¹Division of Infectious Diseases, Department of Internal Medicine, Center for the Study of Emerging and Reemerging Pathogens, University of Texas Medical School, Houston, TX, USA

²The Myeloma and Transplantation Research Center, University of Arkansas for Medical Sciences, Little Rock, AR, USA

³2nd Department of Internal Medicine, Athens, University, Greece

⁴Department of Leukemia, MD Anderson Cancer Center, Houston, TX, USA

Correspondence to: J H Rex, Division of Infectious Diseases, University of Texas Medical School-Houston, 6431 Fannin, 1728 JFB, Houston, TX 77030, USA; Fax: 713 500 5495

^aCurrent address: The Myeloma and Transplantation Research Center, University of Arkansas for Medical Sciences, Little Rock, Arkansas 77205, USA

Leukemia (2002) 16, 1197-1199. DOI: 10.1038/sj/leu/2402495

TO THE EDITOR

Although antifungal prophylaxis is thought to reduce invasive fungal infections in at least some neutropenic patients,¹ limited data exists on its utility in adults receiving chemotherapy for acute myelogenous leukemia (AML). Despite the lack of focused data on its utility in AML, prophylaxis is now commonly used. As the available large studies report patients with a mixture of underlying types of leukemia, we have retrospectively examined the impact of antifungal prophylaxis in a large consecutive series of adults with AML. To evaluate simultaneously a control population that did not receive prophylaxis, we reviewed the medical records, mycological reports and autopsy reports of patients 14 years of age or older treated at the University of Texas MD Anderson Cancer Center for AML from January 1988 to December 1992. As noted below, this period spans the introduction of fluconazole and a local change in practice towards extensive use of antifungal prophylaxis. Only patients who received remission-induction chemotherapy were eligible. Patients were excluded if they had received systemic antifungal therapy or had documented fungal or mould infection within the month prior to study entry or if they had Aspergillus colonization of the nasal cavity.

Three hundred and thirty-one patients with 974 episodes of neutropenia met enrolment criteria. Of these, complete records were available on 851 episodes of neutropenia in 322 patients. After excluding 18 episodes of neutropenia because of an invasive fungal infection during a prior episode of neutropenia, 833 episodes of neutropenia in 322 patients remained. These episodes of neutropenia spanned two significant time frames. Between January 1988 and December 1990 there were 453 qualifying episodes of neutropenia in 186 patients. Fluconazole was largely unavailable during this time, a systemic antifungal agent was given in only 174 (38%) of episodes, and that agent was amphotericin B (49%), fluconazole (34%) and other agents (17%). However, systemic antifungal prophylaxis was given during 305 (80%) of the 380 episodes of neutropenia in 151 patients that occurred after January 1991 (15 patients had episodes both before and after December 1990) and was fluconazole in 304 (99.7%) of episodes. Other than 90 patients who were randomized to receive either fluconazole or amphotericin B prophylaxis between May 1989 and December 1990,² the decision to give prophylaxis during either time frame was at the discretion of the attending physician.

Patients were considered to have received potentially effective antifungal prophylaxis if an antifungal agent was initiated within 4 days of the onset of an episode of neutropenia and if the drug was given by a route that reliably produced meaningful serum and tissue levels of the drug. Oral polyene therapy was not counted as potentially effective prophylaxis. Systemic fungal infection was diagnosed by the presence of fungi in blood or tissue samples. Hematogenously disseminated candidiasis was defined as involvement of one or more internal organs via bloodstream inoculation.³ Candidemia was defined as fungemia without clinically apparent involvement of any organ, retinal or skin site. Infections due to moulds (eg Aspergillus) required a positive culture or pathology for a mould in association with a compatible clinical syndrome (eg pulmonary infiltrate). These definitions are similar to the proven infection categories of recently proposed research criteria for diagnosis of invasive fungal infections.⁴ Relapse in AML was defined as the appearance of 5% blasts or cytopenia of the bone marrow secondary to persistent leukemia (eg <5% blasts, but a current or continuing cytogenic aneuploidy).

Analyses considering only the first episode of neutropenia for each patient and all episodes of neutropenia (considered as separate occurrences) were performed. Assessed risk factors for fungal infection included the status of underlying disease, Zubrod performance score, early risk of mortality (ERM) score,⁵ duration of neutropenia (absolute neutrophil count of <1 ´ 10⁹/l), and use of antibacterial therapy (prophylactic and therapeutic), and the nature of the antibacterial therapy.

The patients given and not given prophylaxis were not significantly different at the first episode of neutropenia (P > 0.05 by Fisher's exact test for 2 ´ 2 comparisons, the ² test for more complex comparisons, or t test for continuous variables) with an age of 53 ± 1 (mean ± s.e.), 53% male, 17 ± 1 days of neutropenia, administration of granulocyte or granulocyte-macrophage colony-stimulating factor in 39% of cases, and 30 ± 1 days in the hospital. The majority (93%) of the patients had AML, including acute monocytic leukemia (1%), acute myelomonocytic leukemia (6%) and acute promyelocytic leukemia (5%), with the remainder having acute undifferentiated leukemia (4%) or refractory anemia with excess of blasts (3%). Many slightly different chemotherapy regimens were used during the study period, but most were based on cytosine arabinoside and there were no meaningful shifts in therapy regimens. The patients receiving prophylaxis did have a slightly lower (better) early risk of mortality score⁵ at the first episode of neutropenia (0.29 ± 0.02 vs 0.22 ± 0.02, P = 0.012).

Fluconazole was the most frequently used agent and was given at a median dose of 400 mg/day (Table 1). Amphotericin B was given intravenously at a median dose of 40 mg/day. Antifungal prophylaxis was associated with fewer invasive fungal infections and less frequent use of empiric amphotericin B therapy (Table 1). Antifungal prophylaxis appeared to act primarily by reducing the rate of yeast infections rather than mould infections (Table 2). Slightly more than 50% of the infections were bloodstream infections.

To assess further the effect of antifungal prophylaxis on risk of invasive fungal infection and in an effort to compensate for differences between the treatment groups, stepwise multivariate logistic regression was used to examine the effect of a variety of factors on risk for fungal infection (Table 3). Factors considered included age, gender, severity of illness (as ERM score and as Zubrod performance score), days of neutropenia, type of underlying leukemia, status of leukemia (initial therapy vs relapse), use of prophylactic broad-spectrum antibiotics (analyzed both as any use and specifically as use of a quinolone- or a sulfa-based regimen), treatment with corticosteroids, year of treatment, type of chemotherapy, and type of antifungal prophylaxis. Only use of prophylaxis, duration of neutropenia, increasing age, and male gender emerged as significant factors. The type of prophylaxis did not seem relevant: both fluconazole and amphotericin B were associated with protection, although the protective effect of fluconazole tended towards greater efficacy than non-fluconazole-based regimens. The effect of gender was surprising, has not been seen in previous studies, but was seen only at the first episode of neutropenia. In a stratified analysis, antifungal prophylaxis was associated with protection in both men and women. For the first episode of neutropenia, antifungal prophylaxis reduced the rate of invasive fungal infection from 30% to 6% for men and from 12% to 2% for women. Likewise, prophylaxis reduced the rate of invasive fungal infection at all episodes of neutropenia from 18% to 6% for men and from 8% to 4% for women. Thus, gender emerged in the logistic regression because of this difference in baseline rates of fungal infection. With the exception of a trend towards a slightly lower (better) early risk of mortality score in the women (0.22 ± 0.02 vs 0.27 ± 0.02, P = 0.09), the men and women were otherwise similar. The cause of the increased baseline rate of invasive fungal infections in men at the first episode of neutropenia is unexplained.

Death within 30 days of the first episode of neutropenia is shown in Table 1 and was also analyzed by step-wise logistic regression using underlying disease, disease status, severity of illness (as measured by ERM score), use of prophylaxis, and gender. Only ERM score was significant, with the likelihood of death increasing with increasing ERM score (odds ratio = 1.02 per 0.01 point, P = 0.0001).

Our results indicate that antifungal prophylaxis is associated with a reduction in the rate of invasive fungal infections in adults undergoing standard chemotherapy for acute myelogenous leukemia. Although our data are from patients cared for almost a decade ago, these results are relevant as they are drawn from a large group of relatively homogeneous patients and because the reasonably common use of antifungal prophylaxis in many settings makes it difficult to acquire new data on patients not given prophylaxis.

As fluconazole was the most commonly used antifungal agent, it is not surprising that the major effect we observed was a reduction in the rate of yeast infections. Fluconazole has no clinically meaningful activity vs Aspergillus and an effect on the rates of this infection would thus not be expected. While increased rates of mould infections in association with reduction in yeast infections have been reported by some,^6,7 we did not observe this.

Our findings with regard to the increased risk for invasive fungal infections in male patients are interesting but of unclear significance. Both men and women appeared to benefit from antifungal prophylaxis. Our study also confirms previous findings of the association between invasive fungal infections and the proportion of time with severe neutropenia.⁸ Similar to other studies in cancer patients, severity of illness was predictive of mortality in this patient population.⁵

Our study suffers from all the limitations of retrospective analyses, including the lack of uniformity in the diagnosis and management of the underlying disease and its complications. Nonetheless, this is the largest study published to date of prophylaxis in a uniform population of adults with AML and its results indicate that at least some patients with AML would benefit from antifungal prophylaxis.

1 Gøtzsche PC, Johansen HK. Meta-analysis of prophylactic or empirical antifungal treatment versus placebo or no treatment in patients with cancer complicated by neutropenia. Br Med J 1997; 314: 1238-1244.

2 Bodey GP, Anaissie EJ, Elting L, Estey E, O'Brien S, Kantarjian H. Antifungal prophylaxis during remission induction therapy for acute leukemia. Fluconazole versus intravenous amphotericin B. Cancer 1994; 73: 2099-2106. MEDLINE

3 Bodey GP, Anaissie EJ, Edwards JE. Definitions of Candida infections. In: Bodey GP (ed.). Candidiasis: Pathogenesis, Diagnosis, and Treatment Raven Press: New York, 1993, 407-408.

4 Ascioglu S, De Pauw B, Bille J, Crokaert F, Denning DW, Donnelly P, Edwards JE, Erjavec Z, Fiere D, Lortholary O, Maertens J, Meis J, Patterson T, Rex JH, Ritter J, Selleslag D, Shah PM, Stevens DA, Walsh TJ. Analysis of definitions used in clinical research on invasive fungal infections: consensus proposal for new, standardized definitions. 39th Interscience Conference on Antimicrobial Agents and Chemotherapy, San Francisco, CA 1999; Abstr. 1639.

5 Estey E, Smith TL, Keating MJ, McCredie KB, Gehan EA, Freireich EJ. Prediction of survival during induction therapy in patients with newly diagnosed acute myeloblastic leukemia. Leukemia 1989; 3: 257-263. MEDLINE

6 van Burik JAH, Leisenring W, Myerson D, Hackman RC, Shulman HM, Sale GE, Bowden RA, McDonald GB. The effect of prophylactic fluconazole on the clinical spectrum of fungal diseases in bone marrow transplant recipients with special attention to hepatic candidiasis - an autopsy study of 355 patients. Medicine 1998; 77: 246-254. MEDLINE

7 Groll AH, Shah PM, Mentzel C, Schneider M, Just-Nuebling G, Huebner K. Trends in the postmortem epidemiology of invasive fungal infections at a university hospital. J Infect 1996; 33: 23-32. MEDLINE

8 O'Donnell MR, Schmidt GM, Tegtmeier BR, Faucett C, Fahey JL, Ito J, Mademanee A, Niland J, Parker P, Smith EP, Snyder DS, Stein AS, Blume KG, Forman SJ. Prediction of systemic fungal infection in allogeneic marrow recipients: Impact of amphotericin prophylaxis in high-risk patients. J Clin Oncol 1994; 12: 827-834. MEDLINE

Table 1 Prophylaxis and outcome

Table 2 Microbiology

Table 3 Factors predicting increased risk of invasive fungal infection by logistic regression