Open Forum Infectious Diseases MAJOR ARTICLE Early Stepdown From Echinocandin to Fluconazole Treatment in Candidemia: A Post Hoc Analysis of Three Cohort Studies E. Moreno-García,1,a P. Puerta-Alcalde,1,a, G. Gariup,1 M. Fernández-Ruiz,2 L. E. López Cortés,3 G. Cuervo,4 M. Salavert,5 P. Merino,6 M. Machado,7 J. Guinea,7 J. García-Rodríguez,8 J. Garnacho-Montero,9 C. Cardozo,1 J. Peman,5 M. Montejo,10 J. Fortún,11 B. Almirante,12 C. Castro,13 J. Rodríguez-Baño,3, J. M. Aguado,2 J. A. Martínez,1 J. Carratalà,4 A. Soriano,1, and C. Garcia-Vidal1, ; Project from GEMICOMED (SEIMC) 1 Hospital Clínic, Universitat de Barcelona, Barcelona, Spain, 2Hospital Universitario “12 de Octubre”, Instituto de Investigación Hospital “12 de Octubre” (imas+12), Universidad Complutense, Madrid, Spain, 3Unidad de Gestión Clínica de Enfermedades Infecciosas, Microbiología y Medicina Preventiva, Hospital Universitario Virgen Macarena, CSIC, Instituto de Biomedicina de Sevilla (IBiS), Seville, Spain, 4Hospital Universitari de Bellvitge, IDIBELL (Institut D’Investigació Biomèdica de Bellvitge), Universitat de Barcelona, Barcelona, Spain, 5Hospital Universitari i Politecnic “La Fe”, Valencia, Spain, 6Hospital Universitario Clínico “San Carlos”, Madrid, Spain, 7Hospital General Universitario Gregorio Marañón e Instituto de Investigación Sanitaria Gregorio Marañón, Universidad Complutense, Madrid, Spain, 8Hospital Universitario “La Paz”, Madrid, Spain, 9Unidad Clínica de Cuidados Intensivos, Hospital Universitario “Virgen Macarena”, Sevilla, Spain, 10 Hospital Universitario “Cruces”, Bilbao, Spain, 11Hospital Universitario “Ramón y Cajal”, Madrid, Spain, 12Hospital Universitari Vall d’Hebron, Universitat Autònoma de Barcelona, Barcelona, Spain, 13Hospital Universitario de Valme, Sevilla, Spain Background. There are no clear criteria for antifungal de-escalation after initial empirical treatments. We hypothesized that early de-escalation (ED) (within 5 days) to fluconazole is safe in fluconazole-susceptible candidemia with controlled source of infection. Methods. This is a multicenter post hoc study that included consecutive patients from 3 prospective candidemia cohorts (2007– 2016). The impact of ED and factors associated with mortality were assessed. Results. Of 1023 candidemia episodes, 235 met inclusion criteria. Of these, 54 (23%) were classified as the ED group and 181 (77%) were classified as the non-ED group. ED was more common in catheter-related candidemia (51.9% vs 31.5%; P = .006) and episodes caused by Candida parapsilosis, yet it was less frequent in patients in the intensive care unit (24.1% vs 39.2%; P = .043), in- fections caused by Nakaseomyces glabrata (0% vs 9.9%; P = .016), and candidemia from an unknown source (24.1% vs 47%; P = .003). In the ED and non-ED groups, 30-day mortality was 11.1% and 29.8% (P = .006), respectively. Chronic obstructive pulmonary dis- ease (odds ratio [OR], 3.97; 95% confidence interval [CI], 1.48–10.61), Pitt score > 2 (OR, 4.39; 95% CI, 1.94–9.20), unknown source of candidemia (OR, 2.59; 95% CI, 1.14–5.86), candidemia caused by Candida albicans (OR, 3.92; 95% CI, 1.48–10.61), and prior surgery (OR, 0.29; 95% CI, 0.08–0.97) were independent predictors of mortality. Similar results were found when a propensity score for receiving ED was incorporated into the model. ED had no significant impact on mortality (OR, 0.50; 95% CI, 0.16–1.53). Conclusions. Early de-escalation is a safe strategy in patients with candidemia caused by fluconazole-susceptible strains with controlled source of bloodstream infection and hemodynamic stability. These results are important to apply antifungal stewardship strategies. Keywords. antifungal; candidemia; de-escalation; invasive candidiasis; outcome. Candidemia is a frequent nosocomial complication, which [3], as well as suboptimal diagnostic tools, have driven the occurs especially in immunocompromised hosts and pa- overuse of antifungal drugs as both therapy and prophylaxis tients hospitalized in intensive care units (ICUs), with a for candidemia. prevalence of 6.9 per 1000 admitted patients according to a Current guidelines recommend echinocandins as initial em- recent survey [1, 2]. High infection-related morbidity and pirical treatment in moderate-to-severe illness due to the im- mortality (attributable mortality ranging from 10% to 49%) proved prognosis in some studies [4], which could be related to its fungicidal effect and broad antifungal spectrum [5, 6]. Echinocandins are preferred to fluconazole, because of pos- Received 16 March 2021; editorial decision 10 May 2021; accepted 13 May 2021. a E. M.-G. and P. P.-A. contributed equally to this work. sible azole resistance in Candida species, which is minimal Correspondence: Carolina Garcia-Vidal, MD, PhD, Department of Infectious Diseases, in Candida albicans and Candida parapsilosis (the most prev- Hospital Clínic de Barcelona, Carrer de Villarroel 170, 08036, Barcelona, Spain (carolgv75@ hotmail.com). alent species in Southern Europe), yet more significant in Open Forum Infectious Diseases®2021 Nakaseomyces glabrata (formerly known as Candida glabrata) © The Author(s) 2021. Published by Oxford University Press on behalf of Infectious Diseases and Candida guilliermondii (up to 45% of tested strains), and Society of America. This is an Open Access article distributed under the terms of the Creative Commons Attribution-NonCommercial-NoDerivs licence (http://creativecommons.org/licenses/ almost universal in Pichia kudriavzevii (formerly known as by-nc-nd/4.0/), which permits non-commercial reproduction and distribution of the work, in any Candida krusei) [7]. medium, provided the original work is not altered or transformed in any way, and that the work is properly cited. For commercial re-use, please contact journals.permissions@oup.com However, there are no universal de-escalation criteria for DOI: 10.1093/ofid/ofab250 fluconazole-susceptible strains. The European Society for Antifungal De-escalation in Candidemia • ofid • 1 Clinical Microbiology and Infectious Diseases (ESCMID) re- Definitions commends a de-escalation strategy at day 3 in stable patients Candidemia was defined as 1 or more blood cultures positive [6], whereas the Infectious Diseases Society of America (IDSA) for Candida species and the presence of clinically apparent guidelines advise de-escalation after 10 days of intravenous signs and symptoms of sepsis. Empirical antifungal therapy was treatment [5]. The level of these recommendations is weak due defined as that initiated before blood cultures yielded any result. to the lack of studies regarding antifungal de-escalation safety. The source of infection was determined by an infectious disease De-escalation provides the possibility of implementing oral specialist who evaluated the patient’s medical history, performed step-down therapy, diminishing costs, toxicity, and drug-drug a physical examination, and assessed results obtained from mi- interaction, and reducing the emergence of resistant strains. crobiological tests and complementary imaging. An intrave- Our hypothesis is that patients with candidemia caused by nous catheter was considered to be the source of candidemia fluconazole-susceptible strains that reach hemodynamic sta- when, in the absence of any other clinically apparent focus, any bility can undergo early de-escalation (ED) to fluconazole. of the following criteria was present: local inflammatory signs Therefore, we studied the mortality of patients who underwent or suppuration at the insertion site; a positive catheter tip cul- ED (within 5 days of a positive blood culture) and compared it ture with the same Candida spp as that isolated in the blood cul- to that of patients who did not de-escalate. tures; and an earlier growth in catheter-drawn blood culture by at least 2 hours when compared with venipuncture-drawn cul- tures. An abdominal source was defined when candidemia was MATERIALS AND METHODS simultaneous with peritonitis or a Candida spp isolate in ab- Patients, Setting, Data Collection, and Study Design dominal drainage. A urinary tract source was considered when This a retrospective post hoc study that included subjects from 3 candidemia occurred in a patient with urinary tract symptoms prospective, collaborative cohorts of patients with candidemia. and concomitant candiduria due to the same Candida spp. First, we included patients from a prospective candidemia sur- When no focal infection could be demonstrated, the source was veillance cohort from Hospital Clinic (a 700-bed tertiary uni- categorized as unknown. Comorbidity was defined as a disease versity hospital in Barcelona). Second, we included patients or treatment that could predispose patients to infection, alter from a prospective Spanish cohort used in a prior study to create defense mechanisms, or cause functional impairment, such a simple prediction score to estimate the risk of candidemia as the following items: diabetes, chronic liver disease, chronic caused by fluconazole-nonsusceptible strains [8]. Third, we in- renal failure, active neoplastic disease, severe chronic obstruc- cluded patients prospectively collected from a prior study that tive pulmonary disease, severe cardiac disease, severe dementia, aimed to assess the epidemiology of breakthrough candidemia and administration of immunosuppressive drugs, including in the era of broad-spectrum antifungal therapies [9]. corticosteroids. Septic shock was defined as a systolic pressure All patients were adults diagnosed with candidemia between under 90 mmHg that was unresponsive to fluid treatment or 2007 and 2016. No interventions to optimize the outcomes of required vasoactive drug therapy [10]. these patients had been undertaken. All patients were followed up until day 30 after candidemia, death, or discharge. For the present Microbiological Methods study, the following patients were excluded: those with fluconazole- The microbiological diagnosis followed in the whole cohort was nonsusceptible strains; those receiving empirical monotherapy with similar. In brief, 2 sets of 2 blood samples were collected from fluconazole or without antifungal treatment; those with an uncon- patients with a suspected bloodstream infection. The blood trolled source of candidemia; those with persistent fungemia; and samples were processed using either a BACTEC 9240 system those who died within the first 5 days of candidemia. (Becton-Dickinson Microbiology Systems, Franklin Lakes, NJ) The following data were obtained for all patients: age, sex, or BacTAlert (BioMérieux SA, Marcy L’Etoile, France) with a pre-existing comorbidities, prognosis of the underlying dis- 5-day incubation period. If yeast cells were observed after mi- ease, prior antimicrobial therapy, prior surgery (within the croscopic examination of a Gram stain, blood bottles were last month), Charlson comorbidity index, immunosuppressive subcultured into Sabouraud agar plates (BD BBL Stacker Plates, drugs, source of candidemia, leukocyte count, length of hos- Heidelberg, Germany) and chromogenic media (ChromAgar pitalization before candidemia diagnosis, ICU admission, need BioMerieux SA, Paris, France). Yeast isolates were identified for mechanical ventilation, empirical and definitive antifungal by conventional methods (biochemical methods in the first treatment, presence of shock, and mortality. years, and matrix-assisted laser desorption ionization time-of- Patients were divided into 2 groups: those who underwent ED flight or pan-fungal polymerase chain reaction after 2010). The (within 5 days of the fungemia) of any antifungal to fluconazole, antifungal susceptibility of the isolates was determined in ac- and those who did not. For the purpose of this study, these cordance with the Clinical and Laboratory Standards Institute groups will be referred to as the ED group and non-ED group. M27-S3 document [11]. In vitro antifungal activity was studied There was no official hospital policy concerning de-escalation. by applying a commercial microdilution method (YeastOne 2 • ofid • Moreno-García et al Sensititre, TREK Diagnostic Systems, Independence, OH) or an human immunodeficiency virus-positive serology. The propen- Etest (bioMérieux SA, Marcy L’Etoile, France). The fluconazole sity score for ED was then used as a covariate in a multivariable minimum inhibitory concentration (MIC) was defined as the analysis to adjust for potential confounding factors. The goodness lowest drug concentration to inhibit 50% of growth compared of fit of the final multivariable model was assessed again with the with the growth control after 24-hour incubation at 35°C for all Hosmer-Lemeshow test and the area under the receiver operating Candida species [12]. However, N glabrata was an exception, characteristic curve. Finally, a multivariable regression model which was determined after 48 hours to prevent misclassifi- (step-forward procedure) was used to identify independent risk cation bias among the isolates [13]. Of note, Candida species factors for mortality in episodes with candidemia. All data were isolates with an MIC ≥4 mg/L to fluconazole were considered analyzed using SPSS software version 20.0 (IBM SPSS, Chicago, nonsusceptible to fluconazole, except for N. glabrata, which was IL). Statistical significance was established at α = 0.05, and all re- considered susceptible-dose dependent to fluconazole when ported P values are 2-tailed. the MIC value was ≤32 mg/L. In addition, P. kudriavzevii was considered nonsusceptible to fluconazole, regardless of the MIC Patient Consent Statement value. Quality controls were performed in each center using C. This observational study was conducted in accordance with parapsilosis American Type Culture Collection (ATCC) 22019 the Declaration of Helsinki. The design of the work has been and P. kudriavzevii ATCC 6258. approved by local ethical committees of each participating hospital. To protect personal privacy, identifiable informa- tion in the electronic database was encrypted for each patient. Statistical Analysis Informed consent was waived due to the retrospective nature Quantitative variables are reported as median with interquartile of the study, the fact that no intervention was involved, and no range (IQR); categorical variables are reported as absolute num- patient-identifiable information was included. bers and percentages. To detect significant differences between groups, we used the χ2 test or Fisher’s exact test for categorical vari- RESULTS ables, and the Student’s t test or Mann-Whitney U test for contin- uous variables, when deemed appropriate. Factors associated with Demographics, Epidemiology, and Microbiology mortality were evaluated by univariate and multivariable analyses, In the overall cohort, 1023 episodes of candidemia were docu- with the multivariable analysis including all significant variables mented over the study period. Of those, 235 patients met the (P < .05) from the univariate analysis. Given the lack of randomi- inclusion criteria: 54 (23%) were classified into the ED group zation of initial therapies, a propensity score for antifungal de-esca- and 181 (77%) into the non-ED group. Figure 1 shows the study lation was estimated using a backward stepwise logistic regression flowchart. Median time to de-escalation in the ED-group was model that included variables related with de-escalation with 3.5 days (IQR, 3–5). Table 1 summarizes the demographic and P ≤ .125 in the univariate analysis. The following variables were clinical characteristics of the patients, as well as microbiological included based on their presence (yes/no) during the candidemia results, source of candidemia, and outcomes by study groups. episode: Candida species (C. albicans, C. parapsilosis, N. glabrata); No significant differences were documented in baseline demo- central catheter at diagnosis; catheter-related and unknown graphics between groups. ED was less common in patients who source of fungemia; ICU patient; breakthrough candidemia; and developed candidemia in the ICU (24.1% vs 39.2%; P = .043). Total cohort: 1023 episodes 788 excluded: • Fluconazole-non-susceptible strains: 166 • Initial monotherapy with fluconazole: 525 • No antifungal treatment: 48 • Death within the first 5 days of candidemia: 38 • Uncontrolled source of candidemia: 11 235 eligible cases Non-Early Early de-escalation de-escalation group: group: 54 181 Figure 1. Study flowchart. Antifungal De-escalation in Candidemia • ofid • 3 Table 1. Main Characteristics and Outcomes of Patients Divided by Study Groups Characteristics ED Group n = 54 (%) Non-ED Group n = 181 (%) Total N = 235 (%) P Value Demographics Male sex, n (%) 36 (66.7) 101 (55.8) 137 (58.3) .155 Age (years), median (IQR) 65.5 (55–75) 62.2 (48–71) 62.9 (49–72) .617 Age ≥65 years 28 (51.9) 83 (45.9) 111 (47.2) .439 Comorbidities, n (%) Diabetes mellitus 14 (25.9) 37 (20.4) 51 (21.7) .391 COPD 7 (13) 32 (17.7) 39 (16.6) .414 Solid organ malignancy 20 (37) 58 (32.0) 78 (33.2) .494 Hematological malignancy 5 (9.3) 26 (14.4) 31 (13.2) .331 Hematopoietic stem cell transplantation 3 (5.6) 11 (6.1) 14 (6.0) 1.000 Solid organ transplantation 5 (9.3) 21 (11.6) 26 (11.1) .630 Chronic liver disease 6 (11.1) 29 (16) 35 (14.9) .374 Chronic renal failure 10 (18.5) 42 (23.2) 52 (22.1) .467 HIV infection 0 (0.0) 10 (5.5) 10 (4.3) .122 Age-adjusted Charlson comorbidity index, median (IQR) 2 (1–5) 3 (2–4) 3 (2–4) .949 Pitt score >2 14 (31.8) 64 (42.1) 78 (39.8) .220 Risk factors for candidemia Neutropenia (<500/mm3) 5 (9.3) 21 (11.6) 26 (11.1) .630 Parenteral nutrition 15 (27.8) 75 (46) 90 (38.3) .157 Immunosuppressant therapy 15 (27.8) 42 (23.1) 57 (24.3) .336 Surgical patients 14 (25.9) 41 (22.6) 55 (23.4) .632 Central venous catheter 49 (90.7) 173 (95.6) 222 (94.5) .073 Previous antibiotic therapy (last month) 41 (75.9) 151 (83.4) 192 (81.7) .265 ICU hospitalization 13 (24.1) 71 (39.2) 84 (35.7) .043 Source of Candidemia Unknown 13 (24.1) 85 (47) 98 (41.7) .003 Catheter-related 28 (51.9) 57 (31.5) 85 (36.2) .006 Urological 5 (9.3) 9 (5) 14 (6.0) .243 Abdominal 5 (9.3) 18 (9.9) 23 (9.8) .882 Other sources 3 (5.6) 12 (6.6) 15 (6.4) 1.000 Candida Species Candida albicans 20 (37) 91 (50.3) 111 (47.2) .087 Nakaseomyces glabrataa 0 (0.0) 18 (9.9) 18 (7.7) .016 Candida parapsilosis 23 (42.6) 40 (22.1) 63 (26.8) .003 Candida tropicalis 12 (22.2) 27 (14.9) 39 (16.6) .205 Other species 2 (3.7) 5 (2.8) 7 (3.0) .662 Severity Indicators Shock 14 (25.9) 58 (32.2) 72 (30.8) .379 Outcomes Length of hospital stay (days), median (IQR) 19.5 (14–53) 20 (10–37) 20 (10–39) .947 Related mortality (7 days) 3 (5.6) 11 (6.1) 14 (6.0) 1.000 Overall mortality (30 days) 6 (11.1) 54 (29.8) 60 (25.5) .006 Bold values indicate variables that were statistically significant. Abbreviations: COPD, chronic obstructive pulmonary disease; ED, early de-escalation; HIV, human immunodeficiency virus; ICU, intensive care unit; IQR, interquartile range. a Formerly known as Candida glabrata. Most cases of candidemia were caused by C albicans (111 group (P = .006). In the ED-group, exact days until de-esca- cases, 47%). Patients in the ED group more frequently had in- lation had no impact on mortality (P = .665). The evolution fection due to C. parapsilosis (42.6% vs 22.1%; P = .003) and of the rates of ED and mortality throughout the years is in- catheter-related candidemia (51.9% vs 31.5%; P = .006), yet cluded in Supplementary Table 1 and Supplementary Table 2, they less frequently presented with an infection caused by N. respectively. glabrata (0% vs 9.9%; P = .016) and candidemia from an un- known source (24.1% vs 47%; P = .003). Risk Factors for Mortality In addition, 30-day mortality was 25.5% (60 patients Table 2 summarizes univariate and multivariable anal- of 235): 11.1% in the ED group and 29.8% in the non-ED ysis of risk factors related with 30-day mortality. Chronic 4 • ofid • Moreno-García et al obstructive pulmonary disease (OR, 3.97; 95% CI, 1.48–10.61), DISCUSSION Pitt score > 2 (OR, 4.39; 95% CI, 1.94–9.20), unknown source The results in this study suggest that antifungal de-escalation of candidemia (OR, 2.59; 95% CI, 1.14–5.86), and candidemia could be a safe strategy in patients with candidemia caused by caused by C albicans (OR, 3.92; 95% CI, 1.48–10.61) were in- fluconazole-susceptible strains when source of infection is con- dependent predictors of mortality, whereas prior surgery trolled and the patient is hemodynamically stable. Such de-es- (OR, 0.29; 95% CI, 0.08–0.97) was protective. The Hosmer- calation strategy could be implemented in patients admitted to Lemeshow test value was 0.164. The discriminatory power of both ICU and regular wards. the model, as evaluated by the area under the receiver operating Some studies, with substantial methodological differences characteristic curve, was 0.824 (95% CI, 0.755–0.893), showing in comparison with ours, have tried to evaluate the impact a strong ability to predict overall mortality. Similar results were of de-escalation in patients with candidemia or other forms found when the propensity score for receiving ED was incor- of invasive candidiasis. Vazquez et al [14] evaluated an early porated into the model. Early de-escalation had no significant step-down strategy from anidulafungin to an oral azole in 102 impact on the multivariable model (Supplementary Table 3 of 250 (40%) patients with candidemia or invasive candidi- shows multivariable analysis after incorporating the propensity asis, showing response rates similar to nonearly switch. The score to the model). Table 3 shows univariate and multivariable population of early step-down strategy was less severely ill analyses of factors associated with early de-escalation. The and de-escalation was performed within the first 7 days. Van goodness of fit of the propensity score was assessed with the der Geest et al [15] described 32 patients with invasive can- Hosmer-Lemeshow test (P = .69). The discriminatory power of didiasis belonging to a cohort of 56 patients admitted to the the de-escalation model, as evaluated by the area under the re- ICU who underwent step-down-to-fluconazole therapy (at ceiver operating curve, was 0.74 (95% CI, 0.66–0.82), showing a a median of 5 days). Regarding the episodes, no differences moderate ability to predict ED strategy. Table 2. Risk Factors for Overall Mortality: Univariate and Multivariable Analyses Risk Factor Univariate OR (95% CI) P Value Multivariable OR (95% CI) P Value Male sex 1.09 (0.60–1.98) .766 - - Age ≥65 years 1.27 (0.71–2.28) .425 - - Diabetes mellitus 1.46 (0.74–2.88) .280 - - COPD 2.75 (1.34–5.64) .005 3.97 (1.48–10.61) .006 Solid organ malignancy 1.23 (0.67–2.27) .508 - - Hematological malignancy 0.52 (0.19–1.42) .198 - - Hematopoietic stem cell transplantation 1.68 (0.54–5.22) .368 - - Solid organ transplantation 1.08 (0.43–2.72) .863 - - Chronic liver disease 2.58 (1.22–5.45) .011 2.47 (0.89–6.84) .082 Chronic renal failure 2.50 (1.30–4.83) .005 2.08 (0.83–5.19) .118 Charlson Comorbidity Index >3 2.24 (1.23–4.08) .008 1.19 (0.48–2.94) .714 Pitt score >2 4.68 (2.34–9.34) <.001 4.39 (1.94–9.20) <.001 Neutropenia (<500/mm3) 1.64 (0.69–3.91) .260 - - Parenteral nutrition 1.07 (0.58–1.98) .837 - - Immunosuppressant therapy 2.16 (1.18–3.96) .012 1.10 (0.46–2.61) .830 Surgical patients 0.21 (0.08–0.57) .001 0.29 (0.08–0.97) .045 Central venous catheter 1.56 (0.33–7.41) .734 - - Previous antibiotic therapy (last month) 1.67 (0.54–5.17) .368 - - ICU hospitalization 2.54 (1.35–4.75) .003 1.28 (0.50–3.31) .607 Unknown source of candidemia 1.89 (1.04–3.41) .034 2.59 (1.14–5.86) .023 Catheter-related candidemia 0.56 (0.29–1.07) .076 - - Urinary source of candidemia 0.79 (0.21–2.91) 1.000 - - Abdominal source of candidemia 0.59 (0.19–1.80) .454 - - Candidemia by Candida albicans 2.20 (1.20–4.00) .009 3.92 (1.74–8.86) .001 Candidemia by Nakaesomyces glabrata 0.34 (0.08–1.54) .171 - - Candidemia by Candida parapsilosis 0.39 (0.18–0.86) .017 1.39 (0.40–4.85) .608 Candidemia by Candida tropicalis 1.01 (0.46–2.21) .986 - - Catheter removal 0.92 (0.41–2.05) .840 - - Early de-escalation (within the first 5 days) 0.29 (0.12–0.73) .006 0.50 (0.16–1.53) .223 Shock 2.44 (1.32–4.51) .004 0.53 (0.18–1.55) .248 Bold values indicate variables that were statistically significant. Abbreviations: CI, confidence interval; COPD, chronic obstructive pulmonary disease; ICU, intensive care unit; OR, odds ratio. Antifungal De-escalation in Candidemia • ofid • 5 Table 3. Univariate and Multivariable Analyses of Factors Associated With Early De-Escalation Risk Factor Univariate OR (95% CI) P Value Multivariate OR (95% CI) P Value Male sex 1.58 (0.84–2.37) .155 - - Age ≥65 years 1.27 (0.69–2.34) .439 - - Diabetes mellitus 1.36 (0.67–2.76) .391 - - COPD 0.69 (0.29–1.67) .414 - - Solid organ malignancy 1.25 (0.66–2.35) .494 - - Hematological malignancy 0.61 (0.22–1.67) .331 - - Hematopoietic stem cell transplantation 0.91 (0.24–3.38) 1.000 - - Solid organ transplantation 0.78 (0.28–2.17) .630 - - Chronic liver disease 0.66 (0.26–1.67) .374 - - HIV infection 0.76 (0.71–0.82) .122 0.01 (0.01-NE) .999 Chronic renal failure 0.75 (0.35–1.62) .467 - - Charlson comorbidity index >3 0.97 (0.51–1.83) .922 - - Pitt score >2 0.64 (0.31–1.31) .220 - - Neutropenia (<500/mm3) 0.78 (0.28–2.17) .630 - - Parenteral nutrition 0.61 (0.30–1.22) .157 - - Immunosuppressant therapy 1.44 (0.68–3.06) .336 - - Surgical patients 1.19 (0.58–2.43) .632 - - Central venous catheter 0.34 (0.10–1.16) .073 0.49 (0.13–1.84) .293 Previous antibiotic therapy (last month) 0.58 (0.22–1.52) .265 - - Breakthrough candidemia 0.43 (0.18–1.03) .053 0.38 (0.15–0.97) .043 ICU hospitalization 0.49 (0.24–0.99) .043 0.51 (0.24–1.08) .079 Unknown source of candidemia 0.36 (0.18–0.71) .003 0.26 (0.12–0.58) .001 Catheter-related candidemia 2.34 (1.26–4.35) .006 1.86 (0.72–4.82) .198 Urinary source of candidemia 1.95 (0.62–6.09) .243 - - Abdominal source of candidemia 0.92 (0.33–2.62) .882 - - Candidemia by Candida albicans 0.58 (0.31–1.09) .087 1.07 (0.44–2.62) .885 Candidemia by Nakaesomyces glabrata 0.75 (0.70–0.81) .016 0.01 (0.01-NE) .998 Candidemia by Candida parapsilosis 2.61 (1.37–4.98) .003 3.00 (1.43–6.30) .004 Candidemia by Candida tropicalis 1.63 (0.76–3.49) .205 - - Catheter removal 0.85 (0.37–1.95) .695 - - Shock 0.74 (0.37–1.46) .379 - - Bold values indicate variables that were statistically significant. Abbreviations: CI, confidence interval; COPD, chronic obstructive pulmonary disease; HIV, human immunodeficiency virus; ICU, intensive care unit; NE, not evaluable; OR, odds ratio. were observed in characteristics or risk factors, and response or included different infection profiles, making extrapolation rates and 28-day and 90-day mortality were similar in both to patients with candidemia difficult. Nevertheless, this is the patient groups. Furthermore, Bailly et al [16] reported a largest-to-date study focusing specifically on patients with multicenter, observational study of 647 nonneutropenic pa- candidemia. tients with suspected or documented invasive candidiasis ad- The importance of de-escalation is the possibility of avoiding mitted to French ICUs in 2012–2013. ED was implemented in adverse events, lower treatment costs [18], and reduce selection 142 (22%) patients; 70 presented with documented invasive pressure. It stands to reason that de-escalation to oral or intra- candidiasis, 32 of whom had candidemia. No differences in venous fluconazole—as soon as when antifungal susceptibility 28-day mortality were documented among patients with or testing states the nonresistance of an isolated strain—should not without early de-escalation. Only 2 prior investigations have worsen the outcomes in terms of morbidity or mortality. In cen- included solely patients with candidemia. Bal et al [17] re- ters that were unable to perform on-site susceptibility testing, this ported 19 patients with candidemia who underwent de-es- form of de-escalation could also be considered when the isolated calation from echinocandin or voriconazole to fluconazole species is typically susceptible. More clinical data would be nec- within a median of 5 (3–9) days. One patient relapsed with essary to further support this strategy. In addition, independent a N. glabrata infection. Finally, in the study by Garnacho- of antifungal susceptibility, echinocandins have been shown to be Montero et al [4], 44 ICU patients underwent de-escalation superior to azoles in candidemia due to their increased fungicidal from echinocandin to fluconazole, presenting lower 30-day activity [19, 20]. However, this fungicidal effect is perhaps most and 90-day mortality. However, most of these studies had ei- important at infection onset when fungal load reaches its peak ther too low of a number of patients to draw any conclusion and decreases thereafter. 6 • ofid • Moreno-García et al Nonetheless, our study has a clear and good methodology References with some limitations. First, it was a nonrandomized, obser- 1. Kett DH, Azoulay E, Echeverria PM, Vincent JL. Candida bloodstream infections in intensive care units: analysis of the extended prevalence of infection in inten- vational, retrospective, post hoc study. The absence of con- sive care unit study. Crit Care Med 2011; 39:665–70. founding factors cannot be assumed despite propensity score 2. Horn DL, Neofytos D, Anaissie EJ, et al. Epidemiology and outcomes of candidemia in 2019 patients: data from the prospective antifungal therapy alli- adjustment. Second, these results are only applicable in a ance registry. Clin Infect Dis 2009; 48:1695–703. selected group of patients. Finally, these results should be con- 3. Gudlaugsson O, Gillespie S, Lee K, et al. Attributable mortality of nosocomial candidemia, revisited. Clin Infect Dis 2003; 37:1172–7. firmed in a randomized clinical trial. 4. Garnacho-Montero J, Díaz-Martín A, Cantón-Bulnes L, et al. Initial antifungal strategy reduces mortality in critically ill patients with candidemia: a propensity score-adjusted analysis of a multicenter study. Crit Care Med 2018; 46:384–93. CONCLUSIONS 5. Pappas PG, Kauffman CA, Andes DR, et al. Clinical practice guideline for the management of candidiasis: 2016 update by the Infectious Diseases Society of In summary, in adult patients with candidemia caused by America. Clin Infect Dis 2016; 62:e1–50. fluconazole-susceptible strains who are clinically stable 6. Cornely OA, Bassetti M, Calandra T, et al.; ESCMID Fungal Infection Study and have an appropriate source control, early de-escala- Group. ESCMID* guideline for the diagnosis and management of Candida dis- eases 2012: non-neutropenic adult patients. Clin Microbiol Infect 2012; 18 (Suppl tion (within 5 day from the candidemia) seems to be a safe 7):19–37. strategy that allows for oral step-down treatment, avoids 7. Pfaller MA, Moet GJ, Messer SA, et al. Candida bloodstream infections: compar- ison of species distributions and antifungal resistance patterns in community- side effects, reduces costs, and limits the emergence of re- onset and nosocomial isolates in the SENTRY Antimicrobial Surveillance sistance to drugs. Program, 2008–2009. Antimicrob Agents Chemother 2011; 55:561–6. 8. Cuervo G, Puig-Asensio M, Garcia-Vidal C, et al. A simple prediction score for estimating the risk of candidaemia caused by fluconazole non-susceptible strains. Supplementary Data Clin Microbiol Infect 2015; 21:684.e1–9. Supplementary materials are available at Open Forum Infectious Diseases 9. Cuervo G, Garcia-Vidal C, Nucci M, et al. Breakthrough candidaemia in the era online. Consisting of data provided by the authors to benefit the reader, of broad-spectrum antifungal therapies. Clin Microbiol Infect 2016; 22:181–8. the posted materials are not copyedited and are the sole responsibility 10. Levy MM, Fink MP, Marshall JC, et al. 2001 SCCM/ESICM/ACCP/ATS/SIS International Sepsis Definitions Conference. Intensive Care Med 2003; 22:530–8. of the authors, so questions or comments should be addressed to the 11. CLSI. M27-A3. Reference Method for Broth Dilution Antifungal Susceptibility corresponding author. Testing of Yeasts; Approved Standard. 3rd ed. Wayne, PA; Clinical and Laboratory Standards Institute; 2008. 12. Guinea J, Zaragoza Ó, Escribano P, et al. Molecular identification and antifungal Acknowledgments susceptibility of yeast isolates causing fungemia collected in a population- We thank Anthony Armenta for providing medical editing assistance for based study in Spain in 2010 and 2011. Antimicrob Agents Chemother 2014; 58:1529–37. this manuscript. 13. Ostrosky-Zeichner L, Rex JH, Pfaller MA, et al. Rationale for reading fluconazole Disclaimer. The funding bodies did not have a role in the study design, MICs at 24 hours rather than 48 hours when testing Candida spp. by the CLSI data collection and analysis, decision to publish, or preparation of the M27-A2 standard method. Antimicrob Agents Chemother 2008; 52:4175–7. manuscript. 14. Vazquez J, Reboli AC, Pappas PG, et al. Evaluation of an early step-down strategy Financial support. This research forms part of an activity that has re- from intravenous anidulafungin to oral azole therapy for the treatment of ceived funding from EIT Health. EIT Health is supported by the European candidemia and other forms of invasive candidiasis: results from an open-label Institute of Innovation and Technology (EIT), a body of the European trial. BMC Infect Dis 2014; 14:97. Union that receives support from the European Union´s Horizon 2020 15. van der Geest PJ, Rijnders BJ, Vonk AG, Groeneveld AB. Echinocandin to Research and Innovation Program. This study has been cofunded by the fluconazole step-down therapy in critically ill patients with invasive, susceptible Candida albicans infections. Mycoses 2016; 59:179–85. European Regional Development Fund. E. M.-G. (PI18/01061), P. P.-A. 16. Bailly S, Leroy O, Montravers P, et al. Antifungal de-escalation was not associated with (“Rio Hortega” contract CM18/00132), M. F.-R. (“Miguel Servet” con- adverse outcome in critically ill patients treated for invasive candidiasis: post hoc ana- tract CP18/00073), and C. G.-V. (FIS PI18/01061) have received research lyses of the AmarCAND2 study data. Intensive Care Med 2015; 41:1931–40. grants from the Ministerio de Sanidad y Consumo, Instituto de Salud 17. Bal AM, Shankland GS, Scott G, et al. Antifungal step-down therapy based on Carlos III. hospital intravenous to oral switch policy and susceptibility testing in adult pa- Potential conflicts of interest. C. G.-V. has received honoraria for talks tients with candidaemia: a single centre experience. Int J Clin Pract 2014; 68:20–7. on behalf of Gilead Science, MSD, Novartis, Pfizer, Janssen, Lilly as well 18. Masterton RG, Casamayor M, Musingarimi P, et al. De-escalation from as a grant from Gilead Science and MSD. A. S. has received honoraria for micafungin is a cost-effective alternative to traditional escalation from fluconazole talks on behalf of Merck Sharp and Dohme, Pfizer, Novartis, Angellini, in the treatment of patients with systemic Candida infections. J Med Econ 2013; 16:1344–56. as well as grant support from Pfizer. P. P-A. has received honoraria for 19. Reboli AC, Shorr AF, Rotstein C, et al. Anidulafungin compared with fluconazole talks on behalf of Merck Sharp and Dohme, Pfizer, Gilead and Alexion. for treatment of candidemia and other forms of invasive candidiasis caused by J. A. M. has received honoraria for talks on behalf of Merck Sharp and Candida albicans: a multivariate analysis of factors associated with improved out- Dohme, Pfizer, Novartis and Angellini. All authors have submitted the come. BMC Infect Dis 2011; 11:261. ICMJE Form for Disclosure of Potential Conflicts of Interest. Conflicts 20. Kullberg BJ, Viscoli C, Pappas PG, et al. Isavuconazole versus caspofungin in the that the editors consider relevant to the content of the manuscript have treatment of candidemia and other invasive Candida infections: the ACTIVE been disclosed. trial. Clin Infect Dis 2019; 68:1981–9. Antifungal De-escalation in Candidemia • ofid • 7