Fiber connectivity density mapping in end-stage renal disease patients: a preliminary study

Abnormal brain structural connectivity of end-stage renal disease(ESRD) is associated with cognitive impairment. However, the characteristics of cortical structural connectivity have not been investigated in ESRD patients. Here, we study structural connectivity of the entire cerebral cortex using a fiber connectivity density(FiCD) mapping method derived from diffusion tensor imaging(DTI) data of 25 ESRD patients and 20 healthy controls, and between-group differences were compared in a vertexwise manner. We also investigated the associations between these abnormal cortical connectivities and the clinical variables using Pearson correlation analysis and multifactor linear regression analysis. Our results demonstrated that the mean global FiCD value was significantly decreased in ESRD patients. Notably, FiCD values were significantly changed(decreased or increased) in certain cortical regions, which mainly involved the bilateral dorsolateral prefrontal cortex(DLPFC), inferior parietal cortex, lateral temporal cortex and middle occipital cortex. In ESRD patients, we found a trend of negative correlation between the increased FiCD values of bilateral middle frontal gyrus and serum creatinine, urea, parathyroid hormone(PTH) levels and dialysis duration. Only the white matter hyperintensity(WMH) scores were significantly negatively correlated with the global FiCD value in multifactor regression analysis. Our results suggested that ESRD patients exhibited extensive impaired cortical structural connectivity, which was related to the severity of WMHs. A compensation mechanism of cortical structural recombination may play a role in how the brain adapts to maintain optimal network function. Additionally, the serum creatinine, urea and PTH levels may be risk factors for brain structural network decompensation in ESRD patients.


Introduction
Patients with end-stage renal disease(ESRD) have a higher prevalence of cognitive impairment (Drew et al., 2019). Studies have shown that even younger dialysis patients have severe overall cognitive impairment and that all age groups receiving dialysis have poorer cognitive function than their peers (Chu & McAdams-DeMarco, 2019). The development of cognitive impairment in ESRD patients is considered multifactorial. ESRD usually coexists with multiple comorbidities. Arteriosclerosis, hypertension and hyperlipidemia are pathophysiological continuums that occur in chronic kidney disease(CKD) and gradually increase as renal function declines (Georgianos et al., 2018). CKD has been identified as an independent risk factor for cerebrovascular disease, particularly small-vessel cerebrovascular disease, leading to cognitive dysfunction (Yamamoto et al., 2011). Next, anemia (Kurella et al., 2011) and secondary hyperparathyroidism (Lourida et al., 2015) were also considered associated with cognitive impairment and dementia. Another noteworthy factor is dialysis, which may indirectly affect cognition because it can lead to rapid fluid transfer and swings in blood pressure (Daugirdas, 2001;Eldehni & McIntyre, 2012).
The underlying neuropathologic processes of cognitive impairment in ESRD patients are complex and remain unclear. Traditional MRI studies have shown that the risk of ischemic white matter(WM) disease in patients with CKD is higher than that in the general population and mainly manifests as confluent WM hyperintensity(WMH) and subcortical lacunar infarcts (Ağildere et al., 2001). Furthermore, most diffusion tensor imaging(DTI) studies investigating ESRD have consistently demonstrated that reduced WM integrity and disruptions of brain structural connection networks are intimately associated with neurological complications in ESRD patients (Drew et al., 2017;Eldehni et al., 2019;Mu et al., 2018;Yin et al., 2018). These results seem to indicate a "disconnection" of WM fiber pathways, possibly providing the anatomical basis for abnormal brain functional interactions.
Despite promising results revealing WM abnormalities in ESRD patients, current DTI studies have major shortcomings. A major limitation is the inability to provide direct and comprehensive analysis of the structural connectivity of the whole cortex. For example, the real shortcomings of both region of interest(ROI) and diffusion tensor tractography(DTT) methods depend on prior defined ROIs and the consequent bias. Similarly, the results of graph analysis methods are limited by the selection of prior templates. Next, measuring the structural connectivity corresponding to the cerebral cortex using the tract-based spatial statistics(TBSS) method is inherently challenging. Additionally, the mean fractional anisotropy(FA) for most DTI studies in quantitative comparisons is not a direct measurement of the designated connectivity of the brain cortex but an indicator of fiber integrity for a certain white matter region, primarily because it does consider the number of fibers or ROI size (Huang & Ding, 2016).
A recent technique combining a diffusion tractography technique and cortical surface-based analysis-named fiber connectivity density(FiCD) mapping (Liu et al., 2017)was proposed to quantify the structural connectivity of the whole cortex, making it possible to accurately locate the pathologically disconnected areas of the cortex (Liu et al., 2016). Specifically, in the FiCD method, vertexwise multiple statistical analyses were used to compare the whole-cortex connectivity between groups. This data-driven approach can automatically identify brain regions with significant differences at the whole-brain level without defining ROIs. Second, the generated FiCD maps were precisely matched to the common brain surface, allowing the WM abnormalities to be located precisely in the cortex. Finally, the FiCD values simultaneously consider fiber numbers, the FA values and the size of the corresponding cortical unit, which may be a more suitable indicator to measure structural connectivity strength and would likely provide more pathophysiological information that can lead to a better understanding of the mechanisms of cognitive impairment in ESRD patients.
In the present study, we described the first application of FiCD mapping to a cohort of participants clinically diagnosed with ESRD. This study aimed to explore the underlying structural connectivity across the entire cerebral cortex in ESRD patients. We additionally investigated whether associations were found between these abnormal cortical connectivities and the clinical variables in patients with ESRD.

Participants
The study was approved by the Medical Ethics Committee of the Affiliated Hospital of Qingdao University, China, and informed consent was obtained from all the participants. All the procedures involving human participants conformed to the 1964 Declaration of Helsinki and subsequent revisions or similar ethical standards. For this hospital-based prospective case-control study, 30 ESRD patients who were diagnosed with renal failure, defined by a glomerular filtration rate(GFR) less than 15 mL/min/1.73 m 2 , and who had undergone regular hemodialysis were recruited from the nephrology and renal transplantation department at our hospital between January 2019 and January 2020. Concurrently, 20 healthy, age-and sex-matched volunteers were recruited from the local community. To avoid possible coupling effects, all the subjects in the present study were right-handed and aged between 20 and 50 years. The demographic and clinical data of each ESRD patient were acquired from the electronic medical records at our hospital. All the ESRD patients completed a laboratory examination after hemodialysis and before MR imaging within 24 h, that included the serum creatinine level, urea level, hemoglobin level, hematocrit level, cholesterol and triglyceride level, serum potassium, serum sodium, serum calcium and parathyroid hormone(PTH) level evaluations. All the subjects completed a questionnaire from the Mini-Mental State Examination(MMSE) before MR imaging.
WMH severity was evaluated using the grading scale presented by Fazekas on FLAIR images based on the agreement of two experienced neurologists(mild, Fazekas score 1-2; moderate, Fazekas score 3-4; severe, Fazekas score 5-6). Periventricular hyperintensity(PVH) and deep white matter hyperintensity(DWMH) were scored separately, and the Fazekas score was the sum of the PVH and DWMH scores. The PVH scores were as follows: 0, absent; 1, caps or pencil-thin lining around ventricles; 2: smooth halo around ventricles; and 3: irregular PVH lesions extending into the deep white matter. The DWMH scores were as follows: 0, absence; 1, punctate foci; 2, beginning confluence of foci; and 3, large confluent areas.
The shared exclusion criteria for patients and control subjects were as follows: (a) history of severe head injury or obvious brain lesions(including severe WMHs and lacunar infarction) on T2-fluid-attenuated inversion recovery(FLAIR) images; (b) neurodegenerative diseases(e.g., epilepsy, Parkinson's disease, or Alzheimer's disease); (c) acute cerebrovascular disease or peripheral arterial occlusion; (d) chronic liver failure or heart failure; (e) a history of psychiatric disorders in any control subject or a history of major psychiatric disorders in any subject; (f) severe metabolic diseases (e.g., primary hyperparathyroidism or diabetes); (g) substance abuse, including drugs, alcohol or cigarettes; (h) pregnancy or lactation at the time of the study; (i) contraindications to MRI. Three ESRD patients were excluded because of obvious WMH lesions (Fazekas score 5-6), and two ESRD patients were excluded because of lacunar infarction. The final study population included 25 ESRD patients and 20 healthy controls(HCs). Of the 25 ESRD patients, 15(60%) also had renovascular hypertension, 10(40%) had hyperlipidemia, 18(72%) had anemia, and 24(96%) had high PTH levels.

Data preprocessing
First, a gray matter-white matter(GM-WM) interface of the whole cortex was constructed for each subject based on 3D-T1WI using the FreeSurfer software package (Fischl, 2012) (Version 5.3.0, http:// surfer. nmr. mgh. harva rd. edu/) by performing the following steps (Supplementary Fig. 1): correction of uneven fields, removal of nonbrain tissue, automated Talairach transformation, subcortical WM and deep GM segmentation (Fischl et al., 2002), tessellation of the GM-WM boundary, automated topology correction (Segonne et al., 2007), and surface deformation (Dale et al., 1999). These steps were performed according to the intensity gradients to optimally place the GM/WM and GM/cerebrospinal fluid boundaries at the greatest changes in intensity, thus defining transitions to other tissue classes. Second, the diffusion images were corrected for head movement and eddy current, and the brain tissue was extracted using the FSL toolbox (Jenkinson et al., 2012).

FiCD mapping
By combining a diffusion tractography technique and cortical surface-based analysis, "FiCD mapping" is based on a set of internal algorithms developed by Liu et al. (2017) to measure the structural connectivity of the entire cortex, shows good intra-and interindividual reproducibility and can accurately reflect the involved cortical regions. FiCD mapping primarily includes the following steps: construction and division of the GM-WM interface, matching of the GM-WM interface and diffusion space, fiber track constructions, acquisition of the whole-cortical FiCD map, spatial normalization and smoothing(see Supplementary Materials and Supplementary Fig. 1).

Group-level comparison of FiCD maps
To identify regions with significant differences between the groups, vertexwise statistical comparisons of the FiCD value of the common brain surface between the groups were performed using the General Linear Model(GLM) of the Query Design Estimate Contrast(Qdec) module in the Free-Surfer software package. Statistical significance was set to p < 0.01(Monte Carlo Null-Z simulation for multiple comparison correction with 10,000 iterations). The mean global FiCD values of the two groups were subsequently extracted, and the differences between the groups were compared using two-sample t test in SPSS 22.0 software with a significance threshold of p < 0.05.

Group differences in demographic and clinical data
The differences in the demographic and clinical data between the groups were compared using two-sample t test and χ 2 test in SPSS 22.0 software(SPSS Inc., Chicago, IL, USA). Statistical significance was set to p < 0.05.

Correlation analyses
Pearson correlation analysis was used to assess the relationships between FiCD values(after Z transformation) and clinical variables(MMSE scores, dialysis duration, serum creatinine level, urea level, hemoglobin level, hematocrit level, cholesterol and triglyceride level, serum potassium, serum sodium, serum calcium and PTH level) in the ESRD patients. To further investigate whether the observed between-group differences in global FiCD were dependent on the cerebrovascular abnormality or directly attributable to the abnormal circulating/metabolic status in ESRD patients, the correlation between global FiCD values and WMH scores, dialysis duration, serum creatinine levels, hemoglobin levels, and PTH levels was performed using multifactor linear regression analysis. Multicollinearity was tested using SPSS when conducting multiple linear regression analyses. Statistical significance was defined at p < 0.05.

Demographic and clinical information of the participants
The demographic and clinical information of all the patients and healthy subjects is shown in Table 1. No significant differences were found in age, sex or education level between the groups (p > 0.05). The ESRD group had significantly lower MMSE scores than the healthy controls(HCs) (p < 0.001).

Group differences in fiber connectivity density
The mean global FiCD value was significantly deceased in ESRD patients compared with that in HCs (p < 0.05) ( Table 1). The vertexwise intergroup comparison showed that the FiCD values of the bilateral dorsolateral prefrontal cortex[(DLPFC), including the bilateral middle and inferior frontal gyrus], inferior parietal, lateral middle temporal, and right middle occipital cortex were significantly decreased in ESRD patients, whereas some regions in the bilateral DLPFC(including the bilateral middle and right inferior frontal gyrus), lateral temporal(including the bilateral superior and right middle temporal gyrus), and left middle occipital cortex of the ESRD group showed significantly increased FiCD values compared with those

Correlation between functional connectivity and clinical variables
In ESRD patients, we found a trend of negative correlation between the increased FiCD values of right middle frontal gyrus as compared with HCs and serum creatinine (r = −0.473, p = 0.017) and urea level (r = −0.511, p = 0.009). Similarly, a negative correlation was found between the increased FiCD values of the left middle frontal gyrus and parathyroid hormone(PTH) levels (r = −0.577, p = 0.003) and dialysis duration (r = −0.552, p = 0.004) (Fig. 2). No significant correlations were found between FiCD values and other clinical variables (p > 0.05).
In multifactor regression analysis, only the WMH scores (beta: -0.620; 95% CI: −213.110, −51.632) were negatively correlated with the global FiCD value (Table 3), indicating that the severity of WMHs was an independent predictor of impairment of the structural connectivity of ESRD patients.

Discussion
In the present study, we used an integrative framework combining a diffusion tractography technique and cortical surface-based analysis-namely, FiCD mapping-to study the whole cortical structural connectivity in ESRD patients and found that this method can be helpful to characterize microstructural changes that accompany ESRD patients from the unique perspective of GM-WM integration.
The mean global FiCD value of ESRD patients were significantly lower than those of HCs, suggesting extensive disconnection of the brain cortex in ESRD patients. Second, vertexwise statistical comparisons showed that FiCD values in the bilateral DLPFC, inferior parietal cortex, lateral temporal cortex and middle occipital cortex were significantly changed(increased or decreased), possibly suggesting regional reconstruction of the brain structural network to adapt to injury.

Impaired cortical structural connectivity in ESRD patients
Prominently decreased FiCD values were observed in the bilateral DLPFC(including in the bilateral middle and inferior frontal gyrus), bilateral inferior parietal cortex, lateral middle temporal cortex, and right middle occipital cortex, which may be consistent with the development of cognitive impairment in ESRD patients. Many previous investigators have demonstrated that ESRD patients had functional and structural abnormalities in the frontal, parietal, temporal, and occipital regions(i.e., decreased functional connectivity (Zheng et al., 2014), cortical thinning (Chiu et al., 2019) and decreased cerebral blood flow (Cheng et al., 2019) with cognitive decline during the clinically asymptomatic phase.
Decreased WM integrity has also been reported in the DTI literature (Chou et al., 2013;Drew et al., 2017), primarily in the fornix, anterior limb of the internal capsule, corona radiata, and anterior thalamic radiation, all of which have extensive anatomical connections with these four regions. In contrast to these studies, we provide a unique and appropriate basis to integrate GM and WM results because the two metrics can be accurately matched in a unified space using FiCD mapping in the present study.
Neither the current study nor previous studies showed a clear or consistent pattern in a particular region of ESRD patients. However, evidence highlights the role of the prefrontal cortex(PFC). Anatomically and functionally, the PFC area has a unique but overlapping pattern of connections with almost all sensory neocortex and motor systems and a wide range of subcortical structures, possibly allow it to exert a "top-down" influence on a wide range of brain processes (Dixon, 2015). Mechanistically, cognitive impairment is largely secondary to the increased prevalence of cerebrovascular disease, particularly cerebral small-vessel disease(CSVD), in ESRD patients (Krishnan & Kiernan, 2009). Additionally, the severity of WMHs is an independent predictor for impairment of the structural connectivity of ESRD patients in the multifactor regression analysis. Studies have shown that WMH is one of the major imaging features of CSVD (Wardlaw et al., 2013). Extensive structural disconnection in the prefrontal region is a characteristic pattern in CSVD patients with mild cognitive impairment (Liu et al., 2020). Although cognitive impairment in ESRD patients is affected by various comorbidities, the contribution of cerebral vascular diseases seems to be further supported by the impaired structural connectivity in the DLPFC in our study.

Pathophysiological significance
By localizing connectivity impairments in ESRD patients to specific cortical regions, our findings support a structural basis for these previous studies. The FiCD value enables estimation of both the axonal density of all associated fibers connected to a CU and the FA value of each associated fiber. Thus, the decrease in FiCD value may result from the loss of axons leading to atrophy of a fiber bundle across its entire cross-section of a CU (Mito et al., 2018). Another likely scenario is that reduced integrity of the WM is associated with decreased FA values (Alexander et al., 2007). This finding may be related to axon degeneration and WM demyelination. On the one hand, atrophy of a fiber bundle may indicate a reduction in the space occupied by fiber bundles, leading to changes in macroscopic structure and morphology, such as a decreased cortical thickness and GM volume (Mito et al., 2018). On the other hand, the speed and capacity of information transmission between a pair of cortical regions connected by a damaged fiber bundle may be reduced. In particular, a reduction in axon density may indicate that fewer nerve impulses are transmitted and therefore carry less information, whereas demyelination leads to slower nerve impulses and therefore less speed transmission (Rushton, 1951;Zalesky et al., 2011). Thus, the cortical regions exhibiting decreased FiCD values may lead to decreased disconnected functional connectivity due to abnormal WM fiber pathways.
Notably, the decrease in functional connectivity of local cortical regions cannot be fully explained by the interruption of structural connections. The decline in functional connectivity can be ascribed to the destruction of specific WM fibers, neuronal dysfunction, or a combination of the two (Daskalakis et al., 2008;Hagmann et al., 2008). Some regions, such as the medial frontal areas, posterior cingulate cortex, and precuneus, were not identified as showing strong connectivity differences in our study, despite often showing functional reductions in patients. This divergence might reflect that our analysis addresses a unique aspect of neuropathology; that is, connectivity was addressed rather than function. Although the present study was influenced by these limitations, future work will benefit from studying the effects of other pathological injuries in this clinical group.

Compensation mechanism of structural reorganization
Compared with HCs, the regions of increased structural connectivity described in ESRD patients suggest a compensating mechanism of structural recombination in the brain that counteracts the damage in the early stages of chronic progressive disease. Several studies have demonstrated that increases in brain structure properties are linked to the maintenance of function despite aging or continuous damage, as shown in neurodegenerative disorders (Meunier et al., 2014;Meunier et al., 2010). Similar results were found in previous studies, and notably increased local efficiency but decreased global efficiency in the modular structure of functional networks has been recently demonstrated in ESRD patients using functional MRI(fMRI) (Ma et al., 2015), indicating a disruption and reorganization in the normal balance of functional brain networks in ESRD patients. In a DTI study, a significantly increased node-clustering coefficient was found in ESRD patients . A higher clustering coefficient indicated high efficiency of information transfer for specialized processing, suggesting that the brain was likely to have increased the structural connectivity of local nodes to maintain transmission efficiency. Thus, these results highlight that the observed changes in network structural connectivity are not only a result of diffuse brain injury but should be considered an adaptive process for the brain to maintain optimal network function.
In ESRD patients, we found a trend of negative correlation between the increased FiCD values of bilateral middle frontal gyrus and dialysis duration, serum creatinine levels, urea levels and PTH levels, suggesting that compensation mechanisms for cerebral structural networks may be exhausted with progression of the disease. Additionally, the serum creatinine, urea and PTH levels may be important risk factors for cognitive impairment in ESRD patients. Hence, early interventions to prevent disease progression are warranted.

Limitations
Some limitations of the current study should be noted. First, the small sample size leads to a lack of statistical power; thus, the findings must be confirmed in larger studies. Second, the results of the FiCD mapping method are strongly dependent on the algorithm and parameter settings; these factors are adjustable. A major limitation of the deterministic fiber tracking algorithm used in this study is that the method is particularly hampered by voxels that contain multiple fiber orientations(such as crossing, bending, or fan-shaped fibers) (Mori & van Zijl, 2002). These voxels are difficult to interpret or assign to specific fiber pathways; thus, inaccurate results may have been obtained. Third, some comorbidities associated with ESRD(such as anemia, hypertension, and hyperlipidemia) were not excluded in the present study. ESRD cognitive impairment may be a common result of these comorbidities; therefore, making a strict distinction unnecessary. Fourth, the study had a cross-sectional design; future work should include longitudinal studies to address how brain networks adaptively reorganize to compensate for their impaired structural networks. Finally, these correlations between abnormal functional connectivity correlations and clinical variables did not survive stringent correction. Thus, they are considered exploratory and must be validated in a larger sample size.

Conclusion
In summary, ESRD patients showed extensive cortical structural connectivity impairment, which was related to the severity of WMHs. Additionally, a compensation mechanism of cortical structural recombination may play a role in how the brain adapts to maintain optimal network function in ESRD patients. Furthermore, the serum creatinine, urea and PTH levels may be risk factors for brain structural network decompensation in ESRD patients. Our study provides novel insight and more pathophysiological information to improve the present understanding of the mechanism of ESRD-related cognitive impairment.
Authors' contributions Author contributions included the conception and study design (MC JXH and RYD), data collection or acquisition (MC, WCJ and ZT), statistical analysis (MC and GGJ), interpretation of the results (MC, JXH, FAR, GGJ, BPR, QSS, FSL and RYD), drafting of the manuscript or revising it critically for important intellectual content (MC, RYD) and approval of the final version to be published and agreement to be accountable for the integrity and accuracy of all aspects of the work (All authors).
Funding This work was supported by the Medical and Health Science and Technology Development Program of Shandong Province (2016WS0285).

Declarations
Ethics approval The study was approved by the Medical Research Ethics Committee of the Affiliated Hospital of Qingdao University, China. All the procedures in studies involving human participants were performed in accordance with the ethical standards of the institutional and/ or national research committee and with the 1964 Helsinki declaration and its later amendments or comparable ethical standards.

Consent to participate
The patients/participants provided their written informed consent to participate in this study.