Lipid accumulation product, visceral adiposity index and risk of chronic kidney disease

We used data from the REasons for Geographic and Racial Differences in Stroke (REGARDS) study, a population-based prospective cohort of community-dwelling individuals aged 45 years and older. The study was designed to evaluate the risk factors that contribute to disproportionate mortality linked with stroke in the Southeast United States and among African Americans. A total of 30,239 adults were recruited between January 2003 and June 2007. Trained staff collected baseline information during a preliminary phone interview followed by an in-home physical examination. The in-home visit included measurement of blood pressure, anthropometric measurements and collection of fasting (10–12 hour) blood and spot urine specimen. Details of the study have been published previously [27]. Between April 2013 and December 2016, participants were invited to undergo a second in-person visit during which the baseline procedures were repeated. Among the 16,150 participants who participated in the second assessment, 15,938 completed the second telephone-administered assessment and 14,449 had in-person assessments. All participants provided written informed consent and Institutional Review Boards of all participating institutions approved the study.

BMI was measured as weight (kg)/height (m²) and categorized as ≤25.1 (reference group), 25.2–28.3, 28.4–32.5, > 32.5. Waist circumference was measured mid-way between the lowest rib and the iliac crest using a tape measure with the participant standing and was categorized in cm as < 86.3 (reference group), 86.4–95.2, 95.3–105.4, and > 105.4. VAI was calculated as: [WC/39.68 + (1.88 × BMI)] × (TG /1.03) × (1.31/ HDL) for men, VAI = [WC/36.58 + (1.89 × BMI)] × (TG/0.81) × (1.52/HDL) for women [23]. LAP was calculated as: (WC-65) × TG for men, and (WC-58) × TG for women [22]. Serum lipids including total cholesterol (TC), HDL and fasting TG were measured by colorimetric reflectance spectrophotometry.

Coronary heart disease (CHD), stroke, and smoking status were obtained by self-report during the telephone interview. DM was defined as a fasting glucose ≥126 mg/dL, or a non-fasting glucose ≥200 mg/dL, or self-reported use of antidiabetic medications. Hypertension was defined as either self-reported use of antihypertensive medications or a systolic blood pressure ≥ 140 mmHg or a diastolic blood pressure ≥ 90 mmHg measured during the home examination at baseline visit. Serum creatinine was calibrated to an international isotope dilution mass spectroscopic (IDMS)-traceable standard and was measured by colorimetric reflectance spectrophotometry (Ortho Vitros Clinical Chemistry System 950IRC, Johnson & Johnson Clinical Diagnostics, www.​orthochemical.​com). Estimated glomerular filtration rate (eGFR) was calculated using the combined CKD-EPI creatinine and cystatin C equation [28]. Urine albumin and creatinine were measured in a random spot specimen by nephelometry (BN ProSpec Nephelometer, Dade Behring, Marburg, Germany) and Modular-P chemistry analyzer (Roche/Hitachi, Indianapolis, IN), respectively.

The primary outcome was 1) incident CKD, defined as eGFR < 60 ml/min per 1.73 m² and at least 25% decline in individuals with baseline eGFR > 60 ml/min per 1.73 m² between the baseline and second in-home visit. The secondary outcomes were: 1) progressive eGFR decline, defined as > 30% decrease in eGFR between the baseline and second in-home visit. Both these endpoints required participants to return and provide blood at the second visit, thus we also evaluated incident kidney failure defined as initiation of dialysis as ascertained by United States Renal Data System (USRDS) linkage up to June 2014. This endpoint relied on administrative linkage of baseline data to USRDS, such that all participants provided time at risk for the kidney failure endpoint.

For the incident CKD and progressive eGFR analyses, we were limited to participants who had a baseline and follow-up visits and provided blood samples for analysis (Fig. 1). After excluding participants with missing data on serum creatinine or cystatin C, BMI, WC, and fasting triglycerides and/or HDL measurements (n = 2430), who had an eGFR < 15 ml/min/1.73 m² or had kidney failure receiving maintenance dialysis (n = 163), or who had a waist circumference less than 66 cm (for men) and less than 59 cm (for women) in order to avoid having negative values for LAP and VAI (n = 40), a total of 11,538 participants were available for analyses of incident CKD cohort and 12,624 participants were available for analyses of progressive eGFR decline. All individuals who did not meet any exclusion criteria and were successfully linked to USRDS data were included in analyses of incident kidney failure (N = 27,550) (Fig. 1).

Univariate parameters were calculated as mean ± standard deviation (SD) or counts and proportions across quartiles of LAP and VAI. Logistic regression was used to determine the association of VAP and VAI with incident CKD and progressive eGFR decline. All anthropometric measures (LAP, VAI, BMI and WC) were modeled both as continuous variables and quartiles. We used a series of sequential models to adjust for confounders. Model 1 was adjusted for age, sex, and race. Model 2 included model 1 plus presence of comorbid conditions (CHD, stroke, hypertension, and DM), and smoking status. Finally, model 3 included model 2 plus baseline eGFR and urine albumin to creatinine ratio (UACR). We tested for interaction between the novel markers of obesity and sex and risk for incident CKD.

Cox proportional hazards models were used to determine the independent associations between LAP, VAI, BMI, and WC with risks of kidney failure using a similar modeling strategy as indicated above. Additional sensitivity analysis for the kidney failure outcome were performed stratifying on baseline eGFR categories (> 60, 45–59, and < 45 ml/min/1.73m²). Follow-up time was censored at development of kidney failure, death, or date of last follow-up phone contact, whichever occurred first. We also performed analyses treating death as a competing risk.

We used Wald test statistics to assess the relative importance of adiposity measures for the prediction of kidney failure.

We used logistic regression to evaluate the association between LAP, VAI, BMI and WC with risk of incident albuminuria using the sequential models previously mentioned.

Lastly, since triglycerides and HDL are components of VAI and LAP equations, we performed logistic regression to evaluate if the associations between VAI and LAP and the outcomes were driven by WC and BMI, or whether triglycerides and HDL were also associated with kidney outcomes.

The mean age was 65 years, 54% were women, 21% had DM, mean baseline eGFR was 82 ml/min/1.73m² and the median baseline UACR was 7 mg/g. Baseline participant characteristics stratified by LAP and VAI levels are presented in Tables 1 and 2 respectively. There were significantly fewer African Americans in the top quartiles of LAP and VAI compared to bottom quartiles. Higher quartiles of LAP and VAI had a greater prevalence of CHD, myocardial infarction, stroke, DM, and hypertension. In addition, higher LAP and VAI quartiles had lower eGFR and higher UACR.

Between the baseline and follow-up visits (median time 6.3 years), there were 1127 cases of incident CKD and 1452 cases of progressive eGFR decline. Each two-fold higher of VAI was associated with 31% higher odds (95% CI 1.24, 1.40) of developing incident CKD in the unadjusted model (Table 3). This association was attenuated but remained statistically significant after adjusting for all the covariates, including baseline eGFR and UACR. Similar associations were seen for LAP where each two-fold higher level associated with 21% higher odds (95% CI 1.13, 1.29) of CKD (Table 3). In categorical analyses, the top quartiles of VAI and LAP were associated with higher odds of incident CKD as compared to the lowest quartile (OR 1.26; 95% CI 1.02, 1.55 and OR 1.51; 95% CI 1.22, 1.87 respectively) in fully adjusted models. BMI and WC were also associated with increased risk of incident CKD in fully adjusted models when compared to the lower quartiles (OR 1.81; 95% CI 1.46, 2.25 and OR 1.64; 95% CI 1.33, 2.01 respectively) as shown in Table 3. We did not find an interaction between VAI or LAP with sex (0.62 and 0.59, respectively).

In the unadjusted model, each two-fold higher level of VAI was associated with 23% higher odds (95% CI 1.17, 1.31) of progressive eGFR decline and this association remained statistically significant after full multivariable adjustment (OR 1.11; 95% CI 1.04, 1.18, Table 4). Similar findings were observed on categorical analyses (Table 4) with the top quartile of VAI being associated with 20% higher odds of eGFR decline as compared to the lowest quartile. Each per two-fold higher and the top quartile of LAP were also associated with odds of progressive eGFR, 18% higher (95% CI 1.11, 1.26) and 36% (95% CI 1.13, 1.63) higher respectfully (Table 4). Each two-fold higher of BMI and WC were also associated with increased risk of progressive eGFR across the models (Table 4).

There were 353 cases of incident kidney failure over 10 years of follow up. Incidence rates of kidney failure were higher with ascending quartiles of VAI (0.11%/year in Q1 to 0.24%/year in Q4) and LAP (0.08%/year in Q1 to 0.29%/year in Q4). Each two-fold higher of VAI was associated with a 38% higher risk (95% CI 1.25, 1.53) of incident kidney failure in unadjusted models (Table 5). This finding remained statistically significant despite adjusting for age, sex, race, prevalent CHD, stroke, hypertension, diabetes mellitus, and smoking status (HR:1.33 and 95% CI 1.19, 1.49). However, addition of eGFR and UACR to the model attenuated these associations almost completely. Similarly, although, each two-fold higher of LAP was associated with greater risk of incident kidney failure in unadjusted, and demographic adjusted models, the inclusion of prevalent co-morbidities, eGFR and UACR rendered this association no longer statistically significant (Table 5) on the continuous per two-fold scale. Similar findings were also seen in categorical analyses where the top quartile of VAI (HR: 1.94; 95% CI 1.37, 2.76) and LAP (HR: 2.03; 95% CI 1.39, 2.97) were associated with risk of incident kidney failure only before but not after adjustment for eGFR and UACR. In comparison, BMI and WC were also associated with risk of kidney failure in an unadjusted and demographic adjusted models. However, further adjustment for eGFR and UACR resulted in a statistically significant inverse relationship between higher BMI and WC and risk of incident kidney failure (Table 5). Supplemental tables 1 and 2 shows the association of VAI and LAP, respectively, and the rest of the covariates with kidney failure before and after adding albuminuria and baseline eGFR to the analysis.

In sensitivity analysis stratifying by eGFR categories, we found statistically non-significant higher risk of kidney failure per two-fold VAI, LAP and kidney failure after multivariable adjustment in persons with eGFR < 45 ml/min/1.73m². The association of BMI and WC with kidney failure in these GFR categories was statistically non-significant, the point estimates were in the opposite direction from VAI and LAP. We found similar results when treating death as a competing risk (Supplemental Table 3). Among those with eGFR < 45 ml/min/1.73m², higher LAP, BMI and WC were all associated with lower risk of kidney failure after multivariable adjustment (Supplemental Table 4).

Wald test statistics only demonstrated a significant contribution by WC when compared to VAI, LAP, and BMI (0–10%) in the prediction of kidney failure (Supplemental Fig. 1).

Lastly, we did not find a significant association between HDL with any of the outcomes whereas we found weak associations between triglycerides and incident CKD and progressive eGFR decline (Supplemental Table 5).

Between the baseline and follow-up visits, there were 1397 cases of incident albuminuria. Each two-fold higher of VAI was associated with 21% higher odds (95% CI 1.14, 1.28) of developing incident albuminuria in the unadjusted model (Supplemental Table 6). This association was attenuated and no longer statistically significant after adjusting for all the covariates, including baseline eGFR and baseline UACR (OR 1.04; 95% CI 0.97, 1.11). Similarly, in the unadjusted model, a two-fold higher level of LAP was associated with 18% higher odds (95% CI 1.12, 1.24) of incident albuminuria, but this association was no longer significant after adjusting for all covariates (OR 0.98, 95% CI 0.93, 1.05) as shown in Supplemental Table 6.

Our study has limitations. First, there was a relatively short follow up time to adequately assess development of kidney failure. Second, although studies have shown that VAI and LAP are correlated with visceral adiposity our study lacks direct measures of visceral adiposity such as CT and MRI measures of adiposity. Strengths of this study include a large cohort of adults from different racial backgrounds with CKD and a longitudinal approach to evaluating the association of LAP and VAI with clinical endpoints.

In conclusion, VAI and LAP, newer visceral adiposity measures, are associated with incident CKD and eGFR decline after adjusting for multiple potential confounders including eGFR and albuminuria; however, BMI and WC, traditional measures, were more strongly associated with these outcomes. Future studies should compare measures of visceral adiposity by direct and indirect methods to evaluate if the risk of adverse kidney events are different and whether better noninvasive markers of visceral adiposity are needed to differentiate the subgroup of patients who are at higher risk of developing worse outcomes.