhood-onset schizophrenia: rare but worth studying

Childhood-onset schizophrenia (defined by an onset of psychosis by age 12) is a rare and severe form of the disorder that is clinically and neurobiologically continuous with the adult-onset disorder. There is growing evidence for more salient risk or etiologic factors, particularly familial, in this possibly more homogeneous patient population. For the 49 patients with very early onset schizophrenia studied to date at the National Institute of Mental Health, there were more severe premorbid neuro-developmental abnormalities, a higher rate of cytogenetic anomalies, and a seemingly higher rate of familial schizophrenia and spectrum disorders than later onset cases. There was no evidence for increased obstetric complications or environmental stress. These data, while preliminary, suggest that a very early age of onset of schizophrenia may be secondary to greater familial vulnerability. Consequently, genetic studies of these patients may be particularly informative and may provide important etiologic information.

CAG repeat length in the hKCa3 gene and symptom dimensions in schizophrenia

BACKGROUND

Long CAG repeats in the hKCa3 potassium channel gene have been associated with schizophrenia. We sought evidence for associations between this polymorphism and aspects of the schizophrenia phenotype.

METHODS

Associations were investigated between CAG repeat length and gender, age of illness onset, and psychotic symptom dimensions in 203 unrelated individuals with DSM-IIIR schizophrenia.

RESULTS

No association was found between CAG repeat length and gender or age of onset. Long CAG repeats were associated with higher negative symptom dimension scores.

CONCLUSIONS

This study provides preliminary evidence that genetic liability to negative symptoms in schizophrenia may be partly mediated through the hKCa3 gene.

Schizophrenia susceptibility loci on chromosomes 13q32 and 8p21

Schizophrenia is a common disorder characterized by psychotic symptoms; diagnostic criteria have been established. Family, twin and adoption studies suggest that both genetic and environmental factors influence susceptibility (heritability is approximately 71%; ref. 2), however, little is known about the aetiology of schizophrenia. Clinical and family studies suggest aetiological heterogeneity. Previously, we reported that regions on chromosomes 22, 3 and 8 may be associated with susceptibility to schizophrenia, and collaborations provided some support for regions on chromosomes 8 and 22 (refs 9-13). We present here a genome-wide scan for schizophrenia susceptibility loci (SSL) using 452 microsatellite markers on 54 multiplex pedigrees. Non-parametric linkage (NPL) analysis provided significant evidence for an SSL on chromosome 13q32 (NPL score=4.18; P=0.00002), and suggestive evidence for another SSL on chromosome 8p21-22 (NPL=3.64; P=0.0001). Parametric linkage analysis provided additional support for these SSL. Linkage evidence at chromosome 8 is weaker than that at chromosome 13, so it is more probable that chromosome 8 may be a false positive linkage. Additional putative SSL were noted on chromosomes 14q13 (NPL=2.57; P=0.005), 7q11 (NPL=2.50, P=0.007) and 22q11 (NPL=2.42, P=0.009). Verification of suggestive SSL on chromosomes 13q and 8p was attempted in a follow-up sample of 51 multiplex pedigrees. This analysis confirmed the SSL in 13q14-q33 (NPL=2.36, P=0.007) and supported the SSL in 8p22-p21 (NPL=1.95, P=0.023).

Identification of sequence variants and analysis of the role of the catechol-O-methyl-transferase gene in schizophrenia susceptibility

BACKGROUND

Deletions of 1.5-2 MB of chromosome 22q11 have been previously associated with schizophrenia. The deleted region includes proximally the region harboring genes involved in DiGeorge and velocardiofacial syndromes. Distally, it includes the gene for catechol-O-methyl-transferase (COMT), an enzyme that catalyzes the O-methylation of catecholamine neurotransmitters, including dopamine, and which therefore is considered a candidate gene for schizophrenia.

METHODS

We address the issue of a direct involvement of the COMT gene in the development of schizophrenia by employing the first extensive mutational analysis of this gene in a sample of 157 schizophrenia patients and 129 healthy controls, using single-strand conformation polymorphism and chemical cleavage methodologies.

RESULTS

No mutations were found, but several sequence variants were identified, including the genetic polymorphism that underlies the high/low activity of the enzyme (a Val158-->Met change, which results in the creation of an NlaIII restriction site in the low-activity allele). The distribution of the NlaIII genotypes among subsets of schizophrenia patients was analyzed.

CONCLUSIONS

The results presented here argue against a major role of COMT in schizophrenia in general (although a minor effect could not be excluded) and represent a first step toward a more refined delineation of the phenotype/genotype relationship between 22q11 microdeletions and schizophrenia susceptibility.

Age at onset in schizophrenia and risk of illness in relatives. Results from the Roscommon Family Study

BACKGROUND

For many common medical and neuropsychiatric disorders, early age at onset reflects high familial liability to illness. However, for schizophrenia, most studies do not find such a relationship.

METHOD

Using Cox proportional hazard modes, we investigate this question in the epidemiologically-based Roscommon family study.

RESULTS

No relationship was found between age at onset in schizophrenic probands and the hazard rate for schizophrenia in their relatives. Similar results were obtained when the definition of illness was expanded to include schizoaffective disorder and other non-affective psychoses.

CONCLUSIONS

For schizophrenia, a 'common-sense' model for age of onset (i.e. those with highest familial liability to illness succumb first while those with lower liability survive longer before falling ill) does not seem to apply. Our results are more consistent with a model in which variation in age at onset of schizophrenia is due to random developmental effects or to environmental experiences unique to the individual.

Cox regression analysis of multivariate failure time data: the marginal approach

Multivariate failure time data are commonly encountered in scientific investigations because each study subject may experience multiple events or because there exists clustering of subjects such that failure times within the same cluster are correlated. In this paper, I present a general methodology for analysing such data, which is analogous to that of Liang and Zeger for longitudinal data analysis. This approach formulates the marginal distributions of multivariate failure times with the familiar Cox proportional hazards models while leaving the nature of dependence among related failure times completely unspecified. The baseline hazard functions for the marginal models may be identical or different. Simple estimating equations for the regression parameters are developed which yield consistent and asymptotically normal estimators, and robust variance-covariance estimators are constructed to account for the intra-class correlation. Simulation results demonstrate that the large-sample approximations are adequate for practical use and that ignoring the intra-class correlation could yield rather misleading variance estimators. The proposed methodology has been fully implemented in a simple computer program which also incorporates several alternative approaches. Detailed illustrations with data from four clinical or epidemiologic studies are provided.

Age at onset, sex, and familial psychiatric morbidity in schizophrenia. Camberwell Collaborative Psychosis Study

BACKGROUND

Although a genetic component in schizophrenia is well established, it is likely that the contribution of genetic factors is not constant for all cases. Several recent studies have found that the relatives of female or early onset schizophrenic patients have an increased risk of schizophrenia, compared to relatives of male or late onset cases. These hypotheses are tested in the current study.

METHOD

A family study design was employed; the probands were 195 patients with functional psychosis admitted to three south London hospitals, diagnosed using Research Diagnostic Criteria (RDC), and assessed using the Present State Examination (PSE). Information on their relatives was obtained by personal interview of the mother of the proband, and from medical records. Psychiatric diagnoses were made using Family History-Research Diagnostic Criteria (FH-RDC), blind to proband information.

RESULTS

There was a tendency for homotypia in the form of psychosis within families. The lifetime risk of schizophrenia in the first degree relatives of schizophrenic probands, and the risk of bipolar disorder in the first degree relatives of bipolar probands, were 5-10 times higher than reported population risks. Relatives of female and early onset (< 22 years) schizophrenic probands had higher risk of schizophrenia than relatives of male and late onset schizophrenic probands. However, this effect was compensated in part by an excess of non-schizophrenic psychoses in the relatives of male probands.

CONCLUSIONS

These results suggest a high familial, possibly genetic, loading in female and early onset schizophrenia, but do not resolve the question of heterogeneity within schizophrenia.

Sequential strategy to identify a susceptibility gene for schizophrenia: report of potential linkage on chromosome 22q12-q13.1: Part 1

To identify genes responsible for the susceptibility for schizophrenia, and to test the hypothesis that schizophrenia is etiologically heterogeneous, we have studied 39 multiplex families from a systematic sample of schizophrenic patients. Using a complex autosomal dominant model, which considers only those with a diagnosis of schizophrenia or schizoaffective disorder as affected, a random search of the genome for detection of linkage was undertaken. Pairwise linkage analyses suggest a potential linkage (LRH = 34.7 or maximum lod score = 1.54) for one region (22q12-q13.1). Reanalyses, varying parameters in the dominant model, maximized the LRH at 660.7 (maximum lod score 2.82). This finding is of sufficient interest to warrant further investigation through collaborative studies.

Estimating effects of probands' characteristics on familial risk: I. Adjustment for censoring and correlated ages at onset

Family studies with age at onset of the disease as the endpoint face two important problems: censoring and correlation of age at onset among relatives. We present a multivariate survival model for ages at onset of relatives which incorporates the problems cited above. The interpretations of regression coefficients and association parameter in the context of family studies are emphasized. The present paper describes a statistical method for estimating these parameters. In a companion paper [Pulver and Liang, Genet Epidemiol 8:339-350, 1991] this model is applied to a genetic epidemiologic study of schizophrenia.