1. Adverse Childhood Experiences,
Institute of Epidemiology and Health Care
Adverse Childhood Experiences,
Stress Biomarkers, and Depression in Later Life
Eleonora Iob, PhD Candidate
Department of Behavioural Science and Health
Soc-B CDT (ESRC & BBSRC)
University College London
I’m going to talk about.., which is the main topic of my PhD project.
APS 77th Annual Scientific Meeting “BODY to MIND”

2. Outline
Background
Study 1: Relationship of C-reactive Protein (CRP) and Cortisol with Depressive Symptoms
Study 2: Association of Adverse Childhood Experiences (ACE) with CRP and Cortisol
Study 3: Longitudinal mediation analysis of ACE, CRP, and Depressive Symptoms
Conclusions
This is the outline of what I will discuss today… I will start by providing a brief overview of the background of my research, I will then present the results of three studies that I’ve conducted in this first year of my PhD project using data from ELSA. Finally, I will discuss the main findings of these studies and next steps for my research.

3. Outline
Background
Study 1: Relationship of C-reactive Protein (CRP) and Cortisol with Depressive Symptoms
Study 2: Association of Adverse Childhood Experiences (ACE) with CRP and Cortisol
Study 3: Longitudinal mediation analysis of ACE, CRP, and Depressive Symptoms
Conclusions

4. Adverse Childhood Experiences (ACE) Psychobiological Processes
Background
Adverse Childhood Experiences (ACE)
High prevalence (Bellis et al., 2014)
Associated with several health and disease outcomes (Hughes et al., 2017)
Depression
50% greater risk of developing depression (McLaughlin et al., 2016)
Worse recovery and treatment outcomes (Nanni et al., 2012)
Psychobiological Processes
Hypothalamic-pituitary-adrenal (HPA)-axis: Cortisol
Inflammation: C-reactive protein
(Danese & McEwen, 2012; Danese & Lewis, 2017)
Adverse childhood experiences (e.g. Abuse, maltreatment, family dysfunction) are extremely prevalent both in the uk and across the world, and have been associated with several health and disease outcomes.
Amongst these, depression is one the health outcomes more strongly associated with ACE. Indeed, Ind exposed to ACE are twice more likely to develop depression and also experience worse recovery and treatment outcomes.
The ACE and depression is now well-established. However, the specific biopsychosocial mechanisms through which this may occur remain unclear to date.
In this context, it has been proposed that maladaptive alterations in the function of the hpa-axis and of the inflammatory response system might explain the long lasting effects of early-life stress on depression. Several studies have investigated the associations of biomarkers of these systems, such as cortisol and c-reactive protein, with both ACE and depression.
However, different meta-analyses have provided inconsistent or weak evidence for these associations, possibly due to a considerable variability in the way in which ACE, depression, and the biomarkers have been measured across studies (I will provide more details on this shortly). Moreover, lack of prospective studies and direct evidence for mediation effects. These are all issues that I’ve addressed in the 3 studies I am going to present today.

5. Adverse Childhood Experiences (ACE) Psychobiological Processes
Background
Adverse Childhood Experiences (ACE)
Depression
Psychobiological Processes
Cortisol
C-reactive protein
Limitations
Inconsistent or weak evidence
Methodological differences in measurement of variables
Reverse causality
Lack of evidence for mediation effects
Several studies have investigated the associations of biomarkers of these systems, such as cortisol and c-reactive protein, with both ACE and depression.
However, different meta-analyses have provided inconsistent or weak evidence for these associations, possibly due to a considerable variability in the way in which ACE, depression, and the biomarkers have been measured across studies (I will provide more details on this shortly). Moreover, lack of prospective studies and direct evidence for mediation effects. These are all issues that I’ve addressed in the 3 studies I am going to present today.

6. Outline
Background
Study 1: Relationship of C-reactive Protein (CRP) and Cortisol with Depressive Symptoms
Study 2: Association of Adverse Childhood Experiences (ACE) with CRP and Cortisol
Study 3: Longitudinal mediation analysis of ACE, CRP, and Depressive Symptoms
Conclusions

7. Study 1: Aim and Hypotheses
Aim: To investigate the relationship of cortisol (i.e. HPA-axis) and C-reactive protein (CRP) (i.e. inflammation) with:
Persistent depressive symptoms
2) Cognitive-affective versus somatic dimensions
- Let’s move onto the first study..
- Somatic symptoms: some preliminary evidence indicating that association of these biomarkers with depression might be predominantly driven by its somatic components.

8. Study 1: Methods Sample English Longitudinal Study of Ageing (ELSA)
CRP sample: N = 5,784
Cortisol sample: N = 4,761
Measures
Outcome: Depressive symptoms (CES-D 8), w1-8 (14-year period)
Independent variables: Hair cortisol & plasma C-reactive Protein, w6
Covariates: Demographic, socioeconomic, lifestyle, chronic disease, and medication data
COGNITIVE-AFFECTIVE:
‘‘enjoyed life”,
‘‘depressed”, ‘‘happy”, ‘‘lonely” , ‘sad”
SOMATIC:
‘‘everything was an effort”, ‘‘sleep was restless”, ‘‘I could not get going”
-data from ELSA
-sample of_ for crp and of_ for cortisol

9. Trait-State Occasion (TSO) model
Study 1: Methods
Trait-State Occasion (TSO) model
Statistical Analyses
Trait-State-Occasion (TSO) structural equation modelling
TSO model and missing data estimated using WLSMV estimator in Mplus
Cortisol and CRP (<=10mg/l) were log-transformed
C = cognitive-affective; S = somatic

10. Overall, cognitive-affective, and somatic factors
Study 1: Results
Effects of Cortisol (N= 4,761) and CRP (N=5,784) on persistent depressive symptoms:
Overall, cognitive-affective, and somatic factors
-overall
-cognitive-affective versus somatic: this provides evidence that relationship between hpa-axis and inflammation might be driven by somatic symptoms.

11. Outline
Background
Study 1: Relationship of C-reactive Protein (CRP) and Cortisol with Depressive Symptoms
Study 2: Association of Adverse Childhood Experiences (ACE) with CRP and Cortisol
Study 3: Longitudinal mediation analysis of ACE, CRP, and Depressive Symptoms
Conclusions
After confirming the relationship of cortisol and crp with depression, examined the effect of adverse childhood experience on C-reactive protein and cortisol

12. Study 2: Aim and Hypotheses
Aim: To investigate the effect of exposure to adverse childhood experiences (ACE) on C-reactive protein (CRP) and hair cortisol, distinguishing between:
ACE cumulative exposure
2) ACE dimensions
ACE - For which I will use the acronym ACE throughout this presentation
ACE dimensions –most studies have used cumulative scores, although some evidence indicating that different dimensions may have distinct physiological effects.

13. Study 2: Methods Sample English Longitudinal Study of Ageing (ELSA)
CRP sample: N = 4,198
Cortisol sample: N = 3,357
Measures
Independent variable: Adverse childhood experiences (wave 3)
Outcomes:
(1) CRP, wave 4 (log)
(2) High CRP, wave 4+6 (>3mg/l)
(3) Hair Cortisol, wave 6 (log)
Covariates: Demographic characteristics, childhood and adult socioeconomic factors, BMI, health behaviours, and medications
Sample
Variables
Covariates: all analyses adjusted for relevant confounders.

14. Study 2: Methods CFA model of ACE Statistical Analyses
ACE cumulative exposure: linear and ordered logistic regression (Rstudio)
ACE dimensions: Confirmatory factor analysis (CFA) (Mplus)
Missing data estimated using multiple imputation (R package ‘mice’)
- ACE cumulative exposure
- CFA model – different items available in ELSA, all measured retrospectively at wave 3. I grouped them into 4 main dimensions, namely…
RMSEA: 0.038; CFI: 0.946; TLI: 0.926

15. Study 2: CRP(log) w4 ACE Cumulative score ACE Dimensions (model 3)
Note. Linear regression model, Ref: ACE(0)
Model 1: sex, age, marital status, medications
Model 2: Model 1 + childhood and adult socioeconomic factors
Model 3: Model 2 + lifestyle indicators
-ACE cumulative score: Effects of being exposed to 1,2 and 3 or more adversities on crp at wave 4, compared to reference group = no adversity. I will focus on the fully adjusted models, blue bar. Group of people with 3 or more adversities had sign higher crp. No differences between the other groups.
-ACE dimensions: all dimensions positively associated with crp at wave4, but loss dimension had much larger effects.
C = cognitive-affective; S = somatic

16. Study 2: High CRP w4+w6 (>3 mg/l)
ACE Cumulative score
ACE Dimensions (model 3)
Note. Ordered logistic regression, Ref: ACE(0)
Model 1: sex, age, marital status, medications
Model 2: Model 1 + childhood and adult socioeconomic factors
Model 3: Model 2 + lifestyle indicators
Ordered logistic regression. The estimates in the output are given in units of ordered logits, or ordered log odds. So for ace3+ we would say that we expect people with ACE3+ to have a value of High CRP on the log odds scale, of almost two units larger, when all other variables are held constant.
Risk of having high crp at 0,1,or 2 occasions – very similar pattern of results. Group of people with 3 or more adversities had sign higher crp. Differences between other groups were much smaller.
Dimensions- all associated, but stronger effects for loss.
C = cognitive-affective; S = somatic

17. Study 2: Hair Cortisol w6 ACE Cumulative score
ACE Dimensions (model 3)
Note. Linear regression, Ref: ACE(0)
Model 1: sex, age, marital status, hair data, medications
Model 2: Model 1 + childhood and adult socioeconomic factors
Model 3: Model 2 + lifestyle indicators
- Hair cortisol was not associated neither with the ACE cumulative score, nor with the different dimensions of adversity.
- After a more careful review of the literature, I found some evidence indicating that age might have moderating effects on cortisol

18. Study 2: Hair Cortisol w6
Interaction effect: ACE Cumulative score x Age (Model 3)
Outcome:
Hair cortisol
Est. SE
p-value
ACE(1)*Age
0.003
0.353
ACE(2)*Age
0.000
0.004
0.963
ACE(3+)*Age
0.010
0.005
0.029
Ref: ACE(0)*Age
Positive interaction effect between group of people with 3 or more adversities and age.
Plot: purple line is for 3 or more adversity group. With increasing age, there was a steeper increase in cortisol for those who had 3 or more adversities compared to those with no or fewer adversities.
Model 3: sex, age, marital status, hair characteristics, medications, childhood and adult socioeconomic factors, lifestyle indicators (bmi, smoking, alcohol, physical activity)

19. Outline
Background
Study 1: Relationship of C-reactive Protein (CRP) and Cortisol with Depressive Symptoms
Study 2: Effect of Adverse Childhood Experiences (ACE) on C-reactive protein and Cortisol
Study 3: Longitudinal mediation analysis of ACE, CRP, and Depressive Symptoms
Conclusions

20. Study 3: Aim and Hypotheses
Aim: To test the mediational role of C-reactive protein (CRP) in the relationship between childhood adverse experiences (ACE) and depression over time:
Direct effects of ACE on depression, CRP, and their trajectories
2) Indirect effects of ACE on depression through CRP
For those of you who are not familiar with indirect effects – these are simply the direct effect of the predictor on the mediator multiplied by the direct effect of the mediator on the outcome.

21. Study 3: Methods Sample English Longitudinal Study of Ageing (ELSA)
Sample: N = 4,382
Measures
Independent variable: Adverse childhood experiences (wave 3)
Mediator: CRP (log) waves 2,4,6
Outcome: Depressive symptoms (CES-D 8) waves 6,7,8
Covariates: Demographic characteristics, childhood and adult socioeconomic factors, lifestyle indicators, and medications

22. Study 3: Methods Statistical Analyses
Parallel process latent growth curve modelling
(WLSMV estimator in Mplus)
Indirect effects estimated with bias-corrected bootstrap standard errors
Missing data estimated using multiple imputation
(R package ‘mice’)
Longitudinal Mediation Model

23. Study 3: Methods Statistical Analyses
Parallel process latent growth curve model
(WLSMV estimator in Mplus)
Indirect effects estimated with bias-corrected bootstrap standard errors
Missing data estimated using multiple imputation
(R package ‘mice’)
Longitudinal Mediation Model

24. Study 3: Methods Statistical Analyses
Parallel process latent growth curve model
(WLSMV estimator in Mplus)
Indirect effects estimated with bias-corrected bootstrap standard errors
Missing data estimated using multiple imputation
(R package ‘mice’)
Longitudinal Mediation Model

25. Study 3: Results Total Depressive Symptoms Somatic Depressive Symptoms
Model 3: Adjusted for sex, age, marital status, childhood and adult socioeconomic factors, lifestyle indicators, and medications

26. Outline Conclusions Background
Study 1: Relationship of C-reactive Protein (CRP) and Cortisol with Depressive Symptoms
Study 2: Effect of Adverse Childhood Experiences (ACE) on C-reactive protein and Cortisol
Study 3: Longitudinal mediation analysis of ACE, CRP, and Depressive Symptoms
Conclusions
Both systems activated by stress, and involved in several physical health conditions

27. Conclusions STUDY 1 STUDY 2 STUDY 3
Higher hair cortisol and CRP levels were associated with more persistent depressive symptoms across a 14-year period
These relationships were driven by somatic symptoms
STUDY 2
Greater exposure to ACE was associated with elevated CRP levels
Adversities related to loss of an attachment figure had the strongest effects
Greater exposure to ACE was related to a steeper increase in hair cortisol with age
STUDY 3
Greater exposure to ACE was associated with higher CRP and depression at baseline and with their increase over time
No evidence for mediation effects of CRP
Both systems activated by stress, and involved in several physical health conditions

28. Conclusions Next steps 1. ELSA 2. ALSPAC 3. TEDS
CRP: reverse mediation model
Cortisol: Moderated mediation model for the interaction between ACE and age
Polygenic scores of depression, cortisol, and CRP: main effects and interaction effects with ACE
2. ALSPAC
Associations between ACE, stress biomarkers, and depression
Main and interaction effects of polygenic scores
DNA methylation
3. TEDS
Associations between ACE, salivary cortisol, and depression
ALSPAC and TEDS: move onto ALSPAC and TEDS where I conduct similar analyses but with a greater focus on the timing of adversity since both studies have very rich data on adversity collected retrospectively. I might also examine the role of dna methylation of stress-related genes.

29. Acknowledgments
Supervisors & collaborators: Prof Andrew Steptoe, Dr Rebecca Lacey, Dr Panos Demakakos, Prof Clemens Kirschbaum
Research funders & ELSA participants
Eleonora Iob (PhD Candidate)
Soc-B CDT (ESRC & BBSRC)
University College London