No Seasonal Accumulation of Resistant P. falciparum when High-Dose Chloroquine Is Used

Background Potentially chloroquine resistant P. falciparum, identified by the 76T haplotype in the chloroquine resistance transporter (pfcrt 76T), are highly prevalent throughout Africa. In Guinea-Bissau, normal and double dose chloroquine have respective efficacies of 34% and 78% against P.falciparum with pfcrt 76T and approximately three times the normal dose of chloroquine is routinely taken. Proportions of pfcrt 76T generally increase during high transmission seasons, as P.falciparum with pfcrt 76T commonly survive treatment with normal dose chloroquine. In Guinea-Bissau, there should be no seasonal increase of pfcrt 76T if the high doses of CQ commonly used are effective. Methods and Findings P. falciparum parasite density, age, sex, the proportion of chloroquine resistance associated haplotypes pfcrt 76T and P. falciparum multidrug resistance gene 1 86Y were assessed in 988 samples collected from children between 2002 and 2007. There was no seasonal accumulation of any allele. During the high and low transmission periods the pfcrt 76T proportions were 24% (95% CI, 21–27%) and 26% (95% CI, 20–33%). There was no significant change of pfcrt 76T (OR 1.05, 95% CI; 0.94–1.16 p = 0.39) or pfmdr1 86Y (OR 0.92, 95%CI; 0.83–1.01 p = 0.08) proportions between 2003 and 2007. Lower median parasite density (P.falciparum/µl) was associated with pfcrt 76T (15254 [95% CI, 12737–17772]; n = 164) compared to pfcrt 76K (18664 [95% CI, 16676–20653]; p = 0.003; n = 591). Similarly, pfmdr1 86Y was associated with a lower median parasite density (16320 [95% CI, 13696–18944]; n = 224) compared to pfmdr1 86N, (18880 [95% CI, 16701–21059]; P = 0.018; n = 445). Conclusions In contrast to the rest of Africa, P. falciparum parasites resistant to normal dose chloroquine do not have a selective advantage great enough to become the dominant P.falciparum type in Guinea-Bissau. This is most likely due to the efficacy of high-dose chloroquine as used in Guinea-Bissau, combined with a loss of fitness associated with pfcrt 76T.


Introduction
Chloroquine resistant (CQR) Plasmodium falciparum spread through Africa during the 80's and 90's and was first described in Guinea-Bissau in 1990 [1]. Until June 2008 chloroquine (CQ) remained, by far, the most commonly used antimalarial in the country. In Guinea-Bissau, as in most other areas of Africa, CQR is associated with a mutation in the CQR transporter (pfcrt K76T) [2,3]. Despite the presence of the pfcrt 76T mutation and continued CQ use that should select CQR P. falciparum, the prevalence of CQR P. falciparum is exceptionally low and unchanged in Guinea-Bissau [4].
The median CQ doses prescribed and reportedly taken in Guinea-Bissau were 81 and 77 mgkg 21 , divided into 2-3 doses per day for median 5 days [5]. According to local physicians, CQ has always been prescribed in this way and use of high doses has been recorded since 1994 [4]. We have shown that 50 mgkg 21 of CQ in 2 divided daily doses over 3 days resulted in a 92% PCR corrected efficacy at day 28 [6] whereas treatment with the standard dose of 25 mgkg 21 given during 3 days had an 80% efficacy [6]. In addition, when treating P. falciparum carrying the CQR associated genetic marker (pfcrt 76T), 50 mgkg 21 had a 78% efficacy whilst only 34% were successfully treated with 25 mgkg 21 [2].
The marked reduction of the pfcrt 76T haplotype in Malawi following the withdrawal of CQ [7] suggests that pfcrt 76T identifies a less fit parasite than pfcrt 76K in the absence of CQ. In line with that, CQ resistance has been shown to be an energy dependent process [8]. Studies from Sudan and The Gambia reported increasing pfcrt 76T haplotype prevalence during the high transmission season and decreasing during the low transmission season [9,10]. The probable explanation is that pfcrt 76T carrying parasites survive treatment with standard CQ doses, thereby accumulating during the high transmission season when CQ is frequently used. When less CQ is used during the low transmission season, haplotype prevalences are reversed because of better fitness of the 76K haplotype in the absence of CQ.
If the high dose CQ treatment commonly used in Guinea-Bissau is efficacious as our data suggests, pfcrt 76T carrying parasites will lose much of their survival advantage and therefore not accumulate markedly during the high transmission season. To evaluate this hypothesis we have analysed seasonal variations of the proportions of pfcrt 76K and 76T and pfmdr1 86N and 86Y in Guinea-Bissau.

Methods
The studies were conducted at the Bandim Health Project in Bissau, Guinea-Bissau. The Bandim Health Project is a Demographic Survey Site (DSS) covering approximately 16 km 2 , mainly comprising semi-urban areas. The population is approximately 90 000 including about 12 000 children ,5 years of age. Three primary health care centres, Bandim, Belem and Cuntum serve the population. In addition, the national hospital with 125 beds 5 km away, also serves as a primary contact as well as a referral hospital for patients.
Children develop symptomatic P. falciparum malaria infections all the year round in Guinea Bissau. However, there is a distinct seasonality with higher incidence of malaria between May and December just before, during and after the rainy season that lasts from May/June to October/November. The malaria prevalence has decreased over the years. In 1990, 183/312 (59%) of children (aged 3-6 yrs) had P. falciparum in community surveys during the rainy season compared to, 7/197 (3.6%) of children (aged,5 yrs) in 2004 [11].
Since October 2002 (except for Oct 2006) three clinical trials including collection of blood spotted onto filter-papers at day 0 for genotyping have been conducted back to back at Bandim Health Centre. From November 2006 until December 2008 the last study also recruited patients from Belem and Cuntum health centres. Inclusion criteria were similar in all studies requiring microscopically verified malaria (.20 P. falciparum per 200 white blood cells) with fever (or history of fever in the past 24 hours) in the absence of signs of severe malaria (convulsions, severe anaemia, hyperparasitaemia, clinically poor condition).
Following informed consent a total of 988 (501 boys and 486 girls) children were included between October 2002 and December 2007. The total number of inclusions each month between January and December were 78, 49, 32, 32, 71, 118, 93, 66, 58, 137, 147, 107. We designated January -April as the low transmission season and May to December as the high transmission season. The high transmission season was split into early (May-August) and late (September-December) high transmission season. This split is halfway through the season and also at the point of lowest number of new cases during the high transmission season.
Monthly rainfall data from 2003 to 2007 was provided by the National Meteorological Service and are presented as mean monthly values for the whole period. Haplotypes were analysed as 3 groups (76T, 76K or both for pfcrt and 86Y, 86N or both for pfmdr1). The association between pfmsp families and pfcrt and pfmdr1 haplotypes were assessed using Fishers exact test.

Study population
The median age was 65 months, the inter-quartile range (IQR) was 38-105 months and the median parasitaemia was 15600 (IQR 6800-38400) P. falciparum per micro-litre assuming a white blood cell count of 8000 per micro-litre (Table 1).
Stable proportions of pfcrt 76K and 76T and pfmdr1 86N and 86Y during the year Pfcrt 76K and 76T alleles were successfully identified in 954/988 samples and pfmdr1 86N and 86Y alleles in 958/988. If both alleles were identified at one locus both were included as numerators but only counted as one in the denominator.
There was no continuous monthly trend of changing pfcrt 76T or 76K (figure 1) nor pfmdr1 86Y or 86N (figure 2) allele proportions during the high (May-December) or the low

Fluctuations in parasite densities during and between seasons
During early high transmission season parasite densities were higher compared to low transmission season irrespective of pfcrt K76T haplotype. For P. falciparum with 76K only, parasite densities during early high transmission season were also higher than during late high transmission season while there was no difference between late high and low transmission season. For P. falciparum with 76T only, parasite densities were higher during both early and late high transmission season compared to the low transmission season. (Table 2).
Lower parasite densities were associated with pfcrt 76T and pfmdr1 86Y

Association between pfcrt K76T haplotype proportions, transmission rate and drug pressure (figure 4)
Through out the year the number of treatments and the proportion of pfcrt 76T vary in a remarkably similar fashion. The malaria incidence starts to increase between April and May just before the rainy season whilst the pfcrt 76T haplotype proportion start to increase approximately one month later. The malaria incidence peaks in June whilst the pfcrt 76T proportion peaks one month later.

Discussion
We present continuously collected data from clinical trials with the same basic inclusion criteria conducted between October 2002 and December 2007. The proportions of pfcrt 76K and 76T and pfmdr1 86N and 86Y do not vary significantly between high and low transmission season in Guinea-Bissau and the proportions remained unchanged between 2003 and 2007.
The low and stable proportion of pfcrt 76T indicates that this haplotype (and thus P. falciparum normally resistant to standard dose CQ) does not have a selective advantage great enough to become the dominant haplotype in Guinea-Bissau. We have previously shown that double normal dose CQ has a 78% efficacy against P. falciparum with pfcrt 76T in Guinea-Bissau [2]. We have also shown that the standard total dose prescribed and commonly taken is approximately 3 times the normal dose (75 mgkg 21 ) divided into smaller doses over 5 days [5]. It is therefore likely that P. falciparum with pfcrt 76T are generally treated effectively in Guinea-Bissau. Much of the selective advantage associated with pfcrt 76T during treatment with normal dose CQ [2,14] is therefore lost unless absorption is poor or the prescribed treatment discontinued. This most probably accounts for the lack of cumulative increase of pfcrt 76T in Guinea-Bissau during the high transmission season.
The small, temporary, but significant increase of pfcrt 76T during the beginning of the high transmission season occurs at the same time as the number of treated children increases and approximately one month after the malaria incidence increases. Pfcrt 76T is likely to provide an advantage soon after treatment when CQ concentrations are moderately high [15]. An increased number of treated children will increase the chance of a new infection occurring in a child with a moderately high CQ concentration. This likelihood will also increases as the malaria incidence increases. Thus it is probable that the increased pfcrt 76T proportion is partly caused by selection of pfcrt 76T in new infections due both to an increased incidence of malaria and an increased number of treatments. However, the increased pfcrt 76T proportion is also likely to be partly due to recrudescence. P. falciparum usually recrudesce a few weeks after treatment [2] and the pfcrt 76T proportion should therefore increase a few weeks after an increase in malaria incidence, as it does. Despite this pfcrt 76T does not accumulate over the whole season supporting our hypothesis that only a fraction survive the high dose treatment used.
Monitoring prescription patterns in 2003-2004 showed that, out of 26134 consultations, 17924 (69%) of children below the age of 5 years were diagnosed with malaria while only 13% of these presumptively treated children had microscopically verified malaria [11]. Similar monitoring during 2007 and 2008 found that 13310/34884 (38%, monthly range 30-43%, authors unpublished data) of children (under 15 years) attending health centres were treated for malaria though only 415/13310 (3.2%, monthly range 0.4-8.9%, authors unpublished data) had microscopically verified malaria. In practice, CQ is used for the treatment of fever and the drug selective pressure is high all the year round. There is therefore no period of low CQ pressure during which pfcrt 76K has a distinct selective advantage as described in The Gambia and Sudan [9,10]. This probably explains the surprisingly stable pfcrt 76T and pfcrt 76K proportions during the dry season. Resistance is generally believed to develop from a sensitive parasite that first becomes tolerant and eventually resistant. Tolerance is a long process during which parasites acquire several genetic changes making them gradually more tolerant of CQ, whereas a resistant parasite is unaffected by drug exposure. [15]. Pfcrt 76T identifies a parasite that is very likely to be resistant when normal dose CQ is used, also in Guinea-Bissau [2,14]. However, the common use of more efficacious high doses of CQ in Guinea-Bissau probably turns pfcrt 76T into a marker of tolerance. But, as discussed previously, pfcrt 76T is also associated with a loss of fitness. The probable explanation for the low and stable pfcrt 76T proportions is therefore the loss of fitness associated with pfcrt 76T that negates the advantage 76T provides when CQ concentrations are moderate. No parasite with the ability to survive the high doses of CQ routinely used has become established in Guinea-Bissau despite ''tolerant'' (pfcrt 76T) parasites existing since at least 1992 [4]. This suggests that development of further resistance is a difficult step in line with predictions [16] and the long time it took for CQ resistance to develop in the first place [17].
Mutations in pfmdr1 have been associated with substantial fitness cost [18]. We therefore assessed parasite densities as a proxy marker of fitness. Pfcrt 76T and pfmdr1 86Y were associated with lower parasite densities suggesting a cost of fitness associated with at least one of these haplotypes as they are linked. This supports the fitness argument for why pfcrt 76T has reached fixation at a low prevalence in Guinea-Bissau. Including children above the age of 10 gave similar but not significant results. Differences in immunity might account for the age effect. However, the data should be interpreted with caution as the age effect is difficult to explain and because it is possible that pfcrt 76T identifies a more virulent parasite causing children to go to the health centres earlier.
As in the Gambia [10], there was no seasonal variation of pfmdr1 86N or 86Y proportions, there was linkage disequilibrium between pfmdr1 86Y and pfcrt 76T and no significant seasonal variation of the mean number of genotypes. The mean number of genotypes were lower in Guinea-Bissau suggesting less chance of sexual recombination in mosquitoes, that should slow the pace of resistance development [15].
As in The Gambia, we noted lower parasite densities during the second half of the high transmission season compared to the first half. This might be due to an enhanced malaria specific immunity. The differences in seasonal parasite density pattern between pfcrt 76T and pfcrt 76K parasites is difficult to interpret and is possibly due to small number of P. falciparum with pfcrt 76T (only 37) during the low transmission season. In summary, the pfcrt 76T proportion does not gradually increase throughout the high transmission season and is low and stable between 2003 and 2007. We suggest that this is due to the use of a more efficacious dosage of CQ in Guinea-Bissau combined with a loss of fitness associated with pfcrt 76T. These factors largely remove the selective advantage that P. falciparum with pfcrt 76T have when normal dose CQ is used. The results therefore indicate that CQ can be an effective drug if dosed differently. As CQ is cheap and accessible further research into dosing strategies of CQ should be done.