Blood platelets and sepsis pathophysiology: A new therapeutic prospect in critical ill patients?

Recent definition of sepsis [24] emphasizes the non-homoeostatic host response to infection that drives life-threatening organ dysfunctions. Activation of innate immune responses in sepsis realizes a systemic inflammatory condition. The inflammatory phase is characterized by the production of pro-inflammatory mediators and immune cell activation [25‐29], and sepsis prognosis is linked to the magnitude and duration of this inflammatory response, high circulating cytokine levels being, for example, associated with poor outcome [30‐32]. The triggering of innate immune responses by pathogens and pathogen-associated molecular patterns (PAMPs) has been identified as an early and primary mechanism [2, 31, 33‐36]. Interestingly, mechanisms of non-septic systemic-associated inflammatory response syndrome (SIRS) as met in major surgery, severe trauma, extensive burns or pancreatitis may share common features with sepsis-associated SIRS, taking the form of a comparable early inflammatory storm that is triggered by alarmins released by damaged tissues [37]. However, the role played by this hyper-inflammatory phase in the progression of sepsis and its prognostic is to be understood in the context of an accompanying anti-inflammatory response and immunosuppression state, and much effort is made in elaborating a coherent vision of these opposite and complex events [38‐43].

Platelets are small (2–4 μm), anucleate, discoid-shaped cytoplasmic fragments released in the bloodstream during the fragmentation of polyploid megakaryocytes in bone marrow sinusoidal blood vessels [44]. In humans, a regulated steady platelet supply and clearance maintains numbers of 150,000–400,000 platelets per microlitre of blood. Platelet production is critically dependent on thrombopoietin (TPO) that acts for an important part on megakaryocyte progenitor proliferation/differentiation and on megakaryocyte maturation [45]. Platelets have a short lifespan, of up to 10 days. They are cleared out from the circulation by mechanisms involving lectin–carbohydrate recognition by splenic and liver macrophages and hepatocytes [46, 47].

Platelets harbour a large variety of mediators stored in a pool of morphologically distinct granules [48]. Granule cargo loading is carried out in megakaryocytes. Platelets also transport mediators, such as serotonin, that they uptake from plasma and can deliver at sites of activation. The cataloguing of platelet-derived mediators reflects the remarkable versatility of platelets in haemostasis, thrombosis and immune responses [49, 50].

The secretion of granule content following platelet activation by agonists is central to platelet functions. Platelet activation induces the expression of membrane proteins and the release of mediators via several mechanisms. Many of these mediators are preformed and stored in granules such as cytokines/chemokines and coagulation factors, others can be synthesized by translational pathways, such as IL-1β, and others are released by yet incompletely defined mechanisms such as CD154. Activated platelets also release vesicles, which include platelet microparticles (PMPs) and exosomes [51]. Platelets represent a major source of circulating MPs [52].

In pathological conditions associated with platelet activation, multiple agonists are generated. In fact, apart from classical strong agonists such as thrombin or collagen, there is an expanding list of agonists that can contribute to platelet activation. These additional platelet agonists have allowed a re-appreciation of mechanisms and role of platelet activation in vascular inflammation and thrombotic events associated with a range of infectious and inflammatory conditions [53].

The archetypal function of platelets is haemostasis. Platelets encounter inhibitory signals that prevent their activation in the healthy vasculature, such as nitric oxide and prostacyclin, which are released by endothelial cells (ECs). Platelets circulate in close proximity to the vessel wall, and the disruption of EC lining overcomes inhibitory signals and drives platelet adherence, activation and aggregation, which temporarily plug the damaged vessel. In this process, platelets also activate and confine coagulation at site of damage, particularly via the exposure of an efficient catalytic phospholipidic surface [54].

Besides binding to damaged vessels and preventing bleeding, platelets support a large spectrum of more recently studied functions, as could be reflected by the diversity of platelet mediators [55‐57]. Platelets are activated in conditions that disrupt tissue homoeostasis and exert, directly and indirectly, a complex control over the different stages of inflammation, contributing to pathogen clearance, wound repair and tissue regeneration (Figs. 1, 2). As such, platelets are now acknowledged as essential components of the innate immune response, monitoring and rapidly responding to noxious signals.

Activated platelets secrete a profusion of pro-inflammatory material, cytokines/chemokines, vasoactive amines, eicosanoids, and components of proteolytic cascades that directly or indirectly, through the activation of bystander target cells, fuel inflammation [23, 58, 59]. ECs and leucocytes are prime targets for platelets. Endothelium is a non-adhesive, non-thrombogenic surface in normal conditions; when stimulated by inflammatory mediators, ECs undergo profound changes, collectively designed as “EC activation”, which include the expression of cell adhesion molecules and tissue factor, production of von Willebrand factor, cytokines/chemokines, proteases and vasoactive substances such as nitric oxide. Platelets adhere to activated ECs, following a multi-step process in which glycans play a critical role [60‐62]. Inflammation can also alter the protective EC glycocalyx barrier, favouring platelet adhesion [63, 64]. During the adhesion process, platelets can be activated and in turn activate ECs. Platelet activation in inflammation can alter the vascular tone and lead to deleterious effects on vasculature integrity, by increasing vascular barrier permeability and contributing to the generation of cytopathic signals, for example by mediating reactive oxygen species generation by neutrophils [65]; these effects have to be paralleled with the opposed protective role of platelets (below) [66‐69]. Leucocytes are a second critical target for platelets, the platelet/leucocyte dialogue being essential in inflammation; here, we focus on neutrophils and monocytes. Platelet/leucocyte interactions are a critical step in leucocyte recruitment, activation and migration in inflammation [70]. Platelet/neutrophil or platelet/monocyte interactions can occur at the EC surface, in clot/thrombi and in circulating blood [18, 70, 71], and platelets direct neutrophil/monocyte migration to sites of tissue injury [72, 73]. Moreover, platelets activate neutrophils and monocytes upon interaction, via several mechanisms, including the triggering of TREM-1 on neutrophils, leading to various pro-inflammatory responses [65, 74‐77]. The formation of platelet/leucocyte aggregates in blood depends on platelet activation and is an early phenomenon in sepsis progression. For example, platelet/neutrophils complexes are elevated at early phases, while being reduced in complicated sepsis possibly reflecting peripheral sequestration or sepsis-associated thrombocytopenia [78, 79], and endotoxin administration in humans leads to an increased circulating platelet/neutrophil aggregates that follows a brief decrease [80]. Amplification of inflammation results from the reciprocal activation between platelets and their target cells [66], and circulating monocyte/and neutrophil/platelet aggregates may contribute to disseminate inflammatory signals [81]. Platelets also link several inflammatory cascades; for example, they propagate the activation of the complement system [82]. Commonly, cytokines have an induced expression that is regulated at the transcriptional/translational level. Most of platelet-derived inflammatory mediators are very rapidly released from activated platelets, making platelets instant providers of pro-inflammatory material. Cytokine bioactivity at organs remote from their source is debated as cytokine bioactivity may be hampered in plasma. Platelet transport may protect inflammatory mediators from otherwise degradation. Therefore, platelets play a central role in the inflammatory reaction. Importantly, they also contribute to the control and resolution of inflammation via several mechanisms, including the release of anti-inflammatory cytokines and inflammation pro-resolving mediators [83].

The activation of coagulation and inflammation cascades are consequences of platelet activation, and inflammation and coagulation pathways crosstalk [84]. For example, some platelet mediators have both inflammatory and pro-coagulant properties, such as polyphosphates [85]. Pro-inflammatory cytokines released by platelets can also activate the coagulation cascade at various steps [86]. Conversely, the activation of coagulation by platelets generates a number of inflammatory effectors, such as thrombin. Further, inflammatory mediators can impair anticoagulant and fibrinolysis pathway mechanisms, which may contribute to coagulation dysregulation in sepsis [87‐89]. Platelet inflammatory mediators may thus contribute to sepsis coagulopathy [88‐91]. DIC is a frequent and major complication of sepsis [41], and various mechanisms concur to involve platelets in DIC; only some can be mentioned. First, platelets support the generation of thrombin. Second, platelet links inflammation and coagulation. Third, platelets are major inducers of the release of pro-thrombotic scaffolds neutrophil extracellular traps (NETs) [92‐96].

Notwithstanding, the involvement of PMPs in vascular inflammation and inflammatory disorders, including sepsis, has been emphasized [97‐102]. PMPs retain many pro-inflammatory and pro-coagulant features of parent platelets and are thought to disseminate inflammatory and coagulation signals. Although they represent potential pathophysiological players in inflammatory disorders [52, 99, 100, 103, 104], their role in sepsis remains ill-understood.

In normal wound healing, platelets establish regulatory crosstalks between soluble and cellular actors of tissue repair that concur to the various phases of inflammation and reestablishment of tissue homoeostasis [50, 83, 104, 105]. Platelets accumulate early, are activated at sites of tissue injury and intervene at each stage of tissue repair, the inflammatory, new tissue formation and remodelling stages. In fact, platelet-healing properties are already translated to the clinics [50, 55]. The best studied role of platelets in tissue homoeostasis is the preservation of resting and injured endothelium integrity, a critical point in MOD pathophysiology [71, 106‐108]. The importance of platelets is exemplified by the disruption of the endothelium barrier associated with thrombocytopenia [109]. How platelets contribute is incompletely understood. Mechanisms include gap filling, production of EC mitogenic factors and factors enhancing the vascular barrier [71, 107]. On injured endothelium, platelets adhere to the vascular wall at sites of damage and immediate proximity, a first step in a sequence of events that lead to the initiation and the propagation of haemostasis, thrombosis and bleeding arrest [110, 111]. Platelets provide material for endothelium repair, including EC growth-promoting, antiapoptotic mediators, and attractants for progenitor cells endowed with vascular healing properties [104]. They help restoring the disrupted vascular network, providing positive and negative regulators of angiogenesis and stimulating angiogenic mediator production by target cells. Platelets are also important contributors to extracellular matrix (ECM) repair as they are a rich source of ECM components, ECM remodelling proteins, and fibro-competent cell activators. Platelets have however been found to both promote and prevent vascular permeability in inflammation. The differential regulation of vascular permeability by platelets has been studied for a large part in acute lung injury (ALI) models and will be presented in the corresponding section. Importantly, platelets are highly efficient at preventing bleeding in an inflammatory context [76, 107, 112]. The platelet count threshold needed for vasculoprotection in humans in normal or inflammatory conditions is an important question that remains to be answered [107]. More generally, platelets may play a broader role in organ regeneration. Platelets not only prevent blood loss but also provide key signals for matrix architecture reconstruction and for the recruitment, proliferation, survival and differentiation of cells endowed with new tissue formation, such as fibroblasts, smooth muscle cells and tissue-specific progenitors cells [113]. This is remarkably illustrated by the requirement of platelets in liver regeneration [114]. PMPs are also thought to contribute to vascular repair [115]. Hence, platelet activation is both necessary to tissue integrity and undesirable as it generates tissue-damaging signals. The complex network of signals that organize this fine-tuned equilibrium is only recently being biochemically dissected [55, 104]. Many questions remain on this balanced platelet friend or foe contribution, although they are of key importance to the pathophysiology of microvascular dysfunctions, such as in sepsis [68].

The role of platelets in the defence against infection is increasingly stressed [83, 116]. Platelets are now acknowledged as bona fide pathogen sensors interacting directly or indirectly with a number of bacterial, viral, fungal and protozoan pathogens and their products, contributing to their clearance. Platelet interaction with bacteria depends on the nature and concentration of bacteria, interaction time, and involves multiple mechanisms. Toll like receptors-dependent and independent mechanisms, such as those involving Fcγ receptors, complement receptors or glycoproteins GPIIb-IIIa and GPIbα, contribute to platelet–bacteria interactions. Indirect interactions are also involved, such as via the binding of plasma proteins, including fibrinogen, von Willebrand factor, complement proteins or IgG, that bridge pathogens and platelets or via interaction with bacterial toxins. Interaction with pathogens can lead to platelet adhesion or to their activation, aggregation and release of platelet mediators [83, 117‐120]. Mechanisms of pathogen clearance by platelets may be direct, through the release of various antimicrobial peptides and indirect via the release of platelet-derived mediators that coordinate chemotaxis and activation of immune cells [83, 116‐118, 120, 121]. Infection is commonly associated with tissue injury. Injured and dying cells generate mediators such as alarmins that fuel inflammation [122]. Mediators generated by cell damage such as complement activation products and histones can activate platelets [123, 124]. Notwithstanding, platelets also contribute to the adaptive immune response to infection [17, 18, 22, 125].

The aforementioned role of platelets in the defence against pathogens suggests that they can interfere with the progression of infection. How can these observations be translated to sepsis pathophysiology [120, 124, 126]? Models suggest a protective role of platelets, as, for example, in streptococcal endocarditis, malaria or gram-negative pneumonia [127‐129], and thrombocytopenia could be a risk factor for bacterial or fungal infection. Alternatively, platelets could contribute to spreading infection, via the transport of pathogens [130].

The clinical management of sepsis remains a difficult challenge, and pathophysiological advances have not yet been translated into effective therapeutic protocols [2]. Notably, strategies to counteract the runaway pro-inflammatory state in sepsis, such as inhibition of specific inflammatory mediators, have given disappointing results [243]. However, current knowledge on sepsis pathophysiology, highlighting multiple humoral and cellular factors in the inappropriate inflammatory response to infection, suggests that therapies targeting a single mediator will not demonstrate effectiveness [41]. Additional complexity is linked to individual disease susceptibilities and medical comorbidities that would necessitate individual approaches. Accumulating evidence therefore speaks for an integrated approach of sepsis treatment based on a better knowledge of its natural history.

The recently described involvement of platelets at the crossroads of several immune response pathways has led to the assumption that platelets or platelet-derived effectors represent therapeutic targets in sepsis. Platelet activation can drive multiple inflammatory and coagulation pathways, and targeting platelets offer the theoretical perspective of targeting simultaneously several deleterious pathways. Although the clinical relevance of animal models has many drawbacks, it is of interest that platelet depletion, inhibition of platelet functions and antiplatelet drugs show protection in experimental ALI or AKI [124]. P2Y12 inhibitors reduce inflammatory and pro-thrombotic mechanisms after endotoxin administration in humans [244]. Several observational and retrospective clinical studies have shown that antiplatelet agents such as acetylsalicylic acid, platelet P2Y12 inhibitor clopidogrel or GPIIb/IIIa antagonists reduce mortality or complications in critically ill patients [245‐255]. However, some studies are conflicting [249, 252, 256] (Table 1). There is therefore a strong need for large randomized controlled clinical trials to investigate the effects of antiplatelet therapy in sepsis. The complexity of such studies relates in part to the heterogeneity of sepsis patients in terms of nature of the causal germ, site and severity of infection, multiple comorbidities, gender, age and genetic background. There is also individual variability in the concentration of antiplatelet agents that efficiently inhibits platelet function. A defective response to clopidogrel or aspirin treatment may concern up to 30 or 40% individuals, respectively [257‐259]. Also, antiplatelet treatments have differential effects on platelet functions. Platelets treated with aspirin can still be activated by strong agonists, such as thrombin or ADP. Hence, in a full-blown pro-inflammatory/pro-coagulant condition as met in sepsis, it remains to be determined whether platelet activation is efficiently inhibited by antiplatelet treatments. Platelets are an important blood reservoir of pro-inflammatory molecules and may contribute to the “cytokine storm” that characterizes sepsis. However, many cellular players, including leucocytes and EC, also produce such mediators, and the relative contribution of platelets is not understood. In a recent study, antiplatelet therapy did not significantly reduce plasma pro-inflammatory cytokines levels in sepsis patients [260]. Antiplatelet agents have also been shown to have indirect off-platelet effects, a mechanism which importance is not yet established [261]. Finally, the impairment of platelet function may have undesirable consequences, such as bleeding or the blunting of platelet protective functions.

As mentioned above, elucidating mechanisms of thrombocytopenia in sepsis are essential with reference to transfusion. Platelet transfusion is mostly used to prevent/treat bleeding [228, 229]. The risk of bleeding increases with the severity of thrombocytopenia [222]. The threshold of platelet count ensuring protection may be higher in sepsis patients, reflecting the severity of the disturbance of the vascular beds. Commonly advocated threshold of platelet count is in the range of 10–50 × 10⁹/L, depending on clinical situations, additional bleeding risks, evidence for central thrombocytopenia. The risk of bleeding is, however, not straightforwardly linked to the depth of thrombocytopenia, in the context of a sustained production of platelets, and additional parameters in the critically ill patient may interfere; indeed, the risk of bleeding is also increased for platelet counts between 50 and 100 × 10⁹/L [8, 9, 222, 229]. In fact, there is a poor evidence-based clinical benefit of platelet transfusion in the non-bleeding ICU patient [154, 228‐230, 242]. The lack of a clear understanding of thrombocytopenia causes makes the risk/benefit assessment difficult, as there is a theoretical risk to aggravate the underlying pathophysiology [229]. The main regulator of platelet production, TPO, is elevated in sepsis and related to the platelet count [262, 263], which may be linked to the reduction in platelet mass or stimulation of TPO production by inflammatory mediators. Experimental models show that TPO neutralization reduces the severity of organ damage [264]. However, in the clinics, the potential benefit of TPO administration in thrombocytopenic patients in sepsis has been recently suggested [265]. At this stage, results from randomized controlled trials remain necessary to evaluate TPO interest in sepsis.

If the interpretation of thrombocytopenia in sepsis patients is made difficult by the multiplicity of underlying mechanisms, the platelet count by itself may hold valuable information [242]. The platelet count may represent a surrogate marker of the severity of organ dysfunction. A low platelet count occurring even early in sepsis patients is indeed recognized as a sign of poor prognostic; however, a single platelet count at admission may have little pertinence [266], and the kinetics of platelet counts appears to have a deeper meaning. Two alterations of this kinetics have been shown to be of clinical interest in sepsis patients, suggesting that they must be given specific attention. Both the magnitude of the drop in platelet count rather that thrombocytopenia per se, and the non-resolution of thrombocytopenia are strong predictors of mortality in sepsis [9, 15, 267]. The onset and dynamics of thrombocytopenia have been stressed as potential diagnostic approaches in ICU patients [222].