Improving Prognostication in Empyema: Is Computed Tomography the Answer?

Empyema is a common and often life-threatening condition, with a reported 30-day mortality of up to 10.5% (1) and a 1-year mortality of more than 19% (2, 3). In addition to the survival implications, large-scale studies have estimated the hospital stay associated with empyema or pleural infection to average 19 days (4), with an often comorbid patient demographic and annual costs totaling $500 million annually in the United States (5).

Given this burden on patients and healthcare systems, risk stratification and prognosis in empyema are of utmost importance. To date, the RAPID (renal, age, purulence, infection source, and dietary factors) score is the most robustly studied prognostic score, using age, urea, hospital-acquired infection, and nonpurulence of pleural fluid to predict poor outcome at presentation (6), and it is the only score that has been prospectively validated. The prospective PILOT study (prospective validation of the RAPID clinical risk prediction score in adult patients with pleural infection, n = 542) stratified patients into low-, medium-, and high-risk groups using the aforementioned parameters, with mortality at 3 months of 2.3%, 9.2%, and 29.3%, respectively (2). The next route of investigation after the validation of this score lies in using the RAPID score to triage early management of pleural infection (e.g., offering earlier surgery to those with the highest predicted mortality). Although RAPID uses clinical and biochemical information, the correlation of imaging findings with outcome has been limited to retrospective data or small-scale prospective studies. Chen and colleagues found in retrospective data sets that the presence of septation on ultrasound imaging appeared to predict the need for chest tube drainage and intrapleural enzyme treatment and correlated negatively with survival (7, 8). The survival link was not demonstrated in the PILOT study, however, or in a recent prospective study by Bedawi and colleagues, with septations predicting longer hospital stay only (9). Few reliable data exist on computed tomography (CT) imaging and the features of empyema that may predict outcome and overall prognosis.

To this end, Shiroshita and colleagues (pp. 807–814) report their findings on the prognostic value of CT in empyema in this issue of AnnalsATS (10). The authors conducted a multicenter retrospective cohort study across six centers in Japan, assessing a number of CT findings in patients with confirmed empyema and the difference in 90-day mortality and the duration of inpatient hospital stay. They selected pleural thickness ⩾2 mm, loculation, interlobar pleural effusion, lung abscess, and bronchopleural fistula as key findings to analyze in a binary fashion (present/absent). The relative differences in 90-day mortality between the presence and absence of each finding and the predicted probability of mortality on the basis of the presence of each CT finding were assessed. A total of 711 patients were included in the analysis, and of note, further analysis adjusting for RAPID score and early thoracic surgery was undertaken.

The authors demonstrate that the presence of bronchopleural fistula predicted higher 90-day mortality, with an estimated difference in 90-day mortality of 13.8% (95% confidence interval [CI], 7.7% to 19.9%). This was consistent even when adjusted for RAPID score and early surgical intervention. The presence of pleural thickness ⩾2 mm was associated with a longer hospital stay (5.13 d [95% CI, 1.93 to 8.34 d]). Interestingly, the opposite effect was observed with interlobar pleural effusion, which was associated with lower estimated 90-day mortality (−9.2% [95% CI, 14.6% to −3.7%]), and loculation was associated with shorter hospital stay (−4.58 d [95% CI, −8.02 to −1.60 d]). Pleural thickening did not appear to predict 90-day mortality, and conclusions regarding the impact of lung abscess could not be drawn, likely because of the small numbers present in the study. The authors assessed interobserver variability for each CT finding at a single center (n = 86), finding pleural thickness as the most variably reported finding (42% interobserver agreement), with loculation, interlobar pleural effusion, and lung abscess all reaching greater than 80% interobserver agreement.

The authors should be congratulated for being among the first to evaluate the impact of specific CT findings on patient prognosis and outcome in empyema. This is of great interest to pleural physicians as the availability of CT increases, and at present, the routine use of CT is not widely advocated in empyema management outside of refractory cases or suspected dual pathology (11). There are a number of potential practical applications that result from these findings, should they be validated more widely, including a consideration of earlier CT scanning in empyema. This may result in earlier identification and management in complex conditions such as bronchopleural fistula and the early use of CT imaging to triage treatment according to prognosis and predicted outcome.

However, it should be noted that there are study factors that may limit generalizability and use in clinical practice. Although the biological plausibility of bronchopleural fistula and reduced survival is clear because of the increased requirement for operative management, the reduction in mortality observed with interlobar pleural effusion is not clear, and contrary to other studies (12), although the authors hypothesize that this may be due to earlier, aggressive management strategies’ being deployed in these patients. CT scanning is not routine practice in empyema worldwide, and as such, the patients included in this study may have presented with more severe disease, prompting early cross-sectional imaging. In this study, noncontrast CT scans were used, and this may have limited the utility of analysis. Typically, late venous phase contrast is optimal for pleural pathology and can improve the sensitivity of CT imaging (13). As such, it is possible that pleural thickening was underreported, and pleural enhancement (often seen in empyema) could not be included. Finally, the retrospective nature of the data collection should be acknowledged, with 244 patients lost to follow-up and not all patients having available accurate pleural fluid laboratory analysis, suggesting that prospective validation of these findings is required.

Nonetheless, the aforementioned limitations do not detract from the importance of the work conducted by Shiroshita and colleagues (10). Although ultrasound has become the default early imaging strategy for empyema, because of its point-of-care access and utility in targeted pleural intervention, the use of CT scanning in empyema remains heterogeneously deployed. The results of this study certainly warrant further investigation across a wider population and, if corroborated, should prompt prospective validation. Ultimately, the ability to provide patients with more accurate, personalized prognostic information should be embraced, and this work could provide the basis for not only a prospective assessment of early CT scanning in empyema but an assessment of whether this can change management in a significant number of patients.