Prevalence and outcomes of malaria as co-infection among patients with human African trypanosomiasis: a systematic review and meta-analysis

Human African trypanosomiasis (HAT) is endemic in Africa; hence, the possibility of co-infection with malaria among patients with HAT exists. The present study investigated co-infection with malaria among patients with HAT to provide current evidence and characteristics to support further studies. Potentially relevant studies that reported Plasmodium spp. infection in patients with HAT was searched in PubMed, Web of Science, and Scopus. The risk of bias among the included studies was assessed using the checklist for analytical cross-sectional studies developed by the Joanna Briggs Institute. The pooled prevalence of Plasmodium spp. infection in patients with HAT was quantitatively synthesized using a random-effects model. Subgroup analyses of study sites and stages of HAT were performed to identify heterogeneity regarding prevalence among the included studies. The heterogeneity of the outcome among the included studies was assessed using Cochran’s Q and I2 statistics for consistency. Publication bias was assessed if the number of included studies was 10 or more. For qualitative synthesis, a narrative synthesis of the impact of Plasmodium spp. infection on the clinical and outcome characteristics of HAT was performed when the included studies provided qualitative data. Among 327 studies identified from three databases, nine studies were included in the systematic review and meta-analysis. The prevalence of Plasmodium spp. co-infection (692 cases) among patients with HAT (1523 cases) was 50% (95% confidence interval [CI] = 28–72%, I2 = 98.1%, seven studies). Subgroup analysis by type of HAT (gambiense or rhodesiense HAT) revealed that among patients with gambiense HAT, the pooled prevalence of Plasmodium spp. infection was 46% (95% CI = 14–78%, I2 = 96.62%, four studies), whereas that among patients with rhodesiense HAT was 44% (95% CI = 40–49%, I2 = 98.3%, three studies). Qualitative syntheses demonstrated that Plasmodium spp. infection in individuals with HAT might influence the risk of encephalopathy syndrome, drug toxicity, and significantly longer corrected QT time. Moreover, longer hospital stays and higher treatment costs were recorded among co-infected individuals. Because of the high prevalence of malaria among patients with HAT, some patients were positive for malaria parasites despite being asymptomatic. Therefore, it is suggested to test every patient with HAT for malaria before HAT treatment. If malaria is present, then antimalarial treatment is recommended before HAT treatment. Antimalarial treatment in patients with HAT might decrease the probability of poor clinical outcomes and case fatality in HAT.

Search strategy. The search strategy involved the use of combinations of the following search terms: "(Malaria OR Plasmodium) AND (Trypanosomiases OR "Sleeping Sickness" OR "African Sleeping" OR Nagana OR Trypanosome OR Nannomona) AND (co-infection OR co-infection OR "Co-infection" OR mixed OR concurrent OR Polymicrobial OR multiple OR dual)" (Table S1). All relevant search terms were retrieved from the Medical Subject Heading in The National Center for Biotechnology Information. The searches were performed in PubMed, Web of Science, and Scopus on July 4, 2021, without restriction on language or publication year. Additional searches of the reference lists of the included studies and Google Scholar were performed to assure that all relevant studies were captured in the search protocol.
Eligibility criteria. The eligibility criteria were related to participants (P), phenomena of interest (I), and context (Co) or PICo. The primary outcome of this study was the pooled prevalence of Plasmodium spp. infection among patients with HAT globally. Therefore, P was patients with HAT, I was Plasmodium spp. infection and Co was worldwide context. Original studies (retrospective or prospective observational, cross-sectional studies, cohort, clinical trials, or case-control studies) that reported Plasmodium spp. infection among patients with HAT was included. Conversely, case studies, case series, letters to the editors, commentary, reviews, systematic review, in vitro studies, and in vivo studies were excluded.

Study selection. Study selection was performed independently by two authors (KUK, MK). Discrepancies
between the authors were resolved through discussion until a consensus was reached. After duplicates were removed, the remaining studies were screened for titles and abstracts. After non-relevant studies were excluded, the full text of the remaining studies was examined according to the eligibility criteria. Then, studies that did not meet the eligibility criteria were excluded for specific reasons. Finally, studies that met the eligibility criteria were included for further data extraction. Data extraction. Studies were extracted into a standardized pilot datasheet under the following categories: name of the first author, year of publication, study site, the year when the study was conducted, study design, number of participants enrolled, characteristics of participants, age, male ratio, number of infecting Plasmodium spp. among patients with HAT, number of patients with HAT, malaria diagnostic method, and HAT diagnostic method.

Risk of bias.
The risk of bias among the included studies was assessed using the checklist for analytical cross-sectional studies developed by the Joanna Briggs Institute (JBI) 19  Data syntheses. Data syntheses included qualitative and quantitative syntheses. For quantitative synthesis, statistical analysis was used to determine the pooled prevalence of Plasmodium spp. infection among patients with HAT. Subgroup analyses were performed by types of HAT (gambiense or rhodesiense HAT) to identify the source of heterogeneity regarding prevalence among the included studies. The heterogeneity of the outcome among the included studies was assessed using Cochran's Q and the I 2 statistic for consistency. Cochran's Q less than 0.05 or I 2 greater than 50% indicated substantial heterogeneity of the outcome among the included studies. Pooled analysis of the prevalence was performed using a random-effects model 20 . Publication bias was assessed if the number of the included studies was at least 10 21 . All quantitative analyses were performed using Stata version 14 (StataCorp, College Station, TX, USA). For qualitative synthesis, a narrative synthesis of the impact of Plasmodium spp. infection on clinical and outcome characteristics of patients with HAT was performed when the included studies reported qualitative data.

Results
Search results. Three hundred twenty-seven studies were identified in PubMed (179 studies), Web of Science (94 studies), and Scopus (54 studies). After 95 duplicates were removed, 232 studies were screened for titles and abstracts. After 218 non-relevant studies were excluded, the full text of 14 studies was examined. After screening the full text of the studies, six animal studies, two review articles, and one article in which no patients with HAT had malaria were excluded. Five studies [22][23][24][25][26] were included in the systematic review. An additional four studies 6,27-29 were identified from searches of reference lists. Finally, nine studies 6,[22][23][24][25][26][27][28][29] were included in the systematic review and meta-analysis ( Fig. 1).  (Fig. 2). Two studies 22,25 were clinical trials, two 23,24 were retrospective studies, three 6,28,29 were cohort studies, one 27 was a cross-sectional study, and one 26

Risk of bias.
The risk of bias among the included studies was assessed using the checklist for analytical cross-sectional studies developed by the JBI. A score of 7 points was given for five studies 23,24,[27][28][29] , which were considered to have a low risk of bias, whereas four studies 6,22,25,26 were given scores of 5-6 points and categorized as having a moderate risk of bias (Table S2).  Authors are allowed to use, edit, and modify any map created with mapchart.net for publication freely by adding the reference to mapchart.net.

Effect of Plasmodium spp. infection in HAT.
Seven studies [22][23][24][25][26][27]29 determined the effect of Plasmodium spp. infection in HAT. Blum et al. 22 suggested that malaria might influence the risk of encephalopathy syndrome, as 14 of 16 patients with HAT and encephalopathy syndrome of the coma type were infected with malaria (87.5%) in Angola. Blum et al. 29 demonstrated that patients with HAT who were co-infected with Plasmodium spp. had a significantly longer corrected QT time (QTc). Kagira et al. 23 suggested that malaria might increase the risk of drug toxicity; however, the effect of Plasmodium spp. in HAT in Kenya was not investigated in their study. The study by Kato et al. 24 conducted a retrospective study in Uganda and concluded that malaria did not significantly affect the clinical presentation or death rates of HAT. However, longer hospital stays and higher treatment costs were recorded among co-infected individuals. Kuepfer et al. 25

conducted a clinical trial of patients with HAT in
Tanzania and Uganda and also demonstrated that Plasmodium spp. did not influence the clinical presentation or treatment outcomes of HAT. Maina et al. 27 conducted the cross-sectional study of HAT in Sudan demonstrat-  27 . Nsubuga et al. 26 conducted a case-control study of HAT in Uganda that assessed TNF-α levels and found that Plasmodium spp. infection was linked to higher TNF-α levels in patients with HAT than observed in patients with HAT or malaria mono-infections.

Discussion
Testing for concomitant malaria infection in patients with HAT is not mandatory and is only conducted according to the decisions of the treating personnel 6 . In cases of malaria co-infection, antimalarial treatment is generally given as an ancillary drug before initiating HAT treatment to prevent encephalopathy [30][31][32] . The meta-analysis revealed that the high prevalence of malaria in HAT was high (50%). Nevertheless, the pooled prevalence appeared to be heterogeneous among the included studies. Subgroup analysis by HAT type indicated that the pooled prevalence of Plasmodium spp. infection was similar between patients with gambiense HAT and rhodesiense HAT (46% and 44%), and high heterogeneity was observed in the prevalence. Several reasons might explain the heterogeneity regarding prevalence among the included studies. The differences in the investigated participants might be the source of heterogeneity. For example, the study by Blum et al. 22 examined Plasmodium spp. infection in patients with HAT and encephalopathy alone, whereas Kuepfer et al. 25 examined Plasmodium spp. infection in patients with HAT and various types of complications. The participants in both studies had high rates of Plasmodium spp. infection. In addition to differences among the participants, the differences in the study design might be another cause of heterogeneity in the pooled prevalence. Two clinical trials 22,25 included in the meta-analysis recorded a high prevalence of co-infection (83-88%). Blum et al. 22 examined the blood slides of patients with HAT for malaria in the case of a reaction concurrent with fever, as malaria was suspected to increase the risk of encephalopathy in their participants; hence, the prevalence of malaria was high. Kuepfer et al. 25 assessed malaria per a standard protocol to test co-infection with malaria at baseline and determine whether malaria influenced the clinical presentation or treatment outcomes in a Tanzanian study population; hence, the prevalence of malaria was high. The difference in the prevalence of co-infection might be explained by the difference in the study sites investigated. Blum et al. 22 conducted their study in Angola, whereas the study by Maina et al. 27 was performed in Sudan. The most recent WHO reports illustrated that the malaria burden share was higher in Angola (3%) than in Sudan (1%) 9 . Among studies conducted in East Africa, where T.b. rhodesiense is predominant, the study by Kagira et al., 23 which was conducted in Kenya, recorded a higher prevalence of co-infection (100%) than the studies by Kuepfer et al., 25  www.nature.com/scientificreports/ Kato et al., 24 which was conducted in Uganda (29%), and Nsubuga et al., 26 which was conducted Uganda (47%). A possible explanation of the high prevalence of Plasmodium spp. infection among patients with HAT (35-71%) was that it was mainly linked with the high prevalence of malaria in the general population of Africa, where malaria is highly endemic 8 . The occurrence of Plasmodium spp. infection among patients with HAT suggested that these patients were susceptible to other infections, which might influence the pathogenesis and prognosis of the disease. As HAT is a chronic disease that leads to long-term immunosuppression 27,33 , which could increase the possibility of Plasmodium spp. infection in patients with HAT. Nevertheless, HAT is usually a chronic disease, but rhodesiense HAT often causes a more acute and rapidly progressive disease than gambiense HAT 34 . The role of immune responses in co-infected patients was demonstrated by Nsubuga et al. 26 , who revealed that co-infection by P. falciparum increased TNF-α levels in patients with HAT. Previous studies revealed associations of TNF-α with the progression and severity of HAT 35,36 . The synergistic effect of TNF-α on disease progression was related to increased IFN-γ levels 37,38 . The possibility of Plasmodium spp. infection among patients with HAT was probably high in previous studies 6,22,25,28 that enrolled only patients with late-stage HAT. Although the study by Blum et al. 6 that enrolled patients with late-stage HAT did not report the number of cases of Plasmodium spp. infection among their participants, a malaria prevalence of 50% among patients with HAT was recorded in several endemic countries including Angola, Central African Republic, Côte d'Ivoire, Democratic Republic of Congo, Equatorial Guinea, Republic of Congo, and Southern Sudan.
Testing for concomitant malaria infection in patients with HAT was not mandatory, and it was only performed according to the discretion of the treating personnel 6 . In cases of malaria co-infection, antimalarial treatment is generally given as an ancillary drug before initiating HAT treatment to prevent encephalopathy [30][31][32] . Because the examination of malaria parasites among patients with HAT was not mandatory, the possibility of Plasmodium spp. infection in HAT could not be discounted. The presence of fever in patients with HAT might support the presence of malaria parasites, and further examinations of malaria parasites are required. Fever was reported in half of the patients with HAT, resulting in a febrile reaction during HAT treatment 6 . If malaria parasites are present, then antimalarial treatment is generally given before initiating HAT treatment 31,32 . Antimalarial treatment was described by Blum et al. 22 in which a full course of 1500 mg of chloroquine was given before the start of HAT treatment with melarsoprol. www.nature.com/scientificreports/ Concerning the effect of Plasmodium spp. infection in HAT, a previous study found that the co-infected patients required multiple tests and experienced higher treatment costs because of the use of multiple drugs 23 . In addition, co-infected patients might have a poor prognosis due to prolonged hospital stay, receiving more drugs, and experiencing increased drug toxicity from melarsoprol treatment 23,24 . Kagira et al. 23 observed that malaria could increase the toxicity of drug therapy. Suramin was injected in patients with early-stage rhodesiense HAT, whereas melarsoprol was used to treat patients with late-stage HAT 23 . Previous research demonstrated that the most serious side effect of melarsoprol treatment is post-treatment reactive encephalopathy characterized by increased mental deterioration, coma, and convulsions or death in some cases after treatment 39 . Blum et al. 29 reported increased QTc in patients with HAT and concomitant malaria, suggesting the probability of melarsoprol-induced ventricular dysrhythmia in late-stage HAT. However, one study suggested that malaria and other co-infections did not significantly affect the clinical presentation and case fatality rates of HAT 24 . A similar observation was reported by Kuepfer et al., 25 who conducted a study in Tanzania and Uganda. They revealed that Plasmodium spp. infection did not influence the clinical presentation or treatment outcomes of HAT 25 . However, the outcome of this co-infection might be influenced by several factors such as the stage of infection and age and sex of the patient 40 .
The present study had several limitations. First, the number of studies that examined malaria and trypanosome co-infection was limited. Therefore, the differences in clinical characteristics, laboratory parameters, and treatment outcomes between co-infected patients and patients with malaria or HAT mono-infection could not be investigated. Second, there was high heterogeneity regarding the pooled prevalence of co-infection in the meta-analysis; therefore, the pooled prevalence should be carefully interpreted.

Conclusion
Because of the high prevalence of malaria among patients with HAT and the endemic nature of malaria in Africa, malaria was present in some asymptomatic patients. Therefore, it is suggested to test every patient with HAT for malaria before initiating HAT treatment. If malaria is present, patients should receive antimalarial treatment before HAT treatment is initiated. Antimalarial treatment in patients with HAT might decrease the probability of poor clinical outcomes and case fatality rates.