Diabetic kidney disease (DKD), a serious microvascular complication of diabetes, is defined as decreased renal function (glomerular filtration rate (GFR)) with persistent clinically detectable proteinuria (albuminuria) [1
]; and occurs in approximately 25–40% of patients with diabetes [2
]. Given the dual problems of a significant risk of progression from DKD to end-stage renal disease (ESRD) [3
], increased concomitant cardiovascular disease [4
], and mortality [5
], it is important to identify patients at risk of DKD, understand the underlying pathogenic pathways, and initiate renal and cardiovascular therapies based on the knowledge of these causal mechanisms.
Obesity is rapidly reaching epidemic proportions globally and Asia is no exception, particularly with the adoption of an increasingly westernized diet and sedentary lifestyle [6
]. Obesity is an established risk factor for diabetes and hypertension [7
], both linked with the development of DKD [9
]. According to the World Health Organization (WHO) [10
], there are two separate classifications of obesity namely ‘generalized’, defined as body mass index (BMI, calculated as weight in kilograms (kg) divided by height in meters (m) squared) of ≥30 kg/m2
; and ‘central/abdominal’, assessed using waist circumference (WC) and/or waist-to-hip/height ratio (WHR/WHtR). While there is emerging evidence suggesting that both forms [11
] contribute to the risk of DKD, independent of diabetes and/or hypertension, it is still unclear which one contributes more to the risk of DKD due to their close inter-relationship [18
Our group has previously shown a differential association of BMI and WHR with diabetic retinopathy (DR), a visual microvascular complication of diabetes [21
]. This lack of clarity is particularly problematic in patients with type 2 diabetes (T2DM) as up to half with impaired GFR have no overt albuminuria (non-albuminuric DKD) [22
], which adds an additional level of uncertainty to the role of these two measures of obesity in the pathogenesis of DKD. In fact, recent research has advocated that GFR may be a better indicator of DKD, as urinary albumin-creatinine ratio (UACR) levels do not necessarily reflect actual levels of renal function [1
In this study, we therefore examined the associations between generalized obesity (defined as BMI) and abdominal obesity (assessed using three different measures: WC, WHR and WHtR) with DKD (assessed using UACR and estimated GFR (eGFR)) in a well-characterized sample of Asian patients with T2DM. We also conducted a meta-analysis of studies evaluating the relationship of these two obesity indices with DKD in patients with T2DM to situate our findings in the broader context.
The mean age (SD) of the 405 patients included in analyses was 58 (7.5) years and 277 (68.4%) were male. The mean (SD) BMI was 26.5 (4.2) kg/m2
, 93.3 (10.6) cm for WC, 0.6 (0.1) for WHtR, and 0.9 (0.1) for WHR. Those with DKD comprised 203 (50.1%) of the sample and were likely to be older, male, have a higher total to HDL cholesterol ratio, higher UACR, longer duration of diabetes, higher SBP, greater WHR, and were more likely to be on insulin and have DR (all p
< 0.05; Table 1
In models adjusted for age and gender (Table 2
, Model 1), no associations were found between BMI analyzed continuously and presence of DKD. However, those categorized as overweight/obese were more likely to have DKD (OR: 1.69, 95% CI: 1.12, 2.55), compared to normal/underweight individuals and this association persisted after multivariable adjustment (Table 2
, Model 2). No associations were, however, found between any of the remaining abdominal obesity parameters (WC, WHR and WHtR) with DKD. As being underweight has been found to be associated with DKD in some studies [37
] we conducted additional Supplementary analyses excluding underweight individuals but found no change in the direction, nor significance of the above reported associations (data not shown). In addition, as iterated previously, we reran the analyses for WHR and WHtR using established abdominal obesity thresholds and did not find any change in the direction, nor significance, of the associations (Table S1
). Mutually adjusting for generalized and abdominal obesity exposures showed results similar to that presented in our main tables (Table S1
In models stratified by gender, we found no association of any of the generalized obesity parameters with DKD (Table 3
The meta-analysis synthesized data from 18 studies (including the present one) for a total of 19,755 participants (Table S2
]. For the effects of continuous BMI and obesity (dichotomized BMI) on DKD, we included data from five [11
] and four [12
] studies in the meta-analysis, respectively. We found that every 5 kg/m2
increase in BMI was on average associated with a 43% increase in the odds of DKD (OR: 1.40, 95% CI: 1.27, 1.61, I-squared: 0%), while obesity was associated with a 65% increase in the odds of renal disease (OR: 1.65, 95% CI: 1.15, 2.34, I-squared: 77.2%; Figure 2
A total of 4 and 6 studies were included in the meta-analysis of the effects of continuous waist circumference [39
] and abdominal obesity [43
] (dichotomized waist circumference), respectively. A 1 cm increase in waist circumference was on average associated with a 2% increase in the odds of renal disease (OR: 1.02, 95% CI: 1.01, 1.03, I-squared: 19.4%), while abdominal obesity was associated with an 80% increase in the odds of renal disease (OR: 1.80, 95% CI: 1.39, 2.34, I-squared: 59.1%; Figure 3
Data from six studies (three each) were included in the meta-analysis of the effects of continuous [15
] and categorized [16
] WHR/WHtR. While we found a significant association between increased waist-hip ratio and likelihood of DKD continuously (OR per 0.1-unit increase: 1.47, 95% CI 1.25, 1.74, I-squared: 28.2%), this association became attenuated when analyzed categorically (OR per category increase: 1.10, 95% CI 0.99, 1.23, I-squared: 52.8%; Figure 4
In our clinical study of Asian patients with T2DM, higher BMI was independently associated with greater likelihood of having DKD. WC, WHR and WHtR however, were not independently correlated with DKD presence. While we also found that BMI (i.e., generalised obesity) was associated with greater odds of DKD on our meta-analysis, other parameters of abdominal obesity namely WC, WHR and WHtR, were also associated with a higher likelihood of having DKD. Taken together, our results suggest that both generalized and abdominal obesity may play a role in the pathophysiology of DKD in T2DM, independent of their established roles as major risk factors of hypertension and diabetes, both of which in turn have been demonstrated to be associated with DKD [9
]. As such, public health interventions to reduce both forms of obesity in patients with diabetes may also help reduce the likelihood of developing DKD although longitudinal data are needed to support this claim.
Unlike previous research in T2DM patients showing an independent association between abdominal obesity and DKD, as evident in the overall pooled estimates from our meta-analyses, we found no significant relationship between WC, WHR or WHtR and the presence of DKD in our clinical study. Discrepancies between findings could be related to the presence of non-albuminuric DKD, which can make up approximately 50% of individuals with T2DM and DKD [22
]. Analyses stratified by classification of DKD (via eGFR or UACR alone) appear to support this theory: the effect sizes for the association of abdominal obesity markers with DKD appear to be larger for DKD categorized using UACR (Table S3
) alone versus DKD categorized using eGFR only (Table S4
). As such, larger, longitudinal studies are needed to validate our findings.
We demonstrated that being overweight/obese was associated with increased odds of having DKD in both our sample and meta-analysis of cross-sectional studies, despite the high degree of heterogeneity in the meta-analysis (I-squared value of 76.5%). This heterogeneity is likely due to the difference in the classification of obesity and DKD utilized in the various studies; for instance, Belhatem and colleagues defined obesity as BMI of 30 to <40 kg/m2
, while Low and associates categorized obesity as >25 kg/m2
. Our cross-sectional results are corroborated by data from large-scale prospective studies [48
]. For instance, the Hypertension Detection and Follow-Up Program, a cohort study comprising 5897 patients with hypertension and no kidney disease at baseline, found that the 5-year incidence of kidney disease was 20% higher in obese patients compared to those with normal BMI, even after adjustment for presence of T2DM [48
]. Unfortunately, we were unable to assess the relationship between BMI and the more severe stages of DKD as we had too few cases of severe DKD. This is important as a few studies have reported that higher BMI is associated with greater rates of survival in those with ESRD [49
]; a finding that has been attributed to the phenomenon known as the “obesity paradox” [51
], where persons with larger body mass are also likely to be healthier (greater muscle mass) with less comorbid conditions. Hence objective measures of body fat (e.g., using dual-energy X-ray absorptiometry or Magnetic Resonance Imaging) [52
] in order to enhance our understanding of the nature of the generalized obesity-DKD relationship are warranted.
As our participants were Asian, we also analysed the BMI-DKD relationship using the Asian cut-points defined by WHO in 2003 (Table S5
). We found that although the direction of the association remained similar, the significance became attenuated. This attenuation may indicate the existence of a threshold beyond which a higher BMI contributes significantly to the pathogenesis of DKD, and that this threshold may lie closer to the international than Asian classification of generalized obesity. Further longitudinal studies are warranted to verify our findings.
The mechanisms that underlie the relationship between obesity and DKD, independent of BP and diabetes, are still poorly understood. One hypothesis is that obesity-induced glomerular hyperfiltration, consequent to increased renal tubular sodium reabsorption, results in impairment of renal autoregulation, which then allows for any increase in systemic BP to be transmitted directly to the glomerulus, leading to subsequent renal insult [53
]. Excessive lipid deposition into the kidney as a result of obesity can also lead to accumulation of toxic metabolites derived from fatty acid metabolism, e.g., diacylglycerols, resulting in mitochondrial dysfunction, endoplasmic reticulum stress, apoptosis, and eventual renal dysfunction [54
Strengths of this study include a large clinical sample, a comprehensive and standardized clinical assessment protocol, as well as the use of meta-analysis to synthesize available data on the association between the relevant exposures (BMI, WC, WHR and WHtR) and outcome (DKD). Limitations include the cross-sectional nature of this study limiting causal inferences, as well as the low number of subjects with severe DKD, particularly women, making severity analyses non-viable. In addition, our analyses were conducted in a clinical population, which may affect the generalizability of results. Furthermore, UACR and eGFR were assessed using a single spot measurement, which could have led to non-differential misclassification of albuminuria and CKD status, resulting in over or under-reporting of the true prevalence of albuminuria and CKD in these subjects. Additionally, we were unable to verify if the DKD cases in our study were a consequence of diabetes, or of a non-diabetic nature as kidney biopsies were not performed on participants due to the potential risks associated with the procedure. Lastly, we assessed only the relationships between the anthropometric measures of obesity and DKD in our study sample. Future studies should be conducted using objective body fat measures to verify our findings.