Age-Related Variations of Genetic and Environmental Contributions to the Covariation of Fear, Distress and Externalizing Symptoms: A Twin Study in Childhood and Adolescence

The study of mental disorders in childhood and adolescence is gaining more and more importance in mental health research, given their high prevalence rates in young ages. It has been estimated, in fact, that about 10–20% of young people worldwide are affected by this kind of pathologies, with symptoms that usually persist throughout adulthood [1].

Since the birth of modern psychiatry, mental illnesses have always been categorized as single units, identified on the basis of precise diagnostic criteria (i.e., characteristics and severity of symptoms and their trajectories over time), in order to facilitate the identification of the physical etiological factors specifically linked to each disorder [2].

However, although it has always played a primary role in mental health clinical research and practice over the years, this categorial approach went through numerous criticisms, which started to question its diagnostic reliability. One of the main challenges that this diagnostic system had to face, starting from the DSM-III, was the comorbidity (i.e., co-occurrence) among disorders, a phenomenon that started to be increasingly reported in epidemiological psychiatric studies [3].

The significant frequency of comorbid cases soon led clinicians to reconsider the structure itself of mental disorders, opening to the possibility that these disorders might be read in the light of a more parsimonious model in which they are grouped into macro-categories, rather than considered as independent entities [3]. In one of the first studies that tried to introduce a new interpretation of the pre-existing taxonomic system [4], a confirmatory factor analysis (CFA) was conducted to analyze the correlations among ten of the most common mental disorders, assessed in a probabilistic sample through a structured interview based on DSM-III-R diagnostic criteria. Results showed that the best structural model explaining the comorbidity links among the disorders was indeed a model including three factors, namely Anxious-Misery, Fear and Externalizing, with the first two being highly correlated (r = 0.73) and, therefore, possibly reflecting a single higher-order construct, namely the Internalizing domain. It should be noted that, over the years, the label Anxious-Misery has been replaced with Distress, mainly to emphasize the pervasive yet heterogeneous sensation of subjective stress of individuals whose symptoms fall within this category [5]. More specifically, Anxious-Misery and Fear usually refer to symptoms that lead to an impairment that is less evident from the outside, more strictly intrapsychic, while the label Externalizing is commonly used to cluster those behaviors that negatively affect individuals’ interaction patterns with the environment [6].

Further support of this model came from the results of a meta-analytic work [7]. Analyzing the existing literature and comparing the various structural models through model-fitting procedures, the authors showed that a three-factor model provided the best fit to the data. Moreover, they further stressed the division of Internalizing symptoms into the lower-order categories of Distress, in case they were linked to major depression, dysthymia and generalized anxiety disorder, and of Fear, when they were associated to more specific anxiety conditions.

The co-occurrence of disorders within each of the categories composing the three-factor model is a widely known phenomenon [8, 9]. Nevertheless, recent findings in the epidemiological field also showed consistent comorbidity rates between symptoms belonging to different higher-order variables of the structural model, in children-adolescents and in adults, in both clinical and general population samples [8‐11].

The observed correlational bond between symptoms belonging to Fear, Distress and Externalizing domains led some researchers to hypothesize that, behind these constructs, there might be a single common susceptibility [8, 11]. In particular, Caspi et al. [8] theorized a structural model of psychopathology consisting of a pyramidal structure. At the basis of the pyramid there would be the disorders conceived as independent entities, while the second step of the structure would be characterized by the Internalizing (hence its lower-order domains) and Externalizing categories that eventually, on the top, would merge into a single continuous dimension of psychopathology, that is the so called “p factor”. From a clinical point of view, a higher p-score will correspond to a worse assessment of the individual with respect to the severity of the disorder, the sequential comorbidity, the developmental history, and the impairment in adult life [8]. Although this hypothesis has been supported by several subsequent studies, it still remains to be established whether the common liability is a stable phenomenon or it is affected by variations due to age. There is no doubt that the aging process can cause modifications in developmental trajectories of mental disorders [12], with a general trend characterized by a decreasing of Externalizing problems and an increasing of generalized anxiety and depressive symptoms with age [13]. Developmental trajectories of this type of symptoms have also been targeted by twin studies, which allowed to investigate the nature of the factors responsible for their stability and their changes over time. In particular, some of these studies showed that genetic factors underlying the symptoms can not only influence their stability but also contribute to their age-related modifications [14, 15]. The notion that genes exert a continuous influence on certain phenotypes led several authors to ask whether the genetic and environmental structure of Fear, Distress and Externalizing symptoms could justify their inclusion in the three-factor model independently from age. In this respect, the results of a work by Waszczuk et al. showed substantial age-related changes in the genetic influence on Internalizing symptoms. These authors found that, during childhood, there was only one single common genetic factor behind all the anxiety disorders, which gave an irrelevant contribution to the variance of depressive symptoms; instead, common genetic influences underlying both anxiety (especially generalized anxiety disorder) and depression were found starting from adolescence and became more and more evident in adulthood, suggesting the idea that the effect of common biological influences can be observed concurrently with the onset of puberty [16].

Thus, in the light of previous literature mainly regarding the reformulation of structural models of psychopathology, in our study we set two goals: (1) to test the hypothesis of the existence of a common latent susceptibility behind the co-occurrence between Fear, Distress and Externalizing symptoms; (2) to test the stability of the common susceptibility in relation to individuals’ age.

To these aims, a twin sample of children and adolescents of Caucasian origin, living in different Italian regions, was stratified into two age groups, namely 6–14 and 15–18 years, in order to mirror, as best as possible, the division made by Waszczuk et al. [16] in their work.

Table 1 reports mean values of Fear, Distress and Externalizing scales in twins considered as individuals and divided by gender, zygosity and age group. Comparisons by t-tests showed higher scores in Fear and Distress symptoms for girls compared to boys, with an opposite pattern for Externalizing symptoms. Age-related differences in mean scores were found only for Distress symptoms, higher in the 15–18 group, while no zygosity differences were detected.

Our results supported the co-occurrence between Internalizing (i.e., Fear-Distress) and Externalizing symptoms, with substantial correlation rates among the three analyzed phenotypes in the 6–14 and 15–18 age groups. This was consistent with previous studies of psychopathology in childhood and adolescence [8‐11], and was in line with available theories on the comorbidity bond existing between these diagnostic domains [3]. Multivariate model-fitting analyses showed a null contribution of shared environment on all the traits in both age groups. Although this finding in the 15–18 age group might not seem surprising, the lack of evidence of common environmental effects in the younger group should be discussed, as the influence of shared environment tends to play a predominant role during childhood [36]. One possible explanation could be that the 6–14 age group may not be strictly representative of childhood age, as it includes adolescents of 11–14 years who may have caused a considerable reduction in the shared (familial) influence on the traits. To test this hypothesis, we further stratified our sample by age, identifying three subsamples, namely 6–10, 11–14 and 15–18 years old. As expected, cross-twin/within-trait correlational trends obtained in the three subsamples separately did unravel a possible role of common environment on the traits during infancy, with lower values of correlations in the 6–10 MZ group for all the three phenotypes (data not shown).

With respect to the main aim of our study, results from multivariate modelling suggested the existence of a common susceptibility factor behind the co-occurrence between Fear, Distress and Externalizing symptoms, which, however, seemed to become evident only starting from adolescence. In fact, the best-fitting models found in the two age groups were indicative of an age-related variation of the genetic and environmental influences on the phenotypes. More specifically, the Cholesky AE model that best fitted the data in the 6–14 age group suggested that, although highly correlated among the traits, genetic and environmental influences on Fear, Distress and Externalizing symptoms may still act through distinct patterns at this age. On the other hand, the best-fitting Common Pathway AE model that was found in the 15–18 age group indicated that the etiological influences on all the phenotypes may begin to acquire a common nature over time. These findings are consistent with those of Waszczuk et al. [16], who considered only Internalizing-like symptoms, and suggested a delay in the effect of new genetic influences coming into play with puberty. Thus, our study would show that the same pattern of etiological variation, with a similar interpretation as in the above-mentioned work, might hold when Externalizing symptoms are included in the experimental design, beside the Internalizing dimensions of Fear and Distress. Our finding on common etiological substrates behind the co-occurrence of the three symptoms domains is consistent with the hypothesis on the existence of a unique latent factor predisposing to psychopathology [8]. Similar evidence supporting the “p factor” of psychopathology can be found in Allegrini et al. [37]; in this study, authors reported a common latent variable, underlying the targeted symptoms, that was heritable at 50–60% [37]. A highly interesting consideration about the phenotypic nature of the general psychopathology factor relates this factor to personality traits [3, 37, 38]. In fact, it has been reported by several studies that many childhood disorders belonging both to the internalizing and the externalizing domain might be more faithfully described in terms of personality traits rather than categorical dimensions [39]. From a structural point of view, personality is constituted by two major domains: temperament and character [40]. Temperament can be defined as the portion of personality that is highly heritable, while character is mainly influenced by the environment. Personality is the result of the constant interaction of these two domains, which extensively occurs during development. That said, it appears clear how much personality can be regarded as something that is strictly connected to the neurobiological substratum of individuals, hence to their genes [40]. In the light of this, the most recent psychiatric genetics findings on the highly genetic nature of the general factor of psychopathology undoubtably highlight the importance of taking into account personality when theorizing a structural model of mental disorders that aims to be as accurate as possible [41]. From a developmental perspective, personality traits tend to increase in maturity with age. More specifically, previous studies have found that functional traits, such as conscientiousness and agreeableness, normally increase, whereas dysfunctional ones, such as neuroticism, tend to decrease over time [42]. Nevertheless, during early adolescence, this tendency towards improvement can change in the opposite direction, causing a temporary decline in personality maturation [42]. Adolescence is considered one of the most critical periods for personality development [42, 43], as it is also the usual onset age of personality disorders when in conjunction with particular circumstances, such as Internalizing or Externalizing disorders not properly treated [43, 44]. One of the main tasks of adolescence is, in fact, the development of one’s own identity, mainly through the social comparison with peers as a mean of self-evaluation [43]. From a neurobiological standpoint, this process is supported by the dorsal medial prefrontal cortex (MPFC), a brain region that starts developing during adolescence, and that allows the interpretation of social stimuli in a self-reflective key, making the experience of self-conscious emotions and autonomic arousal possible [43]. Our results seem consistent with this mechanism, as they suggest that the (highly genetically determined) influence of the latent susceptibility, common to Fear, Distress and Externalizing symptoms, may come effectively on-line during adolescence. Trying to be more specific about the psychological nature of the latent factor, numerous studies have identified neuroticism as the most likely personality trait related to Internalizing and Externalizing problems [45, 46]. Neuroticism (or negative affect) is defined as the tendency to experience frequent negative emotions, such as anger, sadness, guilt, and nervousness [47], which usually leads to frequent worry, emotional avoidance, and rumination [48]. High levels of neuroticism could lead to biased interpretations of the social stimuli, making people perceive ordinary situations as unreasonably threatening [49]. A very intriguing hypothesis, deriving from several twin studies that found a significant genetic and environmental overlap of neuroticism with Internalizing and Externalizing problems, suggests that the unique latent susceptibility factor laying behind the comorbidity between these clusters of symptoms might be represented by neuroticism itself [50‐52]. In the light of the developmental trajectories of this dispositional trait, our results could be consistent with this hypothesis; in fact, like other dysfunctional personality traits, neuroticism seems to increase during adolescence, as it has been recently reported for girls [42], and this could explain why, in our study, the latent factor becomes more evident during this age. Some limitations should be taken into account when interpreting our results. First, we used only CBCL/6–18 to assess symptoms. Even though this instrument has proved to be highly reliable in both clinical and non-clinical populations at a multicultural level [20, 53], it can be affected by some measurement biases due to its being a parent-rated questionnaire. Among these systematic errors, the most common is usually an overestimation of the Externalizing symptoms and an underestimation of the Internalizing ones [6]; furthermore, in the specific context of twin studies, the use of CBCL/6–18 to assess the symptoms often leads to detect higher levels of the shared environment contribution to the total variance [54]. In the light of this, although this last bias does not seem to have affected our results, the use and comparison of multiple tests in the assessment of symptoms would have increased results’ reliability. Second, problem behaviors were assessed at one time point only; this is a potential confounder for the estimation of the effects of unique environmental factors and for the distinction of these effects from measurement error [10]. Third, even though the sex-stratified correlational pattern didn’t seem to suggest any role of sex in the gene-environment structure of the single traits and their mutual relationships (data not shown), the relatively small size of our sample conferred limited power to formally address potential sex differences in the etiology of symptoms. The importance of this last limitation stems from the observation that developmental trajectories of neuroticism appear to be highly influenced by gender [42], and that there are marked differences in the prevalence rates of internalizing and externalizing problems, which have also been explained in terms of gender-linked behavioral preferences that drive the expression of the individual’s susceptibility toward general psychopathology [8, 55].

In conclusion, though with some limitations, the present study supports the existence of a common latent susceptibility factor behind Fear, Distress and Externalizing symptoms. This factor is likely to be mainly genetic in origin, and its effects seem to become more evident during puberty. These findings support the idea that diagnostic decisions are to be made considering the complexity that characterizes the psychological profile of the single individual, including possible comorbidity patterns [39].

In this perspective, results of the present work could be of use in the clinical setting, by encouraging clinicians to monitor the symptomatic manifestations of Externalizing problems both at individual and familial level, when in presence of a patient who comes to clinical attention for Internalizing-like problems, and vice versa. Moreover, the idea that the common etiological factor behind these clusters of symptoms might be represented by neuroticism could bring a new insight even regarding the most effective psychological treatment to use. In this regard, transdiagnostic therapies could be an intriguing and promising approach, aiming to target neuroticism itself rather than the single disorders as fragmented categories [49, 56].