Cross-sectional assessment of mild cognitive impairment in pre-dialysis chronic kidney disease and its association with inflammation and changes seen on MRI: what the eyes cannot see

Abstract Introduction: Mild cognitive impairment (MCI) is a prevalent and underdiagnosed condition in chronic kidney disease (CKD), that shares common pathophysiological factors such as chronic inflammation. Objective: To evaluate the association of MCI in CKD stages 1-5 using inflammatory markers and changes by magnetic resonance imaging (MRI). Patients and Methods: Cross-sectional study in adult patients with pre-dialysis CKD. MCI was assessed by the Montreal Cognitive Assessment (MoCA) and the estimated glomerular filtration rate (eGFR) by the Chronic Kidney Disease Epidemiology Collaboration equation. Sociodemographic and clinical data were collected from medical records. The cytokines IL-4, IL-6, IL-17, TNF-α and hs-CRP were determined. Brain MRI was performed in a 1.5 Tesla device, without paramagnetic contrast. A descriptive analysis followed by a comparison of abnormal versus normal MoCA scores among all studied variables. A linear regression analysis was performed using MoCA as a dependent variable, adjusted for confounding factors. Results: Of 111 invited patients, eighty completed the neuropsychological assessment and 56 underwent MRI, and were included in the study. Mean age was 56.3 ± 8.3 years and 51.8% (n = 29) had altered MoCA. When compared to the group with normal MoCA, the group with altered MoCA had higher levels of IL-6 and IL-17. There was no correlation between altered MoCA with eGFR or with MRI abnormalities. Conclusão: MCI assessed by MoCA was prevalent in patients with pre-dialysis CKD, it was associated with inflammation and showed no correlation with MRI changes.


IntroductIon
One of humanity's achievements was the improvement in the diagnosis and treatment of maternal, child and infectious diseases, significantly reducing morbidity and mortality rates. This phenomenon led to the predominance of non-communicable chronic diseases (NCDs), highlighting systemic arterial hypertension (SAH) and diabetes mellitus (DM), a favorable scenario for the increased prevalence of cardiovascular diseases (CVD), chronic kidney disease (CKD) and dementia syndromes 1 .
The prevalence of CKD has increased significantly, being estimated at around 11% of the world population 2 . A similar phenomenon occurs with cognitive impairment (CI) and dementias, particularly in low and middle-income countries, in which CI is associated with population aging, low socioeconomic and educational levels and a higher frequency of potentially risk factors modifiable, such as SAH, sedentary lifestyle, alcohol abuse, etc 3,4 .
In Latin America, dementias are a growing problem, as a result of the high levels of illiteracy and the demographic transition observed in that continent 5 . Among the dementias, Alzheimer's disease accounts for about two thirds of the cases, followed by vascular dementia, mixed dementia, Lewy body dementia and frontotemporal degeneration 6,7 . As a general rule, we use specific protocols to assess CI. In these, the most cited cognitive screening instruments are the Mini Mental State Examination (MMSE) 8 and the Montreal Cognitive Assessment (MoCA) 9 , associated with the assessment of functional activities, usually performed using the functional activities questionnaire by Pfeffer et al. 10 , in addition to laboratory tests for syphilis (VDRL), vitamin B12 dosing, thyroid hormones and human immunodeficiency virus (HIV) serology. These are associated with running neuroimaging exams, represented by computerized tomography (CT) and/ or magnetic resonance (MR) of the brain, aiming to exclude reversible causes of CI 11 .
In recent years, the association of CKD categories 4 and 5 with dementia has been reported by different authors [12][13][14] . Preceding the diagnosis of dementia, initially mild and progressive CI occurs, often undiagnosed. Mild cognitive impairment (MCI), described by Petersen et al. 15 , is defined as cognitive impairment without functional repercussions, being an intermediate condition between normality and dementia. Since then, there have been several reports of CI at different stages of CKD, affecting up to two-thirds of this population.
Etiologically, MCI is attributed to factors such as uremic toxins, inflammation and oxidative stress, among others 16,17 . These factors determine endothelial dysfunction and, later, vascular lesions identifiable by brain MRI and characterized by white matter hyperintensities, subclinical cerebral infarcts or cortical atrophies 18,19 . Therefore, in initial cases of CI in CKD, these changes, of a functional nature, cannot be identified by MRI, making early diagnosis difficult.
We know that CI impacts the treatment of CKD, as it interferes with the understanding of the guidelines from healthcare professionals, reducing adherence to dietary guidelines and pharmacological treatment in this population. Thus, it is essential to diagnose MCI as early as possible 17,19,20 . The hypothesis is that, in CKD, cognitive alterations are initially of a functional nature and, therefore, not identifiable by MRI. Given the above, this study focuses on evaluating the MCI detected by MoCA, inflammatory markers and changes in brain MRI, in patients with CKD under conservative treatment.

PatIents and Methods
This study was approved by the Research Ethics Committee (CEP) of the Federal University of Juiz de Fora, under protocol nº. 183,387. This is a cross-sectional study of a database created for cognitive analysis of patients with pre-dialysis CKD, called NEFROCOG. For this base, we invited the patients scheduled for clinical consultations at the Hiperdia Minas Center in Juiz de Fora-MG (CHM-JF) from March 2013 to December 2014. The patients were contacted by telephone and scheduled for individual neuropsychological evaluation, on a day different from that of their clinical appointment.
We used a convenience sample, consisting of 111 patients seen by the CHM-JF of secondary care in SAH, DM and CKD. Of these, 14 were participated in a pilot project -carried out to calibrate the tests. The following were defined as inclusion criteria: patients with CKD in categories 1 to 5, undergoing conservative treatment (non-dialysis); age > 21 and < 65 years; they all agreed to participate in the study and signed the Informed Consent Form (ICF). We excluded those patients with a previous history of cerebrovascular accident (CVA); infectious or degenerative diseases of the Central Nervous System (CNS); presence of delirium and/or psychotic disorders; previous history of traumatic brain injury; visual and auditory disorders that prevented the tests from being carried out; HIV and/or Acquired Immunodeficiency Syndrome (AIDS); and contraindication or intolerance to performing MRI.
Inflammatory markers represented by interleukins (IL) IL-4, IL-6 and IL-17 and by tumor necrosis factor alpha (TNF-α) were also evaluated. For this purpose, 10 mL blood samples were collected in tubes without anticoagulant. The tubes were centrifuged at 3,500 rpm and the collected sera were frozen at -86°C. The measurements of these markers were performed using the Enzyme Linked Immune Sorbent Assay (ELISA) technique. The samples were quantified by comparison with recombinant standard curves (antibody and recombinant concentrations as per manufacturer's recommendations, PeProtech Inc, New Jersey).

evaluatiOn and study instruments
Two medical professionals (one neurologist and on psychologist) performed the assessments, conducted in a silent environment, in a single session of approximately one hour.
We ran a structured anamnesis and applied clinical depression questionnaires for the tests were applied in the same sequence for all patients and corrected only by the neurologist/psychologist examiner. The sequence of the neuropsychological tests' application considered both the mental wear of the individual when performing each test and the interference of one test with another, which could generate, for example, learning. We must also consider that some tests are distracting.
The Mini International Neuropsychiatric Interview/ Mini International Neuropsychiatric Interview-Plus (M.I.N.I/M.I.N.I-PLUS) 21 tests were used to diagnose depression and MoCA, in the Brazilian version 22 , to diagnose CI. To interpret the MoCA, the cutoff point adopted for the Brazilian population (≤ 25 points) was reduced to ≤ 24 points, as a strategy to adjust the sample's educational level. The choice of MoCA was due to the good sensitivity of the test to detect MCI when compared to the MMSE 17 .
The Pfeffer et al. 10 questionnaire was also used to assess the patient's functionality. This test was applied to the companion simultaneously, with the patient's assessment or over the telephone.
The questionnaires and tests, the functions and domains assessed, as well as the cutoff point used for their correction, are shown in Chart 1. The outcome assessed was CI assessed by the MoCA.

magnetic resOnance evaluatiOn Of the brain
We performed MRI images of the brain using a highfield Siemens/Avanto apparatus (1.5 Tesla), without paramagnetic contrast. For this purpose, we used a specific protocol recommended for the study of cognitive alterations and dementia 23 . We used three visual scales to assess the MRI images -Fazekas 24 , MTA (Mesial Temporal Atrophy) 25 and the GCA (Global Cortical Atrophy) 26 (Chart 2). The image analyses were performed by neuroradiologists blinded in the study.

statistical analysis
We ran a descriptive analysis of the data with the mean, standard deviation, median or percentage according to the characteristic of the variable. We assessed normality using the Kolmogorov-Smirnoff test. A comparison was made between patients with normal MoCA vs. altered sociodemographic, clinical, laboratory, inflammatory and imaging variables, using the Chi-square and/or Student's t tests for independent data, and the Mann-Whitney test to compare medians. The correlations between MoCA and eGFR with all variables were performed using the Pearson or Spearman correlation tests. Finally, we performed a linear regression using the MoCA as the dependent variable; and age, gender and eGFR as confounding variables.
For each model, we used an inflammatory marker as an associated confounding variable, as there was a lot of collinearity between them, preventing them from being placed in a single model. We used the SPSS 17.0 software, Chicago, Illinois®.

results
We invited 111 patients to participate in the study; 13 of whom refused, and four were scheduled but did not attend the evaluations. The remaining 94 signed the Consent Form and underwent the evaluations. Of these, 14 made up the pilot study, leaving eighty complete evaluations. After applying the inclusion criteria, eight patients were excluded and 16 refused to undergo brain MRI, making a final sample of 56 patients ( Figure 1  To present the results, the patients were divided into two groups: normal MoCA (≥ 24 points) and altered MoCA (< 24 points).
Regarding the sociodemographic parameters, there was a tendency to change the MoCA in older patients (58.2 ± 6.9 vs. 54.2 ± 9.2 years; p = 0.06), as well as in patients with schooling < 4 years, when compared to those > 4 years (p = 0.04). Regarding comorbidities, patients with DM tended to have an altered MoCA (65.5% vs. 38.5%, p = 0.06). In other comorbidities, as well as in the different CKD categories, there were no significant changes ( Table 1).
In the laboratory evaluation, plasma sodium (138.0 vs. 139.6 mEq/L; p = 0.06) and ferritin (105.0 vs. 157.7; p = 0.06) were slightly higher in the group of patients with altered MoCA. However, these data were not statistically significant, or clinically relevant. The same was true in relation to triglycerides, which were lower in the group with altered MoCA (148.5 vs. 222.9 mg/dL; p = 0.06).
Importantly, there was no statistically significant difference in the other exams, including eGFR (Table 2 and Figure 2). Furthermore, there were no associations between eGFR and the inflammatory variables ( Figure 3).
Neuroimaging was performed by MRI, the gold standard for evaluating the white matter lesions by detecting microangiopathies or leukoaraiosis, MCI markers of a vascular nature. These lesions, evaluated by the Fazekas scale, 24 had a score of 0 in 17.1% of the cases; and a score of 1 in 60.7%, that is, mild or no changes in the white matter, not explaining a CI of a vascular nature. Moderate microangiopathy or Fazekas 2 were found in only 19.6% of the cases. However, as they are mostly hypertensive patients, SAH per se would justify the presence of these alterations.
There were no changes (MTA 0) in the MTA 25 scale, used to assess mesial changes in the temporal lobes, correlated with Alzheimer's disease, in 81.8% of the cases, and 16.4% had mild changes. Thus, similarly to the previous scale, there were no alterations related to Alzheimer's disease detected using the MTA scale.   The same behavior was observed in the third scale used, the ACG 26 . In 98.2% of the cases, cortical atrophy ranged between 0 (74.5%) and 1 (23.6%), that is, there was no atrophy in most cases and, when present, it was mild. Only 1.8% had severe cortical atrophy. Therefore, there were no statistically significant image changes in the normal MoCA and altered MoCA groups (Table 3). Table 4 and Figure 4, show important differences between the levels of IL-6, IL-17 and TNF-α between patients with altered MoCA, compared to normal.
In the linear regression analysis models (Table 5) using the MoCA as the dependent and continuous numerical variable, all models were adjusted for age, sex and eGFR. The variables IL-4, IL-6, IL-17 and TNF-α were not placed in the same model because of their high collinearity; therefore, each model was adjusted for each of these variables separately.      The model was not adjusted for the DM variable, because this is a population that has a different behavior.
In model 1, adjusted for IL-17, males and IL-17 were independently associated with MoCA, with males being positively associated, and IL-17 being negative. In model 2, adjusted for IL-4, there was no association with MoCA. Model 3, adjusted for IL-6, showed that only this inflammatory marker was an independent predictor of altered MoCA. The last model, adjusted for TNF-α, did not show any association.

dIscussIon
The hypothesis that cognitive alterations in CKD are initially of a functional nature and, therefore, not identifiable by MRI was observed in our study, also the CI assessed by MoCA was prevalent in patients with pre-dialysis CKD and associated with inflammatory markers, the same not occurring in relation to MRI findings.
The sample consisted of patients with a lower mean age (56.3 ± 8.3 years) than that observed in the usual population of the pre-dialysis CKD outpatient clinic (66.2 ± 13.4 years) of the CHM-JF 27 In this study, the population presented, as a non-inclusion criterion, age > 65 years, considering that the increased incidence and prevalence of vascular and degenerative diseases occur from this age group onwards. This approach substantially reduced the age bias in the study sample. On the other hand, we know the action of diseases reported to cause vascular aging and early atherosclerosis in these individuals 18,28 .
Regarding education, the MoCA was lower in those with < 4 years of education when compared to those > 4 years. The sample compared to the National Household Sample Survey (PNAD), 2019, revealed a higher percentage of individuals with more than 4 years of education (89.1% vs 32.2%) 29 . There is evidence in the literature that correlates MoCA with education, being significantly higher in individuals with high education, thus resulting in the cutoff point of 26 used in the original study, contrasting with other studies validated in other countries that used various cut-off points. The higher the educational level, the greater the cognitive reserve and the better the performance on tests, thus acting as a protective factor for dementia 30 .
In this study, we adopted the cutoff point of 24, as it has already been suggested in previous studies in the Brazilian population 22,31 and in patients with CKD 32 .
As for comorbidities related to CKD, there were significant changes found in patients with DM, a population known to be at higher vascular risk. A study by Freedman et al. 33 using secondary data evaluated the relationship between cognition and DM with MRI findings, and reported a worse performance on the MMSE associated with a decrease in gray matter, with a higher albumin/creatinine ratio and a reduction in eGFR. Our study found no association of proteinuria or eGFR with CC assessed by MoCA.
Due to the high prevalence of SAH in our sample, we could not establish its association with CI, although there are numerous citations that correlate SAH as one of the main traditional risk factors for vascular compromise and secondarily to CI and CKD itself 18,19,34 .
Regarding the use of medications, there were significant changes in patients using oral antidiabetics, which, as discussed above, cause worse cognitive performance. Another therapeutic class associated with worse cognitive performance was diuretics. This fact may be related to the greater use of this therapeutic class in patients in more advanced categories of CKD and the likely association of CKD with other comorbidities. A systematic review 34 addressing the use of antihypertensive drugs and cognition in the elderly, despite highlighting the heterogeneity of the studies, showed an association between the use of ARB and improvements in episodic memory, while diuretics and other antihypertensive drugs did not have any association 34 . The same occurred with the use of acetylsalicylic acid, associated with a higher prevalence of CVD and, for this reason, it is believed to be associated with worse MoCA performance.
As for laboratory findings, the relationship between CI and anemia in CKD is a known fact. The drop in ferritin is related to worse functionality, lower MoCA scores in verbal and working memory, and in attention 35 . In this study, higher ferritin values were associated with better MoCA performance, but it seems that this increase is not associated with inflammation, but with iron reserves. In our study, we can speculate that hypertriglyceridemia may be associated with CKD, a condition in which dyslipidemia is observed in up to 75.7% of cases, and hypertriglyceridemia in up to 50% 36 .
Previous studies 20,[37][38] have shown a correlation between impaired renal function and CI, but most of these studies were carried out in populations over 65 years of age, in whom CKD is associated with the aging process. In the literature, it has been reported that CI, including dementia, occurs in 16% to 38% of patients with end-stage CKD 38 . In the REGARDS study 37 , a cohort with 23,405 patients, with a mean of 64.9 years, they reported that, in cases of eGFR < 60 mL/minute, 10 mL reduction in eGFR increased by 11% the prevalence of cognitive decline.
Another prospective cohort 38 showed a correlation between worsening renal function and CI in a multibreed population, with eGFR less than 90 mL/ minute and follow-up for 2.8 years; and they concluded that CKD was linearly associated with cognitive decline in this population, adjusted for multiple risk factors, even in patients with slightly reduced renal function 38 . On the other hand, the CRIC 39 study in younger patients with mild to moderate CKD (57.7 ± 11.0 years), with a mean eGFR of 45.0 ± 16.9 mL and a mean of 6 years of follow-up, concluded that CI is not associated with additional risk of CKD progression. In our study, consisting of patients under 65 years of age and with a predominance of earlier CKD categories, there was an important prevalence of CI, but this was not linearly associated with eGFR.
A relevant finding was the association of inflammatory markers with MoCA, especially in linear regression, with significant associations of dosages in the increase of pro-inflammatory cytokines IL-6 and IL-17, as well as a statistical trend in TNF-α. The same was did not happen for the anti-inflammatory cytokine IL-4. These findings corroborate those in the literature that report the importance of inflammation in the context of CKD, and the likely involvement of the endothelium and, secondarily, of the bloodbrain barrier (BBB) 19,28 , initially leading to functional changes and, subsequently, to the worsening of CKD and structural vascular lesions. A systematic review 40 on the association between IL-6 and CKD concluded that IL-6 is directly related to the progression of CKD, which may be a possible marker of kidney damage to be used both in the diagnosis and in the follow-up of this population. Similarly, it can be speculated that, based on the data from the present study, IL-6 may also be a marker of CI in CKD.
Numerous publications 16,18,19,24,28,33 correlate CI in CKD with vascular lesions, especially those located in the white matter, secondary to compromised microcirculation (microangiopathic or small vessel vascular disease) resulting from this fact, the use of neuroimaging as a diagnostic resource, with emphasis on MRI, which, when compared to CT, has greater sensitivity in the early detection of these lesions, considered specific markers of CI of a vascular nature. Studies correlating white matter lesions or microangiopathies (leukoaraiosis) associated or not with microinfarcts, microbleeds and cortical atrophies with CI 16,18,19,28,33 are frequent.
Differently in the present study, there was no correlation between altered MoCA and MRI findings, as the images were within normal limits or with very subtle alterations, not justifying the CI detected in 51.8% of the patients evaluated. This discrepancy could be explained by the functional nature of these alterations that precede the anatomical ones reported in previous studies. On the other hand, when CI was compared to inflammatory markers, a direct relationship was found with pro-inflammatory cytokines, especially IL-6. These findings suggest the hypothesis of the functional nature of CI in the population evaluated.
Among the limitations of the present study is the fact that this is a cross-sectional study, carried out in a single center and with a small sample size. On the other hand, it is worth mentioning as positive points the performance of MRI in all patients, in addition to the dosage of inflammatory markers.
A fact of clinical relevance was the demonstration of the MoCA usefulness, a simple, easy-to-perform and low-cost tool that can be applied by non-specialized professionals 9,31 . The addition of this tool to the clinical routine of nephrologists may contribute to the early diagnosis of CI and enable the implementation of preventive and therapeutic measures in patients with chronic kidney disease.

conclusIon
The MCI assessed by MoCA was prevalent in patients with pre-dialysis CKD, was associated with inflammatory markers and did not show any correlation with changes seen in the MRI.

authors' contrIbutIon
LAP: Conception, design and interpretation of data; writing of the article, critical review and approval of the version to be published. ALMA: Application and evaluation of research instruments; interpretation of data; critical review and approval of the version to be published. MAP: Application and evaluation of research instruments; interpretation of data; critical review and approval of the version to be published. JOAC: Laboratory analysis and approval of the version to be published. DSD: Laboratory analysis and approval of the version to be published. NSF: Statistical analysis and data interpretation. DGAE: Data Interpretation; writing the article; critical review and approval of the version to be published. RBP: Conception, design and interpretation of data; writing of the article, critical review and approval of the version to be published. NMSF: Conception, design and interpretation of data; writing of the article, critical review and approval of the version to be published.