点击显示 收起
1 From the Division of Nephrology (MLT and TAI) and the Department of Biostatistics (AS), Vanderbilt University Medical Center, Nashville, TN, and the Division of Nephrology, Maine Medical Center, Portland, ME (JH)
2 Supported by National Institutes of Health grants no. R01 DK45604 and K24 DK62849 and Diabetes Research and Training Center grant no. DK-20593 from the National Institute of Diabetes, Digestive and Kidney Diseases; grant no. R01 HL070938 from the National Heart, Lung, and Blood Institute; and grant no. M01 RR-00095 from the National Center for Research Resources (to the Vanderbilt General Clinical Research Center). 3 Reprints not available. Address correspondence to TA Ikizler, Division of Nephrology, Vanderbilt University School of Medicine, Medical Center North, S-3223, 1161 21st Avenue, Nashville, TN 37232-2372. E-mail: alp.ikizler{at}vanderbilt.edu.
ABSTRACT
Background: Insulin resistance has been noted in patients with chronic kidney disease (CKD). The determinants of insulin resistance have not been well-studied in CKD patients.
Objective: The objective of this study was to examine the degree and determinants of insulin resistance in persons without diabetes but with stage 3–4 CKD.
Design: Demographic characteristics, metabolic hormones, and inflammatory markers were measured in 95 nonobese stage 3–4 CKD patients without prior diagnosis of diabetes mellitus and 36 control subjects without CKD. The estimated glomerular filtration rate (eGFR) was measured by using the Modification of Diet in Renal Disease study equation. Insulin resistance was measured with the use of the homeostasis model assessment of insulin resistance (HOMA-IR).
Results: After age and sex adjustments, HOMA-IR scores were significantly and positively correlated with body mass index (BMI) and percentage body fat. After control for age, race, adiponectin concentrations, sex, and eGFR in a multivariate regression model, BMI remained as the only significant predictor of insulin resistance (standardized regression coefficient = 0.55; P < 0.001). When substituted for BMI, percentage body fat also was an independent predictor of insulin resistance. The prevalence of abnormal HOMA did not differ significantly between CKD patients (98%) and BMI-matched control subjects (94%).
Conclusion: Whereas insulin resistance is highly prevalent in stage 3–4 CKD, the primary determinant of insulin resistance in this population is BMI, specifically, fat mass.
Key Words: Insulin resistance • chronic kidney disease • homeostasis model assessment • body mass index • adiposity
INTRODUCTION
Both the incidence and prevalence of chronic kidney disease (CKD) and end-stage renal disease (ESRD) continue to increase at an alarming rate in the United States. Much investigation has been focused on ESRD patients, but an increasing recognition of the high prevalence of moderate-to-severe CKD has redirected the attention to this patient population to identify risk factors associated with hospitalization, death, and progression to ESRD. Indeed, studies have shown that there is a greater risk of atherosclerotic events and a higher risk of death in patients with mild-to-moderate CKD than in those without kidney disease (1). Furthermore, CKD is accompanied by numerous metabolic derangements such as oxidative stress, chronic inflammation, and endothelial dysfunction (2).
Insulin resistance (IR) in advanced kidney disease has been well recognized since the seminal work by DeFronzo et al (3) using hyperinsulinemic euglycemic clamp techniques. IR was reported to be an independent risk factor for cardiovascular morbidity and mortality in patients with ESRD (4). IR associated with mild-to-moderate CKD has also been described, albeit in reports mainly from European and Japanese populations. To our knowledge, few studies have investigated IR in CKD patients in the United States, where 11% of the adult population is estimated to have CKD (5, 6), and potential determinants of IR in the US population have not been studied in detail. Greater attention is being focused on the role of inflammation, adiposity, and its associated adipokines such as adiponectin in the general and CKD population in the United States; however, their potential relation to IR and cardiovascular disease risk has yet to be clearly defined. The growing prevalence of obesity and metabolic syndrome in the United States, the complex relation of both conditions with CKD, and their association with cardiovascular disease risk underlie the importance of recognizing and defining the risk factors for IR in this patient population.
In the present study, we aimed to evaluate potential determinants of IR in a population of patients without diabetes but with stage 3–4 CKD. We hypothesized that the estimated glomerular filtration rate (eGFR) and body mass index (BMI; in kg/m2) would each be closely associated with levels of IR in persons with CKD. To test this hypothesis, we examined the relation among eGFR, BMI, and insulin resistance, as determined by using the homeostasis model assessment of IR (HOMA-IR) in 95 nondiabetic persons with moderate-to-severe (stage 3–4) CKD. We compared results in this group with those in a group of 36 subjects with normal kidney function who were frequency matched for race, sex, and BMI.
SUBJECTS AND METHODS
Patients
Subjects were recruited from among the patients attending the outpatient nephrology clinics at the Maine Medical Center (Portland, ME) and the Vanderbilt University Medical Center (Nashville, TN). Criteria for study participation included age > 18 y and CKD due to any cause, being followed in one of the above nephrology clinics, and stage 3–4 CKD as defined by an eGFR between 15 and 59 mL/min. The eGFR was calculated by using the abbreviated equation described in the Modification of Diet in Renal Disease (MDRD) study (7):
RESULTS
Baseline characteristics
The baseline characteristics of the CKD patients and the matched control subjects are shown in Table 1; the CKD patients were subdivided by disease stage. Race, sex, and BMI did not differ significantly between the groups. CKD patients were significantly older than the control subjects. Both the CKD and the control groups were composed of overweight and obese persons (BMI range: 25–32 and 26–30, respectively). Mean %BF for the CKD and control groups also did not differ significantly. In this nondiabetic population, the mean HOMA-IR score was 3.7 ± 3.2 in CKD patients and 3.1 ± 1.8 in control subjects (P = 0.73). As expected, baseline eGFR was significantly (P < 0.001) lower in the CKD patients than in the control subjects. The only significant (P < 0.001) difference between patients with stage 3 CKD and those with stage 4 CDK was in baseline eGFR. Significant differences were noted by sex in baseline weight, %BF, and adiponectin concentrations; therefore, subsequent analyses were adjusted for age and sex.
View this table:
TABLE 1. Baseline characteristics1
Correlation between homeostasis model assessment of insulin resistance and markers of inflammation, body mass index, adiposity, and kidney function
The potential relation between IR and inflammation among study participants was examined by comparing HOMA-IR scores with concentrations of inflammatory cytokines and with BMI, %BF, and eGFR (Table 2). In the unadjusted analysis, there was a significant (P < 0.05) negative correlation between HOMA-IR and plasma concentrations of IL-1β, IL-8, and TNF-. No significant correlations were found between HOMA-IR and other proinflammatory cytokines. The inverse correlation between HOMA-IR and adiponectin concentrations was significant (P = 0.007), but no significant correlation was found between HOMA-IR and resistin concentrations. However, IL-10, BMI, and %BF remained significantly associated after adjustment for age and sex.
View this table:
TABLE 2. Association between homeostasis model assessment of insulin resistance and inflammatory cytokines and adipokines, body mass index, body fat percentage, and estimated glomerular filtration rate (eGFR) in study participants with and without chronic kidney disease1
In both the CKD patients and the control subjects, HOMA-IR showed a highly significant (P < 0.001) correlation with BMI. When BMI was divided into tertiles of <24.9, 25–30, and >30, HOMA-IR scores differed significantly (P < 0.001, test for linear trend) in the CKD group (Figure 1). BMI and %BF also were significantly correlated with CRP in CKD patients (rs = 0.308, P = 0.002 and rs = 0.235, P = 0.025, respectively). CRP concentrations also differed significantly according to BMI tertile in the CKD and control groups (P = 0.008 and 0.007, respectively; Kruskall-Wallis test).
FIGURE 1.. Homeostasis model assessment of insulin resistance (HOMA-IR) versus BMI ranges in patients with chronic kidney disease (CKD) and control subjects. Mean HOMA-IR scores in CKD patients and control subjects by BMI ranges of normal (<24.9), overweight (25–30), and obese (>30) differed significantly between the 3 ranges (P < 0.001, linear trend test). Error bars represent 95% CIs. BMI <24.9 data from control subjects were omitted because n = 3.
Predictors of insulin resistance in chronic kidney disease patients by multivariate analysis
In a multivariable regression model after control for age, African American race, adiponectin, sex, and eGFR, BMI was a significant predictor of IR in CKD patients but not in control subjects. In a separate analysis, %BF measured by bioelectrical impedance analysis was substituted for BMI and was an independent predictor of IR. Because of the high correlation between BMI and %BF, data reduction methods were performed to combine these 2 variables into a single variable. This combination variable was also found to be a significant predictor of IR in multivariable analysis after control for age, eGFR, race, sex, and adiponectin concentrations (Table 3). The individual cytokines found to be significant in the age-adjusted analysis were also combined by data reduction methods and were placed in the model. Even with the addition of this variable, BMI and %BF remained significant predictor of HOMA-IR (data not shown).
View this table:
TABLE 3. Results of 3 separate multivariable linear regression models and significant predictors of insulin resistance with BMI, percentage body fat, and combined BMI and percentage body fat among patients with chronic kidney disease (CKD) and control subjects1
DISCUSSION
It has long been recognized that a complex relation exists among uremia, glucose dispersion, and insulin function. Alterations in insulin function associated with CKD were reported as early as 1951 (13, 14), and the effects of kidney disease on renal uptake and excretion of insulin were reported as early as 1970 (15). In a seminal series of studies, DeFronzo et al (3, 16) and DeFronzo (17) used euglycemic insulin clamp techniques to characterize uremic IR in patients with ESRD who required dialysis. Thus, the pathophysiology of uremic IR in patients undergoing dialysis has been relatively well recognized for many years. Dialysis-dependent patients are under severe physiological stress, and it is likely that additional metabolic abnormalities contribute to uremic IR in these patients. In contrast, few investigations have focused on understanding IR in the larger population with less severe CKD.
In the present study, we examined the determinants of IR in nondiabetic patients with stage 3–4 CKD on the basis of the hypothesis that worsening kidney function would be associated with increasing IR. To our surprise, our results show that IR in this CKD patient cohort is primarily determined by BMI and not by eGFR. Furthermore, %BF, when substituted for BMI, was also predictive of IR in CKD patients. Further analysis indicated that the %BF of BMI is the relevant component predicting IR (ie, HOMA-IR); this relation was maintained in multivariate regression analysis. Adjustment for IL-6 and TNF- did not change these results. Whereas a significant association between BMI and IR is well recognized in the general population, the relation between body composition (in particular, %BF and IR) in stage 3–4 CKD patients has been less well studied, and it constitutes a novel aspect of the present study. To our knowledge, this study is one of the first to describe BMI as the primary determinant of IR in CKD patients, and it is the first to evaluate the relative contribution of fat mass in this relation.
Recent studies have suggested a complex relation between IR and CKD. A cross-sectional study utilizing participants from the third National Health and Nutrition Examination Survey (NHANES III) examined associations between metabolic syndrome and CKD and found that a person's odds of having kidney disease increased as the number of metabolic syndrome components possessed by him or her increased. This association remained significant after adjustment for the presence of hypertension and diabetes, 2 well-known causes of CKD (5). Kurella et al (18) conducted a prospective study using the Atheroslcerosis Risk in Communities study cohort to establish the metabolic syndrome as an independent risk factor for CKD in nondiabetic adults. Their data indicated that obesity and other components of the metabolic syndrome may contribute to the development or progression of CKD, but the data did not indicate whether the development of CKD also contributes to IR.
We observed that, in our stage 3–4 CKD patient group, eGFR did not correlate with the degree of IR. This has also been noted by other investigators who evaluated the presence of IR in CKD (19, 20), regardless of the method by which IR or GFR was measured. Previous studies that examined IR in kidney disease patients are summarized in Table 4. Kobayashi et al (22) described a relation between eGFR and IR that was calculated with the use of the hyperinsulinemic euglycemic clamp technique, but that relation was not maintained in multivariate analysis. Thus, whereas IR is present in these patients, the severity of underlying CKD does not seem to be the principal cause of the metabolic derangement—at least in our study population.
View this table:
TABLE 4. Summary of literature evaluating insulin resistance in chronic kidney disease patients without diabetes1
The prevalence of obesity continues to increase in the United States, and thus much attention is being focused on the role of adipose tissue—in particular, visceral adipose tissue—as an active secretory organ modulating endocrine systems. The adipokine adiponectin is known for its role in regulating insulin sensitivity (24). Although adiponectin is secreted by adipocytes, its concentrations are lower in obese subjects than in lean subjects (25). This counterintuitive relation is not completely understood, but feedback inhibition of adiponectin's production by inflammatory cytokines such as TNF- (26), which are higher with greater visceral obesity, may contribute to it (27). Low adiponectin concentrations have been associated with the development of IR in mouse models of obesity (28). The correlation between IR and adiponectin in the present study approached significance only in the adjusted analysis.
Visceral fat contains greater amounts of inflammatory mediators—including CRP, IL-6, and TNF-—than does subcutaneous fat, and these mediators are thought to contribute to the development of IR (29). In the ESRD population, Axelsson et al (30) found an association between the inflammatory biomarkers and regional fat distribution, in which greater truncal fat mass correlated with higher concentrations of IL-6 and CRP. It is interesting that the data in the present study showed a negative correlation between HOMA-IR and the concentrations of individual cytokines. The cause of this counterintuitive relation is not clear, and that lack of clarity calls for further studies examining the mechanisms underlying these observations.
There are several limitations to our study, in particular the relatively small size of the study population. In addition, the cross-sectional nature of this study, although showing an association between BMI and IR, does not provide information regarding causal relations. Moreover, the control subjects were significantly younger than the CKD patients, which may have accounted for the differences noted. However, subsequent analysis was adjusted for age and sex. Rather than direct measurement, the GFR in both groups was estimated by the use of the abbreviated equation described in the MDRD study (7). This equation has been found to be an adequate predictor of GFR when 24-h creatinine clearance or inulin clearance is not available (31). Ideally, the use of the hyperinsulinemic euglycemic clamp would have provided the best measure of IR in this study, but its use is laborious and time-consuming for a study of this size. Shoji et al (9) showed that HOMA-IR scores correlate well with the hyperinsulinemic euglycemic clamp as a measure of IR in individuals with a wide range of GFRs. Whereas we provide intriguing data regarding body composition and IR, the %BF measured in the CKD patients and in the control subjects in the current study did not differentiate between truncal and nontruncal fat, which may have resulted in an underestimation of the relation of %BF, adiponectin concentrations, and IR. Finally, limiting our study population to persons with stage 3 or 4 CKD limited the extrapolation of our findings to other stages of kidney disease and hindered the detection of a correlation between GFR and IR over a wider range of kidney functions.
In summary, our data show that BMI measures, particularly %BF, are the major determinant of IR in nondiabetic stage 3–4 CKD patients. Whereas the IR of uremia may be seen in the population of ESRD patients undergoing dialysis, who experience greater metabolic stress (3), body composition likely plays a more significant role in the development of IR in patients with less severe renal disease. Prospective studies are needed to more clearly define this relation and to determine whether interventions targeting IR in this patient population can decrease cardiovascular morbidity and mortality, as well as progression to ESRD.
ACKNOWLEDGMENTS
The authors thank Karen Majchrzak, Cindy Booker, Andrew Vincz, and the Vanderbilt General Clinical Research Center nursing staff and Jane Kane at Maine Medical Center for their excellent technical assistance.
The authors’ responsibilities were as follows—TAI and JH: contributed equally to designing the experiment, collecting and analyzing the data, and writing the manuscript; MLT: analyzed the data and wrote the manuscript; and AS: performed data analyses. None of the authors had a personal or financial conflict of interest.
REFERENCES