Prediction model for cognitive impairment in maintenance hemodialysis patients: the role of diabetes

Objective: The aim of this study was to identify risk factors for cognitive impairment (CI) in maintenance hemodialysis (MHD) patients and to develop a predictive model.

Methods: A total of 151 MHD patients from our hospital were recruited between July 2020 and April 2021. Sociodemographic, clinical, and laboratory data were collected. Cognitive function was assessed using the Mini-Mental State Examination (MMSE), whereas physical performance was evaluated using handgrip strength, the Timed Up and Go Test (TUGT), and 4-m gait speed. Univariate and multivariate logistic regression analyses identified risk factors, which were used to develop original and simplified predictive models.

Results: CI was present in 43 patients (28.5%). The simplified model demonstrated discriminatory ability comparable to that of the original model (AUC: 0.737; 95% CI: 0.648–0.818) and was easier to use. A robust nomogram was developed on the basis of the simplified model. Decision curve analysis (DCA) confirmed the clinical utility of both models. Diabetes was identified as an independent risk factor, whereas dialysis duration was not associated with CI.

Conclusions: This study provides a simple predictive model for CI in MHD patients, which could aid in clinical decision-making.

Introduction

Chronic kidney disease (CKD), with a global prevalence of 14.3%, has become a serious public health problem worldwide (1). End-stage kidney disease (ESKD) is the final stage of CKD, in which renal replacement therapy, such as kidney transplantation, hemodialysis (HD), and peritoneal dialysis, is needed. According to the Chinese Research Data Services Platform (https://www.cnrds.com), 916,647 patients with ESKD were undergoing hemodialysis by the end of 2023. Cognitive impairment (CI) is commonly observed in patients receiving hemodialysis, whose prevalence has been reported to range between 30% and 80% (2, 3). CI has been linked to several negative outcomes, including a lower quality of life, poor medication adherence, and impaired decision-making capacity. Additionally, it is an independent predictor of all-cause premature mortality in individuals receiving maintenance hemodialysis (MHD) (4–6). Therefore, early identification of CI is crucial for timely treatment and delaying the progression of CI in MHD patients.

Cognitive decline is frequently affected by various factors, including age, educational background, cerebrovascular conditions, and a range of metabolic abnormalities (7, 8). Poor physical performance also accompanies CI in HD patients (9). Engaging in multicomponent physical activities, such as balance training, aerobic exercises and muscle strengthening, has been demonstrated to enhance cognition and reduce the risk of dementia (10). This study was conducted to identify factors associated with CI and to develop a predictive model for CI among MHD patients.

Materials and methods

Study design and patient population

This cross-sectional study enrolled patients who underwent MHD at the blood purification center in Shanghai Pudong New Area People’s Hospital between July 2020 and April 2021. Patients aged 18 years or older who were receiving hemodialysis for 4 hours per session, three times a week, and for more than 3 months were included in the study. Participants with the following conditions were excluded: (1) were diagnosed with severe cerebrovascular disease and had sequelae; (2) were unable to finish the physical performance test; and (3) did not complete the MMSE test (Figure 1). All the investigations were conducted in accordance with the Declaration of Helsinki, and the study was approved by the Ethics Committee of Shanghai Pudong New Area People’s Hospital (PRYLW2018-01). All of the participants in this study provided written informed consent. All the details and privacy of the patients have been deidentified.

Figure 1

Figure 1. The flowchart of enrollment in this study. A total of 206 participants were collected and 151 patients that underwent maintenance hemodialysis (MHD) were enrolled in this study.

Baseline variables

All participants were required to complete a detailed questionnaire on social demographics and clinical characteristics. The demographic characteristics included age, sex, education level, and health behaviors, such as sleep duration and smoking and drinking habits. Clinical characteristics included height, weight, dialysis duration, ultrafiltration volume and concurrent ailments. Their biochemical laboratory indices, including levels of hemoglobin (Hb) and C-reactive protein (CRP), vitamin D, parathyrin, ferritin, trioxypurine, urea, creatinine, alanine aminotransferase (ALT), aspartate aminotransferase (AST), total protein, albumin, cholesterol, triglycerides, high density lipoprotein (HDL), and low density lipoprotein (LDL), were measured before hemodialysis within 1 week of cognitive function assessment. Dialysis adequacy was defined as the total fractional clearance index for urea (Kt/V).

Assessment of cognitive impairment

The Mini Mental State Examination (MMSE) was used to assess cognitive function. It includes a broad set of cognitive domains that measure the following: time orientation (5 points), place orientation (5 points), registration (3 points), attention and calculation (5 points), recall (3 points) and language (9 points) (11). The possible scores ranged from 0 to 30, with higher scores indicating better cognitive status. The cut-off points used for CI were as follows: 17 for illiterate people, 20 for people with only primary school education, and 24 for people with middle school or higher levels of education (12).

Assessment of physical performance

The assessment of physical performance consisted of handgrip strength (HS), the Timed Up and Go Test (TUGT) and 4-m gait speed (GS). HS was assessed on their nonfistula hand or dominant hand with an indwelling dialysis catheter using a dynamometer (GRIP-D; Takei, Niigata, Japan). Participants were asked to exert maximum effort twice, and the better of the two measurements was recorded (13). The TUGT counts the number of seconds needed for an individual to stand up from a chair, walk 3 m, turn around, walk back to the chair and sit down again with their back against the chair (14). To determine GS, the participants were asked to walk 4 m twice at their usual pace on a flat surface. GS was calculated as the time taken (s) to complete the 4-m distance (m/s), and the mean speed from the two tests was used in the study (13). Walking aids were allowed during the TUGT and GS test. These performance-based tests were performed before the hemodialysis session.

Statistical analyses

Data management and statistical analyses were performed in R v4.3.1 (https://www.r-project.org/). The normality of the continuous variables was tested with the Shapiro–Wilk test. Normally distributed data are described as the means ± standard deviations (SDs) and were analyzed by t tests; nonnormally distributed data are presented as medians and interquartile ranges (IQRs) and were analyzed by the Mann–Whitney test. Categorical variables are expressed as percentages and were analyzed through the chi-square test.

Logistic regression was conducted with the R software stats package, and the factors whose p value was < 0.1 were considered related to the CI and then analyzed by multivariate logistic regression. The Hosmer–Lemeshow test, which is a statistical test for goodness of fit for logistic regression models, was conducted with the R software ResourceSelection package. The forest plots were drawn by the R software ggstats package. The nomogram plot and calibration plot were drawn by the R software rms package and ggplot2 package. The ROC analyses were performed by the R software pROC package (15). The 95% confidence interval (CI) for the area under curve(AUC) was calculated using the bootstrap method, a commonly used algorithm in the field of machine learning. The cut-off values for the ROC curve were selected using the Youden index, which maximizes the sum of sensitivity and specificity. The DCA curve was drawn by the R software rmda package. To model the nonlinear relationships between variables, restricted cubic spline regression was employed with the R software rms package. This method offers the advantage of flexibly capturing complex nonlinear associations while maintaining smoothness, without imposing strict assumptions on the functional form. P values < 0.05 were considered to indicate statistical significance.

Results

Baseline characteristics of the included patients

A total of 151 patients who underwent MHD at the blood purification center were included in this study. The baseline clinical, biochemical and physical characteristics are shown in Table 1–1 and Table 1-2. The median age of the study patients was 62.0 years (IQR: 53.0-69.0 years), and males accounted for 53.0%. Among MHD patients, 43 (28.5%) had CI. The median dialysis duration was 54.0 months (IQR: 25.5–99.5 months).

Table 1-1

Table 1-1. Basic characteristics of MHD patients (demographics variables).

Table 1-2

Table 1-2. Basic characteristics of MHD patients (laboratory variables).

Among the patients included in this study, some characteristics differed between those with and without CI. Age significantly differed between the two groups (P = 0.006). The median age in the CI group was 64.0 years (IQR: 61.0–70.5), which was greater than the median age of 59.5 years (IQR: 51.0–67.5 years) in the group without CI. These findings suggest that older age may be associated with an increased risk of CI. Additionally, there was a significant difference in the incidence of diabetes between the two groups (P = 0.003). A greater percentage of patients in the CI group had diabetes (39.5%) than in the group without CI (15.7%). These findings suggest that diabetes may influence cognitive function, although the exact relationship remains unclear.

In addition, physical tests revealed significant differences between the two groups of patients: hand grip strength (P = 0.009), TUGT (P = 0.022) and gait speed (P = 0.006). The median hand grip strength in the group without cognitive impairment was 25.0 kg (IQR: 19.1–30.6 kg), whereas it was 21.0 kg (IQR: 15.9–28.2 kg) in the group with cognitive impairment. The TUGT of patients with CI was 8.0 seconds (IQR: 6.8–11.5 s), which was longer than the 7.1 seconds (IQR: 6.3–8.7 s) recorded in patients without CI. Similarly, the gait speed in the group without CI was 1.1 m/s (IQR: 1.0–1.3 m/s), whereas it was 1.0 m/s (IQR: 0.7–1.2 m/s) in the CI group. Poor physical performance may be associated with poor cognitive function, indicating a potential relationship between physical function and cognitive health.

Furthermore, we compared the sex distribution; 57.4% of the participants without CI were male, whereas 41.9% of the participants with CI were male. Conversely, a greater percentage of females were in the CI group (58.1%) than in the group without CI (42.6%). However, this sex difference was not statistically significant (P = 0.122).

Univariate logistic analysis and multivariate analysis

To identify variables that may be related to CI, we first used univariate logistic analysis for calculation and screening. The outcomes are shown in Table 2. Univariate logistic regression analysis revealed several significant associations between baseline characteristics and cognitive impairment. Age was significantly associated with CI, with an odds ratio (OR) of 1.047 (95% CI: 1.012–1.086; P = 0.01). Compared with males, females had a significantly greater likelihood of having a CI (OR: 1.872; 95% CI: 0.92–3.874; P = 0.086), although this association did not reach statistical significance. Education level showed a nonsignificant trend, with patients with more than 12 years of education being less likely to have cognitive impairment (OR: 0.139; 95% CI: 0.007–0.801; P = 0.069). Physical performance was also directly associated with CI: HS was inversely associated, with an OR of 0.936 (95% CI: 0.889–0.982; P = 0.009); the TUGT was directly associated, with an OR of 1.11 (95% CI: 1.018–1.216; P = 0.019); and GS was directly associated, with an OR of 0.134 (95% CI: 0.034–0.483; P = 0.003). Diabetes was also significantly associated with CI, with an OR of 3.5 (95% CI: 1.572–7.872; P = 0.002). Stroke was also significantly related to CI, with an odds ratio (OR) of 2.334 (95% CI: 0.872–6.132; P = 0.085).

Table 2

Table 2. Related factors for patients with CI.

For other variables, such as body mass index (BMI), dialysis duration, hypertension, coronary heart disease, congestive heart failure, tobacco use, alcohol consumption, Hb, CRP, vitamin D, parathyrin, ferritin, Kt/V, urea, creatinine, liver enzymes (ALT, AST), total protein, albumin, cholesterol, triglycerides, HDL, and LDL, no statistically significant differences were detected in the univariate analysis.

Afterwards, 8 variables, namely, age, sex, stroke status, HS, TUGT results, GS, diabetes status, and education level, were chosen to develop a multivariate logistic regression model (Model 1).

First, the VIF values of these eight variables were calculated to evaluate multicollinearity; the VIF values of these eight variables were all < 2.7. This absence of multicollinearity increases the robustness of the predictive capabilities of the model, underscoring its reliability in capturing the complexities of the underlying relationships among the variables. This model provides important insights into the factors associated with CI in the MHD patient population (Figure 2). Among these factors, it is particularly important to pay attention to diabetes, which remains a significant predictor even after adjusting for other variables, with an OR of 3.013 (95% CI: 1.155–8.062; P = 0.025), indicating that diabetes is an independent risk factor for CI. The Hosmer–Lemeshow (H-L) test indicated good predictive performance of the model (P = 0.497).

Figure 2

Figure 2. The forest plot of model 1, including 8 variables.

A simplified multivariate model and comparisons

In practical applications, simpler models are often more easily accepted. Therefore, to simplify the model while maintaining good predictive performance, in this study, stepwise regression was applied on the basis of AIC values to reduce the complexity of Model 1 using multivariate logistic regression. The final simplified model (Model 2) consisted of four variables: age, sex, diabetes status, and GS. However, the H-L test did not show a good fit for this model. Despite this, when age, sex, and diabetes status were retained, the model still demonstrated good predictive performance, with no substantial increase in the AIC value, indicating that the model complexity was well controlled (Figure 3A). The VIF values of these 3 variables are all < 1.2, suggesting little multicollinearity. Furthermore, a chi-square test comparing the two models revealed no significant difference between them (P = 0.523).

Figure 3

Figure 3. The forest plot of model 2 and comparisons between model1 and model2. (A) The forest plot of model1, including 3 variables. (B) The ROC curve of model1 and cut-off point. (C) The ROC curve of model2 and cut-off point. (D) The DCA curve of model1 and model2.

As demonstrated by the ROC curve of Model 1, the AUC was 0.739 (95% CI: 0.642–0.823, by bootstrap), with an optimal cut-off of 0.274 (Figure 3B). For Model 2, the AUC was 0.737 (95% CI: 0.648–0.818, by bootstrap), with an optimal cut-off of 0.289 (Figure 3C). The ROC curves of both models were similar, suggesting comparable predictive performance. Additionally, a formal comparison of the ROC curves indicated no significant difference between the two models (P = 0.916). This further confirms that the simplified model achieves a predictive performance very similar to that of the original, more complex model.

Moreover, according to the DCA curve, the performance of the two models is remarkably similar. At the cut-off point (0.289), the benefits of both models are comparable, suggesting that Model 2 could serve as a viable substitute for Model 1 (Figure 3D). These observations underscore that these two models not only have similar performance in mathematical calculations but also have comparable practical utility.

Notably, on the basis of our literature review, a model developed by Chen et al. in a similar study (PMID: 37828422) can be deployed in this study; however, it demonstrated an AUC of 0.575 and a specificity of 0.259, with a sensitivity of 0.930. This indicates superior predictive performance for Model 1 and Model 2, which were developed in this study (16).

Development of the nomogram

To enhance the presentation of the multivariate logistic regression model, a nomogram was developed for the diagnosis of CI (Figure 4A). By adding the points of each variable in the first line, a certain score can be obtained. Afterwards, the score can be used to determine the probability of the CI. Additionally, the H-L test of Model 2 also indicated good performance (Figure 4B, P = 0.771). The cut-off was set at 0.289, as determined using the Youden index, with a sensitivity of 0.721 and a specificity of 0.694 (Figure 4C).

Figure 4

Figure 4. The exhibition and evaluation of model2 and RCS plot of dialysis time. (A) The diagnostic nomogram of CI in MHD patients(model2). (B) The calibration plot of model2 and Hosmer-Lemeshow test. Blue: the histogram of patients without CI; orange: the histogram of patients with CI. (C) The confusion matrix of model2; the cut-off value was 0.289, with a sensitivity of 0.721 and specificity of 0.694. (D) Restricted cubic spline plot of dialysis duration.

In summary, in contrast to the original complex model, the simplified model has good performance, simplicity and user-friendliness.

Dialysis duration

The impact of dialysis duration on patient health remains controversial, including in the context of CI. This study also addresses this metric. The dialysis duration for patients without a CI was 55.0 months (IQR: 26.0–105.0 months). For the CI group, the dialysis duration was 46.0 months (IQR: 24.5–83.5 months). Despite the shorter dialysis duration in the CI group, the difference was not statistically significant (P = 0.305). Additionally, univariate analysis revealed that the OR value was 0.996 (0.989, 1.003), which was not statistically significant (p = 0.268). Additionally, to further investigate the impact of dialysis duration on CI, restricted cubic spline regression was employed (Figure 4D).

The results indicate that the relationship between dialysis duration and CI is linear (nonlinear test: P = 0.928). Moreover, the OR remains close to 1, suggesting that dialysis duration does not significantly impact the risk of CI. This study revealed no significant association between dialysis duration and patients’ cognitive function.

Discussion

Maintenance hemodialysis is the primary treatment for patients with ESRD. Currently, the prevention and treatment of dialysis-related complications and comorbidities have received widespread attention, among which CI significantly affects patients’ quality of life and self-care ability. Furthermore, few studies have explored the relationship between physical performance and cognitive function. Predictive models incorporating physical performance as a factor for CI prediction in MHD patients remain scarce. Therefore, this study explores the factors associated with CI, with a particular focus on the relationship between physical performance and CI.

In this study, the incidence of CI among MHD patients was 28.5%, which is lower than the 30–80% range reported in previous studies. This difference may be due in part to the exclusion of patients with severe cerebrovascular disease and sequelae in our study. To ensure the rigor of this study, patients with severe cerebrovascular diseases and sequelae were excluded to control for potential confounding effects. In addition, the clinical heterogeneity in prevalence is likely influenced by variations in population demographics, sample sizes, diagnostic criteria, and assessment methods for CI.

Consistent with findings in the general population, advanced age and female sex were identified as potential risk factors for CI among MHD patients in this study. Although the p values of these two indicators did not reach less than 0.05 in the multivariate analysis involving eight variables (Model 1), they demonstrated significant p-values in the simplified model (Model 2). These two indicators, which should not be ignored, can be used to assist in the diagnosis of CI. Ageing is strongly associated with dementia through mechanisms such as brain atrophy, amyloid-β deposition, and neuroinflammation (17). Additionally, female patients appear to be more vulnerable to CI because of hormonal changes, particularly a decrease in estrogen after menopause, which has been linked to impairments in learning and memory (18, 19). The majority of female MHD patients are postmenopausal and therefore require more attention. These findings align with those of previous studies, including those by Odagiri et al., Kurella Tamura et al. and Lee Heeryong et al., which reported similar associations in MHD populations (20–23).

Diabetes is a common comorbidity, affecting up to 45% of patients with ESKD (24). Moreover, diabetes emerged as a significant, independent risk factor for CI in our study. It can assist in predicting the CI by the original model (Model 1) or simplified model (Model 2). This may be explained by the effects of persistently high blood glucose levels, which induce oxidative stress and generate large amounts of reactive oxygen species. This process can lead to chronic cerebral ischemia, amyloid deposition, and the accumulation of toxic advanced glycation end products, ultimately causing direct neuronal damage (25, 26). Numerous studies have demonstrated that diabetes is associated with an increased risk of general or specific CI in ESKD patients undergoing hemodialysis, peritoneal dialysis, or kidney transplantation (27–29). Similarly, this study revealed that diabetes increased the risk of CI in MHD patients. This may also be related to the absolute or relative insufficiencies of insulin secretion in diabetic patients. Indeed, insulin also plays important roles in neuronal synaptic plasticity and facilitates learning and memory in humans; therefore, impaired insulin signaling can contribute directly to neuronal dysfunction and degeneration (30). Furthermore, insulin regulates cerebral bioenergetics, increases synaptic activity and dendritic spine formation, and increases the turnover of neurotransmitters (31). Insulin receptors and components of the insulin signaling pathway are widely distributed in the brain, especially in cognition-related regions, such as the cerebral cortex, olfactory bulb, hippocampus, and hypothalamus (32). On the basis of these findings, effective diabetes management and brain insulin resistance could be therapeutic targets to help control CI, which could be meaningful for its treatment.

Moreover, this study revealed that lower HS, longer TUGT, and slower GS were associated with CI. Patients with CI exhibited poorer physical performance, suggesting that better physical performance may serve as a protective factor against cognitive impairment. A meta-analysis reported a strong link between CKD progression and reduced walking speed (33). Another meta-analysis of longitudinal observational studies (1–21 years) revealed that regular exercise was associated with a reduced risk of dementia (34). Yu Ho Lee et al. demonstrated that MHD patients with low GS and HS scores exhibited significantly impaired cognitive function, as assessed by the MMSE and KDQOL-SF (35). Similarly, Jyotish Chalil Gopinathan et al. reported that MHD patients with CI had a significantly greater prevalence of frailty, including weak grip strength and slow walking speed (36). Moreover, another study revealed that 12 HD patients who received three months of tablet-based cognitive training during dialysis experienced improvements in MMSE scores, Montreal Cognitive Assessment (MoCA) scores, and executive function (9). However, in the multivariate logistic regression analysis, physical performance did not reach statistical significance. Although physical performance may aid in the diagnosis of CI, it cannot be considered an independent risk factor. Some studies have suggested that low muscle mass and strength can be improved through exercise, and physical activity in MHD patients has been shown to improve vascular function by reducing systemic inflammation, oxidative stress, and arterial stiffness (37). Nevertheless, these conclusions should be interpreted with caution in the context of this study. These differences may arise from variations in the study population and inclusion criteria. Moreover, relevant cohort studies are lacking. Therefore, the relationship between physical performance and CI requires further investigation. However, as this study was confined to individuals with CI and did not include individuals with mild cognitive impairment (MCI), the relationships between these three indicators and MCI warrant further investigation.

In addition, the association between dialysis duration and CI remains controversial. Currently, there is a perspective suggesting that with increasing dialysis duration, repeated circulatory stress during hemodialysis may contribute to ischemic brain injury. This is largely due to recurrent cerebral hypoperfusion during hemodialysis, which can accelerate cognitive decline (38). Ding Chen et al. reported that a dialysis duration of ≥5 years was associated with an increased risk of CI (16). In contrast, there was no statistically significant difference in dialysis duration between the CI group and the non-CI group in this study. Furthermore, these findings were validated through univariate analysis and restricted cubic spline regression. Using multiple analytical approaches, this study ultimately revealed that dialysis duration was not associated with CI. Therefore, controversy remains concerning the association between hemodialysis duration and CI in MHD patients. However, each dialysis session differs in terms of frequency, adequacy, dialysate temperature, and even the type of dialyzer, all of which may influence dialysis outcomes. Therefore, further longitudinal studies may be needed to clarify the relationship between hemodialysis duration and CI.

Beyond these, some other factors also deserve attention. Several studies have shown that lower educational levels are associated with CI (39, 40). However, we found that patients with more than 12 years of education were less likely to have CI (OR: 0.139; 95% CI: 0.007–0.801; P = 0.069). This may be explained by the link between lower education levels and reduced cognitive and functional reserves. However, education does not have statistical significance. Stroke was not a significant risk factor in our cohort, possibly because of the exclusion of patients with severe cerebrovascular disease.

Interestingly, this study revealed no difference in the percentage of ultrafiltration volume between patients with CI and those without CI. Hemodialysis results in rapid fluid shifts that can often lead to wide swings in blood pressure (41). A possible explanation for this could be that timely treatment was administered to patients when their blood pressure decreased. Further exploration of the relationship between changes in circulating volume and their impact on CI in MHD patients may be valuable.

In summary, this study explored the factors related to CI, and a simplified predictive model was developed. This model not only provides a practical and user-friendly tool for clinical application but also offers valuable insights into the potential risk factors for CI. By incorporating multiple analytical approaches, this study enhances the understanding of CI in MHD patients and offers assistance for future research in this area.

Finally, this study has several limitations. First, patients who were unable to complete the physical performance assessments were excluded, which may have resulted in a biased sample. Second, although the MMSE offers high clinical utility and efficiency for global cognitive screening, its well-documented limitations in sensitivity for mild cognitive impairment (MCI) must be acknowledged. Its speed and simplicity come at the cost of insufficient assessment of executive functions and complex attention, leading to potential under detection of early cognitive decline and possible underestimation of effects in this study. Future studies should incorporate more sensitive tools, such as the Montreal Cognitive Assessment (MoCA), or detailed neuropsychological batteries to improve the accurate identification of MCI and ensure a more comprehensive evaluation of cognitive domains. Finally, its single-center, cross-sectional design hinders generalizability and prevents the establishment of causal relationships. Therefore, a prospective cohort study design is recommended. The sample size, although adequate for initial analysis, may limit the statistical power for detecting subtle effects. Due to the limitation of single-center, future multicenter studies with larger cohorts and longitudinal designs are needed to validate these findings and explore causality.

Conclusions

In conclusion, this study revealed a high incidence of cognitive impairment (CI) among MHD patients. A predictive model for CI was developed using eight variables: age, sex, stroke status, HS, TUGT results, GS, diabetes status, and education level. To improve clinical applicability, a simplified model incorporating only diabetes, age, and sex was constructed, which demonstrated good diagnostic accuracy while being more practical for use in clinical settings. Additionally, diabetes was identified as an independent risk factor for CI, whereas dialysis duration was not found to be associated with CI. These findings provide valuable insights into the risk factors for CI in MHD patients and offer a practical tool to aid in early identification and intervention.

Data availability statement

The raw data supporting the conclusions of this article will be made available by the authors, without undue reservation.

Ethics statement

The studies involving humans were approved by The Ethics Committee of Shanghai Pudong New Area People’s Hospital. The studies were conducted in accordance with the local legislation and institutional requirements. The participants provided their written informed consent to participate in this study.

Author contributions

JX: Data curation, Writing – original draft. HL: Writing – review & editing, Formal analysis. HQ: Project administration, Writing – review & editing. XZ: Project administration, Writing – review & editing.

Funding

The author(s) declare financial support was received for the research and/or publication of this article. This study was supported by grants from the Key Discipline Construction Project (no. PWZxk 2022-19) of Pudong Health, the Flagship Department of Integrated Traditional Chinese and Western Medicine in the Pudong New Area Health Commission (YC-2023-0406) and the Family Planning Commission of Shanghai.

Conflict of interest

The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

Generative AI statement

The author(s) declare that no Generative AI was used in the creation of this manuscript.

Any alternative text (alt text) provided alongside figures in this article has been generated by Frontiers with the support of artificial intelligence and reasonable efforts have been made to ensure accuracy, including review by the authors wherever possible. If you identify any issues, please contact us.

Publisher’s note

All claims expressed in this article are solely those of the authors and do not necessarily represent those of their affiliated organizations, or those of the publisher, the editors and the reviewers. Any product that may be evaluated in this article, or claim that may be made by its manufacturer, is not guaranteed or endorsed by the publisher.

Supplementary material

The Supplementary Material for this article can be found onlineat: https://www.frontiersin.org/articles/10.3389/fendo.2025.1594605/full#supplementary-material

References

1. Kalantar-Zadeh K, Jafar TH, Nitsch D, Neuen BL, and Perkovic V. Chronic kidney disease. Lancet. (2021) 398:786–802. doi: 10.1016/S0140-6736(21)00519-5

PubMed Abstract | Crossref Full Text | Google Scholar

2. Karakizlis H, Bohl K, Ziemek J, Dodel R, and Hoyer J. Assessment of cognitive impairment and related risk factors in hemodialysis patients. J Nephrol. (2021) 35:931–42. doi: 10.1007/s40620-021-01170-3

PubMed Abstract | Crossref Full Text | Google Scholar

3. Cao T, An R, Wang Y, Lei J, Liang J, Wan Q, et al. Risk factors and prevalence of cognitive impairment in maintenance haemodialysis patients: A systematic review and meta-analysis of observational studies. J Advanced Nurs. (2023) 79:3691–706. doi: 10.1111/jan.15746

PubMed Abstract | Crossref Full Text | Google Scholar

4. Griva K, Stygall J, Hankins M, Davenport A, Harrison M, and Newman SP. Cognitive impairment and 7-year mortality in dialysis patients. Am J Kidney Dis. (2010) 56:693–703. doi: 10.1053/j.ajkd.2010.07.003

PubMed Abstract | Crossref Full Text | Google Scholar

5. Kwan E, Draper B, Endre ZH, Harvey SB, and Brown MA. Prevalence, types and recognition of cognitive impairment in dialysis patients in South Eastern Sydney. Internal Med J. (2021) 51:2034–41. doi: 10.1111/imj.14976

PubMed Abstract | Crossref Full Text | Google Scholar

6. Guo Y, Tian R, Ye P, Li X, Li G, Lu F, et al. Cognitive domain impairment and all-cause mortality in older patients undergoing hemodialysis. Front Endocrinol. (2022) 13. doi: 10.3389/fendo.2022.828162

PubMed Abstract | Crossref Full Text | Google Scholar

7. Posner HB, Tang MX, Luchsinger J, Lantigua R, Stern Y, and Mayeux R. The relationship of hypertension in the elderly to AD, vascular dementia, and cognitive function. Neurology. (2002) 58:1175–81. doi: 10.1212/WNL.58.8.1175

PubMed Abstract | Crossref Full Text | Google Scholar

8. Arvanitakis Z, Wilson RS, Bienias JL, Evans DA, and Bennett DA. Diabetes mellitus and risk of Alzheimer disease and decline in cognitive function. Arch Neurol. (2004) 61:661–6. doi: 10.1001/archneur.61.5.661

PubMed Abstract | Crossref Full Text | Google Scholar

9. Bogataj Š, Trajkovi´ N, Pajek M, and Pajek J. Effects of intradialytic cognitive and physical exercise training on cognitive and physical abilities in hemodialysis patients: study protocol for a randomized controlled trial. Front Psychol. (2022) 13:835486. doi: 10.3389/fpsyg.2022.835486

PubMed Abstract | Crossref Full Text | Google Scholar

10. Piercy KL, Troiano RP, Ballard RM, Carlson SA, Fulton JE, Galuska DA, et al. The physical activity guidelines for americans. Jama. (2018) 320:2020–8. doi: 10.1001/jama.2018.14854

PubMed Abstract | Crossref Full Text | Google Scholar

11. Zhang J, Tang G, Xie H, Wang B, He M, and Fu J. Higher adiposity is associated with slower cognitive decline in hypertensive patients: secondary analysis of the China stroke primary prevention trial. J Am Heart Assoc. (2017) 6. doi: 10.1161/JAHA.117.005561

PubMed Abstract | Crossref Full Text | Google Scholar

12. Zhang MY, Katzman R, Salmon D, Jin H, Cai GJ, Wang ZY, et al. The prevalence of dementia and Alzheimer’s disease in Shanghai, China: impact of age, gender, and education. Ann Neurol. (1990) 27:428–37. doi: 10.1002/ana.410270412

PubMed Abstract | Crossref Full Text | Google Scholar

13. Hooghiemstra AM, Ramakers I, Sistermans N, Pijnenburg YAL, Aalten P, Hamel REG, et al. Gait speed and grip strength reflect cognitive impairment and are modestly related to incident cognitive decline in memory clinic patients with subjective cognitive decline and mild cognitive impairment: findings from the 4C study. J Gerontol A Biol Sci Med Sci. (2017) 72:846–54. doi: 10.1093/gerona/glx003

PubMed Abstract | Crossref Full Text | Google Scholar

14. Shimada H, Suzuki T, Suzukawa M, Makizako H, Doi T, and Yoshida D. Performance-based assessments and demand for personal care in older Japanese people: a cross-sectional study. BMJ Open. (2013) 3. doi: 10.1136/bmjopen-2012-002424

PubMed Abstract | Crossref Full Text | Google Scholar

15. Robin X, Turck N, Hainard A, Tiberti N, Lisacek F, Sanchez JC, et al. pROC: an open-source package for R and S+ to analyze and compare ROC curves. BMC Bioinf. (2011) 12:77. doi: 10.1186/1471-2105-12-77

PubMed Abstract | Crossref Full Text | Google Scholar

16. Chen D, Xiao C, Xiao W, Lou L, Gao Z, and Li X. Prediction model for cognitive impairment in maintenance hemodialysis patients. BMC Neurol. (2023) 23:367. doi: 10.1186/s12883-023-03407-z

PubMed Abstract | Crossref Full Text | Google Scholar

17. Jia L, Du Y, Chu L, Zhang Z, Li F, Lyu D, et al. Prevalence, risk factors, and management of dementia and mild cognitive impairment in adults aged 60 years or older in China: a cross-sectional study. Lancet Public Health. (2020) 5:e661–71. doi: 10.1016/S2468-2667(20)30185-7

PubMed Abstract | Crossref Full Text | Google Scholar

18. Rahman A, Schelbaum E, Hoffman K, Diaz I, Hristov H, Andrews R, et al. Sex-driven modifiers of Alzheimer risk: A multimodality brain imaging study. Neurology. (2020) 95:e166–78. doi: 10.1212/WNL.0000000000009781

PubMed Abstract | Crossref Full Text | Google Scholar

19. Yan W, Wu J, Song B, Luo Q, and Xu Y. Retraction Note to: Treatment with a brain-selective prodrug of 17β-estradiol improves cognitive function in Alzheimer’s disease mice by regulating klf5-NF-κB pathway. Naunyn Schmiedebergs Arch Pharmacol. (2021) 394:1989. doi: 10.1007/s00210-021-02121-2

PubMed Abstract | Crossref Full Text | Google Scholar

20. Drew DA, Weiner DE, Tighiouart H, Duncan S, Gupta A, Scott T, et al. Cognitive decline and its risk factors in prevalent hemodialysis patients. Am J Kidney Dis. (2017) 69:780–7. doi: 10.1053/j.ajkd.2016.11.015

PubMed Abstract | Crossref Full Text | Google Scholar

21. Odagiri G, Sugawara N, Kikuchi A, Takahashi I, Umeda T, and Saitoh H. Cognitive function among hemodialysis patients in Japan. Ann Gen Psychiatry. (2011) 10:20. doi: 10.1186/1744-859X-10-20

PubMed Abstract | Crossref Full Text | Google Scholar

22. Kurella Tamura M, Larive B, Unruh ML, Stokes JB, Nissenson A, Mehta RL, et al. Prevalence and correlates of cognitive impairment in hemodialysis patients: the Frequent Hemodialysis Network trials. Clin J Am Soc Nephrol. (2010) 5:1429–38. doi: 10.2215/CJN.01090210

PubMed Abstract | Crossref Full Text | Google Scholar

23. Lee H, Kim K, Ahn J, Lee DR, Lee JH, and Hwang SD. Association of nutritional status with osteoporosis, sarcopenia, and cognitive impairment in patients on hemodialysis. Asia Pac J Clin Nutr. (2020) 29:712–23.

PubMed Abstract | Google Scholar

24. Kopple JD. National kidney foundation K/DOQI clinical practice guidelines for nutrition in chronic renal failure. Am J Kidney Dis. (2001) 37:S66–70. doi: 10.1053/ajkd.2001.20748

PubMed Abstract | Crossref Full Text | Google Scholar

25. Moulton CD, Stewart R, Amiel SA, Laake JP, and Ismail K. Factors associated with cognitive impairment in patients with newly diagnosed type 2 diabetes: a cross-sectional study. Aging Ment Health. (2016) 20:840–7. doi: 10.1080/13607863.2015.1040723

PubMed Abstract | Crossref Full Text | Google Scholar

26. Biessels GJ, Staekenborg S, Brunner E, Brayne C, and Scheltens P. Risk of dementia in diabetes mellitus: a systematic review. Lancet Neurol. (2006) 5:64–74. doi: 10.1016/S1474-4422(05)70284-2

PubMed Abstract | Crossref Full Text | Google Scholar

27. Li Y, Tian X, Xiong ZY, Liao JL, Hao L, Liu GL, et al. Performance of the modified mini-mental state examination (3MS) in assessing specific cognitive function in patients undergoing peritoneal dialysis. PloS One. (2016) 11:e0166470. doi: 10.1371/journal.pone.0166470

PubMed Abstract | Crossref Full Text | Google Scholar

28. Gupta A, Mahnken JD, Johnson DK, Thomas TS, Subramaniam D, Polshak T, et al. Prevalence and correlates of cognitive impairment in kidney transplant recipients. BMC Nephrol. (2017) 18:158. doi: 10.1186/s12882-017-0570-1

PubMed Abstract | Crossref Full Text | Google Scholar

29. Cui L, Chen W, Yu X, and Ju C. The relationship between cognitive function and having diabetes in patients treated with hemodialysis. Int J Nurs Sci. (2020) 7:60–5. doi: 10.1016/j.ijnss.2019.12.003

PubMed Abstract | Crossref Full Text | Google Scholar

30. Verdile G, Fuller SJ, and Martins RN. The role of type 2 diabetes in neurodegeneration. Neurobiol Dis. (2015) 84:22–38. doi: 10.1016/j.nbd.2015.04.008

PubMed Abstract | Crossref Full Text | Google Scholar

31. Kellar D and Craft S. Brain insulin resistance in Alzheimer’s disease and related disorders: mechanisms and therapeutic approaches. Lancet Neurol. (2020) 19:758–66. doi: 10.1016/S1474-4422(20)30231-3

PubMed Abstract | Crossref Full Text | Google Scholar

32. Cui Y, Tang TY, Lu CQ, and Ju S. Insulin resistance and cognitive impairment: evidence from neuroimaging. J Magn Reson Imaging. (2022) 56:1621–49. doi: 10.1002/jmri.28358

PubMed Abstract | Crossref Full Text | Google Scholar

33. Zemp DD, Giannini O, Quadri P, and de Bruin ED. Gait characteristics of CKD patients: a systematic review. BMC Nephrol. (2019) 20:83. doi: 10.1186/s12882-019-1270-9

PubMed Abstract | Crossref Full Text | Google Scholar

34. Livingston G, Huntley J, Sommerlad A, Ames D, Ballard C, Banerjee S, et al. Dementia prevention, intervention, and care: 2020 report of the Lancet Commission. Lancet. (2020) 396:413–46. doi: 10.1016/S0140-6736(20)30367-6

PubMed Abstract | Crossref Full Text | Google Scholar

35. Lee YH, Kim JS, Jung SW, Hwang HS, Moon JY, Jeong KH, et al. Gait speed and handgrip strength as predictors of all-cause mortality and cardiovascular events in hemodialysis patients. BMC Nephrol. (2020) 21. doi: 10.1186/s12882-020-01831-8

PubMed Abstract | Crossref Full Text | Google Scholar

36. Gopinathan JC, Hafeeq B, Aziz F, Narayanan S, Aboobacker IN, and Uvais NA. The Prevalence of Frailty and its Association with Cognitive Dysfunction among Elderly Patients on Maintenance Hemodialysis: A Cross-Sectional Study from South India. Saudi J Kidney Dis Transpl. (2020) 31:767–74. doi: 10.4103/1319-2442.292310

PubMed Abstract | Crossref Full Text | Google Scholar

37. Bronas UG. Exercise training and reduction of cardiovascular disease risk factors in patients with chronic kidney disease. Adv Chronic Kidney Dis. (2009) 16:449–58. doi: 10.1053/j.ackd.2009.07.005

PubMed Abstract | Crossref Full Text | Google Scholar

38. Polinder-Bos HA, García DV, Kuipers J, Elting JWJ, Aries MJH, Krijnen WP, et al. Hemodialysis induces an acute decline in cerebral blood flow in elderly patients. J Am Soc Nephrol. (2018) 29:1317–25. doi: 10.1681/ASN.2017101088

PubMed Abstract | Crossref Full Text | Google Scholar

39. Joseph SJ, Bhandari SS, and Dutta S. Cognitive impairment and its correlates in chronic kidney disease patients undergoing haemodialysis. J Evol Med Dent Sci. (2019) 8:2818–22. doi: 10.14260/jemds/2019/611

PubMed Abstract | Crossref Full Text | Google Scholar

40. Gesualdo GD, Duarte JG, Zazzetta MS, Kusumota L, Say KG, Pavarini SCI, et al. Cognitive impairment of patients with chronic renal disease on hemodialysis and its relationship with sociodemographic and clinical characteristics. Dement Neuropsychol. (2017) 11:221–6. doi: 10.1590/1980-57642016dn11-030003

PubMed Abstract | Crossref Full Text | Google Scholar

41. Daugirdas JT. Pathophysiology of dialysis hypotension: an update. Am J Kidney Dis. (2001) 38:S11–7. doi: 10.1053/ajkd.2001.28090

PubMed Abstract | Crossref Full Text | Google Scholar