From Granulomas to Malignancy: Validating the Sydney System in a Tuberculosis-Endemic Landscape.

BACKGROUND

The Sydney System for reporting lymph node cytopathology standardizes lymph node fine-needle aspiration cytology (FNAC) reporting and improves risk-based clinical stratification. Validation data from tuberculosis (TB)-endemic regions remain limited.

MATERIALS AND

METHODS

This retrospective observational multicentric study analyzed 427 consecutive lymph node FNAC cases from two tertiary care centers in Eastern India during January-December 2025. Three independent pathologists, reviewing slides at separate time points, categorized aspirates according to the Sydney System (L1-L5), blinded to clinical details and to each other's assignments. Histopathology and Ziehl-Neelsen staining served as reference standards. Diagnostic accuracy parameters with 95% confidence intervals (Wilson score method), receiver operating characteristic (ROC) analysis, category-wise risk of malignancy (ROM), TB correlation with a complete 2 × 2 matrix, and interobserver agreement (Cohen's kappa) were assessed.

RESULTS

Cervical lymph nodes constituted 357 (83.6%) cases. Reactive, granulomatous, and necrotizing lymphadenitis accounted for the majority of diagnoses; malignant lesions accounted for 39 cases (9.1%). The L2 (benign/reactive) category constituted the overwhelming majority of cases, with a mean of 384 (89.9%) across pathologists out of 427 cases. Per-pathologist sensitivity ranged from 79.5% to 84.6%; specificity was 99.7% across all three observers; positive predictive value was 96.9-97.1%; negative predictive value was 98.0-98.5%; and overall accuracy was 97.9-98.4%. ROC analysis demonstrated excellent discriminatory performance (AUC range: 0.935-0.974). ROM increased progressively from L2 to L5. Interobserver agreement was substantial to almost perfect (κ = 0.79-0.87). Among L2 cases, the sensitivity for TB capture was 98.6%; however, the PPV of L2 for TB was only 19.2%, and specificity was 13.0%, with 307 (80.8%) cases of granulomatous L2 findings attributable to non-TB aetiologies.

CONCLUSIONS

The Sydney System demonstrates excellent diagnostic accuracy, reproducibility, and clinically meaningful ROM stratification in a TB-endemic setting. Granulomatous cytomorphology alone is insufficient for TB diagnosis; microbiological confirmation is mandatory before assigning an L2 classification to TB.