Background
Despite the availability of new antifungal agents, systemic candidiasis (i.e., candidemia and other forms of invasive candidiasis) continue to contribute to excess morbidity, greater mortality, prolonged hospitalizations, and increased costs [
1,
2]. Epidemiological data demonstrate that the frequency of
Candida infections is rising [
3‐
6], along with an increase in the proportion of infections caused by non-
albicans Candida species that are intrinsically resistant or variably susceptible to fluconazole [
7‐
10]. However,
C. albicans continues to be the most common
Candida species isolated [
6,
11,
12].
The vast majority of
C. albicans isolates from bloodstream infections remain fully susceptible to fluconazole, which has been the treatment of choice for these infections in most settings [
13]. However, updated IDSA guidelines now favor an echinocandin (anidulafungin, caspofungin, or micafungin) as first-line treatment for systemic candidiasis in moderately severe to severely ill patients or those with prior azole exposure [
14]. Echinocandins have several potential advantages over fluconazole for the treatment of systemic candidiasis. They have a broader spectrum of activity (encompassing fluconazole-resistant
C. glabrata and
C. krusei) and exhibit potent fungicidal activity against most
Candida species [
15‐
17]. Echinocandins are highly active in vitro against
C. albicans, with lower MICs than those of fluconazole [
8,
18]. While fluconazole and the echinocandins have similarly favorable safety profiles, the latter do not require dose adjustment in patients with renal insufficiency [
19‐
21].
A phase III, randomized, double-blind study compared anidulafungin with fluconazole as primary treatment of systemic candidiasis in adult patients infected with any
Candida species, except
C. krusei [
22]. Global response rates at the end of IV study treatment in mITT patients were significantly higher with anidulafungin (76%) than fluconazole (60%). The superiority of anidulafungin was questioned because of a potential center effect. However, a number of robust statistical tests all failed to show the presence of such an effect [
22].
Of note,
C. albicans was identified as the cause of infection in the majority of cases (62%) and the difference in global responses among this clinically important subgroup was remarkable: 81% with anidulafungin versus 62% with fluconazole (
p = 0.02) [
22]. Since
C. albicans isolates were almost uniformly susceptible to fluconazole [
22], the excess failures in the fluconazole group could not be attributed to antifungal resistance. However, due to the lack of a multivariable analysis, it was unclear whether these differences were potentially related to unknown host factors. We therefore conducted a post-hoc analysis using data collected from that trial, in order to explore the factors associated with the better global response of anidulafungin, specifically in patients with
C. albicans infections.
Discussion
This study represents the first clinical analysis of differential response in patients with systemic candidiasis caused by C. albicans to an echinocandin versus fluconazole, conducted post-hoc in the relevant subpopulation from a prospective clinical trial. The significance of this analysis is that it enabled us to directly compare the efficacy of a fungicidal with that of a fungistatic drug in patients with C. albicans infections, without potential confounding by differences in susceptibility. Despite all isolates of C. albicans being susceptible to the antifungal agent received by each patient, response to anidulafungin was significantly greater than to fluconazole. Logistic regression analysis, using the Akaike Information Criterion methodology, demonstrated that study drug and baseline APACHE II score were the principal independent variables determining this outcome, reflecting pharmacological and host factors, respectively. Study drug remained a significant predictor of treatment success even after adjustment for APACHE II scores and center variability, with an estimated OR of 2.6. While there was notable variation in this odds ratio, in part due to the relatively small sample size, it remained significant from a statistical perspective and should therefore be considered important. These results suggest a pharmacological benefit of anidulafungin in the treatment of systemic candidiasis due to C. albicans, since no other factor, such as potential baseline imbalances and center variability, was shown to have impacted the outcome.
Understanding the factors determining clinical outcomes in patients with
C. albicans infections was important, since treatment differences among this most common subpopulation in turn drove the treatment differences observed in the overall clinical trial [
22]. Among patients infected with
C. albicans, anidulafungin resulted in better global response at end of IV study treatment, at the end of all treatment, and at the 2-week follow-up. Moreover, patients treated with anidulafungin also had more rapid eradication of yeast from the blood, as well as lower rates of persistent
C. albicans infection.
Whether the better response of anidulafungin in the subpopulation of patients with systemic candidiasis caused by
C. albicans emerged as the result of an imbalance in characteristics between the two treatment arms was not evident from the previously available data. However, the present analyses show that the efficacy differences were not due to imbalances in baseline clinical or demographic characteristics between treatment arms. There was also no imbalance in the proportion of patients whose central venous catheters remained in place during the study; this factor is considered important, because removal or replacement of central venous catheters may lead to better outcomes [
26,
27]. Moreover, the lower response to fluconazole could also not be explained by in vitro resistance. Finally, first-line therapy with anidulafungin was identified as a significant and independent predictor of successful treatment outcome and remained so after adjustment for APACHE II score.
Even though we did not specifically test fungicidality, the recognized fungicidal properties of the echinocandin may have contributed to this outcome. In fact, the more rapid clearance of blood cultures and greater patient survival during the early phase of therapy are consistent with the pharmacodynamic properties of a fungicidal agent [
15,
16,
28‐
30]. Similar results have also been observed in pre-clinical studies. For instance, a neutropenic murine model of disseminated
C. albicans infection demonstrated that anidulafungin yielded significantly greater reduction of fungal burden in the kidneys than fluconazole [
31] and, in a neutropenic rabbit model of systemic candidiasis, anidulafungin cleared
C. albicans from tissues more effectively than fluconazole [
32]. Organism-mediated tissue injury appears to be an intrinsic component of
C. albicans pathogenesis [
33‐
37], and failure to quickly control candidemia can lead to disseminated infection and poor outcomes [
38,
39]. Considered together, these results suggest that the enhanced fungicidal activity of echinocandins may have an impact on treatment outcomes in invasive
C. albicans infections.
Although there was a clear clinical benefit with anidulafungin, this did not translate into a difference in long-term survival. We can merely speculate why this may have occurred. It is possible that later deaths were attributable to underlying illnesses rather than to systemic candidiasis. In a matched case-controlled study among candidemia-exposed and candidemia-unexposed patients, slightly less than half of the overall mortality was caused by candidemia [
2]. Another possible explanation is that optimal management of patients with invasive
Candida infections includes other interventions besides antifungal therapy, and these factors were not controlled for in our study. Also, the clinical trial may have selected for patients with better prognoses, thus making it more difficult to demonstrate a survival benefit.
Our study has several potential limitations. Although the comparison of global response in patients with C. albicans infection was planned a priori, the analyses described in this manuscript were performed post-hoc in a subset of patients, albeit using prospectively collected data from a double-blind, randomized clinical trial. Owing to the post-hoc nature of these analyses, the two treatment groups did not have the same sample size, which was not chosen to be statistically powered. However, this slight disproportion should not impact our overall results, as all relevant differences in baseline factors were adjusted for using multivariate logistic regression; the odds ratio for the treatment effect is thus corrected for any baseline imbalances. Of note, despite incorporating a relatively large number of baseline variables into the multivariate analyses, there may have been unidentified confounding variables with a potentially significant impact on response. Since no adjustment was made for multiplicity, the results should be interpreted with some caution. Another potential limitation is the lack of daily blood culture collection during the original trial, which precluded an accurate determination of the exact time to negative blood culture; the corresponding analysis should thus be interpreted with some caution. This shortcoming is somewhat compensated for by the fact that the vast majority of blood cultures were collected on time points (± 1 day) prespecified by the study protocol. Finally, extrapolation of our results to all patients with systemic candidiasis due to C. albicans may not be appropriate, since we focused on patients with C. albicans only and excluded those with mixed (albicans and non-albicans) infections.
Competing interests
Potential conflicts of interest: ACR has received clinical research grant support from Merck and Pfizer, has been a consultant for Merck, Astellas, and Pfizer, and has been a lecturer for Pfizer and Merck. AFS has served as a consultant to, speaker for, and/or has received research support from Astellas, Merck, and Pfizer. CR has received grant/research support from Astellas, Basilea, Johnson & Johnson, Merck, and Pfizer; has been a consultant to Astellas, Bayer, iCo, Merck, and Pfizer; and has served on Speakers' Bureaus for Astellas, Bayer, Johnson & Johnson, Merck, and Pfizer. PGP has received grant/research support from Merck, Pfizer, and Astellas, and has been an ad hoc advisor to Novartis, Merck, Astellas and Pfizer. DHK has received research support from Pfizer, Astellas, and Akers Bioscience; has been a consultant to Pfizer and Astellas; and has served as a speaker for Pfizer, Astellas, and GlaxoSmithKline. TJW has served as consultant to Trius, Novartis, Vestagen, Sigma Tau, and iCo. HTS, ALR, and PB are full-time employees of Pfizer Inc.
Authors' contributions
ACR, CR, PGP, DHK and TJW participated in the original clinical trial. ALR and PB conducted the statistical analyses described in this paper. ACR, AFS, HTS and TJW interpreted the statistical analyses, with subsequent input from CR, PGP, and DHK. ACR and HTS wrote the first draft of the manuscript. All authors critically revised the manuscript for important intellectual content and have read and approved the final version.