We found a good diagnostic yield based on 18-month follow-up data and using ENB as a stand-alone diagnostic tool. This number is in the range of reported pooled yields (58.6–84%),6,23,29,33,35,39,40,45,46,56−59 it represents an average assessment of ENB. Our diagnostic yield is good when taking into account that our mean nodule diameter was smaller than those described in diagnostic ENB studies (Supplementary Table 5). 6,23,29,33,35,39,40,45,46,56−59
Several studies are similar to ours, since they used only ENB for all of their cases.45,56−59 For example, Bertoletti showed a yield of 77.4% but with a much larger nodule diameter (31.2 vs. 16 mm).45 Ozgul et al. examined ENB yield in 56 cases, but r-EBUS was used in 26 of those cases.39 The yield for non-r-EBUS cases was 71.4%, which is similar to our study. Further, Raval and Amir reported an 83.3% yield with a relatively small lesion size (19.3 mm).46 Although only ENB was used, they utilised a tidal volume expiration mapped ENB (Veran Medical Technologies), which limits the comparison.
Some of our independent yield predictors have been previously identified in studies with multivariate analysis.14,16,18,20,22,27,49 For example, smoking history increased our yield by 7-fold. Ost et al. also reported a smoking association, less strong.14 The effect of smoking might be related to the localised distortion effect that it generates at the bronchial architecture level, triggered by the chronic inflammatory effect.47 This could facilitate locatable guide access to nearby solitary pulmonary nodules. The presence of a bronchus sign on CT increased the yield by 4-fold. Seijo et al. reported such an association with an 8-fold increase in yield.42 Our study confirms their findings, as a bronchus sign increased performance by 17% compared to patients without bronchial signs in our sample. Since bronchus sign was only present in about 50% of the patients, for cases without bronchus sign, it is our practice to make strategies, such as modifying the location of the locatable guidewire on at least two occasions, while obtaining transbronchial biopsies to avoid the “all-or-none” diagnostic phenomenon.14 On this wise, samples are collected from various regions close to the lesion. We have yet to verify this fact, but we believe that it could mitigate the effect of not presenting the bronchus sign.
PLLs in the lower lobes decreased our yield by close to 80%.16,22 This could be explained by diaphragmatic movement during inspiration, with a difference in PLLs location of up to 2.5 cm.22.60 Lesion diameter of > 20 mm in its largest axis14,49 and > 30 mm18 correlated with higher diagnostic yield in other studies. For us, a lesion size of ≥ 15 mm in the Z axis increased the yield close to 5-fold. We postulate that a larger lesion on the Z-axis might provide better endobronchial exposure. This requires additional validation. Perifissural lesions decreased the diagnostic yield; this might occur because of a reduction in bronchus size and a more tortuous airway pattern, limiting the advance of the locatable guide. Age of ≥ 75 increased our ENB yield by 4.8-fold. The rationale for this effect is unclear.
One of the clinical factors that negatively modify diagnostic yield is presenting a personal history of previous extrathoracic cancer before ENB.61 Those patients had a higher percentage of metastatic PLLs compared to those without a history of extrathoracic cancer: 25% (13/51) vs. 4.5% (5/109), respectively. In turn, PLLs of metastatic origin decreased the yield by 85%, with an OR of 0.15. Two studies assessed a history of extrathoracic cancer previous to ENB and suspicion of metastatic PLLs as factors decreasing the diagnostic yield of ENB.22,23 We believe the reasons for the decrease in profitability are that most metastases of tumours at the lung level are due to hematogenous dissemination and, additionally, to the development of a metastatic niche that provides the adequate microenvironment for the implantation and growth of disseminated tumour cells.61,62 Tsuboi et al.63 documented a significant difference in the bronchoscopy yield of peripheral lung lesions secondary to primary bronchogenic malignancies versus lung metastases, at 76.5% versus 29.1%, respectively. They found that bronchial airway exposure was present in only 5.1% of the metastases < 2 cm in size. Pulmonary metastases follow a hematogenous spread and are surrounded by non-malignant tissue (fibroblasts, neovasculature, inflammatory cells, and extracellular matrix).61,62 Such dissemination patterns compared to those of bronchogenic carcinoma nodules, plus limited endobronchial exposure, might explain the lower yield observed.63,64 We believe that this fact is of great importance for ENB and at the level of bronchoscopy as a diagnostic technique in pulmonology. Finally, using the independent variables associated with the diagnostic yield, we generated a model to predict the diagnostic yield of ENB with good discriminating capacity (area under the ROC curve: 0.83). We plan to validate this model in a future prospective study.
We also assessed the diagnostic performance of the tools and techniques used during ENB.14 Combination of multiple sampling techniques, particularly TBBx and TBNA, positively impacted the diagnostic yield and a diagnostic yield of 69% was reached when the most common individual techniques were used together: suction catheter + TBBx + BAL + CB + TBNA + bronchial washing (Table 4 and Fig. 2). Chao et al. also noticed a significant yield increase (18%) when TBNA was added to r-EBUS (78.4%) compared to TBBx and bronchial washing without TBNA (60.6%).65 In general, TBNA appears to be underutilised,14 even in cases with pleural distance of ≥ 10 mm. This is likely due to technical difficulties manoeuvring the needle in more distal locations and to concerns about a higher pneumothorax risk. However, in our study, its use was not associated with an increased incidence of pneumothorax. Most needle-associated pneumothorax risk have been extrapolated from CT-guided TTNA data (pneumothorax as high as 23%, up to ¼ requiring chest tube drainage).66 We believe that TBNA is a safe tool for lesions ≥ 10 mm from the pleura, as seen in our study and recently confirmed in the large multicentre NAVIGATE study.10
Finally, since our study used stand-alone ENB under moderate sedation, it is possible that the routine use of additional diagnostic tools or general anaesthesia could increase the yield of ENB.14,22,29,33 For example, Eberhardt et al. reported an 88% yield for combined ENB + r-EBUS versus r-EBUS (69%) or ENB (59%) alone.22 Our study bears several limitations, including its single-centre, retrospective observational nature, which exposes it to the risk of an unmeasured confounder and might limit the generalisability of the results. Further, our samples were analysed by the same pathologist. This can introduce bias to the diagnostic yield of various sampling techniques once an initial sample is diagnostic. We did not use fluoroscopy, r-EBUS, or ROSE, which could have potentially increased our diagnostic yield; therefore, our results apply mainly to studies not using such techniques.