Schizophrenia is a relatively common mental disorder that has a life-time prevalence of approximately 1% in the general population (Kleinman,
1988; Thakker & Ward, 1998). The disorder has been defined in various ways since it was first described by Bleuler and Kraepelin in the early 1900s (Kraepelin, 1904; Andreasen & Carpenter, 1993). At present, the most widely used definition of schizophrenia comes from the 4th edition of the Diagnostic and Statistical Manual of Mental Disorders (DSM-IV, 1994). According to the DSM-IV, schizophrenia is defined by the presence of two or more of the following symptoms for a significant portion of time during a 1-month period: delusions, hallucinations, disorganized speech, grossly disorganized or catatonic behavior, or neg
ative symptoms, that is, affective flattening, alo-gia, or avolition. Research has found that this complex mental disorder is associated with a wide variety of both genetic and environmental risk factors (Cannon, Kaprio, Lonnqvist, Huttunen, & Koskenvuo, 1998; for reviews, see Van Os & Mar-celis, 1998; McGuffin & Owen, 1996; Syvalahti, 1994; and Egan & Weinberger, 1997). Several different quantitative genetic models have been proposed to describe the etiology of schizophrenia (e.g., Slater, 1958; Elston, Namboodiri, Spence, & Rainer, 1978; Kidd & Cavalli-Sforza, 1973; Mat-thysse & Kidd, 1976; Gottesman & Shields, 1967). Behavioral genetic techniques can be used to determine the relative contribution of genetic and environmental influences on the development of schizophrenia.
One of the first things that behavioral geneticists noted in their study of schizophrenia is that the disorder tends to aggregate in families. Indeed, Kraepelin himself emphasized the importance of familial factors in his description of dementia praecox (Kraepelin, 1904). Family studies have indicated that the lifetime prevalence of schizophrenia in close relatives of schizophrenic patients is much higher than the population base rate of 1% noted before. Eight to ten percent of full siblings and 12 to 15% of the offspring of schizophrenic patients will be diagnosed as schizophrenic at some point during their lifetimes (Gottesman, 1993). Although these findings suggest that genetic factors may be involved in the etiology of schizophrenia, it is not possible to differentiate between genetic or environmental influences using data from family studies. These influences can be distinguished by using twin studies, many of which have been conducted over the years.
The twin study of schizophrenia conducted by Franz Kallmann is one of the largest twin studies of schizophrenia. As of 1950, Kallmann had registered a total of 953 schizophrenic index twin cases (Kallmann, 1950; Shields, Gottesman, & Slater, 1967). Four years before this, Kallmann published a more detailed report on 691 twin pairs which has received a great deal of scientific scrutiny over the years (Kallmann, 1946; see Jackson, 1960, and Rosenthal, 1962, for criticisms of Kallmann, 1946). Some of the criticisms that have been leveled against Kallmann's 1946 report have been addressed in subsequent reanalyses of Kallmann's original data (Kety, 1958; Shields et al., 1967). According to Shields et al.'s relatively conserva tive reanalysis of Kallmann's data, 50% of his MZ twins had both received definite diagnoses of schizophrenia, whereas only 6% of DZ twins were concordant for schizophrenia. It must be noted that these pairwise concordance rates may be artificially inflated due to the manner in which the schizophrenic co-twins were ascertained (see Shields et al., 1967, for practical reasons and Smith, 1974, for statistical reasons why this may be so). Indeed, the concordance rates reported by other investigators tend to be lower than Kall-mann's concordance rates (Farmer, McGuffin, & Gottesman, 1987; Kendler & Robinette, 1983; Mo-skalenko & Gindilis, 1981; Onstad, Skre, Torger-sen, & Kringlen, 1991; Torrey, 1992; for reviews, see Gottesman, 1996; Portin & Alanen, 1997; Pulver et al., 1996; Tsuang, 1998; Van Os & Marcelis, 1998). Nevertheless, the concordance rates in MZ twins are higher than concordance rates for DZ twins in every twin study of schizophrenia conducted to date, thus providing strong evidence for the contention that genetic factors contribute to the development of this disorder. In addition, the fact that the concordance rate for MZ twins is far less than 100% suggests that nongenetic (i.e., environmental) factors play an important role in the etiology of schizophrenia.
One difficulty with the early twin studies reviewed before is that most of them used pairwise concordances to summarize the similarity between MZ and DZ twins. Probandwise concordances are superior to pairwise concordances in several ways (McGue, 1992). These include (1) characterizing risk for disorder at the level of the individual, rather than at the level of the twin pair, thus facilitating the comparison of risk to a twin with the risks to other types of relatives; (2) the similarity of probandwise concordances to concordances among other relatives (e.g., parents and their offspring) and to the prevalence of the disorder, whereas pairwise concordances have no direct analogue for non-twin relative pairs and are not directly comparable to the prevalence of the disorder. This renders the interpretation of pair-wise concordances and the comparison of twin and non-twin resemblance difficult at best; (3) a clear and useful statistical basis for the interpretation of probandwise concordances (viz., the probability that a twin will be affected given that their co-twin is affected), which is lacking for pairwise concordances; and (4) valid estimation of the magnitude of twin resemblance under both complete and incomplete ascertainment, and hence comparability across studies regardless of the thoroughness of ascertainment. In contrast, sample pair-wise concordance rates vary as a function of the thoroughness of ascertainment and thus are not comparable across studies unless the studies have similar levels of ascertainment. In addition, Heath (personal communication, 1994) has pointed out that the analysis of tetrachoric correlations estimated from MZ and DZ concordances and population base rates yield more valid inferences regarding the magnitude of genetic and environmental influences than the concordances themselves, given that the interpretation of concordances and their differences depend heavily on the base rate of the disorder.
Other behavioral genetic studies have provided insight into the question of whether a single gene or multiple genes are responsible for the increased liability to develop schizophrenia. For example, a twin study conducted by Kendler and Robinette (1983) compared the single-locus, two-allele model of the transmission of schizophrenia with the poly-genic, multifactorial threshold model (Falconer, 1965; Smith, 1974). The single locus, two-allele model assumes that schizophrenia is due to the effects of a single gene of major effect and that one allele confers a heightened risk for schizophrenia, whereas the other does not (Elston et al., 1978; Kruger, 1973; Slater, 1958). The multifactorial polygenic threshold model, on the other hand, posits that there are multiple genes each of which confers a small degree of risk of developing schizophrenia. According to the multifactorial threshold model, schizophrenia develops when the total liability toward schizophrenia due to the combined genetic and environmental risk factors exceeds a certain threshold. Kendler and Robi-nette (1983) were able to show that results for schizophrenia in their twin registry were not compatible with a fully penetrant, single-locus, two-allele model of schizophrenia with no pheno-copies, nor were they compatible with the more flexible, generalized single-locus model developed by Kidd and Cavalli-Sforza (1973) that allows for incomplete penetrance and the presence of phenocopies. In contrast, their results were consistent with the multifactorial threshold model of the inheritance of schizophrenia. As Kendler and Robinette note, this superiority of the polygenic model to monogenic models is consistent with a more extensive model-fitting study conducted by
O'Rourke, Gottesman, Suarez, Rice, and Reich (1982). Making an assumption of polygenic transmission of schizophrenia allows behavioral geneticists to estimate the broad heritability of schizophrenia (Falconer, 1965). In the Kendler and Robinette study, for example, the estimated h2 = .91 ± 0.16%. This value for h2 is within the range obtained from other twin studies. Values for broad-sense heritability estimated from the concordance rates obtained in six studies cited in Gottesman's review of twin studies of schizophrenia (1991) vary between .52 and .92.
Adoption studies provide another way of disentangling genetic and environmental influences. If the incidence of schizophrenia in the adopted-away offspring of schizophrenic individuals raised by normal parents is higher than that of the adopted-away offspring of nonschizophrenic individuals, this provides strong evidence that genetic influences are involved in the etiology of schizophrenia. Indeed, this has been the case across schizophrenia adoption studies. For example, in the Danish Adoption Study, Kendler, Gruenberg, and Kinney (1994) compared the prevalence of DSM-III diagnoses of schizophrenia in the relatives of schizophrenic and control adoptees. The first-degree biological relatives of schizophrenic adoptees showed an increased risk for schizophrenia compared with the relatives of control adoptees (7.9% vs. 0.9%). This was also true if all interviewed biological relatives were taken into consideration (3.3% vs. 0.3%). The difference in prevalence was larger if the entire schizophrenia spectrum of disorders was considered (23.7% vs. 4.7% in first-degree relatives and 13.0% vs. 3.0% in all interviewed biological relatives).
Another useful method for disentangling genetic and environmental influences on schizophrenia is to compare the rates of schizophrenia in the offspring of MZ twins discordant for schizophrenia. If schizophrenia is primarily due to environmental factors, one would predict that the prevalence of schizophrenia should be considerably higher in the offspring of the schizophrenic twin than in the offspring of the nonschizophrenic co-twin. In contrast, to the extent that schizophrenia is genetically influenced, rates of schizophrenia will be equally high in both sets of offspring. The results of studies using this method are varied. In a follow-up study of Fischer's (1971) Danish twin sample, Gottesman and Bertelsen (1989) found no difference between the prevalence of schizo phrenia in the offspring of schizophrenic (16.8%, N = 47) and nonschizophrenic co-twins (17.4%, N = 24). The rates of schizophrenia in both sets of offspring are well above the prevalence in European populations, which is approximately 1% (Kleinman, 1988; Thakker & Ward, 1998). In a somewhat smaller study using the Norwegian national twin sample, Kringlen and Cramer (1989) found that, whereas 17.9% of the children of schizophrenic twins were later diagnosed with schizophrenia, only 4.4% of the children of non-schizophrenic twins developed schizophrenia. Although this difference does not quite reach statistical significance at the .05 level, the trend for the children of schizophrenic twins who have higher rates of schizophrenia does imply that familial environmental factors may significantly influence the development of schizophrenia in addition to genetic influences.
Considering the results of the disparate behavior genetic designs reviewed before, the overwhelming majority of the evidence indicates that genetic influences play a substantial role in the development of schizophrenia. Indeed, behavioral geneticists have progressed from asking the question of whether genetic factors are important in the etiology of schizophrenia to the question of which model of genetic transmission most accurately describes the inheritance of this disorder (Kendler and Robinette, 1983; McGue, Gottesman, & Rao, 1985; O'Rourke et al., 1982). In general, results of these model-fitting studies favor the multifactorial polygenic threshold model over other models of genetic transmission.
Was this article helpful?