Vaccine 27 (2009) 4747–4753
Contents lists available at ScienceDirect
Vaccine
journal homepage: www.elsevier.com/locate/vaccine
Review
Efficacy of killed whole-parasite vaccines in the prevention
of leishmaniasis—A meta-analysis
Sassan Noazin a,∗ , Ali Khamesipour b , Lawrence H. Moulton c , Marcel Tanner d ,
Kiumarss Nasseri e , Farrokh Modabber b,f,h,k , Iraj Sharifi g , E.A.G. Khalil h ,
Ivan Dario Velez Bernal i , Carlos M.F. Antunes j , Peter G. Smith k
a
World Health Organization, 20 Avenue Appia, CH1211, Geneva 27, Switzerland
Center for Research and Training in Skin Diseases & Leprosy, University of Tehran/Medical Sciences, P.O. Box 14155-6383, Tehran, Iran
c
Department of International Health, Johns Hopkins Bloomberg School of Public Health, Baltimore, MD 21205, USA
d
Swiss Tropical Institute, Department of Public Health and Epidemiology, P.O. Box, 4002 Basel, Switzerland
e
Public Health Institute, California Cancer Registry, 3944 State Street, Suite 330, Santa Barbara, CA 93105, USA
f
Drugs for Neglected Diseases Initiative (DNDi), 1 Place St Gervais, CH1201, Geneva, Switzerland
g
Leishmaniasis Research Center, Kerman University of Medical Sciences, P.O. Box 444, Kerman, Iran
h
Institute of Endemic Diseases, University of Khartoum, Khartoum, Sudan
i
Programa de Estudio y Control de Enfermedades Tropicales, PECET, Universidad de Antioquia, Apartado Aéreo 1226, Calle 62 # 52-59 Medellín, Colombia
j
Departamento de Parasitologia, Instituto de Ciencias Biologicas, Universidade Federal de Minas Gerais, Caixa Postal 486, 31270-901 Belo Horizonte, MG, Brazil
k
London School of Hygiene and Tropical Medicine, Keppel Street, London WC1E 7HT, UK
b
a r t i c l e
i n f o
Article history:
Received 6 February 2009
Accepted 31 May 2009
Available online 18 June 2009
Keywords:
Leishmaniasis vaccine
Meta-analysis
Clinical trial
a b s t r a c t
Despite decades of investigation in countries on three continents, an efficacious vaccine against Leishmania infections has not been developed. Although some indication of protection was observed in some of
the controlled trials conducted with “first-generation” whole, inactivated Leishmania parasite vaccines,
convincing evidence of protection was lacking. After reviewing all previously published or unpublished
randomized, controlled field efficacy clinical trials of prophylactic candidate vaccines, a meta-analysis
of qualified trials was conducted to evaluate whether there was some evidence of protection revealed
by considering the results of all trials together. The findings indicate that the whole-parasite vaccine
candidates tested do not confer significant protection against human leishmaniasis.
© 2009 Elsevier Ltd. All rights reserved.
Contents
1.
2.
3.
4.
5.
Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
1.1.
Historical perspective . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
Data selected for analysis . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
2.1.
Definition of studies . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
2.2.
Information sources . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
2.3.
Selection criteria . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
2.4.
Excluded studies . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
2.5.
Included studies . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
Statistical analysis . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
Results . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
Discussion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
Acknowledgements . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
4747
4748
4748
4748
4748
4749
4749
4749
4750
4750
4752
4752
4752
1. Introduction
∗ Corresponding author. Tel.: +41 22 791 2095; fax: +41 22 791 4777.
E-mail addresses: sassanno@yahoo.com, noazins@who.int (S. Noazin).
0264-410X/$ – see front matter © 2009 Elsevier Ltd. All rights reserved.
doi:10.1016/j.vaccine.2009.05.084
Leishmaniasis is endemic in at least 88 countries, some of
which are among the poorest in the world [1,2]. The estimated
global prevalence of all forms of the disease is 12 million, with
Armijos et al. (1998)
Velez et al. (2005)
BCG
BCG
Merthiolate
3-strain vaccine (L.
guyanensis, L.
brasiliensis, L.
amazonensis)
Ecuador2
I
2
Merthiolate
Colombia 3
H
3
L. guyanensis, L.
brasiliensis, and L.
amazonensis
Colombia—exposure
during military
missions
Rural rainforest,
Ecuador
None
Saline
L. panamensis
Endemic and
non-endemic,
army recruits
Endemic
Khalil et al. (2000)
Antunes et al. (1986)
Gedarif, Sudan
Amazonas,
Brazil—exposure
during military
missions
Autoclaved
Merthiolate
Sudan 2
Brazil 1981
and 1983
F
G
2
2
L. major
5-strain vaccine
(species of brasiliensis
and mexicana
complexes, including L.
guyanensis and L.
amazonensis)
L. amazonensis
BCG
None
BCG
Phosphate
buffer + merthiolate
L. donovani
?
Endemic
Endemic and
non-endemic
Endemic
Endemic and
non-endemicarmy
recruits
Bam, Iran
Esfahan, Iran
L. tropica
L. major
BCG
BCG
Autoclaved
Autoclaved
Bam1
Esf1
D
E
1
1
L. major
L. major
BCG
BCG
Endemic
Bam, Iran
L. tropica
BCG
Autoclaved
Bam3
C
3
L. major
BCG
Endemic
Borkhar, Iran
L. major
BCG
Autoclaved
3
L. major
BCG
Endemic
Zavareh, Iran
L. major
BCG
Autoclaved
L. major
3
Inactivation
method
Bor3
Trial protocols and progress reports for studies in Iran, Sudan
and Colombia were reviewed. For each clinical trial identified, the
principal investigator was requested to provide the original trial
database. Thus, individual-level data for trials conducted in Iran
and Sudan as well as aggregated data for the trial in Colombia were
B
2.2. Information sources
Zav3
All published and unpublished field efficacy trials of prophylactic candidate vaccines against leishmaniasis conducted to date
were considered for inclusion. As a result, publication bias is not a
concern in this meta-analysis.
Table 1
Design characteristics of selected trials.
2.1. Definition of studies
Number of
injections
2. Data selected for analysis
Vaccine parasite
Adjuvant
Control treatment
Causative parasite
Area, Country
To date, the only effective way of inducing immunity against
leishmaniasis in humans is provided by leishmanization (LZ), the
practice of injecting live virulent parasites in healthy individuals
[15]. LZ has been practiced historically in high incidence endemic
foci as a means of controlling the timing and site of the initial lesion,
but it is no longer widely used because of rare complications and
difficulties in standardization of the injected parasites [15–18].
During the first half of the 20th century researchers in Latin
America investigated different antigens as potential vaccines
[19,20]. Beginning in the 1970s and 1980s, Mayrink and colleagues
in Brazil and Convit and colleagues in Venezuela experimented with
the use of whole, killed parasites, both for prophylaxis and therapy. Later studies were conducted with inactivated whole-parasite
vaccines in Ecuador (trivalent vaccine composed of three strains of
locally obtained parasites), Colombia (Biobras single strain L. amazonensis vaccine), Iran and Sudan (autoclaved L. major with BCG
included as an adjuvant: ALM + BCG) [21–28]. With the exception
of the trial by Armijos in Ecuador in which a locally prepared vaccine was used [21], none of the other trials demonstrated significant
protection associated with vaccination [26].
Some investigators observed a lower incidence of leishmaniasis
in the subset of those in the vaccinated group whose Leishmanin Skin Test (LST) had converted (from an induration of <5 mm
to >5 mm) after vaccination [24,25,29]. Also, researchers in Iran
observed significant protection in school age boys but not in girls
[27]. Evidence of potential clinical value of such vaccines for treatment, rather than the prevention, of disease was demonstrated in
trials among leishmaniasis patients in the New World [30,31].
We have re-examined the combined data from all except one
published, and unpublished, randomized, controlled clinical trials
(RCT) of prophylactic first-generation leishmaniasis vaccine conducted to date to evaluate whether, overall, there is evidence of
efficacy, or there is efficacy in some sub-groups of the trial populations.
A
1.1. Historical perspective
BCG
Study population
origin
Investigators (year
published)
1.5–2 million cases of cutaneous leishmaniasis (CL) added annually
(with duration of lesions typically from few months to a year)
and 500,000 cases of visceral form of the disease (with duration
of disease from several months to more than a year) [2,3]. Current control measures, including environmental sanitation and
drug treatment of cases, are expensive and cannot be sustained
effectively by poor countries due to the problems of financing and
implementation [4–6]. Moreover, toxicity associated with some of
the most widely available drug treatments, including injections of
pentavalent antimony compounds, and the resistance developed
by the parasite [7–12], underline the need for development of
effective methods of prevention, especially vaccines [4,13,14].
Khamesipour, et al.
(not published)
Khamesipour, et al.
(not published)
Sharifi et al. (not
published)
Sharifi et al. (1998)
Momeni et al. (1999)
S. Noazin et al. / Vaccine 27 (2009) 4747–4753
Study label
4748
57.5
54.3
0.349
53.3
55.3
0.599
0.0
N/A
0.0
0.0
N/A
0.0
(1) Trial objective: Efficacy of a first-generation vaccine for prevention of leishmaniasis in healthy individuals in an endemic
area.
(2) Study design: Randomized, double blind, controlled clinical
trial designed to estimate vaccine efficacy.
(3) Candidate vaccine: Killed, whole Leishmania promastigotes.
(4) Normal field conditions during follow-up: Sample size and
power calculations are generally based on previously observed
disease rates in the trial area. Unforeseen, major changes in
environmental and climatic factors could lead to a significant
change in disease incidence, affecting the study power and conclusions. Selected clinical trials were conducted under the usual
field conditions that gave rise to the previously observed incidence rates.
On the basis of the above criteria, one study [32] was excluded
due to the unusual climatic changes that were attributed by authors
to the El Nino phenomenon during the study follow-up [32]. These
changes led to significantly lower disease incidence rate than
expected.
46.3
52.4
0.002
48.8
49.6
0.302
2.5. Included studies
Data and reports from the randomized, blinded, controlled efficacy trials listed in Table 1 were used. Trial details are presented in
Tables 1–3. Further details are available in Noazin et al. [26].
Table 3
LST conversion 42–80 days post-vaccination (among participants with LST = 0 prior
to vaccination).
V = vaccine; C = control.
a
SEX
% Female
P-value (Fisher exact)
47.1
51.8
0.003
52.7
52.6
0.496
6
6
13
13
8.2
8.7
0.000
5
5
67
72
18.0
18.8
0.118
6
6
15
15
9.1
9.1
0.822
2.4. Excluded studies
50.2
48.9
0.323
–
18.6
7.2
6.5
0.010
<3
2.3. Selection criteria
6
6
12
12
7.4
7.4
0.132
5
5
59
59
19.0
19.2
0.666
<3
obtained. These data were used to estimate or verify the effect
statistics (relative risk) as well as age and gender composition in
Iran studies. For other studies values reported in the published
articles were used (see Section 2.5).
0.0
–
19.8
18.6
–
18.6
18.6
658
4749
0.0
N/A
5.4
5.7
0.271
19.8
406
438
1295
1302
616
644
667
1151
1155
941
1190
N
1838
1795
1124
2149
2068
1107
1084
1055
C
Sudan2
V
C
V
C
Zav3
Bor3
V
C
Bam3
V
C
V
Ca
Esf1
Bam1
Va
Trial arm
Study
Table 2
Age and gender distribution of participants in trials selected for meta-analysis.
Age (years)
Minimum
Maximun
Mean
P-value (Kruskal–Wallis H)
Ecuador2
C
C
V
C
C
Brazil 1983-2
V
Brazil 1981-2
V
Colombia3
V
S. Noazin et al. / Vaccine 27 (2009) 4747–4753
Trial
Trial arm
N
% LST > 5 mm
Bam1
Va
Ca
1807
1761
16.5
3.3
Esf1
V
C
1168
1104
36.2
7.9
Bam3
V
C
1980
1935
18.2
2
Bor3
V
C
608
538
29.9
6.1
Zav3
V
C
772
901
–
–
Sudan2
V
C
1919
1005
30
7
Brazil 1981A-2
V
C
311
–
33
–
Brazil 1981B-2
V
C
338
–
37
–
Brazil 1983-2
V
C
611
–
68
–
Colombia3
V
C
–
–
–
–
Ecuador2
V
C
–
–
–
–
– = not measured.
a
V = vaccine arm; C = control arm.
S. Noazin et al. / Vaccine 27 (2009) 4747–4753
4750
Table 4
Incidence of leishmaniasis among vaccinated and unvaccinated individuals and estimated vaccine efficacy.
Study
Follow-up (months)
Vaccine
Total N
Vaccine
Cases
Control
Total N
Control
Cases
% Case (vaccine)
% Case (control)
Vaccine efficacy
Bam1
Esf1
Bam3
Bor3
Zav3
Sudan2
Brazil 1981A-2
Brazil 1981B-2
Brazil 1983-2
Colombia3
Ecuador2
24
24
24
36a
24
24
12
12
12
12
12
1838
1190
2082
604
742
1155
322
345
658
1302
333
52
214
81
64
102
133
28
4
4
101
7
1795
1124
2008
561
868
1151
289
356
616
1295
316
60
208
93
63
109
141
32
5
8
88
24
2.83
17.98
3.89
10.60
13.75
11.52
8.70
1.16
0.61
7.76
2.10
3.34
18.51
4.63
11.23
12.56
12.25
11.07
1.40
1.30
6.80
7.60
15%
3%
16%
6%
−9%
6%
21%
17%
53%
−14%
72%
a
Although there were 3 years follow-up in Bor3, only cases from years 2 and 3 are included in this analysis because vaccination was completed (3rd dose) at the end of
year 1.
Individual-level data were used for trials A–E, and published
information was used for trials F–I (Table 1).
Although most trials excluded individuals with LST > 0 mm, in
the Borkhar (Bor3) and Zavareh (Zav3) trials (A and B in Table 1)
volunteers with any LST value at screening were enrolled. Since
an LST > 0 could indicate previous exposure to leishmaniasis and
be associated with immunity, for these two trials we analysed
data only on participants with a pre-vaccination LST of zero.
This excluded 12% of trial participants in Zavareh and 40% in
Borkhar.
The study conducted in Brazil in 1981 was conducted in two separate cohorts [29]. These cohorts were different in several respects,
including the risk of disease, duration of exposure and previous vaccination history, and we have treated them as two separate studies
(identified as Brazil 1981A and Brazil 1981B).
3. Statistical analysis
We used a meta-analysis approach based on relative risk
(RR = incidence in the vaccine arm/incidence in the control arm)
calculated for each study separately and then pooled across studies. Briefly, in calculating the pooled effect, an average of the
trial-specific relative risks was calculated by weighting individual study effects according to their trial size (i.e., weighting by
the relative quantity of information provided by the trial). These
weights can be calculated using a variety of methods, including
the inverse variance (I–V) and Mantel–Haenszel (M–H) methods.
If the relative risks in different studies are not widely different
(i.e., studies are homogeneous), a fixed effect model would be
appropriate. If the variation is more than would be expected by
chance then a random effects model is more appropriate, for which
a common method for calculation of the pooled effect is that of
DerSimonian–Laird (D + L) [33]. To assess heterogeneity, we used
a chi-square test of the Q statistic (Q = sum of squared deviations
of weighted RR’s from their overall mean); with degrees of freedom = k − 1, where k is the number of studies. In addition, due
to limitations of Q [34], I-squared was also used to assess heterogeneity. I-squared measures the percent of variation due to
between-studies variability. A value of zero for I-squared indicates that all variability in relative risks is due to sampling error
[34].
EpiInfo 2002, Stata 9 and MS Excel were used in the analyses. The
“metan” program in Stata 9 was used to calculate relative risk (RR)
estimates and corresponding 95% confidence intervals. The inverse
variance (I–V) and the Mantel–Haenszel (M–H) methods were used
separately to fit the fixed effect model and the DerSimonian–Laird
(D + L) method to fit the random effect model.
4. Results
Age and gender breakdown of participants in vaccine trials
included in this analysis are provided in Table 2. Some trials were
confined to children, whereas other included all ages and the trials
among the military in Brazil and Colombia were confined to adult
males. In two of the trials there was a significant excess of males
in the vaccinated group and in two the vaccinees were significantly
younger, on average.
Participants included in the analysis of vaccine immunogenicity were restricted to those with negative pre-vaccination LST. LST
measurements 42–80 days post-vaccination (depending on the
trial) are displayed in Table 3. LST ≥ 5 is generally accepted as an
indication of a significant skin test response in volunteers after vaccination. In all trials where immunogenicity was assessed in both
vaccinated and unvaccinated participants there was a significantly
higher level of skin test conversion among vaccinated individuals. However, the level of skin test conversion varied substantially
among trials, ranging from only 16% among those vaccinated in the
Bam1 trial to 68% in the Brazil 1983-2 trial. In Table 3 and thereafter,
and in our analysis, we have treated the Brazil 1981-2 study as two
distinct trials: Brazil 1981A-2 and Brazil 1981B-2. This approach
was adopted due to the differences between the two cohorts of volunteers in this study in their duration and timing of exposure as well
as the length of time after yellow fever vaccination that the vaccines
were given (which could affect their immunological response) [29].
The incidence rates of leishmaniasis in vaccine and control
arms in the different trials are summarised in Table 4 (participants in Zav3 and Bor3 with pre-vaccination LST ≥ 0 mm were
excluded from this analysis). Vaccine efficacy (VE) is calculated as
(100 × (1 − RR)). The percent of the trial populations who developed disease varied from around 1% to 18%. In only one of the trials
(Ecuador2) was the difference in incidences between the vaccinated
and unvaccinated group statistically significant.
Table 5 shows the confidence intervals on the relative risk (RR)
estimates from the trials and the relative weights derived for each
study according to the method of pooling the results.
The weights assigned to the Ecuador trial vary substantially
between the three estimation methods. This reflects the tendency of
the fixed effect models (I–V and M–H) to give less weight to smaller
trials.
We sought evidence of heterogeneity in the results from the
different trials. The heterogeneity statistics estimated by the three
methods are very similar, as indicated in Table 6 and in no case was
there evidence of significant heterogeneity, providing justification
for using fixed effect models. A comparison of the RR’s from the
Old World and the New World trials indicated such heterogeneity
as there is may be attributed to the latter group.
S. Noazin et al. / Vaccine 27 (2009) 4747–4753
4751
Table 5
Relative risks (incidence in vaccinated/incidence in unvaccinated), 95% confidence intervals and relative weights for each study according to the method used for meta-analysis.
Study
N
RR
95% Conf. Int.
Bam1
Esf1
Bam3
Bor3
Zav3
Sudan2
Brazil 1981A-2
Brazil 1981B-2
Brazil 1983-2
Colombia3
Ecuador2
3633
2314
4090
1165
1610
2306
611
701
1274
2597
649
0.846
0.972
0.84
0.944
1.095
0.94
0.785
0.826
0.468
1.142
0.277
0.587
0.818
0.628
0.679
0.851
0.753
0.485
0.224
0.142
0.867
0.121
Weight (%)
I–V
1.22
1.155
1.124
1.31
1.408
1.174
1.271
3.049
1.547
1.503
0.633
M–H
6.32
28.37
9.94
7.84
13.34
17.08
3.64
0.49
0.59
11.15
1.23
7.26
25.59
11.32
7.81
12.02
16.89
4.03
0.59
0.99
10.55
2.95
100
100
D+L
8.33
19.66
11.41
9.72
13.66
15.63
5.39
0.86
1.02
12.27
2.05
100
Table 6
Heterogeneity and effect statistics in the three methods of meta-analysis.
Method
I–V
M–H
D+L
Heterogeneity statistics
Effect statistics
Test of RR = 1
Chi-square
d.f.
P
I-squared
Pooled RR
95% Conf. Int.
14.59
14.62
14.62
10
10
10
0.148
0.146
0.146
31.5%
31.6%
31.6%
0.947
0.939
0.928
0.864
0.857
0.821
Pooled RR estimates and the 95% confidence interval (CI) estimated by the three methods (Table 6) are very similar, regardless of
the model used, providing little evidence to reject the hypothesis
of no vaccine effect on leishmaniasis incidence.
Fig. 1 shows the “forest plot” of the findings in the trials. The
area of the gray square boxes represent the relative size of each
trial with the centre dot and the line in the centre of each square
representing the RR and its 95% CI. The overall RR is depicted by
two blank diamond boxes, representing the M–H and the D + L estimates.
While Old World trials are clustered around the vertical line of
RR = 1 (i.e., homogeneous but with minimal efficacy), the results
1.038
1.029
1.049
z-Value
P
1.16
1.34
1.2
0.246
0.179
0.231
from the trials conducted in Latin American tend to be scattered on
the left of that line, suggesting more heterogeneity but also more
efficacious results. The Ecuador trial, the only trial with significant
results, is located in the far left of the forest plot. Despite their
lower individual RR values, these trials have limited impact on the
pooled RR due to their smaller sample sizes (and wide confidence
intervals).
A graphical display of the influence of individual trials on the
pooled RR is presented in Fig. 2. This graph shows the values of the
pooled RR, when studies are omitted one at a time. The reference
line is the overall, pooled RR. Thus, the pooled RR is lower when
Esf1, Zav3 or Colombia3 trials are omitted and the reverse is the case
Fig. 1. Forest plot of vaccine efficacy measures in different leishmaniasis vaccine trials.
4752
S. Noazin et al. / Vaccine 27 (2009) 4747–4753
Fig. 2. Influence of individual trials on pooled RR. The circles indicate the pooled relative risk estimate when each individual trial is omitted. 95% confidence intervals are
also shown.
when any one of the Bam trials, Brazil 1981A-2 and the Ecuador2
are omitted.
5. Discussion
In some trials, vaccinated participants who skin test converted
following vaccination were reported to have a lower incidence of
leishmaniasis than other trial participants [24,25,29]. However, our
meta-analysis clearly demonstrates the overall inability of firstgeneration leishmaniasis vaccines evaluated to date in phase 3
clinical trials to protect vaccinated individuals against infection by
the Leishmania parasite.
The apparent absence of efficacy of these vaccines may be due
to a number of potential factors. First, the immune stimulation
provided by a single dose, or even multiple doses, of inactivated
parasite antigen, even when mixed with BCG as an adjuvant, may
be inadequate. Secondly, BCG was used in the control arm in several of the studies and was also used as a vaccine adjuvant. BCG
stimulates Th1 response and contributes to the immunogenicity
of the vaccine [35]. However, when used as the clinical trial control vaccine (for blinding), and as a vaccine adjuvant, it induces
Th1 response in both arms, thus making it potentially more difficult for any potential vaccine effect to be detected. To the extent
that BCG alone might protect against leishmaniasis, the difference in incidence between the two study arms would be reduced
and the statistical power of the study would be compromised.
Thirdly, LST is an imprecise and highly variable indicator of previous exposure to leishmaniasis. This could lead to misclassification
of some individuals with previous exposure and immunity as unexposed and allow their inclusion in both arms of the clinical trial. If
such persons are still at some risk of leishmaniasis, but the vaccine confers no additional protection in such partially immune
individuals, then the protective efficacy of the vaccine in “unexposed” individuals may be underestimated. Fourthly, it is possible
that some genetically non-responsive volunteers in endemic areas
would show no LST reaction while they could have been exposed
and possibly immune. To the extent that this occurs, the resulting misclassification would contribute to a reduced difference
between the two arms and contribute to misleading efficacy estimation.
Our use of the meta-analytic approach is subject to some limitations. The different vaccine candidates used in the trials were
similar in their dependence on killed parasites, but the composition of the vaccines varied between trials, BCG was used as an
adjuvant in some cases and the ecological setting of the different
trials varied substantially. Thus, it may be argued that combining the results from the trials should be done with great caution.
We would not disagree with this view but have combined the
findings to seek evidence that might encourage further work on
first-generation vaccines. In this respect, our findings are depressing and suggest that other approaches to leishmaniasis vaccine
development should be vigorously pursued.
Acknowledgements
The support to this study provided by the Swiss Tropical Institute
and by Dr Marie-Paule Kieny, Director, WHO/Initiative for Vaccine
Research (IVR) is greatly appreciated. The valuable contributions
made by investigators and trial participants in many countries in
the world, including Brazil, Colombia, Iran, Sudan, Venezuela and
others towards the development of a vaccine against leishmaniasis
is acknowledged.
References
[1] Davies CR, Kaye P, Croft SL, Sundar S. Leishmaniasis: new approaches to disease
control. BMJ 2003;326(7385):377–82.
[2] WHO. Leishmaniasis Home Page. World Health Organization website, 821-2006, http://www.who.int/leishmaniasis/burden/en/. (GENERIC) Ref Type:
Electronic Citation.
[3] WHO. Urbanization: an increasing risk factor for leishmaniasis (Weekly Epidemiological Report), vol. 44. Geneva: World Health Organization; 2002. p. 365
[The Weekly Epidemiological Record (WER)].
[4] Dantas-Torres F, Brandao-Filho SP. Visceral leishmaniasis in Brazil: revisiting paradigms of epidemiology and control. Rev Inst Med Trop Sao Paulo
2006;48(3):151–6.
[5] Gafurov IM. Experience in controlling and preventing zoonotic cutaneous leishmaniasis in Uzbekistan. Med Parazitol (Mosk) 1999;(1):58–9.
[6] Thakur CP, Kumar K. Post kala-azar dermal leishmaniasis: a neglected aspect of
kala-azar control programmes. Ann Trop Med Parasitol 1992;86(4):355–9.
[7] Bryceson A. A policy for leishmaniasis with respect to the prevention and control of drug resistance. Trop Med Int Health 2001;6(11):928–34.
[8] Croft SL, Coombs GH. Leishmaniasis—current chemotherapy and recent
advances in the search for novel drugs. Trends Parasitol 2003;19(11):502–8.
S. Noazin et al. / Vaccine 27 (2009) 4747–4753
[9] Croft SL, Sundar S, Fairlamb AH. Drug resistance in leishmaniasis. Clin Microbiol
Rev 2006;19(1):111–26.
[10] Dube A, Singh N, Sundar S. Refractoriness to the treatment of sodium stibogluconate in Indian kala-azar field isolates persist in in vitro and in vivo
experimental models. Parasitol Res 2005;96(4):216–23.
[11] Hadighi R, Mohebali M, Boucher P, Hajjaran H, Khamesipour A, Ouellette M.
Unresponsiveness to glucantime treatment in Iranian cutaneous leishmaniasis
due to drug-resistant Leishmania tropica parasites. PLoS Med 2006;3(5):659–67.
[12] Khalil EA, el Hassan AM, Zijlstra EE, Hashim FA, Ibrahim ME, Ghalib HW,
et al. Treatment of visceral leishmaniasis with sodium stibogluconate in
Sudan: management of those who do not respond. Ann Trop Med Parasitol
1998;92(2):151–8.
[13] Ahluwalia IB, Bern C, Costa C, Akter T, Chowdhury R, Ali M, et al. Visceral leishmaniasis: consequences of a neglected disease in a Bangladeshi community.
Am J Trop Med Hyg 2003;69(6):624–8.
[14] WHO/TDR. Generating evidence to improve the effectiveness of control programmes (Report of the Scientific Working Group on Leishmaniasis). Geneva,
Switzerland. World Health Organization on behalf of the Special Programme
for Research and Training in Tropical Diseases; 2004.
[15] Khamesipour A, Dowlati Y, Asilian A, Hashemi-Fesharki R, Javadi A, Noazin S, et
al. Leishmanization: use of an old method for evaluation of candidate vaccines
against leishmaniasis. Vaccine 2005;23(28):3642–8.
[16] Nadim A, Javadian E, Tahvildar-Bidruni G, Ghorbani M. Effectiveness of leishmanization in the control of cutaneous leishmaniasis. Bull Soc Pathol Exot
Filiales 1983;76(4):377–83.
[17] Nadim A, Javadian E, Mohebali M. The experience of leishmanization in the
Islamic Republic of Iran. East Mediterr Health J 1997;3(2):284–9.
[18] Khamesipour A, Rafati S, Davoudi N, Maboudi F, Modabber F. Leishmaniasis vaccine candidates for development: a global overview. Indian J Med Res
2006;123(3):423–38.
[19] Genaro O, de Toledo VP, da Costa CA, Hermeto MV, Afonso LC, Mayrink
W. Vaccine for prophylaxis and immunotherapy, Brazil. Clin Dermatol
1996;14(5):503–12.
[20] Handman E. Leishmaniasis: current status of vaccine development. Clin Microbiol Rev 2001;14(2):229–43.
[21] Armijos RX, Weigel MM, Aviles H, Maldonado R, Racines J. Field trial of a vaccine against New World cutaneous leishmaniasis in an at-risk child population:
safety, immunogenicity, and efficacy during the first 12 months of follow-up. J
Infect Dis 1998;177(5):1352–7.
[22] Bahar K, Dowlati Y, Shidani B, Alimohammadian MH, Khamesipour A, Ehsasi
S, et al. Comparative safety and immunogenicity trial of two killed Leishmania major vaccines with or without BCG in human volunteers. Clin Dermatol
1996;14(5):489–95.
[23] Dowlati Y, Ehsasi S, Shidani B, Bahar K. Stepwise safety trial of a killed Leishmania
vaccine in Iran. Clin Dermatol 1996;14(5):497–502.
4753
[24] Khalil EA, el Hassan AM, Zijlstra EE, Mukhtar MM, Ghalib HW, Musa B, et
al. Autoclaved Leishmania major vaccine for prevention of visceral leishmaniasis: a randomised, double-blind, BCG-controlled trial in Sudan. Lancet
2000;356(9241):1565–9.
[25] Momeni AZ, Jalayer T, Emamjomeh M, Khamesipour A, Zicker F, Ghassemi
RL, et al. A randomised, double-blind, controlled trial of a killed L. major
vaccine plus BCG against zoonotic cutaneous leishmaniasis in Iran. Vaccine
1999;17(5):466–72.
[26] Noazin S, Modabber F, Khamesipour A, Smith PG, Moulton LH, Nasseri K, et al.
First generation leishmaniasis vaccines: a review of field efficacy trials. Vaccine
2008;26(52):6759–67.
[27] Sharifi I, FeKri AR, Aflatonian MR, Khamesipour A, Nadim A, Mousavi MR,
et al. Randomised vaccine trial of single dose of killed Leishmania major
plus BCG against anthroponotic cutaneous leishmaniasis in Bam, Iran. Lancet
1998;351(9115):1540–3.
[28] Velez ID, Gilchrist K, Arbelaez MP, Rojas CA, Puerta JA, Antunes CMF, et al.
Failure of a killed Leishmania amazonensis vaccine against American cutaneous leishmaniasis in Colombia. Trans R Soc Trop Med Hyg 2005;99(8):
593–8.
[29] Antunes CM, Mayrink W, Magalhaes PA, Costa CA, Melo MN, Dias M, et al. Controlled field trials of a vaccine against New World cutaneous leishmaniasis. Int
J Epidemiol 1986;15(4):572–80.
[30] Convit J, Castellanos PL, Rondon A, Pinardi ME, Ulrich M, Castes M, et al.
Immunotherapy versus chemotherapy in localised cutaneous leishmaniasis.
Lancet 1987;1(8530):401–5.
[31] Machado-Pinto J, Pinto J, da Costa CA, Genaro O, Marques MJ, Modabber F, et
al. Immunochemotherapy for cutaneous leishmaniasis: a controlled trial using
killed Leishmania (Leishmania) amazonensis vaccine plus antimonial. Int J Dermatol 2002;41(2):73–8.
[32] Armijos RX, Weigel MM, Calvopina M, Hidalgo A, Cevallos W, Correa J. Safety,
immunogenecity, and efficacy of an autoclaved Leishmania amazonensis vaccine plus BCG adjuvant against New World cutaneous leishmaniasis. Vaccine
2004;22(9–10):1320–6.
[33] Deeks J, Altman D, Bradburn MJ, Egger M, Davey Smith G, Altman D, editors.
Systematic reviews in health care: meta-analysis in context. 5. Statistical methods for examining heterogeneity and combining results from several studies in
meta-analysis. 2nd ed. London: BMJ Publishing Group; 2005. p. 43–68.
[34] Huedo-Medina TB, Sanchez-Meca J, Marin-Martinez F, Botella J. Assessing
heterogeneity in meta-analysis: Q statistic or I-2 index? Psychol Methods
2006;11(2):193–206.
[35] Alimohammadian MH, Khamesipour A, Darabi H, Firooz A, Malekzadeh S,
Bahonar A, et al. The role of BCG in human immune responses induced by multiple injections of autoclaved Leishmania major as a candidate vaccine against
leishmaniasis. Vaccine 2002;21(3–4):174–80.
