Association between inflammatory markers and cognitive function in adults with bipolar disorder: A systematic review

Abstract

Introduction:

Bipolar disorder (BD) is frequently associated with cognitive dysfunction, which can significantly impact the quality of life and functional recovery of affected individuals. Growing evidence suggests that inflammation may contribute to the cognitive dysfunction observed in BD.

Methods:

We conducted a systematic review following PRISMA guidelines, searching six databases on March 23, 2023 (PubMed, Embase, Cochrane Library, Web of Science, PsycINFO, and ClinicalTrials.gov), with the aim of identifying studies that examined the relationship between peripheral or central inflammatory markers and cognitive function in adults with BD. Studies involving animals, abstracts, protocols, reviews, and non-English publications were excluded. The quality of included studies was assessed using the Risk of Bias in Non-Randomized Studies – of Exposure (ROBINS-E). A narrative synthesis was completed, stratifying results based on the associations between inflammatory markers and cognitive domains in BD. The review protocol was pre-registered in PROSPERO (CRD42023415437).

Results:

Out of 2,680 identified records, 25 studies involving 3,567 adults with BD (mean age: 43.6 years; 1,839 females and 1,728 males) met the inclusion criteria. Seventeen studies were classified as low risk of bias, seven as having some concerns, and one as high risk. Elevated levels of C-reactive protein (CRP), tumor necrosis factor-alpha (TNF-α), interleukin-6 (IL-6), and interleukin-1 receptor antagonist (IL-1Ra) were most commonly associated with cognitive dysfunction in domains such as executive function, processing speed, and memory. Findings for other inflammatory markers were less consistent. Most studies relied on cross-sectional designs, which limit causal interpretations.

Conclusion:

This review found a consistent association between inflammation and cognitive dysfunction in BD, particularly involving CRP, TNF-α, IL-6, and IL-1RA in areas such as executive function, processing speed, and memory. Targeting inflammation may offer a promising approach to mitigating these cognitive challenges. Future studies should prioritize longitudinal designs, standardized cognitive assessments, and the exploration of central inflammatory markers to better understand the neurobiological processes underlying cognitive dysfunction in BD. These findings may help inform the development of adjunctive anti-inflammatory strategies to support cognitive health in individuals with BD.

1. Introduction

Bipolar disorder (BD) is a devastating and chronic psychiatric illness that affects approximately 40 million people worldwide [1]. The cognitive dysfunction observed in BD, particularly in areas such as processing speed, attention, explicit memory, and aspects of executive function [2], are similar to schizophrenia (SZ), though they tend to be less pronounced [3]. Importantly, this cognitive dysfunction persists even when patients are affectively stable [4–7], and are strongly associated with poor quality of life [8] and disability, which prevent complete functional recovery [3,9–12]. However, there is considerable heterogeneity in the cognitive profiles of individuals with BD [13]. Approximately 12%–40% exhibit global cognitive deficits across multiple domains (i.e., psychomotor speed, attention, verbal memory, and executive function), 29%–40% show selective impairment in attention and psychomotor speed, and 32%–48% are relatively cognitively intact compared to healthy age-matched controls [12,14–17].

Although research on the pattern of cognitive dysfunction in BD has grown over the past decade, the underlying pathophysiological mechanisms driving these observed cognitive dysfunctions in BD are not well understood. Multiple lines of evidence indicate alterations in immune functioning in BD [18–20]. Systematic reviews and meta-analyses have demonstrated that peripheral levels of pro-inflammatory markers are elevated in patients with BD during acute phases of the disease compared to healthy controls (HCs) [21,22]. Interestingly, cytokine levels do not appear to be elevated during periods of inter-episode stability early in the illness [23], but after multiple episodes, at least some patients show chronic low-grade inflammation even when euthymic [22], which overlaps with the trajectory of cognitive dysfunction observed in some patients with BD [24–29].

Previous systematic reviews have identified associations between elevated levels of pro-inflammatory cytokines, such as C-reactive protein (CRP), tumor necrosis factor-alpha (TNF-α), interleukin-6 (IL-6), and interleukin 1 receptor antagonist (IL-1Ra), and cognitive dysfunction in BD [19,20,30]. However, given the rapid advancements in research methodologies and the publication of updated guidelines from the International Society for Bipolar Disorders (ISBD) Task Force on Cognition [31], there is a need for an updated and comprehensive review of the literature. The last major reviews on this topic were published nearly a decade ago [19,20,30], and since then, significant progress has been made in understanding the neurobiological underlying cognitive dysfunction in BD..

2. Aim of the Study

This systematic review aimed to analyze the association between inflammatory markers and cognitive function in individuals with BD. It synthesizes recent evidence to clarify the potential role of inflammation in cognitive dysfunction. It also highlights emerging mechanisms and identifies directions for future research.

3. Materials & Methods

This systematic review was prepared following the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) 2020 guidelines (Supplemental Tables 1 and 2) [32].

3.1. Protocol registration

The protocol for this review was pre-registered in PROSPERO (registration number: CRD42023415437).

3.2. Data sources

Electronic searches for published literature were conducted by a medical librarian (M.L.) using Medline (PubMed), Embase.com (Elsevier), Core Collection (Web of Science), Cochrane Central Register of Controlled Trials (Ovid), PsycINFO (Ovid) and ClinicalTrials.gov on March 23, 2023. Grey literature was not included in this review.

3.3. Search strategy

The search strategy incorporated controlled vocabulary and free-text synonyms for the concepts of BD, inflammatory markers, and cognition. The full search strategy is documented in Supplemental Tables 3–7. No restrictions on language or any other search filters were applied. All identified studies were combined and de-duplicated in a single reference manager [33]. The citations were then uploaded into Rayyan software [34].

3.4. Eligibility criteria

The criteria for the included studies were as follows: (1) studies on patients with BD aged ≥18 years old with a clinical diagnosis of BD, including type I (characterized by the presence of manic episodes), type II (characterized by hypomanic episodes), or unspecified BD, regardless of sex/gender or psychiatric comorbidities; (2) measured inflammatory markers either in the periphery or in the central nervous system; (3) measured cognitive function with objective testing; (4) reported an association between inflammatory markers and cognitive performance in patients with BD; (5) any study design; (6) any year of publication. The exclusion criteria were as follows: (1) non-peer-reviewed articles, abstracts, protocols, reviews, animal studies, and articles published in languages other than English; (2) studies focused on genes or polymorphisms; (3) studies involving interventions, such as pharmacotherapy; (4) studies where information specific to BD could not be extracted due to the inclusion of mixed populations with other psychiatric diagnoses.

3.5. Selection process and data extraction

Four investigators (D.R.A.C., J.N.S.B., M.M., and W.F.V.) screened the titles, abstracts, and then full texts in pairs to select studies that met the inclusion criteria. Any discrepancies or lack of consensus on eligibility criteria were then reviewed, discussed, and resolved by all reviewers. Two reviewers (D.R.A.C. and M.D.L.) independently conducted data extraction using a standardized data extraction spreadsheet. The extraction process was designed to capture a comprehensive range of data from the studies, including the following information: first author; year of publication; country; study design; study population; age with mean and standard deviation (SD); sex; duration of BD; symptoms of BD; inflammatory biomarker assessment and types; inflammatory biomarker findings; cognitive assessment and domains; cognitive assessment findings; covariates analyzed; and findings regarding association between inflammation and cognition in BD. In cases where disagreements occurred among reviewers, a consensus was reached through a detailed discussion of the study among all reviewers.

3.6. Quality assessment

To enhance the credibility of our findings, a comprehensive risk of bias assessment was performed for the included studies. Two reviewers (D.R.A.C. and J.N.S.B.) independently assessed the risk of bias using the Risk of Bias in Non-Randomized Studies – of Exposure (ROBINS-E) [35]. This Cochrane tool evaluates observational studies based on various criteria, including (1) bias due to confounding; (2) bias arising from measurement of the exposure; (3) bias in selection of participants into the study (or into the analysis); (4) bias due to post-exposure interventions; (5) bias due to missing data; (6) bias arising from measurement of the outcome; (7) and bias in selection of the reported result. Studies were categorized based on their risk of bias into three distinct levels: (1) low risk of bias if all domains were judged as having low risk; (2) some concerns of bias if one or more domains were judged as having some concerns and no domains were judged as having high risk; or (3) high risk of bias if one or more domains were judged as having high risk. Any disagreements in categorization between the reviewers were resolved through discussion and consensus.

3.7. Synthesis methods

A narrative synthesis was completed stratifying the results based on the specific inflammatory biomarkers measured (such as CRP, TNF-α and its soluble receptors, IL-6, IL-1Ra, and other biomarkers), and the findings related to cognitive function in BD are presented. While the synthesis focuses on associations between inflammation and cognitive performance within the BD group, we also report biomarker findings for BD and HC groups, as well as cognitive performance outcomes, when available, to provide contextual background and facilitate interpretation of the association results. A meta-analysis was not feasible due to the high level of heterogeneity regarding the assessment of cognitive function across the included studies.

4. Results

4.1. Study selection

A total of 2,680 records were initially retrieved, and 963 duplicates were removed. Four reviewers in pairs (D.R.A.C., J.N.S., M.M., and W.F.V.) screened 1,717 records by title and abstract, and 1,632 records were excluded for not meeting the inclusion criteria. A total of 85 reports were assessed for full-text eligibility, and 60 reports were excluded in this step for specific reasons, as detailed in the PRISMA flowchart (Figure 1). Following the screening process, 25 reports representing 25 independent studies were included in the review [26–29,36–56]. The list of included articles is depicted in Supplemental Table 8, and the list of excluded articles with reasons is shown in Supplemental Table 9.

Figure. 1.

PRISMA flow diagram.

4.2. Overview of the included studies

A total of 65,554 participants (3,567 BD/61,987 HCs) were enrolled across the 25 included studies. Among the BD group, the mean age was 43.6 years, with 1839 females and 1728 males. Among the HC group, the mean age was 57.6 years, with 30,294 females and 31,693 males. The most frequently assessed inflammatory biomarker was TNF-α, along with its soluble receptors, soluble tumor necrosis factor receptor 1 (sTNFR1) and soluble tumor necrosis factor receptor 2 (sTNFR2). Executive function was the most commonly evaluated cognitive domain, with the Wisconsin Card Sorting Test (WCST) being the most frequently used cognitive assessment. The geographic distribution of the studies was as follows: Taiwan (k=5), Brazil (k=4), United States (k=4), Spain (k=3), Turkey (k=2), United Kingdom (k=2), Egypt (k=1), France (k=1), Italy (k=1), Norway (k=1), and Sweden (k=1). Table 1 is organized alphabetically and provides a detailed summary of the included studies, focusing specifically on associations between inflammatory markers and cognitive performance within the BD group. Figure 2 provides a visual synthesis of these associations across biomarkers and studies.

Table 1.

Characteristics of included studies.

Author (Year)	Country	Study Design	Study Population	Age Mean (SD)	Sex N (F/M)	Duration of BD Mean (SD)	BD Symptoms Mean (SD)	Inflammatory Biomarker Assessment and Types	Inflammatory Biomarker Findings Mean (SD)	Cognitive Assessment and Domains	Cognitive Performance Findings Mean (SD)	Covariates Analyzed	Association between Inflammation and Cognition in BD
Arslan et al. (2017) [36]	Turkey	Cross-sectional	BD I euthymia: 36 HC: 38	BD: 38 (11.8) HC: 37.9 (12)	BD: 22 F 14 M HC: 20 F 18 M	Age onset: 25.3 (8)	HAM-D: 2.36 (2.05) YMRS: 0.33 (0.89)	ELISA [Serum]: IL-6 IL-18	No significant differences between BD and HC groups	WCST: Conceptualization, attention, perseveration, working memory, executive function, abstract thinking, and processing Stroop Test BSRG: Ability to focus and sustain attention RAVLT: Learning, recognition, instant recall, and delayed recall	No significant differences for WCST between BD and HC groups Stroop interference: BD 21.3 (11) HC 15.1 (10.2) p=0.014 RAVLT (leaning): BD 16.3 (4.4) HC 19 (3.8) p=0.006 RAVLT (recognition): BD 11.8 (3.2) HC 13.4 (2.4) p=0.007	BMI, HAM-D	No significant correlations between IL-6 and cognition IL-18 [WCST]: Positive correlation with number of categories achieved (r=0.288, p=0.05) IL-18 [WCST]: Negative correlations with perseverative responses (r=−0.264, p=0.05) and errors (r=−0.279, p=0.01) IL-18 [Stroop interference]: Negative correlation (r=−0.235, p<0.05) IL-18 [RAVLT]: Positive correlations with instant recall (r=0.307, p=0.01), delayed recall (r=0.321, p=0.01), and learning scores (r=0.297, p=0.05)
Barbosa et al. (2012) [37]	Brazil	Cross-sectional	BD I euthymia: 25 HC: 25	BD: 50.9 (9.1) HC: 48 (7.1)	BD: 17 F 8 M HC: 14 F 11 M	27.8 (11.8)	17-item HDRS: 1.52 (1.64) YMRS: 1.08 (1.53)	ELISA [Plasma]: BDNF TNF-α sTNFR1 sTNFR2	BDNF levels significantly higher in BD compared to HC (p=0.001) No significant differences for TNF-α, sTNFR1, or sTNFR2 between BD and HC groups	30-item MMSE screener with questions that assess orientation, attention, memory, and language FAB: Assesses executive functioning and consists of 6 subscales-conceptualization, mental flexibility, motor programming, sensitivity of interference, inhibitory control, and environmental autonomy	No significant group differences for MMSE between BD and HC groups FAB (sensitivity of interference): BD 2.4 (1) HC 2.84 (0.6) p=0.02 FAB (inhibitory control): BD 1 (1) HC 1.76 (1.2) p=0.02	Spearman’s correlation analyses	No significant correlations between BDNF and cognition TNF-α [FAB]: Positive correlation with inhibitory control (ρ=0.50, p=0.02) No significant correlations between sTNFR1 or sTNFR2 and cognition
Barbosa et al. (2018) [38]	Brazil	Cross-sectional	Remitted BD I: 20 HC: 25	BD: 43.8 (10.9) HC: 43.5 (11.7)	BD: 15 F 5 M HC: 16 F 9 M	20.2 (12.1)	Remission defined as 17-item HDRS & YMRS total scores ≤ 7 points for ≥ 8 weeks	Flow cytometry [Plasma]: IL-2 IL-4 IL-6 IL-10 IL-17A TNF-α IFN-γ ELISA [Plasma]: sTNFR1 sTNFR2	Significant differences for all inflammatory biomarkers with patients with BD having higher plasma levels than HC	30-item MMSE screener with questions that assess orientation, attention, memory, and language BAC-A: 8 tasks that assess affective interference, verbal memory, working memory, motor speed, semantic fluency, letter fluency, executive functions, and attention and motor speed	No significant differences for MMSE between BD and HC groups Significant group differences for all tasks of BAC-A with patients with BD showing poorer performance than HC	Dura tion of disease, Years of study, MMSE, IL6 plasma levels	IL-6 [BAC-A]: Negative correlation with BAC-A total z-score (rho=−0.461, p=0.047) No significant correlations between other inflammatory biomarkers and cognition
Chou et al. (2012) [39]	Taiwan	Cross-sectional	BD I euthymia: 23 HC: 33	BD: 36.5 (8.9) HC: 37.6 (7.8)	BD: 17 F 6 M HC: 21 F 12 M	5.7 (4.7)	MADRS: 6.1 (1.8) YMRS: 3.1 (0.9)	ELISA [Plasma]: BDNF	No significant differences between BD and HC groups	GO/NO-GO: Attention TAP: Attention WMS-III: Memory CTT: Executive function WCST: Executive function	Faces memory (faces 1 recognition): BD 33.09 (4.5) HC 36.24 (4.2) p=0.01 Faces memory (faces 1 true positive): BD 13.86 (5) HC 18.52 (3.6) p=0.00 WCST (trails administered): BD 110.35 (22.4) HC 90.9 (20.4) p=0.00 WCST (failure to maintain set): BD 1.43 (1.4) HC 0.45 (0.7) p=0.00 No significant differences for other tasks	Age, Gender, Education	No significant correlations
Congio et al. (2022) [40]	Brazil	Case-control	BD: 42 (BD I: 31 BD II: 11) MDD: 27 HC: 40	BD: 39.7 (11.3) HC: 44 (12)	BD: 33 F 9 M HC: 23 F 17 M	BD: 20.3 (12.5)	17-item HDRS: 10.67 (7.93) YMRS: 3.71 (5.10)	Immunonephelometry [Serum]: hs-CRP	No significant differences between BD and HC groups	Phonetic verbal fluency: Language processing Semantic verbal fluency: Language and semantic memory TMT-A: Processing speed TMT-B: Mental flexibility Stroop Test: Selective attention	Significant differences for all tasks between BD to HC, with patients with BD performing worse	Not reported	Phonetic verbal fluency: BD with hs-CRP ≥ 5 mg/L: 8.64 (4) HC with hs-CRP < 5 mg/L: 13.14 (4.4) p<0.01 Semantic verbal fluency: BD with hs-CRP ≥ 5 mg/L: 12.27 (1.8) HC with hs-CRP < 5 mg/L: 16.63 (5.1) p=0.04 TMT-A (seconds): BD with hs-CRP ≥ 5 mg/L: 71.22 (36.5) HC with hs-CRP < 5 mg/L: 40.73 (24.5) p=0.03 No significant difference on Stroop performance among the groups
Dickerson et al. (2013) [41]	USA	Cross-sectional	BD I most recent manic episode: 29 BD I most recent depressive episode: 41 BD I most recent episode mixed: 23 BD II: 14	36.3 (13.4)	76 F 31 M	Age onset: 16.1 (9)	HDRS: 22.3 (10.7) YMRS: 14.0 (9.2)	Enzyme immunoassay [Serum]: hs-CRP	No HC group	RBANS, Form A: Immediate memory, visuospatial/ constructional, language, attention, and delayed memory WAIS-III (Information and Letter Number Sequencing) and TMT-A: Memory	No HC group	Age, Gender, Race, Maternal education, Cigarette smoking status, BMI, HAM-D, YMRS, Seropositivity to HSV-1	RBANS: Negative correlation with total score (t=−2.48, p=0.015), immediate memory (t=−2.16, p=0.033), attention (t=−2.18, p=0.032), and language (t=−2.13, p=0.036) TMT-A: Negative correlation (t=−2.39, p=0.019)
Doganavsargil-Baysal et al. (2013) [26]	Turkey	Case-control	BD I euthymia: 54 HC: 18	BD: 39.5 (11.6) HC: 38.3 (10.8)	BD: 36 F 18 M HC: 13 F 5 M	Age onset: 29.9 (10.6)	Euthymia defined as HDRS total score ≤ 7 and YMRS total score ≤ 5	ELISA [Serum]: TNF-α sTNFR1 sTNFR2	No significant differences for TNF-α between BD and HC groups sTNFR1 BD 1.35 (0.9) HC 0.86 (0.6) p=0.029 sTNFR2 BD 9.17 (4.5) HC 6.40 (3.3) p=0.014	WCST: Executive and visual-motor functions RAVLT: Verbal learning and memory	WCST (number of trial): BD 126.48 (5.1) HC 114.78 (21.1) p=0.008 WCST (total errors): BD 62.44 (17.6) HC 47.22 (23.5) P=0.028 WCST (number of category achieved): BD 2.42 (1.7) HC 3.88 (1.8) p=0.005 WCST (percentage of conceptual level responses): BD 33.76 (16.4) HC 50.49 (21.5) p=0.006 RAVLT (maximum learning score): BD 14.35 (1.1) HC 14.88 (0.5) p=0.032 RAVLT (learning error score): BD 1.11 (1.2) HC 0.33 (0.5) p=0.021	Spearman’s correlation analyses	TNF-α [RAVLT]: Negative correlation with delayed recall score (rho=−0.275, p=0.044) No significant correlations between sTNFR1 or sTNFR2 and cognition
Garés-Caballer et al. (2022) [42]	Spain	Prospective1-year follow-up cohort	BD I: 42 SZ: 30 MDD: 35 T2DM: 30 HC: 28	BD: 50 (9.5) HC: 36.6 (14.5)	BD: 21 F 21 M HC: 18 F 10 M	23.4 (11.5)	HDRS: 6.4 (4.4) YMRS: 3.5 (4.5) Patients diagnosed with MDD and BD had to be clinically stable without presenting an acute affective episode	Flow cytometry [Serum]: IL-6 IL-10 TNF-α Immunonephelometry [Serum]: CRP-us	No significant differences between BD and HC groups	SCWT: Cognitive flexibility and processing speed Color/Word Subtest: Cognitive flexibility WCST: Cognitive flexibility Verbal Fluency Tasks Semantic and Phonemic Forms: Verbal fluency TMT: Working memory and processing speed WMS-III: Working memory and processing speed Finger Tapping Test: Processing speed	BD and SZ had significantly worse executive functions compared to the other groups	Age, Sex, Years of education, BMI	In the BD group, 26.4–49.8% of improvement in executive functioning was correlated with a decrease in the levels of IL-6 and CRP No significant correlations between other inflammatory biomarkers and cognition
Hamdani et al. (2015) [28]	France	Cross-sectional	BD euthymia (type I and II): 42 HC: 36	BD: 46.1 (13) HC: 38.1 (15.2)	BD: 20 F 22 M HC: 24 F 12 M	18.2 (13.1)	MADRS: 9.03 (1.07) YMRS: 4.73 (6.6)	Real-Time PCR [Peripheral blood mononuclear cells]: IL-6 mRNA expression Immunoassay [Serum]: IgM and IgG class antibodies to T. gondii	IL-6 mRNA expression: BD 6.42 (6.5) HC 3.38 (3.6) p=0.015 T. gondii seropositivity: BD 88.1% HC 52.7% p=0.000	CVLT: Episodic verbal memory WAIS-III: Working memory, verbal ability, and cognitive deterioration index (calculated from WAIS subscales)	WAIS-III (speed of processing): BD 7.66 (3.4) HC 10.14 (2.9) p=0.001 WAIS-III (verbal learning): BD 47.35 (11.1) HC 53.08 (7.5) p=0.011 WAIS-III (deterioration index): BD 0.23 (0.3) HC −0.38 (0.7) p=0.000	Age, Gender, Tobacco use	In the BD group infected by T. gondii, cognitive deterioration index was positively correlated to IL-6 mRNA expression (rho=0.43, p=0.01) Among deteriorated patients (scores above 0.10 according to the Weschler’s definition), IL-6 mRNA expression was twice as high (deteriorated: mean 10.17 SD [2.9], not deteriorated: mean 4.25 SD [1], p=0.01)
Hassan et al. (2023) [43]	Egypt	Cross-sectional	BD I euthymia: 30 BD I depression: 24 BD I mania: 29 HC: 20	BD I euthymia: 34.7 (8.9) BD I depression: 39 (10) BD I mania: 34.4 (9.8) HC: 38.6 (8.3)	BD I euthymia: 10 F 20 M BD I depression: 13 F 11 M BD I mania: 9 F 20 M HC: 8 F 12 M	BD I euthymia: median 13 (range: 0.25–35) BD I depression: median 10 (range: 1–25) BD I mania: median 10 (range: 0.25–37)	MADRS: median (range) BD I euthymia: 0 (0–7) BD I depression: 12 (9–25) BD I mania: 0 (0–7) YMRS: median (range) BD I euthymia: 4 (0–12) BD I depression: 0 (0–9) BD I mania: 17 (13–34)	ELISA [Serum]: IL-6	In the BD group (all phases), IL-6 levels were significantly higher compared to HC	MoCA: Short term and working memory, visuospatial and executive function, attention span and concentration, language, and orientation	Not reported separately	Current manic episode, Current depressive episode, Cur rent psychotics features, Current hospitalization, YMRS	No significant correlations
Hope et al. (2015) [29]	Norway	Cross-sectional	BD (type I, II, and not otherwise specified): 111 SZ: 121 HC: 241	BD: 33 (10) HC: 36 (12)	BD: 51 F 60 M HC: 147 F 94 M	4.6 (6)	IDS: 16 (11) YMRS: 5.5 (5)	Enzyme immunoassay [Serum]: sTNFR1 IL-1Ra OPG IL-6 sCD40L hsCRP	BD group had higher levels of sTNFR1 and lower levels of OPG compared to HC No significant differences for IL-1Ra, IL-6, sCD40L, or hsCRP between BD and HC groups	WASI: General cognitive abilities	Not reported	Age, Sex, Diagnostic group	In the BD group, there were negative correlations between sCD40L (β=−0.10, p=0.03) and IL-1Ra (β=−0.10, p=0.03) with general cognitive abilities No significant correlations between other inflammatory biomarkers and cognition
Hua et al. (2021) [44]	Taiwan	Cross-sectional	Remitted BD I: 58 Remitted BD II: 27 HC: 51	BD I: 38.3 (9.1) BD II: 37.3 (11.9) HC: 36.1 (9.5)	BD I: 34 F 24 M BD II: 19 F 8 M HC: 31 F 20 M	BD I: 13.8 (9.5) BD II: 11 (8.1)	BD I MADRS: 2.66 (2.78) BD II MADRS: 2.81 (2.82) BD I YMRS: 2.28 (2.32) BD II YMRS: 1.44 (2.15)	ELISA [Serum]: sIL-6R sTNFR1 CRP	In the BD I group, there were significant increases in all inflammatory biomarkers compared to HC No significant differences between the BD II group and HC	WLMT: Immediate/working memory, and recognition function WCST: Executive function	BD I and BD II had significantly worse performance compared to HC in several tasks of WLMT and all tasks of WCST	Age, Sex, Education, BMI, MADRS, YMRS	sTNFR1 [WLMT]: Negative correlation with word list I recall (β=−0.004, p<0.001) and word list II recall (β=−0.002, p<0.001) sTNFR1 [WCST]: Negative correlation with percent conceptual level responses (β=−0.009, p=0.012) and number of categories completed (β=−0.001, p=0.048) No significant correlations for sIL-6R or CRP
Kai-Lin Huang et al. (2022) [45]	Taiwan	Cross-sectional	Recurrent BD I: 38 First-episode BD I: 31 HC: 43	Recurrent BD: 28.6 (3.8) First-episode BD: 25.3 (5.4) HC: 26.2 (4)	Recurrent BD: 19 F 19 M First-episode BD: 24 F 7 M HC: 27 F 16 M	Recurrent BD age onset: 20.1 (5.1) First-episode BD age onset: 25.1 (5.4)	Recurrent BD I MADRS: 8.13 (7.03) First-episode BD I MADRS: 13.39 (8.62) Recurrent BD I YMRS: 2.97 (3.69) First-episode BD I YMRS: 7.45 (7.63)	ELISA [Serum]: IL-6 TNF-α CRP	No significant differences for IL-6 between BD and HC groups TNF-α levels were significantly higher in the recurrent BD group compared to HC p=0.01 Trends towards CRP levels being higher in the first-episode BD group compared to HC p=0.054	WCST: Executive function	WCST (percent of perseverative errors): Recurrent BD had a significantly higher percent compared to first-episode BD and HC (p=0.046) WCST (percent of perseverative responses): Recurrent BD had a significantly higher percent compared to first-episode BD and HC (p=0.044)	Age, Sex, Education, BMI, MADRS, YMRS	No significant correlations
Liou et al. (2023) [46]	Taiwan	Cross-sectional	BD: 641 BD + AUD: 150 HC: 185	BD: 32.9 (12.4) BD + AUD: 38.3 (10.5) HC: 32.2 (8.9)	BD: 326 F 315 M BD + AUD: 34 F 116 M HC: 81 F 104 M	Age onset: 16.3 (5.3)	BD 17-item HDRS: 16.6 (5.9) BD + AUD 17-item HDRS: 16.2 (7.2) BD YMRS: 11.3 (4.6) BD + AUD YMRS: 11.7 (6.0)	ELISA [Plasma]: IL-8 TNF-α CRP TGF-β BDNF	IL-8: BD + AUD 12.99 (27) BD 3 (3.3) HC 2.39 (3.8) p<0.001 TNF-α: BD + AUD 3.47 (2.5) BD 1.8 (1.7) HC 1.57 (1.7) p<0.001 BDNF: BD + AUD 13.97 (8.7) BD 15.19 (9.5) HC 17.48 (9.4) p=0.003 No significant differences for CRP or TGF- β between BD and HC groups	WCST: Executive function CPT: Sustained attention and vigilance WMS-III: Memory	BD and BD + AUD had significantly worse performance compared to HC in almost all the tests of WCST, CPT, and WMS-III	Age, Sex, Education, HDRS, YMRS	IL-8 [WCST]: Negative correlation with number of completed categories in patients with BD and BD + AUD (r=−0.13, p=0.02) TNF-α [WMS-III]: Negative correlation with visual immediate index in patients with BD and BD + AUD (r=−0.10, p=0.05) No significant correlations between other inflammatory biomarkers and cognition
Lotrich et al. (2014) [27]	USA	Cross-sectional	BD euthymia: 21 HC: 26	BD: 64.8 (9.1) HC: 65.5 (8.4)	BD: 13 F 8 M HC: 14 F 12 M	Not reported	17-item HDRS: 4.48 (2.5) YMRS: 2.62 (2.0)	Enzyme immunoassay [Serum]: IL-1Ra IL-6 IL-10 TNF-α BDNF	IL-1Ra: BD 439 (326) HC 269 (109) p=0.004 No significant differences for IL-6, IL-10, TNF-α, or BDNF between BD and HC groups	Battery of 21 neuropsychological tests (Gildengers et al. 2012): Language, delayed memory, visuomotor ability, and information processing speed/executive function	BD had significantly worse performance in all the tests compared to HC	Age, Gender, Education, BMI, BD diagnosis, BDNF, IL-6	IL-1Ra: Negative correlation with global cognition (r=−0.372, p=0.01), visual cognition (r=−0.312, p=0.03), memory cognition (r=−0.345, p=0.02), and speed/executive cognition (r=−0.404, p=0.005) No significant correlations between other inflammatory biomarkers and cognition
Mao-Hsuan Huang et al. (2022) [47]	Taiwan	Cross-sectional	BD I: 37 BD II: 33 MDD: 25 HC: 54	BD I: 34.9 (8.2) BD II: 31.5 (8.5) HC: 32.1 (7.7)	BD I: 19 F 18 M BD II: 20 F 13 M HC: 34 F 20 M	Not reported	BD I MADRS: 5.3 (5.0) BD II MADRS: 7.8 (5.1) BD I YMRS: 4.1 (5.1) BD II YMRS: 4.5 (4.5)	ELISA [Serum]: sIL-6R sTNF-αR1 CRP	sTNF-αR1: BD I 3.042 (0.1) BD II 2.978 (0.1) HC 2.926 (0.1) p<0.001 No significant differences for sIL-6R or CRP between BD and HC groups	WLMT: Learning and memory WCST: Executive function TAP: Attention and working memory GO/NO-GO: Sustained attention and inhibitory control	BD I had significantly worse performance compared to BD II and HC in all tests	Age, Sex, BMI, Education, Psychotropic medication use, MADRS, YMRS	No significant correlations for sIL-6R sTNF-αR1 [WLMT]: Negative correlation with word list II recall (β=−5.97, p=0.004) sTNF-αR1 [WLMT]: Negative correlation with word list retention (β=−30.851, p=0.037) sTNF-αR1 [TAP]: Negative correlation with auditory divided attention (β=−3.774, p=0.009) CRP [TAP]: Negative correlation with auditory divided attention (β=−0.732, p=0.006)
Millett et al. (2020) [48]	USA	Cross-sectional	BD euthymia: 219 HC: 52	BD: 43.6 (11.9) HC: 38.9 (13.4)	BD: 103 F 116 M HC: 30 F 22 M	21.9 (11.5)	Affective stability was defined as HDRS and/or YMRS ≤ 8	ECL [Serum]: TNF-α sTNF-R1 sTNF-R2	sTNF-R1: Late-stage BD (defined by more than 1 psychiatric hospitalization) 4440 (2967.1) Early-stage BD 3787.5 (1086.1) p=0.047 No significant differences for TNF-α or sTNF-R2 between BD and HC groups	COWAT: Verbal fluency SCWT: Selective attention, cognitive flexibility, processing speed, and executive functioning WCST: Conceptualization, attention, perseveration, working memory, executive functions, abstract thinking, and processing MCCB: Speed of processing, attention and vigilance, working memory, verbal learning, visual learning, reasoning and problem-solving, and social cognition	Not reported separately	Age, Sex, Race, Edu cation, Premorbid IQ	TNF-α [Stroop]: Negative correlation with executive function (r=−0.2, p=0.005) TNF-α [WCST]: Negative correlation with executive function (r=−0.2, p=0.03) sTNF-R2 [Stroop]: Negative correlation with executive function (r=−0.2, p=0.005) No significant correlations for sTNF-R1
Millett et al. (2021) [49]	USA	Cross-sectional	BD I euthymia: 179 BD II euthymia: 43 HC: 52	BD: 44 (11.9) HC: 39 (13.4)	BD: 105 F 117 M HC: 30 F 22 M	22 (11.5)	Affective stability was defined as HDRS and/or YMRS ≤ 8	ECL [Serum]: CRP	No significant differences between BD and HC groups	MCCB: Speed of processing, attention and vigilance, working memory, visual learning, reasoning and problem-solving, and social cognition CVLT: Verbal learning test SCWT: Executive function COWAT: Executive function WCST: Executive function Reading Mind in the Eyes task: Theory of mind	MCCB (composite score): BD 42.7 (7.7) HC 47.3 (7.8) p=0.01	Age, Sex, Race, Education, Number of psychiatric hospitalizations, CRP group	In the analysis comparing the entire sample of BD + HC with high CRP (≥ 5 mg/L) versus those with normal CRP (< 5 mg/dL), there was a significant effect of the high CRP group on the WCST (F(1)=6.98, p=0.009), Stroop (F(1)=4.46, p=0.036), Reading Mind in the Eyes (F(1)=19.9, p<0.001), MCCB speed of processing (F(1)=6.4, p=0.012), and MCCB reasoning and problem solving (F(1)=6.8, p=0.01) There was no main effect of diagnosis or interaction between diagnosis and CRP group in the model Within the BD I group, a significant association was observed between CRP levels and the MATRICS composite score (R = −0.199, p = 0.008), an effect notably stronger than that in the BD II group (R = 0.04, p = 0.83)
Milton et al. (2021) [50]	United Kingdom	Cross-sectional	BD I: 502 BD II: 467 MDD: 22441 HC: 60858	BD I: 55 (8) BD II: 56 (8) HC: 58 (8)	BD I: 233 F 269 M BD II: 208 F 259 M HC: 29657 F 31201 M	Not reported	Not reported	Immunoturbidimetric high-sensitivity analysis [Serum]: CRP	CRP levels were significantly higher in the BD I group compared to HC p=0.017	Tasks designed for the UK Biobank that assess visuospatial memory through reaction time and pairs matching	Not reported separately	Age, Sex, Education, Townsend deprivation score, Current depressive symptoms	CRP [Pair matching]: Negative relationship for BD I (β=−0.162, p=0.001)
Mora et al. (2019) [51]	Spain	Cross-sectional	BD euthymia: 52 BD mania: 32 HC: 49	BD euthymia: 47.5 (11.9) BD mania: 41.3 (12.9) HC: 48.3 (12.1)	BD euthymia: 26 F 26 M BD mania: 14 F 18 M HC: 28 F 21 M	BD euthymia: 22.9 (12.4) BD mania: 11.9 (11.2)	BD euthymia 17-item HAM-D: 2.4 (2.3) BD mania 17-item HAM-D: 8.3 (3.7) BD euthymia YMRS: 1.4 (1.7) BD mania YMRS: 31.3 (5.9)	ELISA [Serum]: IL-6 IL-10 TNF-α BDNF	IL-6: BD mania 1.23 (0.7) HC 0.86 (0.5) p=0.019 BDNF: BD euthymia 40 (9.9) BD mania 35.05 (10.6) HC 45.86 (13.6) p=0.039 (euthymia vs. HC) and p<0.001 (mania vs. HC) No significant differences for IL-10 or TNF-α between BD and HC groups	WMS-III: Memory WCST: Executive function and perseverative behavior SCWT: Selective attention and inhibition capacity FAS: Executive function TMT A and B: Processing speed and cognitive flexibility CPT-II: Sustained attention CVLT: Verbal learning, recall, and recognition RCFT: Visual memory	BD euthymia and BD mania had significantly worse performance in almost all tests compared to HC, with the exception of WAIS-III digit span forward, which did not show significant difference between groups	Age, Premorbid IQ, BMI, Duration of illness, Number of depressive episodes, Number of manic episodes, Number of hospitalizations	No significant correlations for IL-6, IL-10, or TNF-α BDNF [WCST]: Positive correlation with executive functioning (β=0.01, p=0.02) BDNF [CVLT]: Positive correlation with verbal memory (β=0.013, p=0.005)
Poletti et al. (2021) [52]	Italy	Cross-sectional	BD with a depressive episode in course: 76	47.1 (11.3)	53 F 23 M	19.2 (11.5)	21-item HDRS: 20.07 (5.72)	Pro Human Cytokine 27-Plex Immunoassay panel (Bioplex) [Plasma]: IL-1β, IL-1Ra, IL-2, IL-4, IL-5, IL-6, IL-7, IL-8, IL-9, IL-10, IL-12, IL-13, IL-15, IL-17A, IP-10 (CXCL10), MCP-1 (CCL2), MIP-1α (CCL3), MIP-1β (CCL4), RANTES (CCL5), Eotaxin (CCL11), TNF-α, IFN-γ, bFGF, PDGF-BB, G-CSF, GM-CSF, VEGF	No HC group	Brief Assessment of Cognition in Schizophrenia: Verbal memory, working memory, psychomotor speed and coordination, selective attention, semantic fluency, letter fluency, and executive functions	No HC group	Age, Sex, Education, Frequency of mood episodes, BMI, Imipra mine and chlorpromazine equivalent doses, Lithium, HDRS	Higher levels of IL-1β, IL-6, CCL2, CCL4, CCL5, CXCL10, and bFGF were associated with an increased likelihood of having poor cognitive performance overall Higher levels of IL-17 were associated with an increased likelihood of having good performance in verbal fluency No significant correlations between other inflammatory biomarkers and cognition
Rolstad et al. (2015) [53]	Sweden	Cross-sectional	BD euthymia: 78 (BD I: 48 BD II: 30) HC: 86	BD: 38.2 (13.4) HC: 37.8 (13.3)	BD: 47 F 31 M HC: 40 F 46 M	Not reported	MADRS: 4.4 (6.1) YMRS: 1.3 (1.9)	ELISA [Cerebrospinal fluid]: YKL-40 TIMP1 MCP1 sCD14	YKL-40: BD 82610.76 (47643.6) HC 65383.58 (47351.9) p=0.02 TIMP1: BD 34.40 (7.0) HC 32.13 (6.9) p=0.03 MCP-1: BD 528.7 (152.8) HC 455.7 (119.0) p<0.0005 sCD14: BD 52834.09 (24001.6) HC 43057.86 (28138.7) p=0.01	WAIS-III: Speed and attention, learning and memory, and visuospatial and verbal functions D-KEFS: Speed and attention, and verbal and executive functions	Memory functions: BD −0.25 (0.48) HC 0.01 (0.41) p<0.0005 Executive functions: BD −0.67 (0.72) HC −0.35 (0.57) p=0.001 Visuospatial functions: BD −0.35 (1.05) HC 0.02 (0.74) p=0.01 Speed/attention: BD −0.71 (1.27) HC −0.01 (0.90) p<0.0005 Verbal functions: BD −0.45 (0.81) HC 0.02 (0.61) p<0.0005	Age, Bipolar subtype, CGI, MADRS, YMRS, GAF, Medication use	YKL-40 [Executive performance]: Negative correlation (β=−0.99, p<0.0005) No significant correlations between other inflammatory biomarkers and cognition
Sanchez-Autet et al. (2018) [54]	Spain	Prospective 3-year follow-up cohort	BD I: 159 BD II: 61 BD not other specified: 3	47.1 (12.5)	BD I: 98 F 61 M BD II: 46 F 15 M Other: 2 F 1 M	19.5 (12)	HDRS: 7.3 (6.0) YMRS: 2.6 (3.6)	CRP [Not reported how it was measured]	No HC group	SCIP: 5 subscales including working memory, intermediate verbal learning, verbal fluency, delayed verbal learning, and processing speed	No HC group	Age, Education, BMI, Duration of illness, Number of hospital admissions, HDRS, Daily tobacco consumption	CRP [SCIP total score]: In women, negative correlation with cognitive performance (r=−0.270, p=0.002) CRP [Immediate verbal learning subscale]: In women, negative correlation with immediate verbal learning (r=−0.208, p=0.020) CRP [Verbal fluency subscale]: In women, negative correlation with verbal fluency (r=−0.329, p<0.001) CRP [Delayed verbal learning subscale]: In women, negative correlation with delayed verbal learning (r=−0.198, p=0.026)
Strawbridge et al. (2021) [55]	United Kingdom	Cross-sectional	BD euthymia: 44	43.7 (12.8)	31 F 13 M	Not reported	HDRS: 4.0 (2.7) YMRS: 2.7 (2.4)	Meso Scale Discovery V-Plex Kit [Plasma]: BDNF, bFGF, CRP, Eotaxin, Eotaxin-3, Flt-1, ICAM-1, IFN-γ, IL-10, IL-12, IL-15, IL-16, IL-17, IL-1α, IL-6, IL-7, IL-8, IP-10, MCP-1, MCP-4, Mip-1α, Mip-1β, PlGF, SAA, TARC, Tie-2, TNF-α, TNF-β, VCAM-1, VEGF, VEGF-C, VEGF-D	No HC group	WAIS: Processing speed, working memory, verbal learning and memory, and executive functioning FAS: Verbal fluency	No HC group	Age, Gender, Bipolar type, Number of lifetime episodes, Smoking, Medications, Health-related quality of life, Physical illness, FAST	Overall, IL-7, VEGF-C, and PlGF were positively correlated with cognitive dysfunction, as measured by the composite scores defined in the study (both binary and continuous measures) No significant correlations between other inflammatory biomarkers and cognition
Zazula et al. (2022) [56]	Brazil	Case-control	BD euthymia: 31 HC: 27	BD: 39.5 (11.5) HC: 38..7 (13.7)	BD: 25 F 6 M HC: 18 F 9 M	Age onset: 20.5 (7.2)	17-item HDRS: median 11 (IQR: 4, 16) YMRS: median 2 (2, 6)	Luminex MAGPIX® [Serum]: TNF-α sTNF-R1 sTNF-R2	TNF-α: BD 24.7 (129) HC 0.7 (1.4) p=0.043 sTNF-R1: Trend towards significance at the level of p=0.082 between BD and HC groups No significant differences for sTNF-R2 between BD and HC groups	CogState Research Battery with set-shifting task, Groton maze learning task, and two-back task: Executive function and working memory	No significant differences for set-shifting task or Groton maze learning task Two-back test: (t(56)=−2.9, p=0.005)	Age, Gender, BMI, Depressive, anxious, and manic symptoms	sTNF-R2 [Set-shifting task]: Positive correlation (ρ=0.37, p=0.042) sTNF-R2 [Groton maze learning task]: Positive correlation (ρ=0.54, p=0.002) sTNF-R2 [Two-back task]: Negative correlation (ρ=−0.49, p=0.005) No significant correlations for TNF-α or sTNF-R1

AUD: Alcohol Use Disorder; BAC-A: Brief Assessment of Cognition in Affective Disorders; BD: Bipolar Disorder; BD I: Bipolar Disorder Type I; BD II: Bipolar Disorder Type II; BDNF: Brain-Derived Neurotrophic Factor; bFGF: Basic Fibroblast Growth Factor; BMI: Body mass index; BSRG: Basic Sciences Research Group; CGI: Clinical Global Impression; COWAT: Controlled Oral Word Association Task; CPT: Continuous Performance Test; CPT-II: Conners’ Continuous Performance Test II; CRP: C-reactive protein; CRP-us: Ultra-sensitive C-reactive protein; CTT: Color Trails Test; CVLT: California Verbal Learning Test; D-KEFS: Deilis-Kaplan Executive Function System; ECL: Electrochemiluminescence; ELISA: Enzyme-Linked Immunosorbent Assay; F: Female; FAB: Frontal Assessment Battery; FAS: Fluency Assessment Scale; FAST: Functioning Assessment Short Test; Flt-1: Fms-like tyrosine kinase 1; G-CSF: Granulocyte-Colony Stimulating Factor; GAF: Global Assessment of Functioning; GM-CSF: Granulocyte-Macrophage Colony-Stimulating Factor; HC: Healthy Controls; HAM-D: Hamilton Depression Rating Scale; HDRS: Hamilton Depression Rating Scale; hs-CRP: High-sensitivity C-reactive protein; HSV-1: Herpes simplex virus, type 1; ICAM-1: Intracellular Adhesion Molecule-1; IDS: Inventory of Depressive Symptomatology; IFN-γ: Interferon-gamma; IgG: Immunoglobulin G; IgM: Immunoglobulin M; IL-1α: Interleukin-1 alpha; IL-1β: Interleukin-1 beta; IL-1Ra: Interleukin-1 Receptor Antagonist; IL-2: Interleukin-2; IL-4: Interleukin-4; IL-5: Interleukin-5; IL-6: Interleukin-6; IL-7: Interleukin-7; IL-8: Interleukin-8; IL-9: Interleukin-9; IL-10: Interleukin-10; IL-12: Interleukin-12; IL-13: Interleukin-13; IL-15: Interleukin-15; IL-16: Interleukin-16; IL-17A: Interleukin-17A; IL-18: Interleukin-18; IP-10 (CXCL10): Interferon gamma-induced protein 10; IQ: Intelligence quotient; IQR: Interquartile range; M: Male; MCCB: MATRICS Consensus Cognitive Battery; MCP-1 (CCL2): Monocyte Chemoattractant Protein-1; MCP-4: Monocyte Chemoattractant Protein-4; MDD: Major Depressive Disorder; MMSE: Mini-Mental State Examination; MIP-1α (CCL3): Macrophage Inflammatory Protein-1 alpha; MIP-1β (CCL4): Macrophage Inflammatory Protein-1 beta; MoCA: Montreal Cognitive Assessment; mRNA: messenger Ribonucleic Acid; N: Number of Participants; OPG: Osteoprotegerin; PDGF-BB: Platelet-Derived Growth Factor-BB; PlGF: Placental Growth Factor; RAVLT: Rey Auditory Verbal Learning Test; RBANS: Repeatable Battery for the Assessment of Neuropsychological Status; RCFT: Rey-Osterrieth Complex Figure Test; RANTES (CCL5): Regulated on Activation, Normal T Cell Expressed and Secreted; SAA: Serum Amyloid A; sCD14: Soluble Cluster of Differentiation 14; sCD40L: Soluble CD40 Ligand; SCIP: Screen for Cognitive Impairment in Psychiatry; SCWT: Stroop Color and Word Test; SD: Standard Deviation; sIL-6R: Soluble Interleukin-6 Receptor; sTNF-αR1: Soluble Tumor Necrosis Factor Alpha Receptor 1; sTNFR1: Soluble Tumor Necrosis Factor Receptor 1; sTNFR2: Soluble Tumor Necrosis Factor Receptor 2; SZ: Schizophrenia; T2DM: Type 2 Diabetes Mellitus; TAP: Test for Attentional Performance; TARC: Thymus and Activation-Regulated Chemokine; TGF-β: Transforming Growth Factor-beta; Tie-2: Angiopoietin-1 Receptor; TIMP1: Tissue Inhibitors of Metalloproteinases 1; TMT: Trail Making Test; TMT-A: Trail Making Test Part A; TMT-B: Trail Making Test Part B; TNF-α: Tumor Necrosis Factor-alpha; TNF-β: Tumor Necrosis Factor-beta; VEGF: Vascular Endothelial Growth Factor; VEGF-C: Vascular Endothelial Growth Factor C; VEGF-D: Vascular Endothelial Growth Factor D; WAIS: Wechsler Abbreviated Scale of Intelligence; WCST: Wisconsin Card Sorting Test; WLMT: Word List Memory Task; WMS-III: Wechsler Memory Scale-III; YMRS: Young Mania Rating Scale.

Figure 2.

Summary of associations between inflammatory biomarkers and cognitive function in adults with bipolar disorder across included studies. This figure provides a visual summary of the reported associations between specific inflammatory biomarkers and cognitive performance across the included studies. A green plus sign (+) indicates a statistically significant association between the biomarker and cognitive outcomes, while a red minus sign (–) indicates that no significant association was found. Blank cells reflect that the biomarker was not assessed in that study. Notes on the right provide context for studies reporting no significant associations, highlighting factors such as small sample sizes, limited adjustment for confounders, or reliance on less sensitive cognitive measures.

4.3. C-Reactive Protein (CRP)

Eleven studies (k=11) [29,40–42,44–48,50,54] assessed CRP in relation to cognitive domains in patients with BD. Of these, seven studies (k=7) [40–42,47,49,50,54] found significant correlations between elevated CRP levels and cognitive dysfunction, while four studies (k=4) [29,44–46] measured CRP but did not find significant correlations.

Congio et al. (2022) [40] found that patients with BD who had elevated high-sensitivity C-reactive protein (hs-CRP) levels (≥5 mg/L) exhibited poorer cognitive performance, particularly in verbal fluency and processing speed tasks. Specifically, these patients scored lower on phonetic (p<0.01) and semantic verbal fluency tests (p=0.04) and took longer to complete the Trail Making Test Part A (TMT-A) compared to HCs (p=0.03). Similarly, Dickerson et al. (2013) [41] identified negative correlations between hs-CRP levels and various cognitive functions in patients with BD in manic, depressive, or mixed episodes, including total scores on the Repeatable Battery for the Assessment of Neuropsychological Status (RBANS) (t=−2.48, p=0.015), as well as in immediate memory (t=−2.16, p=0.033), attention (t=−2.18, p=0.032), and language (t=−2.13, p=0.036). The study also reported a negative correlation between hs-CRP and TMT-A performance (t=−2.39, p=0.019).

In a prospective 1-year follow-up cohort, Garés-Caballer et al. (2022) [42] demonstrated that lower levels in ultra-sensitive CRP (CRP-us) were significantly associated with a 26.4–49.8% better performance in executive functions among affectively stable BD patients. Additionally, Mao-Hsuan Huang et al. (2022) [47] found that elevated CRP levels were associated with worse performance in auditory divided attention tasks, as measured by the Test for Attentional Performance (TAP) (β=−0.732, p=0.006). Millet et al. (2021) [49] further explored the impact of elevated CRP levels on cognitive function by comparing euthymic BD patients and HCs with high CRP levels (≥5 mg/L) to those with normal CRP levels (<5 mg/L). The study found significant effects of high CRP levels on multiple cognitive domains, particularly on the WCST (F(1)=6.98, p=0.009), Stroop Color and Word Test (Stroop) (F(1)=4.46, p=0.036), Reading Mind in the Eyes (F(1)=19.9, p<0.001), MATRICS Consensus Cognitive Battery (MCCB) speed of processing (F(1)=6.4, p=0.012), and MCCB reasoning and problem solving (F(1)=6.8, p=0.010). However, there was no main effect of diagnosis nor was the interaction between diagnosis and CRP group significant. The authors also found that, within the BD I group, CRP levels were significantly associated with the MATRICS composite score (R=−0.199, p=0.008), an association that was stronger compared to the effect noted within the BD-II group (R=0.04, p=0.83).

Milton et al. (2021) [50] observed that higher CRP levels were significantly associated with worse visuospatial memory in patients with BD, as indicated by reaction time and pairs matching tasks (β=−0.162, p=0.001). Lastly, Sanchez-Autet et al. (2018) [54] found differential correlations between CRP levels and cognitive performance in men and women with BD. Specifically, negative correlations were observed in women across multiple cognitive domains, including overall cognitive performance (r=−0.270, p=0.002), immediate verbal learning (r=−0.208, p=0.020), verbal fluency (r=−0.329, p<0.001), and delayed verbal learning (r=−0.198, p=0.026).

4.4. Tumor Necrosis Factor-alpha (TNF-α), Soluble Tumor Necrosis Factor Receptor 1 (sTNFR1), and Soluble Tumor Necrosis Factor Receptor 2 (sTNFR2)

A total of fifteen (k=15) studies assessed the relationship between TNF-α and/or its soluble receptors (sTNFR1 and sTNFR2) and cognitive domains in patients with BD [26,27,29,37,38,42,44–48,51,52,55,56]. Specifically, twelve studies (k=12) focused on TNF-α [26,27,37,38,42,45,46,48,51,52,55,56], while eight studies (k=8) focused on the soluble receptors sTNFR1 and/or sTNFR2 [26,29,37,38,44,47,48,56].

4.4.1. Tumor Necrosis Factor-alpha (TNF-α)

Among the twelve studies (k=12) [26,27,37,38,42,45,46,48,51,52,55,56] that assessed TNF-α in relation to cognitive functions in patients with BD, four (k=4) [26,37,46,48] reported significant correlations between elevated TNF-α levels and cognitive dysfunction, whereas eight studies (k=8) [27,38,42,45,51,52,55,56] found no significant correlations.

Barbosa et al. (2012) [37] observed a positive correlation between TNF-α levels and inhibitory control in the FAB, particularly within the inhibitory control subdomain (ρ [rho]=0.50, p=0.02). Similarly, Doganavsargil-Baysal et al. (2013) [26] identified a significant negative correlation between TNF-α levels and delayed recall in the Rey Auditory Verbal Learning Test (RAVLT) (r=−0.275, p=0.044), indicating that higher TNF-α levels might be associated with poorer memory retention in euthymic patients with BD.

Liou et al. (2023) [46] also found that elevated TNF-α levels were negatively correlated with the visual immediate index in the WMS-III (r=−0.10, p=0.05). Millet et al. (2020) [48] further reported significant negative correlations between TNF-α levels and executive function, as measured by the Stroop and the WCST. The study showed that higher TNF-α levels were associated with poorer performance on Stroop (r=−0.2, p=0.005) and WCST (r=−0.2, p=0.03). In addition, the study also found that peripheral TNF markers partially mediated the causal relationship between number of prior severe mood episodes and impaired executive function.

4.4.2. Soluble Tumor Necrosis Factor Receptor 1 (sTNFR1) and Soluble Tumor Necrosis Factor Receptor 2 (sTNFR2)

Among the eight studies (k=8) [26,29,37,38,44,47,48,56] that examined the relationship between sTNFR1 and/or sTNFR2 and cognitive performance in patients with BD, four studies (k=4) [44,47,48,56] identified significant correlations between these soluble receptors and cognitive functions, while the remaining four studies (k=4) [26,29,37,38] did not find significant associations.

Hua et al. (2021) [44] found that elevated sTNFR1 levels in patients with remitted BD were negatively correlated with cognitive performance, specifically in tasks assessing immediate memory and executive function, such as the Word List Memory Task (WLMT) (β=−0.004, p<0.001, for word list I recall; β=−0.002, p<0.001, for word list II recall) and the WCST (β=−0.009, p=0.012, for percent conceptual level responses; β=−0.001, p=0.048, for the number of categories completed). Similarly, Mao-Hsuan Huang et al. (2022) [47] observed that elevated sTNFR1 levels were associated with poorer performance in learning and memory tasks, as well as attention and working memory tasks, including the WLMT (β=−5.97, p=0.004, for word list II recall; β=−30.851, p=0.037, for word list retention) and the TAP (β=−3.774, p=0.009, for auditory divided attention).

Millet et al. (2020) [48] also reported significant negative correlations between sTNFR2 levels and executive function, as measured by the Stroop (r=−0.2, p=0.005). Higher sTNFR2 levels were linked to poorer performance on tasks requiring cognitive flexibility. In contrast, Zazula et al. (2022) [56] observed a positive correlation between sTNFR2 levels and performance on executive function tasks, including the set-shifting task (ρ=0.37, p=0.042) and Groton maze learning task (ρ=0.54, p=0.002). However, they also found a negative correlation with the two-back task (ρ=−0.49, p=0.005), indicating that while sTNFR2 might support certain aspects of executive function, it could also be associated with dysfunction in working memory.

4.5. Interleukin-6 (IL-6)

Thirteen studies (k=13) [27–29,36,38,42–45,47,51,52,55] assessed IL-6 in relation to cognitive functions in patients with BD. Of these, four studies (k=4) [28,38,42,52] found significant correlations between elevated IL-6 levels and cognitive dysfunction, while nine studies (k=9) [27,29,36,43–45,47,51,55] measured IL-6 but did not find significant correlations.

Barbosa et al. (2018) [38] found a negative correlation between IL-6 levels and cognitive performance in patients with remitted BD, specifically in the total z-score (rho=−0.461, p=0.047), as measured by the Brief Assessment of Cognition in Affective Disorders (BAC-A). Similarly, in a prospective 1-year follow-up cohort, Garés-Caballer et al. (2022) [42] demonstrated that lower levels of IL-6 were significantly associated with better executive functions. Specifically, in the affectively stable BD group, 26.4–49.8% of better executive functioning was correlated with lower levels of IL-6.

Hamdani et al. (2015) [28] also found a significant positive correlation between IL-6 mRNA expression and cognitive deterioration among euthymic BD patients, particularly in the WMS-III cognitive deterioration index (rho=0.43, p=0.01). This relationship was even more pronounced in patients with Toxoplasma gondii (T. gondii) infection, where IL-6 mRNA expression was twice as high in those who showed cognitive decline (p=0.01). Finally, Poletti et al. (2021) [52] reported that higher levels of IL-6 were associated with an increased likelihood of having poor cognitive performance overall among BD patients with a depressive episode in course, as measured by the Brief Assessment of Cognition in Schizophrenia.

4.6. Interleukin-1 Receptor Antagonist (IL-1Ra)

Three studies (k=3) [27,29,52] examined the relationship between IL-1Ra levels and cognitive functions in patients with BD. Among these, two studies (k=2) [27,29] identified significant correlations between elevated IL-1Ra levels and cognitive dysfunction, while one study (k=1) [52] did not find significant associations.

Hope et al. (2015) [29] reported negative correlations between IL-1Ra levels and general cognitive abilities in patients with BD, as measured by the Wechsler Abbreviated Scale of Intelligence (WAIS). The study found that higher IL-1Ra levels were associated with poorer overall cognitive functioning (β=−0.10, p=0.03). Similarly, Lotrich et al. (2014) [27] found that elevated IL-1Ra levels were negatively correlated with cognitive performance across multiple domains in euthymic BD patients, including global cognition (r=−0.372, p=0.01), visual cognition (r=−0.312, p=0.03), memory cognition (r=−0.345, p=0.02), and speed/executive cognition (r=−0.404, p=0.005).

4.7. Other inflammatory markers

Several other biomarkers have been investigated for their potential relationships with cognitive performance in patients with BD. These biomarkers include BDNF, IL-18, IL-8, monocyte chemoattractant protein (MCP-1/CCL2), YKL-40, and soluble CD40 ligand (sCD40L), and others. Although the evidence is less consistent and often comes from a smaller number of studies, these biomarkers also provide insight into the complex interactions between inflammation and cognition in BD.

4.7.1. Brain-Derived Neurotrophic Factor (BDNF)

A total of six studies (k=6) measured BDNF in patients with BD [27,37,39,46,51,55]. Of these, one study (k=1) [51] found significant positive correlations, while the remaining five studies (k=5) [27,37,39,46,55] did not find significant associations. Specifically, Mora et al. (2019) [51] reported that BDNF was associated with better performance in executive functions, as measured by the WCST (β=0.01, p=0.02), and verbal memory, as assessed by the California Verbal Learning Test (CVLT) (β=0.013, p=0.005)..

4.7.2. Interleukin-8 (IL-8)

One study (k=1) explored the relationship between IL-8 and cognitive performance in patients with BD [46]. Liou et al. (2023) [46] found a significant negative correlation between IL-8 levels and executive function, specifically in the number of categories completed on the WCST (r=−0.13, p=0.02). This suggests that elevated IL-8 levels may be associated with dysfunctions in cognitive flexibility.

4.7.3. Interleukin-18 (IL-18)

Similarly, one study (k=1) explored the role of IL-18 among euthymic BD patients and found mixed results [36]. Arslan et al. (2017) [36] reported that IL-18 was positively correlated with the number of categories achieved in the WCST (r=0.288, p=0.05), indicating better cognitive flexibility. Additionally, IL-18 was positively correlated with instant recall (r=0.307, p=0.01), delayed recall (r=0.321, p=0.01), and learning scores (r=0.297, p=0.05) in the RAVLT. However, IL-18 also showed negative correlations with perseverative responses (r=−0.264, p=0.05) and errors (r=−0.279, p=0.01) in the WCST, as well as with Stroop interference (r=−0.235, p<0.05). These findings suggest that while IL-18 may be associated with improved performance in some cognitive domains, it might also be linked to other aspects of cognitive dysfunction.

4.7.4. Soluble CD40 Ligand (sCD40L)

One study (k=1) explored the relationship between sCD40L and cognitive performance in patients with BD [29]. Hope et al. (2015) [29]. found a significant negative correlation between sCD40L levels and general cognitive abilities, as measured by the WASI (β=−0.10, p=0.03). This suggests that higher levels of sCD40L may be associated with poorer overall cognitive functioning in patients with BD.

4.7.5. YKL-40

One study (k=1) investigated YKL-40, a biomarker found in cerebrospinal fluid (CSF) [53]. Rolstad et al. (2015) [53] found that YKL-40 levels were significantly higher in euthymic BD patients compared to HCs. YKL-40 was negatively correlated with executive performance (β=−0.99, p<0.0005), suggesting that higher levels of this biomarker were associated with worse executive function in euthymic patients with BD.

4.7.6. Additional markers

In a study by Poletti et al. (2021) [52], various other cytokines were examined among BD patients with a depressive episode in course, and several were found to be significantly associated with cognitive performance. Higher levels of IL-1β, CCL2, macrophage inflammatory protein-1 beta (MIP-1β /CCL4), regulated upon activation, normal T cell expressed and secreted) (RANTES/CCL5), interferon gamma-induced protein 10 (IP-10/CXCL10), and bFGF were associated with an increased likelihood of poor cognitive performance overall, indicating that these cytokines may contribute to cognitive dysfunction in BD. Conversely, IL-17 was found to be associated with better verbal fluency, suggesting a potentially protective or compensatory role in this specific cognitive domain. Additionally, Strawbridge et al. (2021) [55] identified positive correlations between IL-7, vascular endothelial growth factor C (VEGF-C), and placental growth factor (PlGF) and cognitive dysfunction, particularly in processing speed, working memory, and executive functioning.

4.8. Risk of bias

In the quality assessment of the included studies, we identified several potential risks of bias, with the most prominent concern being in the domain of risk of bias due to confounding. Specifically, seven out of the 25 studies exhibited some level of bias due to confounding factors [26,28,39,40,42,43,46]. These studies did not adequately adjust for significant within-group confounders that could influence both inflammatory marker levels and cognitive function, such as medication use [26,28,39,40,42,43], body mass index (BMI) [28,46], and alcohol use disorder [46]. However, it is important to note that all 25 studies did control for age, sex, and education, which are well-established confounders affecting inflammation and cognitive dysfunction in BD.

In the domain of exposure measurement, one study [53] was categorized as having some concerns. In this study, CSF sampling was conducted approximately seven months after the neuropsychological examination. Although the authors argued that the participants were in a euthymic state during this period as defined by a score lower than 14 on both Montgomery–Åsberg Depression Rating Scale (MADRS) and Young Mania Rating Scale (YMRS), and thus, the delay likely did not compromise the data, this temporal gap introduces potential bias in exposure measurement. Additionally, in the domain of selection of participants into the study (or into the analysis), one study [40] was categorized as having a high risk of bias. In this case, the control group, composed of staff members, significantly differed from the patients with BD in baseline characteristics, such as years of education.

Finally, in the domain of post-exposure bias, another study [42] was identified as having some concerns. This study did not account for changes in medication or relapses of BD during a one-year period, factors that could significantly affect both inflammation levels and cognitive performance. Additionally, this study raised concerns regarding missing data, with a loss to follow-up of 40 patients. For all other domains assessed in the quality review, no significant risk of bias was detected in the 25 studies. Detailed information on the risk of bias for each study can be found in Supplemental Figure 1. Overall, a total of 17 studies were categorized as having a low risk of bias [27,29,36–38,41,44,45,47–52,54–56], seven as having some concerns [26,28,39,42,43,46,53], and one as high risk [40]. The overall summary of the risk of bias is presented in Supplemental Figure2.

5. Discussion

This systematic review highlights the relationship between inflammation and cognitive dysfunction in BD, with elevated inflammatory markers such as CRP, TNF-α, IL-6, IL-1Ra associated with cognitive impairment in executive function, processing speed, and memory. Notably, most studies were categorized as low risk of bias, strengthening the reliability of the observed associations. Growing evidence suggests that anti-inflammatory therapies improve cognition in conditions such as rheumatoid arthritis and in patients with an inflammatory profile in Alzheimer’s disease and major depressive disorder (MDD) [57–60]. Together, these findings support the potential for targeting inflammation as a therapeutic strategy in BD.

Among the inflammatory markers, elevated levels of CRP were associated with cognitive dysfunction. CRP is an acute-phase protein produced by the liver in response to systemic inflammation, is widely recognized as a marker for inflammatory processes, and has been studied in several neuropsychiatric disorders, including MDD, SZ, obsessive-compulsive disorder (OCD), and Alzheimer’s disease [61–65]. In these conditions, CRP is thought to contribute to cognitive dysfunction through different mechanisms, including disruption of synaptic plasticity and neurogenesis, oxidative stress, and activation of microglial cells, leading to neurodegeneration [66]. Mechanistically, CRP exacerbates inflammation by activating the complement system, releasing pro-inflammatory cytokines, disrupting neuroplasticity, and influencing the blood-brain barrier (BBB), potentially allowing more inflammatory mediators to enter the central nervous system (CNS), which could be implicated in the cognitive dysfunction process observed in BD [61,66–68].

Similarly, elevated levels of TNF-α were also associated with cognitive dysfunction in BD. TNF-α is a pro-inflammatory cytokine with a central role in the immune response, and its relationship with cognitive dysfunction has been studied across other neuropsychiatric conditions [69–71]. TNF-α influences microglia and astrocytes, leading to alterations in synaptic function, neurogenesis, and neuronal survival [72], and chronic elevation of TNF-α has been associated with neuroinflammation [73]. In the context of BD, TNF-α is particularly relevant due to its ability to cross the BBB [74] and modulate neurotransmitter systems [75]. Furthermore, the soluble receptors of TNF-α, sTNFR1 and sTNFR2, regulate the activity of TNF-α [76], either neutralizing or prolonging its activity, depending on the concentrations [77]. The findings indicate that sTNFR2, in particular, is more consistently associated with cognitive dysfunction [48,78]. The role of TNF-α and its receptors in BD may involve multiple pathways, including the activation of microglia with the production of reactive oxygen species (ROS), leading to neuroinflammation and neuronal damage [79,80]. This cascade of events may contribute to the cognitive dysfunction observed in BD.

IL-6 is another cytokine that was associated with cognitive dysfunction in BD. IL-6 is produced by glial cells and affects synaptic plasticity, neurogenesis, and neuronal survival [81]. Chronic exposure to elevated IL-6 levels can lead to a pro-inflammatory state within the brain [82] and elevated IL-6 levels have been linked to cognitive dysfunction in several neuropsychiatric disorders [83–85]. One of the aspects that stands out, as highlighted by Hamdani et al. (2015) [28], is the potential interaction of IL-6 with the parasite T. gondii. Toxoplasmosis has been associated with increased IL-6 levels [86,87] and may contribute to the exacerbation of cognitive dysfunction observed in BD [28]. Notably, T. gondii infection has been linked to cognitive decline in patients with SZ [88], and although preventable, knowledge about the disease and its prevention remains low among pregnant women and healthcare professionals [89]. The mechanism by which IL-6 influences cognitive function may involve its impact on synaptic plasticity and neurogenesis [90], altering the balance between excitatory and inhibitory neurotransmission [91]. Additionally, IL-6 can influence the hypothalamic-pituitary-adrenal (HPA) axis, leading to increased cortisol production, which has been associated with memory impairments and other cognitive dysfunction [92,93].

Moreover, IL-1Ra was also associated with cognitive dysfunction in BD. IL-1Ra is an anti-inflammatory cytokine that functions as an endogenous inhibitor of IL-1, a pro-inflammatory cytokine involved in neuroinflammation and neurodegeneration [94]. Elevated IL-1Ra levels have been observed in inflammatory conditions and may reflect an attempt by the body to counteract excessive inflammation [95]. However, in the context of BD, elevated IL-1Ra levels have been paradoxically associated with cognitive dysfunction [27]. This could be due to the chronic nature of inflammation in BD, where the prolonged elevation of IL-1Ra may indicate ongoing immune activation rather than successful resolution of inflammation [96]. The mechanisms underlying the association between IL-1Ra and cognitive dysfunction are not fully understood, but it is possible that chronic inflammation, even in the presence of elevated IL-1Ra, leads to sustained neuroinflammation and cognitive decline [97]. The role of IL-1Ra in inhibiting IL-1 may also have downstream effects on other cytokines, further complicating its impact on cognitive function [98]. The evidence suggests that IL-1Ra, while generally considered a protective factor, may serve as a marker of cognitive dysfunction in BD.

Other inflammatory biomarkers such as BDNF, IL-18, IL-8, MCP-1, YKL-40, and sCD40L have also been explored for their potential roles in cognitive dysfunction in BD. However, the evidence for these markers is more limited and less consistent. For instance, BDNF, typically associated with neuroprotection and cognitive enhancement [99], has shown complex associations with cognitive function in BD, with some studies suggesting a protective role and others indicating potential impairment. Similarly, IL-18 and IL-8 have shown both positive and negative correlations with cognitive performance. MCP-1 and YKL-40, markers of microglial activation and neuroinflammation [100,101], have been linked to cognitive dysfunction in some studies. sCD40L, a marker of platelet activation and immune response [102], has also been associated with poorer cognitive outcomes, further highlighting the diverse ways in which inflammation can impact brain function in BD.

Our findings align with previous reviews that highlight the role of neuroinflammation in cognitive dysfunction in BD [19,20,30]. Unlike reviews from Bauer et al. (2015) [19], which focused only on aspects of inflammation, or Misiak et al. (2018) [30], which included SZ alongside BD, our review is distinct in that it focuses specifically on the association between inflammation and cognitive function in BD. Additionally, our review involved a broader search strategy, incorporating more databases than the study by Rosenblat et al. (2015) [20]. Of note, as we were completing the review process for this manuscript, another review was published by Mario et al. (2024) [103] that investigated a similar topic. However, Mario et al. (2024) [103] also included neuroimaging studies, adolescents, and studies evaluating inflammation more broadly, not solely its direct role in cognition. Moreover, it is important to mention that our review utilized different databases, potentially leading to the identification of additional studies that evaluated other biomarkers, such as BDNF, MCP-1, YKL-40, and sCD40L.

Strengths of this review include the comprehensive search across six major databases with no language or year restrictions during the search process, the rigorous application of pre-specified eligibility criteria, and a systematic risk of bias assessment using the ROBINS-E tool. Additionally, the use of a standardized data extraction process and independent screening by multiple reviewers enhances the reliability of our findings. The findings from this review suggest that inflammation may be a relevant target for improving cognitive outcomes in BD. While causality remains to be determined, screening for elevated inflammatory markers in clinical settings might help identify patients at greater risk for inflammation-induced cognitive impairment. Additionally, these findings support the rationale for investigating anti-inflammatory interventions in BD, particularly among patients with elevated markers.

Despite these strengths, several limitations must be acknowledged when interpreting the findings of this review. First, a significant proportion of the studies reviewed were cross-sectional in nature, limiting the ability to establish a causal relationship between inflammation and cognitive dysfunction in BD [104]. Second, the heterogeneity in study methodologies further complicates the interpretation of findings, with only a few studies adhering to the ISBD recommended cognitive battery [31]. The timing of blood sample collection for cytokine measurement also varied across studies, with some not specifying the time-of-day samples were taken. Given the known diurnal variations in cytokine levels, this variability could affect the results and introduce confounding factors [105]. Third, most studies focused on peripheral markers rather than central markers from CSF, with only one study in this review examining CSF cytokines [53], limiting our understanding of the direct impact of neuroinflammation on cognitive function in BD. Another important limitation is that the phase of illness in BD, including different subtypes (e.g., BD I, BD II) and phases (e.g., remission, acute episodes, mania, depression), adds a layer of complexity. Cognitive dysfunction in BD can vary depending on the illness phase, with some impairments persisting during remission while others fluctuate with mood episodes [60,106]. Additionally, subsyndromal symptoms, often present during remission, may have influenced findings across studies, potentially confounding the inflammation-cognition relationship. As noted by Sæther et al. (2022) [107], subgroup heterogeneity related to affective state or illness course may obscure consistent patterns and should be addressed in future research. Lastly, some studies in this review did not adjust for certain confounding factors, such as BMI or medication use, raising some concerns of bias.

This review also has its limitations. We only included studies involving adults aged 18 years and older, excluding research on pediatric populations, limiting the generalizability of the findings. Additionally, we excluded articles published in languages other than English. We also did not search the gray literature, which may have led to the omission of relevant unpublished data Lastly, we did not assess potential moderators or mediators of the relationship between inflammation and cognitive dysfunction. The substantial heterogeneity across studies, including differences in biomarker assays, cognitive domains evaluated, clinical characteristics (e.g., phase of illness), and covariate adjustments, limited the feasibility of identifying consistent patterns or conducting such stratified analyses.

6. Conclusion

This systematic review provides evidence that elevated peripheral inflammatory markers are associated with cognitive dysfunction in BD, with CRP, TNF-α, IL-6, and IL-1Ra emerging as key biomarkers associated with specific cognitive domains such as executive function, processing speed, and memory. The involvement of other inflammatory markers also points to a complex relationship between inflammation and cognition in BD. While more longitudinal and mechanistic studies are needed, our findings suggest that targeting inflammation may offer a novel therapeutic avenue to address cognitive impairment in BD. Clinicians should consider the potential relevance of inflammatory processes in the cognitive symptoms experienced by patients with BD, and future research should focus on developing precision medicine approaches that integrate inflammatory profiles into treatment planning. Future studies should also prioritize longitudinal designs, consistent use of ISBD-recommended cognitive batteries, careful control of confounders, and exploring central markers of inflammation to provide more direct insights into the neurobiological mechanisms underlying the cognitive dysfunction in BD.

• Summations

  • Elevated levels of inflammatory markers, including CRP, TNF-α, IL-6, and IL-1RA, are consistently associated with cognitive dysfunction in adults with bipolar disorder, particularly in domains such as executive function, processing speed, and memory.
  • Most included studies were assessed as having a low risk of bias, strengthening the reliability of the observed associations.
  • Targeting inflammation may offer a potential therapeutic strategy to address cognitive dysfunction in bipolar disorder.
• Limitations

  • Most of the included studies relied on cross-sectional designs, limiting the ability to infer causality between inflammatory markers and cognitive function in bipolar disorder.
  • High heterogeneity was observed in the cognitive assessment tools used across studies, along with a predominant focus on peripheral rather than central cytokine measurements.
  • The exclusion of non-English publications and grey literature, along with the exclusive focus on adult populations, may reduce the generalizability of the findings.

Data availability statement:

Data sharing not applicable to this article as no datasets were generated or analyzed during the current study.