Malaria is a globally distributed infectious disease and, in 2020, 241 million cases and 627 thousand deaths from malaria were reported worldwide (1). In Brazil, for the same period, approximately 145,188 cases and 42 deaths were recorded, of these, 82.5% were caused by Plasmodium vivax (1). P. vivax has certain characteristics, such as its ability to remain latent in the liver in the form of hypnozoites that can cause relapses when reactivated, that are an obstacle for its elimination (2, 3).

Primaquine (PQ), a hypnozoiticidal drug, needs to be biotransformed into an active metabolite in order to exert its antimalarial effect (4). Biotransformation occurs through two main pathways: cytochrome P-450 (CYPs) and monoamine oxidase (MAO-A) (5). Biotransformation mediated by CYPs is attributed to the enzymes CYP2C19, CYP2D6 and CYP3A4 (6). The MAO-A pathway generates aldehyde derivatives, such as carboxyprimaquine, a predominant but inert metabolite (6, 7).

The contribution of the human host’s genetics to the results of antimalarial treatment has been demonstrated in recent studies (8–11). Bennet et al. (8) reported therapeutic failure of PQ in the treatment of vivax malaria, which was attributed to the presence of polymorphism in CYP2D6, thus resulting in decreased biotransformation of the drug and, consequently, low levels of the active metabolites. Baird et al. (9) showed that decreased CYP2D6 activity was associated with an increased risk of therapeutic failure, which suggests a relation between CYP2D6 with PQ biotransformation and an increased risk of relapse. The relation between the therapeutic failure of PQ and the presence of polymorphism in CYP2D6 has also been reported in Brazil (10–12).

Pybus et al. (6) demonstrated that MAO-A and CYP2D6 are responsible for 93% of PQ metabolites and that only traces levels of oxidated and dimethylated metabolites are generated by CYP3A4 and CYP2C19. Ariffin et al. (13) associated the presence of the CYP2C19*2 allele with the decrease in PQ biotransformation into carboxyprimaquine (13). This null activity allele has also been related to the decrease in biotransformation of other antimalarial drugs (7, 14, 15). Furthermore, CYP3A4 has a role in the biotransformation of PQ, but no clear genotype-phenotype association for this enzyme, and the effect of its variants remains controversial (6, 16–18).

Polymorphic variants of CYP2C19, CYP2D6 and CYP3A4 have been reported in populations in Brazilian Amazon, as well as incidences of P. vivax malaria infection and recurrent infections by the parasite. Genetic variability of these CYPs is likely to have an impact on the clinical response of patients, and the detection of genotypes related to PQ metabolism could guide individualized treatment protocols in the future in patients with multiple recurrences. As such, this study aimed to investigate the frequency of genotypes and phenotypes of CYP2D6, CYP3A4 and CYP2C19 and their association with recurrence of P. vivax in patients from the Brazilian Amazon.

This study is the first to simultaneously evaluate the influence of CYP2D6, CYP2C19 and CYP3A4 genetic polymorphisms in relation to the recurrence of P. vivax in the Brazilian Amazon. Since some patients followed up in this study come from urban areas, where the vector mosquito is absent, the cause of these recurrences could be attributed to P. vivax relapses (24). This causes problems in the elimination of malaria, since these individuals can directly contribute to the maintenance of local transmission. In addition, hypnozoites are difficult to eradicate, as they can only be eliminated by treatment with primaquine, which may have its effectiveness reduced due to altered metabolism of CYP2D6 (6, 8) With the constant decline in malaria transmission in some areas of the Brazilian Amazon (25, 26), protocols for personalized therapy could become more viable and their implementation could be possible in referral centers.

In this study, after genotyping of 11 SNPs, no significant difference was found between CYP2D6 allele frequencies and patients with or without recurrence (p=0.646). In addition, increased function alleles were only observed in patients without recurrence (p=0.018). In general, the frequency of these alleles is low and since the majority of the studies are conducted to evaluate the association of impaired genotypes and PQ failure, this is the first study to report a high frequency of CYP2D6 multiplication in individuals without recurrence.

Furthermore, the gUM phenotype presented a higher frequency in those without recurrence (p=0.057). Some studies have shown that the presence of the gUM phenotype is related to the high risk of toxicity of the prodrug codeine, and a high risk of failure of tricyclic antidepressants (27, 28). However, for individuals with the gUM and gNM phenotype, it is expected that a standard dose of tamoxifen should achieve the therapeutic effect in breast cancer treatment (29). For PQ, it is supposed that individuals with the gUM phenotype would exhibit the fastest metabolism through the CYP2D6 pathway. Gonçalves et al. (30) evaluated the effect of the CYP2D6 activity score (AS) on PQ plasma concentration over time after a single dose of PQ, and showed that the prodrug concentration decreased faster in children with AS=3 than those with AS ≤ 2. Moreover, the clinical effect of the gUM phenotype is still unknown (5). In our study, the frequencies of the CYP2D6 gPM, gIM and gNM phenotypes were similar between those with or without recurrence, similar to what has previously been observed in the Brazilian Amazon region (11).

The influence of decreased CYP2D6 phenotypes (PM+IM) in the recurrence was not associated with high risk of recurrence. No significant effect of predicted CYP2D6 phenotype in recurrence episodes was reported in Australia and Thailand (31, 32). However, the effect of the CYP2D6 impaired phenotype and the increased risk of PQ failure has been demonstrated in other studies, thus suggesting that PQ metabolism is dependent on CYP2D6 (8, 9, 11, 33).

The Brazilian Amazon is a region that is endemic for malaria; therefore, there is a possibility that part of the unidentified recurrences could be reinfections and relapses due to primaquine failure. This could be the cause of the disagreements between the results of this study and other previously published studies (8–11). Furthermore, there are differences regarding the system for CYP2D6 phenotype prediction from the genotype; specifically, regarding the activity index and the cut-off points used to predict CYP2D6 activity levels (22, 34). Some studies have already reported inhibition of CYP2D6 with the use of chloroquine, which would affect the actual phenotype of the enzyme and the response to primaquine (35, 36).

In this study, we did not find any association between the CYP2D6 predicted phenotype and gametocytemia or asexual parasitemia clearance. A recent study in the Amazon region also did not show any association between the presence of polymorphisms in CYP genes and early asexual clearance of P. vivax (37). However, Pett et al. (38) demonstrated the effect of the CYP2D6 gPM/gIM phenotype in P. falciparum gametocyte clearance, and it was also demonstrated that the biotransformation of PQ is not a condition for the eradication of the blood stages (asexual and sexual) of P. berghei (39). The CYP2D6 predicted phenotype was also not associated with the number of recurrence episodes and neither with the time to the first episode of recurrence, similar to previously observed (11, 33).

In our study, the frequency of this allele is in agreement with what is already known for northern Brazilians (40, 41) ( Table 2 ). Even though the PM phenotype was more often observed in individuals with recurrence, the frequencies of PM and IM phenotypes were similar between the groups. Furthermore, this study demonstrated that there is no association between the CYP2C19 polymorphism and the recurrence episodes neither with asexual parasitemia clearance nor gametocyte clearance. Pybus et al. (6) showed that CYP2C19 plays a role in PQ metabolism, although the influence of this CYP in the production of PQ’s hemolytic and/or therapeutic metabolite is not clear.

For CYP3A4*1B, no statistical difference was found in the allele frequency between the individuals with or without recurrence (p=0.762). Several studies have been conducted to explain the importance of polymorphisms in the CYP3A4 genotype–phenotype relationship, but the functional effect of the polymorphism remains contentious (15, 42). Although CYP3A4 is known to have a role in PQ biotransformation (6), we could not show the effect of CYP3A4*1B in the recurrence episodes in individuals treated with PQ, neither in the asexual parasitemia clearance nor in gametocyte clearance. A previous study conducted in the Amazon also showed no association between this allele and the recurrence episodes nor between the presence of CYP3A4*1B and the clearance time (37).

Because PQ is not metabolized by only one enzyme (6, 12), a multigenic analysis might be more suitable for determining the effect of the CYP450 polymorphism in PQ failure. In our study, we found a single individual representing the CYP2D6 gPM (*4/*4) + CYP2C19 gIM (*1/*2) that had had 2 recurrence episodes. Despite this, a CYP2D6 + CYP2C19 analysis was not associated with a higher risk of recurrence.

This genotype-based prediction assumes that the genotyping precisely corresponds the metabolic activity (42, 43). However, the prediction of phenotype from the genotype has not been correlated with complete certainty, since the gene splicing, SNPs, epigenetics, microRNA, transcription regulation and multiple gene copies are factors that can modify the enzymatic activity (44–47). Furthermore, phenoconversion may occur, whereby a genotypic gNM can be converted into a transient phenotypic IM or PM, which is mainly caused by extrinsic factors, such as drug-drug interactions and some pro-inflammatory cytokines in the inflammatory process (43, 44, 48).

This study had some limitations. The phenotype was predicted by the genotype. Recently, Baird et al. (9) showed a highly significant correlation between the genotype-determined activity score and the measured phenotypes when using dextromethorphan as a probe drug. Primaquine metabolites were not assessed to confirm the impact of CYP2D6 in the pharmacokinetics of the drug. Drug-drug interactions and the presence of other inflammatory conditions were not evaluated. It was not possible to certify factors related to the quality of the CQ and PQ drugs administered to patients and PQ administration was not supervised. The low sample size for individuals with recurrence found here is in agreement with other studies (9, 11). Moreover, data regarding recurrence episodes were accessed using the SIVEP-Malaria platform, which is susceptible to underreporting issues.

In this study, it was not possible to establish an association between CYP2D6, CYP3A4, and CYP2C19 decreased metabolizing genotypes and recurrence of P. vivax. The investigation of other causes of recurrence for these individuals is necessary since it is a region where there is a high risk of re-infection and many asymptomatic infections (49).

This prospective cohort study demonstrated no influence of CYP2D6, CYP2C19 and CYP3A4 polymorphisms on malaria recurrence. Despite this, our findings suggest that the CYP2D6 predicted ultrarapid phenotype was less susceptible to recurrence (p=0.057). Future studies are warranted in order to understand the association of PQ and CYP450 considering factors such as the metabolite concentration, the drug–drug interactions and the presence of some inflammatory conditions, and these may guide the individualized radical treatment of vivax malaria in reference centers that treat patients with multiple recurrences.

The original contributions presented in the study are included in the article/supplementary material. Further inquiries can be directed to the corresponding author.

The studies involving human participants were reviewed and approved by Fundacao de Medicina Tropical Dr. Heitor Vieira Dourado. The patients/participants provided their written informed consent to participate in this study.

Conceived and designed the experiments: GM, WM, ML, VS, and JC. Sample processing: JC and YS. Performed the experiments: JC, AA and LB. Data entry and analyses: JC, MR, VS, and AA. Wrote the paper: JC, GM, YS, LB, VS, FR-S, WM, and ML. All authors read and approved the final manuscript.

This research was funded by the Conselho Nacional de Desenvolvimento Científico e Tecnológico (CNPq) via the Universal (MCTIC/CNPq 2018), and Fundação de Amparo à Pesquisa do Estado do Amazonas (FAPEAM) (Pró-Estado Program—#002/2008, POSGRAD 2021 Program and PCTI-EMERGESAÚDE Program—#005/2020).

The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

All claims expressed in this article are solely those of the authors and do not necessarily represent those of their affiliated organizations, or those of the publisher, the editors and the reviewers. Any product that may be evaluated in this article, or claim that may be made by its manufacturer, is not guaranteed or endorsed by the publisher.