Efficacy and safety of methylene blue in the treatment of malaria: a systematic review

All available and accessible data published until 28 February 2017 on the efficacy and/or safety of MB in the treatment of malaria were considered. Studies reporting data on MB given to humans of any age as monotherapy or in combination with other antimalarials and irrespective of drug formulations and sample size were included. In vitro and animal studies with MB, reports on the use of MB in humans for indications that were not malaria and review papers were excluded.

Standard electronic databases were searched for scientific papers on the subject, but also hand searches were done in scientific books of major libraries. The following databases were searched for studies published in any language: Medline, Embase, BIOSIS, Cochrane Central Register of Controlled Trials, Cochrane Library, Web of Science, and the China National Knowledge Infrastructure and the Chinese Biomedical databases. The search used combinations of the terms “malaria” and “methylene blue” as both medical subject headings and key or free text words and included a broad range of derivations to ensure as wide a search strategy as possible. A list of the detailed search strategy used is available online as Additional file 1. We also retrieved and manually searched articles with relevant titles but an unclear or no abstract. Bibliographies of reports were searched and additional relevant references identified and, where appropriate, included in the review.

Predetermined study characteristics were defined for extraction and documentation by two of the investigators (GL and OM), first independently and then in a consensus procedure. Reports selected were stratified according to study type. The primary outcome was the reported cure rate. Secondary outcomes were fever and parasite clearance rates, the effect on malaria gametocytes and adverse events (AEs).

Two of the investigators (GL and OM) evaluated the quality of included studies in a consensus procedure. The risk of bias of randomised controlled trials (RCTs) was evaluated by the tool described in the Cochrane Handbook for Systematic Reviews of Interventions [51]. The risk of bias in other controlled studies was evaluated by MINORS, for which 12 methodological items are reported [52]. Each domain was scored 0 (not reported), 1 (reported but inadequate) or 2 (reported and adequate). The maximum ideal score was 24.

Quality criteria for case series and case reports were developed from MINORS and the 2016 Critical Appraisal Checklist of the Joanna Briggs Institute for case reports and case series [52‐54]. Seven methodological domains were evaluated (score 0 = not reported, score 1 = reported but inadequate or score 2 = reported and adequate). The maximum ideal score was 14. Where necessary, study authors have been contacted for clarification. Due to a lack of homogeneity among the studies, a meta-analysis was not possible.

Figure 1 summarises the selection process. After removing duplicates, the literature search identified 474 records of which 151 were selected for full text assessment (Additional file 2). Of these, 129 were excluded leaving 22 publications reporting on 21 studies for data extraction and further analysis [29, 55‐75].

The most common reasons for exclusion were as follows: the manuscripts were reviews or comments or guidelines (n = 47), studies were not retrievable despite exhaustive efforts due to very old date (n = 38), manuscripts provided insufficient information (n = 32), preclinical studies (n = 5), MB was not used for treatment of malaria (n = 4), and patients were not treated with MB (n = 3).

The 21 studies included are summarised in Table 1. They originated from Africa (n = 7), Asia (n = 3), the Americas (n = 3) and Europe (n = 8) and reported data on 1504 malaria patients (2/3 were children). The majority of reports were historical studies from around the year 1900 (n = 15), one was from the year 1949, and the remaining (n = 5) were published after the year 2000. The historical studies reported on malaria types diagnosed microscopically and/or clinically, which were not always well specified (of 368 cases, 30 were diagnosed as quartan malaria, 87 as tertian malaria, 183 as falciparum malaria, one as a double infection with tertian malaria and falciparum malaria, and the remaining were not specified). The more recently conducted controlled trials reported microscopically well determined outcomes on falciparum malaria (n = 5) and on vivax malaria (n = 1).

Fifteen studies were historical case reports (n = 5) or case series (n = 10), while six studies were non-randomised controlled trials (n = 3) or RCTs (n = 3) from more recent years.

Summaries of treatment regimens are shown in Table 1. In the historical studies, MB was usually given in divided doses (e.g. five times per day) of 300–1000 mg per day in adults and 20–300 mg per day in children for 3–90 days as daily or interrupted (e.g. a 2-day pause after several days of treatment) oral monotherapy, with largely varying follow-up periods [29, 62, 64‐67, 69]. In the controlled study on vivax malaria, MB was given to adults at a dose of 500 mg per day for 14 days in combination with isopentaquine (IQ) or in combination with IQ and quinine (Q) [61]. In the very recent RCTs on falciparum malaria, MB was given to children in various combination regimens [MB- CQ, MB-amodiaquine (AQ), MB-artesunate (AS) or MB-AQ-AS] in doses ranging from 4 mg/kg per day to 24 mg/kg per day, initially as 2–4 divided doses per day, and always for 3 days [56, 58, 60]. The latest dose regimen applied was 15 mg/kg per day in a single dose over 3 days [55]. In the only recently conducted MB monotherapy study, MB in adults was administrated at a fixed dose of 780 mg per day for 3, 5 or 7 days [59]. In the controlled studies, comparator regimens were CQ, AS-AQ, IQ and IQ-Q [56, 58, 60, 61].

Tables 2, 3 and 4 show the quality assessment of the studies included. The three RCTs had a low risk for bias and were consequently considered to be of high quality. Of the three non-RCTs, the two recently conducted studies scored rather high and were, thus, considered of high quality, while the study from 1949 had a relatively low quality score. The quality of the case series and case reports finally included in our review was overall high. The mean quality score was 10.53, and the majority of case series and case reports provided a clear description of malaria treatment and efficacy outcomes. The information most frequently missing from case series and reports was study participant characteristics (33.3%, 5/15) and safety outcomes (26.7%, 4/15).

In MB monotherapy studies conducted around the year 1900, 337/373 (90%) of malaria cases were reported to be cured. Reported potential reasons for treatment failure were: unsuccessful quinine pretreatment (n = 8), vomiting of treatment (n = 3), low dose of MB (n = 6) and unknown (n = 19). In a more recent trial, MB monotherapy in 60 West African adults with falciparum malaria (a fixed dose of 780 mg for 3 days) showed a 100% cure rate when MB was given for 7 days compared to a 85% cure rate when MB was given for a shorter time period; this difference was close to being statistically significant [59].

Like MB in the combination treatment, a controlled study on vivax malaria among US prisoners from 1949 reported a cure rate of 6/9 (67%) for MB plus IQ and of 3/3 (100%) for MB plus IQ plus quinine [61]. In the recent RCTs among West African children with falciparum malaria, MB-based regimens (n = 391) showed superior efficacy compared to control regimens (n = 207). In the first trial, the cure rate of MB (4 mg/kg per day) plus CQ (56%) was slightly but non-significantly higher compared to that of CQ alone (46%) [58]. A subsequent dose-finding study demonstrated an improved efficacy (77%) of MB-CQ when higher doses of MB (12–24 mg/kg per day) were used [60]. A third study compared the regimens MB-AQ, MB-AS (MB at 20 mg/kg per day) and AS-AQ. The cure rates were 95%, 62% and 82%, respectively, and these differences were statistically significant [56]. Finally, a trial that compared AS-AQ-MB (MB at 15 mg/kg per day) with AS-AQ produced similar cure rates in both regimens (80% vs 85%) [55].

Parasite clearance with MB appeared to be rather slow. In African adults treated with MB monotherapy, 9% were still parasitaemic on day 3 [59]. In children treated with MB-CQ or CQ, the median parasite clearance time did not differ (91.3 and 86.4 h) [58]. In children treated with MB-AS, only 2% were parasitaemic on day 3, i.e., a significantly lower proportion than following treatment with AS-AQ (5%) or MB-AQ (17%) [56]. The clearance time for the P. falciparum parasite was non-significantly shortened in children receiving AS-AQ-MB compared to the AS-AQ group (medians 43.7 vs 47.0 h) [55]. Fever was cleared rapidly in all of these studies without differences between MB and comparator regimens [55, 56, 58].

Gametocyte clearance following MB-based treatment was investigated in two RCTs. The first study was a secondary analysis of data from an RCT in children. Compared to AS-AQ, both MB-containing regimens were associated with significantly reduced gametocyte carrier rates during follow-up days 3, 7 and 14 [57]. The second study, in which post-treatment gametocyte prevalence was the main outcome variable, demonstrated a significantly lower figure in children treated with MB-AS-AQ compared to AS-AQ on day 7 of follow-up (microscopically, 1% vs 9%; by QT-NASBA, 37% vs 63%) [55].

Treatment of malaria with MB was consistently associated with green-blue discoloration of urine. AEs concerning the urogenital system (urethritis) and the gastrointestinal system (vomiting) were reported to be associated with MB treatment. These AEs were more frequently reported with higher doses of MB [65, 66, 69, 75]. Vomiting in children was often associated with the bitter taste of MB, depending on the type of formulation. In two RCTs, vomiting in MB-containing regimens ranged from 24% to 68% [55, 56]. Urethritis was reported in 78% of adults and in 55% of older children treated with MB, but this symptom was not reported from RCTs in preschool children [56, 59]. In the controlled studies, severe adverse events (SAEs) were rarely reported and none were attributed to MB [60]. In particular, there were no cases of severe haemolysis associated with MB.

Despite some AEs being clearly associated with MB treatment, one study reported no differences between study groups in acceptance rates by parents and caregivers of children [55].

In our analysis of 15 case reports and case series, 339/373 (91%) of patients were cured, which provides evidence for the high efficacy of MB monotherapy in the treatment of malaria. Treatment failures were attributed to previously unsuccessful quinine treatment, vomiting and a low and/or short MB dosing regimen, which helps to explain early controversies on whether quinine or MB was more effective against malaria [63‐65, 77]. Conducted in tropical as well as in non-tropical areas and including all types of human malaria, these early studies were of overall good external validity. Interestingly, the MB dosing schedules used in these early studies were rather similar to what has been found to be effective in a MB dose-finding study in African children as well as in a proof of principle MB monotherapy study in African adults [59, 60].

In the recent studies, MB was usually given in combination with other antimalarials and tested in RCTs. In an area of high CQ resistance, MB-CQ was more effective than CQ alone but not sufficiently so [58, 60]. Combining MB with AS, AQ or AS-AQ and increasing MB doses in these combinations did not lead to a substantial increase in efficacy [55, 56]. This indicates that the curative efficacy of MB in eliminating asexual parasitaemia appears to be limited in the study region of Burkina Faso compared to the cure rates observed roughly a hundred years ago in a variety of settings. There is no evidence for and—because of the lack of drug pressure—no reason to expect that MB resistance has emerged meanwhile. The actual causes of the comparatively lower efficacy of MB-containing regimens in the recent studies, thus, remains unclear. Total dosage and a generally longer treatment period in the older studies might be relevant, however.

Despite this limitation, other properties of MB are notable and promising. Adding MB to an ACT reinforced the particular beneficial effects of the artemisinins, i.e., it further accelerated the elimination of asexual P. falciparum parasites and reduced P. falciparum gametocytes [55‐57]. The epidemiological relevance of the latter observation has recently been confirmed by a phase II study in Mali, which showed a 100% reduction of mosquito infectivity by day 7 both with PQ- and MB-containing drug regimens through membrane feeding assays [78]. This supports findings from preclinical studies on an existing synergy between MB and artemisinins as well as on the very strong effects of MB on P. falciparum gametocyte reduction [41, 44, 45, 79]. Interestingly, the efficacy of MB against gametocytes had been observed in historical studies [30]. However, further studies to identify the lowest effective dose for the gametocytocidal effect of MB in falciparum malaria should be conducted.

In the historical studies, MB was usually given orally and often in high doses and for prolonged periods of time, both in children and in adults, and without reports of major safety problems. During World War I for example, some European soldiers received more than 400 g of MB over several weeks without major side effects, apart from moderate urogenital symptoms [31]. Brazilian children were reported to tolerate 20–50 mg/kg per day of MB very well for long periods of time [68]. Also, in the recent RCTs conducted in West Africa, MB treatment was not associated with SAEs. However, while MB given orally seems to be largely well tolerated, MB given intravenously must be applied with caution. Intravenous MB is often routinely given as the first-line treatment for acute acquired methemoglobinemia in doses of 1–2 mg/kg, but a dose of 7 mg/kg can lead to severe gastrointestinal symptoms [9]. Moreover, a dose of 5 mg/kg has been reported to be associated with an altered mental status during parathyroidectomy [80]. For sheep, the LD50 of MB was found to be 42 mg/kg when applied intravenously [81].

AEs shown to be associated with MB treatment are mild gastrointestinal symptoms, which may manifest as vomiting, and mild urogenital symptoms, which usually manifest shortly after drug intake. This has consistently been reported, both in historical studies and in more recently conducted RCTs. Pure MB powder or MB dissolved in water has a very bitter and metallic taste. The gastrointestinal symptoms are clearly influenced by the formulation of MB, which suggests it is better to use taste-masked formulations. As the World Health Organization is now also recommending solid formulations for small children, mini-tablets, which should also be taste-masked, will be preferred over liquid formulations [82]. MB given with food and with a small amount of grated nutmeg has consistently been described as being effective in suppressing the frequently reported gastrointestinal and urogenital AEs [30, 60, 66]. Finally, both MB and PQ were recently shown to be well tolerated in males in Mali [78].

MB is on the list of drugs potentially dangerous for patients with G6PD deficiency, but the clinical importance of this is still controversial [83, 84]. In this review, no association between MB and severe haemolysis has been detected. Nevertheless, in a pooled analysis of all recent studies conducted with MB against falciparum malaria in West African children (including one unpublished RCT), small effects were seen. In 844 MB-treated African children, two patients developed severe anaemia (Hb < 5 g/dL) during the first days of treatment and both were G6PD deficient. Minimal Hb concentrations following MB treatment did not significantly differ in children with and without G6PD deficiency. However, when modelling the Hb time course, an MB dose-dependent effect of lowering Hb concentrations was observed for children with a full G6PD defect (hemi- and homozygous deficiency; prevalence, 10%). The maximal difference compared to non-deficient peers was estimated as −0.9 g/dl on day 5 of the follow-up, which is, however, considered to be of limited clinical relevance [85]. In sub-Saharan Africa, the moderate A minus type of X-chromosomally inherited G6PD deficiency (15–25% enzyme activity) dominates and affects 10–25% of the population. G6PD-deficient red blood cells have increased sensitivity to oxidative stress originating from various triggers including antimalarial drugs [86]. Whether or not MB confers an increased risk in malaria patients with more severe variants of G6PD deficiency needs to be evaluated in future studies, e.g. in South-East Asia. There are, however, reports of MB being associated with severe haemolytic reactions in neonatal G6PD deficiency [87, 88]. Finally, MB at low concentration is a strong inhibitor of monoamine oxidase A and therefore, should not be given together with serotonin reuptake inhibitors [89].

Reassuringly and despite the characteristic AE profile, MB has been well accepted by patients and caregivers of children in West Africa, which was also supported by an anthropological study in Burkina Faso [90]. However, the acceptance of its colouring properties also needs to be studied in other populations and cultures.

The strength of this review is that it includes all studies ever conducted with MB or MB-containing regimens in humans with malaria irrespective of study design and sample size and irrespective of the language of the report. Moreover, it can be considered as beneficial that a number of authors of this review have been the investigators of all recently conducted studies on this topic. Limitations of this review are the lower quality of the historical studies included compared to recent RCTs, the availability of three reports only in an abbreviated form from other sources [73, 91, 92] and the non-availability of a number of studies due to their old age despite extensive searches.