Risk Factors for Candida tropicalis Fungemia in Patients with Cancer Dimitrios P. Kontoyiannis, Irfan Vaziri, Hend A. Hanna, Maha Boktour, Jack Thornby, Ray Hachem, Gerald P. Bodey, and Issam I. Raad Department of Infectious Diseases, Infection Control and Employee Health, The University of Texas M.D. Anderson Cancer Center, Houston
The risk factors for and presentation of Candida tropicalis fungemia, in comparison with those of Candida albicans, have been incompletely characterized. We compared 43 cases of C. tropicalis fungemia with 148 cases of C. albicans fungemia. In univariate analysis, patients with C. tropicalis fungemia were more likely to have leukemia (P p .0006), prolonged neutropenia (P p .03 ), and a positive blood culture for more days (P p .02). The 2 groups did not differ with regard to baseline Acute Physiology and Chronic Health Evaluation (APACHE) II score, frequency of catheter-associated fungemia, or response to antifungals. In multivariate analysis, patients with C. tropicalis fungemia were more likely to have leukemia (P p .02 ), previous neutropenia (P p .002), and a longer stay in the intensive care unit during the infectious episode (P p .01 ). Also, the response of the breakthrough C. tropicalis fungemia was lower (P p .05 ). In conclusion, the host determinants associated with susceptibility to C. tropicalis are leukemia and prolonged neutropenia. Candida tropicalis has been reported to be one of the leading Candida species other than Candida albicans to cause fungemia in patients who have malignancy [1, 2]. During the 1970s and 1980s, several studies showed that C. tropicalis fungemia was common in patients with leukemia and in those who received a bone marrow transplant [1–3]. The introduction in the early 1990s of fluconazole, an effective triazole against most Candida species along with a selection of other non-albicans Candida species, altered the frequency and distribution of Candida species in patients with cancer . The recent reports of C. tropicalis in patients with cancer have been limited by the relatively small number of subjects studied . Because both C. tropicalis and C. albicans are intrinsically susceptible to fluconazole in vitro , it is important to reexamine the current contri-
Received 15 March 2001; revised 4 June 2001; electronically published 24 September 2001. Reprints or correspondence: Dr. Dimitrios P. Kontoyiannis, Dept. of Infectious Diseases, Infection Control, and Employee Health Box 402, The University of Texas M.D. Anderson Cancer Center, 1515 Holcombe Blvd., Houston, TX 77030 ([email protected]
). Clinical Infectious Diseases 2001; 33:1676–81 2001 by the Infectious Diseases Society of America. All rights reserved. 1058-4838/2001/3309-0009$03.00
1676 • CID 2001:33 (15 November) • Kontoyiannis et al.
bution of C. tropicalis as a pathogen in patients with cancer at high risk for fungemia and contrast it with that of C. albicans. To that end, we conducted this retrospective study, which covered a 6-year period, to identify the risk factors and other relevant prognostic data associated with C. tropicalis fungemia in patients with cancer, and we compared the findings with those associated with C. albicans fungemia in the same population.
PATIENTS AND METHODS We identified 43 consecutive patients with cancer who experienced C. tropicalis fungemia from 1 January 1993 through 31 December 1998 in the microbiology records of The University of Texas M.D. Anderson Cancer Center (Houston). The 43 patients were compared with 148 patients with cancer in whom C. albicans fungemia developed during the study period. Patients with mixed fungemia were excluded from both cohorts. A case of C. tropicalis fungemia was defined as one that had ⭓1 blood culture positive for C. tropicalis in the presence of signs and symptoms of systemic infection. In addition, the patients’ medical records were reviewed for the following demographic and potential risk factors:
age; sex; underlying disease; Acute Physiology and Chronic Health Evaluation (APACHE) II score on the day of the first positive blood culture; duration of hospitalization; duration of intensive care unit (ICU) stay; and receipt of steroids, total body irradiation, chemotherapy, cyclosporin A, tacrolimus, iv hyperalimentation, or prophylactic antifungals; and presence of neutropenia. These clinical characteristics were noted within 30 days of occurrence of the first positive blood culture and during the course of 1 infection. Neutropenia was defined as a neutrophil count of !500 cells/mm3 at the time of onset of infection. Finally, “response to antifungal treatment” was defined as the resolution of all clinical manifestations of fungemia with blood cultures negative for C. tropicalis, and “failure” was defined as persistence of the clinical signs and symptoms of the fungal infection in the absence of an intercurrent infection, blood cultures positive for C. tropicalis, or both. For categorical data, we used Fisher’s exact text when we compared C. tropicalis fungemia cohorts with C. albicans cohorts and evaluated the response of C. tropicalis fungemia to antifungals. For numerical data, we used Student’s t test and the Wilcoxon rank sum test. Also, we used logistic regression with a stepwise backward elimination to determine multiple discriminators between C. tropicalis and C. albicans and predictors of response by C. tropicalis. Significance was assigned for P values of ⭐.05, and no adjustments were made for multiple comparisons. All analyses were performed using SAS software (STAT, version 8; SAS Institute).
RESULTS Many (44%) of the 43 patients with C. tropicalis fungemia had hematologic malignancy as an underlying disease (table 1). In contrast, the majority of patients who had C. albicans fungemia (76%) had an underlying solid tumor (P p .02). Neutropenia before or at the onset of fungemia was much more common in the cohort of patients who had C. tropicalis fungemia (60% vs. 26%; P p .0004). Also, the mean duration of neutropenia was longer in this group (6 days vs. 2 days; P p .03). Similarly, more patients with C. tropicalis fungemia remained neutropenic during their infection (P p .00009) and were given growth factors (P p .007) or WBC transfusions during their infection (P p .01). Finally, more patients with C. tropicalis fungemia had received systemic antifungals either empirically or prophylactically within 30 days before entry into the study (26% vs. 16%; P p .18). The 2 patient groups did not differ in other characteristics, such as baseline APACHE II score, response to antifungals, or frequency of central venous catheters, including catheter-related fungemia (data not shown). The presence of leukemia and neutropenia before the first positive blood culture was an independent risk factor for the occurrence of C. tropicalis fungemia in multivariate analysis
(table 2). Most (53%) of the 43 patients with C. tropicalis fungemia had a positive blood culture on more than 1 day. Ten (44%) of those 23 patients had fungemia for 15 days. All but 2 of those 10 patients with persistent C. tropicalis fungemia were treated with fluconazole (400 mg/d). In addition, the duration of ICU stay tended to be longer during the infectious episode in patients with C. tropicalis fungemia (OR, 3.1; 95% CI, 1.2–7.7; P p .017; table 2). The treatment failure rate was higher in patients with persistent neutropenia (55%) than it was in those without neutropenia or those who had recovered from neutropenia (P p .047; table 3). Also, the treatment failure rate was higher in patients who had leukemia and myelodysplastic syndrome (53%) than it was in those who did not (P p 0.012; table 3). Persistent neutropenia was the only independent factor for poor outcome of C. tropicalis fungemia in multivariate analysis (OR, 6.9; 95% CI, 1.6–29.7; P p .01). On the other hand, the majority (88%) of patients with C. tropicalis fungemia who had a solid tumor as underlying disease had a response to antifungal treatment (table 3). Most of the patients with C. tropicalis fungemia (39 [91%] of 43) received therapeutic antifungals (amphotericin B deoxycholate, lipid formulations of amphotericin B, or fluconazole) at some time during the treatment of their infection (table 1). Clinical and microbiologic response of the infection to the antifungals was observed in 27 (69%) of the 39 treated patients. Eleven (73%) of the 15 patients with breakthrough C. tropicalis fungemia had underlying leukemia. However, only 6 (14%) of 43 patients developed breakthrough C. tropicalis fungemia if they had previously received fluconazole prophylaxis. The response rate to fluconazole in patients with C. tropicalis fungemia who had not previously received prophylaxis with fluconazole was comparable with that to treatment with amphotericin B or with lipid formulations of amphotericin B. However, when C. tropicalis fungemia represented a breakthrough infection despite previous prophylactic or empiric antifungal treatment, the response rate was lower compared with C. tropicalis fungemia representing a de novo infection (table 4). Eleven percent of all the fungemia cases that occurred during the study period were caused by C. tropicalis, making C. tropicalis the fifth most common bloodstream Candida isolate after C. albicans (34%), Candida glabrata (19%), Candida parapsilosis (18%), and Candida krusei (12%). The incidence of C. tropicalis fungemia has remained stable, going from 0.6 cases per 1000 hospital admissions in 1993 to 0.8 cases per 1000 hospital admissions in 1998 (P p NS).
DISCUSSION This study reports what is to our knowledge the largest recent series of patients with cancer with C. tropicalis fungemia at a Candida tropicalis Fungemia • CID 2001:33 (15 November) • 1677
Patient demographics and clinical characteristics.
Age, median years (range)
Sex, no. male/no. female (% male/% female)
Acute leukemia and MDS
Lymphoma and myeloma Solid tumor
52 (2–80) 68/80 (46/54) 16/148 (11)
NS NS .0006 NS .02
BMT within past year
ICU before first positive blood culture
Previous total body irradiationb
Tacrolimus Cyclosporin Ab Total parenteral nutritionb
Duration of previous neutropenia, mean days (range) Duration of prior hospitalization, mean days (range) Central venous catheterc
Duration of ICU stay before study entry, mean days (range)
.035 NS NS
Neutropenia during study period
Mean APACHE II (range)c
Growth factor during infection
ICU stay for 110 days during the infection
WBC transfusion during infection Duration of positive blood cultures for 110 days Previous use of antifungals
Response to initial antifungal therapy
NOTE. Data are n/N (%), unless otherwise indicated. APACHE, Acute Physiology and Chronic Health Evaluation; BMT, bone marrow transplant; ICU, intensive care unit; MDS, myelodysplastic syndrome; NS, not statistically significant. a b c d
Determined by means of univariate analysis. Within 30 days before the first positive blood culture or at study entry. At study entry (day of the first positive blood culture). Assessable patients.
single institution. We found that C. tropicalis fungemia remains a significant cause of fungemia in patients with hematologic malignancies because almost half of the patients had a hematologic malignancy (mainly leukemia) as an underlying disease. Similarly, neutropenia was found in the majority (60%) of cases of C. tropicalis fungemia. In fact, as shown in table 2, leukemia and neutropenia, were, according to multivariate analysis, independent factors favoring C. tropicalis fungemia, as has been reported elsewhere [1–4, 7–9]. In contrast to other studies that have reported the frequent occurrence of C. tropicalis fungemia in critically ill patients with high APACHE II scores , in our study, patients with both C. tropicalis and C. albicans fungemia had similarly low severity scores of illness at the onset of their infections (table 1). However, in comparison with patients who had C. albicans funge-
1678 • CID 2001:33 (15 November) • Kontoyiannis et al.
mia, those who had C. tropicalis fungemia were more likely to have had persistent fungemia and a longer ICU stay during the course of their infection. This may signify a higher virulence of C. tropicalis when compared with C. albicans in the setting Table 2. Multivariate logistic regression contrasts between Candida tropicalis and Candida albicans fungemia. Variable Intensive care unit stay during infection 110 days
OR (95% CI)
Neutropenia before first posia tive blood culture
Within 30 days before fungemia.
Table 3. Univariate analysis of predictors for poor outcome in Candida tropicalis fungemia. Response, n/N (%)
OR (95% CI)
Leukemia and MDS
Other hematogenous malignancy Yes
Solid tumor Yes
ICU stay before or after first PBC Yes
No neutropenia or recovery from Yes Persistent neutropenia
26/32 (81) 5/11 (45)
Growth factors at study entry Yes
Growth factors during infection Yes
NOTE. ICU, intensive care unit; MDS, myeloplastic syndrome; NS, not statistically significant; PBC, positive blood culture. a b c d
OR for poor response if the variable is “yes.” Compared with other hematogenous malignancies and solid tumors. Compared with leukemia and solid tumor. Compared with hematoxic malignancies.
of neutropenia, as has been suggested in other studies [1, 3, 4, 7]. For example, the high pathogenicity of C. tropicalis was illustrated by the observation that as many as 60%–80% of colonized patients with neutropenia eventually develop invasive infection [1, 10]. However, whether C. tropicalis does indeed have higher virulence is debatable. C. tropicalis is not more virulent than is C. albicans when introduced into the bloodstream in animal models [1, 11, 12]. However, C. tropicalis was more virulent than was C. albicans when introduced in the gastrointestinal tract in the setting of neutropenia and mucositis in animals [13–15]. Those results suggest that C. tropicalis does not have higher virulence but, rather, that it takes advantage of the host. In addition, our study showed that persistent neutropenia is a strong prognostic factor for poor response. Previous riskstratification models of candidemia in patients with cancer have described the poor prognostic significance of the presence of leukemia, neutropenia, or a high baseline APACHE II score [2,
4, 7, 9, 16]. We failed to identify the severity of illness score as a prognostic factor, perhaps because of the relatively small number of C. tropicalis fungemia episodes and because the affected patients had relatively low baseline APACHE II scores. Our antifungal therapy response data should be viewed with caution because decisions about management were not controlled. The limited data comparing fluconazole with amphotericin B in patients with C. tropicalis fungemia suggest that both agents are comparable, at least in stable patients or those lacking neutropenia . Nevertheless, our data support the notion that breakthrough C. tropicalis fungemia in this patient population is associated with suboptimal efficacy of subsequent antifungal therapy. In contrast, others have suggested that this may not be true if one controls for other severity measures of infection, such as the APACHE II score . However, such an analysis was not feasible because of the small number of breakthrough fungemia cases. Only a minority of the cases of C. tropicalis fungemia in our
Candida tropicalis Fungemia • CID 2001:33 (15 November) • 1679
Table 4. Response to initial antifungal therapy in patients with Candida tropicalis fungemia.
Characteristic Prophylaxis or empiric therapy
No. of patients
No. (%) of responses
AmB or lipid AmB empiric therapy No prophylaxis Fluconazole therapy AmB or lipid AmB therapy AmB and fluconazole therapy Total response rate NOTE.
8 4 39
5 4 27 (69)
AmB, amphotericin B.
One of the 7 patients received no subsequent therapy. b Lipid AmB indicates AmB lipid complex or lipid-associated AmB. One patient received AmB and fluconazole as empiric therapy. All patients received subsequent AmB or lipid AmB as specific therapy. c We were unable to assess the response to antifungal therapy in 4 of 43 patients. d For 22/28 vs. 5/11, P p .05.
study developed in the setting of fluconazole prophylaxis. There have been conflicting reports regarding whether there has been an emergence of resistance of this Candida species to fluconazole, because cases of breakthrough fluconazole-resistant C. tropicalis fungemia in patients receiving fluconazole have been reported [8, 19, 20]. C. tropicalis has been shown to rapidly acquire a stable, broad spectrum of resistance to azoles after exposure to increasing concentrations of fluconazole in vitro through upregulation of efflux transporters . However, recent population-based surveillance studies of candidemia have suggested that C. tropicalis has negligible level of fluconazole resistance [22–24]. Finally, our data demonstrated that although the rate of C. tropicalis fungemia remained constant, its relative frequency has been continuously decreasing in our institution since 1990 . More specifically, when compared with the period of 1988–1992, in which fluconazole use at the M.D. Anderson Cancer Center was not widespread, C. tropicalis shifted from the second to the fifth leading cause of fungemia. This trend also coincided with the emergence of other non-albicans Candida species resistant to fluconazole, both in our institution and elsewhere [4, 8, 22–25]. In fact, absence of fluconazole prophylaxis was shown to be an independent predictor of C. tropicalis fungemia in this patient population in an earlier study at our institution . We did not find such an association in our study, most likely because we used neutropenia as the first independent variable in regression analysis. Therefore, because the majority of patients with prolonged neutropenia were receiving prophylactic or empiric antifungals at the onset of their
1680 • CID 2001:33 (15 November) • Kontoyiannis et al.
C. tropicalis fungemia, the independent contribution of the lack of use of antifungals, especially that of fluconazole, could not be demonstrated. Nevertheless, the contribution of fluconazole use to the changing epidemiology of fungemia has been questioned by some researchers . Our study has several limitations as a result of its retrospective nature that did not allow an assessment of other, potentially important differences between the 2 cohorts, such as the presence of mucositis, myositis, type and duration of previous or concomitant antibacterials, dissemination, and underlying disease activity. In addition, our study did not look at types of invasive C. tropicalis infection other than fungemia, such as meningitis; C. tropicalis is the leading cause of Candida meningitis in children with cancer . Finally, we did not provide data concerning C. tropicalis–attributable mortality. Some studies [5, 28]—but not all [7, 9, 29]—have suggested that C. tropicalis sepsis results in a higher mortality rate in this patient population. In summary, C. tropicalis, even though it is less common than are other Candida species, remains an important pathogen, especially in patients with neutropenia who have hematologic malignancies, including those patients who are receiving fluconazole. There are distinct differences in presentation and risk factors between C. tropicalis and C. albicans fungemia. Poor prognostic factors for the outcome of C. tropicalis fungemia are having leukemia, persistent neutropenia, and, possibly, breakthrough fungemia. References 1. Wingard JR. Importance of Candida species other than C. albicans as pathogens in oncology patients. Clin Infect Dis 1995; 20:115–75. 2. BodeyGP. Hematogenous and major organ candidiasis. In: Bodey GP, ed. Candidiasis: pathogenesis, diagnosis, and treatment. 2d ed. New York: Raven Press, 1993:279–329. 3. Wingard JR, Merz WG, Saral R. Candida tropicalis: a major pathogen in immunocompromised patients. Ann Intern Med 1979; 91:539–43. 4. Abi-Said D, Anaissie E, Uzun O, Raad I, Pinzcowski H, Vartivarian S. The epidemiology of hematogenous candidiasis caused by different Candida species. Clin Infect Dis 1997; 24:1122–8. 5. Viscoli C, Girmenia C, Marinus A, et al. Candidemia in cancer patients: a prospective, multi-center surveillance study by the Invasive Fungal Infection Group (IFIG) of the European Organization for Research and Treatment of Cancer (EORTC). Clin Infect Dis 1999; 28:1071–9. 6. Hoban DJ, Zhanel GG, Karlowsky JA. In vitro susceptibilities of Candida and Cryptococcus neoformans isolates from blood cultures of neutropenic patients. Antimicrob Agents Chemother 1999; 43:1463–4. 7. Komshian SV, Uwaydah AK, Sobel JD, Crane LR. Fungemia caused by Candida species and Torulopsis glabrata in the hospitalized patient: frequency, characteristics, and evaluation of factors influencing outcome. Rev Infect Dis 1989; 11:379–90. 8. Nguyen MH, Peacock JE, Morris AJ, et al. The changing face of candidemia: emergence of non–Candida albicans species and antifungal resistance. Am J Med 1996; 100:617–23. 9. Fraser VJ, Jones M, Dunkel J. et al. Candidemia in a tertiary care hospital: epidemiology, risk factors, and predictors of mortality. Clin Infect Dis 1992; 5:414–21.
10. Sanford GR, Metz WG, Wingard JR, Charache P, Saral R. The value of fungal surveillance cultures as predictors of systemic fungal infections. J Infect Dis 1980; 142:503–9. 11. Louria DB, Buse M, Brayton RG, Finkel G. The pathogenesis of Candida tropicalis infections in mice. Sabouraudia 1966; 5:14–25. 12. Wingard JR, Dick JD, Merz WG, Sanford GR, Saral R, Burns WH. Differences in virulence of clinical isolates of Candida tropicalis and Candida albicans in mice. Infect Immun 1982; 37:833–6. 13. De Repentigny L, Phaneuf M, Mathieu LG. Gastrointestinal colonization and systemic dissemination by C. albicans and C. tropicalis in intact and immunocompromised mice. Infect Immun 1992; 60: 4907–14. 14. Wingard JR, Dick JD, Merz WG, Sanford GR, Saral R, Burns WH. Pathogenicity of Candida tropicalis and Candida albicans after gastrointestinal inoculation in mice. Infect Immun 1980; 29:808–13. 15. Walsh TJ, Merz WG. Pathologic features in the human alimentary tract associated with invasiveness of Candida tropicalis. Am J Clin Pathol 1986; 85:498–502. 16. Anaissie EJ, Rex JH, Uzun O, Vartivarian S. Predictors of adverse outcome in cancer patients with candidemia. Am J Med 1998; 104: 238–45. 17. Kontoyiannis DP, Bodey GP, Mantzoros C. Fluconazole versus amphotericin B for the management of candidemia in adults: a metaanalysis. Mycoses 2001; 44:125–35. 18. Uzun O, Ascioglu S, Anaissie EJ, et al. Risk factors and predictors of outcome in cancer patients with breakthrough candidemia [abstract 705]. In: Program and abstracts of the 40th Interscience Conference on Antimicrobial Agents and Chemotherapy (Toronto). Washington, DC: American Society for Microbiology, 2001:364. 19. Barnes AJ, Wardley AM, Oppenheim BA, et al. Fatal Candida tropicalis
27. 28. 29.
fungemia in a leukemic patient receiving fluconazole prophylaxis. J Infect 1996; 33:43–5. Jandourek A, Brown P, Vazquez JA. Community-acquired fungemia due to a multiple-azole–resistant strain of Candida tropicalis. Clin Infect Dis 1999; 29:1583–4. Barchiesi F, Calabrese D, Sanglard D, et al. Experimental induction of fluconazole resistance in Candida tropicalis ATCC 750. Antimicrob Agents Chemother 2000; 44:1578–84. PfallerMA, Jones RN, Doern GV, et al., and The Sentry Participant Group. Bloodstream infections due to Candida species: SENTRY antimicrobial surveillance program in North America and Latin America, 1997–1998. Antimicrob Agents Chemother 2000; 44:747–51. St Germain G, Laverdiere M, Pelletier R, et al. Prevalence and antifungal susceptibility of 442 Candida isolates from blood and other normally sterile sites: results of a 2-year (1996 to 1998) multicenter surveillance study in Quebec, Canada. J Clin Microbiol 2001; 39:949–53. Kao AS, Brandt ME, Pruitt WR, et al. The epidemiology of candidemia in two United States cities: results of a population-based active surveillance. Clin Infect Dis 1999; 29:1164–70. Abbas J, Bodey GP, Hanna HA, et al. Candida krusei fungemia. Arch Intern Med 2000; 160:2659–64. White MH. The contribution of fluconazole to the changing epidemiology of invasive candidal infections. Clin Infect Dis 1997; 24: 1129–30. McCullers JA, Vargas SL, Flynn PM, Razzouk BI, Shenep JL. Candidal meningitis in children with cancer. Clin Infect Dis 2000; 31:451–7. Martino P, Girmenia C, Micozzi A, et al. Fungemia in patients with leukemia. Am J Med Sci 1993; 306:225–32. Nguyen MH, Peacock JE, Tanner DC, et al. Therapeutic approaches in patients with candidemia: evaluation of multicenter, prospective, observational study. Arch Intern Med 1995; 155:2429–35.
Candida tropicalis Fungemia • CID 2001:33 (15 November) • 1681