J Obes Metab Syndr 2023; 32(3): 236-246
Published online September 30, 2023 https://doi.org/10.7570/jomes23037
Copyright © Korean Society for the Study of Obesity.
1Faculty of Medicine, Pelita Harapan University, Tangerang; 2Department of Nephrology and Hypertension, Faculty of Medicine, Gadjah Mada University, Yogyakarta, Indonesia
Timotius Ivan Hariyanto
Faculty of Medicine, Pelita Harapan University, Boulevard Jendral Sudirman Street, Karawaci, Tangerang 15811, Indonesia
This is an Open Access article distributed under the terms of the Creative Commons Attribution Non-Commercial License (http://creativecommons.org/licenses/by-nc/4.0/) which permits unrestricted non-commercial use, distribution, and reproduction in any medium, provided the original work is properly cited.
Background: Chronic kidney disease (CKD) is a leading cause of death worldwide and has a high cost of treatment. Studies have indicated that a combination of waist circumference (WC) and triglyceride (TG) levels can be used to determine the risk of CKD. This study analyzes the risk of CKD using four phenotype models based on WC and TG.
Methods: This meta-analysis analyzes 113,019 participants from 13 studies. We conducted relevant literature searches in the Europe PMC, Medline, and Scopus databases using specific keywords. The results obtained were pooled into odds ratios (ORs) with 95% confidence intervals (CIs) using random-effects models.
Results: Our pooled analysis revealed that the highest significant independent association was between CKD and the high WC-high TG phenotype (adjusted OR, 1.61; 95% CI, 1.39 to 1.88; P<0.00001; I2=59%), followed by the high WC-normal TG phenotype (adjusted OR, 1.33; 95% CI, 1.12 to 1.57; P=0.001; I2=67%), and the normal WC-high TG phenotype (adjusted OR, 1.20; 95% CI, 1.06 to 1.37; P=0.005; I2=29%) when the normal WC-normal TG phenotype was taken as the reference.
Conclusion: Our study suggests that phenotype models based on WC and TG can be used as screening tools to predict the risk of CKD. Our results also indicate that WC plays a larger role than TG in the CKD risk. Further prospective studies are needed to confirm the results of our study.
Keywords: Triglycerides, Waist circumference, Renal failure, Metabolic syndrome, Obesity
Chronic kidney disease (CKD) is defined as damage to the kidneys in structure or function that lasts for 3 months or more.1 It can be classified based on etiological factors, the glomerular filtration rate, and the degree of albuminuria.1 Globally, the prevalence of CKD continues to increase, from only 11.8% from 1988 to 1994 to 14.2% from 2015 to 2016.2 As one of the leading causes of death worldwide, CKD caused around 1.3 million deaths in 2019.3 In addition, treatment for CKD is costly.4 A study from Italy showed that the annual cost of treating CKD patients who do not yet require dialysis is around 11,000 euros (EUR), and that amount increases 53,000 EUR for patients undergoing dialysis, indicating the importance of early prevention and detection.4
Age, hypertension, cardiovascular disease, diabetes, and a family history of CKD are widely recognized as the risk factors for CKD.5 Obesity is also an important risk factor associated with an increased incidence of CKD.5 Apart from being related to hypertension and diabetes, obesity directly affects the occurrence of CKD through adiposity of the kidneys, where adipose tissue produces adiponectin, leptin, and resistin, triggering inflammation, oxidative stress, abnormal fat metabolism, and activation of the renin-angiotensin-aldosterone system, which contribute to the development of kidney damage.6 Therefore, visceral adiposity is likely to be directly related to CKD.7 Typically, obesity is measured using the body mass index (BMI) alone because it is easy and practical to calculate.8 However, an examination using BMI alone cannot describe a body’s visceral fat.8 A bodybuilder who has a fairly high muscle mass can be classified as obese in a BMI examination, but a more comprehensive examination would show that she does not necessarily have high visceral fat.8 Waist circumference (WC), which is often used to define central obesity, is a better predictor of visceral body fat than BMI.9-11 Several observational studies have shown that WC can predict CKD better than BMI.9-11 A study from Korea has even shown that WC, but not BMI, is associated with decreased kidney function.12
Obesity and visceral adiposity are both closely related to triglyceride (TG) levels.13 Patients with CKD show a nearly linear increase in TG levels.14 Studies have shown that increasing TG and decreasing high-density lipoprotein cholesterol are independently associated with a higher-than-average risk of developing CKD stages 4–5.15 Therefore, the combination of high WC and TG levels, i.e., the hypertriglyceridemic waist phenotype, could be a useful screening tool for predicting CKD risk. Several studies have tested the relationship between the hypertriglyceridemic waist phenotype and the incidence of CKD, but they showed conflicting results.16,17 A cross-sectional study conducted in China in 2012 showed that the hypertriglyceridemic waist phenotype is associated with the occurrence of CKD.16 On the other hand, a prospective analysis by Ramezankhani et al.17 using data from 1999 to 2014 showed that the hypertriglyceridemic waist phenotype cannot be used to predict the development of CKD. Given those inconsistencies, a systematic review and meta-analysis could be helpful. A previous meta-analysis on this topic did not satisfactorily address all of the available models based on WC and TG: high WC-high TG (HW-HT), high WC-normal TG (HW-NT), and normal WC-high TG (NW-HT).18 Moreover, recent observational studies were not considered by the previous meta-analysis.18 Therefore, in this systematic review and meta-analysis, we summarize the latest evidence about the utility of the hypertriglyceridemic waist phenotype for predicting the risk of CKD.
This review was written based on guidelines from the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) statement.19 The literature we selected for inclusion in this review all met the following inclusion criteria: (1) population: non-hospitalized (community-based) adult individuals older than 18 years at the time of recruitment; (2) exposure: individuals with the HW-HT, HW-NT, or NW-HT phenotype; (3) control: individuals with the normal WC-normal TG (NW-NT) phenotype; (4) outcome: data on the presence of CKD among participants with different hypertriglyceridemic waist phenotypes; and (5) study design: cohort (either prospective or retrospective), case-control, or cross-sectional.
Literature was also excluded from this review if one or more of the following exclusion criteria were met: (1) pediatric patient population (<18 years old); (2) hospitalized participants with an advanced stage of CKD at the time of recruitment; (3) incomplete data; (4) unavailability of the full-text (abstract-only); and (5) review articles, case-series, and case-reports.
A systematic literature search was conducted on three databases (Europe PMC, Medline, and Scopus) by two independent authors from the date of database inception until February 25th, 2023. The search was limited to English-language literature. We used the following keyword combinations to elicit relevant literature: “(hypertriglyceridemic waist phenotype OR high waist circumference-high triglycerides OR enlarged waist high triglycerides OR HW OR HTW OR HTGW OR HTHW OR EWHT) AND (chronic kidney disease OR chronic renal insufficiency OR chronic kidney failure OR chronic renal failure OR chronic kidney insufficiency OR CKD OR CRF OR CKF OR reduced eGFR OR decreased eGFR OR albuminuria OR urine albumin excretion).” The initial step in identifying articles for inclusion in this review involved eliminating duplicates and screening the articles based on their titles/abstracts. If an article passed the initial screening process, it was evaluated in full-text form to assess its suitability according to the pre-formulated eligibility criteria. Any discrepancies in this review process were resolved through discussion with the third author. All article identification processes were carried out independently by two authors.
Two independent authors performed data extraction and tabulated the data in Microsoft Excel 2019. The following data were extracted: authors’ names, country, study design, number of samples, WC cut-off value, TG cut-off value, baseline characteristics of the study participants (age and sex), variables adjusted for the analysis, and the outcomes of interest in each group.
The outcome of interest in this study is the presence of CKD, defined according to criteria from the Kidney Disease Improving Global Outcomes guidelines, which stipulate the presence of any of the following: (1) estimated glomerular filtration rate (eGFR) <60 mL/min/1.73 m2 and (2) albuminuria, as evidenced by an albumin-to-creatinine ratio >30 mg/g or an albumin excretion rate >30 mg/day.1 The presence of CKD was compared across three phenotypes based on participants’ WC and TG: (1) HW-HT vs. NW-NT (primary analysis); (2) HW-NT vs. NW-NT (secondary analysis); and (3) NW-HT vs. NW-NT (secondary analysis).
Two independent authors performed a risk of bias assessment of the included cohort studies using the Newcastle-Ottawa Scale (NOS), which assesses three aspects of each study: (1) selection of an exposed and non-exposed cohort; (2) comparability between groups of participants; and (3) measurement of the outcome.20 We also used the modified NOS to assess the quality of cross-sectional studies according to the: (1) representativeness of the sample and ascertainment of exposure; (2) comparability between subjects in different outcome groups; and (3) assessment of the outcomes.21 The results of evaluations using these tools are given as numbers from 0 to 9, with studies with a total score ≥7 categorized as ‘good quality.’20,21
The presence of CKD in the primary analysis (HW-HT vs. NW-NT), which was a dichotomous variable outcome, was calculated into an odds ratio (OR) with a 95% confidence interval (CI) using the Mantel-Haenszel formula. The adjusted ORs for the primary analysis (HW-HT vs. NW-NT) and the secondary analyses (HW-NT vs. NW-NT and NW-HT vs. NW-NT) were calculated using the generic inverse-variance formula. Random-effects models were chosen in this review because we expected significant heterogeneity among studies due to differences in the study population characteristics. In this review, we used the I-squared (I2) statistic to assess heterogeneity between studies, with I2 values of ≤25%, 26%–50%, and >50% categorized as low, moderate, and high heterogeneity, respectively.22 Subgroup analyses of the primary analysis were also performed based on the study design, country of origin, mean age, and sex. Publication bias was analyzed using a funnel-plot whenever more than 10 studies were used for each outcome of interest. All statistical analyses were conducted using the Review Manager 5.4 application from the Cochrane Collaboration.
This is a systematic review and meta-analysis study. The Faculty of Medicine, Pelita Harapan University Research Ethics Committee has confirmed that no ethical approval is required.
The initial literature search process on the three databases yielded 148 studies. After eliminating duplicates and screening based on titles or abstracts, 128 studies were excluded, leaving 20 studies for full-text assessment. Of those 20 full-text articles, seven were excluded for the following reasons: six articles did not have data on the outcome of interest, and one article was not in the English language. Therefore, the remaining 13 articles,16-18,23-32 which together had 113,019 participants, were included in the analyses (Fig. 1). Ten of the 13 studies were conducted in China, one was in Brazil, one was in Iran, and one was in both China and Australia. Most of the included studies had cross-sectional designs. The sample sizes of the included studies ranged from 538 to 40,674 individuals. The cut-off values used for WC varied, but most studies used 90 cm in males and 80 cm in females to differentiate between HW and NW. The cut-off values for TG levels also varied, but most of the included studies categorized ≤1.70 mmol/L as a NT level and >1.70 mmol/L as a HT level. Further details about the baseline characteristics of the included studies are summarized in Table 1.
Our assessment of study quality using the NOS tool showed that both included cohort studies had good quality, with one study23 having a total score of 8 and the other study24 having a total score of 9. In addition, all the included cross-sectional studies were good quality, with total scores of 8 or 9 on the modified NOS tool. Summaries of the study quality assessment are presented in Tables 2 and 3.
In the unadjusted analysis model, based on our pooled analysis of 13 datasets from 12 studies (n=42,990), individuals with the HW-HT phenotype had a higher risk of developing CKD than those with the NW-NT phenotype (OR, 2.17; 95% CI, 1.86 to 2.52;
In the adjusted analysis model, our meta-analysis of 16 adjusted datasets from 12 studies showed that the HW-HT phenotype was independently associated with a higher risk of CKD than the NW-NT phenotype (adjusted OR, 1.61; 95% CI, 1.39 to 1.88;
Our pooled analysis of 10 adjusted datasets from eight studies showed that the HW-NT phenotype was independently associated with a higher risk of CKD than the NW-NT phenotype (adjusted OR, 1.33; 95% CI, 1.12 to 1.57;
Our meta-analysis of 10 adjusted datasets from eight studies also showed that the NW-HT phenotype was independently associated with a higher risk of CKD than the NW-NT phenotype (adjusted OR, 1.20; 95% CI, 1.06 to 1.37;
We performed subgroup analyses on the unadjusted HW-HT versus NW-NT model (primary analysis) based on the study design, country of origin, age, and sex. The subgroup analysis based on the study design revealed a higher OR for CKD in cross-sectional studies (OR, 2.27; 95% CI, 1.92 to 2.69;
The subgroup analysis based on the country of origin showed a higher OR for CKD in studies from China (OR, 2.34; 95% CI, 1.98 to 2.76;
The subgroup analysis based on the mean age of the participants showed that studies whose participants had a mean age <60 years had a higher OR for CKD (OR, 2.26; 95% CI, 1.83 to 2.80; P<0.00001; I2=79%, random-effects models) than studies whose participants had a mean age >60 years (OR, 2.02; 95% CI, 1.75 to 2.32;
Finally, the subgroup analysis based on sex distribution revealed a higher OR for CKD in studies in which the proportion of male participants was <40% (OR, 2.31; 95% CI, 1.74 to 3.07;
The publication bias analysis was performed for all adjusted analyses that compared the HW-HT, HW-NT, and NW-HT phenotypes with the NW-NT phenotype. The funnel-plot analysis revealed a relatively symmetrical inverted plot for all of the outcomes, indicating no publication bias (Supplementary Fig. 5).
The results of our systematic review and meta-analysis have shown an independent relationship between abnormalities in both WC and TG levels and a higher risk of developing CKD than is seen in individuals with a NW and TG. Among the WC- and TG-based phenotype models, the OR for CKD was highest in individuals with the HW-HT phenotype, followed by those with the HW-NT phenotype and those with the NW-HT phenotype, compared with the NW-NT phenotype. That could indicate that WC plays a larger role than TG levels in the risk of CKD. The subgroup analyses for the HW-HT versus NW-NT comparison found higher ORs for CKD in studies with a cross-sectional design, performed in China, and with a mean age of participants <60 years and a male proportion <40%.
The results of our meta-analysis are in line with those of the previous meta-analysis by Chen et al.,18 which also showed a significant relationship between the HW-HT phenotype and CKD risk. However, our current study and the previous meta-analysis by Chen et al.18 differ in some important ways.23
First, the previous meta-analysis by Chen et al.18 involved only 14 datasets from 11 studies when analyzing the relationship between the HW-HT phenotype and CKD risk, and one33 of those 11 studies was an abstract-only study. Including data from abstract-only studies in a meta-analysis is not recommended unless evidence is scarce.34,35 Data from an abstract usually reflect preliminary results published for a congress or competition, and they generally include very limited information.34,35 Those limitations prevent the researchers performing the meta-analysis from assessing the quality or risk of bias of that study as a whole and thus from determining whether the study is worthy of being included in the meta-analysis.34,35 We could not find the abstract-only study by Zhang et al.33 in Medline, Scopus, or even CrossRef. Furthermore, the doi link for it included in the reference section of Chen et al.18 cannot be opened, which calls the validity of the data derived from it into question. On the other hand, our study analyzed 17 datasets from 13 full-text studies that present complete data, which allowed us to assess their risk of bias. All the studies we included in our analysis have good quality, ensuring that our evidence is valid and reliable.
Second, the previous meta-analysis by Chen et al.18 performed only one type of analysis: pooled adjusted results for the relationship between the HW-HT phenotype and CKD risk compared with the NW-NT phenotype. However, four phenotype models can be derived from WC and TG measurements: HW-HT, HW-NT, NW-HT, and NW-NT. The HW-NT and NW-HT phenotype models were not considered in the previous meta-analysis.33 Our current study conducted three analyses using all four phenotype models to provide a more complete picture of the relationship between them and CKD risk. Our analyses show that among the four phenotype models, the greatest risk of developing CKD is found among individuals with the HW-HT phenotype, followed by the HW-NT phenotype, and then the NW-HT phenotype, indicating that WC plays a larger role in CKD risk than the TG level.
Excessive WC reduces the mobilization and utilization of free fatty acids and increases the volume of blood lipids.36 Evidence from previous studies suggests that visceral adiposity is closely related to insulin resistance, hypertension, and dyslipidemia (high triglycerides), which are all well-established risk factors for CKD.36 Cytokines seem to be a major regulator of adipose tissue metabolism, and cytokines inside adipose tissue arise from adipocytes, preadipocytes, and other cell types.36-38 mRNA expression studies show that adipocytes can produce tumor necrosis factor-α and several interleukins (ILs), particularly IL-1β and IL-6.36-38 The production of those cytokines produces oxidative stress and upregulates inflammation.36,38 It also induces abnormal fat metabolism and activate the renin-angiotensin-aldosterone system, which can contribute to the development of kidney damage.36-38 Because of the close relationship between high lipid content, excessive WC, and CKD pathophysiology, HT and HW can be used as screening tools to predict the risk of CKD.36
Despite its advantages, our study also has some limitations. First, our analyses are based mostly on cross-sectional studies, so direct causal relationships between the phenotypes and the risk of CKD cannot be ascertained. Second, most of the studies included in our analyses came from China, so we cannot be sure whether the results of this study are applicable to populations outside China, especially to non-Asian populations. Third, other factors that could affect the risk of CKD, such as dietary intake and a history of nephrotoxic drug use, were not considered in the included studies, so we could not adjust for them. Data from large prospective cohort studies are thus needed to confirm the results of our study.
Our results suggest that abnormalities in both WC and TG values are significantly associated with the risk of CKD. Therefore those values could be used as a screening tool to predict an individual’s CKD risk. The risk of developing CKD is highest in individuals who have HW and HT, followed by those with HW and NT, and then by those with a NW and HT, all compared with those who have both a NW and NT. Our results thus imply that WC plays a larger role than TG levels in the risk of CKD. Large prospective studies are still needed to confirm a direct causal relationship between both WC and TG levels and the risk of CKD. Studies conducted in non-Asian populations are also needed to determine whether our study results apply to those populations.
The authors declare no conflict of interest.
Study concept and design: KVJ, TIH, and MSM; acquisition of data: KVJ and TIH; analysis and interpretation of data: TIH and MSM; drafting of the manuscript: KVJ, TIH, and MSM; critical revision of the manuscript: KVJ, TIH, and MSM; statistical analysis: KVJ, TIH, and MSM; obtained funding: MSM; administrative, technical, or material support: KVJ and TIH; and study supervision: MSM.
Characteristics of the included studies
|Author (year)||Country||Design||Sample size||WC cut-off (cm)||TG cut-off (mmol/L)||Age (yr)||Male (%)||Variables adjusted on analysis|
|Borges et al. (2021)3||Brazil||Cross-sectional||788||102||88||1.7||1.7||61.6 ±11.1||37.3||-|
|Chen et al. (2022)18||China||Prospective cohort||7,406||90||85||1.7||1.7||59.0 ±8.9||45.3||Baseline age; sex; residence; education level; smoking and drinking status; BMI; HDL-C; history of DM, hypertension, and CVD; use of hypoglycemic agents, anti-hypertensive agents, and lipid-regulating agents; baseline eGFR levels; and physical activity|
|Chen et al. (2021)24||China||Cross-sectional+ retrospective cohort||2,205||90||85||2.0||1.5||66.8 ±5.6||45.5||Baseline age, sex, smoking status, alcohol intake, high-salt diet, physical activity, hypertension, and diabetes|
|Huang et al. (2015)25||China||Cross-sectional||1,828||90||80||1.7||1.7||52.6 ±14.5||37.3||Baseline age, history of coronary heart disease, history of stroke, history of malignancy, current smoker, current alcohol use, physical activity, educational status, hypertension, and diabetes|
|Li et al. (2014)16||China||Cross-sectional||1,534||90||85||2.0||2.0||57.2 ±10.9||37||Baseline age, history of coronary heart disease, history of stroke, history of malignancy, current smoker, current alcohol use, physical activity, educational status, hypertension, and diabetes|
|Ma et al. (2017)26||China||Cross-sectional||538||90||85||1.7||1.7||55.4 ±13.5||53.3||Baseline age, sex, BMI, hypertension, history of diabetes, and HbA1c|
|Qiu et al. (2020)27||China||Cross-sectional||31,296||90||80||1.7||1.7||55.6 ±11.3||40.5||Baseline age, sex, marital status, educational status, working status, smoking status, alcohol consumption, physical activity, BMI, hypertension, LDL-C, HDL-C, TC, and T2DM|
|Ramezankhani et al. (2017)17||Iran||Cross-sectional||12,012||90||85||2.0||2.0||42.3 ±14.8||43.8||Baseline age, smoking status, educational level, marital status, family history of diabetes, BMI, TC, hypertension, and diabetes|
|Su et al. (2020)28||China||Cross-sectional||40,674||90||85||2.0||1.5||58.1 ±9.2||29.7||Baseline age, sex, education status, smoking habits, drinking habits, CVD status, diabetes history, hypertension history, use of diabetes or hypertension drugs, BMI, eGFR, HDL-C, LDL-C, TC, AST, ALT, FBG, PBG, SBP, and DBP|
|Xuan et al. (2022)29||China||Cross-sectional||4,254||90||80||1.7||1.7||66.6 ±8.9||46.6||Baseline age, sex, BMI, current smoking, current drinking, TC, LDL-C, HbA1c, UA, FBG, SBP, DBP, use of anti-diabetes agents, hypertension, and use of anti-hypertension agents|
|Yu et al. (2018)30||Australia China||Cross-sectional||7,210||90||85||2.0||1.5||52.0 ±11.2||38.7||Baseline age and BMI|
|Zeng et al. (2016)31||China||Cross-sectional||2,102||90||80||1.7||1.7||71.6 ±6.5||40.3||Baseline age, sex, education, marital status, physical exercise, smoking, drinking, family history of CVD, hypertension, and diabetes|
|Zhou et al. (2018)32||China||Cross-sectional||1,172||90||85||2.0||2.0||51.7 ±15.6||33.3||Baseline age, sex, history of stroke, history of coronary heart disease, current smoker, current alcohol use, physical inactivity, educational attainment, CRP, UA, BMI, hypertension, and diabetes|
WC, waist circumference; TG, triglyceride; BMI, body mass index; HDL-C, high-density lipoprotein cholesterol; DM, diabetes mellitus; CVD, cardiovascular disease; eGFR, estimated glomerular filtration rate; HbA1c, glycosylated hemoglobin; LDL-C, low-density lipoprotein cholesterol; TC, total cholesterol; T2DM, type 2 diabetes mellitus; AST, aspartate transaminase; ALT, alanine transaminase; FBG, fasting blood glucose; PBG, postprandial blood glucose; SBP, systolic blood pressure; DBP, diastolic blood pressure; UA, uric acid; CRP, C-reactive protein.
Newcastle-Ottawa quality assessment of observational studies
|Author (year)||Study design||Selection*||Comparability†||Outcome‡||Total score||Result|
|Chen et al. (2022)18||Cohort||***||**||***||8||Good|
|Chen et al. (2021)24||Cohort||****||**||***||9||Good|
*(1) Representativeness of the exposed cohort; (2) selection of the non-exposed cohort; (3) ascertainment of exposure; and (4) demonstration that outcome of interest was not present at start of study; †(1) Comparability of cohorts on the basis of design or analysis, (maximum two stars); ‡(1) Assessment of outcome; (2) follow-up was long enough for outcomes to occur; and (3) adequacy of follow-up.
Modified Newcastle-Ottawa quality assessment of cross-sectional studies
|Author (year)||Study design||Selection*||Comparability†||Outcome‡||Total score||Result|
|Borges et al. (2021)23||Cross-sectional||***||**||***||8||Good|
|Chen et al. (2021)24||Cross-sectional||****||**||***||9||Good|
|Huang et al. (2015)25||Cross-sectional||****||**||***||9||Good|
|Li et al. (2014)16||Cross-sectional||****||**||***||9||Good|
|Ma et al. (2017)26||Cross-sectional||***||**||***||8||Good|
|Qiu et al. (2020)27||Cross-sectional||****||**||***||9||Good|
|Ramezankhani et al. (2017)17||Cross-sectional||****||**||***||9||Good|
|Su et al. (2020)28||Cross-sectional||***||**||***||8||Good|
|Xuan et al. (2022)29||Cross-sectional||***||**||***||8||Good|
|Yu et al. (2018)30||Cross-sectional||***||**||***||8||Good|
|Zeng et al. (2016)31||Cross-sectional||****||**||***||9||Good|
|Zhou et al. (2018)32||Cross-sectional||****||**||***||9||Good|
*(1) Representativeness of the sample, (2) sample size, (3) non-respondents, and (4) ascertainment of exposure (maximum four stars); †(1) Comparability of subjects in different outcome groups (maximum two stars); ‡(1) Assessment of outcome and (2) statistical test (maximum three stars).