Comparative performance of endobronchial ultrasound-guided sampling techniques in patients with mediastinal lesions: a network meta-analysis.

Comparative evidence on the diagnostic yield of endobronchial ultrasound (EBUS)-guided sampling techniques for mediastinal lesions remains limited. We conducted a network meta-analysis (NMA) to evaluate EBUS-guided transbronchial needle aspiration (EBUS-TBNA), intranodal forceps biopsy (EBUS-IFB), transbronchial needle core biopsy (EBUS-TBNB) and transbronchial mediastinal cryobiopsy (EBUS-TMC). Random-effects NMA was performed within Bayesian and frequentist frameworks. Frequentist estimates were reported as risk ratios (RRs) with 95% CIs and, where applicable, 95% predictive intervals (PrIs). Bayesian ranking was assessed using surface under the cumulative ranking curve (SUCRA) values. 35 studies involving 3751 patients were included. EBUS-TMC (RR 1.27, 95% CI 1.16 to 1.39; 95% PrI 0.89-1.81) and EBUS-IFB (RR 1.21, 95% CI 1.07 to 1.38; 95% PrI 0.84-1.75) may provide higher diagnostic yield than EBUS-TBNA, whereas EBUS-TBNB did not show a clear advantage (RR 1.10, 95% CI 0.98 to 1.23). However, all 95% PrIs crossed unity, suggesting uncertainty about reproducibility in future settings. SUCRA rankings favoured EBUS-TMC followed by EBUS-IFB, EBUS-TBNB and EBUS-TBNA but should not be interpreted as certainty of superiority. Subgroup analyses suggested greater benefits of EBUS-TMC and EBUS-IFB in benign diseases, sarcoidosis and lymphoma but similar performance in lung cancer. Most adverse events were mild to moderate, although EBUS-TMC had a higher reported adverse-event rate. EBUS-TMC and EBUS-IFB may improve diagnostic yield over EBUS-TBNA in selected diseases and experienced centres. Given the very low certainty of evidence, wide PrIs and exploratory SUCRA rankings, these findings should be viewed as hypothesis-generating rather than definitive evidence. CRD420251236286.