Predicting Major Bleeding in Native Kidney Biopsies: From External Validation of the Universal Bleeding Score to a New Clinical Tool

Vollständiger Abstract

Worum geht es in dieser Arbeit?

Objectives: A percutaneous native kidney biopsy remains the gold standard for diagnosing glomerular, tubulointerstitial and vascular kidney diseases. Although generally safe, it may cause major haemorrhagic complications. Using the French national registry, Kaczmarek et al. developed a simplified bleeding risk score (anaemia, female sex, heart failure, acute kidney injury; range: 0–5), achieving an AUC of 0.755 in native biopsies. We aimed to externally validate this score in a Romanian cohort and to assess whether additional clinically relevant predictors could improve discrimination for major bleeding. Methods: We conducted a retrospective cohort study of all consecutive adults undergoing an ultrasound-guided percutaneous native kidney biopsy at a tertiary nephrology centre in Romania between January 2008 and December 2024. The primary outcome was a composite major bleeding event: clinically significant haematoma or haemorrhage, blood transfusion, angiographic intervention or nephrectomy. The Kaczmarek score was validated using a receiver operating characteristic (ROC) analysis. Candidate predictors were assessed by multivariable logistic regression, and a simplified score was derived from the regression coefficients. Results: Among 3081 patients, 162 (5.3%) experienced major bleeding. The Kaczmarek score showed modest discrimination (AUC: 0.624, 95% CI: 0.585–0.663). In the multivariable analysis (n = 2506 complete cases), anaemia, heart failure, solid neoplasm and fibrinogen were independently associated with major bleeding. An eight-component score (anaemia, heart failure, solid neoplasm, and fibrinogen < 511 mg/dL; female sex, hypertension, liver disease, and eGFR < 30 mL/min/1.73 m2) achieved an AUC of 0.676 (95% CI: 0.631–0.717), outperforming the Kaczmarek score. Conclusions: The French score performed only modestly in our cohort. Adding comorbidities, kidney function and haemostatic parameters, particularly fibrinogen, improved the risk discrimination.