Characterization of Treatment Failure in Efficacy Trials of Drugs against Plasmodium vivax by Genotyping Neutral and Drug Resistance-Associated Markers

Approximately 40% of the world's population is at risk of vivax malaria (7). Recent interest in this infection has been heightened by the emergence of chloroquine (CQ) resistance in 1989 (21) and subsequent reports of severe disease (6, 25). There has been a resultant increase in monitoring of Plasmodium vivax drug sensitivity through efficacy trials (2, 12, 19, 20, 22) and identification of molecular markers of resistance (1, 8, 16, 24).

Recurrent P. vivax parasitemia in intervention trials may indicate not only treatment failure but also activation of liver-stage hypnozoites (relapses) or a new infection (17). Two studies of patients not at risk of reinfection found that most relapses were genetically distinct from the primary infections (4, 10). Standardized genotyping protocols characterizing treatment failure have not yet been developed for antimalarial trials for vivax malaria. However, candidate markers on genes coding for surface proteins (2, 9, 14) or neutral markers such as microsatellites (10, 13, 14) could, if sufficiently polymorphic, allow discrimination between strains in assessing posttreatment recurrence in a way analogous to that established for falciparum malaria (26). In a recent study of small numbers of children in Papua New Guinea (PNG) treated with amodiaquine or CQ plus sulfadoxine-pyrimethamine (SP), the authors recommended use of two highly polymorphic markers associated with a very low probability of independent infections carrying the same alleles (14). We have utilized this approach in a retrospective analysis of samples taken from a larger number of PNG children participating in an efficacy trial comparing CQ-SP and three artemisinin combination therapies (ACTs) (12), and we performed a complementary analysis of 4-aminoquinoline and SP drug resistance markers (3, 5, 15, 24).

The study was conducted in Madang and East Sepik Provinces between 2005 and 2007 (12) and involved 195 children aged 0.5 to 5 years with >250 P. vivax asexual forms/μl and no features of severe malaria who were randomly assigned to CQ-SP, artesunate-SP (ARTS-SP), dihydroartemisinin-piperaquine (DHA-PIP), or artemether-lumefantrine (AL) arms of the trial. The non-PCR-corrected clinical and parasitologic failure rates were 49.0%, 48.7%, 15.8%, and 51.5%, respectively, after 28 days of follow-up and 87.0%, 66.7%, 30.6%, and 69.7% after 42 days. There was no difference between the rate of recurrent P. vivax parasitemia between the CQ-SP, ARTS-SP, and AL arms (P = 0.28, log rank test) (Fig. 1A).

Genotyping based on length polymorphism of a region of msp1 (msp1F3) and a microsatellite, MS16, was performed; this combination has a probability of <0.25% for two isolates to carry the same alleles (14). Recurrent infections occurring during 42 days of follow-up that contained at least one genotype present at baseline were classified as recurrent infections with the same genotype, and recurrences with a different genotype were classified as new infections. Since P. vivax genotyping was not prespecified and in view of limited sample volumes, usable blood samples on the day of recurrent parasitemia were available for 70.1% and 70.3% of the samples to days 28 and 42, respectively. The present substudy was approved by the PNG IMR Institutional Review Board (approval 1029).

During 28 days of follow-up, there were no significant differences between CQ-SP, ARTS-SP, or AL in the rates of either recurrent parasitemia with the same genotype (Fig. 1B) (P = 0.74) or new infections (Fig. 1C) (P = 0.59). Up to day 42, there were significantly more cases of recurrent parasitemia with the same genotype in the CQ-SP arm but not in the number of new infections (Table 1). At days 28 and 42, fewer infections with either the same or different genotypes were observed after DHA-PIP (Table 1). Day 42 treatment failure rates were 87.0%, 66.7%, 69.7%, and 30.6% for CQ-SP, ARTS-SP, AL, and DHA-PIP, respectively, and 51.4%, 28.1%, 22.2%, and 9.7% after PCR correction.

To better understand these differences, we screened mutations in two P. vivax genes related to SP or 4-aminoquinoline resistance, namely, dhfr (5) and mdr1 (3, 23). Of patients allocated to the CQ-SP or ARTS-SP groups, 32 (69.6%) and 38 (71.8%), respectively, were infected with at least one triple or quadruple dhfr mutant parasite at enrolment (57L, 58R, and 61 M, ± 117T). While the small sample size did not allow firm conclusions regarding selection of mutant parasites, an increase in the frequency of triple/quadruple dhfr mutants was observed in the recurrent parasitemias with the same genotype (based on msp1/MS16 genotyping) in the CQ-SP (15/18; 83.3%) and ARTS-SP (9/9; 100%) arms. No such increase was observed in the AL arm (20/32 [62.5%] versus 4/6 [66.7%]). In the CQ-SP arm, the presence of parasites with triple/quadruple dhfr (57L, 58R, 61M/117T) plus mdr1 976F was associated with treatment failure (recurrent parasitemia with same genotype) with an odds ratio of 3.9 (95% confidence interval, 0.6–28.2; P = 0.08). Given the overall high levels of triple/quadruple dhfr mutants, this adds to emerging evidence that mdr1 (976F) mutations may be involved in reduced P. vivax CQ sensitivity (16, 24).

Despite comparable non-PCR-corrected adequate clinical and parasitologic responses (ACPRs), the genotyping/molecular marker data reveal between-treatment differences in recurrent parasitemia that reflect the pharmacodynamic and pharmacokinetic profiles of the antimalarial drugs (11, 18, 27). Given the high prevalence of quadruple dhfr mutant parasites and the relatively short elimination half-lives of the components, SP is likely to have contributed little to either initial parasite clearance or to prevention of new (or relapsing) infections during follow-up in the CQ-SP and ARTS-SP arms. Thus, ARTS would have been primarily responsible for successful initial clearance in the latter arm. The predominantly late occurrence of recurrent parasitemia irrespective of origin in the CQ-SP arm indicates that CQ remained partially effective despite the positive selection of mdr1 mutant (976F) parasites. The difference in efficacy between DHA-PIP and AL may largely reflect the terminal elimination half-lives of PIP (3 to 4 weeks) and lumefantrine (4 to 6 days), with long-lasting PIP suppression of reinfections and relapses regardless of genotype and plasma lumefantrine concentrations beyond 2 weeks posttreatment that were insufficient to prevent recurrence (20).

The present preliminary data highlight the important potential contribution that genotyping/molecular marker typing can make to improved characterization of recurrent parasitemia in P. vivax intervention trials. Since genotyping cannot differentiate between true failures and relapses with the same genotype, the primary endpoint should remain ACPR without genotyping. However, ACPR after genotyping based on epidemiologically appropriate markers could be added as a secondary endpoint. Further research may promote harmonization of P. vivax genotyping protocols and the adoption of consensus recommendations.

The intervention trial was sponsored by WHO Western Pacific Region, Rotary Against Malaria (PNG), and the National Health and Medical Research Council (NHMRC) of Australia (grant 353663). The present substudy was funded by NHMRC project grant 1010203. T.M.E.D. was supported by an NHMRC Practitioner fellowship.