catheter drainage, or tube thoracostomy can be guided by ultrasound.

Computed tomography (CT) has become the most reliable radiographic technique used to characterize pleural effusions. Intravenous contrast material can de ne pulmonary vascular anatomy and enhance the parietal pleura. Parapneumonic effusions and empyema have abnormally high Houns eld units (−20 HU) compared to transudative effusions (−100 HU). Differentiating empyema, lung abscess, transudative pleural fuid, and subdiaphragmatic fuid (ascites) is often dif cult without CT. Lung abscesses generally appear as air-fuid spherical lesions forming acute angles with the lung parenchyma. The lung appears destroyed rather than compressed. There is an abrupt cutoff of vessels and bronchi. Empyema appears laterally, pushing or compressing adjacent lung parenchyma, vessels, and bronchi. The shape is not uniform, and angles with the pleura are acute. Lateral lung abscesses or abscesses in the basilar segments of the lung near the diaphragm may be dif cult to distinguish from empyema.

Biochemical Analysis

In health, pleural fuid is typically low in volume (<1 mL), consisting of a small number of cells including mesothelial cells, macrophages, and lymphocytes [1]. Pleural fuid tends to contain more bicarbonate compared to serum, typically with a pH of 7.6, with lower levels of sodium than the serum, and similar levels of glucose [1]. As described in the section on pathophysiology, the rst changes seen in simple, uncomplicated parapneumonic effusions refect the pathophysiologic changes seen in the exudative phase. The fuid rapidly increases in volume and develops a neutrophilic exudate. Ancillary testing at this stage can, for the patient with recent retching and rapidly progressing pleural fuid, also include fuid amylase to evaluate for potential esophageal rupture [1].

The typical biochemical evaluation to distinguish simple, uncomplicated parapneumonic effusion with sterile pleural fuid from compli-

cated parapneumonic effusion include pleural pH, pleural glucose, and pleural LDH. The most sensitive pleural fuid measurement that indicates a parapneumonic effusion is complicated and should be drained is the pH, which drops to 7.20 before the glucose drops or LDH rises to greater than three times the upper limit of normal [19, 23]. Neutrophil phagocytosis and bacterial death fuel an infammatory milieu which results in increased production of pleural lactic acid and carbon dioxide production, accompanied later with increased glucose metabolism, and as these leukocytes die, a production in pleural LDH. Per both British Thoracic Society and American Association for Thoracic Surgery guidelines, therefore, complicated parapneumonic effusions are de ned by a pleural pH < 7.20, glucose <40 mg/dL (or 2.2 mmol/L), and LDH > 1000 IU/L [1, 2].

Despite being the earliest and most sensitive indication for complicated pleural effusion, pH measurement can also be fraught with error. Besides needing to be evaluated in a blood-gas analyzer as opposed to pH paper or a pH indicator strip, pleural loculations, residual lidocaine or heparin in the pleural space, residual air, and delays in time to analysis have all been found to affect the diagnostic accuracy of pleural pH [24, 25]. Pleural glucose, the second most-sen- sitive biochemical measure (receiver operating characteristic with area under the curve (AUC) of 0.84 compared to 0.92 of pleural pH), is not affected by measurement technique and represents a reasonable alternative for pleural fuid testing.

To eliminate some of this measurement bias and time to results, several studies have measured the ef cacy of point-of-care (POC) pleural fuid testing at the bedside. In 2000, Kohn et al. demonstrated agreement between tabletop blood-gas analyzers with laboratory evaluation, with an absolute difference of 0.024 U [26]. In a similar vein, to evaluate other POC-systems, Abdo et al. evaluated the use of bedside POC pleural glucose via ACCU-CHEK glucometers, nding agreement between lab-measured and POC pleural glucose, particularly for values <80 mg/dL, and earlier diagnosis by nearly 2 h, however, with a

mean difference between lab-measured and POC glucose of 14.8 mg/dL [27].

Besides pleural pH, glucose, and LDH, other biomarkers have been evaluated or potential diagnostic utility. Infammatory cytokines (TNF-­ a, IL-8, IL-16, and IL-1B), enzymes (neutrophil elastase, myeloperoxidase, metalloproteinases, lipopolysaccharide binding protein, soluble triggering receptor expressed on myeloid cells-1 STREM-1, and CRP) have all been evaluated, and have yet to outperform traditional criteria [5]. Zou et al. performed a meta-analysis for pleural procalcitonin and CRP, nding while CRP was slightly more speci c compared to procalcitonin (77% compared to 70%, respectively), both performed poorly with poor sensitivity (54% and 67%, respectively) [28]. High-throughput proteomics may represent the next phase for identifying pleural fuid biomarkers, with one study using i-TRAQ-based mass spectrometry nding four new potential biomarkers (BP1, NGAL, AZU1, and calprotectin) with excellent sensitivity and speci city [29]. BP1, a neutrophil granule protein with antimicrobial properties, had the best sensitivity and speci city of the four (AUC 0.966, sensitivity 97%, speci city 91.4%), and when combined with LDH, an even higher sensitivity of 100%, and may even represent disease severity, with levels in empyema found to be twofold compared to those in parapneumonic effusions, although prospective validation is still pending.

Microbiology

The standard bacteriology of parapneumonic effusions and empyema were, classically, those refected in the etiologies of pneumonia. CAP-­ associated pleural infections were traditionally caused by streptococcal species, most common S. pneumoniae, while HAP-associated parapneumonic effusions and empyema were more closely related with staphylococcal and gram-negative bacteria. After all, the pathogen that lead to the empyema outbreak of 1917 that lead to the original formation of the Empyema Commission was Group A Streptococcus [13]. Over time, however,

through the development of pneumococcal vaccines, the bacteriology has started to evolve. When Grijalva reviewed causes of pleural infection from 1996 to 2008, he found stable rates of pneumococcal empyema, but an increase in both streptococcal and staphylococcal related empyema (1.9 and 3.3-fold, respectively) [14]. This has had profound effects on hospitalization and mortality, as staphylococcal-related empyema is associated with longer hospitalization and the highest in-hospital case fatality ratio [14]. Of these causes of empyema, however, 62.4% of empyema in the period from 1996 to 2008 had after routine microbiologic evaluation to have no identi ed infectious cause [14].

Through better detection techniques and changing patterns of infection and vaccination, the MIST-1 study group also found a shift away from pneumococcal causes of parapneumonic effusion and empyema, shifting instead to the Streptococcus milleri group (encompassing S. constellatus, S. intermedius, and S. mits), seen in 23% of community-acquired isolates [16, 18]. In this same community acquired group, they also found anaerobic infection to be remarkably common, found in 20% of the identi ed causes of pleural infection. Furthermore, with the use of bacterial identi cation via nucleic acid ampli cation in addition to routine gram stain and culture reduced unknown causes for infection from 42% to 26% of their total population. Of those patients initially found to be culture negative, 16% had an identi ed etiology via nucleic acid ampli cation [18]. Of note, they did nd that patients who received antibiotics prior to pleural fuid sampling were more likely to be culture negative (61%). Finally, they found that parapneumonic effusion and empyema infection were microbiologically distinct from that of pneumonia, postulating that bacteria that thrive in the low pH and PO2 environment of the pleura, like the S. milleri group, are more likely to cause pleural infection.

In addition to nucleic acid ampli cation, several other methodologies have been proven and proposed for better identi cation of the infective cause. One small study of 57 patients found that inoculation of pleural fuid into blood aerobic and anaerobic BacTec culture bottles in addition

to standard culture increased bacterial isolation from 37.7% to 58.5% [30]. Interestingly, this study also evaluated the incidence of bacteremia in these patients, founding only 11.8% of patients with blood cultures had bacteremia. Another small study, the AUDIO study, evaluated the feasibility of pleural biopsy at the time of tube thoracostomy to increase microbiologic diagnostic yield. Using a 18-gauge Temno cutting needle with a throw of 2 cm, they performed six to eight biopsies at the site with >3 cm of pleural fuid, and found it increased diagnostic yield by 25% [31]. Furthermore, in patients who had ­previously received antibiotics, they found that pleural biopsy increased diagnostic yield from by 27%. Finally, they trialed nucleic acid ampli cation of pleural biopsy samples, nding that 16S rRNA ampli cation and qPCR-based pathogen detection feasible and a potential method for rapid, sensitive microbiological detection, but yet requiring broader investigation.

Finally, uncommon causes of parapneumonic effusion and empyema can be infuenced by geography and severe immunocompromised status. Throughout Thailand, for example, up to 22% of patients present with pulmonary melioidosis caused by Burkholderia pseudomallei, and places with high infection of Entamoeba histolytica can present with pleuropulmonary amoebiasis following rupture of a hepatic collection with transdiaphragmatic spread [1]. Fungal empyema can also be seen, albeit <1% of cases, and is typically from Candida species with resultant mortality up to 73% [1].

Non-operative Management

The mainstay of non-operative management of parapneumonic effusions and empyema is the selection of appropriate antibiotics based on local microbiology and antibiotic resistance patterns. The appropriate selection of antibiotics, however, should include broad spectrum antibiotics, as the secondary analysis of MIST-1 demonstrated pleural bacteriology is distinct compared to pneumonia. While 50% of community acquired parapneumonic effusion isolates

were penicillin-­sensitive streptococci, the other 50% of isolates were typically penicillin resistant species, including penicillin-resistant staphylococci, Enterobacteriaceae, and anaerobes [18]. Antibiotic selection for hospital acquired parapneumonic effusion and empyema should include consideration for multi-drug resistant organisms as gram negative bacteria are much more common than in communityacquired infection. The length of antibiotic therapy has never been studied in comparative trials, with guidelines recommending anywhere from 2 to 6 weeks pending drainage, clinical improvement (i.e., defervescence), radiological, and laboratory (i.e., CRP) improvement [1, 2].

Appropriate supportive care is also recommended in all patients with parapneumonic effusion and empyema. This traditionally includes nutritional support, as malnutrition is known to correlate with poor outcomes, and deep vein thrombosis prophylaxis [1, 17].

In addition to appropriate selection of antibiotics, there are several options for the management of pleural infection, namely, observation, therapeutic thoracentesis, tube thoracostomy with or without intrapleural instillation of brinolytics, pleuroscopy, video-assisted thoracoscopic surgery (VATS) with decortication, thoracotomy with decortication, and open drainage. Non-interventional therapy is rarely effective, and often contraindicated for management of parapneumonic effusions and empyema, particularly those with continued signs of sepsis.

Thoracentesis can be both diagnostic and therapeutic. For simple, exudative parapneumonic effusions, thoracentesis, observation, and culture-­ sensitivity-­based antibiotic therapy are appropriate and generally successful. Knowing when to perform diagnostic thoracentesis can be subtle, although most argue for thoracentesis in patients with pleural effusions and persistent signs of sepsis [1]. More objective indications have included patients with free-fowing fuid greater than 1 cm from the inside of the chest wall to the pleural line on a lateral decubitus view, but with the advent of thoracic computed tomography and ultrasound, the recognition of the effusion may be earlier [6]. When the parapneumonic effusion is moderate

and free fowing, the initial “diagnostic” thoracentesis using a vacuum bottle or other drainage system can evacuate the pleural space completely, thereby permitting pulmonary re-­expansion. If the lung expands suf ciently and the fuid does not reaccumulate, no further intervention beyond clinical observation is required. If the fuid reaccumulates, repeat thoracentesis can aid diagnostically, but serial therapeutic thoracentesis is less desirable due to patient discomfort from repeated procedures and the possibility of incomplete drainage leading to lung entrapment and the need for surgical intervention, and is no longer recommended in the American Association for Thoracic Surgery consensus guidelines [2]. Finally, for patients with complicated, brinopurulent or organized parapneumonic effusions, however, therapeutic thoracentesis is rarely successful, and it is crucial not to delay drainage, as the fuid will become more dif cult to drain as loculations form.

Historically, tube thoracostomy using a large-­ bore chest tube (32 Fr to 38 Fr) was the initial intervention when the diagnosis of empyema was established. These chest tubes were later converted to an “empyema tube” at 14–21 days when pleural symphysis had occurred, to be slowly withdrawn slowly over several weeks. Patients often were discharged with the tube connected to a drainage bag. When the initial drainage by chest tube was unsuccessful at eliminating all loculations or lung entrapment was present, open surgical drainage with decortication was performed. An empyema tube is less frequently used today due to earlier use of antibiotics, improved diagnosis by CT, image-guided drainage, and earlier use of VATS.

Today, tube thoracostomy is performed by image-guided catheter placement via Seldinger technique with a smaller (8–20 Fr), more fexible catheter. The theory for large-bore chest tubes was that brin or the viscosity of pus would impede drainage via small-bore chest tubes, inhibiting timely drainage and increasing treatment failure [32]. While there are no randomized, prospective, comparative trials of small versus large-bore chest tubes, a secondary analysis of MIST-1 dichotomized chest tube size into “small” (<14 Fr) and

“large” (>14 Fr), and analyzed both clinical outcomes and perceived pain. Small-bore chest tubes were found to have no difference in the combined outcome of death or surgery at 1 year, individual outcomes of death or surgery, length of stay, or 3-month forced vital capacity (FVC), forced expiratory volume (FEV1), and chest radiography [33]. Patients with large-bore chest tubes, however, had increased perception of pain during placement and while in-situ. Most guidelines thus recommend tube thoracostomy with small (10–14 Fr)-bore chest tubes, with the BTS Pleural Infection guidelines speci cally recommending routine fushing of small-bore chest tubes with 20–30 mL of saline every 6 hours via 3-way stopcock to ensure continued patency [1]. Chronic indwelling pleural catheters have even been suggested, with one small study even describes placement of 15.5 Fr indwelling pleural catheters for the management of patients with chronic pleural infection who are poor surgical candidates [34].

If drainage is incomplete or lung entrapment has occurred, intrapleural brinolytic therapy should be considered. This strategy relies on cleavage o intrapleural brinous septations to better facilitate chest tube drainage. Observational data had initially suggested that intrapleural administration of brinolytic drugs reduced the frequency of failed drainage and subsequent surgery. The debate between which brinolytic agent continues, with initial efforts studying streptokinase, and subsequent studies suggesting urokinase was more ef cacious and less likely to cause a febrile or allergic reaction. The rst large, multi-center trial, prospectively randomized, placebo-­controlled trial evaluating the clinical effectiveness of streptokinase, MIST-1, however, demonstrated no difference in death or surgical drainage at 3 months, nor any clinical bene t in length of stay, residual pleural thickness (RPT), or post-recovery spirometry [16]. Later case series supported the use of a different direct-­ acting brinolytics agent, recombinant tissue plasminogen activator (t-PA). Based on a theory that the presence of extracellular DNA and other bacterial components in the pleural space may increase viscosity and permit bio lm formation, a second multi-center, prospective, double-­