Brain function and metabolism in patients with long-term tacrolimus therapy after kidney transplantation in comparison to patients after liver transplantation

Background About 50% of the patients 5–7 years after kidney transplantation show impairment of memory, attention and executive function. Tacrolimus frequently induces neurological complications in the first few weeks after transplantation. Furthermore, tacrolimus treatment is associated with impaired cognitive function in the long-term in patients after liver transplantation. We hypothesize that long-term tacrolimus therapy is associated with cognitive dysfunction and alterations of brain structure and metabolism in patients after kidney transplantation. Methods Twenty-one patients 10 years after kidney transplantation underwent cognitive testing, magnetic resonance imaging and whole brain 31-phosphor magnetic resonance spectroscopy for the assessment of brain function, structure and energy metabolism. Using a cross-sectional study design the results were compared to those of patients 1 (n = 11) and 5 years (n = 10) after kidney transplantation, and healthy controls (n = 17). To further analyze the share of transplantation, tacrolimus therapy and kidney dysfunction on the results patients after liver transplantation (n = 9) were selected as a patient control group. Results Patients 1 and 10 years after kidney transplantation (p = 0.02) similar to patients 10 years after liver transplantation (p<0.01) showed significantly worse cognitive function than healthy controls. In contrast to patients after liver transplantation patients after kidney transplantation showed significantly reduced adenosine triphosphate levels in the brain compared to healthy controls (p≤0.01). Patients 1 and 5 years after kidney transplantation had significantly increased periventricular hyperintensities compared to healthy controls (p<0.05). Conclusions Our data indicate that cognitive impairment in the long-term after liver and kidney transplantation cannot exclusively be explained by CNI neurotoxicity.


Methods
Twenty-one patients 10 years after kidney transplantation underwent cognitive testing, magnetic resonance imaging and whole brain 31-phosphor magnetic resonance spectroscopy for the assessment of brain function, structure and energy metabolism. Using a cross-sectional study design the results were compared to those of patients 1 (n = 11) and 5 years (n = 10) after kidney transplantation, and healthy controls (n = 17). To further analyze the share of transplantation, tacrolimus therapy and kidney dysfunction on the results patients after liver transplantation (n = 9) were selected as a patient control group. inclusion criteria were age between 18 and 80 years, German as native language and stable tacrolimus therapy in standard dosage (stable tacrolimus trough levels above 5μg/l). Only patients with tacrolimus therapy were included because it is the standard immunosuppressive drug used after KT [11]. Exclusion criteria were additional transplantation of other organs, kidney re-transplantation (>3 months after first KT), neurological or psychiatric diseases, regular intake of drugs affecting brain function, contraindications for magnetic resonance imaging (MRI), acute transplant-rejection or acute infection and decompensated heart-, liver-or kidney function at study inclusion. After application of the inclusion and exclusion criteria 51 of the patients remained available for study participation. Of these, 21 agreed to participate (Fig 1). For the cross-sectional analysis it was intended to assess 10 patients 1 and 5 years after KT as patient control groups, respectively. Thus, all patients registered in the kidney transplantation outpatient clinic database about 1 or 5 years after KT were screened to find comparable subjects to the 21 patients 10 years after KT concerning age, sex, education and time on dialysis before KT (Fig 1).
Finally, 42 patients after KT were included: 21 patients about 10 years (KT10), ten patients about 5 years (KT5) and eleven patients about 1 year after KT (KT1) ( Table 1). All KT patients were treated with standard dose tacrolimus. However, all patients received at least one further drug for immunosuppression: 32 patients were additionally treated with prednisolone and mycophenolic acid, seven additionally with prednisolone and three additionally with mycophenolic acid. Eight patients of KT10 had been treated with ciclosporin for 261.4±490.4 days before the immunosuppression was changed to tacrolimus.
Nine patients about 10 years after liver transplantation (LT) were selected as a further patient control group. These patients are a subset of 85 patients who participated in a study analysing cognitive function and brain alterations in patients after liver transplantation [8]. The subgroup of patients after liver transplantation was selected according to time since transplantation, treatment with standard dose tacrolimus and stable tacrolimus trough levels above 5μg/l, age, education and sex to be comparable to the KT patient cohort. The underlying liver disease of the nine selected patients was primary sclerosing cholangitis (n = 5), polycystic liver disease (n = 2) and hepatitis B virus infection induced cirrhosis (n = 2). Seven liver transplantation patients were treated additionally with at least one further immunosuppressant: three with prednisolone, two with prednisolone and azathioprine, one with prednisolone and mycophenolic acid and one with mycophenolic acid. Five of these 9 patients had been treated with ciclosporin for 91.4±137.8 days before the immunosuppression was changed to tacrolimus.
Data of 33 healthy controls was already available. Of these, 17 (HC) adjusted for age, gender and education were selected and served as a control group (Table 1).
2 KT patients (n = 1 of KT10 and KT1, respectively) had incomplete MRI measurement and thus were excluded from MRI analysis. All subjects gave written informed consent. The study was approved by the local ethics committee at Hannover Medical School and performed according to the World Medical Association Declaration of Helsinki of 1975 (as revised in 2013). No donor organs were obtained from executed prisoners or otherwise institutionalized persons. None of the transplant donors were from a vulnerable population and all donors or next of kin provided written informed consent that was freely given.

Methods
All subjects underwent a standardised physical neurological examination (H.P.). Age, sex, years of education, underlying kidney disease (diseases were grouped according to polycystic kidney disease, nephropathy, nephritis and other), presence of arterial hypertension, diabetes mellitus or hypercholesterolemia, glomerular filtration rate (GFR) at the time of study inclusion in ml/min, bilirubin total in μmol/l at the time of study inclusion, medication, years since transplantation, living or deceased donor, time on dialysis before KT, tacrolimus dosages and tacrolimus trough levels of each visit at the outpatient clinic were assessed and documented from case records. The GFR was used to classify patients into patients with or without chronic kidney disease grade III. The mean tacrolimus trough level and the total tacrolimus dosage for each patient were calculated with last observations carried forward as previously described [8].

Psychometric testing and Beck's Depression Inventory
Cognitive function was assessed by the Repeatable Battery for the Assessment of Neuropsychological Status (RBANS) [17][18][19]. Indications of depression were determined by the Beck's Depression Inventory [20].

Magnetic resonance imaging and-spectroscopy
All subjects underwent Magnetic resonance imaging and-spectroscopy examinations at 3T (Verio, Siemens, Erlangen, Germany). Cerebrovascular lesions and alterations of brain volume were assessed by semi quantitative assessment with a 12-channel phased-array head coil. The MRI protocol consisted of a T2 weighted turbo spin echo sequence (TSE) with triple echos All scans were made in axial section with an acceleration factor of 2. A field of view (FOV) with 256x208 mm 2 and a voxel size (VS) with 1x1x3 mm 3 were applied for TSE and GRE scans, while a FOV with 256x224 mm 2 and a VS with 1x1x1 mm 3 for MPRAGE scan, and a FOV with 230x194 mm 2 and a VS with 1x0.9x5 mm 3 for FLAIR scan were used. The details of the sequences are displayed in S1 Table. Periventricular hyperintensities (PVH) and white matter hyperintensities (WMH) were assessed visually using the FLAIR and GRE sequences and semi quantitative graded by the Scheltens Scale [21]. Furthermore, the ventricular width at the level of the caudate nucleus (VWCN) and semioval centre (VWSC) were measured in mm using the FLAIR sequence. The assessed structural MRI values are displayed in Fig 2. For 31-phosphor magnetic resonance spectroscopy (31P-MRS) a non-localized single pulse 31P-MRS free induction decay sequence (TR/TE = 2000/0.23 ms, 64 acquisitions, flip angle of 50˚) was used. A double-tuned 1H/31P transmit/receiver volume head coil (Rapid Biomedical, Würzburg, Germany) was used for 31P-MRS scans. The data of 31P-MR spectra were processed with an adapted LCModel version, with the spectral basis sets calculated by the VeSPA program [22] to estimate global concentrations of the brain phosphorous metabolites Adenosine triphosphate (ATP), Phosphocreatine (PCr), Inorganic phosphate (Pi), Nicotinamide adenine dinucleotide (NAD), Phosphomonoester (PME) and Phosphodiester (PDE). The obtained 31P metabolite values were subsequently corrected for T1 saturation as previously described [23]. An exemplary spectrum of 31P-MRS of a patient (male, 57 years old) together with overlaid LCModel analysis is shown in Fig 3. Each 31P metabolite concentration was determined as a ratio to the sum of total phosphorus metabolite values without unit. The Cramer-Rao lower bound (CRLB) of the spectral analysis was used as quality criterion for estimated metabolite values, as recommended by the author of LCModel [24], i.e. only metabolites estimated with a CRLB less than 25% were considered for further analyses. All MRI and MRS were visually inspected by two experienced neuroradiologists to exclude subjects with morphological abnormalities or artefacts.

Statistical methods
Kolmogorov-Smirnov test was applied to assess normality of distribution. Kruskal-Wallis-Test and Mann-Whitney-U test for abnormally distributed values and Analysis of variance (ANOVA) with post-hoc Bonferroni correction for normally distributed values were applied to test for significant group differences. Categorical variables were assessed by Chi-squared test. Pearson test (normal distribution) and Spearmen rank test (abnormal distribution) were used for correlation analysis. A backward multivariate linear regression analysis was applied to identify independent prognostic factors for cognitive function considering the RBANS Total scale as dependent variable. Independent variables for the analysis including all subjects were age, diagnosis (KT, liver transplantation or healthy control), PVH occipital, ATP and PME. The regression coefficient, p value and confidence interval (CI) are displayed.
Normally distributed values are shown as mean ± standard deviation, abnormally distributed values are shown as median with interquartile (IQ) range. A p-value �0.05 was considered significant for all tests applied. The statistical calculations were performed with SPSS 24 (IBM, Armonk, NY).

Characteristics of patients and controls
The physical neurological examination was normal in all subjects. Age, sex and years of education did not significantly differ between the patient groups and between patients and healthy controls. In accordance with the design of the study, the time since transplantation differed significantly between the patient groups. The KT1 patients had a significantly higher tacrolimus mean trough level than all other patient groups. Directly after transplantation higher doses of tacrolimus are needed leading to high trough levels. With time after transplantation the tacrolimus dose is reduced. Thus, the time since transplantation explains the significantly higher mean tacrolimus trough level in the KT1 patient group. The time on dialysis before KT, underlying kidney disease, the presence of arterial hypertension, diabetes mellitus, hypercholesterolemia or chronic kidney disease grade 3 according to the GFR did not differ between the KT patient groups. Interestingly, more patients in KT10 and KT1 had received their organ from a deceased donor compared to KT5 (p = 0.02). All patients after liver transplantation had received organs from deceased donors. Compared to the KT patients arterial hypertension (p<0.001) and hypercholesterolemia (p = 0.02) were less often present in liver transplantation patients while diabetes mellitus was equally frequent (p = 0.99). Liver transplantation patients had a significantly higher GFR at the time of study inclusion and in contrast to all three KT patient groups none of the liver transplantation patients had a chronic kidney disease grade 3 (p<0.001) ( Table 1).

RBANS and Beck's Depression Inventory
All patient groups and controls had a median Beck's Depression Inventory score within the normal range (mean�6) and the groups did not differ significantly (p = 0.76).
Three (14%) patients of KT10 as well as one patient of KT1 (9%) and LT (11%), respectively, had a pathological RBANS test result (percentile <10%). The mean RBANS results of the index scores immediate memory (p = 0.02), visuospatial/constructional (p = 0.001) and delayed memory (p = 0.02) as well as the RBANS Total scale (p = 0.001) differed significantly between the groups. The pairwise group analysis showed that patients 1 year after KT achieved significantly worse mean results in the index scores immediate memory (91.0±16.  (Table 2 and Fig 4). In conclusion, all patients-after KT and liver transplantation-showed an impaired visuospatial/constructional ability compared to healthy controls. Interestingly, only the patients after liver transplantation showed impaired immediate memory and reduced attention compared to healthy controls.
The RBANS results of the KT patients did not correlate with the tacrolimus total dose (S1

Magnetic resonance imaging and spectroscopy
Although data showed higher periventricular hyperintensities and white matter hyperintensities values in general in the patients irrespective of the time interval since transplantation a significant difference from controls could be shown only for a few locations: patients 1 year after KT had a significantly increased extent of occipital and total periventricular hyperintensities compared to healthy controls (p<0.05 and p<0.01, respectively). Furthermore, the patients 5 years after KT showed a significantly increased extent of total periventricular hyperintensities compared to healthy controls (p = 0.01). Interestingly, the KT patients 10 years after transplantation showed no significant differences compared to healthy controls in all assessed MRI parameters. Concerning white matter hyperintensities and the ventricular widths no significant group differences were found (Table 3 and Fig 5A-5C).
The phosphor spectroscopy showed significant differences between the groups for ATP (p<0.001). ATP concentrations in all 3 KT patient groups were significantly reduced compared to healthy controls (10 years after KT p<0.001, 5 years after KT p = 0.001, 1 year after KT p = 0.01) ( Table 4 and Fig 6). Concerning the metabolites PCr, Pi, NAD, PDE and PME no significant differences were found between the patient groups and healthy controls. The correlation analysis including only patients after KT, however, showed a significant positive correlation between PME levels and the RBANS domain score Immediate memory (r = 0.43, p<0.01)   grade III, diabetes mellitus and hypercholesterolemia had no effect on the MRI/MRS parameters of the KT patient groups.

Linear regression
A backwards linear regression analysis with the RBANS Total scale as dependent and age, diagnosis (KT, liver transplantation or healthy control), PVH occipital, ATP and PME as independent factors was applied. PME (n = 65, regression coefficient 356.

Discussion
This single center cross-sectional observational study investigated cognitive function, brain structure and metabolism in patients with standard dose tacrolimus therapy 10 years after  kidney transplantation. The results were compared to patient control groups consisting of patients 1 and 5 years after KT and tacrolimus treated patients who underwent liver transplantation about 10 years ago as well as healthy controls. Both, patients after KT and patients after liver transplantation showed cognitive impairment compared to healthy controls adjusted for age and education. Furthermore, KT patients showed an increased extent of periventricular hyperintensities and reduced brain ATP concentrations compared to healthy controls. Tacrolimus is currently the standard immunosuppressive drug used after kidney transplantation to prevent rejection [11]. However, long-term tacrolimus therapy is accompanied by several significant adverse effects such as renal dysfunction, cardiovascular disease and malignancy [12]. Furthermore, tacrolimus therapy is associated with neurological complications in the first few weeks after transplantation [13] and these are expected in the long-term as well. Several possible pathomechanisms are discussed: Tacrolimus induced cardiovascular risk factors [12] enhance atherosclerosis and microangiopathy which may lead to cerebrovascular events and subsequently impaired brain function, tacrolimus might cause neurodegeneration by inhibiting the cerebral immune system [15] and tacrolimus might impair the cerebral energy metabolism [14].
While the improvement of cognitive dysfunction associated with severe chronic kidney disease and hemodialysis within 1 year after KT is well described [3,4], interestingly, only few studies investigated the long-term neurological outcome after KT. Troen and colleagues assessed cognitive function in 183 patients approximately 7 years after KT. One third of these patients showed an impaired memory, about half impaired attention and mental processing speed and about 40% impaired executive function [6]. In another study Gelb and colleagues examined cognitive function in 42 patients approximately 5 years after KT and compared their results to 45 patients with chronic kidney disease and 49 healthy controls [7]. Both patient groups showed worse verbal memory and worse executive functioning skills than controls, while there was no significant difference between the two patient groups. Both studies indicate that cognitive dysfunction is present in KT patients in the long-term after transplantation. However, the underlying cause was not addressed. Data showing that cognitive function significantly improves within the first year after KT indicate that a secondary decline of cognitive function might occur in the long-term. Cognitive deterioration after transplantation was previously described in liver transplantation patients [9] and we recently showed cognitive impairment in a subset of patients in the long-term after liver transplantation who had been treated with low dose CNI after they had developed significant renal dysfunction with standard CNI therapy. We hypothesized that these patients might be hypersensitive against CNI toxicity [8].
Corresponding to the literature our results show that cognitive impairment is present in patients after KT compared to healthy controls. Especially visuospatial/constructional abilities seem to be affected in the long-term. The impairment of immediate memory in patients 1 year after KT might be a residue from dialysis related encephalopathy as patients on dialysis were shown to have especially memory and attention deficits [25]. The impairment of visuospatial/ constructional ability in the long-term after transplantation was previously described in patients 10 years after liver transplantation who were on a reduced CNI dose due to CNI induced kidney injury [8]. The liver transplantation patients included in this study who were on a standard dose immunosuppressive therapy regimen showed a similar visuospatial/constructional ability as the KT patients, however, probably due to a higher variability of the test results no significant difference compared to healthy controls was found. Taken together, similar to patients after liver transplantation, also patients after KT show cognitive impairment in the long-term after transplantation. This might be related to long-term CNI therapy. All patient groups-KT and liver transplantation patients-showed worse visuo-constructive abilities than controls, while language seemed preserved in all groups and attention seemed preserved especially in the KT patients. Thus deficits of visuo-constructional abilities might be related to tacrolimus therapy-a variable that is shared by all patient groups, while the deficit in alertness could be a residuum of hepatic encephalopathy.
Arterial hypertension and impaired kidney function might have an influence on cognitive function. However, it can be assumed that the impact of both on cognitive function in our study cohort is less important than that of the assumed vulnerability towards tacrolimus. For example, patients 5 years after KT showed only slight deficit in the visuospatial/constructional ability and no overall cognitive impairment despite having arterial hypertension and impaired kidney function just like the other KT groups. Furthermore, the patients after liver transplantation showed cognitive impairment although they had a normal kidney function and less often arterial hypertension than KT patients.
To address two of the possible tacrolimus associated pathomechanisms underlying cognitive dysfunction in our patients we performed MRI and 31P-MRS to assess cerebrovascular injury and cerebral mitochondrial energy metabolism. Structural brain alterations have been described in patients with significant chronic kidney disease with and without dialysis [26] , [27]. However, the effect of KT upon brain structure is not well described. Gupta and colleagues performed MRI and psychometric testing in eleven patients before and three months after KT [28]. They showed an improvement of cognitive function accompanied by an improvement of brain white matter integrity measured by diffusion tensor imaging in tracts associated with memory and executive function after KT. Zhang and colleagues measured white matter structural connectivity using diffusion tensor imaging in 21 patients before and 1 month after KT [29]. Although the MRI values of patients did not normalize, the diffusion tensor imaging results showed a significant improvement after KT. In both studies the authors concluded that structural brain alterations in chronic kidney disease patients might be at least partially reversible after KT.
Considering the assumed neurotoxicity of tacrolimus and the increased prevalence of cardiovascular risk factors in patients after KT [12], alterations of brain structure have to be expected. Accordingly, we found an increased extent of periventricular hyperintensities in patients about 1 and 5 years after KT compared to healthy controls. Surprisingly, KT patients about 10 years after transplantation did not significantly differ from healthy controls and none of the KT patient groups differed significantly from healthy patients concerning white matter hyperintensities or ventricular widths. Other groups discussed that structural brain alterations are at least partially reversible after KT [28,29], however the lack of a difference between the patients 10 years after KT and controls in our study might well be due to the variability of these parameters and the limited number of subjects included into the study.
Another tacrolimus associated pathomechanism underlying cognitive dysfunction in patients after KT is the possible impairment of the cerebral energy metabolism. Illsinger and colleagues showed in vitro that clinically relevant tacrolimus concentrations impair the mitochondrial energy metabolism in human umbilical endothelial cells [14]. Furthermore, tacrolimus induced alterations of mitochondrial function was described in human cell lines [30]. Both studies indicate that tacrolimus impairs the mitochondrial metabolism. Based on these in vitro studies we hypothesized that tacrolimus associated impairment of mitochondrial function might be connected to long-term cognitive impairment in patients after transplantation. We found significantly reduced ATP concentrations in patients after KT compared to healthy controls. Furthermore, PME concentrations correlated significantly to cognitive function.
Because ATP is the main energy source for brain cells and represents mitochondrial activity [31] the reduced ATP concentrations in KT patients indicate altered mitochondrial function and might be associated to cognitive impairment. PME concentrations represent membrane turnover of the brain [32] and consequently are connected to brain metabolism. In conclusion, our results indicate an impaired brain metabolism in KT patients. Interestingly, the liver transplantation patient control group which as well received tacrolimus for immunosuppression showed no alterations of the brain energy metabolism compared to controls. Thus, other factors besides tacrolimus must be involved. Kidney dysfunction might play a significant role in this respect. It might reduce the ability of the kidney to protect from CNI toxicity or it leads to an altered pattern of tacrolimus metabolites inducing neurotoxicity. Significant impairment of cognitive function was recently described in patients long-term after liver transplantation who had a history of kidney dysfunction and still decreased GFR [8]. The liver transplantation patients selected for this study had no kidney dysfunction at study inclusion (Table 1). This might explain why the liver transplantation patients in this study had no altered mitochondrial energy metabolism despite receiving tacrolimus therapy in similar doses gaining similar blood levels. Thus, the individual vulnerability towards tacrolimus associated toxicity needs to be considered as well. In conclusion, the tacrolimus mean trough level or tacrolimus dose alone are not a sufficient explanation for altered brain energy metabolism. In accordance, in our study neither was correlated to cognitive function or the cerebral energy metabolism.
Several limitations apply to our study. It has a limited transferability to other centres and unfortunately no data from before KT was available. The results of our study might be influenced by underlying diseases such as diabetes, other immunosuppressants, especially prednisolone, and most patients were taking several other drugs besides tacrolimus. Unfortunately, patients after KT who only received tacrolimus since transplantation as well as patients with a CNI-free or prednisolone-free maintenance therapy since transplantation were not available. The sample size limits statistical power and inhibits a statement on clinical impact. Furthermore, the different underlying diseases of patients after liver transplantation and KT limit comparability, but also allow the conclusion that cognitive impairment in the long-term after liver and kidney transplantation cannot exclusively be explained by CNI toxicity.

Conclusions
In conclusion our results indicate that KT is associated with cognitive impairment and alterations of the brain energy metabolism in the long-term. The underlying pathomechanism seems to be complex and besides many other factors tacrolimus might be involved. Further investigation in a bigger study cohort is needed to clarify the role of tacrolimus and the impact of several variables such as concomitant disorders, extent of kidney dysfunction, sex and others. Furthermore, brain ATP in patients with chronic kidney disease and cognitive impairment before transplantation needs to be analysed in the future.
Supporting information S1 Dataset. This dataset contains all data underlying our results. (XLSX) S1 Table. This table contains