https://orcid.org/0000-0002-8107-7338
ABSTRACT
Chronic kidney disease is associated with cognitive impairment although the underlying mechanisms are still not fully understood. Characterization and efficient monitoring of the cognitive impact of kidney disease and ensuing therapies are critical for the accurate clinical management of patients. A vast array of imaging modalities, biomarkers, and sensors have shown relevance for the assessment of cognitive impairment. Knowing the potential and limitations of these paraclinical techniques is a necessary condition to improve the understanding of this phenomenon and to design monitoring protocols and guidelines applicable to this clinical population. The goal of this review is to provide an overview of current imaging modalities and biomarker sources available to the community, for the benefit of the research and clinical community.
Watch the video of this contribution at https://academic.oup.com/ndt/pages/author_videos
INTRODUCTION
Chronic kidney disease (CKD) has been known for a long time to be associated with cognitive impairment, but many questions remain about the underlying causes and how to establish assessment and monitoring approaches that are robust, reproductible, and feasible. A wide variety of imaging and physiological biomarkers have been considered as potentially useful for the cognitive assessment of CKD patients, including early diagnosis of subclinical cognitive effects. This review covers magnetic resonance and isotopic imaging, electroencephalography, cerebrospinal fluid and blood biomarkers, as well as sensor-based devices, aiming to present a comprehensive, albeit far from exhaustive, summary of available resources applicable to the investigation and monitoring of CKD-related cognitive effects, as well as the relevant biomarkers that each modality allows to extract (Fig. 1). We strongly advocate that the research and clinical community should collaborate towards exploring the potential and complementary nature of the described modalities.
Graphical overview of the modalities covered in this review.
MAGNETIC RESONANCE IMAGING
Today, magnetic resonance imaging (MRI) studies represent an essential part of the diagnostic approach to neurocognitive disorders (NCD) [1]. In addition to differential diagnosis, MRI biomarkers may detect early, subclinical changes. Advanced brain MR techniques allow the non-invasive assessment of different aspects of brain physiology that can be combined in a single multi-modal session [2, 3]. Because no administration of contrast media is needed, these techniques are of particular interest in patients with kidney disease. In the following, the most important modalities and their application in CKD patients will be discussed.
Structural MRI
Structural brain MRI (sMRI) evaluates brain anatomy. It allows the quantitative assessment of whole-brain or region-specific cortical thickness by MR volumetry using high resolution T1-weighted sequences as well as the qualitative assessment of microbleeds and white matter hyperintensities, all of which have been associated with cognitive decline [4]. In CKD stage 2–5 and kidney transplant recipients, sMRI has consistently revealed signs of small vessel disease including silent brain infarcts, white matter hyperintensities and microbleeds in addition to brain atrophy. Overall, anomalies were correlated with kidney function parameters and improved after kidney transplantation [5–22].
Diffusion MRI
Diffusion MRI relies on a sequence sensitive to the Brownian motion of water molecules. The diffusivity of water protons depends on the adjacent tissue microstructure and may be affected by various disease processes [23]. Main clinical indications of diffusion-MRI therefore include acute brain ischemia, evaluation of brain tumors, and detection of white matter disease. While diffusion-weighted imaging evaluates tissue diffusivity, diffusion tensor imaging (DTI) additionally assesses diffusion directionality and evaluates the integrity of white matter tracts. The main diffusion-derived parameters include the apparent diffusion coefficient and fractional anisotropy (FA), respectively. Similar to the general NCD setting [24, 25], in CKD patients, DTI has shown higher diffusivity and lower FA values, which correlated with clinical cognition markers and were improved after kidney transplantation [12, 16, 22, 26–35].
Arterial spin labeling
Arterial spin labeling (ASL) evaluates cerebral blood flow (CBF) using magnetically labelled water protons as an endogenous tracer through generation of a subtraction image [36]. Altered perfusion patterns may be seen in various disease states including cerebrovascular disease, brain tumors, and epilepsy. Given the relationship between cerebral perfusion, brain metabolism, and activity, ASL-derived CBF has been evaluated as a non-invasive functional cognition marker [37]. Indeed, region-specific CBF changes have been reported in dementia subtypes and corresponding pre-stages [38, 39]. Among patients with CKD and end-stage kidney disease (ESKD), ASL has revealed increased CBF compared with healthy controls and decreased CBF after dialysis start and kidney transplantation. In most studies, CBF correlated with cognitive performance but was, however, strongly influenced by hemoglobin level or hematocrit [16, 35, 40–43].
Functional MRI
Functional magnetic resonance imaging (fMRI) is widely used for the assessment of brain function through the BOLD effect (blood oxygen level dependent). Resting-state fMRI (rs-fMRI) is particularly attractive in this context, because it does not involve the execution of tasks that may prove challenging for patients. A 2018 review [44] highlighted the potential of rs-fMRI to probe the mechanisms behind cognitive deficits in Parkinson's disease. Functional connectivity (FC), i.e. the correlation between BOLD time-series extracted from regions of interest, is showcased as a powerful metric. Similar approaches have been applied to the study of other neurodegenerative diseases, including correlation with sMRI and diffusion MRI [45].
Regarding application to CKD, local differences between the activation patterns of children with and without CKD performing a working memory task were demonstrated [46]. Many studies have explored metrics such as regional homogeneity (ReHo) and (fractional) amplitude of low-frequency fluctuations (ALFF), which reflect local physiological effects closely related to connectivity. Liang et al. [47] showed local ReHo changes in ESKD patients when compared to controls, whereas Yu et al. [48] demonstrated localized differences in ReHo and ALFF between dialysis-dependent and non-dialysis-dependent patients. These studies hinted at a correlation between the imaging metrics and urea and creatinine levels. Correlation-based FC has been shown to differ between young CKD patients and controls, suggesting connectivity deficits in attention circuits, which may represent an early marker of cognitive decline [49].
Two independent groups [50, 51] have shown a correlation between MoCA scores and rs-fMRI metrics in hemodialysis (HD) and non-dialysis patients, respectively. It should be noted that both studies resorted to sophisticated analysis strategies, including FC between resting-state networks, dynamic FC, and structural-functional network coupling.
ISOTOPIC IMAGING
Isotopic imaging enables the non-invasive location of molecular and pathophysiological processes in the body, after injection of a molecule of interest coupled with a radioactive isotope (radiotracer).
Disruption of the blood–brain barrier (BBB) is a complex and important common mechanism to several neurodegenerative diseases and systemic diseases associated with cognitive impairment and could play a major role during cognitive impairment associated with CKD [52]. Brain single-photon emission computed tomography coupled with CT scanner (SPECT/CT) with 99mTc-diethylene-triamine-penta-acetic acid (DTPA) enables the disruption of the BBB to be visualized and quantified. DTPA is a small aminopolycarboxylic acid, which does not normally cross the intact BBB, and the presence of DTPA in the brain parenchyma reflects BBB rupture. 99mTc-DTPA brain SPECT/CT is the most used isotopic imaging in pre-clinical studies to evaluate BBB permeability [53, 54], but also in clinical studies [55–57], notably in stroke. Cerebral positron emission tomography (TEP) with 68Ga-EDTA has also been used to quantify BBB disruption in patients [58].
Several models of CKD in rodents showed increased BBB permeability. The BBB disruption, quantified using brain 99mTc-DTPA SPECT/CT imaging, was correlated with both cognitive impairment and indoxyl sulfate levels [54]. Gupta et al. found an increased BBB permeability in ESKD patients [59] compared with control participants, using 99mTc-DTPA SPECT/CT imaging, but association between cognitive impairment and BBB permeability in CKD patients is yet to be proven. The BREIN study (NCT04328415) [60] confirmed the existence of an increased BBB permeability in ESKD patients undergoing HD and previously undiagnosed for cognitive impairment compared with healthy volunteers matched in age, sex, and education, and its association with cognitive impairment.
Modifications of cerebral perfusion were also described in patients with cognitive impairment using SPECT/CT imaging with various radiotracers such as 99mTc-ethyl cysteinate dimer (ECD) [61] or 99mTc-hexamethyl propyleneamine oxime (HMPAO) [62]. In ESKD patients, HD sessions were associated with a mean decrease of 10%±15% of global and regional CBF, evaluated using 15O-H2O PET-CT imaging [63].
Brain positron emission tomography (PET) with 18F-flurodeoxyglucose (FDG) enables the study of brain metabolism. It has proven to be useful for early diagnosis of neurodegenerative diseases, and to differentiate between different subtypes of dementia, which display distinct spatial patterns of brain metabolism [64]. For example, patients with Alzheimer's disease (AD) display cortical hypometabolisms, particularly in the frontal, posterior temporal, and parietal regions [65]. It can also help to differentiate between mild cognitive impairment and early-stage AD [66, 67]. However, very few studies have evaluated brain metabolism during CKD, and its association with cognitive impairment in both pre-clinical and clinical studies. Interestingly, patients with pre-dialysis stage 5 CKD and depressive symptoms displayed decreased cerebral glucose metabolism in 18F-FDG PET imaging, especially in the prefrontal cortex [68].
These imaging techniques are not routinely used in clinical practice in CKD patients. However, we suggest that 99mTc-DTPA SPECT/CT imaging could be used to evaluate the BBB disruption in CKD patients, notably in the context of cognitive troubles, and the risk of neurological events. Indeed, BBB disruption could also increase drug neurotoxicity and predisposition to stroke in these patients [52]. Also, brain PET with FDG could help differentiate CKD-induced MCI from other neurodegenerative diseases like AD, along with MRI, in CKD patients displaying cognitive impairment.
ELECTROENCEPHALOGRAPHY
In comparison to other imaging methods, electroencephalography (EEG) stands out as a wearable cost-effective tool. Event-related potentials (ERPs) are stimulus-evoked signals hidden in natural EEG, with early peaks reflecting detection and basic processing stages and later ones reflecting higher-level cognitive processes such as attention, memory, and executive function. In cases of cognitive decline, ERP features will become disrupted.
EEG has been used to study the central nervous system functioning of CKD patients for decades. Marsh et al. [69] observed abnormally high latency and low amplitude of later ERP components in CKD patients compared with normal subjects, suggesting less effective cognitive processing. Subsequent ERP studies have revealed reduced latency and increased amplitude after the HD treatment session [70, 71]. In a study conducted by Madan et al. [72], ERPs were measured two hours before and after the HD treatment and juxtaposed with data from healthy controls. The healthy group demonstrated the lowest mean auditory P3 latency, while the CKD patients exhibited a significantly prolonged latency pre-HD, a disparity that was diminished substantially following the HD. While the shorter latency indicates better cognitive processing, Michałowski et al. [73] observed that behavioral and ERP data indicated action preparation deficits for dialysis patients, which were not observed in kidney transplant and healthy controls.
These findings hint at a potential utility of ERP measures as early markers of cognitive decline. However, integrating cognitive testing into EEG study requires complex equipment and data processing. A resting state EEG approach may offer a more feasible solution for cognitive assessment. Florea et al. [74] reported that HD caused slowing of the EEG background activity. However, the changes were related to younger age, recent initiation of HD therapy and the level of uremia, which seem to refer to the dialysis disequilibrium syndrome (DDS). The patients’ mild symptoms were not classified as DDS, suggesting that EEG enables the detection of even subtle changes below the threshold of DDS. The authors suggest that EEG indicates that the dialysis induced pathomechanisms of CNS [74].
Lizio et al. [75] designed a study protocol to investigate the impact of uremic toxins with resting-state EEG signals from four groups: CKD patients at stages 3–4 with mild cognitive impairment (MCI) (CKDMCI-3&4), representing a model of high-level uremic toxins; CKD patients undergoing HD with MCI (CKDMCI-H), representing low/medium levels of uremic toxins; MCI patients with cerebrovascular disease (CVMCI); and healthy controls, both without abnormal levels of uremic toxins. EEG recordings for the HD group were performed between treatment days to avoid acute effects of dialysis. All MCI groups indicated decreased occipital alpha power compared with healthy controls, with the lowest parietal alpha power observed in the CKDMCI-3&4. Additionally, the CKDMCI exhibited increased frontal, central, and parietal delta-theta power compared with the control and CVMCI. The authors concluded that circulating uremic toxins affect brain activity and EEG measures could be used to monitor brain function and the effects of interventions.
The presented studies employed traditional linear methods for EEG signal analysis. However, EEG signals are inherently complex and nonlinear. Päeske et al. [76] compared various linear and nonlinear measures and recommended Higuchi's fractal dimension (HFD) as a preferred EEG-based measure due to its ability to capture a broad spectrum of brain disorders. As far as we know, HFD has not been applied to EEG signals of CKD patients. Further research is needed, given the simplicity of single-channel EEG recording and the growing accessibility of certified portable EEG devices.
Today clinicians require support systems providing personalized information to guide treatment decisions for individual patients. Consequently, future research should emphasize longitudinal studies and personalized models. The availability of wearable devices allows EEG data collection during dialysis in the clinic and at home.
CEREBROSPINAL FLUID AND BLOOD BIOMARKERS
Biomarkers [77] are tools used to study MCI, AD, and other neurodegenerative diseases, including: (i) Amyloid plaque markers: Amyloid β peptides 42 (Aβ42) and the ratio of amyloid β peptides 42 to 40 (Aβ42/Aβ40). Low levels of Aβ42 in CSF are associated with the formation of amyloid plaques in the brain, a hallmark of AD. (ii) Tau pathology markers: Total tau (T-tau) and phosphorylated tau (P-tau) are markers of AD-related tau metabolism and phosphorylation. The three main phosphorylated tau proteins are P-tau181, P-tau217, and P-tau231. Hyperphosphorylation causes tau protein to accumulate in neurofibrillary tangles between neurons, thereby disrupting synaptic communication. P-tau181 is a classic AD biomarker. (iii) Neurodegenerative markers: Neurofilament light chain (NfL) is enriched in the axons of large-caliber neurons. Increased concentrations of NfL are seen in multiple neurodegenerative diseases and are considered biomarkers of disease severity. (iv) Glial activation markers: Glial fibrillary acidic protein (GFAP), a marker of astrocyte activation and astrogliosis, is increased in multiple neurodegenerative diseases. Initially, these biomarkers could only be measured in CSF, but advances in technology now make the entire amyloid/tau/neurodegeneration (ATN) [78] panel of AD biomarkers and GFAP available via blood tests [77].
CKD and CSF biomarkers
Studies have shown that CKD increases the risk of MCI, AD, and have suggested a causal direction [79]. The association between CKD and biomarkers is an emerging area of research that is important for understanding the interaction between CKD and neurological diseases, facilitating early diagnosis and tracking disease progression. However, research on the relationship between CKD and CSF biomarkers is limited and has yielded inconsistent results. For instance, two studies—one involving AD patients [80] and the other involving cognitively normal individuals [81]—found that CKD was associated with decreased CSF Aβ and increased CSF levels of T-tau. Conversely, a community-based study from the Gothenburg H70 cohort [82] found no differences in CSF biomarkers (Aβ, Tau, and NfL) between CKD and non-CKD individuals. However, among participants exhibiting Aβ pathology [83], correlations were observed for CSF Aβ42, Aβ42/40, t-tau, and estimated glomerular filtration rate (eGFR).
CKD and blood biomarkers
Several studies have previously explored the relationship between CKD and plasma Aβ levels. In the Mayo Clinic Study of Aging cohort (MCSA) [84], CKD correlated with elevated levels of Aβ42, Aβ40, and Aβ42/40 ratio. Similarly, a study performed under the Atherosclerosis Risk in Communities Study cohort [85] revealed a doubling of the urine albumin-to-creatinine ratio and lower eGFR linked to increased plasma levels of Aβ1–40 and Aβ1–42. Gronewold et al. [86] found that plasma Aβ accumulation corresponded to CKD stage, indicating the kidney's pivotal role in Aβ elimination from plasma. Kitaguchi et al. [87] investigated plasma tau levels in CKD patients undergoing HD. Regardless of HD, CKD patients exhibited elevated plasma tau concentrations, with tau observed to increase 1-hour post-dialysis. These findings suggest that the kidneys play a role in tau degradation and excretion. However, HD appears ineffective in removing tau from the blood. Two studies from the MCSA cohort demonstrated that CKD correlates with T-tau [84], as well as with P-tau181 and P-tau217 [88] after adjusting for age and sex. In the BALTAZAR study of 476 MCI participants, CKD or eGFR were independently associated with plasma P-tau181 levels, which should be taken into account when using P-tau181 [89] as a diagnostic marker for AD. Several larger studies have previously reported an association between NfL and CKD, including participants from the cognitively normal ESTHER study [90] and the Mayo Clinic Study of Aging [83]. Additionally, in a cohort of cognitively normal Swedish individuals [82], there was a statistically significant correlation between NfL and eGFR. However, in contrast, a recent study of patients from the French MEMENTO [91] cohort did not find any influence of CKD and other vascular complications on the predictive value of circulating NfL for the incidence of dementia during a 5-year follow-up. These findings suggest that the relationship between NfL and CKD may vary across different populations and study designs, indicating the need for further research. The ESTHER study [90] demonstrated stronger associations between eGFR and plasma GFAP, only in men after adjusting for age and sex. GFAP levels and kidney function have been shown to vary according to sex [92].
For many years, the diagnosis of AD was based on clinical criteria. The 2024 NIA-AA criteria [93] now include blood biomarker testing. The future of precision medicine will likely rely more on blood biomarkers to identify the high-risk population, diagnose diseases, and assess the effectiveness of treatments. Future research on the connection between CKD and blood biomarkers of AD could focus on several areas. For instance, investigating the potential mechanisms by which CKD affects the levels and clearance of blood biomarkers (such as Aβ42, Aβ40, and tau), since understanding these mechanisms can help refine biomarker assessments in CKD patients at risk for AD and adjust relevant cutoff points. Conducting longitudinal studies with representative cohorts to track changes in biomarker levels over time in CKD patients and relate them to cognitive decline and AD onset could clarify causal links and reveal early intervention points. In addition, exploring alternative or improved dialysis technologies that may effectively remove tau and other neurodegenerative biomarkers from the blood, studying the variability of CKD-AD biomarker relationships across different populations, including ethnic and geographic differences, and investigating how CKD impacts AD biomarkers differently in men and women, which could lead to more personalized management strategies for both CKD and AD, are viable research alternatives.
SENSORS
Sensor techniques offer an opportunity for non-invasive, real-time measurement of cognitive impairment parameters-biomarkers in patients with CKD. Wearable and wireless technologies would be advantageous since they allow for unobtrusive and seamless measurements [94]. Parameters related to monitoring cognitive impairment on CKD patients include cerebral hypoperfusion [95], arterial stiffness [96], mobility performance, and levels of uremic toxins [97]. In everyday routine, pervasive remote tools, such as watches and trackers, may provide details on physical activity, estimated caloric expenditure, metabolic data, sleep quality, and vital parameters [98].
Intradialytic cerebral hypoperfusion can occur during routine hemodialysis (HD) due to circulatory stress. Appropriate intradialytic cerebral perfusion monitoring can be seen as a part of cognitive decline prevention strategy. Cerebral tissue oxygenation, measured by near-infrared spectroscopy (NIRS), is known to be decreased in HD patients and is associated with weaker cognitive performance [99]. The NIRS-derived HD cerebral oxygen demand was shown to be associated with intradialytic mean arterial pressure, heart rate, and volume removal in HD patients [100]. Cerebral regional oxygen saturation (rSO2), which can be monitored by NIRS, is associated with cognitive function in non-HD CKD patients [101], and was shown to be affected by eGFR, serum albumin, and sodium concentrations in CKD patients [102]. Also, Hb levels revealed a positive and significant association with cerebral rSO2 and the relationship was proven to be linear [103]. However, patient age and sensor location can affect rSO2 values [104]. Intradialytic cerebral ischemia from cerebral oxygenation during HD, but not hypotension from continuous blood pressure (BP) measurements, correlated with decreased executive cognitive function [105].
The consistent evidence linking systemic arterial stiffness to cognitive decline and dementia has been highlighted by several studies, albeit with inconsistencies due to different study designs (cross-sectional vs. longitudinal), heterogeneity of target populations and cognitive screening tools. In general, higher pulse wave velocity (PWV) estimating systemic arterial stiffness, relevant also for CKD patients [106], predicts poor cognitive performance and a steeper cognitive decline [96].
Mobility performance, as a potential digital biomarker of cognitive impairment among HD patients, including cumulated posture duration (sitting, lying, standing, and walking), daily walking performance (step and unbroken walking bout), as well as postural-transition (daily number and average duration), can be measured using a validated pendant-sensor for a continuous period of 24-hour during a non-dialysis day [107]. Also, speed of ankle rotation, monitored by a wearable sensor during a simple intradialytic cognitive-demanding exercise, has been proposed as a digital biomarker for screening cognitive frailty in HD patients [108]. Motor capacity can be quantified by assessing standing balance and gait performance [107].
Uremic toxins are known to have an impact on cerebrovascular diseases and/or cognition [109–111]. An earlier study suggested that disturbed purine nucleotide metabolism, resulting in hyperuricemia in CKD, is a risk factor for cognitive impairment [97]. Cognitive impairment in ESKD patients in HD, assessed as an increased BBB permeability, could be associated with serum uremic toxins like indoxyl sulfate (NCT04328415), [112] and indole-3-acetic acid [113]. Multicomponent, real-time, and on-line monitoring of uremic toxins can facilitate studies of the associations between cognitive impairment and uremic toxin removal efficiency. A technology for multi-component uremic toxins’ intradialytic optical monitoring of spent dialysate [114, 115] demonstrated that optical dialysis monitoring can simultaneously reveal intradialytic removal patterns of low molecular weight and middle uremic retention solutes without any blood or dialysate sampling [116–118]. The small uremic retention solute uric acid is one of the main chromophores [114], and protein-bound uremic retention solute indoxyl sulfate is one of the main fluorophores [115]. Furthermore, time-averaged concentration estimation of uremic toxins, based on intradialytic spent dialysate measurements, has been proposed [119].
Large-scale longitudinal studies have found an association between glycemic control and cognitive decline [120]. Continuous glucose monitoring (CGM) offers benefits in terms of detecting hypoglycemia by measuring glucose in interstitial fluid every few minutes and can provide more reliable method of glycemic assessment, including asymptomatic hypoglycemia and hyperglycemic episodes [121]. Hyperglycemia metrics (mean glycose value and time in range) derived from CGM are shown to be associated with cognitive functions, especially with executive function and working memory, in older adults with type 2 diabetes [122]. However, assessing glycemic control in patients with advanced CKD and on dialysis (G4-5) can be challenging because laboratory biomarkers, such as glycated hemoglobin (HbA1c), may be biased by abnormalities in blood hemoglobin, use of iron therapy and erythropoiesis-stimulating agents, and chronic inflammation due to uremia, etc. [121], and data are scarce in this population. Further studies are needed to investigate in more detail how cognitive decline might be prevented by glycemic control in CKD patients.
CONCLUSION
In this review we aimed to provide an overview of the currently available range of biomarkers and imaging modalities that have shown potential for the detection of cognitive impairment in CKD patients. It is hoped that this review will contribute to collaborative efforts towards the pursuit of pre-clinical and clinical research on this topic and the incorporation of several modalities and biomarkers into protocols, procedures, and guidelines for the reproducible, precise, and clinically relevant assessment of cognitive status in patients with CKD.
FUNDING
AA is funded by Fundação para a Ciência e a Tecnologia via the FCT project UIDB/00645/2020 (https://doi.org/10.54499/UIDB/00645/2020).
This work is partially financed by Estonian Centre of Excellence in Well-Being Sciences (EstWell), funded by the Estonian Research Council grant nr TK218.
This article is based upon work from COST Action CA19127, supported by COST (European Cooperation in Science and Technology). www.cost.eu. COST (European Cooperation in Science and Technology) is a funding agency for research and innovation networks. Our Actions help connect research initiatives across Europe and enable scientists to grow their ideas by sharing them with their peers. This boosts their research, career and innovation.
CONFLICT OF INTEREST STATEMENT
None declared
DATA AVAILABILITY STATEMENT
No new data were generated or analysed in support of this research.
APPENDIX
CONNECT collaborators
Giovambattista Capasso, Alexandre Andrade, Mustafa Arici, Maie Bachmann, Matthew Bailey, Michelangela Barbieri, Mickaël Bobot, Annette Bruchfeld, Inga Arune-Bumblyte, Daiva Rastenytė, Antonello Calcutta, Giovanna Capolongo, Sol Carriazo, Michele Ceccarelli, Adrian Constantin Covic, Ananya De, Pilar Delgado, Nicole Endlich, Matthias Endres, Fabrizio Esposito, Michele Farisco, Quentin Faucher, Ana Carina Ferreira, Andreja Figurek, Denis Fouque, Casper Franssen, Ivo Fridolin, Sebastian Frische, Liliana Garneata, Loreto Gesualdo, Konstantinos Giannakou, Olivier Godefroy, Aleksandra Golenia, Dimitrios Goumenos, Eugenio Gutiérrez Jiménez, Gaye Hafez, Ewout Hoorn, Pedro Henrique Imenez Silva, Raafiah Izhar, Dearbhla Kelly, Shelli Kesler, Aleksandra Klimkowicz-Mrowiec, Samuel Knauss, Justina Kurganaite, Hélène Levassort, Sophie Liabeuf, Jolanta Malyszko, Laila-Yasmin Mani, Gianvito Martino, Ziad Massy, Christopher Mayer, Armida Mucci, Alma Mutevelic-Turkovic, Rikke Nielsen, Dorothea Nitsch, Alberto Ortiz, Vasileios Panagiotopoulos, Despoina Karasavvidou, Giuseppe Paolisso, Bojana Pejušković, Marion Pepin, Alessandra Perna, Andrea Perrottelli, Vesna Pešić, Pasquale Pezzella, Merita Rroji (Molla), Ivan Rychlík, Giorgos Sakkas, Mariadelina Simeoni, Maria José Soler Romeo, Goce Spasovski, Ana Starčević, Gioacchino Tedeschi, Francesco Trevisani, Robert Unwin, Evgueniy Vazelov, Carsten Alexander Wagner, Franca Wagner, Christoph Wanner, Andrzej Wiecek, Hong Xu, Miriam Zacchia, Lefteris Zacharia, Irene Zecchino, Carmine Zoccali, Francesco Mattace-Raso, Karl-Hans Endlich, Norberto Perico, Giuseppe Remuzzi, Francesco Trepiccione, Mark Okusa, Vincenzo Di Marzo, Peter Blankestijn, Kai-Uwe Eckardt, Maximilian Konig, Ron Gansevoort, Hassan Askari, Brian Hansen, Sunna Snaedal, Elena Cuiban, Edoardo Caporusso, Vincenzina Lo Re, Jonathan Roiser, Kerry Rosenberg, Alvino Bisecco, Laura Denby, Onkar Prakash Kulkarni, Kumar Sharma, Subrata Debnath, Afaf Jaafar, Anna Capasso, Michele Mulholland, Biruh Workeneh, Anna Iervolino, Simon Fraser, Isabelle Frey-Wagner, Annachiara Pastore, Romaldas Mačiulaitis, Antonio De Donato, and Ana Farinha.
REFERENCES
Author notes
All authors contributed equally.
Comments