A range of endophenotypes characterise psychosis, however there has been limited work understanding if and how they are inter-related.

This multi-centre study includes 8754 participants: 2212 people with a psychotic disorder, 1487 unaffected relatives of probands, and 5055 healthy controls. We investigated cognition [digit span (N = 3127), block design (N = 5491), and the Rey Auditory Verbal Learning Test (N = 3543)], electrophysiology [P300 amplitude and latency (N = 1102)], and neuroanatomy [lateral ventricular volume (N = 1721)]. We used linear regression to assess the interrelationships between endophenotypes.

The P300 amplitude and latency were not associated (regression coef. −0.06, 95% CI −0.12 to 0.01, p = 0.060), and P300 amplitude was positively associated with block design (coef. 0.19, 95% CI 0.10–0.28, p < 0.001). There was no evidence of associations between lateral ventricular volume and the other measures (all p > 0.38). All the cognitive endophenotypes were associated with each other in the expected directions (all p < 0.001). Lastly, the relationships between pairs of endophenotypes were consistent in all three participant groups, differing for some of the cognitive pairings only in the strengths of the relationships.

The P300 amplitude and latency are independent endophenotypes; the former indexing spatial visualisation and working memory, and the latter is hypothesised to index basic processing speed. Individuals with psychotic illnesses, their unaffected relatives, and healthy controls all show similar patterns of associations between endophenotypes, endorsing the theory of a continuum of psychosis liability across the population.

Morphometric endophenotypes which have been proposed for psychotic disorders include lateral ventricular enlargement and hippocampal volume reductions. Genetic epidemiological studies support an overlap between schizophrenia and bipolar disorder, and COMT, BDNF, 5-HTT, NRG1 and DTNBP1 genes have been implicated in the aetiology of both these disorders. This study examined associations between these candidate genes and morphometric endophenotypes for psychosis.

A total of 383 subjects (128 patients with psychosis, 194 of their unaffected relatives and 61 healthy controls) from the Maudsley Family Psychosis Study underwent structural magnetic resonance imaging and genotyping. The effect of candidate genes on brain morphometry was examined using linear regression models adjusting for clinical group, age, sex and correlations between members of the same family.

The results showed no evidence of association between variation in COMT genotype and lateral ventricular, and left or right hippocampal volumes. Neither was there any effect of the BDNF, 5-HTTLPR, NRG1 and DTNBP1 genotypes on these regional brain volumes.

Abnormal hippocampal and lateral ventricular volumes are among the most replicated endophenotypes for psychosis; however, the influences of COMT, BDNF, 5-HTT, NRG1 and DTNBP1 genes on these key brain regions must be very subtle if at all present.