Figure

Figure. Rainfall (bars) and monthly incidence of Plasmodium falciparum clinical malaria cases (curve) at the French Military Hospital – CHA Bouffard (circle) and in the dispensaries of the city (Department of Epidemiology...

The study was conducted at the Centre Hospitalier des Armées Bouffard, a French military hospital in Djibouti serving military and civilian natives from the entire city, and at other public health facilities of Djibouti. From 1997 to 2002, clinical malaria in the hospital shows the same temporal fluctuations as in dispensaries in the city (Figure). The incidence of patients with P. falciparum malaria admitted to the hospital increased >10-fold from March to May 1999 compared with the same period in 1997, 1998, and 2000–2002. In contrast, the number of admissions, consultations at the outpatient clinic, or blood counts performed for other causes than fever did not vary over the same period. The meteorologic station of the international airport of Djibouti recorded heavy rainfall the month before the epidemic. However, similar rainfall in 1997 or autumn 1999 was not followed by such a dramatic increase in malaria incidence in the ensuing months (Figure). When annual averages were compared, no particular variations in minimal or maximal mean air temperatures were found to occur during the months preceding the epidemic.

Forty-six blood samples were collected from September 14 to December 31, 1998 (period 1), 61 from April 12 to April 30, 1999 (period 2), and 32 from March 15 to May 15, 2002 (period 3), from patients with P. falciparum clinical cases who had not travelled outside the city of Djibouti during the preceding month and declared not having taken any antimalarial drug before the blood sampling. The study was cleared by the Djibouti Ministry of Health. Informed oral consent was obtained from patients before blood collection. Venous blood was collected before treatment administration in Vacutainer EDTA tubes (Becton Dickinson, Rutherford, NJ, USA). Thin blood smears were stained with an RAL kit (Réactifs RAL, Paris, France). Parasitemia was expressed as the proportion of P. falciparum–infected erythrocytes. Aliquots of fleshly collected blood were kept at –20°C until DNA extraction.

P. falciparum genetic diversity was investigated by using msp1 and msp2 encoding highly polymorphic loci from merozoite surface protein genes. Msp1 and msp2 were genotyped by using nested polymerase chain reaction (PCR), as described (7), except that family-specific fluorescent primers were used in the nested PCR for assignment to the K1-, Mad20-, or Ro33-type msp1 family and to the 3D7- or FC27-type msp2 family. Fragment length was analyzed by the Genescan technology. Approximately 50% of the blood samples contained multiple msp1 or msp2 genotypes. The mean multiplicity of infection, i.e., the number of genotypes present in the blood sample, was ≈1.5 concurrent P. falciparum infections per person, with a decreasing tendency over the study period (Table 1). For each locus, multi-infection cases were excluded from analysis of genetic diversity. We identified 9 msp1 alleles in 83 isolates and 17 msp2 alleles in 108 isolates. The genetic diversity estimated by the unbiased expected heterozygocity (8), i.e., the probability that 2 randomly chosen genotypes are different in the sample, before, during, and after the 1999 outbreak was 0.79 (n = 23), 0.37 (n = 39), and 0.64 (n = 21) at the msp1 locus and 0.83 (n = 31), 0.34 (n = 47) and 0.63 (n = 30) at the msp2 locus, respectively. During the epidemic, Ro33-131 accounted for 79% of the msp1 allele and FC27-408 accounted for 81% of the msp2 alleles. Both alleles were present before and after the epidemic but with a much lower prevalence. They accounted for 26% of the msp1 and 35% of the msp2 alleles in 1998 and 14% of the msp1 and 10% of the msp2 alleles in 2002 (Table 2).

To look for resistance-associated point mutations and haplotypes, the complete coding region of Pfdhfr (dihydrofolate reductase) and Pfdhps (dihydropteroate synthase) was amplified and sequenced (ABI 3100 Genetic Analyser, Applied Biosystems, Courtaboeuf, France) as described (9). We focused the analysis on point mutations of Pfdhfr codons 16, 51, 59, 108, and 164 and Pfdhps codons 436, 437, 540, 581, and 613, which have been associated with resistance to pyrimethamine and proguanil metabolite and to sulfadoxine, respectively (10). The prevalences of the Pfdhfr and Pfdhps mutations are shown in Table 1. No mutant was detected for Pfdhfr codons 16 and 164 and Pfdhps codon 581. A single isolate collected in period 2 harbored the Pfdhps A613S mutation. No isolate harbored the quintuple mutant haplotype (Pfdhfr S108N, N51I, and C59R and Pfdhps K540E and A437G) or the Pfdhfr C59R and Pfdhps K540E combination that predicts sulfadoxine-pyrimethamine clinical failure (9). One isolate containing at least 2 P. falciparum populations harbored 3 Pfdhfr mutations (S108N, N51I, and C59R) and the Pfdhps K540E mutation.

From 1998 to 1999, the frequency of isolates with mutated Pfdhfr codons 51, 59, and 108 decreased (not significantly), and Pfdhps allelic frequency did not differ significantly. The prevalence of isolates harboring the Pfdhfr N51I, Pfdhfr S108N, Pfdhps A437G, and Pfdhps K540E mutations increased from 1998–1999 to 2002 (Fisher exact test, p < 0.001 each). Presence of the chloroquine resistance–associated K76T mutation of Pfcrt (chloroquine-resistance transporter) (11) was analyzed by nested allele–specific PCR. Over the study period, 93% of the isolates harbored the Pfcrt K76T mutation (Table 1), without any significant temporal variation.

Twenty seven P. falciparum isolates collected during the 1999 epidemic with a 0.05%–5.0% parasitemia were transported at 4°C to our laboratory in Marseille, France, and analyzed for in vitro drug sensitivity by using an isotopic microtest (12). Among them, 93% were classified as resistant to chloroquine (Table 3). No isolate was resistant to amodiaquine. In vitro resistance was 4% for both pyrimethamine and cycloguanil.

Before and after the 1999 epidemic, P. falciparum genetic diversity in Djibouti was large, with ≈80% and 63% heterozygocity. This finding is somewhat surprising for an area where disease endemicity is low (13) and probably reflects importation of strains from neighboring areas such as Ethiopia or Somalia (1,5). P. falciparum genetic diversity was diminished during the epidemic, reflecting the circulation of a restricted number of strains during that period. Most of these strains harbored an msp1 and msp2 genotype that was detected before the epidemic. The prevalence of Pfcrt, Pfdhfr, and Pfdhps mutant genotypes did not vary significantly from 1998 to 1999. Thus, our data do not support the hypothesis of a sudden increase in the drug resistance of the local P. falciparum population as causing the epidemic. Our data are also not consistent with massive invasion by a single strain/genotype but rather suggest expansion during the epidemic of a few strains that were already prevalent. Further genotyping is needed to establish how many strains were circulating and their possible origin. What could have caused this sudden amplification? One possibility is a temporary increase in vector density. Unfortunately, no vectors were captured at that time, and this hypothesis is difficult to explore retrospectively.

The low prevalence of Pfdhfr and Pfdhps resistance mutations in 1998 and 1999 and of proguanil or pyrimethamine in vitro resistance in 1999 may explain the very low incidence of clinical malaria among the French soldiers stationed in Djibouti who were taking chloroquine-proguanil chemoprophylaxis. However, the sharp increase of Pfdhfr and Pfdhps resistance mutations observed in 2002 threatens sulfadoxine-pyrimethamine efficacy in the near future, even more so since the limited acquired immunity is unlikely to contribute to sustained drug efficacy (14). Molecular and in vitro assays point to a very high prevalence of chloroquine resistance. This finding calls for an urgent in vivo assessment of the antimalarials presently used in Djibouti in order to consider a rapid change in first-line treatment policy.