ers’ bureaus for Abbott, Aventis, Bayer, GlaxoSmithKline, Merck, Ortho-McNeil, Pfizer, and
Wyeth. All other authors: no conflicts.
Lionel A. Mandell,1 John G. Bartlett,2
Scott F. Dowell,3 Thomas M. File, Jr.,4
Daniel M. Musher,5,6 and Cynthia Whitney3
1
Division of Infectious Disease, McMaster
University, Henderson Hospital, Hamilton, Ontario,
Canada; 2Johns Hopkins University School of
Medicine, Baltimore, Maryland; 3Centers for
Disease Control and Prevention, Atlanta, Georgia;
4
Summa Health System, Northeastern Ohio
Universities College of Medicine, Akron, Ohio;
5
Baylor College of Medicine and 6Veterans Affairs
Medical Center, Houston, Texas
References
1. Yu VL, Ramirez J, Roig J, et al. Legionnaires
disease and the updated Infectious Diseases Society of America guidelines for communityacquired pneumonia. Clin Infect Dis 2004; 37:
1734 (in this issue).
2. Mandell LA, Bartlett JG, Dowell, SF, File TM,
Musher DM, Whitney C. Update of practice
guidelines for the management of community
acquired pneumonia in immunocompetent
adults. Clin Infect Dis 2003; 37:1405–33.
3. Vergis EN, Akbas E, Yu VL. Legionella as a cause
of severe pneumonia. Semin Resp Crit Care
Med 2000; 21:295–304.
4. Myburgh J, Nagel GJ, Petschel E. The efficacy
and tolerance of a three-day course of azithromycin in the treatment of communityacquired pneumonia. J Antimicrob Chemother
1993; 31(Suppl E):163–9.
5. Kuzman I, Soldo I, Schonwald S, Culig J. Azithromycin for treatment of community-acquired
pneumonia caused by Legionella pneumophila:
a retrospective study. Scand J Infect Dis 1995;
27:503–5.(Stout JE and Yu VL. NEJM 1997;
337:562)
6. Stout JE, Yu VL. Legionellosis. N Engl J Med
1997; 337:682–7.
Reprints or correspondence: Dr. Lionel A. Mandell, Henderson
Hospital, 711 Concession St., Hamilton, ON L8V 1C3, Canada
([email protected]).
Clinical Infectious Diseases 2004; 39:1737–8
2004 by the Infectious Diseases Society of America. All
rights reserved. 1058-4838/2004/3911-0037$15.00
Empirical Antifungal Therapy
in Febrile Neutropenic
Patients: Caution about
Composite End Points and
the Perils of P Values
In his recent article, Wingard [1] points
out several issues that continue to be a
source of debate in clinical trials of empirical antifungal therapy in febrile neutropenic patients [2]. The basic objective
of administering antifungals in clinical trials to these patients is the treatment of
occult fungal infections, not prevention of
infection; this is reflected in the US Food
and Drug Administration indication that
reads “empirical therapy for presumed
fungal infections in febrile neutropenic
patients” [3]. “Breakthrough” infections
may represent progression of an occult
fungal infection or a new infection. In
clinical practice, this distinction is less important, but in clinical trials, one must
differentiate between treatment and prevention of disease, because the designs of
such trials are quite different. It seems illogical to consider that a patient who is
receiving fluconazole prophylaxis and
who becomes febrile while neutropenic
changes to another drug to receive additional prophylaxis. The clinician changes
drugs because of concern for the presence
of an occult infection—that is, failure of
prophylaxis. One cannot prevent an infection the patient is presumed to have
already.
The presence of fever in these studies is
considered indicative of the presence of an
occult fungal infection. If the specificity of
fever for occult fungal infection is low,
then it calls into question the need for
empirical therapy in all patients in these
studies. Perhaps better selection criteria
and enrollment of patients who are more
likely to have occult infections would help
address this issue. Better diagnostic testing
would decrease the need for empirical
treatment. In any case, the process of randomization gives an equal probability of
the distribution of these nonspecific causes
of fever between the treatment arms of
the trial, so that any observed differences
would be associated with the drugs
administered.
When evaluating subgroup analyses [4]
of the components of the composite end
point, one should not consider a P value
of .05 to be significant. With a single comparison (the composite end point as a
whole), the chance of accepting a falsepositive result is 5%. When making multiple comparisons among the 5 individual
1738 • CID 2004:39 (1 December) • CORRESPONDENCE
components, the chance of accepting a
false-positive result increases to 20%.
Therefore, one should correct the P value
for multiple comparisons [5]. With 5
comparisons, a P value of !.01 is considered to be statistically significant. Differences in breakthrough infections do not
reach this degree of significance in any of
the empirical therapy trials.
Lastly, one cannot consider the components of the composite end point in
isolation. This is part of the reason for
having a composite end point [6]. One
cannot evaluate breakthrough infections
without examining discontinuations of
therapy and deaths, because breakthrough
infections cannot be measured in patients
who are no longer in the trial. Deaths may
be due to occult fungal infection or to
drug toxicity, which are difficult to ascertain with certainty in these patients. Again,
the process of randomization gives an
equal probability that deaths due to causes
other than fungal infection will be distributed randomly between the treatment
arms of the trial. Regardless of cause, if a
patient dies while receiving therapy, it is
difficult to argue that the drug was beneficial to that patient. In the voriconazole
trial [7], if one counts deaths as “breakthroughs,” the incidence of breakthrough
infections is 9.2% for each drug, showing
no difference between treatments [8].
Acknowledgment
Potential conflict of interest. J.H.P.: no
conflict.
John H. Powers
US Food and Drug Administration,
Rockville, Maryland
References
1. Wingard JR. Empirical antifungal therapy in
treating febrile neutropenic patients. Clin Infect
Dis 2004; 39(Suppl 1):S38–43.
2. Bennett JE, Powers J, Walsh TJ, et al. Forum
report: issues in clinical trials of empirical antifungal therapy in treating febrile neutropenic
patients. Clin Infect Dis 2003; 36(Suppl 3):
S117–22.
3. AmBisome [package insert]. Foster City, CA:
Gilead, 2004.
4. Freemantle N. Interpreting the results of secondary end points and subgroup analyses in
clinical trials: should we lock the crazy aunt in
the attic? BMJ 2001; 322:989–91.
5. Bland JM, Altman DG. Multiple significance
tests: the Bonferroni method. BMJ 1995; 310:
170.
6. Lubsen J, Kirwan BA. Combined endpoints:
can we use them? Stat Med 2002; 21:2959–70.
7. Walsh TJ, Pappas P, Winston DJ, et al. Voriconazole compared to liposomal amphotericin
B for empirical antifungal therapy in patients
with neutropenia and persistent fever. N Engl
J Med 2002; 346:225–34.
8. Powers JH. Voriconazole NDA 21–266 and
21–267, empirical antifungal therapy of febrile
neutropenic patients. In: Proceedings of the US
Food and Drug Administration Anti-Viral
Drugs Advisory Committee Meeting. 4 October
2001. Available at: http://www.fda.gov/ohrms/
dockets/ac/01/slides/3792s2_03_Powers.ppt.
Reprints or correspondence: Dr. John H. Powers, Office of
Drug Evaluation IV, Center for Drug Evaluation and Research,
US Food and Drug Administration, HFD-104, 9201 Corporate
Blvd., Rockville, MD 20850 ([email protected]).
Clinical Infectious Diseases 2004; 39:1738–9
This article is in the public domain, and no copyright is
claimed 1058-4838/2004/3911-0038
Reply to Powers
I agree with the comments by Powers [1],
but those considerations do not undermine my arguments about empirical antifungal therapy and how to apply the lessons of the trials to today’s practice.
There are (at least) 3 problems with the
designs of studies of empirical antifungal
therapy trials. First, neutropenic fever is a
poor surrogate for invasive fungal infection (IFI), yet fever has been routinely
used as the chief entry criterion in trials.
Causes of neutropenic fever are myriad,
and the frequency of IFI as a cause of fever
is too low. This is especially the case in
patients receiving fluconazole prophylaxis,
among whom the rate of documented IFI
as a cause of persistent fever is only 1%
[2]. Today, we know that the more likely
and more frequent causes of neutropenic
fever are the effects of proinflammatory
cytokines released by cellular injury from
cytotoxic chemotherapy; other causes include occult or slowly responding bacterial
or viral infections and drug-related fever.
Trials to test the efficacy of antifungal
agents should include only patients with
a higher level of IFI documentation. Without having some assurance that the patient
has IFI, one cannot truly evaluate the utility of antifungal therapy.
Second, use of fever as a criterion to
judge response to treatment is problematic: if the fever that prompted patient entry was never due to IFI, then the resolution of fever is also a poor gauge of
response to treatment. The routine use of
myeloid growth factors in leukemia and
oncology practice today and new sources
of stem cells in hematopoietic cell transplantation have led to the quickening of
neutrophil recovery, which frequently outraces the effects of antimicrobial agents
and is the chief reason for defervescence.
Third, the prevalent use of antifungal
prophylaxis complicates the interpretation
of study results. The target fungal pathogen, the time of IFI onset, and the magnitude of risk differ according to whether
prophylaxis was administered. For persons
who did not receive antifungal prophylaxis, the chief fungal pathogen was Candida species, the modal time of onset was
the second week of neutropenia, and the
risk of IFI was sizable (varying between
8% and 30%, depending on case mix, type
of chemotherapy regimen, and duration
of neutropenia). For persons who received
antifungal prophylaxis, the chief fungal
pathogen was Aspergillus species, the time
of onset was later (third and subsequent
weeks), and the risk was lower (as low as
1%).
So, if prior trials have fallen short, how
can future trials help? I offer 3 suggestions.
First, trials should use more stringent
entry criteria. In addition to fever, the
presence of signs and symptoms of IFI [3],
CT images [4], and rapid diagnostics [5,
6] all should be used. Admittedly, none of
the criteria are perfect, but each criterion
adds assurance that an IFI is present and
deserving of therapy.
Second, we should not mix patients
who have received prophylaxis with patients who have not in the study cohort.
As noted above, the target pathogens, the
time of onset, and the magnitude of risk
are too different. A drug may be quite
useful for one group but not for another.
Third, the “success” criterion used in
the past no longer serves as an adequate
surrogate for treatment response. If morestringent criteria for study entry are used,
we have additional information to
strengthen the judgment of response.
Acknowledgments
Potential conflicts of interest. J.R.W. has received grant support from Pfizer, Merck, Schering,
Ortho, and Fujisawa; consultation fees from Pfizer,
Merck, and Enzon; and lecture honoraria from
Pfizer, Merck, and Enzon.
John R. Wingard
Division of Hematology/Oncology, Department
of Medicine, Blood and Marrow Transplant
Program, University of Florida Shands Cancer
Center, Gainesville, Florida
References
1. Powers JH. Empirical antifungal therapy in febrile neutropenic patients: caution about composite end points and the perils of P values
[letter]. Clin Infect Dis 2004; 39:1738⫺9 (in
this issue).
2. Goodman JL, Winston DJ, Greenfield RA, et
al. A controlled trial of fluconazole to prevent
fungal infections in patients undergoing bone
marrow transplantation. N Engl J Med 1992;
326:845–51.
3. Gerson SL, Talbot GH, Hurwitz S, et al. Discriminant scorecard for diagnosis of invasive
pulmonary aspergillosis in patients with acute
leukemia. Am J Med 1985; 79:57–64.
4. Caillot D, Couaillier JF, Bernard A, et al. Increasing volume and changing characteristics of
invasive pulmonary aspergillosis on sequential
thoracic computed tomography scans in patients with neutropenia. J Clin Oncol 2001; 19:
253–9.
5. Maertens J, Verhaegen J, Lagrou K, et al. Screening for circulating galactomannan as a noninvasive diagnostic tool for invasive aspergillosis
in prolonged neutropenic patients and stem cell
transplantation recipients: a prospective validation. Blood 2001; 97:1604–10.
6. Odabasi Z, Mattiuzzi G, Estey E, et al. Beta-dglucan as a diagnostic adjunct for invasive fungal infections: validation, cutoff development,
and performance in patients with acute myelogenous leukemia and myelodysplastic syndrome. Clin Infect Dis 2004; 39:199–205.
Reprints or correspondence: Dr. John R. Wingard, PO Box
100277, 1600 S.W. Archer Rd., Gainesville, FL 32610-0277
([email protected]).
Clinical Infectious Diseases 2004; 39:1739
2004 by the Infectious Diseases Society of America. All
rights reserved. 1058-4838/2004/3911-0039$15.00
CORRESPONDENCE • CID 2004:39 (1 December) • 1739