Abstract

Introduction

Criminal violence is one of the most burdensome public-health issues worldwide, with staggering socio-economic repercussions (Krug et al. 2002). The urgent need to develop effective strategies for the prevention of criminal violence has recently given impetus to new research efforts aimed at identifying its psychobiological causes. These investigations have revealed that the propensity towards criminal violence is underpinned by a complex interplay of genetic and socio-environmental factors (Rhee and Waldman, 2002; Gottschalk and Ellis, 2009); the exact nature of these interactions, however, remains poorly understood, also in consideration of the ethical, legal and logistic concerns raised by genetic studies in criminal offenders.

Among the genes that have been implicated in the predisposition to violence, emerging evidence has highlighted a key role for MAOA, which encodes for monoamine oxidase A. This enzyme serves a primary role in the metabolism of neurotransmitters implicated in the regulation of aggression, such as serotonin, norepinephrine and dopamine (Bortolato et al. 2008). A loss-of-function mutation of MAOA gene has been shown to result in a clinical syndrome characterized by overt proclivity to engage in violent acts in response to minor stressors (Brunner et al. 1993).

The role of MAO A in the neurobiological bases of violence is also confirmed by preclinical studies. In particular, these lines of research have shown that, in mice, the deficiency of this enzyme leads to marked aggressiveness, maladaptive defensive reactivity, defects in information processing and perseverative responses (Cases et al. 1995; Bortolato et al. 2011; Godar et al. 2011; Bortolato et al. 2012; Bortolato et al. 2013).

The link between MAOA and violent behavior has been further investigated with respect to its allelic variants, and in particular MAOA-uVNTR, a 30-bp functional polymorphism located upstream of its transcription initiation site (Sabol et al. 1998). Six MAOA-uVNTR variants have been characterized, based on their different number of repeats (2, 3, 3.5, 4, 5 and 6) (Huang et al. 2004); of these, the 2- and 3-repeat variants are associated to lower transcriptional efficiency and enzymatic activity (Sabol et al. 1998; Deckert et al. 1999; Denney et al. 1999). The low-activity alleles (L-MAOA) have been linked to a higher risk of impulsive aggression (Oreland et al. 2007; Buckholtz and Meyer-Lindenberg, 2008), as well as maladaptive processing of affect (Lee and Ham, 2008). In particular, the 2-repeat variant, albeit extremely rare in the general population, has been associated with delinquent and violent behavior (Guo et al. 2008; Beaver et al. 2013; Beaver et al. in press). Several independent studies have also shown that carriers of L-MAOA variants with a history of maltreatment during childhood have a significantly higher risk to develop impulsive aggression (Caspi et al. 2002; Kim-Cohen et al. 2006; Williams et al. 2009; Fergusson et al. 2011; but see Haberstick et al. 2014 for conflicting results).

Recently, Beaver and colleagues (2010) documented the association of L-MAOA variants with multiple aspects of violent criminal activity, such as gang membership and weapon use. This background suggests that L-MAOA alleles may play a role in the predisposition to criminal violence. To the best of our knowledge, however, no study has examined whether criminal violence in prison inmates could be predicted by MAOA-uVNTR genotype. Thus, here we investigated this possibility in a sample of male convicts incarcerated for violent and non-violent acts, and verified whether this relation may be paralleled by alterations in other psychological traits related to violence, namely impulsivity and aggression.

Methods

Results

Distribution of MAOA genotypes across races and violent convicts

In substantial agreement with previous data on the MAOA allelic distribution in the general population (Sabol et al. 1998), we found a trend (P=0.08) toward a significantly higher frequency of African-American carriers of low-activity MAOA variants, as compared with their Caucasian counterparts (Fig. 1A). Notably, low-activity MAOA variants were displayed by 61.22% violent and 20% non-violent offenders, indicating a robust association of these alleles with violent crime (P<0.0001; Fisher’s exact test) (Fig.1B). The odds ratio and relative risk for violent behavior in convicts with low-activity MAOA alleles were 6.31 and 2.12, respectively. Fisher’s exact test revealed a significantly disproportionate distribution of MAOA alleles between violent and non-violent convicts in the Caucasian subgroup (P<0.01) (Fig. 1C). Conversely, only a marginally significant difference (P=0.08) was found in the proportion of low-activity MAOA alleles in African-Americans violent and non-violent convicts (Fig. 1D).

Open in a separate window

Fig.1

Comparisons of the frequencies of MAOA-uVNTR allele carriers with respect to violent crime and ethnicity. L-MAOA, low-activity MAOA-uVNTR variants (2 and 3 repeats); H-MAOA, high-activity MAOA-uVNTR variants (4 repeats).**, P<0.01; ***, P<0.001 in comparison with L-MAOA. For more details, see text.

Comparisons and correlations of impulsivity, aggression and early-trauma scores

The comparison between genotypes revealed that carriers of low-activity MAOA variants exhibited a weak, yet significant increase in BIS-11 total scores (Fig. 2A) [Main effect for genotype: F(1,85)= 4.52, P<0.05; partial η²=0.05] and BIS-11 attentional-impulsiveness subscale scores (Fig. 2B) [Main effect for genotype: F(1,85)= 4.76, P<0.05; partial η²=0.05]. No main effects for ethnicity or race × genotype interactions were identified. Furthermore, no other statistically significant effects were found for other BIS-11 scores, BPAQ and CTQ scores (Fig. 2 and Suppl. Table 1). ANOVA failed to reveal any significant difference across the four subgroup defined by the combinations of violent crime and MAOA genotype with respect to CTQ, BIS-11 and BPAQ scores (Supplementary Fig. 1). Nevertheless, statistical trends were found for the effect of MAOA genotype on BIS-11 total score [F(1,85)=3.63; P=0.06; Main effect], BIS-11 attentional-impulsiveness subscale score [F(1,85)=3.51; P=0.06; Main effect] and BIS-11 motor-impulsiveness subscale score [F(1,85)=3.15; P=0.08] (Supplementary Fig.1 G–I). In line with previous results (Garcia-Forero et al. 2009), total BIS-11 and motor-impulsiveness substance scores were significantly correlated with BPAQ total scores, as well as Anger and Hostility subscale scores (P<0.0002 for each correlation; Bonferroni correction for multiple testing) (Table 2). ANCOVA identified that none of these correlations was influenced by violent crime charge (Fig.3C), MAOA genotype (Fig.3F) or ethnicity (data not shown). CTQ scores did not correlate with either BIS-11 or BPAQ scores (Table 2 and Fig. 3).

Open in a separate window

Fig.2

Comparisons of BIS-11 and BPAQ scores across MAOA-uVNTR genotype and ethnicity. A, African-Americans; C, Caucasians. L-MAOA, low-activity MAOA-uVNTR variants (2 and 3 repeats); H-MAOA, high-activity MAOA-uVNTR variants (4 repeats). *, P<0.05 for comparisons with L-MAOA variants (Main effect). All values are represented as means ± SEM. For more details, see text.

Open in a separate window

Fig.3

Correlations of total CTQ, BIS-11 and BPAQ scores across MAOA-uVNTR genotypes and violence of the crime charge. For more details, see text.

Table 2

Synoptic tables of correlations between CTQ, BIS-11 and BPAQ scale scores for all participants (irrespective of genotype and violent crime).

CTQ TOT
CTQ EA	0.82
CTQ PA	0.82	0.66
CTQ SA	0.72	0.49	0.65
CTQ EN	0.73	0.50	0.33	0.30
CTQ PN	0.67	0.34	0.46	0.25	0.58
BIS-11 TOT	0.15	0.19	0.09	−0.15	0.23	0.20
BIS-11 ATT	0.18	0.19	0.10	−0.05	0.24	0.18	0.87
BIS-11 MOT	0.08	0.23	0.06	−0.16	0.11	0.01	0.81	0.63
BIS-11 NP	0.13	0.05	0.06	−0.14	0.23	0.32	0.79	0.59	0.35
BPAQ TOT	0.05	0.21	0.13	−0.14	0.03	−0.05	0.48	0.45	0.56	0.20
BPAQ PHYS	0.05	0.15	0.14	−0.11	0.00	−0.01	0.37	0.36	0.43	0.14	0.88
BPAQ VOC	−0.05	0.14	0.10	−0.14	−0.10	−0.23	0.29	0.25	0.36	0.10	0.78	0.60
BPAQ ANG	0.03	0.09	0.11	−0.10	0.04	−0.02	0.52	0.48	0.54	0.26	0.86	0.76	0.51
BPAQ HOST	0.11	0.28	0.09	−0.13	0.11	0.01	0.43	0.38	0.53	0.15	0.83	0.52	0.63	0.59
TOTAL	CTQ TOT	CTQ EA	CTQ PA	CTQ SA	CTQ EN	CTQ PN	BIS11 TOT	BIS ATT	BIS MOT	BIS NP	BNAQ TOT	BNAQ PHYS	BPAQ VOC	BPAQ ANG	BPAQ HOST

Open in a separate window

Significant correlations (P<0.0002; Bonferroni correction for multiple testing) are marked in bold. The values indicated in the tables are the β coefficients of each correlation.

Abbreviations: TOT: total score; EA, emotional abuse; PA, physical abuse; SA, sexual abuse; EN, emotional neglect; PN, physical neglect; ATT, attentional; MOT, motor; NP, non-planning; PHYS, physical aggression; VOC, vocal aggression; ANG, anger; HOST, hostility.

Discussion

The main result of the present study was that, in a sample of incarcerated male offenders, violent crime charges were significantly more frequent in carriers of L-MAOA alleles. This association was replicated in Caucasian inmates, while it only reached a marginal statistical trend in African-American convicts; this apparent divergence, however, is likely to be due to the demographic characteristics of our sample, which featured a majority of Caucasian convicts. The high frequency of L-MAOA alleles in violent offenders is in line with previous reports documenting that L-MAOA alleles are associated with an increased risk to commit violent acts (Reif et al. 2007). Furthermore, our results are in keeping with recent evidence highlighting L-MAOA as a vulnerability genetic factor for use of weapons and proclivity to violence among members of criminal gangs (Beaver et al. 2010). While our data strongly suggest the implication of MAOA in the genetic bases of the predisposition to criminal violence, it is worth noting that this association is likely to hold only in relation to offenders, and not to the general population. L-MAOA alleles may predispose to criminal violence by interacting with other known environmental factors linked to a higher predisposition to commit crimes, such as uninvolved and/or permissive parenting style in the family of origin, exposure to early traumas and/or specific forms of maltreatment, abuse of substances etc. Given the extremely small number of subjects in our sample, we could not reach a sufficient statistical power needed to detect interactions between MAOA alleles and environmental factors; future studies on much larger samples of inmates and non-convicted subjects will be necessary to identify the environmental factors that may interact with MAOA to predict for a higher risk of criminal violence.

The role of MAOA in violence may reflect neurodevelopmental alterations in the homeostatic regulation of serotonin, dopamine and norepinephrine neurotransmission, in consideration of the involvement of these systems in the early regulation of aggression circuitry (Bortolato and Shih, 2011). Recent evidence on MAOA-deficient mice has shown that the role of MAO A in aggression is mediated by alterations in N-methyl-D-aspartate glutamate receptor (Bortolato et al. 2012) in the prefrontal cortex, which may underpin the impairments in social information processing and environmental reactivity. Accordingly, male carriers of L-MAOA alleles have been shown to exhibit pronounced anatomical and functional alterations of the prefrontal cortex and its limbic connections (Meyer-Lindenberg et al. 2006; Buckholtz et al. 2008), which likely lead to a negative cognitive bias in the interpretation of ambiguous social cues (Buckholtz and Meyer-Lindenberg, 2008), as well as to exaggerated responses to provocation (Kuepper et al. 2013).

Carriers of low-activity MAOA-uVNTR variants were also found to exhibit a very mild, yet significant enhancement in BIS-11 total and attention-impulsiveness subscale scores. The very small effect size of these differences underscores the need for further investigations to confirm the association between MAOA genotype and impulsivity in offenders. It is worth noting that, in line with previous reports (Barratt, 2000), the average BIS-11 scores in our sample of male inmates (Total: 70.1 ± 12.7; Attentional: 19.1 ± 3.9; Motor: 23.0 ± 5.9; Non-planning: 28.0 ± 5.7) were markedly higher than those reported in the general population (Barratt, 2000), and therefore potentially less suitable to capture subtle differences due to genetic contributions. Nevertheless, previous reports showed that, in the general population, BIS-11 scores are not predicted by L-MAOA alleles, but rather by their epistatic interaction variants of the 5-HTTLPR polymorphism of the SLC6A4 gene (encoding for the serotonin transporter) (Stoltenberg et al. 2012). This finding is particularly interesting, in consideration of the role of the 5-HTTLPR as a genetic determinant for antisocial personality in convicts (Garcia et al. 2010).

In contrast with the associations of the L-MAOA genotypes with both violent crime and impulsivity, the lack of correlations between these two latter factors suggests that MAOA exerts multiple, mutually independent influences on these behavioral outcomes, possibly through the interplay with different genetic and/or environmental vulnerability factors.

Although previous studies have occasionally shown a direct association between aggression and L-MAOA alleles (Aslund et al. 2011; Sjöberg et al. 2007; Prom-Wormley et al. 2009), this relation was not apparent in our sample. These findings are in line with previous studies, which failed to identify an influence of MAOA-uVNTR genotype on BPAQ scores (Hurd et al. 2011). Although previous evidence has documented an association between L-MAOA alleles and high indices of antisocial personality, including anger and hostility (Yang et al. 2007; McDermott et al. 2009; Williams et al. 2009), ample evidence has shown that L-MAOA alleles do not confer an inherent predisposition to aggression (Fowler et al. 2007), but their influence on this trait is likely mediated by interactions with early exposure to traumatic experiences (Caspi et al. 2002; Kim-Cohen et al. 2006; Williams et al. 2009; Fergusson et al. 2011) and/or other concurrent risk factors (Beaver et al. 2011), including parenting style (Pickles et al. 2013).

Our analyses revealed a lack of association between criminal violence and either impulsivity or aggression scores. These results are in keeping with previous findings by Williams et al (1996), who failed to identify any significant differences in BPAQ scores between violent and non-violent offenders. While BPAQ total scores have been shown to predict the proclivity to engage in fights and violent acts in the general population (Buss and Perry, 1992; Archer et al. 1995), it should be noted that the nature of the crime does not accurately reflect the criminal history and pattern of aggressive conduct of inmates. This idea is indirectly supported by previous evidence documenting that, while the intensity of bullying behavior in incarcerated offenders is associated with BPAQ total scores (Palmer and Thakordas, 2005), it is not strongly predicted by the nature of the criminal charge (Ireland and Ireland, 2000). In addition, it should be observed that the relation between aggression levels and the violent nature of the crime may be modified by the actual exposure to the prison environment, which has been shown to significantly affect the social behavior of inmates (Bottoms, 1999; Edgar et al. 2003; Homel and Thompson, 2005).

L-MAOA variants have been typically associated with reactive aggression (McDermott et al. 2009), a trait characterized by irritability, reduced self-control, as well as maladaptive perceptions of ambiguous social cues and information processing deficits (Dodge and Coie, 1987; Dodge, 1991; Crick and Dodge, 1996; Scarpa and Raine, 1997; Volavka, 1999; Coccaro et al. 2007; Wilkowski and Robinson, 2008). Conversely, L-MAOA variants in inmates may be negatively associated with proactive aggression (Tikkanen et al. 2011), which features callous and unemotional conduct and instrumental violence (Frick and Ellis, 1999; Frick and White, 2008). Given the predominance of proactive-aggressive traits in violent offenders (Hare and McPherson, 1984), this background is apparently at variance with the robust association between L-MAOA and criminal violence documented by our results. Nevertheless, it should be noted that reactive and proactive aggression are often correlated (Vitaro and Brendgen, 2005) and share common genetic bases (Baker et al. 2008). Indeed, these overlapping heritable factors may be specifically related to the predisposition to physical aggression, a common underlying form between reactive and aggression (Little et al. 2003; Brendgen et al. 2006). Capitalizing on this notion, our results may indicate that the endophenotypic anomalies associated with L-MAOA alleles may lead to a higher predisposition for exteriorizing aspects of physical violence, which may be common to both reactive and proactive aggression.

Several limitations should be considered in the interpretation of the results of this study, such as the exclusive employment of self-reported measures, which may raise issues on the reliability of our findings. In addition, the conclusions of the study are limited by the relatively small sample size and the inclusion of only male convicts; nevertheless, it should be noted that recruitment of prison inmates for research purposes is extremely problematic, in view of significant legal and logistical hurdles, as well as important ethical limitations (Gostin et al. 2007). From this perspective, the number of subjects recruited for our study was comparable or higher than other recent genetic and investigations in this population (Garcia et al. 2010; Aluja et al. 2011).

While these caveats need to be addressed in future larger studies, it is important to underscore that our results provide the first demonstration of an association between L-MAOA variants and criminal violence in incarcerated offenders. While this evidence is in support of the importance of heritable biological components in violent crime, future studies on larger cohorts of subjects and non-convicted individuals are necessary to clarify the role of MAOA polymorphism in the predisposition to violence. Until then, it is clear that any application of our findings to the criminal justice system should be done with extreme caution, particularly in view of recent controversial judicial decisions in Italy and USA, in which genetic evidence on MAOA-uVNTR variants was used to justify sentence reductions (Baum, 2013). This type of interpretation appears premature, given our poor understanding of the neurobiological and cognitive underpinnings of violence predisposition. Thus, we strongly advocate extreme caution against any potential misrepresentations of the present results. Nevertheless, it appears that the present data support the idea that a significant fraction of violent crimes might be related to genetic predispositions, involving genes involved in serotonergic and dopaminergic neurotransmission. The incorporation of this concept in our current criminological framework could lead to a significant improvement in the development of interventional strategies for violent crime. In particular, the identification of potential biomarkers for risk of criminal violence may help enact preventive programs for highly predisposed youth, and result in a significant reduction of the staggering socio-economic burden of this important problem.

Supplementary Material

Acknowledgements

References