Candida Krusei Fungemia an Escalating Serious Infection in Immunocompromised Patients

ORIGINAL INVESTIGATION Candida krusei Fungemia An Escalating Serious Infection in Immunocompromised Patients

Jalal Abbas, MD; Gerald P. Bodey, MD; Hend A. Hanna, MD, MPH; Masoud Mardani, MD; Essam Girgawy, MD; Dima Abi-Said, PhD; Estella Whimbey, MD; Ray Hachem, MD; Issam Raad, MD

Background: Candida krusei is inherently resistant to fungemia more often had leukemia than patients with fluconazole and is emerging as a frequent cause of funge- C albicans (77% vs 11%; P=.02), whereas catheter- mia in patients with hematologic malignant neoplasms. related infections were more common among patients with C albicans fungemia (42% vs 0%; PϽ.001). Objective: To determine the risk and prognostic fac- Patients with C krusei fungemia had a lower response tors associated with C krusei fungemia in comparison with rate (51% vs 69%; P=.05), largely because they more Candida albicans fungemia in patients with cancer. frequently were neutropenic and had disseminated infection. Mortality related to fungemia was 49% in Methods: Retrospective study of 57 cases of C krusei the cases with C krusei vs 28% in C albicans. Multiple fungemia occurring at the M. D. Anderson Cancer Cen- logistic regression analysis showed that persistent neu- ter, Houston, Tex, from 1989 to 1996. The C krusei cases tropenia (P=.02) and septic shock (P=.002) were pre- were compared with 57 cases of C albicans fungemia with dictors of poor prognosis. respect to demographics, underlying cancer, Acute Physi- ology and Chronic Health Evaluation II score, immuno- Conclusion: In neutropenic patients, C krusei funge- suppression status, chemotherapy, and the use of cen- mia is associated with high mortality. It should be tral venous catheters, as well as fluconazole prophylaxis. suspected in patients with leukemia who are receiving fluconazole prophylaxis and should be treated aggres- Results: At our institution, C krusei accounted for 5% sively with an amphotericin B regimen. of fungemias during 1989 through 1992 and for 10% during 1993 through 1996. Patients with C krusei Arch Intern Med. 2000;160:2659-2664

OR MANY years, nearly all termine risk factors that were unique for cases of candidemia were the acquisition of C krusei fungemia. caused by Candida albicans. During the past decade, Can- RESULTS dida tropicalis emerged as an important pathogen, especially among pa- FEATURES OF C krusei INFECTION F 1-3 tients with cancer. More recently, other Candida species have proved to be respon- There were 57 cases of C krusei fungemia sible for an increased proportion of can- from 1989 to 1996. The frequency in- didemias, and Candida krusei has be- creased significantly from 2.5 per 10000 come one of the most frequent of these hospital discharges during 1989 through emerging pathogens, especially among pa- 1992 to 5.7 per 10000 during 1993 tients with acute leukemia.2,4 Candida kru- through 1996 (P=.004). Candida krusei ac- From the Sections of Infection sei is inherently resistant to fluconazole, counted for 5% of all candidemias during Control and Infectious Diseases, and many of the infections caused by this the former period and 10% during the lat- Department of Internal organism have been associated with the ter period. Nearly all patients had hema- Medicine Specialties prophylactic or therapeutic use of this an- tologic malignant neoplasms, especially (Drs Abbas, Bodey, Hanna, tifungal agent.5,6 Since most institutions acute leukemia (Table 1). Only 8 infec- Mardani, Girgawy, Whimbey, Hachem, Raad) and have had limited experience with C kru- tions (14%) were considered to be com- Department of Neurosurgery sei fungemia, we reviewed all cases occur- munity acquired. (Dr Abi-Said), The University ring at our institution during a 7-year pe- Forty-nine patients (86%) with C kru- of Texas M. D. Anderson riod. A control group of patients with sei fungemia had received fluconazole pro- Cancer Center, Houston. C albicans fungemia was selected to de- phylaxis; this proportion increased from

(REPRINTED) ARCH INTERN MED/ VOL 160, SEP 25, 2000 WWW.ARCHINTERNMED.COM 2659

Downloaded From: https://jamanetwork.com/ on 09/28/2021 PATIENTS AND METHODS neutropenic patient in whom a catheter was not the source of fungemia. Neutrophil counts were categorized as less than The records of the Microbiology Laboratory of The Uni- 0.50ϫ109/L and less than 0.10ϫ109/L. Neutropenia was versity of Texas M. D. Anderson Cancer Center, Houston, defined in this study as a neutrophil count of less than were reviewed for the period from January 1, 1989, through 0.50ϫ109/L, since the results were similar for the 2 cat- December 31, 1996, to identify all cases of hematogenous egories of neutropenia. Shock was defined as a decrease in candidiasis. Of the 885 cases, 60 were caused by C krusei, systolic blood pressure of 40 mm Hg or more, or a systolic and the medical records of 57 of these cases were available blood pressure less than 90 mm Hg in a previously nor- for review. A case-control study was conducted by select- motensive patient, not related to other possible causes. An ing 57 patients with C albicans fungemia as the control infection was considered to be catheter related if the cath- group. The two groups were matched with respect to time eter tip was notably colonized with the same Candida spe- of the first positive blood culture for C krusei and C albi- cies as the bloodstream (Ն15 colony-forming units by the cans fungemia within each of the two periods, 1989 through roll plate semiquantitative method) or the quantitative blood 1992 and 1993 through 1996. culture collected through the central venous catheter had The medical records of all cases and controls were re- a 10-fold greater colony count than the concurrent periph- viewed and the following information was determined for eral venous quantitative blood culture specimen. the 30 days before collection of the first positive blood cul- Response was defined as the resolution of all clinical ture and subsequently during the course of infection: age, manifestations with blood cultures negative for Candida spe- sex, underlying malignant neoplasm, Acute Physiology and cies for at least 1 week after therapy. Failure to respond Chronic Health Evaluation II score (at onset of funge- was defined as the persistence of the clinical signs and symp- mia), duration of hospitalization, duration of stay in in- toms of the fungal infection or persistent candidemia caused tensive care unit, and duration of neutropenia. In addi- by the same Candida species after onset of therapy. Death tion, we recorded during this interval the administration was attributed to Candida infection if the patient had dem- of antibiotics, adrenal corticosteroids, total body irradia- onstrated no response to therapy and there was no other tion, cancer chemotherapeutic agents, immunosuppres- obvious cause of death, such as major hemorrhage or other sive agents (cyclosporine, tacrolimus), parenteral alimen- infection. tation, and antifungal prophylaxis. DATA ANALYSIS DEFINITIONS The incidence density of hematogenous candidiasis was A patient was determined to have candidemia if C krusei or computed as the number of cases during a period per 10000 C albicans was isolated from at least 1 blood culture speci- hospital discharges. Frequencies and descriptives of de- men, associated with fever or signs of organ infection. mographic and clinical characteristics of the patient popu- Fungemias occurring between 1989 and 1992 have been lation were performed. The ␹2 test and the Fisher exact test included in previous reports.2,7 Patients were considered to were used to determine categorical predictors of infec- have disseminated infection if, in addition to fungemia, tions and outcome. Continuous variables were compared there was clinical or histopathological evidence of infec- by means of the t test. Multiple predictors were examined tion in at least 1 internal organ and a Candida species was for significance by multiple logistic regression with the use identified, or the same species was isolated from a tissue of backward elimination by the Wald test. P values of less specimen. Infection was also considered to be dissemi- than .05 were considered statistically significant. All sta- nated when the Candida species was isolated from blood tistical analyses were performed with SPSS, version 8.0 for culture specimens collected on 3 separate days from a Windows (SPSS, Inc, Chicago, Ill).8

78% to 90% during the 2 periods, and the mean dura- days). Twenty-one (37%) of the patients had dissemi- tion of prophylaxis increased from 12 days (range, 4-32 nated infection, and 18 (86%) of these patients were neu- days) to 19 days (range, 6-47 days). The response rate tropenic at the onset of their infection. None of theC kru- for the 49 patients who received fluconazole prophy- sei fungemias was considered to be catheter related. laxis was similar to that for the 8 patients who did not Twenty-nine (51%) of the patients responded to (51% [25 patients] vs 50% [4 patients]). One of the 8 pa- therapy, and the remaining 28 patients died with their tients was treated empirically with fluconazole and re- infection (Table 2). Twenty-four of the latter 28 pa- sponded. Eleven patients (all with neutropenia) devel- tients were considered to have died primarily of their in- oped C krusei fungemia while receiving amphotericin B fection. Among those patients who died, 22 (79%) re- for persistent fever presumed to be caused by fungal in- mained neutropenic throughout their infection and 18 fection. The dose administered was 0.35 to 0.5 mg/kg per (64%) had disseminated infection. The mortality rate was day for 2 to 36 days (mean, 13 days). Only 3 of the pa- 81% (17 patients) for patients with persistent neutrope- tients recovered from their infection, which was associ- nia and 86% (18 patients) for patients with dissemi- ated with resolution of neutropenia and an increase in nated infection. Only 11 patients (31%) who were not the dose of amphotericin B. neutropenic or whose neutropenia resolved during Fifty-one patients (89%) were neutropenic at the on- therapy died of their infection. None of the patients who set of their infection, and the mean duration of neutro- had solid tumors or were in remission of their malig- penia before onset of infection was 18 days (range, 3-95 nant neoplasm died of their infection.

(REPRINTED) ARCH INTERN MED/ VOL 160, SEP 25, 2000 WWW.ARCHINTERNMED.COM 2660

Downloaded From: https://jamanetwork.com/ on 09/28/2021 Table 1. Patient Demographics and Clinical Characteristics*

Candida krusei Candida albicans Patient Characteristics (n = 57) (n = 57) P Demographics Age, median (range), y 50 (13-84) 57 (1-85) .05 Sex, No. (%) M 37 (65) 27 (47) .06 F 20 (35) 30 (53) Clinical characteristics Underlying disease, No. (%) Leukemia 44 (77) 6 (11) Other hematologic malignant neoplasm 9 (16) 8 (14) Ͻ.001 Solid tumor 4 (7) 43 (75) APACHE II score, median (range) 14 (5-28) 14 (7-22) .85 Received a BMT within past year, No. (%) 16 (28) 2 (4) Ͻ.001 ICU within month before fungemia, No. (%) 12 (21) 10 (18) .15 Duration of ICU stay, median (range), d 9 (1-90) 7 (2-28) .51 Immunosuppression status before infection, No. (%) Neutrophils Ͻ0.50 x 109/L 51 (89) 17 (30) Ͻ.001 Chemotherapy 47 (82) 28 (49) Ͻ.001 Adrenal corticosteroids 29 (51) 19 (33) .06 Cyclosporine 9 (16) 0 (0) .001 Tacrolimus 2 (4) 0 (0) .25 Methotrexate 10 (18) 2 (4) .02 Total body irradiation, No. (%) 5 (9) 0 (0) .03 Duration of previous neutropenia, mean (range), d 18 (3-95) 10 (2-30) .02 Central venous catheter, No. (%) 54 (95) 50 (88) .19 Catheter-related fungemia, No. (%) 0 (0) 24 (42) Ͻ.001 WBC transfusion, No. (%) 4 (7) 0 (0) .06 G-CSF, No. (%) 41 (72) 11 (19) Ͻ.001 Dissemination, No. (%) 21 (37) 12 (21) .06 Total parenteral nutrition, No. (%) 16 (28) 25 (44) .79 Use of prophylactic fluconazole, No. (%) 49 (86) 6 (11) Ͻ.001 From 1989-1992 14/18 (78) 2/18 (11) Ͻ.001 From 1993-1996 35/39 (90) 4/39 (10) Ͻ.001 Duration of prophylactic fluconazole, mean (range), d From 1989-1992 12 (4-32) 9 (9-9) .58 From 1993-1996 19 (6-47) 26 (3-76) .32

*APACHE indicates Acute Physiology and Chronic Health Evaluation; BMT, bone marrow transplant; ICU, intensive care unit, WBC, white blood cell; and G-CSF, granulocyte colony-stimulating factor.

Nine patients were initially treated with fluconazole, and 2 responded to this therapy. Both of these pa- Table 2. Differences in Outcome Between tients had neutropenia at the onset of their infection that Candida krusei and Candida albicans resolved during therapy. The remaining 7 patients’ therapy Candida Candida was changed to amphotericin B, usually when the fun- krusei albicans gus was identified as C krusei. Forty-nine patients re- Outcome Event (n = 57) (n = 57) P ceived amphotericin B deoxycholate or a lipid formula- Overall response, No. (%) 29 (51) 36 (69)* .05 tion of amphotericin B, and 25 (51%) of the cases Mortality related to fungemia, 28 (49) 16 (28) .08 responded. Among the 25 responders, 21 (84%) were not No. (%) neutropenic or had recovered from neutropenia. Response among patients treated 2/9 (22)† 23/29 (79)‡ .003 Univariate analysis indicated that a poor response was with fluconazole, No. (%) associated with persistent neutropenia, disseminated in- Response among patients treated 25/49 (51) 10/14 (71)‡ .18 with any amphotericin B fection, septic shock, other concomitant fungal infection, preparation, No. (%) and immunosuppressive therapy (Table 3). Multiple logistic regression analysis identified that persistent neu- *Response could not be determined in 5 cases. tropenia (P=.02) and septic shock (P=.002) were the only †Eight of these patients had also received fluconazole as prophylaxis, and significant multiple predictors for a poor prognosis. only 1 of them responded to fluconazole therapy. ‡Nine patients with C albicans never received any antifungal therapy. DEMOGRAPHIC AND CLINICAL GROUP CHARACTERISTICS (PՅ.05) were identified between the patients with C krusei and C albicans fungemia. Patients in the former group The characteristics of the 2 patient groups are described were somewhat younger and were predominantly pa- in Table 1. Several statistically significant differences tients with acute leukemia. A greater proportion of these

(REPRINTED) ARCH INTERN MED/ VOL 160, SEP 25, 2000 WWW.ARCHINTERNMED.COM 2661

Downloaded From: https://jamanetwork.com/ on 09/28/2021 Table 3. Effect of Different Prognostic Factors on Response of Candida krusei Infection*

Factor Present Factor Absent

No. (%) No. (%) Factor No. of Patients Who Responded No. of Patients Who Responded P Underlying disease .12 Leukemia 44 20 (45) 13 9 (69) .13 Other hematologic malignant neoplasms 9 5 (56) 48 24 (50) .52 Solid tumor 4 4 (100) 53 25 (47) .06 Dissemination, No. (%) 21 3 (14) 36 26 (72) Ͻ.001 Septic shock 19 1 (5) 38 20 (53) Ͻ.001 Concomitant fungal infection 6 0 51 28 (55) .01 Any immunosuppressive agent† 31 11 (36) 26 18 (69) .01 Prophylactic fluconazole† 49 25 (51) 8 4 (50) .65 Persistent neutropenia 21 4 (19) 36 25 (69) Ͻ.001 Inserted CVC 54 26 (48) 3 3 (100) .24 G-CSF† 41 22 (54) 16 7 (44) .5 APACHE II score, median (range) 13 (7-18)‡ 14 (10-22)§ .07

*CVC indicates central venous catheter; G-CSF, granulocyte colony-stimulating factor; and APACHE, Acute Physiology and Chronic Health Evaluation. Results by univariate analysis are given. †Within 30 days before infection. ‡For patients who responded. §For patients who did not respond.

patients had received cancer chemotherapy or total to 4.5% of fungemias.2,9,10 In a literature review of can- body irradiation, received a bone marrow transplant didemias, Wingard1 found that only 4% were caused by within the preceding year, were neutropenic, and had C krusei. Our analysis represents the largest series of prolonged neutropenia. The proportion of patients who C krusei fungemias from a single institution. The largest received fluconazole prophylaxis was significantly previous report was a collated review of 62 cases.6 Ap- higher among patients with C krusei fungemia. Catheter- proximately 90% of cases have occurred in patients with related fungemia occurred only among patients with C acute leukemia or aplastic anemia or bone marrow trans- albicans fungemia. Disseminated infection occurred plant recipients, and very few cases have been reported more often in patients with C krusei fungemia (37% vs in patients who did not have cancer. Neutropenia has been 21%; P=.06). Interestingly, 18 (86%) of the 21 patients present in 80% to 90% of patients with this infec- with disseminated C krusei infection were neutropenic, tion.6,9,10 In our matched-pair analysis, acute leukemia and compared with only 4 (33%) of the 12 patients with dis- neutropenia were significant predisposing factors for seminated C albicans infection. The majority of both C C krusei fungemia. krusei and C albicans infections were considered to be The frequency of C krusei fungemia has increased nosocomial. substantially at several institutions, beginning in the A significantly greater proportion of patients with early 1990s. Candida krusei is inherently resistant to flu- C krusei fungemia had leukemia (P=.02) and had re- conazole, and this increase in infections has been attrib- ceived antifungal prophylaxis (PϽ.002), most often with uted to the use of fluconazole, especially for fungal pro- fluconazole (P=.03). On the other hand, a significantly phylaxis. Wingard et al5 first reported this association greater proportion of patients with C albicans fungemia among neutropenic patients treated in 1989 to 1990. had catheter-related infections (PϽ.001). Candida krusei colonization was found in 41% of The response rate was higher among patients with patients who received fluconazole prophylaxis com- C albicans infections (69% vs 51%; P=.05), largely be- pared with 17% of patients who received no prophy- cause more patients with C krusei fungemia were neu- laxis. The frequency of disseminated infection was 8% tropenic and had disseminated infection (Table 2). Among and 1%, respectively. They also found that norfloxacin patients who died, a higher proportion of patients with antibacterial prophylaxis increased the risk of C krusei C albicans fungemia were not neutropenic or had recov- colonization. However, Merz et al11 reported an increase ered from neutropenia (11 patients [69%] vs 11 [39%]; in C krusei colonization and infection among neutrope- P=.06). Fluconazole and amphotericin B were equally nic patients treated at the same institution from 1977 to efficacious for treatment of C albicans fungemia (79% vs 1985, before fluconazole was available. At that time, the 71%). A trend was found to suggest that amphotericin B frequency of colonization was 12.4%, and 1.2% devel- was more effective for the treatment of C albicans than oped fungemia. Some of these patients had received C krusei infection (71% vs 51%; P=.18). miconazole prophylaxis. At our institution, the frequency of C krusei funge- COMMENT mia doubled from 5% in 1989 to 1992 to 10% in 1993 to 1996. In 1993, fluconazole became widely used as a pro- Candida krusei is an infrequent cause of fungemia. At most phylactic agent in patients with hematologic malignant institutions, this Candida species accounts for only 3.5% neoplasms. In addition, fluconazole usage was a signifi-

(REPRINTED) ARCH INTERN MED/ VOL 160, SEP 25, 2000 WWW.ARCHINTERNMED.COM 2662

Downloaded From: https://jamanetwork.com/ on 09/28/2021 cant predisposing factor in our analysis. In the review by vitro than isolates of C albicans.18,19 Berrouane et al20 Goldman et al,6 82% of the patients had received flu- considered C krusei to be a multiple-drug–resistant conazole before their infection. Nguyen et al10 reported pathogen. In our study, 19% of C krusei fungemias first 34 cases of candidemia occurring while patients were occurred while patients were receiving 0.35 to 0.5 mg receiving fluconazole, and 7 were caused by C krusei. of amphotericin B per kilogram per day, and Blumberg They found fluconazole use to be an important predis- and Reboli21 reported breakthrough fungemia when posing factor for C krusei fungemia. However, not all even higher doses of amphotericin B were administered. studies have linked fluconazole use to increased coloni- In our study, response rates have been higher when zation and infection by this yeast. Iwen et al12 found that daily doses of 1 mg or more of amphotericin B per kilo- 4% of invasive Candida infections were caused by C kru- gram were used. It is somewhat difficult to interpret sei and none of the patients had received fluconazole. In these data because most infections occur in neutropenic 2 multi-institutional randomized studies of fluconazole patients and response to amphotericin B is dependent prophylaxis in neutropenic patients, the frequency of on neutrophil recovery.16 Although C krusei is inher- C krusei colonization and infection was similar among ently resistant to fluconazole, occasional infections have patients receiving fluconazole and placebo.13,14 Likewise, responded to fluconazole, including those in 2 neutro- in a randomized study of bone marrow transplant recipi- penic patients in our study whose neutrophil counts ents who received fluconazole or placebo for 75 days recovered during therapy. Itraconazole usually is active after transplantation, there was no significant difference against C krusei in vitro, but its use for the treatment of in colonization rates and no infections in either group.15 serious infections has been limited because of the lack It may be that there are nosocomial sources of C krusei of an intravenous preparation and the variability of within some hospitals and that colonization and infec- absorption of the initial oral preparation.22 tion are facilitated by the selective effect of fluconazole Candida species other than C albicans now account usage. for nearly half of the cases of hematogenously dis- Several studies have found that about 70% of pa- seminated candidiasis at many institutions.1,10 Although tients develop colonization by C krusei before the onset C krusei accounts for only a modest proportion of these of infection.6,9,10 Merz et al11 found that the gastrointes- infections, it is increasing substantially at some leukemia tinal tract was the most frequent site of colonization, fol- and bone marrow transplant centers where fluconazole lowed by the respiratory tract. Goldman et al6 found mu- is being used extensively for prophylaxis. While the cosal damage of the gastrointestinal tract in 68% of infected association is obviously of concern, it is offset by the patients. Hence, the sequence of events leading to C kru- overall beneficial effect of this antifungal prophylaxis sei infection may be colonization facilitated by flucona- against other Candida species.2,15 In patients with hema- zole administration, local infection caused by mucosi- tologic malignant neoplasms, C krusei fungemia is asso- tis, and subsequent hematogenous dissemination caused ciated with a high frequency of dissemination and mor- by neutropenia. Although the majority of patients have tality. Since C krusei may be less susceptible to intravascular catheters inserted before infection, our study amphotericin B than other Candida species, the daily and that of Merz et al11 did not find these catheters to be dose of amphotericin B used as empirical antifungal a source of fungemia. therapy probably should be 1 mg/kg at institutions In our study, only 37% of patients were considered where C krusei infections are prevalent. to have disseminated infection, which may be an under- estimate. The true frequency of disseminated infection Accepted for publication February 1, 2000. is often difficult to ascertain because this infection is usu- We thank Jack Thornby, PhD, for his revision of the ally manifested by many small abscesses involving mul- study and for his valuable comments. tiple organs that do not cause organ-specific symptoms. Reprints: Hend A. Hanna, MD, MPH, The University Hence, many disseminated infections are detected only of Texas M. D. Anderson Cancer Center, 1515 Holcombe at autopsy examination. In 2 studies among patients with Blvd, Houston, TX 77030. C krusei fungemia who died and underwent autopsy examination, 67% and 100% were found to have dissemi- REFERENCES nated infection.9,11 The mortality rate in our study among patients with C krusei fungemia was 49% overall but was 93% among 1. Wingard JR. Importance of Candida species other than C albicans as pathogens in oncology patients. Clin Infect Dis. 1995;20:115-125. patients with persistent neutropenia and 83% among those 2. Abi-Said D, Anaissie E, Uzun O, Raad I, Pinzcowski H, Vartivarian S. The epide- with disseminated infection. The mortality rate was lower miology of hematogenous candidiasis caused by different Candida species. Clin among patients with C albicans fungemia (28%), which Infect Dis. 1997;24:1122-1128. was because of a lower frequency of neutropenia and a 3. Komshian SV, Uwaydah AK, Sobel JD, Crane LR. Fungemia caused by Candida higher frequency of catheter-related fungemias. Previ- species and Torulopsis glabrata in the hospitalized patient: frequency, characteristics, and evaluation of factors influencing outcome. Rev Infect Dis. 1989; ous studies of candidemia have found mortality rates of 11:379-390. 90% or higher among patients with cancer who have per- 4. Wingard JR. Infections due to resistant Candida species in patients with can- sistent neutropenia or disseminated infection.16,17 Other cer who are receiving chemotherapy. Clin Infect Dis. 1994;19(suppl 1):S49- studies have found mortality rates of 60% to 80% in pa- S53. 6,9,11 5. Wingard JR, Merz WAG, Ronald MG, Johnson TR, Karp JE, Saral R. Increase in tients with C krusei fungemia. Candida krusei infection among patients with bone marrow transplantation and Several studies have demonstrated that isolates of neutropenia treated prophylactically with fluconazole. N Engl J Med. 1991;325: C krusei may be less susceptible to amphotericin B in 1274-1277.

(REPRINTED) ARCH INTERN MED/ VOL 160, SEP 25, 2000 WWW.ARCHINTERNMED.COM 2663

Downloaded From: https://jamanetwork.com/ on 09/28/2021 6. Goldman M, Pottage JC, Weaver DC. Candida krusei fungemia: report of 4 cases phylaxis for fungal infections after marrow transplantation: a prospective, ran- and review of the literature. Medicine. 1993;72:143-150. domized, double-blind study. J Infect Dis. 1995;171:1545-1552. 7. Anaissie EA, Rex JH, Uzun O, Vartivarian S. Predictors of adverse outcome in 16. Bodey GP. Hematogenous and major organ candidiasis. In: Bodey GP, ed. Can- cancer patients with candidemia. Am J Med. 1998;104:238-245. didiasis: Pathogenesis, Diagnosis and Treatment. 2nd ed. New York, NY: Raven 8. Norusis MJ. Statistical Package for the Social Sciences. Chicago, Ill: SPSS Inc; Press; 1993:279-329. 1994. 17. Goodrich JM, Reed E, Mori M, et al. Clinical features and analysis of risk factors 9. Horn R, Wong B, Kiehn TE, Armstrong D. Fungemia in a cancer hospital: chang- for invasive candidal infection after marrow transplantation. J Infect Dis. 1991; ing frequency, earlier onset, and results of therapy. Rev Infect Dis. 1985;7:646- 164:731-740. 655. 18. Pfaller MA. In vitro susceptibilities of clinical yeast isolates to three antifungal 10. Nguyen MH, Peacock JE Jr, Morris AJ, et al. The changing face of candidemia: agents determined by the microdilution method. Mycopathologia. 1995;130: emergence of non–Candida albicans species and antifungal resistance. Am J Med. 3-9. 1996;100:617-623. 19. Pfaller MA, Rex JH. Antifungal susceptibility testing of isolates from a random- 11. Merz WG, Karp JE, Schron D, Saral R. Increased incidence of fungemia caused ized, multicenter trial of fluconazole versus amphotericin B as treatment of non- by Candida krusei. J Clin Microbiol. 1986;24:581-584. neutropenic patients with candidemia. Antimicrob Agents Chemother. 1995;39: 12. Iwen PC, Kelly DM, Reed EC, Hinrichs SH. Invasive infection due to Candida kru- 40-44. sei in immunocompromised patients not treated with fluconazole. Clin Infect Dis. 20. Berrouane YF, Hollis RJ, Pfaller MA. Strain variation among and antifungal sus- 1995;20:342-347. ceptibilities of isolates of Candida krusei. J Clin Microbiol. 1996;34:1856-1858. 13. Winston DJ, Chandrasekar PH, Lazarus HM, et al. Fluconazole prophylaxis of fun- 21. Blumberg EA, Reboli AC. Failure of systemic empirical treatment with ampho- gal infection in patients with acute leukemia: results of randomized placebo- tericin B to prevent candidemia in neutropenic patients with cancer. Clin Infect controlled, double-blind, multicenter trial. Ann Intern Med. 1993;118:495-503. Dis. 1996;22:462-466. 14. Goodman JL, Winston DJ, Greenfield RA, et al. A controlled trial of fluconazole 22. Cartledge J, Midgley J, Gazzard B. Itraconazole cyclodextrin solution: the role of to prevent fungal infections in patients undergoing bone marrow transplanta- in vitro susceptibility testing in predicting successful treatment of HIV-related tion. N Engl J Med. 1992;326:845-851. fluconazole-resistant and fluconazole-susceptible oral candidiasis. AIDS. 1997; 15. Slavin MA, Osborne B, Adams R, et al. Efficacy and safety of fluconazole pro- 11:163-168.

(REPRINTED) ARCH INTERN MED/ VOL 160, SEP 25, 2000 WWW.ARCHINTERNMED.COM 2664

Downloaded From: https://jamanetwork.com/ on 09/28/2021