Nilsson KW, Åslund C, Comasco E, Oreland L
J Neural Transm (Vienna) 125 (11) 1601-1626 [2018-11-00; online 2018-06-07]
Since the pioneering finding of Caspi and co-workers in 2002 that exposure to childhood maltreatment predicted later antisocial behaviour (ASB) in male carriers of the low-activity MAOA-uVNTR allele, frequent replication studies have been published. Two meta-analyses, one in 2006 and the other in 2014, confirmed the original findings by Caspi and co-workers. In the present paper, we review the literature, note some methodological aspects of candidate gene-environment interaction (cG×E) studies and suggest some future directions. Our conclusions are as follows. (1) The direction of the effect in a cG×E model may differ according to the positive and negative environmental background of the population. (2) There is a predictor-intersection problem such that when measuring one type of maltreatment in a person, other kinds of maltreatment often co-occur. Other forms of abuse are implicitly considered in statistical models; therefore, it is difficult to draw conclusions about the effects of timing and the severity of different forms of stressful life events in relation to ASB. (3) There is also an outcome-intersection problem because of the major intersection of ASB and other forms of mental health problems. It is likely that the G×E with MAOA is related to a common unmeasured factor. (4) For the G×E model, in which the effect of the gene on the outcome variable is dependent on other predictor variables, theoretically, hypothesis-driven statistical modelling is needed.