**7. Discussion and remarks**

This prospective candidemia surveillance study represents one of the largest multicenter studies conducted in Spain and provide one of the most representative data on the epidemiology of candidemia to date. The first remarkable finding of our study was the higher incidence of candidemia than those reported from centers located in the Northern Hemisphere which ranged between 0.28 to 0.96 per 1,000 admissions (Banerjee et al., 1991, Doczi et al., 2002, Marchetti et al., 2002, Pfaller et al., 1998, Pfaller et al., 2004, Richet et al., 2002, Sandven et al., 1998, Tortorano et al., 2002, Tortorano et al., 2004) and including those published before in Spain (0.76 to 0.81 per 1,000 admissions) (Almirante et al., 2005, Alonso-Valle et al., 2003, Pemán et al., 2002, Pemán et al., 2011, San Miguel et al., 2005). Although the reasons for this high rate are not entirely clear, it is possible that this may be related to a

Neoplasia 2.9 1.4 – 5.9 ≤ 0.05 Prior fluconazole use 3.8 1.7 – 8.2 ≤ 0.01

Table 6. Summary of multivariate statistical analysis of risk factors for candidemia caused

Caspofungin and anidulafungin resistance was low (16 cases for caspofungin and 8 cases for anidulafungin) (Table 4). Despite this low rate of in vitro resistance to echinocandins of the isolates studied MICs from *C. parapsilosis* and *C. guilliermondi* were higher compared to the MIC obtained from other *Candida spp*. as it has been described in others

At the time of diagnosis and inclusion in this study, 122 case patients (12.3%) were receiving a systemic antifungal agent and were considered breakthrough infections (fluconazole, 78 patients, amphotericin B, 31 patients, itraconazole and voriconazole, 3 patient each; and echinocandins, 5 patients). Although the reason for this high rate of breakthrough infections is not clear, it is possible that other factors besides the antifungal resistance have got a role in the explanation of this phenomenon. A deep analysis of these 122 cases showed that either the antifungal therapy duration or the election of the antifungal drug was inadequate. A total of 536 case patients (54.5%) received antifungal therapy, started at a median of 3 days

The crude mortality rate was 24.10%, but the mortality rate among children was significantly lower. (see Table 3) As it has been described in different studies published in the medical literature candidemia due to *C. parapsilosis* had a lower mortality rate than the rate due other Candida species (Morgan et al., 2005, Pappas et al., 2004, Pemán et al., 2002), but no statistically significant result when analyzing the death rate of patients infected by a susceptible isolate (54%) or a less-susceptible isolate (SDD or resistant) (64%) among

This prospective candidemia surveillance study represents one of the largest multicenter studies conducted in Spain and provide one of the most representative data on the epidemiology of candidemia to date. The first remarkable finding of our study was the higher incidence of candidemia than those reported from centers located in the Northern Hemisphere which ranged between 0.28 to 0.96 per 1,000 admissions (Banerjee et al., 1991, Doczi et al., 2002, Marchetti et al., 2002, Pfaller et al., 1998, Pfaller et al., 2004, Richet et al., 2002, Sandven et al., 1998, Tortorano et al., 2002, Tortorano et al., 2004) and including those published before in Spain (0.76 to 0.81 per 1,000 admissions) (Almirante et al., 2005, Alonso-Valle et al., 2003, Pemán et al., 2002, Pemán et al., 2011, San Miguel et al., 2005). Although the reasons for this high rate are not entirely clear, it is possible that this may be related to a

a Only statistically significant variables are summarized in the table.

by fluconazole susceptible isolates vs. resistant or SDD ones a

from the *Candida* isolation or onset of candidemia.

patients who received fluconazole as treatment. (*P* ≤ 0.44).

**7. Discussion and remarks** 

studies.

**6. Mortality** 

**5. Antifungal treatment** 

Odds ratio 95 percent confidence Limits P - value

combination of multiple factors, including differences in medical care resources, transplantation programs, implementation of infection control measures in hospitals, empirical antifungal therapy and prophylaxis for high-risk patients. Another possibility is that our series may not reflect the current trends, but the data from a prospective study held in Spain and recently published by Peman *et al.* support our data (Pemán et al., 2011, Pfaller & Diekema, 2007) (see Table 7).


Table 7. Summary of geographical differences in species distribution in Candida BSI isolates. (Adapted and modified from Pemán et al., 2001 and Pfaller & Diekema, 2007).

Despite this fact, it seems that probably a combination of factors may have affected the overall rates of fungemia cases in Spain. The differences appeared in the average age of our patients when compared to other surveillance series from the United States (Ostrosky-Zeichner et al., 2003, Pappas et al., 2003), are probably due to the high proportion of children in our study, especially in the Spanish hospitals located in the Southern part of the country (32% of children in the Spanish hospitals from the Southern part of the country compared to 9% described in the study from the United States) (data not shown). Therefore, this condition reflects that there are great differences among patients of different geographical locations as it was mentioned in the introduction and the demographical composition and lastly the risk factors, could be very different from one population to another. (Table 7). In fact, the number of cases of invasive candidemia was not homogeneous across the country., the distribution of the clinical isolates obtained during the study period along four different

Epidemiology of Bloodstream *Candida* spp. Infections

Observed During a Surveillance Study Conducted in Spain 25

immunocompromised patients, where clinical responses are poorer the picture is not good. Summarizing, we are concerned about the use of echinocandins alone based on the identification of non-albicans *Candida* specie. Grouping these agents into one treatment scheme is difficult due to the variability not only in the susceptibility of the isolates, as well

Regarding the susceptibility of the studied isolates, antifungal resistance was an infrequent finding in our study and was restricted to a few isolates, and none of them were resistant to amphotericin B. This condition is similar to the findings published in three recent studies (Almirante et al., 2005, Messer et al., 2009, Pemán et al., 2011). Our proportion of fluconazole-resistant isolates (6.32%) was low, similarly to the rate observed with Spanish (Almirante et al., 2005, Pemán et al., 2011) European (5.2%) and North American isolates (6.6%) (Messer et al., 2009, Richardson & Lass-Flörl et al., 2008) (see Table 7). Mixing the ideas exposed above we can argue that the differences in the activity and susceptibility of the antifungal compounds studied suggest that azole drugs and echinocandins have got a complementary susceptibility profile. While azoles has got excellent in vitro activity to *C. albicans*, *C. parapsilosis* and *C. tropicalis* bloodstream isolates, echinocandins showed excellent activity against *C. glabrata* and *C. krusei* which are are associated with higher azole MICs. On the contrary, species with high MICs to echinocandins such as *C. parapsilosis* and *C.* 

Amphotericin B Fluconazole Voriconazole Caspofungin Anidulafungin

as the microbiological responses seen between different echinocandins.

*guilliermondii* showed excellent activity to azole agents. (Table 8).

dose-dependent; R, resistant.

the one reflected in our work.

infections (Imhof et al., 2004, Pfaller et al., 2004).

Specie of *Candida* Susceptibility to antifungal agent

C. albicans S S S S S C. parapsilosis S S S S - I S – I C. glabrata S – NS S – SDD – R S – NS S S C. tropicalis S S S – NS S S C. krusei S R S S S a S, susceptible; NS, non-susceptible (intermediate for CLSI M27-A2 clinical breakpoints); SDD, sensitive

b The clinical breakpoints adopted in this table are those reflected in the CLSI M27-A2 methodology. No new clinical breakpoints or epidemiological cut-offs were used, in order to make the data comparable to

Table 8. Summary of commonly associated in-vitro susceptibility profiles for *Candida* spp.

These ideas are of great importance because previous exposure to fluconazole was a strong and independent factor associated with candidemia caused by fluconazole non-susceptible isolates as it had been reported previously by Marr *et al.* and Lin and colleagues and higher voriconazole MICs tended to be associated with prior exposure to fluconazole. Although these obtained results are statistically significant, they must be taken with some caution because the low resistant proportion of isolates in our study. However, they depict a situation of concern and illustrate the potential problem of cross-resistance between azoles with a direct impact in treatment failure and the outcome of the patient. Moreover, the potential for voriconazole resistant *C. glabrata* to emerge as a threat in people receiving voriconazole therapy and or prophylaxis has been raised in reports of breakthrough

BSI isolates. (Adapted and modified from Richardson & Lass-Flörl, 2008) a, b

geographical areas in Spain,. (North, Center, East and South). *C. albicans* covers almost half (49.08%) of the global cases, remaining as the most frequently isolated specie, but the rates between the four different areas were not homogenous., for instance, in the Southern part of the country the rate of isolates of C. albicans (39.2%) was similar to the one of *C. parapsilosis* (37.4%).

Some studies have reported a shift in the etiology of candidemia reporting an increase of candidemia cases caused by non-*albicans Candida* species during the last decade (Colombo et al., 2006, Richet et al., 2002, Tortorano et al., 2004) (Table 7). Although C. albicans remain the most frequently isolated specie, reasons for the emergence of non-*albicans* species remain unclear, but some medical conditions may explain increasing incidence of candidemia due to non-*albicans* species. It has been noted in previous reports that infections due to *C. tropicalis* candidemia is associated with neoplasia and neutropenia (Komshian et al., 1989) and those attributable to *C. parapsilosis* are often associated with the presence of intravascular catheters and are not influenced by exposure to fluconazole or other antifungal agents (Clark et al., 2004, Girmenia et al., 1996, Levy et al., 1998, Sandven et al., 1998). The last situation, may us to consider *C. parapsilosis* as an exogenous pathogen and breaches of catheter care and of infection control practice should be investigated and revised within institutions where this species has become a common blood culture isolate.

The increasing incidence some of non-*albicans Candida* species with reduced susceptibility to azoles, such as *C. glabrata*, creates new therapeutic challenges and leads to another important question such as the influence of previous antifungal therapy in the development of non-*albicans* species candidemia. Recent studies, such as those published by Marr *et al.* and Tortorano *et al.* had addressed this question and showed interesting results about the association with previous exposure to azoles and the risk of development of fungemias due to *C. krusei* and / or *C. glabrata* (Marr et al., 2000, Tortorano et al., 2004). In fact, the differences in antifungal susceptibilities among isolates between different regions in Spain was almost entirely attributable to high-level resistance to azoles observed among *C. glabrata* and *C. krusei* isolates (Table 4).

We are not aware of these epidemiological changes mentioned in the paragraph above and our findings from our study are supportive of them. *C. parapsilosis* fungemia account for the large majority of non-*albicans* species (in the same manner that that been described for other European countries) and candidemia due to *C. krusei* is rare in Spain as it has been described in other series (Almirante et al., 2005, Alonso-Valle et al., 2003, Ostrosky-Zeichner et al., 2003, Pemán et al., 2002, Pemán et al., 2011, Tortorano et al., 2004). The explanation for this species distribution in Spain which is more or less similar to other Latin American countries is not clear and perhaps many factors are involved. However, this species distribution has got a great importance in the developing of therapeutic schemes and in the prevention of antifungal resistance.

While most non-albicans *Candida* species are associated with high fluconazole minimum inhibitory concentrations (MIC), *C. parapsilosis* is typically susceptible to most antifungals, although is associated with higher echinocandin MIC that vary between different agents. Moreover, the rate of persistently positive fungemia in patients treated with caspofungin was reported as almost double for *C. parapsilosis* versus other Candida species. Even more, if we consider clinical patients at risk of suffering a candidemia episode, such as

geographical areas in Spain,. (North, Center, East and South). *C. albicans* covers almost half (49.08%) of the global cases, remaining as the most frequently isolated specie, but the rates between the four different areas were not homogenous., for instance, in the Southern part of the country the rate of isolates of C. albicans (39.2%) was similar to the one of *C. parapsilosis*

Some studies have reported a shift in the etiology of candidemia reporting an increase of candidemia cases caused by non-*albicans Candida* species during the last decade (Colombo et al., 2006, Richet et al., 2002, Tortorano et al., 2004) (Table 7). Although C. albicans remain the most frequently isolated specie, reasons for the emergence of non-*albicans* species remain unclear, but some medical conditions may explain increasing incidence of candidemia due to non-*albicans* species. It has been noted in previous reports that infections due to *C. tropicalis* candidemia is associated with neoplasia and neutropenia (Komshian et al., 1989) and those attributable to *C. parapsilosis* are often associated with the presence of intravascular catheters and are not influenced by exposure to fluconazole or other antifungal agents (Clark et al., 2004, Girmenia et al., 1996, Levy et al., 1998, Sandven et al., 1998). The last situation, may us to consider *C. parapsilosis* as an exogenous pathogen and breaches of catheter care and of infection control practice should be investigated and revised

within institutions where this species has become a common blood culture isolate.

The increasing incidence some of non-*albicans Candida* species with reduced susceptibility to azoles, such as *C. glabrata*, creates new therapeutic challenges and leads to another important question such as the influence of previous antifungal therapy in the development of non-*albicans* species candidemia. Recent studies, such as those published by Marr *et al.* and Tortorano *et al.* had addressed this question and showed interesting results about the association with previous exposure to azoles and the risk of development of fungemias due to *C. krusei* and / or *C. glabrata* (Marr et al., 2000, Tortorano et al., 2004). In fact, the differences in antifungal susceptibilities among isolates between different regions in Spain was almost entirely attributable to high-level resistance to azoles observed among *C. glabrata*

We are not aware of these epidemiological changes mentioned in the paragraph above and our findings from our study are supportive of them. *C. parapsilosis* fungemia account for the large majority of non-*albicans* species (in the same manner that that been described for other European countries) and candidemia due to *C. krusei* is rare in Spain as it has been described in other series (Almirante et al., 2005, Alonso-Valle et al., 2003, Ostrosky-Zeichner et al., 2003, Pemán et al., 2002, Pemán et al., 2011, Tortorano et al., 2004). The explanation for this species distribution in Spain which is more or less similar to other Latin American countries is not clear and perhaps many factors are involved. However, this species distribution has got a great importance in the developing of therapeutic schemes and in the prevention of

While most non-albicans *Candida* species are associated with high fluconazole minimum inhibitory concentrations (MIC), *C. parapsilosis* is typically susceptible to most antifungals, although is associated with higher echinocandin MIC that vary between different agents. Moreover, the rate of persistently positive fungemia in patients treated with caspofungin was reported as almost double for *C. parapsilosis* versus other Candida species. Even more, if we consider clinical patients at risk of suffering a candidemia episode, such as

(37.4%).

and *C. krusei* isolates (Table 4).

antifungal resistance.

immunocompromised patients, where clinical responses are poorer the picture is not good. Summarizing, we are concerned about the use of echinocandins alone based on the identification of non-albicans *Candida* specie. Grouping these agents into one treatment scheme is difficult due to the variability not only in the susceptibility of the isolates, as well as the microbiological responses seen between different echinocandins.

Regarding the susceptibility of the studied isolates, antifungal resistance was an infrequent finding in our study and was restricted to a few isolates, and none of them were resistant to amphotericin B. This condition is similar to the findings published in three recent studies (Almirante et al., 2005, Messer et al., 2009, Pemán et al., 2011). Our proportion of fluconazole-resistant isolates (6.32%) was low, similarly to the rate observed with Spanish (Almirante et al., 2005, Pemán et al., 2011) European (5.2%) and North American isolates (6.6%) (Messer et al., 2009, Richardson & Lass-Flörl et al., 2008) (see Table 7). Mixing the ideas exposed above we can argue that the differences in the activity and susceptibility of the antifungal compounds studied suggest that azole drugs and echinocandins have got a complementary susceptibility profile. While azoles has got excellent in vitro activity to *C. albicans*, *C. parapsilosis* and *C. tropicalis* bloodstream isolates, echinocandins showed excellent activity against *C. glabrata* and *C. krusei* which are are associated with higher azole MICs. On the contrary, species with high MICs to echinocandins such as *C. parapsilosis* and *C. guilliermondii* showed excellent activity to azole agents. (Table 8).


a S, susceptible; NS, non-susceptible (intermediate for CLSI M27-A2 clinical breakpoints); SDD, sensitive dose-dependent; R, resistant.

b The clinical breakpoints adopted in this table are those reflected in the CLSI M27-A2 methodology. No new clinical breakpoints or epidemiological cut-offs were used, in order to make the data comparable to the one reflected in our work.

Table 8. Summary of commonly associated in-vitro susceptibility profiles for *Candida* spp. BSI isolates. (Adapted and modified from Richardson & Lass-Flörl, 2008) a, b

These ideas are of great importance because previous exposure to fluconazole was a strong and independent factor associated with candidemia caused by fluconazole non-susceptible isolates as it had been reported previously by Marr *et al.* and Lin and colleagues and higher voriconazole MICs tended to be associated with prior exposure to fluconazole. Although these obtained results are statistically significant, they must be taken with some caution because the low resistant proportion of isolates in our study. However, they depict a situation of concern and illustrate the potential problem of cross-resistance between azoles with a direct impact in treatment failure and the outcome of the patient. Moreover, the potential for voriconazole resistant *C. glabrata* to emerge as a threat in people receiving voriconazole therapy and or prophylaxis has been raised in reports of breakthrough infections (Imhof et al., 2004, Pfaller et al., 2004).

Epidemiology of Bloodstream *Candida* spp. Infections

*Clin. Microbiol*. 40:1298–1302.

and 2000*. J. Med. Microbiol*. 51:677–681.

mortality. *Clin. Infect. Dis*. 15:414–421.

fluconazole. *Microb. Drug Resist*. 8:311–319.

*Medicine* (Baltimore) 81:425–433.

infection. *Diagn Microbiol Infect Dis*. 7:103-109.)

*Med.* 91:86S–89S.

Observed During a Surveillance Study Conducted in Spain 27

[6] Banerjee, S. N., T. G. Emori, D. H. Culver, R. P. Gaynes, W. R. Jarvis, T. Horan, J. R.

[7] Berenguer J, M Buck, F Witebsky, *et al.* 1993. Lysis-centrifugation blood cultures in the

[8] Blumberg, H. M., W. R. Jarvis, J. M. Soucie, J. E. Edwards, J. E. Patterson, M. A. Pfaller,

Edwards, J. Tolson, T. Henderson, W. J. Martone, et al. 1991. Secular trends in nosocomial primary bloodstream infections in the United States, 1980–1989*. Am. J.* 

detection of tissue-proven invasive candidiasis. disseminated versus single-organ

M. S. Rangel-Frausto, M. G. Rinaldi, L. Saiman, R. T. Wiblin, R. P. Wenzel, et al. 2001. Risk factors for candidal bloodstream infections in surgical intensive care unit patients: the NEMIS prospective multicenter study. *Clin. Infect. Dis*. 33:177–186. [9] Clark, T. A., S. A. Slavinski, J. Morgan, T. Lott, B. A. Arthington-Skaggs, M. E. Brandt, R.

M. Webb, M. Currier, R. H. Flowers, S. K. Fridkin, and R. A. Hajjeh. 2004. Epidemiologic and molecular characterization of an outbreak of *Candida parapsilosis*  bloodstream infections in a community hospital*. J. Clin. Microbiol*. 42:4468–4472. [10] L. Colombo, M. Nucci, B. J. Park, S. A. Nouér, B. Arthington-Skaggs, D. A. da Matta, D.

Warnock, and J. Morgan for the Brazilian Network Candidemia Study. 2006. Epidemiology of candidemia in Brazil: a nationwide sentinel surveillance of

Herwaldt, and M. A. Pfaller. 2002. Epidemiology of candidemia: 3-year results from the emerging infections and the epidemiology of Iowa organisms study. *J.* 

of yeast bloodstream infections in a Hungarian university hospital between 1996

Candidemia in a tertiary care hospital: epidemiology, risk factors, and predictors of

trends of candidemia over 12 years in adult patients at a tertiary care hospital.

and A. Cassone. 1996. Rising incidence of *Candida parapsilosis* fungemia in patients with hematologic malignancies: clinical aspects, predisposing factors, and differential pathogenicity of the causative strains*. Clin. Infect. Dis*. 23:506–514. [17] Gudlaugsson, O., S. Gillespie, K. Lee, B. J. Vande, J. Hu, S. Messer, L. Herwaldt, M.

Pfaller, and D. Diekema. 2003. Attributable mortality of nosocomial candidemia, revisited. Clin. Infect. Dis. 37:1172–1177. 19. Hsueh, P. R., L. J. Teng, P. C. Yang, S. W. Ho, and K. T. Luh. 2002. Emergence of nosocomial candidemia at a teaching hospital in Taiwan from 1981 to 2000: increased susceptibility of *Candida* species to

[13] Doczi, I., E. Dosa, E. Hajdu, and E. Nagy. 2002. Aetiology and antifungal susceptibility

[14] Fraser, V. J., M. Jones, J. Dunkel, S. Storfer, G. Medoff, and W. C. Dunagan. 1992.

[15] Garbino, J., L. Kolarova, P. Rohner, D. Lew, P. Pichna, and D. Pittet. 2002. Secular

[16] Girmenia, C., P. Martino, B. F. De, G. Gentile, M. Boccanera, M. Monaco, G. Antonucci,

candidemia in eleven medical centers *J. Clin. Microbiol*. 44:2816–2823. [11] Costa, S. F., I. Marinho, E. A. Araujo, A. E. Manrique, E. A. Medeiros, and A. S. Levin. 2000. Nosocomial fungaemia: a 2-year prospective study. *J. Hosp. Infect*. 45:69–72. [12] Diekema, D. J., S. A. Messer, A. B. Brueggemann, S. L. Coffman, G. V. Doern, L. A.

Despite these concerning matters exposed above, voriconazole was the azole which exhibited the best in vitro antifungal activity in our study, and only one of six fluconazoleresistant isolates was cross-resistant to voriconazole. The combination of a third generation azole such as voriconazole or posaconazole with an echinocandin could be of benefit for some patients, especially in those with persistent candidemia.

The crude mortality rate observed in our study was similar to that reported in other series (Almirante et al., 2005, Colombo et al., 2006, Gudlaugsson et al., 2003, Pappas et al., 2003, Pfaller et al., 2004). Adults had higher mortality rates than pediatric patients (24.10% to 16%). Similar to other reports, patients with *C. parapsilosis* candidemia had the lowest death rates (Nucci et al., 1998, Pappas et al., 2003).

Summarizing, the epidemiological and susceptible data described along the text, document important differences and similarities in the epidemiology of candidemia in Spain compared to updated reports from other countries. This report shows that candidemia is a source of significant morbidity and mortality with high associated healthcare costs. Although our high rates of candidemia may be related to many factors, reasons for them are not clear and further study is necessary. Determining them may lead to identify potential measures that can help in disease prevention. In addition, our data support that fluconazole nonsusceptibility could be associated with prior fluconazole exposure and suggest that such exposure could lead to other new azoles cross-resistance and complicate the clinical outcome of some patients.
