The patient is a 38-year-old man from northeast Africa. In December 2008, he experienced a febrile illness in Sudan that was diagnosed as vivax malaria. He was treated with chloroquine (CQ) but did not receive PQ. The patient recovered and moved to Canada in mid-January 2009. One month after his primary infection, he sought treatment at a hospital in Canada with fever, chills, and malaise. P.vivax malaria was diagnosed by microscopy and real-time polymerase chain reaction. He was treated with CQ (600 mg base immediately, 300 mg base at 6, 24, and 48 h), followed by 14 d of PQ (30 mg by mouth daily). His estimated weight was 60 kg. The patient’s symptoms resolved, and smears were negative for Plasmodium on day 16. The patient experienced a second episode of symptomatic P. vivax malaria 3 months later. He was treated with CQ as before, followed by 28 days of PQ (30 mg by mouth daily). Smears were negative 2 days later. Nearly 30 months later, the patient had a third episode of P. vivax malaria. He had not traveled outside North America since his arrival in Canada. He was treated with CQ for 3 d, then PQ for 14 d (30 mg by mouth daily). Smears on days 2 and 9 after CQ treatment were negative. The importance of adherence was emphasized at each clinic visit, and the patient affirms that he took the full course of PQ treatment at the same time every day.

To identify mutations in parasite genes that are potentially associated with primaquine tolerance, we performed whole genome sequencing on P. vivax DNA obtained from patient samples at each relapse (EAC01–03). In total, 55,517 high-confidence single nucleotide variants (SNVs) were genotyped (Technical Appendix). The 3 parasite isolates were genetically related, but not identical, and they have been proposed to be meiotic siblings (A.T. Bright et al., unpub. data).

In addition, the 3 strains contained SNVs in genes homologous to known P. falciparum drug-resistance genes, including pvdhps, pvmdr, and pvmrp (6–8). Variants compared to the P. vivax reference strain SalI, presumed to be primaquine sensitive, were found at 27 of 39 sites within 5 known and putative drug resistance genes (Table). All 3 isolates possessed a double mutant antifolate-resistant genotype in pvdhfr (6). The SNVs within the putative drug-resistance genes in each of the patient’s 3 samples were identical except at amino acid positions 976 and 1393 of the pvmdr1 gene. The parasite genomes were also compared to the BrazilI strain of P. vivax, which was obtained from a patient who had multiple malaria episodes in a malaria-nonendemic country despite primaquine treatment (9). Comparison of the genotypes at the 5 genes demonstrated similar profiles. All strains exhibit intermediate to high levels of antifolate resistance on the basis of mutant genotypes identified in pvdhfr and pvdhps. In addition, the parasite strains obtained in this study share variant alleles with BrazilI in 2 multidrug resistance–associated transporters, pvmdr and pvmrp.

Host pharmacogenetics may also contribute to PQ failure by affecting drug metabolism. Genetic polymorphisms in the CYP gene family are associated with poor or intermediate metabolism of many drugs used to treat tropical infections (10) and several of these enzymes are specifically implicated in the metabolism of PQ (11) and other antimalarial drugs (12). We, therefore, determined whether the patient carried alleles that might also explain the failure of treatment. Based on allele frequencies in northeasernt African populations, polymorphisms within 4 of the 60 CYP genes were selected for genotyping: CYP1A2*1C, CYP2B6*6, CYP3A4*1B, and CYP2D6*4 (Technical Appendix). The patient was homozygous for the wild-type allele at all 4 loci.

Lastly, we examined whether the patient metabolized PQ to carboxy-primaquine (CPQ), the main PQ metabolite found in plasma. Drug levels were measured with a stereoselective bioanalytical LC-MS/MS method (W. Hanpithakpong et al., unpub. data). A plasma sample was collected on day 12 of treatment of the 3rd relapse, at 12–15 h post-dose. The total PQ and CPQ concentrations were 90 ng/mL and 1,042 ng/mL, respectively. The measured PQ concentration was similar to simulated maximum concentrations at steady-state in healthy male volunteers (96 ng/mL) and patients with vivax malaria (88 ng/mL). Concentration-time profiles for CPQ could not be simulated because of limited published information. These data demonstrate appropriate absorption of PQ and metabolism into CPQ.

Although this case highlights the challenges in managing patients with P. vivax who relapse after high doses of PQ, it also provides a unique opportunity to clarify the mechanisms underlying PQ tolerance. The multiple relapses in this patient result from previously acquired hypnozoites that likely possessed a genetic profile rendering them tolerant to PQ. Genotyping did not identify any mutations within 4 of the CYP loci potentially responsible for the antiparasite effect of PQ and plasma measurements demonstrated adequate levels of PQ and CPQ. However, this study presents a limited screen of polymorphisms in the CYP2D6 gene (13), and we cannot exclude the possibility that other alleles contribute to PQ tolerance.

Parasite genotype data demonstrate that the 3 isolates contain mutations in several putative drug- resistance genes. All 3 isolates are resistant to antifolates and harbor mutations in the ABC transporter genes that are implicated in resistance to numerous antimalarial drugs. Of particular interest are the mutations in the pvmrp1 gene that encodes a putative multidrug resistance-associated protein. Studies from P. falciparum implicate PfMRP1 in glutathione efflux, consistent with the predicted mode of action of PQ in disrupting mitochondrial function (14). Furthermore, gene knockouts of pfmrp1 have increased sensitivity to several antimalarial drugs, including PQ, which suggests this protein may play a role in transporting antimalarial agents out of the parasite (15).

This case study demonstrates the feasibility of using molecular tools to better understand therapeutic responses to PQ. Genetic analysis of SNVs in putative resistance genes may identify molecular markers of parasite resistance or correlate with known variations in PQ sensitivity of strains from different geographic areas. Clarification of the role of genetic factors involved in PQ efficacy cannot be readily addressed in populations in which endemic transmission occurs because relapses cannot be distinguished from reinfections. Genetic studies of relapses that occur in nontransmission settings provide a unique opportunity to answer questions about this human pathogen.