Rapid dissemination of Plasmodium falciparum drug resistance despite strictly controlled antimalarial use

Abstract

Background: Inadequate treatment practices with antimalarials are considered major contributors to Plasmodium falciparum resistance to chloroquine, pyrimethamine and sulfadoxine. The longitudinal survey conducted in Dielmo, a rural Senegalese community, offers a unique frame to explore the impact of strictly controlled and quantified antimalarial use for diagnosed malaria on drug resistance.

Methodology/principal findings: We conducted on a yearly basis a retrospective survey over a ten-year period that included two successive treatment policies, namely quinine during 1990-1994, and chloroquine (CQ) and sulfadoxine/pyrimethamine (SP) as first and second line treatments, respectively, during 1995-1999. Molecular beacon-based genotyping, gene sequencing and microsatellite analysis showed a low prevalence of Pfcrt and Pfdhfr-ts resistance alleles of Southeast Asian origin by the end of 1994 and their effective dissemination within one year of CQ and SP implementation. The Pfcrt resistant allele rose from 9% to 46% prevalence during the first year of CQ reintroduction, i.e., after a mean of 1.66 CQ treatment courses/person/year. The Pfdhfr-ts triple mutant rose from 0% to 20% by end 1996, after a mean of 0.35 SP treatment courses/person in a 16-month period. Both resistance alleles were observed at a younger age than all other alleles. Their spreading was associated with enhanced in vitro resistance and rapidly translated in an increased incidence of clinical malaria episodes during the early post-treatment period.

Conclusion/significance: In such a highly endemic setting, selection of drug-resistant parasites took a single year after drug implementation, resulting in a rapid progression of the incidence of clinical malaria during the early post-treatment period. Controlled antimalarial use at the community level did not prevent dissemination of resistance haplotypes. This data pleads against reintroduction of CQ in places where resistant allele frequency has dropped to a very low level after CQ use has been discontinued, unless drastic measures are put in place to prevent selection and spreading of mutants during the post-treatment period.

Figure 1. Amount of antimalarials used in Dielmo from 1990 to 1999.

The mean number of person-years surveyed was 124 in 1990 (from June to December) and 295, 235, 238, 262.8, 260.1, 257.1, 292.5, 288.7, 299.3 and 297.8 from 1991 to 1999, respectively. The cumulated survey days were 51745, 81120, 82583, 90761, 89551, 88908, 100706, 95917, 100942 and 99083 from 1990 to 1999, respectively. The monthly quinine, CQ and SP intake was calculated from the recorded drug prescriptions in the data base. For <5% of cases, the actual prescribed dose was not available and was extrapolated from the standard treatment course at that time in the village. The peaks in 1992 and 1994 correspond to eradication therapies administered to 100, 118, and 59 persons, respectively , . There were 155, 262, 567, 310, 428, 435, 710, 649, 681 and 823 treatment courses administered from 1990 to 1999, respectively.

Figure 2. Temporal variation of the multiplicity of infection in Dielmo (A), frequency of Pfcrt codon 76 and Pfdhfr-ts codon 108 genotypes (B) and frequency of infections with only mutant type detected (C).

The number of isolates typed at each locus is indicated in Table 1. (A). Multiplicity of infection is depicted separately for each locus. For Pfmsp1 block2, the figures derive from nested PCR analysis using family-specific primers and allele identification based on allelic family assignment and size polymorphism. For Pfcrt and Pfdfhr-ts, it is based on K76T and S108N genotypes determined by molecular beacons, respectively. Symbols used: (Red triangles): Pfcrt codon 76 genotype; (green squares): Pfdhr-ts codon 108 genotype, (blue open circles) Pfmsp1 block2. B) Allelic frequency of resistance genotypes, calculated as percentage of mutant genotype within the total number of alleles detected for each locus. Symbols used as in A. C) Percentage of isolates containing only the mutant type. Symbols used as in A.

Figure 3. Temporal variation of the relative Pfdhfr-ts gene polymorphism in Dielmo during 1990–99.

The 1.8 kb PCR fragment corresponding to the full length Pfdhfr-ts coding sequence was sequenced on both strands for a total of 204 isolates. The yearly distribution of the various genotypes is shown, using the colour code shown to the right of the figure. The alleles presenting synonymous mutations were omitted from the colour coding. The C59Y non-synonymous substitution was a TGT to TAT mutation. The synonymous mutations (CTA to TTA or CTC for codon 40, GGA to GGC for codon 241) are not depicted. No mutation was detected at codon 16 and 164, and no bolivia repeat type was observed. Overall there were 155 isolates with wild type coding sequence and 49 isolates with non synonymous mutations (76% and 24%, respectively). There were 15 (7.3%) single mutants [51I (0.5%), 59Y (1%), 108N (5.8%)], one (0.5%) 51I 108N double mutant and 33 (16.2%) 51I 59R 108N triple mutants.

Figure 4. Frequency distribution of Pfcrt intron 4 microsatellite types by codons 72–76 and 220 haplotype.

247 isolates were typed (19, 23, 22, 33, 28, 22, 29, 17, 39 and 15 isolates in 1990, 1991, 1992, 1993, 1994, 1995, 1996, 1997, 1998 and 1999, respectively) for the intron 4 microsatellite by gene sequencing (see Materials and Methods). There were 31 CVIETS haplotypes and 216 wild type haplotypes. The haplotype codes are listed in Table S2.

Figure 5. Frequency distribution of Pfdhfr-ts −4.4 kb/−01 kb+0.5 kbmicrosatellite haplotypes by coding the sequence mutant type.

The microsatellites were determined as described in Materials and Methods for 81 isolates. The haplotype codes are listed in Table S3.

Figure 6. Temporal distribution of CQ and SP drug pressure and drug resistance in Dielmo in 1990–9.

The drug pressure is expressed as No of treatments/person/year (first graph) and as overall No of treatment courses administered per year (second graph). Panels A and B refer to CQ and SP, respectively. The prevalence of the Pfcrt mutant alleles was calculated from molecular beacon studies (N = 324) (see Figure 2), while the prevalence of the Pfdhfr-ts triple mutant was calculated from the full gene sequences available (N = 202) (see Figure 3). In vitro susceptibility assays were carried out in 1990–4 during the rainy season (N = 26) and from 1995 onwards for the last 2–3 months of the year, namely from 7/11/1995–26/12/1995 (N = 46) ; 6/01/1996–3/12/1996 (N = 59); 27/10/97–15/121997 (N = 26) ; 10/01/1998–15/11/1998 (N = 54) and 29/09/1999–08/11/1999 (N = 25). The proportion of interpretable CQ and pyrimethamine susceptibility tests was 68–81% and 72–81%, respectively, depending on the year. The prevalence of resistance is expressed as the percent of interpretable assays presenting a CI₅₀ for CQ >100 nM or a CI₅₀ for pyrimethamine>2000 nM. The occurrence of a second clinical malaria episode within 7, 14, 21 and 28 days of treatment was calculated as described in Materials and Methods. The bars correspond to the 95% confidence interval. The years before implementation of CQ and SP (1990–4) are grouped together. A. CQ pressure, Pfcrt 76T resistance mutation, CQ in vitro resistance and prevalence of clinical attacks following a CQ treatment B. SP pressure, Pfdhfr-ts triple mutant, pyrimethamine in vitro resistance and prevalence of clinical attacks following a SP treatment Colour codes: 1990–4: grey; 1995: purple; 1996: yellow; 1997: light green; 1998: light blue; 1999: orange.