Identification of Plasmodium falciparum and host factors associated with cerebral malaria: description of the protocol for a prospective, case-control study in Benin (NeuroCM).

INTRODUCTION: In 2016, an estimated 216 million cases and 445 000 deaths of malaria occurred worldwide, in 91 countries. In Benin, malaria causes 26.8% of consultation and hospitalisation motif in the general population and 20.9% in children under 5 years old.The goal of the NeuroCM project is to identify the causative factors of neuroinflammation in the context of cerebral malaria. There are currently very few systematic data from West Africa on the aetiologies and management of non-malarial non-traumatic coma in small children, and NeuroCM will help to fill this gap. We postulate that an accurate understanding of molecular and cellular mechanisms involved in neuroinflammation may help to define efficient strategies to prevent and manage cerebral malaria. METHODS AND ANALYSIS: This is a prospective, case-control study comparing cerebral malaria to uncomplicated malaria and non-malarial non-traumatic coma. This study takes place in Benin, precisely in Cotonou for children with coma and in Sô-Ava district for children with uncomplicated malaria. We aim to include 300 children aged between 24 and 71 months and divided in three different clinical groups during 12 months (from December 2017 to November 2018) with a 21 to 28 days follow-up for coma. Study data, including clinical, biological and research results will be collected and managed using CSOnline-Ennov Clinical. ETHICS AND DISSEMINATION: Ethics approval for the NeuroCM study has been obtained from Comité National d'Ethique pour la Recherche en santé of Benin (n°67/MS/DC/SGM/DRFMT/CNERS/SA; 10/17/2017). NeuroCM study has also been approved by Comité consultatif de déontologie et d'éthique of Institut de Recherche pour le Développement (IRD; 10/24/2017). The study results will be disseminated through the direct consultations with the WHO's Multilateral Initiative on Malaria (TDR-MIM) and Roll Back Malaria programme, through scientific meetings and peer-reviewed publications in scientific or medical journals, and through guidelines and booklets. © Author(s) (or their employer(s)) 2019. Re-use permitted under CC BY-NC. No commercial re-use. See rights and permissions. Published by BMJ.

This case-control study aims to identify the causative factors of neuroinflammation in the context of cerebral malaria.

This study will inform on the aetiologies and management of non-malarial non-traumatic coma in small children.

The final products of NeuroCM are expected to feed the pipeline of new therapeutical (immune intervention) and preventive (vaccine) strategies that will improve cerebral malaria outcome.

This study does not have the power to investigate all aetiologies of fever in Benin. Contrary to the malaria groups, there is no information on the frequency of non-malarial non-traumatic coma admissions, and no certainty on the number of children who will be included in the non-malarial non-traumatic group.

According to the limited number of patients, conclusions will further need to be confirmed in larger studies.

Introduction

Malaria is triggered by an apicomplexan parasite, Plasmodium spp. Six Plasmodium species can infect humans, with Plasmodium falciparum (P. falciparum) being the most frequent in Sub-Saharian Africa (99.7% of estimated cases in 2017). P. falciparum is the agent of severe malaria and responsible for most malarial deaths.

In 2017, an estimated 219 million cases and 435 000 deaths of malaria occurred worldwide, in 91 countries.1 Despite a recent decrease in malaria mortality due to extensive malaria control through insecticide impregnated bednets and increased use of artemisinin derivatives, 275 000 children still die every year from malaria. Most cases and deaths were in African region (respectively 92% and 93%). Severe malaria occurs mostly in non-immune patients and in Sub-Saharian Africa and 90% of severe malaria affect young children.2 In endemical states, malaria is one of the three major causes of hospitalisation in children under 5 years old.

Benin is in West Africa, along the Gulf of Guinea In 2018, 11.5 million people live in Benin, most of them in South Benin. Malaria transmission occurs seasonally, during rainy seasons (from May to August and October). According to the Beninese health department in 2016, malaria is responsible for 26.8% of disease reports in consultation and hospitalisation in the general population and for 20.9% in children under 5 years old.3 It is also the first morbidity cause in the general population with a prevalence of 39.7% in 2013, followed by respiratory infections in 12.4% cases and gastro-intestinal disease for 6.4%.4

According to the WHO, severe falciparum malaria is defined by the association between P. falciparum asexual parasitaemia and the presence of one or more of the clinical or laboratory features (with no other confirmed cause for their symptoms) presented in table 1. Cerebral malaria (CM) is defined by the presence of asexual form of P. falciparum associated with Blantyre score ≤2 (table 2). CM is a coma which persists for >1 hour after a seizure irrespective of anticonvulsant medications. Clinical criteria for CM diagnosis are currently debated, and it has been highlighted that a P. falciparum parasitaemia can be observed in comatose children with coma related to a non-malarial central nervous system disease,5 leading to a possible overestimation of CM cases. A recent study in Malawi found that 25% of CM cases were misdiagnosed and that many children may have had a viral meningoencephalitis concomitant to a malarial infection.5 Implementation of fundoscopic examination (in order to look for malaria retinopathy signs)6 and microbiological investigations (blood culture, cerebrospinal fluid (CSF) culture, CSF multiplex polymerase chain reaction (PCR)) could limit the overestimation of CM diagnosis, but fundoscopic examination requires trained physicians and microbiological investigations are expensive. Clinical research needs to focus on new clinical or diagnostical tools designed to help physicians in order to better diagnose CM.

Table 1

Clinical and laboratory criteria for severe malaria (from (4))

Clinical manifestations	Prognosis value	Frequency in children
Impaired consciousness	+++	+++
Respiratory distress	+++	+++
Multiple convulsions	+	+++
Prostration	+	+++
Shock	+++	+
Pulmonary oedema (radiology)	+++	+/−
Abnormal bleeding	+++	+/−
Jaundice	++	+

Table 2

Blantyre score (from (4))

	Score
Best motor response
Localises painful stimulus	2
Withdraws limb from pain	1
Non-specific or absent response	0
Verbal response
Appropriate cry	2
Moan or inappropriate cry	1
None	0
Eye movement
Directed	1
Not directed	0
Total	0–5

NeuroCM study aims to collect data (such as blood and cerebrospinal fluid culture, fundoscopic examination) on coma’s aetiologies in Beninese young children.

Without treatment, CM is invariably fatal. Even with parenteral artemisinin use, severe malaria death rate is 20%.7 In case of severe malaria or CM, patients should be hospitalised in an intensive care unit. Conscience status, blood pressure, heart and respiratory rates, diuresis and oxygen saturation have to be monitored every 6 hours.8 Parenteral artemisinin instead of quinine is recommended at a posology of 2.4 mg/kg by injection (3 mg/kg in children under 20 kg of body weight) at diagnosis, 12 and 24 hours later and then daily until patients can take oral drugs.7 Benzodiazepine should be used to treat convulsions (diazepam 0.3 mg/kg, midazolam or lorazepam). It seems accepted that CM surviving patients generally do not present any neurological sequelae and fully recover their neurological capacity. However, immediate neurological after-effect is described in 6.7% to 11.6% of cases7 9 and a recent meta-analysis found a relation between CM and neurological disease.10 The NeuroCM study will collect data on children’s clinical recovery at discharge and 21 to 28 days later.

Tools for malaria control are less and less effective. On one hand, P. falciparum drug resistance is a growing concern. Resistance to chloroquine, one of the main anti-malarial drugs, appeared during the sixties in South-East Asia and then spread to Africa.11 12 Artemisinin-combined therapy became the treatment of choice for malaria with the aim to reduce the risk of parasites developing resistance,13 but resistance appeared in South-East Asia in 200814 and was confirmed in others studies.15 It has not, hitherto, spread to Africa but this is a real concern for the WHO.16 On the other hand, mosquitos become more and more resistant to insecticides, making antivectorial prevention more and more difficult.17 Thus, research for new therapies is needed.

Pathophysiology of CM is complex and multifactorial, based on both parasite and host immune factors. It is currently believed that CM is caused by dedicated parasite variants that specifically localise in brain through interaction between parasite proteins expressed on the surface of the infected erythrocytes (iE) and brain endothelium. This sequestration occurs with erythrocytes infected with late stage of P. falciparum (trophozoites and schizonts). Binding of iE to endothelial vascular cells is mediated by variant surface antigens (VSA). Plasmodium virulence is linked to its ability to express VSA.18 VSA include three different multigenic families: var, rifin and stevor. More specifically, var genes coding for Plasmodium falciparum erythrocyte membrane protein 1 (PfEMP1) proteins are highly polymorphical and present in sixty copies in P. falciparum genome. Those PfEMP1 are expressed on iE surface and are responsible for endothelial receptors binding, such as chondroitin sulfate A (CSA) and endothelial protein C receptor (EPCR) or intercellular adhesion molecule (ICAM) respectively involved in placental malaria19 and severe malaria.20 PfEMP1 family has been clearly associated with binding of iE to the microvascular endothelium of every organ and tissue.21 We now better understand iE’s binding on placenta22 and vaccine development to prevent gestational malaria seems an achievable goal. By contrast, research is still needed to understand which type of proteins specifically binds to cerebral endothelial receptor. In a previous study conducted in Benin, we identified several proteins associated with CM.23

The finding of a PfEMP1 variant specifically related to CM could pave the way to the development of a vaccine targeting this specific protein. Studying the transcriptomical and proteomical profiles of plasmodial strains involved in CM compared with strains involved in uncomplicated malaria (UM) is a first step to better understand related mechanisms to cerebral endothelium binding.

The host immune aspect of the pathophysiology of CM are the consequences of microvascular sequestration of iEs and rosettes (non-infected red blood cells) in brain. Such sequestration leads to blood flow obstruction, ischaemia/hypoxia and local inflammation resulting in neuroinflammation.24–28 The way this cascade of events is linked and the reasons why they result in death or inflammation resolution remains to be elucidated. Acute cerebral ischaemia is known to drive microglia activation and influx of myeloid immune cells to the brain. Resident microglia and infiltrating monocytes/neutrophils have a critical role in initiating, sustaining (M1 polarisation) and resolving (M2 polarisation) post-ischaemic inflammation.29–31 Another important immune aspect of neuroinflammation during CM is redox equilibrium. The production of reactive oxygen species both by parasites (haemoglobin digestion) and monocytes/macrophages are thought to cause damages to neurons.32 This process leads to blood-brain barrier (BBB) permeability and neurodegeneration.33 34 In order to counterbalance the excess of oxidants, oxidant scavengers and antioxidant enzymes may be produced. NO bioinsufficiency and subsequent vasoconstriction constitute other important aspects of CM pathophysiology.35 Haem and superoxide anion release during infection leads to NO mobilisation for detoxification, depriving vascular smooth muscle cells in NO and leading to inflammation-related vasospasm.35 36 Although vasospasm has not been clearly associated to death risk during CM,37 NO pathway deserves a better understanding during CM pathophysiology. In the NeuroCM study, we intend to better understand mechanisms of neuroinflammation and its resolution in a context of CM, by comparing data collected in children presenting with CM, in children hospitalised for non-malarial non-traumatic coma and in children with UM. We will focus our studies on markers of immune cell migration and polarisation (towards inflammatory or resolutive phenotypes), of pro- or anti-oxidant response and of pro- or anti-inflammatory response through urine and blood samples analysis at inclusion, 3 and 21 to 28 days post-inclusion.

Study objectives

The main objective is to identify the causative factors of neuroinflammation in the context of CM. There are currently very few systematic data from West Africa on the aetiologies and management of non-traumatic coma in small children, and NeuroCM will bring new information on these aspects. We postulate that an accurate understanding of molecular and cellular mechanisms involved in neuroinflammation may help to define efficient strategies to prevent and manage CM.

There are three distinct objectives in this study.

To identify parasitological factors associated with P. falciparum CM or UM

We expect to identify and validate P. falciparum virulence factors associated with CM by comparison with UM. Once proteins of interest will be found, functional studies will help to better understand their role in CM.

To identify immune host factors associated with fatal of favourable outcome of CM

We expect to better understand which mechanisms trigger neuroinflammation and its resolution during CM by comparing three groups of children: presenting with CM, hospitalised for non-malarial non-traumatic coma and presenting with UM. We aim to identify therapeutical molecular targets involved in neuroinflammation resolution.

To describe coma’s aetiology in Sub-Saharian Africa

We expect to improve knowledge in non-malarial non-traumatic coma’s aetiologies in Sub-Saharian Africa in order to improve young children’s coma management and inform health public policies on the role played by infections that could be prevented by vaccination.

Methods and analysis

Design

This is a prospective, case-control study comparing CM to UM and non-malarial non-traumatic coma. Patients will be recruited in South Benin, in two different hospitals for coma and in a dispensary for UM, as UM is rarely detected in hospitals where children with coma are managed. This study is conducted by one Beninese research team (CERPAGE, Centre d’Etude et de Recherche sur le Paludisme Associé à la Grossesse et à l’Enfance) and three French research teams (UMR D216 MERIT in Paris carrying the project, UMR D152 PHARMADEV in Toulouse, UMR S1094 NET in Limoges).

Study environment

This study takes place in Benin, precisely in Cotonou and Calavi for the hospital’s recruitment. UM recruitment takes place in Sô-Ava district. Cotonou is the largest city and economical centre of Benin, with an estimated population of 679 012 habitants in 2013.

The two recruitment hospitals are the CHU-MEL (CHU-Mère et Enfant de la Lagune) and Hôpital de zone de Calavi. Routine laboratory analysis, except bacteriology, are performed on site for children included with coma and at IRCB (Institut de Recherche Clinique du Bénin) for children with UM. Bacteriological analyses are performed in the microbiology laboratory of CNHU (Centre National Hospitalier Universitaire). Research analyses are done in the CERPAGE laboratory.

Participants

We aim to include three different clinical groups of 100 children between 24 and 71 months during 12 months (from December 2017 to November 2018). This duration has been determined according to previous studies in Benin.38

In the first group, a diagnosis of CM will be defined as follows: positive P. falciparum thin blood smear with a Blantyre score ≤2 with exclusion of patients presenting: positive bacteraemia, meningitidis proved or suspected (leucocytes in cerebrospinal fluid >1000 per microlitre and/or Gram positive in CSF and/or CSF bacterial culture positive and/or PCR positive for any bacteria or virus).

In the second group, a diagnosis of non-malarial non-traumatic coma will be defined as follows: Blantyre score ≤2 and no Plasmodium infection as detected by thin blood smear.

In the third group UM will be defined as follows: (1) fever at inclusion or within 24 hours before, (2) no clinical or biological sign of severe malaria (table 1), no danger signs and no other obvious cause of fever and (3) P. falciparum parasitaemia between 1000 to 500 000 parasites per microlitre.

Inclusion and exclusion criteria

For all children, the first inclusion criterion is parental acceptance that their child participate in the study after information has been given (see section ‘Ethics and safety considerations’). Inclusion criteria for coma (CM and non-malarial non-traumatic coma) are: age between 24 to 71 months, Blantyre score ≤2, negative HIV rapid diagnostic test (RDT). Non-inclusion criteria are: pre-existent neurological disease and traumatical or toxical coma.

Inclusion criteria for UM are: age between 24 to 71 months, fever >38°C at inclusion or within 24 hours before and no clinical severity/danger sign, positive malaria RDT and negative HIV RDT.

Exclusion criteria for coma are: thick and thin blood smear not realised at day 0 (D0) and/or biological blood test not realised at D0 and/or research blood test not realised at D0.

Exclusion criteria for UM are: thick and thin blood smear not realised at D0 and/or biological blood test not realised at D0 and/or research blood test not realised at D0 and/or laboratory indices for severe malaria and/or thick and thin blood smear negative for P. falciparum and/or parasite density under 1000 parasite per microlitre or higher than 500 000 parasites per microlitre. To evidence a significant difference between CM and UM groups in the ratio of endogenous mediators associated with inflammation resolution, we estimated that a sample size of 100 subjects per group was sufficient to reach the main study target, that is, by linear regression analysis involving a maximum of six predictors and an R2 value of 0.400, ensuring an 80% power and a 5% probability of type I error. This sample size also complies with the requirements of the reverse transcription quantitative PCR (RT-qPCR) analysis used to validate the discrimination of CM and UM samples obtained by SARTools, and finally with the overall funding request of the project.

Recruitment process

Step 1: enrolment/screening

For CM and non-malarial non-traumatic coma group, every young child with neurological symptoms is screened for eligibility. For UM group, every child presenting at the outpatient clinic with fever or fever during the previous 24 hours is screened. The first step is patients’ screening to confirm study eligibility and provide participants with information about the study. A questionnaire assessing eligibility will inform on home addresses, sociodemographical data (number of children in the family, ethnical group…), clinical history, use of mosquito net and vaccination status. Informed consent is then obtained from the parents or caregivers.

The following tests are performed to screen for malaria and to rule out HIV infections: a RDT detecting histidin-rich protein 2 (HRP2) for P. falciparum detection and Determine HIV-1/2 set, Alere for HIV detection.

Step 2: clinical examination and biological sample/analysis

A clinical examination is performed by a study physician for children hospitalised with coma, and by a study nurse for UM. In the coma group, a fundoscopic assessment is performed (Eyepax 1.0 Dioptrix) and pictures are captured and downloaded on an online database. The clinical data entry is performed on an online case report form.

In order to allocate children to their respective groups, biological analyses according to severe malaria are needed. For coma’s inclusion: thick and thin blood smear analysis, complete blood count (CBC) (Sysmex KX-21N, Sysmex XT-1800i and ABS Micro ES 60), biochemistry analysis (Na+, K+, Cl-, Ca++, HCO3 -, albumin, urea, creatinine, glucose, lactate) with Piccolo Sysmex and ALAT plus bilirubin (Biolabo Kenza Max biochemistry and Mindray BS-200) are performed on site. Blood culture, Gram staining and bacterial culture for CSF are realised in a university hospital reference laboratory. Biomérieux Biofire FilmArray Meningitis/Encephalitis Panel multiplex PCR (looking for E. coli, H. influenzae, L. monocytogenes, N. meningitidis, S. agalactiae, S. pneumoniae, cytomegalovirus, enterovirus, Epstein-Barr virus, herpes simplex virus (HSV) 1, HSV2, human herpesvirus 6, paraechovirus, varicella zona virus and Cryptococcus neoformans and C. gattii) will be further performed in France. The required following samples are needed: one Etylenediaminetetraacetic Acid (EDTA) tube (2 mL) for CBC and malaria diagnostic, one heparin tube (2 mL) for biochemistry analysis, one cerebrospinal sample (1 mL), one blood sample for blood culture (5 mL) for routine analyses, two additional EDTA tubes (6 mL) and 50 mL of urine for research analyses.

For UM inclusions: severe malaria was ruled out according to results from blood cell count (Sysmex XS500i), biochemistry analysis (bilirubin, glucose, creatinine) on Selectra pro automate (Elitech group) and thick and thin blood smear. The following samples are needed: one EDTA tube (2 mL), one heparin tube (2 mL) and one fluorinated tube (2 mL) for routine analyses, two additional EDTA tubes (6 mL) and 50 mL of urine for research analyses.

Step 3: research analyses

A part of research analyses is realised in CERPAGE laboratory. Blood, 1 to 1.5 mL, is cultured in supplemented Roswell Park Memorial Institute 1640 medium with Albumax (Gibco) for less than 48 hours until parasites reach the mature stage (from young trophozoite to schizont), then purified using magnetic-activated cell sorting (MACS; Milteny Biotec, Bergisch Gladbach, Germany) for binding and endothelial cell activation assay. The resulting mature stage are stored at −80°C for further mass spectrometry protein analysis. Whole blood samples of 200 µL are conserved at −20°C for DNA analysis, 200 µL are transferred in TRIzol reagent (Life technologies, France) and stored at −80°C for further RNA extraction39 and 200 µL in liquid nitrogen for parasite cryoconservation. Plasma samples are conserved at −20°C and −80°C respectively for immune response analysis and dosage of biomarkers. Peripheral blood mononuclear cells (PBMC) are separated from red blood cells by Ficoll density gradient and stored in liquid nitrogen. Finally, urines are stored at −80°C for further analysis. See table 3 for detailed research planning.

Table 3

Detailed research planning

Task	Calendar
	2017			2018												2019												2020
	T4			T1			T2			T3			T4			T1			T2			T3			T4			T1			T2			T3
Cohort recruitment and follow-up
Area preparation
Inclusion
Follow-up
Biological samples organisation
Parasite factors
Parasite whole genome sequencing
Parasite RNA-sequencing
Mass spectrometry analysis
Identified protein validation
Protein’s role on endothelium activation
Host factors
Macrophage M2 kinetics apparition in mice brain
Endogenous mediator role in neuroinflammation
Neuroinflammation markers identification in cerebral malaria patients
Results exploitation
Database validation
Data analysis
Dissemination

Parasite factors analyses will be performed in several ways. We will compare CM and UM isolates with whole genome DNA sequencing; RNA-sequencing and by quantitative MS analysis. Highly polymorphical var genes will be assembled and BLASTed against peptide hits from the MS approach. Nucleotide primers will be designed with DNA-sequencing data and used in RT-qPCR to validate the RNA-seq data. Associations between gene polymorphisms and modifications in RNA nature and quantity detected by RNA-seq will be investigated. Then, we will use recombinant protein and P. falciparum genome modification by gene disruption to study proteins’ role.

Host factors study will be based on the analysis of PBMC, plasmas and urine samples of the three groups of children. PBMC analysis will focus on the phenotyping of monocytes to distinguish M1 and M2-like phenotypes. Flow cytometry will be used to measure expression levels of CD11b and CD16 as M1 markers, and CD163 and CD206 as M2 markers. The assessment of gene expression levels of cytokines, chemokines and their receptors by RT-qPCR will complete phenotype analysis. Plasmas and urine samples will allow to measure redox (L-arginine and biopterins), pro-/anti-inflammatory (cytokines, chemokines and lipid mediators such as eicosanoids) and pro-resolving mediators (such as prostaglandins and lipoxins) by ELISA or EIA. We will first compare data from the group of CM to the two other groups in order to identify the biological markers best related to inflammation and neurological impairment during CM. Second, we will analyse data obtained with the two coma groups at inclusion (Day 0), at Day 3 and Day 30 to understand the kinetics of immune events and its relation to death or favourable outcome. Finally, we will search for severity and death risk factors within the CM groups.

Step 4: coma follow-up

In children presenting with coma, both clinical data and blood samples are collected at day 3 (D3) and day 21 to 28 (D21-28) on disease outcome, and for research purpose. One EDTA tube (6 mL) and 50 mL urine will be sampled. In order to prevent losses, parents/guardians are called a few days before D21-28 to remind them of follow-up visit. No follow-up visit is scheduled for children with UM.

Data management

Data, including clinical, biological and research results are collected and managed using CSOnline-Ennov Clinical (https://ufrcb.chu-limoges.fr/crfonline/). It is a secure, web-based application designed to support data capture for research studies. Study participants are identified by a code and have their own account. The two physicians and the nurse were trained to entry the data on included children in the database. Nobody can delete a patient created in the base, except the data manager.

Data & Safety Monitoring Board, composed by two physicians specialised in infectious disease and one statistician, will review allocation of children to the pre-defined study groups and discuss possible deviations from the expected number of subjects in the groups.

Data analysis

In a first step, descriptive statistics will be realised by calculating mean and SD for quantitative variables, and proportion for qualitative variables to determine the main characteristics of the three clinical groups.

Focusing on cerebral malaria and UM children the MS/MS data will be searched against the databases (UNIPROT and PlasmoDB,40 the proteins will be considered as positive hits with at least two peptides. The MaxQuant software will be used to compare malaria protein expression between isolates of these two clinical groups. Transcriptomical data will be analysed with Galaxy (https://usegalaxy.org/) and R software (https://www.r-project.org/).41 The raw data will be trimmed with Trimmomatic tool for Phred Quality Score Qscore >20, read length >30 bases and ribosome sequences will be removed with tool sortMeRNA. Reads will be mapped against the P. falciparum 3D7 reference genome combined with var transcript sequences from 7 P. falciparum genomes. Differential expression analysis on RNAseq data will be performed using the DESeq242 package considering a one log-fold increase as significant using adjusted p value <0.05. Data normalisation will be realised with DESeq2 software, with hypothesis that there exists genes overexpressed and underexpressed and that majority of genes are not expressed in a differential way. Transcript expression levels (evaluated with RT-qPCR) will be compared by t-tests and analysis of variance of transformed outcomes.

Regarding immune response analysis, potential markers related to inflammation and neurological symptoms will be compared using variance analysis in samples from children from the three groups, CM, UM and non-malarial non-traumatic group. The groups will be compared two by two with a linear regression, with a special attention to CM/UM comparison. Adjustment variables such as age, sex, ethnical group, time to hospital transfer, body temperature and co-morbidities will be taken into account in the model. The threshold for significance level will be 0.05, and a Bonferroni correction will be applied to take into account multiple testing. It will be further determined if a global comparison between the three groups will be made. Generally speaking, the non-malarial non-traumatic coma group will be used as a comparator to analyse specific effect of malaria in neuroinflammation development. The second major question to be answered to is, within the CM group, whether the changes of the inflammation markers between D0 (admission) and D3 are predictive of the outcome (survival/death). A logistical model (univariate then multivariate) will be used for this analysis. The same adjustment variables will be used as in the comparison between groups. The dependent variable will be the outcome survival/death.

The last model (also a logistical regression) will study the changes in inflammation markers between D3 and D21 in the survivors in order to determine if they are predictive of a favourable evolution. The dependent variable will be the outcome, here the discharge from the hospital without apparent sequelae.

Missing data are not expected to affect more than 10% of the records for the main factors that will be analysed. Should they be over 5%, an imputation method such as the minimum information for publication of quantitative real time pcr experiments (MIQE) method will be applied, as the errors can be considered at random.43 No proteomic analysis for immune marker will be done.

Patient and public involvement

From patients’ experience and preference, follow-up of children admitted with coma was scheduled in order to be able to detect neurological sequelae. The diagnostical workup proposed to all children included into the study, although not affordable to all patients in routine practice, met parent’s expectations on what heath facilities should provide to all patients. All patients were recruited in health facilities where they usually seek care, and to that respect patients were involved in their recruitment process. Finally, results will not be disseminated directly to study participants but through peer-reviewed scientific journal and conference presentations.

Ethics and dissemination

Safety considerations

Parents/guardians will be given an oral information by the physician or the nurse and an opportunity to ask question and refuse the protocol. Patient’s confidentiality will be ensured and anonymity guaranteed by anonymous coding given at the inclusion.

Dissemination

The main benefits of NeuroCM are targeted on people living in malaria endemic countries. The study results will be disseminated through a variety of instruments to ensure that a broad range of both specialists and non-specialists are informed and can properly benefit from the findings. First, through the direct consultations with the WHO’s TDR-MIM, Roll Back Malaria programme to reach the wider public health audience; through scientific meetings and peer-reviewed publications in scientifical or medical journals to reach the scientifical/medical/public health communities; through guidelines targeting the medical and paramedical staff for optimisation of severe malaria management, through booklets (eg, first aid procedures and adapted behaviour in case of emergency) elaborated and adapted to the population of Benin.

Discussion

CM is the most life-threatening form of malaria with high mortality rate in young children. Mortality related to malaria is still high in children population and accurate CM diagnosis remains challenging. Among CM surviving children, up to 25% have long-term neuro-cognitive deficits (visual/hearing/cognitive/language impairment/ataxia/hemiparesis/motor deficit…), and 10% show evidence of mental health disorders.44 As CM might be one of the more common causes of epilepsy in malaria-endemic regions, the burden of CM neurological sequelae may be largely underestimated, but difficult to estimate because diagnosis is challenging in malaria-endemic regions. Bacterial or viral central nervous system infections may occur in children with malarial infection; this may not only originate overdiagnosis of CM, but also may overlook potential bacterial and viral central nervous system infections.

Patients were included in different areas reflecting the healthcare system in Benin. UM patients could not be included in hospital centres such as the CHU-MEL (Cotonou) hospital, and Calavi’s hospital, because outpatients with UM rarely seek care in these centres. In 2014, a pilot study aimed to include UM patients in the Cotonou CHU-MEL, and highlighted the absence of UM cases in hospitals. However, patients from the Sô-Ava areas are referred to the main hospital centres when patients present severe malaria (or any severe illness that cannot be monitored and managed in dispensary). In 2016, we aimed to include patients suffering from cerebral malaria in the Sô-Ava, and realised that first, patients were directly sent to the main hospitals, and second, that it would not be ethical to include severe malaria cases in these health structures due to the facility itself. A multicentre study for UM cases inclusion, using the main patient’s origin from the corresponding hospital, would have been more even accurate. This represents a possible limitation of our study.

The expected impact of NeuroCM is of great magnitude. First, NeuroCM is an attempt to propose improvements for the diagnosis of CM. It will provide as far as possible, for the first time in West Africa, an identification of the causes of coma in the study area. Second, thanks to DNA, RNA and protein analyses, NeuroCM will allow to identify new parasite targets for a vaccine to prevent CM. Third, NeuroCM will provide data on the kinetics of appearance of inflammatory and pro-resolving molecular and cellular events in brain during CM. The role of endogenous mediators in neuroinflammation resolution during CM will be clarified, with emphasis on pro-oxidant components and lipid mediators. NeuroCM will also identify markers allowing the definition of an immunological state in the process of neuroinflammation resolution in CM patients. Our experimental murine model will allow the formulation of new hypothesis while proof of concept will be achieved through the correlation of our proposed targets with patient morbidity and mortality parameters. In the future, it may allow clinicians to better manage CM, with specific pro-resolving drugs for instance.

The final products of NeuroCM are expected to feed the pipeline of new therapeutical (immune intervention) and preventive (vaccine) strategies to improve CM outcome, as well as other diseases involving neuroinflammation.