Endobronchial ultrasound-guided transbronchial mediastinal cryobiopsy in the diagnosis of mediastinal lesions: safety, feasibility and diagnostic yield – experience in 50 cases

Discussion

This study tested and analysed the diagnostic yield, feasibility and safety of TMC as a novel procedure for the diagnosis of mediastinal and hilar lesions. Zhang et al. [14] conducted a randomised trial that included a total of 197 patients who underwent EBUS-TBNA and TMC in the same procedure to assess the diagnostic yield and safety of this technique. For TMC they performed a small incision in the tracheobronchial wall adjacent to the mediastinal lesion using a high-frequency needle-knife; the knife was then replaced by the cryoprobe and biopsies were performed by cooling down for 7 s. In their trial the overall diagnostic yield was 79.9% and 91.8% for TBNA and TMC, respectively, while TMC was more sensitive than TBNA in uncommon tumours (91.7% versus 25%) and benign disorders (80.9% versus 53.2%) [14]. In our study the overall diagnostic yield was 82% and 96% for TBNA and TMC, respectively.

An important difference in our report is the way we perform the procedure. We have shown that the high-frequency needle-knife is not essential; we eliminated this step of the process by directly introducing the 1.1-mm cryoprobe always guided by the ultrasound image through the puncture site created by the EBUS-TBNA needle, allowing us to perform the procedure in a faster way [13]. The overall mean±sd procedure time in our series was 33.96±5.86 min (range 21–42 min). In the Zhang et al. [14] trial using the high-frequency needle-knife, the overall mean±sd procedure time was 31.9±9.1 min. In our study the mean±sd TMC procedure time was 9.54±2.10 (range 6–15 min); we believe these times seem acceptable in consideration of the benefits of a more intact and bigger biopsy sample, which avoids repetitions of EBUS-TBNA and invasive techniques such as mediastinoscopy.

Since EBUS-TMC is a new technique, only 18 cases have been published in the literature to date without using the high-frequency needle-knife and without relevant complications [13–16]. One of the perceived limitations of the EBUS technique is the small size obtained by the 22-G needle. The small amount of material obtained makes it difficult for the pathologist to diagnose benign and haematological diseases. Previous studies have demonstrated that EBUS-guided mediastinal forceps biopsy compared with TBNA improves specimen quality and diagnostic yield, particularly in patients with sarcoidosis and lymphoma [2, 17–20]. However, one of the largest prospective trials of transbronchial forceps biopsy reported a lower diagnostic sensitivity of this technique compared with TBNA in malignant lymph nodes [19].

The diagnostic value of EBUS-TBNA, which provides groups of cells for a cytological evaluation instead of tissue for histological analysis in the majority of the cases, is less well established in patients with sarcoidosis or other benign diseases [21–23]. The reason for a suboptimal diagnostic hit rate seems to be the small specimen size by EBUS-TBNA that does not allow adequate pathological analysis because the general tissue architecture cannot be evaluated. Previous studies have reported that EBUS-TBNA can have a diagnostic yield in sarcoidosis of 54–93%, at least in experienced centres [22, 23]. A meta-analysis by Agarwal et al. [9] in 553 patients suggests an excellent overall diagnostic yield of EBUS-TBNA in this pathology. In our study, 12 patients were diagnosed with sarcoidosis, with a concordance between TBNA and TMC of 100%. Although multiple studies have shown potentially limited ability of EBUS-TBNA to diagnose sarcoidosis, we demonstrated no significant difference in diagnostic yield between both techniques regarding this condition. It is important to mention that our pathologists reported that the quality of the samples obtained by TMC was optimal for obtaining a definitive diagnosis with greater certainty.

Lymphoid neoplasia is diagnosed on the basis of the World Health Organization (WHO) classification system. The WHO recommends the use of multidimensional diagnostic modalities, including cytomorphological studies, immunophenotyping, cytogenetic analyses and molecular studies to accurately subtype lymphoma [24, 25]. Surgical excision and core biopsy are the current preferred sampling techniques for diagnosing lymphoma, but there is growing interest in using less invasive endoscopic techniques with lower complication rates [4]. In patients who present with intrathoracic adenopathy that is suspicious for lymphoma, EBUS-TBNA is in most cases the first diagnostic option, given the risks of surgical sampling [26]. The inherent disadvantages of EBUS-TBNA include lesser sampling of core tissue and an inferior negative predictive value [27]. The low-volume samples can be a major concern, particularly in diagnosing marginal zone and follicular lymphoma [5], in which studies have suggested a high level of discordance between cytological and histological specimens [28]. This has led many authors to conclude that EBUS-TBNA is not usually a clinically appropriate diagnostic tool if lymphoma is suspected [29]. Several studies have demonstrated that the use of mini-forceps guided by EBUS increases the diagnostic yield with respect to Hodgkin disease, non-Hodgkin lymphoma and sarcoidosis [2, 30]. However, these studies have been performed mainly in subcarinal nodes, and the blunt forceps did not pass the perforated mucosa and penetrate the lymph node in all cases. In our study, TMC established the diagnosis of lymphoma in all patients in whom this pathology was suspected (100%) and all cases were successfully subclassified; meanwhile, with TBNA no definitive diagnosis of lymphoma was reached. Although there were only four cases of lymphoma in our study, the mediastinal cryobiopsy was able to avoid EBUS-TBNA repetition and the potential risks of mediastinoscopy in this group of patients.

EBUS-TBNA has an excellent yield in normal sized and enlarged mediastinal and hilar lymph nodes, with a sensitivity of 84–94% [21, 31, 32]. With specific regard to NSCLC, the number of potentially actionable oncogenic drivers is constantly increasing. While the first acknowledged driver was the epidermal growth factor receptor, novel biomarkers of therapeutic interest are emerging, this requiring increasing quantity of tumour tissue for testing multiple biomarkers at once. This is especially relevant considering the role that next-generation sequencing (NGS) is acquiring in current clinical practice; in this context, the tissue collected at diagnosis needs to be adequate both in terms of quantity and quality, in order to prevent repeated procedures [33]. Retrospective data on a small subset of cancer patients suggest that cryobiopsy performed on peripheral lesions might provide more adequate samples for NGS compared with EBUS-TBNA on mediastinal stations, allowing to analyse more genes [34]. In this study, the diagnostic yield of EBUS-TBNA was marginally lower than reported before in bigger studies. In our study, no differences were found in diagnostic yield between both techniques in the diagnosis of small cell lung carcinoma; however, TMC diagnosed six cases of lung adenocarcinoma and four cases of squamous cell lung carcinoma that had been previously classified by TBNA as insufficient sample (n=6), carcinoma of unknown origin (n=3) and unclassified nonsmall cell lung carcinoma (n=2). The two patients that had an insufficient diagnosis by TMC were diagnosed by TBNA (two lung adenocarcinomas). In all cases in which a definitive diagnosis of lung cancer was not obtained by TBNA, the cryobiopsy allowed optimal molecular testing to be performed.

We performed EBUS-TMC under conscious sedation and it was well tolerated by all patients, suggesting that moderate sedation is sufficient and safe for performing TMC. There were no relevant complications during the procedure, nor 24 h later, nor at the 2-week follow-up. The most predominant adverse event was minor bleeding and it was more frequent when performing TBNA. There was no pneumothorax, pneumomediastinum or mediastinitis in our study.

Several limitations need to be considered in the present study. In five cases we had to make five passes of TBNA prior to performing the cryobiopsy because it was not possible to penetrate the bronchial wall. This problem occurred in patients with suspected silicosis, where the bronchial wall is more rigid and cartilaginous. TMC was not performed in three lymph node stations (10L, 10R and 11Rs). In our series, we did not have the necessity to make a diagnosis at these lymph node stations, so its feasibility in these three stations is unknown. Two patients with lung cancer were not diagnosed by TMC, despite the acquisition of more material compared with TBNA. We believe that this may have been due to the fact that cryobiopsies were always conducted at the same site of the lesion and as such may have missed the affected tissue. Another probable limitation of the study is the lack of ROSE of the TBNA samples in all procedures, which could have improved the diagnostic yield of EBUS-TBNA, and also the fact that the pathologist was not blinded between the samples obtained by the two techniques.

A strength of our study is that it is the first report in which EBUS-TMC was performed in two centres without using the high-frequency needle-knife with this number of patients.