In sub-Saharan Africa, severe malarial anaemia (SMA) is a frequent complication of Plasmodium falciparum infections in young children 1 and is one of the main causes of severe anaemia, with a case-fatality rate reaching 23% in malaria holoendemic areas 2 . Pathogenesis of SMA is not well understood, although destruction of the infected erythrocytes accompanied by clearance of uninfected erythrocytes, erythropoietic suppression and dyserythropoiesis, can all contribute to anaemia 3 4 . SMA is associated with elevated levels of myelo-suppressive cytokines, such as TNF, but this is not specific to the SMA syndrome, as children with cerebral malaria (CM) also have highly elevated TNF plasma levels 5 . Previous analysis showed that SMA can be distinguished from CM on the basis of an elevated ratio of TNF to its potent anti-inflammatory regulator IL-10 suggesting a central role for the TNF-IL-10 balance in SMA pathogenesis 5 6 . This is supported by the observation that anaemia is increased in IL-10 knockout mice infected with Plasmodium chabaudi 7 and reversed upon TNF neutralization 7 8 .

TNF alone and in concert with multiple other cytokines and chemokines is a potent inhibitor of haematopoietic stem cells 9 . Elevated levels of TNF in patients with chronic inflammation 10 , aplastic anaemia 11 or inherited anaemic disorders 12 have been associated with inhibition of erythropoiesis. Multiple underlying mechanisms have been reported, including the caspase-mediated cleavage of the major erythroid transcription factor GATA-1 13 , impairment of cell cycle progression 14 , and remodelling of the extracellular matrix within erythroid niches 15 . The high TNF/IL-10 ratio characteristic of SMA patients might reflect an insufficient production of IL-10 in SMA patients to prevent or counteract the inhibition of erythropoiesis and the increase of erythrophagocytosis induced by TNF and/or to mitigate other pro-inflammatory stimuli.

Monocytes and T cells are generally recognized as the main source of TNF and IL-10 in vivo. However, little is known about their activation status and their contribution to the TNF/IL-10 imbalance associated with SMA. Data on monocyte or T cell status in SMA patients are scarce, although indirect lines of evidence suggest activation of the monocyte/macrophage compartment in SMA patients 16 17 18 19 . To date, no published study has compared T cell and monocyte status in SMA and CM.

The work reported here sought at investigating whether children with SMA or CM have distinct TNF and IL-10 production capacities accounting for their different TNF/IL-10 ratios. To gain insight into the cytokine production capacity and the cellular subsets involved, Ghanaian children with non-overlapping SMA or CM syndromes were recruited. Their T cell and monocyte activation statuses were compared. In addition, their intrinsic TNF and IL-10 secretion capacity was investigated using whole-blood stimulation assays with LPS and PHA, used as surrogate of monocyte and T cell stimulant, respectively. Similar analyses were performed on children with uncomplicated malaria (UM), considered as a control group for acute malaria and asymptomatic children (AC) living in the same area, recruited as a reference group for population baseline. The data show that the low IL-10 level in SMA cannot be attributed to a defective IL-10 response capacity and point to a specific dysregulation. Moreover, the parameters investigated provide interesting novel insights into the distinct inflammatory statuses in SMA and CM.

The high TNF/IL-10 ratio observed in SMA suggests an imbalanced production of inflammatory cytokines that could contribute to anaemia 5 6 . Whether such an imbalance is an intrinsic characteristic of children with SMA or reflects a specific response pattern to malaria infection involving particular cellular sources has profound implications on the design of intervention strategies to prevent SMA. The data reported here show that SMA patients indeed displayed low spontaneous IL-10 production at admission resulting in higher TNF/IL-10 ratios than CM cases. These findings are consistent with previous studies in Ghana 5 21 , but also more recent studies in Southern Zambia 28 . Interestingly, in response to a monocyte or T cell stimulus IL-10 production readily increased in both CM and SMA patients, but SMA patients were characterized by a much higher amplitude of the IL-10 and TNF monocyte response to LPS compared to CM, possibly reflecting different monocyte priming status. However, the absolute levels of IL-10 reached after PHA-stimulation remained modest, much lower than for CM or any other group. This indicates that children experiencing SMA have no inherent incapacity to produce IL-10 and therefore that the imbalanced cytokine response at admission and upon further stimulation in vitro likely reflects a specific immunological pattern rather than an intrinsic predisposition to a deficient IL10-production.

The data also provide interesting insights into the immune status of children with CM. Although expression levels of surface activation markers on both lymphocytes and monocytes were similar in CM and SMA, CM patients presented a distinct cytokine expression profile, characterized by spontaneous production of high levels of both TNF and IL-10 but limited increase in TNF production after monocyte or T cell stimulation, suggesting an overall relative low-responsiveness to further stimulation. This points to a distinct functional status of circulating T cells and monocytes in SMA and CM, which both differed from the functional status in UM.

UM children had lower lymphocyte counts, limited monocyte deactivation, balanced IL-10 and TNF levels at admission (both lower than CM, IL-10 higher than SMA) and strong responsiveness to monocyte and T cell stimulation. Thus, based on analysis of circulating cells, the three clinical groups had specific response profiles to the ongoing infection and to further monocyte or T cell stimulation. IL-10 and TNF responses to a T cell stimulus were higher in AC than in any of the three clinical malaria groups, suggesting an impaired T cell responsiveness in malaria (regardless of the clinical presentation) as observed by others 29 . Parasite-related factors may explain the specific IL-10 production profile of SMA patients. Some studies 30 31 but not others 32 have found that haemozoin phagocytosis triggered the production of IL-10 by monocytes and induced a state of monocyte “anergy/reprogramming” associated with a deregulated production of pro-inflammatory cytokines such as TNF 33 34 . However, the lack of association of IL-10 plasma levels with the number of circulating haemozoin-containing monocytes observed here, including in CM patients who have the highest IL-10 levels, does not support a direct impact of haemozoin load on IL-10 production by circulating leukocytes.

There was no significant correlation between the number of pigmented monocytes and haemoglobin. This contrast with results from Casals-Pascual et al., although they found a moderately positive correlation (r² = 0,29) 35 . This reflects the unclear relationship between pigmented leukocytes and the disease manifestation or the parasite biomass. The number of circulating pigmented monocytes depends on a complex clearance kinetics 36 , which may differ depending on whether anaemia is consecutive to an acute infection or results from a protracted infection.

CD36-dependent adhesion of infected erythrocytes to monocytes may modulate the inflammatory cytokine secretion profile, including IL-10 production 37 38 39 and the low IL-10 plasma levels in SMA patients may reflect the low CD36-binding capacity of their infected erythrocytes 40 . This is supported by the lower proportion of haemozoin-containing monocytes in SMA relative to CM patients, possibly reflecting the reduced phagocytosis of infected erythrocyte subsequent to CD36 binding 41 but may also merely reflect differences in parasite biomass.

The discrepant IL-10 levels in SMA and CM could result from different types or proportions of IL-10 producing cells. Recent studies suggest that various subsets of CD4⁺ T cells including Tr1 and Th1 are important contributors 42 43 . Compared to CM and UM, SMA cases produced lower absolute levels of IL-10 but higher levels of TNF in response to T cell stimulation. This suggests a T cell functional impairment specific for IL-10 production in children with SMA. Whether this reflects an infection-related cytokine expression programming or an effector/regulatory T cell subset imbalance is unclear. Additional work is needed to elucidate this question, especially since depletion of CD4+ T cells significantly alleviates anaemia in a murine model 44 .

Beside T cells, two monocyte subpopulations with different IL-10 producing capacity upon LPS stimulation are now recognized: the regular CD14^brightCD16^-/dull monocytes producing both TNF and IL-10 and the CD14^dim CD16^bright monocytes producing high levels of TNF and little or no IL-10 45 . Although the latter subset was recently found to be enriched in SMA children 46 , comparable CD14/CD16 cytograms and monocyte CD14 MFI (CM: 234.1 ± 262.3; SMA: 162 ± 95) and CD16 MFI (CM: 10.8 ± 12.2; SMA: 13.9 ± 12.1) were observed here.

A significant but transient down-regulation of HLA-DR expression of circulating monocytes was observed in children with severe malaria, irrespective of the clinical form (SMA or CM). HLA-DR down-regulation has been described for dendritic cells in Kenyan children with acute malaria, but was observed in both mild and severe cases 47 . Phagocytosis of haemozoin and exposure to IL-10 both induce down-regulation of monocyte HLA-DR surface expression 25 48 . However, in the CM and SMA patients studied here, HLA-DR expression was independent from the number of pigmented monocytes and did not correlate with circulating IL-10 levels. The observed HLA-DR down-regulation rather results from the complex integration of multiple anti-inflammatory signals, as observed in severe inflammatory syndromes where it is generally associated with an impaired TNF production capacity in response to further LPS stimulation 49 reflecting a general cellular reprogramming phenomenon of acute inflammatory injuries 50 51 52 . The impaired TNF production after LPS stimulation observed in CM, but not in SMA, is reminiscent of this cellular reprogramming and suggests interference between the monocyte signalling pathway involved in the overwhelming cytokine production associated with CM and the LPS-triggered MD2 signalling pathway 53 54 . Thus, CM appears as an acute inflammatory syndrome with excessive TNF production by monocytes/macrophages rapidly inducing a high counter-regulatory IL-10 production. In contrast, the high TNF levels observed in SMA would result from a more chronic/sustained production of TNF maintained by an impaired IL-10 regulatory feedback reflecting a specific leukocyte polarization/programming state in SMA.

The data reported here point towards a specific programming of monocytes and T cells in SMA patients where low IL-10 levels are not due to intrinsic production incapacity but rather to a specific polarization/programming pathway. Differences in the environmental context as well as the intensity and duration of malaria-associated inflammatory stimuli may explain this specific cytokine response. Directions for future work include identification of the parasite factors implicated in the polarization of the immune response and ex vivo dynamic functional analysis of specific T cell subsets in children with CM or SMA. Dissecting the infection-acquired changes in cytokine expression profiles associated with SMA and exploring additional upstream and downstream mediators as well as cells possibly implicated in dyserythropoiesis or erythrophagocytosis is of major interest to design intervention strategies.

The authors have no conflict of interest. Jørgen Kurtzhals has received project funding for unrelated studies from Vifor Pharma, Switzerland and Novo Nordisk, Denmark.

BG, JK, BDA, LH and CB designed the study. GOA, BG and JK recruited the participants. PB, SL, GAA, JT and MMA generated the data. PB, OMP and CB analysed the data and wrote the manuscript. CB supervised the research and secured the funding. All authors approved the final version of the manuscript.