Beyond the epithelioid percentage: Adverse-feature burden for prognostic stratification of renal epithelioid angiomyolipoma — A cohort study and individual patient data meta-analysis

Renal epithelioid angiomyolipoma (EAML) has unpredictable malignant potential, and the prognostic value of the epithelioid proportion underlying the 80% diagnostic threshold is unproven. We analyzed an institutional cohort of 29 renal EAMLs and performed a systematic review with individual patient data (IPD) meta-analysis of 123 published cases assembled from 17 studies; four patients with synchronous metastases at presentation were excluded, leaving 119 patients (34 progressions) for the primary analysis (PROSPERO CRD420261460370). A prespecified five-item adverse-feature score (AFS: size >7 cm, high mitotic activity, marked nuclear atypia, vascular invasion, Ki-67 >10%) was compared head-to-head with operationalized Brimo, Nese, and Folpe criteria. In the institutional cohort (5 progressions, 4 deaths), epithelioid proportion did not stratify progression (log-rank p = 0.220; C-index 0.64–0.73) and was uncorrelated with adverse-feature burden; the AFS separated three tiers with 0%, 0%, and 45.5% progression (log-rank p = 0.0006; exact p = 0.0039) and captured all five progressors. These single-center estimates are hypothesis-generating. In the IPD primary cohort (34 progressions), the four-item morphologic approximation showed a stepwise risk gradient across tiers 0, 1, and ≥2 (3.2%, 14.3%, and 58.7%; three-tier log-rank p = 4.7 × 10 −10 ; C-index 0.819, 95% CI 0.764–0.876), with numerically higher discrimination than epithelioid percentage (C = 0.682), operationalized Brimo (C = 0.693), or Nese (C = 0.790) criteria, and was stable across sensitivity analyses, including reinclusion of the four synchronous-metastasis patients (n = 123; C = 0.825). Necrosis was strongly outcome-associated in the primary cohort (85% vs 48%) and is proposed as a candidate sixth item. Within the limits of a small single-center series and a case-report–dominated literature, we found no evidence that epithelioid percentage carries prognostic information in renal EAML; a simple adverse-feature count was associated with progression risk in a graded manner across both data sources. Prospective validation of the score is required before clinical use.