Comparison of Renal Function Decline in Proteinuric and Non-Proteinuric Diabetic Kidney Disease: A Retrospective Single-Centre Study

Introduction

Diabetes mellitus is a leading cause of CKD globally and accounts for a substantial proportion of patients progressing to end-stage kidney disease (ESKD).^1,2 Diabetic kidney disease (DKD) has classically been described as a progressive condition characterized by increasing albuminuria, declining GFR, and heightened cardiovascular risk. Albuminuria has therefore been regarded as both a diagnostic hallmark and a strong predictor of renal disease progression.³

This paradigm has been challenged by recognition of a non-proteinuric DKD (NP-DKD) phenotype. Population-based studies have demonstrated that ∼20–40% of patients with diabetes and reduced estimated GFR (eGFR) do not exhibit significant albuminuria.^4-6 This phenotype is believed to reflect predominantly vascular, tubulointerstitial, orischemic renal injury rather than classic glomerular diabetic lesions.⁷

The prognostic implications of NP-DKD remain incompletely defined. While several studies report a slower rate of renal function decline in NP-DKD compared with P-DKD, cardiovascular morbidity and mortality appear to be comparable or, in some cohorts, higher.^8,9 The increasing prevalence of NP-DKD has been attributed to widespread use of renin-angiotensin system blockade, improved blood pressure and glycemic control, introduction of sodium-glucose cotransporter 2 (SGLT2) inhibitors, an ageing diabetic population, and survival bias allowing patients to live long enough to develop vascular-dominant CKD.^10-12

In Sri Lanka, diabetes and CKD represent major and growing public health challenges. Community-based studies estimate the prevalence of CKD among Sri Lankan adults to be ∼14%, with diabetes mellitus and hypertension identified as leading contributors outside regions affected by CKD of unknown aetiology.¹³ Hospital-based and registry data indicate that diabetes accounts for ∼36–60% of non-dialysis-dependent patients with CKD attending tertiary nephrology services.^14,15 Despite this substantial burden, local data describing DKD phenotypes and their renal progression remain limited, with most studies focusing on advanced or overtly proteinuric disease.

This retrospective single-center study aimed to compare renal function decline between P-DKD and NP-DKD. The primary objective was to compare the decline in renal function between patients with P-DKD and NP-DKD attending the Nephrology Clinic of Sri Jayewardenepura General Hospital between August 2017–2019.

Materials and Methods

This retrospective, descriptive, single-center study was conducted at the Nephrology Clinic of Sri Jayewardenepura General Hospital, Colombo, Sri Lanka. Patients with DKD registered between August 2017 and August 2019 were eligible for inclusion.

DKD was defined as the presence of diabetes mellitus (type 1 or type 2) accompanied by persistent abnormalities in kidney structure or function, including reduced (eGFR <60 mL/min/1.73 m²) and/or proteinuria, in the absence of alternative primary renal diseases.

Proteinuria was defined as urinary protein excretion ≥300 mg/day or a urinary albumin-to-creatinine ratio (UACR) ≥300 mg/g. Persistence was confirmed by at least two measurements obtained ≥3 months apart. Proteinuria assessment was based on either 24-hour urinary protein quantification or spot urine UACR, depending on clinical availability. Patients were classified according to baseline proteinuria status into P-DKD and NP-DKD groups, in accordance with KDIGO 2022 Clinical Practice Guidelines. Baseline classification was adopted to minimize misclassification due to temporal variability in proteinuria.

A total of 400 patients were initially screened. Of these, 98 were excluded due to incomplete baseline data (n = 42), inadequate follow-up duration of less than 12 months (n = 38), or alternative renal diagnoses (n = 18). The final study cohort comprised 302 patients, including 182 with P-DKD and 120 with NP-DKD [Supplementary Table 1].

Participants were followed for a median duration of 24 months (interquartile range [IQR] 18–24 months). Renal function was assessed using serial serum creatinine measurements, and eGFR was calculated using the CKD-EPI equation at ∼3–6 monthly intervals during routine clinical care. The median number of eGFR measurements per patient was six.

Renal function decline was evaluated as the absolute change in eGFR over the follow-up period. Given the variability in measurement timing inherent to retrospective data, slope-based or mixed-effects modelling was not performed. While this approach does not provide precise estimates of progression rates, it allows consistent comparison across groups. Accordingly, findings should be interpreted as an overall change in renal function rather than the true trajectory of decline.

Baseline cardiovascular comorbidities, including ischemic heart disease (IHD), cerebrovascular accident (CVA), and peripheral vascular disease (PVD), were recorded at study entry and were not assessed as incident outcomes. IHD was defined as documented myocardial infarction, angina, prior coronary revascularization, or physician-diagnosed coronary artery disease. CVA was defined as a prior ischemic or hemorrhagic stroke confirmed clinically and/or radiologically. PVD was defined as documented peripheral arterial disease or prior vascular intervention. These data were obtained from clinical records, discharge summaries, and documented diagnoses in patient charts.

Collected data included demographic characteristics, duration and control of diabetes, comorbidities, medication use, and microvascular and macrovascular complications. Continuous variables are presented as mean ± standard deviation, and categorical variables as frequencies and percentages. Baseline characteristics between P-DKD and NP-DKD groups were compared using SMD, with values >0.2 considered indicative of meaningful imbalance [Table 1].

To account for potential confounding, a multivariable linear regression model was constructed to evaluate the association between proteinuria status and change in eGFR. Clinically relevant covariates were selected a priori and included age, sex, baseline eGFR, duration of diabetes mellitus, blood pressure, use of renin-angiotensin-aldosterone system inhibitors (RAASi), and presence of diabetic retinopathy. Given the retrospective design and variability in measurement timing, the adjusted analysis was intended to assess the direction and consistency of associations rather than precise effect estimates.

Ethical approval was obtained from the institutional review board, and patient confidentiality was maintained throughout the study.

Results

Patients with NP-DKD were older (68 ± 9 vs. 62 ± 10 years, p <0.001) and had a lower prevalence of diabetic retinopathy (18% vs. 42%, p <0.001). Duration of diabetes was longer in P-DKD (14 ± 6 vs. 11 ± 5 years, p <0.01). Glycemic control was suboptimal in both groups, with no statistically significant difference in HbA1c (8.1 ± 1.2% vs 7.8 ± 1.1%, p = 0.08).

Hypertension prevalence was similar between groups; however, Patients with NP-DKD demonstrated better blood pressure control and more favorable lipid profiles. They also had lower baseline rates of IHD (12% vs. 25%, p = 0.01) and cerebrovascular disease (7% vs. 18%, p = 0.02).

Baseline renal function was higher in NP-DKD (eGFR 38 vs 30 mL/min/1.73 m², p <0.001). Over two years, renal decline was slower in NP-DKD (38 → 30 mL/min/1.73 m²) compared with P-DKD (30 → 24 mL/min/1.73 m²), p <0.001. Patients with NP-DKD also had higher serum albumin levels and fewer microvascular complications.

Although the absolute decline appeared greater in the P-DKD group, a formal statistical comparison of the rate of eGFR decline between groups was not performed. Therefore, no definitive conclusions can be made regarding differences in progression rates. Renal outcomes were worse in P-DKD, with a lower baseline eGFR (∼30 mL/min/1.73 m²) declining to ∼24 mL/min/1.73 m² at 2 years, indicating more rapid progression to advanced CKD. In contrast, patients with NP-DKD had a higher baseline eGFR (∼38 mL/min/1.73 m²) declining to ∼30 mL/min/1.73 m², consistent with slower disease progression.

A multivariable linear regression model was constructed to assess factors associated with change in eGFR over the follow-up period, including age, sex, baseline eGFR, duration of diabetes, systolic blood pressure, RAASi use, and diabetic retinopathy. After adjustment, proteinuria remained independently associated with a greater decline in renal function (β = −3.8 mL/min/1.73 m²; 95% CI: −5.6 to −2.0; p <0.001). Increasing age, longer diabetes duration, and higher systolic blood pressure were also associated with greater eGFR decline [Table 2]. RAASi use was associated with a smaller decline in eGFR; however, this finding should be interpreted cautiously due to potential confounding by indication. Given variability in measurement timing, results are presented as overall change in renal function rather than slope-based progression estimates.

Progression in P-DKD was primarily associated with higher baseline albuminuria, hypertension, and dyslipidemia. In contrast, progression in NP-DKD appeared more closely related to older age, vascular comorbidities, and blood pressure control. These findings suggest that while P-DKD is driven predominantly by glomerular injury, NP-DKD may reflect a more vascular phenotype, highlighting the importance of aggressive risk factor modification in both groups.

Discussion

This study provides one of the first analyses of P-DKD and NP-DKD in a Sri Lankan tertiary care population. Our findings support the well-established role of proteinuria as a predictor of renal function decline, consistent with international evidence.^3,6,8

Patients with NP-DKD were older, had better baseline renal function, and demonstrated slower renal decline over two years. This supports the concept that NP-DKD represents a predominantly vascular or tubulointerstitial phenotype rather than classic glomerular injury.⁷ Although Patients with NP-DKD in this cohort had a lower baseline prevalence of cardiovascular disease, their older age suggests a potentially higher long-term vascular risk, underscoring the need for continued cardiovascular risk assessment and management.

The observed association between RAAS inhibitor use and slower decline in eGFR is likely influenced by confounding by indication and should not be interpreted as causal. Nevertheless, the findings highlight the importance of optimizing vascular and metabolic risk factors in NP-DKD, where progression may be driven more by age, hypertension, and dyslipidemia than by proteinuria alone.

These observations are consistent with current KDIGO recommendations supporting individualized risk stratification and management across DKD phenotypes.^11,12 Substantial baseline imbalances between groups (standardized mean difference >0.4 for age, RAAS inhibitor use, and diabetic retinopathy) indicate a high risk of confounding. Although multivariable adjustment was performed, residual confounding is likely and limits causal interpretation of the observed associations.

This study has several limitations. First, its retrospective single-center design limits generalizability and introduces potential selection bias. Second, renal function decline was assessed as a change in eGFR rather than slope-based modeling due to variability in measurement timing. Using overall change in eGFR rather than slope-based modeling limits precise estimation of renal progression and may introduce bias from baseline differences and regression to the mean. Third, residual confounding remains possible despite multivariable adjustment, particularly given baseline imbalances between groups. Fourth, data on key variables such as body mass index, smoking status, and medication adherence were unavailable. Fifth, hard renal outcomes such as progression to ESKD or mortality were not assessed. The lack of kidney biopsy data precludes definitive characterization of underlying histopathology.

In conclusion, P-DKD and NP-DKD may represent heterogeneous clinical phenotypes within the same disease spectrum. Proteinuria is a marker of renal progression. These findings need confirmation in larger studies and may inform risk stratification and management in DKD.