TBLCs are useful for the diagnosis of DPLDs
81 patients with TBLCs performed for the diagnosis of a DPLD were included between April 2015 and December 2019. Their main clinical characteristics are summarized in the table 1. Four TBLCs were performed in the most affected lobe with a mean size of 21 mm2 (ranging from 9 to 44 mm2) (Table 1). The quality of TBLC tissue samples was good to excellent. A specific pathological pattern was observed following TBLCs in 64% of the cases (84% including NSIP patterns). Complications following TBLCs were mainly hemorrhage (mild or moderate in a majority of the cases, severe in 5% of the cases) and pneumothoraxes (21%, a half of them requesting chest drainage) (Table 2).
From these 81 patients, an undetermined or NSIP pattern was observed in 13 and 16 patients respectively (Table 2). Comparison of patient's characteristics between patients with nondiagnostic TBLCs and the others showed no differences in term of HRCT pattern distribution but a significant difference in the size of the cryobiopsies. Indeed, the total surface area per patients of nondiagnostic specimens was smaller than those given a specific diagnosis (65 ± 24 vs 82 ± 27 mm2, mean ± SD; p < 0,01) (Table 1). We further determined that using a cut-off of a total surface area of 110 mm2, the risk to have nondiagnostic biopsies was lower than 5%.
SLB provides critical additional information in case TBLCs are inconclusive or show a pattern suggestive of a NSIP
Among the 29 patients without specific diagnosis following TBLCs, 7 did not consent to have a SLB (further classified as unclassifiable ILD), and 8 received a provisional diagnosis following a novel MDD (5 chronic HP, 2 IPAF, 1 idiopathic NSIP). A SLB was performed for the other 14 patients (six with a suspected NSIP pattern and eight without pathological diagnosis following TBLCs) (Fig. 1). As expected, SLB provides ten-fold bigger sample than TBLC (76 ± 27 mm2 for TBLCs vs 676 ± 307 mm2 for SLB, mean ± SD) and subpleural area was always present for pathological examination (Fig. 2). In 11 out of the 14 patients, a UIP pattern was observed in the SLB (79%) whereas a NSIP pattern was only confirmed in 1 of the 6 patients with a suspected NSIP on TBLC (17%). Finally, a HP pattern was observed in two other patients without pathological diagnosis following TBLC. Following another MDD discussion taking into account SLB data, the final diagnosis changed in 93% of those patients (83% of those with a NSIP pattern on TBLC). In other words, our data showed that if a SLB is performed in a patient with a NSIP pattern on TBLC, another diagnosis will be observed in 83% of the cases changing significantly the final diagnosis and eventually the subsequent treatment.
Inter-observer agreement and diagnostic confidence are higher for SLBs than TBLCs
Three ILDs experienced pathologists further retrospectively and blindly reviewed all biopsies from the 14 patients who benefited from both procedures. Each pathologist was asked to propose a most probable diagnosis, even with a low confidence level. Major determinants underlying differences between TBLCs and SLBs were depicted. This analysis demonstrated a higher diagnostic concordance among pathologists for SLB (86%) compared to TBLCs (43%) (Fig. 3). Moreover, the level of confidence of the histological diagnoses was much higher for SLBs than for TBLCs (high diagnostic confidence in 88% of SLB contrasting with low and very low level of confidence in 97% of TBLCs) (Fig. 3). The benefit to perform SLB after TBLC in this selected population was not only to improve the diagnostic confidence as it changed the final diagnosis in 60% of the cases in the retrospective pathological analysis (table 3). Accordingly, the overall agreement between paired TBLC and SLB for specific histopathological pattern was 40%. The major determinants underlying this benefit reported by the three pathologists were the size of the biopsies in 90%, a better representation of the subpleural area in 50% and the localization of the biopsies in different lobes in 23%.
Concerning the identification of a NSIP pattern on TBLC, there was only an agreement among the three pathologists for two patients. Of note, a pattern suggestive of a NSIP on TBLC was systematically proposed without high diagnostic confidence (Table 4). Surprisingly, SLB performed in these two patients only confirmed a NSIP pattern for one pathologist whereas a NSIP was proposed on SLB by another pathologist for which the corresponding TBLC was not suggestive of NSIP (Table 4).