Nephrotic‑range proteinuria in patients with Type 2 diabetes mellitus (T2DM) can result from classic diabetic nephropathy or from non‑diabetic glomerular disease (NDGD). Distinguishing between these etiologies is clinically important because it affects prognosis and management, but confirmation requires kidney biopsy, which carries procedural risk and is not universally available. The study aimed to identify clinical predictors of NDGD in T2DM patients presenting with nephrotic‑range proteinuria and to develop a practical prediction score to help guide biopsy decisions.
This investigation was a single‑center retrospective study conducted at a tertiary hospital in Thailand. The study period spanned 2014 to 2022 and included T2DM patients who underwent native kidney biopsy for evaluation of nephrotic‑range proteinuria. Biopsy findings were categorized by the presence or absence of NDGD.
Among 289 T2DM patients who had native kidney biopsies for nephrotic‑range proteinuria, 142 patients (49.1%) had NDGD, alone or coexisting with diabetic nephropathy. The most common pathologic diagnosis within the NDGD group was IgA nephropathy. These results indicate that nearly half of biopsied T2DM patients with heavy proteinuria had a non‑diabetic glomerular process contributing to their presentation in this cohort.
Multivariable logistic regression was used to identify independent clinical features associated with NDGD. The predictors retained in the final model were:
In addition, higher hemoglobin was retained in the prediction model despite a borderline statistical association. These variables reflect readily available clinical data that can be collected without invasive testing.
Using the independent predictors from multivariable analysis, the authors derived a weighted clinical prediction score intended to estimate the likelihood that a T2DM patient with nephrotic‑range proteinuria has NDGD rather than pure diabetic nephropathy. The abstract does not provide the exact point assignments for each predictor; those details were not reported in the source abstract and would need to be consulted in the full article.
Discrimination of the prediction score was assessed using the area under the receiver operating characteristic curve (AUROC). The score demonstrated good discrimination with an AUROC of 0.85 (95% confidence interval, 0.80–0.89).
Example threshold performance reported in the cohort:
These thresholds illustrate the trade‑offs between sensitivity and specificity: a low threshold favors sensitivity and would identify most cases of NDGD but with low specificity, whereas a higher threshold improves specificity at the expense of sensitivity.
The study shows that NDGD is common among T2DM patients with nephrotic‑range proteinuria in this single‑center Thai cohort, and that a clinical prediction score using routinely available variables can discriminate between NDGD and diabetic nephropathy with good accuracy (AUROC 0.85). The score could potentially support selection of patients for kidney biopsy and prioritize those at higher likelihood of NDGD.
However, the derivation cohort was single‑center and retrospective. The abstract highlights the need for external validation before the score can be recommended for clinical decision‑making. The abstract does not report the exact point weights, calibration metrics, or subgroup performance; these details would need to be reviewed in the full text to assess generalizability and operationalization in other settings.
In this retrospective study of 289 T2DM patients undergoing kidney biopsy for nephrotic‑range proteinuria, nearly half had NDGD, most commonly IgA nephropathy. A multivariable clinical prediction score incorporating age, diabetes duration, diabetic retinopathy, hypertension history, eGFR, hematuria, and hemoglobin demonstrated good discrimination (AUROC 0.85) for identifying NDGD. The authors conclude that the score requires external validation before it can be used to guide biopsy decisions in routine practice.