The association between malaria and non-typhoid Salmonella bacteraemia in children in sub-Saharan Africa: a literature review

Malaria

A number of studies have reported malaria and NTS bacteraemia in the same populations of children in sub-Saharan Africa [4–6, 13–17, 20, 21, 23, 26, 27, 62–66], although only a few of them [14–17], were specifically aimed to evaluate the effect of malaria on the risk of NTS bacteraemia (Table 1). In the majority of these studies that evaluated the effect of malaria on NTS bacteraemia [15–17], health facility controls were used, and could be subject to selection bias. Thus, comparisons of prevalence of parasitaemia between NTS cases with other predefined hospitalized patients may over or underestimate the effect of malaria on risk of NTS bacteraemia. In malaria endemic areas, the rate of NTS bacteraemia increases during the high transmission/rainy season, and it is believed that malaria probably contributes to the seasonality of NTS [13, 15, 17, 19, 21, 30, 62, 67]. A more precise description of the association between malaria and NTS is also made difficult by variation in how different studies have defined malaria exposure. Current malaria is usually defined as asexual stages of P. falciparum on a blood film together with compatible clinical symptoms, whilst recent malaria has been variably defined based on: the presence of malaria pigment (haemozoin) in blood leucocytes; the presence of gametocytes but not asexual parasites on the blood film; or a positive rapid diagnostic test (RDT) in the presence of a negative blood film. However, many studies have defined malaria exposure in a way that may represent a composite of these groups. Surprisingly few studies have explicitly reported the independent associations of current or recent malaria with NTS bacteraemia.

Only two studies that explicitly reported the association of current malaria with NTS bacteraemia in children were identified [13, 17]. A study in urban Malawi showed that those with current malaria infection (parasitaemia) were more likely to have NTS bacteraemia (about two times more) compared to those without malaria parasitaemia (Table 1) (RR =2.4, 95% CI 1.46-3.96) [13]. In this health facility-based study with a cross-sectional design, the controls were admitted children who did not have positive blood culture for NTS. Another study that evaluated the effect of current malaria infection in an urban area of The Gambia showed that patients with NTS bacteraemia had higher prevalence of malaria parasitaemia compared to other bacteraemic patients (X² = 9, p < 0.01) [17].

Similarly, four studies that explicitly reported an association with recent malaria were identified [6, 14–16]. In a study conducted in Tanzania, the proportion of those with a positive RDT and a negative blood film was higher in those with NTS bacteraemia (20/45, 44%) compared to 13/97 (13.4%) in those with other bacteraemia [6]. In a study conducted in rural Kenya, children with NTS bacteraemia in a health facility were compared to community controls in a matched case–control study [14]. Those with haemozoin (malaria pigment) visible on blood films were close to 17 times more likely to develop NTS bacteraemia compared to those without visible malaria pigment. In addition, HbAS genotype was associated with protection against bacteraemia, which was mediated by the known protection of HbAS against malaria. In the same report, which included a longitudinal study, there was a reduction in the incidence of NTS bacteraemia associated with a reduction in malaria infection [14]. Children with detectable P. falciparum histidine rich protein 2 (Pf HRP2) and negative blood film were about two to four times more likely to have NTS bacteraemia compared to those without any evidence of recent malaria [15, 16]. A health facility-based study in Tanzania showed that children with detectable Pf HRP2 and negative blood film were more likely to have NTS bacteraemia compared to those without (adjusted OR =4.13, 95% CI 2.66-6.44) [16]. Another study conducted in rural Kenya showed that children with detectable Pf HRP2 and negative blood film had a higher risk of NTS bacteraemia compared to those without (OR = 1.8, 95% CI 1.0-3.1) [15]. It should be noted that this study also compared children with NTS bacteraemia to those with other bacteraemia who were admitted in a health facility. A recent study in health facilities in rural Burkina Faso reported that the proportion of children with RDT-positive malaria was higher in those with NTS bacteraemia (81%) compared to those with other bacterial infections (31%) (p < 0.001) [68], although it was unclear whether these RDT-positive participants had a negative blood film. In studies that explicitly evaluated both current and recent malaria, the association with NTS bacteraemia was observed more [6], or solely [15], for recent malaria. This may indicate that the mechanism by which malaria causes susceptibility to NTS lags behind the acute infection, or it may simply reflect the time at risk, since a current malaria episode is limited relatively quickly by seeking treatment, whereas malaria antigens and haemozoin may persist for weeks to months, respectively.

Despite some limitations, the aforementioned studies have provided accumulating body of evidence that has been consistent with an association between symptomatic/recent malaria and NTS bacteraemia in sub-Saharan African children. Whilst there is evidence for the association between symptomatic malaria and NTS bacteraemia, few data on the association between asymptomatic malaria infections and NTS bacteraemia, were found. There is some evidence that among children with severe malaria, those with severe malarial anaemia have a higher risk of NTS bacteraemia [24], suggesting that this might be due to haemolysis. It should be noted that this is based on relatively few studies. There is lack of data on the risk of NTS in those with severe malaria compared to those with mild malaria.

Some studies have reported an association between malaria parasitaemia and any Salmonella bacteraemia (Typhoidal or NTS) [23, 69, 70]. One of the studies, which involved health facilities in a rural area in The Gambia, showed that those with haemozoin in their leucocytes were more likely to have invasive Salmonella infections compared to those without haemozoin in their leucocytes (RR = 4.05, 95% CI 1.15-14.42) [70].

Host defence against invasive NTS

Studying the pathogenesis of NTS bacteraemia in humans is difficult, but findings from animal models have suggested a plausible sequence of events from the arrival of NTS in the intestine to invasion, bacteraemia and dissemination to other organs. Invasive disease begins with invasion of epithelial and M cells of the intestine, facilitated by the proteins of the Salmonella Pathogenicity Island 1, Type Three Secretion System (T3SS1), which induce internalization of bacteria by massive rearrangement of the host cell membrane [51]. Bacteria then invade, or are phagocytosed by, macrophages and dendritic cells in the submucosa and mucosa-associated lymphoid tissues, where another set of proteins of the T3SS2 are activated to promote intracellular survival and replication [71, 72]. Bacteria begin to disseminate through the lymphatic system draining the intestine, and may also pass directly into the blood stream. These bacteria may be carried intracellularly in migrating cells, or become extracellular after pyroptosis (inflammatory cell death) of infected cells. Ultimately they reach the phagocytic cells of the reticuloendothelial system (liver, spleen and bone marrow) and establish new foci of infection from which they can disseminate [71]. Early containment of infection in the submucosa of the intestine is dependent on activation of the innate immune response and early production of interleukins (IL) IL-1β, IL-18, IL-23 and Tumour Necrosis Factor (TNF). This early response solicits local chemokine production, influx of monocytes and neutrophils and interferon-γ production by T-cells, which is critical for control of intracellular bacteria [73, 74]. Killing of bacteria at this early stage is particularly dependent on the oxidative burst in phagocytic cells [75], and it is likely that antibodies against NTS are also important in limiting this initial infection and dissemination of bacteria [74]. Cell-mediated immunity is likely to have a major contribution as infection proceeds, primarily by a Th1 response (most prominently IFN-γ) facilitating intracellular killing mechanisms [74, 76]. One important caveat, is that the commonly used model system, S. Typhimurium infection in the mouse, has been primarily used as a model of Typhoid [71], and may not represent all of the features of the most invasive ST313 strain in African children. However, in broad terms the conclusions about mechanisms of host defence against NTS are supported by observations of susceptibility to NTS in humans with a variety of primary and acquired immunodeficiencies including: innate immune defects affecting the IL-12/IL-23/Interferon-γ axis [77]; chronic granulomatous disease (in which phagocytic cells cannot mount an effective oxidative burst) [78]; MHC class II deficiency; treatment with TNF neutralizing antibodies (for inflammatory diseases such as Crohn’s disease and rheumatoid arthritis) [74]; and HIV (in which impaired mucosal integrity, CD4 lymphopaenia, immune dysregulation and abnormal antibody production are all implicated [1, 79, 80].

Modulation of host defence against NTS by malaria

Numerous mechanisms have been proposed to explain the susceptibility to NTS induced by malaria, including impairments of: mucosal barrier function; macrophage function; neutrophil function; and, antibody production. Plasmodium falciparum is particularly distinguished by its propensity for cyto-adherence of infected red blood cells to vascular endothelium, a phenomenon known as sequestration [53]. This is believed to permit the generation of very high parasite loads in the human body, by avoiding clearance in the spleen, and also to contribute to pathology by obstruction of flow in small blood vessels. Extensive sequestration of parasitized red blood cells in the microvasculature of the intestine is a common finding in post-mortem studies of fatal P. falciparum malaria [54]. Impaired microvasculature blood flow might directly lead to impaired mucosal barrier function, and reduce resistance to invasion by NTS [22, 81]. In addition, L-arginine deficiency which is induced by malaria, has been shown to increase intestinal permeability and bacterial translocation, and may exacerbate this effect [82].

Macrophage dysfunction has been one of the most attractive explanations for the susceptibility to NTS in malaria [17], because macrophages are one of the most important cells harbouring NTS in standard models of infection. A variety of reasons for macrophage dysfunction have been proposed including haemozoin ingestion [83], erythro‒/haemo‒phagocytosis [84] or reduced production of cytokines such as IL‒12 (which is necessary to facilitate the killing of intracellular NTS) [85]. Erythropoietin has recently been described to have a role in the impairment of resistance to NTS in mice [86]. Erythropoietin is the main regulator of erythropoiesis in bone marrow, but its receptors are also expressed on other cell types and appear, amongst other effects, to regulate the inflammatory function of macrophages [87]. Consistent with this, erythropoietin levels are generally elevated in severe malarial anaemia, the major risk factor for NTS, as would be expected as part of the homeostatic response to severe anaemia [88]. However, an important caveat is that it is unclear to what extent the standard model of pathogenesis applies during malaria-NTS co-infection, and particularly whether macrophages are the main cell type harbouring NTS. As detailed below, neutrophils appear to provide a new niche for bacterial replication during co-infection and may be more important than macrophages in this respect.

Relatively little attention has focused on the role of humoral immunity and the effect of malaria on the natural acquisition of antibodies against NTS. Antibodies play a clear role in protection against NTS bacteraemia, and are usually acquired during the second year of life in African children [76]. Malaria can suppress the generation of heterologous antibody responses, and recurrent episodes of malaria during early childhood might suppress natural acquisition of antibodies to NTS, just as they suppress antibody responses to Salmonella capsular polysaccharide vaccine [89]. Lack of antibody may also reduce the efficiency of neutrophil-mediated oxidative killing of NTS [90]. Disruption of splenic architecture during malaria is one likely mechanism for humoral immune dysfunction [89], but may also directly impair the splenic clearance of NTS [91].

Despite these many possibilities, one of the most consistent and compelling observations to explain the relationship between malaria and susceptibility to NTS is the causal role of haemolysis. Apart from malaria, several other haemolytic diseases are associated with susceptibility to NTS, namely sickle cell disease and Carrion’s disease (acute Bartonellosis) [92, 93]. Consistent with this, any cause of haemolysis (including malaria infection) will induce susceptibility to NTS in mice [94–96]. Interestingly, in this model, bacteraemia is a prominent feature, as seen in human NTS-malaria co-infections, and replicating bacteria are particularly found concentrated inside neutrophils in the blood [97]. The mechanism underlying this susceptibility was shown to be an indirect consequence of host mechanisms that promote survival in malaria. Haemolysis can be extremely damaging if its extent overwhelms host mechanisms to prevent liberation of free haem from haemoglobin [98, 99]. Haem oxygenase-1 (HO-1) is the inducible haem-degrading enzyme, which protects against haem-mediated toxicity and is essential for survival in malaria in mice. In addition to protecting against haem toxicity, HO-1 and the haem-degradation products, carbon monoxide and biliverdin, all have diverse immunomodulatory and cytoprotective effects. Amongst these effects, HO-1 induction has been associated with suppression of oxidative burst activity, which could benefit the host by preventing oxidative cytotoxicity in severe malaria, but might be disadvantageous when combating NTS. Indeed, malaria was shown to cause HO-1 dependent impairment of the neutrophil oxidative burst in mice, and this was apparent during neutrophil maturation in the bone marrow [97]. However, susceptibility was also dependent on haem, possibly as a source of iron for the siderophilic NTS, and could be reversed by competitive inhibition of HO-1. A similar defect in neutrophil oxidative burst activity was subsequently demonstrated in Gambian children with P. falciparum malaria, which correlated with the extent of haemolysis and HO-1 induction, and interestingly persisted for weeks beyond the clearance of acute malaria infection [100]. No studies have yet been performed to investigate prospectively whether the extent of impairment of neutrophil oxidative burst function associated with malaria correlates with risk of subsequent invasive NTS infection.

Clinical presentation

There is much literature on the clinical presentation of bacteraemia but little specifically on NTS and malaria co-infection, which is the subject of this review. Generally, the signs of NTS bacteraemia are clinically difficult to distinguish from other causes of bacteraemia in children [21, 101], characterised by non-focal sepsis. Nevertheless, data on the clinical presentation of NTS bacteraemia in children indicate that apart from the usual signs of sepsis, the following may be prominent: anaemia, splenomegaly, respiratory, and gastrointestinal signs [6, 13, 15, 21, 101]. Only one study that compared clinical features in children with co-infection with those in children with malaria alone, was identified [6]. Anaemia, particularly severe anaemia, tends to occur more frequently in those with NTS bacteraemia compared to those with malaria, or those without bacteraemia, or those with other pathogenic bacteria [6, 13, 15, 21]. A recent review of 25 studies across 11 countries in sub-Saharan Africa confirmed that a higher proportion of children with severe malaria anaemia had NTS bacteraemia [102]. Splenomegaly is more frequent in children with NTS bacteraemia compared to non-bacteraemic or other causes of bacteraemia [13, 15]. Other clinical features such as fever, jaundice, hypoglycaemia, malnutrition, and diarrhoea have also been reported to be more frequent in NTS bacteraemia compared to those with malaria only or other causes of bacteraemia [6, 27]. Respiratory features have been described both clinically and radiologically in conjunction with NTS bacteraemia, and include signs of lower respiratory tract infection and effusion or consolidation [15, 27], however whether these are caused by NTS itself or another co-infecting pathogen remains to be determined. White blood cell counts appear to be higher in children with NTS compared to others [27].

Diagnosis

Blood culture is the most widely used method for diagnosis of NTS bacteraemia in most studies, although PCR methods have been developed and may prove useful in future studies [103]. The sensitivity of blood culture for NTS in sub-Saharan African children was not found in the published literature, but a study in HIV-infected Malawian adults suggested that the viable bacterial load in blood is low (about 1 CFU/mL), and so small volume blood culture samples that are often taken in children may have limited sensitivity [80]. There are no similar data in humans with malaria and NTS co-infection, but in mice with co-infection, bacteria were found to be particularly concentrated in blood [97].

Prevention

The most pragmatic way to prevent malaria and NTS co-infection may be to prevent malaria. There is abundant evidence that conventional strategies to reduce malaria transmission result in reductions in both malarial and non-malarial morbidity and mortality [106–108]. Furthermore, reductions in malaria transmission (at least partly due to control measures), in at least three distinct epidemiological settings, have been associated with reductions in the incidence of NTS bacteraemia [14, 18, 19]. However, most of this evidence is from observational studies, and to date there have been no controlled trials to confirm that reducing malaria transmission is an effective method to reduce NTS bacteraemia.

To prevent NTS bacteraemia in children who already have malaria, targeted antibiotic prophylaxis for children at high risk of subsequent NTS infection, for example those with severe anaemia, may represent an effective strategy. Another intriguing possibility is inhibition of haem oxygenase activity with tin protoporphyrin [97, 109], and which might be administered after treatment for malaria in order to reverse haem oxygenase-1-mediated neutrophil dysfunction. Neither of these has yet been evaluated in a clinical trial, but they may not be difficult to combine with existing malaria treatment strategies.

Management

Prior to the availability of antibiotics, it was reported that some adults with malaria-NTS co-infection were cured with quinine alone, implying that treatment of malaria was sufficient to allow the host response to clear the NTS bacteraemia [12]. Such a management strategy would no longer be acceptable, and combined anti-malarial and antibiotic treatment would be essential. Malaria can usually be diagnosed rapidly (by blood film or RDT), but diagnosis of NTS bacteraemia requires at least 24–48 hours for growth in blood cultures (which are unavailable in many resource-poor settings), and so a major problem is deciding which patients with malaria should receive empirical antibiotic therapy [4]. Recent work by Hendrikssen et al. indicates that a quantitative measure of PfHRP2 – which is already the basis of many RDTs – might help to identify those most at risk of having co-incident bacteraemia [116]. Children appearing to have severe malaria, with either very high or very low levels of PfHRP2 had the highest risk of bacteraemia, and might be targeted for empirical antibiotics. Arising from this is a second problem – what empirical antibiotic is most appropriate? Although NTS is often reported as the most common bacterial pathogen causing co-infections in children with malaria, other bacteria also need to be considered, including: Streptococcus pneumoniae, Escherichia coli, Haemophilus influenzae, Acinetobacter species and Pseudomonas aeruginosa[22].

However, antimicrobial resistance among NTS isolates is an increasing problem worldwide, particularly with spread of the ST313 clone of S. Typhimurium. Multidrug resistance to commonly used antimicrobials, such as ampicillin, chloramphenicol and cotrimoxazole, has been reported to be common in a variety of African settings [26, 27, 30, 34, 117, 118]. Selections of empirical antibiotic treatment should be made based on local epidemiology and resistance data wherever possible, but it is clear that many of the older (cheaper) antimicrobials may no longer be adequate, and treatment with a third generation cephalosporin, or combination therapy including ciprofloxacin, may be necessary to give a broad enough coverage and reasonable chance of cure in children with malaria and suspected bacterial co-infection [1, 4].