To the Editor: Corwin et al. (Sept. 6 issue)1 report that theuse of epoetin alfa in critically ill patients does not reducethe incidence of red-cell transfusion but may reduce mortalityamong patients with trauma. However, the drug was given subcutaneously.It is well known that skin perfusion is impaired in patientswith critical illness. Did the investigators conduct a pharmacokineticevaluation in a subgroup to ensure adequate erythropoietin absorption?In addition, a relatively large erythropoietin dose was usedin the study. Such a large dose could be associated with endothelialdysfunction, a proinflammatory state,2 and increased thrombogenicityby inducing the expression of tissue factor.3 Giving smallerdoses of erythropoietin at more frequent intervals in patientsundergoing dialysis often improves responsiveness to the drugand decreases the total erythropoietin requirement. Designingsimilar studies with the use of smaller doses of erythropoietin,administered at more frequent intervals intravenously, mightaddress these points.
Dr. Mikhail reports having received honoraria from Roche, Amgen,and Astellas. No other potential conflict of interest relevantto this letter was reported.
References
Corwin HL, Gettinger A, Fabian TC, et al. Efficacy and safety of epoetin alfa in critically ill patients. N Engl J Med 2007;357:965-976. [Free Full Text]
Tobu M, Iqbal O, Fareed D, et al. Erythropoietin-induced thrombosis as a result of increased inflammation and thrombin activatable fibrinolytic inhibitor. Clin Appl Thromb Hemost 2004;10:225-232. [Free Full Text]
Fusté B, Serradell M, Escolar G, et al. Erythropoietin triggers a signaling pathway in endothelial cells and increases the thrombogenicity of their extracellular matrices in vitro. Thromb Haemost 2002;88:678-685. [Web of Science][Medline]
To the Editor: Corwin et al. report both desired and unwantedeffects of erythropoietin in their trial involving patientsin the intensive care unit. Paradoxical effects of erythropoietinhave been reported in other groups of patients and conditions,and the dose of erythropoietin may be important in this respect.
The statement that erythropoietin failed to decrease red-celltransfusions between days 1 and 29, however, may be based ona false premise, since it is generally accepted that in thefirst 14 days of its administration, this agent does not increasehemoglobin and thus should not influence the need for transfusion.1
We are concerned about whether iron treatment in the participatingpatients was adequate and similar in the two groups. Among patientswith chemotherapy-related anemia who receive similar amountsof epoetin, intravenous, but not oral, iron supplementationimproves the response to epoetin.2 In addition, oral iron supplementsfail to maintain adequate iron stores in epoetin-treated patientsundergoing hemodialysis.3 Therefore, it is possible that inthe study by Corwin et al., patients in the epoetin alfa groupreceived more parenteral iron than patients in the placebo group.Parenteral iron, as well as the high dose of erythropoietin,might have played a role in the observed, undesired effectsof erythropoietin.
Karien van der Putten, M.D. Meander Medical Center 3800 BM Amersfoort, the Netherlands
Branko Braam, M.D., Ph.D. University of Alberta Edmonton, AB T6G 2B7, Canada
Carlo A. Gaillard, M.D., Ph.D. Meander Medical Center 3800 BM Amersfoort, Netherlands cajm.gaillard{at}meandermc.nl
Drs. van der Putten, Braam, and Gaillard report serving as investigatorsin the EPOCARES study (ClinicalTrials.gov number, NCT00356733
[ClinicalTrials.gov]
),which is funded by the Dutch Heart Foundation and Roche. Noother potential conflict of interest relevant to this letterwas reported.
References
van Iperen CE, Gaillard CA, Kraaijenhagen RJ, Braam BG, Marx JJ, van de Wiel A. Response of erythropoiesis and iron metabolism to recombinant human erythropoietin in intensive care unit patients. Crit Care Med 2000;28:2773-2778. [Web of Science][Medline]
Auerbach M, Ballard H, Trout JR, et al. Intravenous iron optimizes the response to recombinant human erythropoietin in cancer patients with chemotherapy-related anemia: a multicenter, open-label, randomized trial. J Clin Oncol 2004;22:1301-1307. [Free Full Text]
Macdougall IC, Tucker B, Thompson J, Tomson CR, Baker LR, Raine AE. A randomized controlled study of iron supplementation in patients treated with erythropoietin. Kidney Int 1996;50:1694-1699. [Web of Science][Medline]
To the Editor: Corwin et al. state in their article, "The mostimportant finding in the current trial is the reduction in mortalityamong patients who received epoetin alfa as compared with thosewho received placebo, which was most apparent in the traumapatients." I believe that the most important finding concernsthe prespecified primary end point of the trial: epoetin alfahad no significant benefit with respect to the percentage ofpatients receiving any red-cell transfusion between days 1 and29. The next most important findings were the lack of any significantbenefit of epoetin alfa with respect to the prespecified secondaryend points, which were the number of red-cell units transfused,mortality at day 29 and at day 140, and the change in the hemoglobinconcentration from baseline to day 29. The subgroup analysisshowing decreased mortality among patients with trauma may bean artifact of multiple comparisons or the play of chance. Atmost, this finding might be considered hypothesis-generatingfor a future study of epoetin alfa in a population of patientswith trauma.
Lewis A. Eisen, M.D. Montefiore Medical Center Bronx, NY 10467 leisen{at}gmail.com
To the Editor: Survival is not the same as mortality. Corwinet al. state, "Mortality at day 29. . . was significantlylower in the epoetin alfa group than in the placebo group (8.5%vs. 11.4%, P=0.02)." However, they report no statistical testingof this hypothesis. A Cox analysis, which tests hazard ratios,showed no significant difference. The significant P value reportedis based on the Greenwood formula (which does not compare thetotal survival experience of the two groups) instead of on thelog-rank test, because of concern about censoring of data. Weare not aware of this limitation of the log-rank test.1 We wouldsubmit that the appropriate test to compare the 8.5% and 11.4%mortality is Fisher's exact test, which yields a P value of0.066 (P=0.051 for the subgroup of patients with trauma). Unfortunately,critical care clinical trials frequently confuse survival timeand mortality by reporting the percentage of patients who dieat a specific time with P values from a survival analysis thattests an entirely different hypothesis.2,3,4 Although we mayplace too much emphasis on arbitrary thresholds, the reportedreduction in mortality at day 29 that was attributed to epoetinalfa was not statistically significant in the overall populationor in the trauma subgroup.
Colin R. Cooke, M.D., M.Sc. University of Washington Seattle, WA 98104 crcooke{at}u.washington.edu
Gordon D. Rubenfeld, M.D., M.Sc. University of Toronto Toronto, ON M4N 3M5, Canada
References
Leung KM, Elashoff RM, Afifi AA. Censoring issues in survival analysis. Annu Rev Public Health 1997;18:83-104. [CrossRef][Web of Science][Medline]
Amato MBP, Barbas CSV, Medeiros DM, et al. Effect of a protective-ventilation strategy on mortality in the acute respiratory distress syndrome. N Engl J Med 1998;338:347-354. [Free Full Text]
Annane D, Sébille V, Charpentier C, et al. Effect of treatment with low doses of hydrocortisone and fludrocortisone on mortality in patients with septic shock. JAMA 2002;288:862-871. [Free Full Text]
Schoenfeld D. Survival methods, including those using competing risk analysis, are not appropriate for intensive care unit outcome studies. Crit Care 2006;10:103-103. [Medline]
The author replies: Both Mikhail and van der Putten et al. raiseissues regarding epoetin alfa dosing and response. My colleaguesand I believe that the absence of a reduction in the transfusionrate reflects transfusion practice, not a lack of hematopoieticeffect. The hemoglobin concentration increased significantly,suggesting a hematopoietic effect with epoetin alfa. In ourprior study,1 epoetin alfa administered in a similar mannerresulted in a reduction in the transfusion rate at day 29. Thehemoglobin response was identical in our prior and current studies;the data for the epoetin alfa and placebo groups began to separateby day 7. Parenteral iron use was not increased with epoetinalfa. A recent study2 of the pharmacokinetics and pharmacodynamicsof the subcutaneous administration of an identical epoetin alphadose in critically ill patients showed increased erythropoietinlevels and a hematopoietic effect. Whether a different dose,dosing schedule, or route of administration would alter theclinical outcome is unknown.
We disagree with Eisen's characterization of the trauma subgroup.The outcome in this subgroup replicates the results of our priortrial.1 In the current trial, randomization was stratified accordingto the admission group (trauma, medicine, or surgery) to ensurethat the treatment groups were comparable within each subgroup.Thus, we have now demonstrated a mortality reduction among patientswith trauma in two randomized trials. The analysis of the traumasubgroup was an appropriate subgroup analysis.3,4
We agree with Cooke and Rubenfeld's statement that survivalis not the same as mortality. Although the raw mortality ratesare reported in our article, the P value (0.02) for the overallpopulation was based on the comparison of Kaplan–Meierestimates within each stratum (trauma, surgical, and medical)and the estimate of the standard deviation with the use of theGreenwood formula. Mortality rates based on the Kaplan–Meiermethod account for censoring before day 29. For the overallpopulation, stratified analysis is more appropriate to preservethe randomization within each stratum. This is accommodatedby the use of a stratified Kaplan–Meier approach. Thevery same approach was used for the comparison within the traumasubgroup, except that the stratified test was not applicable,since the analysis involved a single stratum. The differencein mortality rates was more pronounced in the trauma subgroup,but the P value was slightly higher (0.039) because of the smallersample as compared with the overall population. Although theabove analyses were the most appropriate, we performed severalsensitivity analyses to assess the robustness of the findings.For the overall population, the log-rank P value was 0.049,the stratified log-rank P value was 0.056, and the Cochran–Mantel–HaenszelP value was 0.053. For the trauma subgroup, the log-rank P valuewas 0.039.
Howard L. Corwin, M.D. Dartmouth–Hitchcock Medical Center Lebanon, NH 03756 howard.l.corwin{at}hitchcock.org
References
Corwin HL, Gettinger A, Pearl RG, et al. Efficacy of recombinant human erythropoietin in critically ill patients: a randomized controlled trial. JAMA 2002;288:2827-2835. [Free Full Text]
Vincent JL, Spapen HD, Creteur J, et al. Pharmacokinetics and pharmacodynamics of once-weekly subcutaneous epoetin alfa in critically ill patients: results of a randomized, double blind, placebo-controlled trial. Crit Care Med 2006;34:1661-1667. [CrossRef][Web of Science][Medline]
Oxman AD, Guyatt G. When to believe a subgroup analysis. In: Guyatte G, Rennie D, eds. Users' guides to the medical literature: a manual for evidence based clinical practice. Chicago: American Medical Association, 2002:553-65.
Corwin HL. Erythropoietin use in critically ill patients: forest and trees. CMAJ 2007;177:747-749. [Free Full Text]
Previous
