Defining Plasmodium falciparum Treatment in South West Asia: A Randomized Trial Comparing Artesunate or Primaquine Combined with Chloroquine or SP

The study recruited patients from five Afghan refugee villages within an 80 km radius of Peshawar, Khyber Province (formerly North West Frontier Province), Pakistan. The 5 villages were established in the early 1980s and each has a history of falciparum transmission. Three (Adizai, Naguman and Yakka Ghund) are situated on the banks of the Kabul river, and two (Mohammed Khoja and Kotki) are situated to the south of Peshawar in Kohat district. Malaria transmission is seasonal with vivax malaria occurring from March to November and falciparum from July to December.

Participants were Afghan refugees who were permanently resident in the Pakistan villages. A minority might have acquired their infections in Afghanistan, but admission criteria required 4–6 weeks of follow-up which would have excluded the more regular travellers.

The study was conducted through three sites with well-managed clinics run by Non-Governmental Organisations (NGO) and government agencies. Site 1 was Adizai village which also served the population of nearby Naguman. Site 2 was Yakka Ghund village. Site 3 was Kotki which also serviced Mohammed Khoja village. Health staff from the NGO HealthNet TPO were seconded to each clinic to manage the recruitment and follow-up of patients.

The study took place over three malaria seasons from 2000 to 2003. In the first season only site 1 was used, in the second season site 2 was added, and in the third season site 3 was added. The stepped inclusion of sites resulted from unexpectedly low recruitment rates at the first study site, and is summarized in Table 1. An earlier in vivo survey across all refugee camps along the length of the 1000 km North-South axis of Khyber Province showed little heterogeneity in resistance frequency to chloroquine and SP [30]. (Note that SP is the only drug name abbreviated throughout the text. Abbreviations for chloroquine (CQ), primaquine (PQ) and artesunate (AS) are used when referring to specific study arms (e.g. CQ+PQ) but not at other points in the text.)

Study design and procedures. Individuals presenting with clinical symptoms and diagnosed microscopically with falciparum malaria were referred to trial staff for further assessment. Inclusion criteria were: over two years of age, not pregnant, P. falciparum mono-infection, more than 1 asexual parasite per 10 fields, no other serious disease, resident in the refugee village for the full period of follow-up, no verbal report of antimalarial use during the last 21 days, and no signs of severe malaria.

Individuals meeting the inclusion criteria and giving informed consent were allocated using randomization tables stratified by age and sex to the following treatment arms: chloroquine (CQ¹) (25 mg/kg) over 3 days; CQ (25 mg/kg) over 3 days plus primaquine (PQ) (0.5 mg/kg) on the last day of treatment; CQ (25 mg/kg) over 3 days plus artesunate (AS) (4 mg/kg per day) over 3 days; sulfadoxine (25 mg/kg) and pyrimethamine (1.25 mg/kg) (SP) on day 0; SP (25:1.25 mg/kg) plus PQ (0.5 mg/kg) on day 0; SP (25:1.25 mg/kg) on day 0 plus AS (4 mg/kg per day) over 3 days. The primaquine co-treatment regimens were based on recommendations for use of primaquine as a gametocytocide treatment for falciparum malaria [31]. On treatment days when any specific arm did not require a partner drug, a placebo was substituted.

Patients were not allocated to all six treatment arms at each site (Table 1). During years two and three, at any one site, patients were allocated to either one of the three SP treatment arms or one of the three chloroquine treatment arms; only in year one were patients allocated to all six treatment arms in the original site.

For each of the study sites patients were stratified by age and sex and assigned consecutive patient numbers at enrolment. Treatment groups were pre-assigned to patient numbers using simple randomization lists generated using Excel (Microsoft Corp., Seattle, USA). The assignment of treatment group to patient number was concealed until after enrolment. The random allocation sequence was generated by two of the investigators (KK, MR) neither of whom enrolled or assessed patients.

The brands and manufacturers/suppliers were: SP (Fansidar ®, Roche), chloroquine (Nivaquine ®, Beacon), primaquine (IDA), artesunate (Plasmotrim ™, Mepha). Although chloroquine resistance was known to exist in the study area, chloroquine was the first line treatment policy of the Government of Pakistan and the UN refugee agency (UNHCR) at the time and was therefore included.

The trial manager, who allocated the treatment arms and administered treatment, was not blind to the study drugs and allocations. Recruited patients, microscopists and health workers were partially blinded; full blinding was not achieved given (i) the appearance of the different drugs, and (ii) the different times of follow up for the SP arms compared to the chloroquine arms.

Patients were given directly observed treatment (by the trial manager), monitored for 30 minutes and re-dosed if vomiting occurred. Health workers then recorded the other day 0 biomedical parameters.

Patients were asked to return on each day of treatment (days 0, 1, 2) and on days 3, 7, 14, 21 and 28 post treatment. Patients in the three SP arms were also followed up on days 35 and 42 for detection of late recrudescence [32]. On each day of follow up thick and thin smears were taken, clinical symptoms recorded and blood spots collected on Whatman No. 1 filter paper for parasite typing and differentiation of recrudescent from new infections. Haematocrits and blood spots were taken either on the day of failure or on day 28. Patients not presenting at the clinic were followed up at home. Criteria for withdrawal were reported administration of additional anti-malarial drugs (protocol violation), emergence of any concomitant febrile illness that interfered with outcome classification (including a non-falciparum malaria, for which the withdrawn study patient would be given appropriate first line treatment), parasitaemia still present on day 7, or signs of severe malaria developing before completing the initial treatment course.

Patients found to be parasitaemic on any day after day 3 were treated with SP (the official second line treatment) or with SP and mefloquine (Fansimef, Roche) for those whose initial treatment was SP based. SP resistant parasites were sensitive to the mefloquine component of Fansimef, which was used in lieu of mefloquine monotherapy which was unavailable in the study area. Those who developed severe malaria, severe anaemia or other complications were referred to Khyber Hospital, Peshawar for treatment.

Laboratory tests. Thick and thin blood smears were stained with 2% Giemsa solution. All slides were read on the day of collection by a microscopist based at each site. Differential diagnosis of vivax and falciparum (trophozoites and gametocytes) was by examination of thick and thin smears according to standard microscopy methods [33]. Trophozoites and gametocytes were counted against 200 white blood cells (WBC) from the thick blood smear on the assumption of a WBC count of 8000/µl. A smear was declared negative if no parasites were seen after examining 100 fields. Slides from sites 2 and 3 were re-examined for accuracy by the site 1 microscopist. Comparison was made between the parasite counts made by the three microscopists. The mean variation in parasite count for trophozoites and gametocytes was less than 5%. Owing to electricity supply deficiency, the haematocrit microcentrifuge was only used at sites 1 and 3.

Outcome measures. Patients completed the trial if treatment was administered fully and all follow-up appointments conducted, or if they failed treatment on any day of follow-up. The primary endpoint of the trial was any clinical or parasitological failure up to day 28, although a subset of patients in the SP treatment arms was also followed until day 42. Patient outcomes were classified under the WHO parasitological classification system of sensitive (S) or resistant (RI, RII, RIII) infections with those whose outcomes were classified as S (sensitive) being treatment successes and those with outcomes of RI, RII or RIII being classified as treatment failures [34]. This classification system was still commonly used at the time of the study and the ethical clearance for this study (received in 2000) was based on a protocol using this classification system. Patients were also classified against the newer WHO treatment outcome system [35], [36]. Presenting the outcomes under both of these systems allows better comparability with data past and present [37]. For the WHO classification system standard definitions were applied: adequate clinical and parasitological response (ACPR), early treatment failure (ETF) or late treatment failure (LTF) which incorporated standard definitions of late clinical and parasitological response [35]. Those cases with ETF or LTF were classified as treatment failures, and ACPR was classified as treatment success if they completed the follow-up period.

Other parameters examined were time to fever resolution (with fever defined as axillary temperature ≥37.5°C), asexual parasite clearance, gametocyte clearance and gametocyte carriage on or after day 7.

Molecular characterisation. PCR genotyping was conducted on a subset of samples to distinguish recrudescent from new infections using the protocol described by Brockman et al [38]. P. falciparum genes msp-1, msp-2 and glurp were genotyped according to polymorphisms present at variable loci on day 0 and day of failure. Approximately 50% of cases were available for genotyping.

PCR corrected outcomes were applied as a secondary analysis using redefined outcomes based on the number and frequency of true recrudescent infections and reclassifying cases with new infections as treatment successes, excluding those with indeterminate outcomes or negative PCR results. PCR testing was performed at the Shoklo Malaria Research Unit, Thailand.

We also conducted an analysis at the London School of Hygiene and Tropical Medicine (LSHTM), UK, for known resistance-associated mutations to chloroquine and antifolate drugs. Samples collected at enrolment were tested for known mutations in the pfcrt, pfmdr1, pfdhfr and pfdhps genes [39], [40]. PCR and sequence specific oligonucleotide probe assays were used to analyse nucleotide polymorphisms at pfcrt codon 76, pfmdr1 codons 86 and 184, pfdhfr codons 16, 51, 59 and 108 and pfdhps codons 436, 437, 581 and 613 [41].

The sample size required to detect a difference in treatment failure or gametocyte carriage with 80% power and 95% confidence (two-sided significance level for type 1 error of 0.05) were calculated using the following predictions. In the chloroquine arms (i) the estimated frequency of failure in chloroquine monotherapy arm was 30% and in each of the combination arms (CQ+PQ or CQ+AS) it was 10% or less; (ii) the estimated proportion of gametocyte positive individuals after 7 days in the chloroquine arm was 50% and in each of the combination arms it was 25%. In the SP arms (i) the estimated frequency of failure in the SP arm was 10% and in each of the combination arms (SP+PQ or SP+AS) it was 1%; (ii) the estimated frequency of gametocyte positive patients after 7 days in the SP arm was 50% and in each combination arm it was 25%. The estimated samples sizes were 64 per arm in the chloroquine arms and 121 per arm in the SP arms. For gametocyte carriage the required sample sizes were 65 per arm. The target sample sizes for the study were 65 for the chloroquine arms and 121 for the SP arms. To allow for 15% loss to follow up the targets for recruitment were 76 for the chloroquine arms and 142 for the SP arms.

The primary aims of the study were to evaluate: 1) the relative efficacy of the combination drugs in providing parasite clearance with no recrudescence compared to monotherapy, and 2) the effect on gametocyte clearance. The secondary aim was to determine the proportion of individuals in each arm with classifiable treatment failure.

The primary outcome was the proportion of individuals in each treatment group classified as a treatment failure during the 28 days follow-up compared to the respective monotherapy. The odds of treatment failure between each of the study groups at day 28 were estimated after adjusting for potential confounders (age, sex, parasitaemia, PCV, study site) using logistic regression analysis or Mantel-Haenszel χ² test. Treatment failure was also analysed using the definitions of ACPR and ETF or LTF, and parasitological classifications using the S-R scale in addition to the PCR corrected analysis [35]. Time to event data (time to treatment failure) were analysed with Kaplan Meier survival analysis.

For assessing potential effects on gametocyte carriage, the primary outcome was the presence or absence of gametocytes on day 7 with a secondary analysis examining the presence or absence of gametocytes and the geometric mean gametocyte density on each day of follow-up.

All data were entered in Microsoft Excel (1997) by one data clerk and the database checked patient by patient against paper records by a second data clerk. Analysis was performed using STATA 10.0 (STATA Corp, College Station, TX, USA).

The study protocol was approved by the ethics committees of the Pakistan Medical Research Council and LSHTM, UK. All patients or their guardians gave written informed consent to participate. The trial was registered under clinicaltrials.gov [NCT00959517↗].

A total of 355 cases of microscopically confirmed falciparum malaria were enrolled into the study between July 2000 and December 2002. Table 2 shows enrolment characteristics and Figure 1 shows the trial profile. Patients were recruited to the chloroquine arms in line with recruitment targets. Insufficient numbers were recruited to the SP arms. This was due to considerably lower than expected malaria cases occurring in these locations during the study period (in contrast to the years leading up to the study). Additional sites were added over the three-year study period to increase recruitment but it was not possible to continue the trial for longer because of resource constraints. chloroquine arms met the recruitment targets because of the site allocations and unexpected pattern of case loads in the third year.

Characteristics of patients recruited into the 3 chloroquine arms were broadly similar, as were the three SP arms. The SP+AS arm showed slightly higher asexual parasite densities on enrolment than other groups (Table 2).

Of 355 patients enrolled, 38 (10.7%) were either withdrawn or lost to follow up leaving 317 for possible inclusion (Figure 1). Of these 292 (82%) were evaluable for parasitological outcomes because 25 patients classified as RII parasitological failure did not meet the definition of clinical failure, i.e. they did not have fever on day 3. Likewise, 308 (87%) were evaluable for clinical outcomes because 9 patients who had classifiable Early Treatment Failure did not fit the definition for parasitological failures (they did not reach day 7 of the trial). Of the 266 patients followed up for 42 days, 221 (83%) were evaluable for clinical and parasitological outcomes. All treatments were well tolerated and no severe or serious adverse events were recorded during the study.

Clinical Outcomes. The 28-day failure rate in the CQ and CQ+PQ arms was 81% and 73% respectively (Table 3, Figure 2A). The addition of artesunate improved the treatment response, with a treatment failure rate of 27%. The failure rates in the SP groups were 16% or less (Table 3) with the combination of SP+AS having the lowest failure rate (1/41 [2.4%]). Neither SP+AS nor SP+PQ arms were significantly different to the SP arm at the 28-day end point (Figure 2B).

None of the potential confounding factors examined (age, sex, parasitaemia, study site, PCV) were associated with treatment outcome so adjustments were unnecessary in the final regression analysis. Crude odds ratios are presented.

Clinical and parasitological outcomes are shown in Table 4. Only 32/133 (24%) of the individuals who failed had fever on the day of failure; the majority of ‘late treatment failures’ therefore fell into the category of ‘late parasitological failure’. Some recent history of fever may have gone unreported.

Trophozoite clearance time (days until negative smear) was lowest for the artesunate combination arms. Addition of primaquine to either chloroquine or SP did not appear to affect clearance times. Clearance times for each arm (median (interquartile range)) were: CQ, 3 days (2–7); CQ+PQ, 3 days (2–7); CQ+AS, 2 days (1–2); SP, 2 days (2–3); SP+PQ, 2 days (1.5–3); SP+AS, 1 day (1–2).

Of the 133 failures, 79 (47%) matched pairs were available for PCR evaluation of MSP-1, MSP-2 and GLURP. Matched pairs were collected on day 0 and on the day of failure. Outcomes of the PCR are shown in Table 5, which excludes those with indeterminate results. Most re-infections took place at one site (Site 3, Yakka Ghund) where malaria transmission was higher than in the other villages. Treatment outcomes were reclassified for those failures where matched pairs were available and used to give adjusted failure rates for the sample (Table 5). Although the number of failures in the chloroquine arms was reduced by the correction, the failure rates remained high at >50% for chloroquine monotherapy. The PCR adjusted treatment failure rate of the CQ+AS combination (9%) was considerably lower that the in vivo rate (28%).

In the subset of patients followed-up for 42 days, recrudescence between day 28 and 42 was rare: only 8/317 (2.5%) of individuals classified as adequate clinical and parasitological response (ACPR) or sensitive (S) at day 28 went on to fail by day 42 (1 in the CQ arm, 2 in the CQ+PQ arm, 3 in the CQ+AS arm, 1 in the SP arm and 1 in the SP+PQ arm). Failure rates at day 42 were therefore similar to day 28. The failures between 28-day and 42-day rates were not corrected by PCR and hence could have resulted from new infections.

Molecular outcomes. Molecular analysis of drug resistance-associated alleles was conducted on samples taken on day 0 (Table 6). The major marker of chloroquine resistance, pfcrt-76T, was fixed at 100%. Mutations in pfmdr1 are known to modulate resistance to quinoline and other drugs. The prevalence of chloroquine-resistance associated allele pfmdr1-86Y was 13.6% in this sample. Mutations associated with pyrimethamine resistance (on the pfdhfr gene) were seen frequently; 108 N was found in all but one of 74 typable samples (98.6%) and 59R occurred in 66/76 (86.8%) samples. However, 51I was only detected in 5/75 (6.7%) samples. Therefore pfdhfr ‘double mutants’ were common (at codons 108+59) but ‘triple mutants’ were rare. Interestingly, we observed one sample with the pfdhfr mutations 16V+108T. These are reportedly selected by the antifolate drug cycloguanil, rather than pyrimethamine [42]. No mutations on the dhps gene were seen. These PCR results broadly match what would be expected from the clinical outcomes. The pfdhr/pfdhps mutation profile suggests that the parasites would be killed by a full therapeutic dose of SP, but that some tolerance to lower doses existed.

The addition of artesunate or primaquine to chloroquine or SP reduced gametocyte carriage on day 7 (Table 3). Combining chloroquine or SP with artesunate succeeded in eliminating gametocytaemia from more individuals by day 7 compared to monotherapy or co-treatment with a single-dose of primaquine.

Figure 3 shows the proportion of individuals with patent gametocytaemia over 28 days of follow-up. In the chloroquine and SP arms, gametocyte carriage persisted at day 28 in >30% of individuals in the CQ arm and >70% of individuals in the SP arm. The addition of primaquine reduced gametocyte carriage; the effect was more pronounced with chloroquine than with SP. The proportion of gametocytaemic individuals was lowest in the artesunate arms and the difference was evident within two days of the start of treatment. Peak gametocyte densities occurred 7 days after the start of treatment (Figure 4). Gametocyte density was higher after SP than after chloroquine treatment. Both primaquine and artesunate reduced density to low levels by day 7.

Most patients did not have gametocytaemia on day 0 (291/355 (82.0%) had no gametocytes on day 0). However, the presence or absence of sexual stage parasites on day 0 was an important explanatory variable in the secondary analysis; patients who presented with gametocytes on day 0 were more likely to be gametocytaemic on day 7 than individuals who were not gametocytaemic on day 0 regardless of treatment group (Mantel-Haenszel OR: 3.5 [95%CI: 2.0–6.4], p<0.001). We therefore conducted further analysis comparing those who presented with gametocytes on day 0 and those who did not.

The artesunate treatments were more effective at preventing the emergence of gametocytaemia between day 0 and 7 than in removing established gametocytaemia already present on day 0 (Table 7). For example, among the CQ+AS group who did have patent gametocytaemia on day 0, 8/9 were still gametocytaemic on day 7 after treatment. This was not significantly different from the proportion gametocytaemic on day 7 (11/12) after treatment with chloroquine monotherapy (Fisher's exact test p = 1.0). By contrast, in the chloroquine arms of the patients who did not have patent gametocytaemia on day 0, only 11% (7/63) were gametocytaemic on day 7 when treated with CQ+AS as compared to 47/56 (84%) among those treated with chloroquine monotherapy (χ² p<0.001). Similar trends were evident in the SP arms treated with artesunate; prevalence of gametocytaemia on day 7 was significantly lower in the SP+AS group that was not gametocytaemic on day 0 than in the group that was (Fisher's exact test p<0.001). This indicates that artesunate is less active against older gametocytes than against those newly emerged or immature forms that are not yet emerged. Primaquine showed no such trend: the proportion of patients gametocytaemic on day 7 was not significantly different for patients who were or were not gametocytaemic on day 0. Primaquine therefore appears to be active against gametocytes of all ages.

Primaquine was more effective at controlling gametocytaemia when combined with CQ than with SP, regardless of whether individuals were gametocytaemic at the start of treatment (MH-OR = 0.22, P = 0.001) (Table 7). By contrast, artesunate seemed as effective when combined with chloroquine as it was when combined with SP (MH-OR = 1.04, P = 0.83). Though, as discussed, artesunate was more effective at preventing development of patent gametocytaemia than in clearing existing gametocytaemia. Artesunate was better than primaquine at preventing patent gametocytaemia among those that were not gametocytaemic at day 0 enrolment (CQ: OR = 0.30, P<0.001; SP: OR = 0.18, P<0.001).

Using the presence of gametocytes on day 7 as the secondary endpoint, adjusted for the presence of gametocytes at day 0, both primaquine and artesunate with chloroquine or SP were more effective than their respective monotherapies at reducing the presence of gametocytes (Table 8).

Compared to monotherapy with chloroquine or SP, co-treatment with artesunate was negatively associated with having gametocytes on day 28 whether the accompanying drug was chloroquine or SP (Table 8). The presence of gametocytes on day 0 was not associated with the presence of gametocytes on day 28 (MH-OR, adjusted for treatment arm: 1.0, p = 1.0) suggesting that older gametocytes had cleared by day 28, regardless of treatment.