Fetal programming effects of testosterone on the reward system and behavioral approach tendencies in humans

Abstract

Background: Sex differences are present in many neuropsychiatric conditions that affect emotion and approach-avoidance behavior. One potential mechanism underlying such observations is testosterone in early development. Although much is known about the effects of testosterone in adolescence and adulthood, little is known in humans about how testosterone in fetal development influences later neural sensitivity to valenced facial cues and approach-avoidance behavioral tendencies.

Methods: With functional magnetic resonance imaging we scanned 25 8-11-year-old children while viewing happy, fear, neutral, or scrambled faces. Fetal testosterone (FT) was measured via amniotic fluid sampled between 13 and 20 weeks gestation. Behavioral approach-avoidance tendencies were measured via parental report on the Sensitivity to Punishment and Sensitivity to Rewards questionnaire.

Results: Increasing FT predicted enhanced selectivity for positive compared with negatively valenced facial cues in reward-related regions such as caudate, putamen, and nucleus accumbens but not the amygdala. Statistical mediation analyses showed that increasing FT predicts increased behavioral approach tendencies by biasing caudate, putamen, and nucleus accumbens but not amygdala to be more responsive to positive compared with negatively valenced cues. In contrast, FT was not predictive of behavioral avoidance tendencies, either through direct or neurally mediated paths.

Conclusions: This work suggests that testosterone in humans acts as a fetal programming mechanism on the reward system and influences behavioral approach tendencies later in life. As a mechanism influencing atypical development, FT might be important across a range of neuropsychiatric conditions that asymmetrically affect the sexes, the reward system, emotion processing, and approach behavior.

Figure 1

Correlations between behavioral approach system (BAS) score across all subscales, behavioral inhibition system (BIS) punishment subscale, and fetal testosterone (FT). This figure presents correlation matrices representing (A) zero-order correlations and (B) partial correlations after partialing out chronological age across FT, BIS, and all BAS subscales. BASd, BAS drive subscale; BASi, BAS impulsivity/fun-seeking subscale; BASr, BAS reward responsivity subscale.

Figure 2

Association between neural response to valenced information (Happy > Fear) and fetal testosterone (FT). This figure shows areas within the striatum and amygdala where Happy > Fear activation is positively correlated with FT. Numbers indicate z-slice coordinate in Montreal Neurological Institute space.

Figure 3

Path diagrams of relationships between fetal testosterone (FT), neural mediators in the striatum or amygdala, and total behavioral approach system (BAS) score summing across all subscales. (A) Path diagram when nucleus accumbens (NAcc) Happy > Fear response is the mediator between FT and BAS. (B) Path diagram when putamen Happy > Fear response is the mediator between FT and BAS. (C) Path diagram when caudate Happy > Fear response is the mediator between FT and BAS. (D) Path diagram when amygdala (Amyg) Happy > Fear response is the mediator between FT and BAS. Path a is the relationship between the predictor (FT) and the mediator (region of interest [ROI]). Path b is the relationship between the mediator (ROI) and the outcome (BAS), controlling for the predictor (FT). Path c' is the relationship between the predictor (FT) and the outcome (BAS) controlling for the mediator (ROI). Path c is the total effect of the relationship between the predictor (FT) and the outcome (BAS), irrespective of the mediator. Path a*b is the difference between path c and path c'. Path coefficients and standard errors (in parentheses) are noted for each path. *p < .05, **p < .01.