Protocol for Developing and Validating a Multimarker-Clinical Prediction Model of SGLT2 Inhibitor-Induced Acute eGFR...

Introduction: SGLT2 inhibitors reduce renal composite endpoints and proteinuria, yet RCTs uniformly show an acute eGFR dip within 2 weeks to 2 months after initiation. However, demographic and clinical predictors of an acute eGFR dip demonstrate considerable heterogeneity across studies. This study aims to identify urinary protein biomarkers of this early eGFR dip and integrate them with routine variables to build a clinically actionable prediction model. Methods and analysis: This three-stage proteomics study includes retrospective discovery, prospective internal validation, and external validation cohorts (total n ≈ 600-700). DIA mass spectrometry will screen for urinary proteins associated with ≥10% eGFR decline at 1 month post-SGLT2i initiation in CKD stages 3-4. Top candidates (FDR < 10%, FC > 1.5, ion intensity > 1 × 10⁴, unique gene families) will be validated by ELISA. A LASSO-logistic regression model will integrate the top three proteins with seven routinely available clinical variables: age, BMI, diabetes status, heart failure, systolic blood pressure, baseline eGFR, and diuretic use. Model performance will be assessed using the C-statistic, NRI, IDI, and calibration metrics. Adaptive stopping rules are pre-specified. Ethics and dissemination: Approved by the Ethics Review Committee at Chinese PLA General Hospital (S2025-859-02, 2025KY126-KS002), all participants will provide written informed consent prior to enrollment, and the study will adhere to the Declaration of Helsinki. Data will be pseudonymized and stored securely according to institutional regulations. Findings will be published in peer-reviewed journals and presented at international nephrology conferences. Trial Registration: Registered Report Identifier: ChiCTR2600119772. Date of registration: 3 March 2026.