Aspirin as a potential treatment in sepsis or acute respiratory distress syndrome

There is a close evolutionary association between the inflammatory cascade and the haemostatic system, and a single trigger mechanism of activation for both systems goes back 450 million years [21]. The common initiators for both pathways, such as endotoxin, highlight the complex interplay and overlap of these pathways. Historically, haemostasis formed a crucial part in the innate immune system where walling off of the pathogen was accomplished by the formation of fibrin, platelet and leucocyte clot which forms the basis of the “haemostatic containment” hypothesis [21]. Although this co-stimulation may have evolutionary advantages, severe sepsis is characterised by microvascular thrombosis which can contribute to multi-organ dysfunction (MOD) [22]. Interventions to reduce the haemostatic defect have been shown to improve organ function and reduce mortality in experimental models [23]. Similar changes in the pulmonary microcirculation are demonstrated in autopsy of lungs with ARDS [24]. An increase in pulmonary vascular dead space, which may reflect pulmonary microcirculation thrombosis, is associated with worse outcomes in patients with ARDS [25]. Platelet leucocyte interaction also enhances the production of inflammatory cytokines such as interleukin (IL)-1β, IL-8, monocyte chemotactic protein 1 and tumour necrosis factor alpha (TNFα), which propagates inflammation further.

Platelet activation by endotoxin and platelet-activating factors such as thrombin plays an important role in sepsis [26]. The sepsis and ARDS complications due to platelets are secondary to enhancement or dysregulation (or both) of their thrombotic and inflammatory actions [20, 27]. Once activated, the platelets alter shape, upregulate the expression of receptors like P-selectin, degranulate and aggregate [28]. This process promotes platelet adhesion with the endothelium, other platelets and leucocytes, leading to the formation and release of inflammatory and thrombotic agents, further leucocyte recruitment, oedema formation and production of neutrophil extracellular traps (NETs).

Intravascular NETs are protrusions of granulated chromatin with the purpose of capturing pathogens and result from the combination of activated neutrophils and platelets. Platelet interactions with the neutrophils are essential for their production as demonstrated by platelet depletion or disruption of the platelet neutrophil aggregation in mice [29]. Although the principle aim of NETs is entrapment of pathogens, over-production is associated with direct tissue and organ damage [30]. Furthermore, specifically in ARDS, the high concentration of pro-inflammatory factors in the alveoli can lead to excessive NET production, and the protrusions themselves can be a cause for direct mechanical injury to the lung tissue [30]. Recently, it has been reported that the NETs themselves can activate further platelets, promote fibrin deposition and act as supports for thrombosis formation, thus further perpetuating the inflammatory thrombotic process [31], ultimately resulting in MOD.

Platelets play a significant role in leucocyte recruitment, vascular permeability and resultant oedema formation. In a murine model of ARDS due to sepsis, platelet-depleted mice had reduced infiltration of neutrophils, reduced pulmonary oedema formation and better outcomes [32], which was felt to be secondary to the diminished leucocyte recruitment. Interestingly, platelet depletion resulted in significant reduction of pulmonary oedema in a transfusion-related model of acute lung injury while not influencing neutrophil migration [33]. Platelet depletion not only inhibited platelet activation and aggregation but also resulted in improved oxygenation, reduced pulmonary hypertension and less interstitial pulmonary oedema [34]. Furthermore, antagonising the effects of specific platelet-derived chemokines, namely CCL5 and CXCL4, reduces neutrophil migration, pulmonary oedema formation and tissue damage in the lungs [35].

These pre-clinical models highlight the importance of platelet activation in sepsis and ARDS and suggest that platelet depletion or inhibition of the platelet or platelet-specific chemokines can reduce platelet neutrophil aggregates and platelet sequestration and ultimately improve outcomes.

The evidence from pre-clinical research in sepsis is conflicting. In a murine model of Staphylococcus aureus septicaemia there was significantly increased bacterial burden, organ dysfunction and cytokine levels in platelet-depleted mice [36]. In addition, a Klebsiella pneumoniae-driven sepsis model, with significant platelet depletion, was associated with worse mortality and haemorrhage at the primary site of infection but with no influence on neutrophil recruitment to the lungs [37].

Activated platelets are found in significant quantities in the organs of patients with sepsis and septic shock [38]. A study in critically ill patients with sepsis concluded that this enhanced and uncontrolled adhesion of the platelets to leucocytes and the endothelium leads to their accumulation in the micro-circulatory system and eventual thrombosis formation contributing to MOD [39] in sepsis [22].

Platelets have been shown to accumulate in the lungs of patients with ARDS. In fact, platelet activation, migration and accumulation in the alveoli are major features of ARDS. Typically, bronchoalveolar lavage fluid from patients with ARDS has excessive concentrations of platelet-specific alpha granule proteins, suggesting high platelet activity [40], and following initial injury, leucocyte platelet aggregates form and increase dramatically in the alveolar tissue [41]. This enhanced and unregulated platelet activity leads to increasing leucocyte concentrations in the alveolar tissue and ultimately lung tissue damage. This was confirmed in lung biopsies from patients with diffuse alveolar damage which were found to have an exaggerated number of leucocytes within the small airways because of excessive platelet activation [42].

Initially used by the ancient Greeks in the form of willow leaf tea and later refined in Germany in the 19th century, by Felix Hoffman, aspirin has become one of the most commonly used drugs today [43]. Aspirin is a non-selective inhibitor of the enzyme cyclooxygenase (COX), has a half-life of approximately 20 minutes and is subject to significant first-pass metabolism, and most of its action occurs in the portal circulation of the liver [43]. Aspirin has previously been used in high doses for the treatment of rheumatic fever, but currently low-dose aspirin continues to be used in both primary and secondary prevention in cardiovascular medicine.

There are several mechanisms in which aspirin can manipulate the processes involved in both sepsis and ARDS (Fig. 1): 1) inhibition of COX [43]; 2) inhibition of nuclear factor kappa B (NFκB) [44]; 3) production of nitric oxide (NO) [45]; and 4) lipoxin production [46].

Aspirin has been shown to be effective in murine models of sepsis and ARDS. Mice injected with Salmonella enteritidis endotoxin were pre-treated with aspirin 30 minutes prior to infection at varying dosages. A significant 24-hour survival rate benefit was demonstrated with 3.75, 15 and 30 mg/kg of aspirin [61].

Zarbock et al. confirmed that platelet neutrophil aggregates are a significant feature in ARDS and demonstrated that inhibition of this aggregation with 1 mg/g aspirin reduced neutrophil recruitment and improved gas exchange and survival in a mouse model of ARDS induced by intra-tracheal hydrochloric acid [62]. In support of this, a two-hit model of ARDS induced in mice that were exposed to LPS for 24 hours, then injected with major histocompatibility complex 1 mAB and were either depleted of platelets or pre-treated with aspirin 100 μg/g intraperitoneally found that both interventions reduced lung injury and mortality [63]. It should be noted, however, that the dose used would equate to a relatively large therapeutic dose in humans. Interestingly, Grommes et al. [35] and Looney et al. [63] both found that blockage of the P selectin receptor or glycoprotein IIb/IIIa receptors on the platelet did not confer the same benefits, possibly suggesting that the outcomes are not simply related to platelet neutrophil aggregation but that aspirin itself may confer additional benefits.

Leucocyte infiltration and oedema formation in a cantharidin-generated dermal blister was assessed in healthy male subjects. Initially, the blisters were induced on a forearm, without aspirin, and assessed over the course of 72 hours for inflammatory response, cell count and period of resolution. In a crossover study, the same volunteers had blisters induced on the other forearm following a 10-day course of 75 mg aspirin. The study concluded that low-dose aspirin, through the production of ATL and NO, significantly reduced the accumulation of macrophages and neutrophils at the site of the blister but had no effect on oedema formation [49].

Supporting the pre-clinical studies, several observational studies have assessed a possible association with pre-hospital anti-platelet (the majority of which is aspirin) therapy and sepsis or ARDS. These are almost exclusively single-centre retrospective observational cohort studies and range from over 600,000 patients [64] to 22 patients [65] (Table 1). Patients admitted with community-acquired pneumonia on anti-platelet therapy have a lower admission rate to the ICU and shorter hospital stay [66]. In a general population of ICU admissions, those on anti-platelet therapy have a decreased mortality [66, 67] and have a decreased risk of developing ARDS [68‐70] and multi-organ failure [71]. Furthermore, in those ICU patients with septic shock or ARDS being treated with anti-platelet therapy, there was a reduction in mortality [64, 66, 72‐74]. Finally, in a prospective cohort study of patients admitted to a medical or surgical ICU investigating pre-hospital statin and aspirin therapy and the development of sepsis or ARDS and mortality in ICU patients, aspirin alone did not significantly affect the development of sepsis or ARDS, but there was a trend toward reduced mortality. In addition, the group prescribed the combination of aspirin and a statin had the lowest rate of sepsis or ARDS; however, this was not statistically significant [75].

Not all the observational studies found a correlation between anti-platelet therapy and outcomes in sepsis and ARDS. A multi-centre analysis of association between pre-hospital aspirin use and development of ARDS across the USA and Turkey found no significant association [76]. This study involved 20 American hospitals and two hospitals in Turkey; this imbalance makes it difficult to adjust for confounding factors and differences in treatment protocols. Also, a study looking at the incidence of ARDS in patients with post-aortic valve replacement concluded that there was no significant difference between those admitted on aspirin and those not; however, the population was small (22 patients) [65].

It is important to highlight the limitations of observational studies. Owing to the observational nature of the study design, there is no control over the administration, dose and compliance to treatment. There could be further confounding factors such as access to health care, in that patients prescribed the medication may have better management of their chronic conditions or the fact that medications including aspirin may be discontinued in the sickest of the critically ill cohort.

There is a paucity of randomised clinical trials investigating aspirin in sepsis or ARDS. There is one large randomised control trial examining the role of NSAIDs, specifically intravenous ibuprofen for 48 hours, in critically ill patients with sepsis. This study included 455 patients and concluded that ibuprofen reduced levels of pyrexia and tachycardia but did not prevent the development of septic shock or ARDS or improve mortality [77]. Because aspirin, unlike other NSAIDs, can exert COX-independent effects, additional studies evaluating the benefits of aspirin are needed.

Sepsis and ARDS are commonly associated with thrombocytopenia. It is seen in 20–40 % of critically ill patients [78], but the incidence varies depending on the cutoff used to define thrombocytopenia. In a study of patients with thrombocytopenia in the ICU, the risk of any type of bleeding increased from 4.1 % in patients without thrombocytopenia to approximately 53 % in patients with a platelet count of less than 100 × 10⁹/l [79]. Bleeding was described as major (life-threatening, compromising haemodynamic status or requiring urgent intervention), moderate (requiring non-urgent transfusions) or minor (all other events). There were 29 major, 21 moderate and 11 minor episodes of bleeding in this cohort. Thrombocytopenia was deemed as causative/contributory to 23 events out of these 61 events. There was no analysis comparing the severity of the bleeding with the platelet count. More recently, a retrospective analysis of thrombocytopenia in mixed ICU admissions concluded that there is no significant association between 28-day mortality and thrombocytopenia, defined as a platelet count of less than 150 × 10⁹, but the patients with thrombocytopenia had an increased incidence of major bleeding at 14.4 % compared with 3.7 % [80]. Major bleeding, however, was defined as intracranial or retroperitoneal or as any overt bleeding or a fall in haemoglobin greater than 2 g/dl. Although thrombocytopenia was defined, the incidence of major bleeding (or the individual components of this composite measure) is not reported according to the degree of severity of thrombocytopenia. Furthermore, in patients admitted to the ICU with community-acquired pneumonia, only a platelet count of less than 50 × 10⁹/l was associated with a significant increase in mortality.

The safety of aspirin in patients with thrombocytopenia has not been studied in critically ill patients in a prospective randomised controlled trial. However, in a retrospective cohort study of patients in a mixed ICU, there was a mortality benefit from anti-platelet drugs irrespective of their bleeding risk. Bleeding was defined as any incident described as ‘bleeding’ in the clinical notes [67]. In a non-ICU population, there was no increased risk of bleeding in cancer patients who received aspirin in the setting of acute coronary syndrome, even in patients with a platelet count of less than 100,000/μl [81]; in this study, the median platelet count in the thrombocytopenic group was 32 × 10⁹/l and the range was 4 to 100 × 10⁹/l. These studies suggest the benefit of aspirin even in patients with thrombocytopenia, but caution is needed given the limits of the observational design of these studies.

Several clinical trials are currently exploring aspirin both as a preventative agent and as a treatment in sepsis and ARDS. In Australia, the ANTISEPSIS trial (Aspirin to Inhibit Sepsis, ACTRN12613000349741) will assess whether 100 mg of aspirin daily for 5–7 years reduces severity of sepsis by preventing admissions to the hospital or ICU and improving mortality. In Brazil, Aspirin for the Treatment of Sepsis (NCT01784159) is a phase 2 trial investigating the effect of 200 mg of aspirin on organ dysfunction, Sequential Organ Failure Assessment score and duration of ventilation in patients with sepsis. The US Critical Illness and Injury Trials Group (USCIITG) is conducting a multi-centre, double-blind, randomised control trial testing the hypothesis that early treatment with aspirin will prevent ARDS (Lung Injury Prevention Study With Aspirin (LIPS-A)) [82]. Patients were randomly assigned to receive either placebo or an initial dose of 325 mg of aspirin followed by 7 days of 81 mg, with the primary outcome being the development of ARDS. The study has completed recruitment. Finally, in the UK, a randomised, double-blind, allocation-concealed, placebo-controlled phase 2 single-centre trial of aspirin as a treatment for ARDS (aSpirin as a Treatment for ARDS, STAR Trial, NCT02326350) has recently started. Patients, within 72 hours of a diagnosis of ARDS, will be randomly assigned to either aspirin 75 mg or placebo once daily for a maximum of 14 days. The primary outcome measure is oxygenation index at day 7.