Upload
others
View
2
Download
0
Embed Size (px)
Citation preview
WHO Advisory Committee on serological responses to Expanded Programme on Immunization vaccines in infants receiving Intermittent Preventive Treatment for malaria (IPTi)
FINAL REPORT
October 8, 2009
Contents
Executive summary ...........................................................................................................1
Background ..............................................................................................................3
EPI Serology Study Design...............................................................................................5
Study sites and study population .................................................................................5
Study design ..............................................................................................................6
WHO Advisory Committee .......................................................................................11
Results – Pooled Analysis ...............................................................................................12
Pooled SP vs Placebo comparison for measles: ........................................................12
Pooled drugs (SP, LapDap, SP-ART, AQ-ART, MQ) vs placebo comparison for
measles ............................................................................................................13
Pooled Analysis for other antigens............................................................................15
Results – Yellow Fever (Navrongo) ...............................................................................16
Conclusions of the WHO Advisory Committee............................................................18
Annex 1: Report of the Technical Consultation on Intermittent Preventive
Treatment in Infants (IPTi), Technical Expert Group (TEG) on Preventive
Chemotherapy, April 23-24, 2009 Geneva.. ..........................................................19
Annex 2: Membership of WHO Advisory Committee on serological responses to
Expanded Programme on Immunization vaccines in infants receiving
Intermittent Preventive Treatment for malaria...................................................20
Annex 3: Final Pooled Analysis: Assessment of Serological Responses to Expanded
Programme on Immunization Vaccines in Infants Receiving Intermittent
Preventive Treatment (v.3 submitted July 3, 2009) .............................................21
Annex 4: Summaries of Individual Study Results .......................................................22
Navrongo, Ghana.......................................................................................................22
Manhiça, Mozambique ..............................................................................................24
Bungoma, Kenya .......................................................................................................26
Kisumu, Kenya ..........................................................................................................29
Kilimanjaro, Tanzania ...............................................................................................32
1
Executive summary
Intermittent Preventive Treatment in infants (IPTi) is the administration of a full course
of an antimalarial treatment to infants at specified time points, regardless of the presence
of parasites. The objective of IPTi is to reduce the infant malaria burden in the first year
of life. Administering IPTi at the time of routine immunization is proposed to be the main
delivery strategy. It is therefore of critical importance to confirm that IPTi does not have
any adverse interaction with the serological responses to EPI vaccines.
As a partner of the IPTi Consortium, and with funding from the Bill & Melinda Gates
Foundation, WHO undertook serology assessments in five IPTi efficacy trials conducted
in Ghana (Navrongo), Kenya (Bungoma and Kisumu), Mozambique (Manhica), and
Tanzania (Kilimanjaro). Infants eligible for the serology studies were selected from the
larger study population at each site.
To oversee and guide this work, in 2003 WHO established a 5-member, independent Ad
Hoc Advisory Committee (see Annex 2 for Terms of Reference and Membership). The
Advisory Committee assisted with the design of the project, selection of the sub-
contractors, and review of all the data arising from five EPI serology studies. The
Advisory Committee guided WHO accordingly, and issued an Interim Report in July
2006. The Interim Report, based on data at the time from the Navrongo (Ghana), Manhica
(Mozambique) and Bungoma (Kenya) studies, concluded that IPTi-SP did not have an
adverse impact on serological responses to vaccination against measles, diphtheria,
tetanus, pertussis, polio serotypes 1 and 3, Haemophilus influenzae type b and hepatitis B.
The studies using other drug combinations and analysis of all the samples have now been
completed and reviewed. This Final Report of the Advisory Committee describes the
methodology of the IPTi serology studies, summarizes the results of the pooled statistical
analysis, and presents the Advisory Committee's final conclusions based on a compilation
of all the data available. The report is primarily based on the final pooled analysis of the
results, and supported by selected analyses of the individual trials where no pooled
analyses were planned (i.e. yellow fever and Hib vaccines).
While it cannot be held responsible for the conduct of the field trials or laboratory work,
the Advisory Committee concludes the following:
1. Serological data from studies in Navrongo, Manhica, and Kilmanjaro provide strong
evidence that IPTi with SP, does not have an adverse impact on serological responses
to measles vaccine;
2. Though very limited, the available serological data from Navrongo provides evidence
that SP does not have a negative impact on antibodies following vaccination against
yellow fever;
3. Serological data from studies in Kisumu and Kilimanjaro provide strong evidence
that IPTi with SP-ART, AQ-ART, MQ, or LapDap do not have an adverse impact on
serological responses to measles vaccination;
2
4. Serological data from Manhica and Kisumu provide strong evidence that IPTi with
SP, SP-ART, AQ-ART, or LapDap do not have an adverse impact on serological
responses to DTP, polio, Hib, and HepB vaccines.
5. The pooled analyses provide further evidence that IPTi treatments do not impair
serological responses to EPI antigens.
Overall conclusion:
Studies have demonstrated that there is no adverse impact on the serological responses to
DTP, polio, Hib, HepB, and measles vaccines when the IPTi drugs SP, SP-ART, AQ-
ART, and LapDap are administered to infants at the time of routine vaccination.
Concomitant administration of IPTi-SP at the time of yellow fever vaccination, and IPTi-
MQ at the time of measles vaccination, have also been shown to have no negative effect
on the serological responses.
3
Background
Intermittent Preventive Treatment in infants (IPTi), is the administration of a full course
of an antimalarial treatment to infants at specified time points, regardless of the presence
of parasites. The objective of IPTi is to reduce the infant malaria burden in the first year
of life. Administering IPTi at the time of routine immunization is one of the delivery
strategies.
Over the past 10 years, IPTi has been the focus of a number of research efforts. With
funding from the Bill & Melinda Gates Foundation, in 2002 the IPTi Consortium (a
collaboration of 17 research institutions plus WHO and UNICEF) was formed to
undertake a coordinated and comprehensive research agenda on IPTi in order to inform
policy development1. This has included clinical trials in southern-African countries
(Gabon, Ghana, Kenya, Tanzania, Mozambique) and large-scale pilot implementation
studies involving over 300,000 infants a year in districts of Benin, Ghana, Madagascar,
Mali, Malawi, Tanzania, and Senegal.
The available evidence on the safety, efficacy and other relevant aspects of IPTi with
sulfadoxine-pyrimethamine (SP-IPTi) delivered through the Expanded Programme on
Immunization (EPI) has been reviewed independently in 2008 by the Institute of
Medicine (IOM)2 and three times by WHO's Technical Expert Group (TEG) on
Preventive Chemotherapy3 most recently in April 2009 (Annex 1).
On the basis of a pooled analysis of 6 published studies4, IPTi-SP was found to be safe
and decreased:
• incidence of clinical malaria episodes by 30% (95% CI: 19.8%; 39.4%) (similar to
the levels of efficacy observed with the use of insecticide-treated bednets);
• anaemia (<8g/dl) overall by 21.3% (95% CI: 8.3%; 32.5%);
• all cause hospital admissions in the first year of life by 23% (95% CI: 10.0%; 34.0%)
(noting that admissions were not, however, all due to severe malaria).
1 See web site: www.ipti-malaria.org and IPTi Fact Sheet (Feb. 2009) for more information.
2 Institute of Medicine. Assessment of the Role of Intermittent Preventive Treatment for Malaria in
Infants: Letter Report. Committee on the Perspectives on the Role of Intermittent Preventive
Treatment for Malaria in Infants, July 11, 2008.
(http://www.iom.edu/CMS/3783/48783/56178,aspx)
3 WHO. Report of the Technical Consultation on Intermittent Preventive Treatment in Infants (IPTi),
Technical Expert Group (TEG) on Preventive Chemotherapy, April 23-24, 2009 Geneva. (Annex 1).
4 Aponte, John. J, Schellenberg, David, et al. Intermittent Preventive Treatment for Malaria Control in
African Infants: Pooled analysis of safety and efficacy in six randomized controlled trials. Accepted for
publication by Lancet 2009, forthcoming.
4
Both the IOM and WHO's TEG have recommended that IPTi-SP be considered for
implementation through EPI as an additional malaria control intervention in countries in
Africa south of the Sahara with moderate to high malaria transmission, and where drug
resistance to SP is low.
As part of the IPTi Consortium research to support this policy recommendation, since
2002 WHO has been coordinating a project designed to investigate the impact of IPTi
with SP and a range of other antimalarial drugs5 on infant serological responses to EPI
vaccines. Confirmation that IPTi does not have any adverse interaction with the
serological responses to vaccination was critical to ensuring that EPI could safely be used
as the delivery mechanism for IPTi. Early results from the first IPTi-SP trial in Ifakara,
Tanzania6 raised concerns that seropositivity to measles and pertussis vaccines was lower
in infants receiving IPTi compared to placebo.
To oversee and guide this area of work, WHO established a 5-member, independent Ad
Hoc Advisory Committee (see Annex 2 for Terms of Reference and Membership). The
Advisory Committee assisted with the design of the project, selection of the sub-
contractors, and review of all the data arising from five EPI serology studies. The
Advisory Committee guided WHO accordingly, and issued an Interim Report in July
20067. The Interim Report, based on data at the time from the Navrongo (Ghana),
Manhica (Mozambique) and Bungoma (Kenya) studies, concluded that IPTi-SP did not
have an adverse impact on serological responses to vaccination against measles,
diphtheria, tetanus, pertussis, polio serotypes 1 and 3, Haemophilus influenzae type b and
hepatitis B.
This Final Report of the Advisory Committee describes the methodology of the IPTi
serology studies, summarizes the results of the pooled statistical analysis, and presents the
Advisory Committee's final conclusions based on all the data available.
5 See Table 1 for list of drugs studied and their abbreviations in the footnote of the table.
6 Schellenberg, D. et al. (2001). Intermittent treatment for malaria and anaemia control at time of routine
vaccinations in Tanzanian infants: a randomized, placebo-controlled trial. Lancet, 357: 1471-1477.
7 The Interim Report summarized the results as of 2006 from Navrongo, Manhica and Bungoma, a pooled
analysis of Navrongo and Manhica data, and the conclusions of the Advisory Committee. This Final
Report includes all data from all trials and as such supercedes the analysis of the Interim Report.
5
EPI Serology Study Design
Study sites and study population
The EPI serology work was designed as "nested sub-studies" within five randomized
controlled studies of the IPTi Consortium and one other on-going trial to assess the
protective efficacy of IPTi against episodes of clinical malaria and anemia. The EPI
serology assessments were included in the IPTi trials conducted in Ghana (Navrongo)8,
Kenya (Bungoma9 and Kisumu), Mozambique (Manhica), and Tanzania (Kilimanjaro).
Infants eligible for the serology studies were selected10
from the larger study population
at each site.
Infants were randomized into: i) placebo or SP in Navrongo, Manhica and Bungoma, ii)
placebo, SP-ART, AQ-ART and LapDap in Kisumu and iii) placebo, SP, MQ and
LapDap in Kilimanjaro. All sites performed serological testing for measles vaccine. Only
Navrongo undertook serological assessment for yellow fever vaccine. Bungoma,
Manhica, and Kisumu performed serological testing on all other antigens.
Table 1 summarizes the specific details of each of the serological studies – including the
drugs used, the IPTi dosing schedule, sample size, timing of blood samples, and EPI
antigens assessed.
In each of the serological studies, IPTi (using a number of antimalarial drug options) or
placebo was administered on three occasions11
during the first year of life at the time of
routine EPI vaccination (doses 2 and 3 of diphtheria, tetanus and pertussis (DTP); and
measles vaccination). Detailed information was obtained on the vaccination status
(including BCG and birth doses of polio) of each infant recruited to the serology studies.
8 Note: The Navrongo study was nearly completed when the protocol for the EPI serology studies was
designed. Since Navrongo provided a unique opportunity to obtain information on serological
responses to yellow fever vaccination, a decision was made to retrospectively measure the serological
responses on a selection of blood samples already taken.
9 This trial was supported by WHO/TDR and not part of the IPTi Consortium.
10 Notes on sample selection: Navrongo: As this trial was already completed, the EPI serology
study used stored samples. Those children with pre and post vaccination paired samples of
sufficient volume were selected. Bungoma: Samples from all children in the study were used unless
there was insufficient volume or sample was not taken. Kisumu: This study was designed with a
sample size of 379 per arm at the beginning. When the serology study was proposed the trial was
unable to increase the sample size to 500 per arm for budgetary and operational reasons. The
serology study used samples collected from all children unless the sample was of insufficient
volume or missed. Manhica: The trial was underway before the EPI serology study was developed.
Serological responses to EPI vaccines were assessed in a subsample of study infants consecutively
selected as they came to the clinic until the estimated sample size of 300 children per arm (for the
main trial) was completed.
11 The Navrongo trial included a fourth dose of IPTi at 12 months of age which was not linked with EPI.
6
All studies were approved by the relevant local and international (including WHO)
Ethical Review Boards. Written informed consent was obtained from the parents or
guardians of all participating infants.12
Study design
The serological assessments were designed as non-inferiority studies, with the objective
to demonstrate that IPTi does not reduce serological responses to EPI vaccines by more
than a proportion deemed to be clinically important.
Primary endpoint: The primary endpoint was serological responses to measles vaccine. If
the proportion of infants attaining protective levels (> 120IU/L) of measles antibody post-
vaccination was reduced by ≤ 5% in the IPTi group compared to the placebo group, it
would not be considered that IPTi had an adverse impact on serological responses to
measles. The sample size of 50013
per arm was calculated with the aim of rejecting, with
80% power and a one-sided significance level of 5%, the null hypothesis that the
difference between the groups was > 5%.
Secondary endpoints: The secondary endpoints were serological responses to all other
EPI vaccines. If the proportion of infants attaining protective levels (see Table 3) of
antibody post-vaccination was reduced by ≤ 10% in the IPTi group compared to the
placebo group, it would not be considered that IPTi had an adverse impact on serological
responses to EPI vaccines. The sample size of 250 per arm was calculated with the aim of
rejecting, with 80% power and a one-sided significance level of 5%, the null hypothesis
that the difference between the groups was > 10%.
Plotting of reverse cumulative distribution functions (RCDF) of the proportion of infants
attaining different antibody titres was an important graphic tool for visualizing the full
distribution of data and comparison of results. Close agreement of RCDFs from infants
given IPTi with those given placebo supports the conclusion that antibody responses are
similar in the two populations over the full range of antibody concentrations. Post-
vaccination geometric mean concentrations in the treatment and placebo groups were
compared.
12 See Annex 4 page 27 for details about Bungoma trial having to re-do written consent retrospectively. 13 At the outset the Kisumu sample size was planned to be less (379) because they knew it was not possible to
obtain a target of 500 before the conclusion of the trial.
7
Table 1. IPTi studies that include assessment of serological responses to EPI vaccines
Study site
Study
design
Drug(s)
for IPTi
Age at IPTi drug dosing
Sample size
EPI immunization schedule
Timing of
blood samples
Serological information
available from each
study
Navrongo,
Ghana
RCT
SP
Placebo
10 weeks (DTP2)
14 weeks (DTP3)
9 months (measles)
12 months
500 SP
500 placebo
BCG: birth
DTP, hepB, Hib: 6, 10, 14
weeks
Polio: birth, 6, 10, 14 weeks
Measles, yellow fever: 9
months
9 months
12 months
Measles, yellow fever
Bungoma,
Kenya
RCT
SP
Placebo
10 weeks (DTP2)
14 weeks (DTP3)
9 months (measles)
500 SP
500 placebo
BCG: birth
DTP, hepB, Hib: 6, 10, 14
weeks
Polio: birth, 6, 10, 14 weeks
Measles: 9 months
6 weeks
18 weeks
9 months
10 months
DTP, polio, hepatitis B,
Hib, measles
8
Manhica,
Mozambique
RCT
SP
Placebo
12 weeks (DTP2)
16 weeks (DTP3)
9 months (measles)
500 SP
500 placebo
BCG: birth
DTP, hepB: 8, 12, 16 weeks
Polio: birth, 8, 12, 16 weeks
Measles: 9 months
20 weeks
9 months
12 months
DTP, polio, hepatitis B
measles
Kisumu, Kenya
RCT SP-ART
AQ-ART
LapDap
Placebo
10 weeks (DTP2)
14 weeks (DTP3)
9 months (measles)
379 x 3 intervention
379 placebo
As for Bungoma
6 weeks
18 weeks
9 months
12 months
DTP, polio, hepatitis B,
Hib, measles
Kilimanjaro,
Tanzania
RCT
SP
MQ
LapDap
Placebo
8 weeks (DTP2)
12 weeks (DTP3)
9 months (measles)
500 x 3 intervention
500 placebo
BCG: birth
DTP, hepB: 4, 8, 12 weeks
Polio: birth, 4, 8, 12 weeks
Measles: 9 months
9 months
10 months
Measles
Legend: RCT: randomised controlled trial; SP: sulfadoxine-pyrimethamine; SP-ART: sulfadoxine-pyrimethamine plus artesunate; MQ: mefloquine; LapDap: chlorproguanil-
dapsone; AQ-ART: amodiaquine plus artesunate; DTP: diphtheria, tetanus, pertussis; hepB: hepatitis B; Hib: Haemophilus influenzae type b
9
It was planned that each study site would attain the requisite sample size for the primary and
secondary endpoints, and that data from all sites be pooled to provide overall summary
measures of even greater precision.
Blood sampling: A minimum of 0.5mls of whole blood was taken by finger prick or venous
sampling at intervals indicated in Table 2. Samples were centrifuged, and sera frozen at
minus 20 degrees Centigrade.
Serological assays: Following a tendering process, the WHO Advisory Committee assessed
quotations from four internationally recognized public health reference laboratories for
carrying out the serological assays for these studies. The Health Protection Agency (HPA),
UK, was chosen on the basis of its competitive pricing, expertise in functional (plaque
reduction neutralization) assays, and capacity to deal with large numbers of samples within
the requisite timeframe. The laboratory was blind to the allocation status (IPTi or placebo)
of all samples. Geometric Mean Titres (GMTs) were measured by plaque reduction
neutralization (measles and yellow fever), microneutralization (polio serotypes 1 and 3) and
enzyme linked immunosorbent assay (ELISA - all other EPI antigens). All assays were run
in duplicate, using standardized reagents or validated test kits. For measles and yellow fever,
pre-vaccination blood samples were assayed to check whether subjects had been exposed to
these diseases prior to vaccination. For all other EPI antigens, measurement was confined to
the post-vaccination blood sample14
. In instances where there was insufficient blood to carry
out all of the assessments, assays were carried out in the following order:
18–20 week sample
i. Haemophilus influenzae type b
ii. Diphtheria
iii. Polio serotype 3
iv. Hepatitis B
v. Pertussis toxin
vi. Tetanus
vii. Polio serotype 1
viii. Pertussis filamentous haemagglutinin
10–12 month sample
i. Measles
ii. Yellow fever
14 Pre-vaccination samples were collected, but as these were likely to contain maternal antibodies they were
stored in the event that that the post-vaccination results were equivocal and analysis of the pre-vaccination
samples was needed.
10
Table 2. Serological assays
Vaccine Pre or post-
vaccination
sample
Timing of blood
samples
Type of test Protective level
Measles
Pre
9 months Plaque reduction
neutralisation (PRN) 1-2
dilutions
To check for presence
of measles antibodies
pre-vaccination
Post 10 months or 12
months
PRN 6 dilutions (GMT) 120 IU/l
Yellow fever Pre 9 months PRN 1-2 dilutions To check for presence
of YF antibodies pre-
vaccination
Post 10 months or 12
months
PRN 6 dilutions (GMT) 1:5 (PRN titre)
Diphtheria Pre At time of DTP1 Store in freezer*
Post One month post
DTP3
Quantitative ELISA
(GMT)
0.1 IU/ml
Tetanus Pre At time of DTP1 Store in freezer*
Post One month post
DTP3
Quantitative ELISA
(GMT)
0.1 IU/ml
Pertussis (PT and
FHA only)
Pre At time of DTP1 Store in freezer*
Post One month post
DTP3
Quantitative ELISA
(GMT)
Protective levels not
defined
Polio (serotypes 1
and 3 only)
Pre At time of DTP1 Store in freezer*
Post One month post
DTP3
PRN 6 dilutions (GMT) 1:8 (PRN titre)
Haemophilus
influenzae type b
Pre At time of DTP1 Store in freezer*
Post One month post
DTP3
Quantitative ELISA
(GMT)
0.15 µg/ml (1.0 µg/ml
will be used as a
secondary descriptive)
Hepatitis B Pre At time of DTP1 Store in freezer*
Post One month post
DTP3
Quantitative ELISA
(GMT)
10 IU/l
* Pre-vaccination GMTs will only be obtained if the post-vaccination results are equivocal; PRN: Plaque reduction neutralisation; GMT: Geometric mean titre; PT: Pertussis toxin; FHA: Filamentous haemagglutinin
11
This ranking was based on the relative immunogenicity of each antigen, with those of lower
immunogenicity being assigned a higher rank, so as to increase the likelihood of detection of
any potential interference. Haemophilus influenzae type b (Hib) was assigned top rank since
regulatory authorities are particularly concerned about serological responses to Hib vaccine.
Interference with responses to this vaccine have been demonstrated in different settings,
particularly in relation to fever15
.
Any infants failing to attain protective levels of antibody post-vaccination were offered
revaccination.
Statistical analysis: A competitive tender was won by the London School of Tropical
Medicine and Hygiene (LSHTM) Tropical Epidemiology Group. LSHTM was sent the
results of all the laboratory serology, relevant clinical data from each study site plus the
randomization codes. Post-vaccination geometric mean antibody concentrations (GMCs)
and the proportion of infants attaining the protective level for each EPI antigen were
compared in the IPTi and placebo groups. Reverse cumulative curves were plotted for each
EPI antigen.
The results for each of the trials are provided in individual statistical reports for each study.
These final reports are on file with WHO and have been shared with the study teams for
inclusion in the publication of their trial results.
A final pooled analysis of all the study results was prepared in early 2009. Data from the
trial in Bungoma were excluded due to previously established concerns about data quality.
The final pooled analysis (March 27, 2009) that was accepted by the Advisory Committee is
provided in Annex 3.
WHO Advisory Committee
The Committee met several times over the course of this project, particularly in 2003 to
advise on the design of the protocol and selection of the laboratory. In later years, the
Committee has used teleconferences to conduct its work (See Annex 2 for complete listing
of dates of meetings and teleconferences).
The Advisory Committee reviewed and discussed each of the statistical reports of the five
studies. Often the Committee requested revisions or additional statistical analyses in order to
facilitate their interpretation and conclusions. A complete record of the minutes of these
teleconference is on file with WHO. Brief summaries of the results for each of the five
studies can be found in Annex 4.
Prior to issuing their final conclusions, the Advisory Committee requested an independent
audit of the laboratory work carried out by HPA. A post study audit was performed on the
laboratory records in Jan and March 2009. Several discrepancies were noted and HPA was
requested to take corrective action on these points. Subsequently, a second follow-up audit
was conducted in June 2009 with the objective to further investigate the discrepancies noted
in the first audit and to evaluate the corrective actions taken. Overall, the HPA labs were
15 Usen S, Milligan P, Ethevenaux C, Greenwood B, Mulholland K. Effect of fever on the serum
antibody response of Gambian children to Haemophilus influenzae type b conjugate vaccine. Pediatr
Infect Dis J 2000;19(5):444-9.
12
found to be operating in a manner that meets relevant GLP criteria for this GCP study. The
audit satisfied the Advisory Committee that the analyses were carried out appropriately.
This Final Report presents the final conclusions of the Advisory Committee and is primarily
based on the final pooled analysis of the results, and supported by selected analyses of the
individual trials where no pooled analyses was planned (i.e. yellow fever and Hib vaccines).
Results – Pooled Analysis
The pooled analysis (see Annex 3 for full report) was conducted on data from four of the
study sites – Navrongo, Manhica, Kisumu, and Kilimanjaro. Data from the fifth trial,
conducted in Bungoma, was excluded due to concerns about data quality. Data from
different trials and treatment groups were pooled only after establishing, by means of
appropriate interaction tests, that treatment effects were not heterogeneous. If there was
evidence of heterogeneity, pooling was not done and trial-specific treatment effects were
reported.
Both Intention-to-Treat (ITT) and According-to-Protocol (ATP) analyses were undertaken.
All children with matched pre and post measles vaccination samples were included in the
ITT analysis. Children with incomplete drug dosing were excluded from the ATP analysis.
Hence, only Navrongo children with all four drug doses taken, and Manhica, Kisumu and
Kilmanjaro children with all three drug doses taken were considered for the ATP.
Separate analyses for measles were undertaken excluding children with: i) detectable and ii)
protective pre-vaccination levels. For all other antigens all children with a post vaccination
sample were included in the ITT analysis, whereas children with incomplete drug dosing
were excluded from the ATP analysis.
For all antigens analyses were conducted on post-vaccination antibody concentrations,
using: i) the continuous concentration variable, summarized by its geometric mean (GMC),
and ii) the binary protected/unprotected variable, based on whether antibody concentrations
were above or below the pre-defined threshold of protection for each antigen, where
appropriate.
Pooled SP vs Placebo comparison for measles:
Pooled data from Navrongo, Manhica and Kilmanjaro provided matched pre and post
vaccination measurements for 2,015 children (997 in placebo and 1,018 in SP groups).
Comparison of the geometric mean concentrations (GMC) for measles antibodies in the two
treatment groups before and after vaccination, as well as the median post-vaccination
concentration, found no evidence of a difference in the GMC between SP and placebo
groups in any of the sub-population investigations16
. (Annex 3)
The formal test of non-inferiority of the null hypothesis (that the difference in percentage
unprotected between groups was > 5%) gave strong evidence that SP is not inferior to
placebo in ITT and ATP analyses, excluding children with detectable/with protective pre-
vaccination concentration levels. For example, for the ATP analysis (excluding those
16 With and without detectable concentration pre-vaccination; with and without protective measles antibody level
pre-vaccination; all children.
13
children with detectable concentration at pre-vaccination) the actual difference (SP minus
placebo) between the groups was -0.15% with a 95% CI (-2.33, 2.04) (p<0.0001).
Finally, the reverse cumulative distribution function for measles antibody concentrations for
the ITT cohort (excluding children with detectable antibody levels at pre-vaccination),
shows that the curves for placebo and SP are nearly identical (Figure 1).
FIGURE 1: Pooled Analysis (Navrongo, Manhica, Kilimanjaro)
Reverse cumulative distribution function for measles antibody concentrations for the ITT
cohort excluding children with detectable antibody levels at pre-vaccination
0.2
.4.6
.81
1 2 3 4 5log(10) measles antibody concentration post-vaccination
Placebo SP
(vertical line indicates assumed protective threshold)
Reverse empirical cumulative distribution function
Pooled drugs (SP, LapDap, SP-ART, AQ-ART, MQ) vs placebo comparison for
measles
For the Kilimanjaro (SP+MQ+LapDap) measles comparison data were pooled for both
analyses, resulting in 397 children in placebo and 1,141 in the single combined treatment
group. No evidence was found to support a difference in the GMC between the combined
treatment group and placebo. The formal test of non-inferiority, gave strong evidence to
reject the null hypothesis that the proportion unprotected was at least 5% higher than in the
placebo group. The actual difference (combined treatment minus placebo) found in the ATP
analysis, excluding those with detectable concentration at pre-vaccination, was 0.21% with a
95% CI (-2.67, 3.08) (p=0.0001).
For the Kisumu (SP-ART+AQ-ART+LapDap) measles comparison, the treatment groups
were pooled for the analysis of GMCs, resulting in 284 children in placebo and 838 in the
combined treatment group. No evidence was found to support a difference in the measles
GMC between the combined treatment group and placebo. For the analysis of proportions
14
unprotected, however, the three treatment groups were heterogeneous (owing to the much
lower proportion unprotected in the LapDap group), so the three treatment groups were not
pooled (the separate analyses for the three treatment groups all rejected the null hypothesis
that the proportion unprotected is at least 5% higher than for placebo; see Annex 4).
A secondary pooled analysis of the Kisumu measles data excluding the LapDap group was
undertaken, to compare SP-ART + AQ-ART versus placebo. This strongly rejected the null
hypothesis that treatment is inferior to placebo.
A final pooled analysis for measles combining all treatment groups (except the Kisumu
LapDap group, for the reasons given above) versus placebo across the four trials (Navrongo,
Manhica, Kisumu and Kilimanjaro) resulted in 1,281 children in placebo and 2,363 in the
combined group. No evidence was found of a difference in the measles GMC between the
all combined treatment group and placebo. The formal test of non-inferiority (difference in
unprotected proportion of children is >5%), provided strong evidence that the combined
group was not inferior to placebo. For the ATP analysis, excluding those children with
detectable concentration at pre-vaccination, the actual difference (combined treatments
minus placebo) was 0.53% with a 95% CI (-1.18, 2.22) (p=0.0001).
The reverse cumulative distribution function of measles antibody concentrations for the ITT
cohort, excluding children with detectable pre-vaccination antibody levels ,provides
additional support that there is little difference between the intervention group (all
treatments except LapDap Kisumu) and placebo (Figure 2).
FIGURE 2: Pooled Analysis All Sites (Navrongo, Manhica, Kisumu and Kilimanjaro), All Treatments Combined (except Kisumu LapDap)
Reverse cumulative distribution function for measles for the ITT cohort excluding children with detectable antibody levels at pre-vaccination
0.2
.4.6
.81
1 2 3 4 5log(10) measles antibody concentration post-vaccination
Placebo Combined
(vertical line indicates assumed protective threshold)
Reverse empirical cumulative distribution function
15
Pooled Analysis for other antigens
Manhica (SP) and Kisumu (SP-ART, AQ-ART, LapDap) were the only sites assessing
antigens other than measles and yellow fever (polio types 1 & 3, diphtheria, tetanus,
hepatitis B, pertussis FHA & Toxin)17. Serology data for these antigens was available for a
total of 634 for the ITT and 567 for the ATP analysis. Complete data on all antigens was
available for 150 children.
The comparison of placebo groups in Manhica and Kisumu found that Kisumu had a
significantly higher percentage of unprotected children for polio types 1 & 3 and diphtheria
when compared to Manhica. For tetanus and HepB there was no evidence of a difference.
These results were the same in ITT, ATP, and other sub-population analyses. For diphtheria,
pertussis toxin and FHA, tetanus and HepB Manhica children had on average a higher
geometric mean concentration after vaccination when compared to Kisumu children.
In this context, the Advisory Committee observed that there appeared to be marked variation
across the study sites, and that some study sites seemed to be "high responders" and others
(particularly Kisumu) were "low responders". The Committee speculated that there could be
differences in the health status and weight of children between the sites, or that the vaccines
were not in their best condition in the ‘low responder’ sites (perhaps owing to cold chain
issues)18
.
As data on antigens other than measles and yellow fever is available from only two sites,
Manhica and Kisumu, the individual analyses for these studies is summarized.
In Manhica, the proportion of infants achieving protective titres post-vaccination was similar
in the SP and placebo groups for diphtheria, tetanus, polio type 1 & 3, and HepB (Hib
vaccine was not at the time included in the Mozambique vaccination schedule). The test of
non-inferiority (with a 10% threshold) was significant for each antigen, providing evidence
that SP treatment had no adverse impact on serological responses to EPI vaccines. There are
no know serological correlates of protection for pertussis. For all antigens, post-vaccination
GMCs and the reverse cumulative curves were similar in both the SP and placebo groups.
In Kisumu (Placebo, SP-ART, AQ-ART, LapDap):
• the post-vaccination GMC's for diphtheria, tetanus, Hib and pertussis toxin were similar
in the four treatment groups in both ITT & ATP analyses.
• For HepB, both in the ITT & ATP analyses, there was evidence to reject the null
hypothesis of equal GMC in all four groups.
• The hypothesis that the proportions unprotected in IPTi-treatment groups were at least
10% higher than in the placebo group was rejected, in both ITT & ATP analyses, for
diphtheria, tetanus, Hib, HepB and polio type 1.
17 Hib was assessed only in the Kisumu trial. Other antigens were also assessed in Bungoma, but these data were
excluded from pooled analyses as stated previously. 18 Differences were also observed in the proportions of children protected against measles within the placebo
groups in Kilimanjaro, Kisumu, Manhica and Navrongo. The proportion unprotected was highest in
Manhica and lowest in Navrongo.
16
• For polio type 3 the null hypothesis of inferiority (with respect to proportions
unprotected) of the LapDap and SP-ART groups compared to placebo for both ITT &
ATP analyses could not be rejected.
• The reverse cumulative distribution functions were similar for all antigens, except
HepB.
Given the difficulties of interpreting the HepB results an additional post-hoc analysis which
used 100 IU/L (instead of 10 IU/L)19 for inferiority was conducted.
The HepB analysis was compromised by the small sample size (n=222 which meant only
around 50 in each arm). There was no evidence of any problem with SP-ART, some
evidence that AQ-ART was better than placebo, and some very weak evidence that LapDap
might be worse than placebo, the proportions under 100 IU/L being quite high. However, no
definite conclusions could be drawn owing to the lack of power.
In Kisumu, a comparison of GMCs in the combined treatment group (SP-ART+AQ-
ART+LapDap) versus placebo group was conducted for diphtheria, pertussis FHA & toxin,
and tetanus, but not for hepatitis B. The GMCs in the combined treatment group were
similar to those for the placebo group. The reverse cumulative distribution functions were
also similar.
For proportions unprotected,, it was possible to pool the treatment groups for polio type 1 &
3 and diphtheria. The null hypothesis that the proportion unprotected is more than 10%
greater in the treatment group than in the placebo group was rejected for all antigens, in ITP
and ATP analyses.
Results – Yellow Fever (Navrongo)
Only one study site, Navrongo (Ghana) was able to provide data on serological responses to
yellow fever vaccination in infants randomized to receive SP or placebo. Unfortunately,
many of the samples selected had insufficient volumes for testing, and additionally there
was a miscommunication with the lab about the dilution, which further reduced the available
samples to only 136, when the protocol specified 250.
Owing to many complicating factors (small sample size; lack of clarity about the correlation
between protective titres and IU/ml concentration values; suspected interference with cross-
reactive antibodies for other flaviviruses) the analysis of the yellow fever data was limited to
differences between the groups post-vaccination..
Focusing only on the post-immunization samples, a comparison of the two groups was
undertaken on the entire cohort (not excluding any of the children with inconclusive
replicate samples, or detectable or protective antibody levels prior to immunization).
The proportions protected were similar in both groups. The formal test of non-inferiority
rejected the null hypothesis that the proportion unprotected in the treatment groups was
more than 10% greater than in the placebo group; the actual difference (SP minus placebo)
was 4.21%. Three methods adjusting and not for clustering produced similar results. The
19 100IU/L is the level at which long term immunity is conferred; 10 IU/L is the level for seroconversion.
17
geometric mean concentrations were found to be very similar in both groups. Finally, the
reverse cumulative distribution functions suggest that there is no reduction in antibody
concentrations in the SP group.
18
Conclusions of the WHO Advisory Committee
The Advisory Committee was established 6 years ago to review data generated by existing
studies. While it cannot be held responsible for the conduct of the field trials or laboratory
work, the Committee concludes the following:
1. Serological data from studies in Navrongo, Manhica, and Kilmanjaro provide strong
evidence that IPTi with SP, does not have an adverse impact on serological responses to
measles vaccine;
2. Though very limited, the available serological data from Navrongo provides evidence
that SP does not have a negative impact on antibodies following vaccination against
yellow fever;
3. Serological data from studies in Kisumu and Kilimanjaro provide strong evidence that
IPTi with SP-ART, AQ-ART, MQ, or LapDap do not have an adverse impact on
serological responses to measles vaccination;
4. Serological data from Manhica and Kisumu provide strong evidence that IPTi with
SP20
, SP-ART, AQ-ART, or LapDap do not have an adverse impact on serological
responses to DTP, polio, Hib, and HepB vaccines.
5. The pooled analyses provide further evidence that IPTi treatments do not impair
serological responses to EPI antigens.
20 Data from Bungoma, although weaker, also suggest that there is no negative interaction between SP
and EPI antigens.
Overall conclusion:
Studies have demonstrated that there is no adverse impact on the serological responses to
DTP, polio, Hib, HepB, and measles vaccines when the IPTi drugs SP, SP-ART, AQ-
ART, and LapDap are administered to infants at the time of routine vaccination.
Concomitant administration of IPTi-SP at the time of yellow fever vaccination, and IPTi-
MQ at the time of measles vaccination, have also been shown to have no negative effect
on the serological responses.
19
Annex 1: Report of the Technical Consultation on Intermittent
Preventive Treatment in Infants (IPTi), Technical Expert Group
(TEG) on Preventive Chemotherapy, April 23-24, 2009 Geneva..
Technical Expert Group meeting on Preventive chemotherapy
Report of the Technical Consultation on Intermittent Preventive Treatment in Infants (IPTi)
WHO HEADQUARTERS, GENEVA, 23–24 APRIL 2009
WHO HEADQUARTERS, GENEVA, 23–24 ApRil 2009
Technical Expert Group meeting on Preventive chemotherapy
Report of the technical consultation on Intermittent Preventive Treatment in Infants (IPTi)
Contents
1. Background ...............................................................................................................................................................................................1
2. Conclusions...............................................................................................................................................................................................3
3. Recommendations...........................................................................................................................................................................5
4. Other recommendations ........................................................................................................................................................7
5. References ...................................................................................................................................................................................................8
6. List of participants ......................................................................................................................................................................10
© World Health Organization 2009. All rights reserved.
The designations employed and the presentation of the material in this publication do not imply the expression of any opinion whatsoever on the part of the World Health Organization concerning the legal status of any country, territory, city or area or of its authorities, or concerning the delimitation of its frontiers or boundaries. Dotted lines on maps represent approximate border lines for which there may not yet be full agreement.
The mention of specific companies or of certain manufacturers’ products does not imply that they are endorsed or recommended by the World Health Organization in preference to others of a similar nature that are not mentioned. Errors and omissions excepted, the names of proprietary products are distinguished by initial capital letters.
All reasonable precautions have been taken by the World Health Organization to verify the information contained in this publication. However, the published material is being distributed without warranty of any kind, either expressed or implied. The responsibility for the interpretation and use of the material lies with the reader. In no event shall the World Health Organization be liable for damages arising from its use.
WHO Library Cataloguing-in-Publication Data
Technical expert group meeting on preventive chemotherapy : report of the technical consultation on intermittent preventive treatment in infants (IPTi) Geneva, 23-24 April 2009. 1.Malaria, Falciparum - prevention and control. 2.Malaria, Falciparum - drug therapy. 3.Infants. 4.Drug adminis-tration schedule. 5.Pyrimethamine - therapeutic use. 6.Sulfadoxine - therapeutic use. 7.Treatment outcomes. 8.Meta-analysis 9.Guidelines. I.World Health Organization.
ISBN 978 92 4 159858 2 (NLM classification: WC 765)
WHO Technical Expert Group on Preventive Chemotherapy 1
Intermittent preventive treatment in infancy (IPTi) is defined as:
the administration of a full course of an effective antimalarial treatment at specified
time points to infants at risk of malaria, regardless of whether or not they are parasitaemic, with the objective of reducing the infant malaria burden.
In October 2006 and October 2007, WHO convened meetings of the Technical Expert Group (TEG) on Intermittent Preventive Treatment in Infants to review the available evidence on the safety, efficacy and other relevant aspects of IPTi with sulfadoxine-pyrimethamine (SP-IPTi) delivered through the Expanded Programme for Immunization (EPI). At the time, six randomised, placebo-controlled clinical trials with SP-IPTi were being, or had been, conducted in areas of Africa, south of the Sahara with relatively high malaria endemicity (see Table). TEG 2006 concluded that SP-IPTi held promise as a potential malaria control intervention, noting that three of the studies were yet ongoing or unpublished.1 At the TEG 2007, at which time the six studies had been completed, the committee concluded that, though IPTi remains a potential intervention for malaria control, the use of SP-IPTi cannot be recommended as a strategy for general deployment based on the assessment of the risks and benefits, and advised a future review of further evidence when available.2
In the current expert review of the evidence on SP-IPTi, TEG 2009 reviewed the evidence available on SP-IPTi including additional data that were generated since the TEG-2007 meeting, with a view to making a definitive policy recommendation on this intervention for malaria control.
The new information reviewed was the following:
1. An in-depth analysis conducted by the IPTi Consortium3, of the severe skin reactions associated with SP-IPTi reported previously.4,5
2. Two additional randomized placebo controlled trials on the safety and efficacy of IPTi, which have been submitted for publication.6,7
3. The experience of implementation studies conducted by UNICEF in selected districts of six countries in Africa, south of the Sahara8 and another by the IPTi Consortium9–11 with respect to the feasibility of implementation, and its safety, monitored through active and passive observations on adverse reactions.
1. Background
2 WHO – Geneva, 23–24 April 2009
Table – Summary of site-specific information on the six published SP-IPTi trials considered for the pooled analysis*
Study site Study period Transmission pattern
Iron supplementation
# Infants studied SP/placebo**
Ifakara, UR Tanzaniai
1999–2000 perennial Yes 350/351
Navrongo, Ghanaii
2000–2004 Seasonal Yes 1183/1203
Manhica, Mozambiqueiii
2002–2004 perennial/seasonal peaks
None 748/755
Kumasi, Ghanaiv
2003–2005 perennial None 535/535
Tamale, Ghanav
2003–2005 perennial/seasonal peaks
None 600/600
Lambarene, Gabonvi
2004–2005 perennial/seasonal peaks
None 504/507
i ) Schellenberg D et al. (2001). Intermittent treatment for malaria and anaemia control at time of routine vaccinations in Tanzanian infants: a randomised, placebo-controlled trial. Lancet, 357:1471–1477.
ii) Chandramohan D et al. (2005). Cluster randomized trial of intermittent preventive treatment for malaria in infants in area of high, seasonal transmission in Ghana. Br Med J, 331:727–733.
iii) Macete E et al. (2006). Intermittent preventive treatment for malaria control administered at the time of routine vaccinations in Mozambican infants: a randomized, placebo controlled trial. J Inf Dis, 194:276–285.
iv) Kobbe R. et al. (2007). A randomized controlled trial of extended intermittent preventive antimalarial treatment in infants. Clin Infect Dis, 45:16–25.
v) Mockenhaupt FP et al. (2007). Intermittent Preventive Treatment in Infants as a Means of Malaria Control: a Randomized, Double-Blind, Placebo-Controlled Trial in Northern Ghana. Antimicrob Agents Chemother, 51: 3273–3281.
vi) Grobusch M. et al. (2007). Intermittent preventive treatment in infants against malaria in Gabon – a randomised, double-blind, placebo-controlled trial. J Inf Dis, 196: 1595–1602.
* The pooled analysis excludes the most recent study6 which was accepted for publication after the meeting, although its data were made available to the meeting.
** Who received at least the first dose of SP-IPTi
WHO Technical Expert Group on Preventive Chemotherapy 3
2. Conclusions
The TEG 2009 concluded that:
1. The previous safety concerns about SP-IPTi, specifically with respect to the reported severe skin reactions were mitigated by the evidence from the larger observational studies and retrospective in-depth examination by the Consortium of the severe skin reactions reported in previous studies.
2. The benefits of SP-IPTi in areas where SP remains effective against Plasmodium falciparum malaria parasites, were upheld as providing a 30% (95% CI: 19.8%–39.4%) overall protection against clinical malaria episodes and a variable reduction (overall 21.3%) (95% CI: 8.3%– 32.5%) in anaemia (< 8 g/dl) in a pooled analysis of data from 6 published studies (see Table).12 The reduction in all cause hospital admissions by 23% (95% CI: 10.0%–34.0%), was noted as a potential benefit. The admissions, however, were not all due to severe malaria, and this therefore cannot be equated to a similar reduction in the incidence of severe malaria. The pooled analysis excludes the most recent study6 which was accepted for publication after the meeting, although its data were made available to the meeting. The protective efficacy of SP-IPTi against clinical malaria episodes in this study was – 6.7% (95% CI: – 45.9–22.0).
3. Where effective, SP-IPTi offers a personal protection against clinical malaria for a period of approximately 35 days following the administration of each dose. There is no evidence for an individual cumulative protective effect beyond this period until the next dose. The mechanism of action appears to be predominantly chemoprophylaxis related to the half-life of the medicine and the susceptibility of the prevalent malaria parasites.
4. The protective efficacy of SP-IPTi is dependent upon the efficacy of SP, to which there is increasing parasite resistance in Africa and worldwide, but the threshold of parasite resistance to SP at which IPTi ceases to be effective is still not known. SP-IPTi was reported to provide benefit when the in vivo therapeutic failure rate of SP at day 14 was 31% (measured
4 WHO – Geneva, 23–24 April 2009
in children with symptomatic malaria)13 and the population prevalence of Pfdhps + Pfdhfr quintuple mutants (molecular markers of parasite resistance to SP) was 50%14 but there was no benefit when the in vivo SP therapeutic failure rate was 82% at day 28, and the prevalence of the quintuple mutants was 90%.6
5. Uncertainties remain on the potential impact, or lack thereof, of SP-IPTi on the incidence of severe malaria or malaria mortality.
6. Uncertainties also remain on the impact of SP-IPTi at low levels of malaria transmission (either natural or resulting from effective control interventions).
7. A rebound effect by way of greater susceptibility to malaria following the termination of SP-IPTi was not evident in the pooled analysis. However, this warrants further observation in view of the fact that three of the studies reported an increase in either malaria infections associated high density parasitaemia15; anaemia (< 7.5 g/dl)4 ; or severe malaria and severe malarial anaemia (Hb < 5g/dl)5 during the post-intervention period in children who had received SP compared to the placebo group.
8. SP-IPTi was deemed a safe addition to EPI because there was no evidence of an adverse effect of SP-IPTi on infants’ serological response to EPI vaccines against DTP, Polio, Hepatitis B, Hib, yellow fever and measles.16
Limited implementation studies suggest that SP-IPTi incurs only marginally additional costs to EPI, and that it has a favorable effect on EPI coverage.
The panel comprised of 15 independent Experts. The Consultation was attended by observers from UNICEF, the Bill and Melinda Gates foundation, and the IPTi Consortium (Appendix 1, List of participants).
WHO Technical Expert Group on Preventive Chemotherapy 5
3. Recommendations
Considering that the benefits of SP-IPTi to infants are in providing a protection against clinical malaria from – 6.7% to 59.4%, and in view of increasing parasite resistance to SP, the TEG 2009 recommended that,
1. SP-IPTi delivered through EPI be considered for implementation as an additional malaria control intervention in countries in Africa, south of the Sahara under the following specific conditions,
a. In areas with moderate to high transmission (Annual Entomological Inoculation Rates [EIR] beyond 10).
b. When parasite resistance to SP in the area is not high. Precise cut-offs cannot be defined on the basis of available data. More information is needed on the relationship between the prevalence of molecular markers (mutations in Pfdhfr and Pfdhps) and the duration of malaria protection provided by SP-IPTi.
c. If its implementation does not detract from efforts to scale-up access to artemisinin-based combination therapies (ACT) for early treatment, and to insecticide-treated bednets (ITN) and indoor residual spraying (IRS) as preventive measures, all of which have significantly greater efficacy in malaria control.
2. Where SP-IPTi is used,
a. Continuous surveillance of parasite resistance to SP must accompany the implementation of SP-IPTi as a surrogate measure of its efficacy. Methodologies for monitoring the efficacy of SP-IPTi should be developed urgently to guide countries on when the intervention should no longer be deployed.
6 WHO – Geneva, 23–24 April 2009
b. SP-IPTi should not be given to infants receiving a sulfa medication for treatment or prophylaxis against an infection, including co-trimoxazole (trimethoprim-sulfamethoxazole) which is widely used as a prophylactic against opportunistic infections in HIV-infected infants.
c. Surveillance for drug safety must be strengthened with effective pharmaco-vigilance systems to monitor serious adverse reactions to SP which may be exacerbated because SP-IPTi is likely to be implemented against a background of co-trimoxazole use for the treatment of acute respiratory infections in infants and for prevention of opportunistic infections in HIV-infected infants.
WHO Technical Expert Group on Preventive Chemotherapy 7
4. Other Considerations
The Technical Expert Group,
• Consideredtheseasonalityofmalariatransmission,andwhetherareasofseasonal malaria should be excluded for the implementation of SP-IPTi given that the optimal protective effect lasts for 35 days post treatment, but concluded that the evidence base did not support such an inference.
• RecognizedtheneedtodeveloptoolstomonitortheeffectivenessofSP-IPTi and validate a measure of parasite resistance to SP which can define a threshold at which SP-IPTi should not be implemented, and recommended that WHO GMP be financially supported to undertake this development as a priority.
• UrgedthedevelopmentofnewalternativemedicinesforIPTiasreplacementsfor SP, with properties conferring an optimum therapeutic profile for IPTi (single dose, excellent tolerability and long half-life) and reliably provide prophylaxis for a period of at least 4 weeks; new medicines for IPTi should preferably be different from those deployed for chemotherapeutic purposes, and should also be suitably formulated for the paediatric age group.
• Warned that the efficacy of SP is decreasing in many areas. Thoughdesirable, the development of a paediatric formulation for SP might take longer than its useful residual therapeutic life for IPTi and hence the simpler option of producing scored SP tablets should be considered.
• Requestedthatthefollowingquestionsbeaddressed:
– What are the optimum pharmacokinetic and pharmacodynamic properties required for medicines used for IPTi?
– What are the optimum ages for administering IPTi taking into account operational realities and burden of disease in relation to transmission intensity?
• Recommended that the assessment of impact on mortality should beconsidered among the endpoints for efficacy of the next candidate medicine for IPTi.
8 WHO – Geneva, 23–24 April 2009
5. References
1. WHO (2006). Report of Technical Consultation on Intermittent Preventive Treatment for Malaria in Infancy (IPTi), Global Malaria Programme, World Health Organization, Geneva, 25-27 October, 2006.
2. WHO (2007). Report of the Technical Expert Group meeting on Intermittent preventive therapy in infancy (IPTi), Geneva, 8–10 October 2007. http://malaria.who.int/docs/IPTi/TEGConsultIPTiOct2007Report.pdf.
3. IPTi consortium (2009). Consortium Safety Panel (CSP) Report – April 2009.
4. Kobbe R. et al. (2007) A randomized controlled trial of extended intermittent preventive antimalarial treatment in infants. Clin Infect Dis, 45:16-25.
5. Mockenhaupt FP et al. (2007) Intermittent Preventive Treatment in Infants as a Means of Malaria Control: a Randomized, Double-Blind, Placebo-Controlled Trial in Northern Ghana. Antimicrob Agents Chemother, 51:3273-3281.
6. Gosling R et al. Protective efficacy and safety of three antimalarial regimens for intermittent preventive treatment for malaria in infants: a randomised, placebo-controlled trial. Lancet, in press.
7. Odhiambo F et al. Intermittent Preventive Treatment in Infants using long and short-acting drug combinations for the prevention of malaria and anaemia in rural western Kenya: a randomized, double-blind placebo controlled trials (unpublished).
8. UNICEF Operational Research on Intermittent Preventive Treatment of malaria in infants (IPTi): pharmacovigilance preliminary report, New York, March 2009 (unpublished).
9. Manzi F et al (2009) Intermittent preventive treatment for malaria and anemia control in Tanzanian infants; the development and implementation of a public health strategy. Trans R Soc Trop Med Hyg, 103:79-86.
10. Pool R et al. (2008) The acceptability of intermittent preventive treatment of malaria in infants (IPTi) delivered through the expanded programme of immunization in southern Tanzania. Malar J, 7:213.
WHO Technical Expert Group on Preventive Chemotherapy 9
11. Maokola W et al. The safety of sulphadoxine-pyrimethamine for intermittent preventive treatment of malaria in infants: evidence from large-scale operational research in southern Tanzania (unpublished).
12. Statistical Working Group of the IPTi Consortium: Pooled analysis of the IPTi trials with SP, 12 April 2009.
13. Schellenberg D et al. (2001) Intermittent treatment for malaria and anaemia control at time of routine vaccinations in Tanzanian infants: a randomised, placebo-controlled trial. Lancet, 357:1471-1477.
14. Mayor A et al. (2008) Molecular markers of resistance to sulfadoxine-pyrimethamine during intermittent preventive treatment for malaria in Mozambican infants. J Infect Dis, 197:1737-1742.
15. Chandramohan D et al. (2005) Cluster randomized trial of intermittent preventive treatment for malaria in infants in area of high, seasonal transmission in Ghana. Br Med J, 331:727-733.
16. WHO (2006). Interim report on IPTi with SP. WHO Advisory Committee on Serological responses to EPI vaccines in Infants receiving IPTi.
10 WHO – Geneva, 23–24 April 2009
6. List of Participants
MEMBERS OF THE TECHNICAL EXPERT GROUP ON PREVENTIVE CHEMOTHERAPY
Dr Willis AKHWALEDirector, Communicable Diseases Division (former manager of National Malaria Control Programme)Ministry of Health KENYA.
Dr Karen BARNES*Associate ProfessorDivision of Clinical PharmacologyUniversity of Cape TownSOUTH AFRICA
Professor Fred BINKA (co-Chairperson)School of Public Health,University of GhanaGHANA
Professor Anders BJÖRKMANDivision of Infectious DiseasesKarolinska University HospitalSE-171 76 StockholmSWEDEN
Professor Ogobara DOUMBO*Director, Malaria Research and Training CentreBamakoMALI
Dr Issa MAKUMBI Head of Epidemiology and Surveillance,(former EPI Programme Manager) Ministry of HealthUGANDA
WHO Technical Expert Group on Preventive Chemotherapy 11
Dr Anne McCARTHY* Director, Tropical Medicine and International Health ClinicOttawa Hospital General CampusOttawa CANADA
Dr Theonest K. MUTABINGWA Associate Member, InternationalSeattle Biomedical Research InstituteMOMS ProjectMorogoroUR TANZANIA
Professor Olayemi OMOTADEDirector, Institute of Child HealthCollege of Medicine, University College HospitalIbadanNIGERIA
Professor Nick WHITE (co-Chairperson)Faculty of Tropical MedicineMahidol UniversityBangkokTHAILAND
Dr Abdoulaye DJIMDE (co-opted Member)*Malaria Research and Training CentreUniversity of BamakoMALI
Dr Feiko ter KUILE (co-opted Member)Liverpool School of Tropical MedicineLiverpoolUNITED KINGDOM
Dr Rick STEKETEE (co-opted Member)MACEPA PATHBâtiment Avant-CentreFerney-VoltaireFRANCE
12 WHO – Geneva, 23–24 April 2009
Observers
Chair of Technical and Research Advisory Committee Professor Barry BLOOM
Members of the IPTI Consortium Dr Pedro ALONSO Dr John APONTE Dr Alasdair BRECKENRIDGE Dr Andrea EGAN Dr David SCHELLENBERG
Bill and Melinda Gates Foundation Dr David BRANDLING-BENNET Dr Erin SHUTES UNICEF Dr Alexandra DE SOUSA
WHO Secretariat
Global Malaria Programme Dr Sergio SPINACI - Associate Director, GMP Dr Kamini MENDIS Dr Peter OLUMESE Dr Pascal RINGWALD Dr Marian WARSAME
Expanded Programme on Immunization Dr Tracey GOODMAN
Special Programme for Research and Training in Tropical Diseases Dr Melba GOMES
WHO Regional Office for Africa Dr Georges A. KI-ZERBO
20
Annex 2: Membership of WHO Advisory Committee on serological
responses to Expanded Programme on Immunization vaccines in
infants receiving Intermittent Preventive Treatment for malaria
The Committee comprises two pediatricians with extensive experience in the laboratory
assessment of serological responses (CAS and DG), one expert in vaccine safety (OW), a
clinician with expertise in clinical trials of antimalarial drugs (PFo), and a biostatistician
with extensive experience in the statistical design of non-inferiority trials and analysis of
serological data (PFa).
Professor Claire-Anne Siegrist (Chairperson)
Head, WHO Collaborating Centre for Neonatal Vaccinology, Centre Médical Universitaire,
1 Rue Michel-Servet, 1211 Geneva, Switzerland
Professor Paddy Farrington
Department of Statistics, Open University, Milton Keynes, MK7 6AA, United Kingdom
Professor Peter Folb
Chief Specialist Scientist, Medical Research Council, Tygerberg 7505, Cape Town, South
Africa
Professor David Goldblatt
Professor of Vaccinology and Immunology, Immunobiology Unit, Institute of Child Health,
30 Guildford Street, London WC1N 1EH, United Kingdom
Dr Omala Wimalaratne
Head, Department of Rabies and Vaccines, Medical Research Institute, PO Box 527,
Colombo 8, Sri Lanka
Meeting: 19 May 2003
Meeting: 13 June 2003
Meeting: 11 June 2005
Teleconference: 30 September 2005
Teleconference: 8 November 2005
Teleconference: 5 April 2006
Teleconference: 28 April 2006
Teleconference: 6 October 2006
Teleconference: 8 March 2007
Teleconference: 14 August 2008
Teleconference: 14 November 2008
Teleconference: 11 February 2009
Teleconference: 11 March 2009
Teleconference: 17 April 2009
Teleconference: 7 July 2009
21
Annex 3: Final Pooled Analysis: Assessment of Serological
Responses to Expanded Programme on Immunization Vaccines in
Infants Receiving Intermittent Preventive Treatment (v.3 submitted
July 3, 2009)
Final pooled analysis V.3 Navrongo, Manhica, Kisumu, Kilimanjaro 03/07/2009
1
ASSESSMENT OF SEROLOGICAL RESPONSES TO EXPANDED PROGRAMME ON IMMUNISATION
VACCINES IN INFANTS RECEIVING INTERMITTENT PREVENTIVE TREATMENT
FINAL POOLED ANALYSIS V.2
<V.1 submitted February 26th 2009>
<V.2 submitted March 27th 2009>
<V.3 submitted July 3rd 2009>
Table of contents
Executive Summary
1. Background
2. Definitions
3. Statistical methods, samples, datasets and strategies of analysis
4. Proportions protected and distributions of antibody concentrations in the placebo groups
4.1 Measles
4.1.1 Analysis of post-vaccination measles antibody concentrations
4.1.2 Proportion of children with protective concentrations post-vaccination
4.1.2.1 Excluding those with protective measles concentrations in their pre-
vaccination sample
4.1.2.2 Excluding those with detectable measles concentrations in their pre-
vaccination sample
4.2 Diphtheria, Tetanus, Polio type 1 & 3, Pertussis toxin and FHA, and Hepatitis B responses
4.2.1 ITT analysis; post-vaccination concentrations & proportion protected
4.2.2 ATP analysis; post-vaccination concentrations & proportion protected
4.2.3 Children with protective concentrations to multiple antigens
5. Proportions protected and distributions of antibody concentrations in the SP groups
5.1 Measles
5.1.1 Analysis of post-vaccination measles antibody concentrations
5.1.2 Proportion of children with protective concentrations post-vaccination
5.1.2.1 Excluding those with protective measles concentrations in their pre-
vaccination sample
5.1.2.2 Excluding those with detectable measles concentrations in their pre-
vaccination sample
6. Proportions protected and distributions of antibody concentrations in the LapDap groups
6.1 Measles
6.1.1 Analysis of post-vaccination measles antibody concentrations
6.1.2 Proportion of children with protective concentrations post-vaccination
6.1.2.1 Excluding those with protective measles concentrations in their pre-
vaccination sample
6.1.2.2 Excluding those with detectable measles concentrations in their pre-
vaccination sample
Final pooled analysis V.3 Navrongo, Manhica, Kisumu, Kilimanjaro 03/07/2009
2
7. Summary of within group comparisons across sites
8. Pooled SP Vs placebo comparison for measles
8.1 Investigation of effect modification by site
8.2 Measles
8.2.1 Analysis of post-vaccination measles antibody concentrations
8.2.2 Proportion of children with protective concentrations post-vaccination
8.2.2.1 Excluding those with protective measles concentrations in their pre-
vaccination sample
8.2.2.2 Excluding those with detectable measles concentrations in their pre-
vaccination sample
9. Pooled LapDap Vs placebo comparison for measles
9.1 Investigation of effect modification by site
9.2 Measles
9.2.1 Analysis of post-vaccination measles antibody concentrations
10. Pooled combined treatment Vs placebo comparison for Kisumu site
10.1 Investigation of combined Vs individual treatment group comparison with placebo
10.2 Measles
10.2.1 Analysis of post-vaccination measles antibody concentrations
10.3 Diphtheria, Tetanus, Polio type 1 & 3, Pertussis toxin and FHA, and Hepatitis B responses
10.3.1 ITT analysis; post-vaccination concentrations & proportion protected
10.3.2 ATP analysis; post-vaccination concentrations & proportion protected
11. Pooled combined treatment Vs placebo comparison for Kilimanjaro site
11.1 Investigation of combined Vs individual treatment group comparison with placebo
11.2 Measles
11.2.1 Analysis of post-vaccination measles antibody concentrations
11.2.2 Proportion of children with protective concentrations post-vaccination
11.2.2.1 Excluding those with protective measles concentrations in their pre-
vaccination sample
11.2.2.2 Excluding those with detectable measles concentrations in their pre-
vaccination sample
12. Appendices
12.1 Summary to the IPTi Consortium Trials
12.2 Post-hoc analysis I; pooled combined treatment group excluding LapDap Vs
placebo for Kisumu site
12.2.1 Investigation of combined Vs individual treatment group comparison with
placebo
12.2.2 Measles
12.2.2.1 Analysis of post-vaccination measles antibody concentrations
12.2.2.2 Proportion of children with protective concentrations post-
vaccination
Final pooled analysis V.3 Navrongo, Manhica, Kisumu, Kilimanjaro 03/07/2009
3
12.2.2.2.1 Excluding those with protective measles concentrations
in their pre-vaccination sample
12.2.2.2.2 Excluding those with detectable measles concentrations
in their pre-vaccination sample
12.3 Post-hoc analysis II; pooled combined treatment group excluding LapDap Kisumu Vs
placebo for all sites
12.3.1 Investigation of combined Vs individual treatment group comparison with
placebo
12.3.2 Measles
12.3.2.1 Analysis of post-vaccination measles antibody concentrations
12.3.2.2 Proportion of children with protective concentrations post-
vaccination
12.3.2.2.1 Excluding those with protective measles concentrations
in their pre-vaccination sample
12.3.2.2.2 Excluding those with detectable measles concentrations
in their pre-vaccination sample
Babis Sismanidis and Paul Milligan,
Tropical Epidemiology Group,
Infectious Disease Epidemiology Unit,
London School of Hygiene and Tropical Medicine
Final pooled analysis V.3 Navrongo, Manhica, Kisumu, Kilimanjaro 03/07/2009
4
Executive summary
Five randomized placebo controlled trials across Africa were undertaken to investigate the
potential of malaria intermittent preventive treatment in infants (IPTi) having an adverse
impact on serological responses to Expanded Programme on Immunization (EPI) vaccines.
Serological response to measles vaccine was the primary outcome with diphtheria, tetanus,
polio type 1 & 3, pertussis toxin & FHA and hepatitis B being the secondary outcomes.
Analyses have been done separately for each trial. The current report presents analyses on
pooled data from the four trials conducted in Navrongo, Manhica, Kisumu and Kilimanjaro.
Data from the fifth trial, conducted in Bungoma, have been excluded due to concerns about
data quality. Infants were randomized into: i) placebo or SP in Navrongo and Manhica, ii)
placebo, SP-ART, AQ-ART and LapDap in Kisumu and iii) placebo, SP, MQ and LapDap in
Kilimanjaro. All four sites performed serological testing for measles. Only Manhica and Kisumu
performed serological testing on all other antigens.
Both Intention-to-Treat (ITT) and According-to-Protocol (ATP) populations have been
defined and analysed in this report. All children with pre and post measles vaccination
samples were included in the ITT analysis. From the ATP analysis, children with incomplete
drug dosing were excluded. Hence, only Navrongo children with all four drug doses taken and
Manhica, Kisumu and Kilimanjaro children with all three drug doses taken have been
considered for the ATP. For the measles analyses we excluded children with: i) detectable
and ii) protective pre-vaccination levels. For all other antigens all children with a post
vaccination sample were included in the ITT analysis, whereas children with incomplete drug
dosing were excluded from the ATP analysis. For all antigens we included analyses on two
outcomes: i) the continuous geometric mean concentration (GMC) post-vaccination and ii) the
binary protected/unprotected based on antigen levels above or below a pre-defined threshold
of protection for each antigen, where appropriate.
Firstly, we investigated within treatment group across trial comparisons.
i) The placebo group comparison across trials for all antigens presented in section 4. For
measles, post-vaccination GMC were highly significantly different across trials with
highest in the Navrongo trial (n=284), followed by Manhica (n=316) and finally Kisumu
(n=284) and Kilimanjaro (n=397) with the lowest and very similar antibody
concentrations. This finding remained consistent for the different sub-population
investigations (1. all children, 2. children without protective measles antibody
concentration level pre-vaccination and 3. children with undetectable concentration level
pre-vaccination). For the binary outcome of unprotected children post-vaccination the
formal test for comparison of percentages found no evidence to suggest a difference in
proportion across trials, for both ITT and ATP. For all other antigens placebo groups from
Final pooled analysis V.3 Navrongo, Manhica, Kisumu, Kilimanjaro 03/07/2009
5
Manhica and Kisumu were available for analysis. Kisumu had a significantly higher
percentage of unprotected children for polio types 1 & 3 and diphtheria when compared
to Manhica. For tetanus and hepatitis B there was no evidence for a difference in the
respective percentages. These results were the same for ITT, ATP, excluding children
with detectable, excluding children with protective pre-vaccination concentration levels.
For diphtheria, pertussis toxin and FHA, tetanus and hepatitis B Manhica children had on
average a higher geometric mean concentration after vaccination when compared to
Kisumu children.
ii) The SP group comparison across Navrongo (n=336), Manhica (n=308) and Kilimanjaro
(n=374) for measles presented in section 5. Post measles vaccination geometric mean
concentration values were highest in the SP group from Navrongo site. This finding
remained consistent throughout the different sub-population investigations. Manhica also
consistently remained the second highest site followed by Kilimanjaro. Evidence found for
these differences was very strong as evidenced by ratios comparing GMC across trials
using Navrongo as reference. Whereas, no evidence was found to support differences in
percentages unprotected children across trials, for ITT, ATP, excluding children with
detectable, excluding children with protective pre-vaccination concentration levels.
iii) The LapDap group comparison across Kisumu (n=260) and Kilimanjaro (n=387) for
measles presented in section 6. Weak evidence was found to support post-measles
vaccination geometric mean concentration values were highest in the LapDap group from
Kisumu site when compared to Kilimanjaro. This finding remained consistent throughout
the different sub-population investigations. Strong evidence was found to support a
higher percentage of unprotected children in Kilimanjaro, compared to Kisumu for ITT,
ATP, excluding children with detectable, excluding children with protective pre-
vaccination concentration levels.
Subsequently, we investigated between treatment groups across trial comparisons. Data from
different trials were pooled only after establishing, by means of appropriate interaction tests,
that treatment effect was not modified in each trial. If there was evidence of effect
modification, pooling was not done and trial-specific treatment effects were reported.
i) The SP versus placebo comparison for Navrongo, Manhica and Kilimanjaro for measles
presented in section 8. No evidence of effect modification was found and hence data
were pooled across sites, resulting in 997 children in placebo and 1,018 in SP group. No
evidence was found to support a difference in the GMC between SP and placebo, for any
of the sub-population investigations. The formal test of non-inferiority of the null
hypothesis that the difference in percentages unprotected between the groups is 5% or
more, gives strong evidence suggesting SP is not inferior to placebo for ITT, ATP,
Final pooled analysis V.3 Navrongo, Manhica, Kisumu, Kilimanjaro 03/07/2009
6
excluding children with detectable, excluding children with protective pre-vaccination
concentration levels.
ii) The LapDap versus placebo comparison for Kisumu and Kilimanjaro for measles
presented in section 9. No evidence of effect modification was found for the GMC
outcome and hence data were pooled across trials resulting in 681 children in placebo
and 647 in LapDap group. After pooling data across trials no evidence was found to
support a difference in the GMC between LapDap and placebo. However, for the binary
outcome of unprotected children evidence for effect modification across trials was found
and hence data were not pooled. The odds of being unprotected when in the LapDap
group were 1.26 (95% CI 0.69-2.31) times the odds of the placebo group for Kilimanjaro
and 0.32 (95% CI 0.10-1.01) for Kisumu, when excluding children with detectable pre-
vaccination concentration levels. When excluding children with protective pre-vaccination
concentration levels, the respective odds ratios were 1.46 (95% CI 0.84-2.55) for
Kilimanjaro and 0.29 (95% CI 0.09-0.88) for Kisumu.
iii) A post-hoc analysis of combining all treatment groups (excluding LapDap Kisumu) versus
placebo for Navrongo, Manhica, Kisumu and Kilimanjaro presented in appendix 12.3. No
evidence of heterogeneity by trial was found, therefore, data across the four trials were
pooled to study the overall combined treatment effect, resulting in 1,281 children in
placebo and 2,363 in the combined treatment group. No evidence was found to support a
difference in the measles GMC between the single combined treatment group and
placebo for any of the sub-population investigations. The formal test of non-inferiority, of
the null hypothesis that the difference in proportion unprotected between the
intervention groups is 5% or more, gives strong evidence suggesting combined group is
not inferior to placebo for both ITT, ATP, excluding children with detectable, excluding
children with protective pre-vaccination concentration levels.
Finally, we investigated combined treatment versus placebo group comparisons for individual
sites. Data from different treatment groups were combined only after establishing, by means
of appropriate interaction tests, that individual treatment effects were not different. If there
was evidence of effect modification, pooling was not done and individual treatment-specific
effects were reported.
i) (SP-ART + AQ-ART + LapDap) versus placebo in Kisumu for all antigens presented in
section 10. For measles, evidence was found against combining treatment groups for the
outcome of unprotected children and will not be presenting pooled estimates. Using
placebo as the reference group odds ratios for the comparison with SP-ART, AQ-ART and
LapDap were respectively: 1.21 (95% CI 0.57-2.56), 1.13 (95% CI 0.53-2.43), 0.32 (95%
CI 0.10-1.01) when excluding children with detectable pre-vaccination concentration.
When excluding children with protective pre-vaccination concentration respective odds
Final pooled analysis V.3 Navrongo, Manhica, Kisumu, Kilimanjaro 03/07/2009
7
ratios were very similar. However, for the GMC outcome evidence supports the
combination of the three treatment groups into one and its comparison to placebo
resulting in 284 children in placebo and 838 in combined treatment group. No evidence
was found to support a difference in the measles GMC between the combined treatment
group and placebo for any sub-population investigations. For other antigens, evidence
was found supporting the pooled combined treatment group use for diphtheria, pertussis
FHA & toxin and tetanus, but not for hepatitis B, for the GMC outcome. Evidence also
supported the use of the single combined group for polio type 1 & 3 and diphtheria for
the binary outcome of unprotected children. There was no evidence found to support the
combined group as inferior by 10% or more compared to placebo. This result was
consistent across antigens and for both the ITT and ATP. For diphtheria, pertussis toxin
and FHA and tetanus children in the combined treatment group had on average similar
geometric mean concentration after vaccination when compared to placebo children.
ii) (SP + MQ + LapDap) versus placebo in Kilimanjaro for measles presented in section 11.
We found evidence supporting pooling treatment groups for both GMC and binary
outcome of unprotected children, resulting in 397 children in placebo and 1,141 in the
single combined treatment group. No evidence was found to support a difference in the
GMC between the combined treatment group and placebo. The formal test of non-
inferiority, of the null hypothesis that the difference in proportion unprotected between
the intervention groups is 5% or more, gives strong evidence suggesting combined group
is not inferior to placebo for both ITT, ATP, excluding children with detectable, excluding
children with protective pre-vaccination concentration levels.
iii) A post-hoc analysis of (SP-ART + AQ-ART) versus placebo in Kisumu for measles
presented in appendix 12.2. For both the binary outcome of unprotected children and the
GMC outcome evidence supported the combination of the two treatment groups into one
and its comparison to placebo, resulting in 284 children in placebo and 578 in the single
combined treatment group. No evidence was found to support a difference in the measles
GMC between the single combined treatment group and placebo for any of the sub-
population investigations. The formal test of non-inferiority, of the null hypothesis that
the difference in proportion unprotected between the intervention groups is 5% or more,
gives strong evidence suggesting combined group is not inferior to placebo for both ITT,
ATP, excluding children with detectable, excluding children with protective pre-
vaccination concentration levels.
Final pooled analysis V.3 Navrongo, Manhica, Kisumu, Kilimanjaro 03/07/2009
8
1. Background
The aim of the study is to test the hypothesis that intermittent preventive treatment in infants
(IPTi) with sulfadoxine-pyrimethamine (SP) and selected other drug combinations does not
have an adverse impact on serological responses to Expanded Programme on Immunization
(EPI) vaccines, since it would be extremely damaging if the large scale introduction of IPTi
were followed by an increase in vaccine-preventable disease and death. The potential impact
of IPTi on recently introduced quadrivalent and pentavalent vaccines incorporating
Haemophillus Influenzae Type b and Hepatitis B respectively, also needs to be addressed. It
would also be valuable to document an increase in immune responsiveness to EPI antigens
that IPTi may induce through a reduction in malaria incidence. In the randomised trial of IPTi
by Schellenberg et al.1, seropositivity to measles and pertussis vaccines was 10% lower
among infants who received sulfadoxine-pyrimethamine (SP) compared to placebo, and
although analysis of further samples was more reassuring, there is a need for more extensive
data on the impact of IPTi on serological responses especially to measles vaccination.
Analyses have been done separately for each trial – Navrongo, Manhica, Bungoma, Kisumu
and Kilimanjaro – and presented in a series of interim reports to the Advisory Committee.
This report presents the definitive analysis using data from four of the five studies. Bungoma
data have been excluded from this final analysis due to concerns about data quality.
The tables below summarize the intervention groups children were randomized into and the
serological information available from each trial.
TABLE 1 Randomized intervention groups by site for measles
Navrongo Manhica Kisumu Kilimanjaro Placebo X X X X SP X X X SP-ART X AQ-ART X MQ X LapDap X X
SP – sulfadoxine-pyrimethamine; ART – artesunate; AQ – amodiaquine; MQ – mefloquine; LapDap – chlorproguanil-dapsone
TABLE 2 Randomized intervention groups by site for other antigens*
Manhica Kisumu Placebo X X SP X SP-ART X AQ-ART X LapDap X
* Polio type 1 & 3, Diphtheria, Tetanus, Hepatitis-B, Pertussis FHA & Toxin
1 Schellenberg D, Menendez C, Kahigwa E, et al. Intermittent treatment for malaria and anaemia control at time of routine vaccinations in Tanzanian infants: a randomised, placebo-controlled trial. Lancet 2001;357(9267):1471-7
Final pooled analysis V.3 Navrongo, Manhica, Kisumu, Kilimanjaro 03/07/2009
9
See Appendix 12.1 for a useful summary to the trials.
2. Definitions
Protective concentrations are defined2 in the table below.
TABLE 3 Protective concentrations levels by antigen Measles 120 IU/L Polio Type 1 & 3 1:8 PRN titre Diphtheria 0.1 IU/ml Tetanus 0.1 IU/ml Hepatitis B 10 IU/L
There is no defined protective level for pertussis toxin and FHA.
3. Statistical methods, samples, datasets and strategies of analyses
For the purposes of the present pooled analysis report we have combined the four datasets
from Navrongo, Manhica, Kisumu and Kilimanjaro. The Navrongo study randomized clusters
of children, rather than children individually, into treatment groups. For the individual
Navrongo analysis report clustering was taken into consideration in all relevant analyses.
Manhica, Kisumu and Kilimanjaro on the other hand were individually randomized trials. To
see if we can ignore clustering for the Navrongo data (in order to pool with data from
Manhica) we have run the primary analysis (comparison of proportions protected in the two
intervention groups) twice; first time taking clustering into consideration and second time
ignoring it (results from these analyses are not reported here). There is no clustering effect in
the Navrongo data therefore it is ignored for the purposes of this analysis.
Both Intention-to-Treat (ITT) and According-to-Protocol (ATP) populations have been defined
and analysed in this report. All children with pre and post measles vaccination samples were
included in the ITT analysis. From the ATP analysis, children with incomplete drug dosing
were excluded. Hence, only Navrongo children with all four drug doses taken and Manhica,
Kisumu and Kilimanjaro children with all three drug doses taken have been considered for the
ATP. For the measles analyses we excluded children with: a) detectable and b) protective pre
vaccination levels.
Clinical, serology and randomization codes from all four trials were received between April
2005 and July 2008. Extensive data checking and cleaning took place with the help of the
laboratories and trial investigators. A draft statistical report was submitted on 26 February
2009.
For all antigens we included analyses on two outcomes: a) the continuous geometric mean
concentration (GMC) post-vaccination and b) the binary protected/unprotected based on
antigen levels above or below a pre-defined threshold of protection for each antigen, where
appropriate.
2 WHO Advisory Committee on EPI Serology in Relation to IPTi, Minutes, Meeting #2, Friday June 13, 2003, WHO Geneva
Final pooled analysis V.3 Navrongo, Manhica, Kisumu, Kilimanjaro 03/07/2009
10
Firstly, we investigated within treatment group across site comparisons. This investigation
resulted in: i) the placebo group comparison across Navrongo, Manhica, Kisumu and
Kilimanjaro for all antigens, ii) the SP group comparison across Navrongo, Manhica and
Kilimanjaro for measles and iii) the LapDap group comparison across Kisumu and Kilimanjaro
for measles.
Subsequently, we investigated between treatment groups across site comparisons. The
appropriate analyses for this were: i) SP versus placebo comparison for Navrongo, Manhica
and Kilimanjaro for measles, ii) LapDap versus placebo comparison for Kisumu and
Kilimanjaro for measles and iii) a post-hoc analysis of combined treatment groups (excluding
LapDap Kisumu) versus placebo for Navrongo, Manhica, Kisumu and Kilimanjaro. Data from
different sites were pooled only after establishing, by means of appropriate interaction tests,
that treatment effect was not modified in each site. If there was evidence of effect
modification, pooling was not done and site-specific treatment effects were reported.
Finally, we investigated all-treatment-combined-into-one versus placebo group comparisons
for individual sites. The appropriate analyses for this were: i) (SP-ART + AQ-ART + LapDap)
versus placebo in Kisumu for all antigens, ii) (SP + MQ + LapDap) versus placebo in
Kilimanjaro for measles and iii) a post-hoc analysis of (SP + MQ) versus placebo in
Kilimanjaro for measles. Data from different treatment groups were combined only after
establishing, by means of appropriate interaction tests, that individual treatment effects were
not different. If there was evidence of effect modification, pooling was not done and
individual treatment-specific effects were reported.
Final pooled analysis V.3 Navrongo, Manhica, Kisumu, Kilimanjaro 03/07/2009
11
4. Proportions protected and distributions of antibody concentrations in the placebo groups
4.1 Measles
In total there were 1,281 children assigned to the placebo groups with available serology
data in the four trials. We are restricting our analysis only on placebo groups and investigate
between site comparisons.
4.1.1 Analysis of post-vaccination measles antibody concentrations
223 children had detectable concentrations in their pre-vaccination sample, 103 of these had
protective (≥ 120IU/L) concentration of measles antibodies in their pre-vaccination sample.
The next table summarises the geometric mean concentrations (GMC) in the four sites before
and after vaccination, and the median post-vaccination concentration.
TABLE 4 Antibody concentration response to the measles vaccination by treatment group GMC
pre-vaccination (95%CI)
GMC post-vaccination (95%CI) *
Median post-vaccination
Children with detectable concentrations pre-vaccination Navrongo (n=64) 241 (178, 327) 2260 (1660, 3075) 2141 Manhica (n=77) 222 (149, 331) 1832 (1299, 2584) 1984 Kisumu (n=36) 56 (37, 86) 755 (500, 1141) 945
Kilimanjaro (n=46) 76 (53, 108) 704 (494, 1003) 761 Children with undetectable concentrations pre-vaccination
Navrongo (n=220) 1439 (1213, 1707) 1424 Manhica (n=239) 962 (800, 1156) 896 Kisumu (n=248) 611 (529, 707) 633
Kilimanjaro (n=351) 621 (555, 696) 700 Ratio Manhica/Navrongo 0.67 (0.53, 0.84) Ratio Kisumu/Navrongo 0.42 (0.34, 0.53) P-value1 <0.0001 Ratio Kilimanjaro/Navrongo 0.43 (0.35, 0.53) Children with protective measles antibody level pre-vaccination
Navrongo (n=43) 417 (297, 586) 3010 (2083, 4350) 3050 Manhica (n=40) 714 (414, 1233) 2590 (1438, 4665) 3232
Kisumu (n=7) 536 (302, 952) 1130 (590, 2163) 1523 Kilimanjaro (n=13) 373 (194, 716) 1265 (604, 2651) 1588
Children without protective measles antibody level pre-vaccination Navrongo (n=241) 1422 (1210, 1671) 1422 Manhica (n=276) 998 (846, 1176) 975 Kisumu (n=277) 619 (538, 711) 637
Kilimanjaro (n=384) 616 (552, 689) 697 Ratio Manhica/Navrongo 0.70 (0.57, 0.87) Ratio Kisumu/Navrongo 0.43 (0.35, 0.54) P-value1 <0.0001 Ratio Kilimanjaro/Navrongo 0.43 (0.36, 0.53) All children
Navrongo (n=284) 1593 (1371, 1851) 1639 Manhica (n=316) 1126 (955, 1327) 1127 Kisumu (n=284) 628 (548, 720) 650
Kilimanjaro (n=397) 630 (566, 702) 701 Ratio Manhica/Navrongo 0.71 (0.58, 0.86) Ratio Kisumu/Navrongo 0.39 (0.32, 0.48) P-value1 <0.0001 Ratio Kilimanjaro/Navrongo 0.40 (0.33, 0.48)
*26 children had non-detectable concentrations post-vaccination. These undetectable levels were assigned a value of 11IU/L, half of the minimum concentration detected in the pooled dataset, for purposes of the analysis in this table (this affects the GMC but not the median). 1 P-value is calculated from the likelihood ratio test comparing the regression model with site as the only factor to its nested null model. H0=all ratios equal to 1
Final pooled analysis V.3 Navrongo, Manhica, Kisumu, Kilimanjaro 03/07/2009
12
Post measles vaccination geometric mean concentration values were highest in the placebo
group from Navrongo site. This finding remained consistent throughout the different sub-
population investigations. Manhica also consistently remained the second highest site. Finally,
Kisumu and Kilimanjaro produced very similar antibody measles concentrations.
FIGURE 1 Reverse cumulative distribution function for measles for all children
0.2
.4.6
.81
1 2 3 4 5log(10) measles antibody concentration post-vaccination
Navrongo ManhicaKisumu Kilimanjaro
(vertical line indicates assumed protective threshold)Reverse empirical cumulative distribution function
4.1.2 Proportion of children with protective concentrations post-vaccination
4.1.2.1 Excluding those with protective measles concentrations in their pre-vaccination sample
103 children had protective concentrations of measles antibody in their pre-vaccination
sample. These children have been excluded from this point on, leaving 1,178 individuals.
TABLE 5 ITT proportion of children protected from measles; excluding those protected at pre-vaccination Protected Unprotected Total Percent unprotected Navrongo 233 8 241 3.32% Manhica 261 15 276 5.43% Kisumu 262 15 277 5.42% Kilimanjaro 361 23 384 5.99%
H0: Percentages unprotected equal across sites, 23Χ =2.28, P-value=0.52
Final pooled analysis V.3 Navrongo, Manhica, Kisumu, Kilimanjaro 03/07/2009
13
A further 31 children are excluded from the ATP analysis, leaving 1,147 individuals.
TABLE 6 ATP proportion of children protected from measles; excluding those protected at pre-vaccination Protected Unprotected Total Percent unprotected Navrongo 225 7 232 3.02% Manhica 250 15 265 5.66% Kisumu 259 14 273 5.13% Kilimanjaro 355 22 377 5.84%
H0: Percentages unprotected equal across sites, 23Χ =2.69, P-value=0.44
No evidence has been found to support differences in the percentages of unprotected
children across sites for the ITT or the ATP analysis.
FIGURE 2 Reverse cumulative distribution function for measles for the ITT cohort excluding those protected at pre-vaccination
0.2
.4.6
.81
1 2 3 4 5log(10) measles antibody concentration post-vaccination
Navrongo ManhicaKisumu Kilimanjaro
(vertical line indicates assumed protective threshold)Reverse empirical cumulative distribution function
4.1.2.2 Excluding those with detectable measles concentrations in their pre-vaccination sample
223 children had detectable concentrations of measles antibody in their pre-vaccination
sample. These children have been excluded from this point on, leaving 1,058 individuals.
TABLE 7 ITT proportion of children protected from measles; excluding those detectable at pre-vaccination Protected Unprotected Total Percent unprotected Navrongo 212 8 220 3.64% Manhica 224 15 239 6.28% Kisumu 235 13 248 5.24% Kilimanjaro 331 20 351 5.70%
H0: Percentages unprotected equal across sites, 23Χ =1.78, P-value=0.62
A further 24 children are excluded from the ATP analysis, leaving 1,034 individuals.
Final pooled analysis V.3 Navrongo, Manhica, Kisumu, Kilimanjaro 03/07/2009
14
TABLE 8 ATP proportion of children protected from measles; excluding those detectable at pre-vaccination Protected Unprotected Total Percent unprotected Navrongo 204 7 211 3.32% Manhica 216 15 231 6.49% Kisumu 234 12 246 4.88% Kilimanjaro 326 20 346 5.78%
H0: Percentages unprotected equal across sites, 23Χ =2.58, P-value=0.46
No evidence has been found to support differences in the percentages of unprotected
children across sites for the ITT or the ATP analysis.
FIGURE 3 Reverse cumulative distribution function for measles for the ITT cohort excluding children with detectable antibody levels at pre-vaccination
0.2
.4.6
.81
1 2 3 4 5log(10) measles antibody concentration post-vaccination
Navrongo ManhicaKisumu Kilimanjaro
(vertical line indicates assumed protective threshold)Reverse empirical cumulative distribution function
Final pooled analysis V.3 Navrongo, Manhica, Kisumu, Kilimanjaro 03/07/2009
15
4.2 Diphtheria, Tetanus, Polio type 1 & 3, Pertussis toxin and FHA, and Hepatitis B responses in placebo groups
For the purposes of this analysis data from Manhica and Kisumu are being used. Serology data for these antigens is available for a total of 634 for the ITT
and 567 for the ATP analysis. Complete data on all antigens was available for 150 children.
4.2.1 ITT analysis; post-vaccination antibody concentrations & proportion protected
TABLE 9 ITT analysis for Diphtheria, Tetanus, Polio type 1 & 3, Pertussis toxin and FHA, and Hepatitis B; post-vaccination concentrations & proportions unprotected Polio Type 1
(n=499) Polio Type 3 (n=499)
Diphtheria (n=590)
Pertussis FHA (n=551)
Pertussis Toxin (n=388)
Tetanus (n=426)
Hep-B (n=318)
Protective threshold 1 in 8 PRN 1 in 8 PRN 0.1IU/ml Not defined (US units)
Not defined (US units)
0.1 IU/ml 10 IU/L
Percent unprotected (below threshold)
Manhica 27/250 10.80% 45/250 18.00%
4/331 1.21%
3/247 1.21%
9/250 3.60%
Kisumu 49/249 19.68% 95/249 38.15% 10/259 3.86% 1/179 0.56%
0/68 0.00%
P-value£ P=0.006 P<0.0001 P=0.04 P=0.49 P=0.11 GMC (95%CI)
Manhica 1.66 (1.52,1.82) n=331
30.1 (26.9,33.7) n=316
184 (157,217) n=227
5.46 (4.64,6.42) n=247
1068 (820,1391) n=250
Kisumu 0.75 (0.66,0.85) n=259
14.7 (12.9,16.7) n=235
67 (50,89) n=161
2.97 (2.52,3.52) n=179
505 (332,767) n=68
Ratio Kisumu/Manhica α (95% CI) P-value*
0.45 (0.39,0.52) P<0.0001
0.49 (0.41,0.58) P<0.0001
0.36 (0.27,0.49) P<0.0001
0.54 (0.43,0.54) P<0.0001
0.47 (0.27,0.82) P<0.0001
£ P-value is based on a chi-square comparison of the percentages of unprotected children in Manhica and Navrongo; α All ratio values are estimated from regression models with intervention arm as
the only factor;* This is the Wald test P-value from the regression model with site as the only factor. 1: 10 =Η r , where 1r is the ratio of ln GMCs for Kisumu compared to Manhica.
Final pooled analysis V.3 Navrongo, Manhica, Kisumu, Kilimanjaro 03/07/2009
16
4.2.2 ATP analysis; post-vaccination concentrations & proportion protected
TABLE 10 ATP analysis for Diphtheria, Tetanus, Polio type 1 & 3, Pertussis toxin and FHA, and Hepatitis B; post-vaccination concentrations & proportions unprotected Polio Type 1
(n=450) Polio Type 3 (n=450)
Diphtheria (n=528)
Pertussis FHA (n=497)
Pertussis Toxin (n=343)
Tetanus (n=385)
Hep-B (n=290)
Protective threshold 1 in 8 PRN 1 in 8 PRN 0.1IU/ml Not defined (US units)
Not defined (US units)
0.1 IU/ml 10 IU/L
Percent unprotected (below threshold)
Manhica 26/228 11.40% 41/228 17.98%
2/298 0.67%
3/224 1.34%
7/226 3.10%
Kisumu 41/222 18.47% 79/222 35.59% 8/230 3.48%
0/161 0.00%
0/64 0.00%
P-value£ P=0.04 P<0.0001 P=0.02 P=0.14 P=0.15 GMC (95%CI)
Manhica 1.69 (1.54,1.85) n=298
29.1 (26.0,32.7) n=387
184 (155,219) n=203
5.43 (4.57,6.46) n=224
1101 (842,1440) n=226
Kisumu 0.76 (0.66,0.87) n=230
14.9 (13.0,17.0) n=210
64 (48,86) n=140
2.88 (2.44,3.41) n=161
495 (319,769) n=64
Ratio Kisumu/Manhica α (95% CI) P-value*
0.45 (0.38,0.53) P<0.0001
0.51 (0.43,0.61) P<0.0001
0.35 (0.25,0.48) P<0.0001
0.53 (0.41,0.68) P<0.0001
0.45 (0.26,0.78) P<0.0001
£ P-value is based on a chi-square comparison of the percentages of unprotected children in Manhica and Navrongo; α All ratio values are estimated from regression models with intervention arm as
the only factor; * This is the Wald test P-value from the regression model with site as the only factor. 1: 10 =Η r , where 1r is the ratio of ln GMCs for Kisumu compared to Manhica.
Final pooled analysis V.3 Navrongo, Manhica, Kisumu, Kilimanjaro 03/07/2009
17
Kisumu had a significantly higher percentage of unprotected children for polio types 1 & 3
and diphtheria when compared to Manhica. For tetanus and hepatitis B there was no
evidence for a difference in the respective percentages. These results were the same for both
ITT and ATP. For diphtheria, pertussis toxin and FHA, tetanus and hepatitis B Manhica
children had on average a higher geometric mean concentration after vaccination when
compared to Kisumu children.
FIGURE 4 Reverse cumulative distribution function for diphtheria
0.2
.4.6
.81
-1.5 -1 -.5 0 .5 1log(10) diphtheria antibody concentration IU/ml
Manhica Kisumu
(vertical line indicates assumed protective threshold)Reverse empirical cumulative distribution function
Final pooled analysis V.3 Navrongo, Manhica, Kisumu, Kilimanjaro 03/07/2009
18
FIGURE 5 Reverse cumulative distribution function for pertussis FHA
0.2
.4.6
.81
0 1 2 3log(10) pertussis FHA antibody concentration IU/ml
Manhica Kisumu
Reverse empirical cumulative distribution function
FIGURE 6 Reverse cumulative distribution function for pertussis toxin
0.2
.4.6
.81
-1 0 1 2 3log(10) pertussis toxin antibody concentration IU/ml
Manhica Kisumu
Reverse empirical cumulative distribution function
Final pooled analysis V.3 Navrongo, Manhica, Kisumu, Kilimanjaro 03/07/2009
19
FIGURE 7 Reverse cumulative distribution function for tetanus
0.2
.4.6
.81
-2 -1 0 1 2log(10) tetanus antibody concentration IU/ml
Manhica Kisumu
(vertical line indicates assumed protective threshold)Reverse empirical cumulative distribution function
FIGURE 8 Reverse cumulative distribution function for hepatitis B
.2.4
.6.8
1
0 1 2 3 4log(10) Hepatitis B antibody concentration IU/L
Manhica Kisumu
(all children are above assumed protective threshold)Reverse empirical cumulative distribution function
Final pooled analysis V.3 Navrongo, Manhica, Kisumu, Kilimanjaro 03/07/2009
20
4.2.3 Children with protective concentrations to multiple antigens
In the ITT analysis, 217 children had results for diphtheria, tetanus, polio type 1, polio type 3
and hepatitis B. In Manhica, 73.3% (126/172) had protective level for all five vaccine types,
compared to 48.9% (22/45) in Kisumu, a difference of -24.4% (90% CI -37.8%, -10.9%).
There was strong evidence found to suggest the proportion of these children with protective
concentrations of antibodies to all five when compared to those protected in four or less was
not the same in the two sites (test of association 21Χ =9.8, P=0.002).
TABLE 11 Number with protective concentration to 4 or less & 5 antigens: N (%) 4or less 5 TOTAL Manhica 46 (67) 126 (85) 172 (79) Kisumu 23 (33) 22 (15) 45 (21) TOTAL 69 (100) 348(100) 217 (100)
Final pooled analysis V.3 Navrongo, Manhica, Kisumu, Kilimanjaro 03/07/2009
21
5. Proportions protected and distributions of antibody concentrations in the SP groups
5.1 Measles
In total there were 1,018 children assigned to the SP groups in the three (Navrongo, Manhica
and Kilimanjaro) trials. We are restricting our analysis in this section to SP groups, and
investigate between site comparisons.
5.1.1 Analysis of post-vaccination measles antibody concentrations
182 children had detectable concentrations in their pre-vaccination sample, 92 of these had
protective (≥ 120IU/L) concentration of measles antibodies in their pre-vaccination sample.
The next table summarises the geometric mean concentrations (GMC) in the three sites
before and after vaccination, and the median post-vaccination concentration.
TABLE 12 Antibody concentration response to the measles vaccination by treatment group GMC
pre-vaccination (95%CI)
GMC post-vaccination (95%CI) *
Median post-vaccination
Children with detectable concentrations pre-vaccination Navrongo (n=82) 249 (191, 323) 2435 (1953, 3036) 2245 Manhica (n=54) 200 (125, 321) 1564 (970, 2520) 1485
Kilimanjaro (n=46) 71 (54, 92) 519 (345, 782) 444 Children with undetectable concentrations pre-vaccination
Navrongo (n=254) 1549 (1305, 1839) 1724 Manhica (n=254) 1008 (848, 1199) 1158
Kilimanjaro (n=328) 686 (611, 769) 733 Ratio Manhica/Navrongo 0.65 (0.52, 0.81) P-value1 <0.0001 Ratio Kilimanjaro/Navrongo 0.44 (0.36, 0.54) Children with protective measles antibody level pre-vaccination
Navrongo (n=54) 425 (311, 580) 2578 (1961, 3389) 2317 Manhica (n=27) 631 (315, 1266) 3293 (1720, 6305) 2606
Kilimanjaro (n=11) 254 (165, 389) 1172 (500, 2746) 608 Children without protective measles antibody level pre-vaccination
Navrongo (n=282) 1603 (1367, 1879) 1746 Manhica (n=281) 979 (829, 1156) 1075
Kilimanjaro (n=363) 651 (582, 729) 705 Ratio Manhica/Navrongo 0.61 (0.50, 0.75) P-value1 <0.0001 Ratio Kilimanjaro/Navrongo 0.41 (0.33, 0.50) All children
Navrongo (n=336) 1730 (1502, 1992) 1915 Manhica (n=308) 1089 (923, 1285) 1205
Kilimanjaro (n=374) 663 (592, 741) 704 Ratio Manhica/Navrongo 0.63 (0.52, 0.77) P-value1 <0.0001 Ratio Kilimanjaro/Navrongo 0.38 (0.32, 0.46)
*19 children had non-detectable concentrations post-vaccination. These undetectable levels were assigned a value of 11IU/L, half of the minimum concentration detected in the pooled dataset, for purposes of the analysis in this table (this affects the GMC but not the median). 1 P-value is calculated from the likelihood ratio test comparing the regression model with site as the only factor to its nested null model. H0=all ratios equal to 1
Post measles vaccination geometric mean concentration values were highest in the SP group
from Navrongo site. This finding remained consistent throughout the different sub-population
investigations. Manhica also consistently remained the second highest site followed by
Kilimanjaro.
Final pooled analysis V.3 Navrongo, Manhica, Kisumu, Kilimanjaro 03/07/2009
22
FIGURE 9 Reverse cumulative distribution function for measles for all children
0.2
.4.6
.81
1 2 3 4 5log(10) measles antibody concentration post-vaccination
Navrongo ManhicaKilimanjaro
(vertical line indicates assumed protective threshold)Reverse empirical cumulative distribution function
5.1.2 Proportion of children with protective concentrations post-vaccination
5.1.2.1 Excluding those with protective measles concentrations in their pre-vaccination sample
92 children had protective concentrations of measles antibody in their pre-vaccination
sample. These children have been excluded from this point on, leaving 926 individuals.
TABLE 13 ITT proportion of children protected from measles; excluding those protected at pre-vaccination Protected Unprotected Total Percent unprotected Navrongo 274 8 282 2.84% Manhica 261 20 281 7.12% Kilimanjaro 342 21 363 5.79%
H0: Percentages unprotected equal across sites, 22Χ =5.44, P-value=0.07
A further 26 children are excluded from the ATP analysis, leaving 900 individuals.
TABLE 14 ATP proportion of children protected from measles; excluding those protected at pre-vaccination Protected Unprotected Total Percent
unprotected Navrongo 264 8 272 2.94% Manhica 253 20 273 7.33% Kilimanjaro 334 21 355 5.92%
H0: Percentages unprotected equal across sites, 22Χ =5.34, P-value=0.07
No strong evidence has been found to support differences in the percentages of unprotected
children across sites for the ITT or the ATP analysis.
Final pooled analysis V.3 Navrongo, Manhica, Kisumu, Kilimanjaro 03/07/2009
23
FIGURE 10 Reverse cumulative distribution function for measles for the ITT cohort excluding those protected at pre-vaccination
0.2
.4.6
.81
1 2 3 4 5log(10) measles antibody concentration post-vaccination
Navrongo ManhicaKilimanjaro
(vertical line indicates assumed protective threshold)Reverse empirical cumulative distribution function
5.1.2.2 Excluding those with detectable measles concentrations in their pre-vaccination sample
182 children had detectable concentrations of measles antibody in their pre-vaccination
sample. These children have been excluded from this point on, leaving 836 individuals.
TABLE 15 ITT proportion of children protected from measles; excluding those detectable at pre-vaccination Protected Unprotected Total Percent unprotected Navrongo 246 8 254 3.15% Manhica 237 17 254 6.69% Kilimanjaro 311 17 328 5.18%
H0: Percentages unprotected equal across sites, 22Χ =3.37, P-value=0.19
A further 25 children are excluded from the ATP analysis, leaving 811 individuals.
TABLE 16 ATP proportion of children protected from measles; excluding those detectable at pre-vaccination Protected Unprotected Total Percent unprotected Navrongo 236 8 244 3.28% Manhica 230 17 247 6.88% Kilimanjaro 303 17 320 5.31%
H0: Percentages unprotected equal across sites, 22Χ =3.27, P-value=0.20
No evidence has been found to support differences in the percentages of unprotected
children across sites for the ITT or the ATP analysis.
Final pooled analysis V.3 Navrongo, Manhica, Kisumu, Kilimanjaro 03/07/2009
24
FIGURE 11 Reverse cumulative distribution function for measles for the ITT cohort excluding children with detectable antibody levels at pre-vaccination
0.2
.4.6
.81
1 2 3 4 5log(10) measles antibody concentration post-vaccination
Navrongo ManhicaKilimanjaro
(vertical line indicates assumed protective threshold)Reverse empirical cumulative distribution function
Final pooled analysis V.3 Navrongo, Manhica, Kisumu, Kilimanjaro 03/07/2009
25
6. Proportions protected and distributions of antibody concentrations in the LapDap groups
6.1 Measles
In total there were 647 children assigned to the LapDap groups in the four trials. We are
restricting our analysis in this section only on LapDap groups and investigate between site
comparisons.
6.1.1 Analysis of post-vaccination measles antibody concentrations
66 children had detectable concentrations in their pre-vaccination sample, 25 of these had
protective (≥ 120IU/L) concentration of measles antibodies in their pre-vaccination sample.
The next table summarises the geometric mean concentrations (GMC) in the two sites before
and after vaccination, and the median post-vaccination concentration.
TABLE 17 Antibody concentration response to the measles vaccination by treatment group GMC
pre-vaccination (95%CI)
GMC post-vaccination (95%CI) *
Median post-vaccination
Children with detectable concentrations pre-vaccination Kisumu (n=33) 111 (60,205) 761 (452, 1282) 806
Kilimanjaro (n=33) 88 (58,135) 479 (305, 753) 554 Children with undetectable concentrations pre-vaccination
Kisumu (n=227) 712 (630, 805) 702 Kilimanjaro (n=354) 608 (540, 686) 678
Ratio Kisumu/Kilimanjaro 0.85 (0.71, 1.02) P-value1 0.08 Children with protective measles antibody level pre-vaccination
Kisumu (n=13) 559 (200,1564) 1448 (431, 4866) 1768 Kilimanjaro (n=12) 314 (166, 596) 1090 (610, 1948) 1127
Children without protective measles antibody level pre-vaccination Kisumu (n=247) 692 (616, 777) 675
Kilimanjaro (n=375) 585 (520, 658) 666 Ratio Kisumu/Kilimanjaro 0.85 (0.71, 1.00) P-value1 0.06 All children
Kisumu (n=260) 718 (634, 813) 704 Kilimanjaro (n=387) 596 (531, 670) 667
Ratio Kisumu/Kilimanjaro 0.83 (0.70, 0.99) P-value1 0.04 *11 children had non-detectable concentrations post-vaccination. These undetectable levels were assigned a value of 11IU/L, half of the minimum concentration detected in the pooled dataset, for purposes of the analysis in this table (this affects the GMC but not the median). 1 P-value is calculated from the likelihood ratio test comparing the regression model with site as the only factor to its nested null model. H0=all ratios equal to 1
Post measles vaccination geometric mean concentration values were highest in the LapDap
group from Kisumu site when compared to Kilimanjaro. This finding remained consistent
throughout the different sub-population investigations with some evidence for statistical
significance.
Final pooled analysis V.3 Navrongo, Manhica, Kisumu, Kilimanjaro 03/07/2009
26
FIGURE 12 Reverse cumulative distribution function for measles for all children
0.2
.4.6
.81
1 2 3 4log(10) measles antibody concentration post-vaccination
Kisumu Kilimanjaro
(vertical line indicates assumed protective threshold)Reverse empirical cumulative distribution function
6.1.2 Proportion of children with protective concentrations post-vaccination
6.1.2.1 Excluding those with protective measles concentrations in their pre-vaccination sample
25 children had protective concentrations of measles antibody in their pre-vaccination
sample. These children have been excluded from this point on, leaving 622 individuals.
TABLE 18 ITT proportion of children protected from measles; excluding those protected at pre-vaccination Protected Unprotected Total Percent unprotected Kisumu 243 4 247 1.62% Kilimanjaro 343 32 375 8.53%
H0: Percentages unprotected equal across sites, 21Χ =13.1, P-value<0.0001
A further 14 children are excluded from the ATP analysis, leaving 608 individuals.
TABLE 19 ATP proportion of children protected from measles; excluding those protected at pre-vaccination Protected Unprotected Total Percent unprotected Kisumu 240 4 244 1.64% Kilimanjaro 334 30 364 8.24%
H0: Percentages unprotected equal across sites, 21Χ =12.1, P-value=0.001
Strong evidence has been found to support a statistically significantly higher percentage of
unprotected children in Kilimanjaro compared to Kisumu, for both the ITT and ATP analysis.
Final pooled analysis V.3 Navrongo, Manhica, Kisumu, Kilimanjaro 03/07/2009
27
FIGURE 13 Reverse cumulative distribution function for measles for the ITT cohort excluding those protected at pre-vaccination
0.2
.4.6
.81
1 2 3 4log(10) measles antibody concentration post-vaccination
Kisumu Kilimanjaro
(vertical line indicates assumed protective threshold)Reverse empirical cumulative distribution function
6.1.2.2 Excluding those with detectable measles concentrations in their pre-vaccination sample
66 children had detectable concentrations of measles antibody in their pre-vaccination
sample. These children have been excluded from this point on, leaving 581 individuals.
TABLE 20 ITT proportion of children protected from measles; excluding those detectable at pre-vaccination Protected Unprotected Total Percent unprotected Kisumu 223 4 227 1.76% Kilimanjaro 329 25 354 7.06%
H0: Percentages unprotected equal across sites, 21Χ =8.2, P-value=0.004
A further 10 children are excluded from the ATP analysis, leaving 571 individuals.
TABLE 21 ATP proportion of children protected from measles; excluding those detectable at pre-vaccination Protected Unprotected Total Percent unprotected Kisumu 220 4 224 1.79% Kilimanjaro 322 25 347 7.20%
H0: Percentages unprotected equal across sites, 21Χ =8.3, P-value=0.004
Strong evidence has been found to support a statistically significantly higher percentage of
unprotected children in Kilimanjaro compared to Kisumu, for both the ITT and ATP analysis.
Final pooled analysis V.3 Navrongo, Manhica, Kisumu, Kilimanjaro 03/07/2009
28
FIGURE 14 Reverse cumulative distribution function for measles for the ITT cohort excluding children with detectable antibody levels at pre-vaccination
0.2
.4.6
.81
1 2 3 4log(10) measles antibody concentration post-vaccination
Kisumu Kilimanjaro
(vertical line indicates assumed protective threshold)Reverse empirical cumulative distribution function
Final pooled analysis V.3 Navrongo, Manhica, Kisumu, Kilimanjaro 03/07/2009
29
7. Summary of within group comparisons across sites
TABLE 22 Summary of measles results from comparisons for each treatment across sites where appropriate
GMC post-vaccination ratios $ Percentage Unprotected £
MEASLES Excluding detectable
pre-vaccination
Excluding protected
pre-vaccination
Excluding detectable
pre-vaccination
Excluding protected
pre-vaccination Placebo
Navrongo Manhica Kisumu
Kilimanjaro
1
0.67 0.42 0.43
1
0.70 0.43 0.43
No difference across sites
No difference across sites
SP Navrongo Manhica
Kilimanjaro
1
0.65 0.44
1
0.61 0.41
No difference across sites
No difference across sites (P=0.07)
LapDap Kisumu
Kilimanjaro
No difference across sites (P=0.08)
No difference across sites (P=0.06)
1.76% 7.06%
1.62% 8.53%
$ Geometric mean concentrations, post measles vaccination, are compared across sites. Comparisons are done with ratio GMC values drawn from regression models with site as the only factor; £ Percentages of unprotected (<120 IU/L) children post measles vaccination are compared across sites; ITT population Percentages of unprotected children post measles vaccination were not found to be different
across sites for the placebo or the SP within group comparisons. In the LapDap groups
however, Kisumu had a significantly lower percentage of unprotected children when
compared to the Kilimanjaro group.
Geometric mean concentration ratio comparisons found strong evidence of differences across
sites for both the placebo and SP groups. We also found weak evidence of a difference in
GMC between Kisumu and Kilimanjaro for the LapDap comparison.
Final pooled analysis V.3 Navrongo, Manhica, Kisumu, Kilimanjaro 03/07/2009
30
8. Pooled SP Vs Placebo comparison for measles
In this section of the report we will investigate the treatment effect of SP compared to
placebo using data from three sites; Navrongo, Manhica and Kilimanjaro. Before we pool data
from these three sites we will investigate whether the effect of SP Vs placebo is modified by
site. If effect modification is found pooling of the data is not appropriate and will not be done.
8.1 Investigation of effect modification by site
TABLE 23 SP Vs Placebo group comparison for measles by site and investigation of effect modification
GMC post-vaccination ratios $ OR Unprotected £
MEASLES Excluding detectable
pre-vaccination (n=1,646)
Excluding protected
pre-vaccination (n=1,827)
Excluding detectable
pre-vaccination (n=1,646)
Excluding protected
pre-vaccination (n=1,827)
Treatment Placebo
SP
1
1.08 (0.95-1.22)
1
1.05 (0.94-1.18)
1
0.95 (0.62-1.48)
1
1.06 (0.70-1.61) Site
Navrongo Manhica
Kilimanjaro
1
0.66 (0.56-0.77) 0.44 (0.38-0.51)
1
0.65 (0.56-0.76) 0.42 (0.36-0.48)
1
1.99 (1.07-3.67) 1.65 (0.90-3.00)
1
2.13 (1.16-3.90) 1.99 (1.11-3.57)
Interaction Treatment X Site P=0.94* P=0.66* P=0.92* P=0.70*
$ Geometric mean concentrations ratios of post measles vaccination are compared. Comparisons are done with ratio GMC values drawn from regression models with site and treatment group as the only factors £ Odds ratios of unprotected (<120 IU/L) children post measles vaccination are compared with site and treatment group as the only factors; ITT population * Likelihood ratio test comparing the model with to the model without the interaction terms between site & treatment
We have found no evidence for effect modification. Therefore, we will pool data across the
three sites to study the overall SP treatment effect.
8.2 Measles
In total 2,015 children are included in this analysis; 620 in Navrongo, 624 in Manhica, and
771 in Kilimanjaro. Measles vaccination is given at 9 months of age in all three sites. Children
were bled at 9 months of age before measles vaccination was administered to establish
baseline measles serology levels. Vaccine responses were measured by a subsequent blood
sample drawn at 10 months (Kilimanjaro) and 12 months of age (Manhica and Navrongo).
Measles serology data with matched pre and post vaccination measurements was available
for 2,015 children. Randomisation resulted in 997 in Placebo and 1,018 in SP groups. All
2,015 children were included in the ITT analyses. From the ATP analyses children were
excluded for not having received all drug doses.
TABLE 24 Numbers of children by study and treatment group for measles
Placebo SP TOTAL Navrongo 284 336 620 Manhica 316 308 624 Kilimanjaro 397 374 771 TOTAL 997 1,018 2,015
Final pooled analysis V.3 Navrongo, Manhica, Kisumu, Kilimanjaro 03/07/2009
31
8.2.1 Analysis of post-vaccination measles antibody concentrations
Characteristics of the 2,015 children, sampled at 9 months, included in the ITT analysis are as
follows:
TABLE 25 Baseline characteristics of children included in the ITT analysis Placebo
N=997 SP
N=1,018 Age in months at vaccination (mean, range) 9 (8, 12) 9 (8, 12)
%male : %female 50%:50% 51%:49%
369 children had detectable concentrations in their pre-vaccination sample, 188 of these had
protective (≥ 120IU/L) concentration of measles antibodies in their pre-vaccination sample.
The next table summarises the geometric mean concentrations (GMC) in the two treatment
groups before and after vaccination, and the median post-vaccination concentration.
TABLE 26 Antibody concentration response to the measles vaccination by treatment group GMC
pre-vaccination (95%CI)
GMC post-vaccination (95%CI) *
Median post-vaccination
Children with detectable concentrations pre-vaccination Placebo (n=187) 175 (140,219) 1556 (1268, 1909) 1522
SP (n=182) 170 (138,208) 1445 (1162, 1796) 1487 Children with undetectable concentrations pre-vaccination
Placebo (n=810) 888 (812, 971) 910 SP (n=836) 988 (903, 1080) 1048
Ratio SP/Placebo 1.08 (0.95, 1.22) P-value1 0.22 Children with protective measles antibody level pre-vaccination
Placebo (n=96) 514 (387, 683) 2514 (1850, 3417) 2354 SP (n=92) 449 (341, 591) 2521 (1936, 3282) 2294
Children without protective measles antibody level pre-vaccination Placebo (n=901) 893 (821, 971) 919
SP (n=926) 970 (890, 1057) 1026 Ratio SP/Placebo 1.05 (0.94, 1.18) P-value1 0.38 All children
Placebo (n=997) 987 (908, 1072) 1001 SP (n=1,018) 1057 (973, 1149) 1091
Ratio SP/Placebo 1.03 (0.92, 1.16) P-value1 0.55 *38 children had non-detectable concentrations post-vaccination. These undetectable levels were assigned a value of 11IU/L, half of the minimum concentration detected in the pooled dataset, for purposes of the analysis in this table (this affects the GMC but not the median). 1 P-value is calculated from the likelihood ratio test comparing the regression model with treatment group and site as the only factors to the model with only site. H0= ratio equal to 1
No evidence has been found to support a difference in the GMC between SP and placebo
groups.
Final pooled analysis V.3 Navrongo, Manhica, Kisumu, Kilimanjaro 03/07/2009
32
FIGURE 15 Reverse cumulative distribution function for measles for all children
0.2
.4.6
.81
1 2 3 4 5log(10) measles antibody concentration post-vaccination
Placebo SP
(vertical line indicates assumed protective threshold)Reverse empirical cumulative distribution function
8.2.2 Proportion of children with protective concentrations post-vaccination
8.2.2.1 Excluding those with protective measles concentrations in their pre-vaccination sample
188 children had protective concentrations of measles antibody in their pre-vaccination
sample. These children have been excluded from this point on, leaving 1,827 individuals.
TABLE 27 ITT proportion of children protected from measles; excluding those protected at pre-vaccination Protected Unprotected Total Percent unprotected Placebo 855 46 901 5.11% SP 877 49 926 5.29% Null hypothesis difference SP-Placebo of 5% Actual difference 0.18%, P<0.0001, z=-4.33 90%CI (-1.52, 1.89) 95%CI (-1.85, 2.22) 99%CI (-2.49, 2.86)
A further 53 children are excluded from the ATP analysis, leaving 1,774 individuals.
TABLE 28 ATP proportion of children protected from measles; excluding those protected at pre-vaccination Protected Unprotected Total Percent unprotected Placebo 830 44 874 5.03% SP 851 49 900 5.44% Null hypothesis difference SP-Placebo of 5% Actual difference 0.41%, P<0.0001, z=-4.07 90%CI (-1.33, 2.15) 95%CI (-1.66, 2.48) 99%CI (-2.31, 3.13)
Final pooled analysis V.3 Navrongo, Manhica, Kisumu, Kilimanjaro 03/07/2009
33
The formal test of non-inferiority3, of the null hypothesis that the difference in proportion
unprotected between the intervention groups is 5% or more, gives strong evidence
suggesting SP is not inferior to placebo for both ITT and ATP.
FIGURE 16
Reverse cumulative distribution function for measles for the ITT cohort excluding those protected at pre-vaccination
0.2
.4.6
.81
1 2 3 4 5log(10) measles antibody concentration post-vaccination
Placebo SP
(vertical line indicates assumed protective threshold)Reverse empirical cumulative distribution function
8.2.2.2 Excluding those with detectable measles concentrations in their pre-vaccination sample
369 children had protective concentrations of measles antibody in their pre-vaccination
sample. These children have been excluded from this point on, leaving 1,646 individuals.
TABLE 29 ITT proportion of children protected from measles; excluding those with detectable concentration at pre-vaccination Protected Unprotected Total Percent unprotected Placebo 767 43 810 5.31% SP 794 42 836 5.02% Null hypothesis difference SP-Placebo of 5% Actual difference -0.29%, P<0.0001, z=-4.50 90%CI (-2.08, 1.51) 95%CI (-2.42, 1.85) 99%CI (-3.10, 2.53)
3 The one-sided P-values were calculated by referring the statistic: d = (pB-pA-d0)/√v0
to the standard normal distribution, P=Prob(z<d). pA is the observed proportion unprotected in placebo and pB the proportion unprotected in SP-ART, and d0 is the difference under the null hypothesis (d0=0.05 for measles and 0.1 for the other antigens), and v0 is the estimated variance of pB-pA under the null hypothesis. (In terms of the proportions protected qA and qB, d = (qA-qB-d0)/√v0 ). 100(1-α)% confidence intervals for the risk difference were calculated as pB-pA±z1-α/2√v where v=pA(1-pA)/NA+pB(1-pB)/NB. In the tables, percentages (100pA% etc) are presented rather than proportions.
Final pooled analysis V.3 Navrongo, Manhica, Kisumu, Kilimanjaro 03/07/2009
34
A further 47 children are excluded from the ATP analysis, leaving 1,599 individuals.
TABLE 30 ATP proportion of children protected from measles; excluding those with detectable concentration at pre-vaccination Protected Unprotected Total Percent unprotected Placebo 746 42 788 5.33% SP 769 42 811 5.18% Null hypothesis difference SP-Placebo of 5% Actual difference -0.15%, P<0.0001, z=-4.30 90%CI (-1.99, 1.68) 95%CI (-2.33, 2.04) 99%CI (-3.03, 2.72)
The formal test of non-inferiority, of the null hypothesis that the difference in proportion
unprotected between the intervention groups is 5% or more, gives strong evidence
suggesting SP is not inferior to placebo for both ITT and ATP.
FIGURE 17 Reverse cumulative distribution function for measles for the ITT cohort excluding children with detectable antibody levels at pre-vaccination
0.2
.4.6
.81
1 2 3 4 5log(10) measles antibody concentration post-vaccination
Placebo SP
(vertical line indicates assumed protective threshold)Reverse empirical cumulative distribution function
Final pooled analysis V.3 Navrongo, Manhica, Kisumu, Kilimanjaro 03/07/2009
35
FIGURE 18
Forest plot of the SP versus Placebo comparison for measles across sites
-0.40%
0.83%
-0.47%
-0.15%
-6.00% -4.00% -2.00% 0.00% 2.00% 4.00% 6.00%SP-Placebo difference
Navrongo Manhica Kilimanjaro POOLED
Final pooled analysis V.3 Navrongo, Manhica, Kisumu, Kilimanjaro 03/07/2009
36
9. Pooled LapDap Vs placebo comparison for measles
In this section of the report we will investigate the treatment effect of LapDap compared to
placebo using data from two sites; Kisumu and Kilimanjaro. Before we pool data from these
two sites we will investigate whether the effect of LapDap Vs placebo is modified by site. If
effect modification is found pooling of the data is not appropriate and will not be done.
9.1 Investigation of effect modification by site
TABLE 31 LapDap Vs Placebo group comparison for measles by site and investigation of effect modification
GMC post-vaccination ratios $ OR Unprotected £
MEASLES Excluding detectable
pre-vaccination (n=1,180)
Excluding protected
pre-vaccination (n=1,283)
Excluding detectable
pre-vaccination (n=1,180)
Excluding protected
pre-vaccination (n=1,283)
Treatment Placebo LapDap
1
1.05 (0.93-1.19)
1
1.02 (0.90-1.15)
Kisumu Placebo LapDap
1
0.32 (0.10-1.01)
1
0.29 (0.09-0.88) Site
Kisumu Kilimanjaro
1
0.93 (0.82-1.06)
1
0.92 (0.81-1.04)
Kilimanjaro Placebo LapDap
1
1.26 (0.69-2.31)
1
1.46 (0.84-2.55) Interaction
Treatment X Site P=0.18* P=0.19* P=0.03* P=0.006* $ Geometric mean concentrations ratios of post measles vaccination are compared. Comparisons are done with ratio GMC values drawn from regression models with site and treatment group as the only factors £ Odds ratios of unprotected (<120 IU/L) children post measles vaccination are compared with site, treatment group and their interaction as the only factors; ITT population * Likelihood ratio test comparing the model with to the model without the interaction terms of site & treatment group
We have not found evidence for effect modification for the GMC outcome. Therefore, we will
pool data across the two sites to study the overall LapDap treatment effect. However,
evidence of effect modification has been found for the binary outcome of unprotected
children. We will not pool data for this outcome, and report site specific LapDap effects,
found in previously submitted analyses for Kisumu and Kilimanjaro sites separately.
9.2 Measles
In total 1,328 children are included in this analysis; 544 in Kisumu, and 784 in Kilimanjaro.
Measles vaccination is given at 9 months of age in all three sites. Children were bled at 9
months before measles vaccination was administered to establish baseline measles serology
levels. Vaccine efficacy was measured by a subsequent blood sample drawn at 10 months.
Measles serology data with matched pre and post vaccination measurements was available
for 1,328 children. Randomisation resulted in 681 in Placebo and 647 in LapDap groups.
Final pooled analysis V.3 Navrongo, Manhica, Kisumu, Kilimanjaro 03/07/2009
37
TABLE 32 Numbers of children by study and treatment group for measles
Placebo LapDap TOTAL Kisumu 284 260 544 Kilimanjaro 397 387 784 TOTAL 681 647 1,328
9.2.1 Analysis of post-vaccination measles antibody concentrations
Characteristics of the 1,328 children, sampled at 9 months, included in the ITT analysis are as
follows:
TABLE 33 Baseline characteristics of children included in the ITT analysis Placebo
N=681 LapDap N=647
Age in months at vaccination (mean, range) 9 (8, 18) 9 (8, 16)
%male : %female 49%:51% 51%:49%
148 children had detectable concentrations in their pre-vaccination sample, 45 of these had
protective (≥ 120IU/L) concentration of measles antibodies in their pre-vaccination sample.
The next table summarises the geometric mean concentrations (GMC) in the two treatment
groups before and after vaccination, and the median post-vaccination concentration.
TABLE 34 Antibody concentration response to the measles vaccination by treatment group GMC
pre-vaccination (95%CI)
GMC post-vaccination (95%CI) *
Median post-vaccination
Children with detectable concentrations pre-vaccination Placebo (n=82) 67 (51,87) 726 (558, 944) 870 LapDap (n=66) 99 (69,142) 604 (429, 849) 618
Children with undetectable concentrations pre-vaccination Placebo (n=599) 617 (564, 675) 673 LapDap (n=581) 647 (593, 706) 679
Ratio LapDap/Placebo 1.05 (0.93, 1.19) P-value1 0.44 Children with protective measles antibody level pre-vaccination
Placebo (n=20) 423 (272, 658) 1216 (744, 1989) 1556 LapDap (n=25) 424 (237, 760) 1263 (666, 2400) 1321
Children without protective measles antibody level pre-vaccination Placebo (n=661) 617 (566, 672) 674 LapDap (n=622) 625 (574, 681) 670
Ratio LapDap/Placebo 1.02 (0.90, 1.15) P-value1 0.81 All children
Placebo (n=681) 629 (578, 685) 683 LapDap (n=647) 642 (590, 700) 677
Ratio LapDap/Placebo 1.02 (0.91, 1.15) P-value1 0.72 *26 children had non-detectable concentrations post-vaccination. These undetectable levels were assigned a value of 11IU/L, half of the minimum concentration detected in the pooled dataset, for purposes of the analysis in this table (this affects the GMC but not the median). 1 P-value is calculated from the likelihood ratio test comparing the regression model with treatment group and site as the only factors to its nested model with only site. H0= ratio equal to 1
Final pooled analysis V.3 Navrongo, Manhica, Kisumu, Kilimanjaro 03/07/2009
38
No evidence has been found to support a difference in the GMC between LapDap and placebo
groups.
FIGURE 19 Reverse cumulative distribution function for measles for all children
0.2
.4.6
.81
1 2 3 4log(10) measles antibody concentration post-vaccination
Placebo Lapdap
(vertical line indicates assumed protective threshold)Reverse empirical cumulative distribution function
FIGURE 20 Reverse cumulative distribution function for measles for the ITT cohort excluding those protected at pre-vaccination
0.2
.4.6
.81
1 2 3 4log(10) measles antibody concentration post-vaccination
Placebo Lapdap
(vertical line indicates assumed protective threshold)Reverse empirical cumulative distribution function
Final pooled analysis V.3 Navrongo, Manhica, Kisumu, Kilimanjaro 03/07/2009
39
FIGURE 21 Reverse cumulative distribution function for measles for the ITT cohort excluding children with detectable antibody levels at pre-vaccination
0.2
.4.6
.81
1 2 3 4log(10) measles antibody concentration post-vaccination
Placebo Lapdap
(vertical line indicates assumed protective threshold)Reverse empirical cumulative distribution function
Final pooled analysis V.3 Navrongo, Manhica, Kisumu, Kilimanjaro 03/07/2009
40
10. Pooled combined treatment Vs placebo comparison for Kisumu site
In this section of the report we will investigate, and present where appropriate, pooling the
three treatment groups (SP-ART, AQ-ART and LapDap) into one combined and compare
against placebo.
10.1 Investigation of combined Vs individual treatment group comparison with placebo
TABLE 35 Combined treatment effect in Kisumu site; measles
GMC post-vaccination ratios $ OR Unprotected £
MEASLES Excluding detectable
pre-vaccination (n=986)
Excluding protected
pre-vaccination (n=1,087)
Excluding detectable
pre-vaccination (n=986)
Excluding protected
pre-vaccination (n=1,087)
Treatment Placebo SP-ART AQ-ART LapDap
1
1.11 (0.94-1.30) Single combined treatment group
1
1.06 (0.91-1.24) Single combined treatment group
1
1.21 (0.57-2.56) 1.13 (0.53-2.43) 0.32 (0.10-1.01)
1
1.21 (0.60-2.45) 1.18 (0.58-2.39) 0.29 (0.09-0.88)
LRT * P=0.69 P=0.67 P=0.02 P=0.006 $ Geometric mean concentrations ratios of post measles vaccination are compared. Comparisons are done with ratio GMC values drawn from regression models with treatment group as the only factor £ Odds ratios of unprotected (<120 IU/L) children post measles vaccination are compared with treatment group as the only factor; ITT population * Likelihood ratio test between: i) the model comparing each treatment group to placebo and ii) the model comparing a single combined treatment group to placebo
We have found evidence against combining treatment groups for the outcome of unprotected
children and will not be presenting pooled estimates. However, for the GMC outcome
evidence supports the combination of the three treatment groups into one and its comparison
to placebo.
Final pooled analysis V.3 Navrongo, Manhica, Kisumu, Kilimanjaro 03/07/2009
41
TABLE 36 Combined treatment effect in Kisumu site; other antigens
OTHER ANTIGENS GMC post-vaccination
ratios $ OR Unprotected £
Treatment Placebo SP-ART AQ-ART LapDap
1
0.91 (0.63-1.31) Single combined treatment group Po
lio t
ype
1
LRT* P=0.80 Treatment
Placebo SP-ART AQ-ART LapDap
1
1.03 (0.76-1.38) Single combined treatment group Po
lio t
ype
3
LRT* P=0.15 Treatment
Placebo SP-ART AQ-ART LapDap
1 1.04 (0.89-1.22) Single combined treatment group
1
1.26 (0.62-2.58) Single combined treatment group D
ipht
heria
LRT * P=0.87 P=0.86 Treatment
Placebo SP-ART AQ-ART LapDap
1 0.92 (0.79-1.07) Single combined treatment group Pe
rtus
sis
FHA
LRT * P=0.17 Treatment
Placebo SP-ART AQ-ART LapDap
1 1.11 (0.81-1.50) Single combined treatment group
Pert
ussi
s to
xin
LRT * P=0.16 Treatment
Placebo SP-ART AQ-ART LapDap
1 1.08 (0.89-1.30) Single combined treatment group
There is only 1 unprotected child in the placebo group
Teta
nus
LRT * P=0.82 Treatment
Placebo SP-ART AQ-ART LapDap
1
1.05 (0.55-2.00) 2.35 (1.27-4.38) 0.84 (0.45-1.56)
There are no unprotected children
Hep
atiti
s B
LRT * P=0.004 $ Geometric mean concentrations ratios of post measles vaccination are compared. Comparisons are done with ratio GMC values drawn from regression models with treatment group as the only factor £ Odds ratios of unprotected (<120 IU/L) children post measles vaccination are compared with treatment group as the only factor; ITT population * Likelihood ratio test between: i) the model comparing each treatment group to placebo and ii) the model comparing a single combined treatment group to placebo We have found evidence supporting the pooled combined treatment group use for diphtheria,
pertussis FHA & toxin and tetanus, but not for hepatitis B, for the GMC outcome. Data also
support the use of the single combined group for polio type 1 & 3 and diphtheria for the
binary outcome of unprotected children.
Final pooled analysis V.3 Navrongo, Manhica, Kisumu, Kilimanjaro 03/07/2009
42
10.2 Measles
Measles serology data with matched pre and post vaccination measurements was available
for 1,122 children. Randomization resulted in 284 children in placebo, 285 in SP-ART, 293 in
AQ-ART and 260 in LapDap treatment groups.
10.2.1 Analysis of post-vaccination measles antibody concentrations
136 children had detectable concentrations in their pre-vaccination sample, 35 of these had
protective (≥ 120IU/L) concentration of measles antibodies in their pre-vaccination sample.
The next table summarises the geometric mean concentrations (GMC) in the two treatment
groups before and after vaccination, and the median post-vaccination concentration.
TABLE 37 Antibody concentration response to the measles vaccination by treatment group GMC
pre-vaccination (95%CI)
GMC post-vaccination (95%CI) *
Median post-vaccination
Children with detectable concentrations pre-vaccination Placebo (n=36) 56 (37,86) 755 (500,1141) 945
Combined (n=100) 69 (51,92) 609 (475, 782) 648 Children with undetectable concentrations pre-vaccination
Placebo (n=248) 611 (529, 707) 633 Combined (n=738) 677 (625, 733) 680
Ratio Combined/Placebo 1.11 (0.94, 1.30) P-value 1 0.22 Children with protective measles antibody level pre-vaccination
Placebo (n=7) 536 (302, 952) 1130 (590, 2163) 1523 Combined (n=28) 464 (264, 816) 1109 (597, 2059) 1189
Children without protective measles antibody level pre-vaccination Placebo (n=277) 619 (538, 711) 637
Combined (n=810) 657 (609, 709) 665 Ratio Combined/Placebo 1.06 (0.91, 1.24) P-value 1 0.44 All children
Placebo (n=284) 628 (548, 720) 650 Combined (n=838) 668 (619, 721) 676
Ratio Combined/Placebo 1.06 (0.91, 1.24) P-value 1 0.42 *19 children had non-detectable concentrations post-vaccination. These undetectable levels were assigned a value of 11IU/L, half of the minimum concentration detected in the pooled dataset, for purposes of the analysis in this table (this affects the GMC but not the median). 1 P-value is calculated from the likelihood ratio test comparing the regression model with treatment group as the only factor to its nested null model. H0= ratio equal to 1
No evidence has been found to support a difference in the measles GMC between the
combined treatment group when compared to placebo.
Final pooled analysis V.3 Navrongo, Manhica, Kisumu, Kilimanjaro 03/07/2009
43
FIGURE 22 Reverse cumulative distribution function for measles for all children
0.2
.4.6
.81
1 2 3 4log(10) measles antibody concentration post-vaccination
Placebo Combined
(vertical line indicates assumed protective threshold)Reverse empirical cumulative distribution function
FIGURE 23 Reverse cumulative distribution function for measles for the ITT cohort excluding those protected at pre-vaccination
0.2
.4.6
.81
1 2 3 4log(10) measles antibody concentration post-vaccination
Placebo Combined
(vertical line indicates assumed protective threshold)Reverse empirical cumulative distribution function
Final pooled analysis V.3 Navrongo, Manhica, Kisumu, Kilimanjaro 03/07/2009
44
FIGURE 24 Reverse cumulative distribution function for measles for the ITT cohort excluding children with detectable antibody levels at pre-vaccination
0.2
.4.6
.81
1 2 3 4log(10) measles antibody concentration post-vaccination
Placebo Combined
(vertical line indicates assumed protective threshold)Reverse empirical cumulative distribution function
Final pooled analysis V.3 Navrongo, Manhica, Kisumu, Kilimanjaro 03/07/2009
45
10.3 Diphtheria, Tetanus, Polio type 1 & 3, Pertussis toxin and FHA, and Hepatitis B responses
10.3.1 ITT analysis; post-vaccination antibody concentrations & proportion protected TABLE 38 ITT analysis for diphtheria, tetanus, polio type 1 & 3, pertussis toxin & FHA and hepatitis B; post-vaccination antibody concentrations & proportion unprotected Polio Type 1
(n=968) Polio Type 3 (n=968)
Diphtheria (n=1,006)
Pertussis FHA (n=893)
Pertussis toxin (n=623)
Tetanus (n=700)
Protective threshold 1 in 8 PRN 1 in 8 PRN 0.1IU/ml Not defined (US units)
Not defined (US units)
Percent unprotected (below threshold)
Placebo 49/249 19.68%
95/249 38.15%
10/259 3.86%
Combined 131/719 18.22%
279/719 38.80%
36/747 4.82%
Difference Combined-Placebo -1.46% 0.65% 0.96% Null hypothesis difference Combined-Placebo
10% 10% 10%
Test of non-inferiority Z=-4.41, P<0.0001+
Z=-2.70, P=0.004+
Z=-6.62 P<0.001+
90%CI (-6.23, 3.31) (-5.23, 6.53) (-1.40, 3.31) 95%CI (-7.15, 4.23) (-6.36, 7.66) (-1.85, 3.76) 99%CI (-8.93, 6.02) (-8.56, 9.86) (-2.73, 4.64)
GMC (95%CI) Placebo 0.75
(0.66,0.85) n=259
14.7 (12.9,16.7) n=235
67 (50,89) n=161
2.97 (2.52,3.52) n=179
Combined 0.78 (0.72,0.85) n=747
13.5 (12.5,14.6) n=658
74 (63,86) n=462
3.20 (2.92,3.51) n=521
Ratio Combined/Placebo α (95% CI) P-value*
1.04 (0.89,1.22) P=0.59
0.92 (0.79,1.07) P=0.28
1.11 (0.81,1.50) P=0.52
1.08 (0.89,1.30) P=0.44
+ 1.0: 210 =−Η pp , where 21, pp are proportions unprotected (below threshold) in the Combined Vs Placebo groups respectively; α All ratio values are estimated from regression models
with treatment group as the only factor; * This is the Wald test P-value from the regression model with treatment group as the only factor. 1: 10 =Η r , where 1r is the ratio of ln GMCs for
combined treatment group compared to placebo.
Final pooled analysis V.3 Navrongo, Manhica, Kisumu, Kilimanjaro 03/07/2009
46
10.3.2 ATP analysis; post-vaccination concentrations & proportion protected
TABLE 39 ATP analysis for diphtheria, tetanus, polio type 1 & 3, pertussis toxin & FHA and hepatitis B; post-vaccination antibody concentrations & proportion unprotected Polio Type 1
(n=880) Polio Type 3 (n=880)
Diphtheria (n=913)
Pertussis FHA (n=813)
Pertussis toxin (n=563)
Tetanus (n=636)
Protective threshold 1 in 8 PRN 1 in 8 PRN 0.1IU/ml Not defined (US units)
Not defined (US units)
Percent unprotected (below threshold)
Placebo 41/222 18.47%
79/222 35.59%
8/230 3.48%
Combined 117/658 17.78%
254/658 38.60%
32/683 4.69%
Difference Combined-Placebo -0.69% 3.01% 1.21% Null hypothesis difference Combined-Placebo
10% 10% 10%
Test of non-inferiority Z=-3.98, P<0.0001+
Z=-1.92, P=0.03+
Z=-6.27 P<0.0001+
90%CI (-5.62, 4.25) (-3.12, 9.15) (-1.18, 3.60) 95%CI (-6.57, 5.19) (-4.30,10.33) (-1.64, 4.06) 99%CI (-8.42, 7.04) (-6.60,12.63) (-2.54, 4.95)
GMC (95%CI) Placebo 0.76
(0.66,0.87) n=230
14.9 (13.0,17.0) n=210
64 (48,86) n=140
2.88 (2.44,3.41) n=161
Combined 0.78 (0.72,0.85) n=683
13.5 (12.4,14.6) n=603
73 (62,86) n=423
3.16 (2.87,3.48) n=475
Ratio Combined/Placebo α (95% CI) P-value*
1.03 (0.87,1.22) P=0.73
0.91 (0.77,1.06) P=0.23
1.14 (0.83,1.58) P=0.42
1.10 (0.90,1.33) P=0.35
+ 1.0: 210 =−Η pp , where 21, pp are proportions unprotected (below threshold) in the Combined Vs Placebo groups respectively; α All ratio values are estimated from regression models
with treatment group as the only factor; * This is the Wald test P-value from the regression model with treatment group as the only factor. 1: 10 =Η r , where 1r is the ratio of ln GMCs for
combined treatment group compared to placebo
Final pooled analysis V.3 Navrongo, Manhica, Kisumu, Kilimanjaro 03/07/2009
47
There was no evidence found to support the combined group as inferior by 10% or more
compared to placebo. This result was consistent across antigens and for both the ITT and
ATP. For diphtheria, pertussis toxin and FHA and tetanus children in the combined treatment
group had on average similar geometric mean concentration after vaccination when
compared to placebo children.
FIGURE 25 Reverse cumulative distribution function for diphtheria
0.2
.4.6
.81
-2 -1 0 1 2log(10) diphtheria antibody concentration IU/ml
Placebo Combined
(vertical line indicates assumed protective threshold)Reverse empirical cumulative distribution function
Final pooled analysis V.3 Navrongo, Manhica, Kisumu, Kilimanjaro 03/07/2009
48
FIGURE 26 Reverse cumulative distribution function for pertussis FHA
0.2
.4.6
.81
0 1 2 3log(10) pertussis filamentous agglutinin (FHA) concentration US units
Placebo Combined
Reverse empirical cumulative distribution function
FIGURE 27 Reverse cumulative distribution function for pertussis toxin
0.2
.4.6
.81
-1 0 1 2 3 4log(10) pertussis toxin concentration US units
Placebo Combined
Reverse empirical cumulative distribution function
Final pooled analysis V.3 Navrongo, Manhica, Kisumu, Kilimanjaro 03/07/2009
49
FIGURE 28 Reverse cumulative distribution function for tetanus
0.2
.4.6
.81
-2 -1 0 1 2log(10) tetanus antibody concentration IU/ml
Placebo Combined
(vertical line indicates assumed protective threshold)Reverse empirical cumulative distribution function
Final pooled analysis V.3 Navrongo, Manhica, Kisumu, Kilimanjaro 03/07/2009
50
11. Pooled combined treatment Vs placebo comparison for Kilimanjaro site
In this section of the report we will investigate, and present where appropriate, pooling the
three treatment groups (SP, MQ and LapDap) into one combined group and compare against
placebo.
11.1 Investigation of combined Vs individual treatment group comparison with placebo
TABLE 40 Combined treatment effect in Kilimanjaro site for measles
GMC post-vaccination ratios $ OR Unprotected £
MEASLES Excluding detectable
pre-vaccination (n=1,377)
Excluding protected
pre-vaccination (n=1,493)
Excluding detectable
pre-vaccination (n=1,377)
Excluding protected
pre-vaccination (n=1,493)
Treatment Placebo
SP MQ
LapDap
1
1.01 (0.89-1.16) Single combined treatment group
1
0.98 (0.86-1.12) Single combined treatment group
1
1.05 (0.62-1.76) Single combined treatment group
1
1.20 (0.75-1.94) Single combined treatment group
LRT* P=0.24 P=0.28 P=0.54 P=0.35 $ Geometric mean concentrations ratios of post measles vaccination are compared. Comparisons are done with ratio GMC values drawn from regression models with treatment group as the only factor £ Odds ratios of unprotected (<120 IU/L) children post measles vaccination are compared with treatment group as the only factor; ITT population * Likelihood ratio test between: i) the model comparing each treatment group to placebo and ii) the model comparing a single combined treatment group to placebo
We have found evidence supporting pooling treatment groups for both GMC and binary
outcome of unprotected children. Therefore, we will be combining the three treatment groups
into one and comparing to placebo.
11.2 Measles
Measles serology data with matched pre and post vaccination measurements was available
for 1,538 children. Randomization resulted in 397 children in placebo, 374 in SP, 380 in MQ
and 387 in LapDap treatment groups. For the purposes of section 11 we will refer to
‘combined’ as the single pooled SP, MQ and LapDap group.
11.2.1 Analysis of post-vaccination measles antibody concentrations
161 children had detectable concentrations in their pre-vaccination sample, 45 of these had
protective (≥ 120IU/L) concentration of measles antibodies in their pre-vaccination sample.
The next table summarises the geometric mean concentrations (GMC) in the two treatment
groups before and after vaccination, and the median post-vaccination concentration.
Final pooled analysis V.3 Navrongo, Manhica, Kisumu, Kilimanjaro 03/07/2009
51
TABLE 41 Antibody concentration response to the measles vaccination by treatment group GMC
pre-vaccination (95%CI)
GMC post-vaccination (95%CI) *
Median post-vaccination
Children with detectable concentrations pre-vaccination Placebo (n=46) 76 (53,108) 704 (494,1003) 761
Combined (n=115) 76 (64,91) 451 (352, 577) 521 Children with undetectable concentrations pre-vaccination
Placebo (n=351) 621 (555, 696) 700 Combined (n=1026) 631 (590, 674) 679
Ratio Combined/Placebo 1.01 (0.89, 1.16) P-value1 0.83 Children with protective measles antibody level pre-vaccination
Placebo (n=13) 373 (194, 716) 1265 (604, 2651) 1588 Combined (n=32) 270 (204, 356) 855 (570, 1282) 623
Children without protective measles antibody level pre-vaccination Placebo (n=384) 616 (552, 689) 697
Combined (n=1109) 604 (565, 645) 665 Ratio Combined/Placebo 0.98 (0.86, 1.12) P-value1 0.76 All children
Placebo (n=397) 630 (566, 702) 701 Combined (n=1141) 610 (571, 651) 665
Ratio Combined/Placebo 0.97 (0.85, 1.10) P-value1 0.60 *19 children had non-detectable concentrations post-vaccination. These undetectable levels were assigned a value of 11IU/L, half of the minimum concentration detected in the pooled dataset, for purposes of the analysis in this table (this affects the GMC but not the median) 1 P-value is calculated from the likelihood ratio test comparing the regression model with treatment group as the only factor to the null model. H0= ratio equal to 1
No evidence has been found to support a difference in the GMC between the single combined
treatment group when compared to placebo.
FIGURE 29 Reverse cumulative distribution function for measles for all children
0.2
.4.6
.81
1 2 3 4log(10) measles antibody concentration post-vaccination
Placebo Combined
(vertical line indicates assumed protective threshold)Reverse empirical cumulative distribution function
Final pooled analysis V.3 Navrongo, Manhica, Kisumu, Kilimanjaro 03/07/2009
52
11.2.2 Proportion of children with protective concentrations post-vaccination
11.2.2.1 Excluding those with protective measles concentrations in their pre-vaccination sample
45 children had protective concentrations of measles antibody in their pre-vaccination
sample. These children have been excluded from this point on, leaving 1,493 individuals.
TABLE 42 ITT proportion of children protected from measles; excluding those protected at pre-vaccination Protected Unprotected Total Percent unprotected Placebo 361 23 384 5.99% Combined 1,030 79 1,109 7.12% Null hypothesis difference Combined-Placebo of 5% Actual difference 1.13%, P=0.001, z=-2.99 90%CI (-1.23, 3.50) 95%CI (-1.68, 3.94) 99%CI (-2.57, 4.83)
A further 71 children are excluded from the ATP analysis, leaving 1,422 individuals.
TABLE 43 ATP proportion of children protected from measles; excluding those protected at pre-vaccination Protected Unprotected Total Percent unprotected Placebo 355 22 377 5.84% Combined 972 73 1,045 6.99% Null hypothesis difference Combined-Placebo of 5% Actual difference 1.15%, P=0.002, z=-2.95 90%CI (-1.22, 3.52) 95%CI (-1.68, 3.98) 99%CI (-2.56, 4.86)
The formal test of non-inferiority, of the null hypothesis that the difference in proportion
unprotected between the intervention groups is 5% or more, gives strong evidence
suggesting combined group is not inferior to placebo for both ITT and ATP.
Final pooled analysis V.3 Navrongo, Manhica, Kisumu, Kilimanjaro 03/07/2009
53
FIGURE 30
Reverse cumulative distribution function for measles for the ITT cohort excluding those protected at pre-vaccination
0.2
.4.6
.81
1 2 3 4log(10) measles antibody concentration post-vaccination
Placebo Combined
(vertical line indicates assumed protective threshold)Reverse empirical cumulative distribution function
11.2.2.2 Excluding those with detectable measles concentrations in their pre-vaccination sample
161 children had protective concentrations of measles antibody in their pre-vaccination
sample. These children have been excluded from this point on, leaving 1,377 individuals.
TABLE 44 ITT proportion of children protected from measles; excluding those with detectable concentration at pre-vaccination Protected Unprotected Total Percent unprotected Placebo 331 20 351 5.70% Combined 965 61 1,026 5.95% Null hypothesis difference Combined-Placebo of 5% Actual difference 0.25%, P=0.0001, z=-3.77 90%CI (-2.12, 2.62) 95%CI (-2.58, 3.07) 99%CI (-3.46, 3.96)
A further 62 children are excluded from the ATP analysis, leaving 1,315 individuals.
TABLE 45 ATP proportion of children protected from measles; excluding those with detectable concentration at pre-vaccination Protected Unprotected Total Percent unprotected Placebo 326 20 346 5.78% Combined 911 58 969 5.99% Null hypothesis difference Combined-Placebo of 5% Actual difference 0.21%, P=0.0001, z=-3.70 90%CI (-2.21, 2.62) 95%CI (-2.67, 3.08) 99%CI (-3.58, 3.99)
Final pooled analysis V.3 Navrongo, Manhica, Kisumu, Kilimanjaro 03/07/2009
54
The formal test of non-inferiority, of the null hypothesis that the difference in proportion
unprotected between the combined treatment group is 5% or more, gives strong evidence
suggesting non-inferiority compared to placebo for both ITT and ATP.
FIGURE 31 Reverse cumulative distribution function for measles for the ITT cohort excluding children with detectable antibody levels at pre-vaccination
0.2
.4.6
.81
1 2 3 4log(10) measles antibody concentration post-vaccination
Placebo Combined
(vertical line indicates assumed protective threshold)Reverse empirical cumulative distribution function
Final pooled analysis V.3 Navrongo, Manhica, Kisumu, Kilimanjaro 03/07/2009
55
12. Appendices
12.1 Summary to the IPTi Consortium Trials
IPTi studies that include assessment of serological responses to EPI vaccines Study site
Study design
Drug(s) for IPTi
Age at IPTi drug dosing
Sample size
EPI immunisation schedule
Timing of blood samples
Serological information available from each study
Navrongo, Ghana
RCT
SP Placebo
10 weeks (DTP2) 14 weeks (DTP3) 9 months (measles) 12 months
500 SP 500 placebo
BCG: birth DTP, hepB, Hib: 6, 10, 14 weeks Polio: birth, 6, 10, 14 weeks Measles, yellow fever: 9 months
9 months 12 months
Measles, yellow fever
Bungoma, Kenya
RCT
SP Placebo
10 weeks (DTP2) 14 weeks (DTP3) 9 months (measles)
500 SP 500 placebo
BCG: birth DTP, hepB, Hib: 6, 10, 14 weeks Polio: birth, 6, 10, 14 weeks Measles: 9 months
6 weeks 18 weeks 9 months 10 months
DTP, polio, hepatitis B, Hib, measles
Manhica, Mozambique
RCT
SP Placebo
12 weeks (DTP2) 16 weeks (DTP3) 9 months (measles)
500 SP 500 placebo
BCG: birth DTP, hepB: 8, 12, 16 weeks Polio: birth, 8, 12, 16 weeks Measles: 9 months
20 weeks 9 months 12 months
DTP, polio, hepatitis B measles
Kisumu, Kenya
RCT SP/Art
AQ/Art LapDap Placebo
10 weeks (DTP2) 14 weeks (DTP3) 9 months (measles)
379 x 3 intervention 379 placebo
As for Bungoma
6 weeks 18 weeks 9 months 12 months
DTP, polio, hepatitis B, Hib, measles
Kilimanjaro, Tanzania
RCT SP
MQ LapDap Placebo
8 weeks (DTP2) 12 weeks (DTP3) 9 months (measles)
500 x 3 intervention 500 placebo
BCG: birth DTP, hepB: 4, 8, 12 weeks Polio: birth, 4, 8, 12 weeks Measles: 9 months
9 months 10 months
Measles
Legend: RCT: randomised controlled trial; SP: sulfadoxine-pyrimethamine; SP/Art: sulfadoxine-pyrimethamine plus artesunate; MQ: mefloquine; LapDap: chlorproguanil-dapsone; AQ/Art: amodiaquine plus artesunate; DTP: diphtheria, tetanus, pertussis; hepB: hepatitis B; Hib: Haemophilus influenzae type b
Final pooled analysis V.3 Navrongo, Manhica, Kisumu, Kilimanjaro 03/07/2009
56
Serological assays Vaccine Pre or post-
vaccination sample
Timing of blood samples
Type of test Protective level
Measles
Pre
9 months Plaque reduction neutralisation (PRN) 1-2 dilutions
To check for presence of measles antibodies pre-vaccination
Post 10 months or 12 months
PRN 6 dilutions (GMT) 120 IU/l
Yellow fever Pre 9 months PRN 1-2 dilutions To check for presence of YF antibodies pre-vaccination
Post 10 months or 12 months
PRN 6 dilutions (GMT) 1:5 (PRN titre)
Diphtheria Pre At time of DTP1 Store in freezer*
Post One month post DTP3
Quantitative ELISA (GMT)
0.1 IU/ml
Tetanus Pre At time of DTP1 Store in freezer*
Post One month post DTP3
Quantitative ELISA (GMT)
0.1 IU/ml
Pertussis (PT and FHA only)
Pre At time of DTP1 Store in freezer*
Post One month post DTP3
Quantitative ELISA (GMT)
Protective levels not defined
Polio (serotypes 1 and 3 only)
Pre At time of DTP1 Store in freezer*
Post One month post DTP3
PRN 6 dilutions (GMT) 1:8 (PRN titre)
Haemophilus influenzae type b
Pre At time of DTP1 Store in freezer*
Post One month post DTP3
Quantitative ELISA (GMT)
0.15 µg/ml (1.0 µg/ml will be used as a secondary descriptive)
Hepatitis B Pre At time of DTP1 Store in freezer*
Post One month post DTP3
Quantitative ELISA (GMT)
10 IU/l
* Pre-vaccination GMTs will only be obtained if the post-vaccination results are equivocal; PRN: Plaque reduction neutralisation; GMT: Geometric mean titre; PT: Pertussis toxin; FHA: Filamentous haemagglutinin
Final pooled analysis V.3 Navrongo, Manhica, Kisumu, Kilimanjaro 03/07/2009
57
12.2 Post-hoc analysis I; pooled combined treatment group excluding LapDap Vs placebo
for Kisumu site
In this section of the report we will investigate, and present where appropriate, pooling the
two treatment groups (SP-ART and AQ-ART) into one combined and compare against placebo
for Kisumu site. We are excluding LapDap group from all analyses in this section.
12.2.1 Investigation of combined Vs individual treatment group comparison with placebo
TABLE 46 Combined treatment effect in Kisumu site; for measles antigen
GMC post-vaccination ratios $ OR Unprotected £
MEASLES Excluding detectable
pre-vaccination (n=759)
Excluding protected
pre-vaccination (n=840)
Excluding detectable
pre-vaccination (n=759)
Excluding protected
pre-vaccination (n=840)
Treatment Placebo SP-ART AQ-ART
1
1.08 (0.91-1.29) Combined group
1
1.04 (0.88-1.23) Combined group
1
1.17 (0.60-2.27) Combined group
1
1.19 (0.64-2.22) Combined group
LRT * P=0.66 P=0.67 P=0.89 P=0.94 $ Geometric mean concentrations ratios of post measles vaccination are compared. Comparisons are done with ratio GMC values drawn from regression models with treatment group as the only factor £ Odds ratios of unprotected (<120 IU/L) children post measles vaccination are compared with treatment group as the only factor; ITT population * Likelihood ratio test between: i) the model comparing each treatment group to placebo and ii) the model comparing a single combined treatment group to placebo
For both the binary outcome of unprotected children and the GMC outcome evidence
supports the combination of the two treatment groups into one and its comparison to
placebo.
12.2.2 Measles
Measles serology data with matched pre and post vaccination measurements was available
for 862 children. Randomization resulted in 284 children in placebo, 285 in SP-ART and 293 in
AQ-ART treatment groups.
Final pooled analysis V.3 Navrongo, Manhica, Kisumu, Kilimanjaro 03/07/2009
58
12.2.2.1 Analysis of post-vaccination measles antibody concentrations
103 children had detectable concentrations in their pre-vaccination sample, 22 of these had
protective (≥ 120IU/L) concentration of measles antibodies in their pre-vaccination sample.
The next table summarises the geometric mean concentrations (GMC) in the two treatment
groups before and after vaccination, and the median post-vaccination concentration.
TABLE 47 Antibody concentration response to the measles vaccination by treatment group GMC
pre-vaccination (95%CI)
GMC post-vaccination (95%CI) *
Median post-vaccination
Children with detectable concentrations pre-vaccination Placebo (n=36) 56 (37,86) 755 (500,1141) 945
Combined (n=67) 55 (40,75) 546 (414, 721) 566 Children with undetectable concentrations pre-vaccination
Placebo (n=248) 611 (529, 707) 633 Combined (n=511) 662 (598, 733) 675
Ratio Combined/Placebo 1.08 (0.91, 1.29) P-value 1 0.38 Children with protective measles antibody level pre-vaccination
Placebo (n=7) 536 (302, 952) 1130 (590, 2163) 1523 Combined (n=15) 395 (199, 786) 880 (463, 1673) 804
Children without protective measles antibody level pre-vaccination Placebo (n=277) 619 (538, 711) 637
Combined (n=563) 642 (583, 707) 653 Ratio Combined/Placebo 1.04 (0.88, 1.23) P-value 1 0.67 All children
Placebo (n=284) 628 (548, 720) 650 Combined (n=578) 647 (588, 712) 667
Ratio Combined/Placebo 1.03 (0.87, 1.22) P-value 1 0.72 *18 children had non-detectable concentrations post-vaccination. These undetectable levels were assigned a value of 11IU/L, half of the minimum concentration detected in the pooled dataset, for purposes of the analysis in this table (this affects the GMC but not the median). 1 P-value is calculated from the likelihood ratio test comparing the regression model with treatment group as the only factor to its nested null model. H0= ratio equal to 1
No evidence has been found to support a difference in the measles GMC between the single
combined treatment group and placebo.
Final pooled analysis V.3 Navrongo, Manhica, Kisumu, Kilimanjaro 03/07/2009
59
FIGURE 32 Reverse cumulative distribution function for measles for all children
0.2
.4.6
.81
1 2 3 4log(10) measles antibody concentration post-vaccination
Placebo Combined
(vertical line indicates assumed protective threshold)Reverse empirical cumulative distribution function
12.2.2.2 Proportion of children with protective concentrations post-vaccination
12.2.2.2.1 Excluding those with protective measles concentrations in their pre-vaccination sample
22 children had protective concentrations of measles antibody in their pre-vaccination
sample. These children have been excluded from this point on, leaving 840 individuals.
TABLE 48 ITT proportion of children protected from measles; excluding those protected at pre-vaccination Protected Unprotected Total Percent unprotected Placebo 262 15 277 5.42% Combined 527 36 563 6.39% Null hypothesis difference Combined-Placebo of 5% Actual difference 0.97%, P=0.006, z=-2.49 90%CI (-1.83, 3.79) 95%CI (-2.37, 4.32) 99%CI (-3.42, 5.37)
Final pooled analysis V.3 Navrongo, Manhica, Kisumu, Kilimanjaro 03/07/2009
60
A further 9 children are excluded from the ATP analysis, leaving 831 individuals.
TABLE 49 ATP proportion of children protected from measles; excluding those protected at pre-vaccination Protected Unprotected Total Percent unprotected Placebo 259 14 273 5.13% Combined 523 35 558 6.27% Null hypothesis difference Combined-Placebo of 5% Actual difference 1.14%, P=0.008, z=-2.42 90%CI (-1.63, 3.91) 95%CI (-2.16, 4.44) 99%CI (-3.19, 5.48)
The formal test of non-inferiority, of the null hypothesis that the difference in proportion
unprotected between the intervention groups is 5% or more, gives strong evidence
suggesting combined group is not inferior to placebo for both ITT and ATP.
FIGURE 33
Reverse cumulative distribution function for measles for the ITT cohort excluding those protected at pre-vaccination
0.2
.4.6
.81
1 2 3 4log(10) measles antibody concentration post-vaccination
Placebo Combined
(vertical line indicates assumed protective threshold)Reverse empirical cumulative distribution function
Final pooled analysis V.3 Navrongo, Manhica, Kisumu, Kilimanjaro 03/07/2009
61
12.2.2.2.2 Excluding those with detectable measles concentrations in their pre-vaccination sample
103 children had protective concentrations of measles antibody in their pre-vaccination
sample. These children have been excluded from this point on, leaving 759 individuals.
TABLE 50 ITT proportion of children protected from measles; excluding those with detectable concentration at pre-vaccination Protected Unprotected Total Percent unprotected Placebo 235 13 248 5.24% Combined 480 31 511 6.07% Null hypothesis difference Combined-Placebo of 5% Actual difference 0.83%, P=0.006, z=-2.50 90%CI (-2.08, 3.73) 95%CI (-2.64, 4.29) 99%CI (-3.72, 5.37)
A further 7 children are excluded from the ATP analysis, leaving 752 individuals.
TABLE 51 ATP proportion of children protected from measles; excluding those with detectable concentration at pre-vaccination Protected Unprotected Total Percent unprotected Placebo 234 12 246 4.88% Combined 476 30 506 5.93% Null hypothesis difference Combined-Placebo of 5% Actual difference 1.05%, P=0.008, z=-2.41 90%CI (-1.79, 3.89) 95%CI (-2.33, 4.44) 99%CI (-3.40, 5.50)
The formal test of non-inferiority, of the null hypothesis that the difference in proportion
unprotected between the combined treatment group is 5% or more, gives strong evidence
suggesting non-inferiority compared to placebo for both ITT and ATP.
Final pooled analysis V.3 Navrongo, Manhica, Kisumu, Kilimanjaro 03/07/2009
62
FIGURE 34 Reverse cumulative distribution function for measles for the ITT cohort excluding children with detectable antibody levels at pre-vaccination
0.2
.4.6
.81
1 2 3 4log(10) measles antibody concentration post-vaccination
Placebo Combined
(vertical line indicates assumed protective threshold)Reverse empirical cumulative distribution function
Final pooled analysis V.3 Navrongo, Manhica, Kisumu, Kilimanjaro 03/07/2009
63
12.3 Post-hoc analysis II; pooled combined treatment group excluding LapDap Kisumu Vs
placebo for all sites
In this section of the report we will investigate the treatment effect of the pooled combined
treatment group when compared to placebo using data from all four sites; Navrongo,
Manhica, Kisumu and Kilimanjaro. We have already established that pooling of the two
treatment groups, excluding LapDap, is appropriate for Kisumu site. We have also already
established that pooling of the three treatment groups is appropriate for Kilimanjaro site.
Before we further pool data from these four sites we will investigate whether the effect of
combined Vs placebo is modified by site. If heterogeneity is found pooling of the data is not
appropriate and will not be done.
12.3.1 Investigation of combined Vs individual treatment group comparison with placebo
TABLE 52 Combined Vs Placebo group comparison for measles by site and investigation of effect modification
GMC post-vaccination ratios $ OR Unprotected £
MEASLES Excluding detectable
pre-vaccination (n=3,103)
Excluding protected
pre-vaccination (n=3,413)
Excluding detectable
pre-vaccination (n=3,103)
Excluding protected
pre-vaccination (n=3,413)
Treatment Placebo
Combined
1
1.05 (0.96-1.15)
1
1.02 (0.93-1.11)
1
1.06 (0.76-1.48)
1
1.18 (0.87-1.62) Site
Navrongo Manhica Kisumu
Kilimanjaro
1
0.66 (0.57-0.77) 0.43 (0.37-0.49) 0.42 (0.37-047)
1
0.65 (0.56-0.75) 0.42 (0.37-0.48) 0.40 (0.35-0.45)
1
1.99 (1.08-3.68) 1.75 (0.97-3.14) 1.77 (1.02-3.07)
1
2.13 (1.17-3.91) 2.00 (1.13-3.56) 2.25 (1.31-3.86)
Interaction Treatment X Site P=0.95* P=0.71* P=0.97* P=0.91*
$ Geometric mean concentrations ratios of post measles vaccination are compared. Comparisons are done with ratio GMC values drawn from regression models with site and treatment group as the only factors £ Odds ratios of unprotected (<120 IU/L) children post measles vaccination are compared with site and treatment group as the only factors; ITT population * Likelihood ratio test comparing the model with to the model without the interaction terms between site & combined treatment We have found no evidence of heterogeneity by site. Therefore, we will pool data across the
four sites to study the overall combined treatment effect.
12.3.2 Measles
In total 3,644 children are included in this analysis; 620 in Navrongo, 624 in Manhica, 862 in
Kisumu and 1,538 in Kilimanjaro. Measles vaccination is given at 9 months of age in all four
sites. Children were bled at 9 months of age before measles vaccination was administered to
establish baseline measles serology levels. Vaccine responses were measured by a
subsequent blood sample drawn at 10 months (Kisumu, Kilimanjaro) and 12 months of age
(Manhica and Navrongo).
Final pooled analysis V.3 Navrongo, Manhica, Kisumu, Kilimanjaro 03/07/2009
64
Measles serology data with matched pre and post vaccination measurements was available
for 3,644 children. Randomisation resulted in 1,281 in Placebo and 2,363 in combined group.
All 3,644 children were included in the ITT analyses. From the ATP analyses children were
excluded for not having received all drug doses.
TABLE 53 Numbers of children by study and treatment group for measles
Placebo Combined TOTAL Navrongo 284 336 620 Manhica 316 308 624 Kisumu 284 578 862 Kilimanjaro 397 1,141 1,538 TOTAL 1,281 2,363 3,644
.
12.3.2.1 Analysis of post-vaccination measles antibody concentrations
Characteristics of the 3,644 children, sampled at 9 months, included in the ITT analysis are as
follows:
TABLE 54 Baseline characteristics of children included in the ITT analysis Placebo
N=1,281 Combined N=2,363
Age in months at vaccination (mean, range) 9 (8, 18) 9 (8, 19)
%male : %female 50%:50% 52%:48%
541 children had detectable concentrations in their pre-vaccination sample, 231 of these had
protective (≥ 120IU/L) concentration of measles antibodies in their pre-vaccination sample.
The next table summarises the geometric mean concentrations (GMC) in the two treatment
groups before and after vaccination, and the median post-vaccination concentration.
Final pooled analysis V.3 Navrongo, Manhica, Kisumu, Kilimanjaro 03/07/2009
65
TABLE 55 Antibody concentration response to the measles vaccination by treatment group GMC
pre-vaccination (95%CI)
GMC post-vaccination (95%CI) *
Median post-vaccination
Children with detectable concentrations pre-vaccination Placebo (n=223) 146 (119,179) 1384 (1149,1668) 1444
Combined (n=318) 113 (97,132) 896 (760, 1055) 934 Children with undetectable concentrations pre-vaccination
Placebo (n=1058) 814 (753, 879) 840 Combined (n=2045) 756 (717, 798) 783
Ratio Combined/Placebo 1.05 (0.96, 1.15) P-value 1 0.31 Children with protective measles antibody level pre-vaccination
Placebo (n=103) 516 (395, 673) 2381 (1782, 3183) 2209 Combined (n=128) 409 (328, 509) 1816 (1440, 2291) 1706
Children without protective measles antibody level pre-vaccination Placebo (n=1,178) 819 (762, 881) 848
Combined (n=2,235) 737 (700, 776) 768 Ratio Combined/Placebo 1.02 (0.93, 1.11) P-value 1 0.67 All children
Placebo (n=1,281) 893 (830, 959) 899 Combined (n=2,363) 774 (735, 815) 795
Ratio Combined/Placebo 1.00 (0.92, 1.09) P-value 1 0.92 *74 children had non-detectable concentrations post-vaccination. These undetectable levels were assigned a value of 11IU/L, half of the minimum concentration detected in the pooled dataset, for purposes of the analysis in this table (this affects the GMC but not the median). 1 P-value is calculated from the likelihood ratio test comparing the regression model with treatment group and site as the only factors to the model with only site. H0= ratio equal to 1
No evidence has been found to support a difference in the measles GMC between the single
combined treatment group and placebo.
Final pooled analysis V.3 Navrongo, Manhica, Kisumu, Kilimanjaro 03/07/2009
66
FIGURE 35 Reverse cumulative distribution function for measles for all children
0.2
.4.6
.81
1 2 3 4 5log(10) measles antibody concentration post-vaccination
Placebo Combined
(vertical line indicates assumed protective threshold)Reverse empirical cumulative distribution function
12.3.2.2 Proportion of children with protective concentrations post-vaccination
12.3.2.2.1 Excluding those with protective measles concentrations in their pre-vaccination sample
231 children had protective concentrations of measles antibody in their pre-vaccination
sample. These children have been excluded from this point on, leaving 3,413 individuals.
TABLE 56 ITT proportion of children protected from measles; excluding those protected at pre-vaccination Protected Unprotected Total Percent unprotected Placebo 1,117 61 1,178 5.18% Combined 2,092 143 2,235 6.40% Null hypothesis difference Combined-Placebo of 5% Actual difference 1.22%, P<0.0001, z=-4.76 90%CI (-0.14, 2.58) 95%CI (-0.40, 2.84) 99%CI (-0.91, 3.35)
Final pooled analysis V.3 Navrongo, Manhica, Kisumu, Kilimanjaro 03/07/2009
67
A further 118 children are excluded from the ATP analysis, leaving 3,295 individuals.
TABLE 57 ATP proportion of children protected from measles; excluding those protected at pre-vaccination Protected Unprotected Total Percent unprotected Placebo 1,089 58 1,147 5.06% Combined 2,012 136 2,148 6.33% Null hypothesis difference Combined-Placebo of 5% Actual difference 1.27%, P<0.0001, z=-4.64 90%CI (-0.10, 2.65) 95%CI (-0.36, 2.91) 99%CI (-0.87, 3.42)
The formal test of non-inferiority, of the null hypothesis that the difference in proportion
unprotected between the intervention groups is 5% or more, gives strong evidence
suggesting combined group is not inferior to placebo for both ITT and ATP.
FIGURE 36
Reverse cumulative distribution function for measles for the ITT cohort excluding those protected at pre-vaccination
0.2
.4.6
.81
1 2 3 4 5log(10) measles antibody concentration post-vaccination
Placebo Combined
(vertical line indicates assumed protective threshold)Reverse empirical cumulative distribution function
Final pooled analysis V.3 Navrongo, Manhica, Kisumu, Kilimanjaro 03/07/2009
68
12.3.2.2.2 Excluding those with detectable measles concentrations in their pre-vaccination sample
541 children had protective concentrations of measles antibody in their pre-vaccination
sample. These children have been excluded from this point on, leaving 3,103 individuals.
TABLE 58 ITT proportion of children protected from measles; excluding those with detectable concentration at pre-vaccination Protected Unprotected Total Percent unprotected Placebo 1,002 56 1,058 5.29% Combined 1,928 117 2,045 5.72% Null hypothesis difference Combined-Placebo of 5% Actual difference 0.43%, P<0.0001, z=-5.59 90%CI (-0.98, 1.84) 95%CI (-1.25, 2.11) 99%CI (-1.78, 2.64)
A further 103 children are excluded from the ATP analysis, leaving 3,000 individuals.
TABLE 59 ATP proportion of children protected from measles; excluding those with detectable concentration at pre-vaccination Protected Unprotected Total Percent unprotected Placebo 980 54 1,034 5.22% Combined 1,853 113 1,966 5.75% Null hypothesis difference Combined-Placebo of 5% Actual difference 0.53%, P<0.0001, z=-5.39 90%CI (-0.90, 1.95) 95%CI (-1.18, 2.22) 99%CI (-1.71, 2.76)
The formal test of non-inferiority, of the null hypothesis that the difference in proportion
unprotected between the combined treatment group is 5% or more, gives strong evidence
suggesting non-inferiority compared to placebo for both ITT and ATP.
Final pooled analysis V.3 Navrongo, Manhica, Kisumu, Kilimanjaro 03/07/2009
69
FIGURE 37 Reverse cumulative distribution function for measles for the ITT cohort excluding children with detectable antibody levels at pre-vaccination
0.2
.4.6
.81
1 2 3 4 5log(10) measles antibody concentration post-vaccination
Placebo Combined
(vertical line indicates assumed protective threshold)Reverse empirical cumulative distribution function
22
Annex 4: Summaries of Individual Study Results
Navrongo, Ghana
Introduction
This was a cluster randomised controlled comparison of IPTi-SP or placebo given at DTP2,
DTP3 and measles vaccination, with an extra dose given at 12 months of age not linked with
vaccination. All infants received one month of iron supplementation at the time of drug
administration. There were 48 clusters (comprised of 100 households) in each arm and a
total of 2,485 children. Unlike the other IPTi studies, this study was conducted in a region in
which yellow fever vaccine was administered concurrently with measles at 9 months of age.
It should be noted that the Navrongo study was nearing completion at the time that the
protocol for the EPI serology studies was being designed. Since the study provided a unique
opportunity to obtain information on serological responses to yellow fever vaccination, a
retrospective decision was made to measure serological responses to measles and yellow
fever on a selection of blood samples taken at 9 and 12 months. Paired pre- and post-
vaccination blood samples on infants documented as having received concurrent measles
and yellow fever vaccination at 9 months were therefore selected from each cluster for
yellow fever serology. Additional paired sera for measles serology were selected from
children that had received measles vaccine alone.
Results
For detailed results see Interim Report Navrongo (May 17, 2006) and Final Yellow Fever
Serology Report Navrongo (Final Draft November 20, 2006).
Measles
Of 2,485 infants enrolled, blood samples for measles serology were obtained from 336
infants who received SP (from 48 clusters) and from 284 infants who received placebo
(from 47 clusters). Post-vaccination geometric mean antibody titres (GMT) in 146 infants
that had detectable levels of measles antibody prior to vaccination were much higher (see
statistical report) than the post-vaccination GMTs in infants with undetectable levels of
measles antibody prior to vaccination. These infants were therefore excluded from the main
analysis, leaving 474 infants (254 SP, 220 placebo).
Following vaccination, 458/474 (97%) infants achieved a concentration of measles antibody
equal to or above the protective threshold, and the proportion of infants unprotected was
similar (3.1% IPTi, 3.6% placebo) in both groups. P-values for the formal test of non-
inferiority in both the intention to treat (ITT) and according to protocol (ATP) analyses were
very small (≤ 0.01), indicating that the difference (SP minus placebo) in proportions
unprotected was not greater than the defined clinical threshold of 5%. The upper 99%
confidence limit was <5%, making it possible to state with a high degree of confidence that
IPTi-SP does not impair serological responses to measles vaccine. The two statistical
methods that were used to adjust for the cluster design of the study produced very similar
results, and the confidence limits from both methods agreed closely. Reverse cumulative
curves were similar for both the IPTi-SP and placebo groups.
23
Yellow fever
Unfortunately, many of the samples selected for yellow fever had insufficient volumes for
testing. Additionally, there was a miscommunication with the lab about the dilution, which
further reduced the available samples to only 136 (64 in placebo and 72 in SP), of which
only 37 had matching pre-vaccination samples (17 in placebo and 20 in SP).
Owing to the very small sample size and numerous complicating factors concerning the
absolute antibody level for protection, the analysis was limited to a comparison of the two
groups only using the post-immunization samples and for the entire cohort (not excluding
any of the children with inconclusive replicate samples or detectable or protective antibody
levels prior to immunization). Two different methods (t-test and generalized estimating
equation GEE) were used to adjust for clustering and both gave similar results. A t-test
analysis comparing the proportions protected in the two arms ignoring clustering was also
conducted.
133/136 (97.8%) children had a yellow fever antibody titre at or above the protective
threshold in their post vaccine samples. The p-values for the formal test of non-inferiority
were small across all three methods (adjusting and not for clustering) confirming the
rejection of the null hypothesis and that the difference in proportions unprotected was
significantly greater than 10% in the SP group. The actual difference SP-placebo found was
4.21%; p=0.004 (GEE), 3.13%;p=0.0003 (t-test ), 4.17%; p=0.01 (t-test ignoring clustering).
The upper confidence limits were consistently less than 10% indicting that the intervention
did not impair the serological responses to yellow fever vaccine.
Comments (WHO Advisory Committee)
A surprisingly high proportion (97%) of infants achieved protective levels of measles
antibody post vaccination*. The augmented post-vaccination response in infants with
detectable antibody pre-vaccination is more suggestive of previous exposure to measles than
persistence of passively acquired immunity. Despite failing to reach the calculated sample
size of 500 per group, a statistically robust result has been achieved with this data.
* This was subsequently discussed with the laboratory, and is likely to reflect reduced
specificity of the PRN assay used at the time.
Conclusions (WHO Advisory Committee)
Serological data from the Navrongo study strongly suggest that IPTi with SP does not have
an adverse impact on serological responses to measles vaccine.
Taking into consideration the analysis of the limited data that are available and the
difficulties with its interpretation, the reverse cumulative distribution functions are very
reassuring and suggest that there is no negative interference of SP on yellow fever
vaccination. The analysis not allowing for clustering produces similar (although more
conservative) results as the other methods, which confirms rejection of the null hypothesis
of a greater than 10% difference.
24
Manhiça, Mozambique
Introduction
This was an individually randomized, controlled comparison of IPTi-SP and placebo given
at the time of DTP2, DTP3 and measles vaccination. Serological responses were measured
in relation to all EPI vaccines except Haemophilus influenzae and yellow fever, since these
are not included in the Mozambique vaccination schedule.
Results
For detailed results see Annex: Final Report Manhiça (May 17th 2006).
Measles
Of 1503 children randomized, 748 received SP and 755 received placebo. Paired (pre- and
post-vaccination) samples were available for 624 infants (308 SP, 316 placebo). Post-
vaccination geometric mean antibody titres (GMT) for 131 infants with detectable levels of
measles antibody prior to vaccination were higher (see statistical report) than the post-
vaccination GMTs in infants with undetectable levels of measles antibody prior to
vaccination. Infants with detectable levels of measles antibody pre-vaccination were
therefore excluded from the main analysis, leaving 493 infants (254 SP, 239 placebo).
Following vaccination, 461/493 (93.5%) infants achieved a concentration of measles
antibody equal to or above the protective threshold, and the proportion of infants
unprotected was similar in both groups (6.7% IPTi, 6.3% placebo). For both the ITT and
ATP analyses, the formal test of non-inferiority was significant (p ≤ 0.03) at the 5% level
and the upper limit of the 90% confidence interval for the difference did not exceed 5%,
indicating that IPTi-SP had no adverse effect on responses to measles vaccine. Reverse
cumulative distribution functions were similar for both the IPTi-SP and placebo groups.
Other EPI antigens (DTP, polio serotypes 1 and 3, hepatitis B)
688 serology samples were available for statistical analysis. Complete data (on all EPI
antigens) were available for 237 infants. The proportion of infants achieving protective titres
post-vaccination was similar in the IPTi-SP and placebo groups for diphtheria, tetanus, polio
serotypes 1 and 3, and hepatitis B. The test of non-inferiority, with a 10% threshold, was
significant in each case, indicating that the difference (SP minus placebo) in the proportion
of infants that did not attain protective antibody concentrations was not greater than the
defined clinical threshold of 10%. There are no known serological correlates of protection
for pertussis. For all antigens, post-vaccination geometric mean titres (GMTs) and reverse
cumulative distribution functions were similar in the IPTi-SP and placebo groups.
Comments (WHO Advisory Committee)
Measles
As for Navrongo, a high proportion (93.5%) of infants achieved protective levels of measles
antibody post vaccination (see section on measles PRN, below). The augmented post-
vaccination response in infants with detectable antibody pre-vaccination is more suggestive
of previous exposure to measles than persistence of passively acquired immunity. Despite
failing to reach the calculated sample size of 500 per group, a statistically robust result has
been achieved with this data.
25
Other EPI antigens
The data look very reassuring.
Conclusions (WHO Advisory Committee)
Serological data from the Manhiça study strongly suggest that IPTi with SP does not have an
adverse impact on serological responses to measles, diphtheria, pertussis, tetanus, polio or
hepatitis B vaccines.
26
Bungoma, Kenya
Introduction
This study was an individually randomized, controlled comparison of IPTi-SP and placebo
given at the time of DTP2, DTP3 and measles vaccination. Serological responses were
measured in relation to all EPI vaccines except yellow fever. The study was sponsored by
WHO/TDR, and had been running for over a year when the decision (approved by the WHO
and local Kenyan Ethical Review Committees) was made to expand the sample size in order
to collect EPI serology data. The study took place in a busy maternal and child health clinic
in western Kenya, many of its short-comings reflect its "real-world" setting, and for this
reason it might most accurately be described as a community effectiveness study. Due to the
relative inexperience of the principal investigator and insufficient clinical monitoring (the
first monitoring visit took place after recruitment of all 934 patients), conduct of the trial
was sub-optimal in a number of important respects, which subsequently raised doubts about
the validity of the clinical and serological data. The main problems were as follows:
Informed consent
The consent form that was used for children in the EPI serology study did not mention the
need for multiple venepuncture, although the procedure was explained to all parents and
guardians. This omission was discussed with the WHO Ethical Review Committee (ERC)
and the local Kenyan Institutional Review Board (IRB). Both Committees considered that
the principal investigator and her team should re-contact all participating parents or
guardians and obtain written affirmation that they had consented to venous blood sampling.
According to the WHO ERC, "lack of documentation of the process cannot be condoned,
but, by re-contacting each mother and obtaining an affirmation from them of the consent that
they had provided, this omission is being mitigated". Written affirmation of consent was
obtained from >90% of parents or guardians.
Drug dosing and allocation
The clinical monitor discovered half tablets in some of the unused randomization envelopes,
although each envelope should have contained a complete tablet. It subsequently transpired
that, at the time of randomization, WHO/TDR had placed half tablets in some of the
envelopes when it appeared that they might run out of placebo. From the monitor's report, it
appears that a total of five children had been significantly under-dosed as a consequence. All
of the unused study drugs (SP or placebo) were subsequently sent to London, where 50%
(84 tablets) were selected for analytical determination of content. The contents were verified
against the randomization code by an independent statistician, and allocation was found to
be correct in 100% of cases.
Database
The database was in disarray at the time of the first monitoring visit, as the data manager
had been absent for several months. All of the data pertinent to the EPI serology study were
subsequently source verified by an external monitor and double-entered into a database that
had been constructed by an experienced external data manager.
27
In view of these and other concerns, the Committee concluded that it would not be
appropriate to include data from the Bungoma study in a pooled analysis of EPI serology
along with data from the other IPTi study sites. They did, however, think that it was
appropriate to review and report on the data since: a). Blood samples had already been taken
from human subjects; b). Serological tests had already been carried out; c). The Bungoma
study represented a unique opportunity to obtain data on serological responses to H.
influenzae in infants given IPTi-SP. d). The study provided serological data from a different
geographic setting (Kenya) from that of the other two IPTi-SP serology studies
(Mozambique and Ghana).
The measles PRN tests for the Bungoma and subsequent serology studies were carried out
using a modified and improved standard operating procedure. An interim analysis was
performed on the measles results that were available as of January 2006 and this was used in
the Advisory Committee's Interim Report, July 2006. A revised final analysis including all
the measles results was completed in January 2007.
Results
For detailed results, see Final report Bungoma Final Report May 17, 2006 and updated
Measles Report January 31, 2007.
Measles
Measles serology data were available for 353 children (163 placebo and 190 SP) only, and
not 500 as originally planned. All 353 children were included in the ITT analyses. From
the ATP analyses children were excluded for not having received all three drug doses.
The GMC post vaccination was similar in both treatment groups.
Ninety-five children had detectable antibody concentrations in their pre-vaccination sample,
of which 17 had antibody concentrations above the protective level of 120 IU/l. Excluding
those with detectable concentrations pre-vaccination left 258 children. Following
vaccination, 222/258 (86.0%21) achieved a concentration of measles antibody equal to or
above the protective threshold. The proportion of infants unprotected was similar in both
groups (13.71% placebo; 14.18% SP).
For both the ITT and ATP analyses, the formal test of non-inferiority was non-significant
with a 5% threshold (p ≥ 0.15) and the confidence intervals are very wide reflecting the
inadequate sample size. Reverse cumulative curves were, however, similar for both the SP
and placebo groups.
Other EPI antigens (DTP, polio serotypes 1 and 3, Haemophilus influenzae b, hepatitis
B)
429 serology samples were available for statistical analysis. The proportion of infants
achieving protective titres post-vaccination was similar in the IPTi-SP and placebo groups
for diphtheria, tetanus, polio serotypes 1 and 3, Haemophilus infuenzae b and hepatitis B.
The test of non-inferiority, with a 10% threshold, was significant (p≤ 0.001) in each case,
indicating that the difference (SP minus placebo) in the proportion of unprotected infants
not greater than 10%. There are no known serological correlates of protection for pertussis.
21 This provides further confirmation of the increased specificity of the new SOP compared to that used for the
Navrongo and Manhica studies in which seroconversion was 97% and 94% respectively.
28
For all antigens, post-vaccination geometric mean titres (GMTs) and reverse cumulative
curves were similar in the IPTi-SP and placebo groups.
Of 411 infants in the ITT analysis with complete data on all EPI antigens, 29.9% (63/211) in
the IPTi-SP group had sub-protective responses to one or more antigens, compared to 27.5%
(55/200) in the placebo group, a non-significant difference of 2.4%.
Comments (WHO Advisory Committee)
Measles
The proportion of infants (86%) achieving post-vaccination concentrations of measles
antibody equal to or above the protective threshold is comparable to that observed in other
populations. The Bungoma study was substantially under-powered for measles, so the non-
significance of the test of non-inferiority is not surprising. The reverse cumulative
distribution functions are reassuring, however, since curves for SP and placebo are closely
aligned at the protective level of 120 IU/L and across a wide range of antibody
concentrations.
Other EPI antigens
The non-inferiority tests are significant (p ≤ 0.001), allowing rejection of the null hypothesis
that SP has an adverse impact on serological responses to EPI antigens. The reverse
cumulative distribution functions are reassuring for all antigens. The similarity between the
intention to treat (ITT) and according to protocol (ATP) analyses, the proportion of infants
attaining the protective level for each EPI antigen and the comparison of the SP and placebo
groups do not suggest that there are specific problems with the dataset.
Conclusions (WHO Advisory Committee)
Data from the Bungoma study suggest that IPTi-SP does not have an adverse impact on
serological responses to diphtheria, tetanus, pertussis, polio serotypes 1 and 3, Haemophilus
influenzae b and hepatitis B vaccines.
Given that the Bungoma analysis is substantially under-powered for measles, the non-
significance of the test of non-inferiority is not surprising. This combined with the
difficulties encountered implementing the trial limit the ability of the Committee to make a
strong conclusion. However, with these provisos, on the basis of the available results from
Bungoma, the Committee finds no reason to believe that there is any negative impact of SP
when given with measles vaccination.
29
Kisumu, Kenya
Introduction
This was an individually randomized, controlled comparison of three IPTi drug groups, SP-
ART, AQ-ART, and LapDap and a placebo group. The target sample size for each of the
four groups was 37922
. The interventions were given at the time of DTP2, DTP3 and
measles vaccination. Serological assessment was conducted for all of the EPI antigens -
DTP, polio, measles, hepatitis B and Haemophilus influenzae b which are in the national
immunization schedule in Kenya. This was the only trial to assess the serological responses
of IPTi given with Hib vaccine.
Results
For detailed results see Final report Kisumu, March 17, 2009.
Measles
1,516 children were enrolled in the study. Paired measles serology data was only available
for 1,122. Of these, 284 were in placebo, 285 in SP-ART, 293 in AQ-ART, and 260 in
LapDap groups. All 1,122 children were included in the ITT analyses. Children not having
received all three drugs were excluded from the ATP analyses.
The ratios comparing GMC post vaccination titres for each of SP-ART, AQ-ART, and
LapDap to the placebo group were not significantly different from 1, or from one another.
The same result was observed for: i) all children, ii) children without protective measles
antibody level pre-vaccination; and iii) children with undetectable concentrations pre-
vaccination.
Following vaccination 95.1% achieved a concentration of measles antibody equal to or
above the protective threshold of 120 IU/L. The proportion of infants unprotected was
5.24% placebo; 6.25% SP-ART; 5.88% AQ-ART; and 1.76% LapDap. Test of non-
inferiority rejected the null hypothesis that the difference (ITPi treatment minus placebo) in
proportions unprotected is 5% or more. In the ATP analyses the actual differences were SP-
ART 1.03% (p=0.03); AQ-ART 1.25% (p=0.03) and LapDap -3.09% (p<0.0001), providing
evidence to suggest that all three IPTi drugs are not inferior to placebo. This is further
supported by the reverse cumulative distribution functions which are very similar.
Other EPI antigens (DTP, polio serotypes 1 and 3, HepB, and Hib)
Serology data on any one of the antigens was available for a total of 1,135 children for the
ITT and 1,028 for the ATP analysis. The target sample size for each group and each antigen
was 250 children. Number of samples tested:
22 The Kisumu trial was designed with a sample size of 379 per arm at the beginning. When the serology study
was added the study was unable to increase the sample size to 500 per arm for budgetary and operational
reasons.
30
TABLE 9
Number of samples tested by antigen
Placebo SP-ART AQ-ART LapDap TOTAL TOTAL in ATP
Diphtheria 259 255 254 238 1,006 913
Tetanus 179 168 180 173 700 636
Pertussis toxin 161 157 154 151 623 563
FHA 235 220 222 216 893 813
Hib 251 252 255 239 997 897
Polio Type 1 249 240 252 227 968 880
Polio Type 3 249 240 252 227 968 880
Hepatitis B 68 52 59 59 238 222
Complete data on all antigens was available on 107 children.
For diphtheria, tetanus, Hib, and pertussis toxin the GMC after vaccination was similar in
the four treatment groups in both ITT and ATP analyses. For Hep B there was evidence to
reject the null hypothesis of equal GMC between all four groups.
The null hypothesis that the difference (ITPi treatment minus placebo) in proportions
unprotected is more than 10% was rejected in the AQ-ART and LapDap groups for
diphtheria, tetanus, Hib, HepB, and polio type 1. For polio type 3 the null hypothesis that the
difference in proportion unprotected is at least 10% could not be rejected for the LapDap
and SP-ART groups for either ITT or ATP analyses.
The reverse cumulative curves were reassuring except for HepB, possibly as a result of the
small sample size available for this antigen.
To investigate HepB responses further a post-hoc analysis using 100IU/L (the level at which
long term immunity is conferred) instead of 10IU/L (level for seroconversion) as a cut-off.
The null hypothesis of inferiority of the SP-ART and LapDap groups compared to placebo
for both the ITT & ATP analyses could not be rejected. The comparison of AQ-ART with
placebo found evidence to reject the null hypothesis of inferiority in both ITT and ATP
analyses.
31
Comments (WHO Advisory Committee)
Measles
Despite the difficulties collecting the samples and with the caveat that the children included
are a representative sample, the Committee found it reassuring that the results for the four
study groups were very similar in the protective levels of measles antibodies.
Other EPI antigens
The post-hoc analysis of HepB using 100IU/L was seriously marred by the small sample
size. However, there was no evidence of any problem with SP-ART, some evidence that
AQ-ART might be better than placebo, and some very weak evidence that LapDap might be
worse than placebo (the proportions under 100IU/L being quite high). However, nothing
more definite can be said owing to the lack of power.
Overall, the Committee was impressed with the robustness of the response of all the
antigens across all the drug combinations.
Conclusions (WHO Advisory Committee)
In spite of the much smaller sample sizes than expected, the Committee was satisfied that
the null hypothesis was supported by the results and that the three drug combinations, SP-
ART, AQ-ART, and LapDap, were not inferior to placebo.
32
Kilimanjaro, Tanzania
Introduction
This was an individually randomized, controlled comparison of three IPTi drug groups, SP,
MQ, and LapDap and a placebo group. The interventions were given at the time of DTP2,
DTP3 and measles vaccination. Serological responses were measured for measles vaccine
only.
Results
For detailed results see Second Draft (actually Final Report) Kilimanjaro, November 21,
2008.
Measles
2,419 eligible were enrolled in the study, and paired pre and post samples were collected
from 1,812 children. Due to insufficient volume or sub-optimal quality of the samples,
paired measles serology data was only available for 1,538 children. There were no
significant differences between treatment groups in the samples lost due to sample quality or
quantity. Of the 1,538 children usable pre and post samples, 397 were in placebo, 374 in SP,
380 in MQ, and 387 in LapDap groups. All 1,538 children were included in the ITT
analyses. For the ATP analyses children were excluded for not having received all three
drug doses.
The GMC values post-vaccination were similar in all four treatment groups.
161 children had detectable measles antibody concentrations in their pre-vaccination
sample. Excluding those left 1,377 children. Following vaccination, 1296/1377 (94.1%)
achieved a concentration of measles antibody equal to or above the protective threshold. The
proportion of infants unprotected was similar in both groups (5.70% placebo; 5.18% SP;
5.52% MQ; and 7.06% LapDap).
Formal tests of non-inferiority, of the null hypothesis that the difference (ITPi group minus
placebo) in proportions unprotected is 5% or more, rejected the null hypothesis in each case
in ITT analyses. Owing to the smaller than expected numbers of children included in this
analysis and the resulting loss of power the evidence is not sustained at the 1% significance
level for the LapDap and placebo group comparison.
In the ATP analyses (further 62 children were excluded for not having received all three
drug doses, leaving 1,315) the formal test of non-inferiority (for all three comparisons),
rejected the null hypotheses SP ( treatment minus placebo difference in proportion
unprotected -0.47%; p=0.002), MQ (-0.48%; p=0.003) and LapDap (1.42%; p=0.03) were
inferior to placebo.
The reverse cumulative distribution functions for all interventions and placebo were very
similar.
33
Comments (WHO Advisory Committee)
Measles
The Committee agreed that all three formal tests of non-inferiority (the null hypothesis that
the difference in proportions unprotected between each of the investigational products with
placebo is 5% or more) give results that provide strong evidence rejecting the null
hypothesis for SP and MQ. However, due to the smaller than expected numbers of children
included in this analysis and the resulting loss of power the evidence found is not sustained
at the 1% significance level for the LapDap and placebo group comparison.
Conclusions (WHO Advisory Committee)
The Committee was satisfied with the robustness of the results and in particular the
concordance of the reverse cumulative distribution functions. Taken together the Committee
concludes that this evidence supports that IPTi treatment with SP, MQ, and LapDap in this
study had no influence on the serological response to measles vaccination.