Inverse Relationship of Perinatal Complications and Eye Tracking Dysfunction in Relatives of Patients With Schizophrenia: Evidence for a Two-Factor Model

The prevalence of dysfunction in smooth pursuit eye tracking is significantly elevated in patients with schizophrenia and their relatives, including mono~zygotic co-twins (1–3). Genetic modeling of data from the families of patients with schizophrenia suggests that eye tracking dysfunction is a sensitive behavioral indicator of an autosomal dominant gene that markedly increases the risk for schizophrenia (4). The model suggests that approximately 53% of gene carriers have eye tracking dysfunction, whereas less than 10% develop schizophrenia. Rates of obstetrical complications, particularly perinatal ones, are also higher in patients with schizophrenia than in comparison subjects (5, 6), including siblings of patients with schizophrenia (6) and monozygotic co-twins.

A two-factor model of schizophrenia is suggested by inverse associations between neurological signs and psychopathology found in the nonschizophrenic relatives of patients with schizophrenia (7). This model proposes that two major etiologic factors run in the families of patients with schizophrenia, and increase the risk for schizophrenia markedly when they occur together, but are statistically independent of each other in the general population.

This model predicts that, although patients with schizophrenia have elevated levels of each risk factor, the joint presence of these two factors discriminates patients with schizophrenia particularly well from subjects without schizophrenia, including relatives of patients with schizophrenia. More important, it predicts that nonschizophrenic relatives of patients with schizophrenia will have significantly higher rates of individuals with one or the other—but not both—risk factors (7). Because the two risk factors are distributed independently, measures associated with one or the other underlying risk factor will be inversely related among subjects without schizophrenia, particularly in relatives of patients with schizophrenia; Matthysse provided a mathematical proof that this is a logical corollary of the model (8).

Findings on perinatal obstetrical complications and Trail Making, a neuropsychological measure, fit the model well: the combination of perinatal obstetrical complications and poor Trail Making performance significantly discriminated patients with schizophrenia from subjects without schizophrenia, and poor Trail Making and perinatal obstetrical complications were significantly and inversely related among the nonschizophrenic siblings of patients with schizophrenia (8). Complementary research (5) found that the adult offspring of patients with schizophrenia had fewer obstetrical complications if the offspring were diagnosed as having borderline schizophrenia rather than psychiatrically normal, as expected if obstetrical complications often interact with genetic liability to produce schizophrenia (9).

Eye tracking dysfunction is of particular interest for the two-factor model because research suggests that eye tracking dysfunction indexes genetic liability for schizophrenia (1–4). We hypothesized that 1) the combination of eye tracking dysfunction and perinatal obstetrical complications discriminates patients with schizophrenia significantly from subjects without schizophrenia, including relatives of patients with schizophrenia, and 2) eye tracking dysfunction and obstetrical complications are inversely related among subjects without schizophrenia, particularly siblings of patients with schizophrenia.

METHOD

Probands with schizophrenia and mood disorders were recruited from patients consecutively admitted to state and private inpatient psychiatric hospitals. Diagnoses were made by experienced investigators who were blind to obstetrical and eye tracking data and who used DSM-III or DSM-III-R criteria as well as information from chart reviews, family informants, and, for most subjects, structured interviews. Written informed consent was obtained from all study participants after study procedures were fully explained. Data on eye tracking dysfunction, diagnosis, and obstetrical complications were available for 83 adults: 18 patients with schizophrenia, 16 of their nonschizophrenic siblings from nine families, and 49 comparison subjects with no personal or family history of schizophrenia. Comparison subjects included 26 normal subjects (normal probands and their well relatives) and 23 others, including four probands with major depression, three psychiatrically ill relatives of comparison probands, and 16 first-degree relatives of probands with bipolar disorder or major depression.

Mean ages and male-female ratios were 29.0 years (SD=5.8) and 16:2 for patients with schizophrenia, 32.1 years (SD=7.9) and 8:8 for the siblings of patients with schizophrenia, and 26.5 years (SD=7.2) and 22:27 for all other subjects. All subjects were born in the United States and, except for three African American comparison subjects, all were Caucasian. Mean illness duration for patients with schizophrenia was 6.2 years (SD=3.7).

Procedures for recording eye tracking dysfunction were standard ones used in eye tracking studies and are described elsewhere (2, 4). Eye tracking records were blindly reviewed by two independent raters, who assigned qualitative scores of normal or abnormal. Obstetrical data in maternity hospital records on subjects' gestations and births were scored blind to diagnostic and eye tracking dysfunction data; the raters applied published scales developed by an eminent obstetrician and used in studies by various investigators (e.g., references 5 and 6). The scale generated a summary score for each subject that was the algebraic sum of all labor and delivery complications present, with more severe complications weighted more heavily. Scoring with these scales yields satisfactory interrater agreement in our laboratory (6). Spearman rank-order correlations (and two-tailed tests) were used to investigate relations between eye tracking dysfunction and obstetrical complication scores, which are ordinal data.

RESULTS

The combination of eye tracking dysfunction and a higher perinatal obstetrical complication score (rating of 3.00 or greater, corresponding to more than one or two mild complications) occurred in 22% (N=4) of patients with schizophrenia but only 1.5% (one normal comparison subject) of 65 subjects without schizophrenia, who included 20 subjects with nonschizophrenic disorders (p=0.006, Fisher's exact test). Mean obstetrical complication scores were 3.00 (SD=2.95) for patients with schizophrenia and 1.75 (SD=1.53) for the 16 siblings—but only 0.50 (SD=0.84) for the six siblings with eye tracking dysfunction—and 2.16 (SD=2.04) for all 49 comparison subjects, including 2.46 (SD=2.32) for 26 normal comparison subjects and 1.83 (SD=1.67) for the 23 other comparison subjects.

Spearman correlations between eye tracking dysfunction and perinatal obstetrical complication scores were, respectively, r_s=–0.07 (n.s., N=18) for patients with schizophrenia, r_s=–0.22 (p<0.05, N=65) for all subjects without schizophrenia, and r_s=–0.66 (p=0.005, 95% confidence interval=–0.24 to –0.87, N=16) for the nonschizophrenic siblings of patients with schizophrenia. These correlations did not reflect demographic variables, which were not significantly correlated with eye tracking dysfunction or obstetrical complications.

DISCUSSION

The combination of eye tracking dysfunction and a high perinatal obstetrical complication score significantly discriminated patients with schizophrenia from subjects without schizophrenia, including siblings of patients with schizophrenia. Moreover, eye tracking dysfunction and perinatal obstetrical complications were significantly and negatively correlated among the nonschizophrenic siblings of patients with schizophrenia—a distinctive prediction of the two-factor model of schizophrenia.

These results complement evidence for a two-factor model from research on patients with schizophrenia involving other variables, including neurological signs (7) and neuropsychological deficits (8, 10). The findings are thus theoretically intriguing but need replication.

Received March 31, 1997; revision received Jan. 20, 1998; accepted Jan. 30, 1998. From the Department of Psychiatry, Harvard Medical School, Boston, and the Laboratories for Psychiatric Research, the Psychology Research Laboratory, and the Brain Imaging Center, McLean Hospital. Address reprint requests to Dr. Kinney, McLean Hospital, NBG28, 115 Mill St., Belmont, MA 02178. Supported by NIMH grants MH-31154, MH-49487, MH-31340, and MH-01021. The authors thank Steven Matthysse, Ph.D., for advice and Rebecca Bloom, Clara Lajonchere, Christian LaVancher, and David Medoff for assistance.