Amphotericin B for Treatment of Visceral Leishmaniasis: Systematic Review and Meta-Analysis of Prospective Comparative Clinical Studies Including Dose-Ranging Studies

Clinical Microbiology and Infection 24 (2018) 591e598

Contents lists available at ScienceDirect

Clinical Microbiology and Infection

journal homepage: www.clinicalmicrobiologyandinfection.com

Systematic review Amphotericin B for treatment of visceral leishmaniasis: systematic review and meta-analysis of prospective comparative clinical studies including dose-ranging studies

* C. Rodrigo 1, , P. Weeratunga 2, S.D. Fernando 3, S. Rajapakse 2

1) Department of Pathology, School of Medical Sciences, Faculty of Medicine, University of New South Wales, Sydney, Australia 2) Department of Clinical Medicine, Faculty of Medicine, University of Colombo, Sri Lanka 3) Department of Parasitology, Faculty of Medicine, University of Colombo, Sri Lanka article info abstract

Article history: Objectives: To evaluate the evidence for use of different formulations of amphotericin B (AmB), minimum Received 4 September 2017 effective dose for each formulation and its comparative efficacy against other drugs in achieving Received in revised form definitive cure of visceral leishmaniasis. 2 November 2017 Methods: This systematic review and meta-analysis included following data sources: PubMed, Embase, Accepted 5 November 2017 Scopus, Web of Science and CINAHL. Controlled prospective clinical trials (randomized or non- Available online 11 November 2017 randomized, including dose-ranging studies) conducted between 1996 and 2017 with at least one Editor: L Leibovici treatment group receiving AmB were included (published data only). The primary outcome was definitive cure at 6 months. Adverse events and mortality were assessed as secondary outcomes. The Keywords: PROSPERO registration number for this review is CRD42017067488. Amphotericin B Results: Thirty-one studies (26 from India) that enrolled 6903 patients into 84 study groups met the Clinical trials selection criteria. In India, liposomal AmB was not inferior to AmB deoxycholate (relative risk 1.00, 95% Liposomal amphotericin confidence interval (CI) 0.96e1.03, two randomized controlled trials (RCTs), 514 participants, high- Systematic review quality evidence), and a single dose of the earlier formulation as low as 3.75 mg/kg achieved a cure Visceral leishmaniasis rate of over 89% (95% CI 70.6e97.2). AmB deoxycholate was as effective as miltefosine (relative risk 0.99, 95% CI 0.95e1.03, two trials, 523 participants, high-quality evidence) and may be better than paromomycin (relative risk 1.04, 95% CI 1.02e1.07, one trial, 667 participants, low-quality evidence) in achieving definitive cure. Conclusions: AmB is an efficacious drug in the Indian subcontinent. Further evidence is needed from prospective clinical trials in other endemic geographical regions. C. Rodrigo, Clin Microbiol Infect 2018;24:591 © 2017 European Society of Clinical Microbiology and Infectious Diseases. Published by Elsevier Ltd. All rights reserved.

Introduction for many decades. However, these drugs have inherent toxicity, and on top of that, parasitic resistance is now frequently encountered in Visceral leishmaniasis (VL), which is caused by the unicellular India [2]. Amphotericin B (AmB) is a macrolide polyene antifungal parasite Leishmania donovani, is potentially fatal if untreated [1]. agent that has become the preferred choice in areas with antimo- Over 90% of VL cases in the world are reported from six coun- nial resistance. triesdBrazil, Ethiopia, Somalia, South Sudan, Sudan and Indiad AmB is not soluble in saline and has to be mixed with a deter- with approximately 200 million residents at risk of the infection. gent, sodium deoxycholate. Soon after entry into plasma, the drug Pentavalent antimonial compounds were the ﬁrst line of therapy dissociates from the detergent and binds to circulating lipoproteins [3]. The lipid-soluble property of AmB has been used to develop several lipid-soluble variants that have less toxicity and a better

* Corresponding author. C. Rodrigo, Room 207, Level 2W, Wallace Wurth Building, therapeutic index. These preparations are liposomal AmB (L-AmB), School of Medical Sciences, UNSW, NSW, Sydney, 2052, Australia. AmB lipid complex (ABLC) and AmB emulsion (ABLE) [3]. However, E-mail address: [email protected] (C. Rodrigo). https://doi.org/10.1016/j.cmi.2017.11.008 1198-743X/© 2017 European Society of Clinical Microbiology and Infectious Diseases. Published by Elsevier Ltd. All rights reserved. 592 C. Rodrigo et al. / Clinical Microbiology and Infection 24 (2018) 591e598 it must be noted that any of these formulations are not homoge- that included any formulation of AmB by itself or in combination nous entities and the efficacies may vary depending on the with another drug in any one of the trial arms. Case series, obser- manufacturer. vational studies and retrospective analyses were excluded. Given AmB's increasing importance as a preferred therapeutic option over the last two decades, the objectives of this systematic Information sources and search strategy review and meta-analysis were to evaluate the evidence for use of different formulations of AmB, to discover the minimum effective PubMed, Embase, Scopus, Web of Science and CINAHL were dose for each formulation and to assess its comparative efficacy searched for relevant articles with the keywords ‘leishmania*’ in against other drugs in achieving definitive cure of VL. any field and ‘amphotericin’ in any field without any language re- strictions. Articles published over the last 20 years (1 January Methods 1996e20 May 2017) in all languages were included. Only published data were considered. We used Endnote X7 software (Thomson Protocol and registration Reuters, Carlsbad, CA, USA) to filter articles. The flow diagram for article selection is shown in Fig. 1. The PROSPERO registration number for this review is CRD42017067488. The review protocol is available online in Study selection Supplementary File S1. Two authors (CR and PW) performed the search and coded the Eligibility criteria studies according to the inclusion criteria. After reading all ab- stracts, key articles were identified by consensus. Full articles were We performed a systematic review and meta-analysis of all obtained for all studies meeting the inclusion criteria (including published prospective clinical trials (randomized or non- articles of undetermined status) for further assessment. Bibliogra- randomized, including dose-ranging studies) on treatment of VL phies of selected articles were also searched to identify relevant studies. The final list of included studies had the concurrence of all authors.

Data collection and data items

The data items extracted from each study included participant demographics, intervention and control groups, drug doses, locality and dates of study, immediate and definitive cure rates, adverse events, relapses during follow-up and mortality statistics. The primary outcome in the analysis was definitive cure, defined as patients being free of parasites as demonstrated by splenic, lymph node or bone marrow aspiration at least 6 months after completion of therapy. Adverse events and mortality were analysed as secondary outcomes. All studies were assessed according to an intention-to-treat model.

Risk of bias

Risks of bias were assessed by the Cochrane risk of bias assessment tool [4].

Summary measures and synthesis of results

Comparable trials with regard to the intervention and the outcomes were combined in a meta-analysis using Review Manager 5 software [4]. Dichotomous data were compared by relative risk (RR) and 95% conﬁdence intervals (CI). Heterogeneity was assessed with the I2 statistic and visual inspection of forest plots [5]. Heteroge- neity was also minimized by combining studies that only assessed comparable interventions and outcomes in the meta-analysis. A ﬁxed-effect model was used for the analysis, which was converted to a random-effect model if a high level of heterogeneity was encountered (I2 statistic >70%) [5]. Quality of evidence was assessed by GRADEPro software (McMaster University, Hamilton, Canada). Funnel plots to assess publication bias could not be created because there were only a few studies per comparison. This research was not funded.

Results

Thirty-one studies that enrolled 6903 patients in 84 study Fig. 1. PRISMA ﬂowchart of study selection process. groups met the inclusion criteria (Fig. 1). All studies except ﬁve C. Rodrigo et al. / Clinical Microbiology and Infection 24 (2018) 591e598 593

[6e10] were conducted in India, and two studies included HIV- coinfected patients [7,8]. All trials excluded pregnant and breast- feeding women. The findings of these studies are discussed in the following sections according to the main themes of comparisons. The risk of bias in each of the studies is highlighted in Fig. 2. All trials diagnosed VL by identifying parasites in splenic, lymph node or bone marrow aspirates, and all confirmed immediate and definitive cure after the completion of treatment and at 6 months (unless stated otherwise), respectively, by the same method. Any new parasitaemia observed between immediate cure and definitive cure was considered a relapse. A summary of dosing regimens, comparisons, and percentages of immediate and definitive cure (±95% confidence intervals), including the absolute numbers of patients, are provided in Tables 1e7. Characteristics of all included studies are provided online in Supplementary File 2.

Comparison of different formulations of AmB

Regarding the comparison of different formulations of AmB, four prospective studies (all performed in Bihar, India) including one nonrandomized study met the inclusion criteria (Table 1) [11e14]. L-AmB as a single dose (>10 mg/kg) was as effective as AmB deoxycholate 15 to 20 mg/kg (three trials) or as ABLC 10 mg/kg (one trial) [11,15]. A meta-analysis showed that 10 mg/kg of L-AmB was not inferior to AmB deoxycholate 15 mg/kg (two RCTs, 514 participants, high-quality evidence; Supplementary Fig. S1). Regarding adverse events, chills and fever were signiﬁcantly less with L-AmB compared to AmB deoxycholate (RR 0.44, 95% CI 0.21e0.92, two RCTs, 514 participants, moderate-quality evidence, p 0.03; Supplementary Fig. S2) [11,12]. One trial also reported that worsening anaemia and hypokalaemia occurred signiﬁcantly more often with AmB deoxycholate [12], and the second trial reported two deaths in AmB deoxycholateetreated patients. A single randomized open-label trial showed that 15 mg/kg of L- AmB was superior to the same dose of ABLE (RR 0.87, 95% CI 0.82e0.92, p 0.0002) [13]. Both groups reported similar rates of adverse events (mild), but there were two deaths in the ABLE- treated group [13].

Dose-ranging studies

For each formulation, multiple dose-ranging studies were available. For AmB deoxycholate (Table 2), a dose of 1 mg/kg per day for 15 to 20 days was superior to a dose of 0.75 mg/kg per day or 0.5 mg/kg per day (definitive cure of 99% (95/96) vs. 91% (87/96) and 82% (79/95), respectively, p <0.05) without any significant in- crease in adverse events, which mostly comprised renal impair- ment, drug hypersensitivity and gastrointestinal events [16]. There were multiple dose-ranging studies for L-AmB [9,10,12,17e23]; they were conducted in India, Kenya, Ethiopia, Sudan and Brazil (Table 3). The doses tested ranged from 3.5 mg/kg to 21 mg/kg administered as single or split doses. The largest dose administered as a single dose was 15 mg/kg [14]. There was a clear geographical bias in response rates, with patients in India having a definitive cure rate of 89% to 100% (Table 3) with doses as low as 3.75 to 6 mg/kg [17e19]. However, a locally manufactured (in India) version of L-AmB had to be administered at a dose of 10 mg/kg to achieve a definitive cure of 90% (Table 3) [20,22,23], and a meta- analysis showed that increasing the dose to 15 mg/kg of this version did not improve the outcome (two studies, 150 participants, moderate-quality evidence; Supplementary Fig. S3) [20,23]. Outside India, the success of L-AmB is less impressive. In Kenya, definitive cure of over 90% (13/14) was achieved with a dose of >10 mg/kg, and in Brazil a dose up to 20 mg/kg was necessary for a fi Fig. 2. Risk of bias assessment of selected studies. de nitive cure of 87% (13/15). This evidence comes from a 594 C. Rodrigo et al. / Clinical Microbiology and Infection 24 (2018) 591e598

Table 1 Comparison of different formulations of AmB

Study (year) Study country Trial arm and drug dosage Apparent cure Deﬁnitive cure 95% CI for Mortality deﬁnitive cure

Thakur (2001) [14] India L-AmB single dose 15 mg/kg 17/17 (100%) 17/17 (100%) 77.1e100 0 AmB deoxycholate 1 mg/kg per day for 20 days 17/17 (100%) 17/17 (100%) 77.1e100 0 Sundar (2004) [11] India AmB deoxycholate 1 mg/kg per day for 15 days (every other day) 49/51 (96%) 49/51 (96%) 85.4e99.3 2 L-AmB 10 mg/kg (2 mg/kg 5 days) 50/51 (98%) 49/51 (96%) 85.4e99.3 0 AmB lipid complex 10 mg/kg (2 mg/kg 5 days) 51/51 (100%) 47/51 (92%) 80.2e97.4 0 Sundar (2010) [12] India L-AmB 10 mg/kg (single dose) 304/304 (100%) 291/304 (96%) 92.6e97.6 0 AmB deoxycholate 1 mg/kg per day for 15 days (every other day) 106/108 (98%) 104/108 (96%) 90.2e98.8 0 Sundar (2014) [13] India AmB lipid emulsion 15 mg/kg (single dose) 354/376 (94%) 317/376 (84%) 80.1e87.8 2 L-AmB 15 mg/kg (single dose) 122/124 (98%) 120/124 (97%) 91.4e99.0 0

AmB, amphotericin B; CI, conﬁdence interval; L-AmB, liposomal amphotericin B.

Table 2 Dose-ranging studies of AmB deoxycholate

Study (year) Study Study group and drug dosage Apparent cure Deﬁnitive cure 95% CI for Mortality country deﬁnitive cure

Thakur (1996) [16] a India AmB deoxycholate 1 mg/kg per day for 15 days 96/96 (100%) 95/96 (99%) 93.5e99.9 0 AmB deoxycholate 0.75 mg/kg per day for 15 days 92/96 (96%) 87/96 (91%) 82.5e95.3 0 AmB deoxycholate 0.5 mg/kg per day for 15 days 84/95 (88%) 79/95 (82%) 73.8e89.8 0 Thakur (1998) [42] India AmB deoxycholate 1 mg/kg per day for 20 days 65/65 (100%) 64/65 (99%) 90.6e99.9 0 AmB deoxycholate 20 mg/kg of total dose in incremental doses 65/65 (100%) 64/65 (99%) 90.6e99.9 0 over 20 days (on alternate days) Sundar (2007) [43] b India AmB deoxycholate 1 mg/kg per day for 15 doses every other day 237/245 (97%) 237/245 (97%) 93.4e98.5 0 AmB deoxycholate 0.75 mg/kg per day for 15 doses every other day 229/244 (94%) 229/244 (94%) 89.9e96.4 0 AmB deoxycholate 1 mg/kg per day for 15 doses daily 491/500 (98%) 491/500 (98%) 96.5e99.1 0 AmB deoxycholate 0.75 mg/kg per day for 15 doses daily 482/496 (97%) 482/496 (97%) 95.2e98.4 0 Singh (2010) [44] b India AmB deoxycholate 1 mg/kg per day for 15 doses every other day 270/303 (89%) 267/303 (88%) 83.8e91.4 0 AmB deoxycholate 1 mg/kg per day for 15 doses daily 300/302 (99%) 299/302 (99%) 96.9e99.7 0

AmB, amphotericin B; CI, confidence interval. a The higher dose was significantly better in achieving definitive cure (p <0.05). b Combined in a meta-analysis.

Table 3 Dose-ranging studies for liposomal AmB

Study (year) Study country Trial arm and drug dosage Apparent cure Deﬁnitive cure 95% CI for Mortality deﬁnitive cure

Thakur (1996) [17] India Total dose 14 mg/kg (2 7 days) 10/10 (100%) 10/10 (100%) 65.5e100 0 Total dose 10 mg/kg (2 5 days) 10/10 (100%) 9/9 (100%) 65.5e100 0a Total dose 6 mg/kg (2 3 days) 10/10 (100%) 10/10 (100%) 65.5e100 0 Berman (1998) [10] Brazil and Kenya Total dose 6 mg/kg (2 3 days) 2/5 (40%) 1/5 (20%) 10.5e70.1 0 Total dose 10 mg/kg (2 5 days) 14/14 (100%) 13/14 (93%) 64.2e99.6 0 Total dose 14 mg/kg (2 7 days) 22/23 (96%) 18/23 (78%) 55.8e91.7 0 Total dose 20 mg/kg (2 10 days) 15/15 (100%) 13/15 (87%) 58.4e97.7 0 Bodhe (1999) [21] India Total dose 21 mg/kg (1 21 days) 15/17 (88%) 15/17 (88%) 62.3e97.9 1 Total dose 20 mg/kg (2 10 days) 11/11 (100%) 11/11 (100%) 67.9e100 0 Total dose 21 mg/kg (3 7 days) 9/10 (90%) 9/10 (90%) 54.1e99.5 0 Total dose 15 mg/kg (3 5 days) 11/13 (85%) 11/13 (85%) 53.6e97.3 0 Total dose 14 mg/kg (2 7 days) 8/11 (73%) 8/11 (73%) 39.3e92.7 0 Sundar (2001) [18] India Total dose 5 mg/kg (single dose) 45/46 (98%) 42/46 (91%) 78.3e97.2 0 Total dose 14 mg/kg (1 5 days) 44/45 (98%) 42/45 (93%) 80.7e98.3 0 Sundar (2002) [19] India Total dose 3.75 mg/kg (over 5 days) 27/28 (96%) 25/28 (89%) 70.6e97.2 0 Total dose 7.5 mg/kg (over 5 days) 27/28 (96%) 26/28 (93%) 75.1e98.8 0 Total dose 15 mg/kg (over 5 days) 28/28 (100%) 27/28 (96%) 79.8e99.8 0 Mondal (2010) [22] b India Total dose 5 mg/kg (single dose) 10/10 (100%) 6/10 (60%) 27.4e86.3 0 Total dose 7.5 mg/kg (single dose) 10/10 (100%) 5/10 (50%) 20.1e79.9 0 Total dose 10 mg/kg (double dose) 10/10 (100%) 9/10 (90%) 54.1e99.5 0 Khalil (2014) [9] Ethiopia and Sudan Total dose 21 mg/kg (3 7 days) 54/63 (86%) 46/63 (73%) 60.1e83.1 0 Total dose 10 mg/kg (single dose) 29/40 (73%) 23/40 (58%) 41.0e72.6 0 Total dose 7.5 mg/kg (single dose) 11/21 (52%) 8/21 (38%) 19.0e61.3 0c Sundar (2015) [20] d India Total dose 10 mg/kg (single dose) 15/15 (100%) 14/15 (93%) 66.0e99.7 0 Total dose 15 mg/kg (single dose) 15/15 (100%) 14/15 (93%) 66.0e99.7 0 Goswami (2016) [23] b,d India Total dose 15 mg/kg (double dose) 60/60 (100%) 60/60 (100%) 92.5e100 0 Total dose 10 mg/kg (double dose) 60/60 (100%) 54/60 (90%) 78.8e95.9 0

AmB, amphotericin B; CI, confidence interval. a One death was reported in this group 2 months after completing treatment from an unrelated cause. b Higher dose was significantly better in achieving definite cure (p <0.05). c One unrelated death due to snakebite. d Combined in a meta-analysis. C. Rodrigo et al. / Clinical Microbiology and Infection 24 (2018) 591e598 595 nonrandomized trial with incremental dosing and a small sample cardiotoxicity was more common (6/44 vs. 0, p 0.02) in the anti- size [10]. A trial in Ethiopia and Sudan was terminated early moniate group but nephrotoxicity was more common (2/44 vs. 16/ because neither of its treatment arms (single L-AmB dose of 7.5 mg/ 45, p <0.01) in the AmB group. There were five deaths each in the kg or 10 mg/kg vs. 3 mg/kg for 7 days) was successful in obtaining a two treatment groups. In the second study, the antimoniate group definitive cure of >73% (Table 3) [9]. As a result of this obvious had serious adverse events including pancreatitis (21.1%, 4/19), geographical bias in response rates and the dissimilarity of study worsening anaemia (15.8%, 3/19) and three deaths [8]. The risk of designs, these studies were not combined in a meta-analysis. The premature discontinuation of the study drug was significantly adverse effects related to L-AmB infusion were mild (mostly chills higher in this group (9/19 vs. 3/38, p <0.001). Regarding immu- and fever during infusion) when the total dose was <15 mg/kg nocompetent adults and children, a prospective open-label non- [17e20]. There was no evidence of significant hepatotoxicity, randomized trial in India showed that AmB deoxycholate was though there were several instances of transient increases in serum significantly better than sodium antimoniate gluconate (definitive creatinine [20,24]. In Brazil, where a higher dose of L-AmB was used cure 100% (60/60) vs. 46.6% (28/60), p <0.001) [15]. This reflects the (20 mg/kg), three patients developed nonfatal respiratory distress increasing antimoniate resistance in India. In Brazil, an open-label and cardiac arrhythmias associated with the infusion [10]. The trial trial of immunocompetent paediatric patients showed that AmB in Sudan and Ethiopia reported that 2% (the denominator is un- deoxycholate was as effective as meglumine antimoniate in clear) of all treatment-emergent adverse events were severe, but achieving definite cure (Table 5) [6]. In both these trials, significant the study did not elaborate further [9]. hepatotoxicity [6,15] and cardiotoxicity [15] (p <0.05) were A meta-analysis of dose-ranging studies for ABLC (all from Bihar, observed in the antimoniate groups. There were two deaths re- India; Table 4) showed that a total dose of 5 mg/kg was not inferior ported in the sodium antimoniate group in one trial [15]. to 10 or 15 mg/kg (two trials, 137 participants, low-quality evidence) [2,25]. AmB in combination with miltefosine or paromomycin

AmB vs. antimonial compounds Miltefosine is an orally administrable treatment option for VL that has been assessed in several phase 1 and 2 clinical studies AmB has been compared to antimonial compounds in four trials [26e28]. In India, two open-label randomized trials (Table 6) per- (Spain, Brazil and India), but none can be combined in a meta- formed a head-to-head comparison of AmB deoxycholate vs. mil- analysis because of the differences in dosing regimens (Table 5) tefosine, and a meta-analysis showed that both options were [6e8,15]. Two trials in Spain in patients with dual infection of HIV equally effective (RR 0.99, 95% CI 0.95e1.03, two RCTs, 523 partic- and VL showed that, firstly, AmB deoxycholate (0.7 mg/kg per day ipants, high-quality evidence; Supplementary Fig. S4) [29,30]. The for 28 days, n ¼ 45) was as effective as meglumine antimoniate (20 AmB group reported significantly more infusion-related transient mg/kg per day for 28 days, n ¼ 44), with a relatively high relapse rigors (RR not estimable, one RCT, 90/99 vs. 1/299), while mild probability (16e18%) at 6 months [7]; and that, secondly, ABLC in gastrointestinal adverse events (vomiting and diarrhoea) were total doses of 15 to 30 mg/kg was as ineffective as meglumine more common in the miltefosine group (RR 3.41, 95% CI 2.45e7.45, antimoniate (Table 5) [8]. In the first study, significant two RCTs, 523 participants, high-quality evidence) [29,31].

Table 4 Dose-ranging studies for amphotericin B lipid complex

Study (year) Study country Study group and drug dosage Apparent cure Deﬁnitive cure 95% CI for deﬁnitive cure Mortality

Sundar (1997) [25] a India Total dose 5 mg/kg (1 5 days) 19/19 (100%) 16/19 (84%) 59.5e95.8 0 Total dose 10 mg/kg (2 5 days) 20/20 (100%) 18/20 (90%) 66.9e98.2 0 Total dose 15 mg/kg (3 5 days) 21/21 (100%) 21/21 (100%) 80.8e100 0 Sundar (1998) [2] a India Total dose 5 mg/kg (single dose) 24/27 (89%) 19/27 (70%) 49.7e85.5 0 Total dose 10 mg/kg (days 1 and 6) 24/24 (100%) 19/24 (79%) 57.3e92.1 0 Total dose 10 mg/kg (days 1 and 2) 24/26 (92%) 21/26 (81%) 60.0e92.7 0 Sundar (1999) [45] India Total dose 7.5 mg/kg 26/28 (93%) 22/28 (79%) 58.5e91.0 0 Total dose 10 mg/kg 30/30 (100%) 27/30 (90%) 72.3e97.4 0

a Combined in a meta-analysis.

Table 5 AmB vs. antimonial compounds

Study (year) Study Study group and drug dosage Apparent Deﬁnitive 95% CI for Mortality country cure cure deﬁnitive cure

Laguna (1999) [7] Spain Meglumine antimoniate (20 mg pentavalent antimony/kg per day 29/44 (66%) NA NA 5 28 days) AmB deoxycholate (0.7 mg/kg per day 28 days) 28/45 (62%) NA NA 5 Laguna (2003) [8] Spain ABLC total dose: 15 mg/kg (3 5 days) 6/18 (33%) 4/18 (22%) 7.37e48.1 0 ABLC total dose: 30 mg/kg (3 10 days) 8/20 (40%) 3/20 (15%) 4.00e38.9 0 Meglumine antimoniate (20 mg pentavalent antimony/kg per day 7/19 (37%) 4/19 (21%) 7.00e46.1 3

28 days) Thakur (2004) [15] a India Sodium antimoniate gluconate (20 mg/kg per day 28 days) 34/60 (57%) 28/60 (47%) 33.9e59.9 2 AmB deoxycholate (1 mg/kg per day for 20 doses) 60/60 (100%) 60/60 (100%) 92.5e100 0 Borges (2017) [6] Brazil Meglumine antimoniate (20 mg pentavalent antimony/kg per day 48/51 (94%) 48/51 (94%) 82.8e98.5 0 20 days) AmB deoxycholate (1 mg/kg per day for 14 doses) 48/50 (96%) 47/50 (94%) 82.5e98.4 0

ABLC, amphotericin B lipid complex; AmB, amphotericin B; CI, conﬁdence interval; NA, not applicable. a AmB was signiﬁcantly better than pentavalent antimoniate. 596 C. Rodrigo et al. / Clinical Microbiology and Infection 24 (2018) 591e598

Table 6 AmB vs. miltefosine or paromomycin

Study (year) Study Study group and drug dosage Apparent cure Deﬁnitive cure 95% CI for Mortality country deﬁnitive cure

Sundar (2002) [29] a India Miltefosine 2.5 mg/kg per day for 28 days 293/299 (98%) 282/299 (94%) 90.9e96.6 0 AmB deoxycholate 1 mg/kg per day for 15 doses 98/99 (99%) 96/99 (97%) 90.8e99.2 0 Singh (2006) [30] a India Miltefosine 2.5 mg/kg per day for 28 days 64/64 (100%) 60/64 (94%) 84.0e97.0 0 AmB deoxycholate 1 mg/kg per day for 15 doses 61/61 (100%) 56/61 (92%) 81.2e97.0 0 Sundar (2007) [32] India Paromomycin 11 mg/kg per day for 28 days 493/502 (98%) 474/502 (95%) 92.0e96.2 2 AmB deoxycholate 1 mg/kg per day for 15 doses 164/165 (99%) 163/165 (99%) 95.2e99.8 1 Sundar (2008) [31] India L-AmB 5 mg/kg (single dose) 44/45 (98%) 41/45 (91%) 77.9e97.1 0b L-AmB 5 mg/kg (single dose) þ miltefosine 100 mg for 10 days 46/46 (100%) 45/46 (98%) 87.0e99.9 0 L-AmB 5 mg/kg (single dose) þ miltefosine 100 mg for 14 days 45/45 (100%) 43/45 (96%) 83.6e99.2 0 L-AmB 3.75 mg/kg (single dose) þ miltefosine 100 mg for 14 days 45/45 (100%) 43/45 (96%) 83.6e99.2 0 L-AmB 5 mg/kg (single dose) þ miltefosine 100 mg for 7 days 45/45 (100%) 44/45 (98%) 86.8e99.9 0 Sundar (2011) [33] India AmB deoxycholate 1 mg/kg per day for 15 doses 149/157 (95%) 145/157 (92%) 86.8e95.8 1c L-AmB 5 mg/kg (single dose) þ miltefosine 50 mg for 7 days 157/160 (98%) 155/160 (97%) 92.5e98.9 0 L-AmB 5 mg/kg (single dose) þ paromomycin 11 mg/kg 155/158 (98%) 153/158 (97%) 92.4e98.8 0 per day for 10 days Miltefosine 50 mg per day for 10 days þ paromomycin 158/159 (99%) 156/159 (98%) 94.1e99.5 0 11 mg/kg per day for 10 days

AmB, amphotericin B; CI, conﬁdence interval; L-AmB, liposomal amphotericin B. a Combined in a meta-analysis. b One unrelated death (during a ﬂood) was reported in the follow-up period. c One additional unrelated death was reported (car accident) in this group.

Paromomycin is an aminoglycoside antibiotic that is effective Comparisons to other drugs against VL which must be administered via the intramuscular route (11 mg/kg per day for 10 days). In the only head-to-head trial of Regarding other comparisons (Table 7), a single underpowered AmB deoxycholate and paromomycin, the authors concluded that randomized open-label trial in India showed that pentamidine was the latter was noninferior in achieving definitive cure based on less effective than AmB deoxycholate in antimonial-resistant pa- noninferiority testing with a noninferiority margin of 0.1. This tients (RR 0.78, 95% CI 0.64e0.97, one trial, 82 participants, low- margin was set by the authors as a clinically acceptable margin of quality evidence, p 0.04) [34]. There were no serious adverse noninferiority. The percentage difference in achieving definitive events in either study group. A single small open-label randomized cure between the two regimens (4.2%) and the upper bound of the trial showed similar efficacy of sitamaquine (an 8-aminoquinoline) 97.5% CI for this result (6.9%) did not exceed the margin of non- vs. AmB deoxycholate in achieving definitive cure at 6 months (one inferiority. However, relative risk with 95% CI suggested that AmB RCT, 61 participants, low-quality evidence) [35]. Mild transfusion- deoxycholate may be better than paromomycin in achieving related chills and gastrointestinal events were more common in definitive cure (RR 1.04, 95% CI 1.02e1.07, one trial, 667 partici- the AmB group (RR 8.71, 95% CI 3.37e22.51). pants, low-quality evidence) [32]. These results have to be inter- preted in the clinical context of whether a noninferiority margin of Discussion 10% is appropriate. There was no overall difference in adverse events. Two deaths were reported in the paromomycin group, In summarizing the results of the studies we assessed, different while one death was observed in the AmB deoxycholate group. formulations of AmB were comparable to each other when used at AmB, paromomycin and miltefosine have been tried in combi- doses above 15 mg/kg (total dose) for AmB deoxycholate, 10 mg/kg nations with each other in two trials in India, but none of the for L-AmB and 10 mg/kg for ABLC. Antimonial compounds were not subgroups could be combined in a meta-analysis because of dif- superior to AmB deoxycholate but caused more adverse events. ferences in doses [30,33]. The evidence from these individual AmB deoxycholate had comparable efficacy to miltefosine or studies suggests equal efficacy, with cure rates above 90% (Table 6) combinations of miltefosine and paromomycin. AmB deoxycholate in all of the following comparisons: AmB deoxycholate vs. L-AmB is comparable to or perhaps better in achieving definitive cure than plus miltefosine or L-AmB plus paromomycin or miltefosine plus paromomycin alone. Pentamidine may have lower efficacy than paromomycin; and L-AmB vs. L-AmB plus miltefosine (Table 6). AmB deoxycholate, while sitamaquine may have equally efficacy. With AmB deoxycholate, the risk of serious adverse events, The quality of evidence and applicability of these data within including deaths and withdrawals (RR 18.23, 95% CI 4.12e80.57, one India is substantial, as the studies are consistent in their findings RCT, 634 participants, low-quality evidence), and the risk of any over a period of two decades. As mentioned previously, India is one adverse event (RR 1.91, 95% CI 1.71e2.12, one RCT, 634 participants, of the six countries that accounts for 90% of the global burden of VL. low-quality evidence) was significantly higher compared to the These findings are also useful to other South Asian countries such as other groups [33]. Sri Lanka and Bangladesh that have indigenous transmission of VL.

Table 7 AmB vs. other drugs

Study (year) Study country Study group and drug dosage Apparent cure Deﬁnitive cure 95% CI for deﬁnitive cure Mortality

Das (2009) [34] India AmB deoxycholate 15 mg/kg per day for 15 days 39/41 (95%) 38/41 (93%) 79.0e98.1 0 Pentamidine 4 mg/kg per day for 15 days 34/41 (83%) 30/41 (73%) 56.8e85.2 0 Sundar (2011) [35] India AmB deoxycholate 15 mg/kg per day for 15 days 19/20 (95%) 19/20 (95%) 73.1e99.7 0 Sitamaquine 2 mg/kg per day for 21 days 39/41 (95%) 35/41 (85%) 70.1e93.9 0

AmB, amphotericin B. C. Rodrigo et al. / Clinical Microbiology and Infection 24 (2018) 591e598 597

Unfortunately, given the global resistance patterns of the parasite, firstly, L-AmB and AmB deoxycholate infusions are associated with the same recommendations cannot be extended to other high-risk chills, shivering and fever that are mild and transient (confirmed by countries in East Africa, South America or the Mediterranean. In a meta-analysis), and secondly, antimonial compounds are more East Africa, antimonial compounds are the preferred first-line likely to cause serious adverse events (hepatotoxicity, anaemia, choice of treatment for VL [9]. pancreatitis), including treatment-related deaths, compared to In India, L-AmB was superior to AmB deoxycholate or antimo- AmB. There were only a few deaths reported in all trials combined nial compounds in terms of adverse events. Though both these except in the two trials that assessed patients with HIV and VL dual formulations showed similar efficacy (in the dosages tested), in infection. This observation may be biased, as most studies excluded practice, single-dose L-AmB will be more convenient to administer patients with severe VL (Supplementary File S1). The few deaths than a 2- to 3-week course of AmB deoxycholate. The compliance observed were distributed across heterogeneous trials, so they rate and patient acceptance will be better with a single-dose option were not combined in a meta-analysis. because there is less interruption to daily routines and earnings. However, the transition of evidence to practice is mainly limited by Conclusions cost considerations [36]. Currently, L-AmB is available at a dis- counted price for use in developing countries, but even so, the cost Evidence from India suggests that L-AmB is as effective as AmB of a treatment course per patient exceeds US$100 for a single dose deoxycholate. ABLC is as effective as L-AmB but better than ABLE. A of 5 mg/kg [37]. Given that the annual VL incidence in India is single low dose of L-AmB (3.75e5 mg/kg) is as effective as higher 150 000 cases (at the most conservative estimate), these costs are doses of the same formulation, but the version produced locally in fi still signi cant [37]. However, the advantages of a short hospital India needs to be administered at a higher dose of 10 mg/kg for the stay with single-dose administration may partly offset the ex- same effect. The efficacy of paromomycin or miltefosine (by itself or penses incurred [33,38]. The combination of miltefosine and in combination with L-AmB) was similar to that of AmB deoxy- paromomycin would be even cheaper (without the cost of the cholate alone or L-AmB alone. Finally, AmB deoxycholate was su- hospital stay) [33]. However, cost calculations for other countries perior to antimonial compounds in achieving definitive cure in will vary depending on VL burden and pricing agreements with the India. manufacturer [39]. Outside India, the data are too limited to draw well-supported fi A previous review published in 2005 agrees with our ndings conclusions. The available evidence suggests that AmB may be as that AmB deoxycholate is more effective than antimonial com- effective as antimonials in Brazil. More prospective comparative pounds because of emerging resistance to antimonials in India, and trials are needed for patients in these geographical locations as well that L-AmB is as effective as AmB deoxycholate [40]. It also agrees as those with VL and HIV dual infection. that pentamidine is less effective (referring to a study that was not within the time period of our review [40]). A second review pub- Transparency declaration lished in 2016 independently highlights the lack of efficacy of AmB in East Africa [1]. A third review provides a descriptive summary of All authors report no conflicts of interest relevant to this article. clinical trials on VL [41]. The study agrees that most studies on this topic have been conducted in India, not only for AmB but also for other treatment options. In their review for all treatment options, of Appendix A. Supplementary data 145 trials enrolling 26 986 patients, a majority were from the Indian subcontinent (66% of cohorts and 83% of patients). However, this Supplementary data related to this article can be found at review differs from ours because their analysis is only descriptive of https://doi.org/10.1016/j.cmi.2017.11.008. the trials and does not include a comparison of different treatment options against each other, which is understandably a complex References task, given the wide heterogeneity of treatment options, doses and formulations available. Our review focuses on one popular treat- [1] Sundar S, Singh A. Recent developments and future prospects in the treatment e ment option and weighs the comparative efficacy of its different of visceral leishmaniasis. Ther Adv Infect Dis 2016;3:98 109. [2] Sundar S, Goyal AK, More DK, Singh MK, Murray HW. Treatment of antimony- formulations to synthesize the best available evidence for use. unresponsive Indian visceral leishmaniasis with ultra-short courses of Our study has several limitations. Regarding the quality of evi- amphotericin-Belipid complex. Ann Trop Med Parasitol 1998;92:755e64. dence, all studies reviewed (except one [19]) were open-label [3] Hamill RJ. Amphotericin B formulations: a comparative review of efficacy and toxicity. Drugs 2013;73:919e34. studies. However, this is unlikely to affect the primary outcome of [4] Nordic Cochrane Centre. Review manager 5.1 [computer programme]. definitive cure, which is decided by objective criteria (presence or Copenhagen: Cochrane Collaboration; 2011. absence of the parasite). More importantly, there were three non- [5] Higgins JPT, Thompson SG, Deeks JJ, Altman DG. Measuring inconsistency in meta-analysis. BMJ 2003;327:557e60. randomized studies that could have led to selection bias [10,15,20], [6] Borges MM, da Silva Pranchevicius MC, Noronha EF, Romero GAS, Carranza- and one study was underpowered because it could not recruit the Tamayo CO. Efficacy and safety of amphotericin B deoxycholate versus N- required sample size [34]. Regarding the studies from outside India, methylglucamine antimoniate in pediatric visceral leishmaniasis: an open- fi label, randomized, and controlled pilot trial in Brazil. Rev Soc Bras Med the Spanish trials that evaluated the ef cacy of AmB in HIV patients Trop 2017;50:67e74. had a small number of patients in each study group, but given the [7] Laguna F, Lopez-Velez R, Pulido F, Salas A, Torre-Cisneros J, Torres E, et al. difficulty in recruiting patients with a dual infection (patients were Treatment of visceral leishmaniasis in HIV-infected patients: a randomized recruited from multiple centres across Spain), it is unlikely that trial comparing meglumine antimoniate with amphotericin B. Spanish HIV- Leishmania Study Group. AIDS 1999;13:1063e9. larger numbers can be assessed by future studies. The low efficacy [8] Laguna F, Videla S, Jimenez-Mejias ME, Sirera G, Torre-Cisneros J, Ribera E, reported in these studies could have been influenced by con- et al. Amphotericin B lipid complex versus meglumine antimoniate in the founders such as other comorbidities and CD4 cell count [7,8]. Apart treatment of visceral leishmaniasis in patients infected with HIV: a randomized pilot study. J Antimicrob Chemother 2003;52:464e8. from the open-label design, the studies by Borges et al. [6] in Brazil [9] Khalil EA, Weldegebreal T, Younis BM, Omollo R, Musa AM, Hailu W, et al. and Khalil et al. [9] in Africa (Sudan and Ethiopia) did not have any Safety and efficacy of single dose versus multiple doses of AmBisome for other significant bias. Unlike the primary outcome, the reporting of treatment of visceral leishmaniasis in eastern Africa: a randomised trial. PLoS fl Negl Trop Dis 2014;8:e2613. adverse events could have been in uenced by the open-label [10] Berman JD, Badaro R, Thakur CP, Wasunna KM, Behbehani K, Davidson R, et al. design in the trials. Yet there are two consistent observations: Efficacy and safety of liposomal amphotericin B (AmBisome) for visceral 598 C. Rodrigo et al. / Clinical Microbiology and Infection 24 (2018) 591e598

leishmaniasis in endemic developing countries. Bull World Health Organ [29] Sundar S, Jha TK, Thakur CP, Engel J, Sindermann H, Fischer C, et al. Oral 1998;76:25e32. miltefosine for Indian visceral leishmaniasis. N Engl J Med 2002;347: [11] Sundar S, Mehta H, Suresh AV, Singh SP, Rai M, Murray HW. Amphotericin B 1739e46. treatment for Indian visceral leishmaniasis: conventional versus lipid for- [30] Singh UK, Prasad R, Mishra OP, Jayswal BP. Miltefosine in children with mulations. Clin Infect Dis 2004;38:377e83. visceral leishmaniasis: a prospective, multicentric, cross-sectional study. In- [12] Sundar S, Chakravarty J, Agarwal D, Rai M, Murray HW. Single-dose liposomal dian J Pediatr 2006;73:1077e80. amphotericin B for visceral leishmaniasis in India. N Engl J Med 2010;362: [31] Sundar S, Rai M, Chakravarty J, Agarwal D, Agrawal N, Vaillant M, et al. New 504e12. treatment approach in Indian visceral leishmaniasis: single-dose liposomal [13] Sundar S, Pandey K, Thakur CP, Jha TK, Das VN, Verma N, et al. Efficacy and amphotericin B followed by short-course oral miltefosine. Clin Infect Dis safety of amphotericin B emulsion versus liposomal formulation in Indian 2008;47:1000e6. patients with visceral leishmaniasis: a randomized, open-label study. PLoS [32] Sundar S, Jha TK, Thakur CP, Sinha PK, Bhattacharya SK. Injectable paromo- Negl Trop Dis 2014;8:e3169. mycin for visceral leishmaniasis in India. N Engl J Med 2007;356:2571e81. [14] Thakur CP. A single high dose treatment of kala-azar with Ambisome [33] Sundar S, Sinha PK, Rai M, Verma DK, Nawin K, Alam S, et al. Comparison of (amphotericin B lipid complex): a pilot study. Int J Antimicrob Agents short-course multidrug treatment with standard therapy for visceral leish- 2001;17:67e70. maniasis in India: an open-label, non-inferiority, randomised controlled trial. [15] Thakur CP, Narayan S. A comparative evaluation of amphotericin B and so- Lancet 2011;377:477e86. dium antimony gluconate, as first-line drugs in the treatment of Indian [34] Das VNR, Siddiqui NA, Pandey K, Singh VP, Topno RK, Singh D, et al. A visceral leishmaniasis. Ann Trop Med Parasitol 2004;98:129e38. controlled, randomized nonblinded clinical trial to assess the efficacy of [16] Thakur CP, Sinha GP, Pandey AK. Comparison of regimens of amphotericin B amphotericin B deoxycholate as compared to pentamidine for the treatment deoxycholate in kala-azar. Indian J Med Res 1996;103:259e63. of antimony unresponsive visceral leishmaniasis cases in Bihar, India. Ther [17] Thakur CP, Pandey AK, Sinha GP, Roy S, Behbehani K, Olliaro P. Comparison of Clin Risk Manage 2009;5:117e24. three treatment regimens with liposomal amphotericin B (Ambisome) for [35] Sundar S, Sinha PK, Dixon SA, Buckley R, Miller AK, Mohamed K, et al. Phar- visceral leishmaniasis in India: a randomized dose-finding study. Trans R Soc macokinetics of oral sitamaquine taken with or without food and safety and Trop Med Hyg 1996;90:319e22. efficacy for treatment of visceral leishmaniais: a randomized study in Bihar, [18] Sundar S, Agrawal G, Rai M, Makharia MK, Murray HW. Treatment of Indian India. Am J Trop Med Hyg 2011;84:892e900. visceral leishmaniasis with single or daily infusions of low dose liposomal [36] World Health Organization. Control of visceral leishmaniasis. Geneva: World amphotericin B: randomised trial. BMJ 2001;323:419e22. Health Organization; 2010. [19] Sundar S, Jha TK, Thakur CP, Mishra M, Singh VR, Buffels R. Low-dose lipo- [37] Alvar J, Velez ID, Bern C, Herrero M, Desjeux P, Cano J, et al. Leishmaniasis somal amphotericin B in refractory Indian visceral leishmaniasis: a multi- worldwide and global estimates of its incidence. PLoS One 2012;7:e35671. center study. Am J Trop Med Hyg 2002;66:143e6. [38] Olliaro P, Darley S, Laxminarayan R, Sundar S. Cost-effectiveness projections [20] Sundar S, Singh A, Rai M, Chakravarty J. Single-dose indigenous liposomal of single and combination therapies for visceral leishmaniasis in Bihar, India. amphotericin B in the treatment of Indian visceral leishmaniasis: a phase 2 Trop Med Int Health 2009;14:918e25. study. Am J Trop Med Hyg 2015;92:513e7. [39] Meheus F, Balasegaram M, Olliaro P, Sundar S, Rijal S, Faiz MA, et al. Cost- [21] Bodhe PV, Kotwani RN, Kirodian BG, Pathare AV, Pandey AK, Thakur CP, et al. effectiveness analysis of combination therapies for visceral leishmaniasis in Dose-ranging studies on liposomal amphotericin B (L-AMP-LRC-1) in the the Indian subcontinent. PLoS Negl Trop Dis 2010;4. treatment of visceral leishmaniasis. Trans R Soc Trop Med Hyg 1999;93:314e8. [40] Olliaro PL, Guerin PJ, Gerstl S, Haaskjold AA, Rottingen JA, Sundar S. Treatment [22] Mondal S, Bhattacharya P, Rahaman M, Ali N, Goswami RP. A curative immune options for visceral leishmaniasis: a systematic review of clinical studies done profile one week after treatment of Indian kala-azar patients predicts success in India, 1980e2004. Lancet Infect Dis 2005;5:763e74. with a short-course liposomal amphotericin B therapy. PLoS Negl Trop Dis [41] Bush JT, Wasunna M, Alves F, Alvar J, Olliaro PL, Otieno M, et al. Systematic 2010;4, e764. review of clinical trials assessing the therapeutic efficacy of visceral leish- [23] Goswami RP, Goswami RP, Das S, Satpati A, Rahman M. Short-course treat- maniasis treatments: a first step to assess the feasibility of establishing an ment regimen of Indian visceral leishmaniasis with an Indian liposomal individual patient data sharing platform. PLoS Negl Trop Dis 2017;11: amphotericin B preparation (Fungisome). Am J Trop Med Hyg 2016;94:93e8. e0005781. [24] Sundar S, Jha TK, Thakur CP, Mishra M, Singh VP, Buffels R. Single-dose [42] Thakur CP, Kumar P, Kumar N, Singh GN, Singh AK, Narain S. A randomized liposomal amphotericin B in the treatment of visceral leishmaniasis in India: a comparison of classical mode of administration of amphotericin B with its multicenter study. Clin Infect Dis 2003;37:800e4. newer modes of administration in kala-azar. J Assoc Phys India 1998;46: [25] Sundar S, Agrawal NK, Sinha PR, Horwith GS, Murray HW. Short-course, low- 779e83. dose amphotericin B lipid complex therapy for visceral leishmaniasis unre- [43] Sundar S, Chakravarty J, Rai VK, Agrawal N, Singh SP, Chauhan V, et al. sponsive to antimony. Ann Intern Med 1997;127:133e7. Amphotericin B treatment for Indian visceral leishmaniasis: response to 15 [26] Sundar S, Rosenkaimer F, Makharia MK, Goyal AK, Mandal AK, Voss A, et al. daily versus alternate-day infusions. Clin Infect Dis 2007;45:556e61. Trial of oral miltefosine for visceral leishmaniasis. Lancet 1998;352:1821e3. [44] Singh UK, Prasad R, Jaiswal BP, Singh PK, Thakur CP. Amphotericin B therapy [27] Sundar S, Gupta LB, Makharia MK, Singh MK, Voss A, Rosenkaimer F, et al. Oral in children with visceral leishmaniasis: daily vs. alternate dayda randomized treatment of visceral leishmaniasis with miltefosine. Ann Trop Med Parasitol trial. J Trop Pediatr 2010;56:321e4. 1999;93:589e97. [45] Sundar S, Goyal AK, Mandal AK, Makharia MK, Singh VP, Murray HW. [28] Jha TK, Sundar S, Thakur CP, Bachmann P, Karbwang J, Fischer C, et al. Mil- Amphotericin B lipid complex in the management of antimony unresponsive tefosine, an oral agent, for the treatment of Indian visceral leishmaniasis. N Indian visceral leishmaniasis. J Assoc Physicians India 1999;47:186e8. Engl J Med 1999;341:1795e800.