Please cite this article as follows: Myles-Worsley M, Tiobech J, Blailes F, Yano VM, Faraone SV. 2007. Recurrence Risk to Offspring in Extended Multiplex Schizophrenia Pedigrees from a Pacific Island Isolate. Am J Med Genet Part B 144B:41–44.
The etiological importance of genetic transmission in the pathogenesis of schizophrenia has been well established. Schizophrenia clearly aggregates in families. First-degree relatives of schizophrenia patients have a 10–15% risk of developing the illness compared to a general population expectancy of about 1% [Gottesman et al., 1982, 1987, 1991].
McGue et al.  calculated the risk for schizophrenia in relatives based on a Western European family data set, which included 40 studies conducted from 1920 to 1987. The data include the Kay and Lindelius  study of 4,000 relatives in Sweden and the Bleuler  study of 3,000 relatives in Switzerland. The recurrence risk for definite plus probable schizophrenia was 12.84% for offspring (N = 1254) and 3.46% for nieces and nephews (N = 2973). These risk estimates for relatives of schizophrenics, as summarized by Gottesman  in “Schizophrenia Genesis,” have stood the test of time and have been widely cited in the literature. The diagnostic category “definite plus probable schizophrenia” corresponds with the narrow definition of schizophrenia used in the field today because “a probable diagnosis” included individuals who had “most of the signs and symptoms of schizophrenia, were psychotic, and were more likely to be schizophrenic than to have an affective psychosis” [McGue et al., 1983, page 1166].
These Western European families included multiplex families, but did not focus on the extended schizophrenia pedigrees that have been ascertained in genetic isolates such as Finland. In genetically isolated populations, well-documented, extended and multi-generational pedigrees with multiple cases of heritable diseases can be found, and these families represent a powerful resource for understanding the transmission of inherited disorders like schizophrenia [Peltonen, 2000; Peltonen et al., 2000; Arcos-Burgos and Muenke, 2002].
One such population isolate is the Republic of Palau where we are studying the genetic etiology of schizophrenia. Palau is a remote island nation in Micronesia that has remained geographically and ethnically isolated for about 2,000 years, but is not inbred. During the epidemiological phase of our study, we found an elevated lifetime prevalence of schizophrenia with cases clustering in large multi-generational families [Myles-Worsley et al., 1999]. These findings suggest enrichment of susceptibility genes for schizophrenia and increased risk for developing schizophrenia in the offspring of these families.
Complete ascertainment of cases and families in Palau found 160 cases of narrowly defined schizophrenia, which included only chronic schizophrenia and schizoaffective disorder, and 55 cases of other psychotic disorders, predominantly affective psychotic disorders [Myles-Worsley et al., 1999]. The lifetime prevalence in the Palau population was 1.99% (2.77% for males versus 1.24% for females) for narrowly defined schizophrenia, and 2.67% (3.42% for males versus 1.95% for females) for broadly defined schizophrenia. The higher prevalence in males was accompanied by an earlier mean age at onset for males (23.3 years) than for females (27.5 years), a finding that was consistent with prior studies [e.g., Faraone et al., 1994].
The multi-factorial model of schizophrenia proposed by Gottesman and Shields  and supported by family and segregation analysis studies [Faraone and Tsuang, 1985, 1988; Risch, 1990], predicts greater genetic risk with increasing number of affected relatives. This model posits many genes combined with adverse environmental factors create a graded genetic predisposition such that the probability of developing schizophrenia (or a genetically related condition) increases as the genetic predisposition increases [McGue et al., 1983; Gottesman et al., 1987; Gottesman, 1991].
Smith , an animal-breeding geneticist, developed recurrence risk estimates for offspring in multiply affected pedigrees based on the multi-factorial model of disease inheritance. Gottesman  applied these risk estimates to schizophrenia and demonstrated that risk to offspring in multiplex pedigrees depends upon the number of both affected and unaffected relatives and their relationship to other affected family members. For example, recurrence risk to offspring of one affected parent can range from 7.6% if there is at least one other unaffected sibling to 38.0% if the child has two affected siblings and two affected 2nd-degree relatives on the side opposite from the affected parent. These estimates suggest that risk for schizophrenia in Palauan offspring with multiple affected relatives distributed over an extended pedigree may be well above the levels found in smaller Western European families.
The objective of the present study was to estimate risk in the youngest generation of the Palauan schizophrenia families based on empirical recurrence rates in previous generations. We hypothesized that recurrence risk in Palauan offspring with multiple affected relatives would be significantly higher than the risks found in smaller schizophrenia pedigrees from more genetically heterogeneous populations.
In the initial epidemiological phase of the Palau study, 262 individuals were evaluated for psychotic illness based on medical records or referral by a family member. Subjects were interviewed by a board certified, US-trained psychiatrist using a modified version of the Schedule for Affective Disorders and Schizophrenia-Lifetime Version [SADS-L, Endicott and Spitzer, 1978]. The SADS-L interviews were supplemented by a review of psychiatric medical records. Two independent board certified psychiatrists conducted blind reviews of the information compiled for each subject and established consensual diagnoses according to RDC criteria [Spitzer et al., 1978]. Narrowly defined schizophrenia included chronic and subchronic schizophrenia (n = 140) and schizoaffective disorder, mainly schizophrenic course (n = 20). Other psychotic illnesses included in the broad diagnostic classification were acute and subacute schizophrenia (n = 5), schizoaffective disorder, mainly affective course (n = 6), bipolar I disorder with psychotic symptoms (n = 22), psychotic depression (n = 3), and unspecified functional psychosis (n = 19).
Study procedures were approved by the Institutional Review Boards of the University of Utah and the Republic of Palau. All participants provided written informed consent after receiving a thorough explanation of the study in both English and Palauan from a Palauan co-investigator (J.T.).
The present study examined affection rates in the generation aged 40 years and older and their parents, all of whom were completely ascertained during the epidemiological phase of the study. Affected nuclear families were included only if all offspring had already reached 40 years of age, which was considered an appropriate cutoff for age of onset in this population. Nuclear families with no affected members were not included.
Since the epidemiological data were compiled in 1997, additional cases of schizophrenia spectrum disorders have been diagnosed and several families have been connected at the founder level. Based on this new data, we have now identified 168 (233) cases of narrowly (broadly) defined schizophrenia representing 59 separate families, each of which was defined by a single common founder.
Cases of schizophrenia cluster in families. Although we identified 59 separate families with broadly defined schizophrenia, 20 of these families have four or more cases, and these 20 “high density” families contain 80% of all cases of illness in Palau. Only four of the 233 cases have not yet been connected to another schizophrenia patient, indicating that these may be non-familial sporadic cases of the illness.
Many of the 233 cases of broadly defined schizophrenia can be linked together via marriage. Figure 1 shows a composite pedigree illustrating the complex multi-lineal relationships that result from marriage connections among six separate pedigrees with a total of 53 affected individuals. Although schizophrenia is clearly aggregating in these six families, the average genetic relationship between affected individuals is only a first cousin relationship [Camp et al., 2001]. Recently, Klei et al.  constructed an even larger pedigree of this type comprising 2,809 individuals and 126 cases of illness for a genetic linkage study of the Palauan families defined by the initial epidemiological study. Despite its size and its numerous marriage loops, this large pedigree contained no inbreeding loops, and the average degree of relationship between affected individuals was between the fifth and sixth degrees.
The relationship between affected individuals may, however, be moving closer as additional cases are ascertained. Many of the more recently identified cases have at least one affected sibling. Based on a broad definition of schizophrenia, we have now identified 21 affected sib-pairs, eight sib-trios, one sibship with four affected members, and seven half-sib-pairs. Schizophrenic patients with at least one affected sibling or half-sibling (n = 86) now account for 36.9% of all broadly defined cases. Multiply affected sibships occur in 15 of the 20 “high density” families.
Recurrence Risk in Offspring of an Affected Parent
In Palau, 63 affected parents (32 male and 31 female) had a total of 120 offspring, while 23 patients (16 male and 7 female) had no children. Only 2 of the 20 affected parents with progeny had no affected offspring. These 120 offspring include all children of the three affected parents who had more than one mate, even though these additional matings did not result in any cases of psychotic illness. To date, there are no known matings between two affected Palauan parents that have resulted in a live birth.
As shown in Table I, 23.4% (25.8%) of the offspring of affected parents were diagnosed with narrowly (broadly) defined schizophrenia. When the data were examined by gender, the recurrence risks for narrowly (broadly) defined schizophrenia in offspring rose to 27.9% (30.9%) in sons compared to only 17.7% (19.2%) in daughters, a reflection of the significantly elevated lifetime morbid risk for males versus females in Palau. These recurrence rates for offspring in Palauan families yield relative risk estimates for narrowly (broadly) defined schizophrenia of 11.8 (9.7) as shown in Table III.
Table I. Affection Rates in Offspring of a Parent With Narrowly (Broadly) Defined Schizophrenia
Fathers' progeny (N = 32)
Mothers' progeny (N = 31)
Total parents' progeny (N = 63)
Number of progeny affected
Recurrence Risk in Nieces and Nephews
Our sample of nieces and nephews was restricted in size because offspring of unaffected parents were ascertained only in a subset of the Palauan pedigrees. To reach an adequate sample size, we included the offspring of unaffected parents with an affected half sibling. Based on a sample of 292 nieces and nephews of schizophrenia probands, the recurrence risk for narrowly (broadly) defined schizophrenia in these second-degree relatives was 7.8% (7.9%) as shown in Table II. However, recurrence risk for narrowly defined schizophrenia was significantly higher in nieces/nephews with two or more affected aunts/uncles (15.0%) than in nieces/nephews with only one affected aunt or uncle (6.4%). Table III shows that relative risk for narrow schizophrenia in the entire Palauan niece/nephew sample was 3.9, similar to the 3.5 value for nieces/nephews in Western European families.
Table II. Affection Rates in the Nieces/Nephews of Patients With Narrowly (Broadly) Defined Schizophrenia
Number of nieces/nephews
Offspring of an unaffected parent with:
One affected full-sibling
One affected half-sibling
Only one affected full or half sibling
Two or more affected full or half siblings
Table III. Summary Comparison of Risk for Schizophrenia in Relatives
The goal of this study was to estimate risk for schizophrenia in offspring who belong to extended multi-generational pedigrees from an isolated population based on actual recurrence rates in relatives. Recurrence rates in the offspring and nieces/nephews of Palauan schizophrenia patients were approximately double the rates reported by McGue et al.  for a combined sample of smaller Western European families. However, estimates of relative risk, the ratio of risk in relatives to the population risk [Risch, 1990], were comparable, indicating that higher recurrence rates in offspring partially reflect the elevated lifetime morbid risk for schizophrenia in the Palau population.
Additional ascertainment since our original epidemiological paper indicates stronger familial aggregation than previously reported [Myles-Worsley et al., 1999]. New connections between families at the founder level have been revealed, and now 20 of the 59 families contain 80% of broadly defined schizophrenia patients in Palau. Furthermore, as new cases have been ascertained, the number of multiply affected sibships has increased, and this has raised rates of illness in many affected pedigree branches.
Our results support the estimates of recurrence risk for a multi-factorial trait developed by Smith . We found that recurrence risk for nieces/nephews of schizophrenia patients more than doubled when the affected aunt/uncle had at least one other affected sibling, confirming Smith's  prediction that risk to offspring increases with the number of close affected relatives. Because risk to offspring is significantly higher if a parent belongs to a sibship with two or more affected members versus only one affected member, the presence of 30 multiply affected sibships and 7 half-sib-pairs in the current generation of parents may lead to further elevation of risk levels in their offspring.
Another important characteristic of the Palau pedigrees that elevates risk levels for offspring is the frequency of affected relatives on both the maternal and paternal side of the family. Although Palauan doctrine has forbidden marriage between close relatives for centuries, endogamous marriage has been encouraged, and this has led to numerous marriage connections between Palauan schizophrenia families, as illustrated in Figure 1. The resulting series of bilineal transmissions across successive generations has created complex multi-lineal patterns of disease inheritance. According to Smith's  calculations, risks to offspring increase with the presence of affected relatives on the side opposite from the affected parent or aunt/uncle. Thus, the multi-lineal pattern of disease transmission in the Palauan pedigrees could continue to increase risk in future generations.
In summary, our results suggest that the high recurrence risk levels for offspring in past generations reflect the elevated population prevalence, strong familial aggregation, and a multi-lineal transmission pattern of schizophrenia in Palau. Future studies will be able to compare the recurrence rates for offspring in previous generations with empirical affection rates in the most recent generation of young Palauans, data that will eventually be provided by our ongoing Palau Early Psychosis Study (PEPS) of high risk adolescents [Myles-Worsley et al., 2006].
The authors thank all members of the Palauan community who assisted in this research, which was supported by US Public Health Service grant MH54186, a Grable Foundation Award from NARSAD, and a Distinguished Investigator Award from NARSAD.