MAOA, maltreatment, and gene–environment interaction predicting children's mental health: new evidence and a meta-analysis


Previous research on adults has shown that a functional polymorphism in the promoter region of the monoamine oxidase A (MAOA) gene moderates the impact of childhood maltreatment on risk for developing antisocial behavior. Thus far, attempts to replicate this finding have been mixed. The current study (i) presents new data investigating this finding in a sample of 975 seven-year-old boys, and (ii) evaluates the extant data by conducting a meta-analysis of published findings. We replicated the original finding by showing that the MAOA polymorphism moderates the development of psychopathology after exposure to physical abuse, we extended the finding to childhood closer in time to the maltreatment experience, and we ruled-out the possibility of a spurious finding by accounting for passive and evocative gene–environment correlation. Moreover, meta-analysis demonstrated that across studies, the association between maltreatment and mental health problems is significantly stronger in the group of males with the genotype conferring low vs high MAOA activity. These findings provide the strongest evidence to date suggesting that the MAOA gene influences vulnerability to environmental stress, and that this biological process can be initiated early in life.


Children who experience familial adversity such as physical abuse and inter-parental violence show wide variability in their mental health outcomes.1, 2 Many of these children develop behavioral and emotional difficulties, but many others are resilient and exhibit better functioning than predicted given their exposure to adversity. Recent findings3 suggest that one explanation for variability in outcomes among maltreated individuals relates to a gene–environment interaction (G × E) involving a functional polymorphism in the promoter region of the monoamine oxidase A (MAOA) gene. Specifically, maltreated children with the MAOA genotype conferring low levels of the MAOA enzyme more often developed conduct disorder, antisocial personality, and violent criminality in adulthood than maltreated children with a high-activity MAOA genotype. MAOA selectively degrades serotonin, norephinephrine, and dopamine following reuptake from the synaptic cleft, and, therefore, plays a key role in regulating behavior.4, 5 Thus far, efforts to replicate Caspi et al.'s3 results have met with mixed results6, 7, 8, 9 indicating that, as yet, accepting the MAOA by maltreatment G × E hypothesis would be premature without further investigation. The present study tested whether Caspi et al.3 original finding would replicate in a representative birth cohort sample of 7-year-old boys.

Previous studies have examined the moderating influence of the MAOA polymorphism on psychopathology years after the exposure to adversity. Mental health outcomes as a consequence of childhood adversity have been measured at mean ages ranging from 13 to 26 years.3, 6, 7 Whether the moderating effect of MAOA activity might be present in early life and closer in time to the stressful experience is unknown, yet this information is crucial for understanding the role that G × E might play in developmental processes. The present study asked whether MAOA gene activity moderates the impact of maltreatment on mental health outcomes in childhood.

A downward extension of the G × E hypothesis to young children requires a developmentally informed approach for assessing mental health.10 Based on evidence from animal5, 11 and human12, 13 studies that linked MAOA enzyme deficiency with increased aggression, Caspi et al.'s3 original G × E investigation of an adult sample targeted antisocial behavior as the hypothesized outcome. Subsequent studies6, 7, 8, 9 using adolescent samples did the same. However, longitudinal data have indicated that adolescent and adult violence and criminality have childhood origins often of the same kind (i.e., antisocial behavior), but also of a different type (e.g., anxiety, hyperactivity).14 This is because early in development, psychopathology tends to be less differentiated than in later years, when broad behavioral and emotional difficulties develop into more stable and differentiated patterns of maladaptation.15 Moreover, physical maltreatment is known to elevate risk for a variety of mental health outcomes, including but not limited to antisocial behavior.16 In childhood, the most common domains of mental health problems include antisocial behavior such as aggression and destructiveness, attention deficits, impulsivity, and hyperactivity, and emotional problems such as anxiety and social withdrawal.17 More often than not, these domains tend to co-occur in the same child18 and when they do, this signals greater severity of psychopathology as well as a poorer long-term prognosis.19 Accordingly, the combination of these mental health problems in 7-year-old boys was the focus of the present investigation.

In order to apply a rigorous test of a G × E, there must be evidence that the risk factor of interest and the outcome are linked via true environmental causation.10 Otherwise, the association between a putative ‘environmental’ pathogen and a disorder could be mediated by an unknown third variable, which could reflect unidentified genetic influences. Such a correlation between genetic susceptibility and an environmental risk variable is referred to as a genotype–environment correlation (rGE).20 The association between familial adversity and mental health outcomes might be explained by rGE of two different kinds.21 First, aggressive parents may transmit to their children both an adverse rearing environment and a genetic susceptibility toward developing psychopathology (passive rGE). Second, a child with a particular genotype may behave in ways that elicit harsh treatment (evocative rGE). With respect to maltreatment and children's antisocial behavior, evidence of a causal association exists,22 which argues against rGE as the only mediating mechanism. However, it is desirable to account for possible rGE and so far, only one G × E study involving MAOA activity and familial adversity has done so.6 In the present study, we tested for the presence of passive rGE by controlling for mothers' antisocial personality and evocative rGE by testing whether boys' MAOA genotype predicted their exposure to maltreatment.

An extension of the replication strategy is to use meta-analysis systematically to aggregate results over multiple studies, which with regard to this particular G × E hypothesis remains mixed. One possible reason for discrepancy is across-study variability in conceptualizing and measuring the putative environmental risk factor. For instance, in a positive replication by Foley et al.,6 ‘family adversity’ was measured prospectively as parental neglect, domestic violence, and harsh discipline. In the study by Haberstick et al.,7 a partial failure to replicate, maltreatment was assessed retrospectively using different indicators. Meta-analysis is one tool for ascertaining whether a finding transcends such differences across studies. Meta-analysis is an increasingly valuable method in the field of psychiatric genetics, which in recent years has been plagued by non-replications,23 leaving questions about gene effects on behavior in doubt. By pooling data from several studies, meta-analysis maximizes power to detect effects and avoids overemphasizing estimates from any single study.24 The present investigation evaluated the current state of the cumulative evidence regarding MAOA activity, childhood adversity, and G × E via a meta-analysis.

Materials and methods


Participants are members of the Environmental Risk (E-Risk) Longitudinal Twin Study. The E-Risk sampling frame was two consecutive birth cohorts (1994 and 1995) in a birth register of twins born in England and Wales.25 Of the 15 906 twin pairs born in these 2 years, 71% joined the register. Bias from non-response was corrected as follows.

The E-Risk Study probability sample was drawn using a high-risk stratification sampling procedure. High-risk families were those in which the mother had her first birth when she was 20 years of age or younger. We used this sampling (a) to replace high-risk families who were selectively lost to the register via non-response and (b) to ensure sufficient base rates of children growing up in at-risk environments. Age at first childbearing was used as the risk-stratification variable because it was recorded for virtually all families in the register, it is relatively free of measurement error, and early childbearing is a known risk factor for children's problem behaviors.26, 27 The sampling strategy resulted in a final sample in which one-third of Study mothers (younger only; N=314) constitute a 160% oversample of mothers who were at high risk based on their young age at first birth (15–20 years). The other two-thirds of Study mothers (N=802) accurately represent all mothers in the general population (aged 15–48) in England and Wales in 1994–1995 (estimates derived from the General Household Survey28). To provide unbiased statistical estimates that can be generalized to the population of British families with children born in the 1990s, the data reported in this article were corrected with weighting to represent the proportion of young mothers in that population.

The E-Risk Study sought a sample size of 1100 families to allow for attrition in future years of the longitudinal study while retaining statistical power. An initial list of families who had same-sex twins was drawn from the register to target for home visits. Of the families from the initial list, 1116 (93%) participated in home-visit assessments when the twins were aged 5 years, forming the base sample for the study: 4% of families refused and 3% could not be reached after many attempts. Written informed consent was obtained from mothers. With parent's permission, questionnaires were posted to the children's teachers, and teachers returned questionnaires for 94% of cohort children.

A follow-up home visit was conducted 18–24 months after the children's age-5 assessment. Follow-up data were collected for 98% of the 1116 E-Risk Study families. At this follow-up, teacher questionnaires were obtained for 91% of the 2232 E-Risk Study children (93% of those taking part in the follow-up). The E-Risk Study has received ethical approval from the Maudsley Hospital Ethics Committee.

The present study includes 975 of the 1092 total boys in the E-Risk Study who are of Caucasian ancestry and for whom genotypic data were available. (DNA was not available for 36 boys. These boys did not differ significantly from the rest of the sample on risk for physical abuse exposure or level of mental health problems. Genotyping failure occurred in 23 cases.) Boys' single X chromosome yields two straightforwardly characterized MAOA promoter genotypes: high activity (66.3% in this sample) and low activity (33.7%).3 Girls were excluded because, having two copies of the X chromosome, they fall into two homozygous groups, ‘high–high’ (39.1% in this sample) and ‘low–low’ (15.1%), and a third heterozygous group, ‘high–low’ (45.8%). Based on presently available evidence, girls' status on MAOA gene expression cannot be characterized with certainty because of conflicting findings regarding the inactivation status of the MAOA locus on the X chromosome.29, 30 In other words, whether the level of MAOA transcription in female subjects results from one or both copies of the MAOA gene is presently unknown and therefore, inferences about high or low MAOA activity in girls cannot be made.

DNA extraction and genotyping

At ages 5 and 7 years, DNA samples were obtained from study members via buccal swabs and extracted using a procedure described by Freeman et al.31 Primer sequences are described by Sabol et al.,4 namely MAO APT1 (5′-IndexTermACAGCCTGACCGTGGAGAAG-3′) and MAO APB1 (5′-IndexTermGAACGGACGCTCCATTCGGA-3′). MAO APT1 was 5′-labeled with the FAM fluorophore. Polymerase chain reaction (PCR) was carried out on a PTC-225 DNA engine (MJ Research, Hercules, CA, USA), using the following cycling conditions: initial 2 min denaturing step at 95°C, followed by 35 cycles of 94°C for 1 min, 55.5°C for 1 min and 72°C for 2 min, and a final extension phase of 72°C for 5 min. Reactions contained 1 × reaction Buffer IV (Abgene, Epsom, UK), 1.5 mM MgCl2, 50 ng of genomic DNA, 5 pmols of each primer, 0.3 mM dNTPs, and 1.5 U of Native Taq (Promega, Madison, WI, USA) in a volume of 10 μl. PCR products were denatured in highly deionized formamide and analyzed by electrophoresis on an Applied Biosystems 3100 genetic analyzer (Applied Biosystems, Foster City, CA, USA), set up in genotyping mode, using a POP4 polymer and ROX-labeled GS500 size standard (Applied Biosystems). Results were analyzed using GeneScan v3.7 and Genotyper v3.6 software (Applied Biosystems).

Physical abuse exposure

Children's physical abuse exposure was assessed separately for each twin at the first and follow-up assessments by interviewing mothers with the standardized clinical interview protocol from the Multi-Site Child Development Project.32 The interview protocol was designed to enhance mothers' comfort with reporting valid child maltreatment information, while also meeting researchers' legal and ethical responsibilities for reporting. Under the UK Children Act,33 our responsibility was to secure intervention if maltreatment was current and ongoing. Such intervention on behalf of E-Risk families was carried out with parental cooperation in all but one case.

The protocol included standardized probe questions such as, ‘Do you remember any time when (boy's name) was disciplined severely enough that he may have been hurt?’ and ‘Did you worry that you or someone else may have harmed or hurt (boy's name)?’ Questions were carefully worded to avoid implying that the mother was the perpetrator, so mothers might feel more willing to report that a child had been maltreated. In cases where mothers reported any maltreatment, interviewers probed for details about the incident and recorded notes. Based on the mothers' narrative, interviewers coded if the boy had definitely been physically maltreated (N=34). Examples of such maltreatment included being a victim of adjudicated assault by a teenaged sibling, punished by being burned with matches, had injuries (e.g., fractures or dislocations) from neglectful or abusive parental care, and/or were formally registered with a social services child protection team for physical abuse.

A further 147 boys were designated as ‘probable maltreatment.’ The ‘probable maltreatment’ category does not represent a milder form of abuse compared to the ‘definite maltreatment’ category. Instead, this group includes children who could only be suspected of having experienced maltreatment. In these probable cases, the boy had been reported by concerned schools, neighbors, and/or family members to child protective services but the case was not resolved or registered, he seemed afraid of his father during our home visit, the mother reported that he received frequent physical discipline, or she said he had been smacked harder than intended, leaving a mark or bruise. Some of these boys will not have experienced maltreatment, whereas a subset of them has been maltreated but our coding had to remain uncertain. To enhance certainty, we used information about intimate partner violence to re-classify children in the probable category who were at greatest likelihood of having been maltreated. Researchers have documented the high correlation between intimate partner violence and child maltreatment.34, 35

Intimate partner violence in the home was assessed by inquiring about 12 acts of physical violence (e.g., kicking a partner, threatening a partner with a knife) following the protocol of the Conflict Tactics Scale – Form R.36 Mothers were asked about their own violence toward a partner and about any partner's violence toward them in the years of the child's life before the age-5 and the age-7 assessments. A methodological study demonstrated strong reliability and validity for the partner violence scale (i.e., 0.89 internal consistency and 75% inter-partner agreement).37

Children in the ‘probable’ maltreatment category were classified as having physical abuse exposure if their families were in the top 5% of the distribution of partner violence at either the age-5 or age-7 assessments. This method resulted in 28 boys from the ‘probable maltreatment’ group being added to the group of children we classified for this article as having been exposed to physical abuse. The resulting group totaled 62 boys (6.4% of the cohort unweighted; 4.7% weighted to represent the population) who were exposed to physical abuse.

Children's mental health outcomes

At the age-7 assessment, children's behavior problems were assessed with the Child Behavior Checklist38 and the Teacher Report Form,39 supplemented with items from the Rutter Child Scale40 and additional items measuring Diagnostic and Statistical Manual of Mental Disorders, 4th edition41 criteria for attention deficit hyperactivity disorder (ADHD). The CBCL cutoff for the clinical range (top 2% of the standardization sample) was exceeded by 7% of E-Risk cohort boys. Symptoms and behaviors were reported for the preceding 6 months and each item was scored as (0) ‘not true,’ (1) ‘somewhat true,’ and (2) ‘very often true.’ Sample items from the Antisocial Behavior Scale include ‘physically attacks people,’ ‘lying or cheating,’ and ‘destroys things that belong to others’ (M=21.57, s.d.=18.02). Sample items from the Attention Deficit Hyperactivity Scale include ‘very restless,’ ‘cannot concentrate,’ and ‘impulsive or acts without thinking’ (M=16.25, s.d.=12.40). Sample items from the Emotional Problems Scale include ‘too fearful or anxious,’ ‘unhappy, sad, or depressed,’ and ‘withdrawn from social interaction’ (M=11.08, s.d.=8.23). Following recommendations from the test manual,42 we used mother interviews and teacher reports of children's behavior in combination to maximize reliability and validity. Mother and teacher reports of the same behaviors were moderately correlated and alpha reliabilities for all scales exceeded 0.85. Because simple combinatorial rules work as well, or better, than more complicated ones,43 mothers’ and teachers’ reports were averaged to create scales of antisocial behavior, attention-deficit hyperactivity, and emotional problems. Principal Components Analysis (PCA) of the three subscales identified a single Composite Mental Health Problem Scale that accounted for 64.1% of the variance. Each of the component scales loaded adequately on the factor, with factor loadings ranging from 0.61 to 0.90. All scale scores were standardized with a mean of 0 and a standard deviation of 1.

Maternal antisocial personality symptoms

Mothers reported on their own lifetime antisocial history. Questions were derived from the Diagnostic Interview Schedule,44 supplemented by items from the Young Adult Behavior Checklist.45 The items covered illegal behavior, deceitfulness, impulsivity, aggressiveness, recklessness, and irresponsibility. A symptom was considered to be present if the mother reported behavioral items representing the symptom as being ‘very true or often true.’ Symptom counts in this study ranged from 0 to 6 (M=0.61, s.d.=1.07).46

Statistical analyses

First, we used ordinary least-squares regression to test the main effect associations of boys’ physical abuse exposure and MAOA genotype in predicting mental health problems. Second, we tested for an interaction between MAOA activity and physical abuse exposure in predicting boys’ mental health problems. Third, we tested whether boys’ MAOA genotype was associated with risk for physical abuse exposure (i.e., evocative rGE). Fourth, we controlled for a possible passive rGE by adjusting for the effect of maternal antisocial personality symptoms in the regression analyses. Analyzing two boys in each family creates dependence in the data, and thus all regression results are based on the sandwich or Huber/White variance estimator,47 a method available in STATA 9.0,48 which adjusts estimated standard errors to account for that dependence and provides statistical tests that are robust to model assumptions.49 Fifth, we conducted a meta-analysis pooling results from previous G × E studies of the MAOA polymorphism, physical maltreatment, and mental health with findings from the present study.


Allele frequencies

Genotypes in the sample consisted of five variants of the 30-bp repeat sequence: 2- (0.2%), 3- (31.9%), 3.5- (2.1%), 4- (64.2%), and 5-repeats (1.6%). These allele frequencies matched closely the frequencies reported in other Caucasian samples.31 In terms of expression, all studies agree on the functional classification of the two most common alleles, that is, 3-repeats (low activity), and 4-repeats (high activity). Of rare alleles, both Sabol et al.4 and Deckert et al.50 assayed the 3.5-repeat with the same result (high activity), whereas a discrepancy resulted for the 5-repeat. We chose the classification of Sabol et al.4 (i.e., 5-repeat equals low activity) as they assayed three cell lines as opposed to one. The rare 2-repeat, of which only two cases exist in our sample, was classified as low activity based upon precedence in previous studies. Dropping the 18 boys with the 2- or 5-repeat alleles did not alter the pattern of findings nor their significance.3, 6, 7

Predicting composite mental health problems

Sample sizes, group means, and effect size comparisons are presented in Table 1. As expected, there was a significant main effect of physical abuse exposure on children's mental health problems (Table 2). In addition, we found a significant main effect of MAOA activity (P=0.017), such that boys with the high activity allele had a higher level of mental health problems. The test for the interaction between MAOA activity and physical abuse exposure revealed a significant G × E (b=−0.84, s.e.=0.40, t=2.09, P=0.037). This interaction showed that the effect of physical abuse exposure was significantly weaker among boys with high MAOA activity (b=0.61, s.e.=0.22, t=2.83, P=0.005) than among boys with low MAOA activity (b=1.45, s.e.=0.33, t=4.40, P<0.001) (see Figure 1).

Table 1 Sample sizes, means, standard deviations, and effect size (Cohen's d) comparisons of age-7 mental health outcomes in boys by MAOA genotype and physical abuse exposure
Table 2 Results of regression analyses testing gene-by-physical abuse exposure interaction effects on children's mental health outcomes
Figure 1

Gene-by-physical abuse exposure interaction predicting children's composite mental health problems.

Next, we examined whether there might be evidence of a rGE in two ways. First, we tested for the possibility that a child's genotype might be involved in evoking or eliciting physical abuse exposure. We found that it did not. Boys’ MAOA activity was not significantly associated with likelihood of exposure to physical abuse (low-activity MAOA 3.5%; high-activity MAOA 5.3%; F(1,974)=1.89, P=0.170). Second, we tested whether a child's MAOA genotype might be related to the likelihood of experiencing physical abuse indirectly via an association with parental characteristics that increase environmental risk exposure. Maternal antisocial personality symptoms were significantly correlated with children's likelihood of physical abuse exposure (b=1.22, s.e.=0.26, t=4.79, P<0.001). However, when the model was adjusted for the main effects of physical abuse exposure, MAOA genotype, and maternal antisocial personality symptoms, the interaction between MAOA activity and physical abuse exposure remained significant (Table 2).

Do results hold for component mental health subscales?

We repeated the analyses separately predicting each of the component mental health subscales to examine whether the G × E interaction finding would be robust across multiple outcomes. All subscale findings of the interaction were in the predicted direction (Figure 2a–c) and mirrored the findings reported for the composite mental health measure. However, only the ADHD subscale yielded clear statistical significance (b=−0.78, s.e.=0.35, t=2.26, P=0.024) (Table 2). This G × E finding remained significant (P=0.014) after controlling for maternal antisocial personality symptoms.

Figure 2

Gene-by-physical abuse exposure interaction predicting subscales of children's (a) antisocial behavior, (b) attentional problems and hyperactivity, and (c) emotional problems.


We included studies in our meta-analysis if they fulfilled four criteria. First, the study had to be published in a peer-reviewed journal. Second, the study had to include genotypic information on the variable number tandem repeat polymorphism in the promoter region of the MAOA gene. Third, the study had to include a measure of serious familial adversity in childhood that was significantly associated in a main effect fashion with the outcome measure. Fourth, the sample had to be drawn from a non-clinical population. In addition to the present study, we know of four previous studies that meet all criteria.3, 6, 7, 8 A study by Young et al.9 was excluded because their entire sample consisted of adolescents in clinical treatment for serious conduct problems, no matched control group was included, and their measure of lifetime conduct disorder symptoms places the temporal ordering of the risk and outcome variables in question.

Standard methods for effect size conversion were used to translate the results of the different studies into a common metric of correlations.51, 52 The correlations of interest were those reported between family adversity and antisocial behavior within each MAOA genotype group. If papers reported more than one outcome measure, we chose the most general measure of antisocial behavior (e.g., composite measures). Where necessary, authors were contacted for additional results in order to calculate effect sizes. All correlations were z-transformed and their standard errors derived.

Heterogeneity across studies was assessed using the χ2-based Q statistic, as well as the I2 measure of the percentage of across-study variation attributable to heterogeneity.53, 54 Using meta-analysis programs available in STATA 9.0,48, 55 fixed- and random-effects estimates and results were consistent across all analyses; we report only results from the more conservative random-effects model. Meta-analysis of the interaction effect was carried out using the difference in the transformed correlations across the MAOA groups within each study.

Figure 3 depicts results of the meta-analysis conducted in two ways. First, panel a presents a forest plot of the correlations between maltreatment and antisocial behavior by MAOA activity group for all studies. Within the high-activity MAOA group, there was no evidence of significant heterogeneity across studies (χ2df=4=0.79, P=0.940, I2=0.0%), but there was mild–moderate heterogeneity across studies in the low-activity MAOA group as indicated by the I2 measure (χ2df=4=6.29, P=0.179, I2=36.4%). The random effects pooled estimates within MAOA activity groups were as follows: low MAOA activity, 0.32 (95% CI: 0.23, 0.40; P<0.001); high MAOA activity, 0.12 (95% CI: 0.08, 0.17; P<0.001).

Figure 3

Meta-analysis pooling results across studies. Summary correlations between measures of childhood maltreatment and mental health are presented separately in the low- vs high-activity MAOA genotype groups. CI indicates confidence interval. Squares and % weight indicate the size of each study's contribution to the summary correlations indicated by diamonds. (a) Forest plot of correlations between childhood maltreatment and mental health, as a function of MAOA genotype group in five independent studies. (b) Forest plot of the interaction effect based on differences in correlations between childhood maltreatment and mental health observed in low- vs high-activity MAOA genotype groups in five independent studies.

Second, panel b presents a forest plot of the meta-analysis of the interaction effect based on the differences in correlations by MAOA activity group. No significant heterogeneity was detected across the studies (χ2df=4=3.79, P=0.440, I2=0.0). The pooled random-effects estimate of the change in correlations from the high- to low-activity MAOA groups indicated a significant effect of 0.18 (95% CI: 0.10, 0.26; P<0.001).

We carried out two additional sensitivity analyses on the meta-analysis results. First, we removed the study by Caspi et al.3 and re-estimated the pooled effects to rule-out potential bias contributed by the first published study of this hypothesis.24, 56 This resulted in a significant but reduced pooled difference in correlations of 0.16 (95% CI: 0.07, 0.25; P=0.001). Second, we removed the Nilsson et al.8 study, which reported an effect size larger in magnitude than the original study by Caspi et al.3 After both studies were removed, results showed a pooled effect of 0.15 (95% CI: 0.05, 0.24; P=0.002) (Meta-analysis using effect sizes for children's antisocial behavior from the Kim-Cohen et al. study resulted in the same pattern of significant findings. Details are available from first author.).


Since 2002 when evidence first appeared that an MAOA polymorphism moderates the impact of childhood maltreatment on risk for antisocial behavior,3 scientists have awaited supporting data through replication. The present investigation evaluated this G × E hypothesis in two ways. First, using data from a representative birth cohort sample of 7-year-old boys, we replicated and extended Caspi et al.'s3 original results. Among children who were exposed to physical maltreatment, boys with the low-activity MAOA allele had mental health problem scores that were half a standard deviation higher than boys with the high-activity allele. Second, our preliminary meta-analysis found supportive evidence of this G × E effect. Pooling estimates from five studies, we found that the association between early familial adversity and mental health was significantly stronger in the low-activity MAOA vs the high-activity MAOA groups. If replication is the ‘sine qua non for accepting a hypothesis’ (p. 627),57 then evidence from the present study brings the field closer toward confirming that MAOA activity is meaningfully involved in explaining variability in developmental outcomes as a consequence of maltreatment.

Beyond replication and meta-analysis, this study contributes novel information in several additional ways. First, we extended the findings downward to a sample of 7-year-old boys, indicating that variability in young children's mental health is explained significantly by differences in vulnerability to stress as indexed by a genetic polymorphism. This finding suggests that one mechanism by which maltreatment leads to the development of psychopathology relates specifically to MAOA functioning, and that this biological process can be initiated early in life. Eventually, such evidence can inform not only what treatments might help prevent psychopathology in physically maltreated children but also when such intervention might be most successful. Second, we evaluated MAOA activity in relation to children's global mental health outcomes in addition to their antisocial behavior. In the Dunedin sample of adult males, the MAOA polymorphism did not moderate stress effects in predicting depression,58 suggesting that in adults, MAOA activity moderates the impact of stress specifically towards an antisocial outcome. In childhood, the G × E involving MAOA appears to influence ADHD-related symptoms as well as a broader phenotype comprising several domains of mental health that together signal the beginning of a maladaptive trajectory toward the development of adult antisocial behavior. Third, we ruled-out the possibility that the G × E might be explained spuriously by a passive or evocative rGE.

One unexpected finding of the present study was that the MAOA polymorphism predicted children's mental health outcomes in a main effect fashion. Boys with the genotype conferring high MAOA activity had slightly but significantly elevated levels of global mental health problems as well as antisocial behavior and attention-deficits/hyperactivity relative to boys with the low-activity genotype. This finding is not inconsistent with previously published results of a significant main effect of the high-activity MAOA genotype on increasing risk for antisocial behavior and ADHD.59, 60, 61 Moreover, closer inspection of other G × E studies involving the MAOA polymorphism reveals similar findings, albeit at marginally significant levels.3, 6, 7 For instance, Foley et al.6 found a nonsignificant trend (P<0.14) for the high-activity MAOA allele increasing risk for conduct disorder; after adjusting for the interaction between the MAOA polymorphism and maltreatment, the main effect became significant (P=0.04). However, two recent studies found contradictory results with positive prediction from the low-activity MAOA genotype to ADHD.62, 63 Further research to resolve these discrepancies is needed to understand the direction of gene effects on conferring risk for psychopathology in the absence or presence of risk exposure.

Strengths of this study include a meta-analysis as well as an original analysis of data from a large, non-selected sample of boys at a young age when early emerging psychopathology predicts continuing difficulties in later life. Genotype, phenotype, and physical abuse exposure were measured using independent methods. The findings, however, should be interpreted in light of several limitations. First, our sample comprised Caucasian twins living in England and Wales. However, allele frequencies for the MAOA promoter polymorphism3, 4 and base rates of physical abuse exposure in our sample are comparable to those reported elsewhere.34 Future research should test whether our findings replicate in singleton children and in non-Caucasian samples. Second, official records of maltreatment history were not available. However, unlike many large-scale studies that rely upon self-complete questionnaires, we interviewed caregivers face-to-face about violence exposure, a method having evidence of good validity.22 Third, because girls’ MAOA genotype cannot be characterized with confidence, our findings and the findings of previous studies are informative only about males. More progress toward understanding X-chromosome inactivation and MAOA gene expression is needed before the relevance of this finding to female subjects can be evaluated. Fourth, our meta-analysis contained only four studies in addition to Caspi et al.'s3 original study. Therefore, our meta-analysis results should be re-evaluated once further tests of the hypothesis are published.

Ultimately, the goal of genomics research is to prevent mental disorder57 and to refine treatment strategies.64 Reliable G × E findings have considerable potential for informing such efforts.10 However, a statistical interaction between a genotype and an environmental risk factor requires further research to uncover the biological mechanisms involved in the interaction.65 Because the MAOA enzyme selectively metabolizes serotonin, norepinephrine, and dopamine,4, 5 which are involved in multiple brain functions associated with stress regulation,66 it is likely to be one of myriad factors involved in the development of biological sensitivity to stress and the social context.67 A statistical G × E involving MAOA thus represents an important launching pad for developmental neuroscience research into the underlying causal mechanisms involved in the etiology of psychopathology.10, 64, 68, 69, 70 Moreover, it is possible that the MAOA gene may simply be a marker for a behavioral trait, which itself moderates the association between maltreatment and children's mental health.

Eradicating child maltreatment is clearly the preferred way to combat risk for psychiatric problems, and yet large numbers of children in Western societies are abused and exposed to family violence each year.35 Once an adverse experience touches off an otherwise ‘silent’ genetic vulnerability and triggers a cascade of biological events toward atypical development, what can be done to halt or reverse the process? In addition to possible pharmacological interventions, recent research on the serotonin transporter polymorphism suggests that social support can protect even genetically vulnerable children from the negative sequelae of maltreatment.71 As evidence for significant G × E in predicting mental health continues to emerge,70 both scientists and the public are becoming increasingly aware that like many developmental processes,72 the nature of gene effects on behavior, too, is often contingent upon experience.


  1. 1

    Rutter M . The promotion of resilience in the face of adversity. In: Clarke-Stewart A, Dunn J (eds). Families Count: Effects on Child and Adolescent Development. Cambridge University Press: New York, 2006, pp 26–52.

    Google Scholar 

  2. 2

    Widom C, McGloin J . Resilience among abused and neglected children grown up. Dev Psychopathol 2001; 13: 1021–1038.

    Article  Google Scholar 

  3. 3

    Caspi A, McClay J, Moffitt TE, Mill J, Martin J, Craig IW et al. Role of genotype in the cycle of violence in maltreated children. Science 2002; 297: 851–854.

    CAS  Article  Google Scholar 

  4. 4

    Sabol S, Hu S, Hamer D . A functional polymorphism in the monoamine oxidase A gene promoter. Hum Genet 1998; 103: 273–279.

    CAS  Article  Google Scholar 

  5. 5

    Shih J, Chen K, Ridd M . Monoamine oxidase: from genes to behavior. Ann Rev Neurosci 1999; 22: 197–217.

    CAS  Article  Google Scholar 

  6. 6

    Foley D, Eaves L, Wormley B, Silberg JL, Maes H, Kuhn J et al. Childhood adversity, monoamine oxidase A genotype, and risk for conduct disorder. Arch Gen Psychiatry 2004; 61: 738–744.

    CAS  Article  Google Scholar 

  7. 7

    Haberstick B, Lessem J, Hopfer C, Smolen A, Ehringer M, Timerlake D et al. Monoamine oxidase A (MAOA) and antisocial behaviors in the presence of childhood and adolescent maltreatment. Am J Med Genet B Neuropsychiatr Genet 2005; 135B: 59–64.

    Article  Google Scholar 

  8. 8

    Nilsson K, Sjoberg R, Damberg M, Leppert J, Ohrvik J, Alm P et al. Role of monoamine oxidase A genotype and psychosocial factors in male adolescent criminal activity. Biol Psychiatry 2005; 59: 121–127.

    Article  Google Scholar 

  9. 9

    Young S, Smolen A, Hewitt J, Haberstick M, Stallings M, Corley R et al. Interaction between MAO-A genotype and maltreatment in risk for conduct disorder: failure to confirm in adolescent patients. Am J Psychiatry 2005; 163: 1019–1025.

    Article  Google Scholar 

  10. 10

    Moffitt TE, Caspi A, Rutter M . Strategy for investigating interactions between measured genes and measured environments. Arch Gen Psychiatry 2005; 62: 473–481.

    CAS  Article  Google Scholar 

  11. 11

    Cases O, Seif I, Grimsby J, Gaspar P, Chen K, Pournin S et al. Aggressive behavior and altered amounts of brain serotonin and norepinephrine in mice lacking MAOA. Science 1995; 268: 1763–1766.

    CAS  Article  Google Scholar 

  12. 12

    Brunner H, Nelen M, Breakefield X, Ropers H, van Oost B . Abnormal behavior associated with a point mutation in the structural gene for monoamine oxidase A. Science 1993; 262: 578–580.

    CAS  Article  Google Scholar 

  13. 13

    Brunner H, Nelen M, van Zandvoort P, Abeling N, van Gennip A, Wolters E et al. X-linked borderline mental retardation with prominent behavioral disturbance: phenotype, genetic localization, and evidence for disturbed monoamine metabolism. Am J Hum Genet 1993; 52: 1032–1039.

    CAS  PubMed  PubMed Central  Google Scholar 

  14. 14

    Moffitt TE . Life course persistent versus adolescence-limited antisocial behavior. In: Cicchetti D, Cohen DJ (eds). Developmental Psychopathology, 2nd edn, vol. 3. Wiley: New York, 2006, pp 570–598.

    Google Scholar 

  15. 15

    Rutter M . Relationships between mental disorders in childhood and adulthood. Acta Psychiatr Scand 1995; 91: 73–85.

    CAS  Article  Google Scholar 

  16. 16

    Lansford JE, Dodge KA, Pettit GS, Bates JE, Crozier J, Kaplow J . A 12-year prospective study of the long-term effects of early child physical maltreatment on psychological, behavioral, and academic problems in adolescence. Arch Pediatr Adolesc Med 2002; 156: 824–830.

    Article  Google Scholar 

  17. 17

    Crijnen A, Achenbach T, Verhulst F . Problems reported by parents of children in multiple cultures: The Child Behavior Checklist syndrome constructs. Am J Psychiatry 1999; 156: 569–574.

    CAS  PubMed  Google Scholar 

  18. 18

    Caron C, Rutter M . Comorbidity in child psychopathology: concepts, issues and research strategies. J Child Psychol Psychiatry 1991; 32: 1063–1080.

    CAS  Article  Google Scholar 

  19. 19

    Verhulst FC, van der Ende J . ‘Comorbidity’ in an epidemiological sample: a longitudinal perspective. J Child Psychol Psychiatry 1993; 34: 767–783.

    CAS  Article  Google Scholar 

  20. 20

    Rutter M, Silberg J . Gene–environment interplay in relation to emotional and behavioral disturbance. Annu Rev Psychol 2002; 53: 463–490.

    Article  Google Scholar 

  21. 21

    DiLalla L, Gottesman II . Biological and genetic contributors to violence – Widom's untold tale. Psychol Bull 1991; 109: 125–129.

    CAS  Article  Google Scholar 

  22. 22

    Jaffee SR, Moffitt TE, Caspi A, Taylor A . Physical maltreatment victim to antisocial chid: evidence of an environmentally mediated process. J Abnorm Psychol 2004; 113: 44–55.

    Article  Google Scholar 

  23. 23

    Insel TR, Collins FS . Psychiatry in the genomics era. Am J Psychiatry 2003; 160: 616–620.

    Article  Google Scholar 

  24. 24

    Ioannidis J, Ntzani E, Trikalinos T, Contopoulos-Ioannidis D . Replication validity of genetic association studies. Nat Genet 2001; 29: 306–309.

    CAS  Article  Google Scholar 

  25. 25

    Trouton A, Spinath FM, Plomin R . Twins Early Development Study (TEDS): a multivariate, longitudinal genetic investigation of language, cognition, and behavior problems in childhood. Twin Res 2002; 5: 444–448.

    Article  Google Scholar 

  26. 26

    Maynard RA . Kids Having Kids: Economic Costs and Social Consequences of Teen Pregnancy. Urban Institute Press: Washington DC, 1997.

    Google Scholar 

  27. 27

    Moffitt TE, The E-Risk Study Team. Teen-aged mothers in contemporary Britain. J Child Psychol Psychiatry 2002; 43: 727–742.

    Article  Google Scholar 

  28. 28

    Bennett N, Jarvis L, Rowlands O, Singleton N, Haselden L . Living in Britain: Results from the General Household Survey. HMSO: London, 1996.

    Google Scholar 

  29. 29

    Benjamin D, Van Bakel I, Craig I . A novel expression based approach for assessing the inactivation status of human X-linked genes. Eur J Hum Genet 2000; 8: 103–108.

    CAS  Article  Google Scholar 

  30. 30

    Carrel L, Willard H . X-inactivation profile reveals extensive variability in X-linked gene expression in females. Nature 2005; 434: 400–404.

    CAS  Article  Google Scholar 

  31. 31

    Freeman B, Smith N, Curtis C, Huckett L, Mill J, Craig I . DNA from buccal swabs recruited by mail: evaluation of storage effects on long-term stability and suitability for multiplex polymerase chain reaction genotyping. Behav Genet 2003; 33: 67–72.

    CAS  Article  Google Scholar 

  32. 32

    Dodge KA, Pettit GS, Bates JE, Valente E . Social information-processing patterns partially mediate the effect of early physical abuse on later conduct problems. J Abnorm Psychol 1995; 104: 632–643.

    CAS  Article  Google Scholar 

  33. 33

    Department of Health. The Children Act. HMSO: London, 1989.

  34. 34

    Osofsky J . Prevalence of children's exposure to domestic violence and child maltreatment: implications for prevention and intervention. Clin Child Fam Psychol Rev 2003; 6: 161–170.

    Article  Google Scholar 

  35. 35

    Tolan P, Gorman-Smith D, Henry D . Family violence. Annu Rev Psychol 2006; 57: 557–583.

    Article  Google Scholar 

  36. 36

    Straus MA . Measuring intrafamily conflict and violence: the Conflict Tactics (CT) Scales. In: Straus MA, Gelles RJ (eds). Physical Violence in American Families: Risk Factors and Adaptations to Violence in 8145 Families. Transaction: New Brunswick, NJ, 1990, pp 403–424.

    Google Scholar 

  37. 37

    Moffitt TE, Caspi A, Krueger RF, Magdol L, Margolin G, Silva PA et al. Do partners agree about abuse in their relationship? A psychometric evaluation of interpartner agreement. Psychol Assess 1997; 9: 47–56.

    Article  Google Scholar 

  38. 38

    Achenbach TM . Manual for the Child Behavior Checklist/4–18 and 1991 Profile. University of Vermont Department of Psychiatry: Burlington, VT, 1991.

    Google Scholar 

  39. 39

    Achenbach TM . Manual for the Teacher's Report Form and 1991 Profile. University of Vermont Department of Psychiatry: Burlington VT, 1991.

    Google Scholar 

  40. 40

    Sclare I . The Child Psychology Portfolio. NFER-Nelson Publishing Company: Windsor, Berkshire, 1997.

    Google Scholar 

  41. 41

    American Psychiatric Association. Diagnostic and Statistical Manual of Mental Disorders, 4th edn. American Psychiatric Association: Washington, DC, 1994.

  42. 42

    Achenbach TM, Rescorla L . Manual for ASEBA School-Age Forms & Profiles. University of Vermont, Research Center for Children, Youth, & Families: Burlington, VT, 2001.

    Google Scholar 

  43. 43

    Piacentini JC, Cohen P, Cohen J . Combining discrepant diagnostic information from multiple sources: are complex algorithms better than simple ones? J Abnorm Child Psychol 1992; 20: 51–63.

    CAS  Article  Google Scholar 

  44. 44

    Robins L, Cottler L, Bucholz K, Compton W . Diagnostic Interview Schedule for DSM-IV. Washington University School of Medicine: St Louis, MO, 1995.

    Google Scholar 

  45. 45

    Achenbach TM . Manual for the Young Adult Self-Report and Young Adult Behavior Checklist. University of Vermont Department of Psychiatry: Burlington, VT, 1997.

    Google Scholar 

  46. 46

    Kim-Cohen J, Caspi A, Rutter M, Polo Tomas M, Moffitt TE . The caregiving environments provided to children by depressed mothers with or without an antisocial history. Am J Psychiatry 2006; 163: 1009–1018.

    Article  Google Scholar 

  47. 47

    Gould W, Scribney W . Maximum Likelihood Estimation with STATA. Stata Press: College Station, TX, 1999.

    Google Scholar 

  48. 48

    StataCorp. Stata Statistical Software: Release 9.0. Stata Corporation: College Station, TX, 2005.

  49. 49

    Lumley T, Diehr P, Emerson S, Chen L . The importance of the normality assumption in large public health data sets. Annu Rev Pub Health 2002; 23: 151–169.

    Article  Google Scholar 

  50. 50

    Deckert J, Catalano M, Syagailo Y, Okladnova O, DiBella D, Nothen M et al. Excess of high activity monoamine oxidase A gene promoter alleles in female patients with panic disorder. Hum Mol Genet 1999; 8: 621–624.

    CAS  Article  Google Scholar 

  51. 51

    Hunter JE, Schmidt FL . Methods of Meta-analysis: Correcting Error and Bias in Research Findings, 2nd edn, Sage: Newbury Park, CA, 2004.

    Google Scholar 

  52. 52

    Lipsey MW, Wilson DB . Practical Meta-Analysis. Sage: Thousand Oaks, CA, 2001.

    Google Scholar 

  53. 53

    Egger M, Smith GD, Altman DG . Systematic Reviews in Health Care: Meta-analysis in Context, 2nd edn. BMJ Publications: London, 2001.

    Google Scholar 

  54. 54

    Higgins JPT, Thompson SG . Quantifying heterogeneity in a meta-analysis. Statist Med 2002; 21: 1539–1558.

    Article  Google Scholar 

  55. 55

    Sterne J, Bradburn M, Egger M . Meta-analysis in StataTM. In: Egger M, Smith GD, Altman DG (eds). Systematic Reviews in Health Care: Meta-Analysis in Context, 2nd edn, BMJ Publishing: London, 2001, pp 347–369.

    Google Scholar 

  56. 56

    Trikalinos TA, Ntzani EE, Contopoulos-Ioannidis DG, Ioannidis JP . Establishment of genetic associations for complex diseases is independent of early study findings. Euro J Hum Genet 2004; 12: 762–769.

    CAS  Article  Google Scholar 

  57. 57

    Merikangas KR, Risch N . Will the genomics revolution revolutionize psychiatry? Am J Psychiatry 2003; 160: 625–635.

    Article  Google Scholar 

  58. 58

    Caspi A, Sugden K, Moffitt TE, Taylor A, Craig IW, Harrington H et al. Influence of life stress on depression: moderation by a polymorphism in the 5-HTT gene. Science 2003; 301: 386–389.

    CAS  Article  Google Scholar 

  59. 59

    Beitchman J, Mik H, Ehtesham S, Douglas L, Kennedy J . MAOA and persistent, pervasive childhood aggression. Mol Psychiatry 2004; 9: 546–547.

    CAS  Article  Google Scholar 

  60. 60

    Manor I, Tyano S, Mel E, Eisenberg J, Bachner-Melman R, Kotler M et al. Family-based and association studies of monoamine oxidase A and attention deficit hyperactivity disorder (ADHD): preferential transmission of the long promoter-region repeat and its association with impaired performance on a continuous performance test (TOVA). Mol Psychiatry 2002; 7: 626–632.

    CAS  Article  Google Scholar 

  61. 61

    Manuck S, Flory J, Ferrell R, Mann J, Muldoon M . A regulatory polymorphism of the monoamine oxidase-A gene may be associated with variability in aggression, impulsivity, and central nervous system serotonergic responsivity. Psychiatry Res 2000; 95: 9–23.

    CAS  Article  Google Scholar 

  62. 62

    Lawson D, Turic D, Langley K, Pay H, Govan C, Norton N et al. Association analysis of monoamine oxidase A and attention deficit hyperactivity disorder. Am J Med Genet B Neuropsychiatr Genet 2003; 116B: 84–89.

    Article  Google Scholar 

  63. 63

    Domschke K, Sheehan K, Lowe N, Kirley A, Mullins C, O’Sullivan R et al. Association analysis of the monoamine oxidase A and B genes with attention deficit hyperactivity disorder (ADHD) in an Irish sample: preferential transmission of the MAO-A 941G allele to affected children. Am J Med Genet B Neuropsychiatr Genet 2005; 134B: 110–114.

    Article  Google Scholar 

  64. 64

    Kendler K . ‘A gene for…’: the nature of gene action in psychiatric genetics. Am J Psychiatry 2005; 162: 1243–1252.

    Article  Google Scholar 

  65. 65

    Clayton D, McKeigue PM . Epidemiological methods for studying genes and environmental factors in complex diseases. Lancet 2001; 358: 1356–1360.

    CAS  Article  Google Scholar 

  66. 66

    Charney DS . Psychobiological mechanisms of resilience and vulnerability: implications for successful adaptation to extreme stress. Am J Psychiatry 2004; 161: 195–216.

    Article  Google Scholar 

  67. 67

    Boyce W, Ellis B . Biological sensitivity to context: I. An evoluationary-developmental theory of the origins and functions of stress reactivity. Dev Psychopathol 2005; 17: 271–301.

    Article  Google Scholar 

  68. 68

    Cicchetti D, Blender JA . A multiple-levels-of-analysis approach to the study of developmental processes in maltreated children. Proc Nat Acad Sci USA 2004; 101: 17325–17326.

    CAS  Article  Google Scholar 

  69. 69

    Meyer-Lindenberg A, Buckholtz JW, Kolachana B, Hariri AR, Pezawas L, Blasi G et al. Neural mechanisms of genetic risk for impulsivity and violence in humans. Proc Nat Acad Sci USA 2006; 103: 6269–6274.

    CAS  Article  Google Scholar 

  70. 70

    Rutter M, Moffitt TE, Caspi A . Gene–environment interplay and psychopathology: multiple varieties but real effects. J Child Psychol Psychiatry 2006; 47: 226–261.

    Article  Google Scholar 

  71. 71

    Kaufman J, Yang B, Douglas-Palumberi H, Houshyar S, Lipschitz D, Krystal J et al. Social supports and serotonin transporter gene moderate depression in maltreated children. Proc Nat Acad Sci USA 2004; 101: 17316–17321.

    CAS  Article  Google Scholar 

  72. 72

    Pollak SD . Experience-dependent affective learning and risk for psychopathology in children. Ann NY Acad Sci 2003; 1008: 102–111.

    Article  Google Scholar 

Download references


We are grateful to the Study families and teachers for their participation. We thank Robert Plomin and Michael Rutter for their contributions, Thomas Achenbach for kind permission to adapt the CBCL, and members of the E-Risk team for their dedication, insights, and hard work. Terrie E Moffitt is a recipient of a Royal Society-Wolfson Research Merit Award. This research was supported by grants from the National Institute of Mental Health (MH45070 & MH49414), the UK Medical Research Council (G9806489 & G0100527), and the ESRC-SCOPIC Network.

Author information



Corresponding author

Correspondence to J Kim-Cohen.

Rights and permissions

Reprints and Permissions

About this article

Cite this article

Kim-Cohen, J., Caspi, A., Taylor, A. et al. MAOA, maltreatment, and gene–environment interaction predicting children's mental health: new evidence and a meta-analysis. Mol Psychiatry 11, 903–913 (2006).

Download citation


  • monoamine oxidase A
  • promoter polymorphism
  • maltreatment
  • psychopathology
  • antisocial behavior
  • ADHD
  • emotional problems

Further reading