Comparison of the Cumulative Efficacy and Safety of Chloroquine, Artesunate, and Chloroquine-Primaquine in Plasmodium vivax Malaria

Vivax malaria relapses frequently even in low-transmission settings. Chloroquine delays but does not prevent recurrences. Adding primaquine to a slowly eliminated schizonticide significantly reduces recurrences and improves hematocrit, but this advantage is offset by hemolysis in G6PD-deficient females.

Plasmodium vivax, which is becoming the main cause of malaria outside sub-Saharan Africa, exerts considerable morbidity by causing repeated relapses. In Southeast Asia and Oceania, P. vivax relapses early and frequently [1,2]. Prevention of relapse requires radical treatment with primaquine (PMQ), which carries the attendant risk of hemolysis in glucose-6-phosphate dehydrogenase (G6PD)-deficient persons. As G6PD deficiency is common in tropical regions and testing is seldom available, radical-cure PMQ regimens are often not given despite being widely recommended. For nearly 70 years the standard treatment for P. vivax malaria has been chloroquine (CQ), although resistance is increasing in many areas [3]. Chloroquine is eliminated slowly, resulting in posttreatment suppression of emerging P. vivax malaria recurrences for over a month following treatment. The artemisinin derivatives are more potent [4] but are rapidly eliminated and therefore do not provide posttreatment suppression of recurrent infections. An important question is whether slowly eliminated antimalarials delay or prevent early relapses [5]. Killing all asexual parasites emerging from the liver 2 weeks after the acute illness, would reduce the total number of relapses substantially and would be of public health M A J O R A R T I C L E benefit. This would support use of slowly eliminated partner antimalarial drugs if artemisinin combination therapies are used to treat P. vivax malaria. To answer this question and to assess the radical curative efficacy of PMQ (in a context where it was not routinely given for radical cure), a 3-way comparison was conducted of CQ, artesunate (AS), and CQ + PMQ.

Shoklo Malaria
Research Unit (SMRU) is located along the border of northwest Thailand. Malaria transmission in this tropical mountainous region is low and seasonal [6,7]. All patients ≥6 months and weighing ≥7 kg presenting to SMRU outpatient clinics with microscopy-confirmed uncomplicated P. vivax monoinfections were screened. Exclusion criteria included pregnancy, severe malaria, hematocrit <25%, allergies to antimalarial drugs, blood transfusion in the last 3 months, antimalarial use in the last 4 weeks, and inability to comply with study procedures. Written informed consent was obtained from patients or their carers with a literate witness present if unable to read in their preferred language.

Procedures
A full medical history and physical examination were performed. A malaria smear, hematocrit, complete blood count, G6PD fluorescent spot test, and urine β-human chorionic gonadotropin pregnancy test were performed and blood samples taken for human G6PD, parasite genotyping, and parasite culture (for parasitemia >300/500 white blood cells with >80% ring stages [8]). All enrollment malaria blood smears were recounted at the central laboratory; parasite density was the mean of the 2 readings. All follow-up positive slides and 10% of randomly selected negative slides were rechecked. G6PD genotyping for Mahidol variant was performed on all female participants using an established polymerase chain reaction-restriction fragment length polymorphism protocol [9]. Capillary whole blood CQ and desethylchloroquine concentrations were measured using a liquid chromatography-mass spectrometry assay (Supplementary Appendix 1).
Patients with G6PD deficiency were randomized to arms 1 or 2 only. All doses were supervised. For children unable to swallow tablets, whole tablets were crushed and mixed with water, and the correct dose was administered in suspension. The full dose was repeated for vomiting within 30 minutes. Half the dose was repeated for vomiting between 30 and 60 minutes.

Follow-up
Temperature and a malaria blood smear were measured daily until afebrile and parasite negative. Fever clearance was the interval to an aural temperature <37.5°C on 2 consecutive measurements. During follow-up visits, a medical history, physical examination, malaria smear, hematocrit, adverse event record, concomitant medication review, and eligibility assessment were performed. Chloroquine levels were measured on day 6 (±1 day) and at recurrences within 28 days of treatment. Recurrences of P. vivax malaria were confirmed by microscopy, enrollment procedures repeated, and patients re-treated with the same study drug. Follow-up visits and procedures were restarted as if newly recruited. The total follow-up duration remained 1 year from enrollment. If Plasmodium falciparum or mixed infection malaria occurred, standard AS-mefloquine treatment was given and follow-up continued without interruption. Radical cure with PMQ was given after 9 recurrences or earlier if clinically indicated. Subjects were censored if consent was withdrawn or lost to follow-up. Adverse events were recorded daily during treatment and weekly until day 28. Hematinics were given if the hematocrit fell below 30% (<34% in children <2 years old) and a blood transfusion if it fell below 18%. If cyanosis was evident, transcutaneous methemoglobin levels were checked (Masimo Radical-7).

Sample Size Calculation
An estimated 10% of CQ-treated patients were expected to have P. vivax recurrences within 28 days. A sample size of 200 patients per treatment arm allowed a difference of >8% to be detected using the log-rank test with 80% power, 95% confidence, and 20% loss to follow-up. To account for exclusion of patients with G6PD deficiency, an additional 60 patients were added for a final sample size of 660.

Statistical Analyses
The primary endpoint was P. vivax malaria recurrence within 28 days. Secondary endpoints were interval to first recurrence, P. vivax malaria recurrences during 1 year, relationships with CQ levels, and adverse events within 28 days. All evaluable subjects were included in a modified intention-to-treat analysis to assess antimalarial drug efficacy. Associations were assessed using χ 2 or Fisher exact tests and group comparisons and relationships assessed were using Student t test, linear regression, nonparametric K-sample test, Wilcoxon rank-sum test, or logistic regression as appropriate. The multivariate analysis included all variables significant at the 5% level in a univariate model. Steady-state hematocrit was defined as the mean of all measurements during the year after day 42 provided there was no interim malaria recurrence [10]. Efficacy was analyzed using the Kaplan-Meier method with Cox regression used to evaluate the risk factors for recurrence (ie, sex, age, hematocrit, or parasitemia) and the effects of CQ levels. An Anderson-Gill extended Cox model [11] was fitted to estimate the total effect (effect of the treatment + effect of dependency between recurrences) of treatment on the hazard of recurrence. The duration of posttreatment prophylaxis after CQ (difference in median times to recurrent infection between the CQ and AS groups) was estimated as 20 ± 5 days. Confidence intervals (CIs) were calculated by bootstrapping. Statistical analyses were performed using Stata version 13 and 14 software (StataCorp, College Station, Texas).
This study was approved by the Mahidol University Faculty of Tropical Medicine Ethics Committee (MUTM 2010-006) and the Oxford Tropical Research Ethics Committee (OXTREC 04-10), and was registered at ClinicalTrials.gov (NCT01074905).

Second and Subsequent Recurrences
The proportions (95% CI) of patients who had a second P. vivax recurrence were similar in the AS (79% [72%-84%]) and CQ  Figure 3). If the background rate of new infections (estimated from the CQ + PMQ group) is  subtracted from the recurrence rate, then the estimated relapse rate per person-year was 4.25 in the AS group and 3.19 in the CQ group, with 90% of all relapses in the CQ group occurring within 9 weeks (Figure 2 ng/mL) was similar to the first treatment (397 ng/mL; P = .584), but in recurrent infections lower levels were neither associated with risk nor timing of subsequent recurrence. Chloroquine levels exceeded 100 ng/mL in 10 of 225 (4%) infections on the day of subsequent recurrence. Of the 58 patients with a subsequent recurrence before day 28, 16 (28%) had levels >100 ng/mL.

Adverse Events up to Day 28
Adverse event rates were similar in all treatment groups. Abdominal pain was more common in the CQ + PMQ than the CQ group but not the AS group (Table 3). More patients in the CQ + PMQ group were treated for anemia than in the AS  or CQ groups; 4 of 9 (44%) were G6PD-heterozygous females ( Table 3). One death from accidental head injury occurred 18 days after completing AS treatment. No blood transfusions were required.

DISCUSSION
Although there is in vivo and in vitro evidence of CQ resistance on the northwestern border of Thailand [12,13], most P. vivax malaria infections remain sensitive and CQ remains clinically efficacious in treatment. In areas with high-grade CQ resistance in P. vivax, national malaria programs now recommend artemisinin combination therapies. The main therapeutic challenge is the high relapse rate. In Southeast Asia and Oceania, P. vivax relapses frequently and repeatedly. Following rapidly eliminated drugs such as AS or quinine, relapses become patent approximately 3 weeks after starting treatment. More slowly eliminated drugs such as CQ suppress the asexual parasitemia of the first relapse, but there has been uncertainty whether the first relapse observed after these drugs derives from protracted suppression (ie, delay) of the first relapse, or clearance of this infection and emergence of the second relapse. This study, together with previous data [14][15][16][17] (Figure 1), supports the former explanation-that is, relapses are predominantly delayed rather than prevented.
More than 90% of all P. vivax recurrences (mainly relapses) occurred by week 16. This suggests that studies evaluating antirelapse treatments in Southeast Asia and Oceania should have at least 16 weeks (~4 months) of follow-up to capture the majority of relapses. Primaquine (total dose 7 mg/kg) was highly efficacious with an estimated antirelapse (radical cure) overall efficacy of at least 92%. Primaquine also has asexual stage activity [18] which would be expected to augment cure rates in resistant infections. Some of the apparent radical cure failures may have been explained by reduced cytochrome P450 2D6 (CYP2D6) PMQ bioactivation [19]. Approximately 20% of the population in the study area carries the CYP2D6*10 allele [20], so 4%-5% of patients would have been intermediate metabolizers. CYP2D6 genotyping results from this study will be reported elsewhere. The cumulative benefit of providing radical cure in preventing P. vivax malaria is enormous. In the Southeast Asia region, an estimated 4.6 million cases would have been prevented if all P. vivax infections were treated with PMQ radical cure [21].
The main clinical concern both from recurrent P. vivax malaria and from PMQ treatment is anemia. In this low-transmission area with ready access to diagnosis and treatment, recurrent P. vivax malaria caused only small reductions in hematocrit; these reductions lessened with successive recurrences [22]. In contrast, on the island of New Guinea where malaria transmission is high, vivax malaria is associated with life-threatening anemia in young children. However, using PMQ caused substantially greater hematocrit reductions (~8 times the effect of malaria) in G6PD-heterozygous females [23]. Even so, no blood transfusions were required. Thus, the major benefit from radical cure with PMQ (in the higher doses necessary in SE Asia) in preventing recurrent malaria and attenuating any associated anemia is offset by the risk to G6PD-heterozygous females diagnosed as "G6PD normal" with current tests who may still have clinically significant oxidant hemolysis [24,25].
One of the study limitations was incomplete follow-up. This resulted from a combination of migration, environmental factors (flooding), the nature of forest work for young males, and long study duration.
In conclusion, CQ continues to be an efficacious treatment of P. vivax malaria along the Thailand-Myanmar border, although low-grade resistance has emerged. The benefits of slowly eliminated antimalarials in delaying early relapses are small in comparison with high-dose PMQ radical cure, which prevents nearly all relapses. However, when only G6PD qualitative screening is used, this regimen may cause significant hemolysis in G6PD-deficient heterozygous females. Despite this, PMQ should be used more widely. Greater availability of quantitative G6PD testing or development of safer radical curative regimens would ameliorate the risks.

Supplementary Data
Supplementary materials are available at Clinical Infectious Diseases online. Consisting of data provided by the authors to benefit the reader, the posted materials are not copyedited and are the sole responsibility of the authors, so questions or comments should be addressed to the corresponding author.