Introduction

Pleural infections are severe infectious diseases, with a mortality rate ranging from 10 to 20% [1, 2]. Incidence of pleural infections has increased in the last few decades and reached 15 cases per 100 000 per year [3–5]. According to guidelines, the two pillars of treatment are effective antibiotics and drainage of the infected liquid [6–8]. Hippocrates already underlined that ‘if an empyema does not rupture, death will occur’. To date, chest tube insertion is the most widespread technique of drainage and is considered as the gold standard treatment in pleural infections. Nevertheless, another strategy is possible, namely repeated therapeutic thoracentesis (RTT). The theoretical advantages of RTT are: i) targeting spaces where loculated collections are the most voluminous; ii) allowing early ambulation; iii) allowing early pleural physiotherapy; iv) avoiding clogging of chest tube. Several studies reported RTT as an effective alternative to chest tube. We previously described a cohort of pleural infections managed by RTT as first-line treatment, with a clinical success rate beyond 80% and low rate of complications related to RTT [9]. In a comparative study, thoracenteses were not inferior to chest tube, with reduced duration of hospitalization and complications rates [10].

Furthermore, releasing fibrinolytic agents into pleural cavity has shown benefits in improving drainage effectiveness in a large randomized controlled trial [11]. In this study, Rahman et al. demonstrated that intrapleural use of tissue plasminogen activator (tPA) and deoxyribonuclease (DNase) reduces surgical referral at 3 months and length of hospital stay. In a previous trial, administration of a single fibrinolytic agent–tPA–did not improve any outcome [1]. Intrapleural fibrinolytic therapy has only been assessed once agents have been released through the chest tube and retrieved after one hour. In RTT, fibrinolytic agents and DNase released into the pleural space have a prolonged exposure. Though urokinase or tPa have very short half-lives (30–45 minutes), DNase has prolonged half-life (3–4 hours) increasing its activity on loculations [12, 13]. We aimed to compare outcomes in patients with pleural infections managed by RTT with urokinase and DNase (RTT-UD), versus RTT with urokinase (RTT-U).

Methods

Results

Of the 443 patients with suspected pleural infection, 133 fulfilled criteria for complicated parapneumonic effusion (CPPE) or pleural empyema (PE) (Fig 1). Data on patients from the first group were previously published [9].

Download:

• PNG
  larger image
• TIFF
  original image

Fig 1. Flowchart of the study.

RTT-U: Repeated therapeutic thoracenteses with use of intrapleural urokinase only. RTT-UD: Repeated therapeutic thoracenteses with use of intrapleural urokinase and DNase.

https://doi.org/10.1371/journal.pone.0257339.g001

The mean age was 59 years (standard deviation, 17.2), 69.2% were men (92/133), 63.9% (85/133) had at least one chronic underlying disease. Infections were mostly community-acquired (116/133, 87.0%). Characteristics of the two groups are summarized in Table 1. Regarding demographical characteristics and comorbidities, the two groups had no significant differences in age, sex-ratio, chronic obstructive pulmonary disease, heart disease, diabetes mellitus or immunodepression. Malignancy was more frequent (n = 7/52 vs. 25/81, P = 0.02) in the RTT-UD group, and neurological impairment less frequent (n = 19/52 vs. 12/81, P = 0.01).

Download:

• PNG
  larger image
• TIFF
  original image

Table 1. Baseline characteristics of the study population.

https://doi.org/10.1371/journal.pone.0257339.t001

In the RTT-UD group, patients had a more severe initial presentation with more respiratory failure at admission (n = 7/52 vs. 33/81, P<0.001) and severe sepsis (n = 3/52 vs. 19/81, P = 0.008). The risk-groups proportions of the RAPID score, namely low–medium–high, were 50% (n = 26/52), 35% (n = 18/52) and 15% (n = 8/52) in the RTT-U group; whereas proportions in the RTT-UD group were 38.3% (n = 31/81), 48.1% (n = 39/81) and 13.6% (n = 11/81) respectively [2]. Using this score, there were no significant differences in the risk-groups between the two groups.

Biochemical and microbiological parameters of pleural fluid did not differ significantly from one group to the other, except for pleural pH and LDL levels (Table 1). Positive cultures on pleural fluid were found in less than half of cases (n = 56/133, 42.1%). Streptococci were the leading pathogens identified (n = 62/133, 46.6%), followed by Gram-negative bacteria (n = 30/133, 22.6%), strict anaerobic bacteria (n = 22/133, 16.5%), and staphylococci (n = 2/133, 1.5%).

The number of thoracentesis per patient in the two groups were equivalent (4 [3–6] vs. 4 [3–5]; P = 0.49) (Table 2). Ultrasonography-guided procedure was more frequent in the recent cohort (54% vs. 84%; P = 0.007). Complications of RTT such as iatrogenic pneumothorax (n = 9/133, 6.8%) and vasovagal reaction (4 out of a total of 632 thoracenteses) were low in both groups but almost one third of patients (n = 46/133, 34.6%) had at least one blank thoracentesis during the management of the pleural infection. Antibiotics used did not differ between the two groups, with a large proportion of patients treated by amoxicillin/clavulanic acid combination (58% in RTT-U group vs. 60% in RTT-UD group). Amoxicillin alone was prescribed in cases of pleural infection due to S. pneumoniae or oral streptococci (15% in RTT-U group vs. 26% in RTT-UD group). Patients in the RTT-U group received longer antibiotic courses compared to RTT-UD (Table 2).

Download:

• PNG
  larger image
• TIFF
  original image

Table 2. Management and outcomes of the study population.

https://doi.org/10.1371/journal.pone.0257339.t002

Time to apyrexia was significantly shorter in patients treated with the combination of urokinase and DNase (median fever duration: 10 vs. 7 days, P = 0.003). Fever duration after the first thoracentesis was also reduced in patients treated by RTT-UD (P<0.001). In-hospital mortality did not differ between the two groups (P = 0.25). Intensive care unit (ICU) admission rates were similar (15% vs. 15, P = 1).

Among outcomes, there was no difference in RTT failure, in-hospital mortality or one-year mortality between the two groups in the propensity score-weighted analysis (Table 3). RTT failure occurred in 19.2% (n = 10/52) with RTT-U and 17% (n = 14/81) with RTT-UD. Secondary chest tube drainages were less frequent with RTT-UD (4% vs. 15.4%, P = 0.02). Patients in the RTT-UD group had a reduced length of hospital stay (16 [12–24] vs. 23 days [18–41], P<0.001) and a higher total volume of pleural fluid retrieved (1041 mL [513–1600] vs. 1300 mL [655–2312.5]). However, abundance of pleural effusion was significantly lower in the RTT-UD group using an abundance score ranging from 0 to 5 on initial chest X-ray (P = 0.004).

Download:

• PNG
  larger image
• TIFF
  original image

Table 3. Results of the overlap propensity score–weighted analysis to estimate the effect of the combination of intrapleural urokinase & DNase.

https://doi.org/10.1371/journal.pone.0257339.t003

Estimates of weighted odds-ratios (wORs) obtained with the overlap weighting method based on the propensity score are reported in Table 3. Time to apyrexia (wOR = 0.51, 95%CI [0.37–0.72]) and the length of hospital stay (wOR = 0.61, 95%CI [0.52–0.73]) were significantly reduced in the RTT-UD group. The total volume of pleural fluid retrieved was significantly and independently higher with the use of urokinase and DNase (wOR = 1.38, 95%CI [1.02–1.88]). Other outcomes, namely in-hospital mortality, one-year mortality, RTT failure and ICU admission did not differ from one group to the other after adjusted analysis.

Non-purulent pleural fluid is associated with poorer prognosis than pleural infections with purulent fluid in the RAPID-score. In the subgroup of patients with non-purulent pleural fluid, only one RTT failure (n = 1/27) was reported in the RTT-U group and no RTT failure (n = 0/32) was encountered in the RTT-UD group. Event-free survivals according to the purulence of pleural fluid are plotted in Kaplan-Meier curves (Supplementary data, S1 Fig).

Discussion

Intrapleural fibrinolytic and DNase agents in pleural infections have been assessed only on patients drained with chest tube. To our knowledge, this study is the first to evaluate intrapleural fibrinolytic agent–urokinase—and DNase in pleural infections managed by RTT. Our work underlines many important messages: i) Time to apyrexia, length of hospital stay and the total volume of pleural fluid retrieved were optimized with a combination of intrapleural urokinase and DNase, ii) Mortality rates, RTT failures and surgical referral were low and did not differ between the two groups, iii) RTT is safe, with a low rate of iatrogenic complications, iv) in patients treated with RTT-UD who had non-purulent pleural fluid, no failure was reported.

In our center, we opted for urokinase instead of streptokinase or alteplase because it seems that urokinase is associated with a lower incidence of bleeding or allergic reaction [12, 14–16]. Moreover, two randomized placebo-controlled trials found that urokinase decreased duration of hospitalization, with faster defervescence and improved pleural drainage [17–19]. In addition, a meta-analysis supported that urokinase use could reduce surgical intervention (OR = 0.33; 95%CI: 0.14–0.78) [20]. Adding DNase is based on clear-cut benefits demonstrated in the MIST-2 trial [21, 22]. It has also been proven that this treatment strategy is highly cost-effective [23]. Moreover, a combination including DNase was more effective in experimental studies, with a reduction of pleural fluid viscosity and an increased volume of pleural fluid drained [24, 25]. These are the reasons why we used this combination with RTT. In line with previous studies, we found that the total volume of pleural fluid drained was higher with intrapleural urokinase and DNase, even though pleural effusion seemed to be less abundant and more loculated in the RTT-UD group. Of note, loculations were more commonly detected in the more recent group, probably explained by development of pleural ultrasonography in our department. On the other hand, we have no explanation for the fact that abundance of pleural effusion on chest imaging was lower in the RTT-UD group. Our work supports an enhanced effective drainage with the use of urokinase and DNase in pleural infections managed by RTT. In addition, time to apyrexia has been assessed as an outcome parameter in infectious diseases, where inappropriate therapies were associated with longer duration of fever [26, 27]. Moreover, several studies identified prolonged fever as a factor of poor outcome [28–30]. In this study, defervescence was lower in the RTT-UD group, arguing for its determinant role in controlling pleural infections.

In a landmark randomized controlled trial, Rahman et al. showed a reduction in hospital stay and a decrease in surgical referral with the use of t-PA and DNase in patients treated with chest tube. Herein, we did not find differences in mortality or surgical referral, but our study may have been underpowered due to limited sample size. The absence of difference in mortality rates comes as no surprise since no evidence of reduction in overall mortality has ever been proven [15]. In a meta-analysis, Janda and Swiston. brought to light that there was no difference in mortality rates with intrapleural fibrinolytic therapy [31]. In this work, we reported mortality and surgical referral rates concordant with other studies in literature, mostly using chest tube drainage strategy [32].

RTT had previously been identified as an effective alternative to chest tube in complicated parapneumonic effusion [9]. An experimental study focused on thoracentesis in the treatment of pleural empyema in rabbits revealed that thoracentesis was at least as effective as chest tube in terms of mortality [33]. In a recent review of literature, only few and old studies of limited sample size reported low rates of success with this strategy of drainage [34]. In other studies, clinical success rates ranged from 81 to 100% [9, 10, 35, 36]. To our knowledge, our study is the largest to date which assessed RTT in pleural infection management. We reported clinical success in 82% of cases, and only three patients required chest tube as second-line treatment in the RTT-UD group. Additionally, thoracentesis with adjunctive fibrinolytic and DNase could be a strategy avoiding more invasive procedures [37, 38]. Considering the half-life of DNase of 3–4 hours, a greater exposition time may improve effects on loculations [13]. In our study, even if fibrinolytic agents (i.e urokinase) and DNase remained in the pleural space until the next thoracentesis, adverse effects were rare with only one hemothorax and no pneumothorax requiring drainage.

Ferguson et al. found that pleural empyema was associated with failure of RTT [39]. However, in the RAPID-score, non-purulent pleural fluid was a predictor of poor outcome. We reported no failure in patients with non-purulent pleural fluid treated by RTT-UD. It suggests that RTT-UD could be considered as an effective alternative treatment strategy of pleural infections with non-purulent fluid. In case of empyema, switching to chest tube or surgery should be considered if necessary.

Our work has limitations. First, as a single-center study, our findings may differ from others. Second, as data were collected retrospectively over a large period of time (2001–2018), this study has potential biases. Nonetheless, heterogeneity in practice between the two groups was limited to adjunction of intrapleural DNase after 2011 and increased use of pleural ultrasonography. In addition, there were few missing data in the final database. Third, this study focused solely on pleural infections managed by RTT as first-line treatment. Hence, our population did not enroll most patients with severe presentation, such as patients admitted straight to ICU–where chest tube is first-line treatment- or patients who required early surgical referral. Fourth, the study was not designed to compare outcomes between patients receiving intrapleural fibrinolytic agent plus DNase through chest tube and by RTT. Lastly, this study did not assess the impact of the main partner for treatment success: antimicrobial treatment. However, our practices regarding antibiotic use for pleural infections did not significantly change during the study period.

In conclusion, intrapleural use of urokinase and DNase was associated with better outcomes than urokinase only in the management of pleural infections managed with RTT. Pleural infections are hard-to-treat infections, but RTT failures were infrequent, confirming the efficacy of RTT-UD as an alternative strategy to chest tube drainage, especially in complicated parapneumonic effusions.

Supporting information