Ultrasound interpretive consistency in radial endobronchial ultrasound: a prerequisite for interpreting the gains of multimodal biopsy

Dear Editor, we read with interest the recent study by Lee et al. in the Journal of Thoracic Disease evaluating a multimodal biopsy strategy for peripheral lung lesions (PLLs) under radial endobronchial ultrasound (rEBUS) guidance, in which larger forceps biopsy (LFB) or needle aspiration (NA) (PeriView FLEX transbronchial needle aspiration) was added to transbronchial lung biopsy (TBLB) (1). The use of diagnostic tissue acquisition rate (DTAR) alongside diagnostic yield (DY) is particularly valuable, because tissue adequacy has become increasingly relevant to immunohistochemistry and molecular profiling in contemporary lung cancer care, consistent with recent American Thoracic Society (ATS)/American College of Chest Physicians (ACCP) recommendations (2). The reported improvement in both DY and DTAR with combined sampling supports the practical value of a multimodal approach (1). Our concern is not with the clinical rationale for supplementary sampling, but with the interpretive foundation of the rEBUS classification used to support both tissue-adequacy prediction and intraprocedural sampling decisions.

In this study, the distinction between a “within-lesion” pattern and an “adjacent/eccentric within” pattern—especially when coupled with a 2.5-mm probe-to-lesion margin threshold—was central to the analytic framework (1). Yet, this distinction is not a purely static or self-evident one. In routine practice, it depends on dynamic image review, lesion-probe motion, and operator judgment during live scanning. This matters because the rEBUS category was not merely descriptive: it was the only independent predictor of diagnostic tissue acquisition by TBLB [adjusted odds ratio (aOR), 5.008; P<0.001], and it also informed whether operators proceeded with LFB or NA (1). Despite this central role, the study did not report inter-observer agreement, predefined criteria for image capture and freezing, or a standardized approach to measuring the 2.5-mm threshold (1). Prior reports suggest that variation in rEBUS image interpretation can influence sampling success, while additional sampling beyond rEBUS-guided TBLB may improve diagnostic performance (3,4). Without clearer documentation of classification reliability, however, it becomes difficult to determine how reproducibly this upstream imaging variable can be applied across operators and centers.

This issue also affects how the comparative findings should be read. The choice of LFB or NA was not randomized; rather, it was made during the procedure according to rEBUS appearance, bronchoscope outer diameter, guide sheath feasibility, and operator discretion (1). As a result, the adjunctive-modality groups were not fully comparable at baseline. Under these conditions, the observed gains in the LFB arm are likely to reflect two intertwined components: the contribution of the device itself and the cumulative effect of an added sampling opportunity after TBLB. That distinction is important, because prior studies have also shown improved yield when conventional sampling methods are added after rEBUS-guided biopsy (4). The subgroup structure in the present study further limits modality-specific inference, with only 6 patients in the TBLB + NA group and 25 in the TBLB + LFB + NA group (1). Such small and uneven subgroups are especially susceptible to procedural selection effects, including preferential use of NA in technically less favorable settings. For this reason, we interpret the study mainly as evidence supporting the utility of a multimodal workflow, rather than as a definitive comparison demonstrating the superiority of LFB over NA.

The recommendation that LFB should be routinely considered after TBLB regardless of rEBUS visualization subtype should therefore be applied with caution (1). In many procedures, performing LFB requires removal of the guide sheath, after which real-time positional confirmation from rEBUS is no longer maintained. Under those circumstances, access to the target and stability of sampling may reasonably differ according to the original visualization pattern. More broadly, the generalizability of a single-center experience using a specific endoscope-device combination and shaped by local operator practice remains uncertain (5). In our view, the next methodological step is not simply to compare more devices, but to standardize the imaging framework on which those device choices depend: acquisition and freezing criteria, rules for final classification during dynamic scanning, measurement methods for close-distance thresholds, and routine reporting of inter-rater agreement. With those elements in place, prospective matched or randomized studies would be better positioned to clarify the independent contributions of LFB and NA and the relationship among rEBUS classification, sampling allocation, and tissue adequacy (3,4).

In summary, Lee and colleagues provide valuable real-world evidence that supplementary sampling can improve the performance of rEBUS-guided biopsy for PLLs (1). At the same time, the interpretive consistency of rEBUS classification deserves more explicit attention when that classification serves as both a predictor of TBLB adequacy and a trigger for downstream sampling choices. Stronger standardization at this imaging step would make subsequent conclusions about multimodal biopsy gains more robust and more generalizable.

Supplementary

The article’s supplementary files as

Ethical Statement: The authors are accountable for all aspects of the work in ensuring that questions related to the accuracy or integrity of any part of the work are appropriately investigated and resolved.