Safety of primaquine in infants with Plasmodium vivax malaria in Papua, Indonesia

There are almost 12 million cases of Plasmodium vivax malaria each year [1]. The greatest burden of these infections is in poorly resourced communities where the disease is associated with severe anaemia and other adverse health and developmental effects in young children [2‐6]. Much of the morbidity and transmission potential of P. vivax infections is attributable to relapsing disease arising from dormant liver stages of the parasite (hypnozoites) [7, 8].

The only widely available hypnozoitocidal agent for radical cure of P. vivax infection is primaquine (PQ), an 8-aminoquinoline anti-malarial drug [9, 10] which has potential to reduce the adverse consequences of severe anaemia due to recurrent episodes of parasite induced haemolysis [11]. This is particularly important in equatorial regions where the risk of P. vivax relapse is high and the interval between relapses is short [12, 13].

Primaquine results in oxidant stress which can cause severe and potentially fatal, drug-induced haemolysis in glucose-6-phosphate dehydrogenase (G6PD) deficient individuals [14, 15]. G6PD deficiency is common in malaria endemic populations, reaching a prevalence of 40% in some regions [16]. The overall risk–benefit of primaquine therapy for vivax malaria, therefore, hinges upon the dangers of drug induced haemolysis weighed against the dangers of recurrent parasite induced haemolysis and other malaria-related morbidities.

In 2015, the World Health Organization (WHO) anti-malarial guidelines reduced the lower age threshold for recommending the use of primaquine from 4 years of age to those over 6 months old, with close monitoring for signs of haemolysis [10]. Although the WHO recommends that patients be tested for G6PD deficiency prior to administration of primaquine, in reality, G6PD testing is rarely available in poorly resourced endemic regions [17‐19].

This study examined the safety outcomes of pragmatic and discretionary use of primaquine in infants at Mitra Masyarakat Hospital in southern Papua, an area where the risk of severe anaemia associated with P. vivax relapse is high and the prevalence of G6PD deficiency is low [2].

Of the 3432 infants in the analysis, 1208 (35.3%) were less than 6 months of age and 2224 (64.8%) between 6 and 12 months of age. Overall, 93.2% (3432/3681) of infants received no primaquine, 0.9% (33) received LDP, 4.7% (174) received HDP and 1.1% (42) received either a single dose or an unknown dose of primaquine (Table 1 and Fig. 1). The proportion of infants treated with primaquine was similar in outpatients compared with inpatients and those with mixed infections compared to those with P. vivax mono-infections. Infants aged 9–12 months were more likely to receive HDP (n = 137 (10.7%)) than younger babies (n = 37, 1.5%) (Table 1 and Fig. 2). Prior to the change of treatment policy in March 2006, oral quinine was the most commonly prescribed oral schizontocidal agent (n = 484, 89.5%) (Fig. 3). Thereafter, 91.9% (n = 2887) of infants received DHA + PIP.

In view of the risk of haemolysis, limited capacity of cardiorespiratory compensatory mechanisms and the difficulty diagnosing G6PD deficiency in young infants, the WHO does not recommend primaquine for infants under the age of 6 months [10, 26, 27]. However the risks of primaquine use in this age group need to be balanced against the significant risks of recurrent P. vivax infections in the first year of life, which in Papua-Indonesia, also includes severe anaemia and both direct and indirect mortality [2, 28]. Approximately 65% of children presenting for the first time with P. vivax infection to hospital in Timika will represent to hospital within 12 months, 50% will represent with malaria within that time period and 25% will require admission (about 40% of these admissions are attributable to malaria) (Patriani et al., pers.commun.).

At the local hospital in Timika, primaquine was prescribed in 6.8% (249/3681) of infants presenting with P. vivax infection. The proportion of infants requiring admission to the hospital within 3 to 30 days was lower in those treated with HDP (5.7%) compared to infants who did not receive primaquine (8.6%). A plausible explanation for this is that primaquine prevented vivax relapse and thus morbidity associated with recurrent malaria. Previous study has shown that the overall effectiveness of unsupervised primaquine at RSMM is poor, but that there is some benefit in young children, potentially because of parental influence increasing adherence to treatment [25]. Alternatively, these observations could reflect bias caused by purposeful selection of less vulnerable infants for primaquine treatment. However, given the similar proportions of inpatients and outpatients treated with primaquine the latter explanation seems unlikely.

Quantifying the acute hemolytic effect of primaquine in infants is confounded by the haemolytic effects of malaria itself along with any other concomitant morbidities [15, 28]. The most appropriate metric for clinically relevant haemolysis is disputed. Since the fractional fall in hemoglobin is correlated positively with the starting Hb, the metric chosen were both development of severe anaemia (Hb < 5 g/dL) and a composite measure of a high fractional fall (> 25%) to an absolute Hb below 5 g/dL, at which level the risk of mortality rises significantly [28].

The mean Hb concentration in infants at baseline was similar (7–8 g/dL) between treatment groups. The crude frequency of severe anaemia and a clinically significant drop in haemoglobin was higher amongst patients who received HDP (18.5% and 5.3% respectively) compared to those who did not receive primaquine (4.5% and 3.5% respectively) whereas neither of these adverse haematological outcomes occurred in patients treated with low dose primaquine. One infant who had a clinically significant drop in haemoglobin after HDP had an initial Hb level of 6.4 g/dL falling to 4.6 g/dL by day 3, and died 9 days after presentation with bronchopneumonia (Table 3). Since biochemical laboratory tests were not available in this infant, severe hemolysis with acute renal failure and associated pulmonary oedema could not be ruled out.

Profound haemolysis leading to severe anaemia, is likely to prompt representation to hospital and be detected clinically by the attending physician. Hence the incidence of severe anaemia amongst those who did not have a repeat haemoglobin was probably low. If this assumption is true, the true frequencies of new onset severe anaemia or a clinically significant reduction in haemoglobin following high dose primaquine are likely to be substantially lower than those observed (as demonstrated in the sensitivity analyses). Nevertheless, the frequency of severe haematological events at a population level needs to be considered in the context of the prevalence of the susceptible phenotype, namely severe G6PD deficiency, which in Timika is estimated to be about 1.3%.

Three infants treated with LDP or HDP died following treatment with primaquine. All three had significant comorbidity (tuberculosis in one case, severe malnutrition, gastroenteritis, dehydration and acidosis in another case and bronchopneumonia). The initial haemoglobin was less than 8 g/dL in all three infants, and thus a small fall in haemoglobin attributable to either acute malaria or drug, associated with significant co-morbidities, could have contributed to their clinical decline [29].

The relevance of these findings to other endemic locations needs to be considered with caution. In the study population the prevalence of G6PD deficiency (< 30% activity) is low and absent in those of highland ethnicity, who make up more than 70% of patients presenting to the hospital (Pava et al., pers.commun.). Elsewhere in Indonesia, the G6PD deficiency prevalence is much higher, the dominant variants being Vanua Lava, Viangchan and Chatam which are associated with enzyme activity of less than 10% [16, 19]. Use of primaquine without prior G6PD testing in those regions, particularly in infants, would carry significantly greater risk and G6PD testing would be mandatory prior to primaquine administration

This study has several important limitations. Firstly infants who died at home or visited other health facilities following their initial presentation to RSMM with vivax malaria will not have been detected. In a treatment seeking survey conducted in 2013, family members from households in which a member had died within the preceding year, reported that the death had occurred at the RSMM hospital in 30% of cases [30]. Hence whilst many deaths will have been identified by our surveillance, some would have been missed. This attrition bias is likely be similar for infants treated with or without primaquine. Secondly it was not possible to confirm that infants completed a full course of primaquine after discharge from the hospital, and a previous study showed poor effectiveness, which was assumed to be secondary to poor adherence to the 14-day primaquine regimen. Thus, incomplete treatment will have resulted in underestimation of the frequency of adverse events that would be observed in a fully supervised regimen [25]. Finally haemoglobin measures were not undertaken routinely. Most patients had their haemoglobin measured only once and this was generally on the day of admission or representation. Thereafter haemoglobin was only measured when there was clinical concern. In view of this, a composite measure of significant haemolysis was chosen, which comprised a high fractional fall to below an absolute concentration of 5 g/dL—a point at which anaemia would presumably have been apparent clinically, triggering laboratory investigation. The true minimum haemoglobin concentration for each individual may have been lower than the lowest measured concentration recorded in the hospital database.

The clinical decision to give primaquine as radical treatment for P. vivax malaria and mixed infections in this vulnerable group without G6PD testing should be made with great caution, particularly in areas with a higher prevalence of G6PD deficiency. If primaquine is to be prescribed, parents and guardians must be informed of the early signs of haemolysis and advised to seek immediate medical attention should haemolysis occur [15, 31]. Considering the potential life-threatening consequences of haemolysis from either recurrent P. vivax malaria or antirelapse treatment in G6PD deficient infants, there is an urgent need for a highly sensitive point of care G6PD test to ensure that radical cure can be deployed safely to those who need it most.