rr;-,
!
, ji r{ :'i l'
':!'.'".jtMode[ing onchocerciasis transmission and control
Final report for Technical Service Agreement No. 08/181/85 provided by the
World Health Organization on behalf of the Onchocerciasis Control Programme in West Africa for the period June 1
L993to December I
1993October 1994
Centre for Decision Sciences in Tropical Disease Control (CDTDC) Dept. of Public Health, Erasmus University Rotterdam
P.O. Box 1738, 3000 DR Rotterdam, The
) 'F /:) i'\,/-\ r \rr)
Introduction
The proposed activities
for
theTSA
1993 were almost exclusively devotedto
the assessmentof
the effectivenessof
ivermectin treatment asa
meansto
control transmission: either when used aloneor
in addition to vector control.In
PartI
(page3)
we report the model based analysisof
the resultsof
the community treatmentin
Asubende (Ghana).We
have attemptedto quantiff the
model-parametersfor
the effectsof
a treatment on the microfilariae and the adult parasite.An
important question that we triedto
answeris
whether the impact on the adult worm is permanent ('macrofilaricide-like') or reversible. This analysis has a valuein
its own and has been submitted as a scientific paper.Part
II
(page 21) of the report deals with the consequences of the new insightsfor
the roleof
ivermectin based strategies. We have triedto find
those strategies basedon
annual ivermectin treatment andvector
controlwhich,
under certain conditions,will
leadto
eradicationof
the parasite from a given area, i.e.will
give rise to negligible recrudescence risks.Part
III
(page 33) comprises some miscellaneous notes which have been sentto
OCP inview of
the meetingof
EAC15.An
important itemin
these notes concerns the possible useof
entomological criteria in post-larviciding surveillance.
2
Part I
Irreversible effects of ivermectin on adult parasites in onchocerciasis patients
in the Onchocerciasis Control Programme in West Africa
Absffact
Ivermectin (MectizanR) is the drug
of
choicefor
onchocerciasis.It
eliminates microfilariae from the skin and considerably suppresses their reappearance.In
this paperit is
investigatedin
what waythis
suppressionis
associatedwith
the fecundityof
adult Onchocerca volvulus. Predictions basedon
hypotheses about reversible and irreversible effects are compiuedwith
post-treatment microfilarial countsof
114 adult personsfrom
Ghana. These persons were followed during five years of annual community treatment by the Onchocerciasis Control Programme in WestAfrica. It is
shown thatthe trend in
microfilarial counts can onlybe
explained whenit is
assumed that ivermectin,in
additionto
causing a temporaryfall
in the microfilariae production, also affecs the fecundityof
the worms irreversibly. Following each treatment, worms recover during a periodof
about 10
to
11 months, and reach a new stablelevel of microfilariaeproduction which is 30% less than before treatment.Key words: Onchocerciasis, Onchocerca volvulus, Ivermectin, Modelling
4
The registration
of
the anthelminthic drug ivermectin (MectizanR)in
1987 was a landmarkin
the controlof
human onchocerciasisor
river-blindness,a
parasitic disease causedby the
filarial nematode Onchocercavolvulus. Oral
administrationin a
standard doseof
150-200p.glkg body weight is followedby
rapid eliminationof
microfilariae (mf) from the skin and gradual reductionof ocular mf levels [1].
Side effectsare
generallymild
andthis
makes ivermectina
better therapeutic option than Diethylcarbamazine @EC), which is often accompaniedby
severe Mazotti reactions and ocular damage. Ivermectin outperforms DEC alsoin a
longer suppressionof
mf- repopulationof
the skin[2,3].
This additional effect of the drug, which was obviousin all
studies done thusfar,
has initiated researchon
the effectof
the drugon
adult parasites.Adult
female parasites in treated persons showed an interruptionof the normal embryogenesis,but
after a single treatment this appeared to be reversiblefor
mostof
the worms[4,5].
Excess worm mortality has never been observed aftera
single treatment U,2,61. Irreversible effectson
worms were found after many treatmentswith
short intervals varying betweentwo
weeks andsix
months [6-11].These studies revealed significantly more dead and moribund female worms
in
frequently treated patients thanin
notor only
once treated controls. Furthermore, the reproductive activityof
the surviving worms was markedly reduced. Since the examinations were done shortly after the last treatmentit
remains unknownto
what extent the impact on reproduction is transientor
irrevers- ible. Recent research, however, demonstrates that also one and a half year afterfive
six-monthly doses the fecundity of female worms isstill
considerably reduced [12].A
question whichis
importantin
viewof
the limited resources availableto
health services in developing countriesis
whether the findingsfor
short treatment intervals are also applicable to regimeswith
intervalsof
one year, which is the current practicein
nearlyall
control progralnmes[1a]. In
the present paperwe
addressthis
questionby
analyzing datafrom a
large community based studyof
annual treatmentsin
a hyperendemic region (Asubende, Ghana), organisedby
the Onchocerciasis Control Programmein
WestAfrica
(OCP)Il4-17).
The data consistof
microfi-larial (mf)
countsin
skin-snipsfrom
persons who were surveyedover a
periodof
almost five yearsin
orderto
evaluate thefirst five
treatments. We have investigated whether the observed trendin mf
counts can be explainedfrom
short-term and transient effectsof
ivermectinonly,
or whether long-term and irreversible effects on the adult worms are involved too.Materials and Methods
Study area and selection
of cohorts.
The Asubende region is located along the lower reachesof the Pru river
in
Ghana, just west of Lake Volta. OCP started controlof
the vector (the blackfly Simulium damnosum) in this river basin in January 1986. Flies have been collected since 1979 to assess the vectorbiting
rate and the vector infectivity.A
clinical survey was donein
September 1987 among 796 personsliving in a
clusterof
three villagesin the
middleof the
area. Both entomological and clinical findings revealed that the area was hyperendemicfor
the savanna formof O.volwlus.
The skin microfilarial densities were among the highest encounteredin
the OCP area [13]. The communitytrial of
annual ivermectin treatment was startedin
October 1987. In thefirst
roundof
treatment more than 15,000 persons were treated, comprising 6L.5%of
the study population.In
the present analysis we use data consistingof mf
countsin
skin-snips which have been collectedto
evaluate thefirst five
treatments (1987-1991). The organizationof
thetrial,
the trendsin
the skinmf
densities, and thejoint
effectof
vector control and ivermectin treatment on the transmission potentialof
the flies have been described elsewhereII4,l7l.
Treatment dosagevaried between 130 and 2ffipglkg body weight.
From the 796 persons examined at the baseline survey
in
1987,we
selectedthe
114 adult persons()20
years)who
satisfythe
inclusion criteriaof
thefollowing two
cohorts. The first cohort(n=78)
consists of persons who were treated in all five rounds and were re-examinedin
all eight follow-up surveys which were done at 4 and 12 months after thefirst two
treatments,5
and 12 months after thethird,
11 months after the fourth and6
months after thefifth
treatment. The second cohort was selectedfor
a longer follow-up periodof
a single treatment.It
consistsof
36 persons who were treated at thefirst
but not at the second round, and who were re-examineA 24 months after thefirst
treatment. We restricted ourselvesto
adult persons becauseit is only
for older ages that one may expecta
constant infectionlevel
and, hence, excludethe
confounding effectof
ageing[18].
The dark barsin
Figure 1 show the frequency distributionof mf
counts in skin-snips before treatment and at the follow-up surveys.Modelling the effects
of
ivermectin. Hypotheses on the effectof
ivermectin have been testedusing
the
stochastic simulation model ONCHOSIM.This
model describesthe
developmentof
O.volvulus
in
man and flies and the transmissionof
the parasite. Elsewhere we explain the modelin full
and report its validation ll9-221. ONCHOSIM allowsfor
a detailed simulationof
control strategies. This enabled usto
mimic the vector control activities employedin
the Asubende area, thus accountingfor
the observed fluctuationsin
the transmission potentialof
theflies [17].
The microsimulation method,which is
characterizedby
simulatingthe
life-historiesof
individual hypothetical personsin
a dynamic population (birth, acquisitionof
parasites, accumulationof
mf,6
death) and individual parasites (maturation, mating, mf-production, death), allows
for
describing and testing possible effectsof
ivermectin on the levelof
worms and humans. Since the outputof
the calculationsis on
an individual level, we were ableto
select those adult personsfrom
the simulated populationfor
whom the timingof
treatments and surveys satisfiedthe
same criteria usedfor
selectingthe
cohorts, andfor
whomthe
skin-snip count distribution beforethe
first treatment was identical to that of the cohorts.An
assumption used throughout the analysis is that ivermectin treatment eliminatesall mf in
aperson.
Apart from this
immediate effect,we will
investigate one transient andtwo
typesof
irreversible effect on the fecundity
of
adult female worms. The transient effectis
modelled as arecovery period during which the mf production rate increases
from
zeroto
a new stable level. In caseof the first
typeof
irreversible effectof
the drug, called oroductivity reduction,this
new stablelevel will be lower
than before treatmentfor all
femaleworms. The
secondtype of
irreversible effect is the immediate and permanent cessation
of
mf-productionin
a certain fraction of the parasites, beit
by death or by total loss of fecundity (.fecundity loss), while the other worms recoverto
their pre-treatment productivity. The magnitudeof
differencesin
transient and irrever- sible effects between persons and between treatmentsis
called effectvariability
andis
measuredby
the variation coefficient,i.e.
quotient of the standard deviation and the meanof
the effect. We explicitly test whether the increasein
the mf-production during the recovery periodis
linear or not. In the Annex we give a mathematical description of the assumptions.Testing of hypotheses. According to the definitions of the model parameters, we
will
test two basic hypothesesI
andII for
the effectof
ivermectin on adult worms:I.
Treatment has transient effectsonly, i.e.
thereis no
productivity reductionor
fecundityloss; II.
Treatment has alsoirreversible effects.
To
check whether the two typesof
irreversible effectdiffer in
their explana- tionof
the data, wewill
also test them separately. HypothesisIIa
states that an irreversible effectis only
causedby
permanent productivity reduction;in
hypothesisIIb it is only
caused by fecundity loss. When botheffecs
are combined, a fractionof
worms immediately looses fecun-dity,
and the remainder eventually reaches a stable mf-production, buton a
reduced level. Each hypothesis is tested by a x'based comparison of observed and predicted skin-snip distributions for the follow-up surveysof
each cohort. Some skin-snip count categories shownin Figure I
havebeen combined
so that the
numberof
individualsis at
least5 per
category. Estimationof
parameters
is
achievedby
minimizingX'using
a downhill-simplex method 1231. Apartfrom
the parameters directly related to the effectof
ivermectin, wewill
estimate the mf-lifespan, since ttris parameter is an important determinantof
the delay between stabilizationof
the mf-production andstabilization
of the mf{ensity in the skin
(see Annex). Furthermore,we
always assume 3%treatment
failure.
Suchfailure has
been reported before andhas
beenmainly
ascribed tomal absorption (d iarrhoea, vo mitin E, etc. I24l) .
Results
The results
of fitting
the hypotheses about the effectof
ivermectin to the dataof
the surveys afterfive
treatment rounds strongly suggest that the drug affects the mf-productionof
female parasites not only temporarily(P<0.01)
but also irreversibly(P>0.1);
see Table 1. The irreversible effect can be quantified as a productivity reductionof
35%for all
female worms,or
as a total lossof
fecundity (or death)
for
28%of
the worms. The data are explainedwell
by both hypotheses, and detailed biological data would be requiredto
differentiate between them. The goodness-of-fit didnot
improve significantlyby
combining thetwo
typesof
irreversible effect;they
appearto
be interchangeable. Lower valuesfor
productivity reduction have to be compensatedby
higher valuesfor
total lossof
fecundityin
such a way that the combined lossof
productivityin
a patient equals about3?%.The
recoveryin
mf-productionis
estimatedto
take 10-ll
months, andthe
mf-production accelerates during this period (the associated shape parameter, not shown in the Table, has a valueof
1.5with
a 95Vo confidence intervalfrom
1.2to
1.9). Note that the absenceof
an irreversible effect leadsto
larger valuesfor
the recovery period and the mf-lifespan,but
apparently without achieving a sufficiently good explanation of the data.There appears
to
be a considerable variability between patients in the effectof
treatment. The valuesof 0.52
and0.54 for the
effect variabilityimply
an inter-quartile range,i.e. a
rangeof
values which covers the centre half
of
the patients,of
6to
14 monthsfor
the recovery period,of
lTVo to 38% for the fecundity loss, and of 20% to 47% for the productivity reduction.
Table 1 also gives
95%
confrdence intervalsfor
the parameter estimates. Since the recovery period and the mf-lifespanjointly
determine the speedof
mf-repopulationof
the skin, they are to some extend interchangeable, and consequentlyfairly
wide confidence intervals were found for these parameters.The limited
informationon
the lifespanof
microfilariae[25)
suggests that 4 months(our lower limit) is highly
unlikely.This
implies that the associated upperlimit of
168
months for the recovery period is possibly also too high
Figure I
comparesthe
observed mf-count frequency distributionsfor the
post-treatment surveyswith
the predictions from the hypothesisof
permanent productivity reduction (Hyp. IIa).There are no systematic differences between prediction and observations
for
cohort1 (Fig.
1a).The lowest and highest mf-count categories are underestimated
for
the follow-up surveyof
cohort2 (Fig.
1b),which
suggests thatthe
longer-term effectof
treatmentmay vary
more between patients ttran predicted by the model.The fit of the
predictionsto the
survey datais
summarizedin Figure 2. The
data are represented asa
geometric mean mf-count.We fitted
an exponential curveto the
peaksof
the predicted post-treatment trends, which represent the situation just before a new annual treatment.The figure clearly shows that disregarding irreversible effects
of
ivermectin leadsto
an underes- timation of the additional effect of each further treatment.Discussion
Our
analysisof the
resultsof five
consecutive annual treatments provides strong evidence that apartfrom killing
microfilariae, ivermectin also has a significant impact on theviability of
adult female parasites.A
good explanationof
the trendsin
mf-counts was obtainedby
assuming that each treatment is followedby
a periodof
aboutl0
monthsof
gradually increasingmf
productionto
a level whichis
30% lower than before treatment. The datadid
not allow usto
differentiate between the hypothesis that all female parasitesin
a person are affected equally and the hypothesis that a fraction looses fecundity completely (or dies) while the others recover to their pre-treatment mf-production level.The temporary effect is
in
line with the results from earlier studies which showed that in most worms the normal releaseof
mf was interrupted after treatment, but thatI0 to
12 months later the percentageof
wormswith
normal production had increased again significantly 14,5,26). Although noneof
these studies revealed excess mortalityof
wormsby
a single treatment, after one year a considerable fraction (40% lsD was not (yet) releasingmf.
This explains the delayin
repopulationof the skin by mf
11,3,27-291. Even aftertwo
years markedly reduced mf-levels were found [30,31], although most persons had become mf-positive again. Also in our cohort2,the
mean mf-density after
two yqus
wasstill
less than half the pre-treatment level. These observations suggest a lasting treatment effect. On the other hand, the dashed line in Figure2
makes clear that one hasto be careful
in
drawing conclusions from visual inspectionof
mf-trends: also without irreversible drug effects the predicted trend shows an, albeit insufficient, decreasein
the mf-count at the 12- month follow-up surveys (the peaksof
the curves). This predicted decreaseis
exclusively due to the reduced transmissionof
the parasite asa
consequenceof
ivermectin treatment and (partial) vector control [17].Conclusive biological evidence
of
irreversible effectsof
ivermectin was obtained from patients who had receivedfour to
twelve doses at intervalsof two
weeksto six
months. More dead or moribund female parasites were foundin
these patients than controls who received noor
a single treatment [5-7,9,10]. Excess worm mortalitiesof
25to
33% were foundin
patients who received eightto
eleven 3-monthly doses[9].
These percentages are aboutthe
same asthe
irreversible effectwe
foundper
treatment round.This
suggests thatthis
effectis
predominantly dueto
aproductivity reduction
for
all female worms and to a much lesser extent due to total fecundity lossor
death. Analogousto
our results, six months after the last dose (the longest follow-upin
moststudies) most female worms had not resumed normal embryogenesis and the percentage
of
viable worms was markedly reduced. Recent investigations suggest that recovery tofull
productivity after this short follow-up period is unlikely orwill
at least take more than two years [12].Our findings partially rely on the interpretation of entomological data (counts
of
flies and their parasite load). We have taken into account the vector control schedule and the resulting trend in the transmission potentialof
theflies [7]. In
calculating this transmission potential we assumed that the parasite larvaein flies
originatefrom
inhabitantsof
Asubende andwill
again be trans- mitted to them. Such an assumption would not be valid in areas with migrationof
flies. However, Asubende is an isolated focus with a local transmission[3],
and hence ideal to test the impactof
ivermectin.
The data on mf-counts enabled us
to
test hypotheses on the effectof
ivermectin, but do notallow to draw
inferenceson the exact biological
mechanismsinvolved.
Unavoidably, simplifications hadto
be madein
the model. Wewill briefly
discusstwo
simplifications which may be importantfor
the interpretationof our
results.Firstly,
we disregarded any effect on the adult male worms.In
some studiesa
significant reductionof
the numberof
male worms were found[7,9],
whilein
others this could not be concluded [5,6,10]. Although lower countsof
malel0
wonns may be due
to their ability to
leave nodules[32], it
may,at
least temporarily, lead to reduced mating chances. Hence,the
transient and irreversible reductionin
mf-production we found couldpartly be
attributedto
absenceor
reducedviability of
maleworns.
Secondly, we assume no prophylactic effectof
ivermectin. Such drug induced protection against (super)infection would leadto a lower
transmission than expected on the basisof
fly-infection data.In
experi- mentally infected chimpanzeesa
partial prophylactic effect was foundon
L3Jarvae,but not
on later stages[33].
Since the L3-stage lastsfor
only 3to 4
days and ivermectinis
rapidly cleared from the body, no important prophylactic effects are to be expected.We have tested several other assumptions and extensions
of
the model,like
effect-variability between the wormsin
one patient, genetic predispositionof
treatment effect, etc. Noneof
them affectedour
conclusionon the
transient and irreversible natureof
treatment effector led to
asignificantly different quantification.
It
is, however, importantto
stress thatour
conclusions have been based on the resultsof
annual treatment using a doseof
l3D-ZDpglkg body weight. Changes in treatment frequencies or doses may lead to other effects per treatment.Our conclusions have important implications
for
the public health impactof
strategies basedon
annual ivermectin treatment, which is the currently recommended regimen.In
earlier studies we emphasized the potentialof
the drugfor
reducing the burdenof
blindness [20,34]. However, due to lackof
sufficient follow-up data at that time, an irreversible effect of treatment on the adult wormscould not
demonstrated.This
wasthe
main reasonfor
doubts aboutthe
potentialof
ivermectin
for
transmission control and thusfor
our cautiousnessin
designatingit
as the successorof
vectorcontrol. Our new
results certainlymerit a
reconsiderationof this point of view. It
should,
for
example, be noted that when each treatment leadsto
an over-all irreversible reduction of the fecundityof
worms with30Vo, after five treatments thiswill
be more than 80% on averagefor
those female worms that survive the whole treatment period. Translatingthis
figureto
the impactof
a long period(>
10)of
annual treatment is not straightforward.Two
complications are thatnot
everybodywill
be treated(a
coverageof
65-70% would be excellentin
routine health care) and that transmissionwill
continue (albeiton a lower
level)so
that new infectionswill
occur. Preliminary predictions
with
ONCHOSIM indicate that, although the impactof
long term ivermectin strategies is much more pronounced than we concluded earlier, the parasitewill
not be eradicatedwithin a
periodof
15 yearsof
annual treatmentin
an endemic area.We will
start shortlywith
extensively analyzing the possibilitiesof
applying annual ivermectin treatmentto
stop vector control earlier thanoriginally
plannedin
some areasof the
OCP.We will
also further explore the potentialof
the drugto
control recrudescenceof
infectionif it
occurs after stoppingvector control.
Finally,
the promisingresuls of our
studywill
stimulate the assessmentof
the impact of higher doses on the viability of adult parasites.Acknowledgement
We would like
to
thankDr. E.M.
Samba, Director OCP, and the staffof
OCP and the National Oncho Team in Ghana for their support of our work.T2
Annex
If m
denotesthe
mean fraction fecundity loss andyii the
effectivenessof
treatment round! G:1,..,5) in
personj,
then a fraction v;3mof
the wormsin
personj will
permanently cease mf- production immediately after treatment!. vij
is a random variable, whichfor
each treatment!
andeach person
j is
generatedfrom
a gamma probability distributionwith
mean1.0
and a standard deviation equivalentto
the effect-variability.For
those female wonnsin
personj
whichdo
not loose fecundity after treatment!
(a fraction1:,:d
the mf-productivityI1r of
each wormk
at timeI
after treatment is described by:r,,r(t) = r,l.dortr -vfl
= ,,|*(r)r(t -v,d)
for t<vrrTr
for t>v,rTr
"(
)"
(1)t
vrrTr
In this
expressiondg1 is the
mf-productivityof worm k without
treatment.This
basic mf- productivity depends on the age and the mating history of the worm. The parameterd
is the mean irreversible productivity reduction,Tr
is the mean durationof
the recover), period, and Sis
the shapeof the
recovery.If s>1,
thenthe
increasein the
mf-productivityis initially slow
andaccelerates by the end of the recovery period.
If g:1,
then this increase is linear.In
the hypothetical case that the random variable v13 takes such high values that v1i4q and/or v1,dexceed 1 the products are truncated
to
1. Given the estimates of the effect-variability (around 0.5;Table 1) these situations are however highly unlikely to occur.
In
caseof
repeated treatment G> 1),in
the modelit is
not allowed that an ineffective treatment(which implies a short
recoveryperiod)
acceleratesthe
recoveryof a previous
(effective) treatment.In
caseof
total treatment failure (3Voof
treated persons)v;, is
zero andno mf
arekilled. It is
assumed thatmf will in
principle be detectablein the skin
immediately after their releasefrom the worm. No
provisionis
madefor a
delay dueto
dispersalin the
body and penetrationof
skin tissues.Mf
are assumed to have a fixed lifespan CIm).In
this simple concept,if
the mf-production rateof
the wormsin
a person stabilises at time!,
then the mf-densityin
the skin stabilises at timet*Tm.
References
1.
AwadziK,
DadzieKY,
Schulz-KeyH,
Haddock DRW, GillesHM, Aziz MA.
The chemo- therapyof
onchocerciasisX. An
assessmentof
four single dose treatment regimesof
MK-933 (ivermectin) in human onchocerciasis. Ann Trop Med Parasitol 1985;79:63-78.2.
GreeneBM,
TaylorHR,
CuppEW,
etal.
Comparisonof
ivermectin and diethylcarbamazine in the treatment of onchocerciasis.N EnglJ
Med 1985;313:133-8.3.
LarivibreM,
VingtainP, Aziz MA,
etal.
Double-blind studyof
ivermectin and diethylcar- bamazine in African onchocerciasis patients with ocular involvement. Lancet 1985;2:174-7.4.
Schulz-KeyH,
KlagerS,
AwadziK, Diallo S,
GreeneBM,
LarivibreM, Aziz MA.
Treat- mentof
human onchocerciasis: the efficacyof
ivermectin on the parasite.Trop
Med Parasit 1985;36 Suppl II:20.5.
DukeBOL,
Zea-FloresG,
MuflozB.
The embryogenesisof
Onchocerca volvulus over the first year after a single doseof
ivermectin. Trop Med Parasitol 199l:42:175-80.6.
DukeBOL,
Zea-FloresG,
CastroJ,
CuppEW,
MuffozB.
Comparisonof the
effectsof
asingle dose and
four
six-monthly dosesof
ivermectinon
adult Onchocerca volvulus.Am
JTrop Med
Hyg
199l;45:132-7.7.
DukeBOL,
Zea-FloresG,
CastroJ,
CuppEW,
MuffozB. Effecs of
multiple monthly dosesof
ivermectin on adult Onchocerca volvulus. Am J Trop MedHyg
1990;43:65744.8.
DukeBOL,
PacqudMC,
MufiozB,
GreeneBM,
TaylorHR. Viability of
adult Onchocercavolwlus
after six 2-weekly dosesof
ivermectin. Bull Wld Hlth Org 199l;69:163-8.9.
DukeBOL,
Zea FloresG,
CastroJ,
CuppEW, MufrozB.
Effectsof
three-month dosesof
ivermectin on adult Onchocerca volvulus. Am J Trop Med
Hyg
1992;46:189-94.10. Chavasse
DC,
PostRI, Lemoh PA,
WhitworthJAG. The effect of
repeated dosesof
ivermectinon adult female
Onchocercavolvulus in
SierraLeone. Trop Med
Parasitolt4
1992;43:256-262
11. Chavasse
DC,
PostRI,
DaviesJB,
WhitworthJAG.
Absenceof
spermfrom the
seminal receptacleof
female Onchocerca volvulus following multiple dosesof
ivermectin. Trop Med Parasitol 1993;44:155-8.12.
Kllger
S, Whitworth JAG, PostRI,
ChavasseDC,
DownhamMD.
How long do the effectsof
ivermectin on adult Onchocerca volvulus persist. Trop Med Parsitol 1993;44:305-10.13. Remme
J, Baker RHA, De Sole G, et al. A
communitytrial of
ivermectinin
theonchocerciasis focus
of
Asubende, Ghana.I.
Effecton
the microfilarial reservoir and the transmissionof
Onchocerca volvulus. Trop Med Parasitol 1989;40:367-74.14.
World Health
OrganizationExpert
Committeeon
onchocerciasiscontrol. First
report Technical Report Series, in press.15. De Sole G, Awadzi
K,
Remme J, et al.A
community trialof
ivermectin in the onchocerciasis focus of Asubende, Ghana.II.
Adverse reactions. Trop Med Parasitol 1989;40:375-82.16. Dadzie
KY,
RemmeJ, De
SoleG.
Changesin
ocular onchocerciasis aftertwo
roundsof
community-based ivermectin treatment in a holo-endemic onchocerciasis focus. Trans Roy Soc Trop Med
Hyg
1991;85:267-71.17.
Alley
ES, Plaisier AP, BoatinBA,
DadzieKY,
Remmel,
ZerboG,
SambaEM.
The impactof five
yearsof
annual ivermectin treatment on skin microfilarial loadsin
the onchocerciasis focus of Asubende, Ghana. Trans Roy Soc Trop Med Hyg 1994; in press18. Remme J, Ba O, Dadzie
KY,
KaramM. A
force-of-infection model for onchocerciasis and its applicationin
the epidemiological evaluationof
the Onchocerciasis Control Programme in the Volta River basin area. Bull Wld Hlth Org 1986;64:667-81.19. Plaisier
AP, Van
OortmarssenGJ,
HabbemaJDF,
RemmeJ, Alley ES.
ONCHOSIM: amodel and computer simulation program
for
the transmission and controlof
onchocerciasis.Comp Meth Prog Biomed 199O;31:43-56.
20.
HabbemaJDF, Alley
ES, PlaisierAP,
Van Oortmarssen GJ, RemmeJHF.
Epidemiological modellingfor
onchocerciasis control. Parasitol Today 1992;8:99-103.21.
HabbemaJDF, Van
Oortmarssen GJ, PlaisierAP.
The ONCHOSIM model andits
use in decision supportfor
river blindness control.In:
Epidemic models: their structure and relation to data. Cambridge University Press; in press.22.
PlaisierAP,
Van Oortrnarssen GJ, RemmeJ,
HabbemaJDF. The
reproductive lifespanof
Onchocerca
volwlus
in West African savanna. Acta Trop l99L;48:271-84.23. Nelder JA, Mead R. A simplex
methodof function minimization.
Computer Journal 1965;7:308-12.24. De
SoleG,
RemmeJ,
AwadziK, et al.
Adverse reactionsafter
large-scale treatmentof
onchocerciasis
with
ivermectin: combined results from eight communitytrials. Bull Wld
Hlth Org 1989;67:707-19.25. Duke BOL. The
effectsof
drugson
Onchocercavolvulus I.
Methodsof
assessment, population dynamics of the parasite and the effects of diethylcarbamazine.Bull Wld Hlth
Org1968;39:137-46.
26.
AlbiezEI,
Walter G, KaiserA,
Ranque P, Newland HS, WhiteAT,
GreeneBM,
Taylor HR, BtittnerDW.
Histological examinationof
onchocercomata after therapywith
ivermectin. Trop Med Parasit 1988;39:93-9.27
.
AwadziK,
DadzieKY,
Schulz-KeyH,
GillesHM,
FulfordN, Aziz MA.
The chemotherapyof
onchocerciasisXI. A
double-blind comparative snrdyof
ivermectin, diethylcarbamazine,and placebo in human
onchocerciasisin Northern Ghana. Ann Trop Med
Parasit 1,986;80:43342.28. Taylor HR,
MurphyRP,
NewlandHS, White AT, D'Anna SA,
KeyvanLarijani E,
AzizMA.
CuppEW,
GreeneBM.
Comparisonof
the treatmentof
ocular onchocerciasis with ivermectin and diethylcarbamazine. Arch Opthalmol I 986; 1 04:863-7 0.16
29.
Diallo S,Aziz MA,
LariviDreM,
Diallo JS, Diop-MarI, N'Dir O,
Badiane S, PyD,
Schulz-Key H,
GaxotteP,
VictoriusA. A
double-blind comparisonof the
effrcacy and safetyof
ivermectin and diethylcarbamazine
in
a placebo controlled studyof
Senegalese patients with onchocerciasis. Trans Roy Soc Trop MedHyg
1986;80:921-3430.
GreeneBM,
WhiteAT,
Newland HS, Keyvan-LarijaniE,
DukulyF,
GallinMY, Aziz MA, Williams PN, Taylor HR.
Single dose therapywith
ivermectinfor
onchocerciasis. Trans Assoc Amer Phys 1987;C:131-8.31.
Schulz-KeyH,
SoboslayPT,
HoffrnannWH.
Ivermectin-facilitated immunity. Parasitology Today 19921'8:152-53.32.
Schulz-KeyH,
KaramM.
Periodic reproductionof
Onchocerca volvulus. Parasitology Today 1986;2:2844.33. Taylor HR, Trpis M,
CuppEW,
BrotmanB,
NewlandHS,
SoboslayPT,
Greene BM Ivermectin prophylaxis against experimental Onchocercavolwlus
infectionin
chimpanzees Am J Trop MedHyg
1988;39:86-90.34. Remme
J, De
SoleG,
DadzieKY,
Alley ES, BakerRHA,
Habbema JDF, PlaisierAP,
Van OortmarssenGJ,
SambaEM.
Large scale ivermectin distributionand its
epidemiological consequences. Acta Leiden 1990;59: 177-91.Table 1: Estimates
of
parameter valuesfor the
different hypotheses*on the working of
ivermectin on the adult worm.
Parameter
Hypothesis
I
(only transient)
IIa
(also irreversible) IIb
productivity reduction total fecundity loss recovery period (months) effect-variability
(coeffi cient of variation) mfJifespan (months)
goodness-of-fit (P-value
ty'l)
357" (26%40V")
10.4 (7-16) 0.s4 (0.4-0.7)
28Vo Q2Vo-35%\
10.7 0.52 19.0
0.87
l4
0.0013 [39]
e (4-12) 10
0.68 [11.9] 0.42
[ls.s]
Values in brackets represent the 95% confidence interval for these estimates.
A '-'
means that this parameter is not considered and thus has the value 0.* In
HypothesisI
the drug has only transient effects.In
HypothesisIIa
treatment resultsin
an irreversible productivity reductionin
each parasite. Under HypothesisIIb,
treatment causes a certain fraction of the parasites to loose fecundity totally.18
Fig. 1
Frequency distributionsof
microfilarial countsfor
the pre-treatment survey and the post- treatment surveysof
cohorts 1 (a) and2 O).
Dark bars represent the observations. The white bars denote the predictionsfor the
assumption that treatment causesa
permanent productivity reductionof
357o (HypothesisIIa
of Table 1). The numbers at the horizontal il(es represent the lower boundariesof
skin-snip count classes. The numbers in the graphs ofFig. la
(1 to 8) correspond to those in Fig. 2.pre-treatmenl cohort 1
o 2 4 0 i6 32 64 124256 No. ol ml per skin snip
prs-lroatmenl cohorl 2
4-5 months alter treatment
1 1-12 monlhs alter treatmenl
o o.5 2 I a
tr€atment round
round 1
round 2
round 3
round 4/5 2
Eo
sJ
oo a.
!
=o
so 50 40 30 20 'to o
50 40 30 20 10 o
40 30 20 10 o
40 30 20 10 o
o O5 2 .t 6 16 32 64124 o o.5 2 4 A 16 32 6112A
o o.5 2 I a
o o5 2
3
16 32 64 124
5
8
4
16 32 64 128
6
co
!J
oo
o E
=o o
s
40 30 20 10
o o.5 2 4 A 16 32 61129 o o.5 2 4 A 18 32 6412e
o 05 2 I a 16 32 64 126 7
50 40 30 20 10 o 50 40 30
10 o 2
No. ol ml per skin snip
co gJ eo e
!f o o
a o 16 32 64 12a256 24 months
arter lreatment
t9
.g
Eqc
=
anL
o
o- E c;
c c
(I,o
=
Fig.
2
Observed (dots) and predicted (ines) geometric mean skin-snip countsof
study cohort 1.The solid lines represent predictions based on the hypothesis
of
a productivity reductionof
35Vo
(Hyp.IIa). The
dashedline is
basedon the
hypothesisof no
irreversible effects(Hyp.I).
The peaksin
the predicted trends (excluding the pre-treatment survey) have been connectedwith
an exponential curve$=a.e-b'*).
The timesof
thefive
treatments are in- dicated below the horizontal axis100 95
30
20
10
0
\
'88T1
\
'89\
'90\
'91\ '92
T2
T3
T4Calendar year
T5
8
20
5
Part II
Eradication of onchocerciasis infection by vector control and annual ivermectin treatment.
Introduction
The objective
of
the Onchocerciasis Control Programme in WestAfrica
(OCP) isto
eliminate onchocerciasis as a public health problem in the area covered by the progralnme and to ensure that no recrudescenceof
the infection and the diseasewill
occur.Initially,
the only measureof
control was aerial larvicidingof
theriver
basins, the breeding placesof
Simulium damnosum ssp., in orderto
eradicate the vector and block the transmissionof
Onchocercavolvulus. Since the vector control operations are extremely costly, an important question is after how many years they couldbe
ceasedwithout
running unacceptablerisks of fatal
transmission levels whenthe
blackflies return.In
a previous study, we have applied the ONCHOSIM simulation modelto
assess the riskof
recrudescence after several yearsof
successful larviciding. Taking into account the differencesin
pre-control endemicityof the
disease and severalrisk
factors favouring recrudescence, we concluded that 14 years should be suffrcientto
reduce the riskto
less than l Vo evenin
the most afflicted areas @laisieret al.,
L99la).This
has since been adopted :rsa
general guideline. In developingthis
guideline,it
was assumedthat,
giventhe limited
treatment coverage and the presumed limited impact on the adult parasites, the microfilaricidal drug ivermectin-
registred in 1987-
would not play a significant rolein
controlling the transmission and that eradicationof
the parasite should be accomplishedby
vector control alone.In
large partsof
the original OCP area (where larviciding started between 1975 and 1977) OCP has discontinued large scale larviciding and replacedit by a
programmeof
entomological and epidemiological surveillancein
order to detect unforflrnate occurrences of recrudescenceof
infection.For two reasons, the initially adopted guideline has to be reconsidered
for
the extension arqmof
OCP where vector control started later aswell
asthe
areas wherevector
control was less sucessful during thefirst
years. The first reason is that financial resources become more and more limited, pressing OCP to speed up the processof
devolution, i.e. the transferof
responsibilities to the participating countries. The second - forfirnate-
reason is the growing evidence that ivermectinis
not onlya very
effective microfilaricide,but
also has considerableeffecs on
theviability of
female worms @uke
et al.,
1992; Kager etal.,
1993).In
a recent model-based analysisof
datafrom a
communitytrial of
annual treatmentsin
the Asubende area (Ghana)we
have found that,22
\l
following each treatment, female worms are considerably delayed
in
their mf-production and that the ultimate fecunditylevel
stabilizes around 70%of
pre-treatment @laisieret al.,
submitted).Such strong effects
put a
newlight on the
potentialof
community treatmentfor
controlling transmission and, hence,for
shortening the required duration of vector control.This potential
of
the drug is the subjectof
the present paper. Wewill
use the ONCHOSIM simulation modelto
investigate the extent to which annual ivermectin treatment can leadto
earlier cessationof
vector control. Wewill
show how this depends on the attainable treatment coverage, the pre-control endemicity in the area, alternative assumptions on the effectof
ivermectin on adult worms, and the timing of the start of ivermectin treatment.Methods
Basic assumptions
Vector control operations are assumed
to
be 100% effective,i.e. to
reduce the biting rate to zero. Flies are assumed to return immediately after cessationof
larviciding, giving rise to a biting rate equalto
the pre-control level.With
respectto
the effectof
ivermectin treatment we use the following assumptions: following treatment (1)all
microfilariae are eliminated,Q)
female worms recoverfrom total
lossof
fecundityduring,l0.to..11
months and (3) reacha
new stable mf- production levelrvhich is
permanently 35Volower
than before tr.eatlUgg! (Plaisieret al.,
sub- mitted). Both the period neededto
recover and the permanent impactof
the drugvary
between persons and between treatments (var. coeff.a;|!L
An alternative assumption we test is that the irreversible reductionof
fecundity is not 35% but only 25Vo per treatment. This lower percentage can be justifiedfrom
the confidence intervalfor
the estimateof
this parameter. The coverageof
treatment (Vo
of
persons getting the drug)is
oneof
the variablesin
the present analysis. This coverageis
not equivalentto
the probabilityto
be treatedfor
each individualin
the population:depending on age and sex and some persons have a higher chance, others a lower.
For
example, childrenbelow 5 are not
treated. Womenin the fertile
ages havea lower probability
since pregnancy and lactation are exclusion criteria. Furthermore, persons candiffer
considerably intheir
compliancewith
treatment. The age- and sex-specific variationin
treatment coverage has been taken from the resultsof
the Asubende-trial (Alley etal.,
in press).In
the Annex we explainhow, given a
mean population-coverage,for
each individualthe probability to be
treated is calculated.The
assumptionshave been
incorporatedinto the
stochastic microsimulation model ONCHOSIMwhich
has subsequently been usedto
simulatethe
control strategies.A
complete description of the model and parameter quantification is provided in Habbema etal.
(in press). We simulate human populationsof
around 300 persons (natural growthof
the populationis
compen- satedby
migration) which showa
pre-control endemicity level similarto
the villages Tiercoura and Folonzo (both Burkina Faso)with a CMFLI of 70
and30 mf per
skin-snip respectively@laisier et
al.,
1991b). Observations of pre-controlfly
biting rates are lackingfor
this area. Using observationsin
the Pru-river, closeto
the highly endemic Asubende regionin
Ghana (Remme etd.,
1989) we estimate the Annual Biting Ratewithin
Tiercouraat27,
(for adult men) andwithin
Folonzoat
I The maximum$osrre to
fly-bitesis, in
both villages, reachedat the
ageof 15
years. Womenare, on
average, 30% less exposed thanmen.
The variation coefficientof
bites/personwithin
a specific age and sex groupis
estimated at 0.39 for Tiercoura and 0.54for
Folonzo. Since we have previously found that the (age- and sex-indepen- dent) exposure heterogeneity is an important risk factorfor
recrudescence, wewill
also simulate a'Tiercoura-like' village (called
Tiercoura*)
with an exposure variation coefficient of 0.58.Simulation of control strategies
A
control strategyis
completely describedby the
numberof
yearsof
(l00Vo successful) vector control(v), the
numberof
annual ivermectin treatments(i),
the treatment coverage (Vo treated, c) assuming that thiswill
be constantfor
the whole period, and the time-lag between the startof
vector control and the startof
ivermectin treatment(d),
assuming that treatment always starts later. We have tested many combinationsof y
(range:0 to
15 years),i
(range: 0-30), c(range: 35-75%), and,
d
(0,8,
and 16 years). The resultof
each simulated strategy is summarized as 'recrudescence'(l) or 'no
recrudescence'(0).
Recrudescenceis
defined asthe
occurenceof
increasing mf-loads after total withdrawal vector control, ivermectin or both
Statistical analysis of simulatton results
The purpose
of
the analysis isto
estimate the recrudescence risk as a functionof
the strategy- variables. Previously @laisier etal., l99la)
we have achieved thisby
performing many simula- tionsfor
one particular strategy (numberof
years vector control) and countingthe
numberof
simulations resulting
in
recrudescence.By
choosing an appropriate rangeof
durationsof
vector control, those durations where therisk
approaches zero (e.g. 0.01) could be identified. However,I
Community Microfilarial Load; i.e. the geometric mean mf-load in adults 24-tr.9^?lT:€I!i
logistic regress t----\l
c, and d)[
+
with four
strategy-variables involvedthis
approachis not
suitable: the rangeof
vector control- durations wherethe risk
changesfrom - 1 to - 0 is
differentfor
each ivermectin treatment strategy.In
the present analysis we have applied another method which comprises the following two steps: (1)for
a large numberof
ivermectin treatment strategies, an iterative procedure is usedto flnd
the durationof
vector control where therisk
changes mostrapidly; Q)
the techniqueof
ion is applied to estimate the riskfor
each ftrategy (which is a combinationof v, i,
An
exampleof
the iterative procedure is given inFig.
1.In
this example we use the parame-ters
representativefor
Tiercoura.The
ivermectin strategy consistsof 10
yearsof
treatment, startingin the
same ye:tr as vector control and covering 65Voof the
population.The
initial durationof
vector controlis
10 years which resultsin
eradicationof the
parasite.The
initial iteration step is 4 years and hence the next attempt is 6 years vector control. This is (far too) short and results in recrudescence. Now, the iteration step is halved and 8 years vector control is tested.This is still too short and again
try
10 years is tested, again resultingin
eradication. Since we are especially interestedin low
recrudescence risks(e.g.
1%) we take one-thirdof
the iteration step whena "1"
(recrudescence)is
followedby a "0"
(eradication). Hence, the new tested value is9.33 years (nine years and
four
months). Here recrudescence is encountered and we subsequently test9.67
years (halvingthe
interval).A total of
10 iteration stepsis
donein this way.
The smallest interval allowedis
one month. Note that 9.67 yearsis
tested three times, successively resultingin "0", "1",
and"0".
Apparently, around this value therisk is
changing most rapidly.All
iteration-stepsfor all
tested combinationsare
accumulatedin a
data-filefor use in
the regression analysis. Tentative regression results are usedto
repeatthe
procedure using other starting values and other (as a rule smaller) initial iteration intervals.We have used SPSS
to
calculate logistic regression equationsfor
therisk of
recrudescence(ref.).
Separate equations are calculatedfor
the different area-conditions: Tiercoura,Tiercoura*,
and Folonzo, aswell
asfor
the different valuesfor
the delay between the startof
the two control methods(d=0, 8, or
16yqrs;
onlyfor
Tiercoura) and the effectivenessof
ivermectin (35% vs.25% permanent effect; only
for
Tiercoura). The independent variablesof
the regression equations are the durationof
vector control (v), the numberof
annual treatments(i),
the treatment coverage(c),
andlinear,
quadratic and cubic combinationsof v, i
andc.
Since neitheri nor c
has ameaning
on
its own, they are always combinedin
the regression equations.In
the procedure for estimating regression coeffrcients we have used a combinationof
backward and forward selection.Each procedure
is
startedwith
the strategy variables(v
andi. c)
andtheir
linear and quadratic combinations (interaction terms,e.g. i2.c.l).
Coefficientsthat are not
significant (P=0.05,according
to
the Wald-statistic;ref.)
are eliminated while the significant coefficienSfor
the cubic terms(e.g. r/;
anA interaction termswith a
cubic component are included(e.g. i'C'rl;. nn
example
of
estimated regression coefficients is provided in Table 1. Form the resulting regression equations the riskof
recrudescencefor
a givenv, i,
and c can be easily calculated. However, they aretoo
complicatedto
calculate the durationof
vector control which correspondswith a
given recrudescencerisk
(giventhe
valuesof i
andc). This is
done iteratively, usingthe
routine available in the Borland Quattro-Pro spreadsheet progr:Im.Results
Fig. 2
shows the trendin
the recrudescencerisk for
several combinationsof
the numberof
annual ivermectin treatments and the duration
of
vector controlin
aTiercouralike
village.It
isassumed that
the
average treatment coverageis
65% and that both control methods startat
the same moment.The
linesin
the figure represent'iso-risk'
lines, connecting those strategies that resultin
equalrisks
(0.01,0.1, 0.5,
and 0.99). Below (less vector control) andleft of
the line (less annual treatments) therisk
is higher than indicated, otherwiseit
is lower.In
the absenceof
ivermectin treatment (points on the Y-axis) approximately 13 years
of
vector control are requiredto
reducethe
recrudescencerisk to
1%.This
duration reducesto
11 years whenten
yearsof
annual ivermectin treatment
is
added. With the same ivermectin strategy therisk
increasesto
0.1 when vector control is stopped already after 10 years. Recrudescence is certain(risk >
0.99) with only less than6.5
yearsof
vector control. The iso-risk lines diverge considerably as the numberof
treatments increase.With
16 treatments risksof
less than 0.99 are achievedin
the absenceof
vector control, while
still9.3
years of vector control are needed to reduce the risk to 0.01.The effect
of
alternative coveragelevels is shownin Fig.
3a.In
this figure only the 0.01 iso-risk
lines are shown. Especiallyfor
longer periodsof
treatment the impactof
higheror
lower coveragesis
considerably. When treatment is continuedfor
15yqlrs,
then each 10% decrease in coverage correspondswith
approximately9
months more vector controlto
maintaina
recrudes- cencerisk of 0.01; with l0
yearsof
treatment the gainor
lossis only 2-4
months.Fig.
3bdemonstrates
that the
effectivenessof
control strategiesis highly
dependenton the
pre-control endemicityof
the area.With
a treatment coverageof
65Vo,in
a villagelike
Folonzo, 20 yearsof
annual treatment are sufficient to eradicate the parasite without the help
of
vector control. Shorter periodsof
treatment(10 - 15
years)allow
considerable reductionsin the
durationof
vector control.26
The predicted implications
of a
delay between the startof
vector control andthe
startof
annual treatment is shown
in
Fig.4. The results applyto
Tiercoura and a coverage levelof
65%.Up to 20
yearsof
annual treatment,with a
delayof 8
years less vector controlis
required to reduce the risk to 0.01 than with an equal start. For example, with a period of ten yearsof
annual treatment,[0
yearsof
vector control are sufficient; this was almost 11 years when both methods started synchronous. For periodsof
20 yearsof
treatmentor
more, a delayof
8 years is slightly less favourablein
termsof
reducing larviciding. Such long durations correspondwith
less than 8 yearsof
larviciding, and this implias a short period without any meansof
control (treatment starts after cessationof
larviciding). Thisis
the major problemwith
a long delaylike
16 years. Now treatment always starts when larviciding has stopped, which implies thatin
many cases ivermectin treatment is usedfor
the control instead of the prevention of recrudescence.A
summaryof the
resultsis
providedin
Table2. For
eachof the
circumstances and four different coverage levels, the required durationof
vector control which reduces the recrudescence riskto l% ue
shownfor four
different durationsof
annual treatment(5, 10,
15, and 20 years).The numbers in the table are obtained from the regression equations.
Table 1.
Ivermectin treatment
ixc ix*
ixc
i2xc i3xc
Estimates
of the
coefficientsfor the
variables(row x
column)of the
logistic regression equationfor
the recrudescence risk Tiercoura(d = 0).
Priorto
estima-tion, the variables are transformed as follows:
i =
no.of
ivermectin treatments+
10;v =
no.of
years vector control+
10;c =
coverage (Vo)+
50. The resulting equationhas the form: tn(r/1-r)=fi.6 + 22.7.v - 28.3.i + 5.71.i.c
-25.1.i. c,v +
..etc., withr
being the recrudescence risk.Vector control
v
f
22.71
i
n.s 12.g*
5.71
-t2.5
3.86 n.s.
n.s.
-25.1$
4.47 n.s.
4.86 n.s.
-28.3 13.6
n.s -1.79 -0.576
n.s.+
n.s n.s n.s n.s n.s
* the constant of the equation
t
coeffrcientfor
covariable v+ not significant
$ coefficient
for
covariablei.
c. v (interaction term)28