Perirenal Fat Thickness Is Significantly Associated With the Risk for Development of Chronic Kidney Disease in Patients With Diabetes

The burden of chronic kidney disease (CKD) in patients with diabetes is high and rising. CKD not only is the major cause of end-stage kidney disease but also is associated with higher cardiovascular risk and all-cause mortality in people with diabetes (1,2). Obesity is an important risk factor for the development and progression of CKD in patients with diabetes (3–5). BMI, a traditional indicator of general obesity, was reported to be related to impaired kidney function in type 2 and type 1 diabetes (6,7). However, abdominal obesity-related parameters, such as waist circumference (WC) and visceral adipose tissue (VAT), have been shown to be superior to BMI for predicting the development and progression of CKD in patients with diabetes (8–13).

VAT refers to the fat around the abdominal organs, which is mainly distributed in the omentum, mesentery, and retroperitoneum. Perirenal fat is a fat pad adjacent to the kidneys and is a type of visceral fat (14). In recent studies it was found that perirenal fat has origins in both adipocyte precursor cells and mature adipocyte different from origins of other components of visceral fat and that the developmental heterogeneity may determine the function heterogeneity (15,16). However, it is not clear whether the anatomical position and developmental heterogeneity of perirenal fat make it more sensitive to development of CKD than are other fat depots in patients with type 2 diabetes mellitus (T2DM). A previous study by Lamacchia et al. (17) showed that after adjustment for traditional metabolic factors, among 151 T2DM patients, an elevated thickness of perirenal fat (measured by ultrasound) was associated with decreased estimated glomerular filtration rate (eGFR). This result was consistent with that of Fang et al. (18), who used a similar design. These cross-sectional studies preliminarily suggested the correlation between perirenal fat and impaired kidney function. For further investigation of whether the perirenal fat is an independent risk factor for CKD incidence and for its comparison with total, subcutaneous, or visceral fat, a longitudinal study was needed.

In the current study of T2DM patients, both cross-sectional and longitudinal analyses were performed. To better evaluate the fat tissue parameters, we measured perirenal fat as perirenal fat thickness (PRFT) by computerized tomography (CT) and total body fat (TBF), and subcutaneous adipose tissue (SAT) and VAT were assessed by DEXA. The predictive value of perirenal fat for CKD in T2DM patients was compared with that of total, subcutaneous or visceral fat.

The study was conducted with subjects enrolled in the Chongqing Diabetes Registry (CDR) (clinical trial reg. no. NCT03692884, ClinicalTrials.gov) from January 2014 to July 2018. T2DM was diagnosed based on the 1999 World Health Organization diagnostic criteria for T2DM (19). Patients with T2DM who had complete abdominal CT or CT urography imaging data were enrolled in the study. The exclusion criteria were as follows: 1) patients with a diagnosis of acute kidney injury; 2) patients with other CKDs, including renal artery stenosis, ischemic nephropathy, hypertensive nephropathy, chronic pyelonephritis, kidney stones with hydronephrosis, kidney atrophy, and kidney loss; 3) patients with perirenal infection, large renal cysts, and other conditions that affect the measurement of perirenal fat; and 4) patients with a history of infection, immune dysfunction, or malignancy. Based on the inclusion and exclusion criteria, a total of 383 patients were included in the cross-sectional study.

Subjects without CKD (eGFR >60 mL/min/1.73 m²) at baseline were asked to participate in the follow-up; 119 patients refused, and 9 patients were lost to follow-up. A total of 190 patients were included in the longitudinal study, with a median follow-up time of 24.5 months (interquartile range [IQR] 13.7, 42.0). The main outcome was the development of CKD, defined as eGFR<60 mL/min/1.73 m². Informed consent was obtained from all participants, and the Ethics Committee of The First Affiliated Hospital of Chongqing Medical University approved the study (no. 2018-042). The research flowchart is shown in Fig. 1.

The medical history of all patients were reviewed. Anthropometric measurements, including height, weight, WC, systolic blood pressure (SBP), and diastolic blood pressure (DBP) were obtained from all participants. BMI was calculated as weight in kilograms divided by the square of height in meters. Body composition was measured with a DEXA scanner (Hologic Discovery QDR Series; Hologic, Bedford, MA) by a trained technician. All standard procedures were performed as previously described for previous studies (20,21): each subject was reclined in a supine position and was scanned from head to foot in the standard mode. The scanning range width was fixed at 60 cm, and the scanning time was ∼20 min. The Hologic Whole Body DXA Reference Database Software and Hologic Visceral Fat Software were used to estimate the TBF percentage and SAT and VAT volume.

Abdominal CT data sets were acquired using multidetector CT scanners (Discovery CT750 HD; GE Healthcare, Milwaukee, WI, and SOMATOM Definition Flash, Siemens Healthcare, Erlangen, Germany), from the diaphragm to L4. The CT protocol was applied as follows: field of view 350 mm, matrix 512 × 512, slice collimation 128, gantry rotation time 0.5–0.6 s, pitch 1.0, scan time 0.4 s, tube voltage 120 kV, and tube current Auto mA.

The PRFT was measured by a trained radiologist who was blinded to the clinical information. PRFT was measured in the central slice of the renal hilum (defined as the central slice of pelvis) from the following three directions: 1): anterior, the vertical distance from top of anterior border of kidney to the anterior renal fascia or the closest visceral organ; 2) lateral, the vertical distance from top of lateral border of kidney to the lateral perirenal fascia or the closest visceral organ; and 3) posterior, the vertical distance from the top of posterior border of the kidney to the posterior renal fascia (Fig. 2). If the top border of the kidney was stuck closely to perirenal fascia or the closest visceral organ in any direction, the distance was recorded as zero. Pixel density in Hounsfield units (HU) was used to identify adipose tissue based on a window width of −195 to −45 HU, centered on −120 HU.

The correlation between PRFT value on the left and that on the right was 0.783 (P < 0.001). PRFT was measured two times. The intraclass correlation efficient was 0.967 (95% CI 0.956, 0.975; P < 0.001) on the left side and 0.966 (0.956, 0.975; P < 0.001) for the right side.

Glycosylated hemoglobin (HbA_1c) was measured by a high-performance liquid chromatography analyzer. Fasting plasma glucose (FPG), total cholesterol (TC), LDL cholesterol (LDL-C), high-density lipoprotein cholesterol (HDL-C), triglycerides (TGs), and hs-CRP were measured enzymatically by an automatic analyzer (Model 7080; Hitachi, Tokyo, Japan) with reagents purchased from Leadman Biochemistry Co. Ltd. (Beijing, China). Non-HDL-C was simply calculated by subtraction of HDL-C from TC, and the ratios of TC to HDL-C and LDL-C to HDL-C were calculated (22). We defined metabolic syndrome (MetS) by using the International Diabetes Foundation (IDF) criteria of 2005 (23). The IDF definition of MetS includes central obesity (WC ≥90 cm in Chinese men and ≥80 cm in Chinese women) plus any two of the following four factors: 1) high blood pressure, SBP ≥130 mmHg, DBP ≥85 mmHg, or known treatment for hypertension; 2) hypertriglyceridemia, fasting plasma TG ≥1.7 mmol/L; 3) low HDL-C, fasting HDL-C <1.0 mmol/L in men and <1.3 mmol/L in women; and 4) hyperglycemia, fasting glucose level of ≥5.6 mmol/L or known treatment for diabetes. Serum creatinine, uric acid, urinary creatinine, and albumin were measured with an automatic biochemical analyzer (Modular DDP; Roche). The urinary albumin–to–creatinine ratio (UACR) was calculated. The eGFR was calculated using the MDRD study equation (for males, eGFR = 186 × SCr^−1.154 × year^−0.203; females, eGFR = 186 × SCr^−1.154 × year^−0.203 × 0.724) (24). The outcome of the longitudinal study was the occurrence of CKD, defined as eGFR <60 mL/min/1.73 m².

Data are expressed as the mean ± SD when the sample distribution was approximately normal. If the data were not normally distributed, we use the median (IQR). Data for categorical variables are reported as frequencies (proportions), and χ² tests were used for group comparisons. Continuous variables between two groups were compared using Student t test or Mann-Whitney U test. ANOVA or the Kruskal-Wallis test was used to determine the differences among three groups. The intraclass correlation coefficients for agreement were used to test for intraobserver reproducibility of the PRFT measurements on both sides. An intraclass correlation coefficient close to 1 indicates a very good correlation.

Multiple linear regression was conducted to test the correlations between perirenal fat and eGFR in the cross-sectional analysis. In the longitudinal analysis, univariate and multivariate Cox proportional hazards regression was conducted to determine whether parameters of TBF, SAT, VAT, and PRFT could be predictive factors of CKD incidence. The hazard ratios (HRs) and 95% CIs were calculated based on z-transformed TBF (per-SD change in TBF), SAT (per-SD change in SAT), VAT (per-SD change in VAT), and PRFT (per-SD change in PRFT). Predictive ability was assessed using Harrell’s C-statistic. Analyses were performed with SPSS, version 22.0 (IBM, Armonk, NY) and R 4.0.3 (R Foundation for Statistical Computing, Vienna, Austria). Cox proportional hazards regression analysis was conducted using Surv function in the survival package, and proportional hazards assumption was tested on the basis of Schoenfeld residuals using the cox.zph function (25). The results of the Schoenfeld residual test showed that none of the covariates were significant based on a P value threshold of 0.05. The ggsurvplot function in the survminer package was used to plot the cumulative hazard function. R package survcomp were used to calculate the C-statistics (26). P values of <0.05 were considered statistically significant.

The resources generated during and/or analyzed during the current study are available from the corresponding authors on reasonable request.

A total of 383 participants were included in the cross-sectional analysis. The clinical characteristics of the participants are shown in Table 1. In comparisons with patients with lower PRFT values, patients with higher PRFT values were more likely to be male; to have hypertension or MetS; to use insulin combined with oral hypoglycemic medication, antihypertensive drugs, or ACE inhibitors/angiotensin receptor antagonists (ACEI/ARBs); to have higher levels of SBP, DBP, TG, and uric acid; and to have lower levels of HbA_1c. Levels of BMI, WC, TBF, SAT, and VAT were elevated in subjects with higher PRFT, while eGFR was significantly decreased. There was no difference in UACR between subjects in different tertiles.

Table 2 shows the differences in adipose parameters (BMI, WC, TBF, SAT, VAT, and PRFT) between the CKD group (eGFR <60 mL/min/1.73 m²) and the non-CKD group (eGFR ≥60 mL/min/1.73 m²). The levels of BMI, WC, TBF, SAT, and VAT in the CKD group tended to be higher than those in the non-CKD group without statistical significance. In comparison with the non-CKD group, the left, right, and bilateral PRFT of the CKD group were significantly increased. Among the three thicknesses of perirenal fat at the anterior, lateral, and posterior positions, lateral PRFT increased significantly. Multiple linear regression showed that after adjustment for age, sex, diabetes duration, hypertension history, insulin therapy, antihypertensive drugs, and lipid-lowering drugs, eGFR was not significantly associated with TBF (β = −0.032, P = 0.599), SAT (β = −0.018, P = 0.751), or VAT (β = −0.081, P = 0.099). The PRFT was significantly correlated with eGFR (β = −0.214, P < 0.001) in the above multivariate-adjusted model, even after adjustment for TBF, SAT, and VAT (β = −0.270, P < 0.001) (Table 3).

Table 4 showed the baseline characteristics of all patients in the longitudinal study. Of the 190 patients, 29 participants developed CKD after a median follow-up of 24.5 months (IQR 13.7, 42.0). Univariate Cox proportional hazards regression analysis showed that age, SBP, eGFR, insulin therapy, antihypertensive drugs, PRFT, and VAT were associated with increased risk of CKD (Table 4). The covariates with a P value <0.1 in univariate regression analysis were selected for inclusion in the multivariate Cox regression model. After adjustment for age, SBP, HbA_1c, and eGFR at baseline, per-SD increment in PRFT was associated with a higher incidence of CKD (HR 1.64, 95% CI 1.14–2.35; P = 0.007). After further adjustment for TPF, SAT, or VAT, the risk of CKD in individuals with higher PRFT was still elevated (1.63, 1.14–2.35, P = 0.008 for TPF-based adjustment; 1.63, 1.13–2.36, P = 0.009 for SAT-based adjustment; and 1.67, 1.04–2.68, P = 0.036 for VAT-based adjustment). TBF, SAT, and VAT did not increase the risk of CKD after PRFT-based multivariate adjustment (Table 5 and Fig. 3).

For the longitudinal study, we computed Harrell’s C-statistic to evaluate and compare the predictive power of the Cox regression models. The results showed that PRFT predicted CKD with a C-statistic (95% CI) of 0.668 (0.562, 0.774), which was higher than that of TPF [0.535 (0.433, 0.637)], SAT [0.526 (0.434, 0.618)], or VAT [0.602 (0.506, 0.698)].

As far as we know, this was the first longitudinal study to evaluate whether the perirenal fat was a risk factor for CKD in T2DM. After adjustment for confounders, our study demonstrated that perirenal fat could independently predict CKD incidence in patients with T2DM. More importantly, perirenal fat had a higher predictive value for CKD than total, subcutaneous, or visceral fat in T2DM patients.

The relationship between perirenal fat and CKD was previously studied in two studies with relatively smaller samples. Lamacchia et al. (17) measured perirenal and pararenal fat thickness by ultrasound in 151 patients with T2DM and found that higher levels of perirenal and pararenal fat thickness were significantly correlated with lower eGFR after adjustment for traditional risk factors, including BMI and WC. On the contrary, BMI and WC showed no correlation with eGFR after adjustment for perirenal and pararenal fat thickness. Another study, by Fang et al. (18), provided a cross-sectional analysis involving 171 patients with T2DM. PRFT was measured by ultrasound. The results showed a negative relationship between perirenal fat thickness and eGFR. These studies suggested a possible role of perirenal fat in kidney dysfunction in T2DM patients. However, both were cross-sectional studies and adjustments were not made for visceral fat. Previously in studies, visceral fat was found to be closely related to CKD in T2DM (9–13). Therefore, it is crucial to adjust for visceral fat to demonstrate the independent relationship between perirenal fat and CKD. In our cross-sectional analysis, the confounding factors including visceral fat were adjusted for, and the results suggested a strong and independent correlation between perirenal fat and eGFR. Furthermore, the results from longitudinal analysis showed that per-SD increment at baseline PRFT was associated with a higher incidence of CKD after VAT-based adjustment, suggesting that perirenal fat was an independent predictor for the development of CKD.

It is widely known that fat distribution is associated with diabetes and related complications. Previous studies referring to obesity and CKD in diabetes mainly focused on TBF and visceral fat. Several studies have found that higher BMI is associated with nephropathy in patients with type 1 diabetes and patients with T2DM (6,7,27,28). However, some studies have observed that after adjustment for cardiovascular risk factors or VAT, the relationships of BMI, TBF, and fat mass index with diabetic kidney disease disappeared (13,29). Based on these data, it has been suggested that the relationship between obesity and diabetic kidney disease may depend on the distribution of adipose tissue rather than the overall content of adipose tissue. Recent studies showed that a central body fat distribution (mainly excess of visceral fat) was more closely associated with nephropathy in diabetes (9–13). The cross-sectional analysis showed that in our population with T2DM, TPF or SAT was not significantly associated with eGFR. There was a significant negative correlation for VAT with eGFR in univariate analysis. However, the correlation disappeared after adjustment for traditional risk factors, especially for PRFT. By contrast, the relationship between perirenal fat and eGFR remained the same, even after adjustment for VAT. It seems likely that there was a stronger correlation of perirenal fat with eGFR than of total, subcutaneous, or visceral fat. Furthermore, the longitudinal analysis showed that the PRFT could independently predict CKD incidence, while TPF, SAT, and VAT could not. More importantly, PRFT predicted CKD, with a higher C-statistic than TPF, SAT, and VAT, which suggested that perirenal fat was a better predictor for CKD development compared with total, subcutaneous, or visceral fat.

The mechanisms for the increased risk of CKD associated with perirenal fat are unclear. Perirenal fat is active in adipokine synthesis and secretion. Adipokines such as leptin (30,31), adiponectin (31,32), apelin (33), and visfatin (31) and inflammatory factors such as IL-6 (32) and TNF-α (34) are synthesized and secreted by perirenal fat. These adipokines and inflammatory factors not only influence insulin sensitivity and glucose and lipid metabolism but also directly affect renal hemodynamics and renal function (35,36). Ma et al. (34) found that perirenal fat from pigs with obesity-related metabolic derangements showed increased proinflammatory macrophage infiltration and TNF-α expression compared with the same parameters in lean pigs. Endothelial function was further reduced in lean arterial rings by incubation with perirenal fat harvested from pigs with obesity-related metabolic derangements. These effects were restored by inhibition of TNF-α, indicating that TNF-α derived from perirenal fat can directly affect the function of renal vessels. Li et al. (37) found that when glomerular endothelial cells (GECs) were cocultured with perirenal adipocytes, increased proliferation of GECs was observed, and leptin antagonists partially alleviated the proliferation of GECs. Adipokines and inflammatory factors may mediate the effect of perirenal fat on CKD development. At present, we don’t known whether the cytokines secreted by perirenal fat are different from those of other fat depots (especially the other components of VAT). Cell lineage tracing experiments revealed that the progenitors in perirenal fat tissue are different from those of mesenteric, gonadal, and pericardial fat tissues (16), and perirenal fat tissue has a developmental gene expression pattern different from that of other fat tissues (38). The developmental heterogeneity may explain the functional differences between adipocytes in these fat tissues.

In addition, like the effect of pericardial fat on the coronary artery (39,40), the anatomic position of perirenal fat may explain its local effects on kidney. The perirenal fat and the kidney are wrapped together by Gerota fascia (41). The distance between the perirenal fat and the kidney is small enough for the two organs to interact with each other. On one hand, the cytokines synthesized and released by perirenal fat may have effects on kidney cells through local cross talk (14). On the other hand, Gerota fascia can restrict the expansion of adipose tissue when perirenal fat is increased, which may produce physical pressure on renal blood vessels and parenchyma. Renal parenchyma compression may cause changes in renal hemodynamics, including an increase in renal interstitial hydrostatic pressure and a decrease in renal blood flow and tubular velocity (42,43). It has been reported that there is a positive correlation between perirenal fat and the renal resistance index (17), which suggests that perirenal fat may exert direct compression on renal blood vessels.

Our study has the following strengths. Firstly, previous studies that reported the correlation of CKD with perirenal fat were cross-sectional. Our study is the first to provide longitudinal analysis. Secondly, we added comparative evidence on the predictive value of total, subcutaneous, visceral, and perirenal fat for CKD in T2DM. The new information has important implications for the risk prediction of CKD in clinical practice. Thirdly, the evaluations of fat tissue parameters were relatively accurate. PRFT were measured in the central slice of the renal hilum by CT from three directions on both sides for each individual. A study by Favre et al. (44) showed that a single slice of PRFT measurement can reliably evaluate the content of perirenal fat. Our measurements of PRFT showed excellent intra-observer reproducibility, with intraclass correlation coefficients of 0.98 and 0.97 for the left side and the right side, respectively. Besides, the measurements of contents of total, subcutaneous, and visceral fat by DEXA are more accurate than anthropometric measurements, such as BMI and WC (45,46). Fourthly, the CT measurement process and clinical evaluation process were conducted independently, and use of the blinded method makes our results more convincing.

Some limitations of the study are worth mentioning. Firstly, our sample size is relatively small and the follow-up period is relatively short. Secondly, enrolled subjects were from a single center and of the same race; whether the effects of perirenal fat on CKD outcomes are applicable in different populations should be further verified. Thirdly, some possible confounders, such as dietary or lifestyle factors, were not adjusted for, as these data were not available in this population. Fourthly, this is an observational study and as such cannot clarify the causal relationship between PRFT and CKD.

In conclusion, our study demonstrated that perirenal fat was a better predictor for CKD than total, subcutaneous, or visceral fat in T2DM patients. Whether perirenal fat is a good predictor for end-stage renal disease and renal death needs to be investigated next. In addition, future research could focus on intervention studies and experimental studies, which might reveal the possible mechanism of perirenal fat in CKD.

Acknowledgments. The authors thank research nurses Fei Lian, Kun Liao, Xixiang Jiang, Jirong Deng, and Xiurong Liu as well as investigators Kanran Wang, Lina Yang, Liang Chen, and Zhixin Xu from the CDR group at The First Affiliated Hospital of Chongqing Medical University for excellent assistance. The authors thank The Chongqing Key Laboratory of Translational Medicine in Major Metabolic Diseases, The First Affiliated Hospital of Chongqing Medical University.

Funding. This work was supported by the National Natural Science Foundation of China (81700754, 81770851, 81870567, 81800731, 81970720, 81800757); National Key Research & Development Plan, Major Project of Precision Medicine Research (2017YFC0909600, sub-project: 2017YFC0909602, 2017YFC0909603); Bethune Merck Diabetes Research Foundation (G2018030); Technological Innovation and Application Development Project of Chongqing (cstc2019jscx-msxmX0207); Chongqing Science and Health Joint Medical Research Project (2020FYYX141); China International Medical Foundation (Z-2017-26-1902-2); High-end Medical Talents of Middle-aged and Young People in Chongqing (yuweiren[2015]49); Yong and Middle-aged Senior Medical Talents studio of Chongqing (ZQNYXGDRCGZS2021001); Chongqing Outstanding Youth Funds (cstc2019jcyjjq0006); and Outstanding Talents of The First Affiliated Hospital of Chongqing Medical University 2019 (2019-4-22 [to J.H.]).

Duality of Interest. No potential conflicts of interest relevant to this article were reported.

Author Contributions. Q.L. and Z.W. designed the study, oversaw the data collection, and interpreted the data. X.C. conducted the data collection and data analysis and wrote the manuscript. Y.M. conducted radiographic measurement and contributed to the writing of the manuscript. J.H. and L.G. contributed to the study design, provided statistical expertise, and contributed to the writing of the manuscript. S.H. provided statistical expertise and contributed to the writing of the manuscript. T.L., S.Y., and H.Q. contributed to the writing and editing of the manuscript. Y.W., Z.D., and M.M. assisted with the data analysis. L.Z., X.L, Y.T., Q.Z., and Y.Z. assisted with the data collection. J.C.H. edited the manuscript. Q.L. and Z.W. are the guarantors of this work and, as such, had full access to all the data in the study and take responsibility for the integrity of the data and the accuracy of the data analysis.