Introduction

Malaria continues to be a serious public health problem that affects the tropical and subtropical regions of the world. In 2022, an estimated 249 million cases and 608,000 deaths were attributed to this disease. The Bolivarian Republic of Venezuela, Brazil, and Colombia accounted for 73% of all malaria cases in the Americas. In this region, P. vivax is the predominant species, responsible for 72% of all malaria cases1. In Brazil, nearly all malaria cases (99.9%) occur in the Amazon region, where P. vivax is responsible for approximately 83% of reported cases2. An obstacle to the elimination of P. vivax is the recurrence of malaria caused by this species3,4.

The hypnozoiticidal drug primaquine (PQ) is recommended for the radical cure of vivax malaria, but first it needs to be biotransformed to exert its antimalarial effect5. The prodrug is mainly metabolized by monoamine oxidase A (MAO-A) into aldehyde, which is then oxidized to carboxyprimaquine, the major metabolite found in plasma6. Additionally, the metabolism of PQ is dependent on the cytochrome P450 2D6 (CYP2D6) enzyme, with its therapeutic effectiveness and potential adverse effects being associated with the production of hydroxylated derivatives (OH-PQm)7,8,9.

The metabolic activity of key enzymes involved in drug metabolism is influenced by genetic variability, which can significantly impact therapeutic outcomes10. CYP2D6 is a 497-amino acid enzyme from the cytochrome CYP450 superfamily, responsible for metabolizing 20 to 25% of all clinically used drugs. The gene encoding CYP2D6 is located on chromosome 22q13.1 and exhibits significant polymorphism, affecting drug metabolism11. Polymorphisms in the CYP2D6 gene contribute to inter-individual and inter-ethnic differences in drug metabolism12. Monoamine oxidase A (MAO-A) is a mitochondrial isoenzyme that catalyzes the oxidative deamination of dietary amines and neurotransmitters13. The MAOA gene, located on the short arm of the X chromosome (Xp11.23-Xp11.43), consists of 15 exons14. Variants in the MAOA gene have been associated with altered enzymatic activity and various clinical outcomes15,16,17. Similarly, UGT2B7, an isoform of UDP-glucuronosyltransferases encoded by a gene located on chromosome 4q13, catalyzes the glucuronidation of a wide range of endogenous and exogenous compounds. Polymorphisms in UGT2B7 are known to influence the metabolism of several therapeutic agents18,19.

Bennet et al. were the first to report the therapeutic failure of PQ against vivax malaria due to variation in the CYP2D6 gene, which resulted in decreased drug metabolism and subsequent inefficacy20. Furthermore, therapeutic failure of PQ as result of genetic variation in CYP2D6 has also been observed in the Brazilian Amazon21. Chamnanphon et al. found an association between the risk of P. vivax recurrence and decreased CYP2C19, ABCG2, and UGT2B7 activity in combination with intermediate or poor CYP2D6 metabolizer status22. Although the exact role of UGT2B7 in the metabolic pathways of PQ is not fully understood, there is evidence suggesting its involvement in the metabolism of similar compounds and in the phase II conjugation of drugs and related substances. A study highlighted that the formation of the N-carbamoyl glucuronide of PQ, a significant metabolite, may be associated with the activity of UGTs, including UGT2B723.

In vitro studies, MAO-A inhibitors were found to decrease carboxyprimaquine levels24. Furthermore, Ariffin et al. observed that the metabolism of PQ was slower in individuals with homozygous recessive genotype variants when compared to the wild-type genotype in MAOA25. Thus, this study aimed to investigate the influence of genetic variations in CYP2D6, MAOA, and UGT2B7 on the recurrence of P. vivax.

Results

Characteristics of the study population

The study included 72 participants: 18 cases and 54 controls. Of the individuals included 70 were genotyped for MAOA, 72 for UGT2B7, and 61 for CYP2D6. In both groups, the gender distribution was similar. The mean age in the case and control group was similar with 40 years (p = 0.8919). Regarding episodes of malaria recurrence, 77.8% of the patients had only one episode of recurrence during the follow-up (Table 1).

Table 1 Demographic and clinical characteristics of the study subjects.
Full size table

Frequency of CYP2D6, MAOA and UGT2B7 alleles

There were comparable distributions of CYP2D6 star alleles between the groups, with no statistically significant differences (p > 0.05), except for the reduced activity allele *4, which was present in 16.7% of the cases and 4.6% of the controls (p = 0.019). No statistically significant differences were observed for the MAOA (p = 0.159) and UGT2B7 alleles (p = 0.540) (Table 2). Additionally, all SNP frequencies were consistent with Hardy-Weinberg equilibrium, with p-value > 0.05 for each SNP (Supplementary Table S1).

Table 2 CYP2D6, MAOA and UGT2B7 allele frequencies.
Full size table

Frequency of predicted CYP2D6 phenotypes and MAOA and UGT2B7 genotypes

A statistically significant disparity was detected in the gNM, evident in 50.0% of cases and 75.6% of controls (p = 0.039). However, the MAOA and UGT2B7 genotypes exhibited no statistically notable variances between the case and control groups (p > 0.05) (Table 3).

Table 3 Frequency of predicted CYP2D6 phenotype, MAOA and UGT2B7 genotypes.
Full size table

Time to first recurrence

Most recurrences occurred between 61 and 120 days (44.4%) of the initial malaria episode (Table 1). The predicted CYP2D6 phenotypes were grouped into normal metabolizers (gNM) and decreased intermediate/poor (gIM/gPM) metabolizers, and MAOA and UGT2B7 were grouped into reference and alternate genotype (heterozygotes plus alternate homozygotes). Significant difference was observed between the time elapsed from the first episode of vivax malaria to recurrence in the groups to MAOA (Fig. 1) (p = 0.0437).

Fig. 1
figure 1

Kaplan–Meier curve for the length of time for malaria recurrence between individuals with normal and reduced CYP2D6 activity phenotype, reference and alternate MAOA and UGT2B7 genotypes. (A) CYP2D6 phenotypes; (B) MAOA genotypes; (C) UGT2B7 genotypes.

Full size image

Clearance of gametocytemia

Following treatment, 52.6% of the participants demonstrated clearance of gametocytemia by D3 (Table 1). No difference was observed in sexual parasite clearance associated with CYP2D6 phenotypes (p = 0.1754), MAOA genotypes (p = 0.4633) and UGT2B7 genotypes (p = 0.8341) (Fig. 2).

Fig. 2
figure 2

Kaplan–Meier curve for malaria gametocyte clearance between individuals with normal and reduced CYP2D6 activity phenotype, reference and alternate MAOA and UGT2B7 genotypes. (A) CYP2D6 phenotypes; (B) MAOA genotypes; (C) UGT2B7 genotypes.

Full size image

Relative risk (RR) of recurrence

The relative risk (RR) of P. vivax recurrence for gIM and gPM of CYP2D6 was 2.13 (95% CI 0.95–4.78; p = 0.066). Heterozygous + alternate homozygous MAOA (CC and CT) and UGT2B7 (AG and GG) genotypes showed RRs of 0.52 (95% CI 0.23–1.14; p = 0.106) and 1.34 (95% CI 0.56–3.20, p = 0.503) respectively compared to individuals carrying the reference genotype (Table 4).

Table 4 Relative risk for recurrence associated with CYP2D6 phenotype, MAOA and UGT2B7 genotypes.
Full size table

Plasma PQ concentrations (ng/mL) on D7 relative to genotypes and predicted phenotypes

There were no significant differences in PQ levels on day 7 among the CYP2D6 predicted phenotype groups and MAOA and UGT2B7 genotypes (p > 0.05). Concentrations of PQ on D7 were.

142.0 ng/mL (IQR 97.2–196.0) in the case group carrying the MAOA CC + CT genotypes when compared to individuals carrying the MAOA wild-type TT 171.0 ng/ML (IQR 69.0–281.0), although with no statistical difference (p = 0.974) was observed. Likewise, the PQ concentration on D7 in the control group was 159.0 ng/mL (IQR (80.19–198.62)) for gIM + gPM CYP2D6 phenotypes and 140.0 ng/mL (IQR 120.49–179.11) for gNM CYP2D6 phenotypes (p = 0.949) (Table 5; Fig. 3).

Table 5 Parasitemia clearance, drug levels, and malaria recurrence patterns between CYP2D6 predicted phenotype, MAOA and UGT2B7 genotypes.
Full size table
Fig. 3
figure 3

Blood primaquine levels among subjects with normal and reduced CYP2D6 activity phenotype, reference and alternate MAOA and UGT2B7 genotypes. (A) CYP2D6 phenotypes; (B) MAOA genotypes; (C) UGT2B7 genotypes.

Full size image

Discussion

Genetic variability in drug-metabolizing enzymes affects about 30% of all drugs10. The CYP2D6 gene presents a high allelic heterogeneity, which results in significant inter-individual variations that can be found in many geographical locations and ethnic groups. According to Llerena et al., the CYP2D6 *4 allele is more prevalent in Europeans, CYP2D6 *10 in Asians, CYP2D6 *41 in Middle Eastern populations, CYP2D6 *17 in Black Africans, and CYP2D6 *29 in African Americans26. In Brazil, CYP2D6 *1 is common in 38% of the population, followed by CYP2D6 *2 (21.5%), CYP2D6 *4 (9.4%), CYP2D6 *17 (5.6%) and CYP2D6 *41 (5.5%) alleles. The most frequent CYP2D6 predicted phenotype is gNM (Normal metabolizer) (83.5%), followed by the gUM (Ultrarapid metabolizer) phenotype (3.7%) and gPM (Poor metabolizer) (2.5%)27. The frequency of mutated alleles for CYP2D6 found in our study is consistent with other studies involving Brazilian populations21,27,28.

The findings of this study demonstrated a risk of recurrence of vivax malaria in patients that are carriers of the CYP2D6*4 allele and who receive PQ treatment. Furthermore, the predicted normal phenotype (gNM) was more frequent in the control group than the case group. These results were agree with observations by Bennett et al., in which patients with multiple recurrences had an intermediate/poor metabolic phenotype, unlike the non-recurrent individuals with the normal phenotype20. Park et al. reported that individuals with gPM and gIM phenotypes have an increased risk of recurrence of P. vivax infection compared to individuals with gNM and gUM phenotypes29. This study observed increased relative risk of P. vivax malaria in relation to genetic variations in the CYP2D6, specifically comparing gIM and gPM phenotypes. Other studies performed in the Amazon region have shown that individuals with reduced CYP2D6 enzymatic activity had frequent recurrences21,28,30.

The MAO-A enzyme plays an important role in catalyzing several biological pathways, including the metabolism of PQ6,31. In the case of the MAOA gene 1460T > C [rs1137070], the frequencies in Latin America indicate that the C allele occurs in 62.79% of the population, while the T allele is found in 37.21%32. Notably, MAO-A is involved in the formation of carboxyprimaquine, the primary metabolite of PQ6,8,31. While carboxyprimaquine itself lacks direct antimalarial activity, evidence suggests that it may undergo further phase I metabolism mediated by CYP enzymes, producing hydroxylated and quinoneimine metabolites33. In the MAOA 1460 T > C SNP (rs1137070), the C allele of rs1137070 lacks a restriction endonuclease site and is associated with lower MAO-A activity34, but data on the relationship between decreased enzyme activity and recurrences of vivax malaria are scarce. Popovici et al. proposed that polymorphisms that reduce MAO-A activity could lead to a higher concentration of free PQ available for metabolism by CYP2D6, resulting in fewer recurrences of P. vivax35. However, Puça et al. suggested that individuals with altered MAO-A upstream Variable Number Tandem Repeat (uVNTR) may exhibit an additive effect in conjunction with dysfunctional CYP2D6, contributing to the recurrence of P. vivax36.

The survival curve analysis showed that individuals with the mutated MAOA genotype (CC + CT) had a significantly shorter time to recurrence of vivax malaria compared with those with the wild-type genotype (TT), with a p-value of 0.0437. This result suggests that the MAOA gene mutation may be associated with increased susceptibility to early recurrence of P. vivax malaria. Although no direct association between P. vivax recurrences and MAOA genotypes was observed in the overall analysis, it is noteworthy that the frequency of the CC genotype was higher in the case group (63.2%) compared with the control group (60.8%). Ariffin et al. found an association between slower primaquine metabolism in individuals with the MAOA 891G > T variant carrying the alternative homozygous (TT) genotype when compared to the wild-type genotype (p = 0.05; 95% confidence interval − 0.04; −1.27)25. These findings underscore the need to consider specific genetic profiles, such as mutated MAOA, when assessing recurrence risk in patients with P. vivax malaria.

One recent study reported that an increased risk of P. vivax recurrences was associated with decreased activity of CYP2C19, ABCG2, and UGT2B7 in combination with intermediate or poor metabolism of CYP2D6 in patients treated with PQ and CQ22. UDP-glucuronosyltransferase (UGT) enzymes comprise a superfamily of key proteins that catalyze the glucuronidation reaction on a wide range of structurally diverse endogenous and exogenous chemicals18. In Latin America, the frequency of allele A for the UGT2B7 gene (372 A > G [rs28365063]) is 87.73%, while allele G is present in 12.27%37. In the present study, no difference was found in the frequency of UGT2B7 genotypes and alleles between the case group and the control group (p > 0.05).

Assessment of blood PQ levels can provide reliable information about exposure to antimalarials during treatment38. This study found no association between PQ concentrations, predicted CYP2D6 phenotypes, and MAOA and UGT2B7 genotypes (p > 0.05). However, previous studies have demonstrated that decreased CYP2D6 activity results in lower levels of active metabolites7,20. Baird et al. analyzed the dextromethorphan/dextorphan (DM/DX) ratio, which is used to assess CYP2D6 activity, and reported that patients with a higher DM/DX ratio were more prone to recurrence, indicating that reduced CYP2D6 activity results in lower efficacy of PQ39.

This study had limitations, such as its small sample size, lack of supervised PQ treatment, reliance on inferred genotypes to predict CYP2D6 phenotypes, difficulty in measuring carboxyprimaquine, and absence of microsatellite analysis to classify recurrences. Additionally, the limited selection of genes represents a significant constraint, as it does not allow for a comprehensive investigation of genetic factors potentially involved in PQ treatment failure. It is important to note that our study does not molecularly differentiate between relapses and reinfections, which is why we intentionally use the general term “recurrence.” However, we adopted a minimum threshold of 29 days for recurrence to minimize the likelihood of recrudescence, considering the persistence of schizonticidal drugs in circulation. Furthermore, our follow-up period was limited to 180 days to reduce the probability of reinfection.

Although this study identified associations between genetic variants and P. vivax recurrence, it is important to acknowledge that the analysis of CYP2D6, MAOA, and UGT2B7 does not allow for a complete assessment of PQ therapeutic failures. Future research should address these gaps by employing larger sample sizes, supervised PQ treatment to minimize variability, and direct phenotyping methods to improve accuracy in CYP2D6 assessment. Improved metabolite measurement techniques and microsatellite analysis are also necessary to distinguish reinfections from relapses. Expanding genetic analyses to include a broader set of candidate genes, encompassing both host and parasite factors, would provide a more complete understanding of PQ treatment outcomes. Furthermore, the combination therapy of PQ with schizonticides complicates the evaluation of PQ’s specific efficacy in clearing hypnozoites. Future studies could consider designs that isolate PQ’s effects, such as monotherapy trials or stratified analyses, to clarify its genotype-specific impact. Addressing these aspects will enhance understanding of the genetic and therapeutic factors influencing PQ treatment outcomes.

Tafenoquine, a new 8-aminoquinoline with a longer half-life than PQ, emerges as a promising alternative for the treatment and prevention of malaria recurrences. Recent clinical trials have shown that tafenoquine demonstrates efficacy comparable to PQ, particularly at lower doses (total dose of 3.5 mg/kg)40,41. Its key advantage lies in its single-dose regimen, which enhances treatment adherence. In a study comparing tafenoquine to supervised PQ administration, both drugs were found to have similar safety profiles41. Furthermore, a more recent study highlighted that single-dose tafenoquine was more effective than PQ in preventing P. vivax recurrences, further reinforcing its potential as an effective and convenient option for malaria management42.

In conclusion, the results suggest that individuals with normal metabolism status have a lower risk of recurrence of P. vivax malaria. Furthermore, individuals carrying the CYP2D6*4 allele have a higher risk of PQ therapeutic failure and, consequently, a greater likelihood of suffering recurrent episodes of vivax malaria. While these findings are preliminary, they highlight the relevance of assessing CYP2D6 in malaria-endemic regions and suggest the need for a more personalized approach in treatment to prevent disease recurrence. Furthermore, our data on the MAOA gene indicate that mutations in this gene could be associated with increased susceptibility to recurrence, underscoring the importance of considering multiple genetic factors in evaluating recurrence risk. However, these observations require confirmation in larger studies to better understand the underlying molecular mechanisms and to explore potential alternative therapeutic strategies.

Methods

Ethics statement

This study was approved by the Research Ethics Committee of the FMT-HVD, under CAAE 69476017.3.0000.0005 and approval number 2.659.845. All research was conducted following the applicable guidelines and regulations.All participants were informed about the objectives of the study and obtained an informed consent form from all participants and/or their legal guardians. In the case of minors, the parents or legal guardians signed the consent form.

Study subjects

This case-control study used convenience sampling from a previous follow-up of a cohort of patients, which was conducted between 2018 and 2021 at Fundação de Medicina Tropical Doutor Heitor Vieira Dourado (FMT-HVD). The FMT-HVD is a reference center for the treatment of tropical diseases in the Brazilian Amazon, Manaus, state of Amazonas. Individuals of either gender, of over 6 months of age, weighing > 5 kg, symptomatic, with a P. vivax mono-infection and a microscopy-confirmed parasite density of between 250 and 100,000 parasites/µL were included. The exclusion criteria were the use of antimalarials in the last 60 days, P. vivax - P. falciparum mixed infection, pregnancy or breastfeeding, and concomitant or underlying diseases. These patients were treated with schizonticides administered concomitantly with 0.5 mg/kg/day of PQ administered for 14 days, with PQ doses being adjusted according to the patient’s weight. It is important to note that in Brazil, the 7-day regimen is used to improve adherence to PQ, which is reinforced by the fact that the 14-day regimen does not demonstrate superiority. However, the prescription of 14-day PQ regimens is indicated in cases where adherence can be monitored in an attempt to ensure greater clinical efficacy43,44. Clinical and laboratory tests were performed on the day of admission (D1) and on all other follow-up visits (D2, D3, D4, D7, D14, D28, D42, D63, D90, D120, D150 and D180). Participants were divided into case and control groups, with the case group consisting of patients who presented recurrence between 28 and 180 days of follow-up, and the control group of individuals who didn’t present a new episode of malaria during the same follow-up period. The recurrence of P. vivax was defined as a new episode diagnosed microscopically, occurring within a period ranging from 29 to 180 days after the initial episode. We opted to work with a general definition of recurrence, as no genetic or genomic analyses of the parasites were conducted to differentiate between relapse, recrudescence, and reinfection. The control group was matched to the case group by gender, mean age, ethnicity and treatment.

Laboratory procedures

Genomic DNA was extracted from blood samples using the QIAmp Blood Mini kit (Qiagen, Hilden, Germany) according to the manufacturer’s protocol. The selection of MAOA and CYP2D6 was based on previous studies that investigated their influence on the response to PQ21,25. The inclusion of UGT2B7 was supported by evidence suggesting that decreased activity of this enzyme, particularly when combined with CYP2D6 intermediate or poor metabolizer status, may increase the risk of P. vivax relapse22. Eleven single-nucleotide polymorphisms (SNPs) in CYP2D6 were genotyped. These included 2549delA [rs35742686], 2615_2617delAAG [rs5030656], 100 C > T [rs1065852], 1846G > A [rs3892097], 4180G > C [rs1135840], 2988G > A [rs28371725], 3183G > A [rs59421388], 31G > A [rs769258], 1023 C > T [rs28371706], 1584 C > G [rs2820985], 2850 C > T [rs16947]. One polymorphism in MAOA 1460T > C [rs1137070] and one polymorphism in UGT2B7 372 A > G [rs28365063] were also genotyped by real-time PCR using Applied Biosystems 7500 Fast Real-time PCR System software, with specific TaqMan probes for each SNP assay (Applied Biosystems, Foster City, CA, USA). All amplification reactions and cycling parameters were set according to the manufacturer’s protocols. In real-time PCR with TaqMan probes, the zygosity of a SNP is determined by the amplification pattern of the VIC and FAM probes. Amplification of both probes indicates heterozygosity, while amplification of a single probe indicates homozygosity for the SNP. The CYP2D6 gene copy number variation (CNV) was also assessed using TaqMan probes followed by real-time duplex PCR. Relative quantification of the CNV was performed using CopyCaller v2.1 software (Applied Biosystems, Foster City, CA, USA). Median CT values were used for copy number calculations, following the manufacturer’s guidelines. CT thresholds were manually set at 0.2 for consistency across all runs. The corresponding CT values for each sample are provided in Supplementary Table S2.

Determination of alleles, genotypes, and haplotypes

Genotypes and alleles of MAOA and UGT2B7 were inferred using Microsoft Office Excel v16.16.1, without evaluating the phenotype, since there are no guidelines for predicting enzymatic activity. CYP2D6 haplotypes were inferred using the HaploStats package (version 1.7.7) implemented in the R platform (www.r-project.org) and compared with the star allele nomenclature to identify allelic variants (*)45. To ensure consistency and reliability, the widely recognized guidelines of the Clinical Pharmacogenetics Implementation Consortium (CPIC), which serve as the standard reference for scoring CYP2D6 activity, were applied in this process.

Measurement of primaquine levels

Primaquine was quantified in whole blood samples using high-performance liquid chromatography coupled with a diode array detector (HPLC-DAD), performed on a Flexar system from PerkinElmer Inc., Boston, MA, USA. The analyte was extracted from whole blood through liquid-liquid extraction with methyl tert-butyl ether at pH 3. Chromatographic separation was achieved on an RP-18 column (15 cm x 4 mm i.d., PerkinElmer Inc.) using a mobile phase composed of acetonitrile and phosphate buffer (pH 3.5) in a 30:70 (v/v) ratio, at a flow rate of 1.0 mL/min38,46. The method demonstrated linearity within the concentration range of 50 to 900 ng/mL, with detection and quantification limits of 20 and 30 ng/mL, respectively. No significant interferences from mefloquine, chloroquine, desethylchloroquine, carboxymefloquine, or acetaminophen were observed in the analyte detection.

Activity score and CYP2D6 phenotype

The activity score (AS) system was used to translate the CYP2D6 genotype into the phenotype from the sum of the scores assigned to the star (*) alleles according to the functionality of each one. Here, zero (0) depicts null function alleles (*3, *4, *4xN, *5, *6, *7, *8, *11, *12, *36, *40, *42, *56), 0.5 depicts decreased function alleles (*17, *29, *44, *49) and 0.25 (*9, *10, *41), 1 indicates normal function alleles (*1, *2, *35, *43, *45, *39), and 2 indicates increased function alleles (*1xN, *2xN). These scores designated the predicted phenotypes as either poor, intermediate, normal and ultra-rapid metabolizers (gPM, gIM, gNM and gUM, respectively)47,48,49, as shown in Supplementary Table S3. The CNV data were integrated into the calculation of the CYP2D6 activity score by adjusting the scores assigned to alleles with variations in copy number. If an individual has a gene duplication at the CYP2D6 locus, the additional copy is also counted towards the total activity score.

Statistical analysis

Genotype and allele frequencies were assessed using the chi-square test or Fisher’s exact test. The SNP frequencies were tested for deviations from Hardy-Weinberg equilibrium (HWE) using the online calculator by Michael H. Court (2005–2008). The observed counts were compared to the expected estimates, and deviations were computed for each genotype. The sum of these deviations was used to calculate the chi-square test statistic in Excel, enabling a statistical evaluation of the population’s adherence to Hardy-Weinberg equilibrium. Detailed results, including p-values and allele frequencies for each SNP, are provided in Supplementary Table S1. The influence of genotypes and phenotypes on time to recurrence and clearing of sexual forms was estimated using Kaplan-Meier survival analysis. The association between recurrences and the studied genetic variants was determined using multiple log-binominal generalized linear regression. Statistical analyses were performed using STATA v.13. The graphics of drug dosages were performed using GraphPad Prism v. 9.0.0. Statistical significance was considered as p < 0.05 in all analyses.