From January 1, 1994, to June 30, 2000, patients seen at the Toronto General Hospital or the Hospital for Sick Children in Toronto, Canada, with microscopically confirmed falciparum malaria were enrolled. Patient interviews were conducted or medical charts were reviewed for potential risk factors for infection and drug resistance by using a standardized data extraction form. This study was approved by the Institutional Review Boards of the Toronto General Hospital and the Hospital for Sick Children.

Falciparum isolates were characterized by polymerase chain reaction-restriction fragment length polymorphism analysis and sequencing for allelic variants of pfcrt and pfmdr1, dhfr, and dhps as described (5–17). Proportions were compared by using the chi-square test or Fisher exact test, as appropriate. The chi-square test for trend was used as required for variables that involved ordered categories. For the purpose of this analysis, we considered mixed isolates (e.g., isolates containing parasites with mutant and wild-type alleles) to be mutant ones and compared them against those possessing only wild-type alleles.

During the study period, 105 consecutive cases of imported falciparum malaria were recorded (69 males, 36 females; age range 1–70 years [mean 30.5]). The geographic regions in which these persons acquired their infections are shown in Table 1.

The prevalence of mutations in chloroquine-resistance markers (pfmdr1 and pfcrt) was significantly higher in isolates acquired in East Africa compared with West Africa (Table 1). Patients who acquired malaria infection in East Africa had a 4.5-fold higher risk of being infected by an isolate possessing the K76T mutation in pfcrt (odds ratio [OR] 4.53 [95% confidence interval (CI) 1.26 to 16.02]; p=0.03) than those visiting West Africa. The N86Y mutation in pfmdr1 was also found more often in isolates acquired in East Africa than in those acquired in West Africa (OR 3.56 [95% CI 1.16 to 10.80]; p=0.03). A linear trend for increasing prevalence of N86Y mutant isolates was evident during the study period (OR per year =1.21; p=0.07) (Table 2). We also examined the association between past chloroquine exposure and the prevalence of chloroquine-resistance markers. Nineteen (18.1%) patients had used chloroquine for prophylaxis (n=13) or treatment (n=6) while abroad. These persons had a significantly increased risk of being infected by an isolate harboring the K76T mutation when compared with other travelers (OR=4.47 [95% CI 1.13 to 25.45]; p=0.03).

We grouped the parasite dhfr and dhps genotypes into four categories on the basis of the cumulative number of mutations that have been linked to escalating SP resistance (Table 1). More mutations in dhps were found in isolates from travelers returning from West Africa versus East Africa (p=0.001, Fisher exact test). We found that the proportion of isolates with at least two mutations increased during the study period in both dhfr (OR for a 1-unit increase in year = 1.28; p=0.02, chi-square test for trend) and dhps (OR=1.29; p=0.03) (Table 2).

In this study, we demonstrate “proof-of-principle” that a molecular surveillance strategy based on imported malaria in travelers can be used to detect and track drug-resistant malaria. Monitoring travelers for imported drug-resistant malaria is a surveillance strategy that offers several potential advantages. Recommendations regarding treatment regimens and chemoprophylaxis for travelers should ideally be made on the basis of the efficacy of these drugs in nonimmune travelers rather than on partially immune persons residing in malaria-endemic areas. However, to date there has been little information on the rates of drug resistance in cases of imported malaria. Using travelers as a sentinel system provides a mechanism to study large numbers of persons returning from diverse malaria-endemic areas. In contrast, traditional studies have often been based on relatively small numbers of persons residing in geographically restricted areas. Travelers are generally nonimmune, facilitating the interpretation of treatment and prophylaxis studies since outcome measures are not confounded by reinfections and by the varying degrees of immunity present in residents of malaria-endemic areas. Similarly, correlating the molecular mechanisms of drug resistance to treatment outcome in travelers may be more straightforward since these confounding variables can largely be excluded. Knowledge of the resistance genotypes of malaria parasites obtained from returning travelers can provide credible and complementary data for evidence-based recommendations for both chemoprophylaxis and therapy of malaria in travelers.

The high correlation between mutations in DHFR and DHPS and in vitro resistance to pyrimethamine and sulfadoxine, further supported by site-directed mutagenesis and transfection experiments, suggests that the epidemiology of antifolate resistance in P. falciparum can be monitored by molecular techniques (5–7,14–17). Furthermore, evidence exists for an association between a stepwise increase in the number of mutations in DHFR and DHPS and a corresponding increase in the level of clinical resistance to SP. In recent in vivo studies in partially immune persons in Cameroon and Kenya, multiple mutations in DHFR (e.g., triple mutation at codons 108, 59, and 51) were associated with early treatment failure, suggesting that these could be useful markers for predicting the in vivo efficacy of SP (18–20).

Using molecular markers of antifolate resistance, our study provides important data on the appropriateness of drugs such as SP that are currently recommended in North America and Europe for treatment or self-treatment of malaria in travelers. We observed that 75%, 66%, and 28% of consecutive imported isolates had at least one, two, and three mutations in DHFR, respectively. In DHPS, corresponding figures were 80%, 30%, and 5%. Furthermore, we found a linear trend for increasing prevalence mutations in dhfr and dhps during this study. These results suggest that antifolate resistance in imported falciparum malaria is now common and escalating over time. These observations question the rationale of continued recommendation of SP as either standby therapy or combination therapy with quinine for the treatment of P. falciparum malaria in travelers. However, some caution is needed in extrapolating our data to predict the in vivo efficacy of SP. Additional prospective in vivo studies, especially in the nonimmune host, are required to definitively link antifolate molecular markers with in vivo resistance (18–20).

We have also collected data on the occurrence of mutations associated with chloroquine resistance in consecutive imported falciparum isolates. The overall prevalence of the N86Y mutation in pfmdr1 and K76T mutation in pfcrt was 47.0% and 60.2%, respectively. Recent in vivo studies have assessed the association between pfcrt mutations and chloroquine response and determined that the K76 allele correctly predicted successful outcome (9–11). A rapid assay to detect pfcrt K76T in travelers’ malaria may be useful, since the presence of the K76 allele would indicate the probable effectiveness of treatment with chloroquine alone. On the basis of these findings, we anticipate that at least 40% of our patients would have responded to chloroquine. However, in the absence of a rapid test, current recommendations from the World Health Organization and the Centers for Disease Control and Prevention must be applied, and P. falciparum infections acquired in areas of known chloroquine resistance should not be treated with chloroquine.

Chloroquine and SP resistance has been selected by drug pressure (5,6,10,21). In Mali, the K76T mutation in pfcrt was observed in persons with persistent or recurrent infection after chloroquine therapy, indicating selection for this mutation. Our study extends these observations to travelers; those who had taken chloroquine for prophylaxis or treatment had a 4.5-fold higher risk of being infected with an isolate carrying the K76T mutation. Although the number of isolates studied was relatively small, our results also indicate that the prevalence of genotypes associated with chloroquine resistance was significantly higher in isolates acquired in East Africa than in those acquired in West Africa. This observation is consistent with currently reported epidemiologic patterns (22). The distribution of chloroquine- and SP-resistant parasites and their degree of resistance are far from uniform, and regular assessment of the therapeutic efficacy of chloroquine and SP, such as obtained with World Health Organization in vivo studies, is required. Studies such as ours, using travelers as sentinels, can contribute in a novel and complementary way to the continuous monitoring and tracking of geographic drug-resistance patterns. A network of digitally linked sites in the developed world that are performing these analyses in cases of imported malaria could provide global and timely monitoring.

In summary, our study demonstrates that a molecular surveillance strategy based on imported malaria in travelers can be used to detect and track patterns of drug-resistant malaria. Given the high prevalence of observed mutations in dhfr and dhps, our data provide evidence that raises questions about the rationale of continued use of SP to treat falciparum malaria in returned travelers. Our data also indicate that a considerable proportion of imported falciparum infections are still responsive to chloroquine.