For analysis, HPLC grade solvents, e.g., methanol, acetonitrile, and formic acid (FA), were purchased from Thermo Scientific (Rockford, IL, United States). Water used for analysis was purified using Millipore Synergy® UV Ultrapure Water Purification System (EMD Millipore Corporation, MA, United States). Reference standards, primaquine diphosphate was purchased from Sigma (St Louis, MO, United States). Different metabolites of PQ and internal standards were synthesized at the National Center for Natural Products Research, University of Mississippi. Synthesis of reference standards have been described in the previous article; primaquine-5,6-orthoquinone (POQ) (Potter et al., 2015), carboxyprimaquine-5,7-orthoquinone (cPOQ) (Khan et al., 2021), PQ-N-CG (Fasinu et al., 2016), cPQ (McChesney and Sarangan, 1984), 4-methylprimaquine-5,6-orthoquinone (4-MePOQ), and deuterated primaquine (d3-PQ) (Khan et al., 2022b). The racemic PQ, PQ enantiomers and placebo capsules used for the clinical study were prepared with complete characterization, purity, and stability, by ElSohly Laboratories, Inc (ELI), Oxford, MS. Each capsule contained the desired specific amount of PQ base as the diphosphate salt. Structure of all analytes are shown in Figure 1.

All the stored plasma and RBCs samples were thawed in ice and processed to quantify analytes. PQ, cPQ, and PQ-N-CG by a validated ultra-high performance liquid chromatography coupled with Xevo G2-S QToF mass spectrometer (UHPLC-MSMS) (Waters Corp., Milford, MA, United States) (Khan et al., 2022a). As the levels of POQ and cPOQ in plasma were very low, and QToF was not sensitive enough, these two analytes were quantified using Xevo TQ-S triple quadrupole mass spectrometer (UHPLC-TQ-MS, Waters Corp., Milford, MA, United States). The method for quantification of these two analytes was validated as per USFDA guidelines (USFDA, 2018). The method for extraction and analysis of POQ and cPOQ in red blood cells (RBCs) was previously described (Khan et al., 2021). Analysis was carried out using gradient elution and detected by electrospray ionization (ESI) in positive mode. Chromatographic separation was performed using a reversed-phase column (Acquity UPLC HSS T3, 1.8 µm, 2.1 × 100 mm, Waters, United States) set at 35°C. The mobile phase consisted of solvent A (water with 0.05% formic acid) and solvent B (acetonitrile with 0.05% formic acid, v/v). For the analysis of plasma, gradient elution was started with 95% A (0 min), a linear gradient from 95% to 55% A (0–9 min), later up to 0% A (9–10 min). For analysis of RBCs, the gradient elution was started with 15% B at the beginning and increased to 55% B at 3.5 min. To detect analytes, ionization was achieved using electrospray ionization (ESI) operated in positive-ion mode (M + H). The MS/MS transition for POQ, cPOQ, PQ, cPQ, PQ-N-CG, 4-MePOQ and d3-PQ were 260.1 > 175.0, 275.2 > 175.0, 260.1 > 175.1, 275.1 > 175.1, 480.2 > 243.1, 274.2 > 189.0, and 263.2 > 178.1, respectively. All data collected in centroid mode were acquired using MassLynx™ NT 4.1 software. Accurate mass tolerance of molecular ion and major fragments was limited to 5 ppm, while minor fragments of parent ions were tolerated up to 10 ppm.

Blood samples for PQ and its metabolite analysis were drawn at baseline, then on days 3 and 5 pre-dose and at 1 h post-dose, and then on day 7 at 24 h after the last dose. These were analyzed as described above, and the relative plasma exposure of the different enantiomers and racemate were compared, along with the major metabolites.

Statistical analysis was performed using the GraphPad prism software version 22.0 (SPSS Inc., Chicago, IL, United States). All data, including PK and biochemical parameters, were summarized using descriptive statistics. Values are expressed as the mean ± SEM for all parameters.

Eighteen G6PDn subjects were enrolled and 15 completed all four arms, receiving RPQ, SPQ, RSPQ, and placebo, separately, in crossover fashion. In each cycle, subjects received one dose of PQ daily for 7 days. In between each cycle, subjects were given at least a 1 week washout period. The baseline demographic and clinical characteristics of the clinical study subjects of both categories are shown in Table 1. The mean age and body weight (±SEM) of G6PDn subjects were 22.0 ± 0.4 years and 168 ± 7 lb, respectively. No changes in BP, pulse, respiratory rate, oxygen saturation, body temperature and weight were observed. The most consistent findings in the G6PDn subjects were elevations in methemoglobin and very small increases in bilirubin, but neither resulted in clinical symptoms, nor required discontinuation of drug.

No serious adverse events (AEs) or withdrawals due to AEs were observed in G6PDn subjects. The adverse events and their incidence are tabulated in Table 2. Three subjects reported mild AEs. None of these three required withdrawal from the study. Two of the events, in the judgment of the study director and physician, were likely unrelated to the drug treatment. These two showed transient mild elevations in aspartate transaminase (AST) and alanine transaminase (ALT) levels, likely attributable to acute alcohol consumption in one subject (during the SPQ arm) and one with a latent past history of gallbladder trouble (increased AST/ALT during placebo arm). The third subject—who had the most prominent methemoglobin response, developed some dyspnea on exertion and mild cyanosis the evening after the 7^th dose of RPQ. The subject was seen in the clinic the following day, and dyspnea had resolved, and the subject was monitored for several days until methemoglobin returned to normal. This reaction did appear to be drug-related; though the subject did not report any exertional dyspnea with SPQ or RSPQ, methemoglobinemia increased prominently during each of these drug arms. This subject was followed by the study nurse and physician until symptoms dissipated (by the next day) and MetHgb returned to normal over several days. She was also examined by a local primary care physician, but no medical intervention was required. This subject was cleared by the physicians to continue in the study.

In G6PDn subjects, biochemical parameters were not significantly affected except for methemoglobin (MetHgB) and total bilirubin (Figure 2). Significant increases (p < .0001) in methemoglobin (MetHgB) from the baseline were seen in subjects receiving PQ enantiomers (at 22.5 mg), although the maximum values averaged about 5% of HgB, and were not associated with clinical symptoms. The increases in MetHgB were not different with the two enantiomers. Subjects receiving racemic primaquine (45 mg) showed a somewhat higher increase in MetHgB, but still only reached a mean of 7% HgB. During placebo administration, MetHgB did not change. From the baseline level, bilirubin progressively increased during daily PQ administration, but these increases were very modest and still under the normal clinical laboratory limits. There were no differences between the enantiomers, and RSPQ showed somewhat higher MetHgB response than the individual enantiomers. In this study, parameters for kidney, liver, and hematological profiles were also analyzed, and no significant changes were observed (Supplementary Tables S1, S2). Findings for vital signs are also tabulated in Supplementary Table S3.

During daily PQ administration for 7 days (days 0–6), blood samples were taken on day zero pre-dose, on day three pre-dose and at 1 h post-dose, on day five pre-dose and at 1 h post-dose, and on day seven, 24 h after the last dose. The plasma concentrations of PQ on days 3 and 5 were very low at pre-dose, consistent with our prior study (Khan et al., 2022a) and those of others (Mihaly et al., 1984), and increased 1 h after oral administration (Figure 3). Two major metabolites of PQ (Figure 3A), cPQ (Figure 3B) and PQ-N-CG (Figure 3C), were measured in plasma at these time intervals. When in the SPQ arm, the subjects showed high concentrations of PQ and low concentrations of cPQ; the reverse situation; low PQ and high cPQ was seen during the RPQ arm. When the racemic mixture RSPQ was administered, both PQ and cPQ were high, reflecting the contribution of each enantiomer. Although the dose of RSPQ was the sum of the individual enantiomers, levels of PQ and cPQ were not doubled, presumably reflecting at least in part, a higher apparent volume of distribution (Vd) for RPQ as compared to SPQ in mice (Fasinu et al., 2022), monkeys (Saunders et al., 2014) and humans (Khan et al., 2022a), and also an expected much lower Vd for cPQ as compared to PQ (Baker et al., 1984).

PQ-N-CG, which is formed through direct N-glucuronide conjugation of PQ, was observed only after SPQ administration. Though PQ-N-CG was not detected after RPQ administration, subjects that received RSPQ showed consistently higher plasma levels than those receiving SPQ (Figure 3C).

POQ and cPOQ are the orthoquinones of PQ and cPQ, respectively, formed secondarily to hepatic metabolism, and are believed to reflect the redox-active metabolite pathway for PQ. These two metabolites were measured in plasma and RBCs (Figure 4). POQ was detected in both plasma (Figure 4Ai) and red cells (Figure 4Bi) in the general range of 4–10 ng/ml with an approximate equal distribution between plasma and cells.

POQ levels in plasma were higher in RSPQ and SPQ than in RPQ arm samples as expected due to their higher levels of PQ. cPOQ was found at higher concentrations in plasma (range 30–60 ng/ml) and red cells (range 10–20 ng/ml) in subjects that received RPQ or RSPQ. This is consistent with the higher levels of cPQ in these subjects. Of interest, the concentration of cPOQ in plasma (Figure 4Aii) was notably higher than POQ in all arms of the study; whether this results from enhanced formation or slower elimination of cPOQ is unknown. Similarly, the concentration of cPOQ in RBC (Figure 4Bii) was markedly higher (ca. two-fold) than POQ in all arms, implying that cPQ may contribute more to the oxidant stress within red cells than PQ. Overall, RBC exposure to orthoquinone metabolites showed different profiles depending on which PQ enantiomer is dosed, but both enantiomers contribute to the potential for hemolytic injury.

We were able to enroll two healthy human volunteers with G6PDd in a pilot investigation for 5 days of dosing with the individual PQ enantiomers. One subject was a heterozygous female (A-variant with G6PD activity of 7.4 units/gram of hemoglobin), who received RPQ at a dose of 15 mg daily for 5 days, and after a 2 week washout, received SPQ at the same dose. The other subject was a hemizygous male (A-variant with G6PD activity of 1.2 units/gram of hemoglobin), who was also enrolled to receive both enantiomers. However, this subject (TS212) could not complete the 5 days of dosing with either SPQ or RPQ, since his bilirubin level was significantly increased, and dosing was discontinued after the third dose. Both G6PDd subjects showed progressive significant increase in bilirubin levels with both enantiomers (Figure 5); this was especially more prominent in the hemizygous subject, and reached the stop criterion for discontinuation of dosing after the third dose. Except for lactate dehydrogenase, no other biochemical parameters were changed (Supplementary Tables S4, S5).

The G6PDd subjects were closely monitored after dosing, and blood samples were collected at pre-dose and 2 h after administration on days 0, 1, 2, and 3, and then on day 4 (24 h after the last dose). In plasma, PQ, cPQ, PQ-N-CG, and cPOQ were quantified. Like G6PDn subjects, G6PDd subjects also showed a similar pattern of plasma exposure for PQ, cPQ, and PQ-N-CG (Figure 6). Again, PQ exposure in plasma was found to be somewhat higher with SPQ than RPQ, and RPQ showed around 37-fold higher exposure for cPQ than SPQ. POQ could be detected but not quantified in RBCs of G6PDd subjects, likely because of the lower PQ dose. But cPOQ was detectable in plasma and RBCs after RPQ administration (Figure 7A, B), respectively.

Although the 8-aminoquinoline class of antimalarials has an unique place in malaria chemotherapy, their optimal employment is limited due to safety concerns in individuals with G6PDd (Clyde, 1981; Baird and Rieckmann, 2003). It is established that this liability in G6PDd is due to the enhanced sensitivity of their red cells to oxidative stressors. In the case of PQ, this has long been attributed to redox active metabolites (Allahyari et al., 1984; Bowman et al., 2004; Ganesan et al., 2009; 2012), though the precise nature of the metabolites and their pathways/locations in vivo were for many years uncertain (Strother et al., 1984; Marcsisin et al., 2016; Flaherty et al., 2022). cPQ is well known as a major circulating metabolite, likely formed by the actions of monoamine oxidases (Nicholl et al., 1987; Fasinu et al., 2016). cPQ is not redox active. Other pathways are also now well established, including the hydroxylation of PQ by cytochrome P450 (CYP) pathways, observed using isolated enzyme systems, liver microsomes, and in intact human hepatocytes (Avula et al., 2013; Jin et al., 2014; Ariffin et al., 2019). Among these CYPs, CYP 2D6 accomplishes the formation of at least four different metabolites via hydroxylation of the PQ ring at 2-, 3-, four- and five- positions (Pybus et al., 2012; Fasinu et al., 2014; Potter et al., 2015). This CYP 2D6 activity has been shown to be an important determinant of the altered metabolism of PQ in animals (Pybus et al., 2013) and humans (Potter et al., 2015), and poor CYP 2D6 metabolizer phenotypes show a reduced anti-relapse efficacy of PQ (Bennett et al., 2013). These hydroxylated metabolites have proven difficult to detect in plasma, whether by virtue of instability, compartmentalization within tissues, or rapid conjugation and excretion (Mihaly et al., 1984).

Several studies have now shown that 5-OH-PQ is exceedingly unstable, rapidly oxidizing to its quinone-imine, then to 5,6-dihydroxy-PQ, and finally to the 5,6-orthoquinone (POQ) (Camarda et al., 2019; Fasinu et al., 2019). Even the more stable POQ is not typically observed in plasma, though it is detected readily in urine (Spring et al., 2019).

Recently, our group also reported that POQ can be formed directly from PQ on incubation with human red cells, presumably via heme- or hemoglobin-dependent formation of oxygen free radicals (Fasinu et al., 2019); the qualitative importance of this in situ generation of oxidants vis-a-vis that derived from hepatic metabolism is yet to be determined. We have also observed that cPQ can also give rise to an ortho-quinone type metabolite (cPOQ), presumably by analogous metabolic routes, but this does not appear to be accomplished by CYP 2D6 (Khan et al., 2021). The relative contributions of POQ and cPOQ to red cell oxidant stress and subsequent injury in humans is clearly of interest to resolution of the importance of the two enantiomers in the clinical use of PQ.

PQ is a racemic mixture of (R) and (S) enantiomers, and our prior studies have shown vastly different metabolism of the two enantiomers, whether with in vitro drug metabolizing enzymes (Pybus et al., 2012; Fasinu et al., 2014; 2016), in animal models (Baker and McChesney, 1988; Saunders et al., 2014), or in humans (Tekwani et al., 2015; Khan et al., 2022a). RPQ is much more significantly converted via the MAO pathway to cPQ, the major PQ metabolite in human plasma. PQ-N-CG has been recently identified as a PQ metabolite in plasma, notably after administration of SPQ (Khan et al., 2022a). The formation of the hydroxylated products of PQ also showed quite distinct patterns with the two enantiomers, though both RPQ and SPQ gave rise to all of the reported ring-hydroxylated species (Fasinu et al., 2016). Several reported animal studies have shown dramatic differences in the toxicity of PQ enantiomers, including their hematological effects, with SPQ showing greater systemic toxicity than RPQ in mice (Schmidt et al., 1977), but lower hepatotoxicity in primates (Schmidt et al., 1977; Saunders et al., 2014); however, hematological effects of SPQ were more pronounced than RPQ’s effects in both species (Nanayakkara et al., 2014; Saunders et al., 2014); this is in spite of an equivalent radical curative efficacy of the two enantiomers in primates (Schmidt et al., 1977; Saunders et al., 2014).

A critical question that arises, then, for the safe use of PQ in humans with G6PDd is whether the differential metabolic profile of enantiomers will translate to a better therapeutic index. Toward answering that question, we undertook to assess the tolerability and metabolism of the RPQ and SPQ, first in G6PDn human volunteers receiving a single dose (Khan et al., 2022a). SPQ showed higher plasma exposure of PQ as compared to RPQ. cPQ was the predominant plasma metabolite with RPQ administration, many folds higher than the parent drug, but PQ-N-CG was not detected. In contrast, for SPQ, PQ-N-CG was the major plasma metabolite, while much lower levels of cPQ were seen (Khan et al., 2022a). POQ and cPOQ were also identified in the RBCs of subjects receiving single doses (Khan et al., 2021). However, in this study using single doses, there were no identifiable tolerability issues, so we could not distinguish any safety advantage. Typically, the hematological toxicity of PQ in G6PDd subjects appear after a few days dosing.

In the current study we aimed to check the comparative tolerability of racemic PQ and its individual enantiomers in G6PDn subjects, and profile the plasma and RBCs for parent and metabolites during oral administration once daily for seven consecutive days. Two G6PDd individuals also received PQ enantiomers for a 5 day course.

With respect to metabolism, as observed in our single dose study (Khan et al., 2022a), plasma exposure to parent drug was significantly higher for SPQ treated groups than for RPQ, and RSPQ behaved in an additive fashion. Plasma cPQ levels were much higher after RPQ administration than SPQ, likely reflecting the greater vulnerability of RPQ to metabolism by monoamine oxidases (Chaurasiya et al., 2021). This presumably also accounts for the reduced exposure to the parent drug after RPQ administration. Of note, cPQ concentrations were sustained in the blood over the 24 h daily dosing interval, such that it was still very high when subsequent doses were administered. Consistent with this prolonged exposure, higher and sustained levels of cPOQ were observed in plasma and RBC, as compared with POQ, raising the possibility that cPQ may contribute more to the overall oxidant stress and haemolytic activity seen clinically than PQ does directly. If so, and if the two enantiomers show similar redox cycling within the red cell, it would follow that the RPQ enantiomer would contribute more to the hemotoxicity of the parent drug than the SPQ enantiomer, with implications for difference in therapeutic index. Conversely, if the sustained higher levels of cPOQ in red cells is due to lesser capacity to redox cycle, and hence lesser production of active oxygen species, RPQ may have carry less potential for haemolytic injury.

In addition, SPQ is cleared by metabolism to PQ-N-CG via a phase II conjugation pathway. N-carbamoylation and glucuronidation of amines is relatively uncommon, but not unknown (Obach et al., 2006). The analogous metabolite of RPQ was not observed in this study or in our previous single dose study (Khan et al., 2022a). The contribution of this metabolite to the therapeutic efficacy of primaquine is not known, but a possible role implies an additional further differentiation in the therapeutic index of the two enantiomers.

Collectively, these considerations further emphasize the relevance of the stereoselective metabolism of primaquine in the clinical use of this drug.

The orthoquinone metabolite of PQ, derived from 5-hydroxylation (POQ) was observed only at very low concentrations in circulation after PQ administration. Further, POQ concentrations in red cells after SPQ administration were higher than after RPQ, and RSPQ showed an additive response. In other words, POQ in red cells appeared to fluctuate in a manner reflecting PQ plasma concentrations.

Our findings with respect to PQ metabolism in the two G6PDd subjects (one heterozygous female, one hemizygous male) did not show any clear differences from the G6PDn subjects with respect to the metabolic clearance of the enantiomers. These data further support the concept that the greater susceptibility of G6PDd individuals to hemolytic anemia after PQ lies in the lesser protective ability of their red cells and not in enhanced production of oxidant metabolites.

Even though the POQ in RBCs after SPQ is somewhat higher than after RPQ, the latter gives rise to substantially higher amounts of cPOQ in both plasma and RBCs. Although POQ has long been assumed to reflect the redox active pathways for efficacy and toxicity of PQ, any contribution of cPOQ in this respect has not been anticipated.

Both PQ enantiomers were well tolerated in G6PDn individuals with 7 day dosing at 22.5 mg (or with 45 mg of RSPQ). The primary findings of interest here were the progressive elevations of methemoglobin (MetHgB) and total bilirubin observed. MetHgB is formed on oxidation of hemoglobin when the ferrous ion is converted to ferric (Rifkind et al., 2015). Methemoglobinemia is one of the hallmark effects of PQ with multiple days dosing. However, methemoglobinemia, per se, is rarely of serious clinical significance with PQ administration (Carmona-Fonseca et al., 2009). A rise in MetHgB was observed in most of our subjects, and there was no difference between the two enantiomers. RSPQ tended to show a higher response, as the dose was double that with the racemate, but in only one subject—the most sensitive responder—was a clinical impact observed. This subject showed some transient exertional dyspnea on the RSPQ arm that required discontinuation of dosing after the 5^th dose. However, this subject responded to each drug form—RPQ, SPQ and RSPQ with the most robust MetHgB rise of any subject—to above 10% of HgB with each enantiomer and with the racemate. This subject did not show any other hematological or biochemical abnormalities on study endpoints, but displayed an elevated thyroid stimulating hormone (TSH). All other subjects exhibited minimal to modest responses with a gradual rise in MetHgB to an average of 4%–5% of HgB. Our findings are thus consistent with earlier reports that methemoglobinemia is generally not a limiting concern with PQ administration, in that many subjects experience mild increases, with no obvious impact on hemolysis (Carmona-Fonseca et al., 2009).

This study was not designed to address gender differences—nor was it sufficiently powered for such. However, safety parameters and profiles of PQ and metabolites were considered to assess any obvious distinctions. In terms of safety, there did not appear to be anything thing clinically significant. The mean MetHgB response in females is somewhat greater than in males after RSPQ administration, but this seems largely influenced by the one female subject mentioned earlier with an exaggerated response. We did find a significantly increased POQ content in red cells in female subjects, as compared to males. These data are depicted in the supplemental files (Supplementary Figures S1, S2). The significance of this—if any—is unknown, but may be a subject for exploration as further field studies with PQ are pursued.

A somewhat unexpected finding of this study was the sensitivity of the total bilirubin responses in G6PDn subjects. Though the changes were still well below the clinical upper limits of normal and did not approach clinical significance in terms of patient safety, the gradual increase with daily dosing was observed consistently. Placebo administration did not result in any increase in bilirubin. In one subject with severe G6PDd deficiency, the rise in bilirubin was much exaggerated, even with a lower dose (15 mg daily) of either of the enantiomers, such that the preset safety limits were exceeded, and dosing had to be discontinued after 3 days with both RPQ and SPQ. The heterozygous female subject showed enhanced responses to the G6PDn individuals in terms of bilirubin. The increase in total bilirubin after PQ or its enantiomers were confirmed to be due to indirect bilirubin (unconjugated), and thus likely derived from red cell destruction, rather than liver injury. This suggests that a minor fraction of red cells—perhaps the oldest fraction in circulation, which display the cumulative oxidative insults of normal cell aging—are impacted by the addition of the oxidative stress of the PQ metabolites. Early studies of PQ sensitivity suggested that even in G6PDn subjects, subclinical responses in various endpoints can be observed, likely due to a minor fraction of aging red cells with lower G6PD and other glycolytic enzymes (Powell and Degowin, 1965). The changes in bilirubin were not significantly different between RPQ and SPQ, though there was a slight tendency toward a greater RPQ response. Our studies suggest that bilirubin might serve as a better hemolysis-relevant marker in G6PDn individuals as compared to methemoglobin.