OBJECTIVE

To analyze the relationship between time-serial changes in insulin resistance and renal outcomes.

RESEARCH DESIGN AND METHODS

A prospective cohort of subjects from the general population without chronic kidney disease (CKD) underwent a biennial checkup for 12 years (n = 5,347). The 12-year duration was divided into a 6-year exposure period, where distinct HOMA for insulin resistance (HOMA-IR) trajectories were identified using latent variable mixture modeling, followed by a 6-year event accrual period, from which the renal outcome data were analyzed. The primary end point was adverse renal outcomes, defined as a composite of estimated glomerular filtration rate (eGFR) <60 mL/min/1.73 m2 in two or more consecutive checkups or albumin ≥1+ on urine strip.

RESULTS

Two distinct groups of HOMA-IR trajectories were identified during the exposure period: stable (n = 4,770) and increasing (n = 577). During the event accrual period, 449 patients (8.4%) developed adverse renal outcomes, and the risk was higher in the increasing HOMA-IR trajectory group than in the stable group (hazard ratio 2.06, 95% CI 1.62–2.60, P < 0.001). The results were similar after adjustment for baseline clinical characteristics, comorbidities, anthropometric and laboratory findings, eGFR, and HOMA-IR. The clinical significance of increasing HOMA-IR trajectory was similar in three or four HOMA-IR trajectories. The increasing tendency of HOMA-IR was persistently associated with a higher incidence of adverse renal outcomes, irrespective of the prevalence of diabetes.

CONCLUSIONS

An increasing tendency of insulin resistance was associated with a higher risk of adverse renal outcomes. Time-serial tracking of insulin resistance may help identify patients at high risk for CKD.

Chronic kidney disease (CKD) is an important risk factor of mortality (1). CKD affects >10% of the general population worldwide, with a profound socioeconomic burden (2,3). Prevalent metabolic syndrome, obesity, and prediabetes are linked with an increased risk of incident CKD (46). With the increasing number of patients with metabolic risk factors (7), the clinical importance of these populations has also increased in terms of CKD prevention.

Insulin resistance is a key pathogenic process in metabolic diseases. It systemically affects the vasculature via oxidative stress and chronic inflammation, thereby promoting the development of CKD (8). Several markers of insulin resistance, such as HOMA for insulin resistance (HOMA-IR) or triglyceride-glucose index, are associated with an increased risk of incident CKD (4,5). However, previous studies were mainly based on single assessments of insulin resistance or mere cross-sectional association analysis, which may not reflect the long-term influence of insulin resistance on CKD (4,5,912). The influence of insulin resistance on CKD is a dynamic time-dependent process (13); therefore, there is a need to dissect the association between the longitudinal trajectories of insulin resistance and renal outcomes.

In this study, we hypothesized that there are distinct groups of time-serial insulin resistance trajectories that might be differentially associated with renal outcomes. The objective of this study was to group the general population according to the time-serial trajectories of insulin resistance and to analyze its relationship with renal outcomes in a large-scale prospective longitudinal cohort with a 12-year follow-up.

Study Population

The study population was obtained from the Korean Genome and Epidemiology Study (KoGES) (14). A detailed description of this cohort is described in the Supplementary Materials. In brief, KoGES is a prospective cohort study in which participants from the general population aged 40–69 years were enrolled from 2001 to 2002 and followed every 2 years, with the seventh checkup in 2014. Figure 1A shows the study flow of the current study. The follow-up period from the baseline to the seventh checkup was divided into a 6-year exposure period (from baseline to the fourth checkup), where distinct HOMA-IR trajectories were identified, and a 6-year period from the fourth to the seventh checkup, which was defined as the event accrual period (Fig. 1A). A total of 5,347 patients with estimated glomerular filtration rate (eGFR) ≥60 mL/min/1.73 m2 at the baseline visit (first checkup) and at the end of the exposure period (fourth checkup), who underwent at least the baseline and the fourth checkup during the exposure period and at least two checkups during the event accrual period, were included in the current analysis (Fig. 1A). During the exposure period, the proportion of participants who underwent the second, third, and fourth checkups was 98.1%, 95.4%, and 100.0%, respectively. During the event accrual period, the proportion of participants who underwent the fifth, sixth, and seventh checkups was 96.7%, 96.1%, and 90.0%, respectively.

Figure 1

Study flow of the exposure and the event accrual periods, and the cumulative incidence of adverse renal outcomes according to each HOMA-IR trajectory group. *, the proportion of participants who underwent the corresponding checkup is shown in relation to the baseline checkup. A: The 12-year follow-up duration of the study participants was divided into a 6-year exposure period from baseline to the fourth checkup and 6-year event accrual period from the fourth to the seventh checkup. A total of 5,347 subjects with eGFR ≥60 mL/min/1.73 m2 at baseline and the fourth checkup underwent at least the baseline and the fourth checkup during the exposure period. The study participants who underwent at least two checkups during the 6-year event accrual periods were included as the final population of analysis. The cumulative incidence of adverse renal outcomes (B), decreased eGFR (C), and proteinuria development (D) in the stable and increasing HOMA-IR trajectory groups is presented.

Figure 1

Study flow of the exposure and the event accrual periods, and the cumulative incidence of adverse renal outcomes according to each HOMA-IR trajectory group. *, the proportion of participants who underwent the corresponding checkup is shown in relation to the baseline checkup. A: The 12-year follow-up duration of the study participants was divided into a 6-year exposure period from baseline to the fourth checkup and 6-year event accrual period from the fourth to the seventh checkup. A total of 5,347 subjects with eGFR ≥60 mL/min/1.73 m2 at baseline and the fourth checkup underwent at least the baseline and the fourth checkup during the exposure period. The study participants who underwent at least two checkups during the 6-year event accrual periods were included as the final population of analysis. The cumulative incidence of adverse renal outcomes (B), decreased eGFR (C), and proteinuria development (D) in the stable and increasing HOMA-IR trajectory groups is presented.

Close modal

Data Collection

The detailed methods for data collection are described in the Supplementary Materials. The baseline demographic and medical history data were collected at the initial checkup, and consequent medical histories were traced by questionnaires at each checkup. BMI was calculated by dividing the weight by the square of the height in meters. Hypertension was defined as systolic blood pressure (SBP) ≥140 mmHg, diastolic blood pressure ≥90 mmHg, or treatment with antihypertensive medications. Diabetes was defined as fasting plasma glucose levels ≥126 mg/dL after fasting for ≥8 h, a positive 75 g oral glucose tolerance test, glycated hemoglobin (HbA1c) ≥6.5%, or treatment with any antidiabetes drugs. Dyslipidemia was defined as triglycerides ≥200 mg/dL, total cholesterol ≥240 mg/dL, calculated LDL cholesterol (LDL-C) ≥160 mg/dL, HDL cholesterol (HDL-C) <40 mg/dL, or treatment with any lipid-lowering medications (15). LDL-C levels were calculated using the Friedewald equation. HOMA-IR was calculated according to the following formula: fasting insulin (µU/L) × fasting glucose (nmol/L)/22.5 (16). The Jaffe method was used to measure serum creatinine levels using nonisotope dilution-mass spectrometry. Creatinine was converted to isotope dilution-mass spectrometry (17,18). The eGFR was calculated using the Chronic Kidney Disease Epidemiology Collaboration Equation (19). Urine albumin was detected by urine strip, and the levels of absent, trace, 1+, 2+, or 3+ on the urine strip corresponded with urine albumin levels of <10, 10–20, >30, >100, and >500 mg/dL, respectively.

Latent Class Analysis for Classification of the Participants by Time-Serial Insulin Resistance Trajectory

The changing trend of insulin resistance over time was classified using latent variable mixture modeling with HOMA-IR during the exposure period (20). Latent variable mixture modeling is an evolving statistical method for the assessment of time-serial longitudinal data. It identifies homogenous patterns of time-serial measurements and classifies the population into groups with similar patterns, called latent classes. For a given variable, it calculates the probability of each individual belonging to a latent class and assigns each subject into distinct subpopulations (21,22). In this analysis, we used a growth mixture model that included more than one latent class and a random effect (23). Time was considered as a fixed effect and the individual as a random effect (24,25). Assuming the existence of multiple trajectory groups for which probabilities can be estimated, we created models with different numbers of trajectory groups (R 4.1.2, LCMM packages). The spline function was used as a link function to fit the data to the trajectories. The prespecified criteria to estimate the fitness of the model (26) is described in the Supplementary Materials, and two HOMA-IR trajectories (i.e., either stable or increasing) were identified to meet all three prespecified criteria (Supplementary Table 1). As a sensitivity analysis, the association of three and four trajectories was analyzed; however, they did not meet all three prespecified criteria of model fitness.

Renal Outcomes

The primary end point was adverse renal outcomes, defined as a composite of decreased eGFR (i.e., eGFR <60 mL/min/1.73 m2) in at least two consecutive checkups during the event accrual period or proteinuria (i.e., an albumin ≥1+ on the urine strip) during the event accrual period, as used in previous studies (26,27). The association of each HOMA-IR trajectory group with the clinical end points was assessed. As a sensitivity analysis, more strict definitions for renal outcomes were applied: 1) concomitant decreased eGFR and proteinuria, and 2) eGFR <45 mL/min/1.73 m2 in at least two consecutive checkups or albumin ≥3+ on the urine strip. To further corroborate the association between HOMA-IR trajectories and renal function, the annual rate of eGFR decline during the event accrual period was compared according to the HOMA-IR trajectories.

Statistical Analyses

The detailed statistical methods are described in the Supplementary Materials. In multivariate survival analyses, the baseline model (model 1), including age, sex, and baseline eGFR, was constructed. The clinical characteristics at baseline, including obesity (i.e., BMI ≥25 kg/m2), diabetes, hypertension, and dyslipidemia, were added to model 1 to construct model 2. Various definitions of dyslipidemia using different cutoff values and their combinations were used for model 2 as a sensitivity analysis. The anthropometric and laboratory findings, including baseline BMI, SBP, LDL-C, HOMA-IR, and hs-CRP, were added to model 1 for model 3. For the sensitivity analysis to minimize the confounding effect of diabetes at baseline, subgroup analysis in subjects with or without diabetes at baseline was performed. To account for newly developed diabetes during the exposure period, the outcome analysis was also performed in subjects with diabetes at the end of the exposure period (fourth checkup). All probability values were two-sided, and statistical significance was set at P < 0.05. All statistical analyses were performed using R 4.1.2 software (R Foundation for Statistical Computing, Vienna, Austria).

Baseline Patient Characteristics and Time-Serial Trajectories of Insulin Resistance

The baseline characteristics of the study population are presented in Table 1. The mean age of the study subjects was 51.1 ± 8.2 years, 47.8% of the subjects were men, and 13.6% had diabetes at baseline. The mean eGFR was 93.0 ± 12.4 mL/min/1.73 m2, and the mean HOMA-IR was 1.6 ± 1.1.

Table 1

Baseline characteristics by HOMA-IR trajectory

Total (n = 5,347)HOMA-IR trajectory
Stable (n = 4,770)Increasing (n = 577)P value
Age, years 51.1 ± 8.2 51.0 ± 8.2 51.7 ± 8.2 0.066 
Male sex 2,557 (47.8) 2,259 (47.4) 298 (51.6) 0.057 
BMI, kg/m2 24.6 ± 3.1 24.4 ± 2.9 26.7 ± 3.1 <0.001 
Waist circumference, cm 82.6 ± 8.6 81.9 ± 8.4 88.2 ± 8.0 <0.001 
SBP, mmHg 120.4 ± 17.5 119.7 ± 17.3 125.7 ± 17.9 <0.001 
DBP, mmHg 80.0 ± 11.1 79.5 ± 11.1 83.7 ± 11.1 <0.001 
ALT, units/L 28.1 ± 31.0 26.4 ± 20.4 42.2 ± 72.2 <0.001 
AST, units/L 29.1 ± 17.6 28.3 ± 12.7 36.1 ± 38.6 <0.001 
BUN, mg/dL 14.2 ± 3.5 14.2 ± 3.5 14.6 ± 3.5 0.006 
Creatinine, mg/dL 0.83 ± 0.17 0.83 ± 0.16 0.84 ± 0.17 0.049 
Cholesterol, mg/dL 190.5 ± 34.5 189.2 ± 33.9 201.2 ± 37.4 <0.001 
HDL-C, mg/dL 44.5 ± 9.9 44.9 ± 9.9 41.2 ± 8.7 <0.001 
Fasting glucose, mg/dL 86.5 ± 19.1 84.2 ± 13.7 107.0 ± 38.5 <0.001 
HbA1c, % 5.7 ± 0.8 5.6 ± 0.6 6.6 ± 1.5 <0.001 
eGFR, mL/min/1.73 m2 93.0 ± 12.4 93.1 ± 12.4 92.2 ± 12.8 0.124 
HOMA-IR 1.6 ± 1.1 1.5 ± 0.9 2.6 ± 1.8 <0.001 
hs-CRP 0.2 ± 0.6 0.2 ± 0.6 0.3 ± 0.4 0.005 
Diabetes 728 (13.6) 452 (9.5) 276 (47.8) <0.001 
Hypertension 1,580 (29.5) 1,321 (27.7) 259 (44.9) <0.001 
Dyslipidemia 2,592 (48.5) 2,198 (46.1) 394 (68.3) <0.001 
Adverse renal outcomes 449 (8.4) 364 (7.6) 85 (14.7) <0.001 
Decreased eGFR 383 (7.2) 317 (6.6) 66 (11.4) <0.001 
Proteinuria 95 (1.8) 66 (1.4) 29 (5.0) <0.001 
Total (n = 5,347)HOMA-IR trajectory
Stable (n = 4,770)Increasing (n = 577)P value
Age, years 51.1 ± 8.2 51.0 ± 8.2 51.7 ± 8.2 0.066 
Male sex 2,557 (47.8) 2,259 (47.4) 298 (51.6) 0.057 
BMI, kg/m2 24.6 ± 3.1 24.4 ± 2.9 26.7 ± 3.1 <0.001 
Waist circumference, cm 82.6 ± 8.6 81.9 ± 8.4 88.2 ± 8.0 <0.001 
SBP, mmHg 120.4 ± 17.5 119.7 ± 17.3 125.7 ± 17.9 <0.001 
DBP, mmHg 80.0 ± 11.1 79.5 ± 11.1 83.7 ± 11.1 <0.001 
ALT, units/L 28.1 ± 31.0 26.4 ± 20.4 42.2 ± 72.2 <0.001 
AST, units/L 29.1 ± 17.6 28.3 ± 12.7 36.1 ± 38.6 <0.001 
BUN, mg/dL 14.2 ± 3.5 14.2 ± 3.5 14.6 ± 3.5 0.006 
Creatinine, mg/dL 0.83 ± 0.17 0.83 ± 0.16 0.84 ± 0.17 0.049 
Cholesterol, mg/dL 190.5 ± 34.5 189.2 ± 33.9 201.2 ± 37.4 <0.001 
HDL-C, mg/dL 44.5 ± 9.9 44.9 ± 9.9 41.2 ± 8.7 <0.001 
Fasting glucose, mg/dL 86.5 ± 19.1 84.2 ± 13.7 107.0 ± 38.5 <0.001 
HbA1c, % 5.7 ± 0.8 5.6 ± 0.6 6.6 ± 1.5 <0.001 
eGFR, mL/min/1.73 m2 93.0 ± 12.4 93.1 ± 12.4 92.2 ± 12.8 0.124 
HOMA-IR 1.6 ± 1.1 1.5 ± 0.9 2.6 ± 1.8 <0.001 
hs-CRP 0.2 ± 0.6 0.2 ± 0.6 0.3 ± 0.4 0.005 
Diabetes 728 (13.6) 452 (9.5) 276 (47.8) <0.001 
Hypertension 1,580 (29.5) 1,321 (27.7) 259 (44.9) <0.001 
Dyslipidemia 2,592 (48.5) 2,198 (46.1) 394 (68.3) <0.001 
Adverse renal outcomes 449 (8.4) 364 (7.6) 85 (14.7) <0.001 
Decreased eGFR 383 (7.2) 317 (6.6) 66 (11.4) <0.001 
Proteinuria 95 (1.8) 66 (1.4) 29 (5.0) <0.001 

All values are presented as n (%) or as mean ± SD. BUN, blood urea nitrogen; DBP, diastolic blood pressure; HbA1c, glycated hemoglobin.

The time-serial HOMA-IR trajectory during the 6-year exposure period was used to classify the study population into the following two groups: stable (89.2%) and increasing (10.8%) insulin resistance. A comparison of the patient characteristics between the two groups of HOMA-IR trajectories is presented in Table 1. The baseline HOMA-IR and the proportion of patients with diabetes tended to be larger in the increasing HOMA-IR trajectory group than in the stable group (P < 0.001). Supplementary Fig. 1A and B shows the mean value and time-serial changes of HOMA-IR in the two HOMA-IR trajectory groups during the exposure period. HOMA-IR increased significantly from 2.61 to 6.92 in the increasing HOMA-IR trajectory group but slightly from 1.53 to 1.87 in the stable HOMA-IR trajectory group.

Association of the Time-Serial HOMA-IR Trajectory Groups With Renal Outcomes

During the 6-year event accrual period, 449 patients developed adverse renal outcomes, including decreased eGFR in 383 and proteinuria in 95. The cumulative events of adverse renal outcomes according to the HOMA-IR trajectory groups are shown in Fig. 1B–D. The risk of adverse renal outcomes was significantly higher in the increasing HOMA-IR trajectory group than in the stable group (hazard ratio [HR] 2.06, 95% CI 1.62–2.60, P < 0.001). In the multivariate analyses adjusted for age, sex, baseline eGFR, BMI ≥25 kg/m2, prevalent diabetes, hypertension, and dyslipidemia, the increasing HOMA-IR trajectory group was associated with a 1.6-times higher increased risk of adverse renal outcomes compared with the stable group (Table 2). The risk of the increasing HOMA-IR trajectory group was consistently higher regardless of various definitions of dyslipidemia (Supplementary Table 2). The results were similar after adjustment for age, sex, baseline eGFR, BMI, SBP, LDL-C, HOMA-IR, and hs-CRP (Table 2).

Table 2

Risk of adverse renal outcomes in the increasing HOMA-IR trajectory group

HOMA-IR trajectoryUnadjusted HR (95% CI)P valueModel 1 HR (95% CI)P valueModel 2 HR (95% CI)P valueModel 3 HR (95% CI)P value
Adverse renal outcomes (a composite of decreased eGFR and proteinuria) 
 Stable 1 (Reference) NA 1 (Reference) NA 1 (Reference) NA 1 (Reference) NA 
 Increasing 2.06 (1.62–2.60) <0.001 1.86 (1.47–2.36) <0.001 1.57 (1.21–2.04) <0.001 1.76 (1.33–2.33) <0.001 
Decreased eGFR (eGFR <60 mL/min/1.73 m2 in at least two consecutive checkups) 
 Stable 1 (Reference) NA 1 (Reference) NA 1 (Reference) NA 1 (Reference) NA 
 Increasing 1.80 (1.38–2.34) <0.001 1.60 (1.23–2.09) <0.001 1.38 (1.03–1.84) 0.032 1.47 (1.07–2.03) 0.017 
Proteinuria (albumin ≥1+ on urine strip) 
 Stable 1 (Reference) NA 1 (Reference) NA 1 (Reference) NA 1 (Reference) NA 
 Increasing 3.89 (2.51–6.02) <0.001 3.72 (2.40–5.76) <0.001 2.90 (1.77–4.76) <0.001 3.44 (2.05–5.76) <0.001 
HOMA-IR trajectoryUnadjusted HR (95% CI)P valueModel 1 HR (95% CI)P valueModel 2 HR (95% CI)P valueModel 3 HR (95% CI)P value
Adverse renal outcomes (a composite of decreased eGFR and proteinuria) 
 Stable 1 (Reference) NA 1 (Reference) NA 1 (Reference) NA 1 (Reference) NA 
 Increasing 2.06 (1.62–2.60) <0.001 1.86 (1.47–2.36) <0.001 1.57 (1.21–2.04) <0.001 1.76 (1.33–2.33) <0.001 
Decreased eGFR (eGFR <60 mL/min/1.73 m2 in at least two consecutive checkups) 
 Stable 1 (Reference) NA 1 (Reference) NA 1 (Reference) NA 1 (Reference) NA 
 Increasing 1.80 (1.38–2.34) <0.001 1.60 (1.23–2.09) <0.001 1.38 (1.03–1.84) 0.032 1.47 (1.07–2.03) 0.017 
Proteinuria (albumin ≥1+ on urine strip) 
 Stable 1 (Reference) NA 1 (Reference) NA 1 (Reference) NA 1 (Reference) NA 
 Increasing 3.89 (2.51–6.02) <0.001 3.72 (2.40–5.76) <0.001 2.90 (1.77–4.76) <0.001 3.44 (2.05–5.76) <0.001 

Model 1, adjusted for age, sex, and baseline eGFR. Model 2, model 1 + BMI ≥25 kg/m2, diabetes, hypertension, and dyslipidemia. Model 3, model 1 + baseline BMI, SBP, LDL-C, HOMA-IR, and hs-CRP. NA, not applicable.

For a sensitivity analysis, we also excluded those who developed adverse renal outcomes during the 6-year exposure period. During the 6-year exposure period, 221 patients experienced adverse renal outcomes, including 78 with decreased eGFR (eGFR <60 mL/min/1.73 m2 in at least 2 consecutive checkups) and 147 with proteinuria (albumin ≥1+ on urine strip). Again, an increasing HOMA-IR trajectory was still associated with an increased risk of adverse renal outcomes in 5,126 patients without adverse renal outcomes during the exposure period (Supplementary Fig. 2A and Supplementary Table 3). These results were also consistent in 704 patients with eGFR ≥90 mL/min/1.73 m2 at baseline and at the fourth checkups (Supplementary Fig. 2B and Supplementary Table 4).

Sensitivity Analyses With Different HOMA-IR Trajectories or Other Definitions of Renal Outcomes

In the sensitivity analysis with three HOMA-IR trajectories, the study population could be characterized as follows: stable with low HOMA-IR (79.1%), stable with high HOMA-IR (14.1%), and increasing (6.8%) HOMA-IR groups. A comparison of the patient characteristics among the three groups is presented in Supplementary Table 5, and the time-serial changes of HOMA-IR in these three groups are presented in Supplementary Fig. 3A and B. Compared with the stable with low HOMA-IR trajectory group, the risk of the adverse renal outcomes significantly increased in the increasing HOMA-IR trajectory group (HR 1.98, 95% CI 1.47–2.67, P < 0.001) (Supplementary Fig. 3C).

In another sensitivity analysis with four HOMA-IR trajectories, the study population could be characterized as follows: stable with low HOMA-IR (59.8%), stable with high HOMA-IR (7.7%), slightly increasing HOMA-IR (24.0%), and rapidly increasing HOMA-IR (8.5%) groups. Supplementary Table 6 describes the baseline patient characteristics of the four groups, and Supplementary Fig. 4A and B shows the time-serial changes of HOMA-IR in these four groups. Compared with the stable with low HOMA-IR trajectory group, the risk of adverse renal outcomes significantly increased in the slightly increasing group (HR 1.34, 95% CI 1.07–1.67, P < 0.001) and to a higher degree in the rapidly increasing group (HR 2.24, 95% CI 1.70–2.95, P < 0.001) (Supplementary Fig. 4C).

Different definitions of renal outcomes revealed similar results. Among 449 patients who developed adverse renal outcomes during the 6-year event accrual period, 29 patients had decreased eGFR and proteinuria concomitantly, the risk of which was significantly higher in the increasing HOMA-IR trajectory group than in the stable HOMA-IR trajectory group (Supplementary Fig. 5A). Moreover, increasing HOMR-IR trajectory was associated with a three-times higher increased risk of eGFR <45 mL/min/1.73 m2 in at least two consecutive checkups or albumin ≥3+ on the urine strip (Supplementary Fig. 5B) and portended a higher annual rate of eGFR decline during the event accrual period than in the stable HOMA-IR trajectory group (Supplementary Table 7).

Time-Serial HOMA-IR Trajectory Groups and Renal Outcomes in Subjects With and Without Diabetes

Because the degree of insulin resistance is closely associated with the presence of diabetes at baseline, we performed a subgroup analysis to investigate whether the time-serial HOMA-IR trajectory is a determinant of renal outcome according to the presence of diabetes at baseline. Baseline patient characteristics by HOMA-IR trajectory in subjects with and without diabetes are shown in Supplementary Table 8.

Supplementary Fig. 6 shows the association of time-serial changes of HOMA-IR in the stable and increasing HOMA-IR trajectory groups with adverse renal outcomes during the event accrual period in subjects with or without diabetes. The increasing HOMA-IR trajectory group was persistently associated with a higher risk of adverse renal outcomes both in subjects with and without diabetes (Supplementary Fig. 6). These trends toward an increased risk were similar after multivariate adjustments (Table 3). Additionally, in subjects with diabetes at the end of the exposure period, the increasing HOMA-IR trajectory group still showed a higher risk of adverse renal outcomes than the stable HOMA-IR trajectory group (Supplementary Fig. 7 and Table 3).

Table 3

Risk of adverse renal outcomes in the increasing HOMA-IR trajectory group according to the prevalent diabetes

HOMA-IR trajectoryUnadjusted HR (95% CI)P valueModel 1 HR (95% CI)P valueModel 2 HR (95% CI)P valueModel 3 HR (95% CI)P value
Subjects without prevalent diabetes at baseline 
 Stable 1 (Reference) NA 1 (Reference) NA 1 (Reference) NA 1 (Reference) NA 
 Increasing 1.75 (1.24–2.47) 0.001 1.77 (1.25–2.50) 0.001 1.65 (1.16–2.35) 0.005 1.63 (1.13–2.35) 0.008 
Subjects with prevalent diabetes at baseline 
 Stable 1 (Reference) NA 1 (Reference) NA 1 (Reference) NA 1 (Reference) NA 
 Increasing 1.46 (1.00–2.13) 0.051 1.55 (1.06–2.27) 0.025 1.52 (1.03–2.23) 0.034 1.71 (1.04–2.82) 0.034 
Subjects with prevalent diabetes at the end of exposure period (fourth checkup) 
 Stable 1 (Reference) NA 1 (Reference) NA 1 (Reference) NA 1 (Reference) NA 
 Increasing 1.50 (1.08–2.07) 0.015 1.56 (1.13–2.16) 0.007 1.51 (1.08–2.10) 0.015 1.79 (1.19–2.71) 0.005 
HOMA-IR trajectoryUnadjusted HR (95% CI)P valueModel 1 HR (95% CI)P valueModel 2 HR (95% CI)P valueModel 3 HR (95% CI)P value
Subjects without prevalent diabetes at baseline 
 Stable 1 (Reference) NA 1 (Reference) NA 1 (Reference) NA 1 (Reference) NA 
 Increasing 1.75 (1.24–2.47) 0.001 1.77 (1.25–2.50) 0.001 1.65 (1.16–2.35) 0.005 1.63 (1.13–2.35) 0.008 
Subjects with prevalent diabetes at baseline 
 Stable 1 (Reference) NA 1 (Reference) NA 1 (Reference) NA 1 (Reference) NA 
 Increasing 1.46 (1.00–2.13) 0.051 1.55 (1.06–2.27) 0.025 1.52 (1.03–2.23) 0.034 1.71 (1.04–2.82) 0.034 
Subjects with prevalent diabetes at the end of exposure period (fourth checkup) 
 Stable 1 (Reference) NA 1 (Reference) NA 1 (Reference) NA 1 (Reference) NA 
 Increasing 1.50 (1.08–2.07) 0.015 1.56 (1.13–2.16) 0.007 1.51 (1.08–2.10) 0.015 1.79 (1.19–2.71) 0.005 

In subjects without and with prevalent diabetes at baseline, adverse renal outcomes occurred in 341 and 108 patients, respectively. In subjects with prevalent diabetes at the end of exposure period, 149 patients developed adverse renal outcomes. Model 1, adjusted for age, sex, and baseline eGFR. Model 2, model 1 + BMI ≥25 kg/m2, hypertension, and dyslipidemia. Model 3, model 1 + baseline BMI, SBP, LDL-C, HOMA-IR, and hs-CRP. NA, not applicable.

Incidence of Renal Outcomes in Relation to Baseline Covariates

We also evaluated the interaction between baseline covariates and the association between HOMA-IR trajectory groups and renal outcomes. Supplementary Fig. 8 presents the HRs of the increasing HOMA-IR trajectory group compared with the stable group in subgroups by age, sex, BMI, hypertension, dyslipidemia, baseline HOMA-IR, and baseline eGFR. The risks of adverse renal outcomes were significantly increased in the increasing HOMA-IR trajectory group, independent of the baseline clinical characteristics.

In the 1:1 matched pairs by the baseline HOMA-IR between the stable and increasing groups, the baseline HOMA-IR was not different between the two groups (2.14 ± 1.45 vs. 2.12 ± 1.47, P = 0.896). However, the cumulative incidence of adverse renal outcomes was again higher in the increasing HOMA-IR trajectory groups (Supplementary Fig. 9).

In the current study, using a large-scale prospective longitudinal cohort with a 12-year follow-up, we grouped the general population into groups with distinct time-serial trajectories of insulin resistance and found that long-term aggravation of insulin resistance is associated with worse renal outcomes. The main findings can be summarized as follows:

First, by applying a latent variable mixture model, two groups of stable and increasing insulin resistance trajectory groups were identified, and the increasing insulin resistance trajectory group was significantly associated with deterioration of eGFR or new-onset proteinuria. These results were consistent with the sensitivity analysis using three or four HOMA-IR trajectories, with both slightly or rapidly increasing HOMA-IR trajectories significantly associated with adverse renal outcomes.

Second, the increasing insulin resistance trajectory was consistently associated with an increased risk of adverse renal outcomes in subjects both with and without diabetes.

Third, the association of increasing HOMA-IR trajectory with adverse renal outcomes was persistently observed in subgroups based on clinical characteristics and even in the matched pairs based on the baseline HOMA-IR.

Metabolic syndrome is closely associated with incident CKD (28). Several phenotypes of metabolic diseases, such as high BMI, abdominal obesity, dyslipidemia, and prediabetes, are robust risk factors for CKD and a rapid decline in kidney function (46,11,29). Given that metabolic risk factors can be modified by preventive measures, such as intensive lifestyle modification, it can be postulated that CKD development can be prevented in a large proportion of patients with metabolic disease. However, although the development of CKD is a process that involves the time-dependent metabolic risk pervading on the renal vasculature, prior studies have only analyzed metabolic risk factors measured at a single time point, thus falling short of providing insights into whether the long-term information accumulated by time may predict adverse renal outcomes (4,5,912,2931).

Recent studies revealing longitudinal trajectories of single metabolic risks, persistent obesity, or an increasing trend toward obesity were associated with an increased risk of CKD development (32) as well as increasing SBP over time (26). Obesity and high blood pressure are but one phenotype of metabolic disease; however, the longitudinal association of time-serial changes in insulin resistance, a central pathogenic mechanism underlying metabolic diseases (33), with renal outcomes has not been investigated yet. Considering the significant time lag between the onset of metabolic derangement and insidious renal damage, our prospective cohort with 12 years of follow-up enabled us to investigate the dynamic and long-term relationship between insulin resistance and CKD.

With this unique prospective cohort of long-term follow-up for 12 years, we found that patients whose HOMA-IR increased over time had a significantly higher risk of developing adverse renal outcomes than those whose HOMA-IR was stable. Our finding aligns with a previous publication reporting the association of higher levels of HOMA-IR with a rapid decline in eGFR, supporting the dynamic association of insulin resistance with CKD development (4). In addition, a recent study showed that the triglyceride-glucose index, another marker for insulin resistance, mediated half of the total association of BMI with end-stage kidney disease with a mean follow-up of 22.7 years (5), suggesting the independent role of insulin resistance in worsening renal outcomes. These findings indicate that a possible longitudinal association of time-serial changes in insulin resistance with renal outcomes may exist and that an increasing pattern of HOMA-IR may also contribute to CKD development.

Notably, when the whole population was classified into four HOMA-IR trajectories in which rapidly increasing, slightly increasing, stable with high HOMA-IR, and stable with low HOMA-IR groups were identified, the risk of adverse renal outcomes was increased even in subjects with a slightly increasing HOMA-IR trajectory. Although the degree of HOMA-IR change between the slightly increasing and stable with low HOMA-IR groups was small, it was associated with a significant difference in the long-term renal outcome. This gradually pervading effect of insulin resistance was also seen in the subjects without diabetes, where the increase in HOMA-IR was not as dramatic as in the subjects with diabetes, but the difference in the renal outcome remained significant. This demonstrates that even a small increase in insulin resistance would lead to significant damage to the kidney when exposed to a significant length of time.

Several studies have demonstrated that insulin resistance precedes the diagnosis of overt diabetes (34) and that diabetes is a leading cause of CKD development. A notable finding of our present study is that the increasing HOMA-IR trajectory is associated with an increased risk of CKD, even in subjects without diabetes. This suggests that the trend toward an increased risk of CKD may begin before the overt presentation of diabetes, which is in line with previous findings on the association of insulin resistance with albuminuria development or CKD progression in individuals without diabetes (10,35). Moreover, an increasing HOMA-IR pattern was also a risk factor for adverse renal outcomes in subjects with diabetes at baseline or at the end of the exposure period, and a previous study reporting the association of the dose-response effect of HOMA-IR on the increased risk of microalbuminuria supports our finding (9). Although the mechanism needs to be further investigated, overproduction of lipoprotein and enhanced glucolipotoxicity by insulin resistance promotes renal insufficiency. Systemic inflammation and oxidative stress are the central pathophysiologic pathway by insulin resistance, causing direct damage to the mesangial cells and renal vasculature (36). Increasing insulin resistance aggravates renal hemodynamics by activating the sympathetic nervous system and sodium retention, and the decrease in Na+/K+-ATPase activity and altered insulin signaling pathway is linked to the endoplasmic reticulum stress leading to podocyte damage associated with proteinuria (8). This pathophysiological basis explains why the aggravation of insulin resistance predicts the future risk of CKD independent of various clinical characteristics and the baseline degree of insulin resistance in the current study. Thus, it is important to trace the longitudinal trend as well as the baseline level of insulin resistance, especially for patients who are currently under long-term regular follow-up in routine practice to detect potentially high-risk patients for CKD development. In particular, individuals without diabetes showing an increasing trend of insulin resistance should be followed with meticulous surveillance of renal function and intensive measures for control of risk factors to prevent future adverse renal outcomes.

Our study has some limitations. First, the observational design of the current study could not fully demonstrate the causal relationship between time-serial changes in HOMA-IR and renal outcomes, and the possibility of bidirectional association should be considered. To mitigate the effect of CKD on insulin resistance, we divided the study period into the exposure period and the event accrual period and included patients with eGFR ≥60 mL/min/1.73 m2 at the start and end of the exposure period. In addition, the results were persistent after excluding those who developed renal outcomes in the exposure period and even in patients with eGFR ≥90 mL/min/1.73 m2 at the start and end of the exposure period.

Second, proteinuria was estimated by a urine dipstick, which is a semiquantitative method at best and may seem crude for defining adverse renal outcomes. Nevertheless, prior literature has demonstrated the high sensitivity and specificity of the urine dipstick test to detect proteinuria (37,38) and its association with rapid kidney function decline or long-term mortality in a large population (39,40).

Third, since we included the patients who underwent at least one or more checkups during the exposure and at least two checkups during the event accrual period, those who were lost to follow-up may have contributed to selection bias.

Fourth, given that the study population was not randomized, there might be potential confounders unequally distributed between the HOMA-IR trajectory groups that cannot be fully adjusted for by multivariate analysis or the subgroup analysis.

Lastly, conclusions cannot be drawn for a possible relationship between the increasing tendency of insulin resistance and the worsening CKD in those with renal insufficiency at baseline since these were excluded from the target population of analysis.

In conclusion, a time-serial increase in insulin resistance is significantly associated with worsening renal outcomes. Long-term serial assessments of insulin resistance in clinical practice may help identify patients at a higher risk of adverse renal events in the future.

S.Ka. and S.-P.L. contributed equally as last authors.

This article contains supplementary material online at https://doi.org/10.2337/figshare.19209897.

Acknowledgments. The authors thank the Medical Research Support Services of Yonsei University College of Medicine for the artistic support and Editage for English language editing.

Funding. This work was supported by a National Research Foundation of Korea grant funded by the Korean government (Ministry of Science and ICT, No. 2019R1A2C2084099).

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

Author Contributions. S.Y., S.Kw., H.S.L., S.Ka., and S.-P.L. conducted the statistical analysis. S.Y., S.Ka. and S.-P.L. contributed to the conception and design of the analysis. S.Y., S.Ka., and S.-P.L. drafted the initial version of the manuscript. S.Kw., Y.-H.S., S.S.H., S.Ka., and S.-P.L. acquired the data. S.Kw., Y.-H.S., S.S.H., and H.S.L. revised the manuscript critically for important intellectual contents. All authors reviewed and approved the final version of the manuscript. All authors agree to be accountable for all aspects of the work in ensuring that questions related to the accuracy or integrity of any part of the work are appropriately investigated and resolved. S.-P.L. is the guarantor of this work and, as such, had full access to all the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis.

1.
Chronic Kidney Disease Prognosis Consortium
;
Matsushita
K
,
van der Velde
M
,
Astor
BC
, et al
.
Association of estimated glomerular filtration rate and albuminuria with all-cause and cardiovascular mortality in general population cohorts: a collaborative meta-analysis
.
Lancet
2010
;
375
:
2073
2081
2.
Hill
NR
,
Fatoba
ST
,
Oke
JL
, et al
.
Global prevalence of chronic kidney disease—a systematic review and meta-analysis
.
PLoS One
2016
;
11
:
e0158765
3.
Xie
Y
,
Bowe
B
,
Mokdad
AH
, et al
.
Analysis of the Global Burden of Disease study highlights the global, regional, and national trends of chronic kidney disease epidemiology from 1990 to 2016
.
Kidney Int
2018
;
94
:
567
581
4.
Huh
JH
,
Yadav
D
,
Kim
JS
, et al
.
An association of metabolic syndrome and chronic kidney disease from a 10-year prospective cohort study
.
Metabolism
2017
;
67
:
54
61
5.
Fritz
J
,
Brozek
W
,
Concin
H
, et al
.
The triglyceride-glucose index and obesity-related risk of end-stage kidney disease in Austrian adults
.
JAMA Netw Open
2021
;
4
:
e212612
6.
Echouffo-Tcheugui
JB
,
Narayan
KM
,
Weisman
D
,
Golden
SH
,
Jaar
BG
.
Association between prediabetes and risk of chronic kidney disease: a systematic review and meta-analysis
.
Diabet Med
2016
;
33
:
1615
1624
7.
Saklayen
MG
.
The global epidemic of the metabolic syndrome
.
Curr Hypertens Rep
2018
;
20
:
12
8.
Gluba
A
,
Mikhailidis
DP
,
Lip
GY
,
Hannam
S
,
Rysz
J
,
Banach
M
.
Metabolic syndrome and renal disease
.
Int J Cardiol
2013
;
164
:
141
150
9.
Hsu
CC
,
Chang
HY
,
Huang
MC
, et al
.
Association between insulin resistance and development of microalbuminuria in type 2 diabetes: a prospective cohort study
.
Diabetes Care
2011
;
34
:
982
987
10.
Jang
CM
,
Hyun
YY
,
Lee
KB
,
Kim
H
.
Insulin resistance is associated with the development of albuminuria in Korean subjects without diabetes
.
Endocrine
2015
;
48
:
203
210
11.
Kim
J
,
Bae
YJ
,
Shin
SJ
,
You
HS
,
Lee
JW
,
Kang
HT
.
The ratio of triglycerides to high-density lipoprotein cholesterol is associated with the risk of chronic kidney disease in Korean men
.
Lipids
2021
;
56
:
475
483
12.
Schroijen
MA
,
de Mutsert
R
,
Dekker
FW
, et al
.
The association of glucose metabolism and kidney function in middle-aged adults
.
Clin Kidney J
2021
;
14
:
2383
2390
13.
Webster
AC
,
Nagler
EV
,
Morton
RL
,
Masson
P
.
Chronic kidney disease
.
Lancet
2017
;
389
:
1238
1252
14.
Kim
Y
;
KoGES group
.
Cohort profile: The Korean Genome and Epidemiology study (KoGES) consortium [published correction appears in Int J Epidemiol 2017;46:1350]
.
Int J Epidemiol
2017
;
46
:
e20
15.
National Cholesterol Education Program (NCEP) Expert Panel on Detection, Evaluation, and Treatment of High Blood Cholesterol in Adults (Adult Treatment Panel III)
.
Third report of the national cholesterol education program (NCEP) expert panel on detection, evaluation, and treatment of high blood cholesterol in adults (Adult Treatment panel III) final report
.
Circulation
2002
;
106
:
3143
3421
16.
Matthews
DR
,
Hosker
JP
,
Rudenski
AS
,
Naylor
BA
,
Treacher
DF
,
Turner
RC
.
Homeostasis model assessment: insulin resistance and beta-cell function from fasting plasma glucose and insulin concentrations in man
.
Diabetologia
1985
;
28
:
412
419
17.
Levey
AS
,
Coresh
J
,
Greene
T
, et al.;
Chronic Kidney Disease Epidemiology Collaboration
.
Expressing the Modification of Diet in Renal Disease Study equation for estimating glomerular filtration rate with standardized serum creatinine values
.
Clin Chem
2007
;
53
:
766
772
18.
Joffe
M
,
Hsu
CY
,
Feldman
HI
,
Weir
M
,
Landis
JR
;
Chronic Renal Insufficiency Cohort (CRIC) Study Group
.
Variability of creatinine measurements in clinical laboratories: results from the CRIC study
.
Am J Nephrol
2010
;
31
:
426
434
19.
Levey
AS
,
Stevens
LA
,
Schmid
CH
, et al.;
CKD-EPI (Chronic Kidney Disease Epidemiology Collaboration)
.
A new equation to estimate glomerular filtration rate
.
Ann Intern Med
2009
;
150
:
604
612
20.
Sinha
P
,
Calfee
CS
,
Delucchi
KL
.
Practitioner’s guide to latent class analysis: methodological considerations and common pitfalls
.
Crit Care Med
2021
;
49
:
e63
e79
21.
Berlin
KS
,
Williams
NA
,
Parra
GR
.
An introduction to latent variable mixture modeling (part 1): overview and cross-sectional latent class and latent profile analyses
.
J Pediatr Psychol
2014
;
39
:
174
187
22.
Berlin
KS
,
Parra
GR
,
Williams
NA
.
An introduction to latent variable mixture modeling (part 2): longitudinal latent class growth analysis and growth mixture models
.
J Pediatr Psychol
2014
;
39
:
188
203
23.
van der Nest
G
,
Lima Passos
V
,
Candel
MJJM
,
van Breukelen
GJP
.
An overview of mixture modelling for latent evolutions in longitudinal data: modelling approaches, fit statistics and software
.
Adv Life Course Res
2020
;
43
:
100323
24.
Ohn
JH
,
Kwak
SH
,
Cho
YM
, et al
.
10-year trajectory of β-cell function and insulin sensitivity in the development of type 2 diabetes: a community-based prospective cohort study
.
Lancet Diabetes Endocrinol
2016
;
4
:
27
34
25.
Lai
D
,
Xu
H
,
Koller
D
,
Foroud
T
,
Gao
S
.
A multivariate finite mixture latent trajectory model with application to dementia studies
.
J Appl Stat
2016
;
43
:
2503
2523
26.
Joo
YS
,
Lee
C
,
Kim
HW
, et al
.
Association of longitudinal trajectories of systolic BP with risk of incident CKD: results from the Korean Genome and Epidemiology Study
.
J Am Soc Nephrol
2020
;
31
:
2133
2144
27.
Lee
J
,
Oh
S
,
Kang
H
, et al
.
Environment-wide association study of CKD
.
Clin J Am Soc Nephrol
2020
;
15
:
766
775
28.
Thomas
G
,
Sehgal
AR
,
Kashyap
SR
,
Srinivas
TR
,
Kirwan
JP
,
Navaneethan
SD
.
Metabolic syndrome and kidney disease: a systematic review and meta-analysis
.
Clin J Am Soc Nephrol
2011
;
6
:
2364
2373
29.
Lu
JL
,
Molnar
MZ
,
Naseer
A
,
Mikkelsen
MK
,
Kalantar-Zadeh
K
,
Kovesdy
CP
.
Association of age and BMI with kidney function and mortality: a cohort study
.
Lancet Diabetes Endocrinol
2015
;
3
:
704
714
30.
Wu
M
,
Shu
Y
,
Wang
L
, et al
.
Metabolic syndrome severity score and the progression of CKD
.
Eur J Clin Invest
2022
;
52
:
e13646
31.
Luk
AO
,
So
WY
,
Ma
RC
, et al.;
Hong Kong Diabetes Registry
.
Metabolic syndrome predicts new onset of chronic kidney disease in 5,829 patients with type 2 diabetes: a 5-year prospective analysis of the Hong Kong Diabetes Registry
.
Diabetes Care
2008
;
31
:
2357
2361
32.
Zhang
HS
,
An
S
,
Ahn
C
,
Park
SK
,
Park
B
.
Obesity measures at baseline, their trajectories over time, and the incidence of chronic kidney disease: a 14 year cohort study among Korean adults
.
Nutr Metab Cardiovasc Dis
2021
;
31
:
782
792
33.
Cornier
MA
,
Dabelea
D
,
Hernandez
TL
, et al
.
The metabolic syndrome
.
Endocr Rev
2008
;
29
:
777
822
34.
Tabák
AG
,
Herder
C
,
Rathmann
W
,
Brunner
EJ
,
Kivimäki
M
.
Prediabetes: a high-risk state for diabetes development
.
Lancet
2012
;
379
:
2279
2290
35.
Kobayashi
H
,
Tokudome
G
,
Hara
Y
, et al
.
Insulin resistance is a risk factor for the progression of chronic kidney disease
.
Clin Nephrol
2009
;
71
:
643
651
36.
Sarafidis
PA
,
Ruilope
LM
.
Insulin resistance, hyperinsulinemia, and renal injury: mechanisms and implications
.
Am J Nephrol
2006
;
26
:
232
244
37.
Lim
D
,
Lee
DY
,
Cho
SH
, et al
.
Diagnostic accuracy of urine dipstick for proteinuria in older outpatients
.
Kidney Res Clin Pract
2014
;
33
:
199
203
38.
White
SL
,
Yu
R
,
Craig
JC
,
Polkinghorne
KR
,
Atkins
RC
,
Chadban
SJ
.
Diagnostic accuracy of urine dipsticks for detection of albuminuria in the general community
.
Am J Kidney Dis
2011
;
58
:
19
28
39.
Kwon
Y
,
Han
K
,
Kim
YH
, et al
.
Dipstick proteinuria predicts all-cause mortality in general population: a study of 17 million Korean adults
.
PLoS One
2018
;
13
:
e0199913
40.
Nichols
GA
,
Déruaz-Luyet
A
,
Brodovicz
KG
,
Kimes
TM
,
Rosales
AG
,
Hauske
SJ
.
Kidney disease progression and all-cause mortality across estimated glomerular filtration rate and albuminuria categories among patients with vs. without type 2 diabetes
.
BMC Nephrol
2020
;
21
:
167
Readers may use this article as long as the work is properly cited, the use is educational and not for profit, and the work is not altered. More information is available at https://www.diabetesjournals.org/journals/pages/license.