The patient presented with fever 10 days after the end of his trip, and a diagnosis of falciparum malaria was made. Treatment with atovaquone-proguanil was well tolerated. Isolated fever in association with falciparum parasites appeared 23 days after therapy. The patient was successfully treated with quinine.

In vitro susceptibility tests performed on blood samples from day 0 and day 23 showed a 50% inhibitory concentration (IC₅₀) value for atovaquone that was more than 100-fold greater on day 23 than on day 0 (Table 1). In addition, the IC₅₀ for cycloguanil was increased by 18-fold.

Sequencing of the cyt b gene, encoding the atovaquone target (12), showed a wild-type P. falciparum strain on day 0 and a Y268S mutation on day 23.

Genotyping of the dhfr gene, encoding the proguanil target (5), showed a double mutation, C59R and S108N, on day 0, while a triple mutation, N51I, C59R, and S108N, was observed on day 23. However, proguanil likely does not act by itself in atovaquone-proguanil treatment but only facilitates the atovaquone activity (11).

Genotyping of the Pfcrt gene (wild type, K76), encoding a transport protein involved in chloroquine resistance, and of the dhps gene (wild type, S436, A437, K540, A581, and A613), encoding the sulfadoxine target (5), showed wild-type, identical alleles.

The genotyping of the two isolates, using three of six microsatellite loci (7A11, Pf2802, C4M79, Pf2689, TRAP, and C4M69) (1), msp1, and msp2 (5), showed differences between days 0 and 23 (Table 1).

The day 23 parasites presented a high IC₅₀ for atovaquone associated with a Y268S mutation in Cyt b. Since 2002, fewer than 20 cases of genetically confirmed clinical resistance to atovaquone-proguanil had been reported (2-4, 6, 7, 9, 13, 14). Clinical failures were associated with in vitro-increased IC₅₀s for atovaquone between day 0 and the failure day only in five isolates (4, 7, 9). In some cases, the increased IC₅₀ was moderate (7, 8). An in vitro atovaquone threshold of 1,900 nM was recommended to discriminate resistant isolates (10). Considering our results, this cutoff must be adjusted to >350 nM.

We were unable to detect Cyt b mutations on codon 268 and high IC₅₀ to atovaquone in the pretreatment isolate. Reinfection was excluded because the patient was treated after returning to France. The atovaquone-resistant strain was probably present in the initial isolate but in the minority, making it undetectable by classical genotyping methods and in vitro testing. The isolate was polyclonal on day 0 and monoclonal on day 23.

This is the first observation of the clinical failure of atovaquone-proguanil treatment of P. falciparum infection in a traveler in Comoros, an area where the in vitro prevalences of isolates with reduced susceptibilities to classical antimalarial drugs were <7% (12).

Although clinical failures of atovaquone-proguanil therapy remain rare in travelers, an increased vigilance is required during their treatment followup, and surveillance of the parasite population should be reinforced as well.