“The Cold Steel of a Surgeon or Some Fool of a Physician?”: The Debate Continues

Almost 20 years ago, the first British Thoracic Society (BTS) guideline on the management of pleural infection published in 2003 contained two important statements within its first paragraph. The first detailed that “there is great variation worldwide in the management of patients with pleural infection”, and the second that “overall, 20% of patients with empyema die” (1). It is rather tragic and, to some degree, a failure on our part as a community caring for these patients that both these statements remain true today.

The importance of early sampling and chest tube drainage has long been a principle in treatment through the adage “the sun should never set on a parapneumonic effusion”, coined by two pioneers of the field of pleural disease (Steven Sahn and Richard Light) (2). However, beyond this, the current treatment paradigm has changed relatively little. In most settings, clinicians start antibiotics, insert a chest tube, and “wait” for medical treatment failure to occur before promoting more aggressive intervention. Data published in the last 3–4 years have convincingly shown that time to intervention is one of the key contributors to adverse outcomes (3, 4). The question that follows, and to date remains unclear, is which intervention?

The role of surgery is well established in pleural infection, and some current guidelines advocate for it to be the primary treatment strategy (5). Indeed, modern Video Assisted Thoracoscopic Surgery (VATS) outcomes have shown considerable success with fewer complications (4). However, there remains a risk associated with the requirement for general anesthetic and a significant rate of conversion to open thoracotomy. In addition, the limitations in the evidence base for early surgery, including selection bias and lack of standardized criteria in the only two small randomized studies of surgery (6, 7) and the multiple case series, should not be overlooked.

The use of combination intrapleural enzyme therapy (IET) using tissue plasminogen activator (tPA) with deoxyribonuclease (DNase) as an alternative potential “rescue” treatment has been a much-needed and practice-changing addition to the landscape in the last decade, driven by the second Multicenter Intrapleural Sepsis (randomized placebo controlled) Trial (MIST-2) (8). Because of its publication, the MIST-2 regimen has been bolstered by a substantial amount of real-world series data demonstrating safety and efficacy in more than 2,000 and 700 patients, respectively (9–15). However, IET has not to date been directly compared head-to-head with surgery.

In current practice, the initial choice of intervention is often on the basis of clinician preference or local access, neither of which are particularly patient-centered. It is, therefore, timely that in this issue of Annals ATS, Wilshire and colleagues (pp. 1827–1833) report their data comparing outcomes of initial surgical versus initial IET (tPA and DNase) management (16). This was a large retrospective multicenter cohort study conducted in 18 hospitals over 5 states and comprising 566 patients with pleural infection (complicated parapneumonic effusion or frank empyema). Patients were managed with either initial surgery (n = 311 [55%]) or initial IET (n = 255 [45%]) with patients identified from retrospective billing data.

The primary study endpoints chosen are highly relevant to practice and included treatment failure and crossover. Treatment failure criteria were well defined and comprised evidence of ongoing infection, including fever or leucocytosis and a persistent pleural collection requiring intervention with additional treatment. Crossover was defined as receiving any dose of IET after surgery or receiving surgery after any dose of IET.

The groups were well matched for age and RAPID score (17) at baseline. Surgery appeared to perform favorably in terms of both treatment failure (7% vs. 29%) and crossovers (4% vs. 19%) compared with IET. Chest tube duration (median, 5 d vs. 7 d) and overall hospital length of stay (median, 10 d vs. 12 d) favored surgery despite a longer time to initiation of surgery compared with initial IET (3 d vs. 2 d). The surgery group had fewer additional treatments and readmissions, with a similar proportion of adverse events. However, when the surgical group was compared specifically with the subset of patients with IET who received the full MIST-2 dosing regime (147/255 [58%], defined as 5–6 doses of tPA 10 mg and DNase 5 mg), no difference in treatment crossover was observed. Importantly, there was no significant difference in mortality at 30 and 90 days between an initial IET or surgery management strategy.

The authors should be commended on this important study; to date, there have been no comparative studies of IET versus surgery, and as such, these data represent an important step in the right direction. Interventional studies in pleural infection pose many challenges, not least of which is clinician equipoise, and this is evident in the presented data. Within the group that was treated with initial surgery, 87% were admitted under a surgical team, compared with a 13% rate of surgery when the primary management team was medical. It is notable that the admitting team was surgical in 69% of patients overall, which may reflect U.S. practice and somewhat limit the external validity of the findings to other healthcare settings.

The treatment failure rate in the IET arm (29%) is considerably higher than previously published data (8–11, 18) and close to failure rates seen in standard care (18), which questions whether the practice was biased in favor of early surgery. Despite similar proportions of CT evidence of loculation and pleural rind or thickening in both groups, the presence of frank pus (empyema) appeared to steer clinicians toward early surgery (62% surgery vs. 38% IET). This is intriguing in the context of the RAPID score, which as the only validated prediction model of mortality in pleural infection, demonstrates the absence of fluid purulence as one of the five independent predictors of an unfavorable outcome (17).

The IET group was demonstrated to have a higher proportion of comorbidities, which highlights the other significant issue seen in nonrandomized studies of pleural infection. We cannot ascertain whether these patients had been considered for surgery but declined in view of their fitness or, conversely, how their outcomes would have compared had they been considered for surgery. The use of inverse probability of treatment weighting to address confounding by indication by the authors represents a praiseworthy effort to account for this bias in the context of a retrospective study.

The results from Wilshire and colleagues highlight some of the challenges in evaluating these two treatment strategies in pleural infection. If early surgery is indeed associated with improved patient outcomes, then healthcare systems should be universally restructured to allow earlier surgical evaluation as the treatment of choice, particularly for those with high RAPID scores who are at risk of the worst outcomes. However, as the data from Wilshire and colleagues suggest and as they conclude, further prospective randomized controlled trials are urgently needed in this area before firm conclusions on optimal initial treatment can be made.

Such efforts are already underway; the MIST-3 trial recently completed recruitment as a feasibility randomized controlled trial randomizing all-comers with ongoing evidence of infection on Day 1 to either early IET or early surgical management versus continued standard care (ISRCTN18192121) and should publish later this year. The FIVERVATS (FIbrinolytics VERsus VATS) study protocol from Denmark has been recently published and has started recruitment (19). These studies, among others on the horizon, will be critical additions to the evidence base and perhaps answer the question of the optimal initial management strategy in this challenging disease.

Among his many well-documented aphorisms, Sir William Osler (1849–1919), the eminent Johns Hopkins and Oxford physician, is quoted to have remarked that “empyema needs a surgeon and 3 inches of cold steel, instead of a fool of a physician.” Almost a century after his passing, we have finally begun to critically appraise his statement through increasing quality evidence. In the coming years, we can hopefully begin to impact treatment paradigms and standardize care resulting in better outcomes for our patients.