|
Correction
to
Bernard et al.,
|
| |||||||||||||||||||||||||||||||||||
There is abundant evidence that arachidonic acid metabolites act as endogenous regulators of cytokine production.2 Prostaglandin E2 inhibits the release of interleukin-1 and tumor necrosis factor 
(TNF-).3 We tested the effect of ibuprofen on serum levels of TNF- and interleukin-6 in humans.4 After the injection of lipopolysaccharide, the respective serum levels of interleukin-6, TNF-, and elastase were 4.2, 1.7, and 1.5 times as high in subjects who received two doses of ibuprofen (800 mg each) as in controls.4 In addition, high TNF- concentrations primed neutrophils for degranulation in vitro.4 Since the mortality rate associated with sepsis correlates with high interleukin-6 and TNF- levels, the use of prostaglandin inhibitors in sepsis may be harmful.
As a rationale for the study, Bernard et al.1 cite only the animal models and clinical studies in which the effects of nonsteroidal antiinflammatory drugs were favorable. However, in five controlled trials of antipyretic agents in nonhuman mammals with severe infection, mortality was increased in animals given prostaglandin inhibitors (relative risk, 2.04; 95 percent confidence interval, 1.19 to 4.51).5 In the study by Bernard et al.,1 ibuprofen did not have serious side effects. However, prostaglandin inhibitors raise cytokine levels during endotoxinemia, and high levels of cytokines correlate with mortality. Therefore, prostaglandin inhibitors should be used with caution in patients with sepsis.
Werner Zimmerli, M.D.
Andreas F. Widmer, M.D.
University Hospitals Basel
CH-4031 Basel, Switzerland
References
, interleukin-6, and elastase during acute endotoxinemia. J Infect Dis 1991;163:89-95. [Medline]
To the Editor: Drs. Zimmerli and Widmer fear that treatment of fever with nonsteroidal antiinflammatory drugs (NSAIDs) may be harmful in the light of data on five animal models.1 Without discussing the virtues and limitations of each of these models, we acknowledge that there is potential harm from such treatment. On the other hand, as we stated in our paper, many more reports show significant benefit, including improved survival. When we began our trial, the substantial data on ibuprofen in animals, on balance, favored a beneficial effect of ibuprofen.
Concern that ibuprofen may augment TNF (and other biologically active compounds) is well taken. However, even if NSAIDs raise TNF levels in patients with sepsis, it is not clear that physiology or survival is adversely affected. Perhaps TNF levels are physiologically irrelevant in the presence of cyclooxygenase blockade. In the endotoxin experiments conducted by Zimmerli and Widmer as well as by others,2 all clinical effects of endotoxin administration, including headaches, nausea, chills, fever, and myalgia, were ameliorated by concomitant administration of ibuprofen. The subjects could not tell whether they had received endotoxin or placebo.
Arguments for and against the relative benefit of TNF antagonism in sepsis from an immunomodulatory perspective are ongoing and were discussed in the editorial accompanying our article.3 Zimmerli and Widmer and others2 present data that suggest that ibuprofen may increase circulating TNF levels in sepsis. However, it remains open to question whether this is harmful or beneficial, since in at least one trial TNF antagonism appeared to worsen the outcome in patients with sepsis.4 Though we know of no additional studies of ibuprofen that are planned or under way, several trials involving various methods of TNF inhibition are under way that may answer some of these questions. Preclinical experiments cannot answer the risk–benefit question with respect to humans with sepsis; they can only suggest where potential problems and efficacy may lie. Hence, clinical trials such as ours are designed to determine the net result of the inhibition of each component of the human response to severe sepsis, be it TNF, prostaglandins, or something else.
Table 2 of our report included incorrect confidence intervals for the black race and hypothermia subgroups. The mortality rate was 42 percent (95 percent confidence interval, 30 to 54 percent) among ibuprofen-treated black patients and 57 percent (95 percent confidence interval, 43 to 70 percent) among black patients given placebo (P = 0.12). The mortality rate was 54 percent (95 percent confidence interval, 33 to 74 percent) among ibuprofen-treated patients with hypothermia and 90 percent (95 percent confidence interval, 67 to 98 percent) among placebo-treated patients with hypothermia (P = 0.02).
Gordon R. Bernard, M.D.
Arthur P. Wheeler, M.D.
Brian Christman, M.D.
Vanderbilt University School of Medicine
Nashville, TN 37232
References
| |||||||||||||||||||||||||||||||||||
HOME | SUBSCRIBE | SEARCH | CURRENT ISSUE | PAST ISSUES | COLLECTIONS | PRIVACY | HELP | beta.nejm.org Comments and questions? Please contact us. The New England Journal of Medicine is owned, published, and copyrighted © 2008 Massachusetts Medical Society. All rights reserved. |
Previous
