Efficacy of azole therapy for tegumentary leishmaniasis: A systematic review and meta-analysis

Background Several controlled and uncontrolled studies addressing azole antifungal drugs for cutaneous and mucosal leishmaniasis have been published with inconclusive results. We conducted a systematic literature review of studies evaluating the efficacy and toxicity associated with azole therapy for tegumentary leishmaniasis. Methodology PRISMA guidelines for systematic reviews and the Cochrane manual were followed, and the review methodology was registered (PROSPERO; CRD42016048668). Sources included the EMBASE, Web of Science, MEDLINE, LILACS, and IBECS databases along with a manual search of references from evaluated studies. Additional resources such as Google Scholar and clinicaltrials.gov were also searched. We included all studies reporting cure rate after cutaneous or mucosal leishmaniasis treatment with systemic azole drugs, regardless of their design. R software was used to estimate global rates of success and adverse events with each drug. The main outcome of interest was clinical cure, defined as complete re-epithelialization of all lesions. Results A total of 37 studies involving 1259 patients that reported outcomes after fluconazole (9), ketoconazole (14) and itraconazole (15) treatments were included. Only 14 (38%) were randomized controlled trials (RCT). The pooled azole final efficacy rate was 64% (CI95%: 57–70%) for all studies and 60% (CI95%: 50–70%) (p = 0.41) if only RCTs studies were considered. Twenty-four studies were conducted in the Old World and 13 studies in the Americas. The final efficacy rate according to New and Old World were 62% (CI95%: 43–77%) and 66% (CI95%: 58–73%), respectively. The final efficacy rate of azoles according to species were 89% (CI95%: 50–98%) for L. mexicana; 88% for L. infantum (CI95%: 27–99%); 80% for L. donovani; 53% (CI95%: 29–76%) for L. major; 49% for L. braziliensis (CI95%: 21–78%); and 15% (CI95%: 1–84%) for L. tropica. The cure rates were similar among the fluconazole, ketoconazole and itraconazole group arms (p = 0.89), specifically 61% (CI95%: 48–72%), 64% (CI95%: 44–80%) 65% (CI95%: 56–72%), respectively. Adverse events during fluconazole, itraconazole and ketoconazole therapy were reported in 7% (CI95%: 3–14%), 12% (CI95% 8–19%) and 13% (CI95%: 6–29%) of treated patients, respectively, without difference among them (p = 0.35). This systematic review included studies with small samples and both non-comparative and non-randomized studies and the main limitation was the low quality of the available studies. Conclusions Available evidence suggests that fluconazole, ketoconazole and itraconazole have similar and modest efficacy rates for tegumentary leishmaniasis treatment. There is insufficient evidence to support the exclusive use of azole therapy as a single agent for leishmaniasis treatment.


Introduction
Tegumentary leishmaniasis (TL), comprising cutaneous (CL) and mucosal leishmaniasis (ML), is a parasitic infection caused by protozoa that annually affects 0.7 to 1.2 million people worldwide [1]. Although TL is a non-fatal disease, it is characterized by a broad clinical spectrum involving single or multiple localized skin lesions, severe diffuse and mucosal lesions. TL leads to considerable morbidities and is associated with physical deformities and psychological effects [2][3][4]. The control of leishmaniasis remains a serious problem, and the few, and often toxic, therapeutic options available are the primary challenge for the disease approach.
There are few drugs available for TL treatment, and some systematic reviews [5][6][7][8][9] have tried to achieve a consensus about an optimal drug treatment for patients using different interventions. Unfortunately, no ideal or universally applicable therapy for leishmaniasis has been identified. The evidence to support the efficacy of different treatments for CL in the Old World is still very limited. In this context, systemic pentavalent antimony (SbV) has remained the first-line treatment for leishmaniasis for decades, achieving a cure rate of 76.5% for American CL [6].
Although SbV is the gold standard therapy, pentavalent antimony cannot be considered a satisfactory option because it requires a daily dosage of injections for 20 to 30 days, it cannot be used in pregnant women, and it has led to severe side effects such as cardiotoxocity and renal failure in several patients [10,11]. Thus, better treatment alternatives are urgently needed.
There is interest in orally administrable antileishmanial agents, and azole therapy comprising fluconazole, ketoconazole and itraconazole may be an alternative that fulfills this requirement. These agents have been shown to be highly efficacious against Leishmania spp. in in vitro tests [12][13][14]. Some in vivo studies also reported an effective response to this medication group [15][16][17]. However, the efficacy rate of azole therapy, to the best of our knowledge, has never been compiled.
The aim of this study was to systematically assess, through an evidence-based approach, the efficacy and safety of azole therapy for TL.

Objectives
Our main objective was to assess the efficacy of azole therapy for TL, following the PICO question: Population: subjects with TL; Intervention: use of any systemic azole drug; Comparison, if applicable: any other therapy, placebo or no treatment; Outcome: cure rate. Secondary objectives were to assess the adverse events reported with azole treatment for TL, to verify whether responses to azole therapy are dose-dependent, species-dependent or associated with the disease geographical distribution (New and Old World), and to determine the rates of relapse and late mucosal involvement after treatment.

Materials and methods
The review methodology was registered in the International Prospective Register of Systematic Reviews (PROSPERO; CRD42016048668; http://www.crd.york.ac.uk/PROSPERO/display_ record.asp?ID=CRD42016048668), and recommendations of the Cochrane Handbook [18] and of the PRISMA statement [19] were followed. Structured searches were conducted independently by two reviewers (ELG and GFC) in EMBASE, Web of Science, PubMed (MED-LINE) and VHL (LILACS and IBECS) using a comprehensive list of key terms that were adapted to each database through October 2016 (updated July 2017). The initial search was complemented by a manual search of reference lists from retrieved articles. Furthermore, a search of gray literature (Google Scholar) and in clinicaltrials.gov were performed for identifying potential ongoing studies. The detailed search strategies are described in S1 File.
There were no restrictions on the publication language, date of publication or study design. The papers were included if cure rate after systemic fluconazole, itraconazole or ketoconazole therapy for TL were reported. Studies involving non-human participants, studies with less than ten patients in the azole therapy arm, and those addressing azole topical treatment or azole therapy combined with another anti-Leishmania active drug were excluded. Furthermore, review articles and letters to the editor were also excluded. Concomitant antibiotic use was not considered exclusion criteria. All studies matching the inclusion criteria were reviewed by the authors, and disagreements on inclusion were resolved by consensus.
After analysis of title and abstract, the selected studies were read in full to confirm their eligibility and to extract the data. The following information was recorded: country; year of publication, design of the study, predominant Leishmania species, therapeutic schedule, and outcomes using a standardized data collection form. The outcome of interest was clinical cure, defined as complete ulcer healing, and was assessed at three time points counted from the first day of the treatment: (1) "initial response," assessed at 30-73 days; (2) "initial cure," assessed at 74-100 days; and (3) "definitive cure," assessed at 101-194 days. The intervals for the cure evaluation were an adaptation of the current recommendations for CL trials [20] to include the outcomes reported in the original studies. The final efficacy rate analysis considered the last cure rate available in the studies within six months. Relapse was assessed only for patients who were treated and were considered cured.
The quality of the randomized studies was evaluated using the following criteria: 1) doubleblind; 2) concealment of treatment allocation; 3) blinding of outcome assessment; and 4) intention-to-treat analysis. Concealment of treatment allocation was considered adequate if the patients and enrolling investigators could not predict assignment. Outcome assessment was blinded if the investigator who assessed the outcome had no knowledge of treatment assignment. The analysis was performed according to the intention-to-treat principle if all randomized patients were included in the analysis and maintained in their originally assigned groups. The Newcastle-Ottawa Scale (NOS) [21] was used to assess the quality of nonrandomized studies. On this scale, the studies were measured considering three dimensions: 1) selection of study groups, 2) comparability of groups, and 3) determination of the results of interest. For randomized and non-randomized studies, it was assumed inadequate if there was not enough information to assess the quality.

Quantitative data synthesis
Data analyses were performed using R software with "meta" and "metafor" packages, except for the meta-regression analyses that were conducted using Comprehensive Meta-Analysis, version 3. Two types of statistical analysis were performed: meta-analysis of main effects (for each one of the azole drugs in three time points) and subgroup analysis to test for effect modification by the one categorical covariate (cure rate according to therapy schedule, to Leishmania species, to geographical localization and adverse events). The threshold for statistical significance was 0.05. Direct comparisons between azoles and another comparator arm were performed when available. Forest plots are presented to illustrate the effects of global estimation and sub-analyses on meta-analysis results. The inconsistency (I 2 ) statistic was used to evaluate heterogeneity. A random model was chosen for all analyses due to considerable heterogeneity across studies.
Considering that the association between effect and dose is evidence of the action of a drug, with the intention to identify an ideal dosage regimen, we attempted to analyze efficacy according to the azole dose used. The daily intake of azoles in milligrams was categorized as low ( 200 mg), intermediate (201 to 400 mg) and high (> 400 mg) dosage. A further analysis was performed considering doses categorized as less than 400 mg or equal to or greater than 400 mg. A meta-regression to determine whether the dosage could explain the variation in effect size values between studies was also performed. Studies that did not present a defined dosage were excluded from these analyses. Publication bias was assessed by observing the symmetry of funnel plots and the Begg's adjusted rank correlation test [22].

Results
Our search identified 33 articles from EMBASE, Web of Science, PubMed and VHL databases. A manual search using a reference list of selected papers added two additional papers. Finally, through Google Scholar, a potentially relevant thesis and one article were also identified. An   [40,42,49]. The process of study selection and reasons for exclusion are summarized in a PRISMA flow diagram (Fig 1). After numerous attempts, we were unable to obtain the Saleem et al. (2007) study in its entirety [47]. Thus, for this study, only the data available in the abstract were used, so these results were included in the meta-analysis but not in all tables that detailed the characteristics of the studies.
Among the 37 studies included, 18 were prospective non-randomized studies, 5 were retrospective studies and 14 were randomized clinical trials (RCT). Of these, 24 studies were conducted in the Old World and 13 in the New World. A study depicting the evolution of American soldiers infected during the Iraq war was allocated along with Old World studies due to the Leishmania species involved (L. major) [44].
The cure criteria defined by the authors varied widely among studies, and eight studies did not report the criteria used [23][24][25][26]44,50,53,55]. Most studies defined cure as complete re-epithelialization of all lesions. However, in one study, cure was defined by "Leishmania donovani bodies negativity" [29], while another considered "reduction in the size of lesions by 80% up to complete clearance" [26], and a third considered "more than 90% re-epithelization and negative smear for Leishmania parasites" [33]. Systemic therapeutic regimens with azoles   Tables 1 and 2. The population we gathered included young adults, and only one study included children exclusively [46]. In general, the mean number of lesions per patient with regard to men: women ratio and time of disease were higher in Old World studies compared to those in the New World (Tables 3 and 4). The mean follow-up time was relatively short (maximum 24 months), and the late mucosal involvement rate was not evaluated in studies.
The concomitant use of antibacterial therapy was mentioned only by two studies [17,55] intending to treat lesions with secondary bacterial infection. The azole therapeutic regimens varied significantly among the studies. The daily dose ranged from 100 to 800 mg, and the treatment duration ranged from 14 to 84 days. The mean treatment length was similar in Old (40.3 days) and New World studies (45.5 days). The therapeutic schedules are presented in Tables 5 and 6. Only nine studies (24%) reported Leishmania species characterization [28,29,32,34,42,45,52,53,55]. L. braziliensis was the species more prevalent in the Americas, while L. tropica and L. major were usually reported in the Old World. The summarized cure rates according to the intention-to-treat analysis and adverse events rates are shown in Tables 7 and 8.
The relapse rate after cure was reported in only 43% of studies. Among 282 patients cured in these studies, 10 (4%) relapsed. Only two studies (5.4%) did not mention the occurrence of side effects. However, in general, side effects were poorly described. Many authors reported only the absence of serious events leading to treatment interruption. Adverse events that were reported as reasons to discontinuation of azole therapy were increased liver enzymes, epigastric pain, nausea and vomiting, increased creatinine, headaches, skin rash, and jaundice. An overview of adverse events reported by the authors of included studies are reported in S1 and S2 Tables.

Methodological quality
According to the score system adopted for RCT quality assessment (S3 Table), the most compromised domain was related to blindness, and the least compromised was related to intention-to-treat analysis. The methodological quality assessment of the 22 non-randomized studies using Newcastle-Ottawa Scale is presented in S4 Table. Overall, studies presented low methodological quality, and none of the studies obtained the maximum score corresponding to nine stars. The risk of bias across studies for a given outcome is presented in S5 Table. No evidence of publication bias was detected in the meta-analysis of initial response for fluconazole (z = 0.29, p = 0.79), ketoconazole (z = -0.45, p = 0.65) and itraconazole (z = -0.05, p = 0.96) or for meta-analysis of final efficacy rate for these same azoles, namely, fluconazole (z = 0.27, p = 0.79), ketoconazole (z = 0.33, p = 0.74) and itraconazole (z = -0.75, p = 0.45). For final efficacy rate from meta-analysis for Old World and New World, there was also no indication of publication bias (z = 0.02, p = 0.98 and z = 0.85, p = 0.39, respectively). Publication bias  analyses were not performed for the other comparisons because of the small number of studies. Visual inspections of funnel plots did not show any substantial asymmetry (S1, S2 and S3 Figs).

Global analysis
In a global analysis including studies addressing all azole medicines, regardless of their origin, the initial response rate was 53% (CI95%: 43-62%, I 2 = 85.3%) and the final efficacy rate was 64% (CI95%: 57-70%, I 2 = 78%). If only RCT studies were considered, final efficacy rate, the only outcome that could be calculated, was 60% (CI95%: 50-70%, I 2 = 82%), which means no difference between RCT and non-RCT studies (p = 0.41). No significant difference was observed in the final efficacy rate (of all azoles and each azole group) of studies shared according to the methodological quality. There were three studies exclusively involving patients with mucosal leishmaniasis. All of them were conducted in the New World and addressed itraconazole therapy. The final efficacy rate observed for these 38 patients was 52% (CI95%: 23-79%, I 2 = 69.5%).
In general, the studies compared different interventions; thus, few direct comparisons were possible. Final efficacy rate using itraconazole was significantly higher compared to a placebo group in studies conducted in the Old World (OR = 3.07, CI95%: 1.2-7.8) (Fig 5).

Discussion
No previous systematic review have specifically addressed our research questions. The present study, which was based on a total population of 1259 patients, identified an overall efficacy rate of 64% (95% CI: 57-70%) in the treatment of TL with azole drugs. This compiled efficacy rate required a critical interpretation by considering all analyzes of subgroups we have done in order to overcome the limitations imposed by the high heterogeneity between studies context. Taking this important observation into account, it is possible to draw a parallel between this result and the efficacy obtained using the gold standard drug in the treatment of the disease: antimony derivatives. The efficacy rate of this first line therapy is approximately 76.5%, according to studies gathered in a systematic review for the New World [6]. Corroborating the same interpretation, an analysis of the two studies, that directly compared antimony derivatives with azoles, specifically with fluconazole, confirms the superior efficacy of the former in the treatment of American tegumentary leishmaniasis (OR = 9.33), despite the heterogeneity across the studies and the large confidence interval found.
Even with a cure rate no higher than 80% in large studies, antimony derivatives continue to be the first-line treatment recommended for all forms of leishmaniasis in many countries. In addition to a suboptimal efficacy, antimony treatment has additional disadvantages including serious adverse effects [10] and the need for long-term parenteral use and laboratory and cardiac monitoring during treatment. Therefore, the identification of new therapeutic approaches for leishmaniasis is considered a priority, and oral medication emerges as an attractive option. In addition to miltefosine, which is already included in recommendations for the treatment of TL [11], allopurinol, macrolides and various azoles have been described for the treatment of the disease. With use first proposed in the 1980s [43], fluconazole, ketoconazole and itraconazole have since been indicated for the treatment of TL in many, generally small studies of poor methodological quality [15][16][17]. In addition, many case reports and series with fewer than ten patients that explored the efficacy and safety of this class of medications have also been published [57][58][59][60][61][62]. In this context, most of the studies included in this systematic review were not randomized. Nonetheless, our results reveal no significant differences between the cure rates found in the randomized and non-randomized studies.
The anti-leishmanial effects of azole antifungal agents are associated with the inhibition of cytochrome P-450 mediated 14α-demethylation of lanosterol in fungi, which blocks ergosterol synthesis and causes the accumulation of 14α-methyl sterols. The inhibition of sterol biosynthesis is associated with the inhibition of leishmaniasis growth [12]. In vitro studies addressing the effects of azoles on sterol biosynthesis have revealed that, for most Leishmania strains, itraconazole was slightly more inhibitory than ketoconazole, and fluconazole was much less inhibitory than the other azoles [13]. In this review, it was not possible to observe differences in efficacy among the different azoles, namely, itraconazole (65%), ketoconazole (64%) and fluconazole (61%). Similarly, the impact of concomitant use of antibacterial therapy cannot be assessed in this review due the small number of patients undergoing this therapy and the absence of a uniform criterion for its use.
Itraconazole was the azole chosen for treatment in all studies that exclusively evaluated patients with the mucosal form of leishmaniasis [41,43,50]. Its better bioavailability and ability to achieve higher tissue concentrations compared to the other azoles, including in the mucosa, may explain this choice [63][64][65]. This review identified a cure rate of 52% for mucosal leishmaniasis in the Americas. However, the rate of late mucosal involvement after treatment with azoles was not assessed by the studies reviewed here, which generally presented a short followup period. This is a relevant concern for assessment, particularly in species prevalent in the Americas [66].
In 2010, azole drugs were identified by WHO as a treatment option for L. major and L. mexicana [11]. Considering the insufficient number of studies addressing the cure rate according to the Leishmania species, additional evidence is necessary to draw conclusions on the efficacy of azoles against each species of Leishmania. In contrast, our results do not reveal a difference in the final efficacy rate with azole therapy between the Old and New Worlds, which would be expected because of the different efficacy rates reported for the different Leishmania species [67,68]. Specifically, the final efficacy rate for L. braziliensis, a predominant species in the Americas, was 49%, based on an analysis involving only 138 patients. Based on only one or two studies with efficacy data for each of the other Leishmania species, the cure rate ranged from 15% for patients with L. tropica to 89% for L. mexicana. These observations reinforce that the actual efficacy of azoles treatment still needs to be determined and that the gathering of non-comparative and methodologically fragile studies is an imperfect strategy for understanding the usefulness of this class of drugs. Among the studies included in this review, only one reported exclusively on the cure rate in children [46], preventing an overall summary analysis of efficacy in this age group. Future clinical trials should be conducted considering the therapeutic response related to the clinical presentation of the disease and the host response according to age to support more specific recommendations by patient subgroup.
In general, adverse events were not observed systemically and were not reported in a standardized way by the authors. Whereas some studies counted the number of events, others reported the rate of patients affected by these events. Based on most of the available data, the overall rate of individuals affected by adverse events due to azoles in this review was 11%. Despite this attempt to synthesize the rate of adverse events, our analysis included studies that did not measure this outcome or used different drugs (fluconazole, itraconazole and ketoconazole) and varied dosing regimens; thus, this result should be interpreted with caution.
The dose-effect association observed with fluconazole can be understood, in addition to evidence of the action of the drug, as an indication that the full efficacy may not yet have been achieved with the doses assessed. In summary, our results for the pooled overall efficacy rate of azoles reveal the fragility of this analysis strategy but, on the other hand, although modest, are corroborated by the direct comparison performed between itraconazole versus placebo (OR = 3.07, 375 patients), what could be taken as an indicative of activity of this class of drugs in the treatment of TL. The identified activity and dose-effect association reinforces the need for further studies evaluating the efficacy and safety of azoles at higher doses, particularly in the New World, where the rate of spontaneous cure is low [68] and the available data suggest a reasonable efficacy against L. mexicana.
The main limitation of this review is the low quality of the available studies. Many studies were conducted at a time prior to the standardization of criteria for clinical studies [69], when the requirements for the performance and reporting of methods were less strict. This systematic review included studies with small samples and both non-comparative and non-randomized studies; moreover, only four studies performed a sample size calculation [17,36,43,54]. The heterogeneity found here is evident in the various treatment schemes with the azoles and Leishmania species involved. Assessing the overall risk of bias for a given outcome across studies, a high risk of bias was observed in the comparison of efficacy between the different azoles, mainly due to differences in the cure criteria adopted by the groups. Related to the same domain (cure criteria), a moderate risk of bias was observed in the comparison between endemic regions across groups. For comparison of treatment efficacy according to Leishmania species, the main limitation is the small number of studies gathered. All freely available databases were evaluated covering a period of 30 years (1986-2016). In addition, studies with different designs were included, which made this review comprehensive and useful in revealing the lack of evidence required to support a therapeutic recommendation. On the other hand, there was no evidence to suggest publication bias, as estimated by the funnel plots. These results should not be understood as evidence to support TL management recommendations but as relevant information to guide future clinical studies. Trials designed in accordance with the current criteria and outcome standardization for CL [20] evaluating higher doses of the different azoles in different patient subgroups in different regions of the world, including in combination, may in the future base recommendations for its incorporation into the limited arsenal of alternatives for the treatment of leishmaniasis. At present, the quality of the available data does not allow to conclude on the benefit of the azole therapy for the treatment of the tegumentary form of leishmaniasis.