Impact of adverse life experiences on pain, depression, anxiety, and comorbidities: a youth longitudinal sample

Supplemental Digital Content is Available in the Text.

The present study explored the links between adverse life events and pain, depression, and anxiety comorbidities from adolescence to young adulthood, including the role of brain structure.

Abstract

Introduction:

Exposure to adverse life experiences (ALEs) renders individuals vulnerable to the emergence of pain, depression, and anxiety. It remains unclear to what extent these symptom categories share common ALEs, especially in cases of comorbidity, and how these relationships manifest in developmental trajectories and neural pathways.

Objectives:

In this study, we investigated the impact of ALEs, considering their timing, quality, and quantity, as well as structural brain changes, on pain, depression, and anxiety symptoms, and their comorbidity from adolescence to young adulthood.

Methods:

We used prospective and retrospective questionnaires and magnetic resonance imaging data from a large European longitudinal cohort (N = 1700) spanning from 14 to 25 years. We conducted Latent-Class-Growth-Analysis for symptom levels of pain, depression, and anxiety, subsequent logistic regressions to explore prediction of ALEs on symptom classes, and mediation analysis to examine the role of insula in this association.

Results:

Physical illness unrelated to pain, bullying, and abuse-maltreatment were associated with pain; sexual abuse, bullying, parental violence, and deprivation with depression; bullying and deprivation with anxiety; substance abuse in the household and abuse-maltreatment with pain–depression comorbidity; deprivation with pain–anxiety comorbidity. Smaller insula volume in late adolescence was a significant mediator for the association between deprivation-related ALEs and the pain–anxiety, but not the pain–depression comorbidity.

Conclusion:

Together, although various and mainly different types of ALEs strongly impact pain, depression, and anxiety symptoms and their comorbidity, insula volume impacts are specific for pain–anxiety comorbidity. These findings may inform early screening and prevention for individuals affected by ALEs.

1. Introduction

In recent years, adverse life experiences (ALEs) have been identified as risk factors for physical and mental health problems.⁹⁴ The concept of ALEs is complex and multifaceted including a wide range of challenging experiences, from highly traumatic to daily life-related, from one-time powerful incidents to chronic or repetitive experiences.⁴⁶ Traditional frameworks, like adverse childhood experiences, have narrowly focused on a limited subset of events, typically occurring before the age of 18 and often limited to 10 original categories. In contrast, ALEs adopt a broader life-course perspective, integrating developmental and environmental influences from adolescence into early adulthood.^9,68 Therefore, results on how ALEs affect mental and physical health are diverse and varied. Although ALEs have been linked primarily to mental disorders like depression and anxiety,^56,59,64 pain is also a common outcome, with exposure to physical and social adversities leading to 50% to 100% increased risk of chronic widespread pain in adulthood.⁵² Moreover, research emphasizes distinct mechanisms through which ALEs are associated to comorbid presence of these symptoms.^53,85,90 Despite these insights, research remains limited,^1,16,33,104 and a comprehensive understanding in respect to the effects of timing, type, and quantity of ALEs is needed. The timing of ALEs during sensitive neurobiological and psychosocial periods^55,61,88 significantly affect their impact. Research exists on adult and clinical samples, whereas epidemiological longitudinal studies during sensitive periods, such as adolescence, are scarce.^50,58,82,88 In respect to type of adversity, existing literature⁹² highlights the need to classify them into dimensions such as deprivation (lack of expected inputs) and threat (risk to physical integrity), or physical and emotional, as they distinctly shape neurodevelopmental trajectories and have different impacts in health outcomes.⁶³ Finally, evidence suggests that also quantity of ALEs plays a decisive and differential role with total amount significantly predicting adult health conditions.¹⁰⁶ However, how this is then related to the comorbidity of affective and pain symptoms is still not clear. Recent advances in the field have highlighted the utility of trajectory analysis in understanding the developmental course of symptoms over time and their co-concurrence.^8,100 This approach better captures the dynamic interplay between various symptom types and sequence of events. Last, sex is expected to differentially affect both individual symptoms,^5,43,102 and comorbidity,⁶⁷ likely due to developmental, biological, and psychosocial factors.⁷ Especially when brain undergoes crucial structural and functional changes, chronic stress and adversity disrupt key neurodevelopmental processes, such as synaptic pruning and myelination, impairing proper neural circuit formation and connectivity. These weaken brain's capacity to manage stress and pain,¹⁹ underscoring the importance of examining the interplay between ALEs, brain development, and subsequent symptomatology through a mediation perspective. Alterations in subcortical and cortical areas involved in emotional, cognitive, behavioral, and stress responses or coping processes,^4,21,48 such as insula, amygdala, hippocampus, thalamus, and prefrontal cortex^{20,36,40,62,77,86} have been shown to mediate the effect of ALEs on symptomology. Among these, the insula stands out as potentially the most crucial brain region, as, given its unique role, it may be capable of driving associations between ALEs and pain comorbidities. The insula regulates aversive motivational salience, integrating sensory, emotional, and cognitive inputs from various brain areas,⁵⁷ and is pivotal in fear responses to adverse or pain experiences.³⁷ It processes pain salience into perceptual decisions¹⁰³ and is targeted in pain treatment strategies.¹⁰ It is also central to self-pain processing and negative affect integration, making it particularly sensitive to ALEs.²³ Notably, under negative psychological influence, the insula mediates the transition from acute to chronic pain, amplifying pain perception through maladaptive neuroinflammatory and sensitization mechanisms.³⁰ Although reductions in insula volume have been noted in depression, anxiety, and chronic pain,¹⁴ as well as in adversities,³ to our knowledge, no prior study has specifically investigated how the insula mediates the link between ALEs and the comorbidity of these symptoms.

In the present study, we capitalized on data from a large longitudinal European cohort spanning from adolescence to early adulthood. This allowed us to consider variations in timing, type, and quantity of ALE exposures from prenatal period to adulthood about the symptoms of pain, depression, and anxiety. We examined ALEs triggered pathways into mental and pain symptom-comorbidities and the role of the insula as potential mediator. Given the established role of ALEs in sensitizing stress–response systems and promoting maladaptive emotional and sensory processing, we hypothesize that diverse levels of pain, anxiety, and depression will be associated with ALE exposure during childhood and adolescence, varying by sex. Furthermore, we anticipate distinct pathways for single and comorbid symptoms. Specifically, we hypothesize that ALE-induced structural changes in the insula disrupt its ability to regulate sensory-emotional interactions, thereby predisposing individuals to comorbid symptom presentations.

2. Methods

2.1. Sample

We used data from the IMAGEN project,^89,96 a prospective longitudinal cohort study that comprises 8 measurement sites across 4 different European countries (Germany-UK-Ireland-France). Adolescent participants were initially recruited from high schools, ensuring a diverse sample in socioeconomic status, academic achievement, and emotional or behavioral functioning. The parents were included in the first assessment through their children's participation.⁹⁶ Exclusion criteria included contraindications for magnetic resonance imaging (MRI) examinations, serious medical, neurological, or developmental conditions, severe pregnancy complications, and absence of psychiatric treatments. Adolescents and their legal guardians provided written informed consent. The study adhered to the Declaration of Helsinki and was approved by each local ethics committee.

The longitudinal structure consists of 4 assessments: 14 to 15 years (baseline [BL]), 16 to 17 (follow-up 1 [FU1]), 18 to 19 (follow-up 2 [FU2]), and 22 to 24 (follow-up 3 [FU3]). The final sample consisted of 1700 participant dyads, comprising 893 females and 807 males and their respective parents.

2.2. Adverse life experiences

For ALEs, we considered prospective and retrospective questionnaires⁵¹ from both parents and adolescents. These included adolescents and parental substance abuse and health, pre- and perinatal environment, household violence and conflicts, peer problems, bullying, negative life events, abuse, and maltreatment, assessed using the following questionnaires: Drug Abuse Screening Test³⁵; Michigan Alcoholism Screening Test⁹¹; Fagerström Test for Nicotine Dependence⁴⁵; Alcohol Use Disorders Identification Test⁸⁷; Genetic Screening and Family History of Psychiatric Disorders Interview; Pregnancy and Birth Questionnaire⁷⁹; Conflict Tactics Scale⁹³; Bully Questionnaire⁷⁶; Strengths and Difficulties Questionnaire³⁹; Life-Events Questionnaire⁷³; and Childhood Trauma Questionnaire (Short Form).¹¹

We then followed a multiple-step approach (according to Ref. 51). First, 137 single items were selected from the 11 above-mentioned questionnaires, categorized into one of the 20 most common ALEs⁴⁹ and binary recoded for adversity presence/absence following different rules. Details about categorization process, items used, recoding, and whether reported by adolescents, parents, or both, can be found in Supplementary S1, http://links.lww.com/PR9/A380. Second, for each of the resulting 14 categories at each time point, a sum score of the corresponding single items was computed. Third, the categories at baseline were analyzed to extract factors. Finally, the single categories from the respective time point were summed resulting in 4 total scores. The comprehensive set of scores enables analysis across multiple dimensions: timing through various time points, type via categorized factors covering a broad spectrum of experiences, and quantity based on total scores.

2.3. Pain symptoms

For pain, we used the Children's Somatization Inventory-revised (CSI-24),^66,101 which measures perceived severity of different nonspecific somatic symptoms over the previous 2 weeks. For this study, as in previous ones,⁷⁰ we used a pain-specific score, ie, a score of all pain items of the CSI (item 1: headache, item 3: heart/chest pain, item 5: back pain, item 6: rheumatic pain, item 16: stomach pain, item 33: knees/elbows/joint pain, item 34: arms/legs pain) from 3 follow-up time points, not including baseline.

2.4. Depression and anxiety symptoms

For depression and anxiety, we used symptom scores for generalized anxiety and depression from Development and Well-Being Assessment Interview (DAWBA),³⁸ which measures psychopathology based on total symptom scores and generates International Classification of Diseases/Diagnostic and Statistical Manual of Mental Disorders diagnoses. Comorbidity was determined by the presence of pain levels, high or moderate, and the manifestation of mental symptoms of depression and anxiety either moderate or high.

2.5. Structural brain data

Magnetic resonance imaging data were acquired at 3 time points (BL-FU2-FU3), during a single session with 3-Tesla equipment. All sites used the same scanning protocol: 4 used GE/Philips scanners with 8-channel coils, and 4 used Siemens scanners with 12-channel coils. High-resolution anatomical MRIs included a three-dimensional T1-weighted magnetization prepared gradient echo sequence based on Alzheimer's Disease Neuroimaging Initiative protocol,² and T2-weighted fast- (turbo-) spin-echo and fluid attenuated inversion recovery scans for visual assessment. To ensure comparability across the sites, scan parameters were optimized and standardized.⁸⁹

2.6. Statistical analyses

All statistical analyses were conducted using SPSS Version 27.0 (IBM Corp., Armonk, NY, USA), R 4.2.0 (R Foundation for Statistical Computing, Vienna, Austria), and Python-3.10 (Jupyter Notebook).

2.6.1. Sample description and factor analysis

In SPSS, we computed descriptive statistics and adopted a stepwise approach to ensure a robust and data-driven factor structure. As reference data, following previous methodologies,⁵¹ we used categories from item allocation at baseline, supplemented by a retrospective measure of childhood abuse and neglect (Childhood Trauma Questionnaire [Short Form]) from FU2. Principal component analysis was used strictly as a preparatory step to determine the optimal factor count, using eigenvalues and scree plot. Subsequently, exploratory factor analysis was performed with a constraint on the identified number of factors, allowing to refine factor solution and examine factor loadings for the selected items. Finally, confirmatory factor analysis was run in Python, to further assess the robustness of our factor structure, after randomly splitting the sample into 5 equally sized groups, a 5-fold cross-validation procedure was applied. The factors were chosen to have a model fitting the data according to literature statistical parameters (comparative fit index, Tucker–Lewis index, and root mean square error of approximation).³⁴

2.6.2. Symptom-based trajectory analysis

In R, we used lcmm package for latent class growth analysis to classify individuals into distinct homogenous subgroups and model longitudinal trajectories, using CSI and DAWBA at each available time point. The model selection, followed literature recommendations, prioritizing lower Goodness-of-fit statistics indexes (Akaike Information Criterion and Bayesian Information Criterion), and avoiding excessively small classes.⁶⁹ Subsequent analyses only used classes depicting high and moderate levels of the symptoms as variables, with comorbidity aspects limited to pain-related conditions.

2.6.3. Symptom-based regression and mediation analyses

Using Python-3.0, both regression and mediation analyses were run. Regression investigates whether ALE variables collected at all time points as well as total scores and factors are associated with class membership, and thus symptomatology. Mediation traces the linking pathway between ALE factors, insula volume, and pain comorbidities. We applied a widely used approach for mediation analysis⁶ selecting only the mediations with both significant direct and indirect effects.

3. Results

3.1. Sample

The sample is composed by 52.5% female and 47.5% male, over 90% of parents of European descents, with a significant portion having completed secondary education or higher. For details, see Table 1.

Table 1.

Sample characteristics.

Sample characteristics	Observed data (N = 1700)
Sociodemographic characteristics
Language child
German	49.4%
English	37.8%
French	12.8%
Sex child
Female	52.5%
Male	47.5%
Mother ethnicity
European descent	90.9%
Non-European descent	8.3%
Missing	0.8%
Father ethnicity
European descent	92.8%
Non-European descent	6.3%
Missing	0.9%
Level of education father (years of schooling)
16+ y1	15.8%
15–16 y2	22.1%
13–15 y3	12%
14 y4	11.8%
12–13 y5	14.4%
11 y6	18.8%
Did not go to school or completed primary school education only7	1.1%
None of the above8	3%
Missing	1.1%
Level of education mother (years of schooling)
16+ y1	11%
15–16 y2	22.4%
13–15 y3	15.4%
14 y4	14.9%
12–13 y5	17.6%
11 y6	14.2%
Did not go to school or completed primary school education only7	0.8%
None of the above8	2.5%
Missing	1.1%

EN: 1. Professional qualification eg, PhD, MD, Master's degree. 2. Bachelor degree eg, BA, BSc. 3. Advanced diploma. 4. A levels or a BTEC national diploma. 5. NVQ or GNVQ. 6. O levels, GCSEs or CSEs. 7. Did not go to school or completed primary school education only. 8. None of the above.

DE: 1. Promotion. 2. Hochschulausbildung (z.B. Diplom). 3. Berufsakademie, Fachhochschule. 4. Abitur, Fachhochschulreife. 5. Mittlere Reife. 6. Besuch der Hauptschule, Realschule oder gymnasialen Unterstufe. 7. Kein Schulbesuch oder nur Grundschule beendet. 8. Nichts davon ist zutreffend.

FR: 1. Diplôme universitaire: Doctorat, Master. 2. Licence. 3. Bac + 2 (DEUG, BTS, DUT). 4. Bac général ou Bac professionnel. 5. BEP. 6. BEPC. 7. N'a pas été à l'école ou s'est arrêtée à la fin du primaire. 8. Autre.

3.2. The determination and integration of adverse life experiences

The factor analysis resulted in a 4-factor model, with 43.76% of variance explained. The 4 factors covered, respectively: (1) Abuse and maltreatment; (2) Material, Health, and Emotional Deprivation; (3) Social and household environment; and (4) Parental Health and Family Dynamics (see Supplement, http://links.lww.com/PR9/A380). These factors provided the basis for regression and mediation analyses.

3.3. The investigation of symptom-based trajectories of pain, depression, and anxiety

We found a significant 3-class model for all 3 symptom variable, see Figure 1.

Figure 1.

The investigation of symptom-based trajectories of pain, depression, and anxiety. The latent class growth analysis resulted in the following trajectories: (A) Pain: Class 1—low or no pain symptoms at all assessments (82%); Class 2—moderate in adolescence and low in early adulthood (15%); Class 3—high in adolescence and moderate in early adulthood (3%). (B) Depression: Class 1—very low or no depressive symptoms at all assessments (85%); Class 2—moderate in adolescence and low in early adulthood (8%); Class 3—high in adolescence and low in early adulthood (6%). (C) Anxiety: Class 1—very low anxiety symptoms at all assessments (84%); Class 2—moderate at all assessments (9%); Class 3—high in adolescence and moderate in early adulthood (7%). The values reported are mean scores and standard deviations at different time points derived from Children's Somatization Inventory-revised (CSI-24) for pain (FU1, FU2, FU3), and from Development and Well-Being Assessment Interview (DAWBA) for anxiety and depression (BL, FU1, FU2, FU3). BL, baseline; FU1, follow-up 1; FU2, follow-up 2; FU3, follow-up 3.

3.4. The investigation of adverse life experiences impact on symptom classes and comorbidities

Descriptive statistics can be found in Supplementary S10, http://links.lww.com/PR9/A380. Table 2 shows the significant results for high and moderate trajectories. Full regression results, including nonsignificant findings, are pictured in Figure 2 and Supplement, http://links.lww.com/PR9/A380.

Table 2.

Adverse life experiences and symptom-based trajectories of pain, depression, anxiety, and comorbidities.

	Coef	SE	T	Pval	R2	Adj R2	CI [2.5%]	CI [97.5%]	VIF	Pval_corrected
ALEs and high-level symptom-based trajectories
Pain
Physical_illness_child FU1	0.029	0.009	3.402	0.001	0.143	0.094	0.012	0.046	5.640213	0.039
Total ALE score FU2	0.022	0.007	3.024	0.003	0.025	0.02	0.008	0.036	5.111497	0.015
Abuse and maltreatment—Factor1	0.023	0.006	3.570	0.000	0.025	0.021	0.010	0.036	1.860628	0.000
Depression
Sexual_abuse	0.023	0.006	4.120	0.000	0.097	0.055	0.012	0.034	1.209255	0.000
BullyingBL	0.028	0.006	4.765	0.000	0.097	0.055	0.016	0.039	1.590381	0.000
Total ALE score FU1	0.016	0.005	2.934	0.003	0.016	0.012	0.005	0.027	6.084844	0.015
Anxiety
BullyingBL	0.034	0.008	4.513	0.000	0.111	0.07	0.019	0.049	1.590381	0.000
Total ALE score FU1	0.023	0.007	3.425	0.001	0.034	0.031	0.010	0.037	6.084844	0.005
Total ALE score FU2	0.018	0.007	2.677	0.008	0.034	0.031	0.005	0.031	4.457266	0.040
ALEs and moderate-level symptom-based trajectories
Pain
BullyingFU1	0.060	0.014	4.300	0.000	0.111	0.059	0.033	0.087	1.419223	0.000
Total ALE score FU1	0.048	0.014	3.348	0.001	0.047	0.043	0.020	0.076	8.055105	0.005
Depression
Parents_violentFU1	−0.035	0.010	−3.349	0.001	0.085	0.043	−0.056	−0.015	1.892365	0.039
Total ALE score FU3	0.031	0.009	3.531	0.000	0.02	0.017	0.014	0.048	4.478934	0.000
Factor 2	0.019	0.006	3.023	0.003	0.021	0.018	0.007	0.031	3.706052	0.015
Anxiety
BullyingFU3	0.041	0.012	3.485	0.001	0.09	0.048	0.018	0.063	1.198944	0.039
Material, health, and emotional deprivation—Factor 2	0.041	0.010	4.192	0.000	0.019	0.016	0.022	0.061	3.706052	0.000
ALEs and symptom-based trajectories of pain comorbidities
Pain–depression comorbidities
Substance abuse in the household FU2	0.021	0.006	3.409	0.001	0.065	0.019	0.009	0.033	1.304255	0.039
Abuse and maltreatment—Factor1	0.015	0.004	3.453	0.001	0.018	0.014	0.007	0.024	1.880138	0.005
Pain–anxiety comorbidities
Material, health, and emotional deprivation—Factor 2	0.015	0.006	2.662	0.008	0.019	0.015	−0.004	0.027	3.635520	0.040
Pain–depression–anxiety comorbidities
Material, health, and emotional deprivation—Factor 2	0.011	0.004	3.008	0.003	0.016	0.013	0.004	0.018	3.680576	0.015

The regression models for ALEs categories at all time points BL (Baseline), FU1 (Follow-up 1), FU2 (Follow-up 2), FU3 (Follow-up 3) and the high and moderate symptoms classes of pain, depression, and anxiety and comorbidities. The table reports only results that retained statistical significance after correction. For high symptom trajectories, physical illness, bullying, sexual abuse, total ALE scores, and specific factors emerged as significant for pain, depression, and anxiety. Moderate symptom trajectories revealed again that bullying, parental violence, total ALE scores, and abuse and maltreatment factor were also significant, with distinct patterns for each symptom. Pain–depression comorbidities were significantly associated with household substance abuse and abuse-maltreatment factor. Pain–anxiety comorbidities were linked to deprivation-related factors, as well as the combined pain–depression–anxiety comorbidity. These findings underscore the differentiated pathways through which ALEs contribute to comorbid symptom development. Full details and expanded analyses are provided in Supplement, http://links.lww.com/PR9/A380. In the corresponding order Coefficients, Standard Errors, T-values, P-values, R-squared, Adjusted R-squared, Confidence Intervals, VIF, and Bonferroni-corrected P-values. Age ranges for participants at each time point were 14 to 15 y (BL), with follow-ups at 16 to 17 (FU1), 18 to 19 (FU2), and 22 to 24 (FU3). Bold font indicates statistically significant P-values.

ALE, adverse life experiences; VIF, variance inflation factor.

Figure 2.

The investigation of ALEs impact on symptom classes. The correlations resulting from regression analysis between the (A) high and (B) moderate symptoms classes of pain, depression, and anxiety, and all the ALEs categories at all time points (BL, FU1, FU2, FU3). The plot shows coefficients, CI [2.5%] and CI [97.5%]. For those marked with an asterisk symbol (*) P-values were significant (P < 0.05), those marked with a red asterisk symbol (*) were still significant (P < 0.05) following the correction for multiple comparisons. ALE, adverse life experience; BL, baseline; FU1, follow-up 1; FU2, follow-up 2; FU3, follow-up 3.

3.5. Sex differences

Across all symptoms, females were more often classified in moderate/high classes than males (see Supplement, http://links.lww.com/PR9/A380). To assess sex differences, regression models were run separately. After Bonferroni correction, male models were not significant, females retained positive effect of bullying at baseline on high anxiety and bullying at FU3 on moderate anxiety; on depression of sexual abuse and bullying at baseline.

3.6. The investigation of insula mediation for adverse life experiences and pain comorbidities

The strongest mediation effect was observed for the deprivation factor on pain–anxiety comorbidity through left and right insula at FU2 (Fig. 3). Although minimal, additional significant indirect effects included deprivation and social environment factors. To avoid confounding effects of the timing component, we conducted additional analyses controlling for previous experiences and using comorbidities scores at FU2 and FU3 instead of classes. These analyses yielded consistent results, reinforcing the robustness of our findings and demonstrating that temporal dynamics do not undermine our primary conclusions (see Supplementary S9, http://links.lww.com/PR9/A380). No significant model was found for comorbidity of the 3 symptoms together (see Supplementary S5, S6, S7, http://links.lww.com/PR9/A380).

Figure 3.

The investigation of insula mediation for deprivation factor and pain–anxiety comorbidity. The path diagrams show the coefficients for most robust models resulting from mediation analysis between the deprivation factor derived from factor analysis and comorbidity of pain–anxiety, through (A) right and (B) left insula volume at FU2. Red colour indicates statistically significant effects (P < 0.05). FU2, follow-up 2.

4. Discussion

In the present study, we investigated the relationship between ALEs and pathways into pain, depression, anxiety, and their comorbidities capitalizing on longitudinal data from a large sample of adolescents and young adults. We identified specific and overlapping characteristics of ALEs from multiple time points linked to mental and pain symptomatology, with distinct ALE patterns for pain comorbidities. A total ALE effect was found, suggesting that risk factors usually do not emerge in isolation but occur and tend to cluster together.¹⁵

In more detail, bullying victimisation was consistently associated with higher depression, anxiety, and pain. This confirms that social and peer-related experiences are crucial for a diverse set of health consequences,^12,15,42 and further underlines the sensitivity of early life stages. Interestingly, bullying was, however, not associated with comorbidities indicating different entities and clusters when it comes to symptom overlaps.

Among the factors relatively specific to individual symptoms, sexual abuse positively correlated with depression, aligning with previous research.^44,71,74 Interestingly, with increased level of parental violence, the depression levels tended to decrease. Probably because the resulting adverse daily living context of household violence fosters a reactive coping response in youth, potentially counteracting depressive symptoms.

For pain symptomatology, general physical illness showed positive correlations, suggesting a possible interplay of these bodily symptoms with somatization.¹³ Although distinct from pain, somatization may heighten it through psychological vulnerability and stress reactivity. Although this might overlap conceptually with somatization, we align existing literature in cautioning against using this label in the absence of clinical verification, as in our case.²⁴ The diathesis-stress model helps explaining how ALEs amplify these vulnerabilities, fostering the co-occurrence of somatic and pain symptoms. Factors like coping strategies, illness beliefs, socioeconomic status, and health care access further shape these outcomes, particularly during developmental stages, with adolescents still forming patterns of emotional response, health-related identities, and especially sensitive to social influences.

One of our main aims was to increase the understanding of mental and pain symptom-comorbidities. Previous literature showed that individuals suffering from comorbid disorders have higher levels of ALEs, severe or recurrent symptomatology and treatment resistance predisposing them to highest disease risk, poorest prognosis, and greatest impairment.^60,75,80 Although we observed significant associations between ALEs and individual symptoms and specific ALE-related risk factors for comorbidities, at the brain level, a significant model was identified only for pain–anxiety comorbidity. This indicates a dynamic interplay of mechanisms rather than a straightforward accumulation of symptoms, varying across contexts and individual history.^{27,84,95,97,99} This is also supported by our trajectory analysis. Even if small in percentage (3%–7%), the high classes trajectories are informative, resembling the recent trend of increasing prevalence of mental disorders and pain with decreasing age of onset. Our results indicate that childhood adversities are more strongly associated with emergence rather than persistence of these disorders.^41,65 Another important finding in this respect comes from sex differences. Within our cohort, females comprised 81% of subjects in high and 71% in moderate classes, exhibiting significantly higher levels of ALEs and their repercussions compared to males. This is in line with prior literature indicating sex as a risk factor for ALE consequences, particularly during adolescence,^18,29,32,98 confirming our initial hypothesis. These stronger associations in females may be due to their increased exposure to interpersonal adversities and heightened stress sensitivity. Also, our focus on internalizing symptoms may explain the lack of significant findings in males, as well as, sample size limitations and reduced statistical power in sex-stratified analyses.

We need to emphasize that our study delves into moderate and high symptom trajectories of adolescents from the general population that may show symptom levels approaching and/or fulfilling the clinical forms but is not a clinical population.⁹⁵ Given that these often-overlooked symptoms still impede daily functioning and trigger help-seeking behaviors but are not reaching services,⁸³ such information is important for more nuanced and personalized treatment approach.²⁶ Our study thus adds to previous literature on patient samples and can particularly inform early screening and prevention initiatives. Although this was beyond the scope of the present study, an interesting follow-up analysis could focus on low-symptom groups not as reference categories but as participants with minimal or no symptomatology, exploring mechanisms that mitigate symptoms and protective rather than risk factors.

Our study integrates a multidimensional perspective of ALEs, symptoms, and development, aligning with the biopsychosocial model, emphasizing how pain experiences develop and are maintained through neurobiological and psychosocial pathways in youth. By including factors such as allostatic load, maladaptive cognitive appraisals, and social dysregulation, this approach illustrates the pathways linking early adversities to long-term outcomes.⁷² In this respect, our models demonstrate that part of the variance in pain comorbidities explained by ALEs is mediated through the insula volume, showing the existence of neurodevelopmental mechanisms. The temporal effect observed in earlier analyses is substantiated, with mediations concentrated during late adolescence, as well as the differential effect of ALEs types. Deprivation-related experiences emerged as the most common among significant models, followed by the abuse-maltreatment, and social environment. Lack of stimuli and social interaction is often reported in conjunction with reduced brain volume, cortical thickness, and connectivity.^28,105 Our results not only indicate that reductions in both left and right insula volumes mediate the effects of deprivation-related ALE factors for pain–anxiety comorbidity, but also that these are long lasting up to early adulthood. The insula is involved in processing stress and integrating emotional and sensory information and prolonged exposure to stress may lead to alterations in insula function and structure, potentially contributing to changes in emotional regulation and pain perception.^57,78 In supplemental analyses, to explain specificity of the insula, we investigated other brain structures, including amygdala, hippocampus, thalamus, and frontal pole (Supplementary S8, http://links.lww.com/PR9/A380). We found significant effect for ALEs on pain–anxiety comorbidity via left hippocampus contrasting previous literature linking ALEs to depression via hippocampal volume.⁸¹ No effect for amygdala and frontal areas was found, likely due to limitations of total volume measurements, not capturing subregion-specificity of amygdala,¹⁰⁷ and ongoing plasticity and resiliency counter-effects in frontal areas.¹⁷ Although multiple brain regions are frequently implicated in emotional regulation and stress response, their precise roles in mediating pain processes are not yet fully understood. By focusing on the insula, we targeted a region uniquely positioned to integrate sensory and emotional processes, whereas we acknowledge the need for a more comprehensive investigation of broader neural networks with specific aims and hypothesis-driven.

Our study holds some limitations, including reliance on secondary data, limited generalizability, absence of clinical diagnoses, potential unmeasured confounders, use of symptom trajectories, subjective item mapping and partial construct coverage, lack of independent validation, and methodological trade-offs in factor derivation. Although our approach has already been adopted,^47,70 the IMAGEN project was not aimed at indexing pain; this resulted in a single pain measure, failing to capture the intricate nature, severity, and frequency of pain syndromes.¹³ Specifically, we acknowledge that interpreting these bodily symptoms as somatization may be misleading as CSI reflects multiple physical complaints. In line with existing literature, who found that most pain studies equate somatization with a mere count of symptoms, we avoid overinterpreting this measure.²⁴ Moreover, although absence of earlier data prevents us from determining whether symptoms decreased, peaked, or adhered to different patterns, the observed reduction in symptoms from adolescence to early adulthood aligns with existing literature.^41,65 This, along with other methodological constraints, highlights the inherent challenges of conducting secondary analyses on preexisting longitudinal data, which, although rich in phenotypic detail and developmental insights, also come with trade-offs in data availability and structure. Consequently, despite the hypothesized model is supported by theory and fits the data well, the temporal order remains equivocal. Recent frameworks for advancing causal thinking in observational research²⁵ emphasize the importance of explicitly modeling confounding variables and tools to establish clearer causal pathways. Although our approach minimized confounding, it is still subject to the inherent limitations of observational data, like unmeasured confounders and potential reverse causality, particularly for ALEs occurring later in development. The lack of control for confounders and reliance on observed correlations prevent us from drawing strong causal conclusions about adversity leading to symptom trajectories. Future studies with advanced methodologies could better disentangle predictive and causal pathways. In addition, the results are restricted to European-descent and highly functioning participants, and dataset did not include self-reported data on race and ethnicity. This reduces the ability to explore potential interactions with critical social factors, known to influence both ALEs and symptomatology, and thus, results may not be extended to broader populations or control for group disparities.³¹ Similarly, our cohort consisted of adolescents from the general population with self-reported measures for the symptom and information from the DAWBA interview. Participants did not undergo other follow-up clinical diagnostics nor was the presence of a diagnosis a prerequisite for study inclusion. Although exclusion criteria for participating in the IMAGEN study ruled out subjects with undergoing psychiatric treatments and clinical diagnosis, the presence of clinical forms of pain, depression, and anxiety as defined by the assessment tools was not an exclusion criterion, but a key criterion of the population sample with individual spanning the subclinical to the clinical symptom spectrum. Finally, the methodological decisions were informed by and aligned with approaches used in previous research,⁵¹ guided by theoretical considerations and practical constraints to best address our specific research questions; however, we recognize that alternative methodologies could have yielded different insights. For instance, we applied latent class growth analysis to each outcome (pain, depression, anxiety) separately, identifying distinct trajectory classes for each domain. This univariate, outcome-specific approach means we did not model multisymptomatic trajectories or use a combined latent class model that accounts for the coevolution of pain and emotional symptoms together.¹⁰⁰ It is possible that certain subgroups of adolescents exhibit coupled changes that a multisymptom analysis could detect. In addition, the mapping of individual questionnaire items into broader categories was complex and involved some subjective decisions. Different researchers might classify certain adversities differently, meaning our data-driven grouping is one of several plausible mappings.^72,92 We used factor analysis to guide the grouping of ALE items, which helps in dimensionality reduction at the cost of some information loss, as not all items within a category contribute equally to outcomes. Furthermore, our adversity constructs were constrained by the IMAGEN dataset, potentially omitting certain stressors or trauma types and limiting the comprehensive coverage of theoretical ALE domains. We also note that we used principal component analysis primarily to determine the number of ALE factors. Although principal component analysis is fundamentally a data-reduction technique rather than a true latent factor extraction, it is a practical and reliable tool for identifying dimensionality especially combined with exploratory factor analysis. However, the resulting ALE categories should still be interpreted with caution, given the limitations of this data-driven process. Indeed, we did not validate our ALE grouping in an independent sample or via replication, as we had data limited to a single sample, but we evaluated model robustness via confirmatory factor analysis on random subsets of the dataset. Although this is not a substitute for replication, the consistent fit indices across all folds suggest that the factor structure is stable and reliable. Nevertheless, papers taking such a broad view of ALEs in a developmental framework and targeting pain comorbidities aspects are lacking.^22,54

Together, the present study provides evidence that ALEs affect mental and pain symptoms and comorbidity in a timing-, type- and dose-specific fashion. Although some ALE-related factors overlap across symptom categories, most are uniquely linked to pain, anxiety, depression, or comorbidity, including associations with brain volume. These findings hope to inform early screening for ALEs and prevention in vulnerable populations.

Disclosures

T.B. served in an advisory or consultancy role for Lundbeck, Medice, Neurim Pharmaceuticals, Oberberg GmbH, Shire. He received conference support or speaker's fee by Lilly, Medice, Novartis, and Shire. He has been involved in clinical trials conducted by Shire & Viforpharma. He received royalties from Hogrefe, Kohlhammer, CIP Medien, Oxford University Press. G.B. has received honoraria from General Electric Healthcare for teaching on scanner programming courses. L.P. served in an advisory or consultancy role for Roche and Viforpharm and received speaker's fee by Shire. She received royalties from Hogrefe, Kohlhammer, and Schattauer. The present work is unrelated to the above grants and relationships. The other authors report no biomedical financial interests or potential conflicts of interest.

Supplemental digital content

Supplemental digital content associated with this article can be found online at http://links.lww.com/PR9/A380.