Microstructural and functional connectivity changes of decision-related brain networks in end-stage kidney disease patients undergoing peritoneal dialysis

Article information

Korean J Nephrol. 2025;.j.krcp.24.146

Publication date (electronic) : 2025 January 17

doi : https://doi.org/10.23876/j.krcp.24.146

Jingjing Yin^1,2,*, Dongxue Luo^3,*, Shuhua Ma^1,2, Chudan Cai⁴, Xiaoyan Shi^1,2, Yanqiang Peng⁴, Jianwei Peng⁴, Lingyu Chen⁴, Yanmin Zheng^1,2, Ruiwei Guo^1,2, Zelin Zhuang^1,2, Zhirong Lin^1,2, Lei Xie^,1,2

¹Department of Radiology, The First Affiliated Hospital of Shantou University Medical College, Shantou, China

²Laboratory of Medical Molecular Imaging, The First Affiliated Hospital of Shantou University Medical College, Shantou, China

³Department of Nuclear Medicine, The First Affiliated Hospital of Shantou University Medical College, Shantou, China

⁴Department of Nephrology, The First Affiliated Hospital of Shantou University Medical College, Shantou, China

Correspondence: Lei Xie Department of Radiology, The First Affiliated Hospital of Shantou University Medical College, 57 Changping Road, Shantou, Guangdong Province 515041, China. E-mail: 13lxie1@stu.edu.cn

*Jingjing Yin and Dongxue Luo contributed equally to this study as co-first authors.

Received 2024 May 27; Revised 2024 September 30; Accepted 2024 October 16.

Abstract

Introduction

Patients with end-stage kidney disease (ESKD), particularly those undergoing dialysis, often experience cognitive impairment as a frequent complication [1–4]. The risk of cognitive impairment in peritoneal dialysis (PD) patients is higher than that in the non-chronic kidney disease (CKD) population, and global cognitive function declines over 2 years [5–7]. According to a study conducted by Griva et al. [8], dialysis patients with cognitive impairment had a lower 7-year survival rate than those without cognitive impairment. The study also concluded that cognitive impairment in the absence of dementia is an autonomous factor that influences mortality in patients with ESKD. The decline in cognitive function can seriously affect the work and daily life of patients with CKD, leading to extended hospitalization and an increased risk of death. Decision-making is an important area of cognitive function that is closely related to an individual’s daily life and work [9,10]. For patients with PD, maintaining good decision-making skills is essential for self-monitoring and effective management of their condition. It also contributes to safe, self-administered dialysis and adherence to complex drug regimens. Given the significance of decision-making in the lives of patients with PD, it is necessary to investigate whether these individuals experience decision-making dysfunction and understand the underlying mechanisms. Further research in this area can provide valuable insights into strategies for improving decision-making abilities among patients with PD, leading to enhanced self-care practices and overall outcomes.

The Iowa Gambling Task (IGT) is a well-known model for assessing uncertain decision-making abilities and has been extensively utilized in clinical and scientific investigations as a highly responsive assessment tool for quantifying compromised decision-making in different neurological and psychiatric conditions [11,12]. Voxel-based morphometry (VBM), a commonly employed technique, is used to examine alterations in the volume and density of the gray matter (GM) in the brain. It can help detect early hidden alterations in brain structure, leading to new ways to diagnose cognitive impairment and understand its underlying causes. Research using VBM has shown that patients with ESKD exhibit a widespread reduction in GM volume (GMV). Moreover, elevated levels of serum urea have been recognized as a possible hazard for the decline of GM in patients with ESKD [13,14]. However, to date, no specific VBM studies have been conducted on patients with PD. Cognitive impairment observed in ESKD has been shown to be associated with abnormal brain function activity [15,16]. According to one study, individuals with both PD and ESKD show reduced levels of amplitude of low-frequency fluctuations in the default mode network (DMN) when compared to healthy individuals. Additionally, PD patients displayed more pronounced spontaneous brain abnormalities, as reported by Luo et al. [16]. However, there is currently limited research on functional connectivity in patients with PD. Overall, the use of IGT and VBM techniques may provide a better understanding of the potential for decision-making disorders and brain structural changes in patients with ESKD and PD.

We hypothesized that decision-related structure and function would be altered in PD patients, and expected a significant association between decision-related structural and functional changes and cognitive/decision-making deficits in PD patients. This study aimed to identify neuroimaging markers to improve the early diagnosis rate of cognitive dysfunction in PD. The influence of PD on decision-making function was also explored by studying the activation of different brain regions among the three groups.

Methods

Participants

Thirty patients with ESKD undergoing PD were recruited from the Department of Nephrology at the First Affiliated Hospital of Shantou University Medical College. Among them, there were 12 male and 18 female patients, with an age range of 24 to 54 years and a median age of 33.50 years (interquartile range [IQR], 30.50–47.25 years). Additionally, 20 ESKD patients who were not undergoing dialysis were also recruited, including 10 males and 10 females, with an age range of 23 to 50 years and a mean age of 38.30 ± 8.24 years.

The inclusion criteria were confirmed CKD stage 5 with stable condition, continuous PD for more than 3 months for patients with PD, education level ≥6 years, age range of 18 to 55 years, right-handedness, and no magnetic resonance imaging (MRI) contraindications. Patients who did not reach stage 5 of CKD, had kidney transplantation or hemodialysis, had a PD duration of <3 months, had a history of brain injury and other brain diseases (such as cerebral hemorrhage, brain trauma, intracranial tumor, Alzheimer disease, etc.) and obvious intracranial organic lesions (such as encephalomalacia lesions, cerebral infarction, cerebral white matter [WM] lesions, etc.) found from T2-weighted imaging that could impact brain structure and function, had previous or present psychiatric disorders, had previous drug abuse or dependence, were illiterate, noncooperative, or had MRI scan contraindications were excluded from the study.

Thirty healthy controls (HC) with normal kidney function (15 male and 15 female patients, median age of 33.50 years [IQR, 28.75–36.50 years], with an age range of 24 to 56 years), who were right-handed, had good physical health status, and matched PD and predialysis CKD stage 5 patients in terms of age, sex, and education level were included in the same period from the outpatient department or physical examination center of our hospital. The exclusion criteria for HC were previous or current renal organic lesions, organic brain disorders, cardiovascular and neuropsychiatric diseases, history of drug or alcohol dependence or abuse, recent sedatives or central nervous system depression, and MRI scan contraindications.

Basic demographic data, disease duration, dialysis duration, and blood biochemical indicators were also collected. General cognitive status was evaluated using the Montreal Cognitive Assessment (MOCA) [17], which has a maximum score of 30. A score below 26 indicates mild cognitive impairment, whereas a score below 19 is used to rule out possible dementia. Trail Making Test (TMT) mainly includes TMT A and TMT B and is used to evaluate execution functions, including decision-making and processing speed [18,19]. Participants were required to connect scattered numbers or letters in sequential order as quickly and accurately as possible. The time taken to complete each test was recorded, with shorter completion times indicating better executive function. Impairment in executive function may be suggested if it takes more than 75 seconds to complete TMT A or more than 180 seconds to complete TMT B. The information is presented in Table 1.

Table 1.

Clinical basic information and neuropsychological tests of the subjects

All participants willingly agreed to participate in the study and provided their signatures on the informed consent form. This research was approved by the Ethics Committee of the First Affiliated Hospital of Shantou University Medical College (No. B-2021-238). This study was performed in strict accordance with the Declaration of Helsinki and was formulated by the World Medical Association.

Results

Iowa Gambling Task results

A 3 × 5 repeated-measures analysis of variance revealed a noteworthy impact of the IGT block (F = 18.669, p < 0.001), a significant effect of groups (F = 4.595, p = 0.02), and a significant interaction between group and IGT block (F = 17.983, p < 0.001). The combined IGT net score of the PD group (p = 0.001) and the predialysis CKD stage 5 group (p = 0.049) exhibited a notable decrease compared to the HC group (Fig. 1A).

Figure 1.

IGT results.

(A) Total IGT net score of the three groups. (B) IGT net scores of each block and the decision-making trend of the three groups. (C) Difference of brain activation between groups during IGT task. a: Brain activation map of PD group vs. HC group during IGT task. b: Brain activation map of predialysis CKD stage 5 group vs. HC group during IGT task. c: Brain activation map of PD group vs. predialysis CKD stage 5 group during IGT task.

CKD, chronic kidney disease; HC, healthy controls; IGT, Iowa Gambling Task; L, left; PD, peritoneal dialysis; R, right.

^*p < 0.05, ^**p < 0.01.

Fig. 1B shows the curve of the net scores for each block in the three groups. While the first and second blocks showed comparable performance among the three groups, notable distinctions were observed between PD and HC (p > 0.001) as well as between predialysis CKD stage 5 and HC (p < 0.05) in the subsequent three blocks.

Functional magnetic resonance imaging data during performance of the Iowa Gambling Task

Statistical comparison of the activation maps showed that, compared with HC, PD patients had reduced activation in some brain regions, such as the bilateral vmPFC, OFC, ACC, and supplementary motor area (SMA). The brain regions with decreased activation in the predialysis CKD stage 5 group compared to the HC group were the vmPFC/OFC/ACC, cingulate gyrus, and precuneus. Compared with the predialysis CKD stage 5 group, the PD group showed enhanced brain activation in the parahippocampal gyrus/amygdala, ACC, vmPFC, and insula (Table 2, Fig. 1C).

Table 2.

Brain activation differences of groups during Iowa Gambling Task

Voxel-based morphometry results

In the VBM analysis based on the whole brain, the normalized GMV of the PD group was the smallest, and there were statistical differences between the HC group and predialysis CKD stage 5 group (t = 5.015, p < 0.001; t = –2.134, p = 0.04).

Furthermore, the VBM study identified extensive GMV reduction in various regions of the brain in the PD group compared with the HC group. The primary areas encompassed the bilateral hippocampal/parahippocampal gyrus/amygdala, bilateral OFC, bilateral vmPFC, dorsolateral prefrontal cortex (DLPFC), right temporal lobe, left insula, and bilateral posterior cingulate cortex. Additionally, decreases were also noted in particular areas, including the left anterior central gyrus, right posterior central gyrus, bilateral putamen, bilateral thalamus, and cerebellar hemisphere (p < 0.05, FDR-corrected) (Table 3, Fig. 2).

Table 3.

Regions of decreased GMV in PD compared to controls (PD < HC)

Figure 2.

Altered GMV among the three groups.

(A) Regions showing decreased GMVs in patients with PD compared to the HC group. (B) Regions showing decreased GMVs in patients with predialysis CKD stage 5 compared to the HC group. (C) GMV matter volume of altered brain regions in the PD group compared to the predialysis CKD stage 5 group. In the right color bar, blue represents voxels with relatively reduced volume, and red represents voxels with relatively increased volumes. p < 0.05 (false discovery rate-corrected).

CKD, chronic kidney disease; GMV, gray matter volume; HC, healthy controls; L, left; PD, peritoneal dialysis; R, right.

Resting-state magnetic resonance imaging results

When the vmPFC was used as the seed point, intragroup analysis of the three groups showed that the brain areas with higher functional connectivity strength than the whole-brain average mainly included the DMN [23,24] (Supplementary Fig. 1, available online). In the PD group, inter-group analysis revealed a decrease in functional connectivity to the vmPFC in several brain regions, including the left superior frontal gyrus (BA10), left inferior frontal gyrus (BA46), right middle and inferior frontal gyrus (BA10, 46), left insula, right inferior temporal gyrus, and right anterior cingulate gyrus, compared to the HC group (p < 0.001) (Fig. 3A).

Figure 3.

Significant differential brain regions of functional connectivity between the PD group and HC group, and the predialysis CKD stage 5 group and HC group.

(A) Based on vmPFC. (B) Based on ACC.

ACC, anterior cingulate cortex; CKD, chronic kidney disease; HC, healthy controls; L, left; PD, peritoneal dialysis; R, right; vmPFC, ventromedial prefrontal cortex.

When the ACC was used as the seed point, the brain regions with higher connections than the average whole brain were mainly concentrated in the salience network (SAN) and executive control network (ECN) (Supplementary Fig. 2, available online). Compared to the HC group, the PD group exhibited decreased functional connectivity with the ACC in various brain regions, primarily involving the bilateral insula (BA13), left DLPFC (BA9), bilateral SMA (BA6), ACC (BA32), and right transverse temporal gyrus (BA13) (p < 0.001) (Fig. 3B).

Correlational analysis

Within the PD group, there was a positive correlation between the net score of the IGT and both the activation intensity and local volume of the GM in the vmPFC (r = 0.733, p < 0.001; r = 0.686, p = 0.002). A strong positive correlation was observed between the GVM of the vmPFC in the local region and the MOCA score (r = 0.535, p = 0.002). The intensity of the vmPFC activation was negatively correlated with TMT A (r = –0.756, p < 0.001) and TMT B (r = –0.656, p = 0.003) (Fig. 4). The intensity of OFC activation was negatively associated with blood creatinine (r = –0.503, p = 0.03), cystatin (r = –0.494, p = 0.04), and parathyroid hormone (PTH) levels (r = –0.482, p = 0.04).

Figure 4.

Scatter plots of the correlation analysis between the BOLD signal changes in the vmPFC, and the IGT net score and the TMT task in the PD group, as well as between the GMVs of vmPFC and the IGT net score.

^*p < 0.05, ^**p < 0.01.

BOLD, blood oxygen level-dependent; GMV, gray matter volume; IGT, Iowa Gambling Task; PD, peritoneal dialysis; TMT, Trail Making Test; vmPFC, ventromedial prefrontal cortex.

Discussion

In this study, decision-making dysfunction and structural and functional impairments in the decision-making–related network (vmPFC-OFC-ACC) were observed in ESKD patients. This study also observed a correlation between activation intensity, GMV, and cognitive/decision-making deficits in patients with PD. The activation intensity and change in GMV of the vmPFC in the PD group were positively correlated with the total net score of the IGT task and the MOCA scores. It was demonstrated that the vmPFC-OFC-ACC network might be a key circuit in decision-making impairment in patients with PD, and the reduction of GMV and activation intensity in the vmPFC might be the key reason for impaired cognitive and decision-making function.

Through the IGT task, our study found that the PD and predialysis CKD stage 5 groups had an impairment of fuzzy decision-making function. Unlike the HC group, neither the PD nor the predialysis CKD stage 5 group showed a learning curve or trend in the decision-making task. Even in the later stages (blocks 3, 4, 5) of the task, the net scores of each block and the total net scores remained mostly negative for both groups, which indicated that participants in these groups were unable to grasp the hidden rules of rewards and punishments in the decision-making module and had difficulty learning from previous mistakes. In summary, this suggests that both the PD and predialysis CKD stage 5 groups have a defect in the learning effect from the decision-making task.

Our study findings demonstrate that both PD and predialysis CKD stage 5 patients had abnormal activation in the vmPFC, OFC, ACC, SMA, cingulate gyrus, and precuneus under decision-making conditions. Previous studies [25,26] have reported that these regions make up the agranular and dysgranular portion of the prefrontal cortex and have extensive connections with the neocortex as well as several subcortical structures. They play a critical role in emotional/social regulation, reward-based learning, attention, and error correction. Through these functions, they contribute to decision-making processes. These brain regions form the vmPFC-OFC-ACC circuit and are involved in decision-making neural circuits, which is consistent with our study results. Moreover, the abnormal activation patterns of this network may lead to cognitive impairment, especially decision-making dysfunction in ESKD patients. Neurological factors related to risk decision-making have been used as brain biomarkers for treatment outcomes [27].

Previous studies have revealed that ESKD patients, especially those undergoing hemodialysis, show reductions in GMV in several brain regions [14,28]. However, there is a lack of VBM studies focusing on PD patients. We suspected that patients with PD might also have volume alterations in GM. The present study demonstrated that patients with PD had extensive GMV reduction, including in the vmPFC, OFC, and so on, which corresponded to abnormal functional activation during decision-making processing. Therefore, this study adds to the existing literature by demonstrating that PD patients also exhibit GMV alterations similar to those observed in ESKD patients undergoing hemodialysis, and the identified structural changes in the vmPFC-OFC-ACC circuit provide further insight into the potential mechanisms underlying the decision-making dysfunction observed in PD patients. This suggests that these alterations may contribute to the cognitive and behavioral impairments associated with PD.

Through correlation analysis, it was found that the activation intensity and the change in GMV of the vmPFC in the PD group were positively correlated with the total net score of the IGT task, indicating that the vmPFC was the key brain area for impaired decision-making function in the PD group. In addition, it was also found that the vmPFC GMV change in the PD group was positively correlated with the MOCA score, which also suggested that cognitive dysfunction in PD patients may be related to the volume reduction of the vmPFC. These findings provide further evidence of the impact of vmPFC injury on cognitive function.

Previous studies have demonstrated that patients with lesions or injuries in the vmPFC and OFC have serious decision-making obstacles [29,30], proving the core role of the vmPFC and OFC in decision-making. The vmPFC plays a crucial role in the decision-making process, as it integrates various options and encodes the value of decision-related information [31]. The OFC is a key structure that promotes decision-making. It can influence and guide decision-making to make long-term favorable choices through emotional signals or integrate feedback on rewards and punishments and transfer the information to the vmPFC for evaluation and calculation of the expected return to guide and control decision-making behavior [32]. In this study, decreased brain activation and GMV of the vmPFC and OFC were found in the PD group, which may be the main causes of decision-making dysfunction in patients with PD.

Through correlation analysis, it was found that the activation intensity of OFC in the PD group was negatively correlated with serum creatinine, cystatin, and PTH levels. Serum creatinine is an important index to detect glomerular filtration rate and also to measure renal function. The higher the blood creatinine level, the lower the activation intensity of OFC, suggesting that the more severe the renal function damage, the more obvious decrease of the blood flow signal of OFC, which also reflects the existence of the “kidney-brain axis.” Cystatin and PTH can cause cognitive function damage by causing neurodegeneration. The higher the concentration of cystatin and PTH, the lower the activation intensity of OFC, that is, the more obvious damage of OFC.

During the decision-making process, the ACC may integrate recent choices and previous results by monitoring behaviors and results and then evaluate and implement the most favorable decisions [33]. SMA/ACC has also been shown to monitor the execution of decision-making behaviors [34]. The decreased ACC/SMA activation in the PD group partly affected the monitoring and execution of individuals during the decision-making process.

Compared to the HC group, the PD group exhibited decreased functional connectivity to the vmPFC in brain regions primarily comprising the DMN. This implies a decline in internal integration function within the DMN in the PD group. The findings of Ni et al. [35], who conducted an independent component analysis of the DMN of patients with ESKD, yielded comparable results to those of this study. Our study suggested that the PD group had reduced integration function within the DMN, which might cause specific dysfunction, including working memory, attention, emotion, and decision-making function [24].

The study showed that the brain regions with higher connective strength to the ACC than the average level of the whole brain were mainly concentrated in the main brain regions of the SAN and ECN [36]. Both the PD and predialysis CKD stage 5 groups showed a decrease in functional connectivity of the SAN and ECN, especially in the PD group. The dysfunction of the internal integration of the network may affect the body to correctly reflect internal and external stimuli, thus causing disorders in the DMN and ECN network activities, leading to the occurrence of mental diseases [37,38]. The weakening of its functional connection may lead to mental problems such as depression and working memory disorders in patients with PD.

A significant reduction in activation was also found in important brain regions related to decision-making, such as the vmPFC, OFC, and ACC in the predialysis CKD stage 5 group compared to the HC group. In addition to the behavioral performance of the predialysis CKD stage 5 group in the IGT task, the overall score was considerably lower than that of the HC group. Moreover, there was no notable improvement in learning, indicating the presence of decision-making impairment in patients with predialysis CKD stage 5. It is further speculated that decision-making dysfunction in patients with predialysis CKD stage 5 may be related to the reduced activation of the above brain regions. Nevertheless, the VBM investigation did not reveal any significant alterations in GMV within decision-making–associated brain regions when comparing the predialysis CKD stage 5 group to the HC group. This indicates that predialysis CKD stage 5 patients have undergone changes in the function of relevant brain regions before structural changes occur.

Notably, through the VBM study, our results showed that PD and predialysis CKD stage 5 patients have GMV reduction and a broader range in PD patients. These would imply that, as the increased duration of disease and dialysis, the GMV have reduced much more in PD patients. However, compared with the predialysis CKD stage 5 group, the PD group showed enhanced activation of some decision-making–related brain regions in the parahippocampal gyrus/amygdala, ACC, vmPFC, and insular lobe. Combining behavioral results, there was no difference in MOCA total score, TMT A/B scores, and IGT-net scores between PD and predialysis CKD stage 5, which would imply that, despite the increased duration of disease and dialysis, there was no significant difference in these cognitive tests. These results suggested that the structural damage of PD patients was more severe, they need more compensatory activation of brain areas (enhanced activation of brain regions) to maintain cognitive function just similar to predialysis CKD stage 5 patients. The brain microstructural damage in PD patients may predate the changes in brain function. These may suggest the comparable capability of PD in preserving one’s cognitive function.

This study has several limitations. Firstly, the study was a cross-sectional study with a relatively small sample size. Although we included ESKD patients without dialysis (predialysis CKD stage 5) and HC groups as a control to increase the interpretability of the results, that’s not enough to explain. In future studies, the sample size should be expanded, and long-term follow-up of PD patients and longitudinal follow-up studies should be conducted to increase the stability and interpretability of the results. By increasing the sample size, we can examine the other potential confounding factors such as KT/V value, all the uremic toxins levels, and it would have been beneficial to examine whether there are cognitive function differences based on dialysis adequacy among PD patients. Secondly, imaging biomarker for clinical diagnosis is not only brain connectivity, for example, ischemic change or microhemorrhage, etc. So, we should include other imaging biomarkers for explaining the alterations of brain networks. Furthermore, this study only claims that PD exhibits alterations of structural and functional networks in decision-related brain and is closely related to cognitive and decision function. While uremic toxins seem to play a role, the introduction suggests an influence of serum urea. However, urea is generally an inert molecule. There are types of uremic toxins that PD can effectively remove and others it cannot. Therefore, future studies should take these factors into account to thoroughly discuss which toxins might contribute to this and whether these toxins are known to affect brain structure or cognitive function as highlighted in this study.

In summary, our study demonstrated that PD patients exhibited alterations of structural and functional networks in the decision-related brain circuits (vmPFC-OFC-ACC), as well as the reduced internal integration function in DMN and SAN and the vmPFC play a key role, which is closely related to cognitive and decision function. The blood serum creatinine, cystatin, and PTH levels may have an impact on the activation of OFC in the vmPFC-OFC-ACC circuit. Moreover, the results suggest that the comparable capability of PD may preserve one’s cognitive function. These provide valuable and objective imaging markers for the clinical diagnosis and potential pathogenesis of cognitive impairment in PD patients.

Supplementary Materials

Article information Continued