HIV/AIDS • CID 2006:43 (1 August) • 357
H I V / A I D SM A J O R A R T I C L E
A Comparison of Miltefosine and Sodium
Stibogluconate for Treatment of Visceral
Leishmaniasis in an Ethiopian Population
with High Prevalence of HIV Infection
Koert Ritmeijer,1Abren Dejenie,1Yibeltal Assefa,2Tadesse Beyene Hundie,1Jo Mesure,1Gerry Boots,1
Margriet den Boer,1and Robert N. Davidson1
1Me ´decins Sans Frontie `res-Holland, Amsterdam, The Netherlands; and
2Tigray Bureau of Health, Mekelle, Tigray, Ethiopia
incidence of human immunodeficiency virus (HIV)coinfectionrequiresalternativesafeandeffectivedrugregimens.
Oral miltefosine has been proven to be safe and effective in the treatment of Indian VL but has not been studied
in Africa or in persons with HIV and VL coinfection.
We compared the efficacy of miltefosine and sodium stibogluconate (SSG) in the treatment of VL
in persons in Ethiopia. A total of 580 men with parasitologicallyand/orserologicallyconfirmedVLwererandomized
to receive either oral miltefosine (100 mg per day for 28 days) or intramuscular SSG (20 mg/kg per day for 30
The initial cure rate was 88% in both treatment groups. Mortality during treatment was 2% in the
miltefosine group, compared with 10% in the SSG group. Initial treatment failure was 8% in the miltefosinegroup,
compared with 1% in the SSG group. Among the 375 patients (65%) who agreed toHIVtesting,HIVseroprevalence
was 29%. Among patients not infected with HIV, initial cure, mortality, and initial treatment failure rates were
not significantly different (94% vs. 95%, 1% vs. 3%, and 5% vs. 1%forthemiltefosineandSSGgroups,respectively).
Initial treatment failure with miltefosine occurred in 18% of HIV-coinfected patients, compared with treatment
failure in 5% of non–HIV-infected patients. At 6 months after treatment, 174(60%)ofthe290miltefosinerecipients
and 189 (65%) of the 290 SSG recipients experienced cure; 30 (10%) of 290 in the miltefosine group and 7 (2%)
of 290 in the SSG group experienced relapse, and the mortality rate was 6% in the miltefosine group, compared
with 12% in the SSG group. HIV-infected patients had higher rates of relapse (16 [25%] of 63 patients), compared
with non–HIV-infected patients (5 [5%] of 131).
Treatment with miltefosine is equally effective as standard SSG treatment in non–HIV-infected
men with VL. Among HIV-coinfected patients, miltefosine is safer but less effective than SSG.
Antimonials are the mainstay of visceral leishmaniasis (VL) treatment in Africa. The increasing
Visceral leishmaniasis (VL; also called “kala-azar”)
caused by Leishmania donovani is endemic in the low-
lands around Humera and Metema in northern Ethi-
opia, with an incidence of 1000–2000 cases annually;
20%–40% of the persons affected are HIV coinfected
[1, 2]. The only antileishmanial drug available in Ethi-
opia is sodium stibogluconate (SSG, a pentavalent an-
Received 19 January 2006; accepted 24 March 2006; electronically published
20 June 2006.
Reprints or correspondence: Dr. Koert Ritmeijer, Me ´decins sans Frontie `res/
MSF-Holland, Plantage Middenlaan 14, 1018 DD Amsterdam, The Netherlands
Clinical Infectious Diseases 2006;43:357–64
? 2006 by the Infectious Diseases Society of America. All rights reserved.
timonial). Although this is not yet a problem in Africa,
in India, primary resistance to antimonials is common
[3, 4]. Although VL is treated similarly in patients with
HIV infection and patients without HIV infection [5,
6], HIV coinfection results in lower VL cure rates,
higher death and relapse rates, and greatertoxicityfrom
antimonials [1, 2, 7]. SSG re-treatment regimens are
lengthy (30–60 days), are difficult to tolerate, and are
not always successful. Miltefosine, a membrane-active
alkyl phospholipid, was developed as an anticancer
agent and was found to have antileishmania activity in
animal VL models [8, 9]. In 1995, ASTA Medica/Zen-
taris and the World Health Organization–Special Pro-
gramme for Research and Training in Tropical Diseases
developed miltefosine for the treatment of VL. Several
358 • CID 2006:43 (1 August) • HIV/AIDS
dose-finding studies were done [10–14]. Phase III clinical trials
in India showed that miltefosine is very effective for treating
VL in both adults and children, including those who experi-
enced failure with antimonials. Miltefosine given orally for 28
days at 100 mg per day (∼2.5 mg/kg per day) had a 95%–100%
initial clinical and parasitological cure rate.
Miltefosine was registered for treatment of VL in India in
2002, in Germany in 2004, and in 5 Latin American countries
in 2005, where it was also registered for treatment of cutaneous
leishmaniasis. We wished to establish the utility of miltefosine
in east African VL, where the parasite strain is different and
patients have severe comorbidities . Our strategy was first to
evaluate miltefosine as monotherapy and to subsequently eval-
uate miltefosine-SSG combinations. The latter might achievethe
ultimate goal of short course treatment, a high cure rate, low
toxicity, low cost, and lower probability of inducing resistance
to either drug.
MATERIALS AND METHODS
is a remote, tropical region where extensive agriculture is per-
formed manually by large numbers of migrant laborers. It is
served by Humera Hospital (Humera, Ethiopia) and surround-
ing clinics, including Mycadra (Tigray, Ethiopia), which fall
under the responsibility of the Tigray Bureau of Health (Me-
kelle, Tigray, Ethiopia). Me ´decins Sans Frontie `res-Holland has
supported the care of patients with VL since 1997. More than
200 patients per month are treated during the peak season
(December–February). More than 80% of patients with VL are
male migrant workers infected with L. donovani while sleeping
in the fields; 19% were previously reported to be coinfected
with HIV .
Males aged ?15 years with parasitologicallyand/
or serologically confirmed VL attending Humera Hospital and
Mycadra Health Center were enrolled in the study. Because of
the potential teratogenicity of miltefosine, females were ex-
cluded. Previous antileishmanialtreatmentwasrecordedathos-
pital admission. Patients were enrolled in the study after giving
informed consent. Potentially eligible patients were only ex-
cluded if they had such severe comorbidity that they were con-
sidered to be likely to die during the month’s treatment.
HIV serostatus of the patient was determined
after voluntary counseling and testing, which is routine at the
hospital. HIV antibodies were detected by parallel testing with
2 rapid tests: HIV-Determine (Abbott Diagnostics) and HIV-
Capillus (Trinity Biotech). In the instance of discordant test
results, a third test (Unigold; Trinity Biotech) was performed.
HIV-infected patients had access to a package of health care
that includes medical follow-up in a dedicated clinic, prophy-
laxis against opportunistic infections, and antiretroviral treat-
Kafta-Humera Woreda (Tigray, Ethiopia)
ment. Participation in the study and HIV testing were not
linked to each other, and participation in either was voluntary.
The World Health Organization case definition
of VL was used for initial screening: a history of fever for 12
weeks (with malaria excluded) in combination with wasting,
and either splenomegaly or lymphadenopathy [16, 17]. For pa-
tients whose illness met this case definition, VL was confirmed
by a high titer leishmania direct agglutination test (DAT [Royal
Tropical Institute]; titer?1:6400) . In patients with an in-
termediary DAT titer (1:800–1:3200), splenic or lymph node
aspiration was performed, and VL was confirmed by micro-
scopic examination. Persons with suspected VL with a negative
DAT titer (?1:400) were evaluated for alternative illnesses and
were retested if signs and symptoms persisted. Severely ill pa-
tients were aspirated without delay, so that a diagnosis could
be made as quickly as possible. Patients with previousantileish-
manial treatment were only admitted if they had a positive
Miltefosine (Impavido, Zentaris) was provided
in foil-wrapped blister packs of 50-mg capsules. Miltefosine
dosage was 100 mg per day for 28 days (all patients weighed
125 kg). Capsules were taken with a meal of high energy, high
protein biscuits, directly observed by the Me ´decins Sans Fron-
tie `res-Holland nurse.
SSG was provided in 30-mL vials, each containing 100 mg/
mL of SSG (Albert David); the vials were provided by Inter-
national Dispensary Association (Amsterdam, TheNetherlands).
SSG dosage was 20 mg/kg per day by intramuscular injection
for 30 days. Patients who had previously received SSG and who
had experienced relapse were randomized to receive eithermil-
tefosine (100 mg per day for 28 days) or SSG (20 mg/kg per
day for 40–60 days) until 2 consecutive weekly aspirates had
Patients who did not respond clinically or parasitologically
to miltefosine treatment or who showed severe symptoms pos-
sibly caused by miltefosine received treatment with SSG (20
mg/kg per day for 30 days). Patients who did not respond to
SSG treatment or who developed intolerable SSG toxicity were
treated exprotocol with amphotericin B deoxycholate. Patients
who relapsed after the study were treated with SSG.
All VL patients received free treatment. Patients who did not
participate in the study were treated with SSG, which is the
standard of care in Ethiopia. Nurses were educatedtoovercome
any preconceived preference for either drug and were super-
vised to ensure accuracy in the recording of data.
Test of cure.
A test of cure (TOC) aspirate was done at day
27–30 to evaluate parasitological cure. In patients with a pal-
pable spleen at the end of treatment, a splenic aspirate was
performed; in other patients, a lymph node aspirate was per-
formed. We have previously found lymph node aspirates to be
comparable to splenic aspirates (data available on request). In
HIV/AIDS • CID 2006:43 (1 August) • 359
Table 1.Patient characteristics at baseline.
to receive miltefosine
(n p 290)
to receive sodium
(n p 290)P
Age, mean years ? SD (median)
Mean body mass indexa? SD (median)
Hemoglobin level, mean g/dL ? SD (median)
Spleen size, mean cm ? SD (median)
Duration of illness, mean no. of months ? SD (median)
No. (%) of migrant workers
No. (%) unable to walk unaided
HIV infected, no. (%)b
Non–HIV infected, no. (%)b
Unknown, no. (%)
29.1 ? 9.9 (26)
17.3 ? 2.1 (17.2)
9.2 ? 2.3 (9)
9.3 ? 5.6 (8.5)
2.6 ? 2.1 (2)
29.5 ? 9.6 (27)
17.4 ? 1.8 (17.4)
9.2 ? 2.3 (9.1)
9.5 ? 5.7 (9)
2.6 ? 2.1 (2)
aCalculated as body weight (kg) divided by height (m2).
bThe numbers of HIV-infected and non–HIV-infected patients reflect only those who were tested.
case of a positive TOC aspirate result, patients were treated
In patients without palpable spleen or lymph nodes, cure could
only be established clinically.
The main outcome of analysis was
the final cure rate at the 6-month follow-up visit. Secondary
outcomes were as follows: (1) initial cure (defined as initial
parasitological clearance as demonstrated by TOC aspirate in
combination with clinical improvement, or clinical cure alone
[clearance of fever, in combination with spleen regression, in-
creased hemoglobin, or weight gain] for a patient for whom a
TOC aspirate could not be performed); (2) initial treatment
failure (defined as parasitological and/or clinical failure after
initial treatment); (3) adverse effects (especially vomiting and
diarrhea); (4) intercurrent events (e.g., death, default [defined
as starting but failing to complete treatment because of reasons
other than death or decision by the clinician], bleeding, di-
arrhea, vomiting, or pneumonia); and (5) relapse (defined as
clinical symptoms of infection with parasitological confirma-
tion within the 6-months follow-up).
Patients were asked to return after 6 months
or sooner if any symptoms of VL recurred. If relapse was sus-
pected (according to clinical case definition) an aspirate was
performed to confirm VL. If no relapse occurred by 6 months
after discharge, the patient was considered to be finally cured.
Any further laboratory investigations to assess final cure were
not possible. We attempted to actively trace those who had not
returned by 6 months.
After the diagnosis of VL was made, and
if the patient satisfied the inclusion criteria, informed oral and
written consent was sought; once given, the patient was ran-
domized to receivemiltefosineorSSGaccordingtoacomputer-
generated number list. The allocation ratio was 1:1. The study
was unblinded; miltefosine is oral medication and SSG is in-
Data were analyzed on an intent-to-
treat basis with Epi Info software, 2002 revision 2 (Centers for
Disease Control and Prevention). Groups were compared by
Yates’ corrected x2test and Fisher’s exact test for categorical
variables and either by t test or the Mann-Whitney U test for
numerical variables, as appropriate. A multivariate logistic re-
gression was done to analyze for independence of risk factors
Formal and ethics approval.
the Tigray regional and Ethiopian national health authorities
and their ethical review boards, as well as by the Me ´decins Sans
Frontie `res-Holland internationalethicsreviewboard.Thestudy
was performed in accordance with the World Medical Asso-
ciation’s Declaration of Helsinki concerning medical research
in humans .
This study was approved by
in the study. Randomization produced groups with no signif-
icant differences in the main baseline characteristics (e.g., age,
body mass index, hemoglobin level, spleen size, duration of
illness, and level of weakness) (table 1).
Diagnosis of VL was confirmed by DAT titer in 449 patients
(77%), and by positive results of aspirate microscopic exami-
nation in 131 (23%). There was no significant difference in
mean DAT titer between HIV-infected and non–HIV-infected
patients (). In patients whose illness was diagnosedpar-
P p .49
asitologically, the parasite density was significantly higher in
HIV-infected patients, compared with non–HIV-infected pa-
tients (). Thirty-four patients (5.9%) experiencedre-P p .0003
A total of 580 adult male VL patients were enrolled
360 • CID 2006:43 (1 August) • HIV/AIDS
Flow diagram of patient outcomes during the study. SSG, sodium stibogluconate.
lapse after previous SSG treatment for VL (17 wererandomized
to each treatment arm). The progress of patients through the
study and the main outcomes are shown in figure 1.
Initial response to treatment.
there were 329 spleen aspirates and 105 lymph node aspirates
performed). In the other 146 patients, no TOC aspirates could
be performed because of absence of palpable spleen or lymph
After evaluation of clinical
nodes. As shown in tables 2 and 3, there was no difference in
initial cure rate between the miltefosine group (88.3%; 95% CI,
84.0%–91.7%) and the SSG group (87.6%; 83.2%–91.2%) (P
p.90). However, initial treatment failure with survivalwasmore
frequent in the miltefosine group (7.9% in the miltefosinegroup
vs. 0.7% in the SSG group; OR, 12.4;
tality was lower in the miltefosine group (2.1% in themiltefosine
group vs. 9.7% in the SSG group; OR, 0.20;
), whereas mor-P ! .0001
).P p .0002
HIV/AIDS • CID 2006:43 (1 August) • 361
Table 2.Baseline characteristics of patients with and without HIV coinfection, by HIV serostatus.
NegativeUnknown All patients
Mean ? SDMedian (range)Mean ? SD Median (range)Mean ? SD Median (range)Mean ? SDMedian (range)
Body mass indexa
33.4 ? 9.532 (18–67)26.3 ? 9.0 27 (16–60)31.2 ? 9.729 (16–65)29.3 ? 9.827 (16–67)
17.2 ? 2.017.2 (11.2–22.2)17.4 ? 1.817.5 (13.3–22.6)17.3 ? 1.917.3 (11.9–22.2)17.4 ? 1.917.3 (11.2–22.6)
Spleen size, cm
DAT titer, wellb
9.4 ? 5.79 (0–22) 10.0 ? 5.810 (0–30) 8.8 ? 5.3 8 (0–25)9.4 ? 5.6 9 (0–30)
8.3 ? 1.48 (6–11)8.3 ? 1.3 8 (4–11)8.4 ? 1.48 (6–11)8.4 ? 1.38 (4–11)
3.7 ? 1.84 (0–6)2.3 ? 1.5 2 (0–6) 2.6 ? 1.8 2 (0–6) 2.7 ? 1.8 2 (0–6)
aCalculated as body weight (kg) divided by height (m2).
bDAT titer is expressed as the highest dilution at which agglutination is still visible: well 4, 1:800 dilution; well 6, 1:3200; well 7, 1:6400; well 8, 1:12800;
well 11, 1:102400.
cThe parasite density score uses a log scale ranging from 0 (no parasites per 1000 oil-immersion fields) to +6 (1100 parasites per oil-immersion field).
DAT, direct agglutination test.
All patients who experienced initial treatment failure (23
patients in the miltefosine group and 2 in the SSG group) were
immediately re-treated with 30 days of SSG treatment. This
strategy increased the end-of-treatment cure rates to 94.1% in
the miltefosine group (95% CI, 90.8%–96.5%), compared with
87.9% in the SSG group (95% CI, 83.6%–91.4%) (
mortality remained significantly lower in the miltefosine group
(2.8% vs. 9.7% in the SSG group; OR, 0.27;
HIV coinfection and initial treatment outcomes.
centage of patients who underwent voluntary counseling and
testing for HIV infection was 64.7% (375 of 590), with no dif-
ference between treatment groups (
who were tested, 107 (28.5%; 95% CI, 24.0%–33.4%) wereHIV-
infected, with no significant difference between both treatment
groups (table 1). HIV coinfection was more prevalent among
migrant workers than among residents (33.7% vs. 17.2%; OR,
2.44;) and was more prevalent among patients withP p .0017
relapse of VL than it was among patients with primary VL
(89.5% vs. 25.3%; OR, 25.1;P ! .0001
the 28 people who died in the SSG group had an unknown
HIV status, as they died before HIV testing. Six (5.6%) of 107
HIV-infected patients received a diagnosis of tuberculosis, and
13 (4.9%) of 268 non–HIV-infected patients and 22 (10.7%)
of 205 patients with unknown HIV status also received a di-
agnosis of tuberculosis.
HIV coinfection was a major determinant of events during
treatment and outcomes (tables 2 and 3) Among non–HIV-
infected patients, there was no significant difference in initial
cure rate, mortality, or initial treatment failure between the
was mainly experienced by HIV-coinfected patients (17.5% vs.
4.6% in non–HIV-infected patients; OR, 4.41;
ilarly, HIV seroprevalence was significantly higher among pa-
tients who experienced initial failure than in patients who ex-
perienced cure (63.2% vs. 26.0%; OR, 4.89;
Tolerability of miltefosineandSSG.
curred during treatment. Death occurred a median of 13 days
);P p .014
P p .001
). Of the 375 patientsP p .30
). Twenty-one (75%) of
). Sim-P p .044
).P p .0001
into treatment (range, 2–30 days), with nosignificantdifference
in time to death between treatment groups. After multivariate
logistic regression, the independent risk factors for death were
determined to be receiving SSG rather than miltefosine (OR,
6.53; 95% CI, 2.53–16.89), being HIV-infected or having an
unknown HIV status(OR, 3.54; 95%CI,1.25–10.06),andvom-
iting (OR, 2.97; 95% CI, 1.28–6.87). Other risk factorsfordeath
(age, body mass index, hemoglobin level,diarrhea,andinability
to walk unaided) were interdependent.
Table 4 shows the incidence of intercurrent events. Vomiting
was more common in the miltefosine group (159 [54.8%] of
290, vs. 93 [32.1%] of 290 in the SSG group; OR, 2.57; P !
) and of slightly longer duration (mean duration, 3.3 vs..0001
2.6 days;). However, vomiting was less severe in mil-P p .02
tefosine patients: only 14 (4.8%) of 290 miltefosine patients
had treatment interrupted for vomiting, compared with 27
(9.7%) of the 290SSG patients (OR, 0.47;
of vomiting inHIV-infectedpatients,comparedwithnon–HIV-
infected patients, were 2.85 (P ! .0001
vomiting was longer (mean duration, 3.5 vs. 2.5 days; P p
). The incidence and duration of diarrhea were similar in.013
both treatment groups, but diarrhea was more commonamong
HIV-infected patients (OR, 2.14;
ference in the incidence of bleeding or pneumonia between the
miltefosine and SSG groups, or betweenHIV-infectedandnon–
HIV-infected patients. One patient discontinued miltefosine
therapy (on day 21) because of an itchy rash; this patient was
lost to follow-up.
Clinical response, as indicated by spleen regression and he-
moglobin level increase, was similar in both treatment groups,
but weight gain was significantly lower in the miltefosinegroup
Final cure at 6 months.
Six months after hospital dis-
charge, 415 patients (79% of treatment survivors) were traced.
For 15 patients (2.8%), death since hospital discharge had been
recorded. All patients who experienced relapse returned for
treatment within 6 months after hospital discharge. There was
). The oddsP p .037
), and the duration of
). There was no dif-P p .015
362 • CID 2006:43 (1 August) • HIV/AIDS
sodium stibogluconate (SSG).
Events during treatment and outcomes of patients with and without HIV coinfection randomized to receive miltefosine or
Non–HIV infectedHIV status unknownAll patients
(n p 63)
(n p 44)
(n p 131)
(n p 137)
(n p 96)
(n p 109)
(n p 290)
(n p 290)
Vomiting41 (65.1)20 (45.5)59 (45.0)26 (19.0) 59 (61.5)47 (43.1)159 (54.8) 93 (32.1)
Duration, days 220.127.116.11.18.104.22.168.58
Drugs withheld, n/N
(%)3/41 (7.3)7/20 (35.0)3/59 (5.1) 3/26 (11.5)8/59 (13.6)17/47 (36.2)14/159 (8.8)27/93 (29.0)
Diarrhea34 (54.0)32 (72.7) 57 (43.5)58 (42.3)58 (60.4)63 (57.8) 149 (51.4)153 (52.8)
Duration, days4.0 22.214.171.124 3.6 4.33.62 3.84
Pneumonia21 (33.3)14 (31.8)31 (23.7)41 (29.9)27 (28.1) 40 (36.7) 79 (27.2)95 (32.8)
Bleeding 10 (15.9)9 (20.5)36 (27.5)30 (21.9) 18 (18.8)25 (22.9) 64 (22.1)64 (22.1)
Initial treatment outcome
Death 1 (1.6) 3 (6.8) 1 (0.8)4 (2.9)4 (4.2) 21 (19.3)6 (2.1)28 (9.7)
Initial cure49 (77.8)40 (90.1)123 (93.8) 130 (94.9) 84 (87.5)84 (77.1)256 (88.3)254 (87.6)
Initial failure11 (17.5) 1 (2.3)6 (4.5)1 (0.7)6 (6.3)0 (0.0)23 (7.9) 2 (0.7)
Discontinuation0 (0.0)0 (0.0)1 (0.8) 0 (0.0)0 (0.0) 0 (0.0)1 (0.3) 0 (0.0)
Default2 (3.2)0 (0.0)0 (0.0)2 (1.5)2 (2.1) 4 (3.7)4 (1.4)6 (2.1)
Death 3 (4.8) 3 (6.8) 1 (0.8)4 (2.9) 4 (4.2) 21 (19.3)8 (2.8) 28 (9.7)
Cure 56 (88.9)40 (90.1)128 (97.7) 131 (95.6)89 (92.7)84 (77.1)273 (94.1)255 (87.9)
Failure1 (1.6)1 (2.3)0 (0.0)0 (0.0)0 (0.0)0 (0.0)1 (0.3)1 (0.3)
Default3 (4.8) 0 (0.0)2 (1.5)2 (1.5)3 (3.1)4 (3.7)8 (2.8)6 (2.1)
Final cure29 (46.0)25 (56.8)99 (75.6)106 (77.4)46 (47.9)58 (53.2)174 (60.0)189 (65.2)
Relapse16 (25.4)5 (11.4) 6 (4.6)0 (0.0)8 (8.3)2 (1.8)30 (10.3) 7 (2.4)
Death7 (11.1)5 (11.4)1 (0.8)6 (4.4)8 (9.4)23 (21.1)17 (5.9)34 (11.7)
Lost to follow-up11 (17.5)9 (20.5)25 (19.1)25 (18.2)33 (34.4)26 (23.9)69 (23.8)60 (20.7)
treatment outcome refers to the outcome after the re-treatment of patients who initially experienced treatment failure, and final outcome refers to the outcome
at the 6-month follow-up visit.
Data are no. (%) of patients, unless otherwise indicated. Initial treatment outcome refers to outcome after the initial course of treatment, end-of-
no difference in follow-up rates between the2 treatmentgroups
(). Final outcomesarepresentedinfigure1andintables
P p .80
2 and 3. The final cure rate in the miltefosine group was 60.0%
(95% CI, 54.1%–65.7%), which was not significantly different
from the rate in the SSG treatment group (cure rate, 65.2%;
95% CI, 59.4%–70.6%) (
P p .23
HIV-infected patients 6 months after treatment was not dif-
ferent between the miltefosine group (cure rate, 75.6%; 95%
CI, 67.3%–82.7%) and the SSG group (cure rate, 77.4%;
69.4%–84.1%) (). Relapse was more common in the
P p .84
miltefosine group (10.3% vs. 2.4% in the SSG group; OR, 5.05;
). This was not wholly caused by HIV coinfection;
P ! .0001
among non–HIV-infected patients, those in the miltefosine
group also had a higher relapse rate (4.6% vs. 0.0% in the SSG
group;). Overall mortality at 6 monthswassignificantlyP p .01
lower in the miltefosine group (5.9% vs. 11.7% in the SSG
group; OR, 0.49; ). When patients lost to follow-upP p .019
are excluded, the final cure rates among non–HIV-infected pa-
). The final cure among non–
tients were 93.4% (95% CI, 86.9%–97.3%) in the miltefosine
group and 94.6% (95% CI, 88.7%–98.0%) in the SSG group.
Of the 30 patients who experienced relapse in themiltefosine
group, 24 experienced cure after re-treatment with a full course
of SSG; in the SSG group, 3 of the 7 patients who experienced
relapse experienced cure after re-treatment. After treatment of
patients with relapse, the final cure rate in the miltefosinetreat-
ment group was 68.3% (198 of 290 patients), compared with
66.2% (192 of 290 patients) in the SSG treatment group (OR,
1.10;). The final death rate after re-treatment of pa-
P p .66
tients with relapse was significantly lower in the miltefosine
group (20 [6.9%] of 290 vs. 37 [12.8%] of 290 in the SSG group;
OR, 0.51; ).
P p .026
No significant difference was observed in final outcomes be-
tween migrant workers and residents, but follow-up rates were
lower among migrant workers than among residents (74.2% vs.
85.3%;). Final outcomes among patients who had re-P p .006
ceived previous treatment for VL were significantly worse, with
HIV/AIDS • CID 2006:43 (1 August) • 363
Table 4. Intercurrent events and response to treatment.
Vomiting, n/N (%)
Duration, mean days ? SD
Drugs withheld, n/N (%)
Diarrhea, n/N (%)
Duration, mean days ? SD
Bleeding, n/N (%)
Pneumonia, n/N (%)
Spleen regression, mean cm ? SD
Hemoglobin increase, mean g/dL ? SD
Weight gain, mean kg ? SD
3.25 ? 2.6
3.62 ? 3.1
2.9 ? 3.1
0.7 ? 2.2
0.13 ? 2.8
2.58 ? 1.8
3.84 ? 3.0
2.9 ? 3.7
0.7 ? 2.3
1.64 ? 3.1
higher relapse rates and death rates. Of previously untreated
patients, 355 (65.0%) of 546 achieved final cure, 41 (7.5%) of
546 died, and 27 (4.9%) of 546 experienced relapse. Among
previously treated patients enrolled in the study, these outcomes
were 8 (24%) of 34, 10 (29%) of 34, and 10 (29%) of 34,
In this, the largest randomized, controlled trial of VL treatment
ever conducted, we have shown that miltefosineisanacceptable
alternative to SSG as treatment for Ethiopian men with VL.
Except for the fact that all were adult males, the patients in
this study were clinically similar in severity to the ∼70,000 pa-
tients with VL treated by Me ´decins Sans Frontie `res-Hollandin
the East African region since 1989, and they were clinicallyvery
different from the patients with VL who have received mil-
tefosine in India [10–14]. For example, the mortality rates in
the Indian miltefosine trials have been !0.2%. Many of our
eligible patients were severely ill with massively enlargedspleens,
anemia, malnutrition, inability to walk unaided, and HIV
coinfection. Death, diarrhea, bleeding, vomiting, and pneu-
monia often complicated their clinical course. We determined
that miltefosine is equivalent to SSG for treatment of VL in
non–HIV-infected patients, and it is probably safer but less
effective for treatment in HIV-coinfectedpatients.Patientswere
treated with a standard miltefosine regimen, as has been rec-
ommended for non–HIV-infected persons in India; optimal
miltefosine dose regimens for African patients with VL and
HIV coinfection still have to be established.
A valid comparison of miltefosine and SSG treatment in
HIV-coinfected patients was made difficult by the fact that 21
(75%) of the 28 patients in the SSG group who died had an
unknown HIV status. HIV testing was done after the patient
was well enough to volunteer for counseling and testing. We
consider it likely that many HIV-coinfected patients receiving
treatment with SSG died before they could be tested for HIV.
The lack of effectiveness of miltefosine in the treatment of
HIV-VL–coinfected patients (defined as experiencing initial
treatment failure and/or relapse) was outweighed, in our view,
by a far lower mortality rate, compared with the mortality rate
in the SSG treatment group. Because VL is currently often
incurable in HIV-coinfected patients and many coinfected per-
sons will ultimately experience relapse , the safety profile of
miltefosine makes it a preferred drug for the treatment of HIV-
The final cure rate at 6 months is probably better than in-
dicated in the intent-to-treat analysis, in which patients lost to
follow-up are counted as having experienced treatment failure;
many of the patients lost to follow-up might, in fact, have been
cured. Considering only patients who could be traced, the final
cure rate among non–HIV-infected patients is ∼94% in both
the miltefosine and the SSG treatment groups.
Our studyconfirms thegoodtolerabilityofmiltefosinefound
in the Indian studies [10–14], with only gastrointestinal symp-
toms being common; these were not prolonged and only of
It has been repeatedly shown that antimonials are poorly
tolerated among patients coinfected with HIV and VL in Eu-
rope [20, 21]. The 6-fold higher odds (Pp.0003) of mortali-
ty associated with SSG treatment, compared with miltefosine
treatment, among patients who were either HIV positive or
whose HIV status was unknown strongly indicates that much
of the mortality among HIV-infected patients was caused by
the SSG treatment itself. The poor final outcomes in patients
who were enrolled in the study as having experienced relapse
after previous treatment may be attributed to the high HIV
coinfection rate (90%).
In contrast to European experience , we found a high
sensitivity of the serological DAT test in HIV-coinfected pa-
tients. This might reflect the fact that, in Ethiopia, patients are
unlikely to survive with very advanced HIV disease, or that
coinfection with L. donovani, being more virulent than Leish-
mania infantum coinfection, occurs at an earlier stage of HIV
364 • CID 2006:43 (1 August) • HIV/AIDS Download full-text
Although it is convenient because it is an oral drug, wide-
spread use of miltefosine for treatment of VL gives cause for
miltefosine is contraindicated during pregnancy, and women
of childbearing age must use effective contraceptionduringand
for 3 months after treatment. Pioneers in the use of miltefosine
in India have recently expressed concern that unsupervised use
of miltefosine in India might lead to high relapse rates . It
is known that HIV-coinfected patients frequently experience
relapse, and when they do, they may be unresponsive to an-
tileishmanial drugs. In Europe, such patients are highly suscep-
tible to infection transmitted by sandflies, particularly if their
CD4+cell count is low . The long serum half-life of mil-
tefosine (7 days) may favor emergence of resistant mutations.
HIV-coinfected patients who have experienced relapse may be-
come an important reservoir of drug-resistant L. donovani, ei-
ther by being parasitaemic or by having post–kala-azar dermal
leishmaniasis. This is of public health concern in Africa, where
transmission of L. donovani is anthroponotic, and resistance
could spread quickly . Therefore, combination therapies
should be considered to delay the emergence of resistance to
miltefosine. Further research is required into drug combina-
tions that might enhance the effectiveness of miltefosine treat-
ment, thus establishing safe and effective drug regimens for
patients with VL in areas with high HIV coinfection rates.
We thank the Me ´decins Sans Frontie `res-Holland teams in the field in
Ethiopia and the staff at Humera Hospital and Mycadra Health Center,
for their commitment to this study; the Tigray Bureau of Health, for sup-
porting the study; Dr. Herbert Sindermann (Zentaris GmbH; Frankfurt,
Germany), for providing the miltefosine, Professor Piet Kager (University
of Amsterdam; Amsterdam, The Netherlands), for the critical review of
this manuscript; andSimon Collin(LondonSchoolofHygieneandTropical
Medicine; London, United Kingdom), for his statistical support.
Zentaris GmbH (provided free miltefosine for the
treatment of 290 study patients); Me ´decins sans Frontie `res-Holland.
Potential conflicts of interest.
All authors: no conflicts.
1. Ritmeijer K, Veeken H, MelakuY,et al. Ethiopianvisceralleishmaniasis:
generic and proprietary sodium stibogluconate are equivalent; HIVco-
infected patients have a poor outcome. Trans Rl Soc Trop Med Hyg
2. Lyons S, Veeken H, Long J. Visceral leishmaniasis and HIV in Tigray,
Ethiopia. Trop Med Int Health 2003;8:733–9.
3. Thakur CP, Sinha GP, Pandey AK, et al. Do the diminishing efficacy
and increasing toxicity of sodium stibogluconate in the treatment of
visceral leishmaniasis in Bihar, India, justify its use as a first-line drug?
An observational study of 80 cases. Ann Trop Med Parasitol 1998;92:
4. Sundar S, More DK, Singh MK, et al. Failure of pentavalent antimony
in visceral leishmaniasis in India: report from the center of the Indian
epidemic. Clin Infect Dis 2000;31:1104–7.
5. Davidson, RN. AIDS and leishmaniasis. Genitourin Med 1997;73:237–9.
6. Laguna F, Adrados M, Alvar J, et al. Visceral leishmaniasis in patients
infected with the human immunodeficiency virus. Eur J ClinMicrobiol
Infect Dis 1997;16:898–903.
7. Laguna F, Lopez-Velez R, Pulido F, et al. Treatment of visceral leish-
maniasis in HIV-infected patients: a randomized trial comparing meg-
lumine antimoniate with amphotericin B. Spanish HIV-Leishmania
Study Group. AIDS 1999;13:1063–9.
8. Croft SL, Neal RA, Thornton EA, Herrmann DB. The activity of alkyl
phosphorylcholines and related derivates against Leishmania donovani.
Biochem Pharmacol 1987;36:2633–6.
9. Kuhlencord A, Maniera T, Eible H, Unger C. Hexadecyl-phospho-
choline: oral treatment of visceral leishmaniasis in mice. Antimicrob
Agents Chemother 1992;36:1630–4.
10. Sundar S, Rosenkaimker F, Makharia MK, et al. Trial of oralmiltefosine
for visceral leishmaniasis. Lancet 1998;352:1821–3.
11. Sundar S, Gupta LB, Makharia MK, et al. Oral treatment of visceral
leishmaniasis with miltefosine. Ann Trop Med Parasitol 1999;93:
12. Jha TK, Sundar S, Thakur CP, et al. Miltefosine, an oral agent, for
treatment of Indian visceral leishmaniasis. N Engl J Med 1999;341:
13. Sundar S, Makharia A, More DK, et al. Short-course of oralmiltefosine
for treatment of visceral leishmaniasis. Clin Infect Dis 2000;31:1110–3.
14. Sundar S, Jha TK, Thakur CP, et al. Oral miltefosine for Indianvisceral
leishmaniasis. N Engl J Med 2002;347:1739–46.
15. Zijlstra EE, el-Hassan AM. Leishmaniasis in Sudan: visceral leishman-
iasis.Trans R Soc Trop Med Hyg 2001;95(Suppl 1):S27–58.
16. World Health Organization (WHO). Manual on visceral leishmaniasis
[WHO/Leish/96.40]. Geneva, Switzerland: WHO, 1996.
17. Me ´decins Sans Frontie `res (MSF). Manual for diagnosis and treatment
of kala-azar under field conditions.Amsterdam,TheNetherlands:MSF,
18. Meredith SE, Kroon NC, Sondorp E, et al. Leish-KIT, a stable direct
agglutination test based on freeze dried antigen for serodiagnosis of
visceral leishmaniasis. J Clin Microbiol 1995;33:1742–5.
19. World Medical Association. Declaration of Helsinki: Ethical principles
for medical research involving human subjects. 2004. Availableat:http:
//www.wma.net/e/policy/pdf/17c.pdf. Accessed October 2002.
20. Delgado J, Macias J, Pineda JA, et al. High frequency of serious side
effects from meglumine antimoniate given without an upper limitdose
for the treatment of visceral leishmaniasisinhumanimmunodeficiency
virus type 1–infected patients. Am J Trop Med Hyg 1999;61:766–9.
21. Pintado V, Martin-Rabadan P, Rivera ML, Moreno S, Bouza E. Visceral
leishmaniasis in human immunodeficiency virus (HIV)-infected and
non–HIV-infected patients. A comparative study. Medicine 2001;80:
22. Medrano FJ, Canavate C, Leal M, Rey C, Lissen E, Alvar J. The role
of serology in the diagnosis and prognosis of visceral leishmaniasis in
patients coinfected with human immunodeficiency virus type-1. Am
J Trop Med Hyg 1998;59:155–62.
23. Sundar S, Murray HW. Availability of miltefosine for treatmentofkala-
azar in India. Bull World Health Organ 2005;83:394–5.
24. Molina R, Lohse JM, Pulido F, Laguna F, Lopez-Velez R, Alvar J. In-
fection of sand flies by humans co-infected with Leishmania infantum
and human immunodeficiency virus. Am J Trop Med Hyg 1999;60:
25. Bryceson A. A policy for leishmaniasis with respect to the prevention
and control of drug resistance. Trop Med Int Health 2001;6:928–34.