Neurocognition and its association with adverse childhood experiences and familial risk of mental illness

1. Introduction

Adverse childhood experiences (ACEs) have been studied extensively over the last decade (Struck et al., 2021), and linked to a number of negative outcomes concerning mental & physical health (Chang et al., 2019; Sonu et al., 2019). Exposure to ACEs is associated with an increased risk, and higher severity of several psychiatric disorders, including mood & anxiety disorders (Selous et al., 2020), psychosis (Carbone et al., 2019), substance use disorders (SUD) (Hughes et al., 2019) and obsessive-compulsive disorder (OCD) (Visser et al., 2014). The effects of ACEs are also understood to be transdiagnostic; we demonstrated in a previous report that even in individuals with high familial loading/genetic vulnerability, ACEs accelerated the age at onset in many of the above-mentioned disorders (Someshwar et al., 2020).

Several mechanisms are postulated to explain how ACEs might lead to these effects. Earlier research focused on the psychological & social aspects of early life stress, and how they affect attachment patterns, bonding and thus impact emotional processing (Pollak, 2008). However, it is increasingly recognized that the mechanisms that underlie the impact of ACEs are also biological. There is substantial evidence that ACEs lead to premature biological ageing, demonstrated at the cellular level by reduced telomere length (Bürgin et al., 2019; Warner et al., 2020), physiological effects such as earlier pubertal timing (Henrichs et al., 2014; Holdsworth & Appleton, 2020), and various brain changes such as smaller age-related volumes of hippocampus in individuals who report ACEs (Bick & Nelson, 2016; Dannlowski et al., 2012; Herzog et al., 2020; Vythilingam et al., 2002).

Emerging literature suggests that there may be a link between ACEs and neurocognitive functioning. The largest study to date, which analyzed data from the UK National Household Longitudinal Study (n=27,985 adults), found that early-life adversity was associated with poorer performance on word recall, verbal fluency and numerical ability tasks (Bridger & Daly, 2019). Many other studies on smaller samples have also demonstrated that adults who have experienced ACEs perform poorer in various neurocognitive domains. (Cowell et al., 2015; Hawkins et al., 2021; Kavanaugh et al., 2017; Majer et al., 2010).

However, there are also studies that suggest the opposite relationship. One such study, drawing data from the Irish Longitudinal Study on Ageing, (n=6,912 adults aged >50 years) found, surprisingly, that childhood sexual abuse (CSA) was associated with better global cognition, memory, executive function, and processing speed, despite poorer psychological health (Feeney et al., 2013). Two other studies also reported similar findings, of better working memory in adults with a history of adverse/unpredictable childhoods (Mittal et al., 2015; Young et al., 2018). Another large population-based study, the English Longitudinal Study on Ageing (n=5223), did not find any association between ACEs & memory decline (immediate and delayed recall) (O’Shea et al., 2021). A systematic review of studies looking at the association between ACEs and neurocognition in healthy individuals found a high degree of heterogeneity in methods & their findings (Su et al., 2019).

Neurocognitive deficits are considered a core feature of many psychiatric disorders. Patients with psychiatric disorders, who also have a history of ACEs have greater deficits compared to those without history of ACEs (R.-Mercier et al., 2018). A meta-analysis of 23 studies with a total of 3315 patients with psychotic disorders (Vargas et al., 2019), examined the relationship between ACEs and overall neurocognitive functioning for four subdomains (working memory, executive function, verbal/visual memory, and attention/processing speed). Only small negative associations (effect sizes between 0.05 - 0.1) were found between overall cognition, or any of the sub-domains and childhood trauma in individuals with psychotic disorders. Only a few such studies have been done in samples with other psychiatric disorders, which also suggest poor neurocognitive functioning with a history of ACEs, in alcohol use disorder (AUD) (Cheng et al., 2020) and in major depressive disorder (Dannehl et al., 2017)

In the context of psychiatric disorders, neurocognitive deficits have also traditionally been conceptualized as candidate endophenotypic markers. Endophenotypes, by definition, are quantifiable measures postulated to be stable (found preceding the onset of the disorder as well as in remission) and heritable. Research on neurocognitive functioning in clinically healthy first-degree relatives (FDRs) of probands with various psychiatric disorders have found poorer performance on various domains in comparison to controls, with considerable overlap in the deficits identified across disorders (Abramovitch et al., 2021; East-Richard et al., 2020). Neurocognitive deficits are thus postulated to represent genetic risk for psychiatric disorders, which are polygenic and have complex heterogeneous distal phenotypes. In addition to shared genetic factors, FDRs also may share environmental factors, particularly ACEs with probands. Moreover, there is fairly robust evidence that ACEs may worsen neurocognition in those with a psychiatric diagnosis, but the relationship has shown mixed results in healthy individuals. Hence, it may be important to examine how ACEs may influence neurocognitive functioning in FDRs as well.

In this study, we aimed to describe the relationship between ACEs and neurocognitive functioning in at-risk FDRs from multiplex families with major psychiatric disorders - schizophrenia (SCZ), bipolar disorder (BD), alcohol use disorders (AUD) & obsessive-compulsive disorder (OCD). A trans-diagnostic approach could help us to understand the differential effect of ACEs on the neurocognitive profiles of such syndromes. In addition, evaluating individuals from multiplex families with a pre-existing genetic loading for psychiatric disorders might inform us about the nature of neurocognitive endophenotypes and the role of ACEs in increasing neurocognitive deficits in vulnerable individuals. We hypothesized that ACEs would affect neurocognition differentially in FDRs vs healthy controls.

2. Methodology

3. Results

Table 1 shows the demographic details of the sample included in this analysis. As stated previously, FDRs groups were not mutually exclusive, i.e, an individual FDR could be “at-risk” for multiple diagnoses by having family members. Hence, no statistical comparisons were made between these groups. Noticeable differences are in education and socio-economic status - these were higher in controls and least in the AUD FDRs group. Similarly, the prevalence & extent of reported ACEs in the sample are depicted in Table 2. Overall, the group of FDRs of AUD report higher ACEs in nearly every domain. Also, the controls (having no FDRs with any psychiatric disorder in the family) were found to have lesser cumulative ACEs, in terms of dimensional, frequency & binary scores. However, reported sexual abuse was higher in them in comparison to all the other groups.

The results of the multiple linear regression model examining the effect of ACEs on the total sample, along with the effects of covariates are shown in Table 3. Significant effects of ACEs were found for the global neurocognition factor, and working memory (p<0.001 in both). There was a trend towards significance for processing speed (p=0.023), and new learning/recall (p=0.024). The covariates (age & gender especially) were found to be significant in predicting most domains.

Table 4 shows the results of our main analysis, with the main effects of ACEs, familial risk & their interaction. When all “at risk” FDRs were examined as a group, the main effect of familial risk was significant for predicting poor performance in all domains (p <0.01), and their interaction had a significant negative association with global neurocognition, processing speed & working memory. The main effect of ACEs alone was not found significant for any of the neurocognitive domains. When looking at sub-groups of FDRs with risk of different syndromes, we found that main effects of familial risk of schizophrenia & bipolar disorder were found to predict poorer performance in working memory, theory of mind & global neurocognition, and there was no significant effect of the interaction for any domain of neurocognition. The interaction term was found significant only in the subgroup of FDRs at risk for AUD. for global neurocognition, processing speed and working memory. Neither the main effect of familial OCD risk, nor its interaction with ACEs was found to significantly predict any neurocognitive domain. Figure 1 shows the plots depicting the relation between ACEs score and global neurocognitive ability in the whole FDR sample (“any risk”) and sub-samples for each disorder, in comparison to controls with no risk.

• Download figure
• Open in new tab

Figure 1 Relationship between adverse childhood experiences and global neurocognitive ability in ‘At Risk’ vs Healthy subjects

Discussion

The key aim of this study was to dissect the influence of childhood adversities versus genetic vulnerabilities for major mental illnesses, and how they might interact and lead to neurocognitive deficits. Neurocognitive deficits have been extensively researched from the “endophenotype” standpoint, and only recent studies have begun to look at how environmental factors might influence them. Previous such studies have looked at patients affected with schizophrenia, some with large sample sizes, but have yielded mixed results (Kasznia et al., 2021; Mansueto et al., 2018; McCabe et al., 2012; Schalinski et al., 2018; Vargas et al., 2019; Velikonja et al., 2019; Wells et al., 2020). There is hardly any research that has examined this relationship (between ACEs and neurocognition) in other disorders.

To our knowledge, we present the first study to examine how ACEs impact neurocognitive deficits in a large sample of unaffected FDRs. This approach enables us to test our hypotheses (endophenotype vs environmental), without the confounding effect of state-related factors, especially medications. Also, the FDRs included in this study were from a genetically “enriched” transdiagnostic sample of multiplex families, and were evaluated comprehensively for their family history, ACEs, as well as neurocognition. Hence, we were presumably well-poised to pick up signals due to familial risk, as well as childhood adversity across the 4 major psychiatric disorders.

The main findings of our study are as follows: When familial risk is not taken into consideration, ACEs seem to have a negative impact on all domains of neurocognition. However, including a familial risk variable into the model nullifies the effect of ACEs alone, but the interaction between the two remains significant. The domains that remained significant after multiple comparison testing were global neurocognitive ability, working memory, and processing speed. Further, this interaction effect appears to be driven predominantly by the familial risk of AUD. On the other hand, the main effects familial risk of schizophrenia and bipolar disorder were significant, with no impact of ACEs or its interaction in both these sub-groups.

The long-term effects of ACEs are undoubtedly complex, and there are likely to be myriad factors that may confer vulnerability or resilience to individuals who experience ACEs. As shown in our study, studying the effects of ACEs alone without including other moderators can be problematic. Though we found initially that ACEs have an overall negative impact on neurocognition, the effect appears to be nuanced, with differences in those who are “at risk” FDRs versus those without a family history. At the most basic level, this points out to familial risk being an important vulnerability factor for succumbing to the effects of ACEs.

The interaction term (familial risk x ACEs) was found to have a negative impact on two neurocognitive sub-domains, namely working memory and processing speed. In comparison to the other neurocognitive domains, both these have been shown to decline significantly with age (Brockmole & Logie, 2013;

McNab et al., 2015; Pliatsikas et al., 2019), as opposed to other domains such as theory of mind. This is in line with findings reported in the Longitudinal Ageing Study Amsterdam (Korten et al., 2014), where processing speed was found to decline faster over 10-year follow-up of those with ACEs & depressive symptoms, but not ACEs alone. Also, an increased rate of decline was not seen for other cognitive domains which were delayed recall and the total MMSE scores. This fits in well with the “accelerated biological ageing” hypothesis of ACEs’ impact, adding the element of specific vulnerability factors that may moderate this effect. We use familial risk as a proxy for genetic vulnerability, and one might argue that “familial risk” itself carries with it shared psychosocial and environmental factors that would confound our interpretation. Hence, it would be crucial to follow-up these studies using genotyping data, by carrying out gene x environment (G x E) analyses, or by polygenic risk scores (PRS) of various disorders instead of familial risk. Such studies have been done to predict disease risk, but not for neurocognition.

For the two severe mental disorders - schizophrenia and bipolar disorder, only familial risk was found significantly associated with poorer neurocognition, particularly in the domains of working memory and theory of mind. Neurocognitive deficits, particularly working memory as well as social cognition, are considered a core feature of psychotic disorders, and have been very consistently replicated as endophenotypes as well. Bearing in mind the high degree of genetic loading that they carry, it is possible that the effect of ACEs is “overpowered”, and is thus not found significant even as an interaction effect. A similar conclusion was drawn in the previous meta-analysis that examined correlations between ACEs and neurocognition in patients with psychosis (Vargas et al., 2019). This was attributed to factors such as medication use & disease severity, whereas in our study we find that a similar phenomenon is seen even in their FDRs. This further highlights the relevance of neurocognition as an endophenotype, especially in severe mental illnesses.

Based on twin and adoption studies, the influence of environmental & social factors is estimated to be higher in AUD, with lower heritability (Verhulst et al., 2015). It is thus possible that we found more significant signals for ACEs/interaction for the AUD risk. Also, the AUD FDRs were found to have lower education, poorer socio-economic status and may have other shared environment factors as well, which could not have been separately accounted for in our analysis, but are simply a part of “familial risk”. Hence, the endophenotype nature of the main effect of familial risk in this group needs to be interpreted with caution, as most of them could simply be better explained by other environmental factors.

The findings of our study need to be interpreted with the following limitations. Firstly, though we had a large sample of unaffected FDRs, it may be insufficient for certain individual disorders (OCD having the lowest number n=73). Also due to sample size constraints, we did not look for the effects of specific types of ACEs (e.g. abuse vs neglect). As the FDR sample was from multiplex families, many subjects had familial risk for multiple disorders, and there would be “cross contamination” across the various “familial risk” for each disorder. Again, due to sample size constraints we could not use the different familial risks & interaction with ACEs in a single model, as this would require too many main effects and interactions, resulting in difficulties in interpretation and possible Type II errors. Thirdly, our control sample (those without familial risk) were better educated, were from a higher socio-economic status than FDR groups, biases may have thus crept in even though these factors were adjusted for.

4. Conclusions and future directions

Our study brings forth important insights about how the impact of childhood adversity on neurocognition may be moderated by familial risk of psychiatric disorders. It also highlights how genetic/familial vulnerability may play a greater role in disorders such as schizophrenia and bipolar disorder, while the interaction may be more relevant in the case of disorders with greater environmental risk, such as substance use. Replacing the “familial risk” variable in our study with genetic markers or polygenic risk scores would help in examining specific genetic factors. Furthermore, replacing our outcome (neurocognitive performance) with a host of other markers such as personality/temperament, structural & functional imaging, or even cellular phenotypes may yield greater insights about the effects of ACEs. This will help improve how we identify and understand vulnerability & resilience factors, and perhaps better target our interventions towards them.

Supplement

• Download figure
• Open in new tab

• Download figure
• Open in new tab

Figure 1 Results of Confirmatory Factor analysis of the neurocognitive variables