Abstract
A discussion of the strengths and limitations of recent data associating bronchoscopy with mortality in immunocompromised adults with acute hypoxaemic respiratory failure http://bit.ly/31mMpGL
To the Editor:
Acute hypoxemic respiratory failure (AHRF) in immunocompromised patients is a challenging clinical problem associated with mortality rates of 40–60% in children and adults [1, 2]. Thus, we read with great interest the results of a pre-planned secondary analysis of a large multicentre observational cohort of 1611 immunocompromised adults with AHRF, as reported by Bauer et al. [3]. The authors described the diagnostic yield and outcomes of fibreoptic bronchoscopy (FOB) in this group of vulnerable patients with the a priori hypothesis that “bronchoscopy, with limited complications, would reduce the number of unidentified causes of respiratory failure and be associated with reduced hospital mortality.” After a rigorous analysis of a highly annotated dataset, the authors conclude that “bronchoscopy was associated with improved diagnosis and changes in management but also increased hospital mortality.”
We commend the authors for undertaking such a study, which attempts to answer an age-old question regarding the benefits and harms of bronchoscopy in patients with acute respiratory failure. While the results did not confirm the a priori hypothesis, we caution clinicians inclined to take the results as evidence to not perform bronchoscopy in clinically appropriate situations, given the inherent limitations of the study design. Since FOB was not randomised, a selection bias exists for the intervention group. To account for this bias, the authors identified 14 confounding factors measured at the time of intensive care unit (ICU) admission that were both associated with mortality and varied significantly in incidence between the noninvasive testing and FOB groups. Not surprisingly, patients who underwent FOB were at higher risk for both acquiring severe infections and for suffering poor outcomes from undiagnosed infection, factors that likely played into physicians' decision to perform FOB. After propensity score matching using these 14 confounders, patients who received FOB still showed significantly greater hospital mortality than their matched counterparts who underwent noninvasive testing alone.
Several scenarios could explain the finding of the propensity-adjusted analysis. The possibility exists that FOB precipitated events leading to death in an already tenuous clinical situation. While respiratory worsening was associated with 11% of FOB, it remains unclear whether FOB can actually be implicated as the cause for respiratory worsening, given that many patients were likely progressing in their disease course concurrently. On the other hand, the apparent discrepancy in mortality despite accounting for day 1 confounders could be driven by the subset of patients who underwent bronchoscopy later in the ICU course at a point when empiric therapies had failed and illness severity had increased from the baseline Sequential Organ Failure Assessment score and arterial oxygen tension/inspiratory oxygen fraction ratio. The clinical decision to perform bronchoscopy is complex and guided by many factors, including the trajectory of the patient's condition. The authors acknowledge they were unable to account for the time-varying nature of the intervention (FOB) as well as time-varying covariates, namely the severity of illness at the time of the non-randomised intervention, which are likely the most significant confounders not accounted for by propensity score matching (figure 1). The timing of FOB with respect to both the onset of AHRF and the initiation of empirical therapies plays a significant role in microbiological yield [4]; whether patients who underwent early FOB had better outcomes than those who underwent later FOB could not be assessed in this study. As such, future studies that detail both the timing of FOB with respect to illness onset, as well as the illness severity at the time of FOB, are needed before concluding that FOB is harmful to immunocompromised patients with AHRF.
The mortality endpoint, with the above caveats, should not overshadow the important finding regarding the utility of bronchoscopy. When noninvasive testing does not yield an AHRF aetiology, FOB is the best next option for pursuing a specific diagnosis, as surgical lung biopsy is even more invasive and empiric therapy is potentially harmful. In this study, the overall diagnostic yield of FOB was 49%; FOB resulted in addition of a new treatment in 12% and antimicrobial cessation in 26%, for an overall therapeutic yield of 38%. Hence, rather than be dissuaded by these results to perform bronchoscopy in immunocompromised patients with AHRF, we advocate for utilising FOB specifically in cases where identifying or refuting a specific diagnosis will aid in the patient's clinical management.
Taking the long view, additional work is needed to improve the clinical utility of FOB. We and others have demonstrated that novel genomic sequencing techniques can identify infectious culprits in half of previously negative respiratory samples and elucidate microbiome composition in immunocompromised patients [5, 6]. In addition, a significant opportunity exists to interrogate and integrate components of human biology into FOB-based tests. Ongoing work incorporating human gene expression and protein biomarkers may significantly increase the utility of FOB [7]. Further, the ability for FOB to impact management decisions depends on the development of novel targeted treatments for AHRF in immunocompromised patients [8], and the development of such pathobiology-targeted treatments depends on tissue-level biologic data accessible only by FOB [9].
In sum, we propose that the future studies assessing clinical outcomes of patients who undergo FOB should consider the timing of FOB with respect to ICU admission and should consider potential changes in illness severity between ICU admission and the time of FOB. We hope that this study by Bauer et al. [3] will stimulate the respiratory and intensive care community to work towards improving both our approach to the utilisation of FOB as well as the development of novel FOB-based diagnostic tests in order to better diagnose, understand, and treat AHRF in this high-risk population.
Shareable PDF
Supplementary Material
This one-page PDF can be shared freely online.
Shareable PDF ERJ-01950-2019.Shareable
Footnotes
Conflict of interest: M.S. Zinter reports grants from Pediatric Blood and Marrow Transplant Foundation, National Marrow Donor Program (Amy Strelzer Manasevit grant), and National Institutes of Health (NICHD K12HD000850), outside the submitted work.
Conflict of interest: G-S. Cheng has nothing to disclose.
- Received October 3, 2019.
- Accepted October 11, 2019.
- Copyright ©ERS 2019