Função Renal e Disfunção Cognitiva : estudo transversal de utentes inscritos na Unidade de Saúde Familiar-Ponte Renal Function and Cognitive Dysfunction : cross-sectional study of users enrolled at Ponte-Family Health Unit

Introdução: Estudos recentes demonstram o aumento da prevalência de Disfunção Cognitiva em pacientes com Doença Renal Crônica. Objetivo: Avaliar a referida associação nos utentes inscritos na Unidade de Saúde Familiar-Ponte. Métodos: Estudamos uma amostra constituída por 246 idosos. Avaliamos a função cognitiva por meio do Mini Mental State Examination e a Taxa de Filtração Glomerular com recurso à equação Modification of Diet in Renal Disease. Os valores da Taxa de Filtração Glomerular obtidos (ml/min/1,73 m2) foram distribuídos por três categorias: < 60,00, 60-89,99 e ≥ 90. Recolhemos variáveis adicionais do Serviço de Apoio ao Médico e estudamos os dados recorrendo a análises bivariadas e a modelos de regressão logística. Resultados: Os grupos com Taxa de Filtração Glomerular < 60 e ≥ 90 apresentaram maior prevalência de Disfunção Cognitiva, independentemente de outros fatores. Os odds-ratio foram, respectivamente, de 4,534 (IC95%: 1,257-16,356) e 3,302 (IC95%: 1,4347,607). Discussão: Conforme a literatura, verificamos maior prevalência de Disfunção Cognitiva no grupo com Taxa de Filtração Glomerular < 60. A elevada prevalência de Disfunção Cognitiva nos utentes com Taxa de Filtração Glomerular ≥ 90 está descrita em alguns estudos e poderá dever-se a situações que induzam a sobrestimação da mesma taxa, como nos estados de caquexia, ou a situações de hiperfiltração glomerular. Conclusão: Constatamos que a relação entre a função renal e a prevalência de Disfunção Cognitiva não foi linear, mas sim parabólica. Novos estudos são necessários para se explicar o porquê deste achado e para se averiguar a necessidade de vigilância da Disfunção Cognitiva em pacientes com alterações da função renal. Resumo


IntroductIon
The incidence of chronic kidney disease (CKD) has increased steadily, particularly among the elderly.3][4] In 2002, Portugal was the country in Europe with the highest rate of patients on renal replacement therapy (1,097 per million inhabitants). 4][7][8][9][10][11][12][13][14] Although until recently neglected, 5 this association has been observed both in patients with end-stage renal disease (ESRD) and individuals with early-stage kidney disease, and has been shown to exist independently from cardiovascular risk factors.Elias et al. 6 and Etgen et al. 7 reported increased prevalence of cognitive impairment in patients with early-stage CKD, estimated by the glomerular filtration rate (GFR).Kurella et al. 8 found the same association and added that for every decrease of 10 ml/min/1.73m 2 in GFR there was an increase of 11% in the prevalence of cognitive impairment DOI: 10.5935/0101-2800.20150012(CI).However, patients with a GFR ≥ 100 ml/ min/1.73m 2 also had a significantly increased risk of developing CI. 9 Buchman et al. 10 reported significantly steeper cognitive decline over the years in patients with a GFR under 60 ml/ min/1.73m 2 .][13] Other kidney disease indicators have been associated with CI: Jassal et al. 14 showed that albuminuria may predict cognitive decline.This prospective study enrolled 1,345 people aged 30 years and older and revealed that the presence of microalbuminuria was associated with a significant reduction of cognitive function after ten years of follow-up.
Small vessel disease (SVD) may explain the relationship between renal disease and CI.According to Mogi & Horiuchi, 15 lesions in the renal and cerebral microvasculature have a common pathogenic basis, since these vessels share anatomical and vasoregulatory characteristics.The strain-vessel hypothesis 16 cited by the authors compares the juxtaglomerular renal afferent arterioles to the perforating arteries of the brain: hemorrhage and cerebral infarction occur more often in the area of the perforating arteries, as they are exposed to higher pressures.According to the aforementioned theory, the juxtaglomerular afferent arterioles face similar circumstances, since vascular damage occurs earlier and more severely in these vessels.
Mogi & Horiuchi 15 listed a string of evidence to support the finding that patients with CKD are at increased risk of having stroke and are more prevalently affected by white matter lesions, silent stroke, and microhemorrhage.These subclinical lesions have also been related to risk of dementia and cognitive impairment.Thompson et al. 17 argued that broad evidence shows that cerebral SVD is the most prevalent neurological disorder, with incidence being possibly six to ten times greater than that of symptomatic stroke.According to the authors, cerebral SVD has been recognized as the main cause of cognitive impairment, alone or combined with Alzheimer's disease.However, Knopman 18 stated that general consensus suggests that cerebrovascular disease plays a role in dementia, although the importance of such role is yet to be determined.
From the point view of pathophysiology, the Steno hypothesis 19 assumes that the endothelial damage caused to the microvasculature by systemic diseases lies in the root of the establishment of a chronic state of inflammation.Knopman 18 argued that, in the same manner as in nephrosclerosis, endothelial alterations lead to the release of serum proteins in urine, in a process similar to what occurs in the brain when proteins leak into the cerebral extracellular space.
This relationship between the brain and the kidneys allows a glimpse into possible intervention perspectives.One organ's lesions may indicate the occurrence of silent lesions in other organs whose function is more difficult to assess.Mogi & Horiuchi 15 postulated that even patients with CKD may now be followed up with the aid of magnetic resonance imaging (MRI).Knopman 18 cited albuminuria as a useful indicator to screen patients for systemic endothelial dysfunction.Elias et al. 6 stressed the need for specific cognitive tests targeted to patients with moderate decreases in GFR.
Despite the relevance and topicality of the matter at hand, we were unable to find a similar study in Portugal on search engines such as Google, Pubmed or the Index of Portuguese Medical Journals.
This project aimed to look into the existence of a relationship between renal function (estimated by the GFR) and prevalence of cognitive impairment [calculated using the Mini Mental State Examination (MMSE)] in a population of elderly individuals registered with the Ponte Family Health Unit (FHU) and assess whether it occurs independently from cardiovascular risk factors.

Study deSign and pOpulatiOn
This study was carried out at the Ponte FHU, Ave Health Care Center Group II -Guimarães/ Vizela, from June 1, 2011 to December 31, 2011.
The population consisted of individuals aged 65 years and older registered with the Ponte FHU on January 1, 2011 -the start date of another project carried out at the Ponte FHU titled 'The impact of thyroid disorders on the cognition and mood of elderly individuals: a cross-sectional study of health care system users registered with the Ponte FHU'.
The size of the sample was defined randomly, and 263 elderly individuals were picked from a total population of 838 subjects.
All individuals registered with the Ponte FHU aged 65 and older were included in the study.Exclusion criteria: not having a listed phone number; failing to answer three phone calls made on different days; inability to travel to the FHU; visual or hearing impairment; no medical records on the Medical Support Service (MSS); medical records not listing serum creatinine levels or listing serum creatinine levels measured prior to January 1, 2010 (one year prior to the application of the MMSE).Data collection took place for the period of one year from the start date of the study in order to maintain a good level of temporal correspondence between data points.A shorter period was not chosen so as not to compromise the volume of collected data.

prOcedureS
Renal function was estimated based on the GFR, using the Modification of Diet in Renal Disease (MDRD) study equation, a formula with greater predictive power than the Cockcroft-Gault in estimating the GFR. 20ognitive function was considered first as a quantitative variable, in the form of the MMSE total score. 21The MMSE is a widely used neuropsychological test applied to assess various cognitive domains (orientation to time, orientation to place, registration, attention and calculation, recall, language and visual construction).It is recommended as a tool to assess global cognitive function and screen patients for further, more specific examination. 22,23xt, cognitive function was converted into a dichotomous variable -'with CI or without CI' -given the need to normalize MMSE scores for level of education.The cutoff points used in this study were established by Guerreiro et al.  for the Portuguese population. 24Subjects were considered to have CI when their MMSE scores were equal to or lower than 15, 22, and 27 for zero, one to 11, or more than 11 years of schooling, respectively.
Other variables were collected in order to verify whether the relationship between renal and cognitive function -if present -was independent from cardiovascular risk factors and potential confounders.][10][12][13][14] Depression symptoms may confound and interfere with the interpretation of the MMSE, 25 and were, therefore, included in the study.The Hospital Anxiety and Depression Scale (HADS) validated for the Portuguese population was used in this analysis.Only domain B of the scale (HADS-B), covering depression, was used in our study.Presence or absence of thyroid disorders, given its association with dementia, 26 mood disorders, 27,28 and cardiovascular disease (CVD), 29,30 was also considered.
Ethnicity was not considered, as all enrolled subjects were Caucasian.
The working definitions, types, and scale of the variables are described in Annex I.

StatiStical analySiS
First, the data sets were entered onto electronic forms (Statistical Package for Social Sciences version 19) and the population was characterized (data not shown).Subsequently, bivariate analysis was used to compare the variables for GRF and CI.The chi-square and the Kruskal-Wallis tests were used to this end.Lastly, the relationship between GFR and CI was analyzed using univariate and multivariate logistic regression.
Statistical significance was attributed when pvalues were equal to or lower than 0.05 (95% confidence interval).

ethical cOnSideratiOnS
This study was approved by the Health Ethics Committee of the North Regional Health Administration (ARS-N) (report No. 72/2011).

results
Ten of the 263 individuals initially enrolled were excluded for not having serum creatinine levels in their records; six for not having medical records; and one for having serum creatinine levels taken in 2008.The studied population thus included 246 individuals.
GFR levels were grouped based on the classification for CKD published by the NKF. 31 However, only one individual had stage 5 and none had stage 4 disease.Therefore, subjects were divided into three groups based on GFR, considering, however, its clinical significance.The first included patients with renal failure, i.e., GFR < 60.00 ml/min/1.73m 2 (GFR < 60), thus encompassing individuals with CKD stages 3, 4, and 5.The second included patients with a GFR ranging between 60.00 and 89.99 ml/ min/1.73m 2 (GFR60-89), covering individuals with stage 2 CKD described by the NKF as having mild GFR decreases.The third included patients with a GFR ≥ 90,00 ml/min/1.73m 2 (GFR ≥ 90), characterized for having stage 1 CKD and normal or increased GFR. 31 Twentythree subjects (9.3%) were in the GFR < 60 group, 103 (41.9%) in group with a GFR in the 60-89 range, and 120 (48.8%) in the group with a GFR ≥ 90.
The conversion of cognitive function into a dichotomous variable -with CI or without CIstandardized for level of education revealed that 48 patients (19.5%) had CI.
Table 1 shows that patients in the group with a GFR < 60 were mostly women; they were also older than the individuals with a GFR ≥ 90, had a higher prevalence of diabetes than the subjects in the GFR 60-89 group, and a higher prevalence of anemia than the other two groups.The individuals with a GFR < 60 and the subjects with a GFR ≥ 90 had a significantly higher prevalence of CI than the patients in the GFR 60-89 group.
The groups did not have a significantly different level of education, although the p-value (0.056) was quite close to reaching statistical significance; less educated individuals were primarily located in the group with a GFR < 60, whereas more educated subjects were in the group with a GFR ≥ 90.
The occurrence of the conditions included in the 'history of CVD' variable (Annex I) showed no significant differences between renal function groups.
When presence or absence of CI was considered, Table 2 shows that the only statistically significant differences were in age (p = 0.010) and HADS-B (p = 0.011).Patients with CI were older and had a higher prevalence of depressed mood states.
Tables 3 and 4 show the results of the logistic regression models.Bivariate analysis revealed a higher prevalence of CI when the GFR was < 60 and ≥ 90; thus, the GFR 60-89 group was treated as the category of reference.
Table 3 shows that the first and third groups had a significantly higher probability of having CI when compared to the GFR 60-89 group.The group with a GFR < 60 had an odds ratio of 4.461 (95% CI: 1.543 to 12.897) relative to the patients with a GFR in the 60-89 range.The subjects with a GFR ≥ 90 had an odds ratio of 2.665 (95% CI: 1.256 to 5.654) when compared to the reference group.
Multivariate analysis considered the variables that yielded significant differences in the group comparisons, including gender, age, anemia, diabetes mellitus, and HADS-B.Hypertension was also considered, as it is a major cause of cardiovascular disease, along with education, an important socio-demographic factor that was very close from reaching statistical significance in bivariate analysis.As seen in Table 4, the p-values for the GFR < 60 and GFR ≥ 90 groups remained significant regardless of the introduction of other variables.Age and HADS-B scores were significantly affected (p < 0.001 and p = 0.05, respectively) by the presence of CI.

dIscussIon
Group analysis and univariate/multivariate logistic regression showed that the individuals with a GFR < 60 and the subjects with a GFR ≥ 90 had a higher prevalence of CI.The variability of CI explained by these GFR categories remained significant regardless of gender, age, education, diabetes mellitus, hypertension, anemia, or HADS-B score.The probability of a patient underperforming in the MMSE quadrupled in the group with a GFR < 60 and tripled in the group with a GFR ≥ 90 when compared to the individuals with a GFR in the 60-89 range.The Hosmer-Lemeshow test revealed that the fit of the regression models was good (p > 0.05); however, they explained only 6.6% of the CI variability in the analysis including only the GFR, and 20.3% in multifactorial analysis (Nagelkerke R Square).Additionally, sensitivity was null in the first case, although it increased to 11.1% in the latter.Therefore, significant differences were observed in the prevalence of CI according to the GFR, although the model was not ideal.
Age was found to be the variable that best associates with presence of CI (p < 0.001), while the HADS-B score also yielded statistical significance in the regression model (p = 0.050), suggesting an association between depressed mood states and lower scores on the MMSE.][8][9] In our study, the group with a GFR < 60 had more females, a higher mean age, and more individuals with diabetes and anemia.][10][12][13][14] Nevertheless, once diabetes and anemia were treated as dichotomous variables, these conditions may have been more severe in patients with a GRF < 60. Surprisingly, the prevalence of CI was significantly higher in the group with a GFR ≥ 90 than in the group with a GFR ranging between 60 and 89.This finding does not agree with most of the literature reports cited above.However, Kurella et al. 9 described increased risk of CI when the GFR was greater than 100 ml/min/1.73m 2 , a finding maintained when adjusted for various risk factors.
Inrig et al. 32 studied 8,941 individuals with the purpose of assessing the clinical impact of increases in the GFR.The study revealed that every decrease of 10 ml/min/1.73m 2 in the GFR below 100 was associated with a 13% increase in the risk of cardiovascular events, and that every increase of 10 ml/min/1.73m 2 in the GFR above 125 was associated with a nine percent ¶ ρ ≤ 0.05 CI: cognitive impairment (adjusted for years of schooling); CI: Confidence interval; BMI: body mass index; BP: Blood pressure; Hx: History; CVD: Cardiovascular disease; HADS-B: Hospital Anxiety and Depression Scale domain B; § Continuous variables, represented by mean values and standard deviation between brackets.increase in cardiovascular risk.The authors concluded that the relationship between the GFR and cardiovascular events might be parabolic, with patients with lower and higher GFR being at increased risk.Similarly, Mostofsky et al. 33 looked into mortality rates according to renal function in 1,175 inpatients with ischemic stroke and found a U-shaped curve to describe the relationship between the GFR and death.The two studies relate to our study: they described a U-shaped correlation between renal function and cardiovascular events, while our study showed a U-shaped association between renal function and CI, for which cardiovascular events must be an important cause.Inrig et al. 32 and Mostofsky et al. 33 formulated two hypotheses to explain their findings.Firstly, they challenged the validity of the Cockcroft-Gault and MDRD equations used: evidence shows a tendency for GFR overestimation with the first (especially in obese patients) and GRF underestimation with the second.The authors further argued that the cardiovascular events recorded in the group with elevated GFR might be due to cachexia characterized by sharp decreases in muscle mass, which translates into lower serum creatinine and thus falsely elevated GFR.
Another hypothesis is the theory of glomerular hyperfiltration, first postulated by Brenner. 34Brenner and other authors have shown that rats submitted to partial nephrectomy developed hemodynamic changes in their remaining glomeruli, along with increased GFR and elevated transcapillary hydraulic pressure.Subsequently, the rats had proteinuria, glomerulosclerosis and renal failure.Glomerular hyperfiltration was then supposed to be a (mal-) adaptive state resulting from kidney injury and leading to progressive decline in renal function. 346][37] Recent studies have shed new light on the role of glomerular hyperfiltration in obese patients. 38,39t has been argued that patients with a GFR > 90 have abnormally elevated renal function for their ages.According to the NKF, GFR values are usually lower than 90 ml/min/1.73m 2 in elderly individuals.In adults, the GFR peaks at between 20 and 30 years of age, within the range of 118-127 ml/min/1.73m 2 , to then decrease at a rate of one ml/min/1.73m 2 per year.The idea of 'normal' GFR, therefore, still remains controversial. 31The GFR level commonly seen in 70-year-old individuals [approximately the mean age of the subjects enrolled in our study (72.23)] sits around 70 ml/min/1.73m 2 ; values above this threshold are considered high, and thus abnormal.Inrig et al. 32 indicated that the prevalence of hyperfiltration increased from 7.4% to 16.6% when adjusted for age.
Thus, GFR increases (adjusted for age) may also serve as an indicator of vascular injury in other organs.Schmieder et al. 40 found that patients with increased GFR had a higher prevalence of myocardial hypertrophy, suggesting that glomerular hyperfiltration might be an indicator of target organ damage.As supported by Inrig et al. 32 and Mostofsky et al., 33 neither the explanation provided by glomerular hyperfiltration nor the possibility of having a falsely elevated GFR deny that patients with an increased GFR are at greater risk.
The main limitation of this study resided in the fact that the data collected from the MSS may contain errors.More specifically, the history of CVD of some patients may not have been captured in the information system.Additionally, the time interval of one year used to gather the data is significant and may have affected the quality of the association observed between the GRF and CI.The MMSE was an easy-to-use and quick-to-apply method to assess global cognitive function, but it may not have captured cases of mild cognitive impairment effectively.

conclusIon
This study revealed that patients with a GFR < 60 and individuals with a GFR ≥ 90 were more likely to present CI than subjects with a GFR ranging between 60 and 89 ml/min/1.73m 2 .These findings remained statistically significant, independently from other factors.GFR and CI presented a parabolic, nonlinear relationship, as reported by other authors.
The finding that the first group (GFR < 60) had a higher prevalence of CI was consistent with similar studies.The higher prevalence of CI in patients with a GFR ≥ 90 might be due to states of hyperfiltration or GFR overestimation, as seen in cases of individuals with cachexia.
Populations at risk in general, and patients with chronic kidney disease under conservative care in particular, should undergo early screening for cognitive impairment.Patients with diagnosed disease should be referred to neuropsychological cognitive rehabilitation.The study also indicated the need to establish a clearer definition for normal GFR and improve the resources used to interpret outpatient GFR estimates.contInued Annex 1.

tAble 1
VariableS per gFr grOup Hospital Anxiety and Depression Scale domain B; § Metric variables, represented by mean values and standard deviation between brackets.ǂ MMSE final score (not adjusted); † dichotomous variable adjusted for years of schooling.

tAble 2
VariableS FOr ci grOupS