24 June 2009. You might call it the “divide and conquer strategy” for finding the genes responsible for schizophrenia; it involves divvying up the manifestations of the illness into related groupings and then seeking the genes that contribute to each grouping. In a paper in the June Archives of General Psychiatry, Ann E. Pulver and others at Johns Hopkins University in Baltimore, Maryland, take steps in that direction. They sought to identify the dimensions that tie together various signs and symptoms of schizophrenia, including indicators of social functioning, occupational performance, and prodromal features. All nine of the dimensions that they found, including some novel ones, tended to cluster in families.
Some researchers argue that the complexity of schizophrenia is hampering the search for its causes, and that breaking the disorder down into more homogeneous parts, at the level of phenotypes or the even more basic level of endophenotypes, may speed progress (see SRF live discussion; see SRF live discussion; see SRF related news story). There has been some research on discerning the factor structure of schizophrenia, with inconsistent results (for a review, see Peralta and Cuesta, 2001; for more recent studies, see Villalta-Gil et al., 2006; Emsley et al., 2003). In tune with that thinking, those involved in the new study, including first author John A. McGrath, used factor analysis to unearth leads for genetic studies.
According to McGrath and colleagues, schizophrenia “is likely several related disorders with varying and sometimes overlapping genetic underpinnings, some of which affect clinical and course features.” This view led them to examine an unusually broad set of schizophrenia manifestations. Specifically, they tested a set of 73 items, which went beyond positive symptoms, negative symptoms, affect, and disorganized thoughts and behavior. They also included measures of disturbed social functioning during childhood and adolescence, social and occupational decline, and academic performance.
The data came from 1,199 people who met criteria for schizophrenia or schizoaffective disorder. Most were living in the United States, although a few subjects from Canada, Greece, Italy, Poland, and Israel also participated. The sample included 513 outbred subjects and 686 Jews of European ancestry. Most of the outbred subjects were of white European descent; 6 percent were African American. Nearly all of the Jews were of Ashkenazi background.
Information on each subject’s signs, symptoms, and psychiatric history came from clinical interviews, medical records, and proxy reports. At least two diagnosticians reviewed the data to render consensus judgments regarding diagnoses and other key information.
Something old, something new
The researchers found nine dimensions that explain 35 percent of the variance in their data, a smaller percentage of the variance than found in other studies (e.g., Villalta-Gil et al., 2006; Emsley et al., 2001). McGrath and colleagues attribute this, in part, to the way they imputed missing data.
Five of the factors echo those in many other factor analytic studies of schizophrenia, including those tapping hallucinations, negative symptoms, affective symptoms, disorganization, and Schneiderian first-rank symptoms. The last reflects signs and symptoms that psychiatrist Kurt Schneider (Schneider, 1974) considered central to schizophrenia, such as hearing one’s thoughts broadcast to other people, attributing one’s thoughts to insertion by others, and other forms of delusions.
In addition, McGrath and colleagues report four dimensions that they describe as new to factor analytic studies in schizophrenia, although they mirror known features of schizophrenia. For instance, the highest-loading items on the disability/impairment factor pertain to work functioning. That factor correlated only weakly with their scholastic factor, which includes aspects of elementary, high school, and college performance. The prodromal factor encompasses warning signs, such as role impairment, avoidance of social interactions, odd behavior, and bizarre thoughts, which may precede full-blown schizophrenia.
A final factor consisted mainly of items related to childhood and adolescent sociability. This complements past findings of an adult disordered social relations factor (e.g., Emsley et al., 2001; Toomey et al., 1997).
In this study, depression and mania formed one factor rather than separating into two, perhaps due to its inclusion of relatively few mood items, McGrath and colleagues speculate. Furthermore, their affective factor correlated only weakly and inversely with the negative symptoms dimension, which, they write, suggests that they arise independently.
Factoring in family ties
Of course, whether these dimensions will speed the hunt for elusive schizophrenia genes depends, in part, on their heritability. While McGrath and associates did not design their study to test heritability per se, they were able to probe whether the factors they found clustered in families. They used a heritability measure, the ratio of polygenic variance to total variance, to gauge familiality, although they caution against using it to infer heritability in the general population.
To map out family ties of the 1,199 subjects, the study used information on 553 of their family members. Most families (714) had only one member with schizophrenia, but 207 had more than one. The results hint that all of the nine factors run in families. In particular, disability/impairment showed the highest heritability at 0.61, with disorganization coming in at a nearly identical at 0.60. Child and adolescent sociability showed the lowest at 0.27. Heritability for the other factors ranged from 0.36 to 0.53.
These findings contrast with those from a recent review and meta-analysis (Rietkerk et al., 2008). It evaluated the evidence for genetic contributions to three dimensions—namely, reality distortion, as evidenced by hallucinations and delusions; psychomotor poverty, manifested by flat affect, scant speech, and decreased spontaneous movement; and disorganization, which includes formal thought disorder, inappropriate affect, and strange behavior. Only disorganization, one of the most familial factors in the McGrath study, showed clear evidence of genetic influence.
Having generated factors that they hoped would provide leads for genetic studies, the researchers are putting them to the test. They are now using the factor scores as phenotypes in quantitative trait locus linkage and association studies.—Victoria L. Wilcox.
McGrath JA, Avramopoulous D, Lasseter VK, Wolyniec PS, Fallin MD, Liang K-Y, Nestadt G, Thornquist MH, Luke JR, Chen P-L, Valle D, Pulver AE. Familiality of novel factorial dimensions of schizophrenia. Arch Gen Psychiatry. 2009 June; 66(6):591-600. Abstract