Background Abnormal nodal disease is radiologically defined by a short axis (SA) dimension of ≥10 mm, and is the basis of NICE recommendations for sampling using EBUS. We routinely sample nodes down to 5 mm. We examined the utility of this.

Method We retrospectively assessed EBUS nodal diagnoses for all patients referred (including suspected cancer and non-cancer diagnoses). Procedures were performed under light-moderate sedation in ˜98% of cases. A systematic nodal examination was performed prior to sampling, and nodes with a SA of ≥5 mm were sampled when thought to be clinically relevant for decision-making. Samples were obtained using 4 needle passes with an EBUS-22G-O Cook needle. Samples were placed directly into formalin and processed as cell blocks. Samples were defined as non-representative (no lymphoid or diagnostic material), borderline (Pauci-lymphoid material) and representative (good nodal sampling ±diagnostic material).

Results Between November 2014 and June 2018, 763 EBUS were performed, of which 487 (64%) were for suspected primary lung cancer. Of 1553 nodes sampled, the overall non-representative, borderline and representative sampling was seen in 122 (7.9%), 86 (5.5%) and 1344 (86.5%) respectively. In 764 (49%) nodes a positive result was obtained.

Nodal short axis measurements were recorded in 1459 (94%) nodes with a median size of 13 mm (interquartile range 10–20 mm). Of 361 (24.7%) nodes<10 mm, 44 (12.1%) were non-representative, 29 (8.0%) were borderline and 288 (78.9%) were representative. Of the representative 288 nodes, 80 (27.8%) yielded a positive diagnosis, of which 61 (21.2%) were malignant.

Across size difference <7 mm and 7 to <10 mm nodes had a representative yield in 63/91 (69%) and 225/270 (83%) respectively, and were malignant in 10/91 (11%) and 45/270 (16%) respectively.

In the 487 patients with a suspected primary lung cancer 245/1017 (24%) nodes sampled were <10 mm, with representative sampling in 80% and a malignant result in 17.6%.

Conclusions Systematic examination and nodal sampling of ≤10 mm lymph nodes provides representative tissue in 80% of patients, while in lung cancer patients, 1 in 6 demonstrated malignant change.