Role of bronchoalveolar lavage in the diagnosis of acute exacerbations of idiopathic pulmonary fibrosis: a retrospective study

Today, BAL technique is standardized [9] and it is often used in the workup of AEX-IPF. Pesci et al. [10] recommended that BAL should be considered in all IPF patients with suspected infection, malignancy or AEX-IPF. Papanikolaou et al. [11], as well as Wuyts et al. [12] state that BAL should be performed if the patient can tolerate the procedure (DLCO > 30 % and P_aO₂ > 75 mmHg on supplemental oxygen). The official ATS/ERS/JRS/ALAT statement on pulmonary fibrosis does not give clear recommendations on the diagnostic workup for AEX-IPF [13]. Overall BAL is widely considered a part of the diagnostic workup of a patient with IPF presenting with acute respiratory failure, and is performed for nearly every evaluated patient that can tolerate it, although the predictive usefulness and safety of the procedure has not been fully elucidated.

Several other potentially useful roles of the BAL were recently entertained. It has been shown that BAL samples from some AEX-IPF patients have increased level of pepsin [14] and that treatment with proton pomp inhibitors might have a role in the prevention of exacerbations in these selected patients [15]. This suggests that some IPF exacerbations might be triggered by silent aspiration and those patients do not need treatment with broad spectrum antibiotics. Song et al. showed that measuring percentage of neutrophils in the BAL fluid can be a useful tool to discriminate between pulmonary infection and AEX-IPF but this practice has not been routinely recommended and needs further investigation [16]. If clinical suspicion for drug induced alveolitis or other specific etiology, BAL can be performed tailored to that specific diagnoses, in case the BAL fluid differential count would change the management. For AEX-IPF or infection, no such strong data is available, and one should not base treatment decisions on BAL fluid differential count. In addition, the percentage of neutrophils in the BAL fluid is increased during the AEX-IPF episodes compared to stable patients with IPF and controls, which makes this data less reliable to exclude infectious process [17, 18]. It has also been shown that BAL is not a benign procedure and in fact, is an independent risk factor for IPF exacerbation [19‐22]. In a retrospective study it was shown that the risk of AEX-IPF is elevated within 30 days after BAL (RR 4.12; 95 % CI 1.03–12.2), moreover the relative risk of developing AEX-IPF after second or later BAL procedures was estimated to be considerably higher (RR 9.10; 95 % CI 2.27–26.98). In a recent review of the utility of BAL in diffuse parenchymal lung diseases, the role of BAL was critical in the diagnosis of opportunistic infections in patients treated with immunosuppressive therapy [23], but is not necessary in all patients.

In this retrospective study to assess the diagnostic value of bronchoscopy and BAL performed in the work up for suspected AEX-IPF cases we identified patients with a known history of IPF, who presented with acute respiratory failure and were being evaluated for AEX-IPF. AEX-IPF and pulmonary infection were the two most common final diagnoses and a minority of patients were found to have other cardiovascular and pulmonary conditions as a cause of their acute decompensation. The diagnosis of AEX-IPF was not associated with any of the predefined patient characteristics or measurable factors (such as gender, tobacco exposure or vital signs on admission). 38 % of patients had a bronchoscopy with BAL performed as a part of the diagnostic workup and it was more likely to be performed in patients receiving cytotoxic agents. One can only speculate that these patients were considered high risk for pulmonary infection and BAL was done due to high pretest probability. There was no other significant difference between two groups, which allowed further statistical analysis.

It is worth noting that in our cohort only three of 14 patients who had a final diagnosis of pulmonary infection had microbiological confirmation, one each from BAL and blood culture, tracheal aspirate culture and sputum culture. Only prior to admission steroid use was associated with a final diagnosis of pulmonary infection. In most cases, diagnosis of infection was made on the basis of physical examination, clinical history, laboratory/imaging data and clinician judgment. It seems BAL is most helpful when performed in patients with high pretest probability such as patients on steroids or immunosuppressive agents.

Most of our patients were started on broad spectrum antibiotics on admission prior to BAL and completed the course despite negative microbiological workup. We believe this is a common scenario in other centers too and shows that BAL fluid analysis does not change the treatment strategy. In our cohort bronchoscopy with BAL had little influence on the management of the patients which might suggest that patients who present with possible AEP-IPF versus pulmonary infection should be empirically treated with broad spectrum antibiotics and that bronchoscopy with BAL should be performed in selected cases only based on clinical judgment and case scenario, such as current use of steroids or other immunosuppressive agents.

Limitations of our study include relatively small sample size, single center participation and the retrospective nature of the study. Treatment selection biases as well as reliance on expert opinion in many cases for final diagnosis should be considered as well. Cleveland Clinic is a tertiary care center and a patient selection bias also could not be excluded. Our practice is not to use immunosuppressants for maintenance treatment of IFP patients, so our findings may not be translatable in other institutions who have not adopted this practice. Most of our patients were started on antibiotics before BAL could be performed and our conclusions may not be generalized to patient in whom BAL with fluid differential and cultures are done first.

Therefore, based on our findings in this study, we propose the following algorithm in the management of IPF patients presenting with acute respiratory failure (Fig. 1). IPF patients presenting with acute respiratory failure should first be evaluated for identifiable causes for their deterioration including but not limited to pulmonary infection, congestive heart failure decompensation, aspiration, pulmonary embolism, drug induced complications, and AEX-IPF based on clinical presentation. We suggest that all patients should be initiated on broad spectrum antibiotics upon presentation (including coverage for PJP if clinically indicated), ideally after blood, sputum and, in select cases, BAL cultures are obtained. One should not wait for culture results to initiate antibacterial therapy, but it should be used for de-escalation strategy. If BAL cultures are routinely obtained after initiation of antibiotic therapy, false negative results are likely and make further decisions for de-escalation a guess. Accordingly, we do not suggest BAL for all patients. If BAL can be safely obtained before the antibiotics are given, the diagnostic workup might be different and not reflected by our algorithm as most of our patients did get antibiotics before the BAL. If a non-infectious cause is identified, such as pulmonary embolism or pneumothorax, then antibiotics can be safely discontinued. Bronchoscopy with BAL should be performed in immunocompromised patients on steroids and other cytotoxic drugs, but also for selected patients with worsening respiratory failure despite broad spectrum antibiotics and inconclusive or unrevealing workup. In a retrospective study by Song et al., BAL and/or endotracheal aspiration were performed in 52.8 % of 461 patients highlighting the fact that in real life scenarios BAL is not performed for various reasons despite the universal recommendation and that our algorithm will be suitable for these cases [16]. It is based on small sample size, retrospective data and single center experience and should be used with these limitations in mind.