Caspofungin versus fluconazole as prophylaxis of invasive fungal infection in high-risk liver transplantation recipients: A propensity score analysis

Invasive fungal infections (IFIs) are associated with high mortality in liver transplantation recipients (LTRs).1-5 Although most IFIs are due to Candida spp., the risk and severity of invasive aspergillosis (IA) in LTRs is well known.6, 7 Universal prophylaxis, mainly with fluconazole, is controversial in patients with a low risk of IFIs. It has been associated with an increased risk of toxicity and interactions and may have an ecological impact through the selection of resistant strains.8, 9 The administration of prophylaxis according to the presence of high-risk (HR) factors for IFIs is a more efficient strategy.8 The benefit of voriconazole or itraconazole, both of which are active against Aspergillus spp., is limited because of potential liver toxicity.10 Prophylaxis with a lipid preparation of amphotericin B (AmB) significantly reduced the frequency of IFI in HR-LTRs.11, 12 However, prophylaxis with a lipid formulation of AmB may be limited by infusion-related toxicity and nephrotoxicity, particularly in patients with renal failure, a major risk factor for the development of IFI in LTRs.13

Echinocandins are an attractive alternative for prophylaxis of IFI in HR-LTRs because of their specific mechanism of action, minimal nephrotoxicity, and activity against both Candida and Aspergillus species. Winston et al.14 recently published the results of a clinical trial comparing anidulafungin and fluconazole in HR-LTRs. For most patients at high risk of IFI, anidulafungin and fluconazole have similar prophylactic efficacy. In the case of patients who are at high risk of IA or have received fluconazole before transplantation, anidulafungin could prove beneficial.14

The TENPIN study is a recent, international, multicenter trial comparing micafungin with the standard of care (fluconazole, caspofungin, or liposomal AmB) in 344 HR-LTRs.15 The per protocol analysis showed that micafungin was successful (98.6% free of IFI) and not inferior to standard of care (99.3%), although kidney function was better preserved with micafungin.15

A prospective study performed by our group with 81 LTRs in Spain demonstrated that caspofungin was effective in reducing the frequency of IFI.16 Consequently, caspofungin began to be widely used as prophylaxis of IFI in LTRs in Spain. However, the real impact of this measure has not been analyzed in a large cohort of patients.

We performed a multicenter, retrospective study in 9 Spanish hospitals to evaluate the effectiveness of targeted prophylaxis for the prevention of IFI in HR-LTRs. We compared 2 strategies: caspofungin (group 1) and fluconazole (group 2).

Patients and Methods

Results

During the study period (2005-2012), a total of 3784 patients received liver transplantations at the 9 participating institutions; our analysis was based on 195 (5.2%) of these patients. The distribution was as follows: group 1 (caspofungin), 97 patients; group 2 (fluconazole), 98 patients. The median time to initiation of antifungal prophylaxis after transplantation was 3.2 days (1 to 18 days) in group 1 and 2.6 days (1 to 14 days) in group 2. The median duration of prophylaxis in groups 1 and 2 was 22 and 24 days, respectively. Fluconazole was administered at a median dose of 200 mg/day (100-400 mg) and caspofungin at 50 mg/day.

Table 1 shows the baseline characteristics of patients included in the 2 groups. The most frequent risk factors for antifungal prophylaxis were multiple TR (>20 UCBP) in 69 (35.4%) patients, RR therapy in 62 (31.8%) patients, and repeat abdominal surgery in 37 (18.9%) patients. Many patients had more than 1 condition. Allocation to the prophylaxis group was according to the physician's criteria at each center. No differences were observed between the 2 groups for risk factors or type of prophylaxis administered, except for RR therapy, which was more frequent in group 1 (44.7% versus 20.4%; P = 0.01). Fluconazole and caspofungin were administered homogeneously throughout the study period (2005-2012).

Table 2 describes the fungi documented in all cases of IFI, the risk factors, the time to starting prophylaxis, the time to IFI (from day 0), the type of prophylaxis, and overall mortality during follow-up. Three of 5 cases of Candidemia were catheter-related. Intra-abdominal infection was the source of invasive candidiasis in 5 patients (1 with Candidemia). Five out of 6 cases of IA were pulmonary and the other cerebral. In pulmonary aspergillosis (n = 5), bronchoscopy and bronchoalveolar lavage fluid (BALF) confirmed A. fumigatus in culture in all cases, and BALF galactomannan was positive in all. According to European Organization for Research and Treatment of Cancer criteria,17 IA was “probable” in all pulmonary aspergillosis cases. The patient with cerebral aspergillosis was confirmed by culture (A. fumigatus) of cerebral biopsy (“proven” IA). Finally, 2 cases of cutaneous zygomycosis (necrotizing wound infection) were documented in 1 patient during prophylaxis with fluconazole and in a retransplanted patient 70 days after stopping prophylaxis with caspofungin.

Table 3 shows the risk of global IFI (until day 180), breakthrough IFI (during prophylaxis administration), IA, need for antifungal therapy, and death. IFI was documented in 17 (8.7%) patients, 11 (5.6%) of whom became infected during administration of prophylaxis (breakthrough IFI). Although the frequency of global IFI and IA was lower in patients receiving caspofungin, the differences were not significant in the univariate analysis (5.2% versus 12.2%; P = 0.12; and 1% versus 5.1%; P = 0.19, respectively). However, the difference was significant in the case of breakthrough IFI (2.1% versus 9.2%; P = 0.04). Overall mortality was 19.5%, with no differences between the groups (23.7% versus 16.3%, P = 0.21). Mortality attributed to IFI was estimated at 7.5%, with no differences between prophylaxis with caspofungin and prophylaxis with fluconazole.

An analysis focused on patients receiving dialysis (the only baseline condition for which significant differences were detected between both types of prophylaxis) revealed similar results (Table 3). A lower rate of all infections was demonstrated in patients receiving caspofungin, and in the case of breakthrough IFI, this difference was statically significant (0% versus 15%; P = 0.03). In these patients, a significantly higher number of patients receiving prophylaxis with fluconazole required empirical antifungal therapy (47.4% versus 19.0%; P = 0.03).

A propensity score was used to evaluate the impact of the type of prophylaxis on the development of global IFI, breakthrough IFI, and IA in the absence of the influence of confounding variables (Table 4). The propensity score analysis revealed a significant reduction in the absolute risk of IA in patients receiving caspofungin (risk reduction, 0.06; 95% confidence interval [CI], 0.001-0.11; P = 0.044). With respect to the other 2 objectives, the propensity score analysis did not confirm a significant result, although a trend toward reduced frequency of breakthrough IFI was observed (P = 0.079).

To evaluate the safety of the different types of prophylaxis, we compared the means of blood parameters at day 0 and day + 14 of prophylaxis (Table 5). Significant variations were found according to the type of prophylaxis and baseline conditions. A linear regression analysis adjusted for the variables included in the propensity score confirmed a significantly higher decrease in blood levels of ALT in patients receiving caspofungin (group 1; 455.6 IU/mL; 95% CI, 128.6-782.6; P = 0.007). Similarly, a trend was observed in levels of AST (537.9 IU/mL; 95% CI, –71.5 to 1147.3; P = 0.078). However, a significantly higher increase in blood levels of bilirubin was confirmed in patients receiving caspofungin (group 1; –4.2 mg/dL; 95% CI, –6.8 to –1.5; P = 0.002; Table 5). No other significant differences indicative of hepatic or renal toxicity were observed for the remaining parameters. No significant differences were observed in the mean values of blood parameters at day + 90 and day +180.

Discussion

In the absence of antifungal prophylaxis, invasive mycosis typically affects 36%-50% of HR-LTRs.6, 7, 11, 12, 21 The meta-analysis by Cruciani et al.22 confirmed that universal prophylaxis (mainly fluconazole) reduced colonization, total proven fungal infections, and mortality attributable to fungal infection. In addition, clinical practice guidelines recommend fluconazole for prevention of IFI in LTRs.23, 24 However, this strategy raises several doubts. In the absence of risk factors, the frequency of IFI is <4%.8, 25 Universal antifungal prophylaxis with fluconazole for LTRs has been associated with increased hepatotoxicity and interactions with immunosuppressive drugs.8, 9 Additionally, the emergence of nonalbicans Candida spp. and increasing rates of resistance have limited the use of fluconazole.4 A recent retrospective study at the University of Pittsburgh confirmed that, compared with universal prophylaxis, targeted prophylaxis is effective, feasible, and safe.26 Therefore, in many transplant groups, prophylaxis targeting Candida spp. and Aspergillus spp. is based on lipid formulations of AmB or echinocandins. The use of lipid formulations of AmB for targeted prophylaxis has been widely evaluated.11, 12, 27, 28 One potential concern with AmB is impaired recovery of renal function, because renal failure, especially that requiring dialysis, is a relevant risk factor for IFI in LTRs. A recent meta-analysis that evaluated randomized controlled trials—echinocandins were not included—and compared regimens for antifungal prophylaxis in liver recipients confirmed that fluconazole or L-AmB reduces the incidence of IFI.29 The evidence provided by the meta-analysis suggests that the 2 are equally efficacious when compared with placebo.

Echinocandins are active against both Aspergillus spp. and Candida spp., including fluconazole-resistant strains. A recent review of clinical practice in the United States confirmed that 72% of centers apply a targeted strategy in LTRs, and the leading choice of mold-active agents for targeted prophylaxis was echinocandins.9 Winston et al.14 recently reported the results of a clinical trial comparing anidulafungin and fluconazole. Although the rate of IFI was similar in both groups, a lower risk of breakthrough IFI, including IA, was observed in patients receiving prophylaxis with anidulafungin who had a Model for End-Stage Liver Disease (MELD) score of > 30, required RR therapy, were given >15 units of packed red blood cells during transplant surgery, or had received fluconazole before transplantation. In the present study, no significant differences were observed in these subgroups, with the exception of patients requiring dialysis who presented a significant reduction of breakthrough IFI and antifungal therapy and a trend in IA in patients receiving caspofungin in comparison to fluconazole (Table 3). In the international, multicenter TENPIN study, which compared micafungin with the standard of care in HR-LTRs (fluconazole, caspofungin, or liposomal AmB), the per protocol analysis did not confirm differences in the incidence of IFI or IA, although kidney function was better with micafungin.15 Unlike the aforementioned studies, which were performed in a clinical trial setting, the present study analyzes the efficacy of 2 different antifungal prophylaxis regimens in clinical practice. We detected higher rates of IFI and overall mortality (8.7% and 19.5%, respectively) than Winston et al.14 and Saliba et al.15

Consistent with the results of other studies, we found no differences between caspofungin and fluconazole in the prevention of global IFI in HR-LTRs. Nevertheless, the propensity score analysis showed a significantly reduced frequency of IA in patients receiving caspofungin and, consistent with the findings of Winston et al.,14 a trend toward reduced frequency of breakthrough IFI. This benefit is probably more evident in patients on dialysis, although the low number of events and patients on dialysis precluded a more intensive analysis.

The superiority of echinocandins over fluconazole in breakthrough infections stems mainly from their ability to control premature fungal infection rather than from the risk of resistance to azoles. In the present study, only 8.2% of transplant recipients had received prior antifungal therapy, including fluconazole. Only 1 of the 9 Candida spp. infections was produced by C. glabrata (minimum inhibitory concentration [MIC] fluconazole: 16 µg/mL, European Committee on Antimicrobial Susceptibility Testing (EUCAST)); the remaining infections were produced by fluconazole-sensitive strains. The increased activity of echinocandins with respect to fluconazole against IA and the rate of early IA in this series (5 of 6 IA occurred before day 36) may also have contributed to these results. Finally, clearance of echinocandins by dialysis is poor. Fluconazole, on the other hand, is easily eliminated by dialysis and is therefore not a good option for prophylaxis in these patients.

The large number of confounders makes it difficult to perform a safety analysis in such a complicated setting as the immediate posttransplant period. Nevertheless, our safety analysis showed a significant decrease in transaminase levels (ALT and AST) but a significant increase in bilirubin levels in patients who received prophylaxis with caspofungin. Increased bilirubin levels have been observed in other studies with caspofungin, even in healthy volunteers.30 Despite the initial recommendation to avoid caspofungin in advanced liver disease, the findings of the present study show a significant decrease in transaminase levels in patients receiving caspofungin. Other studies performed in patients with liver disease confirm the hepatic safety of caspofungin.10, 16, 31-33

The present study is subject to a number of limitations. First, it is not a clinical trial, and selection of prophylaxis regimens could be conditioned by various circumstances in a specific case or center. In this sense, the propensity score analysis may help to calibrate the impact of other variables on the selection of a specific prophylaxis regimen. We used a propensity score analysis because the number of events per confounder was very low. Other authors have confirmed that the propensity score produces estimates that are less biased, more robust, and more precise than the logistic regression estimates in the presence of ≤7 events per confounder.34 Second, in order to simplify the analysis, we did not evaluate the role of anidulafungin and micafungin, which are widely used echinocandins. Third, the study period was 8 years, during which time protocols changed and therapeutic innovations were incorporated at the centers. Finally, although the number of LTRs included represents only 5.1% of all liver transplantations performed in the 9 centers during the study period, the estimated number of HR-LTRs would probably be close to 10%-20%.8 This difference could be explained by the exclusion of other prophylactic drugs (eg, lipid AmB, anidulafungin, and micafungin) or the lack of prophylaxis in some HR-LTRs.

In summary, despite the limitations cited above, the results of our observational study show caspofungin and fluconazole to be equally efficacious in the prevention of global IFI in HR-LTRs. However, caspofungin could be associated with a lower rate of IA and breakthrough IFI. This benefit is confirmed even after adjusting for confounders and is probably more favorable in patients on dialysis. Caspofungin is safe in HR-LTRs, although bilirubin levels may be increased.