Subclinical Albuminuria is linked to Gray Matter Atrophy in Type 2 Diabetes Mellitus

Abstract

Objective

Microalbuminuria (MA), a marker of renal microvascular disease, is associated with brain atrophy and neurovascular changes in older adults with type 2 diabetes mellitus (DM). We evaluated the relationship between urine albumin-to-creatinine ratio (UACR) and regional brain volumes determine whether subclinical albuminuria may indicate early structural brain changes in type 2 DM.

Materials/Methods

We studied UACR and brain volumes in 85 type 2 DM patients (64.8 ± 8.3 years) and 40 age-matched controls using 3D magnetization prepared rapid acquisition with gradient echo (MP-RAGE) MRI (magnetic resonance imaging) at 3 Tesla. The relationship between UACR and brain volumes was analyzed using the least square models.

Results

In DM patients, UACR ≥ 5 mg/g, UACR ≥ 10 mg/g and clinically significant MA (UACR ≥ 17 mg/g [males] and 25 mg/g [females]) were associated with lower gray matter (GM) volume in the frontal lobe (r² _adj = 0.2-0.4, P= 0.01-0.05) and UACR ≥ 5mg/g was also related to global GM atrophy (r² _adj= 0.1, P= 0.04), independent of DM duration, glucose levels, HbA_1c and hypertension. For UACR ≥ 5mg/g, a lower global GM volume was related to worse executive function (P= 0.04) in the DM group. No associations were found for UACR (< 5 mg/g) and controls.

Conclusions

Subclinical albuminuria (UACR ≥ 5 mg/g) is associated with lower GM volume that has clinical impact on cognitive function in older diabetic patients, and these relationships are independent of DM control and hypertension. Therefore, UACR levels may serve as an additional marker of DM-related brain structural changes.

1. Introduction

Type 2 diabetes mellitus (DM) is associated with endothelial dysfunction affecting the kidney and the brain, but the link between albuminuria and brain tissue damage remains unclear. Endothelial sensitivity to and the time course of vascular changes may differ between the central and the peripheral circulation [1-6].

Albuminuria, a marker of chronic vascular kidney disease, is associated with accelerated cognitive decline in diabetic patients [2]. Cerebromicrovascular disease manifests as impaired vasoreactivity [7, 8], hypoperfusion and decreased metabolism that may lead to hypoxia and brain tissue loss [9-12]. Therefore, cognitive and motor regions such as the frontal and parietal lobes are particularly vulnerable to hyperglycemia because of their high metabolic rates [13, 15]. Older diabetic patients have lower brain volumes, larger ventricular volumes, [9] and faster progression of brain atrophy [16, 17].

The relationships among brain atrophy, hyperglycemia and DM duration have been widely studied [1, 5, 6, 18-20], but the link between albuminuria and structural brain changes has not been well characterized. Therefore, as a systemic vascular dysfunction indicator, albuminuria may be associated with type 2 DM complications in the brain that may be prodromal for cognitive impairment later in life.

We hypothesized that: 1) Albuminuria is associated with gray matter (GM) loss in older people with type 2 DM. Subclinical levels of urine albumin-to-creatinine ratio (UACR) (UACR ≥ 17 mg/g [males], 25 mg/g [females]) may indicate brain structural changes. 2) The relationship between UACR and brain structure is independent of DM control and duration. We aimed to determine the association between UACR and brain structural changes in type 2 DM and control patients.

2 Materials and Methods

2.1 Participants

This study was conducted at the Syncope and Falls in the Elderly Laboratory at the Beth Israel Deaconess Medical Center and at the Center for Advanced Magnetic Resonance Imaging, Department of Radiology.

We reviewed records obtained from January 2002 to July 2012, collected during three studies: 1) Cerebral vasoregulation in diabetes (January 2002-December 2005); 2) Cerebral perfusion and cognitive decline in type 2 diabetes (January 2006-December 2009); and 3) Cerebromicrovascular disease in elderly with diabetes (August 2009-July 2012) (see funding section for details).

Participants were recruited from the local community via advertisements. All subjects signed an informed consent approved by the Institutional Review Board. Inclusion criteria for the diabetes group (men and women diagnosed with type 2 DM > 1 year, normotensive or hypertensive, age > 50 years old) were similar among these studies. For the control group, however, studies 1 and 2 included only normotensive non-diabetic participants, while study 3 included both normotensive and hypertensive participants.

A total of 308 participants signed informed consent (120 non-diabetics, 188 diabetics). One hundred twenty two participants (52 non-diabetics, 70 diabetics) were excluded for one of the following reasons: type 1 DM, stroke or recent myocardial infarction, clinically relevant cardiac disease (arrhythmias, kidney or liver transplant, carotid artery stenosis), neurologic disorder, other systemic disorders, uncontrolled HTN or treatment with 3 or more antihypertensive agents, claustrophobia, metal implants, arterial stents and pacemakers not compatible with 3Tesla MRI, no transcranial Doppler window, withdrew consent, lost to follow-up or other reasons. For this study, we also excluded 61 records with incomplete datasets (missing UACR or GM volumetric measurements). Forty three patients that were excluded had HTN diagnosis. We analyzed data from 125 subjects.

2.2 Study Cohort

The DM group consisted of 85 men and women, aged 50-85 years, who were diagnosed with type 2 DM and treated with oral agents and/or combinations with insulin; they were normotensive (blood pressure [BP] < 140/90 mmHg and had no medical history of hypertension [HTN]) or hypertensive (n= 51) (BP ≥ 140/90 mmHg and/or treated for HTN). Subjects were treated with one or more of the following: insulin (6), oral glucose-control agents (sulfonylurea, second generation agents or a combination) (56), both (15) and/or diet (8). The control group consisted of 40 age-matched (± 5 years) men and women; seven of them treated for HTN. The control group had normal fasting blood glucose and HbA_1c levels. They also had similar cardiovascular risk factor distribution (HTN, hypercholesterolemia, obesity, history of cardiovascular disease) as the type 2 DM group.

2.3 Study Protocol

All participants were screened using medical history, autonomic symptoms and activity questionnaires, ECG, vital signs, height, weight and body mass index (BMI). Blood was drawn to measure fasting blood glucose, HbA_1c, renal panel, lipid profile and complete blood count. Neuropsychological assessments included executive function (Trail Making Tests A and B (TMT); verbal learning and memory (Hopkins Verbal Learning Test) tests, as well as the Instrumental Activities of Daily Living scale. Systolic, diastolic and mean BP measurements were obtained every 20 minutes using a wearable 24-hour home monitoring device (Dynapulse, Inc. Vista, CA).

2.4 Urine albumin creatinine ratio and microalbuminuria

UACR was measured from the first-void clean-catch urine samples collected on the day participants were scheduled for brain MRI. Urinary albumin was measured by immunoassay and urinary creatinine by the modified Jaffe method.

Microalbuminuria (MA) was defined as an UACR ≥ 17- 250 mg/g in men and 25-350 mg/g in women per the sex specific in spot urine sample [21]. These sex-specific cutoffs, when converted to albumin excretion rates, became almost equal at 30 mcg/min [1, 5, 18-20]. We adopted the UACR ≥5 mg/g and UACR of ≥10 mg/g as cut-off ranges for the analyses, based on epidemiological studies that showed their association with an increased risk for HTN and cardiovascular events [22, 24].

2.5 Magnetic Resonance Imaging

Studies were performed on a 3-Tesla GE GHX MRI scanner using a quadrature and eight-channel phase array head coils (GE Medical Systems, Milwaukee, WI). Anatomical images were acquired using 3-D magnetization prepared rapid acquisition with gradient echo (MP-RAGE) and fluid attenuated inversion recovery (FLAIR) sequences. Images were analyzed using tools developed in interactive data language (IDL, Research Systems, Boulder, Colorado, USA) and MATLAB (MathWorks, Natick, Massachusetts, USA). Anatomical magnetic resonance images (MP-RAGE and FLAIR) were co-registered non-linearly to the MNI152 standard template and segmented to calculate regional GM, white matter (WM) and cerebrospinal fluid (CSF) volumes in main anatomical lobes and their subregions (SPM, University College London, UK) [25]. MP-RAGE and FLAIR images were co-registered to a standard template and segmented to calculate regional GM, WM, CSF and white matter hyperintensities (WMHs) volume in the frontal, temporal, parietal and occipital regions using statistical parametric mapping software package (SPM, University College London, UK). WMHs were identified by thresholding of hyperintense pixels on FLAIR images with > 30% increase in signal intensity compared with the global WM average for each patient, and normalized for total brain volume [26].

2.6 Imaging Analysis

All the images were processed on a Linux workstation using tools developed in the IDL programming environment (Research systems, Boulder, CO) [10]. Three dimensional regions of interest corresponding to parenchymal brain were extracted using the “brain extraction tools” algorithm [27] on the MP-RAGE and FLAIR images to segment gray and WM and CSF. Subregions were defined according to the LONI Probabilistic Brain Atlas (LPBA40). Each lobe (e.g. frontal, temporal etc.) was divided according to anatomical divisions and structures (e.g. middle orbitofrontal gyrus, superior frontal gyrus, hippocampus, etc.).

2.7 Statistical Analysis

The JMP software 11 PRO (SAS Institute, Cary, NC) was used for analyses. Descriptive analyses were used for demographic characteristics and one-way ANOVA and Wilcoxon tests for group comparisons between type 2 DM and control groups. Linear regression and least square (LS) models were used to find relationships between albuminuria, brain volumes and WMHs. Global CSF volume was included in all the analyses as co-variant to account for differences in regional atrophy and head size [28].

UACR ≥ 5 mg/g; UACR ≥ 10mg/g and MA [≥ 17mg/g (M), ≥ 25 mg/g (F)] were used as independent variables. The dependent variables were global and regional GM and WM volume and WMHs. The relationships between UACR, MA and brain volumes were analyzed using unadjusted linear regression and four least square models (LS) adjusted for (A-CSF; B-CSF, age and gender; C-CSF, HTN years and glucose levels; D- CSF, diabetes duration and HbA_1c). We present r² adjusted for model covariates and we conservatively selected models with adjusted r² _adj > 0.1, and P < 0.05. Each brain region was modeled separately to minimize repeated measures effects. The effect of smoking, alcohol abuse and other confounders was also evaluated.

3. Results

3.1 Characteristics of the Study Cohort

The characteristics of the study cohort are listed in Table 1. The type 2 DM group had higher BMI, HbA_1c, glucose, 24 hour systolic BP, longer HTN duration, worse cognitive performance, lower GM and WM volumes, as well as a slower preferred walking speed. Other characteristics, such as smoking pack years, C-reactive protein, urine creatinine, urine albumin and WMHs volume were similar between the groups. Control participants had higher cholesterol blood levels due to untreated hypercholesterolemia (11 participants).

Table 1. Demographic and clinical characteristics by group.

Characteristics	Diabetes n=85 (mean, STD deviation)	Control n=40 (mean, STD deviation)	P value
Age (years)	64.8 ± 8.3	67.5 ± 9.9	0.1
Males/Females	47/38	20/20	NS
BMI (kg/m2)	28.4 ± 6.5	24.5 ± 5.1	0.001
Hypertension	51 (60.0%)	7 (17.5%)	< 0.0001
Hypertension (years)	6.7 ± 9.1	2.07 ± 5.01	0.003
Diabetes duration (years)	11.6 ± 9.5	NA	NA
24hr systolic blood pressure (mmHg)	133.6 ± 7.9	128.3 ± 7.6	0.01
24hr diastolic blood pressure (mmHg)	70.9 ± 7.3	68.2 ± 8.0	0.1
Alcohol intake dose/week	4.0 ± 12.8	1.39 ± 2.9	0.2
Smoking pack (years)	14.4 ± 19.8	18.6 ± 19.4	0.3
Hemoglobin (g/dl)	13.2 ± 1.3	13.4 ± 1.3	0.4
HbA1c (%) (mmol/mol)	7.2 ± 1.2	5.6 ± 0.3	< 0.0001
Fasting Glucose (mg/dl)	120.5 ± 45.5	86.2 ± 11.9	< 0.0001
Urine creatinine (mg/dl)	103.4 ± 59.5	90.7 ± 43.7	0.2
Albumin creatinine ratio (mg/g)	15.4 ± 27.4	23.9 ± 47.3	0.2
Microalbuminuria	70.5 ± 53.7	116.3 ± 71.7	0.1
(30-300 mg/g)	n= 10 (8%)	n= 6 (4.8%)
UACR ≥ 17 mg/g (M), ≥ 25 mg/g (F)	49.4 ± 46	103.7 ± 73.5	0.03
	n= 18 (14.4%)	n= 7(5.6%)
UACR ≥ 10 mg/g	33.6 ± 39.4	56.9 ± 66.1	0.1
	n= 31 (24.6%)	n= 15 (11.9%)
UACR ≥ 5 mg/g	21.4 ± 31.9	40.8 ± 59.1	0.06
	n= 57 (45.6%)	n= 22 (17.6%)
UACR < 5 mg/g	3.2 ± 0.8	3.2 ± 1.19	0.9
	n= 29 (23.2%)	n= 18 (14.4%)
C-reactive protein (mg/L)	2.4 ± 4.2	1.4 ± 1.7	0.1
Cholesterol (mg/dl)	170.2 ± 41.3	191.4 ± 46.9	0.01
Retinopathy grading (0-9)	0.6 ± 1.7	0	NA
Global intracranial volume (ml)a	1555.8 ± 240.8	1637.7 ± 240	0.02
Frontal lobe volume gray matter volume (ml)a	97.9 ± 1.1	100.5 ± 1.7	0.2
Global gray matter volume (ml)a	623.46 ± 64.6	651.1 ± 81.5	0.01
Global white matter (ml)a	422.3 ± 58.1	436.7 ± 57.7	0.1
Global WMHs (ml)a	13.1 ± 9.7	12.4 ± 7.1	0.6
Global CSF volume (ml)b	510.2 ± 162.4	549.9 ± 185.4	0.1
Gait Speed (m/s)	1.05 ± 0.1	1.13 ± 0.1	0.01
TMT: T score	47.0 ± 11.6	49.1 ± 10.1	0.3
HVLT Total Recall T-score	45.6 ± 12	54.4 ± 11.1	0.0004
HVLT Delayed Recall T-score	41.8 ± 13.2	50.6 ± 11	0.001

^aVolumes were compared using least squares models adjusted for CSF, age, gender and group (reported parameters: LS mean and STD error).

^bVolumes were compared using least squares models adjusted for age, gender and group (reported parameters: LS mean and STD error).

Abbreviations: standard (STD); Least squares (LS); Urine albumin creatinine ratio (UACR); Males (M); Females (F); White matter hyperintensities (WMHs); Cerebrospinal fluid (CSF); Hopkins verbal learning test (HVLT); Instrumental activities of daily living scale (IADL); Trail making test (TMT).

3.2 Relationship between UACR and MA with Brain Structural Changes

In diabetic participants, increasing UACR levels were associated with global and regional GM atrophy within the frontal and occipital lobes. Figure 1 demonstrates this relationship between UACR ≥ 5 mg/g and global and frontal lobe GM in an adjusted LS models for age and gender in type 2 DM patients. UACR < 5 mg/g was not associated with brain volumes or functional measures.

Figure 1.

Association of UACR ≥ 5 mg/g with lower global (A) (r2adj= 0.3, P= 0.01) and regional (B) (frontal lobe: r2 adj= 0.3, P= 0.005) gray matter volume in diabetic patients independent of global cerebrospinal fluid volume (ml), age and gender. Abbreviations: GM (gray matter); UACR (urine albumin creatinine ratio).

Table 2 shows functional associations among GM volumes and UACR ≥ 5 mg/g using a linear regression (unadjusted) and four models (A-D) adjusted for age, gender, HTN years, glucose levels, DM duration, HbA_1c and CSF volume. Within the DM group, UACR levels ≥ 5 mg/g were associated with global (r² _adj= 0.1, P= 0.04) GM atrophy. Within the frontal lobe, increasing UACR levels were associated with regional GM atrophy (UACR ≥ 5 mg/g [r² _adj= 0.2, P= 0.01]; UACR ≥ 10 mg/g [r² _adj = 0.3, P= 0.05]). In the frontal lobe GM, the strongest association with UACR ≥ 5 mg/g (r² _adj= 0.1, P= 0.009) was found in model C adjusted for CSF, HTN years and glucose levels. These relationships were also significant for models A, B and D. UACR levels ≥10 mg/g were also associated with lower frontal lobe GM in all models (e.g. model B, r² _adj= 0.4, P= 0.01) and lower global GM volume in model B (r² _adj = 0.4, P=0.03). UACR levels ≥ 17 mg/g (men) and ≥ 25 mg/g (women) were associated with lower global (r² _adj= 0.3, P= 0.01-0.02) and regional frontal lobe (r² _adj= 0.4, P= 0.01) GM volumes on models C and D. We also tested log UACR, and detected similar associations with GM volume. UACR < 5 mg/g was not associated with GM changes. UACR was not associated with WM volume or WMHs in the DM group. In controls, UACR was not associated with GM changes.

Table 2. Relationship between UACR and GM volumes (type 2 DM group).

UACR ≥ 5 mg/g		Global GM	Global frontal lobe	Superior frontal gyrusa	Middle frontal gyrusa	Inferior frontal gyrusa	Middle orbitofrontal gyrusa	Lingual gyrusa
Linear	r2 adj	0.07	0.1	0.08	0.1	0.08	0.1	0.07
Regression	P	0.02	0.006	0.01	0.003	0.01	0.008	0.02
LS Model–A	r2 adj	0.1	0.2	0.1	0.2	0.1	0.4	0.3
	P	0.04	0.01	0.02	0.006	0.02	0.01	0.03
LS Model–B	r2 adj	0.3	0.3	0.3	0.2	0.1	0.3	0.4
	P	0.01	0.005	0.008	0.01	0.02	0.01	0.01
LS Model–C	r2 adj	0.1	0.1	0.1	0.1	0.1	0.3	0.3
	P	0.03	0.009	0.01	0.01	0.02	0.01	0.05
LS Model–D	r2 adj	NS	0.2	0.1	0.2	0.1	NS	0.4
	P	NS	0.02	0.05	0.02	0.04	NS	0.02

^aLeft hemisphere

Model A= adjusted for CSF only; Model B= adjusted for CSF, age and gender; Model C= adjusted for CSF, hypertension years and glucose levels; Model D= adjusted for CSF, diabetes duration and HbA1c. Abbreviations: Urine albumin creatinine ratio (UACR); Gray matter (GM); Diabetes mellitus (DM); Least squares (LS); Cerebrospinal fluid (CSF).

Figure 2 shows the GM volume for each UACR range for the DM and the control groups (Figure 2A) and for the entire cohort (Figure 2B). Ten DM participants with clinically significant MA (UACR 30-300 mg/g) had lower global GM volume as compared to diabetic subjects with UACR < 5 mg/g by 8.5% (p=0.03). In the DM group, as UACR increased, GM volume decreased by at least 30 cm as compared to controls (UACR ≥ 5 mg/g [r² _adj= 0.3, P= 0.04]; 612.08 cm³ [type 2 DM], 649.929 cm³ [controls]). DM subjects with MA, had also lower frontal lobe GM volume (r² _adj=0.4, P=0.04), and lower orbitofrontal gyrus GM volume bilaterally (r² _adj= 0.8, P= 0.02-0.03) (model adjusted for CSF, HTN and glucose levels) as compared to controls. For the entire cohort (Figure 2B), GM volume was lower for subjects with UACR ≥ 5 mg/g (p=0.03), UACR ≥10 mg/g (p=0.02) and MA (p=0.03) as compared to subjects with an UACR closer to normal. Urine creatinine ≥ 1.2g) was not associated with structural brain changes.

Figure 2.

A. Mean global gray matter (GM) volume for each urine albumin-to-creatinine ratio (UACR [mg/dl]) subgroup for the diabetes (dotted bars) and the control (black bars) groups. GM volume was lower in the diabetic participants with MA as compared to diabetic participants with UACR<5 [mg/dl]. B. For the entire cohort, GM volume was lower for participants with UACR ≥5 (p=0.03), UACR >10 (p=0.02) and MA (p=0.03) as compared to participants with an UACR closer to normal.NJ

3.3 Relationship between gray matter atrophy, cognition and diabetes control

The type 2 DM group performed worse than controls on cognitive measures of learning and memory (Table 1, HVLT total recall [P= 0.0004], HVLT delayed recall [P= 0.001]). In DM subjects, lower GM volumes were associated with worse executive function performance on TMT T scores (A and B). This relationship was observed for global GM, superior frontal, right middle frontal, right lingual and middle orbitofrontal GM gyri (r² _adj= 0.05-0.2, P= 0.006-0.03). DM severity (i.e. glucose levels, HbA_1c and disease duration) was not associated with global and frontal lobe GM atrophy.

3.4 Relationship between gray matter atrophy and hypertension

In diabetic patients, lower global (r² _adj= 0.2, P= 0.02) and frontal (r² _adj= 0.2, P=0.03) GM volumes were associated with higher UACR ≥ 5 mg/g independent of global CSF volume, HTN diagnosis, and 24h systolic BP. A systolic BP effect on global GM (P= 0.02) and frontal lobe GM (P= 0.04) was observed, but systolic BP was not associated with UACR. In controls, systolic BP correlated with higher UACR ≥ 5 mg/g (r² _adj= 0.09, P= 0.01) but not with GM volumes. Models including the effect of diastolic BP and mean BP were not significant. The volume of global WMHs was not associated with higher UACR in patients with DM, HTN or both.

3.5 Relationship between gray matter atrophy and BMI

BMI was not associated with global gray matter volume in the diabetic group or in the entire cohort (linear regression, LS models). However, higher BMI was associated with slower gait speed in the control group (r²= 0.09, p= 0.0002) and in the entire cohort (r²= 0.04, p=0.0007), but not in the DM group. BMI showed no correlation with 24-hour systolic blood pressure.

4. Discussion

This study reports that UACR ≥ 5 mg/g is associated with lower GM volume in type 2 DM. UACR-GM relationships were observed for global GM measures and regionally (frontal lobe), and they were independent from DM duration, fasting glucose levels, HbA_1c and HTN. The UACR-GM associations were detected even at a UACR ≥ 5 mg/g, which is lower than clinical MA threshold. Diabetic subjects with MA had 8.5% lower GM volume as compared to those with normal UACR. The degree of GM atrophy was clinically relevant and its manifestation was evident as a worse cognitive function in diabetic participants. In particular, we found a correlation between UACR, frontal lobe atrophy and executive dysfunction. This relationship is clinically relevant because DM imposes a risk for cognitive decline and dementia. A urine creatinine value ≥ 1.2g) was not linked to GM atrophy, demonstrating a stronger relationship between UACR and brain volume changes. UACR-GM associations were not detected in the control group, even though they had similar UACR levels and risk factors (e.g. HTN, hyperlipidemia). Therefore, these findings suggest that the adverse effects of endothelial dysfunction associated to chronic hyperglycemia may impact both the kidney and the brain even before MA is clinically diagnosed.

MA associated to endothelial dysfunction in diabetic nephropathy causes diffuse microvascular changes affecting the efficacy of glomerular filtration [29, 30]. Similarly, DM-related cerebral microvascular disease is associated with lower cortical perfusion, and diffuse vascular manifestations in GM, WMHs and lacunar infarcts [31]. In our study, UACR and MA were correlated with lower GM volume, suggesting a similar microvascular pathophysiology affecting the kidney and the brain. This relationship between UACR and frontal lobe GM may be explained by the impact of DM on brain aging. In normal aging, frontal lobe GM volume begins to decline around the fifth decade of life [31, 32]; type 2 DM may accelerate this process [17, 33].

Other contributing factors to the decreased brain volume in the type 2 DM group may be reduced cerebral blood flow, vasoreactivity and metabolism [9] that may lead to structural changes within GM and neurovascular units. As it has been previously reported, we did not find significant associations between UACR and WMHs in type 2 DM participants [16, 31, 34]; our study analyzed structural changes in cerebral cortex and white matter separately. The link between MA and brain volume reduction was only observed on GM structures.

MA, as an early marker of endothelial dysfunction and early predictor of renal disease (usually precedes the decline in glomerular filtration rate), may also be a signal of early-stage microvascular brain damage [35]. Epidemiological studies show that an UACR ≥ 5 mg/g is independently correlated with incidental HTN [22, 23]. An UACR of ≥ 10 mg/g is linked with increased mortality [23]; this risk further increases for every 4 mg/g raise in UACR [24]. In diabetic participants, the negative relationship between UACR ≥ 5 and GM volume was observed independent of 24h systolic BP and WMHs, supporting the hypothesis that DM is the major contributor of microvascular disease and GM loss in this cohort. UACR was not predictive of cognitive decline, but lower gray matter volumes were associated with executive dysfunction and higher UACR; these findings suggest that UACR may potentially hallmark early GM structural changes that may over time manifest as impaired cognitive performance. In our study, the association of UACR ≥ 5 mg/g) with GM atrophy supports the notion that subclinical levels of UACR may have clinical predictive value as an early marker of brain structural changes in type 2 DM. Therefore, the adverse effect of UACR on the brain may have begun even before MA is detected.

There are potential limitations of this study. UACR was calculated from a single spot early morning urine sample; however, this method has been shown to correlate with 24-hour urine albumin excretion rates [36]. Recent data have shown that early morning urine sample is a more reliable predictor of albumin excretion as compared to a random urine sample [37]. In this study, there was a small number of DM patients with MA (n= 10, range 30-300 mg/g), and especially with severe MA (n= 5). However, global GM volume was lower in the DM participants with MA, as compared to participants with DM and normal UACR (Figure 2). This was not observed in the control group. Therefore, in our study, patients with DM and MA were more likely to show reduced GM volumes as compared to control patients with MA.

This retrospective analysis of prospectively collected data included a small number of subjects with clinically significant MA (30-300 mg/g); therefore, UACR ≥ 17 mg/g in men > 25 mg/g in women) for spot urine sex specific MA was used. The number of non-diabetic hypertensive participants was limited due to recruitment difficulties. A prospective study with larger cohort would allow to further validate the observed results and to determine whether UACR can be used as predictor of brain damage and cognitive function loss.

Frontal lobe GM atrophy correlated with cognitive decline in diabetic participants, supporting previously reported relationships between vascular disease, frontal lobe atrophy and dementia in DM [38-42]. In contrast, in non-diabetic patients, similar UACR levels were not associated with significant brain volume loss. Although our correlational study was unable to elucidate the causal link between UACR and brain atrophy [29], other studies support the notion that hyperglycemia-associated microvascular disease contributes to brain loss [43].

5. Conclusions

This study demonstrated a link between UACR and GM volume changes in type 2 DM subjects as compared to age-matched controls. Therefore, even subclinical levels of UACR ≥ 5 mg/g) may hallmark an early decline in brain health. The latter changes present as GM volume loss, which has consequences on cognitive performance in older people with type 2 DM.

Subclinical albuminuria, an inexpensive diagnostic tool, may be used as an early associative marker of brain structural changes; this requires further investigation [35, 37]. Despite the relatively young age prevalent in our cohort, participants suffered from cognitive impairment. Brain volume reduction might be an early effect of microvascular disease, which further leads to functional changes such as cognitive decline and dementia. Therefore, UACR-GM relationships may be prodromal of severe cognitive impairment in the future. UACR levels may serve as an additional marker of DM-related brain structural changes, and thus add to a repertoire of clinically relevant, inexpensive diagnostic tools that can be used in this vulnerable population to identify diabetic patients at a greater risk of functional decline.

Abbreviations

MP-RAGE

3-D magnetization prepared rapid acquisition with gradient echo

r²_adj

Adjusted correlation coefficient

BP

Blood pressure

CSF

Cerebrospinal fluid

DM

Diabetes Mellitus

FLAIR

Fluid attenuated inversion recovery

GM

Gray matter

HTN

Hypertension

LS

Least squares

MRI

Magnetic Resonance Imaging

MA

Microalbuminuria

mcg/min

Micrograms per minute

mg/g

Milligrams per gram

TMT

Trail making test

UACR

Urine Albumin Creatinine Ratio

WM

White matter

WMHs

White matter hyperintensities