Cytochrome P450 2D6 profiles and anti-relapse efficacy of tafenoquine against Plasmodium vivax in Australian Defence Force personnel

ABSTRACT Plasmodium vivax infections and relapses remain a major health problem for malaria-endemic countries, deployed military personnel, and travelers. Presumptive anti-relapse therapy and radical cure using the 8-aminoquinoline drugs primaquine and tafenoquine are necessary to prevent relapses. Although it has been demonstrated that the efficacy of primaquine is associated with Cytochrome P450 2D6 (CYP2D6) activity, there is insufficient data on the role of CYP2D6 in the anti-relapse efficacy of tafenoquine. We investigated the relationship between CYP2D6 activity status and tafenoquine efficacy in preventing P. vivax relapses retrospectively using plasma samples collected from Australian Defence Force personnel deployed to Papua New Guinea and Timor-Leste who participated in clinical trials of tafenoquine during 1999–2001. The CYP2D6 gene was amplified from plasma samples and fully sequenced from 92 participant samples, comprised of relapse (n = 31) and non-relapse (n = 61) samples, revealing 14 different alleles. CYP2D6 phenotypes deduced from combinations of CYP2D6 alleles predicted that among 92 participants 67, 15, and 10 were normal, intermediate, and poor metabolizers, respectively. The deduced CYP2D6 phenotype did not correlate with the corresponding participant’s plasma tafenoquine concentrations that were determined in the early 2000s by high-performance liquid chromatography or liquid chromatography-mass spectrometry. Furthermore, the deduced CYP2D6 phenotype did not associate with P. vivax relapse outcomes. Our results indicate that CYP2D6 does not affect plasma tafenoquine concentrations and the efficacy of tafenoquine in preventing P. vivax relapses in the assessed Australian Defence Force personnel.

areas of Thailand in children less than 15 years of age (9).Overall, relapses account for ≥79% of recurring P. vivax (10).Multiple episodes of P. vivax infections have been shown to increase the risk of hospitalization and death (5).The frequency and burden of relapses highlight the importance of eliminating hypnozoites from an individual's liver when treating P. vivax infections.
Currently, the WHO recommends "radical cure" treatment using a 3-day course of a blood schizontocidal drug plus a 14-day course of primaquine (PQ) to target the hypnozoite reservoir in the liver and prevent relapses and further transmission (11).The 8-aminoquinoline drugs, which include PQ and tafenoquine (TQ), are the only class of drugs found effective against both P. vivax hypnozoites to achieve radical cure and P. falciparum gametocytes to reduce transmission (11,12).However, individual adherence to the 14-day PQ regimen is less than ideal, reducing the effectiveness of radical cure (13,14).
In 2018, the United States Food and Drug Administration and the Australian Therapeutic Goods Administration approved the use of TQ, in combination with a course of schizontocidal drug, as a radical cure for P. vivax and for malaria prophylaxis.TQ offers a promising alternative to PQ for radical cure as it has an elimination half-life of approximately 2 weeks (15) compared to PQ's half-life of approximately 4-6 hours (16,17), effectively eliminating the requirement for daily drug dosing.
TQ can also be taken weekly for up to 6 months as a chemoprophylaxis.Currently, it is not recommended in children (<18 years of age) and pregnant women.As with PQ, TQ is not used in individuals with glucose-6-phosphate dehydrogenase (G6PD) deficiency, a hereditary X-linked condition responsible for drug-induced hemolysis, the severity of which is directly linked to the individual's level of G6PD enzyme activity and the dose of the drug (18,19).
The human cytochrome P450 2D6 (CYP2D6) gene codes for the cytochrome P450 hepatic enzyme, which is responsible for the metabolism of 20%-30% of all clinical pharmaceuticals currently available (20,21).Polymorphisms identified in the CYP2D6 gene result in variations in the level of metabolism of drugs reliant upon this pathway.To date, over 160 allelic variants have been documented, which are separated into four categories: increased function (IF), fully functional (FF), reduced function (RF), and nonfunctional (NF).The combination of these allelic variants for an individual deter mines their CYP2D6 metabolic activity phenotype: ultrarapid metabolizer (UM), normal metabolizer (NM), intermediate metabolizer (IM), and poor metabolizer (PM) (22).
PQ metabolism in humans utilizes two pathways for metabolite conversion: the MAO-A pathway converting PQ to inactive carboxyprimaquine and the CYP2D6 pathway where hydroxylation generates phenolic metabolites with redox activity.The latter is believed to be responsible for the hypnozoiticide efficacy and hemolytic activity of PQ when there is a deficiency in the G6PD enzyme (23).There has been substantial evidence that CYP2D6 phenotype or genotype-predicted activity score is a significant factor in PQ metabolism and an IM or PM is at increased risk of PQ failure with P. vivax relapse (24)(25)(26).
In murine models, CYP2D gene knockout completely blocked the anti-liver stage activity of PQ, and this activity could be partially restored using humanized CYP2D6 knock-in mice (27).Like PQ, mouse models have shown that TQ's activity against liver-stage parasites was suppressed in CYP2D6 knocked-out mice, but partial activity was restored in humanized CYP2D6 knock-in mice although requiring a higher TQ dose (28).CYP2D6 knockout also affected the blood TQ exposure, though with a much less reduction compared to PQ (29).These studies suggest that the CYP2D6 gene may affect TQ pharmacokinetics and anti-liver stage activity, though to a lesser extent than PQ.
Two clinical studies have been conducted to investigate the influence of CYP2D6 activity on the anti-relapse efficacy of TQ (30,31).The first study showed that an individual's CYP2D6 metabolism status was not associated with TQ anti-relapse efficacy (30).The second study demonstrated that the frequency of recurrent P. vivax infections following TQ radical cure was not associated with participants' CYP2D6 activity scores (AS) (31).As the number of participants with CYP2D6-PM status in the TQ arm was zero in the first study and three in the second study, the impact of the human CYP2D6-PM phenotype on the anti-relapse efficacy of TQ remains unanswered.
The aim of this study was to determine whether participants' CYP2D6 activity status influences plasma TQ concentrations and the anti-relapse efficacy of TQ.We exam ined historical plasma samples collected from deployed Australian military personnel treated with TQ to genotype their CYP2D6 status and predict CYP2D6 activities.We then examined if previously determined plasma TQ concentrations and TQ anti-relapse efficacy were reliant on the deduced CYP2D6 phenotype (genotype predicted CYP2D6 activity score and metabolic status).Our results suggest that unlike PQ, TQ concentra tions and anti-relapse efficacy are likely not dependent upon the individual's CYP2D6 activity status.

RESULTS
This was a retrospective study using human plasma samples collected as part of clinical trials of TQ conducted between 1999 and 2001.Plasma samples from the participants who had experienced a P. vivax relapse (relapse group) and from a randomly selected subset of participants who did not experience a P. vivax infection (non-relapse group) after TQ treatment were used to amplify the full-length CYP2D6 gene and sequenced.Of the 174 plasma samples tested, 92 (53%) were successfully amplified to complete the CYP2D6 gene sequence to identify all single nucleotide polymorphisms of interest.Of these 92 samples, 89 (30 relapse and 59 non-relapse) had plasma TQ concentration data available from the earlier studies (32, 33) (Fig. 1).The three samples (one relapse, two non-relapse) that did not have plasma TQ concentrations were used only for CYP2D6 profile and statistical analysis, which did not require TQ concentrations.

CYP2D6 alleles and allele frequency
CYP2D6 sequence analysis of 92 samples revealed 174 alleles, of which 14 different alleles were identified: 11 each in the relapse and non-relapse groups (see Table S1).The predominant allele for both groups was *1, the wild-type fully functioning allele followed by *4, a nonfunctional allele.Overall, the wild-type allele accounted for 44% of alleles in the complete set of samples, with fully functional alleles accounting for 60%.Three gene duplication alleles (*2XN, 1.7%) were detected, and one of the original 174 samples had a gene deletion (0.57%).

Allele combinations and frequency
Table 1 displays the genotypes and frequencies observed within 92 participants.Among these participants, the wild-type genotype *1/*1 had the highest frequency (29.3%), followed by *1/*4 (9.8%) and *4/*4 (7.6%).The frequency of these major genotypes was not significantly different between the non-relapse and relapse groups (P > 0.05).The homologous and heterologous combination of alleles accounted for 59% and 41% of genotypes, respectively.Interestingly, individuals with a homologous combination of alleles had a significantly higher odds ratio of relapse (OR = 2.78, 95% CI: 1.08-7.18).The risk of relapse was even higher for individuals predicted to be IM, with a homologous allele combination (OR = 13.13,95% CI: 1.92-89.52).

Predicted CYP2D6 phenotype
Each participant's plasma sample had their CYP2D6 metabolism status classified based upon the allele combination as either UM, NM, IM, or PM and activity score (AS) calculated from 0 to 4. Among the 92 samples, 1, 67, 15, and 10 were classified as UM, NM, IM, and PM, respectively.The single UM sample was included in the NM group for statistical analysis.Fifty-four plasma samples had an AS >1 and 38 with an AS ≤1 (Table 1).

Plasma TQ concentration versus TQ dosing regimen
Participant's plasma TQ concentrations were compared between the three TQ dos ing regimens.Pairwise comparisons revealed significant differences between all three regimens (P < 0.001).Individuals receiving 400 mg daily for 3 days had higher plasma TQ concentrations (mean 637 ± 156 ng/mL, n = 34) than those of individuals receiving 200 mg daily for 3 days (325 ± 71.5 ng/mL, n = 39), who had higher plasma TQ concen trations than those of individuals receiving a loading dose of 200 mg daily for 3 days followed by a weekly 200-mg dose (236 ± 63 ng/mL, n = 16) (Fig. 2).No association was found between CYP2D6 metabolism status (NM, IM, PM) and plasma TQ concentration (P = 0.493) (Fig. 3), suggesting that CYP2D6 does not affect the plasma concentrations of TQ.

Relationship between plasma TQ concentration and relapse status
Plasma TQ concentrations were compared between participants in the relapse and non-relapse groups.The mean plasma TQ concentrations between participants in the relapse and non-relapse groups were 620 ± 135 ng/mL (n = 10) and 644 ± 166 ng/mL (n = 24), respectively, for the 400-mg TQ daily dose for 3 days.The corresponding values for the 200-mg TQ daily dose for 3 days were 322 ± 89.7 ng/mL (n = 12) and 326 ± 63.7 ng/mL (n = 27).The corresponding values for the loading dose of 200 mg TQ daily for 3 days followed by 200 mg TQ weekly were 221 ± 43.4 ng/mL (n = 8) and 252 ± 78.1 ng/mL (n = 8), respectively.No statistically significant difference was observed in participant's plasma TQ concentrations between the relapse (395 ± 193 ng/mL, n = 30) and non-relapse (445 ± 204 ng/mL, n = 59) groups (P = 0.233, Fig. 4).

Full-Length Text
Antimicrobial Agents and Chemotherapy

Days to relapse and TQ dosing regimens
The number of days from the last TQ dose to relapse ranged from 67 to 354 days, with a mean of 158 days and a median of 139 days.There was no statistical significance between TQ dosing regimens and days to relapse (P = 0.496) (see Fig. S1 in the supple mental material).

CYP2D6 metabolism status and relapse status
There was no significant difference in proportions of CYP2D6 NM, IM, and PM between the relapse and non-relapse groups (P = 0.512, Table 1).There was also no significant association between AS and relapse (P = 0.592).Although the non-relapse group had a higher proportion with AS >1 compared to that in the relapse group (60.7% versus 54.8%), the difference did not reach significance (P = 0.148) (Table 1).These results indicate that CYP2D6 metabolic activity status did not affect the relapse outcome among the participants treated with TQ.

Relationship between CYP2D6 metabolic activity status, TQ dose, and plasma TQ concentration
A general linear model was developed to determine whether the combination of CYP2D6 metabolism status and TQ dose influenced an individual's plasma TQ concen tration.The model was significant (P < 0.001, R 2 68.5%); however, the only significant variable was TQ dose (P < 0.001).This suggests that after accounting for TQ dose, CYP2D6 metabolism status did not affect the plasma TQ concentration.

DISCUSSION
This study aimed to investigate whether CYP2D6 affects plasma TQ concentration and anti-relapse efficacy of the new 8-aminoquinoline drug TQ.To achieve this goal, archived plasma samples from groups of Australian Defence Force (ADF) personnel deployed to PNG and Timor-Leste during 1999 and 2001 who participated in TQ presumptive anti-relapse therapy (PART, also known as terminal prophylaxis or postexpo sure prophylaxis) and prophylaxis studies were used to determine CYP2D6 genotype and deduced phenotype.Plasma TQ concentrations were available from earlier clinical studies and were used in this study to determine the association with CYP2D6 pheno types.The participants deployed to PNG and Timor-Leste were pooled in the analysis for the same reasons outlined in our earlier CYP2D6 study of PQ (34).
The study initially consisted of 174 plasma samples stored since 1999; however, only 92 of these samples yielded good quality and quantity of genomic DNA allowing for a successful amplification of the entire CYP2D6 gene and subsequent sequencing.The remaining 47% of samples failed to amplify either the entire or some fragments of the gene.This was likely due to degradation of DNA over the 20-year plus storage period, despite the samples being stored at −80°C at the Australian Defence Force Malaria and Infectious Disease Institute (ADFMIDI).Furthermore, plasma samples are not an ideal source of genetic material compared to blood samples.Note that the plasma TQ concentrations measured in the early 2000s in samples collected from the 89 participants provide evidence of drug adherence in the PART (30) and chemoprophylactic (31,33) studies of TQ.
For those participants from the relapse cohort, the shortest duration until relapse following TQ dosing was 67 days.Therefore, all relapses are considered true relapses of the previous P. vivax infection and not new infections since the relapse time exceeds four terminal elimination half-lives of TQ and the participants had returned to Australia prior to relapse.Our study also identified and included participants with CYP2D6 IM and PM, fulfilling the gap left by previous studies where no or a low number of subjects with PM status had been identified in the TQ arm (30,31).This, combined with the measurement of plasma TQ concentrations in this study, provides the opportunity to detect possible interactions between CYP2D6 status and TQ anti-relapse efficacy.This study provided several lines of evidence demonstrating that CYP2D6 was not a major contributor affecting plasma TQ concentrations and anti-relapse efficacy of TQ for radical cure and chemoprophylaxis.Firstly, the deduced CYP2D6 AS and metabolism status did not correlate with plasma TQ concentration.Instead, plasma TQ concentra tions were shown to be directly dependent on the TQ dose regimen.Secondly, CYP2D6 genotype frequency, the deduced CYP2D6 AS, and metabolism status did not correlate with P. vivax relapses.Although the non-relapse group had a higher proportion of UM and NM and a lower proportion of IM and PM than the relapse group, the difference did not reach statistical significance.
These findings were reinforced by the linear modeling between participants' TQ dosing regimens and plasma TQ concentrations against relapse outcomes, which showed a relationship between dosing regimens and plasma TQ concentrations, but not between CYP2D6 status and plasma TQ concentrations or between CYP2D6 status and relapse outcomes.
It is noteworthy that, although plasma TQ concentrations were dose-dependent in the PART studies (35), no statistical difference between the two dosing regimens and the risk of P. vivax relapses was found.This was unexpected as TQ's anti-relapse efficacy has a clear dose-response relationship as shown by pooled individual data obtained from preregistration trials of TQ for radical cure (36).Watson et al. (36) using dose-response models showed that the FDA-and TGA-approved single-fixed 300-mg TQ adult dose combined with chloroquine (1,500-mg base given over 3 days) has suboptimal radical cure efficacy, with a mean recurrence proportion of 15.3% (95% CI: 9.6 to 21.7) at 4 months after follow-up.Increasing the fixed dose of TQ to 450 mg reduces the risk of P. vivax recurrence by more than two fold, with a mean recurrence proportion of 6.2% (95% CI: 3.2 to 11.5).Unlike the multicenter Phase 3 studies for radical cure conducted between 2010 and 2017 in 1,073 patients across nine countries (36), in the PART studies (30), the total TQ dose administered to the participants over three consecutive days of 600 mg and 1,200 mg was well above the recommended 450 mg fixed dose, with relapse rates of 5.0% and 6.3%, respectively.As there are no data available on the minimum inhibitory concentrations that prevent relapses of P. vivax infections and variability in hypnozoite susceptibility, the determination of an effective TQ dose for both radical cure and PART is still elusive.
Curiously, the study identified that participants with homologous CYP2D6 alleles had an increased risk of relapse compared to individuals with heterologous CYP2D6 alleles.This is mainly due to a higher risk of relapses in individuals who are predicted to have IM status with a homologous combination of alleles compared to those with a heterologous combination of alleles.In addition, all individuals with PM status resulted from a homologous combination of alleles.Nevertheless, the overall CYP2D6 metabolic status and AS were not associated with the risk of relapse.Another possibility could be due to possible unknown linkages of other genes to CYP2D6 that affect the anti-relapse efficacy of TQ.The relatively small sample size, particularly individuals with PM and IM status could also be a contributor.Further studies with larger sample sizes may provide a clearer answer.
As this was a retrospective study, not originally designed to investigate the relation ship between CYP2D6 and TQ anti-relapse efficacy, the requirement to use plasma samples severely reduced the number of available participants due to poor DNA amplification for sequencing.This may have limited the statistical power of detecting a significant relationship.Future studies with larger sample sizes of blood samples from populations of diverse CYP2D6 activity status would be warranted.
In conclusion, unlike PQ where anti-relapse efficacy is affected by the CYP2D6 metabolic activity status, plasma TQ concentration and anti-relapse efficacy appear to be independent of CYP2D6 activity.While both PQ and TQ's radical cure are restricted to patients with normal G6PD activity, TQ's much longer half-life and non-reliance on CYP2D6 compared to PQ would enhance the adherence and coverage of TQ for radical cure and PART in settings where G6PD phenotypes can be readily screened.

FIG 1
FIG1 Flowchart illustrating the study design, number of participants, and plasma samples collected from the participants.

FIG 2 FIG 3
FIG 2 Box plot and violin plot of the distribution of plasma TQ concentrations in participants receiving three different TQ regimens.The boxes indicate the first and third quartiles, the median is indicated by the horizontal line, and the mean is indicated by a solid black square.The violin plot indicates the distribution.

FIG 4
FIG 4 Comparison of plasma TQ concentrations in the relapse and non-relapse groups.The mean and 95% confidence intervals are represented as horizontal lines.

TABLE 1
CYP2D6 genotype and corresponding deduced metabolic status (MS), activity score (AS), and frequency in participant groups b UM, NM, IM, and PM represent ultrarapid, normal, intermediate, and poor metabolizers, respectively.Homo and hetero represent homologous and heterologous combinations, respectively.
a OR is the odds of relapse for homologous genotypes compared to heterologous genotypes.b