Why do researchers use adoption studies in an effort to reveal genetic influences on personality?

Twin Studies

Twin and adoption studies are natural experiments that can help determine whether the familial aggregation of BP and its related conditions is the result of genetic or environmental factors. Twin studies dating back to the 1920s have observed that monozygotic (MZ) twins are more concordant for mood disorders than are dizygotic (DZ) twins. However, as noted previously, these earlier studied did not distinguish between BP and MD. In a review of these earlier studies, Tsuang and Faraone (1990) combined data from 11 twin studies published between 1928 and 1986, composed of 195 MZ and 255 same-sex DZ pairs. They reported a proband-wise concordance of 78% for MZ and 29% for DZ twins, resulting in an estimate of heritability of 63%.

Since the 1990s, at least four twin studies using a modern concept of BP have been reported. These include studies of 486 twin pairs from the Swedish Psychiatric Twin Register (Kendler, 1993), 224 twin pairs from the Maudsley Twin Register (Cardno et al., 1999), 38 twin pairs from the Finnish Twin Register (Kieseppä, Partonen, Haukka, Kaprio, & Lonnqvist, 2004), and 303 twin pairs from the Norwegian Twin Register (Edvardsen et al., 2008). Estimates from these studies of the heritability of BP ranged from a low of 73% (Edvardsen et al., 2008) to a high of 93% (Kieseppä et al., 2004). Interestingly, all of the studies found that the remaining proportion of variation in risk for BP was explained by individual residual factors and that there was no notable contribution of shared environmental factors. The studies by Cardno et al. (1999) and Edvardsen et al. (2008) both found that the estimates of heritability increased when considering a broader spectrum of BP that included both BPI and BPII. In addition, in their study, Cardno et al. (1999) examined the overlap between BP and SCZ and found evidence for significant genetic correlation between the syndromes, 0.68 for BP and SCZ and 0.88 for BP and schizoaffective disorder (Fig. 30.2).

8 Conclusions

The 1998 CAP longitudinal adoption study of personality is perhaps the most methodologically sound and least environmentally biased study that human behavioral genetics has ever produced. However, the finding of no significant personality correlation between birthparents and their 240 adopted-away biological offspring is largely unknown. In contrast, the popular press and authoritative secondary sources such as psychology textbooks have widely reported claims of important genetic influences on the basis of reared apart twin studies, often accompanied by photographs of reunited twin pairs and a list of their supposed similarities. And yet, a plausible interpretation of the 1998 CAP study is that, in the context of the current “missing heritability” stage of molecular genetic research, it provides additional evidence that family and twin studies of personality and behavior have recorded nothing more than research bias and the impact of environmental influences. This conclusion is consistent with the position that family, social, cultural, economic, religious, and political environments—and not genetics—are the main factors underlying variation in human behavioral traits.

The results of this lost study should be at least as well known as the celebrated (yet greatly flawed) twin studies, and should have led to a serious reevaluation of the genetics of personality and behavior question in the first decade of the twenty-first century. However, it didn’t. Instead, the study has had little impact and does not hold a position of importance in either psychology or behavioral genetics.

Instead of accepting Plomin and colleagues’ final conclusion that their study produced a 14% heritability estimate, and that adoption studies may not be able to detect nonadditive genetic influences, the behavioral sciences would do better to accept the researchers’ preliminary conclusion that the 0.01 birthparent/adopted-away biological offspring correlation, a correlation that they believed “directly indexes genetic influence,” suggests that “genetic factors correlated with parents’ self-reported personality have little effect” (p. 212).

2.1 Behavior and Molecular Genetics

The heritability of E–I has been investigated using twin and adoption studies. Structural equation models fitted to behavior genetic data suggest that broad heritability of E–I is approximately .50. The remaining variance is attributed to the environment, gene–environment interaction, and measurement error. Models suggest that the influence of the unshared environment, unique to each child, is considerably stronger than that of the shared environment related to the family. Molecular genetic studies are beginning to investigate specific alleles that may correlate with E–I. For example, variations at the D4 dopamine receptor gene have been found to correlate with some traits related to E–I, such as excitement seeking and positive emotionality, although the variance explained by any single allele is likely to be small.

3.05.5.4 Exemplar Study: the Danish Adoption Study

An example of the benefits of record linkage in psychological research is the adoption study of schizophrenia in Denmark (Kety, Rosenthal, Wender, Schulsinger, & Jacobsen, 1975). Familial and twin studies of schizophrenia suggested strongly that schizophrenia was inherited, but these studies were open to the interpretation that the inheritance was cultural, not genetic, because family members are raised together, and identical twins may be raised in social environments which are more similar than the environments of fraternal twins. The Danish Adoption Study ingeniously combined the strategy of file linkage with interviews of cases, controls, and their relatives. In Denmark, each individual receives a number at birth which is included in most registration systems. The registration systems are potentially linkable, after appropriate safeguards and clearances. In the adoption study, three registers were used. First, all 5483 individuals in the county and city of Copenhagen who had been adopted by persons or families other than their biological relatives, from 1924 through 1947, were identified from a register of adoptions (Figure 4). These were linked to the psychiatric case register, wherein it was determined that 507 adoptees had ever been admitted to a psychiatric hospital. From case notes in the hospitals, 34 adoptees who met criteria for schizophrenia (about 0.5% of of the total number of adoptees) were selected, and matched on the basis of age, sex, socioeconomic status of the adopting family, and time with biologic family, or in institution, prior to adoption. The relatives of these 68 cases and controls were identified by linkage with yet another register in Denmark which permits locating families, parents, and children. After allowing for mortality, refusal, and a total of three who were impossible to trace, a psychiatric interview was conducted on 341 individuals (including 12 on whom the interview was not conducted but sufficient information for diagnosis was obtained). Eleven of the 118 biologic relatives of the index adoptees were schizophrenic (9%), vs. one in the 35 relatives of adoptive families of the index adoptees (3%), and three of the 140 biologic relatives of the control adoptees (2%). The findings for schizophrenia spectrum disorders (including uncertain schizophrenia and schizoid personality) also show a pattern consistent only with genetic inheritance (26/118 in the biologic relatives of index cases, or 22%, vs. 16/140 biologic relatives of control adoptees, or 14%).

2 The Genetics of Depression

After more than a century of family, twin, and adoption studies we can now state with certainty that almost all forms of behavior, including depressive psychopathology, are influenced by multiple genes. Such genes are called quantitative trait loci (QTLs) because they are likely to result in quantitative (continous) distributors of phenotypes that underlie a predisposition to depression. Specifically, this means that there is no single locus that by itself causes psychiatric syndromes: instead multiple alleles (gene variants) that interact to produce vulnerability are found at multiple loci within the genome. Genetic research in psychiatry has left little room for the supposition that a single gene at a given locus determines any given psychiatric syndrome. In other words, in different families, different gene variants are required to confer increased vulnerability. In the light of the manifold possibilities of gene–gene interaction, genetic research has provided a particularly interesting demonstration that genes do not act alone.

Genes must interact with nongenetic environmental factors to convert vulnerability into illness. This is an important aspect as much research in the past has been devoted to the understanding of the development of depression based entirely on an individual's biography, making depression one of the last hermeneutic resorts in medicine. Many studies have shown that there is little reason to doubt that major stressful experiences, in particular those that occur in early life or even during pregnancy, render an individual more liable to develop depression (Heim et al. 2000). But it needs to be recognized that the conversion of life events into severe psychopathology requires the presence of the appropriate genetic endowment. An additional level of complexity has been introduced by the studies of Kenneth Kendler, who showed that individuals with a high genetic risk of depression are more likely to expose themselves to depression-triggering life events (Kendler et al. 1997). Indeed, several studies have supported this surprising proposition, that genetic factors may indirectly influence the liability to major depression by predisposing an individual to an aversive environment.

This shows that genetic and nongenetic environmental factors are much too inextricably intertwined to be dissected with high precision. Only when functional genomics have ultimately allowed us to establish a broad enough database on the functions of individual genes and gene–gene interactions, will the influence of environmental nongenetic factors on vulnerability to, and the onset and course of, depressive syndromes become available to analysis.

Definition of the depressive phenotype for genetic studies is extremely difficult and findings from these tudies have been largely inconclusive. Another reason for the paucity of genetic studies of unipolar major depression is the psychosocial adversity toward studies that call the stress diathesis into question. The use of association studies, which allow detection of small effects, have for methodological reasons (the need for the whole genome scan) been limited to candidate genes. The most intriguing of these studies is that of Klaus Peter Lesch, who investigated—together with colleagues from the National Institute of Mental Health, Bethesda, MD—the serotonin transporter gene, located on chromosome 17 (Lesch et al. 1996). This serotonin transporter is of particular interest as the majority of antidepressants act on this protein located in the cell membrane of presynaptic nerve terminals. By blocking the reuptake of serotonin molecules these drugs, of which Prozac® is the best known, enhance serotonergic neurotransmission.

Lesch submits that a polymorphism in the transcriptional control region of the serotonin transporter may result in a long and a short version of the promoter polymorphism. The occurrence of the short allele has been shown to be associated not only with decreased transcriptional efficiency resulting in reduced serotonin transporter expression, but also with those personality traits that are likely to be relevant to vulnerability to depression. Most studies aimed at replicating this intriguing finding have failed to do so, but this does not invalidate the experimental approach. Rather, it suggests that the means we currently have to phenotype individuals are perhaps not suited to genetic studies. Researchers have questioned the usefulness of diagnostic algorithms such as the Diagnostic and Statistical Manual of Mental Disorders of the American Psychiatric Association (DSM-IV) or the International Classification of Disease of the World Health Organization (ICD-10) for studies of causality because the psychopathological symptoms that constitute psychiatric diagnoses may be the very remote effects of genes. Therefore, it has been suggested that it would be more fruitful if genetic studies focused on more direct measures of brain function.

Such a strategy is exemplified in the Munich Vulnerability Study conducted at the Max Planck Institute of Psychiatry, where individuals with no history of illness but, being members of families with a high genetic load for depression, are investigated not only according to psychopathology but also according to laboratory measurements and other symptoms that can be objectified. Among these laboratory assessments are neuroendocrine function tests and sleep electroencephalographic-activity recordings. As illustrated in Fig. 1, the hormonal response to a neuroendocrine function test specifically designed to identify impairments in stress hormone regulation is altered in probands with a high familial risk for depression. Their stress hormone release is significantly increased compared with matched controls. What is not yet known is to what extent premorbid stress hormone dysregulation may predict an increased likelihood of developing overt clinical depression.

The idea behind the strategy is to identify patients with impaired stress hormone regulation across diagnostic boundaries and to study their genetic commonalities. Of course, other symptoms or symptom clusters, e.g., psychotic symptoms, need to be considered regardless of the diagnostic category to which the individual patient may belong. The proposed breakdown of the diagnostic description of complex behavior into simpler components including specific symptoms, together with results from neurobiological investigations (e.g., functional brain imaging, electroencephalography, drug treatment response, neuroendocrine assessment, etc.) deserve more attention in future genetic research than do traditional diagnostic attributions.

Bipolar depression, where depressive episodes alternate with manic episodes characterized by elevated mood, ideas of grandiosity, flight of thought, sleeplessness, etc., is much more easily phenotyped. Attempts to understand this disorder in terms of biographic events have also not been very successful, thus facilitating a push towards genetic research. Morbid risk in first-degree relatives is four to six times higher than the population prevalence of 1% for both men and women, making the case for genetic susceptibility particularly obvious. Enormous efforts in the United States and Europe have allowed the study of families large enough to discover a linkage between bipolar disorder and certain areas on several chromosomes, notably 12q, 18p, 18q, 21q, 4p, and X1.

Of particular interest is a study by Nicholas Barden from Quebec, Canada, who completed a genome screen in a large French Canadian pedigree with over 100 individuals (Morissette et al. 1999). This investigator showed a linkage to chromosome 12q 23–24 in some, but not all branches of the pedigree and concluded from his data that more than one locus may be segregating in this isolated population. A smaller pedigree from the same geographic region also confirmed the linkage between bipolar depression and chromosome 12. Similar conclusions were drawn from another investigation into a family where major affective disorder was segregating with Darier's disease. The gene, whose mutation for Darier's disease encodes a calcium ion pump, was recently identified but screening for mutations in a sample of bipolar probands without Darier's disease did not identify any gene mutation, rendering it unlikely that the susceptibility gene for bipolar disorder is in close proximity to the Darier's disease gene.

This example illustrates that, despite promising results and a fairly clear phenotype (although it may well be that a bipolar case is diagnosed very late because a manic episode occasionally only becomes clinically apparent in later life), we are still far away from the identification of susceptibility genes for bipolar disorder. One conclusion to be drawn is that there are likely to be numerous susceptibility genes that influence the risk of unipolar or bipolar depression. Supported by progress in biotechnology, critical regions for each susceptibility gene will be narrowed down and candidate genes will be tested much faster than in the past. These opportunities, together with new techniques allowing the dissection of gene–gene and gene–environment interactions, will ultimately allow the identification of susceptibility genes. The future role of the clinician scientist will be to improve diagnosis-independent phenotyping and to conduct studies focused on the elucidation of potential pathophysiological roles of candidate genes (see Mental Illness, Genetics of).

Genetics

There is substantial evidence from a number of family, twin, and adoption studies for a genetic transmission of alcoholism. The risk for alcoholism is significantly increased in first-degree relatives of alcoholics. Some adoption studies have shown an up to four times increased risk for alcoholism for sons of alcoholics even if they were raised apart from their biological parents. Although the heritability of alcoholism is a topic of numerous biological and genetic studies on the genetic and molecular/biological level, no vulnerability marker or gene for alcoholism is definitely identified yet. It seems most likely that alcoholism is not transmitted by a single but a number of genes. Alcoholism appears to be a polygenic disorder. Also no definite genetic marker for alcoholism is found yet. Numerous genome-wide scans to identify genes mediating the risk for alcoholism have been initiated in recent years, with conflicting results. The most robust findings come from chromosomes encoding for alcohol-metabolizing enzymes, especially genetic variations of ADH and ALDH. Other candidate genes are those mediating the pharmacogenetic response to alcohol, such as mu opioid or gamma-aminobutyric acid (GABA) receptor genes.

There are marked differences not only in alcohol metabolism but also in tolerance. Experimental and follow-up studies have shown that high-risk individuals (children of alcoholic parents) usually tolerate alcohol much better than other individuals. This in part explains the increased risk for alcoholism.

2.4 Sex Differences in the Expression of Impulsivity

Although impulsivity is evident in both men and women, the phenotypic expression might differ. In the Stockholm adoption study, it was demonstrated that there are at least two types of alcoholism (Cloninger et al. 1981). Type I alcoholism was observed in both men and women. Type II alcoholism was only observed in males. It was characterized by early onset and an increased frequency of social difficulties. When the clinical characteristics were defined (von Knorring et al. 1985), it was found that Type II alcoholism was strongly related to impulsivity. It was previously found that females with a genetic predisposition for Type II alcoholism had no increased risk of alcoholism or criminality (von Knorring 1983). Instead, they had an increased risk of somatization disorder. Later, von Knorring et al. (1986) demonstrated a similar biological basis for males with Type II alcoholism and females with somatization disorder. Furthermore, it is well known that the personality disorders in the DSM system related to impulsivity have pronounced sex differences (Ekselius et al. 1996). Thus, APD is much more common among males, while borderline personality disorder is more common among females.

4 Genetic Factors

Genetic factors seem to play a role in the development in anxiety disorders in humans. In familial studies, twin studies and adoption studies show genetic variances of 20 to 40 percent. The same was found for questionnaire variables such as neuroticism. Fear conditioning ability and predisposition to learn fear responses were extensively studied in animals. In rodents, fear processes or anxiety measured in the open field test or avoidance conditioning have revealed a strong genetic basis explaining most of the interindividual variance. However, the genes responsible for that trait have not yet been identified.

The situation is different for antisocial personality disorder and criminal behavior. Assuming that a deficit in anticipatory fear conditioning is the central symptom from which all consequent social problems follow, all evidence ranging from twin to adoption studies and gene sequencing suggests a moderate to strong genetic influence. If sensitivity to alcoholism and novelty-seeking are included as comorbidity factors the genetic variance increases further. Whether this is related to genetic fear conditioning remains to be seen (Plomin et al. 2001).

Genetic and Epigenetic Factors

Genetic factors seem to play a role in the development in anxiety disorders in humans. Familial studies, twin studies, and adoption studies show genetic variances of 20–40%. The same was found for fear conditioning. Fear conditioning ability and predisposition to learn fear responses were extensively studied in animals. In rodents, fear processes or anxiety measured in the open field test or avoidance conditioning have revealed a strong genetic basis explaining most of the interindividual variance. However, the genes responsible for that trait have only partially been identified. Important genes that modulate fear conditioning involve the serotonin transporter (5HTT) and the catechol-O-methyltransferase (COMT) gene. Moreover, genes involved in the regulation of the hypothalamus–pituitary–adrenal axis such as the glucocorticoid receptor gene or the corticotropin-releasing hormone genes, but also the dopamine transporter (DAT1) and the pituitary adenylate cyclase 1 receptor (ADCYAP1R1) genes have been implicated (Lonsdorf and Kalisch, 2011). Controversial data exist for the brain-derived neurotrophic factor (BDNF) gene. There is also mounting evidence that epigenetic mechanisms such as DNA methylation and histone modification contribute to fear conditioning (Zovkic and Sweatt, 2013). For example, histone acetylation and DNA methylation were found to be crucial for fear memory consolidation and synaptic plasticity in the lateral amygdala. Dias and Ressler (2014) showed that prenatal fear conditioning odors led to enhanced startle responding, neuroanatomical changes, and hypomethylation of an odor receptor encoding gene in the offspring of the trained animals, thus explaining transgenerational effects in fear learning.

Molecular Genetic Studies

Once a solid foundation of support for significant genetic influence on a trait has been built by means of behavior genetics (family, twin, and adoption studies), as has been the case for sexual orientation especially with males, molecular genetic studies are a next logical step. The two primary varieties of these studies are association and linkage designs.

Association studies are based on linkage disequilibrium. This means that a gene variant influencing a trait was initially associated with specific alleles of nearby polymorphic loci. As generations (and the meioses that produce sperms and eggs) pass, the trait-influencing gene and marker allele may remain statistically associated because their proximity reduces the number of recombinations or crossing-over events that occur between them. An advantage of association tests is that the chromosomal region examined is usually much smaller than the region examined by testing for linkage in families. Association is often more powerful than linkage in that a valid association may be detected in a sample when linkage is not detectable, even when the gene is playing only a modest role. Most association studies in the past were the population-based type where the allele frequencies of a group of unrelated cases were compared against those of a group of unrelated controls, and this is the only type of association study that has been carried out on male sexual orientation. A potential pitfall of population-based case–control studies is that, although some populations appear homogeneous to superficial examination, they are in reality composed of different ancestral human groups, each one potentially with a different allele distribution at the studied loci. If one or more such groups are represented in a largely different proportion in one of the samples of an association data set (i.e., either in the controls or in the cases), false-negative or false-positive association findings may easily arise due to methodological artifact.

Linkage analysis exploits the key biological phenomenon during generation of sperm and eggs of meiotic recombination, or crossing over, during which both the maternally and paternally derived chromosomes lie in close proximity and undergo exchange of genetic material between the homologous chromosomes, e.g., between a paternally derived chromosome and its maternally derived counterpart. The chance of crossing over between two loci (locations on a chromosome) is referred to as the recombination fraction. Genes and other genetic markers (DNA sequence variations known as polymorphisms) that are close together are less likely to be separated by this process than are those that are farther apart. Therefore, they are usually inherited together by the progeny cells, and are considered genetically linked. Due largely to the complexity of the genetic contributions to male sexual orientation and uncertainty regarding key parameters (mode of inheritance, number of relevant genes, etc.), the type of linkage analyses preferred are nonparametric allele-sharing methods, and more specifically, the affected sibling pair (ASP) method. ASP designs measure the frequency with which a genetic marker allele (or variant) is inherited from a particular parent (referred to as IBD, meaning identical by descent) in a pair of siblings both manifesting the trait. The presence of a trait-influencing gene is revealed when the IBD allele sharing between affected siblings exceeds the expected 50%.