Does resolvin D2 reflect cognitive impairment in older adults? A clinical and biomarker analysis Authors: Serap Boz, Arzu Nevin Dagdemir, Hande Selvi Oztorun, Zeliha Kocaismail, Rana Tuna Dogrul, Esra Fırat Oguz, Kubra Erdogan, Busragul Yılmaz, Fatma Kaplan Efe, Rıdvan Erten, Ertugrul Demirel, Gunes Eken, Kamile Silay Published: 2026 ABSTRACT BackgroundAlzheimer’s disease is a progressive neurodegenerative disorder with an unclear etiology, in which inflammation and impaired resolution pathways may play a role. This study investigated the relationship between serum Resolvin D2 levels and cognitive status in individuals with normal cognition, mild cognitive impairment, and Alzheimer’s disease.MethodsIn this cross-sectional study, 165 adults aged ≥ 65 years who attended a tertiary geriatric clinic between 2022 and 2025 were classified as cognitively normal, mild cognitive impairment, or Alzheimer’s disease. Demographic, clinical, laboratory, and comprehensive geriatric assessment data were collected. Serum Resolvin D2 levels were measured using ELISA and analyzed in relation to cognitive status. Logistic regression and ROC curve analyses were performed to evaluate the potential diagnostic value of Resolvin D2.ResultsResolvin D2 levels did not differ significantly among cognitive status groups. In multivariate analyses, Resolvin D2 was not an independent predictor of dementia or cognitive impairment. ROC curve analyses demonstrated limited diagnostic performance of Resolvin D2 compared with established clinical and laboratory indicators. These findings suggest that serum Resolvin D2 levels may have limited utility as a standalone biomarker, although their role within neuroinflammatory and resolution pathways warrants further investigation.ConclusionSerum Resolvin D2 levels have limited ability to distinguish older adults with cognitive impairment or dementia from cognitively normal individuals. In contrast, S-MMSE, Lawton IADL scores, and comorbidity burden remain independent predictors of cognitive status. While Resolvin D2 shows anti-inflammatory and pro-resolving effects in experimental models, its value as a standalone biomarker for cognitive decline appears limited. Larger, prospective, multi-biomarker studies are needed to clarify its role.Supplementary InformationThe online version contains supplementary material available at 10.1186/s12877-026-07817-9. Introduction Dementia is a neurodegenerative disorder characterized by impairment of cognitive functions, including learning, memory, orientation, language, and personality, leading to significant deterioration in social and occupational functioning. Alzheimer’s disease (AD) accounts for approximately 60 to 80% of all dementia cases. More than 5.5 million individuals in the United States and over 35 million worldwide have been diagnosed with Alzheimer’s disease, which may lead to death within 3–9 years after diagnosis. The most important risk factor for Alzheimer’s disease is age. The incidence of the disease doubles every 5 years after age 65, with 1,275 out of every 100,000 people over the age of 65 receiving a new diagnosis each year [1, 2]. Alzheimer’s disease is believed to begin at least 20 years before the onset of clinical symptoms and progressively worsens over time. The prodromal stage of AD represents a critical “window of opportunity” for therapeutic intervention to delay the onset of AD and slow the progression of neurodegeneration. Therefore, it is essential to investigate novel biomarkers and preclinical diagnostic approaches to improve early detection and intervention [3]. Neuropathologically, AD is characterized by senile plaques composed of extracellular amyloid β protein (Aβ) and neurofibrillary tangles composed of intracellular hyperphosphorylated tau protein [4]. Although the exact role of Aβ in AD is not fully understood, the formation of plaques from accumulated Aβ is considered a central event in disease development. It has been shown that Aβ accumulation, resulting from an imbalance between production and clearance, plays a critical role in AD pathogenesis. Although inflammation-induced neuronal loss, senile plaques, and neurofibrillary tangles are recognized as key pathogenic mechanisms, the precise processes underlying Aβ accumulation and subsequent neuronal degeneration remain unclear [5]. Oxidative stress also plays a key role in pathogenesis. The accumulation of reactive oxygen species damages major cellular components, primarily mitochondria, in specific brain regions [6]. Inflammation plays an important role in the pathogenesis of Alzheimer’s disease. However, peripheral inflammatory markers may not fully reflect central nervous system–specific inflammatory and resolution processes, as neuroinflammation is largely compartmentalized within the brain and cerebrospinal fluid. Activated microglia and reactive astrocytes accumulate in fibrillary plaques, and their biochemical markers increase. Initially, microglial phagocytes degrade beta-amyloid through phagocytosis. Chronically activated microglia initiate the cytokine cascade following chemokine release. Fibrillar Aβ and glial activation trigger the classical complement pathway. Stimulated astroglia release acute-phase reactants that can exert both detrimental and potentially protective effects on the clinical course of the disease [2]. Uncontrolled inflammation is a component underlying many pathologies such as cardiovascular disease, diabetes, and sepsis. The resolution of inflammation is not a passive process but an active phase that progresses through specifically regulated biological mechanisms. This process is orchestrated by a class of autacoids known as “specialized pro-resolving mediators” (SPMs). These mediators are locally produced during the resolution phase of inflammation and actively regulate inflammatory responses. They include lipoxins derived from arachidonic acid, as well as resolvins, protectins, and maresins derived from omega-3 fatty acids such as EPA and DHA. In recent years, it has been demonstrated that resolvins, endogenously synthesized from omega-3 fatty acids, exhibit potent pro-resolving and anti-inflammatory effects during the resolution phase of inflammation. Resolvin D2 (RvD2), biosynthesized from DHA, was first identified during this phase and is considered a key mediator, although its complete stereochemical structure and biological effects are still being elucidated [6, 7]. Among these SPMs, RvD2 was selected due to its regulatory effects on neuroinflammation, particularly via the GPR18 receptor and PI3K/Akt pathways; these pathways have been shown in preclinical studies to be associated with cognitive functions and synaptic plasticity [8, 9]. While other SPMs, such as Resolvin D1 and maresins, also promote resolving, RvD2 has been found to uniquely modulate microglial activation and cytokine resolving in the central nervous system [6, 10]. Furthermore, the reliable measurement of RvD2 in peripheral blood using ELISA increases its applicability to clinical research in older adults. Therefore, in our study, RvD2 was chosen to investigate inflammation-resolving mechanisms associated with cognitive impairment. Currently, there is no treatment available to halt or reverse the progression of AD. Therefore, further research is needed to better understand the etiopathogenesis of AD and to identify potential novel pathological pathways and biomarkers. To our knowledge, no studies have evaluated serum Resolvin D2 (RvD2) levels in individuals with Alzheimer’s disease and compared them with those in cognitively normal individuals or those with mild cognitive impairment. Given its role as a specialized pro-resolving mediator involved in inflammation-resolution pathways, RvD2 may provide insight into the relationship between inflammation and cognitive decline. The aim of this exploratory study was to investigate the association between serum Resolvin D2 levels and cognitive status across different stages of cognitive impairment. Given the limited and inconsistent prior clinical evidence on Resolvin D2 in cognitive impairment, this study was designed as an exploratory, hypothesis-generating analysis. We hypothesized that serum RvD2 levels might differ across cognitive status groups; however, no directional hypothesis was prespecified. The primary outcome was cognitive status, categorized as cognitively normal control, mild cognitive impairment, and dementia. Among specialized pro-resolving mediators, Resolvin D2 (RvD2) has attracted particular interest due to its potent anti-inflammatory and pro-resolving effects, mediated through pathways such as GPR18 signaling and PI3K/Akt activation. These mechanisms are implicated in neuroinflammatory processes relevant to cognitive decline and Alzheimer’s disease. Therefore, this study aimed to investigate whether circulating RvD2 levels are associated with cognitive status across different stages of cognitive impairment. Materials and methods Patient characteristics A total of 165 patients aged ≥ 65 years who presented to the geriatric outpatient clinic and provided written informed consent were included in the study: 60 with Alzheimer’s disease, 52 with mild cognitive impairment (MCI), and 53 without cognitive impairment. The diagnosis was based on a comprehensive clinical evaluation and was not determined solely by Standardized Mini-Mental State Examination (S-MMSE) cut-off values. Patients with inflammatory diseases, active infections, malignancy, heart failure, vascular dementia, Parkinson’s dementia, Lewy body dementia, frontotemporal dementia, and other neurodegenerative diseases were excluded from the study. MCI diagnosis was made according to Petersen’s criteria [11]. After a comprehensive geriatric assessment (CGA) and cognitive evaluation, a diagnosis of dementia was made according to NINCDS-ADRDA [12] and DSM-5 [11] criteria. Magnetic resonance imaging was performed in all patients diagnosed with dementia to exclude other causes of dementia prior to the diagnosis of Alzheimer’s disease. Patient age, sex, living arrangement, educational status, smoking and alcohol use, body mass index (BMI), comorbidities, urinary incontinence, history of falls, and number of medications were recorded. The study design was cross-sectional. Comprehensive geriatric assessment and cognitive assessment As part of a CGA, patients were evaluated for their functional status using the Katz Activities of Daily Living Scale (ADL) [13] and the Lawton-Brody Instrumental Activities of Daily Living (IADL) scale [14], the short form of the Yesavage Geriatric Depression Scale (GDS) [15] for depression screening, and the short form of the Mini Nutritional Assessment (MNA-SF) [16] for malnutrition risk assessment. Cognitive function was assessed using the Standardized Mini-Mental State Examination (S-MMSE) [17, 18]. To assess the cognitive status of study participants, medical histories were obtained from patients and their caregivers, and the S-MMSE was administered. Following the assessment, patients were classified as normal, MCI, or AD according to the specified criteria. Resolvin-D2 measurement Venous blood samples were collected into vacutainer tubes and centrifuged at 1300 × g for 10 min. Separated sera were aliquoted into Eppendorf tubes and stored at − 80 °C until analysis. Serum Resolvin D2 (RvD2) levels were measured using a commercially available ELISA kit (Sunred Biological Technology, Baoshan, China) based on a quantitative sandwich enzyme immunoassay. Optical density (OD) was measured spectrophotometrically at 450 nm using a microplate reader, and concentrations were determined from the standard curve. The detection range of the assay (catalog no: 201-12-9317; lot no: DZE201129317) was 0.05–15 ng/mL, with a sensitivity of 0.042 ng/mL. Intra- and inter-assay coefficients of variation were < 10% and < 12%, respectively. Ethics This study was approved by the Ethics Committee of Ankara Bilkent City Hospital (Approval No: TABED-1-24-643). All participants provided written informed consent prior to participating in the study. Statistical analysis All statistical analyses were performed using IBM SPSS Statistics (Version 26). The distributions of continuous variables were assessed using the Kolmogorov-Smirnov test; none were normally distributed. Therefore, continuous variables were reported as medians and interquartile ranges (IQRs), and categorical variables as counts and percentages. The study population was divided into three groups based on cognitive status:Control (normal cognitive status)Mild Cognitive Impairment (MCI)Dementia Control (normal cognitive status) Control (normal cognitive status) Mild Cognitive Impairment (MCI) Mild Cognitive Impairment (MCI) Dementia Dementia Comparisons between the three groups Continuous variables were compared using the Kruskal–Wallis test. For variables found to be significant, post hoc pairwise comparisons were performed using the Bonferroni-corrected Mann–Whitney U test. No global correction for multiple comparisons across all laboratory variables was applied, given the exploratory nature of the analyses. Categorical variables were evaluated using the chi-square test or Fisher’s exact test, as appropriate. All p-values, whether significant or not, were reported in the tables. Multivariate logistic regression analyses Two separate logistic regression models were created to examine the predictors of cognitive impairment. Model A - factors determining dementia (dementia vs. control + MCI) A hierarchical (block entry) logistic regression three-stage model was established:Model 1 (Clinical model): Age, gender, Charlson Comorbidity Index (CCI).Model 2 (Clinical + Functional + Cognitive): Katz ADL, Lawton IADL, S-MMSE, MNA, GDS, Three Word Test.Model 3 (Laboratory + Resolvin D2-containing final model): Albumin, hemoglobin, ferritin, total cholesterol, iron, TDBK, vitamin B12, Resolvin D2. Model 1 (Clinical model): Age, gender, Charlson Comorbidity Index (CCI). Model 1 (Clinical model): Age, gender, Charlson Comorbidity Index (CCI). Model 2 (Clinical + Functional + Cognitive): Katz ADL, Lawton IADL, S-MMSE, MNA, GDS, Three Word Test. Model 2 (Clinical + Functional + Cognitive): Katz ADL, Lawton IADL, S-MMSE, MNA, GDS, Three Word Test. Model 3 (Laboratory + Resolvin D2-containing final model): Albumin, hemoglobin, ferritin, total cholesterol, iron, TDBK, vitamin B12, Resolvin D2. Model 3 (Laboratory + Resolvin D2-containing final model): Albumin, hemoglobin, ferritin, total cholesterol, iron, TDBK, vitamin B12, Resolvin D2. Model evaluation included the Hosmer–Lemeshow goodness-of-fit test, the − 2 log-likelihood, and the Cox–Snell and Nagelkerke R² values. Adjusted odds ratios (OR) and 95% confidence intervals were reported for all variables. Variables were selected a priori based on clinical relevance and prior literature. However, given the relatively high number of predictors compared to the number of events, these performance metrics may reflect model overfitting and should be interpreted with caution. Model B - factors determining cognitive impairment (MCI + dementia vs. control) The same model structure was used, and all variables were entered simultaneously using the enter (block entry) method rather than stepwise selection. ROC analyses ROC analyses were performed for the Charlson Comorbidity Index (CCI), S-MMSE, Lawton IADL, vitamin B12, and Resolvin D2 to evaluate their diagnostic performance. For both Model A (dementia discrimination) and Model B (cognitive impairment), area under the curve (AUC) values with 95% confidence intervals, sensitivity, specificity, and optimal cut-off values (determined using SPSS) were reported. ROC curves were presented as separate figures for each model (Figs. 1 and 2). Fig. 1ROC curves for Model A (Dementia) ROC curves for Model A (Dementia) Fig. 2ROC curves for Model B (Cognitive Impairment) ROC curves for Model B (Cognitive Impairment) To assess the risk of model overfitting, the events-per-variable (EPV) ratio was calculated. In Model A, with 60 dementia cases and multiple predictors included in the final model, the EPV was below the recommended threshold of 10, indicating a potential risk of overfitting. Multicollinearity among predictors was evaluated using variance inflation factors (VIF), with values < 10 considered acceptable. Sample and power analysis A priori power analysis was performed using G*Power 3.1.9.7 software to assess sample adequacy for the primary (dementia) and secondary (cognitive impairment) outcome variables of the study. For the Kruskal–Wallis test, which examined differences among three independent groups, a large effect size (f = 0.40) was assumed based on Cohen’s effect size classification [19]. When the significance level was set at α = 0.05 and the power at 0.80, the minimum required total sample size was 102 individuals. The final sample of 165 participants exceeded the minimum required and was sufficient to detect large effect sizes, although smaller effect sizes may not have been detected. Results A total of 165 older individuals were included in the study and were classified into three groups based on cognitive status: cognitively normal controls (n = 53), mild cognitive impairment (MCI; n = 52), and dementia (n = 60). Intergroup comparisons (Table 1) The median age of the dementia group was significantly higher than that of both the control and MCI groups (p < 0.001). Gender distribution did not differ significantly among groups (p = 0.629). Living arrangements differed significantly between groups (p < 0.001), with a higher proportion of participants in the dementia group living with their children. Table 1Demographic, geriatric assessment, and laboratory characteristics of the groupsVariableControl (n = 53)MCI (n = 52)Dementia (n = 60)p-valueDemographic Age (years), median (IQR)74 (11)ᵇᶜ78.5 (10)ᵃ82 (10)ᵃ< 0.001 Female, n (%)37 (22.4)32 (19.4)41 (24.8)0.629 Living in community, n (%)52 (31.5)50 (30.3)53 (32.1)0.068Living arrangement, n (%)< 0.001 Alone8 (4.8)18 (10.9)14 (8.5) With spouse27 (16.4)22 (13.3)15 (9.1) With children10 (6.1)11 (6.7)26 (15.8) With spouse and children7 (4.2)1 (0.6)0 (0) With caregiver1 (0.6)0 (0)5 (3)Education level, n (%)0.629 < 5 years32 (19.4)32 (19.4)42 (25.5) 6–8 years3 (1.8)5 (3)5 (3) > 8 years18 (10.9)15 (9.1)13 (7.9)Comorbidities DM, n (%)20 (12.1)24 (14.5)26 (15.8)0.673 HT, n (%)36 (21.8)37 (22.4)39 (23.6)0.785 OP, n (%)4 (2.4)7 (4.2)17 (10.3)0.010 Depression, n (%)7 (4.2)17 (10.3)28 (17)< 0.001 SVO, n (%)1 (0.6)2 (1.2)12 (7.3)< 0.001 Hypothyroidism, n (%)8 (4.8)13 (7.9)9 (5.5)0.305 CKD, n (%)4 (2.4)8 (4.8)13 (7.9)0.113 Smoking, n (%)5 (3)5 (3)1 (0.6)0.146 Alcohol, n (%)1 (0.6)0 (0)0 (0)0.349 CCI4 (2)ᵇᶜ5 (2)ᵃᶜ6.5 (3)ᵃᵇ< 0.001 Number of medications2 (1)ᶜ2 (1)ᶜ2 (0)ᵃᵇ0.002Geriatric assessment Katz ADL6 (1)ᶜ6 (1)ᶜ4 (3)ᵃᵇ< 0.001 Lawton IADL8 (1)ᶜ8 (2)ᶜ2 (4)ᵃᵇ< 0.001 S-MMSE29 (2)ᵇᶜ26 (3)ᵃᶜ19 (9)ᵃᵇ< 0.001 MNA13 (2)ᶜ12 (2)ᶜ10 (3)ᵃᵇ< 0.001 GDS3 (5)ᶜ2 (7)4 (6)ᵃ0.042 Three-word test2 (1)ᵇᶜ2 (1)ᵃᶜ0 (1)ᵃᵇ< 0.001Biomarker Resolvin D2 (ng/mL)1.94 (0.878)1.955 (0.745)2.211 (0.539)0.061Biochemistry FBG (mg/dL)94 (23)100 (49)99 (41)0.269 Urea (mg/dL)39.5 (17)45 (15)43 (17)0.111 Creatinine (mg/dL)0.88 (0.25)1.10 (0.32)1.01 (0.31)0.230 Na (mmol/L)141 (5)141 (3)141 (4)0.190 K (mmol/L)4.50 (0.6)4.55 (0.7)4.50 (0.7)0.501 Ca (mg/dL)9.60 (0.7)9.70 (1.1)9.50 (0.5)0.162 P (mg/dL)3.50 (0.6)3.40 (0.9)3.60 (0.9)0.683 ALT (U/L)20 (11)19 (10)16 (7)0.085 AST (U/L)15 (9)16 (9)17 (11)0.176 GGT (U/L)19.5 (13)19 (12)20 (14)0.834 ALP (U/L)78.5 (24)73 (29)75 (31)0.885 Total protein (g/L)69.5 (7)70 (6)68.5 (7)0.149 Albumin (g/L)44 (4)ᶜ44 (4)ᶜ42 (5)ᵃᵇ0.005 Total cholesterol (mg/dL)188 (56)ᶜ198 (53)ᶜ166 (68)ᵃᵇ0.018 LDL (mg/dL)114 (48)122 (50)95.5 (59)0.053 Triglycerides (mg/dL)117 (63)132 (94)121 (53)0.961 Iron (Fe) (µg/dL)69 (50)ᶜ65 (32)60 (42)ᵃ0.041 Iron saturation (%)22 (14)19 (12)22 (13)0.568 TIBC (µg/dL)321 (99)ᶜ326 (105)ᶜ292 (100)ᵃᵇ0.008 Ferritin (ng/mL)33 (34)ᶜ27 (48)ᶜ50.5 (85)ᵃᵇ0.028 Vitamin B12 (pg/mL)384 (195)ᶜ401 (235)477.5 (329)ᵃ0.039 Folic acid (ng/mL)11 (5)9.5 (5)9.5 (6)0.051 Vitamin D (ng/mL)70 (55)64.5 (60)47.5 (57)0.291 TSH (mIU/L)1.515 (1.31)1.565 (1.28)2.060 (1.34)0.077 Free T4 (ng/dL)1.195 (0.23)1.190 (0.33)1.110 (0.21)0.087Hematology Leukocyte (×10³/µL)6.92 (2.05)6.42 (2.32)6.93 (2.27)0.097 Neutrophil (×10³/µL)4.24 (1.62)3.94 (2.06)4.30 (2.40)0.122 Monocyte (×10³/µL)0.415 (0.18)0.380 (0.12)0.380 (0.16)0.653 Hemoglobin (g/dL)12.65 (1.6)ᶜ13.10 (2.4)13.10 (2.3)ᵃ0.026 Platelet (×10³/µL)257 (107)245 (113)242 (99)0.775 Sedimentation (mm/h)19 (25)20 (25)23 (31)0.521 CRP (mg/L)3.65 (9.9)3.50 (5.6)5.60 (14.9)0.409Continuous variables are presented as median (interquartile range, IQR), and categorical variables as number (percentage). Comparisons between the three groups were performed using the Kruskal–Wallis test for continuous variables and the chi-square test for categorical variables. For variables with overall statistical significance, pairwise comparisons were conducted using the Bonferroni-corrected Mann–Whitney U testSuperscript letters indicate pairwise differences:ᵃ significantly different from the control group,ᵇ significantly different from the MCI group,ᶜ significantly different from the dementia group Table 1Demographic, geriatric assessment, and laboratory characteristics of the groupsVariableControl (n = 53)MCI (n = 52)Dementia (n = 60)p-valueDemographic Age (years), median (IQR)74 (11)ᵇᶜ78.5 (10)ᵃ82 (10)ᵃ< 0.001 Female, n (%)37 (22.4)32 (19.4)41 (24.8)0.629 Living in community, n (%)52 (31.5)50 (30.3)53 (32.1)0.068Living arrangement, n (%)< 0.001 Alone8 (4.8)18 (10.9)14 (8.5) With spouse27 (16.4)22 (13.3)15 (9.1) With children10 (6.1)11 (6.7)26 (15.8) With spouse and children7 (4.2)1 (0.6)0 (0) With caregiver1 (0.6)0 (0)5 (3)Education level, n (%)0.629 < 5 years32 (19.4)32 (19.4)42 (25.5) 6–8 years3 (1.8)5 (3)5 (3) > 8 years18 (10.9)15 (9.1)13 (7.9)Comorbidities DM, n (%)20 (12.1)24 (14.5)26 (15.8)0.673 HT, n (%)36 (21.8)37 (22.4)39 (23.6)0.785 OP, n (%)4 (2.4)7 (4.2)17 (10.3)0.010 Depression, n (%)7 (4.2)17 (10.3)28 (17)< 0.001 SVO, n (%)1 (0.6)2 (1.2)12 (7.3)< 0.001 Hypothyroidism, n (%)8 (4.8)13 (7.9)9 (5.5)0.305 CKD, n (%)4 (2.4)8 (4.8)13 (7.9)0.113 Smoking, n (%)5 (3)5 (3)1 (0.6)0.146 Alcohol, n (%)1 (0.6)0 (0)0 (0)0.349 CCI4 (2)ᵇᶜ5 (2)ᵃᶜ6.5 (3)ᵃᵇ< 0.001 Number of medications2 (1)ᶜ2 (1)ᶜ2 (0)ᵃᵇ0.002Geriatric assessment Katz ADL6 (1)ᶜ6 (1)ᶜ4 (3)ᵃᵇ< 0.001 Lawton IADL8 (1)ᶜ8 (2)ᶜ2 (4)ᵃᵇ< 0.001 S-MMSE29 (2)ᵇᶜ26 (3)ᵃᶜ19 (9)ᵃᵇ< 0.001 MNA13 (2)ᶜ12 (2)ᶜ10 (3)ᵃᵇ< 0.001 GDS3 (5)ᶜ2 (7)4 (6)ᵃ0.042 Three-word test2 (1)ᵇᶜ2 (1)ᵃᶜ0 (1)ᵃᵇ< 0.001Biomarker Resolvin D2 (ng/mL)1.94 (0.878)1.955 (0.745)2.211 (0.539)0.061Biochemistry FBG (mg/dL)94 (23)100 (49)99 (41)0.269 Urea (mg/dL)39.5 (17)45 (15)43 (17)0.111 Creatinine (mg/dL)0.88 (0.25)1.10 (0.32)1.01 (0.31)0.230 Na (mmol/L)141 (5)141 (3)141 (4)0.190 K (mmol/L)4.50 (0.6)4.55 (0.7)4.50 (0.7)0.501 Ca (mg/dL)9.60 (0.7)9.70 (1.1)9.50 (0.5)0.162 P (mg/dL)3.50 (0.6)3.40 (0.9)3.60 (0.9)0.683 ALT (U/L)20 (11)19 (10)16 (7)0.085 AST (U/L)15 (9)16 (9)17 (11)0.176 GGT (U/L)19.5 (13)19 (12)20 (14)0.834 ALP (U/L)78.5 (24)73 (29)75 (31)0.885 Total protein (g/L)69.5 (7)70 (6)68.5 (7)0.149 Albumin (g/L)44 (4)ᶜ44 (4)ᶜ42 (5)ᵃᵇ0.005 Total cholesterol (mg/dL)188 (56)ᶜ198 (53)ᶜ166 (68)ᵃᵇ0.018 LDL (mg/dL)114 (48)122 (50)95.5 (59)0.053 Triglycerides (mg/dL)117 (63)132 (94)121 (53)0.961 Iron (Fe) (µg/dL)69 (50)ᶜ65 (32)60 (42)ᵃ0.041 Iron saturation (%)22 (14)19 (12)22 (13)0.568 TIBC (µg/dL)321 (99)ᶜ326 (105)ᶜ292 (100)ᵃᵇ0.008 Ferritin (ng/mL)33 (34)ᶜ27 (48)ᶜ50.5 (85)ᵃᵇ0.028 Vitamin B12 (pg/mL)384 (195)ᶜ401 (235)477.5 (329)ᵃ0.039 Folic acid (ng/mL)11 (5)9.5 (5)9.5 (6)0.051 Vitamin D (ng/mL)70 (55)64.5 (60)47.5 (57)0.291 TSH (mIU/L)1.515 (1.31)1.565 (1.28)2.060 (1.34)0.077 Free T4 (ng/dL)1.195 (0.23)1.190 (0.33)1.110 (0.21)0.087Hematology Leukocyte (×10³/µL)6.92 (2.05)6.42 (2.32)6.93 (2.27)0.097 Neutrophil (×10³/µL)4.24 (1.62)3.94 (2.06)4.30 (2.40)0.122 Monocyte (×10³/µL)0.415 (0.18)0.380 (0.12)0.380 (0.16)0.653 Hemoglobin (g/dL)12.65 (1.6)ᶜ13.10 (2.4)13.10 (2.3)ᵃ0.026 Platelet (×10³/µL)257 (107)245 (113)242 (99)0.775 Sedimentation (mm/h)19 (25)20 (25)23 (31)0.521 CRP (mg/L)3.65 (9.9)3.50 (5.6)5.60 (14.9)0.409Continuous variables are presented as median (interquartile range, IQR), and categorical variables as number (percentage). Comparisons between the three groups were performed using the Kruskal–Wallis test for continuous variables and the chi-square test for categorical variables. For variables with overall statistical significance, pairwise comparisons were conducted using the Bonferroni-corrected Mann–Whitney U testSuperscript letters indicate pairwise differences:ᵃ significantly different from the control group,ᵇ significantly different from the MCI group,ᶜ significantly different from the dementia group Demographic, geriatric assessment, and laboratory characteristics of the groups Continuous variables are presented as median (interquartile range, IQR), and categorical variables as number (percentage). Comparisons between the three groups were performed using the Kruskal–Wallis test for continuous variables and the chi-square test for categorical variables. For variables with overall statistical significance, pairwise comparisons were conducted using the Bonferroni-corrected Mann–Whitney U test Superscript letters indicate pairwise differences: ᵃ significantly different from the control group, ᵇ significantly different from the MCI group, ᶜ significantly different from the dementia group Regarding comorbidities, osteoporosis (p = 0.010), depression (p < 0.001), and stroke (p < 0.001) were more prevalent in the dementia group. The Charlson Comorbidity Index (CCI) increased progressively across groups and was significantly higher in the dementia group compared with the other groups (p < 0.001). In the comprehensive geriatric assessment, Katz Activities of Daily Living (ADL), Lawton Instrumental Activities of Daily Living (IADL), Standardized Mini-Mental State Examination (S-MMSE), Mini Nutritional Assessment (MNA), Geriatric Depression Scale (GDS), and the Three-Word Test scores differed significantly among the three groups (all p < 0.05), with lower functional and cognitive performance observed in the dementia group. Among laboratory parameters, serum albumin (p = 0.005), total cholesterol (p = 0.018), iron (p = 0.041), total iron-binding capacity (TIBC) (p = 0.008), ferritin (p = 0.028), and vitamin B12 levels (p = 0.039) showed significant differences between groups. In addition, no significant correlations were observed between circulating Resolvin D2 levels and cognitive, functional, or laboratory parameters (all p > 0.05). Although serum Resolvin D2 levels appeared numerically higher in the dementia group, this difference did not reach statistical significance (p = 0.061) and should be considered exploratory. Spearman correlation analysis (Supplementary Table 1) No significant correlations were observed between circulating Resolvin D2 levels and cognitive tests, functional status, nutritional status, depression scores, or biochemical parameters (all p > 0.05). Based on these findings, the independent effects of the variables on cognitive impairment were further evaluated using logistic regression analyses. Logistic regression analyses Model A (dementia vs. non-dementia (control + MCI)) (Table 2) In Model A, a three-stage logistic regression analysis (clinical → laboratory → Resolvin D2) was established to distinguish between dementia and non-dementia groups (control + MCI). Older age, higher comorbidity burden, poorer functional capacity, and lower cognitive performance were independently associated with dementia. Specifically, advanced age, higher Charlson Comorbidity Index, lower Lawton IADL scores, and lower S-MMSE scores remained significant predictors in the final model (Table 2). In contrast, Resolvin D2 levels did not contribute independently to the model (p = 0.385). Table 2Multivariable logistic regression analysis for dementia (Model A)PredictorBp-valueOR95% CI for ORAge (year)-0.4960.0470.609(0.37–0.99)Sex (male vs. female)1.3220.4493.749(0.08–168.3)CCI (Charlson Index)1.6570.0315.244(1.16–23.7)Katz ADL0.2700.7151.310(0.30–5.60)Lawton IADL-1.4510.0060.234(0.08–0.63)S-MMSE-0.7980.0230.450(0.23–0.89)MNA-0.2900.3880.748(0.39–1.42)GDS0.1140.6291.121(0.70–1.78)Three-Words Test-2.2100.0670.110(0.01–1.21)Albumin (g/dL)0.2550.3251.291(0.77–2.16)Hemoglobin (g/dL)0.7480.2331.992(0.59–6.73)Ferritin (ng/mL)0.0110.4691.011(0.98–1.04)Total Cholesterol (mg/dL)0.0060.6961.006(0.97–1.04)Iron (Fe) (µg/dL)0.0240.4341.024(0.96–1.09)TIBC (µg/dL)-0.0130.3540.987(0.96–1.01)Vitamin B12 (pg/mL)0.0050.0991.005(1.00–1.01)Resolvin D2 (ng/mL)0.2620.3851.300(0.72–2.35) Table 2Multivariable logistic regression analysis for dementia (Model A)PredictorBp-valueOR95% CI for ORAge (year)-0.4960.0470.609(0.37–0.99)Sex (male vs. female)1.3220.4493.749(0.08–168.3)CCI (Charlson Index)1.6570.0315.244(1.16–23.7)Katz ADL0.2700.7151.310(0.30–5.60)Lawton IADL-1.4510.0060.234(0.08–0.63)S-MMSE-0.7980.0230.450(0.23–0.89)MNA-0.2900.3880.748(0.39–1.42)GDS0.1140.6291.121(0.70–1.78)Three-Words Test-2.2100.0670.110(0.01–1.21)Albumin (g/dL)0.2550.3251.291(0.77–2.16)Hemoglobin (g/dL)0.7480.2331.992(0.59–6.73)Ferritin (ng/mL)0.0110.4691.011(0.98–1.04)Total Cholesterol (mg/dL)0.0060.6961.006(0.97–1.04)Iron (Fe) (µg/dL)0.0240.4341.024(0.96–1.09)TIBC (µg/dL)-0.0130.3540.987(0.96–1.01)Vitamin B12 (pg/mL)0.0050.0991.005(1.00–1.01)Resolvin D2 (ng/mL)0.2620.3851.300(0.72–2.35) Multivariable logistic regression analysis for dementia (Model A) The overall fit of the model was high (Nagelkerke R² = 0.912), with a classification accuracy of 97.2%. However, given the relatively high number of predictors compared to the number of events, these performance metrics may reflect model overfitting and should be interpreted with caution. Multicollinearity analysis showed that all variables had acceptable VIF values (range: 1.10–2.96), indicating no significant multicollinearity. Parsimonious model analysis To address potential overfitting, a simplified logistic regression model was constructed including age, sex, Charlson Comorbidity Index, and Resolvin D2. In this model, Resolvin D2 was not independently associated with dementia (p = 0.227), while the Charlson Comorbidity Index remained a significant predictor (p < 0.001). Age (p = 0.144) and sex (p = 0.224) were not significantly associated with dementia in this simplified model. These findings were consistent with those of the full model. This approach also minimized potential circularity bias arising from the inclusion of cognitive test variables. Model B (cognitive impairment = MCI+dementia vs. control) (Table 3) In Model B, greater comorbidity burden and lower cognitive performance emerged as independent predictors of cognitive impairment. Higher Charlson Comorbidity Index scores and lower S-MMSE scores were significantly associated with cognitive impairment, while albumin and vitamin B12 were also independently associated. Resolvin D2 was not associated with cognitive impairment. Table 3Multivariable Logistic Regression Analysis for Cognitive Impairment (Model B)PredictorBp-valueOR95% CI for ORAge (year)0.1140.1181.1210.97–1.29Sex (male vs. female)0.8350.3432.3060.41–12.96CCI (Charlson Index)0.8880.0042.4301.32–4.48Katz ADL0.9120.1032.4900.83–7.46Lawton IADL–0.1970.4260.8220.51–1.33S-MMSE–1.6200.0000.1980.09–0.44MNA0.1600.4771.1740.76–1.82GDS0.2020.1251.2230.95–1.58Three-Words Test0.1440.7851.1550.41–3.25Albumin (g/dL)0.3010.0491.3521.00–1.83Hemoglobin (g/dL)0.1940.5561.2140.64–2.31Ferritin (ng/mL)0.0100.3471.0100.99–1.03Total Cholesterol (mg/dL)0.0170.1221.0171.00–1.04Iron (Fe) (µg/dL)0.0170.2161.0170.99–1.04TIBC (µg/dL)0.0090.1801.0091.00–1.02Vitamin B12 (pg/mL)0.0050.0341.0051.00–1.01Resolvin D2 (ng/mL)–0.2100.1780.8100.60–1.10 Table 3Multivariable Logistic Regression Analysis for Cognitive Impairment (Model B)PredictorBp-valueOR95% CI for ORAge (year)0.1140.1181.1210.97–1.29Sex (male vs. female)0.8350.3432.3060.41–12.96CCI (Charlson Index)0.8880.0042.4301.32–4.48Katz ADL0.9120.1032.4900.83–7.46Lawton IADL–0.1970.4260.8220.51–1.33S-MMSE–1.6200.0000.1980.09–0.44MNA0.1600.4771.1740.76–1.82GDS0.2020.1251.2230.95–1.58Three-Words Test0.1440.7851.1550.41–3.25Albumin (g/dL)0.3010.0491.3521.00–1.83Hemoglobin (g/dL)0.1940.5561.2140.64–2.31Ferritin (ng/mL)0.0100.3471.0100.99–1.03Total Cholesterol (mg/dL)0.0170.1221.0171.00–1.04Iron (Fe) (µg/dL)0.0170.2161.0170.99–1.04TIBC (µg/dL)0.0090.1801.0091.00–1.02Vitamin B12 (pg/mL)0.0050.0341.0051.00–1.01Resolvin D2 (ng/mL)–0.2100.1780.8100.60–1.10 Multivariable Logistic Regression Analysis for Cognitive Impairment (Model B) The model demonstrated high classification accuracy (94.4%); however, this finding should be interpreted with caution, given the potential risk of overfitting. ROC analyses In the ROC analysis performed in Model A, the Charlson Comorbidity Index (CCI) was the strongest predictor for distinguishing dementia (AUC = 0.818, 95% CI: 0.754–0.883), indicating good discriminatory performance. In contrast, vitamin B12 (AUC = 0.539, 95% CI: 0.446–0.632) and Resolvin D2 (AUC = 0.610, 95% CI: 0.523–0.698) demonstrated low discriminatory ability. These findings suggest that clinical markers may be more informative than biochemical markers in distinguishing dementia (Table 4). Table 4ROC Curve Analysis for Model-A and Model-BModelVariableAUCBest Cut-offSensitivity1 - SpecificityModel-ACCI0.8204.50.8980.442B120.618283 pg/mL0.8980.827Resolvin D20.6031.745 ng/mL0.9350.874Model-BCCI0.7914.50.7360.340B120.584345.5 pg/mL0.7090.585Resolvin D20.5471.745 ng/mL0.8000.736Optimal cut-off values have been determined based on the Youden index Table 4ROC Curve Analysis for Model-A and Model-BModelVariableAUCBest Cut-offSensitivity1 - SpecificityModel-ACCI0.8204.50.8980.442B120.618283 pg/mL0.8980.827Resolvin D20.6031.745 ng/mL0.9350.874Model-BCCI0.7914.50.7360.340B120.584345.5 pg/mL0.7090.585Resolvin D20.5471.745 ng/mL0.8000.736Optimal cut-off values have been determined based on the Youden index ROC Curve Analysis for Model-A and Model-B Optimal cut-off values have been determined based on the Youden index Similarly, in Model B, CCI remained the strongest marker for distinguishing cognitive impairment (AUC = 0.789, 95% CI: 0.717–0.861). In contrast, vitamin B12 (AUC = 0.570, 95% CI: 0.481–0.660) and Resolvin D2 (AUC = 0.550, 95% CI: 0.455–0.646) showed low discriminatory ability. The consistently low AUC values of Resolvin D2 across both models suggest that this biomarker provides limited diagnostic discrimination. Despite the calculation of optimal cutoff values, the overall low AUCs and overlapping confidence intervals indicate limited clinical applicability of these biomarkers. Discussion To our knowledge, this is the first clinical study to evaluate the relationship between cognitive status and Resolvin D2 levels in older adults. In a cohort including cognitively normal controls, MCI, and dementia, cognitive, functional, laboratory, and biomarker parameters were comprehensively assessed. Resolvin D2 was not independently associated with dementia or cognitive impairment. In contrast, established clinical and functional indicators (CCI, S-MMSE, and Lawton IADL) emerged as stronger predictors, while Resolvin D2 did not provide additional independent predictive value. Consistent with these findings, no significant correlations were observed between Resolvin D2 levels and cognitive, functional, or laboratory parameters. It is also important to note that in multivariable models, strong clinical predictors such as cognitive performance (S-MMSE), functional status, and comorbidity burden (CCI) may overshadow smaller biomarker effects, potentially masking subtle associations of Resolvin D2 with cognitive impairment. This interpretation is also supported by ROC analyses, which show that clinical predictors demonstrated higher discriminative performance than Resolvin D2. Overall, these findings suggest that the role of inflammation in cognitive impairment may be more complex in older adults and that the clinical utility of Resolvin D2 as a biomarker remains limited. Resolvin D2 has been shown to suppress pro-inflammatory cytokines in monocytes following TLR4-mediated stimulation and to increase the expression of anti-inflammatory cytokines such as IL-10. Furthermore, it activates the PI3K/Akt pathway, promoting GSK3β phosphorylation-dependent inactivation and supporting CREB-mediated transcription of anti-inflammatory genes, suggesting a potential regulatory role in chronic inflammatory conditions [20, 21]. Experimental studies further support the biological relevance of RvD2. It has been detected in the brains of stroke-exposed mice, suggesting a role in the resolution of ischemic injury [22]. In addition, RvD2 has been shown to reverse LPS-induced depressive-like behaviors via the mTORC1 signaling pathway and to regulate chronic inflammation in adipose tissue and atherosclerosis models [23, 24]. Consistent with these findings, RvD2 has been demonstrated to limit the progression of atherosclerosis by suppressing inflammatory lipid mediators, reducing macrophage accumulation, and increasing plaque stability [25]. Despite strong experimental and mechanistic evidence supporting the anti-inflammatory and pro-resolving effects of Resolvin D2, the present findings suggest that these effects may not be reflected in measurable differences in circulating RvD2 levels across clinical stages of cognitive impairment. This discrepancy may reflect disease-stage specificity, compartmentalization of resolvin activity within the central nervous system, or compensatory inflammatory responses that are not captured by single-time-point serum measurements. Furthermore, the extent to which circulating Resolvin D2 levels reflect central nervous system inflammatory or resolution processes remains uncertain. Peripheral and central immune responses are partially compartmentalized, and blood–brain barrier dynamics may limit the direct translation of systemic measurements to brain-specific processes [26]. This may partly explain the absence of significant associations observed in the present study. Although the absence of a significant association may reflect a true lack of effect, alternative explanations should also be considered. Peripheral measurements of Resolvin D2 may not fully capture central nervous system processes, and assay-related variability, as well as pre-analytical factors, may have influenced circulating levels. In addition, the cross-sectional design and single-time-point measurement limit the ability to detect dynamic changes in inflammation-resolution pathways [10]. Although the study included clinically well-defined and distinct cognitive groups, enabling meaningful cross-sectional comparisons, inflammation-resolution pathways are inherently dynamic. Therefore, the absence of a significant association may partly reflect the inability to detect time-dependent changes rather than a true lack of biological relevance. Finally, although the study was powered to detect large effects, it may be underpowered to identify smaller differences. Previous studies have explored the relationship between Resolvin D2 and cognitive function. One study demonstrated a direct correlation between postmortem brain and cerebrospinal fluid (CSF) RvD2 levels and cognitive scores in patients with Alzheimer's disease [27]. In addition, in aged rats treated with docosahexaenoic acid, improved memory was observed alongside increased brain RvD2 levels [28]. The anti-inflammatory effects of RvD2 are thought to be mediated by the G protein–coupled receptor GPR18, although the underlying mechanisms remain unclear [29, 30]. Similarly, a study investigating plasma lipoxin A4 (LXA4) levels in post-stroke cognitive impairment (PSCI) reported reduced plasma LXA4 levels and LXA4/LTB4 ratios, which were associated with cognitive performance as measured by S-MMSE. However, when other specialized pro-resolving mediators (SPMs), including RvD1, RvD2, and MaR1, were analyzed, no significant differences were observed between PSCI and post-stroke non-cognitive impairment (PSNCI) groups, consistent with our findings [31]. These results suggest that impaired inflammation resolution may contribute to the pathogenesis of cognitive impairment, while different SPMs may have distinct and context-specific roles. Given that neuroinflammation is a key pathological component in Alzheimer's disease, Resolvin D2 was investigated based on the hypothesis that its resolution-based mechanisms may be involved in the disease process. Despite this theoretical and experimental background, the lack of a significant association observed in our study suggests that the role of Resolvin D2 may depend on disease stage, cellular context, or distinct regulatory mechanisms under in vivo conditions. These findings further indicate that inflammation-resolution pathways in Alzheimer's disease are shaped by complex and context-specific mechanisms. This study has several important strengths. First, it is among the few studies evaluating the relationship between Resolvin D2 levels and cognitive status in older adults, contributing to the limited clinical literature in this field. Second, the sample was structured to include three distinct cognitive groups (healthy controls, MCI, and dementia), allowing comparisons across stages of cognitive impairment within the same study and enhancing the clinical interpretability of the results. Furthermore, a multidimensional dataset was established, including assessments of cognitive performance, functional status, nutritional status, mood, and extensive laboratory parameters, enabling more effective control of potential confounding variables. Finally, the use of both multivariate logistic regression and ROC analyses strengthens the consistency and methodological robustness of the findings. This study has several limitations. First, the cross-sectional design precludes causal inferences regarding whether Resolvin D2 levels are a cause or a consequence of cognitive impairment and limits the ability to capture temporal variability in inflammation-resolution pathways. Given the dynamic nature of these pathways, single-time-point measurements may not fully reflect fluctuations across different stages of cognitive impairment. Second, the study was conducted at a single center with a relatively modest sample size, which may limit generalizability. In addition, the MCI group was not stratified into amnestic and non-amnestic subtypes, which may have contributed to clinical heterogeneity. Third, circulating lipid mediators such as Resolvin D2 are sensitive to pre-analytical conditions. In this study, factors such as fasting status, timing of blood collection, dietary omega-3 intake, and anti-inflammatory medication use were not systematically standardized, which may have contributed to variability in RvD2 levels and attenuated detectable group differences. Fourth, in the regression analyses, the relatively large number of predictors compared to the number of events may have increased the risk of overfitting. Although multicollinearity was not detected, model performance should be interpreted with caution. However, sensitivity analyses using a parsimonious model yielded consistent results, supporting the robustness of the findings. Finally, the a priori power analysis was based on a large effect size (f = 0.40), and the study may be underpowered to detect small-to-moderate differences in Resolvin D2 levels. In addition, no global correction for multiple comparisons across laboratory variables was applied, which may increase the risk of type I error. Clinical implications and future recommendations This study suggests that Resolvin D2 is not independently associated with cognitive impairment in older adults and does not provide additional diagnostic value beyond established clinical and functional measures. Clinical assessments, including S-MMSE and Lawton IADL, as well as comorbidity burden (CCI), remain more informative indicators of cognitive impairment. Accordingly, the clinical utility of Resolvin D2 as a standalone screening or diagnostic biomarker appears limited at this stage. These findings highlight that single peripheral biomarkers may have restricted value when used in isolation. For future research, studies with larger sample sizes and prospective, longitudinal designs are warranted, particularly those that capture the temporal and stage-specific dynamics of inflammation-resolution pathways. These should incorporate multi-biomarker approaches that integrate peripheral and central nervous system–specific measures, such as cerebrospinal fluid markers, along with inflammatory and resolution-related pathways, to more comprehensively assess neuroinflammatory processes and their relationship with cognitive decline. Such approaches would provide clearer insight into the temporal dynamics and clinical relevance of specialized pro-resolving mediators in older adults. Conclusion This study demonstrated that serum Resolvin D2 levels have limited ability to distinguish older adults with dementia or cognitive impairment from cognitively normal controls. In contrast, clinical measures of cognitive and functional status, including S-MMSE and Lawton IADL scores, as well as comorbidity burden assessed by the Charlson Comorbidity Index, were independent predictors of cognitive status. Although Resolvin D2 has well-established anti-inflammatory and pro-resolving properties in experimental models, our findings suggest that its clinical utility as a standalone biomarker for cognitive impairment in older adults is limited. Future longitudinal studies incorporating larger cohorts and multi-biomarker panels of specialized pro-resolving mediators may provide further insight into the role of inflammation-resolution pathways in cognitive decline. Supplementary Information Supplementary Material 1. Supplementary Material 1.