The Genetics of Schizophrenia: A Current, Genetic-Epidemiologic Perspective

VOL. 19, NO.
2, 1993
The Genetics of Schizophrenia: A Current, Genetic-Epidemiologic Perspective

Abstract In the "Special Report on Schizophrenia" published in the Schizophrenia Bulletin in 1987, the genetic basis of schizophrenia was reviewed. Here, we provide our perspective on the current status of this area of investigation, focusing largely but not exclusively on recent findings. Methodologically rigorous family studies have now clearly shown that schizophrenia substantially aggregates in families. Familial factors that predispose to schizophrenia also increase the risk for certain schizophreniarelated personality disorders and probably for some forms of nonschizophrenic nonaffective psychosis. Results from one new twin study and updates from two ongoing adoption studies continue to support the hypothesis that genetic factors play a major role in the etiology of schizophrenia. Little is known about how genetic liability to schizophrenia is transmitted, although statistical models suggest that transmission is probably not due solely to a single major gene. Schizophrenia is clearly a complex disorder in that gene carriers need not manifest the illness (incomplete penetrance), affected individuals need not have the gene (environmental forms or phenocopies), diagnostic uncertainties cannot be avoided, and different families may carry different susceptibility genes (genetic heterogeneity). Therefore, segregation or linkage analyses are far more difficult to perform with schizophrenia than with Mendelian genetic disorders. Given this complexity, it is not too surprising that no replicated positive evidence for linkage to
261
by Kenneth S. Kendler and Scott R. Dlehl
schizophrenia has been reported to date. However, just as linkage analysis of schizophrenia should not be excessively embraced as the only form of viable genetic research in schizophrenia, it also shouldn't be prematurely spurned. If one or several genes of major effect exist for schizophrenia, large samples using new statistical and laboratory methodologies have a good chance of detecting them. The authors thus recommend a balanced research approach to the genetics of schizophrenia that includes traditional methods of family, twin, and adoption studies as well as a major effort in large-sample linkage studies. The status of the genetics of schizophrenia was thoroughly reviewed in 1987 in the "Special Report on Schizophrenia" (Gottesman et al. 1987). In this review, we provide our perspective on the current status of this research field, focusing largely but not exclusively on recent findings. (For other recent perspectives on this subject, see Weeks et al. 1990; Gottesman 1991.) It has been an eventful time for the field. Center stage has shifted from traditional family, twin, and adoption studies to linkage analyses of "high-density" families. The focus of scientific inquiry has moved from determining the magnitude of familial aggregation, disentangling the effects of genetic and shared environmental factors, and understanding the boundaries
Downloaded from http://schizophreniabulletin.oxfordjournals.org/ at Pontificia Universidad Catolica de Chile on October 29, 2012
Reprint requests should be sent to Dr. K.S. Kendler, Dept. of Psychiatry, Box 710, Medical College of Virginia, Richmond, VA 23298-0710.
262
SCHIZOPHRENIA BULLETIN
of the schizophrenia spectrum toward attempts to locate specific susceptibility genes. This is not a new pursuit. In establishing the first modem psychiatric research institute in Munich early in this century, Kraepelin recruited the young psychiatrist/geneticist Ernst Rudin with the specific goal of uncovering the Mendelian basis of schizophrenia. When one of us (KSK) visited with Manfred Bleuler in 1986, Bleuler, then in his early eighties, appeared at the door with an old display case of butterflies. His high school science project had been performing simple crosses and backcrosses in butterflies to uncover the Mendelian basis of wing patterns. He said then that it had become his lifelong dream to discover the Mendelian basis of schizophrenia. The psychiatric genetics community has responded in three ways to the rapidly increasing power of human gene-mapping methods and their astounding successes with classic Mendelian disorders. Some have concluded that these techniques are irrelevant, as the genetic and etiologic complexity of schizophrenia renders such a simple approach useless. At the opposite extreme, others have grasped the world of human gene mapping with unbounded enthusiasm. Overnight, traditional psychiatric genetic methods such as family, twin, and adoption studies have been dismissed as antiquated, no longer "cutting edge," and ready for the dustbin of history. Avoiding either extreme, a third response has been that, while very exciting and holding out the promise of a true revolution in our understanding of the etiology of schizophrenia, gene-mapping approaches continue to represent
but one among many potentially useful techniques in psychiatric genetics. Our long-held view, reflected in this article, is the third response (Kendler 1987). Given the complexity of psychiatric disorders and the slow nature of progress to date, impatience is understandable. The potential power of human gene mapping is therefore seductive, encouraging investigators to focus exclusively on these glamorous new technologies. However, as we hope to demonstrate, the progress we continue to make in our understanding of the etiology of schizophrenia with the more traditional psychiatric genetic methods is substantial, even if it is not revolutionary. Thus, a balanced research plan for this devastating illness should include both traditional and gene-mapping approaches.
Family Studies The first question in the genetic epidemiology of schizophrenia is the degree to which the disorder aggregates (or "runs") in families. Historically, studies of this key issue can be divided approximately into two periods. In the first period, as summarized by ZerbinRudin (1967), studies were nearly all performed nonblind, without control groups, and without the use of structured psychiatric assessments or operationalized diagnostic criteria. In many of these studies, it is unclear how many relatives were seen individually rather than evaluated from secondary sources (e.g., hospital or parish records, reports from relatives). Although many of these studies were performed by thorough and careful researchers, their findings that schizophrenia consistently and
substantially aggregates in familieswere open to methodological criticism. In the early 1980s, two research groups (Pope et al. 1982; Abrams and Taylor 1983) questioned the validity of these studies and argued that the previous evidence for the familial aggregation of schizophrenia arose from methodological deficiencies, particularly an overly broad diagnostic approach to the disorder. When schizophrenia was "narrowly" diagnosed, these groups argued, evidence of familial aggregation might be weak or absent. We are, in 1993, in a good position to evaluate these claims. Since 1980, a substantial number of second-generation family studies of schizophrenia have been completed. These share three key methodological features: (1) a normal control group, (2) structured psychiatric assessment and operationalized diagnostic criteria, and (3) blind assessment and diagnosis. We judged seven studies as meeting these criteria (table 1). Three other valuable studies (Scharfetter and Niisperli 1980; Guze et al. 1983; Onstad et al. 1991a) were eliminated because the only control group consisted of nonschizophrenia psychiatric patients; a fourth (Tsuang et al. 1980) was eliminated because it used a consensus diagnostic procedure not based on a single system of operationalized diagnostic criteria. These seven selected studies were performed in the United States (Baron et al. 1985; Kendler et al. 1985a; Coryell and Zimmerman 1988; Gershon et al. 1988), Greece (Frangos et al. 1985), Germany (Maier et al. 1990), and Ireland (Kendler et al., in press a), with the total number of assessed relatives ranging from 232 (Coryell
Table 1. Summary results for the familial aggregation of schizophrenia in major recent family studies of schizophrenia that included a control group, personal interviews with relatives, and blind diagnosis of relatives
O
ro
First-degree relatives of schizophrenia probands Schizophrenia Schizophrenia BZ n P 0.0001 0.0001 8x10-8 0.37 0.04 0.28 0.04 0.38 0.03 2 0.6 1.1 0.2 6 2 337 536 931 MR % Correlation of llabllltySE Controls BZ 329 5.8 5.4 3.7 478 703 26 26 19 n Normals Normals Screened surgical patients Never-ill volunteers 72 160 0 0 1 Volunteer controls 97 3.1 2 349 3 Never-ill matched controls 463 23 5.0 Unscreened matched controls 276 6.5 428 18 2 294 1 0.6 1.4 NS MR %
First-degree relatives of control probands
First author/ year
Diagnostic criteria
Baron et al. (1985)
RDC DSM-III
Frangos et al. 1 (1985)
DSM-III
Kendler* (1985)
DSM-III
Coryell and Zimmerman (1988)
RDC
0.23 0 . 1 2
Gershon et al. (1988)
RDC chronic schizophrenia
0.038
0.25 0.09
Maier et al. (1990)
RDC
0.3
0.0004
0.38 0.03
Note.BZ o bezugsziffer (total lifetime equivalents of risk); MR = morbid risk; SE = standard error; RDC = Research Diagnostic Criteria (Spitzer et al. 1978); DSM-JII, DSM-///-R o Diagnostic and Statistical Manual of Mental Disorders (American Psychiatric Association 1980, 1987); NS = not significant. 'Includes definite and probable/possible cases. 2 Results Include relatives with only hospital records, based on the original study by Tsuang et al. 1980.
nloaded from http://schizophreniabulletin.oxfordjournals.org/ at Pontificia Universidad Catolica de Chile on October 29, 2012
Kendler et al. 1 (in press a)
DSM-III-R
0.5
0.000004
0.41 0.04
ro
CO
264
and Zimmerman 1988) to 1,634 (Kendler et al. 1985a). Studies used either Research Diagnostic Criteria (RDC; Spitzer et al. 1978) (Coryell and Zimmerman 1988; Maier et al. 1990), modified RDC (Baron et al. 1985; Gershon et al. 1988), DSMIII (American Psychiatric Association 1980) (Frangos et al. 1985; Kendler et al. 1985<J), or DSM-IIIR (American Psychiatric Association 1987) criteria (Kendler et al., in press a). Diagnostically, their results should be broadly comparable, except that the two studies using unmodified RDC criteria (Coryell and Zimmerman 1988; Maier et al. 1990) allow for cases with a duration of illness of less than 6 months. The risk of schizophrenia in relatives of control probands ranged from 0 to 1.1 percent in the seven studies. A x2 analysis suggests that the variation in these results could easily be explained by random statistical effects (x2 = 6.79, df = 6, p = 0.34). It therefore might be appropriate to sum the results across these studies, producing 15 cases of schizophrenia in 3,035 lifetimes at risk for an estimated morbid risk for schizophrenia of 0.5 percent in relatives of controls. The risk of schizophrenia in the aggregate first-degree relatives of schizophrenia probands was much more variable, ranging from 1.4 to 6.5 percent. However, the results of these seven studies again do not significantly differ from one another (x2 = 7.25, df = 6, p = 0.30). Compared with the risk in the aggregate relatives, the risk of schizophrenia in relatives of schizophrenia probands appears somewhat too low in both studies from Iowa (Kendler et al. 1985a; Coryell and Zimmerman 1988) and somewhat too high in the study from Ireland (Kendler et al., in
press a). Summing the results of all seven studies yields 116 cases of schizophrenia in 2,418 lifetimes at risk, or a morbid risk for schizophrenia of 4.8 percent in the first-degree relatives of schizophrenia probands. Table 1 also includes the correlation of liability (Falconer 1965) for schizophrenia calculated for each of these seven family studies. If schizophrenia is due to several genetic and environmental factors that act somewhat additively to influence an individual's liability, this statistic approximately represents the degree of correlation between first-degree relatives for this overall liability to schizophrenia. This statistic, which is calculated only from the risk figures for schizophrenia in relatives of schizophrenia and control probands, is particularly useful because it provides a single estimate of the magnitude of familial aggregation that can be compared across studies. In addition, if familial aggregation is due mostly to genes that are largely additive in their effect, the correlation of liability in first-degree relatives should be approximately half the heritability of liability. The heritability of a disorder, in turn, is an estimate of the proportion of variation in the risk of the disorder in a population that is due to genetic factors. The correlations of liability for schizophrenia in first-degree relatives in these studies range from +0.23 to +0.41. The 95 percent confidence intervals for these estimates nearly all overlap. Summing across these studies, we calculate a best estimate of the correlation of liability for schizophrenia to be +0.35 0.02. Three major conclusions can be drawn from this brief review.
First, the earlier first-generation family studies were substantially correct in concluding that schizophrenia strongly aggregates in families. Their findings were probably not systematically biased by nonblind or unstandardized diagnostic practices, the absence of control groups, or an overly broad classification of the disease. Second, these seven later studies suggest that the degree of familial aggregation of schizophrenia is substantial. In aggregate, the risk for schizophrenia in a first-degree relative of an individual with schizophrenia is 9.7 times as great as that for a relative of a matched control. This figure is strikingly similar to the tenfold increased risk commonly quoted in older textbooks of psychiatry. Third, we have no convincing evidence that the familial aggregation of schizophrenia differs significantly across samples. Although it may be somewhat lower than expected in Iowa and higher than expected in Ireland, these results could be due to chance fluctuations.
Twin Studies The last 10 years have seen relatively little twin research in schizophrenia. Since the previous review (Gortesman et al. 1987), results have been published of one new twin study of schizophrenia, performed in Norway by Onstad and colleagues (1991b). This study has three important methodological strengths. First, the twin sample was obtained from the high-quality Norwegian twin and psychiatric registries and thus should be representative of all treated cases of illness. Second, psychiatric assessment was performed using structured instruments. Third, diagnoses
VOL. 19, NO. 2, 1993
265
were all performed using DSMIH-R operationalized criteria. The study also has three methodological weaknesses. First, the sample size, which included only 52 samesex twin pairs, was small. Second, interviews and diagnoses were performed nonblind. Third, zygosities were based only on selfreport measures. Probandwise concordance for schizophrenia in monozygotic (MZ) and same-sex dizygotic (DZ) twins was 15/31 (48.4%) and 1/28 (3.6%), respectively. Assuming a population risk for schizophrenia of 0.8 percent, the correlation in liability in MZ and DZ twins can be calculated at +0.87 0.08 and +0.24 0.16, respectively. Because the correlation in MZ twins substantially exceeds twice that in DZ twins, the best estimate of broad heritability of liability from this sample is probably the MZ correlation itself. This result is toward the upper end of the estimates obtained from previous twin studies of schizophrenia (Kendler 1983). Overall, the results from this study are reassuringly similar to those obtained from previous studies (Gottesman and Shields 1966; Kendler 1983) in suggesting that genetic factors play a major role in the etiology of schizophrenia. This report is also consistent with previous twin studies in suggesting that familial environment makes little or no contribution to the liability for schizophrenia (Kendler 1983). Adoption Studies Updated findings from two major adoption studies of schizophrenia have appeared since the last review. Preliminary results of a replication of the Copenhagen Adoption Study of Kety and colleagues
have been presented (Kety 1987). In the Copenhagen study, 34 adoptees were identified who the investigators considered to have chronic, acute, or latent schizophrenia. Their biologic and adoptive relatives were assessed through hospital records (Kety et al. 1968) and later by personal interview (Kety et al. 1975). Since then, the same research team examined adoptees outside of Copenhagen (the so-called Provincial sample) and identified 42 other adoptees with chronic, acute, or latent schizophrenia. As with the initial reports from the Copenhagen sample (Kety et al. 1968), hospital records have been searched, blinded, and diagnostically reviewed for the biologic and adoptive relatives of these index adoptees and of a matched group of control adoptees. Both chronic schizophrenia and latent and uncertain schizophrenia were found to be significantly more common in the biologic relatives of index adoptees than in the biologic relatives of control adoptees. Furthermore, the rates of these disorders were low and not different in the adoptive relatives of both groups of adoptees (Kety 1987). Preliminary results are also now available from a blind diagnostic review by Kety and colleagues of personal interviews with both the adoptees and the biologic and adoptive relatives from the Provincial sample (Kety et al., in preparation). The researchers found schizophrenia to be significantly more common in the biologic relatives of all schizophrenia adoptees (4.1%) than in the biologic relatives of all control adoptees (0.5%) (p = 0.01 by one-tailed Fisher's exact test). By contrast, latent and uncertain schizophrenia was not
found to be significantly more common in the biologic relatives of the schizophrenia adoptees than in those of the control adoptees (6.5% vs. 5.5%, respectively). However, if the schizophrenia adoptees were screened to include only those with confirmed chronic schizophrenia and the control adoptees were screened to eliminate any cases with major mental illness, latent schizophrenia would have been significantly more common in the biologic relatives of the former group than in those of the latter group (8.2% vs. 2.4%, p = 0.025). Rates of schizophrenia and related disorders were very low and equal in the two groups of adoptive relatives. The largest study of adoptedaway offspring of schizophrenia and control mothers is being conducted in Finland under the direction of Tienari (1991). As of April 1991, 361 adoptive families containing the adopted-away offspring of a schizophrenia or control mother had been contacted for field study. Results were available for 144 offspring of schizophrenia mothers and 178 offspring of control mothers. To date, 15 psychotic adoptees have been ascertained, of whom 13 are offspring of schizophrenia mothers (13/144 = 9.1%) and 2 are offspring of control mothers (2/178 = 1.1%). This is a highly significant difference (x2 = 11.20, df = 1, p < 0.000). Of the 13 psychotic adopted-away offspring of schizophrenia mothers, by DSM-HI-R criteria, 7 had schizophrenia, 2 had schizophreniform disorder, 2 had delusional disorder, and 2 had psychotic bipolar illness. Both of the psychotic offspring of control mothers had schizophrenia. Therefore, the prevalence of schizophrenia is also significantly greater
266
in offspring of index mothers (7/144 = 4.9%) than in offspring of control mothers (2/178 = 1.1%) (X2 = 4.09, df = 1, p < 0.04). This adoption study has also examined the role of familialenvironmental factors in the etiology of schizophrenia and found a substantial correlation between the functioning of the adoptive family and the psychiatric outcome in the adoptee. Of particular interest, this relationship was greatest when the adoptee was an offspring of a schizophrenia mother. These results are consistent with a model in which schizophrenia emerges when genetically susceptible individuals experience disruptive family environments. However, most of the families were evaluated when the adoptees were well into adult life. Thus, this observed correlation between adoptee and adoptive family functioning could also be due to disturbed adoptees creating disturbed families. A prospective study is now under way in a subset of this unique cohort with the hope of further clarifying this central issue. Summary of Family, Twin, and Adoption Studies Family studies suggest that schizophrenia strongly aggregates in families. Twin and adoption studies continue to suggest that genetic factors account for most of this familial aggregation, although reports from the Finnish adoption study could suggest a significant role for environmental factors in vulnerable individuals. Previous estimates of the heritability of liability to schizophrenia, based on either twin studies alone (Kendler 1983) or twin and family studies (not including any of the family studies reviewed here [McGue et al.
1983]), range from 0.63 to 0.68, and the most recent twin study (Onstad et al. 1991b) suggests a heritability moderately higher than this. If, as indicated by nearly all twin and adoption studies, most of the familial aggregation of schizophrenia is caused by genetic factors, heritability should be about twice the correlation of liability in first-degree relatives. The family study data agree closely with this prediction. The correlation in liability from the recent methodologically strong family studies (around +0.35) is about half of the aggregate estimate of heritability obtained from twin studies. These conclusions, which apply to general populations of individuals with schizophrenia and not to individual cases, are not inconsistent with the hypothesis that in some individuals, schizophrenia is largely environmental in origin, while in others, the disorder is caused largely by genetic factors. Boundaries of the "Schizophrenia Spectrum" An accurate definition of the affected phenotype is crucial in any genetic investigation. A major area of interest in the genetics of schizophrenia has therefore been to use genetically informative designs to determine what psychiatric syndromes reflect the familial liability to schizophrenia. Such an endeavor is also of interest because it can provide insights into the nature of that familial liability and hence into the etiology of schizophrenia itself. We have found it helpful here to articulate four hypotheses for the nature of the transmitted liability to schizophrenia: (1) liability only to typical schizophrenia, (2) liability to schizophrenia and schizophrenia-
like personality disorders, (3) liability to all nonaffecrive psychosis, and (4) broad liability to all psychiatric illness (Kendler 1988). Each of these hypotheses predicts a distinct pattern of illness in relatives of schizophrenia versus control probands. We will not review in this article the important attempts to define the nature of the transmitted liability to schizophrenia using nonclinical measures such as attention, smooth-pursuit eye movements, or evoked potentials. Schizophrenia-Like Personality Disorders. Many clinicians and researchers over the last 100 years have noted that some close relatives of patients with schizophrenia, though never psychotic, had odd or eccentric personalities that were clinically similar to schizophrenia (Kendler 1985). This has led to the second hypothesis articulated abovethat the familial liability to schizophrenia is expressed, at least in part, as a predisposition to a set of schizophrenia-like or schizotypal personality traits. Since this hypothesis was first put to a rigorous test by Kety and colleagues in the Danish adoption studies of schizophrenia (Kety et al. 1968, 1975), work in this area has been strongly influenced by the development in the DSM-1I1 of operationalized diagnostic criteria for schizotypal personality disorder (SPD; Spitzer et al. 1979) and by its elaboration in the DSM-III-R. Operationalized diagnostic criteria were also proposed for other putative schizophrenia-related personality disorders, especially paranoid personality disorder (PPD). Seven family or adoption studies using personal interviews, DSM-III or DSM-III-R criteria, and blind
VOL. 19, NO. 2, 1993
267
diagnoses have examined the lifetime prevalence or risk for SPD and/or PPD in first-degree biologic relatives of schizophrenia and matched control probands (table 2). Several observations on the results of these studies are of interest. First, the rates of SPD/PPD in relatives of patients with schizophrenia are widely variable and highly significantly different across studies (x2 = 101.31, df = 6, p < 0.000). A similar but somewhat less striking variation is also seen in rates in relatives of controls (x2 = 36.75, df = 6, p < 0.000). Rates of SPD/PPD are far more variable across studies than are rates of schizophrenia. Although this may be the result of true population differences, a more likely source of much of this variation is methodological. The mode of assessment of personality disorders varied widely in these studies and correlated with observed rates. The studies with the three lowest rates of SPD/PPD in relatives of schizophrenia probands used either unstructured assessments (Frangos et al. 1985) or general personality disorder interviews mostly done by phone (Coryell and Zimmerman 1988; Gershon et al. 1988). These methods may have low sensitivity for the detection of SPD/ PPD. Studies with the higher rates in relatives used instruments that focused specifically on schizophrenia-related personality traits (Lowing et al. 1983; Kendler and Gruenberg 1984; Baron et al. 1985; Kendler et al., in press b). The second important result of these seven studies is that they all found rates of SPD/PPD that were higher in relatives of schizophrenia probands than in relatives of control probands. With two exceptions (Lowing et al. 1983; Coryell and
Zimmerman 1988), both of which had small sample sizes, these studies produced differences that were statistically significant and, in the two largest sample studies (Baron et al. 1985; Kendler et al., in press b), very highly so. With the exception of one small-sample adoption study (Kendler and Gruenberg 1984), the risk for SPD/PPD in relatives of schizophrenia probands exceeded that found in relatives of control probands by twofold to fivefold. In aggregate, these results provide strong support for the second hypothesis noted above. The familial liability to schizophrenia can, at least in part, express itself as a predisposition to a certain set of schizophrenia-like personality traits. However, the specificity of these traits remains uncertain. Rates of SPD appear to be similarly elevated in offspring of parents with schizophrenia and with affective illness (Squires-Wheeler et al. 1988, 1989). Two studies have now addressed this important question in more typical adult samples. In a small sample of blindly interviewed first-degree relatives of twin probands, Onstad et al. (1991fl) found a significantly higher rate for SPD or PPD in relatives of schizophrenia probands (17/136 = 12.5%) than in those of mood disorder probands (1/46 = 2.2%) (X2 = 4.11, p < 0.04). In the Roscommon Family Study (Kendler et al., in press b), the rate of SPD or PPD in interviewed first-degree relatives of schizophrenia probands (26/319 = 8.2%) was highly significantly greater than that found in relatives of affective disorder probands (12/371 = 3.2%) (x2 = 7.97, df = 1, p < 0.005). More importantly, the rate of SPD or PPD in relatives of affective disorder probands did not significantly exceed
that found in relatives of controls (10/580 = 1.7%) (x2 = 2.28, df = 1, NS). Taken together, these results suggest that SPD and PPD may have some specificity in reflecting a familial vulnerability to schizophrenia rather than to all forms of serious psychopathology. Nonschizophrenic Nonaffective Psychosis. The third hypothesis outlined above (that the familial liability to schizophrenia predisposes to a broad range of psychotic conditions) predicts that all psychotic disorders should coaggregate with schizophrenia in families. All seven of the major recent controlled family studies of schizophrenia reviewed above have addressed this issue to one degree or another (table 3). Consistent with this hypothesis, five of these studies (Baron et al. 1985; Kendler et al. 1985a; Gershon et al. 1988; Maier et al. 1990; Kendler et al., in press d) found the risk for nonschizophrenic nonaffective psychosis (e.g., schizoaffective disorder, schizophreniform disorder, delusional disorder, and atypical psychosis) to be substantially higher in relatives of schizophrenia probands than in relatives of control probands. Other studies using DSM-IH-R or related criteria have found similar results. In the Iowa 500-non-500 series, the risk for schizophrenia was elevated in the relatives of probands with schizophreniform and schizoaffective disorder (Kendler et al. 1986). In a small Italian family study, the risk for schizophrenia was similar in the interviewed relatives of schizoaffective and schizophrenia probands (Maj et al. 1991). In the Irish Roscommon Family Study (Kendler et aL, in press a), relatives of probands with schizoaffective
ro oo
Table 2. Summary results of major family and adoption studies using personal interviews to examine prevalence for DSM-lll and DSM-III-R schizotypai or paranoid personality disorder in first-degree relatives of schizophrenia and normal control probands Relatives of schizophrenia probands Relatives of control probands SPD or PPD Total n n Prevalence SE SPD or PPD Total n n Prevalence SE
First author/year
Study
Lowing et al. (1983) 39 10 329 572 13 3 4.2 2.3 160 4 72 2.7 0.7 694 3 0.6 0.3 2.5 1.2 72 21.9 2.3 337 16 4.7 1.2 3 30.0 14.5 47 1 2.1 2.1 0.02 < 10x10-9 0.004 0.49 6 15.4 5.8 39 3 7.7 4.3 0.29
Adopted-away offspring
Kendler et al. (1984)
Biological relatives of adoptees
Baron et al. (1985)1
Nuclear family study
Frangos et al. (1985)
Coryell and Zimmerman (1988)1 108 319 26 8.2 1.5 580 3 2.8 1.6 380 0 10
Gershon et al. (1988)
0 1.7 0.5
0.01 0.000003
Kendler et al. (in press b)
Note.-OSA4-lll, DSM-/H-fl = Diagnostic and Statistical Manual of Mental Disorders (American Psychiatric Association 1980, 1987); SPD - schizotypai personality disorder, PPD > paranoid personality disorder; Prevalence = lifetime prevalence; SE = standard error.
nloaded from http://schizophreniabulletin.oxfordjournals.org/ at Pontificia Universidad Catolica de Chile on October 29, 2012
'Data only reported age corrected, so that morbid risk rather than prevalence reported.
O I N
CO
o
TJ 3D
m ro z
c m z
VOL. 19, NO. 2, 1993
269
Table 3. Summary of the results of major recent family studies examining the risk of nonschizophrenic nonaffective psychosis in relatives of schizophrenia versus control probands
Risk In first-degree relatives First author/year Baron et al. (1985) Frangos et al. (1985) Kendler (1985) Coryell and Zimmerman (1988) Gershon et al. (1988) Maier et al. (1990) Kendler et al. (in press d) Schizophrenia Control probands probands 1.6 1.7 5.0 0 7.1 5.0 3.7 0.3 1.1 0.4 2.0 0.9 0.5 1.4
Disorders DD, AP AP SF.SA.DD.AP SA SA+AP SA SF,SA,DD,AP
P
0.06 NS 0.000 NS 0.000
(1)
0.05
Note.DD = delusional disorder AP atypical psychosis; SF = schlzophrenrform disorder SA = schizoaffecttve disorder NS = not significant. 'Cannot be calculated and not given In original report.
disorder or other nonaffective psychosis had significantly elevated risks for schizophrenia. Given that suffering from affective illness represents a stress that might precipitate psychosis, one prediction of this hypothesis is that the probability of developing psychosis given affective illness should be increased in relatives of schizophrenia probands. Two studies have confirmed this prediction (Kendler et al. 1985a, in press c). However, other studies have been less consistent with this hypothesis. Three family studies done in the 1980s failed to find an increased rate for schizophrenia in relatives of delusional disorder probands (Watt 1980, 1985; Kendler et al. 1985b). Similarly, no increased risk for schizophrenia spectrum disorders was found in the biologic relatives of the small number of adoptees with schizophreniform disorder, delusional disorder, or atypical psychosis in a DSM-J/J-based rediagnosis of
the Copenhagen Danish Adoption Study (Kendler and Gruenberg 1984). The preponderance of the evidence, however, supports the validity of the third hypothesis that the familial vulnerability to schizophrenia may manifest itself, at least in part, as a predisposition to more broadly defined psychosis. Schizophrenia and Other Psychiatric Disorders. The last hypothesis (that the familial vulnerability to schizophrenia is entirely nonspecific) predicts that the risk for all major forms of psychopathology ought to be increased in relatives of schizophrenia versus control probands. We found nine recent controlled studies that used modem operationalized criteria, blind diagnosis, and control groups to address this issue with respect to three major categories of nonschizophrenic psychopathology: affective illness, anxiety disorders, and alcoholism (table 4). All nine
studies examined the risk for affective illness in relatives of schizophrenia and control probands. The results are inconsistent. One study reports significantly higher rates of affective illness in the relatives of the control probands (Frangos et al. 1985); six studies report no significant differences in the rates of illness in the two groups of relatives (Kendler and Gruenberg 1984; Baron et al. 1985; Kendler et al. 1985a; Coryell and Zimmerman 1988; Maj et al. 1991; Kendler et al., in press c); and two studies report significantly greater risks for affective illness in the relatives of the schizophrenia probands (Gershon et al. 1988; Maier et al. 1990). Overall, these results provide some support for the Kraepelinian dichotomy between dementia praecox and manicdepressive insanity. However, these latter two studies (Maj et al. 1991; Kendler et al., in press c) provide some evidence for a "unitary psychosis" theory most recently articulated by Crow (1990). In both of these studies (Gershon et al. 1988; Maier et al. 1990), when divided by polarity, only unipolar and not bipolar illness was significantly more common in relatives of schizophrenia probands than in relatives of control probands. This is inconsistent with most unitary models for psychosis, which place bipolar illness closer to schizophrenia than unipolar illness. Table 4 also reviews results from studies that have examined the risk for anxiety disorders or alcoholism in relatives of schizophrenia versus control probands. For both these disorders, one study found an increased rate in relatives of controls and the remaining studies found no signifi-
270
CO CO
(0
z z oE
- 0)
I II
cq co
IT) (D
C O
fl
I II
q d
111
"5
(0 * -
c o
E
C O 3 X C O
q co a> C\J Tt cj < NO
I II I II
in in
co
II
D (0 C O m
to
w
i- (S
N >
15
to
D
co co co co a> LO o cvj * co
n o
T - O) O ) CM CO T-;
CO
H
o 3
CQ CO
C OC OO C OO O *- O
C O OC O V oiqcq
co z
I?
co co co co in
cq
CNJ
oc a:
E"S
t--; i n -i-
CN f ^
in in ^ N ib
co in
CM
cant difference in risk between the two groups of relatives (Kendler et aJ. 1984, in press c; Baron et al. 1985; Frangos et al. 1985; Kendler 1985; Coryell and Zimmerman 1988; Gershon et al. 1988; Maier et al. 1990; Maj et al. 1991). The available evidence is providing an increasingly clear picture of the boundaries of the schizophrenia spectrum. The evidence is consistently against the first, narrow hypothesis because several disorders other than classic schizophrenia aggregate in relatives of individuals with schizophrenia. However, substantial evidence supports the second hypothesisthat familial liability to schizophrenia also predisposes to schizophrenialike personality traits. Although some uncertainty exists, most available data also support the third hypothesis^that the familial liability to schizophrenia also influences the risk for other forms of nonaffective psychosis. The evidence is not consistent with the broadest hypothesis, however. Anxiety disorder, alcoholism, and probably most forms of affective illness appear to have little or no familial relationship with schizophrenia. Thus, the familial liability to schizophrenia appears to be neither extremely narrow nor extremely broad in nature. Schizophrenia and Single Gene Models
sf
c *
5 O5 5 o
COQCOCOQ Q l Q Q l
OO5
QQCO CCOCQ
8-
1*
(0 C O * 3 C O O
.co g s s litll
c o _ . -jg
C D
M
i n eg
ll-
(0
Q . <D
C ^ O ) _: C O _. ^ "" C O C O
(a 0 ^
1!!
l- o a.
2 u.
m li-
C S o
If schizophrenia is strongly influenced by genetic factors, the next logical question is, What kind of factors? Are there single genes (sometimes called single major loci, or SML) that substantially influence liability to schizophrenia? Qearly some SML exist that cause human disease, such as classic Mendelian disorders like
VOL. 19, NO. 2, 1993
271
Huntington's disease, muscular dystrophy, or cystic fibrosis. But although SML may exist for schizophrenia, it is now certainly clear that schizophrenia differs from the classic Mendelian disorders in at least four crucial ways (McGue and Gottesman 1989; Risch 1990). First, most Mendelian disorders are fully penetrant. That is, if one inherits a "bad gene," one will always suffer from the disorder if one lives through the period of risk. For schizophrenia, the extreme rarity of families with apparent Mendelian inheritance, the concordance rate in MZ twins of well below 100 percent, and the evidence that offspring of unaffected MZ cotwins of schizophrenia persons are at substantially increased risk for schizophrenia (Gottesman and Bertelsen 1989) all suggest that this disorder exhibits reduced penetrance. That is, one can have high genetic liability to schizophrenia and not manifest the illness. Second, in most Mendelian conditions, in all individuals who express typical symptoms of the disease, the symptoms are due to direct effects of the disease gene. For schizophrenia, however, this is not true. Cases of schizophrenialike symptoms produced by metabolic or neurologic conditions or drug abuse clearly exist; these cases are called phenocopies. Third, most Mendelian conditions are rare and etiologically homogeneous; that is, all cases arise from abnormalities in the same gene. Different mutations in the same gene may cause variants of a disorder, but this does not affect linkage analysis because such allelic variants occur at the same genetic locus. Schizophrenia is, by contrast, a relatively common condition and is likely to involve genetic heterogeneity. This means
that defects at several genes probably influence the risk for forms of schizophrenia that are, at present, clinically indistinguishable though etiologically distinct. Such a perspective is comparable to the modern view of cancer as a collection of disorders identified by a common clinical manifestation (uncontrolled cell division) but caused by very distinct genetic mechanisms (e.g., activation of growthpromoting oncogenes vs. loss of tumor-suppressing genes). Finally, in most Mendelian disorders there is an obvious discontinuity between affected and unaffected individuals. As reviewed above, such diagnostic boundaries are less clearly delineated for schizophrenia. Although there is good evidence to consider an individual with schizoaffective disorder or SPD in a high-density schizophrenia pedigree to be "affected," family members with delusional disorder, atypical psychosis, or schizoid personality are more problematic. Segregation Analysis One approach in genetic epidemiology to evaluating whether SML exist is complex segregation analysis. This method determines a disease's mode of transmission by examining the pattern of disease in a systematically collected sample of nuclear families or extended pedigrees. In contrast to linkage analysis, this approach examines only disease phenotypes and not genetic marker information. The rather inconclusive results for schizophrenia obtained to date with this method are reviewed elsewhere (Kendler 1988; Kendler and Diehl, in press). One application of this method that was reported recently (Vogler et al. 1990)
used the large Swedish family study of Iindelius (1970). A multifactorial model with high heritability (> 80%) provided a substantially better fit to the data than did any SML models. Technical problems prevented the evaluation of a "mixed model" containing both multifactorial and SML components. Nevertheless, and consistent with most previous analyses, it appears unlikely that the distribution of schizophrenia is due solely to the effect of a SML (McGue and Gottesman 1989). Although complex segregation analyses can include incomplete penetrance and phenocopies as well as diagnostic uncertainty (by using different diagnostic "classes"), this method has no real power to examine genetic heterogeneity. Rather, results provide only an "aggregate" mode of transmission averaged across families. Therefore, lack of evidence of SML for schizophrenia should not be interpreted as strong evidence against the possibility that such "major" genes exist in a subset of families. Linkage Analysis The most striking recent development for research into the genetics of schizophrenia has been in linkage analysis, the general principles and methods of which are described elsewhere (e.g., Ott 1991). Before reviewing this area, however, we again emphasize the special challenges in applying this method to psychiatric disorders, which, as discussed above, are likely to be complex and heterogeneous. Linkage analysis in humans was developed originally for simple Mendelian traits. Schizophrenia is clearly not such a simple disorder, and this must be
272
kept in mind when performing and interpreting linkage studies. Reduced penetrance, phenocopies, diagnostic error, and locus heterogeneity clearly make linkage studies both qualitatively and quantitatively more difficult for schizophrenia than for simple Mendelian disordersqualitatively because statistical methods developed for classic Mendelian disorders must be modified before being applied to schizophrenia and quantitatively because necessary sample sizes are likely to be much larger than those sufficient for simple Mendelian disorders. We illustrate these points by an example. Several studies (CavalliSforza and King 1986; Chakravarti et al. 1987; Clerget-Darpoux et al. 1987; Goldin and Gershon 1988; Chen et al. 1992) have examined the impact of genetic heterogeneity on the power of linkage analysis. Two studies (Martinez and Goldin 1989; Chen et al. 1992) have examined conditions similar to those confronted by researchers studying multiplex families in schizophrenia, but both presented results across only a limited number of parameters. Following the methods of one of these reports (Martinez and Goldin 1989), we conducted our own power calculations (table 5) under a series of assumptions (detailed in the footnote). Although these assumptions are unlikely to be exactly correct, we believe they are reasonable guesses as to what must be confronted in attempting to undertake linkage studies of schizophrenia. As expected, the higher the penetrance (the probability of having schizophrenia if one has one or two copies of the predisposing gene), the smaller the number of families required to detect linkage. This makes sense because, as pen-
Table 5. Number of families needed to detect linkage to schizophrenia with 80 percent power, given varying levels of genetic heterogeneity
Penetrance
a
0.5 47 61 76 99 137 196 300 522 1106 4359
0 . 6 43 56 71 93 126 180 277 483

1045 3824
0.7 40 51 64 84 114 160 246 434 922

3550
0.8 34 44 56 73 99 140 213 370 794

2994
1 . 0 0 . 9 0.8 0.7 0.6 0.5 0.4 0.3 0.2 0 . 1
Note.a = proportion of families In which the schizophrenia-predisposing gene is linked to the marker. The power calculations depicted in this table are based on the following assumptions. (1) nudear families consisting of both parents and five offspring, in which both parents were unaffected and the number of affected offspring was modeled to equal that found in our Irish study of high-density schizophrenia pedigrees (Dlehl and Kendler 1989; Su 1991; Su et al. 1993), (2) schizophrenia modeled as a dominant trait with Incomplete penetrance ranging from 50-80 percent, a phenocopy rate of 0.10 percent, and population risk of 1 percent In both the linked and unlinked families; (3) 6 (the recombination fraction between the marker and the disease gene) = 5 percent; (4) power = 80 percent; (5) a LOD score of 4.27 required as evidence for linkage (the level found by MacLean et al. [1993] to equal a LOD of 3.0, given that the LOD score Is maximized with respect to penetrance and phenotyplc definition); and (6) PIC value (polymorphism Information content, a measure of how variable the marker Is In the population) of 0.70. We test for linkage assuming homogeneity because of evidence that, with family structures as found In this project, this is usually as powerful as or more powerful than testing for linkage assuming heterogeneity (Martinez and Qoldin 1989; Risen 1989).
etrance becomes higher, the clinical phenotype becomes a more accurate means to determine the underlying genotype. More important for our discussion here is the impact of genetic heterogeneity on the power to detect linkage. Assuming for a moment that the penetrance of our putative dominant-like schizophrenia gene is 60 percent, if all high-density families were due to this gene, it would require only 43 families (301 individuals) to provide an 80percent chance of detecting linkage. Unfortunately, few would expect schizophrenia to be so etiologically homogeneous. To us, it would seem optimistic to assume even the modest hetero-
geneity reflected in 70 percent of high-density families being linked to one gene, in which case the required sample size would more than double to 93 families (651 individuals). More likely, the single most common gene responsible for schizophrenia is present in only 50 or 30 percent of high-density families, so that required samples would amount to 180 families (1,260 individuals) or 483 families (3381 individuals), respectively. If schizophrenia is highly heterogeneous and the most common single gene accounts for only 10 or 20 percent of high-density families, sample sizes required to detect linkage by current methods may be unobtainable with realistic
VOL 19, NO. 2, 1993
273
resources. Similar results are obtained if, instead of assuming that highdensity families each have one and only one schizophrenia gene in them, one assumes that, in most such families, there exist a modest number of genes (e.g., about five), each of which influences liability to schizophrenia. This model, termed "oligogenic" for postulating a few important genes, also predicts that sample sizes needed for linkage analysis are many times greater than those needed for simple Mendelian traits (Suarez et al, in press). The first linkage study of schizophrenia known to us was performed in Switzerland and published in 1958 (Constantinidis 1958). McGuffin et al. (1983) reviewed this and other early linkage studies of schizophrenia using conventional protein-based markers such as human leukocyte antigens (HLA), blood groups, and enzyme polymorphisms. Most such studies since that time, such as the work of Goldin et al. (1987), focused on the HLA region, although one study examined blood-group markers in 19 multiplex families (Andrew et al. 1987). No replicated positive evidence for linkage to schizophrenia emerged from the use of these protein polymorphisms. In principle, the nature of genetic markers used in linkage studies is immaterial. Markers simply distinguish which of the four parental alleles at some chromosomal location are transmitted to each child in order to detect nonindependent assortment within families of chromosomal segments and a disease. In practice, modern molecular genetics has revolutionized linkage studies in man. Previously, informative markers
were available for only a very small fraction of the human genome. Thus, rare positive findings of linkage were of interest, but negative findings meant little because only a very small portion of the genome could be examined. However, modem molecular genetic techniques have provided both large numbers of markers for virtually all locations in the human genome and great improvements in marker informativeness (i.e., variation among individuals). To determine whether a disease is cosegregating with a chromosomal region, it is necessary to distinguish the chromosomes of maternal and paternal origin as well as the two homologous chromosomes of each parent. A marker with only 2 variants (alleles) is much less likely to provide these distinctions than a marker with 10 common alleles. One goal of the Human Genome Project (already nearly attained) is the development of highly informative deoxyribonucleic acid (DNA)-based markers, spanning the entire human genome, along with technologies for their accurate and efficient characterization (e.g., Ziegle et al. 1992). Certainly the most dramatic series of events in the genetics of schizophrenia in recent years began with the observation (McGillivray et al. 1990) of a young man and his maternal uncle, both of whom had schizophrenia and subtle dysmorphic racial features. Karyotype analysis revealed that both affected individuals had a chromosome 5qll.2ql3.3 trisomy, with the unaffected mother having a balanced inverted insertion. This isolated observation of two relatives with a partial trisomy and schizophrenia could easily have been a chance occurrence. Alternatively, these two rel-
atives could have had a very rare kind of schizophrenia etiologically unrelated to more common forms. However, this finding did represent a potential clue to the location of schizophrenia genes, which was pursued very quickly. In late 1988, Sherrington et al. reported linkage in seven highdensity pedigrees (five from Iceland and two from England) between schizophrenia-related phenotypes and two DNA markers in the region of the 5q trisomy. This article has been extensively reviewed elsewhere (McGuffin et al. 1990; Owen et al. 1990; Watt and Edwards 1991), so our comments here will be limited to its especially striking features. First, the pedigrees themselves were remarkable for both their size and their very high density of schizophrenia. Such pedigrees have rarely been found by other investigators. Second, evidence in favor of linkage of the 5qll-13 region to schizophrenia-related phenotypes was very strong, with a maximum LOD score1 of 6.49. Despite the
'The LOD score (Logarithm of the Odds) is a ratio of the likelihood of the observed pattern of disease and marker phenotypes in the families under study under two alternative hypotheses. The null hypothesis in the denominator is the hypothesis of no linkage (recombination fraction = 0.5), and this is contrasted with a series of alternative hypotheses ranging from very close linkage of the marker and disease loci (recombination fraction = 0.0) to various degrees of looser linkage (e.g., recombination fractions 0.01, 0.05, 0.10 ... < 0.5). Traditionally (based on arguments appropriate only for simple, nonheterogeneous diseases), a LOD score greater than 3.0 (i.e., odds of 1,000:1) is considered statistically significant evidence in support of linkage.
274
fact that at least 18 different linkage tests were performed on these data (using different definitions of illness and penetrance assumptions), this result suggests a very low probability of Type I statistical error. Third, contrary to findings from the family studies described above, evidence for linkage was Table 6. Study St. Clair et al. (1990) Aschauer et al. (1991) Lannfelt et al. (1991) Sherrington et al. (1988) Kennedy et al. (1988) St. Clair et al. (1989) Detera-Wadleigh et al. (1989) Aschauer et al. (1990) McGuffin et al. (1990) Crowe et al. (1991) Byeriy et al. (1991) Su (1991); Su et al. (1991) Mankoo et al. (1991) Wildenauer et al. (1991) Byeriy et al. (1991) Kennedy et al. (1991a) Wildenauer et al. (1991) McGuffin and Sturt (1986 2 ) Moises et al. (1991) Wildenauer et al. (1991) Byeriy et al. (1991) Macciardi et al. (1991) Owen et al. (1991) Su (1991); Su et al. (1991, 1993) Macciardi et al. (1991)
greatest when all psychiatric disorders (including phobia, minor depression, and alcoholism) were used to classify individuals as affected, rather than limiting this classification specifically to schizophrenia or schizophrenia-spectrum phenotypes. Fourth, contrary to expectations of high levels of genetic
heterogeneity, support of linkage was found for all seven pedigrees, which suggests substantial genetic homogeneity. Table 6 contrasts this original report with subsequent studies performed by other researchers using markers on proximal chromosome 5q. Attempts to replicate this find-
Linkage studies of schizophrenia Locus/region1 CR/1q43-11q21 ADM/2q DRD3/3q ADM75q11-13 ADM/5q11-13 ADM/5q11-13 ADM/5q11-13 ADM/5q11-13 ADM/5q11-13 ADM/5q11-13 ADM/5q11-13 ADM/5q11-13 ADM/5q11-13 ADM/5q11-13 DRD1/5q34-35 DRD1/5q34-35 DRD1/5q34-35 HLA/6p21.3 DRD2/11q23 DRD2/11q23 DRD2/11q23 DRD2/11q23 DRD2/ADM/11q23 DRD2/ADM/11q23 Exclusion of entire region assuming > 25%-50% proportion linked DRD4/11p15.5 Exclusion assuming homogeneity Exclusion or support of linkage Suggestion of linkage (for broadly defined psychiatric illness) Weak support assuming homogeneity or heterogeneity Exclusion assuming homogeneity Strong support assuming either homogeneity or heterogeneity Exclusion assuming homogeneity Exclusion assuming homogeneity Exclusion assuming homogeneity Exclusion Exclusion Exclusion Exclusion assuming assuming assuming assuming homogeneity homogeneity homogeneity homogeneity
Exclusion assuming > 25%-50% proportion linked Weak support of linkage assuming heterogeneity Exclusion assuming homogeneity Exclusion assuming homogeneity Exclusion assuming homogeneity Exclusion assuming homogeneity Exclusion assuming homogeneity Exclusion assuming homogeneity Exclusion assuming homogeneity Exclusion assuming homogeneity Weak suggestion of linkage Exclusion of DRD2; suggestion of linkage proximally
VOL 19, NO. 2, 1993
275
Table 6.
Study
Linkage studies of schizophreniaContinued

Locus/region1 DRD4/11p15.5 HRAS/11p15.5 HTR2/13q ADM/17q ADM/19 CYP2D/22 DMD/Xp21 ADM/XYpter ADM/XYpter ADM/XYpter ADM/XYpter ADM/XYpter ADM/Xq27-28 PP/20 loci ADM/150 loci ADM/30 loci ADM/150 loci
Exclusion or support of linkage Exclusion assuming homogeneity Exclusion assuming homogeneity Exclusion assuming homogeneity Exclusion assuming homogeneity Exclusion assuming Exclusion assuming Weak suggestion of Exclusion assuming homogeneity homogeneity linkage with muscular dystrophy homogeneity
Kennedy et al. (19916) O'Neill et al. (1991) Hallmayer et al. (1992) Kennedy et al. (1991c) Wildenauer et al.
(1991) Vallada et al. (1992) Zatz et al. (1991) Wildenauer et al. (1991)
d'Amato et al. (1992) Nonomura et al. (1991) Asherson et al. (1992) Collinge et al. (1991) DeLisi et al. (1991) McGuffin and Shirt
Suggestion of linkage Exclusion assuming homogeneity Exclusion assuming homogeneity Suggestion of linkage Exclusion assuming homogeneity Exclusion assuming homogeneity or inconclusive Exclusion assuming homogeneity or inconclusive Exclusion assuming homogeneity or inconclusive Exclusion assuming homogeneity or inconclusive
(19862)
Byerly et al. (1991) Polymeropoulos et al. (1991) Barr et al. (1991)
Note.Citations in table are presented based on chromosomal location (chromosome 1 through xy, followed by the genomewlde searches). CYP2D cytochrome P450 subfamily IID; DRD = dopamlne receptor D#; HRAS = H-ras oncogens (located In dose vicinity of DRD4); HTR2 = serotonln2 receptor (5-hydroxytryptamlne receptor #); HLA human leukocyte antigen (major hlstocompatibillty complex); DMD = Duchenne muscular dystrophy. 'Locus: Name of locus for markers that are genes or proteins of known function, ADM If anonymous DNA marker, PP If miscellaneous protein polymorphisms such as Isozymes or blood group antigens, and CR If chromosome rearrangement; Region: Chromosomal location of marker loci. 2 Review of several Independent studies
ing have been uniformly negative. Only the reports of Su et al. (Su 1991; Su et al. 1991) formally examined the important question of exclusion of linkage from this candidate region in the presence of possible heterogeneity among families within a study. Using a relatively large sample of Irish highdensity schizophrenia pedigrees, this study ruled out linkage to a disease gene present in any more
than 25 to 50 percent of all families, depending on the genetic model assumed. Although absolutely no positive evidence was found, even this sample, which is considerably larger than all others in table 6, could not eliminate the possibility that a major susceptibility gene in this chromosomal region could be present in up to half of the families studied. This exemplifies the crucial need for
large samples for studies of this complex disorder. A review by McGuffin et al. (1990) examined the question of heterogeneity among studies and found strong evidence that the pedigrees studied by Sherrington et al. (1988) differ significantly from those studied by subsequent investigators. The most recent addition to this controversial story has been contributed by the same investigative
276
team that reported the initial positive finding (Mankoo et al. 1991). Their new results include both retests of the original seven pedigrees using more informative markers and tests of additional high-density pedigrees from both Iceland and England. In the original pedigrees, evidence for linkage has diminished considerably although it has not disappeared entirely. Furthermore, the new families show strong evidence against linkage for this region, and the entire combined sample excludes linkage over the entire trisomic region. How can we explain the initial strong positive evidence of linkage, followed first by uniformly negative results in all independent studies and then by an inability of the original investigators to replicate their own work? There are three plausible explanations (McGuffin et al. 1990). First, the original finding may represent a highly unlikely Type I error. How unlikely is not entirely clear, as the original conceptual framework for interpreting LOD scores was constructed for Mendelian conditions in which the presence of a gene was assured and the mode of transmission was known. However, it is clear that, even after correcting for these effects, the original LOD score above 6.0, if a false positive, was very unlikely. Second, there may be a gene in the 5qll-13 region that influences liability to schizophrenia, but either it is of only modest effect or it is present in only a very small subset of families. This hypothesis also seems unlikely, however, because it is difficult to understand why the gene should be detected in all or most of the seven families in the initial study but essentially in very few or none of the
many other families studied by other investigators. This explanation also fails to account for the loss of most support, even in the original families, upon reanalysis with additional, more highly informative marker loci. Third, the possibility of a systematic source of error in the original report must be acknowledged. Although some limitations in the report by Sherrington et al. (1988) have been noted (Watt and Edwards 1991), there is no evidence of systematic bias that might arise, for example, if psychiatric diagnoses were performed with knowledge of marker genotypes. Thus, we remain in the unsatisfactory position of having no adequate explanation for this puzzling and discouraging series of events. Our own view is that a reasonably common gene of major effect for the liability to schizophrenia almost certainly does not exist in the 5qll-13 region. However, we do not yet possess data with sufficient statistical power to confidently rule out the existence, within this chromosomal region or at any other location in the human genome, of either a quite rare gene of major effect or a more common gene of minor effect on liability to schizophrenia. In addition to the 5qll-13 region, linkage to schizophrenia has been examined for several other candidate regions (table 6). To make table 6 as comprehensive as possible, we included citations to work thus far presented only as abstracts for meetings. Some of these studies are quite small in size and/or are preliminary, and we caution that final reports may offer different conclusions from those indicated in the table. Perhaps of greatest interest are candidate loci associated with the dopamine (DA) system, long con-
sidered to be centrally involved in the pathophysiology of schizophrenia. Three studies (Byerley et al. 1991; Kennedy et al. 1991a; Wildenauer et al. 1991) examined linkage between schizophrenia and the Dj locus located at 5q34-35, six studies (Byerly et al. 1991; Macciardi et al. 1991; Moises et al. 1991; Owen et al. 1991; Wildenauer et al. 1991; Su et al. 1993) evaluated the D2 locus on llq, one study examined the D 3 receptor, and three linkage studies (Kennedy et al. 1991b; Macciardi et al. 1991; O'Neill et al. 1991) evaluated the region of the newly identified D4 locus on lip. All studies considered only exclusion of linkage assuming homogeneity, except the study of Su et al. (Su 1991; Su et al. 1991, 1993), in which linkage to the D2 locus was excluded for at least 50 to 75 percent of families, depending on assumptions used for testing various genetic models. Very weak support of linkage was found for the D 2 locus in one study (Macciardi et al. 1991). Another study (Owen et al. 1991) found results suggestive of linkage on chromosome Ilq23 when evaluating an additive model of disease gene transmission in a region proximal to the location of the D2 DA receptor gene, but this same region was excluded for most families for a nearly identical genetic model in a study reported by Su et al. (Su 1991; Su et al. 1991, 1993). Evidence suggestive of linkage has been reported for chromosome 2q in the one study thus far reporting results for this region (Aschauer et al. 1991), and one report noted weak evidence of cosegregation of schizophrenia with the muscular dystrophy gene located on chromosome Xp21 (Zatz et al. 1991). Linkage between schizophrenia and the serotonin
VOL 19, NO. 2, 1993
277
receptor located on chromosome 13, a possible receptor for neuroleptics, has also been ruled out in a large Swedish pedigree (Hallmayer et al. 1992). Some investigators suggested that affected siblings are more often of the same sex, especially when schizophrenia appears to be paternally transmitted (Crow 1988). This finding would be consistent with a susceptibility gene located in the pseudoautosomal regions of the X and Y chromosomes (table 6). Two studies (Collinge et al. 1991; d'Amato et al. 1992) reported suggestive evidence of linkage to distinctly different locations in this chromosomal region while other studies (Nonomura et al. 1991; Wildenauer et al. 1991; Asherson et al. 1992) were negative. Three groups (Barr et al. 1991; Byerly et al. 1991; Polymeropoulos et al. 1991) reported preliminary results of genomewide searches using a very limited number of families (table 6). Thus far, no strong positive evidence of linkage has been found, which might not be surprising given the likely heterogeneity of this disorder and the need to evaluate many families, as discussed above. Finally, brief mention should be made of studies that provide weak evidence for the cosegregation of translocations involving several chromosomes (e.g., [2, 18], [1, 11], [6, 11], [9, 11]) with various forms of psychosis (Genest et al. 1976; Holland and Gosden 1990; St. Clair et al. 1990; Wagner et al. 1990). These studies may provide helpful leads in defining further potential candidate regions for linkage studies. In summary, no replicated positive findings have yet emerged from efforts to locate individual genetic loci that influence the lia-
bility to schizophrenia. This is not surprising, nor should it be too discouraging. The candidate loci that we have for schizophrenia are relatively weak in that evidence of their involvement in the pathophysiology of schizophrenia (i.e., a site of action of antipsychotic drugs) is indirect at best. The number of genes expressed in the central nervous system almost certainly reaches the tens of thousands, only a minute fraction of which have been identified. A priori, if major genes exist for schizophrenia, it is more likely that they will be among the large majority of previously unknown genes than among the small minority of genes already identified. Furthermore, from the perspective of exclusion mapping and with the possible exception of ongoing studies of a small number of families (Barr et al. 1991; Byerley et al. 1991; Polymeropoulos et al. 1991), the proportion of the genome excluded unless we assume nearly complete genetic homogeneity for schizophreniaremains quite small. If one or more major genes for schizophrenia exist in the human genome, it is unlikely that they would have been identified by all combined research efforts to date. Given the plausible scenario that several such genes exist, some with major effects and others having more minor influences, studies to date must still be considered preliminary. Association Studies While most interest has focused on linkage analysis, association studies represent another viable approach. Linkage examines the cosegregation of markers and disease within families. Association studies usually compare the frequency of in-
dividual genes (more technically alleles) in unrelated affected individuals with that in matched controls. Association studies have two major advantages over linkage studies in schizophrenia. First, finding large numbers of schizophrenia patients and controls is much easier and less costly than finding large numbers of highdensity families. Second, in some circumstances, genetic heterogeneity may impair the statistical power of association tests less than it does the power of tests for linkage. However, compared with linkage studies, association studies have two critical disadvantages. First, linkage studies examine general locations in the genome while association studies examine individual alleles (i.e., specific copies of a gene that reside at a particular location). Although too technical to describe here fully, association studies are most powerful for diseases in which many affected individuals have inherited the same disease gene dating back many generations to a very rare mutation event. For a genetic disease that has arisen from very many independent mutational events, association tests lose much of their power. Given the high frequency of schizophrenia in the population and the low reproductive success (i.e., fitness) of individuals affected by this disorder, genetic theory would suggest that mutational events leading to schizophrenia should be fairly common. Second, except for the case in which the same DNA variation that causes the disease simultaneously causes the difference among marker alleles, association works through linkage disequilibriumthat is, the tendency for alleles at different places very close together on chro-
278
mosomes to be correlated in the population. This occurs because the alleles are so close that crossing over (recombination) between the neighboring locations occurs very rarely. The further back in time the mutational event occurred, the more crossing over will occur in subsequent generations and the smaller the chromosomal distance will be over which linkage disequilibrium still exists. Linkage works on an entirely different principle: the cosegregation of marker and disease genes within families. For practical purposes, linkage can detect a disease gene within 5-10 centimorgans (cM) of the marker (where 1 cM equals a 1-percent probability of a recombination in any single meiosis). Association studies, by contrast, can rarely detect a disease gene further away than 1 cM if that far. Therefore, for genetic diseases arising from one or a few mutational events, association studies can be quite powerful if a marker is very close to the mutation. Consequently, association studies are limited to tests of "candidate genes," which are suspected a priori of having a possible etiologic role in the disease owing to known biochemical properties. When searching for disease genes not yet characterized, however, linkage studies are almost certainly a superior method. Associations of various polymorphisms, including isozymes, bloodgroup antigens, serum proteins, and the HLA region, have generally not been replicated (McGuffin and Sturt 1986; Saha et al. 1990), although sample sizes have not been very large for many of these studies. An association of borderline statistical significance was found with a DNA polymorphism
in the adenosine deaminase gene in a small sample of both bipolar and schizophrenia patients compared with control subjects (Detera-Wadleigh et al. 1987). Recently, two studies (Sanders et al. 1991; Diehl et al., submitted for publication) found associations with the porphobilinogen deaminase (PBGD) gene located distal to the D2 DA receptor gene on chromosome llq. This gene is a potential candidate for schizophrenia because PBGD deficiency causes acute intermittent porphyria, a rare disorder that, like schizophrenia, usually has a postpubertal onset and can present with psychotic symptoms. This disorder is more common than expected in psychiatric patients. One of these studies (Diehl et al., submitted for publication) used the same sample of schizophrenia and unaffected family members to test for both PBGD association and linkage to this chromosomal region. Although modestly significant evidence of association was found, no positive evidence of linkage was observed, and the PBGD map position was excluded for 60 to 80 percent of the families, depending on the genetic model parameters assumed. While the complexities of interpreting this example are beyond the scope of this review, it does suggest that, if the candidate gene approach is viable, a combination of both linkage and association methods may be optimal for attempting to uncover the complexities of a disorder such as schizophrenia.
phrenia has advanced modestly by steady increments in the last several years. The question of the familial aggregation of narrowly defined schizophrenia has been rather definitively settled. Although it is possible that schizophrenia is more familial in some populations than in others, this remains to be firmly established. The ongoing twin and adoption studies continue to confirm, with ever-increasing confidence, the major etiologic role played by genetic factors in the etiology of schizophrenia. Our knowledge of the boundary of the schizophrenia spectrum is coming into sharper focus. However, linkage studies of schizophrenia have yet to live up to their promise. Instead, we have had to face the frustration of nonreplication of apparently strong positive reports. What might the future hold for the genetics of schizophrenia? First, we caution against prematurely abandoning the more tried and true methods of psychiatric genetic research and replacing them entirely with linkage studies. Competition for funds and for new investigators is inevitable in a world of finite resources. It is true that linkage studies provide the possibility of great breakthroughs in our knowledge of the genetics and etiology of schizophrenia at basic biochemical and physiological levels that may never be addressed by traditional methods. However, a complete understanding of schizophrenia, from DNA to phenotype, will undoubtedly require a host of methods, including the traditional family, twin, and adoption studies, which many investigators now consider to be outdated. We reemphasize that, regardless of how sophisticated our molecular genetic and statistical methods become, chances of sue-
Conclusions and Future Directions In the more traditional areas of psychiatric genetics, our understanding of the genetics of schizo-
VOL 19, NO. 2, 1993
279
cess still ultimately depend on our ability to measure and classify psychiatric phenotypes in a manner consistent with their underlying genetic basis. The best analogy might be that of a stock portfolio. Family, twin, and adoption studies are low-risk, slow-growth, dependable investments that will continue, at a modest speed, to provide increasing knowledge about the genetics of schizophrenia. Linkage studies are hot, new, high-risk investments that might produce great breakthroughs but also might stall or even go bust. Individual investors will hold different views as to the optimal balance of these alternative strategies, depending on their objective assessment of the relative chances of success or failure and the rewards associated with each strategy, as well as on their personal comfort level in dealing with varying levels of uncertainty. Most investment counselors would suggest that a portfolio should include at least some of both kinds of investments. As a field, we would do well to follow such advice. Second, it is critical that we avoid premature disillusionment with linkage studies of schizophrenia. The human brain is very complex and quite difficult to access, and schizophrenia is a common and crippling condition. One of the very best hopes for approaching a complete understanding of the pathophysiology of this disorder, which could lead to therapeutic options currently undreamt of, lies in the "positional cloning" strategy (Collins 1992). For a complex disorder such as schizophrenia, this approach would most likely begin with gene mapping by linkage analysis. The aggregate results from twin and adoption studies allow us to con-
clude with some confidence that genes that influence liability to schizophrenia exist somewhere in the human genome. The crucial questions to which we do not have answers are (1) How many such genes are there? (2) How common are they? and (3) How large are their individual effects? If there are any relatively common genes of moderate to large effect, we have a very good probability of detecting them reliably in most study populations if we persevere in our study of large samples and maximize our statistical power to detect linkage under complex modes of inheritance. If there are very many genes, none of which has any more than a small effect on liability, current methods and projected sample sizes are almost certainly inadequate and will yield negative or unreplicated results. The path to replicated linkage results for schizophrenia will probably not be a smooth one. A pattern of tentative findings by one group not replicated by other groups may be likely. It is important that we begin this undertaking well informed of the risks and difficulties as well as of the possible benefits. If immediate returns are expected to come too easily, failure to fulfill this unrealistic expectation may lead to withdrawal of support. Thus, as we previously wrote: Linkage analysis could join the many scientific approaches to schizophrenia which have been characterized by rapid and overly enthusiastic endorsement by the psychiatric community only to be followed by disappointment and precipitous rejection. [Kendler 1987, p. 31] To carry out a truly credible execution of the linkage strategy for a disease as complex and hetero-
geneous as schizophrenia, large numbers of carefully diagnosed families and highly informative markers are required. These resources are just now beginning to be brought into action. While definitely not offering a guaranteed success, this approach, if allowed sufficient time to mature, could yield truly unprecedented insights into the etiology of this disorder. References Abrams, R., and Taylor, M.A. The genetics of schizophrenia: A reassessment using modern criteria. American Journal of Psychiatry, 140:171-175, 1983. American Psychiatric Association. DSM-III: Diagnostic and Statistical Manual of Mental Disorders. 3rd ed. Washington, DC: The Association, 1980. American Psychiatric Association. DSM-III-R: Diagnostic and Statistical Manual of Mental Disorders. 3rd ed., revised. Washington, DC: The Association, 1987. Andrew, B.; Watt, D.C.; Gillespie, C; and Chapel, H. A study of genetic linkage in schizophrenia. Psychological Medicine, 17:363-370, 1987. Aschauer, H.N.; Aschauer-Treiber, G.; Isenberg, K.E.; Todd, R.D.; Knesevich, M.A.; Garver, D.L.; Reich, T.; and Cloninger, C.R. No evidence for linkage between chromosome 5 markers and schizophrenia. Human Heredity, 40:109115, 1990. Aschauer, H.N.; Meszaros, K.; Aschauer-Treiber, G.; Willinger, U.; Fessl, D.; Chaudry, H.R.; Stompe, T.; Fuchs, K.; and Sieghart, W. RFLP-linkage studies of schizophrenia on chromosome 2. Psychiatric Genetics, 2:12, 1991.
280
Asherson, P.; Parfitt, E.; Sargeant, M.; Tidmarsh, S.; Buckland, P.; Taylor, C; Clements, A.; Gill, M.; McGuffin, P.; and Owen, M. No evidence for a pseudoautosomal locus for schizophrenia linkage analysis of multiply affected families. British Journal of Psychiatry, 161:63-68, 1992. Baron, M.; Gruen, R.; Rainer, J.D.; Kane, J.; Asnis, L.; and Lord, A. A family study of schizophrenia and normal control probands: Implications for the spectrum concept of schizophrenia. American Journal of Psychiatry, 142:447-455, 1985. Barr, C.L.; Kennedy, J.L.; Pakstis, A.J.; Wetterberg, L.; Sjogren, B.; Gelernter, J.; Hallmayer, J.; Moises, H.W.; Cavalli-Sforza, L.L.; and Kidd, K.K. Progress in genome scan for linkage in schizophrenia. [Poster] American Journal of Human Genetics, 49(Suppl.):335, 1991. Byerly, W.; Plaetke, R.; Jensen, S.; Holik, J.; Hoff, M.; Myles-Worsley, M.; Wender, P.; Reimherr, F.; Leppert, M.; O'Connell, P.; Lalouel, J.; and White, R. Linkage analysis of six schizophrenia pedigrees with 150 DNA markers. Schizophrenia Research, 4(Special Issue):274-275, 1991. Cavalli-Sforza, L.L., and King, M.-C. Detecting linkage for genetically heterogeneous diseases and detecting heterogeneity with linkage data. American Journal of Human Genetics, 38:599-616, 1986. Chakravarti, A.; Badner, J.A.; and Li, C.C. Tests of linkage and heterogeneity in Mendelian diseases using identity by descent scores. Genetic Epidemiology, 4:255, 1987. Chen, W.J.; Faraone, S.V.; and Tsuang, M.T. Linkage studies of schizophrenia: A simulation study of statistical power. Genetic Epidemiology, 9:123-139, 1992.
Clerget-Darpoux, F.; Babron, M.C.; and Bonaiti-Pellie, C.B. Power and robustness of the linkage homogeneity test in genetic analysis of common disorders. Journal of Psychiatric Research, 21:625-630, 1987. Collinge, J.; DeLisi, L.E.; Boccio, A.; Johnstone, E.C.; Lane, A.; Larkin, C; Leach, M.; Lofthouse, R.; Owen, F.; Poulter, M.; Shah, T.; Walsh, C; and Crow, T.J. Evidence for a pseudo-autosomal locus for schizophrenia using the method of affected sibling pairs. British Journal of Psychiatry, 158:624-629, 1991. Collins, F.S. Positional cloning: Let's not call it reverse anymore. Nature Genetics, 1:3-6, 1992. Constantinidis, J.K. Les marqueurs de chromosomes chez les schizophrenes et la recherche du linkage entre ces caracteres et la schizophrenie par la methode de Penrose. Journal de genetique humaine, 7:189-242, 1958. Coryell, W., and Zimmerman, M. The heritability of schizophrenia and schizoaffective disorder: A family study. Archives of General Psychiatry, 45:323-327, 1988. Crow, T.J. Sex chromosomes and psychosis: The case for a pseudoautosomal locus. British Journal of Psychiatry, 153:675-683, 1988. Crow, T.J. Nature of the genetic contribution to psychotic illnessA continuum viewpoint. Ada Psychiatrica Scandinavica, 81:401-408, 1990. Crowe, R.R.; Black, D.W.; Wesner, R.; Andreasen, N.C.; Cookman, A.; and Roby, J. Lack of linkage to chromosome 5qll-ql3 markers in six schizophrenia pedigrees. Archives of General Psychiatry, 48:357361, 1991. d'Amato, T.; Campion, D.; Gorwood, P.; Jay, M.; Sabate, O ; Petit,
C; Abbar, M.; Malafosse, A.; Leboyer, M.; Hillaire, D.; ClergetDarpoux, F.; Feingold, J.; Waksman, G.; and Mallet, J. Evidence for a pseudoautosomal locus for schizophrenia. II: Replication of a non-random segregation of alleles at the DXYS14 locus. British Journal of Psychiatry, 161:59-62, 1992. DeLisi, L.E.; Crow, T.J.; Davies, K.E.; Terwilliger, J.D.; Ott, J.; Ram, R.; Flint, T.; and Boccio, A. No genetic linkage detected for schizophrenia to Xq27-q28. British Journal of Psychiatry, 158:630-634, 1991. Detera-Wadleigh, S.D.; de Miguel, C; Berrertini, W.H.; DeLisi, L.E.; Goldin, L.R.; and Gershon, E.S. Neuropeptide gene polymorphisms in affective disorder and schizophrenia. Journal of Psychiatric Research, 21:581-587, 1987. Detera-Wadleigh, S.D.; Goldin, L.R.; Sherrington, R.; Encio, I.; de Miguel, C; Berrettini, W.; Gurling, H.; and Gershon, E.S. Exclusion of linkage to 5qll-13 in families with schizophrenia and other psychiatric disorders. Nature, 340:391-393, 1989. Diehl, S.R., and Kendler, K.S. Strategies for linkage studies of schizophrenia: Pedigrees, DNA markers, and statistical analyses. Schizophrenia Bulletin, 15:403-419, 1989. Diehl, S.R.; Su, Y.; Murphy, B.; O'Neill, F.A.; Nie, L.; Burke, J.; Duke, B.W.; Kipps, B.R.; MacLean, C.J.; Walsh, D.; and Kendler, K.S. "Association of Porphobilinogen Deaminase RFLPs With Schizophrenia Without Evidence of Linkage." Submitted for publication. Falconer, D.S. The inheritance of liability to certain diseases, estimated from the incidence among relatives. Annals of Human Genetics, 29:51-76, 1965.
VOL 19, NO. 2, 1993
281
Frangos, E.; Athanassenas, G.; Tsitourides, S.; Katsanou, N.; and Alexandrakou, P. Prevalence of DSM-HI schizophrenia among the first-degree relatives of schizophrenic probands. Ada Psychktrica Scandinavica, 72:382-386, 1985. Genest, P.; Dumas, L.; and Genest, F.B. Translocation chromosomique t (2; 18) (q21; q23) chez un individu. schizophrene et sa fille. Union Medicale du Canada, 105:1617-1681, 1976. Gershon, E.S.; DeLisi, L.E.; Hamovit, J.; Nurnberger, J.I.; Maxwell, M.D.; Schreiber, J.; Dauphinais, D.; Dingman, C.W.; and Guroff, J.J. A controlled family study of chronic psychoses. Archives of General Psychiatry, 45:328336, 1988. Goldin, L.R.; DeLisi, L.E.; and Gershon, E.S. Relationship of HLA to schizophrenia in 10 nuclear families. Psychiatry Research, 20:6977, 1987. Goldin, L.R., and Gershon, E.S. Power of the affected-sib-pair method for heterogeneous disorders. Genetic Epidemiology, 5:35-43, 1988. Gottesman, I.I. Schizophrenia Genesis: The Origins of Madness. New York, NY: W.H. Freeman, 1991. Gottesman, 1.1., and Bertelsen, A. Confirming unexpressed genotypes for schizophrenia: Risks in the offspring of Fischer's Danish identical and fraternal discordant twins. Archives of General Psychiatry, 46:867872, 1989. Gottesman, I.I.; McGuffin, P.; and Farmer, A.E. Clinical genetics as clues to the "real" genetics of schizophrenia: A decade of modest gains while playing for time. Schizophrenia Bulletin, 13:23-47, 1987.
Gottesman, 1.1., and Shields, J. Contributions of twin studies to perspectives on schizophrenia. In: Maher, B.A., ed. Progress in Experimental Personality Research. Vol. 3. New York, NY: Academic Press, 1966. Guze, S.B.; Cloninger, C.R.; Martin, R.L.; and Clayton, P.J. A follow-up and family study of schizophrenia. Archives of General Psychiatry, 40:1273-1276, 1983. Hallmayer, J.; Kennedy, J.L.; Wetterberg, L.; Sjogren, B.; Kidd, K.K.; and Cavalli-Sforza, L.L. Exclusion of linkage between the serotonin2 receptor and schizophrenia in a large Swedish kindred. Archives of General Psychiatry, 49:216-219, 1992. Holland, T., and Gosden, C. A balanced chromosomal translocation partially co-segregating with psychotic illness in a family. Psychiatry Research, 32:1-8, 1990. Kendler, K.S. Overview: A current perspective on twin studies of schizophrenia. American Journal of Psychiatry, 140:1413-1425, 1983. Kendler, K.S. Diagnostic approaches to schizotypal personality disorder A historical perspective. Schizophrenia Bulletin, 11:538-553, 1985. Kendler, K.S. The feasibility of linkage studies in schizophrenia. In: Helmchen, H., and Henn, F.A., eds. Biological Perspectives of Schizophrenia. Chichester, England: John Wiley & Sons, 1987. pp. 1932. Kendler, K.S. The genetics of schizophrenia: An overview. In: Nasrallah, H.A.; Tsuang, M.T.; and Simpson, J.C., eds. Handbook of Schizophrenia: Nosology, Epidemiology, and Genetics. Vol. 3. Amsterdam, The Netherlands: Elsevier, 1988. pp. 437-462.
Kendler, K.S., and Diehl, S.R. The genetics of schizophrenia. In: Kaplan, H.I., and Sadock, B.J., eds. Comprehensive Textbook of Psychiatry. Vol. 4. Baltimore, MD: Williams & Wilkins, in press. Kendler, K.S., and Gruenberg, A.M. An independent analysis of the Copenhagen sample of the Danish Adoption Study of Schizophrenia. VI: The pattern of psychiatric illness as defined by DSM-III in adoptees and relatives. Archives of General Psychiatry, 41:555-564, 1984. Kendler, K.S.; Gruenberg, A.M.; and Tsuang, M.T. Psychiatric illness in first-degree relatives of schizophrenic and surgical control patients: A family study using DSM-HI criteria. Archives of General Psychiatry, 42:770-779, 1985a. Kendler, KS.; Gruenberg, A.M.; and Tsuang, M.T. A DSM-IU family study of the non-schizophrenic psychotic disorders. American Journal of Psychiatry, 143:1098-1105, 1986. Kendler, K.S.; Masterson, C; and Davis, K.L. Psychiatric illness in first-degree relatives of patients with paranoid psychosis, schizophrenia, and medical illness. British Journal of Psychiatry, 147:524531, 1985b. Kendler, K.S.; McGuire, M.; Gruenberg, A.M.; O'Hare, A.; Spellman, M.; and Walsh, D. The Roscommon Family Study. I: Methods, diagnosis of probands and risk of schizophrenia in relatives. Archives of General Psychiatry, in press a. Kendler, K.S.; McGuire, M.; Gruenberg, A.M.; O'Hare, A.; Spellman, M.; and Walsh, D. The Roscommon Family Study. HI: Schizophrenia-related personality disorders in relatives. Archives of General Psychiatry, in press b.
282
Kendler, K.S.; McGuire, M.; Gruenberg, A.M.; O'Hare, A.; Spellman, M.; and Walsh, D. The Roscommon Family Study. IV: Affective illness, anxiety disorders and alcoholism in relatives. Archives of General Psychiatry, in press c. Kendler, K.S.; McGuire, M.; Gruenberg, A.M.; Spellman, M.; O'Hare, A.; and Walsh, D. The Roscommon Family Study. II: The risk of nonschizophrenic nonaffective psychoses in relatives. Archives of General Psychiatry, in press d. Kennedy, J.L.; Bassett, A.S.; Sidenberg, D.G.; Van Tol, H.H.M.; Civelli, O.; Honer, W.G.; Collins, E.J.; Kamble, A.B.; and Barr, C.L. Linkage studies of the dopamine D4 receptor gene and schizophrenia in eastern Canadian families. [Poster] Psychiatric Genetics, 2:81, 1991a. Kennedy, J.L.; Guiffra, L.A.; Moises, H.W.; Cavalli-Sforza, L.L.; Pakstis, A.J.; Kidd, J.R.; Castiglione, CM.; Sjogren, B.; Wetterberg, L.; and Kidd, K.K. Evidence against linkage of schizophrenia to markers on chromosome 5 in a northern Swedish pedigree. Nature, 336:167-170, 1988. Kennedy, J.L.; Honer, W.G.; Gelernter, J.; Wetterberg, L.; Kaufmann, C.A.; Niznik, H.B.; O'Dowd, B.F.; Civelli, O.; and Kidd, K.K. Linkage studies of two new genes of neuropsychiatric interest. [Poster] Schizophrenia Research, 4:280, 1991b. Kennedy, J.L.; Honer, W.G.; Martignetti, J.; Macciardi, F.; Moreno, C; Cavallini, M.C.; and Smeraldi, E. Antibody selection for candidate genes in schizophreniaLinkage studies. Psychiatric Genetics, 2:3132, 1991c. Kety, S.S. The significance of genetic factors in the etiology of
schizophrenia: Results from the national study of adoptees in Denmark. Journal of Psychiatric Research, 21:423-129, 1987. Kety, S.S.; Rosenthal, D.; Wender, P.H.; and Schulsinger, F. The types and prevalence of mental illness in the biological and adoptive families of adopted schizophrenics. Journal of Psychiatric Research, 6:345-362, 1968. Kety, S.S.; Rosenthal, D.; Wender, P.H.; Schulsinger, F.; and Jacobsen, B. Mental illness in the biological and adoptive families of adopted individuals who have become schizophrenic: A preliminary report based on psychiatric interviews. In: Fieve, R.; Rosenthal, D.; and Brill, H., eds. Genetic Research in Psychiatry. Baltimore, MD: Johns Hopkins Press, 1975. pp. 147-165. Kety, S.S.; Wender, P.; Jacobsen, B.; Ingraham, L.J.; Kinney, D.K.; Schulsinger, F.; Rosenthal, D.; Jansson, L.; and Faber, B. "Mental Illness in the Biological and Adoptive Relatives of Adopted Individuals with Schizophrenia. Phase II: The National Sample Outside of Copenhagen." In preparation. Lannfelt, L.; Sokoloff, P.; Martres, M.-P.; Pilon, C; Giros, B.; and Schwartz, J.-C. The dopamine D 3 receptor gene and schizophrenia. Psychiatric Genetics, 2:16, 1991. lindelius, R., ed. A study of schizophrenia. A clinical, prognostic, and family investigation. Acta Psychiatrica Scandinavica, 216(Suppl.):l-125, 1970. Lowing, P.A.; Mirsky, A.F.; and Pereira, R. The inheritance of schizophrenia spectrum disorders: A reanalysis of the Danish Adoptee Study data. American Journal of Psychiatry, 140:1167-1171, 1983.
Macciardi, F.; Cavallini, M.C.; Kennedy, J.L.; Marino, C; Grandy, D.; Van Tol, H.H.M.; Niznik, H.B.; and Smeraldi, E. Testing the hypothesis of genetic linkage between schizophrenia and dopamine receptor genes. Psychiatric Genetics, 2:17-18, 1991. MacLean, C.J.; Bishop, C.T.; Sherman, S.L.; and Diehl, S.R. Distribution of LOD scores under uncertain mode of inheritance. American Journal of Human Genetics, 52:354 361, 1993. Maier, W.; Hallmayer, J.; Minges, J.; and Iichtermann, D. Affective and schizoaffective disorders: Similarities and differences. In: Marneros, A., and Tsuang, M.T., eds. Morbid Risks in Relatives of Affective, Schizoaffective, and Schizophrenic PatientsResults of a Family Study. New York, NY: Springer-Verlag, 1990. pp. 201-207. Maj, J.M.J.; Starace, F.; and Pirozzi, R. A family study of DSM-III-R schizoaffective disorder, depressive type, compared with schizophrenia and psychotic and nonpsychotic major depression. American Journal of Psychiatry, 148:612-616, 1991. Mankoo, B.; Sherrington, R.; Brynjolfsson, J.; Kalsi, G.; Perursson, H.; Sigmundsson, T.; Read, T.; Murphy, P.; Curtis, D.; Melmer, G.; and Gurling, H. New microsatellite polymorphisms provide a highly polymorphic map of chromosome 5 bands qll.2-ql3.3 for linkage analysis of Icelandic and English families affected by schizophrenia. [Abstract] Psychiatric Genetics, 2:17, 1991. Martinez, M.M., and Goldin, L.R. The detection of Linkage and heterogeneity in nuclear families for complex disorders: One versus two marker loci. American Journal of Human Genetics, 44:552-559, 1989.
VOL. 19, NO. 2, 1993
283
McGillivray, B.C.; Bassett, AS.; Langlois, S.; Pantzar, T.; and Wood, S. Familial 5qll.2-ql3.3 segmental duplication cosegregating with multiple anomalies, including schizophrenia. American Journal of Medical Genetics, 35:10-13, 1990. McGue, M., and Gottesman, I.I. Genetic linkage in schizophrenia: Perspectives from genetic epidemiology. Schizophrenia Bulletin, 15:453-464, 1989. McGue, M.; Gottesman, I.I.; and Rao, D.C. The transmission of schizophrenia under a multifactorial threshold model. American Journal of Human Genetics, 35:11611178, 1983. McGuffin, P.; Festenstein, H.; and Murray, R. A family study of HLA antigens and other genetic markers in schizophrenia. Psychological Medicine, 13:31-43, 1983. McGuffin, P.; Sargeant, M.; Hetti, G.; Tidmarsh, S.; Whatley, S.; and Marchbanks, R.M. Exclusion of a schizophrenia susceptibility gene from the chromosome 5qll-ql3 region: New data and a reanalysis of previous reports. American Journal of Human Genetics, 47:524-535, 1990. McGuffin, P., and Sturt, E. Genetic markers in schizophrenia. Human Heredity, 36:65-88, 1986. Moises, H.W.; Gelemter, J.; Giuffra, L.A.; Zarcone, V.; WetterbeTg, L.; Civelli, O.; Kidd, K.K.; CavalliSforza, L.L.; Grandy, D.K.; Kennedy, J.L.; Vinogradov, S.; Mauer, J.; Litt, M.; and Sjogren, B. No linkage between D2 dopamine receptor gene region and schizophrenia. Archives of General Psychiatry, 48:643-647, 1991. Nonomura, Y.; Yoneda, H.; Inayama, Y.; Kono, Y.; Ishida, T.; and Sakai, T. A study of pseudoautosomal region in sib-pairs with
schizophrenia. [Poster] Psychiatric Genetics, 2:89, 1991. O'Neill, F.A.; Hugher, A.; Diehl, S.R.; and Kendler, K.S. Exclusion of the H-ras region from linkage to schizophrenia in Irish families. [Poster] Psychiatric Genetics, 2:9192, 1991. Onstad, S.; Skre, I.; Edvardsen, J.; Torgersen, S.; and Kringlen, E. Mental disorders in first-degree relatives of schizophrenics. Ada Psychiatrica Scandinavica, 83:463467, 1991a. Onstad, S.; Skre, I.; Torgersen, S.; and Kringlen, E. Twin concordance for DSM-Ul-R schizophrenia. Ada Psychiatrica Scandinavica, 83:395401, 1991b. Ott, J. Analysis of Human Genetic Linkage. Revised ed. Baltimore, MD: Johns Hopkins University Press, 1991. Owen, M.; Craufurd, D.; and St. Clair, D. Localization of a susceptibility locus for schizophrenia on chromosome 5. British Journal of Psychiatry, 157:123-127, 1990. Owen, M.; Parfirt, E.; Sargeant, M.; Asherson, P.; McGuffin, P.; Taylor, C; Collier, D.; Powell, J.; Murray, R.; and Gill, M. Linkage studies of schizophrenia with markers of chromosome llq. [Poster] American Journal of Human Genetics, 49 (Suppl.):354, 1991. Polymeropoulos, M.H.; Delisi, L.E.; Poulter, M.; Lofthouse, R.; Crow, T.J.; Xiao, H.; Boccio, A.; Weber, J.L.; and Merril, C.R. Genetic linkage studies in schizophrenia using (CA)n repeat polymorphisms. [Abstract] Schizophrenia Research, 4:283, 1991. Pope, H.G.; Jones, J.M.; Cohen, B.M.; and Iipinski, J.F. Failure to find evidence of schizophrenia in first-degree relatives of schizophrenic probands. American Journal
of Psychiatry, 139:826-828, 1982. Risch, N. Linkage detection tests under heterogeneity. Genetic Epidemiology, 6:473-480, 1989. Risch, N. Genetic linkage and complex diseases, with special reference to psychiatric disorders. Genetic Epidemiology, 7:3-16, 1990. Saha, N.; Tay, J.S.H.; Tsoi, W.F.; and Kua, E.H. Association of Duffy blood group with schizophrenia in Chinese. Genetic Epidemiology, 7:303-305, 1990. Sanders, A.R; Hamilton, J.D.; Fann, W.E.; and Patel, P.I. Association of genetic variation at the porphobilinogen deaminase gene with schizophrenia. [Poster] American Journal of Human Genetics, 49 (Suppl.):358, 1991. Scharfetter, C, and Niisperli, M. The group of schizophrenias, schizoaffective psychoses, and affective disorders. Schizophrenia Bulletin, 6:586-591, 1980. Sherrington, R.; Brynjolfsson, B.; Petursson, R; Porter, M.; Dudleston, K.; Barradough, B.; Wasmuth, J.; Dobbs, M.; and Gurling, H. Localization of a susceptibility locus for schizophrenia on chromosome 5. Nature, 336:164-167, 1988. Spitzer, R.L.; Endicott, J.; and Gibbon, M. Crossing the border into borderline personality and borderline schizophrenia. Archives of General Psychiatry, 36:17-24, 1979. Spitzer, R.L.; Endicott, J.; and Robins, E. Research Diagnostic Criteria: Rationale and reliability. Archives of General Psychiatry, 35:773782, 1978. Squires-Wheeler, E.; Skodol, A.E.; Bassett, A.; and ErlenmeyerKimling, L. DSM-IU-R schizotypal personality traits in offspring of schizophrenic disorder, affective disorder, and normal control par-
284
ents. Journal of Psychiatric Research, 23:229-239, 1989. Squires-Wheeler, E.; Skodol, A.E.; Friedman, D.; and ErlenmeyerKimling, L. The specificity of DSM-HI schizotypal personality traits. Psychological Medicine, 18:757-765, 1988. St. Clair, D.; Blackwood, D.; Muir, W.; Baillie, D.; Hubbard, A.; Wright, A.; and Evans, HJ. No linkage of chromosome 5qll-ql3 markers to schizophrenia in Scottish families. Nature, 339:305-309, 1989. St. Clair, D.; Blackwood, D.; Muii, W.; Carothers, A.; Walker, M.; Spowart, G.; Gosden, C; and Evans, H.J. Association within a family of a balanced autosomal translocation with major mental illness. Lancet, 336:13-16, 1990. Su, Y. "Linkage Analysis of Schizophrenia in Irish Pedigrees: Exclusion of Candidate Regions on Chromosomes 5q and llq." Unpublished Ph.D. dissertation, Virginia Commonwealth University, 1991. Su, Y.; Bray, J.; Burke, J.; Inomata, T.; MacLean, C.J.; Marazita, M.; Murphy, B.; Nie, L.; Ni Nuallain, M.; O'Neill, F.A.; Ploughman, L.; Shinkwin, R.; Walsh, D.; Kendler, K.S.; and Diehl, S. Linkage studies of schizophrenia: Exclusion of candidate regions on chromosomes 5q and llq. [Poster] American Journal of Human Genetics, 49(Suppl.):361, 1991. Su, Y.; Burke, J.; O'Neill, F.A.; Murphy, B.; Nie, L.; Kipps, B.; Bray, J.; Shinkwin, R.; Ni Nuallain, M.; MacLean, C.J.; Walsh, D.; Diehl, S.R.; and Kendler, K.S. Exclusion of linkage between schizophrenia and the D 2 dopamine receptor gene region of chromosome l l q in 112 Irish multiplex families.
Archives of General Psychiatry, 50:205-211, 1993. Suarez, B.K.; Hampe, C.L.; and Van Eerdewegh, P. Problems of replicating linkage claims in psychiatry. In: Gershon, E.S., and Cloninger, C.R., eds. New Genetic Approaches to Mental Disorders. New York, NY: American Psychiatric Association, in press. Tienari, P. Interaction between genetic vulnerability and family environment: The Finnish adoptive family study of schizophrenia. Acta Psychiatrica Scandinavica, 84:460465, 1991. Tsuang, M.T.; Winokur, G.; and Crowe, R.R. Morbidity risks of schizophrenia and affective disorders among first-degree relatives of patients with schizophrenia, mania, depression and surgical conditions. British Journal of Psychiatry, 137:497-504, 1980. Vallada, H.; Collier, D.; Dawson, E.; Owen, M.; Nanko, S.; Murray, R.; and Gill, M. Debrisoquine-4hydroxylase (CYP2D) locus and possible susceptibility to schizophrenia. Lancet, 340:181-182, 1992. Vogler, G.P.; Gottesman, I.I.; McGue, M.K.; and Rao, D.C. Mixed model segregation analysis of schizophrenia in the Lindelius Swedish pedigrees. Behavior Genetics, 20:461-472, 1990. Wagner, C; McPherson, J.; Sugiyama, R.; and Smith, M. Mapping in Hq23 using in-situ hybridization: Identification of a probe which spans the 11/22 constitutional breakpoint. American Journal of Human Genetics, 47(Suppl.):A265, 1990. Watt, D.C, and Edwards, J.H. Doubt about evidence for a schizophrenia gene on chromosome 5. Psychological Medicine, 21:279-285, 1991.
Watt, J.A.G. Paranoid states of middle life: Familial occurrence and relationship to schizophrenia. Acta Psychiatrica Scandinavica, 61:413-^26, 1980. Watt, J.A.G. The relationship of paranoid states to schizophrenia. American Journal of Psychiatry, 142:1456-1458, 1985. Weeks, D.E.; Brzustowicz, L.; Squires-Wheeler, E.; Comblatt, B.; Lehner, T.; Stefanovich, M.; Bassett, A.; Gilliam, T.C.; Ott, J.; and Erlenmeyer-Kimling, L. Report of a workshop on genetic linkage studies in schizophrenia. Schizophrenia Bulletin, 16:673-686, 1990. Wildenauer, D.B.; Schwab, S.; Wurl, D.; Ertl, M.; Ackenheil, M.; Schmidt, S.; Drews, B.; Schmidt, F.; Hallmayer, J.; and Maier, W. Linkage analysis in schizophrenia: Exclusion of 5qll-ql3, 5q34-qter, Ilq22, 23, Xpter and chromosome 19 in 15 systematically ascertained European families. [Poster] American Journal of Human Genetics, 49(Suppl.):363, 1991. Zatz, M.; Melo, M.S.; PassosBueno, M.R.; Valada-Filho, H.; Vieira-Filho, A.H.G.; Vainzof, M.; Rapaport, D.; and Gentil-Filho, V. Association of schizophrenia and Becker dystrophy (BMD): A susceptibility locus for schizophrenia at Xp21 or an effect of the dystrophin gene in the brain? [Poster] American Journal of Human Genetics, 49(Suppl.):364, 1991. Zerbin-Rudin, E. Endogene psychosen. In: Becker, P.E., ed. Humangenetik: ein kurzes Handbuch in fUnf BUnde. Vol. 2. Stuttgart, Germany: Thieme, 1967. pp. 446-513. Ziegle, J.S.; Su, Y.; Corcoran, K.P.; Nie, L.; Mayrand, P.E.; Hoff, L.B.; McBride, L.J.; Kronick, M.N.; and Diehl, S.R. Application of automated DNA sizing technology for
VOL 19, NO. 2, 1993
285
genotyping microsatellite loci. Genomics, 14:1026-1031, 1992.
Acknowledgments This research was supported by USPHS grants MH-41953 and
MH-45390 from the National Institute of Mental Health, and the Rachel Brown Banks Endowment Fund. Table 5 was prepared in collaboration with Charles MacLean, Ph.D. Becky Gander, M.A., assisted in the manuscript preparation.
The Authors Kenneth S. Kendler, M.D., is Professor, and Scott R. Diehl, Ph.D., is Assistant Professor, Departments of Psychiatry and Human Genetics, Medical College of Virginia, Virginia Commonwealth University, Richmond, VA. McGill University and the staff of the Allan Memorial Institute of Psychiatry at the Royal Victoria Hospital. At the same time, he successfully completed studies at the Canadian Psychoanalytic Institute by 1968. When he left that year for McMaster, he was Director of the Family Psychiatry Section at the Allan Memorial Institute. In 1975, Jock became Chair of the Department of Psychiatry at McMaster, which he held until 1983. To increase his knowledge of behavioral neurobiology, he became a visiting professor in the Behavioral Neurology Unit at Beth Israel and Mailman Research Center at Harvard, returning to McMaster to further develop his now well-known studies of positron emission tomography and neurobehavior in schizophrenia patients. Following this period in Boston, he pursued the most important scholarly work of his career. On his return from Boston he worked in the Department of Psychiatry at McMaster where he continued to establish for himself an enviable role in clinical work. The teams that he created are continuing, but his rare gifts of enhancing investigation and inspiring collaboration are very much missed and not easily replaced. Jock had a distinguished career and a rich life professionally, academically, and with his family. The world of psychiatry deeply mourns his passing.
In Memoriam
After a brief illness, Dr. John Marnoch Cleghorn died on June 8, 1992, at his home in Burlington, Ontario, Canada, 3 days before his 59th birthday. Dr. Cleghorn, who was known universally as 'Jock," was Professor Emeritus of the Department of Psychiatry at McMaster University and Professor of Psychiatry and Head of the Schizophrenia Research Program at the University of Toronto. His death occasioned a sense of extreme loss to the field of psychiatry of an outstanding clinician, teacher, scientist, administrator, and fine human being. Jock was born in Scotland where his father was pursuing postgraduate studies in medicine. Jock studied at McGill University and obtained a degree in arts in 1954 and medicine in 1958. His father, by this time, had become a distinguished Professor of Psychiatry, and Jock started his own postgraduate training after an internship at the Royal Victoria Hospital in Montreal. He spent a year in medicine in metabolism before transferring to psychiatry in 1960. After a year at the Douglas Hospital where he was inspired by the work of Dr. Heinz Lehmann, he spent the next 3 years in Cincinnati as a resident and later an instructor and assistant research professor with L.A. Gottschalk. When he returned to Montreal in 1964, he joined the faculty of

The Genetics of Schizophrenia: A Current, Genetic-Epidemiologic Perspective

Uploaded by

Copyright:

Available Formats

The Genetics of Schizophrenia: A Current, Genetic-Epidemiologic Perspective

Uploaded by

Document Information

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

The Genetics of Schizophrenia: A Current, Genetic-Epidemiologic Perspective

Uploaded by

Copyright:

Available Formats

VOL. 19, NO.

The Genetics of Schizophrenia: A Current, Genetic-Epidemiologic Perspective

by Kenneth S. Kendler and Scott R. Dlehl

First-degree relatives of control probands

First author/ year

Baron et al. (1985)

Frangos et al. 1 (1985)

Coryell and Zimmerman (1988)

Gershon et al. (1988)

RDC chronic schizophrenia

Maier et al. (1990)

Kendler et al. 1 (in press a)

VOL. 19, NO. 2, 1993

VOL. 19, NO. 2, 1993

Kendler et al. (1984)

Biological relatives of adoptees

Baron et al. (1985)1

Nuclear family study

Frangos et al. (1985)

Nuclear family study

Nuclear family study

Gershon et al. (1988)

Nuclear family study

Kendler et al. (in press b)

Nuclear family study

VOL. 19, NO. 2, 1993

Disorders DD, AP AP SF.SA.DD.AP SA SA+AP SA SF,SA,DD,AP

q co a> C\J Tt cj < NO

VOL. 19, NO. 2, 1993

0.5 47 61 76 99 137 196 300 522 1106 4359

0 . 6 43 56 71 93 126 180 277 483

0.7 40 51 64 84 114 160 246 434 922

0.8 34 44 56 73 99 140 213 370 794

1 . 0 0 . 9 0.8 0.7 0.6 0.5 0.4 0.3 0.2 0 . 1

VOL 19, NO. 2, 1993

VOL 19, NO. 2, 1993

Linkage studies of schizophreniaContinued

VOL 19, NO. 2, 1993

VOL 19, NO. 2, 1993

VOL 19, NO. 2, 1993

VOL. 19, NO. 2, 1993

VOL 19, NO. 2, 1993

genotyping microsatellite loci. Genomics, 14:1026-1031, 1992.

Acknowledgments This research was supported by USPHS grants MH-41953 and

You might also like

Pfad - The Proxy pFad of © 2024 Garber Painting. All rights reserved.