Chapter 3. Psychiatric Genetics

A scientific revolution has occurred in the field of genetics with the advent of molecular biological techniques. Using these techniques, genes influencing risk for many neuropsychiatric diseases have been identified: initially Mendelian single gene conditions such as Huntington disease were resolved; in the last few years complex genetic conditions such as alcohol dependence have yielded specific genes. Some of this work has been facilitated by the study of endophenotypes or biologic vulnerability markers.

Clinical Epidemiology: Twin, Family, and Adoption Studies

Three types of population genetic studies—twin, family, and adoption studies—are conducted to ascertain whether a particular human phenomenon is substantially genetically influenced.

Twin studies are based on the fact that monozygotic (MZ) or identical twins represent a natural experiment in which two individuals have exactly the same DNA sequence for each of their genes. This is in contrast to dizygotic (DZ) or fraternal twins who share 50% of their DNA sequences and are no more genetically similar than any pair of siblings. A phenomenon that is influenced by genetic factors should be more “concordant” (similar) in MZ twins compared to DZ twins.

Family studies can answer three critical questions concerning the inheritance of a disorder:

• Are relatives of an affected subject at increased risk for the disorder compared to relatives of control subjects?
• What other disorders may share a common genetic vulnerability with the phenomenon in question?
• Can a specific mode of inheritance be discerned?

A family study typically begins with a proband or initially ascertained patient, whose relatives are then studied.

Adoption studies: In adoption studies the risk for the disorder may be evaluated in four groups of relatives: the adoptive and biological relatives of affected adoptees and the adoptive and biological relatives of control adoptees. If the disorder is heritable, one should find an increased risk among the biological relatives of affected subjects, compared to the other three groups of relatives. One can also compare risk for illness in adopted-away children of ill parents versus adopted-away children of well parents.

Segregation analysis may be used to determine whether the pattern of illness in families is consistent with a specific mode of transmission. This is most useful for conditions in which a single gene accounts for a substantial portion of the variance. Most major psychiatric disorders do not appear to fall in this category.

Some of the complexities of major psychiatric disorders are as follows:

• Variable penetrance (some individuals with a genetic predisposition will not manifest the disease)
• Phenocopies (individuals without a genetic predisposition who manifest the symptoms of the disease)
• Genetic heterogeneity (more than one type of genetic cause can produce the same syndrome)
• The diagnostic boundaries of a syndrome may be uncertain.
• Pleiotropy (one gene may be expressed in different ways in different persons).

Linkage Analysis

At any genetic locus, each individual carries two copies (alleles) of the DNA sequence that defines that locus. One of these alleles is inherited from the mother and the other from the father. These alleles will be transmitted with equal probability (i.e., 1/2), one of the two alleles to each offspring. If two genetic loci are “close” to each other on a chromosome, their alleles tend to be inherited together (not independently) and they are known as “linked” loci. During meiosis, crossing over (also known as recombination) can occur between homologous chromosomes, thus accounting for the observation that alleles at linked loci are not always inherited together.

The rate at which crossing over occurs between two linked loci is directly proportional to the distance on the chromosome between them. In fact, the genetic distance between two linked loci is defined in terms of the percentage of recombination between the two loci (this value is known as theta). Loci that are “far” apart on a chromosome will have a 1/2 chance of being inherited together and they are not linked. Thus, the maximum value for theta is 0.5, while the minimum value is 0. Linkage analysis is a method for estimating theta for two or more loci.

The probability that two loci are linked is the probability that theta <0.5. The probability that the two loci are not linked is the probability that theta = 0.5. Thus, a LOD (logarithm of the odds ratio) score is defined as

Although it is possible to perform such calculations by hand (see Ott, 1985), LOD scores are usually calculated using computer programs, such as GENEHUNTER or Merlin. Since a LOD score is a log value, scores from different families can be summed. For complex conditions, collections of affected sib pairs may be studied rather than large families. A LOD score of 1.0 indicates that linkage is 10 times more likely than nonlinkage. For simple genetic conditions, a LOD score of three or greater is evidence for linkage, while a score of −2 or less is sufficient to exclude linkage for the sample studied. For disorders with more complex forms of inheritance (including most psychiatric disorders), a higher positive LOD score is required (3.6 for definite linkage and 2.2 for suggestive linkage).

Association Studies

In association studies one compares allele frequencies for a given locus in two populations, one of which is composed of unrelated individuals who have a disease, while the “control” population is usually composed of ethnically similar unrelated persons who do not have the disease. If a particular allele commonly predisposes individuals to the disease in question, then that allele should occur more frequently in the disease population, compared to the control population.

There are potential pitfalls to the case-control association approach. The locus chosen for study must predispose to illness. Thus, loci chosen for association studies are often known as candidate genes. If the locus does not predispose to illness, then the results of an association study should be negative. However, false positive results can occur if the two populations are not carefully matched for ethnic background. One alternative control group is the parents of affected individuals (the alleles not transmitted to the affected child compose the “control group”—this is known as the Transmission Disequilibrium Test or TDT).

High-Risk Studies

Biochemical studies of individuals with psychiatric diseases are always confounded by the issue of disease effects: are biochemical differences between affected individuals and controls related to the cause of the disorder, or are they related to the effects of the disorder (or its treatment)? When investigating possible biochemical differences for a genetic disease, this difficult issue can be addressed by studying a group of individuals (usually adolescents or young adults) who are at high risk to develop the disorder under study (usually because they have parents and/or other relatives with the disorder). The high-risk group may then be followed over time to assess whether the biochemical abnormalities observed are truly predictive of the disease.

Molecular Genetic Methods

Simple sequence repeat (SSR) markers, also known as microsatellites represent a group of polymorphisms based upon a variable number of a repeated sequence. However, in the case of SSR markers, the repeated sequence consists of 2–5 nucleotides. The repeated sequence is often —(CA)n—, —(AG)n—, or —(AAAT)n—, although other SSR sequences have been described. The region containing the SSR is amplified by using a thermostable DNA polymerase in a polymerase chain reaction (PCR). Microsatellites are commonly used to test linkage.

A different class of DNA markers, single nucleotide polymorphisms (SNPs), are usually used to detect association. An SNP is variation at one base in a DNA sequence (e.g., compare GATACA with GATGCA, in which the fourth nucleotide can be either “A” or “G”). It is estimated that there are ≈3 million common SNPs in the human genome, evenly distributed across the 3 billion bases of the human genome.

SNP detection is now highly automated, permitting the determination of more than one million SNP genotypes on each DNA sample, via DNA “chips” (microscope slide shaped devices) to which are attached up to one million oligonucleotides (short fragments of DNA sequence ≈20–30 bases in length). These oligonucleotides each permit the determination of the presence of an allele at a single SNP. This advance now permits genome-wide association studies for complex disorders, in which the frequency of alleles at each of the one million SNPs are compared in large groups of cases and controls.

Affective Disorders

Genetic Epidemiologic Studies

Family Studies

Family studies in affective disorder have continually demonstrated an aggregation of illness in relatives (Table 3–1). In a study at NIMH, 25% of the relatives of bipolar (BP) probands were found to have bipolar or unipolar (UP) illness, compared to 20% of relatives of UP probands and 7% of relatives of controls. In the same study, 40% of the relatives of schizoaffective probands demonstrated affective illness at some point in their lives. These data demonstrate increased risk in relatives of patients. They also show that the various forms of affective illness appear to be related in a hierarchical way: relatives of schizoaffective probands may have schizoaffective illness themselves, but are more likely to have BP or UP illness. Relatives of BP probands have either BP or (more likely) UP illness.

Age of onset may be useful in dividing affective illness into more genetically homogeneous subgroups. Early-onset probands have an increased morbid risk of illness in relatives in some data sets. Other subphenotypes, such as cycling frequency and comorbid anxiety disorders or substance use disorders, have also been studied.

A birth cohort effect was observed in several family studies, with an: an increasing incidence of affective illness among persons born more recently. The cohort effect was observed among relatives at risk to a greater degree than in the general population. The reasons for this increase in incidence are not yet clear.

Twin Studies

Twin studies show consistent evidence for heritability. On the average, MZ twin pairs show concordance 65% of the time and DZ twin pairs 14% of the time.

Adoption Studies

Several adoption studies have been performed in the area of affective illness. The results have been generally consistent with genetic hypotheses.

The Affective Spectrum

Following are the types of affective disorders and other disorders that are genetically related:

BPI

Classic “manic–depressive illness” with severe mania, generally including episodes of major depression as well.

BPII

This disorder is genetically related to BPI and UP. There is some evidence in recent family studies for an excess of BPII illness in relatives of BPII probands. It has been demonstrated that BPII tends to be a stable lifetime diagnosis, that is, patients do not frequently convert to BPI.

Rapid Cycling

Rapid-cycling BP illness has been the subject of great theoretical and clinical interest. A link with thyroid pathology has been proposed. Rapid-cycling appears to arise from factors which are separable from the genetic vulnerability to BP illness and which do not lead to aggregation within families. However “rapid switching” of mood, which is related, appears to be familial.

UP Mania

This entity includes BPI patients with no history of major depression. This group is not distinguishable from other BPI patients on the basis of family pattern of illness.

Cyclothymia

This condition of repetitive high and low mood swings, generally not requiring clinical attention, is probably genetically related to BP disorder.

Schizoaffective Disorder

A group of patients with intermittent psychosis during euthymia have an increase in affective illness and schizophrenia in relatives. This group may have the highest genetic load (total risk for affective or schizophrenic illness in relatives) of any diagnostic category. They may carry genes related to both BP illness and schizophrenia. Patients with chronic psychosis and superimposed episodes of mood disorder confer risk for both chronic psychosis and mood disorder to relatives but have less overall genetic load.

Schizophrenia

An overlap in linkage areas and vulnerability genes has been identified in recent years; some of this overlap may relate to genes involved in glutamate neurotransmission.

Eating Disorders

Family studies of anorexia and bulimia have generally found excess affective illness in relatives. Relatives of anorexics may have similar risk for affective disorders to that of relatives of BP probands.

Attention-Deficit Disorder

Children with this disorder appear to have increased depression in their relatives. The opposite has not been demonstrated (BP/UP probands have not been reported to have increased risk of attention deficit disorder in their offspring).

Alcohol Dependence

There may be overlapping vulnerability traits. Alcoholism appears to be comorbid with UP and BP disorders (each appears to confer an increased risk for the other within individuals). There is some evidence that alcoholism with affective disorder may itself aggregate within families.

Linkage Studies

Linkage has been demonstrated on 4p, 6q, 8q, 13q, 18p, 18q, and 22q. Other areas are “close” to significant (e.g., 12q, 21q, and Xq).

Endophenotypes

A number of endophenotypic markers have been suggested, including:

• REM sleep induction by cholinergic drugs,
• white matter hyperintensities on MRI,
• amygdala activation on fMRI,
• hippocampal size,
• response to tryptophan depletion, and
• response to sleep deprivation.

Gene Expression Studies

Studies have begun on genome-wide gene expression in animal models of affective disorder, in brain samples from autopsy studies of patients with mood disorders, and in peripheral tissues such as blood. These studies should be helpful in identifying candidate genes for mood disorders.

High-Risk Studies

More offspring of patients than controls have a diagnosed Axis I disorder. Offspring of BP parents may be more prone to respond to dysphoric feeling states with “disinhibitory” behavior.

Association/Candidate Gene Studies

Numerous candidate gene studies are now in the literature for BP illness. A few genes have emerged with replicated findings, or positive meta-analyses from multiple studies. We will feature these here.

G72

This gene is one of the two implicated together in association studies on chromosome 13q. The gene G30 is a DNA sequence that is reverse-transcribed within G72. The association was first identified by Hattori et al. (2003) after work by Chumakov and colleagues in schizophrenia. It has been replicated by three other independent groups. The most recent work shows association not only with BP illness, but also with a subset of subjects who have schizophrenia with clear mood episodes. The function of G72 (also sometimes referred to as DAOA) may be to, oxidize serine, a potent activator of glutamate transmission via a modulatory site on the NMDA (n-methyl-d-aspartate) receptor. Inadequate DAOA function might be hypothesized to lead to problems in modulating the glutamate signal in areas of the brain such as the prefrontal cortex. Evidence from animal studies suggests that glutamate antagonists have antidepressant effects, and that depression is associated with inadequate modulation of glutamate neurotransmission. However a recent study suggests that the major role of G72 may be in maintaining neuronal structure.

Brain-Derived Neurotrophic Factor (BDNF)

This gene is a candidate based both on position (11p14, near reported linkage peaks in several family series) and function (as a neuronal growth factor, it is implicated in several recent theories of depression and BP mood disorder). BDNF has shown significant association with BP illness in three independent reports in family-based data, but not in several case-control series. Two reports have suggested association in child/adolescent-onset BP disorder, and two additional series show association in rapid-cycling BP patients. Several studies have shown that antidepressant administration is associated with increased central BDNF levels in experimental animals, and administration of BDNF itself has been associated with antidepressant-like activity. Depression has been postulated to be associated with decreased neurogenesis in the hippocampus, which is dependent on neurotrophic factors, including BDNF. Mood stabilizing medications used in BP illness are thought to have neuroprotective effects.

Disrupted in Schizophrenia 1 (DISC1)

This gene on chromosome 1q was identified in a Scottish family with a genetic translocation and with multiple cases of psychiatric disorders, primarily schizophrenia. However DISC1 variants were associated with mood disorders in family members as well. Later studies in an independent series of BP patients in Scotland were positive for association as well. A study in Wales of schizoaffective patients showed a linkage peak in the same chromosomal location. This gene is expressed in multiple brain regions, including the hippocampus, where it is differentially expressed in neurons. It is associated with microtubules; in mice, disruption of DISC1 leads to abnormal neuronal migration in the developing cerebral cortex. DISC1 appears to interact with phosphodiesterase 4B, which may play a role in mood regulation.

5HTT, MAOA, COMT

These three genes have been shown in meta-analyses to be associated with BP disorder, even though no strong effects have been shown in any one study. The effect size for each appears to be in the range of 10–20% increase in risk. Each of these genes is associated with other behavioral phenotypes, and each has been reported to interact with environment to increase the risk of specific disorders (major depression, antisocial personality disorder, and schizophrenia respectively). Recent data in BP illness are more positive for 5HTT than for MAOA or COMT.

P2RX7 (aka P2X7, P2X7R)

This gene on 12q24 was identified in a French–Canadian case-control series following linkage studies using large pedigrees from the same population. It is a calcium-stimulated ATPase. The data are suggestive, but await replication in an independent study.

GRK3

This is the only candidate identified using animal model studies (a mouse model employing methamphetamine). The original gene expression studies were followed up by association studies in several samples as well as expression studies in human lymphoblasts. This gene participates in the down-regulation of G-protein coupled receptors.

Empirical Data for Genetic Counseling

Molecular genetic studies hold great promise in the future for families with affective disorder, particularly BP disorder. However genetic counseling currently is based on empirical risk figures.

The lifetime risk for severe (incapacitating) affective disorder is about 7%. Risk is increased to about 20% in first-degree relatives of UP patients, and 25% in first-degree relatives of BP. It appears to be 40% in relatives of schizoaffective patients. The risk to offspring of two affectively ill parents is in excess of 50%. Overall risk figures appear to be rising in recent years, but more so in relatives of patients than in the general population (keeping at about a 3:1 ratio). Average age of onset is about 20 for BP disorder and 25 for UP.

Alcoholism

Epidemiologic Genetic Studies

Twin Studies

Twin studies tend to show heritability of drinking behavior and heritability of alcoholism. A Finnish twin study included interview data on 902 male twins between 28 and 37 years of age. Heritability was 0.39 (i.e., about 39% of the variance between members of a twin pair is due to genetic factors) for frequency of drinking and 0.36 for amount consumed per session. A second Finnish study involved several thousand pairs of twins in the state twin registry. Overall heritability for total alcohol consumption was 0.37 in males and 0.25 in females. A study in which 572 twin families from the Institute of Psychiatry register were examined found that additive genetic factors accounted for 37% of the variance in alcohol consumption among drinkers, when pedigree data are considered together with twin data and the effect of shared environment on twin concordance is accounted for. The critical data from these three large twin studies are strikingly similar, at least in males.

Twin studies of alcoholism itself have generally shown heritability. Kaij studied registration of twin subjects at the Swedish County Temperance Boards. Such registration implies that a complaint was made about a person's behavior while drinking, either by the police or a third party. This would not generally include alcoholics who are socially isolated, though they might be significantly impaired. The registration information was followed up with personal interviews of probands and cotwins. In a total of 205 twin pairs, probandwise concordance was 54.2% in MZ's and 31.5% in DZ's (p <.01). Concordance rates in MZ's increased with the severity of the disturbance. A reanalysis of these data shows heritability to vary from 0.42 to 0.98, the more serious forms of alcoholism being more heritable.

Kendler conducted a population-based study of female twin pairs from the Virginia twin registry. Personal interviews were completed on 1033 of 1176 pairs. MZ concordance varied from 26% to 47% (depending on whether a narrow or broad definition of alcoholism was used) while DZ concordance ranged from 12% to 32%. The calculated heritability was 50–61%. This suggests a substantial genetic influence for alcoholism in women.

Adoption Studies

Goodwin compared 55 adopted-away male children of an alcoholic parent with 78 adoptees without an alcoholic parent. The groups were matched by age, sex, and time of adoption. The principal finding was that 18% of the proband group were alcoholic compared with 5% of the controls (p <0.02). This study also compared adopted-away sons of alcoholics with sons of alcoholics raised by the alcoholic parent. There was no difference.

Bohman used state registers in Stockholm to study 2324 adoptees born in that city between 1930 and 1949. Male adoptees whose fathers abused alcohol (excluding those who were also sociopathic) were more likely to be alcoholic themselves (39.4% vs. 13.6%, p <0.01) compared with adoptees without an alcoholic (or sociopathic) father. Cloninger, Bohman, and Sigvardsson postulated a familial distinction of alcoholics: a milieu-limited (type I) and a male-limited (type II) group. Type I alcoholics usually have onset after age 25, manifest problems with loss of control, and have a great deal of guilt and fear about alcohol use. Type II alcoholics have onset before age 25, are unable to abstain from alcohol, and have fights and arrests when drinking, but less frequently show loss of control and guilt and fear about alcohol use. Cloninger reanalyzed the Stockholm Adoption data using these specified categories. He showed that type I alcoholics were significantly greater in prevalence only among those adoptees with both genetic and environmental risk factors (i.e., alcoholism in both biologic and adoptive parents). Type I was the most common type of alcoholism; however, it was present in 4.3% of controls with no risk factors. Type II alcoholism was present in only 1.9% of the controls but in 16.9–17.9% of adoptees with genetic risk factors. The presence or absence of environmental risk factors (alcoholism in adoptive parents) did not appear to make a difference.

Bohman extended this finding to women adoptees, identifying as particularly important the incidence of alcoholism in the biologic mothers of these adoptees.

Family Studies

There is a concentration of alcoholics in the families of alcoholic probands. Cotton (summarizing 39 studies on families of 6251 alcoholics and 4083 nonalcoholics) reports an overall prevalence of 27.0% alcoholism in fathers of alcoholics and of 4.9% in mothers; 30.8% of alcoholics had at least one alcoholic parent. The same preponderance of alcoholism was not seen in the parents of comparison groups of patients with other psychiatric disorders. The studies of nonpsychiatric controls reviewed in the same study show alcoholism rates of 5.2% in fathers and 1.2% in mothers. A recent report from the Collaborative Study of the Genetics of Alcoholism (COGA) shows significant coaggregation of drug dependence, mood disorders, and anxiety disorders as well as alcohol dependence in the relatives of persons with alcohol dependence.

Disorders Genetically Related to Alcoholism

Winokur reported an increased prevalence of depression in the female relatives of alcoholics, roughly comparable to the increased prevalence of alcoholism in male relatives. Some forms of illness may result from shared vulnerability factors. Recent studies suggest that comorbid disorders (including alcoholism and affective illness) themselves run in families.

Bohman and Cloninger observed that adopted-away daughters of type II (male-limited) alcoholics manifest no increase in alcoholism but do show an increase in somatization disorder.

It is not possible to conclude at this time that a single genetic predisposing factor is manifest as either alcoholism or sociopathy (antisocial personality disorder). However, some sociopathic alcoholics may transmit both alcoholism and sociopathy as part of the same syndrome.

Earls reported an increase in the Diagnostic and Statistical Manual of Mental Disorders, third edition (DSM-III) behavior disorder in general (attention deficit disorder with hyperactivity, oppositional disorder, and conduct disorder) in the offspring of alcoholic parents. The risk is greater for offspring of two alcoholic parents than for those of one alcoholic parent.

Linkage Studies

Several linkage studies have been completed in sizeable populations. Genes predisposing to alcohol dependence appear to be located on chromosomes 1, 2, 4, 7, and 16.

Association Studies

GABRA2

Variants in GABRA2 on chromosome 4p have been shown by Edenberg and colleagues in COGA to be associated with the power of beta oscillations in the EEG (which are inversely related to inhibitory neuronal activity in the cortex) and to alcohol dependence. This association has now been replicated by four other groups. GABRA2 appears to be particularly strongly related to problems with impulse control; the risk allele is also seen in adolescents with conduct disorder and in alcohol dependent persons who are drug dependent. Other GABA receptor genes such as GABRG3 may also be associated with alcohol dependence.

ADH4

ADH (alcohol dehydrogenase) is the major metabolic enzyme for alcohol, catalyzing its breakdown into acetaldehyde, which is then further metabolized by aldehyde dehydrogenase (ALDH). Both ADH and ALDH have variants associated with the “flushing” reaction to alcohol (a feeling of warmth accompanied by reddening of the skin and sometimes nausea and tachycardia). These variants are most common in East Asian populations. They tend to protect against the development of alcohol dependence. In recent studies, single nucleotide polymorphisms in some of the ADH enzymes (genes for several isoenzymes of ADH are located on chromosome 4q) have been associated with alcohol dependence in Caucasian populations and in Native Americans. The strongest finding is in ADH4, which appears to be associated with the early onset of regular drinking.

CHRM2

The M2 muscarinic receptor gene on chromosome 7q was associated with alcohol dependence and major depression in the COGA study, an association that has been replicated. The association with depression recalls the cholinergic-adrenergic balance hypothesis of Janowsky and colleagues from the 1970s (in which a relative increase in central cholinergic activity is associated with depression and a relative increase in central adrenergic activity with mania).

hTAS2R16

This gene, located under the same linkage peak on 7q as CHRM2, codes for a bitter taste receptor. Variants are associated with alcohol dependence, which is consistent with studies showing that relative sensitivity to sweet taste is related to alcohol acceptance in rodent models. The risk gene variant is much more common in African Americans than in European Americans.

DRD2

Originally reported about a decade ago, the literature on DRD2 is still controversial. A meta-analysis of 21 studies shows an increased risk of alcoholism of 50–100% for persons carrying the A1 allele. However, recent work has questioned whether this polymorphism may actually be reflecting variation in a gene next to DRD2.

Etiologic Marker Studies

Major areas of concentration in the search for a potential biologic trait marker of alcoholism include the following: (1) enzymes of alcohol metabolism and other enzymes; (2) EEG and evoked potentials before and after alcohol; (3) psychologic/psychophysiologic differences; and (4) behavioral and neuroendocrine responses to alcohol.

Alcohol is metabolized primarily in the liver by alcohol dehydrogenase (ADH) and aldehyde dehydrogenase (ALDH). Four isozymes of ALDH are known. Three are found in the cytoplasm and one (ALDH2) in the mitochondria. It is the latter that is probably responsible for most acetaldehyde metabolism in vivo. The ALDH2 enzyme is lacking in about 50% of Japanese and apparently in other Oriental groups as well. Such people are subject to the “flushing” reaction from alcohol (similar to a disulfiram reaction). Alcohol elimination is not different in such subjects; however, the alcoholism rate is significantly diminished.

A poorly synchronized resting EEG (lower alpha) has been thought to be related to a predisposition for alcoholism. Change in alpha rhythm following alcohol is more concordant in MZ than DZ twins (as are multiple other EEG parameters). Change in alpha rhythm following alcohol was also found to differentiate young adult subjects at high risk for alcoholism from controls.

Measurements of event-related potentials have shown smaller P300 waves following visual stimuli in 7–13-year-old sons of alcoholics compared to controls. The EEG/ERP area remains one of the more promising in the field of pathophysiologic markers for alcoholism. Schuckit has studied behavioral and neuroendocrine responses to alcohol infusion in a series of high-risk populations. Offspring of alcoholics displayed less subjective intoxication, than controls. A follow-up shows that decreased subjective intoxication is correlated with later development of alcoholism in sons of alcoholics.

Summary

There appear to be hereditary factors operative in normal drinking behavior and in vulnerability to alcohol abuse. The “flushing” reaction to acetaldehyde is one of the clearest instances of a pharmacogenetic variant that influences human behavior, though even here the interaction between genotype and environment in different ethnic groups may result in very different outcomes. Alcoholism clearly runs in families. It is more often manifest in men than in women. The familial preponderance is primarily genetic and the differentiation by gender is primarily the result of sociocultural factors. There may be two distinct types of alcoholism with different patterns of inheritance, type I (or milieu-limited) and type II (or male-limited). Type II alcoholism is more severe and more likely to be strongly influenced by major gene effects. Genetic markers for the vulnerability to alcoholism have yet to be verified. Of most interest are studies suggesting EEG/ERP differences in those at risk, the studies showing decreased responsiveness to alcohol in those at risk, and the neurochemical investigation of appropriate animal models of alcoholism. The recent single gene findings for GABRA2, ADH4, and CHRM2 have provided a stimulus for additional studies of the presumed functional consequences of these genetic variants in alcohol dependent subjects.

Alzheimer Disease

Genetic etiologies are clear for some forms of Alzheimer disease. The specific genes that influence vulnerability have been identified (Table 3–3). Epidemiologic studies, however, do not show high heritability of the disorder. This is partially attributable to multiple etiologic, environmental as well as genetic factors. It is also related to the variable age of onset of the condition. Early-onset cases are more likely to be hereditary, and may be determined by single genes. Later-onset cases are more likely to be multifactorial. Important genetic factors in late-onset cases may be obscured by the fact that mortality from other causes decreases familial aggregation.

Twin Studies

A total of 81 twin pairs, where at least one member is affected with Alzheimer disease (AD), have been reported in the literature. The MZ concordance rate (approximately 45%) is not different from the DZ concordance rate (approximately 35%) in these studies.

Family Studies

Most of the relatives of later-onset AD probands will have died of other causes before passing the age of risk. However, the risk to siblings of probands (with an affected parent) whose age of onset was less than 70 years is close to 50%. The morbid risk may be 40% for first-degree relatives at age 90. Heun reported a 30% incidence of dementia in first-degree relatives of Alzheimer's probands compared to 22% in controls.

A subset of early-onset AD cases is highly familial. At least some AD cases (primarily those with later onset) are sporadic.

Linkage and Association Studies

In 1987, St. George-Hyslop et al. reported linkage of familial AD to RFLP markers on chromosome 21. The peak LOD score (4.25) suggested a causative gene near these markers. Subsequently, certain isolated, rare AD families have been found to have a point mutation in the gene for amyloid precursor protein. These studies suggest that abnormalities in this gene (which produces a proteinaceous material found to accumulate in the extracellular space in brains of persons with AD) can cause the disease by itself. Another cause of early-onset familial Alzheimer's is a gene on chromosome 14. Additional families are linked to a gene on chromosome 1. The genes on chromosome 1 and 14 code for proteins named presenilins and appear to be highly homologous.

Many late-onset families show linkage to a region of chromosome 19 coding for lipoprotein E (usually abbreviated as ApoE), which is also implicated in cardiovascular illness. One copy of the E4 allele will increase risk fourfold compared to that with E2. Two copies of E4 increase risk by a factor of eight. It is not clear at this time whether E4 is a risk factor for all ethnic groups and families or just in selected populations. The molecular mechanisms for the Alzheimer's vulnerability genes are now the subject of intense investigation. There is reason to suspect that they all affect the accumulation of amyloid.

Antisocial Personality Disorder

Epidemiologic Studies: Twin Studies

In a Danish twin study, 32.6% (28/86 pairs) of MZ twins were concordant for criminal behavior, compared to 13.8% (21/152 pairs) of DZ twins (p <0.001). In a Norwegian twin study, Dalgard and Kringlen found a higher MZ concordance (25.8%) for crime compared to DZ concordance (14.8%), but this failed to reach statistical significance (p = 0.11).

Schulsinger and Crowe conducted adoption studies of antisocial personality disorder (AP). In these early studies, there was a consistent observation that the adopted-away offspring of AP biologic parents had a higher risk for antisocial behavior than did control adoptees. For example, 6/46 adoptees of female felons met criteria for AP compared to 0/46 control adoptees. In this study, outcome was unrelated to the length of time the adoptees remained with their biologic mothers. In Schulsinger's study of 57 adoptee AP and 57 control probands, AP was found among 3.9% (12/305) of biological relatives of AP adoptees, compared to 1.4% (4/285) of biological relatives of control adoptees, a highly significant difference. If only the fathers were considered, 9.3% of probands’ biological fathers (5/54) received a diagnosis of AP compared to 1.8% of the biological fathers of control adoptees. In a larger study, using adoption and criminal registries, Mednick reported that when neither the biologic nor adoptive parents had been convicted, adoptees had a conviction rate of 13.5%. If only the adoptive parents were convicted, the adoptee conviction rate rose to only 14.7%. If only the biologic parents were convicted, the adoptee conviction rate was 20%. When both sets of parents had been convicted, the conviction rate for adoptees was 24.5%. The risk for conviction in a male adoptee increased as a function of the number of convictions in the biological and adoptive parents.

These findings were confirmed in a later study of criminality in a cohort of adoptees from Stockholm. Both genetic and “postnatal” influences were detectable in the risk for AP. When postnatal factors predisposed to criminality, 6.7% of male adoptees were criminal compared to 2.9% of male adoptees with nonpredisposing postnatal and genetic backgrounds. When the genetic background, but not the postnatal environment, was predisposing, 12.1% of male adoptees were criminal compared to 2.9% of control male adoptees. When both genetic and postnatal backgrounds were judged to predispose to criminal behavior, 40.0% of male adoptees were criminal. These results are consistent with the additive effects of genes and postnatal influences. The environmental influences implicated were multiple foster homes (for men) and extensive institutional care (for women).

Epidemiologic Studies: Family Studies

Of 223 male criminals, 80% were found to have a diagnosis of AP in a study by Guze. Sixteen percent of interviewed male first-degree relatives also had this diagnosis, while only 2% of female relatives had AP, compared to 3% and 1% in the relatives of controls. Increased rates of alcoholism and drug abuse were also found among the first-degree relatives of these criminals.

A family study of 66 female felons and 228 of their first-degree relatives revealed increased rates for AP (18%), alcoholism (29%), drug abuse (3%), and hysteria (31%), all the hysteria occurring in the female relatives. Predictably, male relatives had a threefold increase in AP (31%) compared to the female relatives (11%). The increased risk for AP among first-degree relatives of female felons (31%) compared to the risk for relatives of male felons (16%) may be related to a greater genetic and social predisposition in the families of the female felons.

Cytogenetic Studies

Several reports have suggested that the prevalence of XYY males in prisons and penal/mental institutions is higher than the prevalence in the general population. The XYY karyotype is associated with slightly lower than normal intelligence, tall stature, and cystic acne. This karyotype is found in approximately 1/1000 of male newborns. Hook found XYY in 1/53 of 3813 males in 20 penal/mental institutions. Witkin surveyed all tall Danish men from a birth cohort, finding 12/4139 (0.29%) who had the XYY anomaly. Five of these 12 XYY men had a criminal record, primarily petty criminality. Witkin suggests that lower than average intelligence may account for the excess of criminal activity among XYY males. This karyotype does not seem to be associated with a predisposition to impulsive violence, as once thought.

Biological Markers

Nielsen identified a variant of the tryptophan hydroxylase gene (which codes for the synthetic enzyme for serotonin) associated with low 5-hydroxyindoleacetic acid (5-HIAA) in cerebrospinal fluid and suicide attempts in violent criminal offenders. Goldman replicated this finding. This deserves follow-up using family-based association methods, though it is now clear that brain serotonin is almost exclusively produced by another metabolic enzyme (TPH2). Low 5-HIAA has been associated with impulsivity and violence in experimental colonies of rhesus monkeys. A Dutch family was reported with lowered monoamine oxidase A activity caused by a point mutation on the eighth exon of the MAOA gene. Males with this mutation (both MAO genes are on the X chromosome) show impulsive aggression, arson, attempted rape, and exhibitionism. It is likely that other familial monoamine defects will be found to be associated with aggressive behavior.

Anxiety Disorders

The increased familial risk for anxiety disorders has been known for over 100 years. Family studies of panic disorder using modern criteria are often complicated by comorbidity with social phobic disorder and generalized anxiety disorder. One family study of pure panic disorder probands found a significantly higher risk of panic episodes among first-degree relatives compared to relatives of controls. There was a fivefold increase in risk for any anxiety disorder. Similarly, an increased (11.6%) risk for agoraphobia has been reported for the relatives of agoraphobic probands, compared to 1.9% for relatives of panic probands and 1.5% for control probands. A study of simple phobia found an increased risk (31%) for simple phobia among relatives of probands with that diagnosis (but no other anxiety disorder) compared to relatives of well probands (11%). A family history study of social phobia demonstrated that relatives of phobic probands are at increased risk for this disorder (6.6%) compared to relatives of panic disorder probands (0.4%) or relatives of controls (2.2%).

A separate genetic transmission for generalized anxiety disorder has not been established. In a family study, Noyes found that relatives of probands had a greater risk than relatives of controls, but this risk is not greater than the risk for relatives of panic disorder probands. Conversely, a separate study reported similar risks for generalized anxiety among relatives of panic disorder probands and relatives of probands with generalized anxiety. Thus, while there is some evidence for familial transmission of generalized anxiety, the transmission may not be specific. In summary, family studies provide evidence that some anxiety disorders are transmitted separately. This is best established for panic disorder and least so for generalized anxiety.

Twin Studies

In a Norwegian sample, the concordance of all anxiety disorders for MZ twins (34.4%) was significantly greater than that for DZ twins (17.0%).

Hamilton and colleagues reported a syndrome linked to chromosome 13q; the complex phenotype included anxiety disorders and urinary tract dysfunction.

Attention Deficit Disorder

Early family studies of attention-deficit disorder noted alcoholism and sociopathy in male relatives and hysteria in female relatives. This same constellation was not manifest in the adoptive parents of adopted ADD children.

Family studies suggest that antisocial personality aggregates in the relatives of ADD children, specifically when the children have conduct or oppositional disorder. Biederman and colleagues found rates of affective illness increased in relatives of their group. ADD itself was also increased in relatives. ADD and antisocial behavior tended to occur together.

Deutsch et al. have reported on the association of minor physical anomalies with ADD. The relationship is consistent, with a genetic latent trait model (an underlying autosomal dominant gene producing either ADD, physical anomalies, or both). Mutations in the gene for the thyroid hormone receptor have been found in one group of subjects with ADD. More recent studies in large cohorts have reported association with several dopamine-related genes, including the dopamine transporter and DRD4.

Autism (Pervasive Developmental Disorder)

The pooled frequency of autism in sibs is about 3%, which is 50–100 times the population rate. Folstein and Rutter reported an MZ concordance of 36% and a DZ concordance of 0%. When the phenotype was expanded to include language and cognitive abnormalities, concordance rates were 82% and 10%. This sample of twins, though carefully selected, was small, but the essential conclusions regarding heritability have been confirmed in later studies.

A segregation analysis in a series of multiplex families was consistent with autosomal recessive inheritance. However, the excess of affected males in the sample suggested sex-specific modifying factors.

What is striking about the genetics of autism is its association with multiple single-gene disorders. The most clearly documented of these disorders is the fragile X syndrome. Perhaps 8% of autistic subjects have the cytogenetic fragile X; 16% of fragile X males are autistic. There are also probable associations between autism and tuberous sclerosis, neurofibromatosis, and phenylketonuria. A variety of other reports of chromosomal anomalies and single-gene associations with the autistic syndrome have been reported.

A number of genome-wide genetic surveys of the autistic phenotype have been reported. All these studies have examined affected pairs of siblings where both twins have a narrowly defined phenotype of autism or where one sib has the narrowly defined autistic phenotype and the other has a defined pervasive developmental disorder. A consistent finding is seen on the long arm of chromosome 7 from 7q22-qter. Folstein reported that the linkage on 7q was specific to families in which the proband had a specific language disorder (usually reading difficulty along with later-onset autism). It is notable that the linked region includes the gene recently dubbed “speech 1” also known as FOXP2 and known to be a transcription factor), which was recently found to be associated with specific language disorder.

A significant site has been found on chromosome 2q32. Other statistically suggestive regions found to date are on chromosomes 5q14, 13q21, and 16p13.3, and near the centromere of chromosome 19.

An autistic phenotype found in individuals with the chromosomal duplications of Prader-Willi/Angelman syndrome on 15q11-q13 has focused considerable research in this region of the genome. Interest has concentrated on the GABRB3 receptor gene in 15q12. GABRB3 shows peak expression both temporally and spatially during pre- and early postnatal murine brain development.

Due to the finding of hyperserotonemia in a proportion of autistic children, it has been suggested that the serotonergic system may play an important role in the etiology of the disease. There are conflicting results regarding the genetic involvement of the serotonin transporter.

Recent data suggests involvement of several of the neuroligin and neurexin proteins (involved in synaptic structure) in genetic predisposition to autism. As more single gene findings emerge, the genetic structure of autism appears quite complex, with cytogenetic abnormalities, common gene variants and rare variants all represented.

Drug Dependence

Genetic Epidemiology: Adoption Studies

In an adoption study of drug abuse, 443 adoptees from Iowa were studied; half were selected for psychopathology in biologic parents and the other half matched for age and sex. Parents were not directly examined but information from adoption records was available. Forty adoptees manifested drug abuse of one kind or another. Antisocial behavior in a biologic relative predicted drug abuse in the adoptee. Alcohol problems also predicted drug abuse in the absence of antisocial behavior. The environmental factors implicated included divorce and significant psychiatric pathology in the adoptive parents.

In an adoption study of smoking behavior, using a large Swedish cohort, the authors observed that adoptees’ status (e.g., current or heavy smoker) is associated with their siblings’ smoking status. Surprisingly, adoptees’ smoking status was not predicted by their biologic or adoptive parents smoking status.

Genetic Epidemiology: Twin and Family Studies

A study of the Vietnam Era Twin Registry revealed evidence for a common genetic factor operating across pharmacologic classes. These authors also found evidence for class-specific genetic factors, especially in opioid dependence, similar to results of a family study. Abuse and dependence for six different classes of drugs was assessed in an adult male twin cohort of Virginia residents (≈1200 twin pairs). The authors report the detection of a common genetic influence for use and abuse/dependence, across the pharmacologic classes of drugs of abuse, as well as pharmacologic class-specific genetic factors predisposing to use. A single common environmental factor predisposed to illicit use. These results confirm their earlier study of female–female twin pairs. Since these were two population-based twin studies, there were few opioid dependent cases and few cocaine dependence cases, a cautionary note. The relatively clear conclusion from these twin, family, and adoption studies is that there are general genetic factors increasing risk for addiction to multiple drugs of abuse, and there are pharmacologic-class specific genetic factors that appear to increase risk predominantly for addiction to a single pharmacologic class of drugs (e.g., opioids).

Linkage Studies

Two linkage studies of opiate dependence have been published, both of which implicate areas on chromosome 17q.

Multiple linkage studies of nicotine dependence have been published in the last 5 years. While many loci have been nominated, most LOD scores have been below genome-wide significance, and confirmation has been less frequent than hoped. There are several promising findings, however. Saccone combined two linkage scans of Australian and Finnish smokers to report a LOD score > 5 (genome-wide significant p value = 0.006) at ≈25 cM on chromosome 22.

Association Studies

Recent genome-wide association studies by Bierut of ≈1100 nicotine dependent probands and ≈900 controls identify the alpha 5 nicotinic cholinergic receptor cluster (including the alpha 5, alpha 3 and beta 4 receptor subunit genes) as associated with nicotine dependence. These results implicate biologically plausible candidate genes. One implicated SNP is a mis-sense variation in the alpha 5 subunit gene. Multiple other candidate genes are expected to emerge from more intensive analysis of these data.

Association studies of candidate genes for nicotine, stimulant (including cocaine), or opioid dependence have nominated literally dozens of alleles as risk factors. Most will eventually prove to be false positives, but, since odds ratios may be less than 1.5 for true positives, large-scale studies in multiple ethnic populations are needed.

Eating Disorders

Family Studies

Controlled family studies have been conducted over the past two decades. These studies suggest that there is considerable familial aggregation. Table 3–4 summarizes the results from these studies. It is difficult to estimate precisely the risk to first-degree relatives because the control samples are not sufficiently large to detect more than one to two affected relatives of controls. However, the overall pattern suggests substantial risk, almost certainly greater than 10 and perhaps much larger. There are increased rates of AN among first-degree relatives of DN probands, and increased rates of BN among first-degree relatives of AN probands. This clustering of eating disorders in families of AN and BN individuals provides strong support for familial transmission of both disorders.

Twin Studies

There have been a number of twin studies of AN. However, many of the studies were small and often had methodological weaknesses. If one examines the twin studies with the largest number of subjects and most appropriate methodology, mean concordance rates are 64% for MZ twins and 14% for DZ twins. Differences between these rates suggest a modest additive heritability with a large influence of nonadditive genetic and/or shared environmental factors. More recent studies have used structural models to estimate the fraction of risk attributable to additive genetic factors. The estimates of heritability range from 0.48 to 0.76. These studies are summarized in Table 3–5.

One of the first twin studies of eating disorders described pairwise concordance of 56% in MZ and 5% in DZ pairs (71% and 10% with probandwise figures). Family history assessment (including additional informant data from parents) showed that 4.9% of the female first-degree relatives and 1.16% of the female second-degree relatives had had anorexia at some point in their lives, a risk considerably higher than the reported population prevalence. The MZ cotwins were much more similar in “body dissatisfaction,” “drive to thinness,” weight loss, length of amenorrhea, and minimum body mass index. Estimates indicate that roughly 58–76% of the variance in the liability to AN, and 54–83% of the variance in the liability to BN, can be accounted for by genetic factors. Although the confidence intervals on these estimates are wide, consistent findings across studies support moderate heritability of these traits. For both AN and BN, the remaining variance in liability appears to be due to unique environmental factors (i.e., factors that are unique to siblings in the same family) rather than shared or common environmental factors (i.e., factors that are shared by siblings in the same family).

Eating disorder symptoms themselves also appear to be moderately heritable. Twin studies of binge eating, self-induced vomiting, and dietary restraint suggest that these behaviors are roughly 46–72% heritable. Likewise, pathological attitudes such as body dissatisfaction, eating and weight concerns, and weight preoccupation show heritabilities of roughly 32–72%. Taken together, findings suggest a significant genetic component to AN and BN as well as the attitudes and behaviors that contribute to, and correlate with, clinical eating pathology.

Molecular Studies

The first AN linkage scan was based on ≈200 multiplex kindreds and revealed a locus on 1p (NPL score = 3.5 at D1S3721, 72.6cM; Grice et al., 2002). Additional genotyping in the region resulted in an increased NPL score of 3.91 at 72.0 cM. Analysis of diagnostic phenotypes, using obsession scale scores and drive for thinness scores as covariates, revealed additional linkage peaks. SNP genotyping at several candidate genes (HTR1D, HCRTR1 and OPRD1) revealed limited evidence for association with the HTR1D and OPRD1 genes (Bergen et al., 2003). These observations were confirmed in an independent population of AN individuals (Brown et al., 2007).

The first BN linkage scan was based on ≈300 multiplex families and yielded a genome-wide significant LOD score of 2.92 on chromosome 10p. When analysis was restricted to those ≈133 multiplex kindreds characterized by self-induced vomiting, the LOD score increase on 10p to 3.39. A promising candidate gene within the 10p linkage peak is glutamic acid decarboxylase (GAD2), a gene implicated in obesity.

Many family-based and case-control association studies of monoamine-related, obesity-related, and neurotrophin-related genes have been published in the past 10 years. These have been small, underpowered, and limited in the numbers of genes (and variants within genes) tested. More recently larger samples sizes have been employed in candidate gene studies, using collaborative, multisite approaches. For example, it was reported that the Met allele of a mis-sense variation in the BDNF gene was associated with AN in Spanish patients. Subsequently, this was confirmed (Ribases et al., 2005) in European collaborative samples totaling greater than 1500 patients.

Mental Retardation

Epidemiologic Studies

Twin studies (performed 40–60 years ago) show an MZ concordance of 100% (N = 83) and a DZ concordance of 55% (N = 10). These would undoubtedly be performed today with separation according to specific causal factors. Adoption studies have not been performed.

Recurrence risk for siblings of a retarded child has been estimated to range from 9.5% to 23% depending on severity of the disorder and the mother's reproductive history. For mothers who have already had more than one retarded child, the risk is 25–50% for sibs.

Specific Etiologic Causes

Many medical syndromes are manifest as mental retardation, such as specific errors of metabolism and chromosomal anomalies. Polani estimated that of the 4% of human conceptions that are chromosomally abnormal, 85–90% are selectively eliminated as spontaneous abortions. Of live births, 6% may have a genetic or developmental abnormality of some type; 0.5% survive with chromosomal abnormality; 4% with another developmental anomaly; and 1.5% with a single gene disorder. Among single gene causes of mental retardation, Koranyi listed five dominant diseases (tuberous sclerosis, neurofibromatosis, Sturge–Weber disease, von Hippel-Lindow, and craniosynostosis), and four recessive diseases (Hurler–Hunter disease, galactosemia, G-6 phosphodehydrogenase deficiency and familial hypoglycemia), as well as three recessive aminoaciduria and three lipid-related disorders. Many more are listed in McKusick's compendium Mendelian Inheritance in Man.

Down syndrome accounts for MR in 1.5 persons per 1000 and is the most common single cause of the condition. The prevalence of Down's varies greatly and is primarily determined by maternal age. Familial microcephaly is present in about 1/40,000 births but may account for a significant proportion of MR because of its effects in heterozygotes (see below). Fragile X syndrome accounts for about 0.5/1000, and other X chromosome syndromes for another 1/1000. All metabolic causes together are responsible for 1/1000 and chromosomal abnormalities for 3/1000.

Down Syndrome

This condition, well studied, is accounted for by a triplication of genetic material on chromosome 21. The area is being localized more and more precisely using molecular techniques combined with cytogenetics. It is probable that sections of 21q22.2 and 21q22.3 are involved, though 21q21 may also be implicated. The areas involved include genes for amyloid and superoxide dismutase. The ETS-2 proto oncogene is near this area as well, and its presence may be related to the well described increased incidence of leukemia in persons with Down syndrome and their relatives. Human 21q21–22.3 is homologous to portions of mouse chromosome 16. A mouse model of Down has been described based on a laboratory-generated reciprocal translocation involving this area.

The reasons for triplication or nondisjunction in Down's are not entirely clear. The likely etiologic factors are environmental rather than genetic. Vulnerability for the condition does not seem to be inherited. A small proportion of Down's patients have a translocation rather than a triplication.

As noted above, the clearest correlate is maternal age. Yet it has been known for some years that the origin of the nondisjunction might be paternal as well as maternal. Serum markers contribute to prenatal determination (decreased alpha-fetoprotein and estriol and increased human chorionic gonadotropin), and aid in the selection of women for referral to amniocentesis.

It has been reported that a familial association exists between Alzheimer's and Down’s, but this is unlikely to be generally true. Recent studies suggest that triplication of a critical region on chromosome 21 is not likely to be the sole cause of clinical variation in Down syndrome, and that other genomic areas are probably important as well.

Fragile X Syndrome

Fragile X syndrome is named after a cytogenetic observation; cultured cells from some patients show chromosomal breakage under appropriate conditions. There are actually multiple “fragile sites” on human chromosomes. The fragile X (breakage at Xq27.3) is merely the best known. The syndrome itself was originally described by Martin and Bell who described a large pedigree with mental retardation segregating in an x-linked recessive pattern.

Fragile X is the most common form of X-linked MR and is, in general, the most common heritable form of MR (Down's being genetic but not inherited). It is estimated that 1/850 persons carry the defect. Of those, 4 out of 5 males will express the clinical phenotype as compared with 1 out of 3 females (some homozygotes are nonpenetrant and some heterozygotes are penetrant). Genetic tests are now available to determine carrier status in nonpenetrant individuals. The precise genetic error in the Xq27.3 region is now known to be a triplet repeat of variable length. Increased numbers of repeats (associated with greater severity of illness) occur as the gene is passed to succeeding generations. When the number of repeats exceeds a threshold, clinical manifestations are seen.

Most female Fragile X heterozygotes do not have MR. However schizotypal features are seen in about one third of a sample of carriers, and there is an association with affective disorders as well. Some subjects with fragile X develop an autistic syndrome.

Obsessive–Compulsive Disorder

Epidemiologic Research

Twin Studies

There are no large twin studies of OCD. Rasmussen and Tsuang reviewed reported series and noted that 32 of 51 (63%) MZ pairs were concordant. Lenane studied 145 first-degree relatives of 46 children with OCD. Of the 90 parents personally evaluated, 15 (17%) received a diagnosis of OCD, compared to 1.5% of the parents of 34 conduct-disordered children who served as a control group. The 17% prevalence rate is significantly higher than the population prevalence rate of about 2%. Fathers were three times as likely as mothers to receive a diagnosis of OCD. Of the 56 siblings personally evaluated, three (5%) met criteria for OCD. When age-correction was applied, the rate rose to 35%. This figure should be viewed with caution because of the magnitude of the age correction for siblings. It should be noted that all probands had severe childhood-onset OCD, and were referred to the authors for treatment. It is possible that childhood-onset OCD represents a more severe form of the OCD spectrum. Nevertheless, this carefully conducted family study reveals an increased risk of OCD among the first-degree relatives of OCD probands.

When OCD occurs in the familial context of Tourette syndrome, it may be part of the spectrum of Tourette syndrome. However, most OCD occurs in individuals who have no first-degree relatives affected by Tourette’s. Occasionally, an individual destined to develop Tourette's will present with symptoms of OCD, and the motor tics appear subsequently. These patients are often diagnosed as having OCD until motor tics develop.

OCD appears to be familial and single gene identification studies are proceeding. Several avenues of research suggest a serotonergic abnormality for OCD patients. Recent data from Goldman shows a rare mutation in the serotonin transporter associated with obsessive–compulsive disorder in a single pedigree. An association was also recently reported with oligodendrocyte lineage transcription factor 2 (OLIG2).

Schizophrenia

Twin Studies

MZ twin concordance is greater than DZ twin concordance in each study of schizophrenia, consistent with genetic hypotheses. Second, the heritability of broadly defined schizophrenia is greater than the heritability of strictly defined schizophrenia. This is consistent with a spectrum concept: some individuals with the genetic loading for schizophrenia manifest a different condition. Third, the amount of discordance is considerable; even in MZ twins using a broad definition of illness the discordance is 51%.

A series of nine MZ twins with schizophrenia who were raised apart from infancy was described. Three were completely concordant and three were partially so.

Twin Study Paradigm to Generate Data Regarding Environmental Effects in Schizophrenia

Examining age of onset in the Maudsley Hospital twin series, it was found that there was a high incidence of illness in the second of a pair of twins within 2 years of onset in the first twin. Further categorizing the group on the basis of whether the twins lived together or apart, he found the 2-year incidence to be primarily in those living together. That is, twins living together are concordant in age of onset, while twins living apart do not. This intriguing finding suggests an environmental factor.

Family Studies

Pooled European family study data show an age-corrected morbid risk of 5.6% in parents, 10.1% in siblings, and 12.8% in children. The lower rate in parents is thought to be related to a relative decrease in fertility among schizophrenic patients. The general population figure concerning risk for schizophrenia is about 1%; thus, all classes of first-degree relatives of schizophrenic patients have a clear increase in prevalence. The risk for offspring of two schizophrenic parents is difficult to estimate because of the small number of cases. It is probably between 35% and 45% (in pooled data it is 46.3%). Among second-degree relatives (uncles, aunts, nephews, nieces, grandchildren), half-siblings, and cousins, the risk is 2–4%.

Thus, close relatives of schizophrenic patients have about a five- to tenfold excess risk for the illness. The risk diminishes in more distant relatives. A further group of first-degree relatives appear to develop “schizophrenic spectrum” disorders (see below). Nevertheless, the majority of close relatives of schizophrenics are psychiatrically normal.

There is no strong evidence for genetic determination of the conventional subtypes of schizophrenia (hebephrenic, catatonic, and paranoid forms). Though there is significant concordance in MZ twins for subtype, this does not hold true in family studies.

The question of the distinctness of schizophrenia and affective disorders is not settled. In a large family study using lifetime diagnoses and separately examining relatives of probands with schizophrenia, chronic schizoaffective disorder, acute schizoaffective disorder, BP affective disorder, UP affective disorder, and controls, it was concluded that there was evidence for overlap in genetic liability. Specifically, an increase in UP disorder was seen in all groups of relatives of patients. Relatives of schizoaffective probands (both chronic and acute) showed both an excess of affective disorders and an excess of chronic psychoses. However, BP probands did not show an excess of schizophrenic relatives, and schizophrenic probands showed no excess of BP relatives. The most parsimonious explanation of these data is that there is a “middle” group of disorders (schizoaffective) that is genetically related to both schizophrenia and affective illness, and that it may not be possible to separate the groups on clinical criteria.

With regard to mode of transmission, the available data have been analyzed extensively. The results have generally been interpreted as favoring a multifactorial rather than a single-locus model.

Adoption Studies

The adoption study methodology was first applied to schizophrenia ≈40 years ago. It was reported that there is an excess of schizophrenia in the adopted-way offspring of schizophrenic women compared to control adoptees. A series of large, systematic studies were carried out by Kety and Rosenthal, who analyzed adoption and psychiatric hospitalization registries in Denmark. In the later studies, subjects were directly interviewed. In all studies, adoptees were separated from their biologic parents at an early age and adopted by non relatives. It was found that there were more schizophrenia and schizophrenia spectrum disorders in the biologic relatives of schizophrenic adoptees than in the biologic relatives of psychiatrically normal adoptees. The prevalence of psychiatric illnesses in the adoptive relatives of the two groups was comparable but small.

The frequency of schizophrenia spectrum disorders is higher in adopted-away offspring of schizophrenic parents than in the adopted-way offspring of normal parents. All of these studies have been criticized on the grounds of selection bias and the validity of diagnosis, and comparisons. However, further independent analysis of the data has confirmed the results: that is, biologic relatives of schizophrenics who have not shared the same environment have a significantly higher prevalence of schizophrenia and schizophrenic spectrum disorders than do biologic relatives of comparable control groups.

Spectrum Studies

Thirty percent of first-degree relatives of schizophrenic patients have associated disorders. The particular DSM-III-R diagnostic categories that seem to be implicated are paranoid personality and schizotypal personality. Schizotypal personality disorder is most likely part of the schizophrenia spectrum, with suggestive evidence for paranoid and schizoid as well. A separate entity characterized by paranoid delusions only (simple delusional disorder) may also exist, for which the inheritance is independent of that for schizophrenia and affective disorder.

Molecular Studies

Multiple linkage scans of the schizophrenic genome have been conducted, using DSM-III-R, DSM-IV, IDC and/or RDC criteria. These linkage scans have been the subject of meta-analyses in which available data have been combined using different methods. Results are only partially convergent. Genomic regions implicated in the meta-analyses include 1q, 5q, 6p, 6q, 8p, 13q, 15q and 22q. Several of these regions will be discussed below with a focus on those that contain highly promising candidate genes.

On chromosome 6p, there is evidence that alleles of the dysbindin (DTNBP1) gene are associated with schizophrenia, although there are some negative reports (for review see Ross et al., 2006). Multiple haplotypes have been associated with schizophrenia in these reports. There is evidence that dysbindin levels are reduced in postmortem schizophrenia brains. An SNP in the 3’UTR of the dysbindin gene may mediate the reduced expression of dysbindin. At least one dysbindin risk haplotype may be associated with decreased cognitive ability. There may be some overlap between psychotic BP disorder and schizophrenia in terms of dysbindin risk alleles.

An 8p candidate gene, neuregulin 1 (NRG1), is associated with schizophrenia (Steffansson et al., 2002). While there is substantial evidence for the role of NRG1 in the genetics of schizophrenia, other investigators (some employing multiplex samples with positive linkage signals in the region) have not confirmed the association of NRG1 with schizophrenia. Others have found evidence for an association of NRG1 alleles with psychotic BP disorder, suggesting a genetic overlap between this entity and schizophrenia.

Alleles at G72 and G30, two novel genes on 13q32 are associated with schizophrenia in multiple populations (for review see Li and He, 2007). G72 and G30 overlap and orient in opposite directions on 13q32. G72 is a primate-specific gene possibly expressed in the caudate and amygdala; (however, some researchers have not been able to detect mRNA in postmortem human brain). Using yeast two-hybrid analysis, evidence for physical interaction was found for G72 and D-amino-acid oxidase (DAO). DAO oxidizes D-serine, a glutamate receptor modulator. Coincubation of G72 and DAO in vitro revealed a functional interaction, G72 enhancing the activity of DAO. As a result, G72 has been named d-amino-acid oxidase activator (DAOA). Associations between DAOA and schizophrenia have been reported in samples from China, Germany, Ashkenazi Jews, South Africa, and the United States. Childhood-onset schizophrenia has been associated with DAOA in a small sample. Various risk alleles and haplotypes have been reported in schizophrenia.

Curiously, independent datasets suggest that the G72 locus contributes to the risk for BP disorder. No clear functional variation has been established.

Several other candidate genes have been implicated repeatedly in the etiology of schizophrenia, including RGS4, COMT, and DISC1, for which a translocation (with a breakpoint at the DISC1 gene) segregates with multiple behavioral disorders in a Scottish family (for review see Mackie et al., 2007).

Endophenotypes in Schizophrenia

The concept of endophenotypes in psychiatric disorders has been developed over the last several decades. The term defines an illness-related characteristic, observable through biochemical testing or microscopic examination. A valid endophenotype should be more closely related to more pathophysiologic genes for the nosologic category, compared to the entire symptom syndrome in a nosologic category.

The utility of endophenotypes in psychiatric research is now more appreciated, because we have a more accurate understanding of the genetic complexity of operationally defined disorders in the current psychiatric nosology. Endophenotypes should create more homogeneous subtypes, which may cut across the current nosologic boundaries. In that case, more rapid advances can be made in understanding these disorders at the molecular level.

Criteria for an Endophenotype

Criteria for an endophenotype have been derived.

• It must be associated with illness in the general population;
• It should be stable and state-independent. In other words, it must be observable despite the fact that the patient is in partial or complete remission;
• It should be heritable;
• It should segregate with illness within families.

Among kindreds in which the proband has the endophenotype, it should be observable at a higher rate among unaffected family members compared to the general population.

There are many reports of attentional deficit measures in schizophrenia. An endophenotype that has been studied extensively in schizophrenia is “working memory.” This term can be defined as the “holding of information in consciousness, in preparation for complex processing.” Working memory can be assessed through multiple different mental tasks, such as N back, Wisconsin Card Sort, and reverse digit span. Deficits in working memory have been described as an endophenotype for schizophrenia. The fraction of individuals with schizophrenia who are designated as having abnormal working memory varies with the tests employed, the sample, and the definition of abnormal (e.g., 1.5 or 2 standard deviation units below the mean for controls). If consideration is given only to studies of large numbers of cases and controls, most reports find 25–50% of persons with schizophrenia as falling in the variably defined “deficit range” for working memory.

Several lines of evidence suggest that working memory deficits are in part heritable. Twin studies of subjects unaffected and discordant for schizophrenia who are MZ and DZ twin pairs indicate that genetic influences in schizophrenia-related working memory deficits are prominent. Multiple studies suggest that a small fraction of the variance in working memory scores is explained by a functional mis-sense SNP (Val/Met) in the COMT gene.

Compared to controls, working memory deficits are more common among the unaffected relatives (compared to controls) of schizophrenic individuals who have deficits themselves. The effect size for this observation is relatively small; substantial sample numbers are required to have adequate power. If only those studies which examined at a minimum more than 50 relatives and 50 controls are considered, the preponderance of data suggests that unaffected relatives of schizophrenic individuals have some of the neuropsychological deficits seen in affected persons. However, there is a negative publication bias, and a great variety of neuropsychological measures have been used (Wisconsin Card Sort, digit span, trail making, tests of verbal and spatial fluency, etc.). The effect size is not large, as evidenced by the fact that many smaller studies have found no significant difference between the relatives of schizophrenic individuals and controls.

The preponderance of data suggests that neuropsychological/cognitive deficits in schizophrenia are present more often among affected persons, compared to controls. There are data to indicate that the measures are heritable. Finally, most larger studies find that the nonpsychotic relatives of schizophrenic individuals score comparably poorly on neuropsychological tests. The different measures of neurocognitive functioning may be valid endophenotypes for schizophrenia.

A promising endophenotype for schizophrenia is an abnormality of the P50 auditory evoked potential. The P50 wave is a positive deflection on by scalp electrodes occurring 50 milliseconds after an auditory stimulus, typically a single click. When two such clicks are presented, (the second, 200 milliseconds or more after the first), the amplitude of the P50 wave after the second click is normally reduced, in comparison to the amplitude of the wave after the first click (see figure below). This phenomenon is considered by some to be an electrophysiologic signature of sensory gating. In some subjects with schizophrenia, the amplitude of the P50 wave for the second click is similar to the amplitude after the first click, as shown below. This has been interpreted as a defect in sensory gating. (Figure 3–1)

The P50 abnormality is found more often among individuals with schizophrenia, compared to controls, although this has not been universally confirmed. The abnormality is also found more frequently among the relatives of persons with schizophrenia, compared to controls. Based on twin studies, it is partially heritable. Heritability is also implied by reports that DNA sequence polymorphisms in and near the alpha 7 nicotinic receptor subunit gene on chromosome 15 explain some of the variance in the P50 abnormality. The chromosome 15 location is a confirmed linkage region for schizophrenia, lending interest to this line of investigation.

While there is ample evidence that the P50 is under partial genetic control, there is substantial evidence that P50 parameters are influenced by environmental forces. Nicotine can “normalize” an abnormal P50 test. This finding becomes more intriguing when it is recalled that about 80% of individuals with schizophrenia are daily smokers. Furthermore, atypical antipsychotic medications can “normalize” abnormal P50 testing. These results indicate a critical point when endophenotypes are considered: environmental influences must be considered, not only as sources of variance (e.g., experimental error, circadian variation, influence of personal habits such as nicotine and caffeine intake), but also as clues to the pathways from gene variants to endophenotypes, or from endophenotypes to key symptom clusters.

In summary, genetic studies of schizophrenia have identified numerous promising candidate genes through linkage and association approaches. These include DAOA, NRG1, dysbindin, DISC1, RGS4, COMT and others. Research has revealed several promising endophenotypes, particularly auditory evoked potential abnormalities and neurocognitive deficit.

Somatization Disorder

In a family history study, Coryell evaluated first-degree relatives of 49 probands with Briquet syndrome. First-degree relatives of non-Briquet syndrome hysteria and affective disorder probands formed the control groups. The risk for a complicated medical history was 8.0% in the first-degree relatives of Briquet syndrome probands compared to control values of 2.3% and 2.5% respectively.

In a family study of Briquet syndrome, Guze and colleagues reported a significantly increased risk for Briquet syndrome among the first-degree female relatives of Briquet syndrome probands (7/105), compared to female relatives of control probands (13/532). Additionally, they reported an increased risk for antisocial personality among the male (18/96) and female (9/105) relatives of the Briquet syndrome probands, compared to the risk for male (44/420) and female (14/532) relatives of controls.

Torgersen studied 14 MZ twin pairs and 21 DZ twin pairs in which one member had a somatoform disorder (somatization disorder, conversion disorder, psychogenic pain disorder or hypochondriasis). 29% of MZ twin pairs were concordant for somatoform disorder compared to 10% of DZ twin pairs. This difference was not significant.

Adoption Studies

In an analysis of a large Swedish adoption cohort, Sigvardsson identified a set of discriminant function variables that distinguished female adoptees with repeated brief somatic complaints and psychiatric disability (“somatizers”) from other female adoptees. In a subsequent analysis, Bohman divided somatizers into two groups, “high-frequency somatizers” (those who have high prevalence of psychiatric, abdominal, or back complaints) and “diversiform somatizers” (those who have a lower frequency of complaints, but multiple, highly variable symptoms). Thirty percent of the high-frequency somatizers had histories of alcohol abuse and/or criminality (based upon the national registries for these behaviors). Their male biological relatives were at increased risk for violent criminal behavior and alcohol abuse. For both types of somatizers, a cross-fostering analysis provided evidence for both congenital and postnatal influences on the development of somatoform disorder. These studies suggest a familial connection between some types of somatoform disorder, alcoholism, and criminality. Torgersen makes the appropriate point that not much research has been carried out in this area in recent years the reason for this is not clear.

Tourette Syndrome

Epidemiologic Research

Twin Studies

Price studied 43 pairs, 30 MZ and 13 DZ same-sex pairs. MZ twin concordance was 77% for any tics, compared to 23% for DZ twins. For Tourette disorder proper, the MZ concordance rate was 53%, compared to the DZ rate of 8%. These are all significant differences.

Pauls studied 338 biological relatives of 38 Tourette disorder probands, 21 adoptive relatives and 22 relatives of normal controls. Among the biological relatives, 8.3% had Tourette disorder, while 16.3% had chronic tics and 9.5% had OCD. These risks are all significantly greater than the risks for the 43 relatives of controls.

Segregation analyses have fit both single locus models and polygenic models to the familial pattern of this disorder.

Linkage Studies

A collaborative effort to use systematic genomic screening to find genes causing Tourette's has been underway for several years. Recent results implicate chromosome 2p. Rare mutations in the dendritic growth protein SLITRK1 (chromosome 13q) have been associated with this condition.

Empirical data for genetic counseling is summarized in Table 3–6. These data assume that the other parent is unaffected. The percentages in parentheses after the family history provide the unselected general population risk. For example, the general population lifetime risk for UP illness (defined narrowly) is 8%. If a subject has a parent with UP depression, the lifetime risk is double that of the general population (16%). Such individuals are also at fourfold increased risk for BP illness.

Some illnesses have narrow age-at-onset distributions in the general population. For example, first episodes of BP illness almost always occur before age 60. Fully 50% of BP individuals have their initial episode (depressive or manic) prior to age 20. This should be considered in a general way when assessing risk. For example, an unaffected 40-year-old son of a parent with BP illness has already passed through most of the age at risk, and thus, his risk is substantially less than 9%. An estimate of 2% would be more accurate in this case.

It is anticipated that genotypic methods will be adapted for use in genetic counseling in the coming years. Such methods are not yet widely applicable aside from use in certain unusual families with single gene conditions. Most experts feel that genotypic screening for persons with multifactorial disorders would still be premature; however some products are already on the market, and it seems likely that the predictive power of such methods will approach clinical utility within the next decade.