• View in gallery

    Selection of the P. falciparum dhfr triple mutation by sulfadoxine-pyrimethamine (SP) monotherapy (dark green) and by SP/artesunate (SP/AS) employed from inception (light green) and as complete replacement therapy for SP when the triple mutant frequency reaches 5% (brown) or 20% (blue). For SP monotherapy, the frequency of the triple mutation reaches a maximum and then decreases as triples are replaced by dhfr quadruple mutants. Each case assumes 100% coverage with the respective treatment. This figure appears in color at www.ajtmh.org.

  • View in gallery

    Selection of the P. falciparum dhfr triple mutation (dhfr108N+51I+59R) by sulfadoxine-pyrimethamine (SP) monotherapy (dark green) and by SP+amodiaquine (SP/AQ) from inception (light green) and as complete replacement therapy for SP when the triple mutant frequency is 5% (brown) or 20% (blue). Each case assumes 100% coverage with the respective treatment. Other model assumptions: see the text. This figure appears in color at www.ajtmh.org.

  • View in gallery

    Selection of the P. falciparum dhfr quadruple mutation (dhfr108N+51I+59R+164L) by sulfadoxine-pyrimethamine (SP) monotherapy (purple) and by SP/amodiaquine (SP/AQ) employed from inception under conditions of complete CT coverage (light green) and 50% CT/50% SP monotherapy (brown). Other model assumptions: see text. This figure appears in color at www.ajtmh.org.

  • View in gallery

    Selection of the P. falciparum dhfr triple mutation (dhfr108N+51I+59R) by Lapdap™ monotherapy (dark green) and Lapdap™-artesunate (CDA), under conditions of complete ACT coverage (light green), and 50% ACT/50% Lapdap™ monotherapy (brown). Starting time point is a triple mutant frequency of 0.2. In the case of Lapdap™ monotherapy the frequency of the triple mutation reaches a maximum and then decreases as triples are replaced by dhfr quadruple mutants. Other model assumptions: see text. This figure appears in color at www.ajtmh.org.

  • View in gallery

    Selection of the P. falciparum dhfr triple mutation (dhfr108N+51I+59R) by Lapdap™ monotherapy (dark green) and Lapdap™-amodiaquine, under conditions of complete CT coverage (light green), and 50% CT/50% Lapdap™ monotherapy (brown). Starting time point is a triple mutant frequency of 0.2. The frequency of the triples eventually starts to fall in the case of Lapdap™ monotherapy because they are being replaced by dhfr quadruple mutants. Other model assumptions: see text. This figure appears in color at www.ajtmh.org.

  • View in gallery

    Selection of the P. falciparum dhfr quadruple mutation (dhfr108N+51I+59R+164L) by Lapdap™ monotherapy (purple) and Lapdap™-artesunate (CDA) from inception (light green), and after periods of Lapdap monotherapy which have increased the dhfr quadruple mutation frequency from 0.0001 to 0.01 (dark green) and to 0.2 (brown). Other model assumptions: see text. This figure appears in color at www.ajtmh.org.

  • 1

    Hastings IM, D’Alessandro U, 2000. Modelling a predictable disaster: the rise and spread of drug-resistant malaria. Parasitol Today 16 :340–347.

    • Search Google Scholar
    • Export Citation
  • 2

    White NJ, Nosten F, Looareesuwan S, Watkins WM, Marsh K, Snow RW, Kokwaro G, Ouma J, Hien TT, Molyneux ME, Taylor TE, Newbold CI, Ruebush TK 2nd, Danis M, Greenwood BM, Anderson RM, Olliaro P, 1999. Averting a malaria disaster. Lancet 353 :1965–1967.

    • Search Google Scholar
    • Export Citation
  • 3

    Hastings IM, Watkins WM, White NJ, 2002. The evolution of drug-resistant malaria: the role of drug elimination half-life. Philos Trans R Soc Lond B Biol Sci 357 :505–519.

    • Search Google Scholar
    • Export Citation
  • 4

    Peters W, 1985. The problem of drug resistance in malaria. Parasitology 90 :705–715.

  • 5

    Nosten F, Brasseur P, 2002. Combination therapy for malaria: the way forward? Curr Opin Drugs 62 :1315–1329.

  • 6

    World Health Organization, 2001. Antimalarial Drug Combination Therapy: Report of a WHO Technical Consultation. Geneva: World Health Organization.

  • 7

    EANMAT, 2001. Monitoring antimalarial drug resistance within National Malaria Control Programmes: the EANMAT experience. Trop Med Int Health 6 :891–898.

    • Search Google Scholar
    • Export Citation
  • 8

    Shretta R, Omumbo J, Rapuoda B, Snow RW, 2000. Using evidence to change antimalarial drug policy in Kenya. Trop Med Int Health 5 :755–764.

    • Search Google Scholar
    • Export Citation
  • 9

    Goodman CA, Coleman PG, Mills AJ, 2001. Changing the first line drug for malaria treatment–cost-effectiveness analysis with highly uncertain inter-temporal trade-offs. Health Econ 10 :731–749.

    • Search Google Scholar
    • Export Citation
  • 10

    Shretta R, Walt G, Brugha R, Snow R, 2001. A political analysis of corporate drug donations: the example of Malarone in Kenya. Health Policy Plann 16 :161–170.

    • Search Google Scholar
    • Export Citation
  • 11

    Hyde JE, 2002. Mechanisms of resistance of Plasmodium falciparum to antimalarial drugs. Microbes Infect 4 :165–174.

  • 12

    International Artemisinin Study Group, 2004. Artesunate combinations for treatment of malaria: meta-analysis. Lancet 363 :9–17.

  • 13

    Trape JF, 2001. The public health impact of chloroquine resistance in Africa. Am J Trop Med Hyg 64 :12–17.

  • 14

    Watkins WM, Mosobo M, 1993. Treatment of Plasmodium falciparum malaria with pyrimethamine-sulfadoxine: selective pressure for resistance is a function of long elimination half-life. Trans R Soc Trop Med Hyg 87 :75–78.

    • Search Google Scholar
    • Export Citation
  • 15

    Sibley CH, Hyde JE, Sims PFG, Plowe CV, Kublin JG, Mberu EK, Cowman AF, Winstanley PA, Watkins WM, Nzila AM, 2001. Pyrimethamine/sulfadoxine resistance in Plasmodium falciparum: What next? Trends Parasitol 17 :582–588.

    • Search Google Scholar
    • Export Citation
  • 16

    Amukoye E, Winstanley PA, Watkins WM, Snow RW, Hatcher J, Mosobo M, Ngumbao E, Lowe B, Ton M, Minyiri G, Marsh K, 1997. Chlorproguanil-dapsone: Effective treatment for uncomplicated falciparum malaria. Antimicrob. Agents Chemother. 41 :2261–2264.

    • Search Google Scholar
    • Export Citation
  • 17

    Nzila AM, Nduati E, Mberu EK, Sibley CH, Monks SA, Winstanley PA, Watkins WM, 2000. Molecular evidence of greater selective pressure for drug resistance exerted by the long-acting antifolate pyrimethamine/sulfadoxine compared with the shorter-acting chlorproguanil/dapsone on Kenyan Plasmodium falciparum. J Infect Dis 181 :2023–2028.

    • Search Google Scholar
    • Export Citation
  • 18

    Talisuna AO, Langi P, Mutabingwa TK, Watkins WM, van Marck E, Egwang TG, D’Alessandro U, 2004. Population-based validation of DHFR gene mutations for the prediction of sulfadoxine-pyrimethamine resistance in Uganda. Trans R Soc Trop Med Hyg 97 :338–342.

    • Search Google Scholar
    • Export Citation
  • 19

    Mutabingwa T, Nzila A, Mberu E, Nduati E, Winstanley P, Hills E, Watkins W, 2001. Chlorproguanil-dapsone for treatment of drug-resistant falciparum malaria in Tanzania. Lancet 358 :1218–1223.

    • Search Google Scholar
    • Export Citation
  • 20

    Watkins WM, Mberu EK, Winstanley PA, Plowe CV, 1997. The efficacy of antifolate antimalarial combinations in Africa: a predictive model based on pharmacodynamic and pharmaco-kinetic analyses. Parasitol Today 13 :459–464.

    • Search Google Scholar
    • Export Citation
  • 21

    EANMAT, 2003. The efficacy of antimalarial monotherapies sulphadoxine-pyrimethamine and amodiaquine in east Africa: implications for sub-regional policy. Trop Med Int Health 8 :860–867.

    • Search Google Scholar
    • Export Citation
  • 22

    Spencer HC, Oloo AJ, Watkins WW, Sixsmith DG, Churchill FC, Koech DK, 1984. Amodiaquine more effective than chloroquine against Plasmodium falciparum malaria on Kenya coast. Lancet 1 :956–957.

    • Search Google Scholar
    • Export Citation
  • 23

    World Health Organization, 2001. The Use of Antimalarial Drugs. Report of a WHO Informal Consultation. Geneva: World Health Organization.

  • 24

    Paget-McNicol O, Saul A, 2001. Mutation rates in the dihydrofolate reductase gene of Plasmodium falciparum. Parasitology 122 :497–505.

  • 25

    Nair S, Williams JT, Brockman A, Paiphun L, Mayxay M, Newton PN, Guthmann JP, Smithuis FM, Hien TT, White NJ, Nosten F, Anderson TJ, 2003. A selective sweep driven by pyrimethamine treatment in southeast Asian malaria parasites. Mol Biol Evol 20 :1526–1536.

    • Search Google Scholar
    • Export Citation
  • 26

    Roper C, Pearce R, Bredenkamp B, Gumede J, Drakeley C, Mosha F, Chandramohan D, Sharp B, 2003. Antifolate anti-malarial resistance in southeast Africa: a population-based analysis. Lancet 361 :1174–1181.

    • Search Google Scholar
    • Export Citation
  • 27

    Cortese JF, Caraballo A, Contreras CE, Plowe CV, 2002. Origin and dissemination of Plasmodium falciparum drug-resistance mutations in South America. J Infect Dis 186 :999–1006.

    • Search Google Scholar
    • Export Citation
  • 28

    Gamage SA, Tepsiri N, Wilairat P, Wojcik SJ, Figgitt DP, Ralph RK, Denny WA, 1994. Synthesis and in vitro evaluation of 9-anilino-3,6-diaminoacridines active against a multidrug-resistant strain of the malaria parasite Plasmodium falciparum. J Med Chem 37 :1486–1494.

    • Search Google Scholar
    • Export Citation
  • 29

    Sirawaraporn W, Sathitkul T, Sirawaraporn R, Yuthavong Y, Santi DV, 1997. Antifolate-resistant mutants of Plasmodium falciparum dihydrofolate reductase. Proc Natl Acad Sci U S A 94 :1124–1129.

    • Search Google Scholar
    • Export Citation
  • 30

    Daramola OO, Alonso PL, O’Dempsey TJ, Twumasi P, McArdle TF, Greenwood BM, 1991. Sensitivity of Plasmodium falciparum in The Gambia to cotrimoxazole. Trans R Soc Trop Med Hyg 85 :345–348.

    • Search Google Scholar
    • Export Citation
  • 31

    Bloland PB, Redd SC, Kazembe P, Tembenu R, Wirima JJ, Campbell CC, 1991. Cotrimoxazole for childhood febrile illness in malaria-endemic regions. Lancet 337 :518–520.

    • Search Google Scholar
    • Export Citation
  • 32

    Kilian AH, Jelinek T, Prislin I, Kabagambe G, Byamukama W, Mpigika G, Korte R, von Sonnenburg F, 1998. Resistance in vivo of Plasmodium falciparum to cotrimoxazole in western Uganda. Trans R Soc Trop Med Hyg 92 :197–200.

    • Search Google Scholar
    • Export Citation
  • 33

    White NJ, 1992. Antimalarial drug resistance: the pace quickens. J Antimicrob Chemother 30 :571–585.

  • 34

    Bloland PB, 2003. A contrarian view of malaria therapy policy in Africa. Am J Trop Med Hyg 68 :125–126.

  • 35

    Dorsey G, Njama D, Kamya MR, Cattamanchi A, Kyabayinze D, Staedke SG, Gasasira A, Rosenthal PJ, 2002. Sulfadoxine/pyrimethamine alone or with amodiaquine or artesunate for treatment of uncomplicated malaria: a longitudinal randomised trial. Lancet 360 :2031–2038.

    • Search Google Scholar
    • Export Citation
  • 36

    Rwagacondo CE, Niyitegeka F, Sarushi J, Karema C, Mugisha V, Dujardin JC, van Overmeir C, van den Ende J, D’Alessandro U, 2003. Efficacy of amodiaquine alone and combined with sulfadoxine-pyrimethamine and of sulfadoxine pyrimethamine combined with artesunate. Am J Trop Med Hyg 68 :743–747.

    • Search Google Scholar
    • Export Citation
  • 37

    Sweetman SC, ed., 2002. Martindale: The Complete Drug Reference. 33rd edition. London: The Pharmaceutical Press.

  • 38

    Krishna S, White NJ, 1996. Pharmacokinetics of quinine, chloroquine and amodiaquine. Clinical implications. Clin Pharmacokinet 30 :263–299.

    • Search Google Scholar
    • Export Citation
  • 39

    Winstanley P, Edwards G, Orme M, Breckenridge A, 1987. The disposition of amodiaquine in man after oral administration. Br J Clin Pharmacol 23 :1–7.

    • Search Google Scholar
    • Export Citation
  • 40

    Alloueche A, Bailey W, Barton S, Bwika J, Chimpeni P, Falade CO, Fehintola FA, Horton J, Jaffar S, Kanyok T, Kremsner PG, Kublin JG, Lang T, Missinou MA, Mkandala C, Oduola AMJ, Premji Z, Robertson L, Sowunmi A, Ward SA, Winstanley P, 2004. Comparison of chlorproguanil-dapsone with sulfadoxine-pyrimethamine for the treatment of uncomplicated falciparum malaria in young African children: double-blind randomised controlled trial. Lancet 363 :1843–1848.

    • Search Google Scholar
    • Export Citation
  • 41

    Nzila AM, Mberu EK, Sulo J, Dayo H, Winstanley PA, Sibley CH, Watkins WM, 2000. Towards an understanding of the mechanism of pyrimethamine-sulfadoxine resistance in Plasmodium falciparum: genotyping of dihydrofolate reductase and dihydropteroate synthase of Kenyan parasites. Antimicrob Agents Chemother 44 :991–996.

    • Search Google Scholar
    • Export Citation
  • 42

    Wilairatana P, Kyle DE, Looareesuwan S, Chinwongprom K, Amradee S, White NJ, Watkins WM, 1997. Poor efficacy of antimalarial biguanide-dapsone combinations in the treatment of acute, uncomplicated falciparum malaria in Thailand. Ann Trop Med Parasitol 91 :125–132.

    • Search Google Scholar
    • Export Citation
  • 43

    White NJ, 1996. Can amodiaquine be resurrected? Lancet 348 :1184–1185.

  • 44

    Adjuik M, Agnamey P, Babiker A, Borrmann S, Brasseur P, Cisse M, Cobelens F, Diallo S, Faucher JF, Garner P, Gikunda S, Kremsner PG, Krishna S, Lell B, Loolpapit M, Matsiegui PB, Missinou MA, Mwanza J, Ntoumi F, Olliaro P, Osimbo P, Rezbach P, Some E, Taylor WR, 2002. Amodiaquine-artesunate versus amodiaquine for uncomplicated Plasmodium falciparum malaria in African children: a randomised, multicentre trial. Lancet 359 :1365–1372.

    • Search Google Scholar
    • Export Citation
  • 45

    Curtis J, Maxwell CA, Msuya FH, Mkongewa S, Alloueche A, Warhurst DC, 2002. Mutations in dhfr in Plasmodium falciparum infections selected by chlorproguanil-dapsone treatment. J Infect Dis 186 :1861–1864.

    • Search Google Scholar
    • Export Citation
  • 46

    Sutherland CJ, Drakeley CJ, Obisike U, Coleman R, Jawara M, Targett GA, Milligan P, Pinder M, Walraven G, 2003. The addition of artesunate to chloroquine for treatment of Plasmodium falciparum malaria in Gambian children delays, but does not prevent treatment failure. Am J Trop Med Hyg 69 :19–25.

    • Search Google Scholar
    • Export Citation
  • 47

    Hastings IM, 1997. A model for the origins and spread of drug-resistant malaria. Parasitology 115 :133–141.

  • 48

    Dye C, Williams BG, 1997. Multigenic drug resistance among inbred malaria parasites. Proc R Soc Lond B Biol Sci 264 :61–67.

  • 49

    Anderson TJ, Haubold B, Williams JT, Estrada-Franco JG, Richardson L, Mollinedo R, Bockarie M, Mokili J, Mharakurwa S, French N, Whitworth J, Velez ID, Brockman AH, Nosten F, Ferreira MU, Day KP, 2000. Microsatellite markers reveal a spectrum of population structures in the malaria parasite Plasmodium falciparum. Mol Biol Evol 17 :1467–1482.

    • Search Google Scholar
    • Export Citation

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

THE SEARCH FOR EFFECTIVE AND SUSTAINABLE TREATMENTS FOR PLASMODIUM FALCIPARUM MALARIA IN AFRICA: A MODEL OF THE SELECTION OF RESISTANCE BY ANTIFOLATE DRUGS AND THEIR COMBINATIONS

View More View Less
  • 1 Department of Pharmacology and Therapeutics, University of Liverpool, Liverpool, United Kingdom; Department of Genome Sciences, University of Washington, Seattle, Washington; Liverpool School of Tropical Medicine, Liverpool, United Kingdom

The extensive data on the relationship between parasite genotype and susceptibility to antifolate drugs can now be coupled with pharmacokinetic information to allow construction of models of the selection and spread of antifolate-resistant Plasmodium falciparum. In this report, we have modeled the effect on resistance selection processes of combinations of antifolate antimalarial drugs with artesunate and with amodiaquine under a variety of conditions that can be defined by the user. The model is intended to assist policymakers in forecasting the useful therapeutic life (UTL) for a range of potential combination treatments. The model is especially designed for use by African malaria programs so that the interactions of key variables can be explored and appropriate combinations of drugs can be chosen for field testing. The model provides some important general conclusions: 1) for optimal extension of UTL, combination therapy must be deployed before either constituent drug is used as monotherapy; 2) even short periods of monotherapy can severely limit the usefulness of subsequent combination therapy; and 3) that adding a second drug to rescue an antifolate antimalarial that is overtly failing is an inappropriate and ultimately wasteful exercise.

INTRODUCTION

The goal of combination therapy (CT) is to delay the emergence and spread of drug resistance. The strategy is supported empirically by the success of CT in treating tuberculosis and human immunodeficiency virus infections, and by mathematical models.1 The rationale for CT is simple. If two drugs have independent mechanisms of action, mutations that confer resistance to each drug will only rarely co-exist in the same parasite. By this logic, drug combinations should both improve treatment cure rate, and delay the emergence of drug resistance.2 In a previous report,3 we have addressed two particular aspects of the selection of resistant parasites: the importance of the elimination half-lives of the drugs, and the effect of a mechanism of resistance that evolves by incremental stages rather than a single step. In our model for sulfadoxine-pyrimethamine (SP), resistance develops in two separate phases. In phase A, parasites with chemosensitivity intermediate between the wild-type and fully resistant genotype are selected by residual drug circulating in previously treated patients. These partly resistant parasites persist at low levels in patients, and spread through the population, replacing the more sensitive genotypes, even though the patients who carry them are still clinically responsive to SP treatment. Phase B commences once parasites that can escape maximum in vivo drug concentrations are selected. In phase B, the selection process accelerates rapidly, and clinical failures increase concomitantly.3

Twenty years ago, Peters demonstrated the potential importance of CT in malaria control in an animal model.4 The practical use of CT in malaria control was pioneered in southeast Asia, where mefloquine-artesunate has been intensively studied and successfully used. The advantages of this artemisinin combination therapy (ACT) have been defined: 1) parasite biomass is reduced rapidly by artesunate, 2) parasites that survive the initial artesunate action are subject to maximal mefloquine concentrations, and 3) there is reduced transmissibility of parasites that survive the double-drug action.5 Thus, ACT can present a powerful stratagem, with clear public health advantages. Drug combinations, especially ACT, have been advocated as the replacement for ineffective and usually monotherapeutic malaria treatments.6

African countries, with the bulk of malaria mortality and morbidity, are the least able to afford effective but expensive ACT. Although a fund to meet this need is now available,* only a handful of African countries have obtained the resources to secure a move from the old, ineffective but affordable treatments to ACT. At the same time, there is a move among African countries to regularly monitor their existing malaria treatments to make evidence-based changes in anti-malarial drug policy.7 Changing antimalarial drug policy is a highly complex issue.8 Cost is a prime consideration, and analysis of cost-effectiveness is clearly important, but the models provide a plethora of choices.9 In addition to domestic financial constraints, policy makers need to consider the agendas of donors and the international agencies,10 and this adds another level of complexity. Thus, considerable uncertainty surrounds the decision process, even when there is convincing scientific evidence that change is needed. Changing the recommended drug is a lengthy process and places demands on National Malaria Control Programs, which are already overburdened.8 These factors demonstrate a key requirement: any new treatment must remain effective for a time span long enough to justify the costs and disruption of the changeover. The epidemiology of drug-resistant malaria is a comparatively new field, but the genetics of parasite resistance and the pharmacologic basis of drug action are sufficiently established to permit simple modeling. The prime objective of this paper is to provide a tool to predict the useful therapeutic life (UTL) of current and potential treatments: one of the most important parameters when a change in drug policy is being considered.

The Global Fund for AIDS, Tuberculosis and Malaria (GFATM): www.globalfundatm.org.

The justification for CT is based on a simple probabilitistic prediction. There are two assumptions: the two drugs have independent mechanisms of action and they target products of unlinked genes. Then, when a mutation to resistance to drug A occurs at a frequency of 10−8 and to drug B at a frequency of 10−7, a parasite resistant to both drug A and drug B will only arise once in every 10−15 asexual divisions.2 This is obviously a very rare event. Furthermore, independent assortment and recombination during the obligate sexual phase of the Plasmodium lifecycle will break apart these rare double-resistant parasites. Both factors ensure that the drug combination AB should have a much longer UTL than either drug alone. However, the actual selection process depends critically upon the pharmacokinetic characteristics of both drugs. If the elimination half-lives differ considerably, the more slowly eliminated drug will at some stage lose the protection provided by the partner drug. Thus, an unprotected component of a CT treatment will exert selective pressure on new infections, exactly as it does when it is used alone. In addition, single mutations to full resistance are not common with antimalarial drugs. Resistance more often arises through a series of mutations or a primary mutation that provides an essential step, augmented by secondary mutations that progressively increase the level of resistance.11 The importance of these factors has been discussed previously.3 In this report, we incorporate them into models assessing drug combinations being considered as first-line malaria treatments in Africa. We demonstrate that the UTL is defined primarily by two factors: the intrinsic antiparasitic activity of each drug (as a varying continuum over time) and the elimination half-life of each component. We then discuss the ways in which the model output can be used in choosing appropriate drug combinations in more general terms.

It has been common practice to consider a change to CT only when failure of a currently used drug is imminent. The expectation is that the addition of a second drug will prolong the UTL of the first. In fact, addition of a second drug to either chloroquine or SP when these drugs were already losing effectiveness has been tried in several situations. For example, adding artesunate (AS) to SP did not greatly increase treatment effectiveness, when the failure rate for SP was already high.12 The most important output of our model predicts that combination therapy is only effective in prolonging UTL when the initial effectiveness of both drugs is high. The addition of a second drug to an already compromised drug fails to add significantly to the UTL.

The specific genetics of SP resistance.

Increasing chloroquine resistance has resulted in increased malaria mortality in Africa over the recent past, and this trend has demonstrated the need to change the recommended treatment.13 Sulfadoxine-pyrimethamine has primarily been seen as the obvious successor to chloroquine on the basis of cost, safety, and availability. At the outset, the long elimination half-life of SP was advantageous, since a single-dose cleared parasites, and prevented re-infection for a period of approximately 50 days, and that allowed patients to recover from the sequelae of malaria infection, particularly anemia. However, the long half-life of SP also had a detrimental effect: the rapid selection of resistance.14

The genetic basis of SP resistance is becoming clear. Pyrimethamine inhibits the key enzyme dihydrofolate reductase (DHFR), and sulfadoxine inhibits dihydropteroate synthase (DHPS). Resistance occurs primarily by sequential accumulation of mutations in the dhfr domain of the dhfrthymidylate synthase gene, with a secondary contribution from mutations in the dhps domain of the dihydrohydroxymethylpterin pyro-phosphokinase gene.15 Currently, the most common pyrimethamine-resistant allele of the dhfr gene in east Africa has three mutations, serine to asparagine at position 108 (S108N) accompanied by mutations from asparagine to isoleucine at amino acid 51 (N51I) and cysteine to arginine at amino acid 59 (C59R). The rate of SP clinical failure depends on many host factors as well as the parasite genotype, so the relationship between the dhfr triple mutant frequency and clinical response cannot be defined precisely. However, there are data to allow us to estimate the relationship. In Kilifi, Kenya, there was a very low SP parasitologic failure rate in the years 1993–1995; 0.7% of the patients treated were slide positive at day 7.16 At that time, the prevalence of the dhfr triple mutant was 30–40%.17 More recently, in Uganda, a 50–66% prevalence of this genotype was associated with an SP-total failure† of 10–15%.18 In Muheza, Tanzania, where the highest SP failure rates in Africa have been recorded, the triple mutant prevalence approaches 100%, and 60% of SP treatments failed to clear parasites by day 7. The total failure rate would have been significantly higher in Muheza with a longer follow-up.19 These three studies illustrate an important difficulty in comparing outcomes from different protocols. The studies used different definitions of drug treatment failure, persistence of parasites in a blood smear in two studies, but a combination of early and late treatment failure based on patient symptoms in the third. Whenever possible, we have used persistence of parasites as our metric, since that reflects the parasite genotype more closely than the patient’s clinical response. Despite this ambiguity, it is apparent that an increasing prevalence of the dhfr triple mutant allele is an effective marker of increasing parasitologic resistance, and that modest increases in its prevalence presage progressive increases in clinical treatment failure.

SP-total failure is defined as the sum of early and late treatment failures in the World Health Organization in vivo test.21

THE BASIC MODEL

The biologic inputs into the model, and their interpretation, size, and justification are discussed below. The resulting equations are relegated to the Appendix 1 for clarity. The model considers only mutations in parasite dhfr. These are fundamental to the antifolate resistance mechanism, although parasite chemosensitivity is further modified by mutations in the parasite dhps gene.15

Assumptions.

Sulfadoxine-pyrimethamine alone.

Parasites that carry the wild type dhfr allele are always killed by a therapeutic dose of SP. Because of the long half-life of SP, parasites that carry the wild-type genotype cannot reinfect the human host, on average, until day 52 after SP treatment.3,14 We refer to this time period as the period of chemoprophylaxis (PoC). Reinfection refers to the reappearance of asexual parasites in the blood, following the hepatic cycle, which is unaffected by SP. Treatment starts on day 0.

Parasites that carry the dhfr 108N mutation are also killed by therapeutic drug concentrations of SP, but because of increased drug tolerance, parasites with this genotype can reinfect a human, on average, on day 12 after SP treatment.20

Parasites that carry alleles with the dhfr 108N + 51I or dhfr 108N + 59R mutations have increased antifolate drug tolerance, and can reinfect the host on average on day 8 after SP treatment.

Parasites carrying the dhfr 108 + 51 + 59 triple mutant genotype are of borderline sensitivity to SP.20 This genotype survives SP treatment at a rate chosen by the user. This parameter will be specific to each situation, since it is strongly influenced by the level of immunity in the patient population. In these examples, that rate is set at 60%: a treatment failure rate that has been recorded at Muheza, Tanzania, where the dhfr triple mutant frequency approaches 100% and patient immunity is expected to be high.19 The dhfr triple mutant parasites that survive treatment recrudesce, and this escape from treatment clearly represents a significant acceleration of the selection mechanism. The Muheza data support the borderline resistance in vivo predicted for the dhfr triple mutant from pharmacokinetic and pharmacodynamic studies.20

A parasite that carries the dhfr 108N + 51I + 59R + 164L quadruple mutant allele is unaffected by the maximum concentrations of SP that can be attained in vivo, and is of borderline sensitivity to chlorproguanil-dapsone (Lapdap™).20 Again, the rates at which this genotype survives treatment are user-defined.

Combinations with SP.

Combination of SP with a second drug will affect treatment efficacy and reinvasion characteristics. Both of these influences are user-defined in the model.

Sulfadoxine-pyrimethamine plus artesunate for three days.

When treatment with SP plus artesunate (SP/AS) is used, clinical efficacy is improved, but the success of reinvading parasites is still defined primarily by SP pharmacodynamics, because artesunate is eliminated very rapidly. The probability that the triple will survive treatment with SP/AS is user-defined, and we have set this rate at 20% since three days of AS added to SP monotherapy reduced parasite reappearance by day 28 (not polymerase chain reaction [PCR] corrected) from 53% to 23% in Kilifi in 1999–2001 (de Boom W, unpublished data). The dhfr triple mutant, with a high degree of tolerance to SP, can reinfect the host on average three days after treatment with SP or four days after the start of treatment with SP/AS (three days of treatment with AS plus one additional day for artesunate elimination in vivo to sub-therapeutic levels).

Sulfadoxine-pyrimethamine plus amodiaquine.

When SP is combined with amodiaquine (SP/AQ), the longer elimination profile of the active metabolite of AQ (desethylamodiaquine) has a significant effect on success of reinvading parasites. For the SP/AQ combination, we have assumed a moderately high sensitivity to the AQ component and set the treatment failure rate for AQ alone at 5%, which corresponds to average AQ failure rates currently observed in east Africa.21 As stated earlier, a parasite carrying the triple mutant genotype can reinfect the host three days after SP alone and four days after SP/AS. This genotype, however, can only reinfect the host 14 days after treatment with SP plus AQ due to the significant chemoprophylaxis associated with AQ treatment.22

Drug pressure.

The drug pressure input has two components: the average number of drug treatments per person per year (which determines the number of humans with residual drug levels and thus the proportion of parasite inoculations that encounter sub-therapeutic drug levels), and the proportion of infections treated (which determines the number of infections experiencing therapeutic levels of drug). Both of these parameters are expected to vary from one site to another, depending on the intensity of transmission. Both of these parameters are used-defined.

Starting frequencies of the dhfr alleles.

It is also necessary to specify the starting frequencies of the various dhfr alleles. This variable has a large impact on the absolute value of the UTL. In this model, the UTL can be defined with some precision for SP monotherapy, where we can estimate that a 60% gene frequency for the dhfr triple mutant is associated with a > 10% clinical failure: a stage at which change in recommended therapy should be imminent.23

The prediction of the UTL for combinations of SP with other drugs is the most important output of the model. This parameter is critically dependent upon the starting prevalence of the dhfr alleles. For example, if AS is added to SP at a late stage, when the selection of triple mutant dhfr alleles is well advanced, the model predicts no significant increase in UTL (see below). If AQ is added at the later stage, the extension of UTL over that of SP monotherapy is dictated by the rate of increase of AQ resistance. However, at dhfr triple mutant frequencies > 60%, we estimate that the SP/AQ combination would be nearing the end of its UTL. We have used a low starting allele frequency (10−5) when we model the UTL of a novel treatment, with virtually no operational use prior to implementation. Alternatively, we have used the dhfr triple mutant starting frequencies of 0.05 and 0.2 where we wanted to examine the effect of implementing a combination after several years of operational use of one of the drugs as monotherapy. The level of resistance in the partner drug, at time zero, is user-defined, as is the rate at which this resistance increases over time. It is assumed that all dosage regimens are standard, and that patient compliance is 100%

The model incorporates addition of mutations to alleles that are already resistant.3 One option is to assume that the malaria population is closed, so that no immigration of dhfr haplotypes occurs. The different haplotypes therefore arise sequentially by mutation. The wild type gives rise to the 108N, which in turn gives rise to the 108N + 51I or 108N + 59R, which in turn give rise to the triple 108N + 51I + 59R genotype. The mutation rate is user-defined and is assumed to be the same at each codon. The alternative is to assume that selection acts simply on pre-existing haplotypes, which may have entered the population through immigration. Once SP is operationally deployed, the frequency of these haplotypes rises, typically to values greater than 0.1, so that input from mutation, which is usually estimated as less than 10−5 and generally less than 10−9,24, is negligible. Recent data from South America, southeast Asia, and east Africa suggest that the immigration of already resistant parasites from outside the study area is an important factor,25–27 and this issue is addressed further in the Discussion.

To track the allele frequencies we must estimate how many infections each allele contributes on average to the next malaria generation. Thus, we can investigate the effect of SP alone and in CT by using simple weighted frequencies. For example if 70% of treatments are SP + AS and 30% are SP alone, then the probability of a triple mutant surviving treatment is (0.7Ya + 0.3Y), where Y is the probability of surviving SP alone and a is the effect of artesunate in reducing the probability that the parasite will survive CT compared with monotherapy. The PoC can be calculated in a similar manner: for the same conditions, the triple can reinvade on average (0.7 × 4 + 0.3 × 3) days after treatment (where the PoC for the triple mutant is four days for SP + AS and three days for SP28).

The parameters and equations (Appendix 1) are encoded in an Excel®(Microsoft, Redmond, WA) spreadsheet that is available for general use at http://pcwww.liv.ac.uk/hastings/CTmodel.xls and includes a table of the default settings of the model.

Applications to specified drugs and drug combinations.

Sulfadoxine-pyrimethamine (SP).

Plasmodium falciparum parasites carrying dhfr and dhps mutations have been isolated from several locations where SP has not been used. There is no evidence that the first three mutations in dhfr carry any biochemical advantage for the parasite. In fact, the DHFR enzyme encoded by the 108N + 59R genotype is significantly less effective, in terms of in vitro rate constants, than the wild-type.29 A much more likely explanation for the low levels of single and double mutant alleles of dhfr is that trimethoprim-sulfamethoxazole (cotrimoxazole) has been widely used for more than 50 years. The efficacy of cotrimoxazole against P. falciparum is variable: it was effective in the early 1990s in The Gambia and Malawi,30,31 but decidedly ineffective in Uganda in the later 1990s.32 Despite a short elimination half-life, which would tend to reduce selective pressure, it seems probable that in many regions of Africa, wide-scale use of cotrimoxazole has effectively selected, and maintained, P. falciparum populations that contain dhfr alleles that confer low-level antifolate resistance. As a result, we have used low but finite starting gene frequencies in modeling the selection of resistance when SP is brought into operational use, setting these at 10−5 for the single, the two double, and the triple mutant haplotypes. For selection of the dhfr quadruple mutant, we have used starting frequencies from 10−10 to 10−4 at a triple mutant frequency of 0.2.

The kinetics of selection are greatly affected by the two major variables: mean number of drug treatments per person per year and the proportion of infections treated. However, differences between the selections effected by monotherapy and combination are generally refractory to differences in these variables, and we have used values of 2 and 0.2, respectively, for all treatments, which we consider to be broadly representative of the African situation. These values can, of course, be varied to fit other situations. Figure 1 shows the increase in dhfr triple mutant frequency, when SP is first brought into use, as selection proceeds over a number of years. Throughout, this is an exponential process although the expansion is essentially unnoticed at low frequencies. As detailed earlier, clinical treatment failure is not obvious below a prevalence of 10–20% triple mutant alleles, but the model shows how little additional time is required for the frequency of the dhfr triple mutant to rise to 60%, which is equivalent to clinical failure of at least 10–15%.18 The overall time frame for this selection is approximately six 6 years, the time point at which clinical failure of SP was observed in southeast Asia.33 If other endpoints are used, parasite persistence or resolution of symptoms for 28 days, for example, the model can be easily modified to accommodate this change.

Sulfadoxine-pyrimethamine plus artesunate.

The ACT theory predicts that the addition of an artemisinin to SP would increase cure rates, reduce gametocytemia and slow the development of resistance. African trials of SP together with a three-day regimen of artesunate (4 mg/kg) have shown that the first two assumptions are correct. In an analysis of randomized trials in which the effect of ACT could be evaluated, treatment failure and gametocytemia were significantly lower at day 14 and day 28 in patients treated with SP/AS than with SP.12 In some trials, the day 28-cure rate was still low in both arms because of high levels of resistance to SP. Provided that ACT is started early (i.e., without significant prior use of SP monotherapy), the model outputs support the theory. Figure 1 shows that with SP monotherapy, a UTL of approximately six years is all that can be expected. With SP/AS, it takes approximately 10 years to reach the same prevalence of the triple mutant. That is a significant lengthening of the UTL of the treatment, although the effect will be reduced proportionately where combination use accounts for less than 100% of treatments. The advantage of SP/AS held true when we varied the key inputs of the model for a mean number of treatments per person per year and a proportion of infections treated. We assumed a very low initial level of biologically fit dhfr quadruple mutants (10−10), where a significant gene frequency is achieved only after 15 years of selection (this is seen in Figure 1 as a reduction in triple mutant frequency, as this genotype is replaced by the dhfr quadruple mutation). We have also assumed no change in susceptibility of parasites to artesunate over the time period modeled.

If SP monotherapy is used for a significant time before SP/AS is used, the picture is very different. In Figure 1, we also show the effect of SP/AS replacing SP monotherapy when the frequency of the triple mutant reaches 5% and when it reaches 20%. A 20% frequency of the dhfr triple mutant approximates that found in the field when the drug has been used widely in a secondary role, but before adoption as first-line treatment.16 Thus, we can compare the degree of protection provided when no significant resistance to SP exists prior to introduction of SP/AS to that provided when resistance to SP already exists. For the latter, the model predicts a very weak protection of SP by AS. This failure is explained by the exponential nature of the increase in dhfr resistance. A dhfr triple mutant prevalence of 0.2 characterizes a population in which SP is already close to the end of its UTL. The addition of AS will slow the final selection stages somewhat, but the combination is deployed too late to make a significant operational difference. At a dhfr triple mutant frequency of 0.6, SP/AS treatments with a short follow-up of 14 days may appear to be effective because of the rapid action of the artesunate component. However, a high proportion of recrudescent infections is shown by follow-up of 28 days or more.5 In east Africa, the dhfr triple mutant frequency now exceeds 0.2 in many, probably most, locations.

If the model predictions are substantially correct, there are important policy implications. If data collected in east Africa is extrapolated to west Africa, where SP is currently more effective, the prevalence of resistant alleles may already exceed the critical value for optimal protection of SP and extension of its UTL by the addition of a partner drug. National programs considering a policy change from chloroquine to SP/AS should carefully assess the prevalence of the DHFR triple mutant. Where this exceeds 5%, it is unlikely that any useful extension of UTL will be achieved by adopting the SP/AS combination. It is crucial that combinations being considered for testing in Africa are analyzed in this way because inappropriate combinations, which are doomed to failure, will affect adversely the uptake of CT or ACT into policy for reasons that are entirely spurious.

Sulfadoxine-pyrimethamine/amodiaquine combination.

In east Africa, SP resistance is now a major public health concern. In the East African Network for Monitoring Anti-Malarial Treatment (EANMAT), the proportion of sentinel sites in the Alert phase (6–15% treatment failure) increased from 5 of 17 pre-2000 to 13 of 32 post-2000. The proportion of sites where treatment failure exceeds 25% increased from 1 of 17 pre-2000 to 7 of 32 post-2000 (P = 0.15).21 In this subregion, the CT concept is widely accepted, and all six countries have now adopted CT as the basis for their antimalarial drug policy. Combination therapy increases the cost of treatment, especially with artemisinin combinations, and this is an important factor for African control programs. This is why more affordable combinations such as SP/AQ have been considered.34 In Uganda, SP/AS and SP/AQ were much more effective than SP monotherapy, and SP/AQ was even more effective than SP/AS in reducing the rate of subsequent treatments over a period of a year.35 Similar advantages for SP/AQ have been reported in a recent Rwandan trial.36 The SP/AQ combination is relevant for Africa because both drugs are widely available and affordable, but the important question is whether the UTLs of the SP/AQ and SP/AS regimens will be significantly different. Here, the model can provide information.

Amodiaquine is a pro-drug that is metabolized extensively to desethylamodiaquine, which has a variable elimination half-life in humans estimated at 1 – 10 days37 or 1 – 3 weeks38 and a mean plasma clearance of 75 mL/hour.39 In east Africa, AQ is generally more effective than SP, with a mean clinical efficacy of 95%.21 Thus, in combination with SP, the comparatively long elimination half-life of desethylamodiaquine will prevent reinfection of the host by parasites that would otherwise tolerate residual SP concentrations, as demonstrated in the Uganda trial.35 Amodiaquine is administered at a dose of 25–30 mg/kg of body weight over a three-day period. We have modeled the effect of adding this AQ dosage regimen to single-dose SP treatment on the rate of selection of the dhfr triple mutant. With user-defined starting gene frequencies, we can investigate the effect of this combination on the selection of antifolate resistant alleles from an early perspective, a point in time prior to widespread use of SP, and also from a time point when resistant alleles have already been selected by SP. The model permits user-defined, variable rates of resistance acquisition in the drug partner. Since AQ is exposed to selective pressure in the SP/AQ combination, we have used a starting resistance of 5% for AQ (see above), increasing at 5% of the initial value per malaria generation. This is a rather high rate of increase for one component of a CT, given that resistance to SP monotherapy only appeared to increase at 5–10% per generation.25,26 Thus, the predictions for AQ combinations are likely to underestimate the true UTL.

With a protective period of 14 days for AQ, and a 5% initial failure rate for AQ (increasing at 5% per malaria generation), the spread of SP resistance is significantly delayed by SP/AQ (Figure 2). This effect is achieved at the expense of increased exposure of AQ to selective pressure, so that AQ resistance will increase over time. As AQ efficacy decreases, the ability of the SP/AQ combination to delay SP resistance will also decrease. The precise rate of change cannot yet be included in the model because the genetic basis of AQ resistance has not yet been defined exactly. Figure 2 shows that if SP/AQ can be brought into use prior to the selection of resistant alleles by SP monotherapy, it will have a UTL of some 10 years. However, this potential is negated by a very low frequency of the triple mutant allele: if the combination is brought into use even at a triple mutant frequency as low as 5%, it is unlikely that the emergence of SP/AQ resistance can be effectively delayed. Furthermore, the selections for resistance in Figure 2 apply to the complete replacement of SP monotherapy by combination therapy. This will be very difficult to achieve in Africa: in practice, both treatments would co-exist, and this would detract from the protective mechanism. The important consideration for SP/AQ in Africa, therefore, is whether any locations remain on the continent where the dhfr triple mutant frequency is sufficiently low (i.e., where SP use has been minimal). This is certainly not the case in east and southern Africa, and may no longer be the case in west Africa. The model demonstrates drug-dependent differences in the way in which artesunate and amodiaquine can delay the spread of SP resistance (Figures 1 and 2), but emphasizes that both combinations need to be brought into operational use as early as possible, before the frequency of SP resistance alleles reaches the critical level.

Experimenting with this model may help countries in west Africa make the most informed choice, when they consider the use of SP in CT or ACT combination. In east Africa, Rwanda changed first-line malaria treatment from chloroquine to SP/AQ in 2001: a policy decision based mainly on direct costs of treatment, and at a stage where SP monotherapy was available through the private, but not the public sector. After two years of first-line use, there is evidence that SP/AQ remains effective in Rwanda.36 Where the SP/AQ combination can be effectively used, selection of the dhfr quadruple mutation is also delayed, even at the comparatively high starting frequency of 10−5, and with CT used in a relatively low proportion of treatments (Figure 3). For the few locations where SP/AQ remains an option, this could be a crucial factor controlling the effectiveness of newer antifolate treatments in the future.

Chlorproguanil-dapsone (Lapdap™) and the Lapdap™-artesunate combination (CDA).

Chlorproguanil-dapsone (Lapdap™) has been developed as a safe, effective, and affordable malaria treatment in Africa.19 This drug has just become commercially available in Africa, at a price comparable to SP, and the development of CDA is well under way.6 Lapdap™ was developed because of superior synergistic action over SP. It is well tolerated, of low cost, and, crucially, it has a short elimination half-life.40 The short half-life requires a three-day regimen for effective treatment.19,41 Critically, the alleles of dhfr that confer high levels of pyrimethamine-resistance also confer a degree of resistance to chlorcycloguanil, so the P. falciparum population already carries a burden of resistance. However, because of its short elimination half-life, Lapdap™ selects resistance less rapidly than SP,20,41 and this effect should substantially increase the UTL.

To model the UTL of Lapdap™ and CDA, we drew estimates of the PoC from studies of chlorcycloguanil-dapsone synergy in vitro against different dhfr genotypes and chlorcycloguanil and dapsone pharmacokinetic studies. For a three-day Lapdap™ treatment, the PoC is estimated to be 8 days for the dhfr wild-type, single, and double mutants, 4.9 days for the dhfr triple mutant, and 3.25 days for the dhfr quadruple mutant.20 Because artesunate is rapidly eliminated, these PoC values remain the same for treatment with CDA, except that for the dhfr quadruple mutant, the PoC increases from 3.25 to 4 days (see earlier). In recent African trials, where the most resistant allele was the dhfr triple mutant, the Lapdap™ 14-day failure rate ranged from 1% to 5%.40 However, the quadruple mutant (108N + 51I + 59R + 164L) is totally resistant to SP, and of borderline sensitivity to Lapdap™, so it is most important that we study the conditions under which the selection of this highly resistant genotype will occur. We estimate that the UTL of Lapdap™ ends when the dhfr quadruple mutation frequency reaches 10%. This genotype was selected by SP use in southeast Asia and South America and it is still common in both areas, but has not yet been detected by standard PCR techniques in Africa. A single trial with chlorproguanil-dapsone in Thailand, using different doses to those currently used,‡ produced a day 28 failure rate of 86%, against infections containing quadruple mutants.42 At the higher CPG dose used in Lapdap™ (2.5 mg/kg of dapsone and 2.0 mg/kg of chlorproguanil) we estimate a failure rate of 50%. For reasons given above, CDA would have greater activity, and we estimate the failure rate at 10% for the quadruple allele. Figure 4 shows the population changes in triple mutant frequency, when Lapdap™ and CDA treatments are used, at a starting triple mutant frequency of 0.2, a quadruple mutant frequency of 10−5, an intrinsic failure rate against the triple mutant of 6% for Lapdap™ and 2% for CDA, and of 50% and 10% respectively, against the quadruple mutant. The efficacy of Lapdap™ monotherapy would be high initially, since the triple mutant is fully susceptible, but would eventually be compromised by the increasing prevalence of more resistant alleles. In this example, the quadruple mutation achieves 10% prevalence after approximately 12 years (seen in Figure 4 as a reducing frequency of triple mutants as they are replaced by quadruple mutants). With 50% coverage, CDA prevents this occurrence for an additional 10 years. This value is based on the currently available estimates of Lapdap™ efficacy that measure treatment success at 14 days. It is likely that a 28-day follow-up would yield a lower estimate of efficacy and a shorter UTL. The model can easily accommodate this change when the 28-day success rates for any particular location are determined.

Dapsone (4 mg/kg) plus chlorproguanil (1.4 mg/day) for a three-day period was used in the Thai study.

The model output shown in Figure 4 makes two important predictions. First, it predicts that a complete replacement of SP by Lapdap™ would essentially arrest the exponential increase in triple mutant prevalence that is now occurring, primarily by eliminating the ability of the triple mutant to survive direct treatment. Second, it predicts that the replacement treatment would have an efficacy approaching 100%, irrespective of the starting prevalence of the triple mutant. This effect is robust, and similar for starting frequencies of the dhfr triple mutant between 0.2 and 0.8. This scenario presents an attractive proposition for African malaria programs, since it can be achieved by an affordable, short-course treatment which is now commercially available. However, the World Health Organization/Roll Back Malaria guidelines require that combination therapies should replace ineffective monotherapy,6 and we consider that both the Lapdap™/AS and Lapdap™/AQ combinations may have applicability in this role. Combination of Lapdap™ with AQ would be highly effective, available, and affordable. Although AQ is considered a reasonably safe drug for malaria treatment,43 a low proportion of treatments give rise to neutropenia,44 and this requires vigilance. Figure 5 compares the selection of the dhfr triple mutation under Lapdap™ and Lapdap™/AQ treatments. For the latter, restricted selection of dhfr resistance is obtained at the expense of increased exposure of the more slowly eliminated AQ component, in contrast to the CDA combination, where AS is protected by the more slowly eliminated Lapdap™, so it is likely that AQ resistance would increase over the time period shown. We have chosen a starting AQ resistance level of 10%, increasing by 5% per malaria generation: the current resistance level and the approximate rate at which AQ resistance has emerged in east Africa over the recent past.21 As shown in Figure 5, the Lapdap™/AQ combination significantly slows the selection of both the dhfr triple and quadruple mutants, although the effect is sensitively dependent upon the starting frequency of the dhfr quadruple mutation. For both the CDA and Lapdap™/AQ combinations, at a starting value of 10−3, rather than 10−5, quadruple mutant frequency increases rapidly after five years of use, and both combinations would be ineffective within eight years.

Given the foregoing, an important question arises, which the model can address. What will be the effect on the UTL of CDA if a country initially uses Lapdap™ monotherapy, and changes to CDA when this treatment becomes available as a fixed-dose formulation? This scenario is described in Figure 6. Essentially, the low selective pressure of Lapdap™ allows more operational leeway than is the case with SP. However, it is apparent that the facilitated selection of the quadruple mutation by monotherapy does impact significantly on the UTL of the combination, even for short periods of monotherapy. If the CDA combination is not used until Lapdap™ failures reach 10%, no significant benefit is obtained.

DISCUSSION

One of the more striking observations of the model and of practical experience is that CT or ACT must be implemented early to achieve optimal extension of the UTL. With hindsight, it is apparent that attempts to delay the spread of SP resistance in east Africa by adoption of SP/AS were seriously compromised by the ubiquitous high frequency of the dhfr triple mutant: a genotype that can survive SP plus three days of artesunate. Unless SP/AS is immediately implemented in those parts of Africa where this allele is still rare (a gene frequency < 5%), no worthwhile extension of UTL will be achieved. Speedy action is required to determine whether any of these locations remain on the continent: if not, SP/AS should be removed from consideration as a feasible ACT for malaria in Africa.

The model supports the use of SP/AQ, where this can be brought into use as the immediate successor to chloroquine (i.e., without a substantial period of SP monotherapy, and, again, a very low triple mutant frequency), and provide an explanation of the advantages of SP/AQ over SP/AS observed in comparisons in east Africa. From the model, there is good evidence that where SP, or other combinations including SP, have been used as first-line treatment for several years, there is little advantage to be gained from a change to SP/AQ. At best, this would extend the UTL by only a few years, and at the expense of a considerable increase in AQ resistance. This reality puts African malaria program managers in an extremely difficult position. The Roll Back Malaria initiative strongly supports the move to CT by African countries, exemplified in the development of a Lapdap™/artesunate combination (CDA). However, Lapdap™ will become available before CDA. Should they use Lapdap™ now as unprotected monotherapy, or wait several more years for CDA, in the expectation that this combination will have a substantially longer UTL? The model allows decision makers to compare outcomes for both scenarios, based upon their local conditions.

Lapdap™ is now available in Africa. Using the most current data for the 14-day clinical efficacy of Lapdap™, the model predicts that chlorproguanil-dapsone, even as monotherapy, has the potential for a long UTL (approximately 10 years) under conditions where the most antifolate-resistant genotypes involve the dhfr triple, but not quadruple, mutation. If Lapdap™ monotherapy were to be widely used, with an equivalent reduction in SP use, this might slow the exponential increase in dhfr triple mutant prevalence that is occurring in east Africa, and probably in other parts of the continent. It is clear that Lapdpap™ does exert selective pressure45 on dhfr. However, it could be assumed that adoption of Lapdap™ would delay the emergence of the dhfr quadruple mutant, if there were a strict sequential requirement in the order of the mutations for biologic fitness. However, there is evidence that dhfr triple mutants from two well-separated African sites, Tanzania and South Africa, share a common ancestry.26 Similar data have been reported from southeast Asia and South America.25,27 This suggests that the increase in the dhfr triple genotype has not resulted from numerous, sporadic, local mutations, but through a gene-sweep, effected by SP-mediated selection pressure. There is no reason why a similar incursion into Africa of a biologically fit dhfr quadruple mutant could not also occur. The model cannot predict the UTL of Lapdap™ with accuracy under these conditions without an estimate of the starting dhfr quadruple gene frequency. Furthermore, if there are novel mutations that encode alleles with high levels of resistance to chlorcycloguanil, but not pyrimethamine, that could greatly affect the outcome. Whatever the uncertainties of the selection process, it is apparent that the rate of selection of Lapdap™ resistance can be significantly reduced by combination with an appropriate partner drug.

The putative effects of natural selection against mutations encoding drug resistance have been omitted from our models because their magnitude is completely unknown. While the omission of natural selection avoids further complexity in the model, it also means that the results presented in this report should be regarded as worst-case scenarios that might be subject to amelioration in reality. This is most easily understood by considering the effects of adding a partner drug to a failing antifolate (Figures 5 and 6). If net selection for resistance is 10% per generation and the partner drug halves this to 5% then the UTL is largely unaffected. A net selection of 10% might comprise, for example, a resistance-mediated selection of 30% per generation and natural selection of −20% per generation (negative because natural selection acts to reduce the frequency of resistance). Halving the failure rate (through drug combination) would decrease the drug selection component to 15%, while natural selection would remain unchanged at −20% (a net selection of −5%), and natural selection would start to eliminate resistance. Since there are no estimates of the magnitude of this ameliorating effect, at present, we present worst-case scenarios, with this caveat.

Several antifolate-based drug combinations have been considered, studied, or brought into use in Africa in the urgent quest for an answer to the emergence and spread of drug-resistant malaria. The artemisinins, with their rapid and highly effective action in malaria treatment, have been advocated as primary candidates for combination with the anti-folates. However, rapid artemisinin elimination means a short protective effect on the partner drug, which may then spend much of its physiologic residence in the host as a sub-therapeutic monotherapy. Thus artesunate combination will not impede the early stages of resistance selection in a partner drug, such as SP, which has a long elimination half-life. Artesunate is unable to eliminate malaria infections reliably in a three-day regimen in Africa, the upper limit of acceptability for outpatient malaria treatment. This means that if resistance in the partner drug has passed a critical level, the ACT will not have adequate clinical efficacy at inception, the current situation in east Africa with SP/AS, and in parts of west Africa with chloroquine/AS.46 There is better potential for the AQ/AS combination in west Africa, where the critical resistance level in AQ does not yet seem to have been reached.

The response of some African malaria programs to this dilemma has been to investigate non-ACT combinations, and we have used the model as a tool to estimate the likely UTL of these treatments. Resistance-selection models of the type reported here may help in the choice of appropriate combinations for testing in Africa. This is particularly important where the proposed regimen includes a partner drug which is not, of itself, sufficient to clear parasites, e.g., artesunate daily for a three-day period. Where resistance to the other partner drug has reached a critical level, as with SP resistance in east Africa, the SP/AS combination cannot achieve an adequate cure rate, and the effect on UTL can be modeled by the methods we describe. In the development of antimalarial drug policy, these models may help to avoid the selection of inappropriate combinations that are doomed to failure, and reduce the risk that combination therapy, as an operational approach, is rejected for the wrong reasons.

APPENDIX 1 EQUATIONS DESCRIBING THE SPREAD OF ANTIFOLATE (AF) RESISTANCE

The equations are based on equation 3 of our earlier paper,3 and where appropriate, we use the same symbolism for consistency. The proportion of infections treated with a drug is x and the number of drug treatments taken per person per year is d. A drug persists in a host for some time after treatment, during which it may prevent new infections establishing themselves; this is known as the period of chemoprophylaxis, PoC, and differs between resistance alleles: the more resistant the allele, the sooner an individual can be reinfected after treatment. The number of successful secondary inoculations from a single infection is k, so this parameter contains all the other biological properties of the system such as transmission intensity, degree of host immunity, etc. These factors are the same for each allele so k cancels out in the subsequent equations; it is included initially purely to clarify the origin of the equations. Resistance to the antifolate is assumed to be encoded solely at the dihydrofolate reductase (dhfr) locus (mutations in dihydropteroate synthase also play a role,15 but we ignore them here for simplicity), and the initial equations are derived for sulfadoxine-pyrimethamine (SP) drug therapy. The partner drugs added to SP to form combination therapy (CT) are assumed to be either artesunate (AS) or amodiaquine (AQ); again this is for illustrative purposes and it is straightforward to investigate other partner drugs.

The dhfr wild-type has a PoC of 52 days after SP and is always killed by drug treatment, so its fitness is the probability that it is not drug treated (1 − x) times the number of secondary inoculations it makes, k, times the proportion of these inoculations of hosts that have not been drug treated in the last 52 days (1 − d/365) giving an overall fitness of (1 − x)k(1 − d/365)52.

The dhfr 108 is always killed by SP treatment and its PoC is 12 days. Its fitness is therefore (1 − x)k(1 − d/365).12 PoC for AS is taken as four days, which is less than the PoC for SP, so its fitness if SP is deployed as a CT with artesunate is again (1 − x)k(1 − d/365).12 However, PoC for AQ is 14 days (i.e., higher than for SP alone), so if SP is deployed with AQ the PoC is longer so the fitness of dhfr 108 decreases to (1 −x)k(1 − d/365).14

(If monotherapy and CT are deployed simultaneously, the PoCs are calculated as a weighted average, and the fitness calculated in the normal manner).

The fitnesses of the dhfr 108 + 51 and dhfr 108 + 59 double mutations are calculated in the same manner as for the dhfr 108, although their PoCs differ.

The fitness of the dhfr 108 + 51 + 59 triple mutation is affected by two properties: the ability to survive SP therapeutic treatment with probability r, and ability to reinfect a host three days after SP treatment. The PoC for SP is 3 days so fitness is [(1 − x) + xr]k(1 − d/365).3

The addition of AS to the treatment means the probability that a triple mutant will survive SP + AS therapy is further reduced by a factor t, and PoC for artesunate is four days after start of treatment so fitness is [(1 − x) + xrt]k(1 − d/365).4

The addition of AQ also reduces the probability that a triple will survive SP + AQ therapy by a factor u. The PoC of AQ is 14 days so the triple can only reinfect 14 days after treatment, giving fitness [(1 − x) + xru]k(1 − d/365).14

If monotherapy and CTs are co-deployed, survival probabilities and PoCs are calculated as weighted averages.

The fitness of the dhfr 108 + 51 + 59 + 164 quadruple mutations is calculated in an analogous manner except that the values of r, and possibly t and u, will change.

These equations enable the frequencies of dhfr mutations to be tracked over the time course of the evolution of resistance as described in more detail elsewhere.4 The time course of evolution is in units of parasite generations, which is translated onto a real time scale in the figures by assuming there are five parasite generations per year. The values of PoCs are user-defined in our calculations, so other antifolate drugs, e.g., chlorproguanil-dapsone, can be investigated in an analogous manner simply by changing PoCs and survival probabilities. We assume in this model that the mutant enzymes do not carry any cost in fitness because of lower enzyme activity. However, that cost in fitness can be incorporated simply by scaling the equations by (1 − s) where s is the strength of natural selection against the mutation.47

There are two main complications encountered when tracking dhfr mutation frequencies when drugs are being deployed as combination therapies. First, the need to track the evolution of resistance to the second drug B. This is achieved by specifying a starting frequency of resistance and basal rate of spread if CT is 100%, the rate of spread then being reduced appropriately if CT is less than 100%. Typically, initial resistance to drug B is 5% and resistance increases (geometrically) by 5% per parasite generation (so if CT is 30%, then rate of spread is 0.05 × 0.3 = 0.015 per parasite generation). The second main problem is that there are six mutations in the dhfr gene that are commonly observed in natural populations.15 In principle, these could occur in all possible combinations. Ideally, all possible parasite haplotypes would be tracked with alleles at the locus encoding resistance to drug B.47 With six dhfr mutations, and two alleles at the drug B resistance locus, there are 12 possible haplotypes in total. This is computationally feasible, but the critical problem lies in incorporating the effects of genetic recombination, which occur during the obligate sexual phase of the Plasmodium falciparum life cycle, and leads to independent assortment of the two loci. Recombination means that the haplotypes are not stable because recombination between two different haplotypes may result in a range of different daughter haplotypes. This instability of resistant haplotypes is one of the chief advantages proposed for the deployment of CT.45,47,48 We have used classic population genetic terminology. The terms recombination and haplotype may differ from the way these terms are now used in molecular biology. Genetic recombination merely indicates that gene combinations are rearranged during meiosis, leading to random reassortment of the alleles (it does not necessarily indicate physical recombination of loci on the same chromosome). Linkage (dis)equilibrium is a statistical description of how commonly alleles occur together and does not imply that the loci are physically linked on the same chromosome. Haplotype describes a complete haploid genotype, not simply a small region of DNA surrounding a specific locus.

The rate of genetic recombination depends on many factors, e.g., the ability of the differing haplotypes to survive treatment with CT and monotherapies, the proportion of CT used, the mismatch of PoCs for each drug, etc. Critically, it is also highly dependent on local epidemiology, with areas of intense malaria transmission having more recombination49 (and references therein). To avoid these complications, and concentrate on the basic mechanism, it is assumed that genetic recombination is sufficiently common that alleles at the dhfr and drug B resistance loci are in linkage equilibrium and are associated in a random manner. Non-random association, or linkage disequilibrium, can in principle be incorporated because an equilibrium is rapidly reached, which only starts to break down as resistance nears fixation.48 It is important to note that assuming linkage disequilibrium to be negligible will underestimate the rate of evolution of resistance to CT, so this is a conservative assumption with respect to the two main conclusions of the study. In the main text we show that concurrent use of monotherapy might substantially reduce the useful therapeutic life (UTL) of CT: our calculations overestimate the UTL of CT so this conclusion is robust. The other major conclusion is that adding a second drug to a failing first-line drug is unlikely to prolong UTL: our calculations overestimate this UTL so again the conclusion is robust.

Figure 1.
Figure 1.

Selection of the P. falciparum dhfr triple mutation by sulfadoxine-pyrimethamine (SP) monotherapy (dark green) and by SP/artesunate (SP/AS) employed from inception (light green) and as complete replacement therapy for SP when the triple mutant frequency reaches 5% (brown) or 20% (blue). For SP monotherapy, the frequency of the triple mutation reaches a maximum and then decreases as triples are replaced by dhfr quadruple mutants. Each case assumes 100% coverage with the respective treatment. This figure appears in color at www.ajtmh.org.

Citation: The American Journal of Tropical Medicine and Hygiene Am J Trop Med Hyg 72, 2; 10.4269/ajtmh.2005.72.163

Figure 2.
Figure 2.

Selection of the P. falciparum dhfr triple mutation (dhfr108N+51I+59R) by sulfadoxine-pyrimethamine (SP) monotherapy (dark green) and by SP+amodiaquine (SP/AQ) from inception (light green) and as complete replacement therapy for SP when the triple mutant frequency is 5% (brown) or 20% (blue). Each case assumes 100% coverage with the respective treatment. Other model assumptions: see the text. This figure appears in color at www.ajtmh.org.

Citation: The American Journal of Tropical Medicine and Hygiene Am J Trop Med Hyg 72, 2; 10.4269/ajtmh.2005.72.163

Figure 3.
Figure 3.

Selection of the P. falciparum dhfr quadruple mutation (dhfr108N+51I+59R+164L) by sulfadoxine-pyrimethamine (SP) monotherapy (purple) and by SP/amodiaquine (SP/AQ) employed from inception under conditions of complete CT coverage (light green) and 50% CT/50% SP monotherapy (brown). Other model assumptions: see text. This figure appears in color at www.ajtmh.org.

Citation: The American Journal of Tropical Medicine and Hygiene Am J Trop Med Hyg 72, 2; 10.4269/ajtmh.2005.72.163

Figure 4.
Figure 4.

Selection of the P. falciparum dhfr triple mutation (dhfr108N+51I+59R) by Lapdap™ monotherapy (dark green) and Lapdap™-artesunate (CDA), under conditions of complete ACT coverage (light green), and 50% ACT/50% Lapdap™ monotherapy (brown). Starting time point is a triple mutant frequency of 0.2. In the case of Lapdap™ monotherapy the frequency of the triple mutation reaches a maximum and then decreases as triples are replaced by dhfr quadruple mutants. Other model assumptions: see text. This figure appears in color at www.ajtmh.org.

Citation: The American Journal of Tropical Medicine and Hygiene Am J Trop Med Hyg 72, 2; 10.4269/ajtmh.2005.72.163

Figure 5.
Figure 5.

Selection of the P. falciparum dhfr triple mutation (dhfr108N+51I+59R) by Lapdap™ monotherapy (dark green) and Lapdap™-amodiaquine, under conditions of complete CT coverage (light green), and 50% CT/50% Lapdap™ monotherapy (brown). Starting time point is a triple mutant frequency of 0.2. The frequency of the triples eventually starts to fall in the case of Lapdap™ monotherapy because they are being replaced by dhfr quadruple mutants. Other model assumptions: see text. This figure appears in color at www.ajtmh.org.

Citation: The American Journal of Tropical Medicine and Hygiene Am J Trop Med Hyg 72, 2; 10.4269/ajtmh.2005.72.163

Figure 6.
Figure 6.

Selection of the P. falciparum dhfr quadruple mutation (dhfr108N+51I+59R+164L) by Lapdap™ monotherapy (purple) and Lapdap™-artesunate (CDA) from inception (light green), and after periods of Lapdap monotherapy which have increased the dhfr quadruple mutation frequency from 0.0001 to 0.01 (dark green) and to 0.2 (brown). Other model assumptions: see text. This figure appears in color at www.ajtmh.org.

Citation: The American Journal of Tropical Medicine and Hygiene Am J Trop Med Hyg 72, 2; 10.4269/ajtmh.2005.72.163

Authors’ addresses: William M. Watkins, 1 Forge Cottages, Mudford, Somerset BA21 5TJ, United Kingdom, E-mail: bwatkins@btinternet.com. Carol Hopkins Sibley, Department of Genome Sciences, Box 357730, University of Washington,, Seattle, WA 98195-7730, E-mail: sibley@gs.washington.edu. Ian M. Hastings, Liverpool School of Tropical Medicine, Pembroke Place, Liverpool L69 3BX, United Kingdom, E-mail: hastings@liverpool.ac.uk.

Acknowledgments: We are grateful to professors S. A. Ward and P. A. Winstanley (University of Liverpool) for their comments and suggestions.

Financial support: Ian M. Hastings is grateful to the Department for International Development-funded Malaria Knowledge Program of the Liverpool School of Tropical Medicine. This work was partly supported by National Institutes of Health grant AI-55604 to Carol Hopkins Sibley. William M. Watkins is grateful to the Wellcome Trust of Great Britain for research and personal support (grant no. 056305).

Disclaimer: The Department for International Development accepts no responsibility for any information or views expressed.

REFERENCES

  • 1

    Hastings IM, D’Alessandro U, 2000. Modelling a predictable disaster: the rise and spread of drug-resistant malaria. Parasitol Today 16 :340–347.

    • Search Google Scholar
    • Export Citation
  • 2

    White NJ, Nosten F, Looareesuwan S, Watkins WM, Marsh K, Snow RW, Kokwaro G, Ouma J, Hien TT, Molyneux ME, Taylor TE, Newbold CI, Ruebush TK 2nd, Danis M, Greenwood BM, Anderson RM, Olliaro P, 1999. Averting a malaria disaster. Lancet 353 :1965–1967.

    • Search Google Scholar
    • Export Citation
  • 3

    Hastings IM, Watkins WM, White NJ, 2002. The evolution of drug-resistant malaria: the role of drug elimination half-life. Philos Trans R Soc Lond B Biol Sci 357 :505–519.

    • Search Google Scholar
    • Export Citation
  • 4

    Peters W, 1985. The problem of drug resistance in malaria. Parasitology 90 :705–715.

  • 5

    Nosten F, Brasseur P, 2002. Combination therapy for malaria: the way forward? Curr Opin Drugs 62 :1315–1329.

  • 6

    World Health Organization, 2001. Antimalarial Drug Combination Therapy: Report of a WHO Technical Consultation. Geneva: World Health Organization.

  • 7

    EANMAT, 2001. Monitoring antimalarial drug resistance within National Malaria Control Programmes: the EANMAT experience. Trop Med Int Health 6 :891–898.

    • Search Google Scholar
    • Export Citation
  • 8

    Shretta R, Omumbo J, Rapuoda B, Snow RW, 2000. Using evidence to change antimalarial drug policy in Kenya. Trop Med Int Health 5 :755–764.

    • Search Google Scholar
    • Export Citation
  • 9

    Goodman CA, Coleman PG, Mills AJ, 2001. Changing the first line drug for malaria treatment–cost-effectiveness analysis with highly uncertain inter-temporal trade-offs. Health Econ 10 :731–749.

    • Search Google Scholar
    • Export Citation
  • 10

    Shretta R, Walt G, Brugha R, Snow R, 2001. A political analysis of corporate drug donations: the example of Malarone in Kenya. Health Policy Plann 16 :161–170.

    • Search Google Scholar
    • Export Citation
  • 11

    Hyde JE, 2002. Mechanisms of resistance of Plasmodium falciparum to antimalarial drugs. Microbes Infect 4 :165–174.

  • 12

    International Artemisinin Study Group, 2004. Artesunate combinations for treatment of malaria: meta-analysis. Lancet 363 :9–17.

  • 13

    Trape JF, 2001. The public health impact of chloroquine resistance in Africa. Am J Trop Med Hyg 64 :12–17.

  • 14

    Watkins WM, Mosobo M, 1993. Treatment of Plasmodium falciparum malaria with pyrimethamine-sulfadoxine: selective pressure for resistance is a function of long elimination half-life. Trans R Soc Trop Med Hyg 87 :75–78.

    • Search Google Scholar
    • Export Citation
  • 15

    Sibley CH, Hyde JE, Sims PFG, Plowe CV, Kublin JG, Mberu EK, Cowman AF, Winstanley PA, Watkins WM, Nzila AM, 2001. Pyrimethamine/sulfadoxine resistance in Plasmodium falciparum: What next? Trends Parasitol 17 :582–588.

    • Search Google Scholar
    • Export Citation
  • 16

    Amukoye E, Winstanley PA, Watkins WM, Snow RW, Hatcher J, Mosobo M, Ngumbao E, Lowe B, Ton M, Minyiri G, Marsh K, 1997. Chlorproguanil-dapsone: Effective treatment for uncomplicated falciparum malaria. Antimicrob. Agents Chemother. 41 :2261–2264.

    • Search Google Scholar
    • Export Citation
  • 17

    Nzila AM, Nduati E, Mberu EK, Sibley CH, Monks SA, Winstanley PA, Watkins WM, 2000. Molecular evidence of greater selective pressure for drug resistance exerted by the long-acting antifolate pyrimethamine/sulfadoxine compared with the shorter-acting chlorproguanil/dapsone on Kenyan Plasmodium falciparum. J Infect Dis 181 :2023–2028.

    • Search Google Scholar
    • Export Citation
  • 18

    Talisuna AO, Langi P, Mutabingwa TK, Watkins WM, van Marck E, Egwang TG, D’Alessandro U, 2004. Population-based validation of DHFR gene mutations for the prediction of sulfadoxine-pyrimethamine resistance in Uganda. Trans R Soc Trop Med Hyg 97 :338–342.

    • Search Google Scholar
    • Export Citation
  • 19

    Mutabingwa T, Nzila A, Mberu E, Nduati E, Winstanley P, Hills E, Watkins W, 2001. Chlorproguanil-dapsone for treatment of drug-resistant falciparum malaria in Tanzania. Lancet 358 :1218–1223.

    • Search Google Scholar
    • Export Citation
  • 20

    Watkins WM, Mberu EK, Winstanley PA, Plowe CV, 1997. The efficacy of antifolate antimalarial combinations in Africa: a predictive model based on pharmacodynamic and pharmaco-kinetic analyses. Parasitol Today 13 :459–464.

    • Search Google Scholar
    • Export Citation
  • 21

    EANMAT, 2003. The efficacy of antimalarial monotherapies sulphadoxine-pyrimethamine and amodiaquine in east Africa: implications for sub-regional policy. Trop Med Int Health 8 :860–867.

    • Search Google Scholar
    • Export Citation
  • 22

    Spencer HC, Oloo AJ, Watkins WW, Sixsmith DG, Churchill FC, Koech DK, 1984. Amodiaquine more effective than chloroquine against Plasmodium falciparum malaria on Kenya coast. Lancet 1 :956–957.

    • Search Google Scholar
    • Export Citation
  • 23

    World Health Organization, 2001. The Use of Antimalarial Drugs. Report of a WHO Informal Consultation. Geneva: World Health Organization.

  • 24

    Paget-McNicol O, Saul A, 2001. Mutation rates in the dihydrofolate reductase gene of Plasmodium falciparum. Parasitology 122 :497–505.

  • 25

    Nair S, Williams JT, Brockman A, Paiphun L, Mayxay M, Newton PN, Guthmann JP, Smithuis FM, Hien TT, White NJ, Nosten F, Anderson TJ, 2003. A selective sweep driven by pyrimethamine treatment in southeast Asian malaria parasites. Mol Biol Evol 20 :1526–1536.

    • Search Google Scholar
    • Export Citation
  • 26

    Roper C, Pearce R, Bredenkamp B, Gumede J, Drakeley C, Mosha F, Chandramohan D, Sharp B, 2003. Antifolate anti-malarial resistance in southeast Africa: a population-based analysis. Lancet 361 :1174–1181.

    • Search Google Scholar
    • Export Citation
  • 27

    Cortese JF, Caraballo A, Contreras CE, Plowe CV, 2002. Origin and dissemination of Plasmodium falciparum drug-resistance mutations in South America. J Infect Dis 186 :999–1006.

    • Search Google Scholar
    • Export Citation
  • 28

    Gamage SA, Tepsiri N, Wilairat P, Wojcik SJ, Figgitt DP, Ralph RK, Denny WA, 1994. Synthesis and in vitro evaluation of 9-anilino-3,6-diaminoacridines active against a multidrug-resistant strain of the malaria parasite Plasmodium falciparum. J Med Chem 37 :1486–1494.

    • Search Google Scholar
    • Export Citation
  • 29

    Sirawaraporn W, Sathitkul T, Sirawaraporn R, Yuthavong Y, Santi DV, 1997. Antifolate-resistant mutants of Plasmodium falciparum dihydrofolate reductase. Proc Natl Acad Sci U S A 94 :1124–1129.

    • Search Google Scholar
    • Export Citation
  • 30

    Daramola OO, Alonso PL, O’Dempsey TJ, Twumasi P, McArdle TF, Greenwood BM, 1991. Sensitivity of Plasmodium falciparum in The Gambia to cotrimoxazole. Trans R Soc Trop Med Hyg 85 :345–348.

    • Search Google Scholar
    • Export Citation
  • 31

    Bloland PB, Redd SC, Kazembe P, Tembenu R, Wirima JJ, Campbell CC, 1991. Cotrimoxazole for childhood febrile illness in malaria-endemic regions. Lancet 337 :518–520.

    • Search Google Scholar
    • Export Citation
  • 32

    Kilian AH, Jelinek T, Prislin I, Kabagambe G, Byamukama W, Mpigika G, Korte R, von Sonnenburg F, 1998. Resistance in vivo of Plasmodium falciparum to cotrimoxazole in western Uganda. Trans R Soc Trop Med Hyg 92 :197–200.

    • Search Google Scholar
    • Export Citation
  • 33

    White NJ, 1992. Antimalarial drug resistance: the pace quickens. J Antimicrob Chemother 30 :571–585.

  • 34

    Bloland PB, 2003. A contrarian view of malaria therapy policy in Africa. Am J Trop Med Hyg 68 :125–126.

  • 35

    Dorsey G, Njama D, Kamya MR, Cattamanchi A, Kyabayinze D, Staedke SG, Gasasira A, Rosenthal PJ, 2002. Sulfadoxine/pyrimethamine alone or with amodiaquine or artesunate for treatment of uncomplicated malaria: a longitudinal randomised trial. Lancet 360 :2031–2038.

    • Search Google Scholar
    • Export Citation
  • 36

    Rwagacondo CE, Niyitegeka F, Sarushi J, Karema C, Mugisha V, Dujardin JC, van Overmeir C, van den Ende J, D’Alessandro U, 2003. Efficacy of amodiaquine alone and combined with sulfadoxine-pyrimethamine and of sulfadoxine pyrimethamine combined with artesunate. Am J Trop Med Hyg 68 :743–747.

    • Search Google Scholar
    • Export Citation
  • 37

    Sweetman SC, ed., 2002. Martindale: The Complete Drug Reference. 33rd edition. London: The Pharmaceutical Press.

  • 38

    Krishna S, White NJ, 1996. Pharmacokinetics of quinine, chloroquine and amodiaquine. Clinical implications. Clin Pharmacokinet 30 :263–299.

    • Search Google Scholar
    • Export Citation
  • 39

    Winstanley P, Edwards G, Orme M, Breckenridge A, 1987. The disposition of amodiaquine in man after oral administration. Br J Clin Pharmacol 23 :1–7.

    • Search Google Scholar
    • Export Citation
  • 40

    Alloueche A, Bailey W, Barton S, Bwika J, Chimpeni P, Falade CO, Fehintola FA, Horton J, Jaffar S, Kanyok T, Kremsner PG, Kublin JG, Lang T, Missinou MA, Mkandala C, Oduola AMJ, Premji Z, Robertson L, Sowunmi A, Ward SA, Winstanley P, 2004. Comparison of chlorproguanil-dapsone with sulfadoxine-pyrimethamine for the treatment of uncomplicated falciparum malaria in young African children: double-blind randomised controlled trial. Lancet 363 :1843–1848.

    • Search Google Scholar
    • Export Citation
  • 41

    Nzila AM, Mberu EK, Sulo J, Dayo H, Winstanley PA, Sibley CH, Watkins WM, 2000. Towards an understanding of the mechanism of pyrimethamine-sulfadoxine resistance in Plasmodium falciparum: genotyping of dihydrofolate reductase and dihydropteroate synthase of Kenyan parasites. Antimicrob Agents Chemother 44 :991–996.

    • Search Google Scholar
    • Export Citation
  • 42

    Wilairatana P, Kyle DE, Looareesuwan S, Chinwongprom K, Amradee S, White NJ, Watkins WM, 1997. Poor efficacy of antimalarial biguanide-dapsone combinations in the treatment of acute, uncomplicated falciparum malaria in Thailand. Ann Trop Med Parasitol 91 :125–132.

    • Search Google Scholar
    • Export Citation
  • 43

    White NJ, 1996. Can amodiaquine be resurrected? Lancet 348 :1184–1185.

  • 44

    Adjuik M, Agnamey P, Babiker A, Borrmann S, Brasseur P, Cisse M, Cobelens F, Diallo S, Faucher JF, Garner P, Gikunda S, Kremsner PG, Krishna S, Lell B, Loolpapit M, Matsiegui PB, Missinou MA, Mwanza J, Ntoumi F, Olliaro P, Osimbo P, Rezbach P, Some E, Taylor WR, 2002. Amodiaquine-artesunate versus amodiaquine for uncomplicated Plasmodium falciparum malaria in African children: a randomised, multicentre trial. Lancet 359 :1365–1372.

    • Search Google Scholar
    • Export Citation
  • 45

    Curtis J, Maxwell CA, Msuya FH, Mkongewa S, Alloueche A, Warhurst DC, 2002. Mutations in dhfr in Plasmodium falciparum infections selected by chlorproguanil-dapsone treatment. J Infect Dis 186 :1861–1864.

    • Search Google Scholar
    • Export Citation
  • 46

    Sutherland CJ, Drakeley CJ, Obisike U, Coleman R, Jawara M, Targett GA, Milligan P, Pinder M, Walraven G, 2003. The addition of artesunate to chloroquine for treatment of Plasmodium falciparum malaria in Gambian children delays, but does not prevent treatment failure. Am J Trop Med Hyg 69 :19–25.

    • Search Google Scholar
    • Export Citation
  • 47

    Hastings IM, 1997. A model for the origins and spread of drug-resistant malaria. Parasitology 115 :133–141.

  • 48

    Dye C, Williams BG, 1997. Multigenic drug resistance among inbred malaria parasites. Proc R Soc Lond B Biol Sci 264 :61–67.

  • 49

    Anderson TJ, Haubold B, Williams JT, Estrada-Franco JG, Richardson L, Mollinedo R, Bockarie M, Mokili J, Mharakurwa S, French N, Whitworth J, Velez ID, Brockman AH, Nosten F, Ferreira MU, Day KP, 2000. Microsatellite markers reveal a spectrum of population structures in the malaria parasite Plasmodium falciparum. Mol Biol Evol 17 :1467–1482.

    • Search Google Scholar
    • Export Citation

Footnotes

*

The Global Fund for AIDS, Tuberculosis and Malaria (GFATM): www.globalfundatm.org.

SP-total failure is defined as the sum of early and late treatment failures in the World Health Organization in vivo test.21

Dapsone (4 mg/kg) plus chlorproguanil (1.4 mg/day) for a three-day period was used in the Thai study.

Save