Autism-associated biomarkers: test-retest reliability and relationship to quantitative social trait variation in rhesus monkeys.

Background: Rhesus monkeys (Macaca mulatta) exhibit pronounced individual differences in social traits as measured by the macaque Social Responsiveness Scale-Revised. The macaque Social Responsiveness Scale was previously adapted from the Social Responsiveness Scale, an instrument designed to assess social and autistic trait variation in humans. To better understand potential biological underpinnings of this behavioral variation, we evaluated the trait-like consistency of several biological measures previously implicated in autism (e.g., arginine vasopressin, oxytocin, and their receptors, as well as ERK1/2, PTEN, and AKT(1-3) from the RAS-MAPK and PI3K-AKT pathways). We also tested which biological measures predicted macaque Social Responsiveness Scale-Revised scores.

Methods: Cerebrospinal fluid and blood samples were collected from N = 76 male monkeys, which, as a sample, showed a continuous distribution on the macaque Social Responsiveness Scale-Revised. In a subset of these subjects (n = 43), samples were collected thrice over a 10-month period. The following statistical tests were used: "Case 2A" intra-class correlation coefficients of consistency, principal component analysis, and general linear modeling.

Results: All biological measures (except AKT) showed significant test-retest reliability within individuals across time points. We next performed principal component analysis on data from monkeys with complete biological measurement sets at the first time point (n = 57), to explore potential correlations between the reliable biological measures and their relationship to macaque Social Responsiveness Scale-Revised score; a three-component solution was found. Follow-up analyses revealed that cerebrospinal fluid arginine vasopressin concentration, but no other biological measure, robustly predicted individual differences in macaque Social Responsiveness Scale-Revised scores, such that monkeys with the lowest cerebrospinal fluid arginine vasopressin concentration exhibited the greatest social impairment. Finally, we confirmed that this result held in the larger study sample (in which cerebrospinal fluid arginine vasopressin values were available from n = 75 of the subjects).

Conclusions: These findings indicate that cerebrospinal fluid arginine vasopressin concentration is a stable trait-like measure and that it is linked to quantitative social trait variation in male rhesus monkeys.