Vascular cell adhesion molecule-1, but not intercellular adhesion molecule-1, is associated with diabetic kidney disease in Asians with type 2 diabetes Jian-Jun Liu, Lee Ying Yeoh, Chee Fang Sum, Subramaniam Tavintharan, Xiao Wei Ng, Sylvia Liu, Simon B.M. Lee, Wern Ee Tang, Su Chi Lim PII: DOI: Reference:

S1056-8727(15)00058-6 doi: 10.1016/j.jdiacomp.2015.02.011 JDC 6407

To appear in:

Journal of Diabetes and Its Complications

Received date: Revised date: Accepted date:

17 December 2014 4 February 2015 23 February 2015

Please cite this article as: Liu, J.-J., Yeoh, L.Y., Sum, C.F., Tavintharan, S., Ng, X.W., Liu, S., Lee, S.B.M., Tang, W.E. & Lim, S.C., Vascular cell adhesion molecule1, but not intercellular adhesion molecule-1, is associated with diabetic kidney disease in Asians with type 2 diabetes, Journal of Diabetes and Its Complications (2015), doi: 10.1016/j.jdiacomp.2015.02.011

This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.

ACCEPTED MANUSCRIPT Vascular cell adhesion molecule-1, but not intercellular adhesion molecule-1, is associated with diabetic kidney disease in Asians with type 2 diabetes

RI P

T

Jian-Jun Liua, Lee Ying Yeohc, Chee Fang Sumb,c, Subramaniam Tavintharanb,c, Xiao Wei Nga, Sylvia Liua , Simon BM Leed, Wern Ee Tangd and Su Chi Limb,c for SMART2D study

SC

a, Clinical Research Unit, Khoo Teck Puat Hospital, Singapore 768828; b, Diabetes Centre, Khoo

NU

Teck Puat Hospital, Singapore 768828; c, Department of Medicine, Khoo Teck Puat Hospital,

MA

Singapore 768828; d, Yishun Polyclinic, National Health Group, Singapore 768796

ED

Corresponding to: Su Chi Lim, Associate Professor, Diabetes Center, Khoo Teck Puat Hospital, 90 Yishun Central, Republic of Singapore 768828, email: [email protected] Tel

AC

CE

PT

+65 6602 2353 Fax +65 66023772

ACCEPTED MANUSCRIPT Abstract Background and Aims: The association of adhesion molecules ICAM-1 and VCAM-1 with

RI P

T

cardiovascular diseases has been well-studied. However, their roles in diabetic kidney disease (DKD) are incompletely understood. We aim to study the association of plasma ICAM-1 and

SC

VCAM-1 with DKD in Asians with type 2 diabetes (T2DM).

NU

Subjects and Methods: 1,950 Asians with T2DM were included in this cross sectional study.

MA

Plasma ICAM-1 and VCAM-1 were measured by immunoassays. Results: Renal filtration function (eGFR) declined and urinary albumin-to-creatinine ratio (ACR)

ED

levels increased progressively with the increase in plasma VCAM-1 levels. In contrast, no significant changes in eGFR and ACR were observed in subjects across different plasma ICAM-1

PT

levels. Both ICAM-1 and VCAM-1 were correlated with ACR (Rho= 0.153, p12%; 6) subjects who

CE

took NSAIDS within the same day of phlebotomy and 7) current user of oral steroids equivalent to prednisolone >5 mg / day. Demographic, clinical and biochemical data as well as history of

AC

medication usage were recorded in a standardized form and entered into a data management server. Retinopathy was ascertained by medical records. By the cut-off time of this study (end of January 2014), 1950 T2DM subjects have been recruited. SMART2D study complies with Helsinki Declaration, has been approved by our domain specific ethical review board and written informed consent has been obtained from all participants. Clinical and biochemical measurements Blood pressure was taken in a sitting position by an automated blood pressure monitor after letting the subjects rest for 5 minutes (Dinamap Pro 100V2, Freiburg, Germany). HbA1c was

ACCEPTED MANUSCRIPT quantified by point-of-care immunoassay analyzer which had met NGSP (National Glycohemoglobin Standardization Program) performance standard (DCA Vantage Analyzer;

RI P

T

Siemens AG, Erlangen, Germany). Total cholesterol, HDL cholesterol and LDL cholesterol were measured by enzymatic methods using Kodak Ektachem chemistry slides. Triglycerides and

SC

creatinine were quantified by enzymatic methods (Roche/Hitachi Cobas C System; Roche Diagnostic GmbH, Mannheim, Germany). Urinary albumin was measure by solid-phase

NU

competitive chemiluminescent enzymatic immunoassay (Immulite, DPC, Gwynedd, UK).

MA

Estimated glomerular filtration rate (eGFR) was calculated based on the Modified Diet in Renal Disease (MDRD) formula, which performed well in subjects with diabetes (Rognant et al., 2011).

ED

High sensitivity c-reactive protein (hsCRP) concentration was quantified by solid phase sandwich ELISA with a minimum detection limit of 0.022 ng/ml and intra- and inter- assay

PT

coefficients of 5.5% and 6.5%, respectively (R&D Systems, Minneapolis, MN, USA). Depending

CE

on hsCRP concentration in each sample, samples were diluted between 50 to 10,000 times to

AC

ensure that the measurements fell within the range of the standard curve. Plasma ICAM-1 and VCAM-1 levels were measured by multiplex immunoassays on a Luminex 200 platform according to protocol provided by the manufacture (Affymetrix, Santa Clara, CA, USA;). Plasma samples were diluted 200 times before multiplex assay. The minimum detection limit for VCAM-1 and ICAM-1 was 1.13 pg/ml and 2.92 pg/ml, respectively. The intra-assay coefficients were 3.1% and 5.6% and the inter-assay coefficients were 5.2% and 7.8% for VCAM-1 and ICAM-1, respectively. Statistical analysis

ACCEPTED MANUSCRIPT Continuous data were presented as mean ± standard deviation (SD) and categorical data were expressed as proportions. Total triglycerides, urinary albumin-to-creatinine ratio (uACR), VCAM-

RI P

T

1, ICAM-1 and hsCRP were reported as median (interquartile range, IQR) and log-transformed before statistical analyses because of skewed frequency distribution. Subjects were divided into

SC

quartiles based on either plasma VCAM-1 or ICAM-1 levels to visualize clinical and biochemical characteristics. One way analysis of variance (ANOVA) or χ² test were used to analyze the

NU

differences in continuous or categorical variables across quartiles where appropriate (Table 1

MA

and Supplementary Table 1). Bivariate correlations between clinical, biochemical variables and plasma VCAM-1 or ICAM-1 levels were examined with Spearman rank correlation test. General

ED

linear regression models were employed to study which variables were independently associated with plasma VCAM-1 or ICAM-1 levels. Plasma VCAM-1 or ICAM-1 level was entered

PT

as dependent variable in the model and age, gender and ethnicities (Chinese, Malay and

CE

Chinese) were entered as main confounders (model 1). Selection of covariates was informed by

AC

pathophysiology of DKD and also based on suggestions from literature. Duration of diabetes, indicators of glycemic control (HbA1c and fasting plasma glucose), waist circumference (indicator of central obesity), systolic blood pressure and lipids profile (HDL, LDL cholesterol and triglycerides) were entered as covariates (model 2). BMI was not included with waist circumferences in the same model due to concerns of multi-collinearity. Renal filtration function (eGFR), uACR and plasma hsCRP were further adjusted in model 3 and 4. In the fully adjusted model, we also included medication usage as suggested by univariate analysis in Table-1 and Supplementary Table-1 (model 5). Finally, we employed backward linear regression on the basis of model 5 to identify the most parsimonious explanatory variables which

ACCEPTED MANUSCRIPT determine plasma ICAM-1 or VCAM-1 variability. Data analysis was performed by SPSS version

AC

CE

PT

ED

MA

NU

SC

RI P

T

21 and two-sided p < 0.05 was considered statistically significant.

ACCEPTED MANUSCRIPT Results The baseline clinical and biochemical characteristics of the study participants were shown in

RI P

T

Table 1. Individuals with VCAM-1 in the higher quartiles were older, had longer duration of diabetes, poorer glycemic control and higher systolic blood pressure (SBP) as compared to

SC

subjects with VCAM-1 in the lower quartiles. There were no significant differences in BMI, waist

NU

circumference, diastolic blood pressure, total cholesterol, LDL cholesterol, triglycerides and hsCRP levels across subjects in different VCAM-1 quartiles. Subjects with higher level of VCAM-

MA

1 were more likely to take insulin and less likely to take metformin. Notably, with the increase of VCAM-1 levels from the lowest to the highest quartile, eGFR declined progressively from 95.7

ED

 30.2 ml/min/1.73m2 to 73.2  36.3 ml/min/1.73m2. Concordantly, urinary ACR levels were

PT

increased with the elevation of plasma VCAM-1 levels. There was no significant difference in

CE

hsCRP levels in subjects across different VCAM-1 quartiles (Table-1). Subjects with higher levels of plasma ICAM-1 also had poorer glycemic control as indicated by

AC

higher levels of HbA1c and fasting plasma glucose as compared to those with lower level of ICAM-1. In contrast to VCAM-1, subjects with higher level of ICAM-1 were younger, had shorter duration of diabetes, higher BMI and waist circumference, lower HDL cholesterol and less likely to take statins as compared to those with ICAM-1 in the lower quartiles (Supplementary Table 1). Urinary ACR levels trended higher with the increase in ICAM-1 from quartile 1 to 4 but the difference did not reach a statistically significant level. Notably, hsCRP levels were significantly increased in subjects with higher levels of ICAM-1 (Supplementary Table 1).

ACCEPTED MANUSCRIPT Bivariate correlation analysis showed that both ICAM-1 and VCAM-1 were positively correlated with HbA1c, fasting plasma glucose, triglycerides and inversely correlated with HDL cholesterol.

RI P

T

However, plasma VCAM-1 and ICAM-1 levels were differentially correlated with some other clinical phenotypes in our subjects with T2DM. As shown in Table 2, VCAM-1 levels were

SC

positively correlated with age, duration of diabetes and SBP but were not significantly correlated with BMI and waist circumferences. On the contrary, ICAM-1 was inversely

NU

correlated with age, duration of diabetes and positively correlated with BMI and waist

MA

circumferences. Both ICAM-1 and VCAM-1 were positively correlated with urinary ACR but only VCAM-1 was correlated with eGFR. Notably, plasma ICAM-1, but not VCAM-1, was correlated

ED

with hsCRP (Figure 1 and Supplementary Figure 1).

PT

To further study the relationship between VCAM-1, ICAM-1 and DKD, we divided subjects into low risk, moderate risk, high risk and very high risk categories based on KDIGO 2012 chronic

CE

kidney disease outcome prognosis (Levin and Stevens, 2014). Plasma VCAM-1 levels were

AC

increased from 78 (64-95) ng/ml in subjects with low risk to 80 (65-96), 84 (70-102) and 95 (81119) ng/ml in subjects with moderate, high and very high adverse outcome risks. On the contrary, plasma ICAM-1 levels did not change significantly in subjects across different risk categories (57(42-76), 57(43-79), 57(44-78) and 60(46-80) ng/ml in patients with low, moderate, high and very high risks, respectively). General linear regression models were employed to study whether plasma VCAM-1 or ICAM-1 was independently associated with indicators of renal injuries in patients with T2DM. As shown in Table 3, both eGFR and urinary ACR were independently associated with VCAM-1 levels after

ACCEPTED MANUSCRIPT adjustment for multiple covariates (Table 3, Model 3 to 5). On the contrary, we did not observe an association of eGFR with ICAM-1 in the multivariable models. Although urinary ACR was

RI P

T

weakly but independently associated with ICAM-1 after adjustment for age, gender, ethnicities, glycemic control, lipids profile, SBP and waist circumference (Supplementary Table 2, Model 3),

SC

this association was abolished after further adjustment for hsCRP (Supplementary Table 2, Model 4). Backward linear regression models revealed that eGFR was the main determinant of

AC

CE

PT

ED

MA

NU

plasma VCAM-1 variability whereas hsCRP was the main determinant of ICAM-1 variability.

ACCEPTED MANUSCRIPT Discussions The associations of ICAM-1 and VCAM-1 with DKD in patients with T2DM are still incompletely

RI P

T

understood. In this large clinical study among Asians with T2DM, we observed that plasma VCAM-1, but not ICAM-1, was independently associated with DKD whereas elevated ICAM-1

SC

might reflect a high systemic inflammation tone. Our data suggested that VCAM-1 and ICAM-1,

NU

though closely related in structure and function, might play non-redundant roles in pathogenesis and progression of DKD in diabetic patients.

MA

Early basic and clinical studies have shown that ICAM-1 and VCAM-1 have some discernible

ED

differences in tissue distribution and receptor specificity (Pradhan et al., 2002). In addition, transcriptional regulation of these two adhesion molecules in response to inflammatory

PT

cytokines may also be different. For example, in endotoxin- stimulated ICAM-1 and VCAM-1

CE

over-expression in endothelial cells, activation of PTEN/Akt, p38 and PKC was essential for high expression of VCAM-1 but not ICAM-1 (Tsoyi et al., 2009). On the other hand, activation of ERK-

AC

1/2 may be more important for TNF-alpha-induced ICAM-1 over-expression in endothelial cells (Nizamutdinova et al., 2008). These data suggest that ICAM-1 and VCAM-1 may play different roles in different stages of the same disease. It is not unexpected that plasma ICAM-1 and VCAM-1 may associate with different clinical phenotypes in patients with T2DM as shown in this study. There are several potential explanations for the different association of VCAM-1 and ICAM-1 with DKD. First, earlier in vitro studies showed that endothelial cells isolated from diabetic patients preferentially express VCAM-1versus ICAM-1 upon stimulation of inflammatory

ACCEPTED MANUSCRIPT cytokines in the presence of high ambient glucose (Haubner et al., 2007). Therefore, it is reasonable to hypothesize that endothelial cells in patients with DKD may preferentially

RI P

T

activate VCAM-1 expression in response to elevated inflammation tone (Endemann and Schiffrin, 2004; Karalliedde and Gnudi, 2011); Second, increased plasma VCAM-1 often indicates

SC

an increased membrane expression of VCAM-1 on neo-vasculature bed at atherosclerotic plaque (Jager et al., 2000; Pradhan et al., 2002). Therefore, higher VCAM-1 in subjects with DKD

NU

may reflect higher atherosclerotic lesions as compared with those without renal impairment.

MA

However, an earlier study showed that VCAM-1 was prospectively associated with cardiovascular events independent of renal filtration function and albuminuria in patients with

ED

T2DM (Jager et al., 2000). In 1,075 subjects with data of CVD history collected in our present study, plasma VCAM-1 levels in those with prior CVD events (history of stroke, acute myocardial

PT

infarction, angina pectoris, coronary revascularization operations by ballooning or by-pass

CE

surgical procedures) were not significantly higher as compared to those who did not have

AC

history of CVD events (90.7  32.6 ng/ml vs 85.2  32.2 ng/ml, p=0.062). Furthermore, exploratory analysis in these 1,075 subjects revealed that further adjustment of CVD history did not attenuate the association of VCAM-1 with eGFR and urinary ACR in the multivariable models. It seems that the association of VCAM-1 and DKD cannot be explained by the potentially high atherosclerotic burden in patients with DKD; Third, we cannot exclude the possibility that elevated plasma VCAM-1 may be secondary to a reduced clearance from the kidney. The catabolism of VCAM-1 is poorly understood but we observed that the correlation of VCAM-1 and eGFR remained statistically significant in subjects with preserved renal filtration function (eGFR  60 ml/min/1.73m2, N= 1566, Spearman’s Rho= - 0.113, p