To identify potential mediators of brain swelling associated with paediatric CM, plasma
metabolite levels of Malawian children with retinopathy positive CM were correlated
to brain volume. Associations were identified between PLA
2
metabolites, including arachidonic acid, with brain volume. Plasma PLA
2
enzymatic activity also correlated to brain volume, suggesting an upregulation of
the PLA
2
pathway in children with high brain swelling. TNF? can upregulate PLA
2
and this study reports that, in conjunction with other inflammatory cytokines, TNF?
is correlated with brain volume. L-urobilin, a haem degradation product and mannitol were also associated with brain
volume. Collectively, these results suggest a higher inflammatory state in children
with increased brain volume. Further studies on how these molecules may be involved
in BBB disruption, and why some children sustain a higher inflammatory state are now
warranted.
Brain swelling has recently been shown to be the strongest predictor of mortality
in paediatric CM, with an adjusted odds ratio of 7.5 (95 % CI 2.1–26.9) for severe
brain swelling among patients who died compared to those who survived 5]. Brain swelling induces increased intracranial pressure that may lead to brain-stem
compromise and respiratory arrest. The mechanism of increased brain volume is unknown.
Multiple mechanisms of increased brain volume may exist in clinical conditions such
as CM 30], including processes that impact BBB function 5], 7], 10]. This study set out to identify circulating biochemical mediators that may contribute
to BBB dysfunction. 7], 10].
Positive correlations between brain volume and several lipids that are PLA
2
metabolites were identified. The fatty acids released by PLA
2
, such as arachidonic acid, can be sources of energy, signaling, and of relevance
to this study, potent mediators of inflammation. Specifically, arachidonic acid, the
precursor of the eicosanoid pathway, can increase the permeability of human brain
microvasculature endothelial cell monolayers via prostaglandin E
2
activation of EP
3
and EP
4
receptors 31]. Lysophospholipids are also generated by PLA
2
and are important in cell signaling and membrane biology. PLA
2
enzymatic activity has been associated with neurological and inflammatory conditions
and inhibitors of the PLA
2
enzyme are being studied to reduce pathologic inflammation 32]–36]. Fatty acids and lysophospholipids can be generated by enzymes other than PLA
2
, therefore it was confirmed in this study that plasma PLA
2
activity was also correlated with brain swelling 37], 38]. These data extend a prior observation where high PLA
2
plasma activity was associated with severe malaria and death in Malawian children
39]. Why some children have higher PLA
2
enzyme activity is unknown. PLA
2
activity is tightly regulated by host responses including TNF? and reactive oxygen
species (ROS), both of which can be elevated during severe malaria 15], 40], 41]. Thus the PLA
2
pathway and its metabolites may be acting directly on brain microvasculature endothelial
cells or indirectly through their effects on cell signaling or energy metabolism.
L-urobilin, a degradation product of haem also correlated with brain volume. Malaria
induces erythrocyte lysis and subsequent release of haemoglobin into the circulation.
This results in increase of bloodstream ROS levels, which can alter BBB permeability
41]–44]. Lysed erythrocytes can increase BBB permeability and result in brain oedema, typically
occurring twenty-four hours after haemolysis 45]–47]. Direct measurement of ROS and other haemoglobin metabolites are needed to explore
the role of erythrocyte lysis in CM associated brain swelling.
Plasma mannitol was also found to have a positive correlation with brain swelling.
Exogenous mannitol has been associated with increased brain oedema or brain weight
in animal models of vasogenic oedema and brain infarct 50], 51]. Mannitol is used therapeutically for the treatment of increased intracranial pressure
following brain injury 48], 49]. Mannitol adjunctive therapy has been studied in CM and was found to either have
no significant effect on clinical outcomes in paediatric CM in Uganda 52] or to potentially have adverse outcomes by prolonging coma resolution time in adults
with CM 53]. Further studies to investigate the role of endogenous and exogenous mannitol in
CM paediatric brain swelling are necessary.
Inflammatory cytokines were also associated with brain volume and these may directly
alter BBB permeability. CCL2 and IL-8 have been previously associated with alterations
in tight junction proteins including occludin, ZO-1 and claudin-5 resulting in increased
brain endothelial cell permeability 54], 55]. In addition, TNF? has been shown to increase retinal endothelial cell permeability
through protein kinase C zeta (PKC?) and the NF?B pathway by reducing the expression
and distribution of claudin-5 and ZO-1 56]. Interestingly, inhibition of TNF? by soluble TNF p55 receptor attenuates status
epilepticus-induced oedema in a rat model, which could be relevant in CM, as seizures
are highly prevalent 3], 57]. Furthermore, TNF? can regulate the transcription of PLA
2
proteins via the NF?B pathway, providing a link between inflammatory cytokines and
the PLA
2
pathway 17]. The mechanism, where inflammatory cytokines, increased PLA
2
activity and its lipid metabolites converge to disrupt the BBB, in the setting of
parasites sequestration to brain endothelial cells in CM, requires further investigation.
If the PLA
2
pathway is found to play a direct role in BBB dysfunction, PLA
2
inhibitors could be evaluated as potential adjunctive therapy.
The higher inflammation and brain volume do not correlate with peripheral parasitaemia
or total parasite body load represented by parasite HRP2 levels 58]. The variability in inflammatory responses during infection could be attributed to
differences in disease duration, host or parasite genetic polymorphism, prior malaria
exposures, uric acid levels or other host or parasite factors 59]–65]. Further studies examining how host and/or parasite parameters are mediating inflammation
and upregulation of the PLA
2
pathway in the setting of increased brain volume could inform therapeutic intervention.
A limitation to the design of this study is that the plasma metabolome was examined
after onset of brain swelling. For this reason, it is difficult to know if the metabolic
profile is a cause of BBB dysfunction or the result of changes secondary to increased
brain volume. A longitudinal study design with repeated metabolic sampling over time
in correlation to brain swelling status would provide a more powerful approach to
identify potential mediators of BBB permeability. Additionally, there were no metabolomics
data from patients with brain volume scores 1, 2 and 8 due to the low prevalence of
those brain volume scores. To develop biomarkers of brain swelling for regions without
MRI, further plasma analyses that include all of the brain volume groups are warranted.
Given the variances observed, larger sample sizes would permit corrections for multiple
comparisons.