Use of primaquine and glucose-6-phosphate dehydrogenase deficiency testing: Divergent policies and practices in malaria endemic countries

Primaquine is the only available antimalarial drug that kills dormant liver stages of Plasmodium vivax and Plasmodium ovale malarias and therefore prevents their relapse (‘radical cure’). It is also the only generally available antimalarial that rapidly sterilises mature P. falciparum gametocytes. Radical cure requires extended courses of primaquine (usually 14 days; total dose 3.5–7 mg/kg), whereas transmissibility reduction in falciparum malaria requires a single dose (formerly 0.75 mg/kg, now a single low dose [SLD] of 0.25 mg/kg is recommended). The main adverse effect of primaquine is dose-dependent haemolysis in glucose 6-phosphate dehydrogenase (G6PD) deficiency, the most common human enzymopathy. X-linked mutations conferring varying degrees of G6PD deficiency are prevalent throughout malaria-endemic regions. Phenotypic screening tests usually detect <30% of normal G6PD activity, identifying nearly all male hemizygotes and female homozygotes and some heterozygotes. Unfortunately, G6PD deficiency screening is usually unavailable at point of care, and, as a consequence, radical cure is greatly underused. Both haemolytic risk (determined by the prevalence and severity of G6PD deficiency polymorphisms) and relapse rates vary, so there has been considerable uncertainty in both policies and practices related to G6PD deficiency testing and use of primaquine for radical cure. Review of available information on the prevalence and severity of G6PD variants together with countries’ policies for the use of primaquine and G6PD deficiency testing confirms a wide range of practices. There remains lack of consensus on the requirement for G6PD deficiency testing before prescribing primaquine radical cure regimens. Despite substantially lower haemolytic risks, implementation of SLD primaquine as a P. falciparum gametocytocide also varies. In Africa, a few countries have recently adopted SLD primaquine, yet many with areas of low seasonal transmission do not use primaquine as an antimalarial at all. Most countries that recommended the higher 0.75 mg/kg single primaquine dose for falciparum malaria (e.g., most countries in the Americas) have not changed their recommendation. Some vivax malaria–endemic countries where G6PD deficiency testing is generally unavailable have adopted the once-weekly radical cure regimen (0.75 mg/kg/week for 8 weeks), known to be safer in less severe G6PD deficiency variants. There is substantial room for improvement in radical cure policies and practices.

adopted the once-weekly radical cure regimen (0.75 mg/kg/week for 8 weeks), known to be safer in less severe G6PD deficiency variants. There is substantial room for improvement in radical cure policies and practices.

Background
Glucose 6-phosphate dehydrogenase (G6PD) deficiency is the most prevalent enzyme deficiency in the world. An estimated 400 million people are affected (reviewed in [1][2][3][4][5]). Within the erythrocyte, G6PD is the only source of NADPH, necessary for provision of reduced glutathione and the function of catalase. Both reduced glutathione and catalase provide essential cellular protection from oxidative stresses. G6PD activity decreases exponentially as red cells age, with a normal half-life of about 50 days, but in most genetic variants, G6PD degrades more rapidly, rendering older erythrocytes increasingly vulnerable to oxidant stresses and, thus, haemolysis. Very severe G6PD deficiencies are sporadic and rare and cause chronic nonspherocytic haemolytic anaemia. Less severe deficiencies are polymorphic, and they are very common in most tropical areas. This is attributed to the protection G6PD deficiency provides against the adverse effects of malaria [6][7][8][9].
As G6PD deficiency is X-linked, males are hemizygous and either phenotypically normal or deficient (although the degree of deficiency depends on the genotype) ( Table 1). Homozygote females are as deficient as the hemizygous males, whereas heterozygous females are mosaics with intermediate levels of deficiency as a result of embryonic random X-chromosome inactivation (lyonisation). Their blood contains a mixture of G6PD normal and G6PD deficient (G6PDd) cells. The proportion in the heterozygous population overall averages 50:50, but some heterozygotes may have a majority of G6PDd erythrocytes. Most qualitative G6PD blood screening tests report enzyme activities >30% as 'normal', so heterozygotes testing 'normal' could still have up to 70% of their red cells G6PDd-and so be vulnerable to oxidant haemolysis [10].
The 8-aminoquinoline primaquine has a unique antimalarial activity spectrum. It has prophylactic, radical curative (Plasmodium vivax and P. ovale), and rapid gametocyte sterilising (P. falciparum) activities ( Table 2). The main concern limiting primaquine use is the risk of dose-dependent acute haemolytic anaemia in G6PDd individuals. Uncertainty over the risks and benefits of primaquine, together with lack of readily available G6PD deficiency testing in malaria-endemic areas, has resulted in a wide and confusing array of policies and practices.

Table 1. G6PD genotypes and phenotypes.
If q is the allele frequency for G6PD deficiency and 1-q is the allele frequency for the wild-type gene, and assuming a Hardy-Weinberg equilibrium, then because G6PD deficiency is X-linked, the prevalence of the G6PD deficient genotype in hemizygous males is q, the prevalence in homozygous females is q 2 , and the prevalence in heterozygote females is 2[q(1-q)]. Thus, the allele frequency equals the proportion of males who are G6PDd, and, although a higher proportion of females carry the mutant allele, only a minority of these will be designated as G6PDd using the usual threshold of <30% of normal enzyme activity. For example, if 10% of men are G6PD-deficient hemizygotes, then 1% of women will be equally deficient homozygotes, and 18% of women will be heterozygotes with variable levels of deficiency (because of lyonisation).
Note: Because G6PD deficiency protects against symptomatic vivax malaria, the prevalence of G6PD deficiency in patients requiring radical treatment with primaquine may be lower than in the general population.
The historical background to these recommendations and the current primaquine and G6PD deficiency testing policies in malaria-endemic countries are reviewed briefly here.

Historical background to current practices
Much of our knowledge about the haemolytic risks of primaquine derives from a comprehensive series of studies conducted in the United States of America on healthy male African American volunteers (presumably G6PD African A-hemizygotes) between the late 1940s and the early 1960s [11][12][13][14]. Haemolysis was shown to be dose dependent, with older red cells haemolysing preferentially. With daily primaquine dosing in G6PD A-subjects, haemoglobin concentrations first fell and then rose again, despite continued drug administration, as young, less G6PDd erythrocytes, which were more resistant to oxidant haemolytic effects, entered the circulation. This was exploited in the 8-week radical cure regimen, in which a single 0.75 mg/kg (adult dose 45 mg) dose is taken once weekly. The safety of this regimen was established only in the G6PDd A-variant, which is generally less severe than most other polymorphic variants.
The widely recommended primaquine radical cure regimen for vivax malaria in G6PDnormal patients (0.25 mg/kg/day for 14 days; adult daily dose 15 mg) was based on experience in the Korean War (1950)(1951)(1952)(1953) with long-incubation P. vivax [15]. This stood unchallenged for 3 decades and, with very few clinical studies, practices began to diverge. In Southeast Asia and Oceania, where the frequently relapsing 'Chesson' phenotype P. vivax is prevalent, the 0.25 mg/kg/day dose was considered insufficient. Higher doses were recommended, culminating today with 0.5 mg/kg/day for 14 days [16]. In contrast, in India, which harbours a substantial proportion of the world's vivax malaria, 5-day primaquine courses were associated with relapse rates of approximately 20% (compared with !50% rates further east) and were recommended for 4 decades until it was discovered that 20% relapse rates occurred whether or not the 5-day regimen was given [17,18]. India now recommends standard 0.25 mg/kg/day 14-day primaquine courses for radical cure. In South America, 7-day 0.5 mg/kg/day regimens are widely recommended.
The gametocytocidal activity of 8-aminoquinolines in falciparum malaria was observed in the first clinical studies nearly 100 years ago and shown soon afterwards to result in rapid sterilisation of the infection [19]. This potent transmission-blocking potential was exploited in recommendations over the past 50 years to add a single dose of primaquine to falciparum malaria treatment, mainly in low-transmission settings, where symptomatic infections contribute substantially to transmission overall. The original single dose was 0.75 mg/kg (adult dose 45 mg), lowered recently to 0.25 mg/kg (single low dose, SLD), which appears to provide the same reduction in transmissibility [20-22] but obviously less haemolytic risk [23].

Malaria epidemiology and current practices
In Africa, P. falciparum predominates, although P. vivax comprises approximately half the malaria in the Horn of Africa and 10% in Madagascar. In the Americas, with the exception of Haiti, P. vivax predominates, and in Asia, prevalences of P. vivax and P. falciparum are approximately equal, with high rates of mixed infection (Table 3). In India, relapse rates following vivax malaria are approximately 20%, and radical treatment is recommended. In East Asia, P. vivax relapse rates exceed 50% in vivax malaria, and approximately 30% of patients experience a presumed P. vivax relapse following falciparum malaria, yet a policy of radical cure has never been considered in acute falciparum malaria. For vivax malaria in the Americas, Asia, and Oceania, radical cure is widely recommended but is often not given because G6PD deficiency testing is unavailable. Use of primaquine as a P. falciparum gametocytocide has been confined mainly to areas of low transmission [20]. In high-transmission settings, the transmission reservoir of asymptomatic individuals is considered too large for primaquine to have any impact. As countries plan for elimination, this transmission blocking recommendation is being more actively supported.

G6PD deficiency and primaquine use
The main factor limiting use of primaquine is concern over dangerous haemolysis in patients with G6PD deficiency. The prevalences of G6PD deficiency in malaria-endemic countries range from very low (e.g., the Democratic Republic of Korea) to over 30% [24]. The most frequent G6PDd variant in Africa (A-) was generally considered 'mild', although in some cases haemolysis can be severe [25]. In the Americas, G6PD A-is also the most common variant, whereas in Asia and Oceania there are numerous variants, most of which are more severe than A-. The G6PD Mediterranean variant, prevalent from Southern Europe to Southeast Asia, is at the severe end of the spectrum, and in people with this form of G6PD deficiency, repeated primaquine administration can cause life-threatening haemolysis [26]. Most estimates of G6PD deficiency prevalence are based on enzyme activity studies in males, although molecular genotyping methods for epidemiological assessments have predominated in recent years. Genotyping assessments inform risks for males, but they are less informative for females as the majority of those carrying the G6PDd allele will be heterozygotes (Table 1), and most of these females will test as G6PD normal with the usual screens (detecting <30% of normal activity) but may still be at risk of significant haemolysis [9].
countries. This is divided into 3 geographic sections: (a) Africa; (b) Asia, Oceania, and the Middle East; and (c) the Americas. Estimates of G6PD deficiency prevalence were obtained from published reviews and studies, including predictive models based on published data. Current policies for G6PD deficiency testing and primaquine treatment in each country were obtained mainly from the World Health Organization (WHO) World Malaria Report 2016 [27] and updated for 2017 [28] and/or publications from each country, and maps were generated based on these reported policies. In most cases, we confirmed WHO-reported current policies and practices in countries by contacting countries' ministries of health (MoHs) or researchers and/or institutions in malaria-endemic areas.

Results
Estimates of G6PD deficiency prevalence and predominant variants are summarised in Tables 4 to 6. These proportions usually refer to the allele frequency, which is equivalent to the proportion of males expected to be G6PDd.
Most countries are in the malaria control phase, except for Algeria, Botswana, South Africa, Cabo Verde, and Swaziland (now in elimination phase), and policy is not to use primaquine as a P. falciparum gametocytocide. The exceptions are Ethiopia, where single low dose (SLD 0.25 mg/kg) is now accepted although not yet widely implemented, and Madagascar, Botswana, Eritrea, Swaziland, Zimbabwe, Mauritania, which recently included single-dose primaquine as policy (2015), and Comoros, Mayotte, Namibia, Sao Tome and Principe, Algeria, and Cabo Verde (Table 4 and Fig 1). Primaquine is seldom available for radical cure of vivax or ovale malaria. In terms of official policy, except for Angola, Cabo Verde, Mauritania, South Africa, Mayotte, Sao Tome and Principe, and Equatorial Guinea, most countries either do not require G6PD deficiency testing before primaquine administration for radical cure or do not have a G6PD deficiency testing policy. Ethiopia is planning to use primaquine for P. vivax in districts targeted for malaria elimination, together with G6PD deficiency testing. Madagascar, Botswana, Cabo Verde, Mayotte, Sao Tomé and Principe, Namibia, and Mauritania have modified their policies recently (2014-2015) to include primaquine for radical cure (0.25 mg/kg for 14 days), but Botswana, Namibia, and Madagascar do not require G6PD deficiency testing before primaquine, while Mauritania, Mayotte, Sao Tomé and Principe, and Cabo Verde do (Table 4 and Fig 2).

Middle East, Asia, and Oceania
The enormous populations of this large region host a heterogeneous array of G6PD variants [30]. While the prevalence of G6PD deficiency in malaria-endemic countries is generally lower than in Africa, it exceeds 13%-17% in 4 countries and is around >10%-13% in 2 others ( Table 5). Most of the variants are at the more severe end of the polymorphic spectrum. Several common Asian variants were categorised as severely deficient (1%-10% residual activity) in the original WHO Working Group severity classification. These include Viangchan, Coimbra, Kaiping, Union, Mediterranean, Valladolid, and Gaohe [24,31-33]. The severe G6PD Mediterranean variant predominates over much of Europe, the Middle East, and West and Central Asia. The Mahidol variant, which is common in Myanmar and adjacent countries, was classed as moderate (Class III; 10%-60% residual activity), but recent studies indicate severity similar to the G6PD Viangchan variant (common in Lao PDR and Eastern Thailand) [34]. Although typical Asian variants would be expected to have a high haemolytic anaemia risk during radical cure treatment with primaquine, apart from the A-, Mahidol, Viangchan, and Mediterranean variants, there is very limited quantitative information on haemolysis in the different G6PD deficiency variants.

Primaquine policies
Most countries in Asia have low seasonal transmission, and in recent years, many have set ambitious targets for malaria elimination within the next decade. The emergence and recent spread of artemisinin-resistant P. falciparum, leading to ACT failure in 5 countries of the Greater Mekong subregion (Cambodia, Lao People's Democratic Republic, Myanmar, Thailand, and Vietnam) [35], threatens these targets. In this context, addition of SLD primaquine to falciparum malaria treatment regimens has been encouraged to reduce transmissibility of the treated infection. Several countries were already recommending the higher single gametocytocidal primaquine dose (0.75 mg/kg) without concerns. Except Papua New Guinea, Vanuatu, Solomon Islands, and Timor-Leste, all countries now recommend adding a single dose of primaquine to ACTs for falciparum malaria (Fig 3 and Table 5). Cambodia has just adopted this policy (SLD 0.25 mg/kg), and Papua New Guinea is updating its policy. Radical cure primaquine treatment policies vary across the Asian region. Some countries state a requirement for G6PD deficiency testing before primaquine administration (Lao People's Democratic Republic, Cambodia, Pakistan, Afghanistan, Malaysia, Saudi Arabia, Sri Lanka, Philippines, Thailand, Yemen, Timor-Leste, Vanuatu, and Solomon Islands), although in practice testing is often not available. Myanmar acknowledges the lack of access to testing at village level, where most malaria cases are diagnosed, and has recently adopted a policy of  weekly primaquine administration for 8 weeks for all patients treated by community malaria volunteers. Iran also recommends this regimen (Table 5). While it is policy to administer the radical cure primaquine regimen as DOT in the Democratic People's Republic of Korea, Malaysia, Iran, Sri Lanka, Philippines, India, Thailand, Vietnam, Timor-Leste, and Afghanistan (Fig 4), this has not been implemented widely in the latter 3 countries. A survey of antimalarial availability and use among physicians, pharmacists, and patients across 6 states in India  Although WHO has recommended the higher daily dose of primaquine (0.5 mg/kg/day) for radical cure of P. vivax in East Asia and Oceania, most countries there still recommend the lower 0.25 mg/kg/day dose except for Malaysia, Myanmar, and the Philippines (Table 5).
Sri Lanka is currently in the prevention-of-reintroduction phase after having successfully eliminated malaria (certified by WHO as a malaria-free country in September 2016) [39]. China, Bhutan, Malaysia, the Republic of Korea, the Democratic People's Republic of Korea, Saudi Arabia, and Iran are currently in the malaria elimination phase.

The Americas
Several countries in Central and South America are well on the path towards malaria elimination (Mexico, Belize, Costa Rica, Dominican Republic, Ecuador, El Salvador). Paraguay has applied to be certified as malaria-free, and Argentina is in the prevention-of-reintroduction phase. In sad contrast, malaria has resurged in Venezuela as the public health services there   41], with a predominance of A- [41]. There are limited data on the distribution of G6PD deficiency variants in areas where malaria is endemic. With the exception of French Guiana, all countries currently recommend primaquine as a P. falciparum gametocytocide (Table 6 and Fig 5), mostly at 0.75 mg/kg. This will be lowered to 0.25 mg/kg in Brazil's revised guidelines, and Colombia has just updated the national malaria treatment guidelines to include primaquine single dose. P. vivax is estimated to have caused approximately 69% of the malaria cases in the Americas in 2015 [27,28]. None of the countries has as policy a requirement for G6PD deficiency testing before radical cure except French Guiana, although Brazil will include this recommendation in upcoming guidelines along with a recommended treatment of 0.75 mg/kg/week for 8 weeks for G6PDd All countries have policies to use primaquine for radical cure of vivax malaria, mainly as 0.25 mg/kg/day for 14 days, with some countries using the weekly regimen (0.75 mg/kg once weekly for 8 weeks). Colours indicate requirement for G6PD deficiency testing before primaquine administration: blue-G6PD deficiency testing required, red-testing not required, with light red or blue for unobserved therapy and dark red or blue for directly observed therapy (DOT). Implementation of these policies, regimens used, and requirement for G6PD deficiency testing vary between countries. Table 5 provides details on countries' policies, implementation, and/or upcoming modifications.

Discussion
There is substantial variation in national policies for the use of primaquine, even among countries with similar patterns of malaria epidemiology, drug resistance, and G6PD deficiency prevalence. This variation is amplified by the disconnect between policy and practice resulting from divergent opinions on risks and benefits. In many cases in this study, respondents confirmed discordance between the national recommendation and the current practice, invariably leading to primaquine underuse. For single-dose primaquine as a P. falciparum gametocytocide, the risk assessment is relatively easy, whilst the assessment of benefit is more difficult. The currently recommended primaquine SLD is 2-3 times lower than the 0.75 mg/kg dose recommended previously and 14-28 times lower than the total radical curative dose. In large studies of mass treatment conducted in areas where the G6PD Mahidol variant predominated, SLD primaquine was not associated with clinically significant haemolysis [42]. Given the information on the safety of multiple-dose radical cure including the implementation of weekly 0.75 mg/kg regimens in areas such as Iran, where the severe G6PD Mediterranean is prevalent, and the characterisation of the dose-response relationship for haemolysis in G6PDd individuals, it seems highly unlikely that SLD primaquine does pose a clinically significant haemolytic risk [21]. G6PD deficiency testing is, therefore, considered unnecessary in this case. It is unclear at what level of transmission intensity it is no longer cost beneficial to give single dose primaquine as a P. falciparum gametocytocide. However, as malaria transmission is reduced actively in an elimination programme, every transmission source must be removed. There is no clear explanation of why SLD primaquine is not deployed in all countries aiming for elimination. However, once the SLD primaquine policy is adopted, every effort should be made to ensure high coverage. Optimised age-based dosing regimens for SLD primaquine are recommended, based on anthropometric data for specific populations [43,44]. An important impediment is the current lack of paediatric formulations [45,46]. Use of primaquine in radical curative regimens carries much greater risks of haemolysis in G6PDd patients, but it also provides substantial benefits to treated patients [4]. The risks are systematically overestimated by surveys in healthy subjects as G6PD deficiency prevalences in malaria are usually lower because of the malaria protective effect, and unsurprisingly, this difference appears to be directly proportional to the degree of deficiency. While G6PD deficiency testing is recommended officially [47] in many countries, in reality, it is not available at point of care, leaving prescribers to make their own decisions. Apart from the Indian subcontinent, there is very little use of primaquine in the private sector [37,38], which in many countries is the main source of antimalarial treatment. Reported adherence to unobserved 14-day radical cure regimens varies between different studies (adherence was considered poor in Papua Indonesia [48], where most patients are children; in 2 studies in Thailand [49,50]; and in studies from India, but good in other studies in Pakistan [an Afghan refugee camp] [51], Brazil [52], and the Thailand-Myanmar border [53]). This suggests that adherence is dependent on context and education practices. Although some countries recommend that primaquine be given as directly observed treatment (DOT) to allow the drug to be stopped if there are signs of haemolysis, the operational difficulties of providing a 14-day DOT may well mean many patients are partially treated or not given primaquine at all.
Where relapse rates are high (i.e., >50%, as in East Asia and Oceania), effective radical cure more than halves the incidence of vivax malaria. The benefit is substantial. If G6PD deficiency testing is unavailable, the correct approach to patient management is unclear as it depends on the prevalence and severity of G6PD deficiency in the patient's original population, the degree of anaemia, and the availability of blood transfusion in case of severe haemolysis, balanced against the probability of relapse and its pathological consequences and the likely dose regimen required for radical cure. More could be done to improve the safety of primaquine administration through better packaging and presentation of the drug as for Coartem (artemether-lumefantrine) [54], which is marketed as age-targeted blister packs, with pictorial instructions on how it should be taken. Concern over haemolytic risk may explain why many countries in East Asia and Oceania recommend a daily dose of 0.25 mg/kg when there is good evidence for the radical cure superiority of the higher dose currently recommended by WHO [18]. For patients with P. vivax or P. ovale malaria who are G6PDd, WHO recommends they should not receive daily primaquine. Instead, these patients may receive once weekly primaquine 0.75 mg/kg for 8 weeks to prevent relapses, provided they are under close medical supervision for signs and symptoms of acute haemolytic anaemia during the first 3 weeks of treatment and they have access to health facilities with capacity for safe blood transfusion [55].
If vivax malaria is to be eliminated, use of radical curative treatment will have to increase. Some heterogeneity in policies for radical cure of vivax malaria is understandable, given the geographic variation in P. vivax relapse risks and the prevalence and severity of G6PD deficiency, but this review suggests there is plenty of room for improvement. The gap between policy and practice is also too wide, and primaquine is underused. Informed review of risks and benefits should help to harmonise treatment policies. As is often the case in malaria, the strength of opinion seems inversely proportional to the quality and quantity of the evidence, so more information is needed on the safety of radical cure regimens in the more severe G6PD deficiency variants to inform policies and practices. Measuring adherence accurately is very difficult, but as it is such a critical determinant of therapeutic response; it deserves further behavioural, pharmacokinetic, and epidemiological studies to try and ensure people do adhere to current and future regimens. Widescale availability of point-of-care G6PD deficiency testing would undoubtedly improve prescribing safety, although the current 30% activity threshold will not identify the majority of female heterozygotes. Safer primaquine regimens could be developed. A new 8-aminoquinoline tafenoquine will be introduced soon. This provides radical cure in a single dose, but as it should not be given to patients with less than 70% of normal G6PD activity, it will require deployment of a new G6PD deficiency test. This will still leave a significant minority of patients (particularly women of childbearing age) requiring a safe primaquine regimen. Policies and practices both need frequent review.

Key learning points.
• Glucose 6-phosphate dehydrogenase (G6PD) deficiency, the most common human enzymopathy, is prevalent in tropical and subtropical areas where malaria is or was endemic. Primaquine, the only drug licensed for radical cure of vivax and ovale malaria, and the only licenced antimalarial with specific gametocytocidal activity against Plasmodium falciparum, causes dose-dependent haemolysis in G6PD-deficient individuals.
• Primaquine is underused (reflected in both policies and practices) because of low availability of G6PD deficiency testing, poor adherence to radical cure regimens, lack of paediatric formulations, and concerns about toxicity.
• The single low dose of 0.25 mg/kg of primaquine recommended by WHO in 2012 to be added to falciparum malaria treatment to reduce transmission in low-transmission settings is safe but has not been adopted by many countries, particularly in Africa.