Corticosteroids for severe community-acquired pneumonia: not for everyone.

Abstract

Corticosteroids are thephysician’s immunomodulatory agent of choice.Theseagentsare inexpensiveandubiquitouslyavailable, have awell-known adverse effect profile, do not require monitoring of drug levels, and, probably most important, mimic the body’s own response to stress. Accordingly, corticosteroidshavebeenusedforawidevarietyofmedical illnesses. Use of corticosteroids for acute illness has alwaysbeen controversial, nomore so than for treating the triad ofcommunity-acquiredpneumonia,1 septic shock,2,3 andacute respiratory distress syndrome.4,5 The study by Torres et al6 in this issue of JAMA adds to this controversy. This group had previously demonstrated that treatment failure in patients with community-acquired pneumonia was associated with increased mortality,7 that a level of C-reactive protein (CRP) greater than 15 mg/dL was associated with treatment failure,8 and that radiographic progression was also associated with both treatment failure and mortality.7,9 In the current study, which is a logical extension of their previous work, the authors identified patients with severe community-acquired pneumonia and a proinflammatory profile at admission (defined as a serum CRP level >15 mg/dL). Patients were randomized to receive 0.5 mg/kg of methylprednisolone (n = 61) or placebo (n = 59) administered intravenously every 12 hours for only 5 days. The study found that corticosteroids decreased the primary end point, which was a composite outcome of treatment failure. Early treatment failure was defined as development of shock, need for invasive mechanical ventilation not present at baseline, or death within 72 hours of treatment, whereas late treatment failurewasdefinedas radiographicprogression, persistence of severe respiratory failure, development of shock, need for invasive mechanical ventilation not present at baseline, or death between 72 and 120 hours after treatment initiation. Treatment failure occurred in 8 patients (13%) in the methylprednisolone group vs 18 patients (31%) in the placebo group (between group difference, 18%; 95% CI, 3%-32%). Of the various criteria included in the definition of treatment failure, the difference between the treatment groups was almost exclusively due to less radiographic progression more than 72 hours after randomization. A nonsignificant difference in in-hospital mortality favoring corticosteroids (6 deaths [10%] vs 9 deaths [15%] for placebo) was also reported. The benefit of steroids remained statistically significant in the multivariable analysis, although baseline imbalances in the 2 groups may explain a portion of the differences observed for the primary outcome. Baseline cortisol levels were not measured and a higher proportion of patients in the placebo group had septic shock and required mechanical ventilation at the time of randomization. Most concerning is that only 24% of patients in the methylprednisolone group and 23% in the placebo group received macrolide combination antibiotic therapy despite multiple studies suggesting better outcomes with this therapy in more severely ill patients with community-acquired pneumonia.10,11 Amore importantquestion iswhatexactlyare steroidspreventing? Because radiographic progression during the period between 72 hours and 5 days was the primary driver of treatment differences, understanding what this clinical finding represents is key to acceptance of the findings. The 2 logical explanations for radiographic progression are uncontrolled pneumonia and development of acute respiratory distress syndrome. Although the latter is supported by a body of literature,2,4 a beneficial effect onuncontrolled pneumonia is less logical. A more intriguing possibility is that corticosteroids block a Jarisch-Herxheimer–like reaction to initiation of antibiotics in patients with high genomic bacterial load.12 The Jarisch-Herxheimer–like reaction is thought tobedue high concentrations of cytokines shortly after initiationof antibiotics, possibly through release of endotoxin or other bacterialmediators in patientswith a high bacterial burden. This possibility would be equivalent to the accepted use of corticosteroids inmeningococcalmeningitis,13 another infection for which high genomic bacterial load is associated with severe clinical manifestations andmortality.14 Despite being an ubiquitous immunemodulator, corticosteroids are not for every patient with severe communityacquired pneumonia. Even though Torres et al6 used somewhat outdated criteria for severe community-acquired pneumonia, only 57% (162/284) of eligible patients whowere not missed or did not have exclusion criteria had a CRP level greater than 15 mg/dL. More importantly, 8 years were required to achieve this enrollment at 3 centers, which is an average of 5 patients per center per year. Accordingly, the patients in this trial represent a small proportionof patientswith severe community-acquired pneumonia, but a group that often has poor outcomes. A larger definitive study is needed to confirm that less radiographic progression does in fact lead to lower mortality. However, the tendency for studies of corticosteroids in critiRelated article page 677 Opinion