Twin study refining psychotic symptom dimensions as phenotypes for genetic research

Abstract

We investigated which psychotic symptom dimensions are likely to be most useful as phenotypes for genetic linkage and association studies. Two hundred twenty-four probandwise twin pairs (106 monozygotic,118 same-sex dizygotic), where probands had psychosis, were ascertained from the Maudsley Twin Register in London. Dimensions were defined as ordinal symptom scores using the OPCRIT checklist, based on previous factor analyses of this and other samples. To qualify as a potentially useful phenotype, dimensions had to show (a) a satisfactory polychoric model fit and significant within-pair correlation in MZ pairs concordant for DSM-III-R psychosis and (b) that they could not be better defined in terms of other combinations of relevant symptoms, single symptoms, or dichotomized dimension scores. Relationships between dimension scores in twin probands and risk of psychosis in co-twins were also investigated. None of the positive or negative dimensions satisfied the phenotypic criteria. The disorganized dimensions showed significant and substantial correlations which maximized for a narrow definition. Combined negative/disorganized dimensions also showed significant and substantial correlations, but did not have advantages over disorganized dimensions. None of the dimensions were significant predictors of psychosis risk in co-twins. We conclude that, of the dimensions analyzed, the narrow disorganized dimension shows most promise as a phenotype for molecular genetic research.

Heritability estimates for psychotic symptom dimensions in twins with psychotic disorders

“…However, the point estimate of 0.34 was similar to that found in affected sib-pairs using the same symptom dimension definition (Spearman r ¼ 0.25) [Cardno et al, 1999a] and closer still using the same statistical approach (PCC ¼ 0.30, CI 0.07-0.51: calculated by AG Cardno), and also similar using other rating scales, analytical approaches and samples (0.19-0.32) [Burke et al, 1996;Loftus et al, 1998;Cardno et al, 1999a;Rietkerk et al, 2008]. The sib-pairs had a narrower range of phenotypes-schizophrenia and schizoaffective disorder-but in this twin sample correlations for the disorganized dimension are similar in MZ pairs concordant for psychotic disorder, and where one or both twins has schizophrenia [Cardno et al, 2008] (there are too few DZ twin pairs concordant for schizophrenia for this comparison). Caveats are the nonsystematic ascertainment of the sib-pair samples and the theoretical possibility of differences in familial aggregation between DZ twins and siblings.…”

“…Initial analysis of the Maudsley twin series [Cardno et al, 2001[Cardno et al, , 2008 showed similar correlations for positive dimension definitions in affected monozygotic (MZ) twin pairs (up Neuropsychiatric Genetics to 0.35), but trends towards higher correlations for some definitions of the negative and disorganized dimensions (up to 0.77). This suggests that there may be notable genetic influences on at least some dimensions, but these analyses did not include formal heritability estimates (see Supplementary Material).…”

“…Symptom dimensions were defined as ordinal symptom scores, based on the most characteristic symptoms found in previous factor analyses of global psychotic symptoms [Liddle, 1987;Andreasen et al, 1995;Peralta et al, 1997;Ratakonda et al, 1998], as previously employed in affected sib-pair analysis [Cardno et al, 1999a] and initial analysis of this twin sample [Cardno et al, 2001[Cardno et al, , 2008. All symptoms were coded as present or absent and we did not include mood symptoms or illness history variables.…”

“…The positive dimension comprised (a) any delusions, and (b) any hallucinations [Cardno et al, 2008], and was scored as 0 if neither symptom was present, 1 if one was present, and 2 if both were present. The negative dimension comprised negative formal thought disorder and restricted affect.…”

“…Analysis needs to take into account the fact that symptom dimensions are usually only measured in individuals with a psychotic disorder, so relatives without psychosis have unknown symptom dimension scores. In this case, genetic influences on a symptom dimension may be viewed as reflecting the level of genetic liability to psychosis and/or being due to modifying factors independent of psychosis liability [Cardno et al, 2001[Cardno et al, , 2008. These two types of influence are expected to result in different patterns of familial aggregation, and analysis needs to take both potential influences into account.…”