Childhood Maltreatment, Serotonin Transporter Gene, and Risk for Callous and Unemotional Traits: A Prospective Investigation

Abstract

Previous studies have reported associations between the serotonin transporter 5-HTTLPR genotype and antisocial and aggressive traits and between child maltreatment and antisocial traits. However, few studies have examined whether 5-HTTLPR moderates the influence of childhood maltreatment on callous and unemotional traits, a hallmark of psychopathy. Using a prospective cohort design, children with documented cases of maltreatment and matched controls were followed up and interviewed in adulthood. DNA was extracted from blood and saliva (N = 414) and callous-unemotional (CU) traits were assessed. Childhood maltreatment predicted higher CU scores in adulthood, whereas the effect of 5-HTTLPR was not significant. The effect of child maltreatment on CU traits did not differ by genetic risk (high or low activity 5-HTTLPR), whereas controls with the LL genotype had higher CU scores than controls with the SS genotype. Similar results were found for females and White, non-Hispanics, but not for males and Blacks. Variations in 5-HTTLPR did not affect the impact of child maltreatment on CU traits in adulthood. Genetic risk had a stronger effect on adults with lower environmental risk (controls). Having a history of child maltreatment or the LL genotype placed participants at risk for higher levels of callous and unemotional trait scores.

1. Introduction

Serotonin (5-HT) is a key central nervous system neurotransmitter involved in regulating a broad range of psychological traits, behaviors, and physical functions as well as cognitive and emotional states (Goldman et al., 2010). A functional polymorphism (5-HTTLPR) consists of a 44-base pair insertion or deletion resulting in two alleles, labeled long (l) and short (s) alleles (Heils et al., 1996). The 14-repeat short (s) allele has been shown to have less transcriptional activity and lower serotonin uptake than the 16-repeat long (l) allele (Heils et al., 1996).

Prior work has examined associations between 5-HTTLPR and antisocial and aggressive behaviors and gene × environment (childhood adversities) interactions predicting violent and antisocial behavior. Studies with individuals with antisocial, aggressive, or impulsive behaviors have reported significant serotonin deficiencies compared to healthy adults (Carver and Miller, 2006). Other studies have reported associations of 5-HTTLPR in incarcerated male criminals and violent offenders (Armstrong et al., 2017; Liao et al., 2004), adolescents with conduct disorder (Sakai et al., 2006), alcoholics (Hallikainen et al., 1999), and individuals with psychopathic traits (Fowler et al., 2009; Glenn, 2011; Sadeh et al., 2010; Sadeh et al., 2013) and antisocial personality disorder (ASPD) (Lyons-Ruth et al., 2007).

Findings have not been consistent. Some research has reported a significant relationship between the L allele and violent/aggressive traits (Armstrong et al., 2017; Canli and Lesch, 2007; Fowler et al., 2009; Glenn, 2011; Gyurak et al., 2013; Sadeh et al., 2010; Sadeh et al., 2013; Twitchell et al., 2001; Zalsman et al., 2001). Others have reported associations between some form of antisocial behavior and the S allele (Hallikainen et al., 1999; Liao et al., 2004; Lyons-Ruth et al., 2007; Sakai et al., 2007). Two studies of children found no relationship between the 5-HTTLPR genotype and aggression (Beitchman et al., 2003; Davidge et al., 2004) and a longitudinal study in adolescents showed no significant relationship between 5-HTTLPR genotype and conduct problems or delinquency (Sakai et al., 2007). Douglas et al. (2011) did not find an association between the low activity S allele and ASPD diagnosis. In a systematic meta-analysis of 19 published studies involving 5-HTTLPR, Vassos et al. (2014) did not find a significant association with aggression and violence.

Another body of research has examined interactions of 5-HTTLPR, antisocial and aggressive behavior and adverse childhood experiences. These studies have linked serotonin function to variations in violent behavior (Ficks and Waldman, 2014; Ortiz and Raine, 2004; Portnoy and Farrington, 2015; Tielbeek et al., 2016). Although not finding a main effect of 5-HTTLPR, Reif et al. (2007) reported that adverse childhood experiences (ACEs) were associated with violent behavior among subjects with one or two copies of the low-activity S allele. In a sample of children, Cicchetti et al. (2012) found that 5-HTT moderated the effects of maltreatment subtype on adult reports of antisocial behavior and the genetic effects were strongest for children who were abused. The findings of other candidate gene by environment interaction studies have generally been inconclusive and typically based on underpowered samples (Dick et al., 2015; Duncan and Keller, 2011; Okbay and Rietveld, 2015). Douglas et al. (2011) reported no overall 5-HTTLPR × ACEs interaction. However, there was evidence of moderation by the 5-HTT genotype in African American women. Sadeh, Javdani, and Verona (2013) found that carriers of the L/L genotype showed higher scores on the affective–interpersonal traits on the Psychopathy Checklist: Screening Version, but did not find that child maltreatment (using self-reports on the Childhood Trauma Questionnaire) interacted with 5-HTT to predict higher psychopathy scores. In a meta-analysis of studies examining interactions among 5-HTTLPR, ACES, and antisocial behavior, Tielbeek et al. (2016) found significant interaction effects; however, they concluded: “methodological constraints of the included studies hampered a confident interpretation of our results, and firm conclusions regarding the direction of the effect” (p. 748).

Noticeably fewer studies have examined associations between 5-HTTLPR and callous and unemotional (CU) traits, although CU traits are thought to be a hallmark of psychopathy and genetically influenced (Blair et al., 2006; Frick and Viding, 2009; Viding and McCrory, 2012). CU traits are characterized by low empathy, callousness, and low interpersonal emotions (Frick and Ray, 2015) and elevated levels of these traits have been found to identify a particularly severe subgroup of antisocial youth (Frick et al., 2014; McMahon et al., 2010) leading to their inclusion in the most recent edition of the DSM (American Psychiatric Association, 2013). In adult samples, the CU dimension appears to be most specific to individuals high on psychopathic traits compared to other antisocial individuals (Cooke and Michie, 1997).

A separate body of literature has examined whether dysfunctional parenting practices are related to CU traits (Barker et al., 2011; Waller et al., 2012). In one prospective longitudinal study of a population-based sample, harsh parenting at age 4 significantly predicted CU traits at age 13, accounting for 10% and 14% of the variance in these traits in boys and girls, respectively (Barker et al., 2011). Harsh, inconsistent, and coercive discipline has been shown to be associated with conduct problems in youths with normative levels of CU traits (Edens et al., 2008; Hipwell et al., 2007; Oxford et al., 2003; Pasalich et al., 2011; Wootton et al., 1997; Yeh et al., 2011). Low warmth in parenting has been associated with conduct problems in youths with elevated CU traits (Kroneman et al., 2011; Pasalich et al., 2011), but not in others (Falk and Lee, 2012). Viding, Fontaine, Oliver, & Plomin (2009) found that negative discipline at age 7 was related to conduct problems at age 12 but not to CU traits at age 12, after controlling for initial levels of outcomes.

In sum, although previous research has reported associations between 5-HTTLPR and antisocial and aggressive behaviors and interactions among 5-HTTLPR, adverse childhood experiences, and antisocial and violent behaviors, findings have not been consistent. Mixed results may have resulted from differences in sample characteristics, statistical power, and definitions of key variables. Adverse childhood experiences have generally been assessed via self-reports, which carry limitations (Baldwin et al., 2019) and little research has examined whether 5-HTTLPR interacts specifically with childhood maltreatment to predict CU traits. Because childhood maltreatment has been associated with higher levels of delinquent and violent criminal behavior (Maxfield and Widom, 1996; Widom, 1989b) and psychopathy (Weiler and Widom, 1996), individuals with histories of child maltreatment should have higher levels of CU traits. And, based on some of the prior literature, it is expected that 5-HTTLPR will moderate the relationship between childhood maltreatment and CU traits.

1.1. Purpose

The purpose of this study was to determine whether 5-HTTLPR moderates the influence of childhood maltreatment on callous and unemotional traits using documented cases of childhood maltreatment in the context of a prospective longitudinal design. As noted in our earlier paper on the role of MAOA in predicting risk for violent and antisocial behavior (Widom and Brzustowicz, 2006), the research design used here offers several advantages over prior studies for testing the gene (5-HTTLPR) × environment (child maltreatment) interaction. First, this design involves a clear operationalization and lack of ambiguity about the child maltreatment in our sample by using court cases with documentation of child abuse and neglect, rather than measures relying on retrospective self-reports. Second, there is a comparison group matched on the basis of age, sex, race/ethnicity and approximate social class background (Widom, 1989b). The comparison group establishes the base rates of pathology we would expect in a sample of adults from comparable circumstances who did not come to court attention in childhood as victims of abuse or neglect. Third, earlier research is limited in terms of age and homogeneity, and here we examine these relationships in a more diverse population, including males, females, Whites, and non-whites. Finally, we have followed our participants beyond adolescence so that we can compare risks for callous and unemotional traits in adulthood.

We have three hypotheses: (1) individuals with documented histories of child maltreatment will have higher levels of callous and unemotional traits compared to individuals without those histories; (2) individuals who are not carriers of the short allele, that is, those with the L/L genotype, will be at risk for higher levels of callous and unemotional traits; and (3) there will be a gene (5-HTTLPR) by environment (child maltreatment) interaction such that individuals with histories of child maltreatment who carry the LL genotype will have higher levels of callous and unemotional traits than those without those genotypes.

2. Methods

2.1. Design and Participants

The data were collected as part of a prospective cohort design study (Leventhal, 1982; Schulsinger et al., 1981) in which abused and neglected children were matched with nonabused and nonneglected children and followed into adulthood. Because of the matching procedure, the participants are assumed to differ only in the risk factor, that is, having experienced childhood maltreatment. Because it is not possible to assign participants randomly to groups, the assumption of equivalency for the groups is an approximation. The control group may also differ from the abused and neglected individuals on other variables nested with abuse or neglect. For complete details of the study design and participant selection criteria, see Widom (1989a).

The original sample of abused and neglected children (N = 908) was composed of all substantiated cases of childhood physical and sexual abuse and neglect processed from 1967 to 1971 in the county juvenile (family) or adult criminal courts of a Midwestern metropolitan area. Cases of abuse and neglect were restricted to children 11 years of age or younger at the time of the incident.

A control group of children without documented histories of childhood abuse or neglect (N = 667) was matched with the maltreated group on age, gender, race/ethnicity, and approximate family social class during the same period. This matching was important, because it is theoretically plausible that any relationship between child abuse and neglect and subsequent outcomes is confounded with or explained by social class differences (Bradley and Corwyn, 2002; Conroy et al., 2010; MacMillan et al., 2001; Widom, 1989a). The matching procedure is based on a broad definition of social class that includes neighborhoods the children were reared in and schools they attended (Watt, 1972). Children who were younger than school age at the time of the abuse or neglect were matched with children of the same gender, race, date of birth (±1 week) and hospital of birth using county birth record information. For children of school age, we used records of more than 100 elementary schools for the same period to find matches with children of the same gender, race, date of birth (±6 months), and class in elementary school during the years 1967–1971. Overall, matches were found for 74% of the maltreated children.

The initial phase of the study was archival and compared the abused and neglected and control children on juvenile and adult criminal arrest records (Widom, 1989b). Subsequent phases involved locating and interviewing both groups during 1989–1995 (N = 1196), 2000–2002 (N = 896), 2003–2005 (N = 808), and 2009–2010 (N = 649).

The current sample is mean age 47.6 years old (SD = 3.5, range = 37–57) and represents those for whom we have both DNA and assessment of CU traits. Approximately half of the sample is male (44.9%), about two-thirds are White, non-Hispanic (63.5%), and 36.5% Black. Participants were asked to self-identify their race and ethnicity. Individuals self-identified as other than Black or White (n = 38) were excluded because this group was too small for meaningful analysis.

Although there was attrition associated with death, refusals, and our inability to locate individuals over the various waves of the study, the composition of the sample has remained about the same. There were no significant differences between the groups in terms of sex, race/ethnicity, age, and maltreatment group status across the phases of the study. To assess the representativeness of the individuals in the current analysis, characteristics of the people included in this analysis were compared with individuals not included. Table 1 shows that there were no significant differences between the included and excluded groups on these demographic characteristics.

Table 1.

Characteristics of included and not included participants

	Remain in the sample	Lost in Int. 4
	N = 414	N = 116	Chi Square	df	p
Maltreated	237 (57.2%)	73 (62.9%)	0.983	1	0.321
Female	228 (55.1%)	61 (52.6%)	0.137	1	0.712
Black	151 (36.5%)	38 (32.8%)	0.395	1	0.530
White	263 (63.5%)	78 (67.2%)
	M (SD)	M(SD)	T score
Age at interview 4	47.6 (3.5)	47.3 (3.5)	0.691	181.76	0.485
Age at petition	6.4 (3.2)	5.8 (3.4)	1.427	176.20	0.155

Note: M = mean; SD = standard deviation; df = degrees of freedom.

2.2. Definition of child abuse and neglect

The maltreatment group in this sample includes court substantiated cases of childhood physical and sexual abuse and neglect. Physical abuse cases included injuries such as bruises, welts, burns, abrasions, lacerations, wounds, cuts, and bone and skull fractures. Sexual abuse cases included felony sexual assault, fondling or touching, sodomy, incest, and rape. Neglect cases reflected a judgment that the parents’ deficiencies in child care were beyond those found acceptable by community and professional standards at the time and represented extreme failure to provide adequate food, clothing, shelter, and medical attention to children.

2.2. Assessment of 5-HTTLPR Genotype

Blood and/or saliva samples were collected from subjects during two waves of interviews (2003 and 2009–2010). In 2003–2005, as part of a medical status examination (including blood collection through venipuncture, at mean age 41, 638 (82%) consented to provide blood and, of these, 623 gave permission for DNA extraction and analyses (Widom and Brzustowicz, 2006). IRB restrictions prevented the collection of blood from an additional 31 study participants who were residing in prisons at the time. DNA was obtained from usable blood samples from 617 study members and extracted using the PureGene (Gentra Systems Inc) system according to the manufacturer’s instructions. In addition, in 2009–2010, participants were asked to provide DNA through saliva using Oragene kits with standard procedures.

When both blood and saliva DNA were available for a given participant, DNA from blood was preferentially used for genotyping. Of those with phenotypic data from both interview waves, 443 participants had available genotypes (404 from blood DNA, 39 from saliva DNA). For this analysis, participants who identified as Hispanic or races other than White or Black were excluded, leaving a total sample of (N=414).

DNA samples were genotyped using a slight modification of the protocol detailed by Wendland et al. (2006) to determine the variants of 5-HTTLPR. In a total volume of 25 ul, 40 ng of genomic DNA were amplified in the presence of 4% dimethyl sulfoxide, 2 U of AmpliTaq Gold polymerase, 1x Buffer II, and 1.25 mM Magnesium Chloride (Applied Biosystems, Foster City, CA) and oligonucleotide primers (Forward: 5’-TCCTCCGCTTTGGCGCCTCTTCC-3’; Reverse: 5’-TGGGGGTTGCAGGGGAGATCCTG-3’, Integrated DNA Technologies, Coralville, Iowa) at final concentrations of 300 nM. Thermal cycling consisted of 15 min of initial denaturation at 95° C followed by 40 cycles of 94° C (30 s), 66° C (90 s) and 72° C (60 s) each with a final extension step of 10 min at 72° C. Subsequently, 7 ul of PCR product were digested by 5 U of HpaII (an isoschizomer of MspI) (New England Biolabs, Ipswich, MA) in a 20 ul reaction assay containing 1 × NEBuffer 1 and 1 × BSA at 37° C for 3 h. Finally, 5 ul of remaining PCR product and 15 ul of restriction enzyme assay solution were loaded onto a 2% GeneMate LE agarose gel (VWR, Radnor, PA), run for 3 h at 150 V in TBE and visualized by SYBR Gold (Invitrogen, Carlsbad, CA).

2.3. Assessment of Callous and Unemotional Traits

CU traits were assessed using the Callous-Unemotional scale of the Antisocial Process Screening Device (APSD) (Frick and Hare, 2001) during the 2009–2010 interviews. This is a widely used instrument to assess characteristics of empathy, lack of remorse, emotional expressiveness, apathy, emotional sensitivity, interpersonal difficulties, and lack of goals. Items were scored as follows: 1 = not true at all; 2 = sometimes true; and 3 = definitely true. There is no established threshold score on the APSD for classification of individuals with CU traits (Frick et al., 1999). The CU scale has demonstrated low to moderate internal consistency in past studies (Loney et al., 2003; Pardini et al., 2003; Poythress et al., 2006). In the current study, internal reliability of the CU scale (Cronbach’s Alpha) was .55, likely due to the small number of items (n = 6) and three-point rating system. The mean CU trait score for the overall sample was 8.41, SD = 1.82, and range = 6 – 16.

2.4. Analysis

Chi square analyses were used to assess characteristics of the sample and the distribution of genotype frequency and t-tests were used to compare callous and unemotional trait scores for maltreated and control participants. A series of linear regression analyses were then conducted to examine main effects and gene (5-HTTLPR) × environment (child maltreatment) interactions predicting CU trait scores. All analyses control for the age, sex, race of the participant, and age at the time of the abuse. Sensitivity analyses were conducted for males, females, Blacks, and Whites separately. Population stratification analysis was undertaken and Bayesian Posterior Probabilities (BPPs) were calculated to assess the probability of being in a given ancestry group. These probabilities were estimated using the method implemented in the STRUCTURE program (Pritchard et al., 2000). A logistic regression was run to determine the (residual) phenotype in which the population sub-structure is controlled. None of the BPPs was significant and the inclusion of BPPs did not change the results.

3. Results

Table 2 shows the overall distribution of 5-HTTLPR genotypes for the maltreated and control groups. Overall, the expected frequencies of the 5-HTTLPR genotype in the sample were 0.156 for “SS,” 0.478 for “SL,” and 0.366 for “LL.” A goodness of fit test revealed no evidence of deviation from Hardy-Weinberg equilibrium (X² = 0.102, df = 1, p = 0.749). There was also no significant difference in the distribution of 5-HTTLPR genotypes between the maltreated and control groups (X² = 3.6113 df = 2, p = 0.1644). Individuals with two copies of the 16-repeat long (l) allele were designated as high activity, labeled “LL”, while individuals with two copies of the 14-repeat short (s) allele were designated as low activity, labeled “SS”. Individuals with one copy of the short and one copy of the long allele were designated as intermediate activity, labeled “LS.”

Table 2.

Overall distribution of 5-HTTLPR genotypes

Genotype	Total	Control	Maltreated	Chi square	p
	N = 414	N = 177	N = 237
LL	155 (37.4%)	70 (39.5%)	85 (35.9%)	3.61	0.164
LS	191 (46.1%)	85 (48%)	106 (44.7%)
SS	68 (16.4%)	22 (12.4%)	46 (19.4%)

Note: LL = high transcriptional activity; LS = intermediate transcriptional activity; SS = low transcriptional activity. Degrees of freedom = 2.

Table 3 shows the CU trait scores for the maltreated and control groups overall and for females, males, White, non-Hispanics, and Black, non-Hispanics separately. For every comparison between the maltreated and control group, the differences in CU scores were significant except for the males, where the results showed a non-significant trend.

Table 3.

Mean callous and unemotional trait scores for maltreated and control groups followed up into adulthood

	Control			Maltreated
Group	N	M	SD	N	M	SD	T score	df	p value
Overall	177	8.13	1.46	237	8.66	2.01	3.094	411.7	0.002
Female	92	7.97	1.18	136	8.59	1.93	3.040	224.2	0.003
Male	85	8.31	1.69	101	8.75	2.12	1.559	183.5	0.121
White	112	7.86	1.25	151	8.36	1.88	2.556	258.1	0.011
Black	65	8.59	1.67	86	9.19	2.13	1.923	148.7	0.056

Note: M = mean; SD = standard deviation; df = degrees of freedom.

Using linear regression analysis, we found that childhood maltreatment predicted higher CU trait scores in adulthood, whereas the effect of 5-HTTLPR was not significant (see Table 4). There was also a significant interaction showing that the effect of child maltreatment on CU trait scores did not differ by genetic risk (high or low activity 5-HTTLPR). However, the results (see Figure 1) show that controls with LL had higher CU trait scores, whereas controls with the short 5-HTTLPR genotype (SS) had significantly lower CU traits (beta = 1.11, SE = 0.54, p = 0.04).

Table 4.

Results of linear regression analyses predicting callous and unemotional trait scores

	Overall Sample			Females			Males			Whites			Blacks
	(N = 414)			(N =228)			(N = 186)			(N = 263)			(N = 151)
	β	SE	p	β	SE	p	β	SE	p	β	SE	p	β	SE	p
(Intercept)	9.35	1.57	0.00	7.64	1.90	0.00	11.24	2.62	0.00	8.25	1.86	0.00	10.45	3.03	0.00
Child maltreatment	0.31	0.28	0.28	0.11	0.34	0.75	0.56	0.49	0.25	0.15	0.38	0.69	0.47	0.45	0.30
Genotype: LS vs LL	0.23	0.29	0.41	0.29	0.37	0.43	0.10	0.46	0.82	0.17	0.35	0.62	0.26	0.52	0.62
Genotype: SS vs LL	−0.56	0.43	0.19	−0.37	0.49	0.45	−0.85	0.78	0.28	−0.50	0.50	0.31	−0.76	0.82	0.36
Female	−0.23	0.18	0.20							−0.25	0.21	0.22	−0.20	0.33	0.54
White	−0.90	0.19	0.00	−1.02	0.23	0.00	−0.73	0.31	0.02
Age	0.01	0.03	0.69	0.01	0.03	0.64	0.04	0.04	0.34	0.01	0.03	0.86	0.03	0.05	0.55
CM * LS	0.18	0.38	0.64	0.59	0.48	0.22	−0.35	0.63	0.58	0.25	0.48	0.60	0.26	0.70	0.71
CM * SS	1.11	0.54	0.04	1.31	0.64	0.04	0.90	0.94	0.34	1.19	0.63	0.06	0.92	1.10	0.40

Note: β = beta; SE = standard error; CM = child maltreatment. LL = high transcriptional activity; LS = intermediate transcriptional activity; SS = low transcriptional activity. Reference group for genotype is LL.

Figure 1.

Scores represent the mean callous or unemotional trait scores as a function of 5-HTTLPR activity and documented history of child maltreatment or control group. 5-HTTLPR is divided into high (“LL”), intermediate (“LS”), and low (“SS”) transcriptional activity based on genotype at the promoter 5-HT polymorphism. Maltreatment status is based on documented court cases. The y axis shows mean callous and unemotional trait scores. There was a significant interaction; however, genotype did not moderate the relationship between child maltreatment and CU trait scores. The interaction showed that controls with the high 5-HTTLPR genotype (LL and LS) had significantly higher CU trait scores, whereas controls with the short 5-HTTLPR genotype (SS) had significantly lower CU trait scores (beta = 1.11, SE = 0.54, p = 0.04).

The results were similar when we restricted the analyses by sex and race (see Table 4 and Figure 2). For females and Whites, controls with the high 5-HTTLPR activity (LL) had higher CU trait scores, whereas controls with the short 5-HTTLPR genotype (SS) had significantly lower CU trait scores (for females, beta = 1.31, SE = 0.64, p = 0.04; for Whites, beta = 1.19, SE = 0.63, p = 0.06). The interaction was not significant for males or Blacks.

Figure 2.

Scores represent the mean of scores of callous or unemotional traits as a function of 5-HTTLPR activity and documented history of child abuse and neglect or control group for females, males, Whites, and Blacks separately. 5-HTTLPR is divided into high (“LL”), intermediate (“LS”), and low (“SS”) transcriptional activity based on genotype at the promoter 5-HT polymorphism. Child maltreatment status is based on documented court cases. The y axis shows mean callous and unemotional trait scores in adulthood. There were significant interactions for females (beta = 1.31, SE = 0.64, p = 0.04) and Whites (beta = 1.19, SE = 0.63, p = 0.06), but not for males or Blacks. For females and Whites, genotype did not moderate the relationship between child maltreatment and CU trait scores. However, controls with LL or LS had higher CU trait scores, whereas controls with the short 5-HTTLPR genotype (SS) had significantly lower CU trait scores.

4. Discussion

Using a prospective cohorts design with children with documented histories of maltreatment and matched controls who were followed up into adulthood, our findings were not entirely as expected. Our first two hypotheses were supported, but the third was not.

First, these new results show clearly that individuals with documented histories are at increased risk for higher levels of CU traits in adulthood compared to individuals without those histories. Although previous research has examined the role of dysfunctional parenting practices on CU traits, this work has been almost exclusively with children and adolescents. These new findings add to the existing literature on the development of CU traits and the “cycle of violence”. Researchers are increasingly examining developmental trajectories for antisocial behavior and factors that may lead to stability and change over time in community samples (Fanti et al., 2017; Fontaine et al., 2010; Hyde et al., 2015; Klingzell et al., 2016) and justice involved youth (Ray et al., 2019). Frick, Ray, Thornton, and Kahn (2014) reviewed 12 studies testing the stability of these CU traits in childhood and adolescence and concluded that there was only modest levels of stability of these traits prior to adulthood. These new findings suggest that maltreated children who manifest CU traits in childhood may continue to manifest these traits in adulthood unless positive interventions or protective factors deflect them from this negative trajectory.

Our findings did not provide support for the hypothesis that the effect of child maltreatment on callous-unemotional traits would be strongest for those at high genetic risk. That is, we did not find that the LL genotype moderated the impact of child maltreatment on CU traits in adulthood. This result is in direct contrast to two previous studies. Cicchetti et al. (2012) studied school-aged children and defined child maltreatment based on official records, similar to the current study, but found that L decreased antisocial behavior when interacting with child maltreatment. Li and Lee (2010) used retrospective reports of maltreatment in a random sample of teenagers and found that L decreased antisocial behavior in females when interacted with maltreatment. One can only speculate on reasons for the differences between the current and these prior studies, but one important factor to consider is that these studies focused on children and youth, whereas here CU traits were assessed in middle adulthood. Another difference is that the current work focused on CU traits, whereas Cicchetti et al. (2012) and Li and Lee (2010) focused on the broader construct of antisocial behavior. A third possibility, suggested by an anonymous reviewer, is that childhood maltreatment may have suppressed heightened emotional reactivity among the SS carriers in the maltreatment group.

Our findings are similar to Sadeh, Javdani, and Verona (2013). The results of the current study suggest that having a history of child abuse and neglect or the LL genotype (associated with higher risk) places them at risk for a higher level of callous and unemotional trait scores. In contrast to our hypothesis, we found a link between the higher transcriptional activity long genotype of 5-HTTLPR and callous and unemotional trait scores, but only in the controls. The controls with the L/L genotype had higher levels of CU traits compared to carriers of the S/S genotype. The CU trait levels for the controls with the LL genotype here were more similar to the higher levels of CU traits shown by individuals with histories of child maltreatment.

These results suggest that genetic risk may have the strongest impact on individuals who are at relatively lower environmental risk. Previous research has not been conducted with children with documented histories of maltreatment and particularly those whose cases have come to the attention of the authorities who by all measures are at high environmental risk. For these children, our results suggest that genetic risk does not confer additional vulnerability -- at least for callous and unemotional traits in adulthood. These new results suggest that environmental (abuse/neglect) or genetic factors (5-HTTLPR) can lead to a similar behavioral pattern, but the presence of both factors do not combine to make the behavior worse.

These results contrast with our earlier work (Widom and Brzustowicz, 2006) where we found that lower levels of MAOA, a gene of similar interest to 5-HTT, were associated with an increased risk of violent and antisocial behaviors in previously abused and neglected individuals. However, the earlier work examined both a different gene (MAOA versus 5-HTTLPR) and different outcome (violent and antisocial behaviors versus callous and unemotional trait scores).

4.1. Strengths and limitations

Despite the many strengths of this study, including the use of a unique data set on the long-term consequences of child maltreatment; a prospective longitudinal design that spans over 30 years from childhood to young adulthood to middle adulthood; a clear operationalization of child maltreatment; and a matched control group of children also followed up, a number of limitations should be noted. Because these cases were identified through the courts, these findings are not generalizable to unreported or unsubstantiated cases of maltreatment. However, the strength of this design is the “objective” nature of the cases that were documented at the time. A recent meta-analysis of studies on child maltreatment outcomes concluded that prospective and retrospective measures cannot be used interchangeably to study associations with health outcomes (Baldwin et al., 2019) and other recent work has reported substantial differences between results when objective reports and subjective recollections of childhood maltreatment are used in relation to psychopathology in adulthood (Danese & Widom, 2020). Because both groups in this study are predominantly from the lower end of the socioeconomic spectrum, these findings cannot be generalized to cases of abuse and neglect occurring in middle or upper class families. This caveat is important because there is an increasing literature on the relationships among family socioeconomic status, 5-HTTLPR, and delinquent behavior (Aslund et al., 2013).

This study represents the experiences of children growing up in the late 1960s and early 1970s in the Midwest part of the United States and may raise concerns about the relevance of these cases to current cases. However, the cases studied here are quite similar to current cases being processed by the child protection system and the courts. One difference is that these children were not provided with extensive services or treatment options as are available today and, thus, the results of this study represent the natural history of the development of maltreated children whose cases have come to the attention of the courts. Although some types of child maltreatment have shown different consequences (e.g., Widom et al., 2012; Widom et al., 2018; Wilson and Widom, 2009), an examination of the differences by type of maltreatment would be worthwhile, but was not possible because of limitations of the sample size by type. Finally, we were not able to control for the potential impact of parental genotype (passive rGE) or the possibility that at a child with a certain genotype might evoke negative parenting (evocative rGE), although this would be an important consideration for the future.

Highlights.

• Although associations have been reported between the serotonin transporter 5-HTTLPR genotype and antisocial and aggressive traits and between child maltreatment and antisocial traits, few studies have examined whether 5-HTTLPR moderates the influence of childhood maltreatment on these traits.

• This unique investigation uses data from a prospective longitudinal design study, documented cases of childhood maltreatment, a demographically matched control group, and assessment of callous-unemotional traits in adulthood.

• Childhood maltreatment predicted higher callous-unemotional traits scores in adulthood, but the effect of child maltreatment on callous-unemotional trait scores did not differ by genetic risk.

• In contrast, matched controls (non-maltreated) with the LL (high activity) genotype had higher callous-unemotional trait scores.

• Having a history of childhood maltreatment or the LL genotype (associated with higher risk) predicted higher levels of callous and unemotional trait scores in adulthood.