White matter damage in maintenance hemodialysis patients: a diffusion tensor imaging study

Anatomic brain abnormalities, including stroke and brain atrophy, are common in patients with kidney failure [1‐5]. Patients receiving maintenance dialysis are also known to have a high burden of white matter disease [1, 5, 6], a finding closely correlated with incident clinical stroke [7‐9]. White matter disease is associated with adverse outcomes in the general population and in CKD [10], including higher risk for overt strokes and cognitive impairment [8, 11].

Through measurement of water diffusivity in the brain, diffusion tensor imaging (DTI) allows for the characterization of alterations to white matter tracts in addition to localization of white matter damage to specific brain regions [12]. In those without kidney disease, DTI shows promise in detailing white matter disease related to stroke [13], aging, [14] Alzheimer’s disease, [15] and multiple sclerosis [16]. Although multiple prior studies have assessed the general burden of white matter disease in kidney disease patients using conventional MRI, very few have utilized DTI [17, 18]. Of those investigating DTI in dialysis patients, firm conclusions have been limited by homogeneous populations, the use of low resolution magnetic resonance imaging (MRI), and the lack of use of tract-based spatial statistics to improve the sensitivity and objectivity of DTI [19].

Dialysis patients are at particularly high risk for stroke [20] and cognitive impairment [21, 22], and white matter disease may be a precursor to each of these adverse outcomes; accordingly, the presence and location of DTI abnormalities in patients receiving dialysis may be of particular interest. We therefore obtained high resolution MR brain imaging in stable maintenance hemodialysis (HD) patients and in participants already scheduled for a brain MRI who had no reported history of kidney disease, utilizing DTI and tract-based spatial statistics to detect and localize damage to white matter bundle pathways.

Thirty-four HD patients and twenty six controls had adequate imaging for analysis. The mean age was similar (52 years vs 51 years for HD vs control), while the HD group had fewer women (38% vs 23%), more participants who were black (38% vs 19%) and a higher rates of diabetes (29% vs 8%), heart failure (29% vs 0%), stroke (15% vs 0%) and hypertension (85% vs 35%) (Table 1). The median dialysis vintage (time since initiation of hemodialysis) was 18 months (7, 40 months for 25th and 75th percentiles, respectively). A variety of etiologies leading to end-stage renal disease was present, with diabetes and hypertension (32%), glomerulonephritis (40%), and hereditary kidney disease (12%) being the most prominent. Fifteen (44%) of the thirty-four HD patients had more than one cardiovascular disease risk factor.

Diffusion tensor imaging in prevalent HD patients demonstrated multiple brain areas with low FA and high MD as compared to a group of participants without kidney disease. These findings are consistent with a loss of white matter integrity, mainly clustered in the anterior of the brain (fronto-temporal connections, genu of corpus callosum and fornix). This pattern of injury seen within HD patients is most similar to that seen with aging [25, 26], despite the HD group having a mean age of 52 years. HD patients with more than one cardiovascular disease risk factors appeared to account for much of the difference in white matter integrity. However, we note that the primary goal of our manuscript is to highlight the significant difference in white matter disease between HD and controls, recognizing that there are many reasons, including comorbidity, for this difference and that we cannot fully adjust for all factors.

Multiple prior studies have demonstrated an increased prevalence and severity of white matter changes in both patients with chronic kidney disease as well as those with end-stage renal disease requiring dialysis [1, 5, 6]. However, these studies have typically relied on conventional MR imaging, which does not allow for direct voxel by voxel comparison of white matter integrity between patients and controls and is poorly suited for the identification of localized differences in white matter integrity.

There are currently limited and inconsistent data on diffusion tensor imaging in patients receiving hemodialysis. Three prior studies of DTI in a population of HD patients have been published. Chou et al. performed DTI in 28 HD patients and 25 age matched controls using a 1.5 T MR scanner, demonstrating increased MD and decreased FA in the HD group versus controls [17]. A study by Kong et al. obtained DTI in 80 HD patients and 80 controls, also showed decreased FA and increased MD for the HD group in comparison with controls [18]. Each of these studies, which were limited to Asian populations, reported widespread increased MD throughout the brain, without a clear pattern across different brain regions or tracts. A third study by McIntyre et al. reported the results of a clinical trial examining if the cooling of dialysate prevented white matter damage in prevalent hemodialysis patients [27]. Participants in the cooling arm (0.5 degrees C below core body temp) showed no change in DTI measures over one year, while those dialyzed at 37 °C demonstrated an increase in FA over one year. This result appears contrary to both our findings and those previously published, which generally have agreed that decreased FA is consistent white matter injury. Our study confirms that HD patients have decreased FA and increased MD compared to controls, but differs from these prior studies by demonstrating a predominantly frontal pattern of white matter injury. The strengths of our study include the use of high resolution 3 T MRI and use of TBSS analysis, a more diverse cohort than prior DTI studies, and ascertainment of cardiovascular risk factors. The inclusion of participants of differing races, as well as those with multiple cardiovascular disease risk factors may account for the spatial differences in white matter damage. In fact, a DTI study of ESRD patients in the United States who ultimately underwent renal transplantation also demonstrated anterior pattern of white matter damage, which showed improvement after transplantation [28]. An anterior pattern of white matter damage appears most similar to that seen when DTI is performed within healthy elderly patients [25, 26]. Limitations to our study include a relatively small number of participants, a younger age than the average U.S. hemodialysis patients, and the use of prevalent HD patients. Taken together, these limitations may limit our ability to generalize these findings to the overall dialysis population. In addition, our control group is overall healthier than the hemodialysis group, reflecting an absence of kidney disease as an inclusion criterion. A greater number and severity of comorbid conditions such as vascular disease in the hemodialysis cohort may be the primary reason for our findings.

There are several possible explanations for our findings. First, white matter damage/disease is reported to be a possible precursor for clinical vascular disease, with strong associations seen between the presence of white matter disease and future risk of transient ischemic attack or stroke [7, 9]. Patients with kidney failure have a high prevalence of vascular disease risk factors (diabetes, hypertension, etc.) and have an increased risk of developing vascular disease [20, 29] which together may predispose towards greater white matter damage. There likely are additional etiologies underlying our findings. Hemodialysis patients, in comparison with PD patients, may be at high risk for recurrent brain injury due to hemodynamic changes during the procedure [30]. However, this cannot explain all of the increased risk as there is also high prevalence of white matter disease seen within those with CKD [10] as well as ESRD patients receiving peritoneal dialysis, [6] neither of whom are subject to sudden hemodynamic shifts. Finally, there is the possibility that kidney failure or kidney disease itself may predispose to brain injury [31], perhaps through retention of harmful toxins (a uremia-like effect) or through dysregulation of existing hormonal pathways, such as with mineral metabolism.

The pattern of anterior white matter damage is most similar to that observed in studies of the effect of aging on white matter integrity [25, 26]. Prior MRI studies have also demonstrated that brain atrophy is common within dialysis patients [2, 5], a finding that is pronounced in elderly individuals [32]. Taken together, these observations may suggest that kidney disease accelerates or mimics the effect of aging on the brain. Alternatively, patients receiving dialysis, particularly those who developed kidney failure from diabetes and/or hypertension, often have developed co-morbid conditions such as vascular disease at a younger age than the general population. A prolonged effect of diabetes and hypertension upon the brain could account for the severity of disease observed among hemodialysis patients in our cohort, despite relatively young ages.

Diffusion tensor imaging performed within a group of patients receiving hemodialysis versus controls without kidney disease demonstrated significant alterations in white matter integrity. Damage was primarily located within anterior brain regions, in contrast to posterior regions which were on average more similar to the control group. These findings may occur either due to a propensity for cerebrovascular disease, through hemodynamic effects mediated by hemodialysis, or be mediated by kidney failure itself. Future studies should focus on the pathophysiology and risk factors that may lead to white matter damage in kidney disease patients as well as methods to prevent or limit brain injury.