Relationship of Estimated Glomerular Filtration Rate (eGFR) and Neuron-Specific Enolase (NSE) with Cognitive Dysfunction in Type 2 Diabetes Mellitus: A Cross-Sectional Analysis

Abstract

Background

Diabetic encephalopathy is a common complication of type 2 diabetes mellitus (T2DM) and a major cause of cognitive dysfunction. The kidneys and the brain share several similar physiological characteristics, which may render them susceptible to analogous pathological processes. Estimated glomerular filtration rate (eGFR) is a widely used indicator of kidney function and may reflect systemic microvascular injury, impaired renal function accelerates cerebral microangiopathy and exacerbates cognitive deterioration. Neuron-specific enolase (NSE) is a well-recognized neuronal injury biomarker, It is a glycolytic rate-limiting enzyme abundant in cerebral neuron cytoplasm, leaks into blood or cerebrospinal fluid upon neuronal membrane damage following central nervous system injury. Existing studies separately explored renal function or neuronal injury indicators in diabetic cognitive dysfunction, while few studies combined eGFR and NSE to jointly predict cognitive dysfunction in T2DM populations, lacking combined diagnostic evidence of kidney-brain damage markers.

Purpose

This study aimed to investigate the associations of eGFR and NSE with cognitive function in T2DM patients and evaluate their predictive value for cognitive dysfunction.

Methods

This cross-sectional study retrospectively enrolled 80 T2DM patients. Demographic, clinical, and laboratory data were collected. Serum NSE and eGFR were assessed. Cognitive function was evaluated using the Montreal Cognitive Assessment (MoCA). Patients were categorized into diabetic cognitive dysfunction group (DCD, MoCA <26, n=40) and Diabetic non-cognitive dysfunction group (DNCD, MoCA ≥26, n=40) groups.

Results

Compared with the DNCD group, the DCD group had significantly lower eGFR and higher NSE, HbA1c, and Hcy levels (all P<0.05). Pearson correlation analysis in the DNCD group demonstrated that eGFR was positively correlated with MoCA scores. In contrast, NSE, HbA1c and Hcy were all negatively correlated with MoCA scores (all P < 0.05). Multivariate regression further confirmed that NSE, glycated hemoglobin, and homocysteine were independently negatively associated with cognitive function, while eGFR was independently positively associated with cognitive function. ROC curve analysis revealed that both eGFR and NSE contribute to the diagnosis of cognitive dysfunction.

Conclusion

Elevated NSE, HbA1c, and Hcy are independently negatively associated with cognitive function, while eGFR was independently positively associated with cognitive function in T2DM patients. Combined detection of eGFR and NSE facilitates the early screening of diabetic cognitive dysfunction. This study compensates for the deficiency of single-organ biomarker research and provides provide a preliminary theoretical reference for future joint prediction model development.

Introduction

With rapid economic development and changes in lifestyle, T2DM has evolved into a prominent global public health concern because of its widespread occurrence and associated complications.¹ It is projected that by 2050, the prevalence rate will rise to 12.96%, with the total number of patients reaching 853 million.² In China, the number of diabetes patients was estimated at 233.03 million in 2023, with a prevalence rate of 15.88%, It is projected to reach 16.15% by 2030.³ Among the complications of diabetes mellitus, Diabetic encephalopathy is a severe and increasingly prevalent neurological disorder.⁴ T2DM is being increasingly acknowledged as a notable risk factor for cognitive dysfunction.^5,6 The progression rate from mild cognitive impairment (MCI) to dementia occurring 1.5 to 3.0 times more frequently in T2DM patients compared to non-T2DM individuals.⁷ The Framingham Heart Study found a 50–100% increased dementia risk in T2DM patients.⁸ Due to the lack of effective treatments for severe cognitive dysfunction, early diagnosis, intervention, and regular cognitive dysfunction screening in T2DM patients are crucial for clinical and outcomes.

Studies have shown that the kidneys and the brain share certain physiological characteristics, which may render them susceptible to similar pathological processes.⁹ eGFR is a clinically validated indicator for assessing renal function, reflecting the status of renal vascular microcirculation. Increasing clinical evidence suggests that chronic kidney injury can also accelerate the onset and progression of cognitive dysfunction.^10,11 A prospective cohort study by Murray et al demonstrated a correlation between eGFR and cognitive function.¹² A meta-analysis of 32,141 participants revealed that cognitive dysfunction prevalence rises as eGFR decreases, with T2DM as a significant risk factor.¹³

NSE is a rate-limiting enzyme in glycolysis, abundantly present in the cytoplasm of brain neurons. When the central nervous system is damaged, the integrity of neuronal cell membranes is compromised, leading to the leakage and release of NSE into the bloodstream or cerebrospinal fluid. NSE can serve as a biomarker reflecting damage to neural tissues.¹⁴ Studies have demonstrated a close association between NSE and cognitive dysfunction. Elsharkawy et al found that patients with diabetic cognitive dysfunction exhibited significantly elevated levels of NSE.¹⁵ Yu et al reported significantly increased NSE levels in patients with diabetic cognitive dysfunctionNSE levels are identified as an independent risk factor for mild cognitive dysfunction in individuals with diabetes.¹⁶

However, the relationship between eGFR, NSE, and cognitive function remains uncertain. For instance, the ADNI cohort study demonstrated that baseline eGFR in elderly individuals with mild to moderate eGFR reduction showed no association with cognitive decline.¹⁷ Gul et al also drew a conclusion that there was no correlation between renal function and cognitive function,¹⁸ Ma et al reported that no statistically significant discrepancy in NSE levels between elderly T2DM patients with cognitive dysfunction and their counterparts without cognitive dysfunction.¹⁹

The unclear and debated connections between eGFR, NSE, and cognitive function in T2DM patients necessitate further research into the links between eGFR, NSE, and cognitive function. Most current studies only focus on the independent effect of a single indicator on cognitive function, while few studies have systematically explored the combined predictive value of peripheral renal function index eGFR and neural biomarker NSE for cognitive dysfunction in T2DM patients. This dual-indicator joint analysis can integrate peripheral microcirculation renal injury and neuronal pathological damage, which is more comprehensive than single-indicator evaluation. The primary objective of this study was to explore the independent and combined correlation of eGFR and serum NSE levels with cognitive function in T2DM patients, and further verify their clinical predictive value for diabetic cognitive dysfunction. Our core hypothesis is that decreased eGFR and elevated NSE levels are closely associated with the decline of cognitive function in T2DM patients, and the combined detection of the two indicators has higher clinical application value for the early identification of diabetic cognitive impairment than single indicator detection.

Materials and Methods

Estimation of Sample Size

This research was designed as a retrospective cross-sectional study. The primary outcome measure was NSE, and the sample size was calculated with the use of PASS 15 software (NCSS, Kaysville, UT, USA). On the basis of the pilot experiment, the value of NSE was 10.160 ± 2.790 ng/mL in the DNCD group versus 12.780 ± 3.170 ng/mL in the DCD group. With a two-tailed significance level of α=0.05 and a power of 0.90, the Two-Sample T-Tests Assuming Equal Variance in PASS 15 software indicated that each group needed at least 32 enrolled cases. Considering a 20% dropout rate, a minimum of 40 cases per group were required, totaling 80 cases. The sample size calculation formula was: . Here, δ represents the mean difference between the two groups, and σ denotes the standard deviation.

Subjects

This study enrolled 80 hospitalized patients with T2DM admitted between October 2025 and April 2026 as study subjects. All participants met the diagnostic criteria proposed by the World Health Organization (WHO) in 1999.²⁰ No acute or chronic diabetic complications were present. To minimize the impact of educational level on cognitive function assessment, participants with an education duration of ≥6 years (primary school) and ≤16 years (university) were selected.

Exclusion criteria included: concomitant other neurological disorders causing cognitive dysfunction (cerebrovascular diseases, central nervous system demyelinating and degenerative diseases, epilepsy, Parkinson’s disease, tumors, etc); history of psychiatric or psychological disorders; severe smoking, alcohol dependence, or drug dependence; severe cardiac, pulmonary, hepatic, or renal dysfunction, as well as hematopoietic system diseases (severe hemolytic anemia etc); thyroid dysfunction; toxic metabolic encephalopathy; and deficiencies in folic acid or vitamin B12;trauma, rhabdomyolysis. To enhance the generalizability of the study findings, no specific restrictions were imposed on medication use.

Clinical Data Collection

Body height and weight measurements, as well as derived body mass index (BMI) values, were documented in medical charts. Baseline demographic data including sex, age, educational attainment and medical history were all archived within electronic medical records. All patients underwent a standard 12-hour overnight fast prior to blood pressure examinations performed at 8:00 a.m. during hospitalization. Fasting cubital venous blood specimens were routinely obtained for complete blood count and biochemical testing, with recorded indicators comprising neutrophil count, lymphocyte count, fasting plasma glucose (FPG), triglycerides (TG), total cholesterol (TC), low-density lipoprotein (LDL), uric acid (UA) and glycated hemoglobin (HbA1c). The neutrophil-to-lymphocyte ratio (NLR) was further calculated based on archived laboratory results.

The creatinine (Cr) levels of all subjects were measured using the creatine oxidase method with a Roche Cobas C702 biochemical analyzer and its corresponding reagent kit. NSE detection was performed via electrochemiluminescence assay using a Roche Cobas ISE analyzer and its matching reagent kit. All procedures followed the manufacturer’s instructions precisely.

All serum biochemical indicators and neuron-specific enolase (NSE) were measured in the standardized clinical laboratory of the Department of Clinical Laboratory in our hospital. Strict internal quality control protocols were followed throughout all testing procedures. Two levels of quality control materials (low and high concentration) were tested before daily sample detection. Test data of the day were only eligible for analysis if quality control results fell within the allowable error range (±2 standard deviations) without drift or trending shifts. The analytical precision of the detection system was as follows: the intra-assay coefficient of variation (CV) for all indicators was less than 5%, and the inter-assay CV was less than 6%. The analytical instrument was calibrated routinely with original manufacturer calibrators for traceability once per month. Immediate recalibration was conducted after instrument maintenance, reagent batch replacement, or out-of-control quality control results. The uniform reference ranges adopted by our laboratory were listed below: NSE: 0–16.3 ng/mL; fasting plasma glucose: 3.9–6.1 mmol/L; TG: 0–1.7 mmol/L; TC: 2.8–5.2 mmol/L; LDL: <3.4 mmol/L; UA: 208–428 μmol/L. Routine blood parameters including neutrophil and lymphocyte counts, as well as glycated hemoglobin, were interpreted according to the adult reference intervals of our clinical laboratory. All testing procedures, calibration and quality control standards complied with the ISO15189 laboratory quality specifications.

Calculation of eGFR: The eGFR was calculated using the simplified MDRD formula (2000 edition), with the specific equation as follows: eGFR [mL/(min·1.73m²)] = 186 × serum creatinine (mg/dL) −1.154 × age (years) −0.203 × (0.742 for females).

Cognitive Function Assessment

This study employed the Montreal Cognitive Assessment (MoCA)²¹ to evaluate the overall cognitive function of participants. The MoCA was used with permission for non-commercial academic research. According to educational attainment criteria, MoCA scores ranged from 0 to 30 points, with a score <26 points indicating cognitive dysfunction. The original MoCA and its Chinese version have been validated and widely used in Chinese populations. To address cultural bias, Participants with an educational duration of less than 12 years are awarded an extra point in their assessment outcomes.

All raters received standardized unified training on MoCA administration and mastered the scoring criteria thoroughly. They were only allowed to conduct assessments after passing a standardized post-training examination. A standardized testing protocol was established with fixed assessment environments, standardized verbal instructions and uniform completion time limits to ensure consistent administration procedures for all participants. All evaluators were kept blinded to patients’ laboratory test results and group assignments throughout the assessment. No medical records including biochemical and renal function data were accessed during scoring, and cognitive ratings were completed independently.

Group

According to MoCA scores, patients were categorized into the diabetic cognitive dysfunction group (DCD group, MoCA<26 points, n=40) and the diabetic non-cognitive dysfunction group (DNCD group, MoCA ≥ 26 points, n=40) based on a cutoff score of 26. To equalize the sample sizes of the DCD and DNCD groups and minimize selection bias, we established the two cohorts using retrospective paired matching. Specifically, we first retrieved all patients who met the diagnostic criteria for DCD, and then screened corresponding DNCD cases in a 1:1 ratio matched for confounders including age and disease duration, so that the two groups ended up with an identical number of participants. Among the DNCD group, there were 26 males and 14 females, with an average age of 52.200 ± 11.301 years and an average education duration of 10.900 ± 2.951 years. In the DCD group, there were 28 males and 12 females, with a mean age of 56.150 ± 8.790 years and a mean education duration of 10.800 ± 2.954 years.

Statistical Analysis

All statistical analyses were performed using SPSS version 26.0 (IBM Corp., Armonk, NY, USA). A complete case analysis was adopted for data preprocessing. All subjects with missing data for any indicator (including eGFR, NSE, MoCA score and confounding covariates) were excluded, and only participants with complete clinical and laboratory data were included for final statistical analysis. The overall missing rate of all variables was less than 5%. After normality testing, measured data were expressed as mean ± standard deviation (SD) if following a normal distribution. Between-group comparisons were performed using independent samples t-test. Categorical data were presented as percentages (%) and compared using chi-square test. Pearson correlation analysis was used to examine the relationships among eGFR, NSE levels, and MoCA scores, while logistic regression assessed the association between these variables and the risk of cognitive dysfunction. Receiver Operating Characteristic (ROC) curve analysis was employed to evaluate the predictive value of eGFR and NSE for cognitive dysfunction, and a P-value of less than 0.05 was considered statistically significant.

Ethics

This study was approved by the Ethics Committee of Jinan Fourth People’s Hospital (approval number: LL20240014). It complied with the principles specified in the Declaration of Helsinki. All participating patients provided informed consent and signed the corresponding informed consent form.

Results

Baseline Characteristics

For measurement data, the Shapiro–Wilk test was applied to judge whether the data conformed to a normal distribution, and the Levene test was used to assess the homogeneity of variance between groups. The baseline indicators compared in this study covered gender, age, years of education, duration of diabetes, prevalence of coronary heart disease, prevalence of carotid atherosclerosis, proportion of patients treated with metformin and insulin, body mass index (BMI), systolic blood pressure (SBP), diastolic blood pressure (DBP), neutrophil-to-lymphocyte ratio (NLR), serum uric acid, low-density lipoprotein cholesterol (LDL), total cholesterol (CHO) and triglycerides (TG). All the above indicators yielded P > 0.05, indicating normal data distribution and homogeneity of variance across groups.

The comparison of baseline characteristics between the two patient groups revealed no statistically significant differences in gender, age, educational background, diabetes duration, prevalence of coronary heart disease, carotid artery atherosclerosis, or medication usage rates for metformin and insulin (all P>0.05). Additionally, no significant differences were observed in physical examination indicators such as BMI, systolic blood pressure (SBP), and diastolic blood pressure (DBP), nor in laboratory indicators including neutrophil-to-lymphocyte ratio (NLR), uric acid, LDL, CHO, and TG levels (all P>0.05). However, the DCD group exhibited significantly lower eGFR compared to the DNCD group (t=−3.055, P=0.003) and significantly higher NSE levels (t=2.947, P=0.004). Additionally, the DCD group had higher HbA1c levels (t=2.093, P=0.040) and higher Hcy levels (t=2.386, P=0.019) than the DNCD group. Cognitive function assessment showed that MoCA scores were notably lower in the DCD group than in the DNCD group, with a significant difference (t=−18.719, P<0.001). See Table 1.

Table 1.

Comparison of Baseline Characteristics of Enrolled Subjects

Variables	DNCD	DCD	t/χ2	P-value
Male/Female	26/14	28/12	0.228	0.633
Age(Years)	52.200±11.301	56.15±8.790	1.745	0.085
Years of education	10.900±2.951	10.800±2.954	−0.151	0.880
Duration of T2DM (Years)	4.550±2.241	4.65±2.486	−0.189	0.851
Coronary disease (%)	18 (45%)	22 (55%)	0.800	0.371
Carotid atherosclerosis (%)	22 (55%)	27 (67.5%)	1.317	0.251
Metformin use (%)	26 (65%)	22 (55%)	0.833	0.361
Insulin use (%)	25 (62.5%)	20 (50%)	1.270	0.260
BMI (kg/m2)	25.925±3.918	26.825±3.707	1.055	0.295
SBP (mmHg)	133.78±11.269	132.38±12.372	−0.529	0.598
DBP (mmHg)	82.180±8.650	85.880±8.259	1.957	0.054
NLR (%)	2.661±2.047	2.375±1.179	−0.766	0.446
eGFR (mL/min/1.73 m2)	139.730±47.665	112.20±31.214	−3.055	0.003
UA (umol/L)	295.95±99.390	292.75±78.475	−0.160	0.873
NSE (ng/mL)	10.158±1.800	12.803±5.384	2.947	0.004
LDL (mmol/L)	2.875±1.137	2.783±.794	−0.422	0.674
HbA1c (%)	9.150±2.704	10.480±2.953	2.093	0.040
CHO (mmol/L)	4.900±1.429	5.178±1.614	0.814	0.418
TG (mmol/L)	2.563±4.996	3.100±3.678	0.548	0.585
Hcy (μmol/L)	12.445±6.437	16.058±7.091	2.386	0.019
MoCA scores	28.150±1.406	19.50±2.562	−18.719	0.000

Notes: Measured data were expressed as mean ± standard deviation (SD). Between-group comparisons were performed using independent samples t-test. Categorical data were presented as percentages (%) and compared using chi-square test. P < 0.05 was considered statistically significant.

Pearson Correlation Analysis

A Pearson correlation analysis was performed on the DCD group, focusing on indicators that showed statistically significant discrepancies (P<0.05) in baseline parameter comparisons. The results demonstrated that NSE (r=−0.351, P=0.026), HbA1c (r=−0.386, P=0.013), and Hcy (r=−0.344, P=0.030) were negatively correlated with MoCA scores, while eGFR (r=0.394, P=0.012) showed a positive correlation with MoCA scores. See Figure 1

Figure 1.

Scatter plots of correlations between clinical indicators and MoCA cognitive scores.

Notes: Pearson correlation analysis was used to examine the relationships among eGFR, NSE, HbA1c, Hcy levels, and MoCA scores. Each panel presents the linear regression fit and bivariate correlation: (A) Positive correlation between eGFR and MoCA scores (r = 0.394, P = 0.012). (B) Negative correlation between NSE and MoCA scores (r = −0.351, P = 0.026). (C) Negative correlation between HbA1c and MoCA scores (r = −0.386, P = 0.013). (D) Negative correlation between Hcy and MoCA scores (r = −0.344, P = 0.030). Regression lines and equations are displayed in each graph.

Logistic Regression Analysis

Multivariate logistic regression analysis was performed to identify independent correlates of diabetic cognitive dysfunction. The variance inflation factor (VIF) was calculated to evaluate multicollinearity across all candidate variables entered into the multivariate logistic regression model. Variables exhibiting VIF > 10 were defined as severe multicollinear variables and excluded from the model. No substantial multicollinearity was observed among variables retained in the final regression model. All clinically essential confounding covariates, including age, gender, education level, diabetes duration, hypertension, vascular complications, etc., were mandatorily adjusted in the model irrespective of their statistical significance in univariate analyses or baseline between-group comparisons, to minimize residual confounding bias. In line with the 10 events-per-variable rule for stable model estimation, only four core predictive biomarkers of interest (eGFR, NSE, HbA1c and Hcy) were incorporated into the model; these confounding covariates were adjusted additionally and not counted against the variable limit. The results showed that after adjusting for confounding factors, NSE (OR=1.424,95% CI: 1.107–1.830, P=0.006), HbA1c (OR=1.281,95% CI: 1.021–1.606, P=0.032), and Hcy (OR=1.1, 95% CI: 1.004–1.205, P=0.040) were independently negatively associated with cognitive dysfunction. eGFR (OR=0.984,95% CI: 0.970–0.997, P=0.021) was an independent positively associated with cognitive dysfunction. See Table 2.

Table 2.

Logistic Regression Analysis of Risk Factors for Cognitive Dysfunction

Variables	B	S.E.	Wald	P-value	OR	95% CI
						Lower	Upper
eGFR (mL/min/1.73 m2)	−0.017	0.007	5.346	0.021	0.984	0.970	0.997
NSE (ng/mL)	0.353	0.128	7.601	0.006	1.424	1.107	1.830
HbA1c (%)	0.248	0.116	4.594	0.032	1.281	1.021	1.606
Hcy (μmol/L)	0.095	0.046	4.222	0.040	1.100	1.004	1.205

Notes: Logistic regression assessed the relationships among eGFR, NSE, HbA1, Hcy levels, and and the risk of cognitive dysfunction. P < 0.05 was considered statistically significant.

ROC Curve Analysis of eGFR and NSE Levels in Predicting Cognitive Dysfunction in Patients with T2DM

ROC curves were plotted with eGFR and NSE as independent variables, respectively, with the occurrence of cognitive dysfunction as dependent variable. The results showed: The area under the curve (AUC) of eGFR for diagnosing cognitive dysfunction was 0.660 (95% CI: 0.540–0.780, P=0.014). The maximum Youden’s index for eGFR was 0.275, corresponding to an optimal cutoff value of 125.50 mL/min/1.73 m², with a sensitivity of 0.725 and specificity of 0.550. For serum NSE, the AUC to identify cognitive dysfunction was 0.723 (95% CI: 0.611–0.834, P=0.001). The maximum Youden’s index for NSE was 0.550, corresponding to an optimal cutoff value of 9.5 ng/mL, with a sensitivity of 0.85 and specificity of 0.40. See Figure 2.

Figure 2.

ROC curve analysis of eGFR and NSE for predicting cognitive dysfunction.

Notes: ROC curve analysis was employed to evaluate the predictive value of eGFR and NSE for cognitive dysfunction. (A) ROC curve of estimated eGFR for the prediction of cognitive dysfunction. The AUC was 0.660 (95% CI: 0.540–0.780, P = 0.014). (B) ROC curve of NSE for the prediction of cognitive dysfunction. The AUC was 0.723 (95% CI: 0.611–0.834, P = 0.001).

Discussion

Cognitive dysfunction in T2DM, as a diabetic complication, is receiving increasing attention. Diabetic individuals complicated with cognitive dysfunction are at a high risk of progressing to dementia without timely diagnosis and intervention, which exerts a substantial burden on families and the wider society. Therefore, the detection of corresponding biochemical markers to predict and assess the likelihood of cognitive dysfunction development holds immense clinical significance.

The study revealed significantly lower eGFR levels in the DCD group compared to the DNCD group (P<0.05). Pearson correlation analysis revealed a positive correlation between eGFR levels and MoCA scores in patients with comorbid cognitive dysfunction (r=0.394, P=0.012), Logistic regression analysis showed that estimated glomerular filtration rate (eGFR) was independently positively associated with cognitive function. Slinin et al found through a 5-year prospective study that cognitive function declined more rapidly when eGFR was at a lower baseline level.²² Similar hemodynamic characteristics were observed in the glomerular vascular bed and the brain.²³ Due to the pathophysiological similarities between the brain and kidneys, their pathological injuries exhibit consistency. Both organs are chronically exposed to high-flow, low-resistance environments, which can easily lead to dysfunction of small vessel endothelial cells, resulting in thickening of the capillary basement membrane and luminal narrowing, increased microcirculatory pressure load, and elevated vascular permeability, ultimately causing target organ damage.²⁴ Microvascular injury in the brain may increase neuronal sensitivity to ischemia and hypoxia, ultimately leading to neuronal apoptosis, which is one of the primary pathological mechanisms underlying cognitive dysfunction in diabetes mellitus.²⁵ A decline in eGFR can lead to the deposition of peripheral neurotoxic substances in cerebral microcirculation, resulting in impaired blood-brain barrier function. Concurrently, cytotoxic effects induced by azotemia may cause leukoencephalopathy and cerebral atrophy, thereby contributing to cognitive dysfunction.²⁶ Abnormal accumulation of urea activates matrix metalloproteinase 2 (MMP2) in cerebral endothelial cells, leading to the degradation of matrix proteins including claudin-5, PECAM-1, and type IV collagen. This damage disrupts blood-brain barrier integrity and facilitates the infiltration of neurotoxic solutes.²⁷ Meanwhile, these toxins can exacerbate diabetes-related cognitive dysfunction by damaging brain mitochondria, inducing oxidative stress, and causing neuronal injury.²⁸

NSE is widely present in neurons and neuroendocrine cells. When neurons are damaged (such as ischemia or oxidative stress), increased cell membrane permeability allows NSE to enter the bloodstream by crossing the blood-brain barrier, thereby elevating peripheral blood NSE levels.²⁹ Research findings suggest a strong association between elevated NSE levels and postoperative cognitive decline (POCD) in elderly individuals.^30,31 The role of NSE as a molecular marker for nerve injury and its association with cognitive dysfunction represent a key research focus in current studies.

This study demonstrated that serum NSE levels were significantly higher in the DCD group compared to the DNCD (P<0.05). Pearson correlation analysis revealed a negative correlation between serum NSE levels and MoCA scores in patients with concomitant cognitive dysfunction (r=−0.351, P=0.026). Logistic regression analysis confirmed that NSE was independently negatively associated with cognitive dysfunction. Aligning with Li et al’s findings.¹⁶ Diabetic cognitive dysfunction involves disrupted energy metabolism, oxidative stress, neuroinflammation, and neuroendocrine dysfunction,⁴ This subsequently leads to neuronal damage, resulting in increased NSE release, with its levels positively correlated with the severity of cognitive dysfunction. Simultaneously, it participates in neuroprotective or injurious processes through signaling pathways such as PI3K/Akt.³² Schmidt et al found a notable rise in NSE levels in the cerebrospinal fluid of Alzheimer’s disease (AD) patients, showing a positive correlation with tau protein,³³ Additionally, excessive tau protein phosphorylation is linked to cognitive dysfunction in diabetes mellitus.³⁴

The ROC curve analysis showed that the area under the curve (AUC) for eGFR was 0.660 (95% CI: 0.540–0.780, P=0.014), with a sensitivity of 0.725 and specificity of 0.550. For NSE, the AUC in diagnosing cognitive dysfunction was 0.723 (95% CI: 0.611–0.834, P=0.001), with a sensitivity of 0.85 and specificity of 0.40. The above results indicate that estimated glomerular filtration rate (eGFR) and neuron-specific enolase (NSE) may serve as auxiliary indicators for preliminary screening. The AUC of NSE reached 0.723, indicating moderate discriminatory ability; the AUC of eGFR was 0.660, representing weak-to-moderate predictive performance. Both AUC values were statistically significant (all P<0.05), confirming that the two indicators can distinguish patients with cognitive dysfunction from those without to a certain extent. NSE exhibited high sensitivity (0.85), which makes it suitable for preliminary population screening to reduce missed diagnoses, though its specificity was relatively low (0.40), leading to a certain rate of false-positive results. eGFR showed balanced sensitivity (0.725) and specificity (0.550), but its overall discriminative power was limited. Combining the characteristics of these two indicators, a “combined screening + stratified diagnosis” strategy can be adopted in clinical practice: NSE (cut-off value=9.5ng/mL) is used as the primary screening indicator to maximize the detection of potential patients with cognitive dysfunction due to its moderate sensitivity. For individuals with positive NSE screening results, further evaluation is conducted using eGFR (cut-off value=125.50mL/min/1.73 m²) to leverage its relatively high specificity, thereby reducing misdiagnosis rates and improving diagnostic accuracy. Finally, definitive diagnosis is established by integrating neuropsychological scales such as MoCA, genetic testing, and imaging examinations.

This study found that HbA1c and Hcy levels were significantly elevated in the DCD group compared to the control group, with statistically significant differences. Correlation analysis revealed negative relationships between HbA1c (r=−0.386, P=0.013) and Hcy (r=−0.344, P=0.030) and the MoCA scores. Regression analysis indicated that HbA1c (OR=1.281,95% CI: 1.021–1.606, P=0.032) and Hcy (OR=1.1, 95% CI: 1.004–1.205, P=0.04) were confirmed to be independent risk correlates associated with cognitive dysfunction. The impact of HbA1c and Hcy on cognition is associated with mechanisms such as vascular risk factors, insulin resistance, inflammatory responses, the buildup of advanced glycation end products (AGEs) along with tau phosphorylation.^35–37 These findings suggest that HbA1c and Hcy may play significant roles in the onset and progression of cognitive dysfunction in T2DM.

Conclusion

In conclusion, this study indicates that both eGFR and NSE have certain clinical diagnostic value for cognitive dysfunction, We acknowledge that the single-indicator AUCs of eGFR and NSE do not reach high diagnostic accuracy (AUC ≥0.8), so neither can serve as an independent diagnostic gold standard for diabetic cognitive dysfunction alone. Given the respective performance characteristics of the two indicators, NSE and eGFR provide a preliminary theoretical reference for future joint prediction model development, and subsequent prospective studies with combined biomarker detection are needed to verify this stratified screening idea. This study has certain limitations.: First, Retrospective single-center cross-sectional design, lacking longitudinal follow-up data to draw causal conclusions;Second, Limited total sample size and limited positive cognitive impairment cases; although we strictly followed the 10 EPV rule and VIF collinearity test to reduce model bias, residual unmeasured confounding factors cannot be completely excluded;Third, Only single indicators eGFR and NSE were analyzed via ROC curves; no combined two-indicator predictive model was constructed, hence the proposed stratified screening strategy remains speculative;Fourth, Single-center patient source may lead to selection bias, limiting the generalizability of our findings to broader populations. Fifth, the MoCA cannot quantify impairment in independent cognitive subdomains. Due to the inherent limitations of cross-sectional design: cross-sectional studies cannot establish causal relationships; we have demonstrated that we only observed a correlational association between decreased eGFR, elevated NSE levels, and cognitive decline, rather than a causal effect. Thus, the results of this research should be regarded as exploratory with preliminary implications. single-center sample of 80 patients has limited generalizability, and larger multi-center cohorts are needed for external verification in future research. Future investigations are warranted to employ a multicenter collaborative design, extend the recruitment period, enlarge the sample size, collect complete and standardized medication data, incorporate a broader range of variables, introduce comprehensive neuropsychological tests and conduct prospective studies to further elucidate the associations of serum eGFR and NSE levels with cognitive function in patients with T2DM.

Funding Statement

This study was supported by the Scientific Research Development Fund of the Affiliated Hospital of Shandong Second Medical University (Teaching Hospital) [Grant No. 2025FYM079, China].

Abbreviations

T2DM, Type 2 diabetes mellituse; GFR, estimated glomerular filtration rate; NSE, neuron-specific enolase;MoCA, Montreal Cognitive Assessment; ROC, receiver operating characteristic; ADNI, Alzheimer’s Disease Neuroimaging Initiative; BMI, body mass index; TG, triglycerides; TC, total cholesterol; LDL-C, low-density lipoprotein cholesterol; UA, uric acid; HbA1c, glycated hemoglobin; NLR, neutrophil-to-lymphocyte ratio; Cr, creatinine; SBP, systolic blood pressure; DBP, diastolic blood pressure.

Data Sharing Statement

The raw data supporting the conclusions of this study are available from the corresponding author upon reasonable request. The data are not publicly available due to ethical reasons.

Ethics Approval and Informed Consent

This study was approved by the Ethics Committee of Jinan Fourth People’s Hospital (approval number: LL20240014). It complied with the principles specified in the Declaration of Helsinki. All participating patients provided informed consent and signed the corresponding informed consent form.

Author Contributions

All authors made a significant contribution to the work reported, whether that is in the conception, study design, execution, acquisition of data, analysis and interpretation, or in all these areas; took part in drafting, revising or critically reviewing the article; gave final approval of the version to be published; have agreed on the journal to which the article has been submitted; and agree to be accountable for all aspects of the work.

Disclosure

The authors declare no conflicts of interest.