Hematopoietic stem cell transplantation (HSCT) was first introduced as a therapeutic modality applicable in oncology clinical practice in the late 1960s, and since then has been used to treat neoplastic and non-neoplastic disorders. Recent data from the National Bone Marrow Transplant Data Registry have documented more than 60,000 transplants since its creation in 1986. Unfortunately, pulmonary complications are often observed in this patient population, which contributes significantly to post-transplant morbidity and mortality.1–3 These complications include those of infectious etiology, such as infections caused by bacterial, viral and fungal pathogens, and those due to non-infectious causes, such as graft-versus-host disease (GVHD), diffuse alveolar hemorrhage, obliterating bronchiolitis, interstitial pneumonitis and pulmonary edema. Pulmonary infiltrates in the HSCT population constitute a difficult therapeutic dilemma. Some experts recommend the use of flexible bronchoscopy (FB) in HSCT patients in whom new infiltrates or worsening of existing infiltrates has been observed,4,5 while others have recommended more empirical treatment with broad-spectrum antimicrobial agents in patients with more localized infiltrates.6,7 An initial approach with broad-spectrum antimicrobials and dieresis has also been proposed, reserving more advanced diagnostic examinations for patients who do not respond, or in whom there are still concerns. There are no hard data which suggest that a certain approach is more effective.

Previous studies have examined the use of FB in the HSCT population with the main diagnostic methods such as bronchoalveolar lavage (BAL), protected specimen brush (PSB) and/or transbronchial biopsy (TBB).8–13 However, most of these studies have various limitations, such as small sample size, the inclusion of other non-transplant patient populations, lack of identification of treatment changes introduced as a consequence of the FB results, and the yield provided by FB in the period prior to the use of prophylaxis.

The aim of our study was to retrospectively evaluate the experience in our center with the use of FB in evaluating pulmonary infiltrates in an adult HSCT population. We planned to evaluate the diagnostic yield of FB, its impact on clinical treatment decisions in hospitalized adult HSCT patients and its complication rate.

Thomas Jefferson University Hospital is a 957-bed tertiary care center, with a 14-bed bone marrow transplant unit (BMTU). Admissions to the BMTU correspond to newly diagnosed hematological malignancy patients, patients undergoing HSCT, or HSCT patients who require hospitalization due to transplant-related complications. All patients admitted to the BMTU in whom FB was performed after a HSCT between 1 January 1996 and 31 December 2009 were identified from hospital discharge codes, through the medical record department. Data were obtained retrospectively by reviewing the medical records and bronchoscopy reports of all patients. In patients who underwent more than one FB, each examination was considered independently. This study was approved by the Thomas Jefferson internal ethical review board.

All patients in the BMTU were admitted under a medical oncologist with a particular interest in bone marrow transplant. The decision to request a respiratory medicine consultation and the time of doing so was at the discretion of the medical oncologist responsible for the patient's care, as in our center there is no standard protocol for evaluating HSCT patients who present pulmonary complications.

FB remains an attractive technique for most clinicians, given its diagnostic ability and ease of use in patients in a critical condition.4–6 Over a period of 14 years in our hospital, 144 of 660 HSCT patients (22%) underwent FB to evaluate pulmonary infiltrates; most were being treated with broad spectrum antimicrobials at the time of the FB. The overall diagnostic yield in our cohort was 52.5%; treatment changes were introduced after the FB in 44% of patients. We believe that these data continue to support the use of FB in the evaluation of pulmonary infiltrates in the HSCT population.

The positive diagnostic yield of FB in the HSCT population was a primary outcome variable in our study. The overall diagnostic yield of FB in our population was 52.5%. BAL was positive in 53.1% of cases and TBB was positive in 81.8% of cases in which it was performed. Most of the results obtained in the BAL corresponded to infectious microorganisms, especially streptococci and candida species; however, alveolar hemorrhage was also observed in 21.1% of positive BALs.

These results agree with those of previous studies, in which the diagnostic ability of FB after transplant was investigated. In the literature, a positive diagnostic yield was observed in 31%–74% of FB.8–13,1 However, as previously indicated, many of the studies had the limitation of small sample size, inclusion of patients without HSCT and the exclusion of prophylactic antimicrobial medication guidelines. The most extensive study conducted in HSCT patients appears to be the study published recently by Shannon et al. from data obtained in the period 1994–1999, in which they attempted to determine the impact of early versus late referrals for bronchoscopy. These authors retrospectively evaluated a total of 598 bronchoscopies with BAL performed in patients treated recently with HSCT, in whom an evaluation was carried out for the presence of pulmonary infiltrates. The overall yield of the technique was 55%, but the authors indicated a 2.5-fold increase in the diagnostic yield when the examinations were performed within 4 days of hospitalization. However, this increased diagnostic yield did not significantly modify the clinical treatment, since there do not seem to have been differences between the population with early use and the population with late use.15

Kasow et al. published a similar study that was more extensive in terms of size and duration, in which they examined a pediatric HSCT population at St. Jude Children's Research Hospital. These authors examined 89 patients over a period of 12 years. Their conclusion was that, in the pediatric population, the use of active immunosuppression or the presence of grade II-IV graft-versus-host disease reduced the probability of obtaining a positive BAL. The clinical impact of the BAL results was not examined, except for the finding that 16 of their patients had an open lung biopsy; however, 14 of these had a positive BAL, which could reduce the need for open lung biopsies.16

A second objective of our study was to assess treatment changes after the FB, defined as the addition or withdrawal of antimicrobial drugs or corticosteroids. Modifications were introduced in the clinical treatment in 44% of our population after the FB; thus we could consider that this is a significant clinical intervention. The changes most frequently identified were the modification of antibiotic treatment guidelines. It is interesting to note that the bronchoscopy provided a result in 52.5% of the examinations, and that treatment changes were made in a further 27 patients, even though the FB result was not diagnostic. It could be argued therefore, that the FB was thus “diagnostic” on discarding significant pulmonary disease and enabling the patient's clinician to possibly “de-escalate” the treatment.

A third objective of our study was to evaluate the complication rate of FB in the HSCT population. FB is considered a safe technique, with complication rates varying between 0.3% and 6% in the general population.17,18 However, in the HSCT population, the complication rates previously described vary between 0 and 41%.19 Our overall complication rate was 30%, with major complications occurring in only 5%, including one death. It is interesting to note that there were no significant differences in the coagulation parameters between patients who had complications after the FB and those who did not. It is of utmost importance that the vast majority of complications consisted of mild hypoxia, which seemed to have little clinical significance.

In view of existing data, many authors recommend the use of FB as the initial diagnostic method of choice in HSCT patients with pulmonary infiltrates. Others also recommend a second FB (possibly with the addition of TBB to the diagnostic arsenal) in the context of an initial non-diagnostic FB.6 However, the need to obtain new diagnostic samples, such as TBB or open lung biopsy samples, remains controversial. We have shown a fairly high diagnostic yield for TBB in our HSCT population, but most of these diagnoses remained non-specific. The real clinical impact of a diagnosis such as non-specific interstitial pneumonitis remains unknown. This diagnosis does not provide the clinician with any information about the specific underlying etiology and process, but it may still provide useful information for treating the patient. An important limitation of our TBB data is the small sample size and the introduction of a sampling bias.

The limitations of our study are multifactorial. Firstly, this is a retrospective study conducted in a single center, which may not be generalizable to all centers. The most important limitation in this respect may be in the clinical practice bias that exists in our hospital for evaluating HSCT patients. Since there is no specific protocol for evaluating HSCT patients, the decision to perform a bronchoscopy, when to do so and the decision to apply certain techniques, including BAL, TBB, PSB, etc. when performing the bronchoscopy are factors that may introduce bias. We believe that the high patient number, as well as the time frame of the present study could be useful for reducing the effect that one or two health professionals and their personal bias could have on the overall study results. It is interesting to note that, over fourteen years, there have been very few identifiable changes in the diagnostic yield obtained in these patients. However, there was a notable reduction in the use of protected specimen brush and transbronchial biopsy. The exact reason is not clear, since there were few major complications in our hospital related with bronchoscopy techniques; however, other changes have appeared in the literature during the study period, which may have influenced the decisions of our professionals when applying certain techniques. Secondly, although we define treatment modifications as those introduced within 3 days after the FB, we do not have any definitive way of ensuring that these changes were a direct consequence of the FB itself, given the retrospective nature of the study. A detailed review of the medical records was carried out, to try to relate the bronchoscopy results with treatment changes, but there may be inaccuracies. Thirdly, radiographic evaluation of the pulmonary infiltrates was not standardized, and could include plain X-rays or computed tomography scans. This may have influenced both the decision to move to a bronchoscopic technique and decision-making during its performance. Finally, as this is a retrospective study, no strong statements can be made regarding the causality with respect to the effect of the FB on the results. These issues may be addressed in a multicenter, randomized trial currently being conducted, in which HSCT patients with new onset infiltrates will be randomly included to early (within 36h) or late FB (within 7 days), in order to determine its effects on the results (ClinicalTrials.gov identifier: NCT00846352).

Diagnostic yield and treatment changes following FB in our HSCT population were 53.5% and 44%, respectively. Our overall complication rate was high, 30%, but the rate of major complications was only 5%. The vast majority of complications corresponded to mild hypoxemia, which is an expected complication after FB. Our documented rate of major complications of 5% is very likely acceptable in the context of the significant morbidity and mortality associated with pulmonary complications in the HSCT population. Based on our results and on available data, we will continue to recommend FB as an initial diagnostic technique in HSCT patients with pulmonary infiltrates. Prospective studies will be required to evaluate the effects of these treatment changes on the clinical results.

No funding was available or used for this project. All the work was carried out at the Thomas Jefferson University Hospital.

The authors do not have any possible conflicts of interest. Dr. Gilbert does not have any conflicts of interest to declare. Dr. Learner does not have any conflicts of interest to declare. Dr. Baram does not have any conflicts of interest to declare. Dr. Awsare does not have any conflicts of interest to declare.