Clinical and histopathological predictors of rapid kidney function decline in patients with biopsy-proven diabetic kidney disease : Chinese Medical Journal

Secondary Logo

Journal Logo


Clinical and histopathological predictors of rapid kidney function decline in patients with biopsy-proven diabetic kidney disease

Zhao, Yiyang1,2; Chang, Dongyuan1,2,; Wu, Liang1,2; Yu, Xiaojuan1,2; Wang, Suxia1,2; Zhao, Minghui1,2; Chen, Min1,2

Editor(s): Li, Jinjiao; Ji, Yuanyuan

Author Information
Chinese Medical Journal 136(10):p 1243-1245, May 20, 2023. | DOI: 10.1097/CM9.0000000000002673

To the Editor: Diabetic kidney disease (DKD) has become the leading cause of end-stage kidney disease (ESKD). According to the Kidney Disease Improving Global Outcomes (KDIGO) guidelines, rapid kidney function decline in DKD is defined as an estimated glomerular filtration rate (eGFR) slope ≤-5 mL∙min -1∙1.73 m -2∙year -1. The prognoses of patients who experience rapid eGFR decline are usually poor. In a multicenter retrospective study, 230 of 377 (61%) biopsy-proven DKD patients were rapid decliners. [1] It was reported that 32% and 42% of 195 type 2 diabetes (T2D) Pima Indians with normoalbuminuria and microalbuminuria, respectively, had a rapid eGFR decline, indicating that proteinuria alone has difficulty in fully predicting rapid kidney function decline in DKD patients with T2D. [2] Therefore, it is of interest to identify more useful clinicopathological predictors of a rapid kidney function decline in T2D patients.

The current study included 467 T2D patients with kidney biopsy-proven DKD from Peking University First Hospital between January 1, 2009 and December 31, 2019. The definition of T2D was according to the American Diabetes Association criteria in 2017. [3] Among the 467 patients, 222 patients with combined non-diabetic kidney disease (NDKD), 50 patients who were lost to follow-up or whose follow-up time was <3 months, and 14 patients with baseline eGFR <15 mL∙min -1∙1.73 m -2 were excluded. Finally, 181 T2D patients with kidney biopsy-proven DKD were included in this study. The study was conducted according to the Declaration of Helsinki and the approval of the Ethics Committee of Peking University First Hospital (No. 2017-1280). Each participant signed a written informed consent form.

Clinical data included sex, age, body mass index (BMI), diabetic retinopathy (DR), duration of diabetes, systolic blood pressure (SBP), diastolic blood pressure (DBP), duration of hypertension, proteinuria per 24 h, urinary albumin-to-creatinine ratio (uACR), urinary N-acetylglutamate synthase (NAG), baseline eGFR, serum creatinine (Scr), hemoglobin A1c (HbA1c), platelet count, hemoglobin (Hb), white blood cell (WBC), serum albumin, serum uric acid, triglyceride (TG), high-density lipoprotein (HDL) cholesterol, low-density lipoprotein (LDL) cholesterol, serum C3, serum C4, and the use of renin–angiotensin–aldosterone system (RAAS) inhibitors and lipid-lowering medications.

Glomerular lesions were evaluated with the Renal Pathology Society (RPS) class I–IV, Kimmelstiel–Wilson (K–W; grade 0–1), subendothelial spacing (grade 0–1), and mesangiolysis/microaneurysm (grade 0–1). [4] For tubulointerstitial lesions, interstitial fibrosis and tubular atrophy (IFTA; 0, none; 1, <25%; 2, 25–50%; 3, >50%) and interstitial inflammation (0, none; 1, infiltration only related to IFTA; 2, infiltration in areas without IFTA) were evaluated. [4] The vascular lesions were evaluated based on the presence of arteriosclerosis (grade 0–2) and arteriolar hyalinosis (grade 0–2). [4] For direct immunofluorescence indices, the intensities of staining of immunoglobulins, including IgG, IgA, and IgM, complements, including C3 and C1q, albumin (Alb), and fibrinogen-related antigen, were scored on a scale of 0–4+. Electron microscopy was performed to rule out some renal morphologic lesions similar to DKD. All renal biopsies were scored independently by two pathologists in an unbiased manner who were blinded to the demographic and clinical data. Repeated reviews of any scoring difference between two pathologists were performed until a consensus was reached.

The eGFR decline slope was calculated as (eGFR [follow-up]–eGFR [baseline])/(follow-up time). According to the KDIGO guidelines, non-rapid decliners were defined as eGFR decline slope >-5 mL∙min -1∙1.73 m -2∙year -1 and rapid decliners were defined as eGFR decline slope ≤-5 mL∙min -1∙1.73 m -2∙year -1.

The kidney endpoint was defined as ESKD or a 40% reduction in eGFR. ESKD was defined as the initiation of hemodialysis/peritoneal dialysis, renal transplantation, or death caused by uremia. [5] Patients who did not reach the endpoint at the last visit were censored during the analysis.

Categorical data are expressed as absolute frequencies and proportions. To compare the non-normally and normally distributed continuous data between two groups, Mann-Whitney U tests and unpaired Student's t-tests were used, respectively. The chi-square test was used to compare two groups of categorical data. The independent predictors for rapid eGFR decline were analyzed by multivariate logistic regression analysis. Variables associated with the risk of rapid eGFR decline ( P <0.05) were selected by the forward likelihood ratio method. The results are shown as odds ratios (OR) and 95% confidence interval (CI). Receiver operating characteristic (ROC) curves, likelihood ratio test (LRT), Hosmer–Lemeshow goodness-of-fit test, the area under the curve (AUC), continuous/category-free net reclassification improvement (NRI >0), and absolute integrated discrimination improvement (IDI) were used to evaluate the statistical power of the models. All data were analyzed by RStudio (v4.1.2; R Development Core Team, Vienna, Austria), IBM Statistical Package for Social Sciences (v26.0; IBM SPSS, Chicago, IL, USA).

The general data of these two groups are presented in Supplementary Tables 1 and 2 ( Rapid eGFR decliners showed a significantly higher frequency of DR than non-rapid decliners (87/108 [81%] vs. 37/73 [51%], P <0.01). Compared with non-rapid decliners, the proteinuria of rapid decliners was significantly worse (5.2 [3.4, 8.1] vs. 2.3 [0.8, 4.1] g/24h, P <0.01). Rapid decliners had significantly lower Hb and serum albumin than non-rapid decliners (112.7 ± 19.3 vs. 120.1 ± 25.2 g/L, P = 0.03; 31.3 [27.8, 35.2] vs. 37.8 [30.4, 41.3] g/L, P <0.01, respectively). Compared with non-rapid decliners, the proportion of RPS class III and class IV in rapid decliners was significantly higher (85/108 [79%] vs. 41/73 [56%], P <0.01). The proportion of K–W nodules and mesangiolysis/microaneurysm in rapid decliners was significantly higher than that in non-rapid decliners (84/108 [78%] vs. 36/73 [49%], P <0.01; 87/108 [81%] vs. 29/73 [40%], P <0.01, respectively). For interstitial and tubular injuries, rapid decliners had a significantly higher proportion of advanced grades of IFTA (grade 2 and 3) than non-rapid decliners (94/108 [87%] vs. 45/73 [62%], P <0.01).

The multivariate logistic analysis of the three models is shown in Table 1. In Model 3, after adjusting for the confounders in Model 1, we found that a higher baseline eGFR, mesangiolysis/microaneurysm, and IFTA were independent risk predictors of a rapid eGFR decline (OR and 95% CI: 1.40 [1.12–1.76], P <0.01; 5.40 [2.37–12.29], P <0.01; and 2.92 [1.46–5.86], P <0.01, respectively; Table 1). Compared with Model 1, Model 3 had significant improvement in the model fit (ΔLRT χ 2 = 37.09, P <0.01), calibration (Hosmer–Lemeshow test, P = 0.78), discrimination (AUC increase from 0.65 to 0.80, P <0.01), sensitivity (decreased from 90% to 42%) and specificity (increased from 39% to 63%; Table 1). Furthermore, the risk classification of Model 3 was significantly improved (overall NRI and 95% CI: 0.83 [0.57–1.18]; Table 1). The IDI also showed that Model 3 had a significant enhancement in predicting probabilities compared with Model 1 (IDI and 95% CI: 19% [13.10–24.80%]; Table 1).

Table 1 - Performance of the clinical, clinical plus proteinuria, and clinical plus proteinuria and histopathological variables prediction models for rapid eGFR decline.

Clinical Model 1

( N = 177)

Clinical Model 2

(Model 1 + proteinuria)

( N = 177)

Clinical and histopathological Model 3

(Model 2 + histopathological variables)

( N = 177)

Male 1.15 (0.53, 2.47), 0.73 1.01 (0.46, 2.20), 0.99 0.85 (0.36, 2.01), 0.71
Age (years) 0.98 (0.95, 1.01), 0.15 1.00 (0.96, 1.03), 0.65 1.02 (0.98, 1.06), 0.31
Higher baseline eGFR (10 mL∙min -1∙1.73 m -2) 1.02 (1.01, 1.06), 0.78 1.11 (0.95, 1.30), 0.21 1.40 (1.12, 1.76), <0.01
SBP (mmHg) 1.02 (1.00, 1.04), 0.04 1.01 (0.99, 1.03), 0.21 1.02 (0.99, 1.04), 0.20
Duration of diabetes (months) 1.00 (1.00, 1.01), 0.07 1.00 (1.00, 1.01), 0.09 1.00 (1.00, 1.01), 0.27
Hb (g/L) 0.98 (0.97, 1.00), 0.05 0.98 (0.97, 1.00), 0.06 0.99 (0.97, 1.01), 0.48
Proteinuria (g/24h) 1.17 (1.05, 1.29), <0.01 1.09 (0.97, 1.21), 0.14
IFTA 2.92 (1.46, 5.86), <0.01
Mesangiolysis/microaneurysm 5.40 (2.37, 12.29), <0.01
Performance measure
LRT χ 2 test 13.54 (<0.01) 23.57 (<0.01) 50.62 (<0.01)
ΔLRT χ 2 test 10.03 (<0.01) 37.09 (<0.01)
H–L test χ 2 17.52 (0.03) 16.63 (0.03) 4.75 (0.78)
Sensitivity (%) 0.90 0.93 0.42
Specificity (%) 0.39 0.33 0.63
Positive predictive value (%) 0.68 0.67 0.63
Negative predictive value (%) 0.74 0.77 0.43
AUC 0.65 (0.56, 0.73) 0.71 (0.63, 0.80) 0.80 (0.73, 0.86)
ΔAUC 0.06 (0.03) 0.15 (<0.01)
NRI >0 0.59 (0.24, 0.89) 0.83 (0.57, 1.18)
NRIR 0.08 (-0.08, 0.31) 0.47 (0.28, 0.67)
NRINR 0.50 (0.24, 0.63) 0.36 (0.20, 0.60)
IDI (%) 6.20 (3.00, 9.40) 19.00 (13.10, 24.80)
Only participants with complete data were included in each model. The most concise model included the basic clinical variables (clinical Model 1). Next, proteinuria was added to Model 1 (clinical plus proteinuria Model 2). Then all significant clinicopathological variables that could predict rapid eGFR decline were forced into the clinical model (clinical plus proteinuria and pathological variables Model 3). Values are given as OR (95% CI) or P value. AUC: Area under the curve; CI: Confidence interval; eGFR: Estimated glomerular filtration rate; Hb: Hemoglobin; H–L: Hosmer–Lemeshow; IDI: Integrated discrimination improvement; IFTA: Interstitial fibrosis and tubular atrophy; LRT: Likelihood ratio test; NRI: Net reclassification improvement; NRINR: NRI in non-rapid decliners; NRIR: NRI in rapid decliners; OR: Odds ratios; SBP: Systolic blood pressure; –: Not applicable.

In the current study, a higher baseline eGFR was an independent clinical predictor for rapid eGFR decliners with T2D. Consistently, a multicenter and cross-sectional study including 334 diabetes patients with rapid eGFR decline had similar results. [6] We also found that mesangiolysis/microaneurysm was another independent histopathological predictor. In this study, mesangiolysis/microaneurysm instead of K–W lesions could better predict rapid kidney function decline, which was not fully consistent with the results of Furuichi et al[1] that both K–W lesions and mesangiolysis/microaneurysm could predict a rapid eGFR decline in type 1 diabetes(T1D) and T2D patients. This difference may be explained by the heterogeneity between T1D and T2D patients, as not all T2D patients have histopathological patterns resembling the typical presentations of the RPS class in T1D patients. In addition, a recent study reported that the nodular changes in db/db mice caused by mesangiolysis derived from chronic hypoxia were different from the classical K–W lesions bearing concentric fibrosis. [7] A state of chronic hypoxia is one of the important contributors to DKD in T2D, which to some extent explains our results that mesangiolysis/microaneurysm rather than K–W lesions could predict a rapid eGFR decline better in T2D patients. A previous study in a European cohort showed that glomerular basement membrane (GBM) thickening as an early pathological change of DKD also had a good predictive ability for kidney endpoints. [8] However, our study did not quantitatively calculate the GBM thickness of the patients; whether it can predict a rapid kidney function decline still needs to be explored in the future.

Several limitations were also identified. Since this was a single-center study with a relatively small sample size, multicenter studies with a larger sample size are needed. In addition, in our center, kidney biopsy was usually applied to patients with diabetes who were suspected of having NDKD, and it was usually not performed in DKD patients with only decreased eGFR without macroalbuminuria. Whether our results could be extrapolated to a broader population of T2D patients with DKD needs further investigation.

In conclusion, the current study has identified that higher baseline eGFR, mesangiolysis/microaneurysm, and IFTA are independent predictors of a rapid eGFR decline in DKD patients with T2D.


This study is supported by grants from the National Natural Science Fund (Nos. 82070748, 82090020, and 82090021), the National Key Research and Development Program (Nos. 2022YFC2502500 and 2022YFC2502502) China International Medical Foundation Renal Anemia Fund, and grants from National High-Level Hospital Clinical Research Funding (Nos. 2022SF01 and 2022XW01).

Conflicts of interest



1. Furuichi K, Shimizu M, Yamanouchi M, Hoshino J, Sakai N, Iwata Y, et al. Clinicopathological features of fast eGFR decliners among patients with diabetic nephropathy. BMJ Open Diabetes Res Care 2020;8: e001157. doi: 10.1136/bmjdrc-2019-001157.
2. Pavkov ME, Knowler WC, Lemley KV, Mason CC, Myers BD, Nelson RG. Early renal function decline in type 2 diabetes. Clin J Am Soc Nephrol 2012;7: 78–84. doi: 10.2215/CJN.07610711.
3. Marathe PH, Gao HX, Close KL. American diabetes association standards of medical care in diabetes 2017. J Diabetes 2017;9: 320–324. doi: 10.1111/1753-0407.12524.
4. Tervaert TW, Mooyaart AL, Amann K, Cohen AH, Cook HT, Drachenberg CB, et al. Pathologic classification of diabetic nephropathy. J Am Soc Nephrol 2010;21: 556–563. doi: 10.1681/ASN.2010010010.
5. Yamanouchi M, Hoshino J, Ubara Y, Takaichi K, Kinowaki K, Fujii T, et al. Clinicopathological predictors for progression of chronic kidney disease in nephrosclerosis: A biopsy-based cohort study. Nephrol Dial Transplant 2019;34: 1182–1188. doi: 10.1093/ndt/gfy121.
6. Yoshida Y, Kashiwabara K, Hirakawa Y, Tanaka T, Noso S, Ikegami H, et al. Conditions, pathogenesis, and progression of diabetic kidney disease and early decliner in Japan. BMJ Open Diabetes Res Care 2020;8: e000902. doi: 10.1136/bmjdrc-2019-000902.
7. Takahashi N, Yoshida H, Kimura H, Kamiyama K, Kurose T, Sugimoto H, et al. Chronic hypoxia exacerbates diabetic glomerulosclerosis through mesangiolysis and podocyte injury in db/db mice. Nephrol Dial Transplant 2020;35: 1678–1688. doi: 10.1093/ndt/gfaa074.
8. Stefan G, Stancu S, Zugravu A, Petre N, Mandache E, Mircescu G. Histologic predictors of renal outcome in diabetic nephropathy: Beyond renal pathology society classification. Medicine (Baltimore) 2019;98: e16333. doi: 10.1097/MD.0000000000016333.

Supplemental Digital Content

Copyright © 2023 The Chinese Medical Association, produced by Wolters Kluwer, Inc. under the CC-BY-NC-ND license.