Clinical pharmacokinetics of quinine and its relationship with treatment outcomes in children, pregnant women, and elderly patients, with uncomplicated and complicated malaria: a systematic review

Standard dosage regimens of quinine formulated for adult patients with uncomplicated and complicated malaria have been applied for clinical uses in children, pregnant women, and elderly. Since these populations have anatomical and physiological differences from adults, dosage regimens formulated for adults may not be appropriate. The study aimed to (i) review existing information on the pharmacokinetics of quinine in children, pregnant women, and elderly populations, (ii) identify factors that influence quinine pharmacokinetics, and (iii) analyse the relationship between the pharmacokinetics and treatment outcomes (therapeutic and safety) of various dosage regimens of quinine. Web of Sciences, Cochrane Library, Scopus, and PubMed were the databases applied in this systematic search for relevant research articles published up to October 2020 using the predefined search terms. The retrieved articles were initially screened by titles and abstracts to exclude any irrelevant articles and were further evaluated based on full-texts, applying the predefined eligibility criteria. Excel spreadsheet (Microsoft, WA, USA) was used for data collection and management. Qualitative data are presented as numbers and percentages, and where appropriate, mean + SD or median (range) or range values. Twenty-eight articles fulfilled the eligibility criteria, 19 in children, 7 in pregnant women, and 2 in elderly (14 and 7 articles in complicated and uncomplicated malaria, respectively). Severity of infection, routes of administration, and nutritional status were shown to be the key factors impacting quinine pharmacokinetics in these vulnerable groups. The recommended dosages for both uncomplicated and complicated malaria are, in general, adequate for elderly and children with uncomplicated malaria. Dose adjustment may be required in pregnant women with both uncomplicated and complicated malaria, and in children with complicated malaria. Pharmacokinetics studies relevant to clinical efficacy in these vulnerable groups of patients with large sample size and reassessment of MIC (minimum inhibitory concentration) should be considered.


Background
Parenteral artesunate, followed by artemisinin-based combination therapy (ACT), is the recommended treatment for severe malaria, when available. Quinine is usually used if artesunate is not available, and is always used in combination with an additional agent, such as doxycycline [1]. Quinine is also used, in combination with clindamycin, for uncomplicated Plasmodium falciparum infections in pregnant women during the first-trimester, as well as for chloroquine-resistant Plasmodium vivax infections [1]. A systematic review in 2012 revealed that artesunate was superior to quinine in reducing mortality rate and adverse drug reactions in severe malaria [2]. A recent systematic review (2020) has suggested a high risk of treatment failure in uncomplicated malaria patients (mostly second-and third-trimester pregnancies) after quinine monotherapy, while the efficacy of quinine-clindamycin combination was comparable with ACT [3].
The current dosage regimens of quinine for children, elderly, and pregnant women, based on adults' regimens, may be suboptimal due to altered physiology and pharmacokinetics [4]. None of the current systematic reviews address the impact of the differences in pharmacokinetics on quinine efficacy and safety in these vulnerable populations [5,6]. Pharmacokinetic investigation of quinine may be required for further dosage optimization in these populations. The objective of this study is to provide a systematic review of the existing information on the dose regimens and pharmacokinetics of quinine in children, pregnant women, and elderly populations, and to determine whether modification of dosage regimens in these populations is required.

Data sources
The Web of Science, Cochrane Library, Scopus, and Pub-Med were used by two independent researchers as search databases up to October 2020. The search terms included 'Quinine' AND 'Pharmacokinetics' AND 'Pregnancy' OR 'children' OR 'Pediatrics' OR 'Neonates' OR 'Newborns' OR 'Infants' OR 'Geriatrics' OR 'elderly' (> 65 years).

Inclusion and exclusion criteria
The inclusion criteria were: the article published in English, and the article involved clinical pharmacokinetic research on quinine in either children, elderly, or pregnant women.

Data extraction and quality assessment
TS and PK independently retrieved and reviewed articles, and any argument or inconsistency was resolved by the third person (KN). The information extracted were: area under the plasma drug concentration-time curve (AUC), maximum plasma concentration (C max ), time to reach maximum plasma concentration (t max ), trough plasma concentration (C trough ), elimination or terminal half-life (t 1/2 or t 1/2β ), apparent volume of distribution (V d , V d /F, V ss , or V ss /F), total clearance (CL or CL/F), efficacy, and toxicity parameters. The AUC, C max , and C trough were normalized with body weight. As there are different values of half-life, volume of distribution, and clearance parameters reported in various studies, the full descriptions of parameter terms are used in the articles. Qualitative data are presented as number and percentage (%). Quantitative data are summarized as mean, mean ± SD, median, median (range), or range values. The volume of distribution and clearance for the intramuscular (im), intrarectal (ir), and oral (po) doses are reported by adjusting to an absolute bioavailability (F) of 0.95, 0.60, and 0.76, respectively.
Quality of the article selection process was assessed using the checklist for the assessment of the methodological quality for both randomized and non-randomized studies of health care intervention [7].
Three pharmacokinetic analysis approaches were applied i.e., non-compartmental analysis (NCA, n = 13) [10, 16-19, 22, 26-28, 30-32, 34], compartmentalanalysis (CA, n = 11) [8, 9, 11-13, 15, 20, 21, 23, 33, 35], and population-based pharmacokinetic analysis (pop-PK, n = 4) [14,24,25,29]. NCA is the standard method which provides pharmacokinetic parameters directly from the observed data giving a gross approximation as the result, and is not the most accurate method for parameter estimation. This analysis approach however, is not suitable for the characterization of pharmacokinetic variability in the populations. CA offers the advantage of allowing for the use of data modeling and simulation for dosage optimization. The pop-PK approach is most applicable to analysis of drug pharmacokinetics in various populations considering both intra-and inter-individual variability. Additional file 1: Table S1-3 summarizes quinine pharmacokinetics (focusing on C max and systemic exposure) reported in children, pregnant women, and elderly with uncomplicated and complicated malaria.

Pregnant women
AUC data was reported in pregnancy with uncomplicated malaria in only 4 articles [27][28][29]31]. A comparison of quinine pharmacokinetics between pregnant and non-pregnant women was reported in 2 articles [27,28] wherein the AUC 0-inf for both groups were comparable (8.24 ± 2.43 vs. 9.90 ± 2.50 µmol/l/h) [27]. Physiological changes during pregnancy and estimated gestational age (EGA) did not appear to significantly affect quinine exposure [29]. It is noted, however, that the number of participants in each group (7 non-pregnant and 16 pregnant women) is small and may not provide adequate statistical power to allow for an accurate conclusion on the effect of pregnancy on quinine pharmacokinetics. C max was reported in all articles, 5 and 2 articles in uncomplicated and complicated malaria, respectively (0.07-0.22 vs. 0.19 µg/ml/kg) [27][28][29]31]. Malaria infection and routes of administration, but not pregnancy status, had significant influence on quinine C max [27]. The C max following iv and po routes were 0.20 ± 0.028 [27,28] and 0.09 ± 0.026 [29,31] µg/ml/kg, respectively.

Elderly
A study in healthy elderly following a single oral dose of 600 mg quinine salt (497 mg base) [34] showed a significantly higher AUC in elderly compared with adults (1.58 ± 0.39 vs. 1.20 ± 0.32 µg/ml/kg), while the C max values were comparable [34].
The volume of distribution of quinine was reported in 8 articles, 6 in complicated [17,19,20,[23][24][25] and 2 in uncomplicated malaria [10,14] (0.53-1.5 vs. 1.4-2.0 l/ kg). Body weight was a significant covariate for the volume of distribution [25]. The effect of malaria disease severity on volume of distribution was contradictory. The volume of distribution in cerebral malaria reported by Pussard et al. [23] was threefold lower than that reported by Frank et al. [17] (0.53 ± 0.1 vs. 1.44 ± 0.01 l/kg, respectively), but was similar to adult patients with cerebral malaria (0.74 ± 0.30 l/kg) [38]. The decrease in the volume of distribution with increasing malaria severity could be due to an increase in AAG and dehydration in patients, and/or a decrease in tissue blood flow in complicated malaria. Nutritional status had less contribution (Vd: 0.58 ± 0.22 vs. 0.53 ± 0.1 l/kg for malnourished vs. nourished children with malaria) [23].
Three out of 19 studies reported models with covariates [14,24,25]. None provided the stratification of age along with pharmacokinetic analysis. Age (range: 6 months to 6.7 years) was not a significant covariate that influenced clearance and/or volume of distribution. Since the activities and expressions of UDP-glucuronosyltransferase 1A1 (UGT1A1) and cytochrome P450 3A4 (CYP3A4), both major quinine metabolizing enzymes, are likely to be steady in individuals aged over 6 months [39], and 24 months [40], respectively, age is unlikely to be a significant covariate. In contrast, body-weight was a significant covariate that influenced both clearance and volume of distribution. It was clear that quinine dosage should be administered based on body-weight rather than age.

Pregnant women
Quinine clearance during pregnancy was reported in 2 articles in uncomplicated [29][30][31] and 1 article in complicated [33] malaria (0.11 ± 0.04 vs. 0.07 l/h/kg). Based on pop-pK analysis, the presence of an effect from EGA status on quinine clearance was unlikely, despite the positive correlation between the EGA associated-pregnancy period and quinine clearance [29]. Since CYP3A4 activity consistently increases throughout gestation, it may not have a sizeable enough impact to be a considered a covariate. Furthermore, the initial body temperature was a significant covariate that influenced clearance due to an increase in hepatic enzyme activity. In patients presenting high body temperature upon admission, plasma concentration of quinine should be monitored. Quinine volume of distribution was reported in 3 articles, 2 in uncomplicated [29,31] and 1 in complicated malaria [33] (2.4 ± 1.3 vs. 0.96 ± 0.27 l/kg). The reported volume of distribution was inconsistent. The volume of distribution values in uncomplicated malaria, based on pop-PK analysis, during the second and third-trimesters were 3.05 (2.66-4.26) and 3.05 (2.75-3.86) l/kg [29], respectively. The volume of distribution estimated based on NCA in pregnant women with uncomplicated malaria was 1.1 (0.62-1.97) l/kg [31], which was comparable to adult patients with uncomplicated malaria (1.26 ± 0.26 l/kg) [42]. The demographic data (e.g., age, body-weight, gestation age, and trimester of pregnancy) of the patients from the two studies were not different [29,31]. However, it is clear that plasma volume in pregnancy during the first, second, and third trimesters increases up to 106%, 129%, and 149% of that reported in non-pregnant women [43]. In addition, an increase in fat composition, extracellular fluid, and total body water during pregnancy may result in an increased volume of distribution as quinine is moderately lipophilic. The reported volume of distribution based on NCA analysis is likely to be inaccurate.
There was no effect of diabetic status on quinine volume of distribution (1.29 ± 0.42 vs. 1.29 ± 0.45 l/kg for non-diabetic vs. diabetic subjects) [35]. The volume of distribution in the elderly was lower than adults [44], due mainly to a decrease in total body water and cardiac output.
The prolongation of quinine half-life in the elderly compared with young adults (19.2 ± 1.1 vs. 10.5 ± 1.6 h) was consistent with the decrease of quinine clearance [34].
Quinine t max was unaffected by age (2.3 ± 1.2 vs. 2.5 ± 0.7 h in elderly vs. young adults) [34], which was at least in part explained by the unchanged gastric emptying rate in the elderly. It is noted that the reports of the gastric emptying rate change in elderly, reported as either delayed [45] or unchanged [46], are discordant. Demographic data (i.e., body-weight, height, body-mass index (BMI), sex, hypertensive status, and glycosylated haemoglobin level) were not evaluated as model covariates.

Ethnicity
Ethnicity can influence quinine clearance due to associated polymorphisms in drug-metabolizing enzymes. Twenty-three studies were carried out in Africans; the other 5 studies were carried out in Cambodians, New Zealanders, and Australian. There were no clinical studies that reported across different ethnic groups on the same malaria severity and vulnerable populations. The conclusion of the influence of ethnicity on quinine exposure is, therefore, limited. However, the distribution of genetic polymorphisms, including those in genes coding for drug-metabolizing enzymes, varies with ethnicity, suggesting the possibility that ethnicity could influence quinine clearance. The allele frequencies of CYP3A4*3, CYP3A4*13, CYP3A4*18, and CYP3A4*19 were 0.021 in European-Americans, 0.011 in Han-Chinese, 0.01 in Japanese, and 0 in Africans [47]. The proportions of patients with impaired quinine clearance in individuals with either CYP3A4*3, CYP3A4*13, CYP3A4*18, or CYP3A4*19 and those with the wild-type genotype CYP3A4*1A were 22.6%, 5.8%, 17.6%, and 43%, respectively [48]. In addition to the CYP3A4/5 polymorphisms in African subjects, four additional polymorphisms including CYP3A4*1B, CYP3A5*3, CYP3A5*6, and CYP3A5*7 were reported. The allele frequencies for CYP3A4*1B, CYP3A5*3, CYP3A5*6, and CYP3A5*7 were 0.66-0.86, 0.04-0.81, 0.05-0.25, and 0-0.21, respectively [67]. All polymorphisms except CYP3A41*B are classified as poor metabolizer (PM) due to a reduction or non-detection of enzyme expression [67]. The decrease in enzyme activity and clearance may result in toxicity. However, results from a recent physiologically-based pharmacokinetic (PBPK) modelling study suggested that the effects of these genetic polymorphisms on quinine exposure and clearance were not clinically significant and dosage adjustment may not be required [49].

Children
Sixteen out of 19 articles reported clinical efficacy of quinine, 8 each in uncomplicated and complicated malaria.
Quinine doses used in complicated malaria treatment varied from 4.74 to 16.35 mg base/kg (with or without a loading dose). Four out of 8 articles reported both FCT [19,[21][22][23] and PCT [19,21,22,24]. The FCT and PCT values in complicated malaria were 25.1-48.6 and 27.4-49.5 h, respectively [19,[21][22][23]. The FCT following im administration [19,22] was shorter than the ir [21] route (27.4 ± 3.6 vs. 48.6 ± 2.7 h). This was in agreement with the relatively short t max following im administration compared with ir routes. Time to regain consciousness was 36-39 h [19]. A recent systematic review and metaanalysis however, concluded no significant difference in PCT, FCT, mortality rate, duration of hospitalization, and time-to-drinking between IR and IM administration [56], but not for the effect of dosage.
Fifteen out of the 19 articles reported C trough at different time points, 8 articles in uncomplicated [8-13, 15, 16] and 6 articles in complicated [17-19, 21, 23, 25] malaria, and 1 article in kwashiorkor [26]. Therapeutic ranges of total quinine concentration for children with complicated and uncomplicated malaria in Africa before and during 1994 ranged from 0.2 to 2.0 µg/ ml [57], and 0.21-0.35 µg/ml unbound quinine [11], respectively. Despite the continuous decline in sensitivity of P. falciparum to quinine over time, the current analysis suggests that total C trough in most studies in both complicated and uncomplicated malaria were maintained above the MIC for 7 days with a clinical efficacy of 100%. Total quinine C trough during 1983-2010 in uncomplicated malaria ranged from 1.42 to 10.43 µg/ml (0.1-0.73 µg/ml free quinine) [8, 9, 11-13, 15, 16]. Results of pharmacokinetic/pharmacodynamic modeling reported in 2003 suggested a total C trough of at least 3.4 µg/ml (0.34 µg/ml free quinine, f u = 0.109 [58]) for curative treatment of uncomplicated falciparum malaria in Thai adults [59]. The consensus meeting in 2007 concluded that the required C trough for uncomplicated malaria was 10-12 µg/ ml 1 (1.09-1.31 µg/ml free quinine) [60], which is equivalent to 9.5-11.50 µg/ml total C trough in children (f u = 0.114, 1.08-1.3 µg/ml free concentrations) [14]. The therapeutic C trough of quinine in children with uncomplicated malaria is therefore, 9.5-11.50 µg/ml. The dosage regimens that provided the required total C trough in children with uncomplicated malaria were a 7-day course of multiple doses of 8 [16] or 10-12 [15] mg base/kg q8h for iv administration and 12 h for im administration. In addition, the suggested ir regimen was multiple doses of 11.84 mg base/kg q8h. Such routes and dose frequency of quinine administration also contribute to quinine exposure and therapeutic outcome. Dosage adjustment in children with uncomplicated malaria may not be required [ Table 1], which is applicable for quinine use in malaria endemic areas where quinine sensitivity is still sufficient. In addition to improved patient compliance, im administration appears most preferable due to its lower frequency of drug administration [ Table 1].
For complicated malaria, the therapeutic range of quinine for adults reported in 1983 was 5-10 µg/ml (0.35-0.7 µg/ml free quinine) [61] or 10-15 µg/ml (0.7-1.05 µg/ml unbound quinine) [62]. Quinine C trough during 1982-2010 in complicated malaria ranged from 2.7 to 12 µg/ml (0.15-0.66 µg/ml free quinine) [17-19, 21, 23, 25]. The predicted C trough based on data analysis from uncomplicated malaria during 1948-1995 was 8-15 µg/ml (0.72-1.05 µg/ml unbound quinine) for curative treatment of complicated malaria [25]. The suggested therapeutic C trough in children with complicated malaria is therefore 14.4-19.09 µg/ml (f u = 0.055 [20], 0.72-1.05 µg/ml free quinine). None of the dosage regimens reported in the current analysis provided sufficient C trough for treatment of complicated malaria in children [ Table 1]. It is noted that the recommended standard dose regimen for complicated malaria was not used in the included articles. Pharmacokinetic/pharmacodynamic studies of quinine in children with complicated malaria following standard dose regimens are required to evaluate their effectiveness.

Pregnant women
Four out of 7 articles reported clinical efficacy of quinine, 3 in uncomplicated [28,30,31] and 1 in complicated [32] malaria during pregnancy. PCT was reported in 1 study (24-57 h) [30]. None of the included studies reported FCT. The cure rates in uncomplicated and complicated malaria were 99.2% [28,31], and 91.67% [32], respectively. Two quinine regimens were used for uncomplicated malaria, i.e., multiple oral doses of 8.3 mg base/kg q8h for 7 days [29][30][31], and a single oral dose followed by artemether (ACT) [27]. In complicated malaria, a standard dose regimen (a loading dose of 16.7 mg base IV over 4 h, followed by 8.3 mg base/kg IV over 4 h q8h) was applied [32,33]. Three articles reported C trough , 2 in uncomplicated [29,31] and 1 in complicated [33] malaria (Additional file 1: Table S4). The required total C trough values for uncomplicated [60] and complicated malaria [61] in pregnant women based on the current information in adults were 13-16 µg/ml, and 12.9-18.8 µg/ml, respectively. The therapeutic ranges of quinine for uncomplicated and complicated malaria during pregnancy, based on a study in Thailand [42] and France [60], were 14.257-17.5 [42] and 15.825-25.40 µg/ml [60], respectively. In addition to the curative treatment, the required total C trough values for uncomplicated and complicated malaria during pregnancy (based on information in children [11] and adults [62]) were 2.6-4.37 and 7-14 µg/ml, respectively. However, total reported C trough values in uncomplicated and complicated malaria were 2-3.9 [29, 31] and 7.1 [33] µg/ml, respectively (Additional file 1: Table S4). Therefore, none of the reported regimens provided adequate total quinine concentrations. Quinine dose adjustment may be required for pregnant women with uncomplicated and complicated malaria (Table 1). It is noted that the contribution of pharmacokinetics on quinine exposure remains inconclusive due to the limitation of sample size (See in Dose-dependent pharmacokinetics (Pregnant women)).

Elderly
With a single oral dose of 600 mg, plasma quinine concentration in elderly subjects was 10% lower than healthy adults [34] (Additional file 1: Table S3). There has been no pharmacokinetic study of this dose regimen in elderly with uncomplicated malaria. The C trough in elderly, estimated from non-elderly adult patients with uncomplicated malaria (4.5 µg/ml) [59], is 4 µg/ml. Quinine plasma protein binding was unaffected by age [63]. An unbound C trough concentration of 0.4 µg/ml in elderly is, thus, considered sufficient for uncomplicated malaria therapy. Plasma quinine concentration at steady-state in adult patients with severe malaria following standard regimen was 14 (10-20) µg/ml [64], which corresponds to 12.46 µg/ml in elderly. This C trough is within the therapeutic range (10-20 µg/ml), and quinine dose adjustment may not be necessary for elderly (Table 1). Clinical study is required to support this argument.
The effects of anatomical and physiological differences in children and elderly, but not pregnant women, do not appear to influence the pharmacokinetics and clinical efficacy of quinine. Standard dose regimens of quinine are likely to be sufficient for effective malaria treatment in children (uncomplicated malaria) and elderly (uncomplicated and complicated malaria). Due to the expansion of plasma volume which may result in inadequate plasma quinine exposure, dose adjustment is likely to be required in pregnant women.

Quinine adverse reactions and toxicity Children
Eight out of 19 articles reported adverse reactions after quinine dosing, 3 in uncomplicated [11][12][13] and 5 in complicated [18-20, 24, 25] malaria. In uncomplicated malaria, no adverse reactions were reported following quinine doses ranging from 8.3 mg base (iv/im) to 11.85 mg base (ir), which produced C max ranging from 4.43 to 9.86 µg/ml (0.31-0.69 µg/ml unbound C max ) [11][12][13]. A C max of over 20 µg/ml (1.4 µg/ml free C max ) was reported in adult patients with cerebral malaria without toxicity [42]. The estimated toxic concentration in children based on healthy adult information is 28 µg/ml. The comparatively high toxic concentration reported in children was likely due to the increase in the fraction of free plasma drug in children.
Hypoglycaemia is a commonly reported adverse reaction to quinine in complicated malaria treatment (1-15%) [18,19,24,25]. One article reported no association with hypoglycaemia in children with uncomplicated malaria [19]. The other 3 reported an association, but without supportive evidence on insulin levels [18,24,25]. Only 1 article reported quinine C max of 25.9 µg/ml at the time of hypoglycaemia [18]. Two articles suggested that the standard glucose dose (3 mg/kg/min or 5% dextrose iv infusion) might be insufficient to correct hypoglycaemia in children [24,25], instead suggesting a dose of 6 mg/kg/min [24]. The risk of quinine-induce hypoglycaemia is increased 3.2-(1.0-9.8) fold and should be a concern in children with pre-existing hypoglycaemia [24]. Since children with complicated malaria are likely to develop hypoglycaemia upon starting treatment, those that do present should have their blood glucose levels monitored throughout the treatment course.
QRS prolongation is one of the most serious concerns of quinine toxicity. One study reported a 10% incidence of QRS prolongation in children with complicated malaria receiving quinine treatment [20]; in this study, two patients experienced QRS prolongation leading to death, the quinine C max of one of them was 16.9 µg/ml [20]. However, no correlation between QRS prolongation and free quinine concentration was found [20]. Although there was no correlation between QRS and free quinine concentration, plasma quinine concentration should be monitored.
Local irritation or pain at the injection site was commonly reported (12%) following im quinine administration but these symptoms resolved within 4 weeks [24]. The incidence of transient neurologic sequelae was 5% (1/18) with the standard quinine regimen in severe malaria [19]; however, there was no information on plasma quinine concentrations. Quinine C max in fatal (15.0 ± 7.8 µg/ml, n = 2) and nonfatal (15.0 ± 3.9 µg/ml, n = 19) cases were similar, although the C max in one fatal case was markedly high (25.9 µg/ ml) [18]. About 5% (4/75) of children with complicated malaria receiving quinine had plasma concentrations over 25 µg/ml without toxicity [25]. The safety level of plasma quinine concentration (MTC) in complicated malaria could be as high as 25 µg/ml due to the higher level of plasma protein binding of quinine [20].

Pregnant women
Three out of 7 articles reported adverse reactions to quinine, 1 in uncomplicated [30] and 2 in complicated [32,33] malaria. Mild-to-moderate tinnitus, headache, and epigastric pain occurred in uncomplicated malaria patients but they had recovered within 2-7 days [30].
Hypoglycaemia was the most commonly reported (50-100%) adverse reaction to quinine in complicated malaria [32,33]. The relationship between plasma quinine concentrations and arterial blood pressure has been reported by Phillips et al. [33]. Quinine appears not to augment uterine contractions nor induce fetal distress.

Elderly
Adverse reactions to quinine were reported in two subjects after a single oral dose of 600 mg, one in an elderly patient with dizziness (4 µg/ml) and another in a young subject with tinnitus (4.3 µg/ml) [34]. Quinine-induced dizziness is of critical concern since it can lead to the fatal injury in the elderly [65]. Therefore, quinine dose administration in elderly, particularly by parenteral route, should be done carefully. Tinnitus was reversible without additional treatment. Since hypoglycaemia was the most commonly reported adverse reaction to quinine in other vulnerable groups, physicians should carefully prescribe quinine and advise patients of its side-effects.

Limitations and suggestions
The MIC or MCOC of quinine for P. falciparum malaria has not been reported since 2007 [55]. As the susceptibility of P. falciparum to quinine changes over time, inaccurate estimation of MIC and MCOC values may lead to inappropriate dose optimization in populations. The sample sizes used in most studies are small, which may not provide adequate power to detect small differences in the parameters under investigation, and thus, lead to incorrect data interpretation and conclusions. The study designs applied in most studies are not double-blind, or randomized controlled trials (RCT), which could result in bias influencing data interpretation and conclusions. Only a few studies applied statistical analysis to draw conclusions on the significant difference between the observed parameters. The heterogeneity of blood sampling frequency for pharmacokinetic investigations makes comparison among various groups or populations difficult. A conclusive understanding of the effect of a single factor (malaria disease, pregnancy status, age) on quinine pharmacokinetics is obscured by the intertwined nature of these factors in the studied populations. Further, pregnancy status (trimester periods) and age groups are not well defined in some studies. As quinine pharmacokinetics varies as a result of the physiological differences between children, elderly, and pregnant women, data obtained from such ill-defined subpopulations could lead to misguided conclusions. Conducting large clinical trials with a sufficient number of participants may have several limitations and ethical concerns in these vulnerable groups of malaria patients. In recent years, PBPK has emerged as a promising pharmacokinetic analysis tool which could be applied for accurate dose optimization for various drug classes with no requirement of large study sample sizes. PBPK is a mathematical model describing drug disposition in the human body based on prior knowledge from both in vitro and in vivo studies [66]. This model consists of various compartments corresponding to real anatomy and physiology of humans and can accurately predict optimal drug dosage regimens in various populations using a large number of virtual populations and applying prior knowledge from past clinical studies [66] (Additional file 1)

Conclusion
The information of quinine pharmacokinetics in children, pregnant women, and the elderly with uncomplicated and complicated malaria are limited. Malaria infection and severity, routes of quinine administration, and nutritional status are the key factors that influence quinine systemic exposure and pharmacokinetics. The recommended dosages for both uncomplicated and complicated malaria are, in general, adequate for the elderly and children with uncomplicated malaria. In pregnant women with either uncomplicated or complicated malaria, and children with complicated malaria however, dose adjustment may be required. The discrepancies of the reported pharmacokinetics, particularly the volume of distribution and clearance, limit accurate dose optimization. Large clinical trials applying pop-PK or PBPK analysis would provide insight on the clinically relevant relationship between pharmacokinetics and clinical outcome parameters following various quinine dose regimens in these vulnerable populations.

Key points
Standard dose regimens of quinine for the treatment of uncomplicated and complicated malaria optimized for the general population have been used in special population (i.e., children, pregnant women, and the elderly) without adjustment, even though these special populations have differences in their physiology and anatomy. Current standard dose regimens for malaria treatment for both uncomplicated and complicated malaria are sufficient for the elderly, but not for pregnant women. In addition to children, only the standard dosage regimen of quinine for uncomplicated malaria treatment is sufficient for malaria treatment. However, no standard dosage regimen of quinine for complicated malaria has been applied in children.
Additional file 1: Table S1. Summary of quinine pharmacokinetic studies in children with malaria (uncomplicated malaria and complicated malaria). Pharmacokinetic parameters (Cmax and systemic exposure) are presented as mean + SD or mean or median (range) or median values. Table S2. Summary of quinine pharmacokinetic studies in pregnant women with malaria (uncomplicated malaria and complicated malaria). Pharmacokinetic parameters (Cmax and systemic exposure) are presented as mean + SD or mean or median (range) or median values. Table S3. Summary of quinine pharmacokinetic studies in the elderly. Pharmacokinetic parameters (Cmax and systemic exposure) are presented as mean+SD or mean or median (range) or median values. Table S4. In vitro quinine susceptibility (IC50) for Plasmodium falciparum isolates collected from Cambodia (2001)(2002)(2003)(2004)(2005)(2006)(2007) and Thailand border (1998)(1999)(2000)(2001)(2002)(2003).