Prophylactic antibiotic treatment

Studies in animal models of ischaemic stroke have demonstrated that preventive treatment with antibiotics reduces the incidence of infections, and improves mortality and neurological function.9 Based on these encouraging findings, a series of clinical trials that tested the safety and efficacy of prophylactic use of antibiotics in patients with stroke have been conducted (table 1). Among these 15 clinical studies, patient inclusion, stroke types, antibiotics selection and treatment duration differ from each other. The majority of these studies treated patients with broad-spectrum antibiotics to cover the most common causative bacteria of pneumonia and urinary tract infections, except for three studies that tested the neuroprotective effects of minocycline, which has inadequate bacterial coverage for patients with stroke; the infection results were not reported in these studies as well.11–13 A meta-analysis of a portion of these studies concluded that preventive treatment with antibiotics could reduce infection rates, but failed to reduce mortality and improve functional outcomes.14 This observation resembles findings in two recently completed phase III trials.7 8 Results from the Preventive Antibiotics in Stroke Study (PASS), which included 2358 patients from 30 Dutch centres, show that preventive antibiotic treatment after stroke was able to reduce poststroke infections, but did not improve functional outcomes at 3 months.8 The other phase III trial, testing prophylactic antibiotics after acute stroke for reducing pneumonia in patients with dysphagia (STROKE-INF), enrolled 1217 patients from 48 stroke units in the UK; it reported that prophylactic antibiotics did not reduce the frequency of pneumonia within 14 days after stroke onset, either as defined by algorithm or diagnosed by a physician. The secondary end point analyses showed that prophylactic antibiotics did not improve functional recovery at 3 months or mortality.7

Although the PASS and STROKE-IFN studies are different in many aspects, such as treatment time window, antibiotics selection and patient inclusion criteria, the results of the two studies are highly consistent, that is, the treatment regimens of both studies failed to reduce the rate of pneumonia. Additionally, a recent study which tested antibiotic treatment in patients selected by the ultrasensitive procalcitonin test did not find a reduction of pneumonia in antibiotic-treated patients either.15 Therefore, since the antibiotic selection and treatment time window in these studies cannot prevent poststroke pneumonia, a possible explanation is that stroke-associated pneumonia might partly be a pneumonitis with more sterile components.16 In light of this speculation, the pathophysiological changes in the lung poststroke warrant further investigations for possible identification of new treatment strategies.

Although these clinical studies indicated that several antibiotics in different patients with stroke are safe, limitations regarding the selection of antibiotics for patients with stroke were also noted. For example, some antibiotics have neurotoxic effects and can worsen outcome. A preclinical study found that enrofloxacin (a fluoroquinolone antibiotic) treatment from day 1 to day 7 in rats undergoing middle cerebral artery occlusion, an ischaemic stroke model, worsened outcomes at 1 month.17 In this context, antibiotics that have neuroprotective properties should be considered in priority, like minocycline12 18 and ceftriaxone.19 However, the benefit of ceftriaxone was not demonstrated in PASS, which yielded to the secondary consideration of antibiotics choice, that is, whether broad-spectrum antibiotics are suitable for intervention in patients with stroke. Exposure to these broad-spectrum antibiotics may cause dysregulation of normal flora in the body, which might influence the stroke outcome, as suggested by a recent animal study in which broad-spectrum antibiotic treatment caused depletion of mice intestinal flora, worsening the outcome of stroke.20 Therefore, the selection of antibiotics for patients with stroke must comprehensively consider the benefits as well as secondary adverse effects of the treatment.

As a severe type of stroke, intracerebral haemorrhage (ICH) accounts for about 10%–15% of all stroke types and is associated with high mortality and morbidity.21 22 Infection is also a common and important complication in patients with ICH, and it can influence neurological function in the acute phase as well as long-term recovery.5 23 The reported infection rates in patients with ICH are similar to those with acute ischaemic stroke (AIS), about 30% of all patients, with higher rates in more severe patients admitted to an intensive care unit.24 25 Currently, studies that have specifically tested antibiotic treatment in patients with ICH are still lacking. However, four of the published studies mentioned above (table 1) also recruited patients diagnosed with ICH. Two studies included only 26 and 11 patients with ICH, respectively, and thus are not suitable for subgroup analysis. In PASS, 269 patients with ICH were enrolled; subgroup analysis indicated that preventive ceftriaxone does not improve outcome in these patients either, however, whether or not preventive ceftriaxone was able to reduce the rate of pneumonia post ICH was not reported. In the STROKE-IFN Study, 125 patients with ICH were included; subgroup analysis indicated that prophylactic antibiotic treatment might reduce the rate of pneumonia in patients with ICH within 14 days after stroke, a noteworthy finding that warrants further confirmation in future studies with larger sample sizes.

Overall, for patients with AIS, prophylactic antibiotic treatment can reduce the rates of both total and urinary tract infections, but no such benefit was seen for stroke-associated pneumonia. In addition, preventive antibiotic treatment does not improve the outcome in patients. Hence, these findings do not support the use of antibiotics in a preventive manner. Preclinical and clinical evidence for the use of antibiotics in patients with ICH requires further enlightenment.

Stroke-induced immunosuppression

The immune system and nervous system crosstalk with each other via multiple facets to maintain the homoeostasis of both systems under physiology conditions.26–30 Severe brain injuries, including stroke, can interrupt such balance and lead to a series of changes in both systems.31 Brain injury after an ischaemic or haemorrhagic stroke leads to the activation and infiltration of inflammatory cells into the brain.32–35 While poststroke inflammation may contribute to the clearance of tissue debris and tissue repair, most published literature indicates that inflammation in the brain during the acute phase of stroke promotes the expansion of stroke lesions and worsens neurodeficits.36–38 Further, inhibition of the immune response in the brain during the acute phase can limit the extent of stroke injury.33 34 38–42 Conversely, the injured brain can reshape peripheral immunity and transition the functional status of the peripheral immune system from competence to suppression after the acute phase of stroke, as manifested by lymphopenia, decreased levels of inflammatory cytokines, monocyte and lymphocyte dysfunction, and atrophy of secondary lymphoid organs.31 36 43 44 Brain lesion size has been considered as an independent risk factor of poststroke immunosuppression and infectious complications.45 It has been postulated that stroke-induced immunosuppression might be an adaptive response to acute brain injury, because systemic immunosuppression may limit the overwhelming inflammation in the brain or reduce the occurrence of autoimmune reactions against neuroantigens.46 47 However, a protective role of immunosuppression after stroke, if any, is uncertain and remains to be defined.

A consequence of immunosuppression has been linked to the increased risk of infection after stroke onset.47 An established concept is that brain-derived neurogenic innervations are in control of systemic immunity. In the context of stroke, brain injury-induced activation of neurogenic pathways, including the sympathetic innervation, hypothalamic-pituitary-adrenal (HPA) axis and parasympathetic innervation work together to influence the magnitude and intensity of systemic immune response.43 48 The central mediators of these pathways include norepinephrine, acetylcholine and glucocorticoid hormones.27 31 Receptors of these molecules are broadly expressed in immune cells.43 49 Activation of the sympathetic nervous system induces the release of catecholamines from sympathetic nerve terminals. Transient increase of catecholamine release often leads to increased blood immune cells by mobilising them from peripheral reservoirs such as the spleen; however, stroke induces a prolonged elevation of circulating catecholamines,44 which promotes the apoptosis of immune cells and leads to a decrease of peripheral immune cells and a bias towards T helper cell 2 (Th2) immune response.50 51 Glucocorticoids, key mediators for HPA axis function, have antiproliferative effects to induce apoptosis of immune cells.52 Additionally, glucocorticoids can promote the production of anti-inflammatory cytokines like transforming growth factor β (TGF-β),53 while inhibiting the production of proinflammatory cytokines, such as interleukin (IL) 1, IL-8 and umor necrosis factor α (TNF-α).54 The anti-inflammatory effects of cholinergic pathways mediated by acetylcholine was recently uncovered; macrophages can be quickly inhibited when exposed to acetylcholine, with decreased secretion of proinflammatory cytokines like TNF-α, IL-1β and IL-18.55 In addition to peripheral immunosuppression, stroke also suppresses immunity in the ischaemic brain. Ischaemic neurons can secrete several neurotransmitters like acetylcholine (Ach), glutamate, serotonin, and so on, which can directly interact with the receptors of peripheral infiltrated lymphocytes and modulate their functional status.43 A model of stroke-induced immunosuppression is  illustrated in figure 1 using natural killer (NK) cells as an example. The identification of the mechanisms underlying stroke-induced immunosuppression offers the possibility of the use of immune modulation as a novel approach to boost host defence and combat poststroke infections.

• Download figure
• Open in new tab
• Download powerpoint

Figure 1

Schematic diagram of mechanisms of stroke-induced immunosuppression with NK cells as an example. In the early stages of stroke (<24 hours), ischaemic neuron-recruited NK cells are swiftly mobilised into ischaemic areas, where they promote neuronal death (not shown in figure).34 Subsequently (>48 hours?), ischaemic neuron-derived signals can turn off NK cells that express neurotransmitter receptors. At the peripheral level, ischaemic brain injury influences the sympathetic, parasympathetic (vagus nerve) and/or hypothalamic-pituitary-adrenal (HPA) axis systems that suppress NK cell-mediated immunity. Differences in the spectrum of neurogenic innervations, immune cell subsets and soluble mediators in the CNS versus the periphery may differentially affect _NK cell deficiency in these two compartments. BBB, blood brain barrier; BDNF, brain-derived neurotrophic factor; CNS, central nervous system; EGF, epithelial growth factor; GABA, gamma-aminobutyric acid; NK, natural killer; UTP, uridine triphosphate.

Potential therapies to combat stroke-associated infection

Animal experiments demonstrated that pharmacological blockade of the sympathetic pathway or HPA axis by β-blockers or glucocorticoid receptor antagonists could reduce stroke-induced immune dysfunction and poststroke infection to improve animals’ functional outcome (table 2).48 56 However, the clinical relevance remains unclear (table 3). A small prospective study indicated that β-blocker use is associated with less severe stroke at presentation, with lower thrombin, haemoglobin A₁C levels, and erythrocyte sedimentation rate.57 Another retrospective analysis of 841 patients with ischaemic stroke, 10.6% of whom received β-blockers during hospitalisation, also implied that β-blockers might be neuroprotective, as the use of β-blockers was associated with reduced risk of early death; however, this association no longer exists when excluding mortality from cardiovascular causes.58 Of interest, a study that included 625 patients, among which 553 patients were admitted with AIS and 72 with ICH, reported that β-blocker exposure, defined as receiving β-blockers prior to stroke and continued treatment during hospitalisation, did not reduce the risk of stroke-associated pneumonia, but reduced urinary tract infection rates. In addition, patients receiving β-blocker therapy showed a higher 30-day mortality than those without exposure to β-blockers.59 However, a recent non-randomised study of 5212 patients with ischaemic stroke suggested that, either treatment with β-blockers before stroke onset or starting treatment after stroke onset was associated with reduced frequency of pneumonia, and β-blocker exposure during stroke was associated with reduced mortality.60 In PASS, 885 patients received β-blockers, but subgroup analysis of this factor was not reported and further analysis regarding the safety and efficacy of such therapy on stroke outcome, as well as in combination with preventive antibiotic treatment, is awaited. Thus, the safety of systemic β-blockade remains controversial in the context of patients with stroke, and randomised trials may be warranted for further determination. In particular, the adverse effects of systemic inhibition of sympathetic input by β-blockers might counteract its protective effect of immune modulation, which warrants further verification. Ultimately, developing selective blockers of lymphocytic β-adrenergic receptors or downstream signalling pathways might avoid the adverse systemic impacts.

Boosting peripheral immunity might serve as another viable approach to countering poststroke infections and avoiding the systemic effects of neurogenic innervations. Preclinical data have demonstrated that adoptive transfer T or NK cells, as well as injection of IFN-γ to animals during the subacute stage of ischaemic stroke (>24 hours) reduces the rate of pneumonia and improves the outcome.43 56 Additionally, treating animals subjected to brain ischaemia with α-galactosylceramide, a specific activator that can promote proinflammatory  cytokines production by the invariant natural killer T cell  (NKT) (iNKT) cell, is able to reduce stroke-associated infections.48 Currently, the safety and efficacy of this approach has not been tested in clinical trials. Nevertheless, before translation to the clinic, the contribution of different cell types and cytokines to restrict stroke-associated infections require further investigations.