Inflammatory Biomarkers and Risk of Psychiatric Disorders

Yu Zeng; Charilaos Chourpiliadis; Niklas Hammar; Christina Seitz; Unnur A Valdimarsdóttir; Fang Fang; Huan Song; Dang Wei

doi:10.1001/jamapsychiatry.2024.2185

. 2024 Aug 21;81(11):1118–1129. doi: 10.1001/jamapsychiatry.2024.2185

Inflammatory Biomarkers and Risk of Psychiatric Disorders

Yu Zeng ^1,^2,³, Charilaos Chourpiliadis ⁴, Niklas Hammar ⁴, Christina Seitz ⁴, Unnur A Valdimarsdóttir ^4,^5,⁶, Fang Fang ^1,^2,^3,⁴, Huan Song ^1,^2,^3,^5,^✉, Dang Wei ⁴

¹Mental Health Center and West China Biomedical Big Data Center, West China Hospital, Sichuan University, Chengdu, China

²Med-X Center for Informatics, Sichuan University, Chengdu, China

³West China Biomedical Big Data Center, West China Hospital, Sichuan University, Chengdu, China

⁴Institute of Environmental Medicine, Karolinska Institute, Stockholm, Sweden

⁵Center of Public Health Sciences, Faculty of Medicine, University of Iceland, Reykjavík, Iceland

⁶Department of Epidemiology, Harvard T.H. Chan School of Public Health, Boston, Massachusetts

Accepted for Publication: June 6, 2024.

Published Online: August 21, 2024. doi:10.1001/jamapsychiatry.2024.2185

^✉

Corresponding Author: Huan Song, MD, PhD, West China Biomedical Big Data Center, West China Hospital, Sichuan University, Guo Xue Lane 37, Chengdu, China (songhuan@wchscu.cn).

Author Contributions: Ms Zeng and Dr Wei had full access to all of the data in the study and take responsibility for the integrity of the data and the accuracy of the data analysis. Drs Song and Wei contributed equally to this article.

Concept and design: Zeng, Chourpiliadis, Fang, Song, Wei.

Acquisition, analysis, or interpretation of data: All authors.

Drafting of the manuscript: Zeng, Wei.

Critical review of the manuscript for important intellectual content: All authors.

Statistical analysis: Zeng, Valdimarsdóttir.

Obtained funding: Fang, Song.

Administrative, technical, or material support: Zeng, Fang, Song.

Supervision: Zeng, Fang, Song, Wei.

Conflict of Interest Disclosures: Dr Hammar reported receiving personal fees from Sobi and being a shareholder in AstraZeneca outside the submitted work. Dr Valdimarsdóttir reported receiving grants from Horizon2020 and from NordForsk during the conduct of the study. Dr Fang reported receiving grants from European Union during the conduct of the study. No other disclosures were reported.

Funding/Support: This work is supported by the 1.3.5 Project for Disciplines of Excellence, West China Hospital, Sichuan University (No. ZYYC21005 to Dr Song) and the EU Horizon2020 Research and Innovation Action Grant (CoMorMent, No. 847776 to Drs Valdimarsdóttir and Fang). Dr Wei was supported by the Swedish Research Council (No. 2023-00206).

Role of the Funder/Sponsor: The funders had no role in the design and conduct of the study; collection, management, analysis, and interpretation of the data; preparation, review, or approval of the manuscript; and decision to submit the manuscript for publication.

Data Sharing Statement: See Supplement 2.

Additional Contributions: We thank Jing Wu, PhD (Institute of Environmental Medicine, Karolinska Institute, Stockholm, Sweden), the team members, and colleagues involved in West China Biomedical Big Data Center, and Institute of Environmental Medicine for their support. No financial compensation was received for these contributions.

^✉

Corresponding author.

PMCID: PMC11339698 PMID: 39167384

This cohort study investigates the association between inflammatory biomarkers and risk of psychiatric disorders.

Key Points

Question

Are inflammatory biomarkers associated with subsequent risk of psychiatric disorders?

Findings

In this cohort study evaluating data of 585 279 individuals from the Swedish Apolipoprotein Mortality Risk (AMORIS) cohort and validated with the data of 485 620 individuals from the UK Biobank, inflammatory biomarkers including leukocytes, haptoglobin, C-reactive protein, and immunoglobulin G were associated with the risk of psychiatric disorders using cohort and nested case-control study analysis. Moreover, mendelian randomization analyses suggested a possible causal link between leukocytes and depression.

Meaning

This study suggests a role of inflammation in the development of psychiatric disorders and may aid in identifying individuals at high risk.

Abstract

Importance

Individuals with psychiatric disorders have been reported to have elevated levels of inflammatory biomarkers, and prospective evidence is limited regarding the association between inflammatory biomarkers and subsequent psychiatric disorders risk.

Objective

To assess the associations between inflammation biomarkers and subsequent psychiatric disorders risk.

Design, Setting, and Participants

This was a prospective cohort study including individuals from the Swedish Apolipoprotein Mortality Risk (AMORIS) cohort, with no prior psychiatric diagnoses and having a measurement of at least 1 inflammatory biomarker. Data from the UK Biobank were used for validation. Longitudinal trajectories of studied biomarkers were visualized before diagnosis of psychiatric disorders in the AMORIS cohort via a nested case-control study. In addition, genetic correlation and mendelian randomization (MR) analyses were conducted to determine the genetic overlap and causality of the studied associations using publicly available GWAS summary statistics.

Exposures

Inflammatory biomarkers, eg, leukocytes, haptoglobin, immunoglobulin G (IgG), C-reactive protein (CRP), platelets, or albumin.

Main Outcomes and Measures

Any psychiatric disorder or specific psychiatric disorder (ie, depression, anxiety, and stress-related disorders) was identified through the International Statistical Classification of Diseases, Eighth, Ninth, and Tenth Revision codes.

Results

Among the 585 279 individuals (mean [SD] age, 45.5 [14.9] years; 306 784 male [52.4%]) in the AMORIS cohort, individuals with a higher than median level of leukocytes (hazard ratio [HR], 1.11; 95% CI, 1.09-1.14), haptoglobin (HR, 1.13; 95% CI, 1.12-1.14), or CRP (HR, 1.02; 95% CI, 1.00-1.04) had an elevated associated risk of any psychiatric disorders. In contrast, we found an inverse association for IgG level (HR, 0.92; 95% CI, 0.89-0.94). The estimates were comparable for depression, anxiety, and stress-related disorders, specifically, and these results were largely validated in the UK Biobank (n = 485 620). Analyses of trajectories revealed that individuals with psychiatric disorders had higher levels of leukocytes and haptoglobin and a lower level of IgG than their controls up to 30 years before the diagnosis. The MR analysis suggested a possible causal relationship between leukocytes and depression.

Conclusions and Relevance

In this cohort study, inflammatory biomarkers including leukocytes, haptoglobin, CRP, and IgG were associated with a subsequent risk of psychiatric disorders, and thus might be used for high-risk population identification. The possible causal link between leukocytes and depression supports the crucial role of inflammation in the development of psychiatric disorders.

Introduction

Psychiatric disorders are some of the leading contributors to the global health burden, with an estimated 970 million cases in 2019.¹ Depression and anxiety disorders are the most common psychiatric disorders, accounting for about 60% (581 million) of estimated cases.¹ Although relatively rare, with an estimated 12-month prevalence of 1%, stress-related disorders are also a public health concern.² Apart from genetic variants, a large body of evidence suggests that psychiatric disorders are strongly associated with psychosocial stress, early-life adversities, obesity (body mass index ≥30; calculated as weight in kilograms divided by height in meters squared), smoking, unhealthy diet, poor sleep, and physical inactivity.³ Inflammation has been proposed as a mechanistic link behind these risk factors for psychiatric disorders.⁴

Inflammation may contribute to the development of psychiatric disorders through several pathways.^4,5 Increasing evidence suggests that inflammatory cytokines could cross the blood-brain barrier and contribute to neuropathology through affecting neurotransmitter metabolism, neuroendocrine function, and neural plasticity,^6,7 which might induce cognition dysfunction and emotional dysregulation. For example, inflammatory cytokines and their metabolites can decrease the synthesis and release (but increase the reuptake) of serotonin, norepinephrine, and dopamine, leading to reduced activities in reward-related regions in the brain.⁵

Despite the growing research interests in the relationship between inflammation and psychiatric disorders, previous studies are limited by mostly cross-sectional design, ie, compare the levels of inflammatory biomarkers between patients with psychiatric disorders and controls.^8,9,10 Although significant differences have been reported between these groups, the directionality and causality of these associations remain unclear. Yet, few cohort studies have explored the association between inflammatory biomarkers (such as C-reactive protein [CRP] and interleukin 6) and subsequent psychiatric disorders^11,12,13,14 with inconsistent findings. Moreover, the absence of a significant time lag between the testing of inflammatory biomarkers and the identification of psychiatric disorders may significantly heighten concerns about reverse causation.^11,12

To this end, taking advantage of a large population-based cohort study with over 30 years of follow-up in Sweden, along with a validation dataset using the UK Biobank, we aimed to investigate associations of blood inflammatory biomarkers with risk of any psychiatric disorder as well as, specifically, depression, anxiety, and stress-related disorders. We further intended to perform genetic correlations analyses and a mendelian randomization (MR) study to evaluate the potential causal relationship between inflammation and the studied psychiatric disorders using publicly available genome-wide association studies (GWAS) summary statistics.

Methods

Study Design

This study was approved by the Regional Ethical Review Board in Stockholm (2018/2401-31) and the Biomedical Research Ethics Committee of West China Hospital (2019-1171). The requirement for informed consent in the Swedish Apolipoprotein Mortality Risk (AMORIS) study was waived due to the large population and time elapsed from baseline, whereas the UK Biobank collected all data after written informed consent was obtained from each participant.

We analyzed the associations between inflammatory biomarkers and subsequent risk of psychiatric disorders by a triangulation approach (Box). First, we examined the associations of interest by a cohort design using data from the AMORIS cohort. We repeated the analyses in a subcohort of AMORIS where the biomarkers were measured by blood sampling at occupational health checkups (screening) to address concerns of indication bias, and we validated the results in the UK Biobank cohort, to evaluate the robustness of our findings. Second, we visualized longitudinal trajectories of the studied biomarkers during the decades preceding the diagnosis of psychiatric disorders through a nested case-control study using the AMORIS cohort. Finally, we analyzed the genetic correlation and causal relationship between inflammatory biomarkers and psychiatric disorders using publicly available GWAS summary statistics (eTable 1 in Supplement 1). Information on race and ethnicity was not gathered in the Swedish national health and population registers and, therefore, was not included in this study.

Box. Study Design.

Associations of inflammatory biomarkers with psychiatric disorders
- 585 297 Exploration dataset: a prospective cohort analysis using the Swedish Apolipoprotein-Related Mortality Risk (AMORIS) study
  - Biomarkers: leukocytes, haptoglobin, immunoglobulin G, albumin, C-reactive protein, platelet, and albumin; lymphocytes, monocytes, neutrophils, and eosinophils; lymphocyte to monocyte ratio, neutrophil to lymphocyte ratio, platelet to lymphocyte ratio, and C-reactive protein to albumin ratio
- 485 620 Validation dataset: a prospective cohort analysis using the UK Biobank
Longitudinal trajectories of changes in inflammatory biomarkers before the diagnosis of psychiatric disorders
- AMORIS data: a nested case-control analysis
- To visualize longitudinal patterns of inflammatory biomarkers before the date of diagnosis (up to 30 y)
- To evaluate the associations of inflammatory biomarkers with common psychiatric disorders in each 2-y period before the diagnosis
Genetic correlation and potential causal relationship between inflammatory biomarkers and common psychiatric disorders
- Genome-wide association study summary statistics: genetic correlation analysis and mendelian randomization (MR) analysis
- Linkage disequilibrium score regression and genetic covariance analyzer approach to detect the pleiotropic effect
- Inverse variance–weighted method, MR-Egger, the weighted median regression method, and MR with correlated horizontal pleiotropy unraveling shared etiology and confounding to estimate the associations

This study followed the Strengthening the Reporting of Observational Studies in Epidemiology (STROBE) reporting guidelines.

Data Source

The AMORIS cohort is a population-based cohort including 812 073 individuals undergoing health examinations during the period 1985 to 1996 who mainly resided in the Stockholm, Sweden, region. Among individuals with at least 1 measurement of inflammatory biomarkers after the age of 20 years, we excluded individuals who had a reused social identification number, any psychiatric disorder before the first measurement of inflammatory biomarkers (health examinations), or conflicting information.

We focused on several common inflammatory biomarkers, including leukocyte, haptoglobin, immunoglobulin G (IgG), albumin, nonsensitive CRP (ie, high-sensitivity CRP was not commonly measured in AMORIS), platelet, and albumin levels as well as subtypes of leukocytes, namely lymphocytes, monocytes, neutrophils, and eosinophils. We also calculated lymphocyte to monocyte ratio (LMR), neutrophil to lymphocyte ratio (NLR), platelet to lymphocyte ratio (PLR), and CRP to albumin ratio (CAR) as additional biomarkers.

We defined the first diagnosis of any psychiatric disorder, depression, anxiety, and stress-related disorders separately through inpatient or outpatient hospital visits recorded in the National Patient Register using the Swedish revisions of the International Classification of Diseases, Eighth, Ninth, and Tenth Revision codes (eTable 2 in Supplement 1). Individuals were followed up from the first measurement of each biomarker until the first diagnosis of studied psychiatric disorders, emigration, death, or the end of follow-up (December 31, 2020), whichever occurred first.

Covariates included fasting status, year at enrollment, sex, birth year, country of birth, marital status, education, income before blood sampling, and Charlson Comorbidity Index (CCI), which includes cancer, myocardial infarction, heart failure, diabetes, dementia, etc, calculated by the Swedish adapted approach.¹⁵

By applying similar criteria in the AMORIS cohort, we included individuals from the UK Biobank in a validation analysis. We obtained GWAS data of the studied traits from Psychiatric Genomics Consortium,¹⁶ iPSYCH¹⁷ and several public databases.^18,19,20,21 Details of these data are presented in the eMethods in Supplement 1.

Statistical Analysis

Prospective Cohort Study

In the AMORIS cohort, we performed a prospective cohort analysis for each biomarker under study. We calculated hazard ratios (HRs) with 95% CIs for the associations between inflammatory biomarkers and risk of psychiatric disorders using Cox regression models. We analyzed any psychiatric disorder as well as depression, anxiety, and stress-related disorders separately. We considered the first measurement of an inflammatory biomarker in this analysis. In the analysis, we adjusted for sex, age at enrollment, country of birth, fasting status at sampling, CCI, marital status, highest educational attainment, and income. We analyzed the biomarkers with categories by their median values and quartiles, as well as standardization by SDs. To investigate whether the strength of the association was different in males and females, we performed a stratified analysis by sex and tested the statistical significance of the interaction between the biomarkers and sex.

To mitigate reverse causality, we reran the main analysis by excluding the first 5 years of follow-up. To address indication bias (ie, a biomarker measured at an outpatient visit might be secondary to the related underlying disease), we repeated the main analysis in the AMORIS screening cohort. To exclude the confounding effect by cancer or anticancer therapies, we performed a sensitivity analysis by excluding individuals with a history of cancer before the first blood sampling. Finally, we validated our results using corresponding data from the UK Biobank.

Nested Case-Control Study

We performed a nested case-control study to describe trajectories of the studied biomarkers up to 30 years preceding the psychiatric disorders diagnosis in the AMORIS cohort. Cases included newly diagnosed psychiatric disorders (ie, any psychiatric disorder, depression, anxiety, or stress-related disorders, respectively) during follow-up, with the first diagnosis date defined as the index date. Using the incidence density sampling method, we randomly selected 5 controls who were free of any psychiatric disorders and still at follow-up for each case individually matched by sex, birth year, and year of enrollment. In the analysis, we included all available blood measurements concerning the biomarkers of interest before the index date. We first plotted the mean values of each studied biomarker among cases and controls during the 30 years before the index date using locally weighted scatterplot smoothing. Further, we used conditional logistic regression models to estimate odds ratios (ORs) and 95% CIs of the studied associations for each 2-year period of the 30 years before the index date. In the case of multiple measurements of biomarkers, the mean level of biomarkers during a 2-year period was included. Further, we also performed stratified analyses by sex. To thoroughly assess the differences in inflammatory biomarker trajectories by disease status, we performed another analysis using linear mixed-effects model that can handle unbalanced data of the AMORIS study (ie, individuals with different measurement time points) (eMethods in Supplement 1).²²

Genetic Correlation and MR Analysis

We used linkage disequilibrium score regression²³ and genetic covariance analyzer²⁴ to estimate the genetic correlations of interest. Further, we tested the association of inflammatory biomarkers with depression, anxiety, and stress-related disorders by several MR analyses (ie, inverse variance weighted [IVW],²⁵ MR-Egger regression,²⁶ weighted median regression,²⁷ and MR with correlated horizontal pleiotropy unraveling shared etiology and confounding [MR-CUE²⁸]) using GWAS summary statistics (eMethods in Supplement 1). We also used MR Steiger to test the appropriate direction.²⁹ We additionally performed a leave-one-out analysis as the sensitivity analysis for significant associations.

Results

Cohort Study Analysis

Among the 585 279 individuals (mean [SD] age, 45.5 [14.9] years; 278 495 female [47.6%]; 306 784 male [52.4%]) in the AMORIS cohort, a total of 93 961 (16.05%) received a diagnosis of psychiatric disorders during a mean (SD) follow-up of 24.0 (9.5) years (eFigure 1 in Supplement 1). Data from 485 620 individuals from the UK Biobank were used for validation studies (eFigure 2 in Supplement 1). Compared with individuals without psychiatric disorders, those with psychiatric disorders were more likely to be female (49 683 [52.9%] vs 228 812 [46.6%]) and single (28 332 [30.2%] vs 133 017 [27.1%]) or divorced or widowed (18 871 [20.1%] vs 79 628 [16.2%]). Conversely, they exhibited lower rates of education exceeding 12 years (23 711 [25.2%] vs 140 635 [28.6%]), higher income (middle income, 58 431 [62.2%] vs 292 281 [59.5%]), and lower participation in occupational health checkups (23 911 [25.5%] vs 168 507 [34.3%]) (Table 1). Characteristics of 254 670 individuals in the screening cohort of occupational health checkups were similar to those of the entire cohort (eTable 3 in Supplement 1).

Table 1. Characteristics of Study Individuals at the First Blood Sampling in the Swedish Apolipoprotein Mortality Risk (AMORIS) Cohort.

Characteristic	With any psychiatric disorder (n = 93 961)	Without any psychiatric disorder (n = 491 318)	Overall (N = 585 279)
Age, mean (SD), y	45.2 (14.9)	45.5 (14.9)	45.5 (14.9)
Sex, No. (%)
Female	49 683 (52.88)	228 812 (46.57)	278 495 (47.58)
Male	44 278 (47.12)	262 506 (53.43)	306 784 (52.42)
Country of birth, No. (%)
Sweden	79 300 (84.40)	422 437 (85.98)	501 737 (85.73)
Other Nordic countries	6302 (6.71)	29 747 (6.05)	36 049 (6.16)
Other countries	8357 (8.89)	39 106 (7.96)	47 463 (8.11)
Unknown	2 (0)	28 (0.01)	30 (0.01)
Fasting status, No. (%)
Overnight fasting	47 865 (50.94)	254 587 (51.82)	302 452 (51.68)
No overnight fasting	45 609 (48.54)	234 366 (47.70)	279 975 (47.84)
Unknown	487 (0.52)	2365 (0.48)	2852 (0.49)
Marital status, No. (%)
Single	28 332 (30.15)	133 017 (27.07)	161 349 (27.57)
Married or cohabiting	46 528 (49.52)	271 747 (55.31)	318 275 (54.38)
Divorced or widowed	18 871 (20.08)	79 628 (16.21)	98 499 (16.83)
Unknown	230 (0.24)	6926 (1.41)	7156 (1.22)
Education level, No. (%)
<9 y	14 000 (14.90)	73 640 (14.99)	87 640 (14.97)
9-12 y	53 669 (57.12)	255 831 (52.07)	309 500 (52.88)
>12 y	23 711 (25.23)	140 635 (28.62)	164 346 (28.08)
Unknown	2581 (2.75)	21212 (4.32)	23 793 (4.07)
Income level, No. (%)
Lowest 20%	19 597 (20.86)	84 914 (17.28)	104 511 (17.86)
Middle	58 431 (62.19)	292 281 (59.49)	350 712 (59.92)
Top 20%	15 703 (16.71)	107 197 (21.82)	122 900 (21.00)
Unknown	230 (0.24)	6926 (1.41)	7156 (1.22)
From occupational health checkups, No. (%)
Yes	23 911 (25.45)	168 507 (34.30)	192 418 (32.88)
No	70 050 (74.55)	322 811 (65.70)	392 861 (67.12)
Charlson Comorbidity Index, mean (SD)	0.123 (0.471)	0.133 (0.522)	0.131 (0.514)
Inflammatory biomarkers
Leukocyte, 10⁹ cells/L
No.	38 643	162 747	201 390
Mean (SD)	6.63 (2.07)	6.61 (2.65)	6.61 (2.55)
Haptoglobin, g/L
No.	66 166	337 104	403 270
Mean (SD)	1.08 (0.33)	1.07 (0.34)	1.07 (0.34)
Immunoglobulin G, g/L
No.	23 950	107 995	131 945
Mean (SD)	11.40 (2.89)	11.59 (3.04)	11.55 (3.01)
C-reactive protein, mg/L
No.	54 748	286 134	340 882
Mean (SD)	5.99 (16.93)	6.08 (17.18)	6.07 (17.14)
Platelet, 10⁹ cells/L
No.	37 100	155 243	192 343
Mean (SD)	263.93 (65.80)	262.29 (67.13)	262.61 (66.88)
Albumin, g/L
No.	80 283	406 947	487 230
Mean (SD)	43.19 (2.87)	43.20 (2.91)	43.20 (2.90)
Subtypes of leukocyte
Lymphocyte, 10⁹ cells/L
No.	10 633	42 334	52 967
Mean (SD)	2.23 (1.13)	2.28 (2.85)	2.27 (2.59)
Monocyte, 10⁹ cells/L
No.	10 632	42 333	52 965
Mean (SD)	0.40 (0.22)	0.41 (0.41)	0.40 (0.38)
Neutrophil, 10⁹ cells/L
No.	5630	21 981	27 611
Mean (SD)	3.97 (1.79)	3.96 (1.78)	3.96 (1.78)
Eosinophil, 10⁹ cells/L
No.	10 633	42 327	52 960
Mean (SD)	0.18 (0.17)	0.18 (0.21)	0.18 (0.20)
Others
Lymphocyte:monocyte ratio
No.	10 533	41 887	52 420
Mean (SD)	7.89 (7.80)	7.97 (8.11)	7.95 (8.05)
Neutrophil:lymphocyte ratio
No.	5630	21 981	27 611
Mean (SD)	2.14 (1.31)	2.16 (1.38)	2.16 (1.36)
Platelet:lymphocyte ratio
No.	9732	38 094	47 826
Mean (SD)	135.13 (54.48)	136.75 (60.86)	136.42 (59.62)
C-reactive protein:albumin ratio
No.	51 784	268 761	320 545
Mean (SD)	0.13 (0.37)	0.14 (0.39)	0.14 (0.38)

Open in a new tab

SI conversion factor: To convert leukocyte count to per microliter, divide by 0.001; C-reactive protein to milligrams per deciliter, divide by 10; haptoglobin to milligrams per deciliter, multiply by 1000 and divide by 10; platelet count to ×10³ per microliter, divide by 1; immunoglobulin G to milligrams per deciliter, divide by 0.01; albumin to grams per deciliter, divide by 1000 and divide by 10.

Individuals with a higher than median level of leukocytes had an associated increased risk of any psychiatric disorder (HR, 1.11; 95% CI, 1.09-1.14). Similar associations were found for the subtypes of leukocytes, namely monocytes, neutrophils, and eosinophils (Table 2). We also found a higher risk of any psychiatric disorders among those with higher levels of haptoglobin (HR, 1.13; 95% CI, 1.12-1.14), CRP (HR, 1.02; 95% CI, 1.00-1.04), albumin (HR, 1.02; 95% CI, 1.01-1.04), NLR (HR, 1.08; 95% CI, 1.02-1.13), and CAR (HR, 1.06; 95% CI, 1.04-1.07). In contrast, an inverse association was observed for IgG level (HR, 0.92; 95% CI, 0.89-0.94), platelet level (HR, 0.98; 95% CI, 0.96-1.00), and LMR (HR, 0.87; 95% CI, 0.84-0.91). We found similar associations of leukocyte, monocyte, haptoglobin, and IgG level, CAR, and LMR with depression, anxiety, and stress-related disorders, specifically (Table 2). Largely similar results were observed when estimating the biomarkers per SD increase or categorizing them by quartiles (eTables 4-5 in Supplement 1).

Table 2. Associations Between Inflammatory Biomarkers (Cutoff Points Set as Median Values) and Psychiatric Disorders.

Biomarker	Cutoff point	Any psychiatric disorder		Depression		Anxiety		Stress-related disorders
Biomarker	Cutoff point	Outcomes (incidence)	HR (95% CI)^a	Outcomes (incidence)	HR (95% CI)^a	Outcomes (incidence)	HR (95% CI)^a	Outcomes (incidence)	HR (95% CI)^a
Inflammatory biomarkers
Leukocyte, 10⁹/L	≤6.3	19 821 (8.44)	1 [Reference]	5372 (2.18)	1 [Reference]	4411 (1.77)	1 [Reference]	3219 (1.29)	1 [Reference]
Leukocyte, 10⁹/L	>6.3	18 822 (9.43)	1.11 (1.09-1.14)^b	5166 (2.44)	1.09 (1.05-1.14)^b	4232 (1.99)	1.07 (1.03-1.12)^b	3073 (1.44)	1.03 (0.98-1.09)
Haptoglobin, g/L	≤1	33 531 (5.97)	1 [Reference]	8761 (1.50)	1 [Reference]	6927 (1.18)	1 [Reference]	5320 (0.90)	1 [Reference]
Haptoglobin, g/L	>1	32 635 (7.10)	1.13 (1.12-1.15)^b	7987 (1.66)	1.16 (1.12-1.19)^b	5717 (1.18)	1.09 (1.05-1.13)^b	3849 (0.79)	1.06 (1.02-1.10)^b
Immunoglobulin G, g/L	≤11	13 544 (7.57)	1 [Reference]	3306 (1.76)	1 [Reference]	2520 (1.33)	1 [Reference]	1621 (0.85)	1 [Reference]
Immunoglobulin G, g/L	>11	10 406 (6.99)	0.92 (0.89-0.94)^b	2696 (1.73)	0.95 (0.90-0.99)^b	2089 (1.33)	0.93 (0.88-0.99)^b	1414 (0.90)	0.93 (0.86-1.00)^b
C-reactive protein, mg/L	≤4	32 254 (6.79)	1 [Reference]	8635 (1.74)	1 [Reference]	6942 (1.39)	1 [Reference]	5095 (1.02)	1 [Reference]
C-reactive protein, mg/L	>4	22 494 (7.09)	1.02 (1.00-1.04)^b	5879 (1.77)	1.02 (0.98-1.05)	4577 (1.37)	1.00 (0.96-1.03)	3355 (1.00)	1.01 (0.96-1.05)
Platelet, 10⁹/L	≤256	18 243 (8.94)	1 [Reference]	4854 (2.25)	1 [Reference]	4045 (1.87)	1 [Reference]	2966 (1.36)	1 [Reference]
Platelet, 10⁹/L	>256	18 857 (8.96)	0.98 (0.96-1.00)^b	5265 (2.37)	1.01 (0.97-1.05)	4262 (1.90)	0.95 (0.91-0.99)^b	3026 (1.35)	0.92 (0.87-0.96)^b
Albumin, g/L	≤43	43 791 (7.17)	1 [Reference]	10 743 (1.69)	1 [Reference]	7948 (1.24)	1 [Reference]	5662 (0.88)	1 [Reference]
Albumin, g/L	>43	36 492 (6.52)	1.02 (1.01-1.04)^b	10 072 (1.72)	1.03 (1.00-1.06)	8201 (1.39)	1.13 (1.09-1.16)^b	6233 (1.06)	1.06 (1.02-1.10)^b
Subtypes of leukocyte
Lymphocyte, 10⁹/L	≤2.07	5257 (9.31)	1 [Reference]	1455 (2.44)	1 [Reference]	1225 (2.04)	1 [Reference]	891 (1.48)	1 [Reference]
Lymphocyte, 10⁹/L	>2.07	5376 (9.25)	1.01 (0.97-1.04)	1535 (2.49)	1.01 (0.94-1.08)	1262 (2.04)	0.97 (0.89-1.05)	982 (1.58)	1.00 (0.91-1.10)
Monocyte, 10⁹/L	≤0.38	5391 (8.76)	1 [Reference]	1502 (2.31)	1 [Reference]	1215 (1.86)	1 [Reference]	944 (1.44)	1 [Reference]
Monocyte, 10⁹/L	>0.38	5241 (9.88)	1.16 (1.11-1.20)^b	1488 (2.65)	1.17 (1.09-1.26)^b	1271 (2.25)	1.29 (1.19-1.40)^b	929 (1.63)	1.22 (1.12-1.34)^b
Neutrophil, 10⁹/L	≤3.6	2896 (10.33)	1 [Reference]	805 (2.71)	1 [Reference]	694 (2.32)	1 [Reference]	539 (1.79)	1 [Reference]
Neutrophil, 10⁹/L	>3.6	2734 (10.98)	1.06 (1.00-1.11)^b	790 (2.98)	1.08 (0.98-1.20)	683 (2.56)	1.09 (0.98-1.21)	501 (1.87)	1.00 (0.88-1.13)
Eosinophil, 10⁹/L	≤0.14	5308 (9.09)	1 [Reference]	1531 (2.48)	1 [Reference]	1303 (2.10)	1 [Reference]	976 (1.56)	1 [Reference]
Eosinophil, 10⁹/L	>0.14	5325 (9.48)	1.05 (1.01-1.09)^b	1459 (2.45)	1.01 (0.94-1.09)	1184 (1.98)	0.98 (0.91-1.06)	897 (1.49)	1.03 (0.94-1.13)
Others
LMR	≤5.5	5216 (9.88)	1 [Reference]	1476 (2.64)	1 [Reference]	1232 (2.19)	1 [Reference]	904 (1.60)	1 [Reference]
LMR	>5.5	5317 (8.79)	0.87 (0.84-0.91)^b	1487 (2.33)	0.85 (0.79-0.92)^b	1236 (1.92)	0.81 (0.75-0.88)^b	955 (1.48)	0.84 (0.76-0.92)^b
NLR	≤1.86	2821 (10.26)	1 [Reference]	795 (2.73)	1 [Reference]	698 (2.38)	1 [Reference]	535 (1.82)	1 [Reference]
NLR	>1.86	2809 (11.03)	1.08 (1.02-1.13)^b	800 (2.95)	1.09 (0.98-1.20)	679 (2.49)	1.05 (0.95-1.17)	505 (1.84)	1.01 (0.90-1.14)
PLR	≤126.45	4888 (9.46)	1 [Reference]	1365 (2.50)	1 [Reference]	1136 (2.07)	1 [Reference]	867 (1.57)	1 [Reference]
PLR	>126.45	4844 (9.41)	0.97 (0.93-1.01)	1374 (2.52)	1.00 (0.93-1.08)	1131 (2.06)	0.99 (0.91-1.08)	820 (1.49)	0.98 (0.89-1.07)
CAR	≤0.09	26 082 (6.65)	1 [Reference]	6958 (1.70)	1 [Reference]	5598 (1.36)	1 [Reference]	4102 (0.99)	1 [Reference]
CAR	>0.09	25 702 (7.31)	1.06 (1.04-1.07)^b	6720 (1.83)	1.05 (1.01-1.09)^b	5268 (1.43)	1.04 (1.00-1.08)^b	3833 (1.03)	1.06 (1.01-1.11)^b

Open in a new tab

Abbreviations: CAR, C-reactive protein to albumin ratio; LMR, lymphocyte to monocyte ratio; NLR, neutrophil to lymphocyte ratio; PLR, platelet to lymphocyte ratio.

^{^a}

HR and 95% CIs were estimated by Cox regression models, adjusting for sex, age at first blood sampling, country of birth, fasting status, CCI, marital status, education level, and income level.

^{^b}

Results with a P value <.05.

We found similar results when restricting the analyses to the AMORIS screening cohort, after excluding the first 5 years of follow-up, after excluding individuals with a history of cancer before the first blood sampling (12 894 individuals accounting for 2.2% of the included study participants) or when repeating the analyses in the UK Biobank study (eTables 6-9 in Supplement 1). The associations were mostly consistent among males and females, although the association for albumin levels was more pronounced in females than in males (HR, 1.08; 95% CI, 1.05-1.10 vs HR, 0.99; 95% CI, 0.97-1.01; P for interaction = 3.75 × 10⁻²²). Conversely, the association for haptoglobin was stronger in males than females (HR, 1.07; 95% CI, 1.05-1.10 in females and HR, 1.20; 95% CI, 1.17-1.23 in males; P for interaction: 1.31 × 10⁻¹⁸) (eTables 10-11 in Supplement 1).

Nested Case-Control Study Analysis

A total of 463 356 participants were included in this analysis. We found that the levels of leukocytes and haptoglobin were consistently higher among patients with any psychiatric disorder than their controls during the 30 years before the index date, whereas the pattern was the opposite for IgG, particularly within the 10 years before the diagnosis (eg, mean value at 10 years before diagnosis, case vs control, 6.67 vs 6.56 ×10⁹ cells/L for leukocyte level [to convert to per microliter, divide by 0.001]; 1.11 vs 1.08 g/L for haptoglobin level [to convert to milligrams per deciliter, multiply by 1000 and divide by 10]; 11.24 vs 11.38 g/L for IgG level [to convert to milligrams per deciliter, divide by 0.01]) (Figure, A). Additionally, higher leukocytes and haptoglobin levels were associated with an increased risk of any psychiatric disorder in each 2-year interval before diagnosis (ORs ranging from 1.05-1.17 for leukocytes and ORs ranging from 1.05-1.25 for haptoglobin during the 30 years before diagnosis) (Figure, B). Largely similar results were observed for depression, anxiety, and stress-related disorders (eFigure 3 in Supplement 1). We also found that individuals with any psychiatry disorder and depression, anxiety, or stress-related disorders, specifically, had a higher level of neutrophils and NLR than their controls during the 5 or 10 years before diagnosis (eFigures 4-6 in Supplement 1). The temporal patterns of inflammatory biomarkers between cases and controls were largely similar between males and females. However, the albumin level was higher in cases than controls among females but lower in cases than controls among males (eFigures 7-8 in Supplement 1). In addition, we observed from linear mixed-effects models that both the level and change rate of leukocytes, haptoglobin and IgG significantly differed between patients with any psychiatric disorder and controls (eFigure 9 and eTable 12 in Supplement 1).

Figure. — A, Mean concentrations of inflammatory biomarkers between cases and controls during the 30 years before the index date. B, Odds ratios and 95% CIs for the associations of inflammatory biomarkers with any psychiatry disorder with the standardization by SD in each 2-year period of the 30 years before the index date. Odds ratios and 95% CIs were estimated by conditional logistic regression models, adjusting for matching factors (sex, year of birth, and calendar period of enrollment to the Swedish Apolipoprotein Mortality Risk [AMORIS] study) and fasting status at sampling.

SI conversion factor: To convert leukocyte count to per microliter, divide by 0.001; C-reactive protein to milligrams per deciliter, divide by 10; haptoglobin to milligrams per deciliter, multiply by 1000 and divide by 10; platelet count to ×10³ per microliter, divide by 1; immunoglobulin G to milligrams per deciliter, divide by 0.01; albumin to grams per deciliter, divide by 1000 and divide by 10.

Genetic Correlation and MR Analysis

We found positive genetic correlations of leukocytes with anxiety as well as CRP with depression and stress-related disorders but negative genetic correlations of platelet and albumin with stress-related disorders (eTable 13 in Supplement 1). The MR analysis showed that F statistics of selected instrumental variables ranged from 34.24 to 427.22, indicating a low possibility of weak instrument bias (Table 3). We found that a higher level of leukocytes was consistently associated with an increased odds of depression across different methods of MR analysis (Table 3 and eFigure 10 in Supplement 1). Moreover, the MR-Egger intercept test for leukocytes and depression suggested no statistically significant effects of horizontal pleiotropy (P for intercept = .14), although a statistically significant heterogeneity was noted among instrumental variables (Cochran Q statistic = 288.77; P = .002) (eTable 14 in Supplement 1). Nevertheless, the association between leukocytes and depression remained stable in the leave-one-out analysis (eFigure 11 in Supplement 1). The relationships of leukocytes with anxiety and stress-related disorders were only observed in the MR-Egger regression. The results from both MR-IVW and MR-CUE suggested an association between CRP and depression. All significant associations passed the MR Steiger test, indicating accuracy of the causal directions, with Steiger P values <0.01. Additionally, each single-nucleotide variant used in the MR analyses for leukocytes and depression passed the Steiger filtering (eTable 15 in Supplement 1).

Table 3. Associations of Inflammatory Biomarkers With Depression, Anxiety, and Stress-Related Disorders According to the Mendelian Randomization (MR) Analysis.

Biomarker	N_SNV	PVE, %	F statistic	MR-IVW		MR-Egger		MR-weighted median		MR-CUE
Biomarker	N_SNV	PVE, %	F statistic	β (95% CI)	P value	β (95% CI)	P value	β (95% CI)	P value	β (95% CI)	P value
Depression
Inflammatory biomarkers
Leukocyte	222	8.61	73.05	0.06 (0.01 to 0.11)	.01	0.14 (0.03 to 0.25)	.02	0.09 (0.01 to 0.16)	.02	0.08 (0.04 to 0.12)	4.64 × 10⁻⁵
Haptoglobin	26	47.71	374.89	−0.01 (−0.02 to 0.01)	.57	0.02 (−0.03 to 0.06)	.48	−0.01 (−0.04 to 0.01)	.37	0 (0 to 0.01)	.18
Immunoglobulin G	2	1.23	66.59	−0.02 (−0.15 to 0.12)	.83	NA^a	NA^a	NA^a	NA^a	−0.01 (−0.03 to 0.02)	.51
C-reactive protein	440	9.19	132.33	0.06 (0.01 to 0.10)	.01	0.03 (−0.04 to 0.10)	.37	0.03 (−0.04 to 0.10)	.37	0.09 (0.05 to 0.13)	9.32× 10⁻⁷
Platelet	329	17.15	104.31	0 (−0.04 to 0.03)	.80	−0.07 (−0.14 to 0)	.05	−0.02 (−0.07 to 0.03)	.50	0.01 (−0.02 to 0.04)	.48
Albumin	382	7.83	79.57	−0.02 (−0.08 to 0.03)	.42	−0.07 (−0.17 to 0.04)	.24	−0.07 (−0.15 to 0.02)	.13	0 (−0.05 to 0.04)	.86
Subtypes of leukocyte
Neutrophil	188	7.92	77.96	0 (−0.05 to 0.05)	.93	−0.06 (−0.17 to 0.05)	.30	−0.01 (−0.08 to 0.06)	.84	0 (−0.04 to 0.05)	.71
Lymphocyte	221	9.10	77.62	0.01 (−0.03 to 0.06)	.60	0.118 (0.01 to 0.22)	.03	0.01 (−0.06 to 0.07)	.86	0.04 (0 to 0.08)	.03
Monocyte	304	16.55	111.20	0.02 (−0.02 to 0.06)	.28	0.03 (−0.04 to 0.10)	.43	0.01 (−0.04 to 0.07)	.60	0.04 (0 to 0.07)	.03
Eosinophil	240	10.27	82.08	0.03 (−0.02 to 0.07)	.25	0 (−0.09 to 0.10)	.93	0.02 (−0.04 to 0.08)	.56	0.01 (−0.04 to 0.05)	.82
Anxiety
Inflammatory biomarkers
Leukocyte	211	8.53	76.12	−0.02 (−0.16 to 0.12)	.78	0.42 (0.06 to 0.77)	.02	0.13 (−0.09 to 0.35)	.25	0.08 (−0.05 to 0.21)	.21
Haptoglobin	15	37.47	427.22	0 (−0.07 to 0.07)	.97	−0.04 (−0.17 to 0.08)	.50	0.01 (−0.08 to 0.10)	.84	0.01 (−0.04 to 0.05)	.67
Immunoglobulin G	2	1.23	66.59	−0.59 (−1.53 to 0.35)	.22	NA^a	NA^a	NA	NA^a	0.01 (−0.07 to 0.09)	.73
C-reactive protein	378	8.02	132.63	0.06 (−0.08 to 0.20)	.43	0.06 (−0.18 to 0.29)	.64	−0.01 (−0.24 to 0.23)	.97	0.06 (−0.06 to 0.19)	.33
Platelet	305	16.46	107.07	−0.03 (−0.13 to 0.07)	.59	0.10 (−0.12 to 0.32)	.37	0.04 (−0.13 to 0.21)	.63	−0.06 (−0.16 to 0.03)	.20
Albumin	356	6.87	74.16	−0.01 (−0.19 to 0.18)	.95	−0.20 (−0.65 to 0.24)	.37	−0.12 (−0.41 to 0.18)	.44	0.05 (−0.10 to 0.19)	.54
Subtypes of leukocyte
Neutrophil	179	7.60	78.38	0.04 (−0.11 to 0.18)	.63	−0.08 (−0.43 to 0.27)	.65	0.14 (−0.09 to 0.38)	.23	−0.03 (−0.16 to 0.11)	.68
Lymphocyte	212	9.01	80.05	−0.07 (−0.21 to 0.07)	.32	0.11 (−0.25 to 0.48)	.54	0.02 (−0.20 to 0.23)	.88	−0.03 (−0.15 to 0.09)	.65
Monocyte	279	14.87	106.71	0.05 (−0.05 to 0.15)	.34	0.26 (0.05 to 0.48)	.02	0.14 (−0.04 to 0.32)	.12	0.04 (−0.07 to 0.15)	.49
Eosinophil	220	9.78	84.77	0.04 (−0.09 to 0.16)	.59	0.03 (−0.28 to 0.35)	.83	0.03 (−0.17 to 0.23)	.76	0.02 (−0.10 to 0.14)	.75
Stress-related disorders
Inflammatory biomarkers
Leukocyte	216	8.44	73.47	0.07 (−0.02 to 0.16)	.14	0.21 (0 to 0.42)	.047	0.11 (−0.04 to 0.25)	.15	0.07 (0 to 0.15)	.06
Haptoglobin	24	44.91	362.87	−0.02 (−0.06 to 0.02)	.25	0.02 (−0.06 to 0.11)	.61	−0.04 (−0.09 to 0.01)	.15	0.01 (−0.02 to 0.04)	.37
Immunoglobulin G	2	0.64	34.24	−0.18 (−0.56 to 0.20)	.35	NA^a	NA^a	NA^a	NA^a	−0.01 (−0.07 to 0.05)	.74
C-reactive protein	409	8.70	133.91	0.08 (−0.01 to 0.17)	.09	−0.04 (−0.18 to 0.10)	.57	0.04 (−0.13 to 0.19)	.67	0 (−0.01 to 0.01)	.99
Platelet	304	16.35	106.60	0.04 (−0.03 to 0.10)	.31	0.02 (−0.13 to 0.16)	.83	0 (−0.11 to 0.11)	.97	0.06 (−0.01 to 0.12)	.08
Albumin	365	7.49	79.32	−0.02 (−0.12 to 0.09)	.73	−0.14 (−0.35 to 0.07)	.18	−0.03 (−0.21 to 0.14)	.73	−0.03 (−0.12 to 0.06)	.52
Subtypes of leukocyte
Neutrophil	180	7.64	78.32	−0.04 (−0.13 to 0.06)	.44	0 (−0.23 to 0.23)	>.99	0.01 (−0.14 to 0.15)	.92	−0.03 (−0.12 to 0.05)	.42
Lymphocyte	211	8.76	77.97	0.03 (−0.06 to 0.12)	.47	0.135 (−0.076 to 0.347)	.21	0.06 (−0.08 to 0.20)	.39	0.07 (−0.01 to 0.14)	.07
Monocyte	288	15.08	105.12	0.04 (−0.03 to 0.10)	.31	0.051 (−0.083 to 0.185)	.46	0.04 (−0.07 to 0.16)	.46	0.04 (−0.02 to 0.12)	.19
Eosinophil	231	9.95	82.32	0.01 (−0.07 to 0.10)	.78	−0.035 (−0.232 to 0.162)	.73	−0.02 (−0.16 to 0.12)	.76	0.02 (−0.05 to 0.10)	.52

Open in a new tab

Abbreviations: CUE, correlated horizontal pleiotropy unraveling shared etiology and confounding; IVW, inverse variance weighted; N_SNV, the number of SNV used in analyses; NA, not available; PVE, the proportion of variance explained; SNV, single-nucleotide variation.

^{^a}

No result because corresponding method requires more than 2 instrumental variants.

Discussion

Through a triangulation approach using multiple study designs and large-scale data from the Swedish AMORIS cohort, the UK Biobank, and GWAS summary statistics, we found that individuals with higher than median levels of leukocytes, haptoglobin, and CRP, or lower than median levels of IgG, were associated with an elevated risk of psychiatric disorders. Individuals with psychiatric disorders had persistently higher levels of leukocytes and haptoglobin, as well as persistently lower levels of IgG, during the 30 years before diagnosis, compared with controls. Furthermore, a higher level of leukocytes was consistently associated with increased odds of depression across different methods of MR analysis, indicating a possible causal relationship between leukocytes and depression.

In line with recent studies from the National Health and Nutrition Examination Survey cohort³⁰ and the Moli-Sani cohort,³¹ we found that the elevated level of leukocytes (including neutrophils) as an indicator for acute or chronic inflammation was associated with a higher risk of psychiatric disorders and that it might have a causal association with depression. This finding was further supported by a large study of the PsycheMERGE network, which suggested that a high polygenic risk score of depression was associated with an increased white blood cell count.³² Previous studies,^11,12 although not all,^13,14 have found that individuals with a higher level of high-sensitivity CRP had elevated risks of depression, anxiety, and stress-related disorders. We observed such associations in the AMORIS cohort (particularly when CRP was categorized in quartiles) and in the UK Biobank cohort where high-sensitivity CRP was available. Moreover, the MR analysis suggested a potential causal relationship between CRP and depression. In addition, we observed haptoglobin, a positive acute-phase protein along with CRP,³³ to maintain consistently higher levels in cases during the 30 years before diagnosis compared with controls, further implying the role of inflammation in psychiatry disorders. Furthermore, for the first time, we found that a higher level of IgG was associated with a lower risk of psychiatric disorders and that there were distinct trajectories of IgG during the 3 decades before the index date between cases and controls. This is in line with the anti-inflammatory role of IgG.^34,35 We found sex differences in albumin levels between cases and controls. Although previous studies reported that sex steroids could bind albumin in a nonlinear manner³⁶ and sex could affect the modification of albumin lysine residues,³⁷ the effect of these biological interactions on psychiatric disorders remains unclear. Nevertheless, CAR, as an inflammatory indicator based on inflammation and nutritional status, was found to be positively associated with the risk of psychiatric disorders in our study, suggesting that individuals with a simultaneous increase in CRP and decrease in albumin level have an associated increased risk of psychiatric disorders. Notably, we also observed associations of LMR and NLR, predictors of inflammation and adaptive immunity,^38,39 with the risks of any psychiatric disorder, depression, anxiety, and stress-related disorders, which have been minimally investigated in previous studies.

The underlying mechanisms for the associations of serum leukocytes, haptoglobin, CRP, and IgG with psychiatry disorders remain unclear. Possible explanations mainly include blood-brain barrier disruption, microglia activation, neurotransmission impairment, and other interactions between inflammations and neuropathology.⁴⁰ For instance, in cases of chronic inflammation, elevated CRP interacts with proinflammatory cytokines produced by neutrophils and monocytes. Peripheral cytokines have the potential to infiltrate the blood-brain barrier.⁴¹ Specifically, in mouse models, neutrophils have been found to migrate to the brain after lipopolysaccharide (LPS)–induced peripheral inflammation, leading ultimately to depressionlike behaviors.⁴² The transmigration of cytokines and immune cells, accompanied by the signals of peripheral inflammation, may activate microglia and astrocytes.⁴ Particularly, the active interaction between transmigrating neutrophils and activated microglia was found in the LPS-induced inflammation of mice.⁴³ The activated microglia and astrocytes (indicators of neuroinflammation) mainly produce the proinflammation helper T cell (Th1) cytokines, subsequently leading to neurotransmission impairment such as increased brain glutamate levels,⁴⁰ known to be associated with mood disorders.⁴⁴ In addition, although scarce, previous studies have reported that haptoglobin and IgG played important roles in lymphocytes proliferation and cognitive function in mouse models.^45,46 Such mechanisms need further investigation.

Strengths and Limitations

The major strengths of our study include the large sample size (more than 1 million participants in total) and the long, complete, and high-quality follow-up through nationwide registers. The prospective and independent assessments of studied blood biomarkers and outcomes greatly reduced the likelihood of reverse causality. All blood samples were analyzed in the same laboratory using a standard method, assuring a high comparability across individuals. Further, we first provided the clear temporal change of inflammatory biomarkers between cases of psychiatric disorders and controls before the diagnosis, which revealed a long-term effect of inflammation on risk of psychiatry disorders. Finally, the triangulation analytical approach using multiple study designs and data sources provided us with robust evidence for the association between leukocytes and depression.

Nevertheless, this study has several limitations. First, as approximately 50% of individuals in the AMORIS cohort donated blood samples during their visits to an outpatient care setting, indication bias is plausible. However, we observed similar associations in the AMORIS screening cohort as well as in the UK Biobank where all participants were recruited without any health-related indication. Second, we identified outcomes according to hospitalizations and specialist care data in the AMORIS cohort, which might entail a delayed diagnosis and underascertainment of mild cases. Thus, the study concerns more severe cases of psychiatric disorders, which may be more relevant and reliable diagnoses. Moreover, the concern of reverse causality was largely alleviated as similar associations were observed in a sensitivity analysis after excluding the first 5 years of follow-up in AMORIS cohort. Third, given that the genetic correlation and MR analysis relied on published GWAS data, replications with more powerful GWAS data for other biomarkers are needed. Fourth, despite adjusting for several sociodemographic factors and comorbidities in the AMORIS cohort, residual confounding due to unmeasured factors, such as smoking, sleep disturbance, and dietary habit, is still possible. Finally, the generalizability of our findings may be limited to European populations because all analyses in the present study were mainly performed among European individuals.

Conclusions

In this cohort study, collectively using large-scale prospective cohorts and publicly available GWAS summary statistics, our findings suggest a role of chronic systemic low-grade inflammation in subsequent risk of psychiatry disorders. We found that serum leukocytes, haptoglobin, CRP, and IgG were associated with the risk of psychiatric disorders in the prospective cohort analyses, whereas a possible causal relationship between leukocytes and depression was suggested by the MR analysis. The distinct trajectories of leukocytes and haptoglobin with higher levels many years before the diagnosis of psychiatric disorders suggest long-term processes and may aid in the identification of individuals at high risk.

Supplement 1.

eMethods

eFigure 1. Flowchart of Study Individuals Using the Swedish AMORIS Study

eFigure 2. Flowchart of Study Individuals Using the UK Biobank Data

eFigure 3. Trajectories of Inflammatory Biomarkers and the Association With Subsequent Depression, Anxiety, and Stress-Related Disorders

eFigure 4. Trajectories of Subtypes of Leukocyte and Derived Inflammatory Biomarkers and the Association With Subsequent Any Psychiatric Disorders

eFigure 5. Trajectories of Subtypes of Leukocyte and the Association With Subsequent Depression, Anxiety, and Stress-Related Disorders

eFigure 6. Trajectories of Derived Inflammatory Biomarkers and the Association With Subsequent Depression, Anxiety, and Stress-Related Disorders

eFigure 7. Trajectories of Inflammatory Biomarkers and the Association With Subsequent Any Psychiatric Disorders by Sex

eFigure 8. Trajectories of Subtypes of Leukocyte and Derived Inflammatory Biomarkers and the Association With Subsequent Any Psychiatric Disorders by Sex

eFigure 9. Trajectories of Inflammatory Biomarkers During the 30 Years Before Diagnosis in Cases of Any Psychiatric Disorders and Their Individually Matched Controls, Using Linear Mixed-Effects Models

eFigure 10. Associations of Leukocytes With Depression—the Mendelian Randomization Analysis

eFigure 11. Associations of Leukocytes With Depression—the Mendelian Randomization Leave-One-Out Analysis

eTable 1. Genome-Wide Association Study Summary Statistics in This Study

eTable 2. International Classification of Diseases (ICD) Codes for Identification of Psychiatric Disorders

eTable 3. Characteristics of Study Individuals in the AMORI Screening Cohort of Occupational Health Check-Ups at the First Blood Sampling

eTable 4. Associations Between Inflammatory Biomarkers and Psychiatric Disorders After the Standardization by SD

eTable 5. Associations Between Inflammatory Biomarkers and Psychiatric Disorders According to Categories of Biomarkers in Quartiles

eTable 6. Associations Between Inflammatory Biomarkers (Cutoff Points Set as Median Values) and Psychiatric Disorders Among Individuals in AMORIS Screening Cohort of Occupational Health Check-Ups

eTable 7. Associations Between Inflammatory Biomarkers (Cutoff Points Set as Median Values) and Psychiatric Disorders After Excluding Those Who Had a Diagnosis of Psychiatric Disorders Within the First 5 Years After Sampling

eTable 8. Associations Between Inflammatory Biomarkers (Cutoff Points Set as Median Values) and Psychiatric Disorders After Excluding Individuals With a History of Cancer Before the First Blood Sampling

eTable 9. Associations Between Inflammatory Biomarkers (Cutoff Points Set as Median Values) and Psychiatric Disorders in the UK Biobank Study

eTable 10. Associations Between Inflammatory Biomarkers (Cutoff Points Set as Median Values) and Psychiatric Disorders Among Males and Females

eTable 11. P Values for the Interaction Term of sex × Biomarker in Stratified Analysis by Sex for the Associations Between Inflammatory Biomarkers (Cutoff Points Set as Median Values) and Psychiatric Disorders

eTable 12. β Coefficient and 95% CIs of the Association Between Inflammatory Biomarkers and Risk of Any Psychiatric Disorders—Analysis Using Linear Mixed-Effects Model

eTable 13. Genetic Correlations Between Inflammatory Biomarkers and Depression, Anxiety, and Stress-Related Disorders

eTable 14. Results of Heterogeneity and Horizontal Pleiotropy Tests for the Mendelian Randomization Analysis

eTable 15. MR Steiger Results of Each SNP Used in the MR Analyses for Leukocyte and Depression

jamapsychiatry-e242185-s001.pdf^{(4.9MB, pdf)}

Supplement 2.

Data Sharing Statement.

jamapsychiatry-e242185-s002.pdf^{(17.1KB, pdf)}

References

1.GBD 2019 Mental Disorders Collaborators . Global, regional, and national burden of 12 mental disorders in 204 countries and territories, 1990-2019: a systematic analysis for the Global Burden of Disease Study 2019. Lancet Psychiatry. 2022;9(2):137-150. doi: 10.1016/S2215-0366(21)00395-3 [DOI] [PMC free article] [PubMed] [Google Scholar]
2.Gradus JL. Prevalence and prognosis of stress disorders: a review of the epidemiologic literature. Clin Epidemiol. 2017;9:251-260. doi: 10.2147/CLEP.S106250 [DOI] [PMC free article] [PubMed] [Google Scholar]
3.Arango C, Dragioti E, Solmi M, et al. Risk and protective factors for mental disorders beyond genetics: an evidence-based atlas. World Psychiatry. 2021;20(3):417-436. doi: 10.1002/wps.20894 [DOI] [PMC free article] [PubMed] [Google Scholar]
4.Bauer ME, Teixeira AL. Inflammation in psychiatric disorders: what comes first? Ann N Y Acad Sci. 2019;1437(1):57-67. doi: 10.1111/nyas.13712 [DOI] [PubMed] [Google Scholar]
5.Miller AH. Beyond depression: the expanding role of inflammation in psychiatric disorders. World Psychiatry. 2020;19(1):108-109. doi: 10.1002/wps.20723 [DOI] [PMC free article] [PubMed] [Google Scholar]
6.Dantzer R, O’Connor JC, Freund GG, Johnson RW, Kelley KW. From inflammation to sickness and depression: when the immune system subjugates the brain. Nat Rev Neurosci. 2008;9(1):46-56. doi: 10.1038/nrn2297 [DOI] [PMC free article] [PubMed] [Google Scholar]
7.Capuron L, Miller AH. Immune system to brain signaling: neuropsychopharmacological implications. Pharmacol Ther. 2011;130(2):226-238. doi: 10.1016/j.pharmthera.2011.01.014 [DOI] [PMC free article] [PubMed] [Google Scholar]
8.Passos IC, Vasconcelos-Moreno MP, Costa LG, et al. Inflammatory markers in post-traumatic stress disorder: a systematic review, meta-analysis, and meta-regression. Lancet Psychiatry. 2015;2(11):1002-1012. doi: 10.1016/S2215-0366(15)00309-0 [DOI] [PubMed] [Google Scholar]
9.Peruzzolo TL, Pinto JV, Roza TH, et al. Inflammatory and oxidative stress markers in posttraumatic stress disorder: a systematic review and meta-analysis. Mol Psychiatry. 2022;27(8):3150-3163. doi: 10.1038/s41380-022-01564-0 [DOI] [PubMed] [Google Scholar]
10.Dowlati Y, Herrmann N, Swardfager W, et al. A meta-analysis of cytokines in major depression. Biol Psychiatry. 2010;67(5):446-457. doi: 10.1016/j.biopsych.2009.09.033 [DOI] [PubMed] [Google Scholar]
11.Milaneschi Y, Kappelmann N, Ye Z, et al. Association of inflammation with depression and anxiety: evidence for symptom-specificity and potential causality from UK Biobank and NESDA cohorts. Mol Psychiatry. 2021;26(12):7393-7402. doi: 10.1038/s41380-021-01188-w [DOI] [PMC free article] [PubMed] [Google Scholar]
12.Eraly SA, Nievergelt CM, Maihofer AX, et al. ; Marine Resiliency Study Team . Assessment of plasma C-reactive protein as a biomarker of posttraumatic stress disorder risk. JAMA Psychiatry. 2014;71(4):423-431. doi: 10.1001/jamapsychiatry.2013.4374 [DOI] [PMC free article] [PubMed] [Google Scholar]
13.Chu AL, Stochl J, Lewis G, Zammit S, Jones PB, Khandaker GM. Longitudinal association between inflammatory markers and specific symptoms of depression in a prospective birth cohort. Brain Behav Immun. 2019;76:74-81. doi: 10.1016/j.bbi.2018.11.007 [DOI] [PMC free article] [PubMed] [Google Scholar]
14.Sumner JA, Chen Q, Roberts AL, et al. Posttraumatic stress disorder onset and inflammatory and endothelial function biomarkers in women. Brain Behav Immun. 2018;69:203-209. doi: 10.1016/j.bbi.2017.11.013 [DOI] [PMC free article] [PubMed] [Google Scholar]
15.Ludvigsson JF, Appelros P, Askling J, et al. Adaptation of the Charlson Comorbidity Index for register-based research in Sweden. Clin Epidemiol. 2021;13:21-41. doi: 10.2147/CLEP.S282475 [DOI] [PMC free article] [PubMed] [Google Scholar]
16.Okada Y, Wu D, Trynka G, et al. ; RACI consortium; GARNET consortium . Genetics of rheumatoid arthritis contributes to biology and drug discovery. Nature. 2014;506(7488):376-381. doi: 10.1038/nature12873 [DOI] [PMC free article] [PubMed] [Google Scholar]
17.Meier SM, Trontti K, Purves KL, et al. Genetic variants associated with anxiety and stress-related disorders: a genome-wide association study and mouse-model study. JAMA Psychiatry. 2019;76(9):924-932. doi: 10.1001/jamapsychiatry.2019.1119 [DOI] [PMC free article] [PubMed] [Google Scholar]
18.Astle WJ, Elding H, Jiang T, et al. The allelic landscape of human blood cell trait variation and links to common complex disease. Cell. 2016;167(5):1415-1429.e19. doi: 10.1016/j.cell.2016.10.042 [DOI] [PMC free article] [PubMed] [Google Scholar]
19.Pietzner M, Wheeler E, Carrasco-Zanini J, et al. Genetic architecture of host proteins involved in SARS-CoV-2 infection. Nat Commun. 2020;11(1):6397. doi: 10.1038/s41467-020-19996-z [DOI] [PMC free article] [PubMed] [Google Scholar]
20.Mbatchou J, Barnard L, Backman J, et al. Computationally efficient whole-genome regression for quantitative and binary traits. Nat Genet. 2021;53(7):1097-1103. doi: 10.1038/s41588-021-00870-7 [DOI] [PubMed] [Google Scholar]
21.Said S, Pazoki R, Karhunen V, et al. Genetic analysis of over half a million people characterises C-reactive protein loci. Nat Commun. 2022;13(1):2198. doi: 10.1038/s41467-022-29650-5 [DOI] [PMC free article] [PubMed] [Google Scholar]
22.Funatogawa I, Funatogawa T. Longitudinal Data and Linear Mixed Effects Models. In: Funatogawa I, Funatogawa T, eds. Longitudinal Data Analysis: Autoregressive Linear Mixed Effects Models. Springer Singapore; 2018:1-26. doi: 10.1007/978-981-10-0077-5_1 [DOI] [Google Scholar]
23.Bulik-Sullivan BK, Loh PR, Finucane HK, et al. ; Schizophrenia Working Group of the Psychiatric Genomics Consortium . LD Score regression distinguishes confounding from polygenicity in genome-wide association studies. Nat Genet. 2015;47(3):291-295. doi: 10.1038/ng.3211 [DOI] [PMC free article] [PubMed] [Google Scholar]
24.Lu Q, Li B, Ou D, et al. A powerful approach to estimating annotation-stratified genetic covariance via GWAS summary statistics. Am J Hum Genet. 2017;101(6):939-964. doi: 10.1016/j.ajhg.2017.11.001 [DOI] [PMC free article] [PubMed] [Google Scholar]
25.Burgess S, Dudbridge F, Thompson SG. Combining information on multiple instrumental variables in mendelian randomization: comparison of allele score and summarized data methods. Stat Med. 2016;35(11):1880-1906. doi: 10.1002/sim.6835 [DOI] [PMC free article] [PubMed] [Google Scholar]
26.Bowden J, Davey Smith G, Burgess S. Mendelian randomization with invalid instruments: effect estimation and bias detection through Egger regression. Int J Epidemiol. 2015;44(2):512-525. doi: 10.1093/ije/dyv080 [DOI] [PMC free article] [PubMed] [Google Scholar]
27.Bowden J, Davey Smith G, Haycock PC, Burgess S. Consistent estimation in mendelian randomization with some invalid instruments using a weighted median estimator. Genet Epidemiol. 2016;40(4):304-314. doi: 10.1002/gepi.21965 [DOI] [PMC free article] [PubMed] [Google Scholar]
28.Cheng Q, Zhang X, Chen LS, Liu J. Mendelian randomization accounting for complex correlated horizontal pleiotropy while elucidating shared genetic etiology. Nat Commun. 2022;13(1):6490. doi: 10.1038/s41467-022-34164-1 [DOI] [PMC free article] [PubMed] [Google Scholar]
29.Hemani G, Tilling K, Davey Smith G. Orienting the causal relationship between imprecisely measured traits using GWAS summary data. PLoS Genet. 2017;13(11):e1007081. doi: 10.1371/journal.pgen.1007081 [DOI] [PMC free article] [PubMed] [Google Scholar]
30.Rengasamy M, Moriarity D, Kraynak T, Tervo-Clemmens B, Price R. Exploring the multiverse: the impact of researchers’ analytic decisions on relationships between depression and inflammatory markers. Neuropsychopharmacology. 2023;48(10):1465-1474. doi: 10.1038/s41386-023-01621-4 [DOI] [PMC free article] [PubMed] [Google Scholar]
31.Gialluisi A, Di Castelnuovo A, Bracone F, et al. Associations between systemic inflammation and somatic depressive symptoms: findings from the Moli-Sani study. Depress Anxiety. 2020;37(9):935-943. doi: 10.1002/da.23070 [DOI] [PubMed] [Google Scholar]
32.Sealock JM, Lee YH, Moscati A, et al. Use of the PsycheMERGE Network to investigate the association between depression polygenic scores and white blood cell count. JAMA Psychiatry. 2021;78(12):1365-1374. doi: 10.1001/jamapsychiatry.2021.2959 [DOI] [PMC free article] [PubMed] [Google Scholar]
33.Gabay C, Kushner I. Acute-phase proteins and other systemic responses to inflammation. N Engl J Med. 1999;340(6):448-454. doi: 10.1056/NEJM199902113400607 [DOI] [PubMed] [Google Scholar]
34.Lux A, Aschermann S, Biburger M, Nimmerjahn F. The pro and anti-inflammatory activities of immunoglobulin G. Ann Rheum Dis. 2010;69(suppl 1):i92-i96. doi: 10.1136/ard.2009.117101 [DOI] [PubMed] [Google Scholar]
35.Böhm S, Schwab I, Lux A, Nimmerjahn F. The role of sialic acid as a modulator of the anti-inflammatory activity of IgG. Semin Immunopathol. 2012;34(3):443-453. doi: 10.1007/s00281-012-0308-x [DOI] [PubMed] [Google Scholar]
36.Zeginiadou T, Kolias S, Kouretas D, Antonoglou O. Nonlinear binding of sex steroids to albumin and sex hormone binding globulin. Eur J Drug Metab Pharmacokinet. 1997;22(3):229-235. doi: 10.1007/BF03189812 [DOI] [PubMed] [Google Scholar]
37.Sikora M, Marczak Ł, Perła-Kajan J, Jakubowski H. Sex affects N-homocysteinylation at lysine residue 212 of albumin in mice. Sci Rep. 2019;9(1):2669. doi: 10.1038/s41598-019-38784-4 [DOI] [PMC free article] [PubMed] [Google Scholar]
38.de Jager CP, van Wijk PT, Mathoera RB, de Jongh-Leuvenink J, van der Poll T, Wever PC. Lymphocytopenia and neutrophil-lymphocyte count ratio predict bacteremia better than conventional infection markers in an emergency care unit. Crit Care. 2010;14(5):R192. doi: 10.1186/cc9309 [DOI] [PMC free article] [PubMed] [Google Scholar]
39.Dentali F, Nigro O, Squizzato A, et al. Impact of neutrophils to lymphocytes ratio on major clinical outcomes in patients with acute coronary syndromes: a systematic review and meta-analysis of the literature. Int J Cardiol. 2018;266:31-37. doi: 10.1016/j.ijcard.2018.02.116 [DOI] [PubMed] [Google Scholar]
40.Najjar S, Pearlman DM, Alper K, Najjar A, Devinsky O. Neuroinflammation and psychiatric illness. J Neuroinflammation. 2013;10:43. doi: 10.1186/1742-2094-10-43 [DOI] [PMC free article] [PubMed] [Google Scholar]
41.Huang X, Hussain B, Chang J. Peripheral inflammation and blood-brain barrier disruption: effects and mechanisms. CNS Neurosci Ther. 2021;27(1):36-47. doi: 10.1111/cns.13569 [DOI] [PMC free article] [PubMed] [Google Scholar]
42.Aguilar-Valles A, Kim J, Jung S, Woodside B, Luheshi GN. Role of brain transmigrating neutrophils in depression-like behavior during systemic infection. Mol Psychiatry. 2014;19(5):599-606. doi: 10.1038/mp.2013.137 [DOI] [PubMed] [Google Scholar]
43.Kim YR, Kim YM, Lee J, Park J, Lee JE, Hyun YM. Neutrophils return to bloodstream through the brain blood vessel after crosstalk with microglia during LPS-induced neuroinflammation. Front Cell Dev Biol. 2020;8:613733. doi: 10.3389/fcell.2020.613733 [DOI] [PMC free article] [PubMed] [Google Scholar]
44.Hashimoto K, Sawa A, Iyo M. Increased levels of glutamate in brains from patients with mood disorders. Biol Psychiatry. 2007;62(11):1310-1316. doi: 10.1016/j.biopsych.2007.03.017 [DOI] [PubMed] [Google Scholar]
45.Huntoon KM, Wang Y, Eppolito CA, et al. The acute phase protein haptoglobin regulates host immunity. J Leukoc Biol. 2008;84(1):170-181. doi: 10.1189/jlb.0208100 [DOI] [PMC free article] [PubMed] [Google Scholar]
46.Zhang C, Wei C, Huang X, et al. MPC-n (IgG) improves long-term cognitive impairment in the mouse model of repetitive mild traumatic brain injury. BMC Med. 2023;21(1):199. doi: 10.1186/s12916-023-02895-7 [DOI] [PMC free article] [PubMed] [Google Scholar]

Associated Data

This section collects any data citations, data availability statements, or supplementary materials included in this article.

Supplementary Materials