Association of Cumulative Average Uric Acid to HDL Cholesterol Ratio with the Risk of Rapidly Declining Renal Function: A Retrospective Community-Based Cohort Study in Shanghai Undertaken on Elderly Subjects with Hypertension and Type 2 Diabetes Mellitus Patients

Introduction

With the global acceleration of population ageing, the concurrence of hypertension and type 2 diabetes mellitus (T2DM) has become increasingly prevalent, constituting a major public-health challenge.¹ These two chronic conditions act synergistically to markedly amplify target-organ damage—particularly renal injury—such that diabetic kidney disease has now emerged as the leading aetiology of end-stage kidney disease.² In older adults, the inherent decline in physiological reserve coupled with multiple comorbidities renders the kidneys especially vulnerable.³ Consequently, early identification of risk factors for rapid kidney function decline (RD) in this high-risk cohort is imperative to retard disease progression and improve patient prognosis.⁴

Conventional metrics of renal function—most notably estimated glomerular filtration rate (eGFR) —typically decline appreciably only after substantial structural damage has occurred, markedly curtailing their utility for early risk stratification. More recently, a panel of easily derivable and cost-effective biomarkers—including the triglyceride–glucose (TyG) index,^5,6 the triglyceride glucose–body mass index (TyG-BMI),⁷ the uric acid to high-density lipoprotein cholesterol ratio (UHR),⁸ the atherogenic index of plasma (AIP),⁹ and single point insulin sensitivity estimator (SPISE Index)¹⁰ —have shown robust promise in quantifying insulin resistance, metabolic dysregulation, and atherogenesis. These same pathophysiological pathways constitute the core mechanistic nexus through which hypertension and diabetes mellitus converge to inflict renal injury.¹¹ Although prior work has linked individual indices to incident kidney damage,^12,13 their comparative strength of association with rapid renal-function decline—and the optimal predictor among them—remain inadequately characterized in the specific demographic of older adults burdened by both hypertension and T2DM.

Accordingly, we conducted a retrospective cohort study enrolling individuals aged ≥60 years with coexistent hypertension and T2DM to delineate the relationship between RD and both baseline and 4-year cumulative average values of the TyG, TyG-BMI, UHR, AIP and SPISE Index. Previous study calculated the cumulative average value using the mean of two time points: the first and the last year.¹⁴ Therefore, this study adopts the same method. Our objective was to identify the strongest predictor among these indices to facilitate early risk stratification and guide targeted clinical intervention.

Subjects and Methods

Study Population

We conducted a retrospective, single-center cohort analysis nested within the electronic health-record system of Chuansha Huaxia Community Health Service Center (Shanghai, China). Consecutive outpatients aged ≥60 years who carried concomitant physician-verified diagnoses of hypertension and T2DM between 1 January 2020 and 31 December 2023 were screened for eligibility. Inclusion criteria: (1) Age ≥60 years at index date; (2) Hypertension, (3) T2DM. Exclusion criteria: (1) Age <60 years, (2) Documented secondary hypertension, (3) Non-type 2 diabetes mellitus (type 1, MODY, gestational, drug- or pancreas-related), (4) Insufficient baseline or follow-up data to ascertain exposure or outcome variables. The study protocol was approved by the Institutional Ethics Committee of Chuansha Huaxia Community Health Service Center and adhered to the Declaration of Helsinki. Written informed consent was obtained from all participants or their legal proxies. (Approval number:PW2019A-14;2019.11.10)

Data Collection and Laboratory Measurements

Baseline demographic and clinical characteristics—including age, gender, height, body weight, fatty liver, duration of hypertension, and duration of diabetes—were systematically recorded for all enrolled participants.

Following a standardized 10- to 12-hour overnight fast, early-morning venous blood and mid-stream urine specimens were obtained. Serum indices comprised alanine aminotransferase (ALT), aspartate aminotransferase (AST), fasting plasma glucose (FPG), triglycerides (TG), total cholesterol (TC), high-density lipoprotein cholesterol (HDL-C), low-density lipoprotein cholesterol (LDL-C), creatinine, urea, and uric acid (UA). Urinalysis quantified urinary glucose and protein by automated dipstick and confirmatory testing.

The estimated glomerular filtration rate (eGFR) was computed with CKD-EPI creatinine equation.¹⁵ Body-mass index (BMI) was calculated as weight (kg) divided by the square of height (m²). Referring to previous study,¹⁴ the cumulative average was simulated based on the mean values of the two time points (2020 and 2023) in this study. The uric acid to high-density lipoprotein cholesterol ratio (UHR) was expressed as:¹⁶ UHR = [UA (mg/dL) / HDL-C (mg/dL)] × 100%. The cumulative average (cum) UHR= (UHR₂₀₂₀+UHR₂₀₂₃)/2. Triglyceride–glucose index (TyG) = ln[TG (mg/dL) × FPG (mg/dL) / 2]. Triglyceride glucose-body mass index (TyG-BMI)=TyG×BMI. cum TyG= (TyG ₂₀₂₀+ TyG ₂₀₂₃)/2.¹⁷ cum TyG-BMI = (TyG-BMI₂₀₂₀+ TyG -BMI₂₀₂₃)/2. Atherogenic index of plasma (AIP)=Log10[TG(mg/dL)/HDL-C(mg/dL)].¹⁸ cum AIP = (AIP₂₀₂₀+ AIP₂₀₂₃)/2. Single point insulin sensitivity estimator (SPISE Index) = 600 × HDL-C^0.185(mg/dL)/ [triglyceride^0.2(mg/dL) × BMI^1.338].¹⁹ cum SPISE Index = (SPISE Index ₂₀₂₀+ SPISE Index ₂₀₂₃)/2.

Diagnostic Criteria and Grouping

Hypertension was diagnosed according to the 2018 Chinese Guidelines for the Management of Hypertension: sustained systolic/diastolic blood pressure ≥140/90 mmHg on ≥2 visits or current antihypertensive use.²⁰ Type 2 diabetes mellitus was defined following the Guideline for the Prevention and Treatment of Type 2 Diabetes Mellitus in China (2020 Edition): fasting plasma glucose ≥7.0 mmol/L, 2h OGTT ≥11.1 mmol/L, HbA1c ≥6.5%, or active glucose-lowering therapy.²¹ Chronic kidney disease (CKD) was identified based on the Guidelines for Early Screening, Diagnosis, Prevention and Treatment of Chronic Kidney Disease (2022 Edition):²² presence of markers of kidney damage—albuminuria, urinary sediment abnormalities, tubular disorders, histological lesions, structural abnormalities on imaging, or history of kidney transplantation—and/or an estimated glomerular filtration rate (eGFR) <60 mL/min/1.73 m² for ≥3 months. Participants with a sustained eGFR decline >5 mL/min/1.73 m²/year,²³ corresponding to a >15 mL/min/1.73 m² reduction over 3 years, were assigned to the rapid renal function decline (RD) group; all others constituted the Non-RD group.

Statistical Analysis

Continuous variables are expressed as mean ± standard deviation (SD) if normally distributed and were compared with the independent-samples t test; otherwise they are reported as median (inter-quartile range, IQR) and compared by the Mann–Whitney U-test. Categorical data are summarised as frequencies and compared by χ² or Fisher’s exact test, as appropriate.

The associations of baseline TyG, cum TyG, TyG-BMI, cum TyG-BMI, UHR, cum UHR, AIP, cum AIP, SPISE index and cum SPISE index with incident rapid kidney function decline (RD) were first examined in univariable logistic regression and subsequently in multivariable models. Each exposure was entered both as a continuous variable (per 1-SD increment) and as a categorical variable (tertiles), with the lowest tertile serving as reference. Three sequentially adjusted models were constructed to evaluate robustness of the estimates: In Model 1, adjustments were made for age and gender. In Model 2, adjustments were made for age, gender, hypertension duration, and diabetes mellitus duration. In Model 3, adjustments were made for age, gender, hypertension duration, diabetes mellitus duration, BMI, fatty liver, ALT, AST, creatinine, and urea to account for potential confounding factors. The results of each model were compared to assess the consistency of the effect estimates under varying levels of covariate adjustment, thereby providing insights into the robustness of the impact. For each multivariable model we computed the area under the receiver-operating-characteristic curve (AUC) and compared AUCs using the DeLong method to quantify the discriminative ability of the respective exposure for predicting incident RD.

We then perform a restricted cubic spline analysis of cum UHR against incident RD, the absolute decline in eGFR, and the decline rate in eGFR from 2020 to 2023. The absolute decline in eGFR from 2020 to 2023 = eGFR₂₀₂₀-eGFR₂₀₂₃. The decline rate in eGFR from 2020 to 2023 =[(eGFR₂₀₂₀-eGFR₂₀₂₃)/eGFR₂₀₂₀]*100%.

Subjects were then stratified into four cum UHR quartiles: Q1 (≤8.40, ≤25th percentile), Q2 (8.41–10.49, 26th–50th percentile), Q3 (10.50–13.29, 51st–75th percentile), and Q4 (≥13.30, ≥76th percentile). We compared among these quartiles the eGFR values, absolute eGFR decline value, and the decline rate of eGFR from 2020 to 2023. eGFR decline value₂₀₂₀ = eGFR₂₀₂₀-eGFR₂₀₂₀. eGFR decline value₂₀₂₁ = eGFR₂₀₂₀-eGFR₂₀₂₁. eGFR decline value₂₀₂₂ = eGFR₂₀₂₀-eGFR₂₀₂₂. eGFR decline value₂₀₂₃ = eGFR₂₀₂₀-eGFR₂₀₂₃. eGFR decline rate₂₀₂₀ = [(eGFR₂₀₂₀-eGFR₂₀₂₀)/eGFR₂₀₂₀]*100%. eGFR decline rate₂₀₂₁ = [(eGFR₂₀₂₀-eGFR₂₀₂₁)/eGFR₂₀₂₀]*100%. eGFR decline rate₂₀₂₂ = [(eGFR₂₀₂₀-eGFR₂₀₂₂)/eGFR₂₀₂₀]*100%. eGFR decline rate₂₀₂₃ = [(eGFR₂₀₂₀-eGFR₂₀₂₃)/eGF]R₂₀₂₀]*100%.

All statistical analyses were performed using R software (version 4.2.2, R Foundation for Statistical Computing, Vienna, Austria), along with the use of MSTATA software (https://www.mstata.com/). P < 0.05 was considered statistically significant.

Results

Baseline Characteristics

A total of 650 participants were enrolled (mean age 71.7 ± 5.5 years; 299 [46.0%] men). The RD group comprised 151 individuals (71 [47.0%] men; mean age 73.4 ± 6.4 years), whereas the Non-RD group included 499 participants (228 [45.7%] men; mean age 71.2 ± 5.2 years).

Compared with the Non-RD group, the RD group exhibited significantly higher values for age, diabetes mellitus duration, cum TyG, UHR, cum URH, cum AIP, fasting plasma glucose, serum creatinine, blood urea, uric acid, and urinary protein (all P < 0.05). Conversely, eGFR, HDL-cholesterol, and LDL-cholesterol were significantly lower in the RD group (all P < 0.05). No significant between-group differences were observed for the remaining parameters (P > 0.05). Details are provided in Table 1.

Table 1 Basic Clinical Data of the 2 Groups

Association Between Various Factors and Rapid Kidney Function Decline

Univariate and multivariate logistic regression analyses were then performed to evaluate the associations of TyG, cum TyG, TyG-BMI, cum TyG-BMI, UHR, cum UHR, AIP, cum AIP, SPISE index and cum SPISE index with incident rapid kidney function decline (RD).

In univariate analysis, cum TyG, UHR, cum UHR and cum AIP were significantly associated with incident RD (all P < 0.05). After adjustment for potential confounders, multivariate logistic regression revealed that UHR (P < 0.001), cum UHR (P < 0.001) and cum AIP (P = 0.015) remained independently related to incident RD (Table 2 and Figure 1).

Table 2 Univariate and Multivariate Analysis of Various Factors

Figure 1 Forest plot of TyG, cum TyG, TyG-BMI, cum TyG-BMI, UHR, cum UHR, AIP, cum AIP, SPISE index and cum SPISE index association with incident rapid kidney function decline by Logistic regression analysis. TyG, triglyceride-glucose index; cum, cumulative average; BMI, body mass index; UHR, uric acid/ high-density lipoprotein cholesterol ratio; AIP, atherogenic index of plasma; SPISE, single point insulin sensitivity estimator.

Multivariable logistic regression revealed that cum TyG, TyG-BMI, cum TyG-BMI, UHR, cum UHR and cum AIP were all independently associated with incident RD (P < 0.05, Table 3). Among these exposures, cum UHR—analysed as either a categorical or continuous variable—achieved the highest discriminative performance, with an area under the receiver-operating-characteristic curve (AUC) of 0.724 (95% CI 0.675–0.773) and 0.711 (95% CI 0.660–0.762), respectively (Table 4 and Figure 2), indicating the strongest relation with incident RD. After adjustment for multiple covariates, higher quartiles of cum UHR remained robustly associated with increased odds of incident RD relative to the lowest quartile (Q1): Q2 OR = 5.11 (95% CI 2.19–11.92), Q3 OR = 7.28 (95% CI 3.08–17.19), and Q4 OR = 12.06 (95% CI 4.96–29.33). Consistently, when considered as a continuous variable, per 1 SD rise in cum UHR was significantly associated with incident RD (OR 1.20, 95% CI 1.12–1.28) (Table 3).

Table 3 Association Between Various Factors and Rapid Kidney Function Decline

Table 4 Area Under Curve of Various Factors

Figure 2 Receiver operating characteristic curve of various factors for incident rapid kidney function decline. (A) TyG as continuous variable; (B) TyG as categorical variable; (C) cum TyG as continuous variable; (D) cum TyG as categorical variable; (E) TyG-BMI as continuous variable; (F) TyG-BMI as categorical variable; (G) cum TyG-BMI as continuous variable; (H) cum TyG-BMI as categorical variable; (I) UHR as continuous variable; (J) UHR as categorical variable; (K) cum UHR as continuous variable; (L) cum UHR as categorical variable; (M) AIP as continuous variable; (N) AIP as categorical variable; (O) cum AIP as continuous variable; (P) cum AIP as categorical variable; (Q) SPISE index as continuous variable; (R) SPISE index as categorical variable; (S) cum SPISE index as continuous variable; (T) cum SPISE index as categorical variable. Red curve: Model 1, adjusted for age and gender. Blue curve: Model 2, adjusted for age, gender, hypertension duration, and diabetes mellitus duration. Green curve: Model 3, adjusted for age, gender, hypertension duration, diabetes mellitus duration, BMI, fatty liver, ALT, AST, creatinine, and urea to account for potential confounding factors. Abbreviations: TyG, triglyceride-glucose index; cum, cumulative average; BMI, body mass index; UHR, uric acid/ high-density lipoprotein cholesterol ratio; AIP, atherogenic index of plasma; SPISE, single point insulin sensitivity estimator; ALT, alanine aminotransferase; AST, aspartate aminotransferase; AUC, area under curve.

Association Between Cum UHR and Kidney Function Decline

When cum UHR was further examined using restricted cubic spline (RCS) modeling against incident RD, the absolute eGFR decline, and the eGFR decline rate from 2020 to 2023, all associations attained high statistical significance (P-overall < 0.001) (Figure 3). RCS curves revealed non-linear relationships of cum UHR with both incident RD (P-nonlinear = 0.008) and the absolute eGFR decline (P-nonlinear = 0.040) (Figure 3A and B). In contrast, the association between cum UHR and the decline rate in eGFR appeared to be linear (P-nonlinear = 0.124) (Figure 3C).

Figure 3 Associations between cum UHR with incident rapid kidney function decline (A), eGFR decline value (B) and rate (C) from 2020 to 2023 by Restricted Cubic Spline. Cum UHR, cumulative average uric acid to high-density lipoprotein cholesterol ratio; eGFR, estimated glomerular filtration rate.

Annual eGFR Decline Across Cum UHR Quartiles

We then stratified all patients into quartiles (Q1–Q4) based on their cum UHR levels and examined annual eGFR changes. From 2020 to 2023, eGFR declined in every patient, with the steepest drop observed in the Q4 group (Figure 4A). Both the absolute eGFR decline and the annual decline rate increased year-over-year, again most pronounced in Q4 (Figure 4B and C).

Figure 4 eGFR value (A) eGFR decline value (B) and eGFR decline rate (C) from 2020 to 2023. eGFR, estimated glomerular filtration rate.

Discussion

In this retrospective cohort analysis, we examined the associations of multiple insulin-resistance and metabolism-related indices with rapid kidney function decline (RD) in hypertensive patients aged ≥ 60 years who also had type 2 diabetes mellitus (T2DM). After adjustment for a comprehensive set of confounders, the cumulative average uric acid to high-density lipoprotein cholesterol ratio (cum UHR) emerged as the strongest predictor of RD.

RD is a well-established harbinger of end-stage kidney disease and is especially prevalent among older individuals harboring overlapping metabolic risk factors such as hypertension and diabetes mellitus.²⁴ A recent Indonesian multicenter cross-sectional study reported a 14.8% prevalence of chronic kidney disease among patients with coexisting diabetes mellitus and hypertension; notably, almost half of these individuals—despite seemingly preserved baseline kidney function—were stratified as being at moderate-to-high or very high risk of adverse renal outcomes.²⁵ These data underscore the substantial renal burden borne by this vulnerable population.

The triglyceride-glucose index (TyG) is a simple yet robust surrogate marker of insulin resistance (IR),²⁶ a core pathogenic driver of renal impairment in patients with T2DM and hypertension.²⁷ IR exerts renal damage through multiple mechanisms, including glomerular hyperfiltration, systemic and intrarenal inflammation, and progressive tubulointerstitial fibrosis.²⁸ TyG-BMI, an integrated metric of both IR and overall adiposity, has been shown to outperform either component alone. In a large prospective cohort leveraging UK Biobank data, TyG and its obesity-adjusted derivatives (TyG-BMI and TyG-waist circumference) were independently and dose-dependently associated with incident cardiovascular disease (CVD),²⁹ a frequent comorbidity that accelerates chronic kidney disease (CKD) and, conversely, is exacerbated by declining kidney function. Low circulating high-density-lipoprotein cholesterol (HDL-C) not only reflects impaired reverse cholesterol transport but also denotes attenuated anti-inflammatory and antioxidant capacities, both of which predispose to renal microvascular injury.³⁰ The atherogenic index of plasma (AIP), calculated as Log10(TG/HDL-C), quantifies the atherogenic potential of the lipid milieu and has emerged as a powerful predictor of cardiorenal events.³¹ In recent years, it has been found that the SPISE Index is also closely associated with IR in both children and adults.¹⁹ Similarly, the uric acid to HDL-C ratio (UHR) integrates pro-oxidant and pro-inflammatory signals driven by hyperuricemia with the vascular protective deficit implied by reduced HDL-C. This “dual-risk” characterization may confer superior discriminative value for early renal microangiopathy compared with either biomarker alone.³² Collectively, these inexpensive and universally available indices are intimately linked to incident and progressive renal dysfunction.¹² Accurate renal-risk stratification remains a research priority. Epidemiological evidence consistently demonstrates that older adults with an estimated glomerular filtration rate (eGFR) of 45–59 mL/min/1.73 m², even in the absence of albuminuria, carry a markedly higher risk of kidney failure, major cardiovascular events, and all-cause mortality.³³

In the present study, UHR, cum UHR and cum AIP were all independently associated with incident RD (Table 2 and Figure 1), with cum UHR displaying the strongest effect size (Table 4 and Figure 2). Moreover, cum UHR exhibited an approximately linear relationship with the rate of eGFR decline, whereas its association with the absolute eGFR decrement was non-linear (Figure 3), implying that cum UHR may capture the dynamic trajectory of renal function loss more sensitively. Consistently, higher cum UHR values corresponded to steeper eGFR slopes (Figure 4). Compared with uric acid or HDL-C alone, UHR has demonstrated superior sensitivity and specificity for predicting metabolic syndrome³⁴ and incident diabetes mellitus;^35,36 however, its relationship with kidney function remains comparatively under-explored. A 2023 cross-sectional survey of 4551 Chinese men with T2DM and post-menopausal women (mean age 67.40 ± 8.71 years) identified UHR as an independent correlate of both CVD and CKD, an association that persisted after adjustment for age, gender and BMI.³⁷ Similarly, a retrospective Turkish study of 287 patients with diabetes mellitus reported that UHR was independently linked to diabetic kidney injury and correlated significantly with serum creatinine, eGFR and urinary albumin-to-creatinine ratio (UACR).³⁸ Collectively, these data indicate a robust connection between UHR and renal impairment in diabetic populations. Our cohort—composed of elderly individuals with co-existent hypertension and T2DM—extends these observations by demonstrating that both baseline UHR and, more strongly, cum UHR predict incident RD. cum UHR, reflecting the average metabolic milieu over the preceding four years, presumably provides a more stable representation of chronic pathophysiological stress than a single measurement, thereby enhancing its predictive capacity. As an inexpensive and readily calculable index, cum UHR is ideally suited for deployment in primary-care settings to flag high-risk individuals at an early, potentially modifiable stage. Given the escalating global burden of CKD, risk-stratification tools anchored to cum UHR could facilitate large-scale community screening and timely initiation of renoprotective interventions.

Several inherent limitations warrant cautious interpretation. The single-center, retrospective design inherently constrains external validity, as the cohort was confined to a single residential community; despite stringent inclusion criteria, subtle selection biases may persist. Although we adjusted for a comprehensive panel of established confounders, the potential for residual confounding—particularly from unmeasured lifestyle factors such as dietary patterns and physical activity—cannot be entirely excluded. Moreover, practical constraints restricted the follow-up duration to four years, precluding insights into longer-term trajectories and rare late-onset events. Future validation through large-scale, multicenter prospective cohorts—integrating high-resolution dietary, exercise, and behavioral data—will be essential to corroborate and extend these preliminary findings.

Conclusions

For elderly patients (aged >60 years) with both hypertension and type 2 diabetes mellitus, cum UHR levels were significantly associated with rapid kidney function decline, suggesting that cum UHR may serve as a promising biomarker.