The blood schizonticidal activity of tafenoquine makes an essential contribution to its prophylactic efficacy in nonimmune subjects at the intended dose (200 mg)

In the pivotal Phase III study [11], 654 nonimmune Australian soldiers were randomly assigned to receive mefloquine (n = 162) or tafenoquine (n = 492) during deployment to Timor, and the P. vivax attack rate during the period of the deployment was retrospectively estimated to be 6.88% [11]. The subjects in this study were of largely European ethnicity, meaning that the rate of deficient CYP2D6 phenotypes may have been as high as 13% [9]. The subjects also had no prior exposure to malaria, meaning that any viable blood stage infections representing a potential risk would have been symptomatic.

The following was assumed for the purposes of conducting a thought experiment:

It can be further assumed that the pivotal study was repeated with the same sample size in the tafenoquine arm and included a placebo group where the P. vivax attack rate was 6.88%. In such a study 1.7, 3.4 or 4.4 P. vivax cases should have been observed in the tafenoquine arm if the background rate of low metabolizer CYP2D6 phenotypes was 5, 10 or 13%, respectively.

However, in the actual Phase III study this was not the case: the point estimate of efficacy (95% confidence intervals [CI]) for tafenoquine was: 100% (93–100%).

In two Phase II studies from which the data were pooled, 519 African residents of high malaria transmission areas were randomly assigned to receive placebo (n = 187), tafenoquine (n = 190), or mefloquine (n = 142) during a period in which the incidence density in the placebo group was (almost completely attributable to P. falciparum) 5.06 cases/person-year [12]. Approximately 20% of individuals of sub-Saharan African ancestry may exhibit 2D6 deficient phenotypes [9].

The following was assumed for the purpose of conducting a thought experiment:

It can be further assumed that the pooled Phase II was conducted with the same sample sizes and a P. falciparum attack rate was 5.06 cases/person-year. In such a study, 22, 32 or 62 P. falciparum cases should have been observed in the tafenoquine arm if the background rate of low metabolizer CYP2D6 phenotypes was 5, 10 or 20%, respectively.

However, this was not the case. The observed number of cases in the actual tafenoquine arm was only 14 and the prophylactic efficacy of tafenoquine was 93.5% (89–96%) [12].

In humans, it is not possible to definitively assess the relative contribution of causal and blood schizonticidal activity to the prophylactic efficacy of tafenoquine. However, animal models provide important clues. Li et al. [13] examined the contributions of causal and suppressive effects of tafenoquine to its prophylactic efficacy in mice utilizing a transgenic P. berghei parasite expressing the bioluminescent reporter protein luciferase. Drug activity against developing liver stages was assessed using a real-time in vivo imaging system and suppressive activity against erythrocytic stages was assessed using flow cytometry. When one dose of tafenoquine (5 mg/kg) was administered 1 day prior to sporozoite inoculation, merozoite leakage was not detected by flow cytometry through day 30 in 10 of 10 animals. However, in these animals real time imaging demonstrated that only 98.6% of exoerythrocytic parasites were eliminated compared to controls at 48 h after sporozoite challenge. Because parasites begin to emerge from the liver to enter the blood at 48 h in this model, some of the remaining 1.4% of liver parasites had entered the blood after that time and the 100% prophylactic efficacy of this regimen was due to elimination of a few blood parasites by tafenoquine remaining in the blood after 48 h. These data imply that at this particular dose in mice, both causal and blood schizonticidal activity are important for the prophylactic efficacy of tafenoquine. It is also true that at higher doses (e.g. 5 mg/kg/day for 3 days), 100% prophylactic efficacy was observed following 100% inhibition of liver schizogony. This means that it might also be possible in humans, assuming it could be tolerated, to administer a dose of tafenoquine at which prophylactic efficacy was mediated only via causal activity.

The first challenge study involving tafenoquine attempted to position the drug as a causal prophylactic. Four nonimmune healthy volunteers were given a single 600-mg dose immediately prior to a mosquito-induced P. falciparum challenge [14]. The intended dose of a 200 mg × 3 load followed by 200 mg weekly essentially maintains this exposure level of tafenoquine at steady state. Three of the four volunteers were protected. The fourth contracted symptomatic malaria on day 31 after the challenge and had drug levels approximately half those of the three protected subjects. Because the study was conducted more than 15 years ago and there was no reason to suspect a correlation between cytochrome P450 enzyme status and activity at the time, the metabolic enzyme profile of the subject who contracted malaria is not known. Nevertheless, the Cmax observed in the single failure (244 ng/ml) was <15% lower than the predicted median Cmax following the intended loading dose of 200 mg × 3 for malaria prophylaxis (~280 ng/ml, see Fig. 1; Table 1). Assuming this effect is generalizable, exposure in humans at the intended dose may produce a pharmacologic effect more similar to that observed following the lower dose used in the mouse studies above. Therefore, the efficacy of continuous chemoprophylaxis can only be explained by a substantial blood schizonticidal effect.

First, tafenoquine administered at doses ≥2 mg/kg/day for 3 days cleared the erythrocytic stages of Plasmodium cynomolgi in Macaca monkeys within 3–4 days and at doses of 6 mg/kg/day for 3 days there was no relapse or recrudescence [15]. Dow et al. reported a median Cmax of approximately 50 ng/ml following administration of 0.6 mg/kg/day for 3 days [15]. Assuming pharmacokinetic parameters are proportional to dose, the median Cmax at the minimum dose (2 mg/kg/day × 3) was 150 ng/ml. This is higher than the lowest 5% Cmax observed following administration of the intended human loading dose (~120 ng/ml, Fig. 1 [16]), meaning it would be expected that the intended loading dose in humans would clear an established blood stage infection. The rate of parasite clearance observed at 2 mg/kg/day was slower than that for a comparator group administered chloroquine and a lower dose of tafenoquine. This means that tafenoquine clears asexual parasitaemia more slowly than established blood schizonticidal drugs.

Second, in Aotus monkeys, doses of 1 mg/kg/day for 3 days cleared the erythrocytic stages of an established chloroquine-resistant P. vivax with recrudescence 20–25 days later [17]. Doses of 3 mg/kg/day for 3 days were required for complete elimination (cure). No pharmacokinetic data are available from that study. However, let us assume that the conversion factor calculated to derive a human equivalent dose (HED) from Macaca monkeys that is presented in Table 1, is the same for Aotus monkeys. Therefore, at an HED of 120 mg per day × 3, tafenoquine is able to clear an established chloroquine-resistant P. vivax infection, while at a dose of around 360 mg per day × 3, the drug is able to cure such an infection. This confirms that clearance of established P. vivax infections should be possible at the anticipated loading dose in humans.

Third, Brueckner et al. [14], in the same challenge study discussed earlier, observed blood stage parasites following sporozoite challenge in one individual on day 31 after that individual had received a single 600-mg dose of tafenoquine. Drug levels at day 31 were approximately 48 ng/ml. Normal patency following a P. falciparum challenge is not usually longer than day 11. A reasonable interpretation of this data is that tafenoquine at concentrations greater than 48 ng/ml are able to suppress the amplification of low-density P. falciparum infections.

Fourth, Edstein et al. [18], in a study conducted in Thai soldiers, reported 3 symptomatic breakthroughs for which drug levels were measurable occurred 6–12 weeks after prophylaxis: two P. vivax cases with drug levels of 20 and 21 ng/ml and a single case of P. falciparum with a tafenoquine concentration of 38 ng/ml (Table 1). During the prophylactic phase of the same study, a single symptomatic P. vivax breakthrough was observed in an individual noncompliant with the medication, and the tafenoquine concentration was determined to be 40 ng/ml [19]. These data are consistent with the hypothesis that tafenoquine concentrations in excess of 40 ng/ml are able to control low density infections caused by multiply drug resistant P. falciparum and P. vivax.

Finally, tafenoquine exhibits greater potency against the blood stages than primaquine in vivo. Against the erythrocytic stages of drug-sensitive strains of P. berghei in vivo, tafenoquine is 9 times more active than primaquine [20]. Against the erythrocytic stages of strains of P. berghei resistant to conventional anti-malarial drugs, tafenoquine is 4 to 100 times more potent than primaquine in vivo [20]. This supports the hypothesis that tafenoquine should exhibit more activity against P. falciparum in vivo in humans than primaquine.

Together, these in vivo studies suggest that tafenoquine is able to clear established (high density) blood stage infections at exposure levels equivalent to the intended loading dose, and to control sub-patent low density infections at relatively low drug concentrations (>50 ng/ml) in vivo.

In a recent study, Milner et al. [21] demonstrated that a 20 mg/kg oral dose of tafenoquine administered exhibited different pharmacokinetic parameters in wild type and 2D6 knockout mice. However, in the same study, the authors also demonstrated that a single dose of 25 mg/kg tafenoquine, administered 4 days following a sporozoite inoculation, cleared asexual parasitaemia in wild type and 2D6 knock out with no difference in time to clearance or course of parasitaemia. These data demonstrate that although it is feasible that defects in 2D6 metabolizer status might affect the pharmacokinetic properties of tafenoquine, the drug is perfectly able to clear established blood stage infections in the absence of a functional CP450 2D6 cluster. Of note, similar effects were observed for primaquine, which implies a class effect.

The most parsimonious explanation for all the available data is that continuous tafenoquine prophylaxis at the intended dose protects non-immune individuals from contracting symptomatic malaria by (i) eliminating most, but not all, developing asexual exoerythrocytic stages following an infectious mosquito bite, and (ii) slowly clearing the few erythrocytic stages that leak out of the liver in some individuals in a manner that is independent of cP450 2D6 metabolizer status. There is, therefore, no need for a cP450 2D6 companion diagnostic.

Based on the association of symptomatic breakthroughs with measured plasma concentration discussed above, Edstein et al. [18] proposed that the minimum inhibitory concentration (MIC) of tafenoquine required to protect nonimmune individuals from a symptomatic breakthrough is approximately 80 ng/ml. The US Army/commercial team developing tafenoquine has modelled all published and internal pharmacokinetic data to model the plasma concentrations of tafenoquine at the intended and lower doses [16]. The current thinking regarding the dose justification for malaria prophylaxis is that the intended dose of 200 mg × 3 followed by weekly maintenance doses generates trough concentrations in excess of the proposed MIC throughout the course of administration in 95% of individuals, whereas lower doses (e.g., a 100-mg regimen) do not (Table 1; Fig. 1).

However, this does not seem to make sense based on in vitro susceptibility data, since the IC50s, and most importantly the IC90/99s, reported for tafenoquine, are generally higher than the proposed MIC in vivo. Ramharter et al. reported that tafenoquine exhibits a mean IC₅₀ of 97 ng/ml and a mean IC₉₉ of 3108 ng/ml against 44 multiple-drug-resistant strains of P. falciparum from the Thai-Burmese border in a schizont maturation assay [22]. Ohrt et al. [23] reported an IC₅₀ of 64 to 110 ng/ml against 2 drug resistant strains of P. falciparum from South East Asia. Tafenoquine was found to exhibit an IC₅₀ of 2041 ng/ml amongst 162 African isolates of P. falciparum [24]. Vennerstrom et al. [25] reported a mean IC50 of 202 ng/ml against seven drug-resistant clones of P. falciparum.

The counter argument is that endpoints measured in vitro, and the conditions utilized to generate them are vastly different from the relevant in vivo pharmacodynamics endpoints for malaria prophylaxis. First, the in vitro assays conducted to date all exhibit short exposure formats (48 h) relative to the parasite clearance time observed in vivo. Second, in vitro assays executed to date utilize much higher initial parasite densities than those after initial merozoite release from the liver in a nonimmune subject who has been administered prophylactic treatment. Third, the in vitro assays conducted do not account for either (i) any metabolic activation outside red blood cells (if this is required for activity) and (ii) cannot replicate possible deleterious effects of tafenoquine on parasites that survive exposure in the liver but escape into the blood stream. In fact, it may not be possible to replicate all the relevant in vivo parameters in an in vitro assay system.

It should be stated that the mechanism(s) of action of tafenoquine against blood stage parasites are not known, and come from a single study reported by Vennerstrom et al. [25]. In that study, Vennerstrom et al. [25] proposed two possible mechanisms of action.

First, tafenoquine, but not primaquine, was shown to inhibit haem polymerization in vitro at concentrations lower than chloroquine (16 v 80 microM [25]). However, as the authors themselves point out, it cannot be concluded that this mechanism is relevant without data showing that tafenoquine accumulates in the food vacuole to the requisite degree at relevant concentrations. Furthermore, this seems implausible since only moderate inhibition is observed in vitro in most parasites lines at concentrations higher than the median Cmax of the intended loading dose (~280 ng/ml, see Fig. 1; Table 1).

Second, Vennerstrom et al. [25] also proposed that 8-aminoquinolines might exhibit their blood schizonticidal effects through oxidative mechanisms. This seems somewhat more plausible since it is well known that 8-aminoquinolines increase oxidative stress in red blood cells with reduced G6PD leading to increased deformability and enhanced splenic and hepatic clearance [26, 27]. A reasonable topic for further study would be to assess whether, in vivo, 8-aminoquinolines, via oxidative mechanisms, increase the deformability of infected red blood cells which are then cleared by the spleen and/or liver.

The existing clinical efficacy database does not include any studies that directly show clearance of P. falciparum in non-immune subjects. The lack of availability of such data has largely been for ethical concerns due to the slow parasite clearance time of tafenoquine. However, two approaches that could generate such data can be envisioned. The first is utilization of a well-characterized blood stage challenge model. Such a model would avoid the confounding effect of the causal activity of tafenoquine that would be encountered with a sporozoite challenge. Additionally, the parasite density following inoculation is more in line with the density that would be encountered after merozoite release during prophylaxis. The second approach would be, in a treatment study enrolling P. falciparum patients, administration of tafenoquine in combination with a dose of artesunate that clears but does not cure. Tafenoquine should clear up the residual parasite burden in a manner analogous to that for mefloquine.