The value of endoscopic ultrasound after bronchoscopy to diagnose thoracic sarcoidosis

Sarcoidosis is a disease of unknown aetiology and affects 5–40 per 100,000 population, making it the most prevalent interstitial lung disease in the Western world 1. There is no single diagnostic test. Its diagnosis is based on a compatible clinical and/or radiological picture, supported by pathological evidence of noncaseating epithelioid cell granulomas in the absence of organisms or particles 1. The diagnosis of sarcoidosis is reasonably certain without biopsy only in patients with Löfgren's syndrome. Otherwise, a biopsy specimen should be obtained from an involved organ that is most easily accessed. Since pulmonary sarcoidosis is the most frequent form, bronchoscopy with tissue sampling is advised as the first step in obtaining a tissue-based diagnosis and to exclude possible alternative diagnoses. Flexible bronchoscopy is cheap and has a reasonable diagnostic yield, especially if transbronchial biopsy (TBB) specimens are taken 2. However, clinicians are frequently confronted with a nondiagnostic result following bronchoscopy, requiring a decision to be made as to whether surgical biopsy should be performed. Cervical mediastinoscopy, and, in some cases, video-assisted thoracoscopic surgery (VATS) procedures are regarded as the next diagnostic step following nondiagnostic bronchoscopy. Although surgery has a high yield 3, 4, it requires general anaesthesia, is costly, is associated with a certain morbidity 5 and invariably results in scars.

Recent reports show that, with either curvilinear transoesophageal endoscopic ultrasound (EUS) with fine-needle aspiration (FNA) (EUS-FNA) or endobronchial ultrasound (EBUS) with transbronchial needle aspiration (TBNA) (EBUS-TBNA), together referred to hereafter as E(B)US, noncaseating granulomas can be demonstrated upon sampling of the intrathoracic nodes and that, as such, these techniques can contribute to the diagnosis of sarcoidosis 6–10. Although these reports indicate feasibility, none of them assessed the value of these techniques in routine daily practice, where bronchoscopy remains the first diagnostic step 1. In addition, these reports invariably originate from tertiary care institutions, where E(B)US was performed by experts in the field, possibly resulting in an overestimation of the yield.

The present article reports on the largest prospective implementation trial, in which patients with suspected thoracic sarcoidosis in 15 hospitals were first sent for conventional bronchoscopy. Only if no definite diagnosis were obtained was the patient offered E(B)US (either EUS-FNA or EBUS-TBNA). It was hypothesised that the addition of E(B)US, after preceding nondiagnostic bronchoscopy, would result in an increase in diagnostic yield and that, using this strategy, the need for surgical biopsy could be reduced in a significant number of patients.

METHODS

RESULTS

Procedures and diagnoses

The procedures performed are summarised in a flow chart (fig. 1⇓). Bronchoscopy was performed in 121 (88%) patients, and resulted in a definite diagnosis in 57 (42%). In 16 (12%) patients, bronchoscopy was cancelled. A total of 80 (58%) patients had no definite diagnosis after bronchoscopy and were, therefore, scheduled for an E(B)US procedure. Either EUS-FNA or EBUS-TBNA was performed in 72 (90%) patients, resulting in a definite diagnosis in 47 (59%). The cancellation of planned bronchoscopy or E(B)US in 16 and eight patients, respectively, was related to a variety of reasons, including refusals upon second thoughts, technical failures and endoscopists judging bronchoscopy or E(B)US to have a very unfavourable risk/benefit ratio in particular cases. As a result, 33 (24%) patients were left without a definite diagnosis after bronchoscopy and E(B)US; 22 underwent a surgical procedure that resulted in a definite diagnosis. In 11 patients, surgical intervention was refused by the patient, or was thought to have no added value over a follow-up strategy.

• Download figure
• Open in new tab
• Download powerpoint

Fig. 1—

Procedures performed in the study population. E(B)US: endoscopic ultrasound in the form of either transoesophageal endoscopic ultrasound-guided fine needle aspiration (FNA) or endobronchial ultrasound-guided transbronchial needle aspiration.

With the current implementation strategy, a definite histopathological diagnosis was obtained in 126 (92%) patients of the study population. The final diagnoses are summarised in table 2⇓. There were 115 (84%) patients with sarcoidosis, whereas acid-fast bacilli were found in five. In six other patients, pneumoconiosis, alveolitis, lymphangitis carcinomatosa and aspecific lymphadenitis were found.

View this table:

• View inline
• View popup

Table 2—

Final pathological diagnosis of the mediastinal lymph nodes in the study population

As summarised in table 3⇓, bronchoscopy was performed mainly under local anaesthesia, although mild sedation was added in a quarter of procedures. TBNA was applied in 21%, whereas EBB was performed in 63%, as was TBB. In 19 patients, the endoscopist decided not to take a tissue sample except for a washing for microbiological and cytological analysis. A definite diagnosis was obtained by bronchoscopy in 57 patients. The majority of these diagnoses were obtained with TBB (72%), which did better than EBB (26%; p = 0.0003) even though both procedures were performed equally. The relative yield for all diagnoses per biopsy method was highest for TBB (41 diagnoses in 76 TBB procedures, or 54%), followed by TBNA (31%) and EBB (20%). Minor complications were encountered with bronchoscopy: minor bleeding, intolerance, and one pneumothorax, the latter being treated with simple manual aspiration.

View this table:

• View inline
• View popup

Table 3—

Procedural details of bronchoscopy

Table 4⇓ shows the procedural characteristics of EUS-FNA and EBUS-TBNA. The majority of procedures were performed under local anaesthesia with mild sedation, whereas general anaesthesia was applied in 10%. EBUS-TBNA was performed three times more than EUS-FNA. Mediastinal lymph nodes were sampled in 95%, whereas hilar lymph nodes alone were approached in only four patients. The yield of EUS-FNA was 94% as compared to 56% for EBUS-TBNA (p = 0.03). No complications were noted in the patients investigated using E(B)US.

View this table:

• View inline
• View popup

Table 4—

Procedural details of endoscopic ultrasound(E(B)US)

The surgical interventions performed in 22 patients were cervical mediastinoscopy in 19 (86%), VATS in two (9%) and open lung biopsy in one (5%). All surgical procedures resulted in a definite diagnosis. One patient developed mediastinitis and was treated with antibiotics, with a favourable course.

DISCUSSION

The most important finding of the present largest implementation trial ever on sarcoidosis is that, in patients with thoracic sarcoidosis, the proposed algorithm of bronchoscopy followed by E(B)US only in those cases in whom no definite diagnosis was obtained yields histological proof of the disease in 84%. Although 45% of the sarcoidosis patients were diagnosed with bronchoscopy, E(B)US provided the diagnosis in an additional 39% of cases.

According to the available guidelines of the ERS, ATS and World Association of Sarcoidosis and Other Granulomatous Disorders, bronchoscopy should be the first diagnostic step 11. Flexible bronchoscopy is a readily available, safe and well-tolerated procedure permitting several modalities of tissue sampling from different anatomical sites 2, 12. TBB has been suggested to be the method of preference, with yields ranging 40–90% in case series 13–15. EBB has added value, even without apparent endobronchial abnormalities 16. The same holds for blind TBNA, especially when added to TBB 17–19. Besides these procedures, bronchial washing is recommended for microbiological analysis, while bronchoalveolar lavage with CD4/CD8 counting is variable and less sensitive 20.

In the present study, it was found that, when bronchoscopy is offered as a first-line tool in a study population with a suspicion of having thoracic sarcoidosis, the overall diagnostic yield is 42%, and the sensitivity for sarcoidosis is 45%. This is at the lower end of what has been reported previously 2, and illustrates the erosion of diagnostic yield once a technique is widely implemented. A possible reason for this is that all of the aforementioned studies were performed in expert centres, where specialised endoscopists investigated very selected patients. This differs greatly from implementation studies, where a variable degree of expertise is present and where some of the procedures were even omitted or only partially carried out because of a variety of patient, endoscopist or technical reasons. In addition, it requires a great deal of expertise to perform all biopsy modalities during a single bronchoscopy session. It is also remarkable to note that only 76 patients underwent TBB; the same number underwent EBB and 26 TBNA. This shows that endoscopists do not always feel comfortable performing some of the biopsy procedures. Nevertheless, when per-protocol analysis of the yield of TBB in the current study was performed, it was found that, in 72% of the procedures, a definite diagnosis was obtained. For sarcoidosis alone, the sensitivity of TBB was 62%. This is a value comparable to those published before 12, 15. As reported recently, there was also no relationship found between TBB yield and the stage of sarcoidosis on chest radiography 15. The safety figures of bronchoscopy appeared very acceptable, with only one case of pneumothorax, which was cured with manual aspiration.

When bronchoscopy does not result in a definite diagnosis, and when no other accessible sites for biopsy are identified, the guidelines suggest that surgical biopsy may be indicated if there are readily identifiable radiological abnormalities 11. The finding of mediastinal adenopathies should, therefore, prompt biopsy by mediastinoscopy before VATS or even open lung biopsy 4. Although these procedures have a superior yield, they are costly and exhibit a comorbidity that cannot be denied. Given the above data obtained with bronchoscopy, it is clear that, in daily practice, new tools that could provide a definite diagnosis in a minimally invasive way are very welcome.

EUS-FNA and EBUS-TBNA are minimally invasive outpatient techniques that have shown good test characteristics in lung cancer staging 21, 22 and have been recommended in the guidelines 23, 24. By consequence, these techniques are also being implemented in non-academic hospitals. To date, a few series have shown that EUS-FNA and EBUS-TBNA can be used to demonstrate sarcoidosis 6–10 in up to 80–90% of highly selected cases. However, what remains unknown is their value following a nondiagnostic bronchoscopy procedure.

We found that E(B)US following negative bronchoscopy provided a definite diagnosis in 59% of the patients in whom a surgical procedure would otherwise have been considered. The sensitivity for diagnosing sarcoidosis was 71%. These figures are lower than those reported in the series in which EUS-FNA or EBUS-TBNA were evaluated mainly as a first-line tool for the diagnosis of sarcoidosis 6–10. The reasons for this are probably comparable to the reasons discussed for the yield of bronchoscopy in an implementation setting. The present calculation also takes into account the fact that not all patients underwent effective E(B)US for a variety of reasons, such as refusal of a second endoscopic procedure. When applying the per-protocol analysis, a sensitivity of E(B)US of 77% for diagnosing sarcoidosis following negative bronchoscopy was found. Recently, it was shown that EBUS-TBNA had a 30% higher yield for finding sarcoidosis than blind TBNA 25. Although calculated in a subgroup, it was found that EBUS-TBNA following false-negative blind TBNA also increased the yield by 30%. This indicates that preceding negative blind TBNA is not a reason to skip E(B)US in patients with presumed sarcoidosis.

The overall strategy, in which patients with sarcoidosis are investigated first with bronchoscopy and then with E(B)US only if the former does not yield the diagnosis, results in an overall sensitivity of 84%, representing an absolute increase of 39% above the yield of bronchoscopy. In the present study, the majority of the E(B)US procedures were performed under mild sedation, allowing the patient to leave the hospital after a short observation period. Although the safety of E(B)US has mainly been described in lung cancer populations, no serious complications were also noted in this series. We, therefore, conclude that the algorithm used here is valuable, safe and useful for daily practice.

Some points should be borne in mind when considering these data. First, a final histopathological diagnosis was pursued in all of the patients, although there were 11 (8%) patients in whom this was not achieved. This figure is probably very reasonable taking into account the study population (mainly patients with benign diseases) and the fact this is an implementation trial. Nevertheless, it could be that there were patients with sarcoidosis that remained undetected. Secondly, it should be noted that the presence of noncaseating epithelioid granulomas without necrosis is not diagnostic per se for sarcoidosis. The diagnosis can only be made by an integration of the clinicoradiological picture, histological data and exclusion of other identifiable causes of granulomatous diseases 1, 11. For example, sarcoid-like inflammation in lymph nodes near lymphomas or carcinomas 26, or in the context of histoplasmosis or tuberculosis, can be misdiagnosed as sarcoidosis. Conversely, histoplasmosis is virtually nonexistent in Northern Europe, and the currently applied methods proved to be sensitive for the diagnosis of mycobacterial disease, making the chance of misdiagnosis probably very low. Finally, the current data do not formally answer the question as to whether bronchoscopy should be continued or whether it is better to carry out or refer for immediate E(B)US when thoracic sarcoidosis is suspected. The only way of finding this out is to perform a randomised controlled trial with a direct comparison between the two. When thinking about this, it should, however, always be remembered that bronchoscopy is cheap and readily available because it is routinely taught to all chest physicians, which is not the case for E(B)US.

In conclusion, we propose a high-yield and safe diagnostic algorithm for patients with thoracic sarcoidosis requiring tissue confirmation, which states that they should first be investigated with bronchoscopy, followed by E(B)US only in those cases in whom no definite diagnosis is obtained.

Clinical Trial

This study is registered at ClinicalTrials.gov (trial number NCT00888212) .

Statement of interest

None declared.