A PRE-OPERATIVE PREDICTIVE MODEL FOR THE CLASSIFICATION OF NEWLY DIAGNOSED RENAL MASSES LESS THAN 5 CM IN DIAMETER AS BENIGN OR MALIGNANT

Abstract

Objective: To develop a predictive model for preoperative differentiation between benign (B) and malignant (M) histology in patients with renal masses (RM) using recursive partitioning. Methods: We analyzed preoperative patient and tumour characteristics in 395 subjects who had surgery for RM suspicious for renal cell carcinoma. Results: The model predicted B vs. M histology with an overall accuracy of 89.6% (95% CI 86.2,92.5). It assigned patients with smaller tumours (<5.67cc) and a predominantly (>45%) exophytic component a high risk of B disease (52.6%). Patients with symptoms, larger tumours (>5.67cc) and larger endophytic component (>35%) have a 0% risk of B disease. Conclusion: B vs. M disease can be predicted accurately. This predictive accuracy is higher than that shown in renal biopsy series. It is hypothesized that for smaller and exophytic RMs, a biopsy is indicated. Symptomatic, larger and endophytic RMs should be removed without further investigations.