Circulating Adiponectin Is Associated with Renal Function Independent of Age and Serum Lipids in West Africans

Adiponectin, a protein secreted by adipose tissue, has been associated with renal dysfunction. However, these observations have not been adequately investigated in large epidemiological studies of healthy individuals in general and in African populations in particular. Hence, we designed this study to evaluate the relationship between adiponectin and renal function in a large group of nondiabetic West Africans. Total adiponectin was measured in 792 participants. MDRD and Cockroft-Gault (CG-) estimated GFR were used as indices of renal function. Linear and logistic regression models were used to determine the relationship between adiponectin and renal function. Adiponectin showed an inverse relationship with eGFR in univariate (BetaMDRD = −0.18, BetaCG = −0.26) and multivariate (BetaMDRD = −0.10, BetaCG = −0.09) regression analyses. The multivariate models that included age, sex, BMI, hypertension, smoking, HDL-C, LDL-C, triglycerides, and adiponectin explained 30% and 55.6% of the variance in GFR estimated by MDRD and CG methods, respectively. Adiponectin was also a strong predictor of moderate chronic kidney disease (defined as eGFR < 60 mL/min/1.73 m2). We demonstrate that adiponectin is associated with renal function in nondiabetic West Africans. The observed relationship is independent of age and serum lipids. Our findings suggest that adiponectin may have clinical utility as a biomarker of renal function.

Serum adiponectin is elevated in persons with impaired kidney function but this becomes normal following kidney transplantation [12]. However, the relationship between adiponectin and kidney function is complex. For example, low adiponectin level is associated with T2D at the onset but adiponectin levels seem to increase with increasing duration of T2D and in diabetic patients with nephropathy [6]. These observations suggest that circulating levels of adiponectin are heavily influenced by the pathophysiologic state of individuals as a result of multiple mechanisms, including 2 International Journal of Nephrology reduced clearance by the kidney [13] and degree of sensitivity or resistance to adiponectin [14,15].
Adiponectin has been suggested as a potential kidney biomarker in chronic kidney disease (CKD) progression [12]. However, the studies linking adiponectin and renal function have two potential limitations. First, the results and conclusions of these studies have been inconsistent and contradictory [8,[16][17][18]. Second, most of the studies investigating the relationship between adiponectin and kidney function have been conducted in persons with existing metabolic disorders (e.g., diabetes). To our knowledge, there are only two studies [16,19] of healthy individuals (i.e., asymptomatic individuals, not previously diagnosed with any serious or chronic illnesses with the exception of hypertension). Given the increasing importance of kidney diseases in sub-Saharan Africa and the young age (20-50 years) at which CKD is often seen [20], investigating the potential utility of adiponectin as a biomarker for renal function in these populations is warranted.
We have previously investigated the relationship between adiponectin levels, obesity, and metabolic markers in West Africans and showed that adiponectin was significantly associated with age, BMI, waist circumference, and serum lipids [21]. Since the relationship between adiponectin and renal function seems to be confounded by disease status, we reasoned that it would be informative to study this relationship in a cohort without any known major morbidity. In the present study, we aimed to investigate the relationship between renal function as measured by estimated glomerular filtration rate (eGFR) and adiponectin in nondiabetic West Africans.

Study Population. This study included nondiabetic West
Africans aged 18-79 years who participated in the Africa America Diabetes Mellitus (AADM) study. The AADM study, a T2D project ongoing in West Africa, is designed to investigate the genetic and environmental determinants of T2D and associated comorbidities, including obesity, kidney function, and several biochemical markers of metabolic disorders [22]. Briefly, the AADM study was conducted in three centers in Nigeria (Enugu, Ibadan, and Lagos) and two centers in Ghana (Accra and Kumasi) as part of an international collaboration between US investigators and West African Scientists. The AADM study enrolled both families and unrelated persons with and without diabetes (controls). The study protocol was approved by the IRB of each participating West African institution, and participants gave their informed consents following the Helsinki guidelines. Participants with an eGFR ≤ 30 mL/ min per 1.73 m 2 were excluded from this study. The final study sample consisted of 792 qualified participants.

Anthropometric and Clinical
Measurements. The study used standardized protocols for anthropometric (height, weight, and waist circumference) and clinical measurements. Weight was measured in light clothing using an electronic scale to the nearest 0.1 kg, height was obtained using a stadiometer to the nearest 0.1 cm, and waist circumference was measured to the nearest 0.1 cm at the narrowest part of the torso. Blood pressure was measured in the sitting position using an oscillometric device (Omron). Fasting serum samples were used to measure number of clinical biomarkers including creatinine, HDL-C, LDL-C, and triglycerides using an autoanalyzer, COBAS Integra 400 plus (Roche Diagnostics, Indianapolis, IN); creatinine was measured using a modified Jaffé reaction. Total serum adiponectin was measured using a commercially available immunoassay kit (R&D Systems, Minneapolis, MN) following the manufacturer's instructions. eGFR was used as measure of kidney function. The estimation of GFR remains controversial especially in Africans [23] because available methods have not been validated in African or Asian populations. In the absence of generalizable methods in African populations, we choose to estimate GFR in this study by using two distinct calculation methods: (1) the Modification of Diet in Renal Diseases (MDRD) formula, that is widely used by clinicians and researchers [24] [26], BSA = (body weight 0.425 (in kg) × height 0.725 (in cm)) × 0.007184. Using the guidelines of the National Kidney Foundation Kidney Disease Outcome Quality Initiative [27], we defined moderate CKD as eGFR < 60 mL/min/1.73 m 2 . The reference group (i.e., participants with "normal" eGFR) was individuals with eGFR ≥ 60 mL/min per 1.73 m 2 .

Statistical Analysis.
All statistical analyses were carried out using SPSS package, version 16.0. Normality of all variables was checked using Q-Q plots. Variables with skewed distributions (adiponectin, creatinine, HDL-C, LDL-C, TG, eGFR MDRD , and eGFR CG ) were logarithmically transformed to prevent violation of normality assumptions. Continuous variables were expressed as mean ± standard deviation unless otherwise specified; Student's-t test was used to compare mean between groups. Partial Pearson's correlation coefficients and simple and multiple linear regression models were used to evaluate the association between renal function (eGFR MDRD and eGFR CG ) and adiponectin. In these models, eGFR MDRD or eGFR CG was the dependent variable, and adiponectin was the explanatory variable. Adjustment was made for factors that are known to affect kidney function specifically age, sex, BMI, hypertension, smoking, and serum lipids (triglycerides (TG), HDL-cholesterol (HDL-C), and LDL-cholesterol (LDL-C)). Hypertension, smoking, and sex were included in the models as categorical variables. Using logistic regression models, we determined the predictors of eGFR MDRD and eGFR CG in moderate CKD (i.e., eGFR < 60 mL/min per 1.73 m 2 ). The first logistic model was an unadjusted model in which only adiponectin levels were the predictor; in the second model, we adjusted for age and sex, and, finally, the third logistic model included the lipid  parameters (HDL-C, LDL-C, and TG). Significance level was set at 0.05 for all analyses.

Results
The 792 subjects comprised 315 men (mean age 48.3 ± 14.2 years) and 477 women (mean age 45.1 ± 12.0 years). The characteristics of this cohort are presented in Table 1.
Overall, men had lower BMI, waist circumference, and adiponectin levels ( To explore the relationships between eGFR, adiponectin, and potential confounders, we first developed scatter plots ( Figure 1) and calculated partial Pearson's coefficients of correlation ( Table 2). The same variables were associated with both eGFR MDRD and eGFR CG . eGFR was negatively associated with adiponectin (P < 0.01), age (P < 0.01), serum lipids (LDL-C, HDL-C, TG), and hypertension (P < 0.01). No association was found between eGFR and smoking (P = 0.68) in this cohort ( Table 2). In contrast, adiponectin was positively associated with age, HDL-C, and hypertension ( Table 2). Interestingly the association between eGFR CG and adiponectin was stronger compared to the association between eGFR MDRD and adiponectin (r p = −0.35 versus r p = −0. 19).
In unadjusted linear regression model, adiponectin explained 3% of the variation in eGFR MDRD (beta = −0.18, P < 0.0001) whereas it explained about 7% of the variation in eGFR CG (beta = −0.26, P < 0.0001). In a multivariate model that included age, sex, BMI, hypertension, and serum lipids, adiponectin remained a significant predictor of eGFR MDRD (beta = −0.10, P = 0.004) and eGFR CG (beta = −0.09, P = 0.001) as well as age (P < 0.0001) and serum lipids (P < 0.0001); this multivariate model explained 30% of the variation in eGFR MDRD (Table 3). Additionally BMI and International Journal of Nephrology 5   Significance level set at 0.05.  CKD coding: 0, reference group represents subjects with eGFR ≥ 60 mL/min/1.73 m 2 ; 1 represents subjects with eGFR < 60 mL/min/1.73 m 2 . * Serum lipids in the model: HDL-C, LDL-C, triglycerides. SE: standard error; B: logistic regression coefficient; df: degree of freedom; OR: odds ratio; AdipoQ: adiponectin. Sex coding-0: male and 1: female. sex were also significantly associated with eGFR CG , and the variance explained why eGFR CG increased to 55.6%. Furthermore, adiponectin was significantly associated with moderate CKD regardless of the estimation method used to define moderate CKD (B MDRD = 1.61, P MDRD = 0.006 and B CG = 1.90, P < 0.0001, (Table 4)), and these associations are independent of the effects of serum lipids, age, and sex.

Discussion
We investigated the relationship between renal function and adiponectin in a large cohort of nondiabetic West Africans. Due to the fact that none of the GFR estimation equations has been formally validated in African populations we implemented the two most commonly used methods (MDRD and the Cockroft-Gault) in this study. The MDRD method is widely used in clinical practice, and the Cockroft-Gault method is the first creatinine-based method adopted by clinicians [28]. Despite the high correlation (r p = 90%) between these two methods in our study, the CGestimated GFR identified more individuals as having low eGFR (<60 mL/min per 1.73 m 2 ) than the MDRD-estimated eGFR (16.3% versus 8.5%). Our finding corroborates with previous observations in other populations [29,30]. For example, Wetmore et al. found that 5.3% of the subjects in their study were classified by the MDRD equation as having CKD (eGFR < 60 mL/min per 1.73 m 2 ) whereas 19.7% were classified by the CG formula [29].
Regardless of the method used to estimate GFR, we observed a negative association between eGFR and circulating adiponectin, as has been previously reported by other investigators [8,13,31,32]. However, most previous observations have been among persons with existing diseases such as type 1 and 2 diabetic, coronary disease and CKD. Thus, apart from confirming previous findings our study demonstrates that the inverse relationship between eGFR and adiponectin extends to persons with no obvious metabolic disorders. Additionally, we observed that the association between eGFR and adiponectin was stronger when Cockroft-Gault calculation was used as marker of renal function. While the reason for the observed difference is not obvious, we posit that it may be due to the fact that the CG equation takes into account anthropometric measurements-which also influence adiponectin levels-while the MDRD formula does not [28,33].
Adiponectin levels alone explained 3% to 7% of the variation in eGFR in this cohort depending on the estimation method. Interestingly, in multiple regression analysis adjusting for additional factors such as smoking, sex, hypertension, and lipids, the variance explained increased by 8-fold to 56% for eGFR CG and by 10-fold to 30% for eGFR MDRD . This dramatic increase in the percent variance explained was due primarily to the inclusion of age, sex, BMI, and serum lipids (HDL-C, LDL-C and triglycerides) in the regression models. The relationship between age, serum lipids and eGFR is well documented [34][35][36][37]. For example, baseline serum lipids (total cholesterol and LDL-C) are independent risk factors for renal diseases as well as powerful predictors of renal function [38]. It is therefore noteworthy that adiponectin remained a significant independent predictor of eGFR in the presence of age and serum lipids.
International Journal of Nephrology 7 We also found adiponectin to be independently associated with moderate CKD (Table 4). These observations suggest that adiponectin may serve as a potential biomarker of kidney function even among individuals who do not have overt T2D. Furthermore, it provides additional evidence to support evaluating adiponectin along with other plasma proteins such as apolipoprotein A-IV, fibroblast growth factor 23, neutrophil gelatinase-associated lipocalin, and the natriuretic peptides as a discriminatory biomarker of CKD progression [12]. To fully understand the potential role of adiponectin in routine clinical pathology, more populationbased studies as well as studies of adiponectin isomers in kidney function are needed. Kawamoto et al. [16] found a positive association between high molecular weight (HMW) adiponectin, and GFR, raising an important question about the role of adiponectin isomers in kidney function. HMW adiponectin has been shown to be the most potent form of adiponectin in its insulin sensitizing function [39,40]. It is also likely that HMW adiponectin may offer more protection in the context of kidney function as suggested by the Japanese study of persons with mild CKD [16].
Despite our careful efforts to evaluate the relationship between adiponectin and renal function in this cohort, the design of the study (cross-sectional) and the use of eGFR as measure of renal function in healthy individuals are potential limitations of the study. A cross-sectional study does not permit the determination of causality in a relationship. Therefore, longitudinal studies are needed to provide such data. Secondly, MDRD-based estimates of eGFR may not be suitable for all populations, especially Africans, Asians [23], and individuals suffering from malnutrition or eating a vegetarian or low-meat diet. While our cohort did not include individuals with malnutrition, our study did not collect dietary data, and we could therefore not control for dietary factors in our analyses. Additionally, our definition of moderate CKD was solely based on eGFR, a creatininebased estimation. No other kidney function markers such as proteinuria or urine albumin/creatinine were available in this study to explain the impaired kidney function seen in the subjects with low eGFR.
In summary, we demonstrated that adiponectin is an independent predictor of both eGFR and moderate CKD in nondiabetic West Africans. This confirms and extends previous findings of an inverse relationship between adiponectin and renal function. The uniqueness of our study lies in the inclusion of a healthier and younger West African population characterized by a lower prevalence of CKD. These attributes allowed us to evaluate the relationship between adiponectin and renal function across a larger age spectrum in a population that is generally understudied.

Ethical Approval
This study was conducted according to the principles expressed in the declaration of Helsinki. Ethical approvals were obtained from participating institutions' IRBs. All study participants provided written informed consent for sample collection and subsequent analysis.