*Review* **Primary Immune Thrombocytopenia: Novel Insights into Pathophysiology and Disease Management**

**Anurag Singh 1,†, Günalp Uzun 2,† and Tamam Bakchoul 1,2,\***


**Abstract:** Immune thrombocytopenia (ITP) is an autoimmune disorder defined by a significantly reduced number of platelets in blood circulation. Due to low levels of platelets, ITP is associated with frequent bruising and bleeding. Current evidence suggests that low platelet counts in ITP are the result of multiple factors, including impaired thrombopoiesis and variations in immune response leading to platelet destruction during pathological conditions. Patient outcomes as well as clinic presentation of the disease have largely been shown to be case-specific, hinting towards ITP rather being a group of clinical conditions sharing common symptoms. The most frequent characteristics include dysfunction in primary haemostasis and loss of immune tolerance towards platelet as well as megakaryocyte antigens. This heterogeneity in patient population and characteristics make it challenging for the clinicians to choose appropriate therapeutic regimen. Therefore, it is vital to understand the pathomechanisms behind the disease and to consider various factors including patient age, platelet count levels, co-morbidities and patient preferences before initiating therapy. This review summarizes recent developments in the pathophysiology of ITP and provides a comprehensive overview of current therapeutic strategies as well as potential future drugs for the management of ITP.

**Keywords:** immune thrombocytopenia; bleeding; platelets; platelet destruction; immune tolerance; megakaryocytes; ITP treatment

#### **1. Introduction**

Primary immune thrombocytopenia (ITP) is a haematological autoimmune disorder characterised by bleeding and a low platelet count of less than 100 × <sup>10</sup>9/L [1–4]. There are several factors contributing to the onset of ITP, and the exact mechanisms behind how host immune response turns against own system (autoimmunity) and leads to ITP are still incompletely understood. There is growing evidence suggesting that the main event during ITP is a misbalanced interaction between effectors and regulatory immune cells [5]. This lack of an equitable response leads to a distorted immune tolerance, resulting in increased platelet clearance by immune cells, as well as an impairment in thrombopoiesis. Earlier studies suggested that a low platelet count is largely a consequence of anti-platelet antibodies opsonizing the cells and hence an increased clearance from the circulation [6–8]. However, lately, it has been demonstrated by many researchers that cytotoxic T cells also play a vital role in ITP pathomechanism by impairing megakaryopoiesis.

During ITP, it has been observed that although brief, spontaneous remissions can occur frequently in children. On the other hand, adult patients rather display a more chronic form of ITP that correlates with significant clinical presentations including bleeding disorders, haemorrhages in skin or mucous membranes, namely purpura, petechiae and rarely intracranial manifestations of the disease [9,10]. Treatment strategies for ITP

**Citation:** Singh, A.; Uzun, G.; Bakchoul, T. Primary Immune Thrombocytopenia: Novel Insights into Pathophysiology and Disease Management. *J. Clin. Med.* **2021**, *10*, 789. https://doi.org/10.3390/ jcm10040789

Academic Editor: Hugo ten Cate Received: 15 January 2021 Accepted: 10 February 2021 Published: 16 February 2021

**Publisher's Note:** MDPI stays neutral with regard to jurisdictional claims in published maps and institutional affiliations.

**Copyright:** © 2021 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (https:// creativecommons.org/licenses/by/ 4.0/).

are mostly prescribed on the basis of clinical symptoms of the patients with a focus on reducing the risk of severe bleeding, and they do not essentially include the boosting of platelet numbers. As per the guidelines of International Working Group [2,11], patients with acute ITP and without a history suggesting severe bleeding risk are advised to be managed with observation strategy (wait and see). On the other hand, ITP patients require urgent treatment if they are prone to a higher risk of bleeding or carry a severe case of chronic thrombocytopenia.

In this review, we discuss the pathomechanisms that lead to platelet destruction in ITP with a particular focus on recent findings regarding various diversifications during thrombopoiesis. Furthermore, we will provide a broad overview regarding various management strategies of ITP patients. We also outline different treatment options including efficacy and safety of therapeutic medicaments, management of bleeding emergencies as well as a summary of different approved drugs as well as drugs under clinical trials for ITP treatment.

#### **2. Pathophysiology of ITP**

One of the crucial steps during pathophysiology of ITP is described as the loss of immunological tolerance to autoantigens on patient's own platelets [12]. Many studies demonstrate that during ITP, a dysregulated T-cell response leads to a distorted balance of helper T cells (Th1/Th2) ratio [13,14], and imbalance further leads to an enhanced number as well as hyperactivity of cytotoxic T cells. Subsequently, this enhanced activity of cytotoxic T cells results in an increase in platelet destruction, combined with improved survival of B cells. An enhanced survival rate of B cells hence facilitates a larger production of autoantibodies, leading to an accelerated rate of platelet clearance. Autoantibodies opsonize platelets leading to enhanced phagocytosis, apoptosis, complement activation and impaired thrombopoiesis [15–17] (Figure 1).

**Figure 1.** Graphical representation of the pathophysiology of immune thrombocytopenia (ITP) illustrating involvement of multiple immune cells. Impairment of regulatory T cells leads to a disruption in regulation of helper T cell-mediated activation of B cells. B cells in turn produce autoantibodies in abundance leading to opsonisation, phagocytosis and complement activation, desialylation and finally destruction of platelets. Autoantibodies further hinder megakaryocyte maturation (megakaryocytopoiesis), and autoreactive cytotoxic T cells destroy megakaryocytes and platelets. (Adapted from Kashiwagi et al. 2013 [18]).

Although platelet destruction in the spleen primarily involves constant fragment (Fc)-dependent mechanisms, various researchers have also described novel mechanisms independent of Fc-mediation [19–21]. In a study, it was shown that ITP-autoantibodies can induce glycan modifications on platelet surface glycoproteins (GPs). Upon further recognition by Ashwell–Morell receptors which are expressed on hepatocytes, this GPs modification leads to accelerated platelet clearance in the liver [22]. CD8+ T cells from ITP patients also induce platelet desialylation and platelet phagocytosis by hepatocytes [23]. This might explain a potential mechanism how splenectomy remains ineffective in some ITP patients. In an intriguing retrospective study with a cohort of 61 ITP patients, it was shown that platelet desialylation and subsequent reduction in response to first line of treatments was independent of any Fc-mediated mechanism [24].

A recent study by Quach and colleagues demonstrated that ITP patients who did not respond to therapy were more likely to produce autoantibodies against the ligand binding domain (LBD) of GPlb/lX [25]. This specific binding leads to activation of GPIb/IX via crosslinking of platelet receptors and unfolding of a mechanosensory domain and platelet destruction, providing further a pivotal evidence of Fc-independent mechanism [25]. Recently, we demonstrated that novel effector functions of autoantibodies in ITP modulate the disease and might interfere with the clinical outcome for patients. We showed that a subgroup of autoantibodies induces cleavage of sialic acid residues from the surface of human platelets and megakaryocytes during ITP. Furthermore, autoantibody-mediated desialylation was found to interfere with the cell–extracellular matrix protein interaction and hence leading to impaired platelet adhesion and megakaryocyte differentiation [26]. This hints towards a potential use of sialidase inhibitors as a treatment approach in combination with other therapies to boost platelet numbers in some patients who have failed to respond to previous therapies.

It is well established that intrinsic apoptotic pathway plays a significant role in platelet life cycle. Many research groups have demonstrated the role of ITP-autoantibodies in regulating platelet apoptosis and pathways involved. There is ample evidence showing that various apoptosis markers including phosphatidylserine (PS) exposure, depolarisation of the mitochondrial transmembrane potential, Bcl-2 family protein expression, activation of caspase-3 as well as of caspase-9 are significantly involved in platelet apoptosis in ITP [27,28]. Immunoglobin infusion was shown to successfully mitigate platelet apoptosis in adult as well as paediatric patients [29,30]. Interestingly, it was shown that apoptotic platelets were not found in ITP patients harbouring anti-GPIa/IIa autoantibodies but only in those who carried anti-GPIIb/IIIa and anti-GPIb autoantibodies [31], indicating a potential role of autoantibody specificity.

Autoantibodies produced during ITP not only affect platelet survival but also platelet formation by megakaryocytes [32]. It has been shown that autoantibodies bind and hinder the megakaryocyte maturation, resulting in reduced platelet formation [33,34]. It was demonstrated in vitro, that autoantibodies inhibit platelet production by impairing megakaryopoiesis and maturation [35–37]. However, the role of megakaryocyte apoptosis still needs to be investigated in terms of involvement in the pathophysiology of ITP. There have been some hints and contradicting claims through results generated via earlier and recent investigations. A study in fact demonstrated that treatment with ITP plasma rather leads to a reduced apoptosis of megakaryocytes [38]. Haematopoietic stem cells (HSCs) isolated from healthy umbilical cord blood were co-cultured with plasma of ITP patients, resulting into a decrease in apoptosis, reduced expression of tumour necrosis factor-related apoptosis inducing ligand (TRAIL) and increased expression of the anti-apoptotic protein Bcl-xL in differentiated megakaryocytes [39]. On the other hand, in contrast to these findings, an earlier in vivo study suggested that megakaryocytes in fact undergo enhanced apoptosis in the presence of autoantibodies [40]. It was observed in biopsies of ITP patients that increased apoptosis involves nuclear fragmentation, chromatin condensation and activation of caspase 3. This further leads to phagocytosis of the polyploid cells by resident macrophages in the bone marrow [40]. Another recent study

showed that an increased megakaryocyte apoptosis occurs in the bone marrow samples obtained from ITP patients [41].

#### **3. Clinical Manifestations**

The overall annual incidence rate of ITP is 1.6–5.3 per 100,000 persons, and it is more frequent in women than men [42–44]. ITP can be classified according to disease duration as acute (<3 months), persistent (3 to 12 months) or chronic (>12 months). Compared to children, adults are more likely to develop chronic ITP disease. While up to 60% of adults develop chronic diseases [45,46], only 20–30% of children have persistent thrombocytopenia at 12 months [47,48].

Most patients are presented with bleeding symptoms such as petechiae, purpura, haemorrhages of the mucous membranes of the mouth and nose, urogenital bleeding or increased menstrual bleeding [49]. Some patients can be asymptomatic at presentation and 30–40% of patients with chronic ITP do not have any bleeding symptom [50]. Bleeding risk is calculated as 8% per year in ITP patients [51].

Major bleedings are associated with a high rate of mortality [52,53]. Reported rates of severe bleeding vary depending on the population studied. In a recent literature review including 108 studies reporting on 10,908 patients, the weighted proportion for intracerebral haemorrhage (ICH) was 1.0% (95% CI, 0.7–1.3) and for non-ICH severe bleeding was 15.0% (95% CI, 9.3–21.8) [54]. Forsthye and colleagues reported a severe bleeding episode that required rescue medication (intravenous immunoglobulin, corticosteroid injections or platelet transfusions) in 10,2% of adult ITP patients within 6 months after starting therapy with thrombopoietin receptor agonists (TRO-RA) [55]. In a retrospective evaluation of the McMaster ITP registry, Arnold et al. found that 56% of ITP patients experience clinically significant bleeding at some point during their disease course and 2.2% had ICH [56].

Compared to ITP patients with normal platelet counts, those with a platelet count between 25 to 49 × 109 /L and <25 × 109 /L had 2.4 fold and 4.5 fold increased bleeding rates, respectively [51]. Furthermore, bleeding requiring a hospital contact within 1 year prior to ITP diagnosis was associated with a 3-fold increased rate of subsequent bleeding [51]. The use of non-steroidal anti-inflammatory drugs (NSAIDs) was found to be associated with any bleeding (OR 4.8, 95% CI 1.1–20.7) and anticoagulant drugs were associated with severe bleeding (OR 4.3, 95% CI 1.3–14.1) [57]. In a large patient cohort, Hato et. al. found that age (>60 years), platelet count (<10 × <sup>10</sup><sup>9</sup> /L), and the presence of haematuria are associated with increased risk for ICH [58].

Fatigue is common in patients with ITP, and its impact on health-related quality of life in ITP patients has been until recently underappreciated [59]. Treatments that increase platelet count also reduce fatigue [60,61]. However, it is also recommended to use treatment strategies that directly target fatigue to improve the health-related quality of life in ITP patients [62].

Paradoxically, an increased frequency of thromboembolic events has been reported in ITP patients [63,64]. Therefore, it is crucial that ITP patients should be aware of the risk of thromboembolic events. Patients should be educated that ITP can increase not only the risk of bleeding but also the risk of venous and arterial thromboembolism [50]. Furthermore, patients at risk of embolic events should be followed more closely. Presence of lupus anticoagulants is related to thrombotic events [65]. The increased levels of prothrombotic, platelet-derived microparticles and complement activation on antibody-coated platelets also contribute to the development of thrombosis in ITP [66]. In addition to disease and patient related factors, ITP treatments such as TPO-RA and splenectomy could also increase the individual risk of thromboembolic events [64,67,68]. Clinical management of thrombocytopenic patients who require anticoagulant or antiplatelet therapy due to cardiovascular comorbidities is a serious challenge. An aggressive treatment of ITP may be required in these patients to achieve a safe platelet count over 50 × <sup>10</sup>9L [50].

The overall mortality rate is slightly higher than general population in ITP patients, predominantly due to increased cardiovascular disease, infection, bleeding and haematological cancer related mortalities [69].

#### **4. Diagnosis**

ITP is usually diagnosed after precluding other potential causes of thrombocytopenia. A diagnosis is performed in patients with a low platelet count (<100 109/L) with no evidence or history of an underlying condition, which can lead to thrombocytopenia, including a physical examination, evaluation of blood counts and visual examination of blood smears. However, since thrombocytopenia may be a multifactorial condition, it is indeed complicated to identity substitute causes, and examining physician needs to have a broad knowledge in platelet disorders. A confirmation of ITP is achieved via detection of characteristic platelet-specific autoantibodies, free in patient serum or bound to own platelets [70]. As per recommendations of various regulatory guidelines, GP-specific assays, for example direct monoclonal antibody immobilisation of platelet antigens (MAIPA test) or direct immunobead assays prove the diagnosis of ITP, and further laboratory tests are deemed unnecessary [71]. However, current ASH-guidelines of 2019 do not give any clear recommendations for antibody evaluation in ITP patients, as there is still lack of strong evidence supporting clinical advantage of the assays [4]. We strongly recommended that as a part of initial assessment, presence of platelet autoantibodies should be evaluated. A positive test result at this stage establishes a sound basis for further diagnostic procedures and paves ways for initiating the treatment. It is notable to mention that although GPspecific tests have shown an excellent specificity, the lack of sensitivity is an important issue to consider. The low sensitivity of the test can often produce negative results, and care needs to be taken while interpretation and subsequent recommendation. Other potential hurdles in implementing antibody testing as a part of mandatory diagnostic regime for ITP also include unavailability of experienced staff, equipment and set up, as well as cost effectiveness.

Therefore, it is recommended to establish an appropriate diagnostic set up to analyse ITP during early phase of patient examination.

#### **5. Treatment of ITP**

The main goals of ITP treatment are to intervene in the case of an acute severe bleeding and to prevent future bleeding events (Figure 2).

#### *5.1. First-Line Treatments and Treatment of Bleeding Emergencies*

The decision to start a treatment in newly diagnosed ITP depends on several factors. Current guidelines recommend a platelet count of 20 to 30 × 103/μL as a cut-off value to start intervention in adult ITP patients [4]. Other than thrombocyte count, patient related factors can help to determine the risk of bleeding such as age (e.g., >65 years), previous bleeding events, comorbidities associated with high bleeding risk (i.e., hypertension, cerebrovascular disease), renal or hepatic impairment, medication with anticoagulants and platelet inhibitors, surgical interventions and risky life style (i.e., combat sports) [51,57,73]. A higher platelet count (>50,000 μL) should be considered for these patient populations.

As emphasized in the current guidelines, the decision regarding ITP treatment should be made in agreement between physician and patient. The patient should be informed about the benefits and possible side effects of treatment options. It should be considered that some side effects of treatments might pose a greater risk for the patient than ITP itself [74]. Advantages and disadvantages of ITP treatments are summarized in Table 1.

**Figure 2.** Treatments for immune thrombocytopenia: Corticosteroids used as first line of treatment modulate Treg, B-cell and FcR function. In combination with or without IVIg and anti-D, they impair antigen presentation and recognition of autoantibody-coated platelets by macrophages. Second-line treatments such as surgical splenectomy remove spleen as a site of platelet destruction, and drugs as rituximab target antibody-producing B cells. TPO-RAs, such as romiplostim and eltrombopag work by stimulating platelet production by megakaryocytes. Fostamatinib impairs Syk-mediated phagocytosis of platelets. FcR: Fc receptors; IVIg: Intravenous immunoglobulin; TPO-RA: Thrombopoietin receptor agonist; Syk: spleen tyrosine kinase. (Adapted from Newland et al. 2018 [72]).




#### 5.1.1. Glucocorticoids

Glucocorticoid treatment is the most-commonly used first-line therapy in patients with ITP [43,75]. The beneficial effects of glucocorticoids include reduction of platelet clearance by reticuloendothelial system [76,77]. Platelet count usually increases within a couple of days after therapy initiation [49]. Two most-commonly used glucocorticoids are prednisone (1 mg/kg orally per day for 2–3 weeks, with a gradual withdraw and discontinuation by 6 to 8 weeks) and high-dose dexamethasone (one or more cycles of 40 mg orally, once daily for 4 days, usually 4 weeks apart) [78]. Current ASH guideline recommends against the use of glucocorticoids longer than 6 weeks [4]. On the other hand, some others suggested that a longer low-dose steroid therapy could be considered to keep the platelet counts over 30 × 103/ml if a response with initial steroid therapy has been achieved [11]. Several studies demonstrated more rapid response with dexamethasone as compared to prednisone, but overall response rates are not significantly different in the long term after 6 and 12 months [79]. Similarly, Wang et al. reported a rapid response with high-dose dexamethasone compared to prednisolone, but sustained response rates were similar at 12 months and later [78]. Of note, dexamethasone seems to have a better

safety profile (fewer Cushing's disease, weight gain and infection rates) in comparison to prednisolone [80].

Despite the high early-response rate, most of the patients do not have a sustained response after the cessation of glucocorticoids. In fact, approximately 80% of patients respond initially to corticosteroids, but only 20 to 40% of these patients achieve sustained response when steroids are discontinued [81,82]. As a predictive factor, Wang et al. measured anti-platelet antibody levels in ITP patients under glucocorticoid treatment [78]. They found that presence of anti-GPIb-IX antibodies predicts a poor initial response to corticosteroids [78]. Further studies are needed to determine the role of antiplatelet antibodies in predicting the corticosteroid response.

It is crucial to closely monitor the patients for possible side-effects of glucocorticoids such as hypertension, hyperglycaemia, sleep and mood disturbances, gastric ulceration, myopathy, glaucoma and osteoporosis [4]. To prevent severe toxicities, corticosteroids should be tapered appropriately and discontinued in non-responding patients. Nonresponders and patients with contraindication to steroid therapy such as (pregnancy, diabetes mellitus, active infection and psychiatric disorders) can be treated with other first-line treatments-IVIG and IV anti-D [4].

To increase the rate of sustained response, combination of dexamethasone with second line treatments such as rituximab have been investigated. A recent meta-analysis compared the effectiveness of the combination of high-dose dexamethasone and rituximab with dexamethasone alone in ITP [83]. Overall response rate at month 3 (RR = 5.07, 95% CI: 2.91–8.86, and *p* < 0.00001) and sustained response rate at 12 months (RR = 1.73, 95% CI: 1.36–2.91, and *p* < 0.00001) was significantly higher in combination arm than that in monotherapy. Furthermore, the rate of adverse events has not significantly increased with combination therapy [83].

#### 5.1.2. Intravenous Immunoglobulin (IVIG)

IVIG has been introduced into the treatment of ITP in 1980s [84]. IVIG is prepared by purification from the pooled plasma of healthy donors [84]. It contains polyvalent IgG (80 to >95%) and irrelevant amount of IgA and IgM. IVIG is thought to inhibit Fcmediated phagocytosis of antibody coated platelets by reticuloendothelial system [85]. Platelet count usually increases within 48 hours after IVIG application [86]. The preferred treatment regime is 1 g/kg per day, which should be repeated for two consecutive days [2]. A lower dose of 0.2–0.4 g/kg/day can also be used for 4–5 days [87]. In a meta-analysis of 13 randomized studies, low dose IVIG regimes were found to be as effective as high dose IVIG, and low-dose-IVIG was associated with a significantly reduced risk of side-effects (OR = 0.39 (95% CI = 0.18–0.83) [88].

Limited number of randomized controlled studies compared the effectiveness of IVIG and corticosteroids as a first line therapy in ITP in adults [89,90]. Godeau et al. demonstrated that IVIG increases platelet count more effectively than high-dose methylprednisolone in adults with newly diagnosed ITP (79% vs. 60% response rate) [90]. In a smaller study, adult ITP patients were treated with oral prednisone (1 mg/kg/day; *n* = 17), high-dose IVIG (400 mg/kg on days 1 through 5; *n* = 13) or a combination of both agents (*<sup>n</sup>* = 13). A platelet response (>50 × 109/L) was achieved in 82%, 54% and 92% of patients, respectively [89].

There may be a relationship between the presence of anti-platelet antibody and the response to IVIG. Peng et al. found that the response rate was significantly higher in patients without anti-GPIb-IX autoantibodies compared to those with anti-GPIb-IX autoantibodies (80.0% vs. 36.4%), while the presence of the anti-GPIIb/IIIa autoantibodies had no effect on response to treatment [91]. However, others failed to show a significant relationship between an autoantibody and nonresponse to IVIG [92].

Most frequent adverse effects of IVIG include headache, pyrexia and vomiting [93]. Severe side effects such as acute kidney injury, aseptic meningitis and thrombotic events are rare [94].

#### 5.1.3. Anti-RhD Immunoglobulin (Ig)

Anti-RhD consists of IgG selectively taken from the plasma of donors immunized to the Rhesus D antigen [85]. Anti-RhD Ig binds to Rh-positive erythrocytes and these antibody-coated erythrocytes competitively inhibit the destruction of antibody-coated platelets by binding and occupying Fc receptors on phagocytes in the spleen [95]. Anti-RhD is therefore only effective in Rh-positive patients with an intact spleen. A single intravenous dose of 50 to 75 μg/kg is recommended [96]. A safe subcutaneous administration in small children or patients is also described in the literature [97]. Side effects include mild infusion reactions such as headache, nausea, chills, fever and mild to moderate haemolysis [98]. However, life-threatening episodes of severe intravascular haemolysis and disseminated intravascular coagulation after Anti-RhD Ig administration have also been reported [99,100]. These reports led to the withdrawal of an Anti-RhD product (WinRho® SDF, Cangene Europe Ltd, London, UK) from European markets in 2009 [50].

#### *5.2. Treatment of Active Bleeding*

In case of clinically relevant bleeding, glucocorticoids, IVIG and platelet transfusion are used alone or in combination to increase the platelet count rapidly [11]. Besides, other interventions such as endoscopy or surgery may be necessary depending on the severity and the site of the bleeding [52]. Furthermore, anticoagulant and antiplatelet medications should be ceased immediately, if possible. Since the effect of platelet transfusion is limited due to rapid clearance of platelets by the circulating autoantibodies, combining platelet transfusion with IVIG or corticosteroids might be useful [11]. Although IVIG increases platelet count in most of the cases within 48 hours, its effect is temporary, and platelet count decreases after 1 to 2 weeks. Therefore, concomitant use of glucocorticoids with IVIG can be considered to achieve a more sustained response than that with IVIG alone [90]. Of note, the recommendations for the treatment of active bleeding in ITP are based on small observational studies, and randomized controlled studies are urgently needed.

The Updated International Consensus Report recommends the use of TPO-RA in the case of a life-threatening bleeding if initial treatments with corticosteroids, IVIG and thrombocyte transfusion fails to increase the platelet count [11]. Roumier et al. used high dose romiplostim (10 μg/kg body weight) together with vinca alkaloids in 30 patients with severe bleeding and compared the results with a historical patient group treated with vinca alkaloids only [101]. Both groups constituted of patients who failed to achieve a response after initial therapy with IVIG, corticosteroids and/or platelet transfusion [101]. At day 7, complete response (60% vs. 29%, *p* < 0.05), and at day 14, both partial response (80% vs. 43%, *p* < 0.05) and complete response (70 vs. 17%, *p* < 0.0001) were significantly higher in the romiplostim plus vinca alkaloid group compared to the vinca alkaloid group alone [101]. Although this study shows the effective use of high dose romiplostim in life-threatening bleeding in ITP patients, two patients (6.6%) treated with high dose romiplostim developed major thromboembolic events. Therefore, the risk over benefit ratio should be carefully assessed for each patient.

Antifibrinolytics (tranexamic acid and aminocaproic acid) are successfully used to control significant bleeding in patients with ITP [102–104]. Oral contraceptives can be used in female patients with menorrhagia. In life threatening bleeding emergencies, recombinant activated factor VII may be a useful supportive treatment [105–107].

#### *5.3. Second-line treatments*

#### 5.3.1. Thrombopoietin-Receptor Agonists (TPO-RA)

Romiplostim is an Fc-peptide fusion protein and administered as a once-weekly subcutaneous injection. The recommended initial dose is 1 μg/kg per week, which can be adjusted by weekly increments of 1 μg/kg according to platelet response to achieve a platelet count of >50 × 109 platelets/L. The maximum dose is 10 <sup>μ</sup>g/kg/week. Romiplostim is indicated in adult ITP patients who have had an insufficient response to corticosteroids, immunoglobulins or splenectomy. Self-administration of romiplostim by patients can help

in reducing healthcare costs and increase patient comfort by eliminating the need to visit the hospital every week for applications [108].

Owing to the effectiveness and safety-profile of TPO-RAs recent studies explored the use of these drugs in other patient groups also. Kuter et al. investigated the effectiveness of romiplostim in patients with ITP for less than 12 months by analysing the data from 9 studies [109]. They found that the number of patients with a platelet response at ≥75% of measurements were higher for romiplostim (74% (204/277)) than for placebo/standard of care (18% (6/34)) in patients with ITP ≤1 year. More importantly the rate of treatment free remission (platelet counts ≥<sup>50</sup> × 109 /l for ≥6 months) was higher in patients with ITP ≤ 1 year compared to those with ITP >1 year [109]. Clinically relevant bleeding-related episodes are significantly lower in patients on romiplostim therapy [110,111]. Kuter et al. followed 292 adult ITP patients receiving romiplostim as weekly treatment and observed that the platelet response is maintained with stable dosing for up to 5 years of continuous treatment [67].

Most frequently observed side effects are headache, arthralgia, myalgia, dizziness and insomnia [112]. Thromboembolism and bone marrow fibrosis are the most feared complications of TPO-RA in ITP patients. Gernsheimer reported that romiplostim does not present an increased risk of thromboembolic events compared to placebo [111]. However, close monitoring of patients for thromboembolic events is recommended. Bone marrow changes were observed in a small proportion of patients receiving romiplostim [113]. But the bone marrow fibrosis is reversed after the end of treatment [114,115].

Eltrombopag, which is a synthetic non-peptide molecule, binds selectively with thrombopoietin receptors on megakaryocytes and induces thrombopoiesis [116]. Eltrombopag is recommended for adult ITP patients who have had an insufficient response to corticosteroids, immunoglobulins or splenectomy. Eltrombopag is administered orally as a daily tablet. Daily dose is 25–75 mg according to the age and hepatic function status of the patient. To ensure an adequate absorption of eltrombopag, it should be taken at least 2 hours before or 4 hours after any medications or products containing polyvalent cations (such as antacids, calcium-rich foods and mineral supplements). Many patients have difficulty meeting these dietary requirements and an alternative intermittent dosing 1–5 times weekly have been recommended [117]. Due to the risk of hepatotoxicity, a dose reduction is necessary in patients with hepatic impairment and a close monitoring of liver enzymes and bilirubin every two weeks throughout the treatment is indicated [118].

Randomized controlled studies showed that eltrombopag achieved early platelet response in 70–80% of the patients and a remission rate of 20–30% [119–122]. In an openlabel extension study, 85% of the patients achieved a platelet response, and 52% of them had a continuous response of 25 weeks or longer [123]. Furthermore, the incidence of bleeding episodes in patients receiving eltrombopag decreased from 57% to 16% at 1 year [123]. Although some patients seem to have a prolonged/complete remission after pausing TPO-RA, no prognostic marker is currently available to identify such patients [124]. However, recently, an inverse relation between TPO level and response to eltrombopag or romiplostim has been shown [125]. Patients with a normal baseline TPO level are more likely to benefit from a therapy with these drugs [125].

Forsthye et al. compared the bleeding related adverse events in patients receiving romiplostim or eltrombopag in a retrospective cross-sectional study. Patients on eltrombopag (*n* = 1617) had significantly fewer bleeding episodes compared to those on romiplostim (*n* = 1140) (7% vs. 14%) [55].

In terms of adverse effects, liver functions, thromboembolism and bone marrow fibrosis have been the areas of concern in the long-term use of eltrombopag [126]. Gastrointestinal symptoms (nausea, vomiting and diarrhoea), mild transaminase elevations and headache are the most commonly observed adverse events in clinical studies [122]. In a prospective safety and efficacy study, thromboembolic events were observed in 6% of patients and hepatobiliary side effects in 15% of patients with a median eltrombopag treatment duration of >2 years [123]. Regular follow-up of patients for these side effects is justified.

Avatrombopag is an orally administered TRO-RA and recently received FDA approval for treatment of resistant ITP in adults. Unlike eltrombopag, avatrombopag can be administered without dietary restrictions. Furthermore, avatrombopag does not require monitoring of liver functions [127]. The phase 3 clinical trial showed a longer median number of weeks with platelet count of 50 × 109/L or higher during the first 26 weeks in patients who received avatrombopag than in those who received placebo [128]. A platelet response (a platelet count ≥<sup>30</sup> × <sup>10</sup>9/L, with at least a two-fold increase in platelet count from baseline and an absence of bleeding) has been observed in 56.3% of the avatrombopag treated patients [128]. The recommended initial dose is 20 mg/day. The doses or dosing frequency should be adjusted individually to maintain platelet count greater than <sup>50</sup> × <sup>10</sup>9/L. The maximum daily dose is 40 mg [127]. The treatment should be discontinued if a platelet response is not achieved in 4 weeks of avatrombopag therapy at a dose of 40 mg/day. Most common side effects are headache, arthralgia, fatigue and diarrhoea. Further studies are needed to ensure the long-term safety of avatrombopag.

#### 5.3.2. Immunomodulators

Rituximab is an anti-CD20 monoclonal antibody that depletes CD20+ B cells and reduces antiplatelet antibody production directly [129]. Rituximab achieves a significantly higher incidence of complete response at 6 months compared to glucocorticoids or placebo in non-splenectomised ITP patients (46.8% vs. 32.5%) [130]. More than one-half of the responders had their response last for at least 1 year, resulting in a 1-year response rate of 38%. Patel et al. reported a 2-year response rate of 31% and a 5-year response rate of 21% in adults treated with rituximab [131]. Sustained platelet response lasts more than 2 years in 50% of patients who have an initial response to rituximab [131,132]. Low dose rituximab therapy has been recommended to avoid treatment related adverse events. A recent systematic review found an overall response rate of 63% and complete response rate of 44% in ITP patients treated with low-dose (100 mg or 100 mg/m2 per week for 4 weeks) rituximab instead of the standard dose of 375 mg/m<sup>2</sup> per week for 4 weeks [133]. Low dose rituximab has a satisfactory efficacy and safety profile [133]. In a long-term follow-up study (median follow-up of 6 years), median duration of response was longer (17 months vs. 11 months), and splenectomy rate was lower (17.2% vs. 26.4) in rituximab-treated patients. However, 70% of the rituximab-treated patients relapsed within two years after response [134]. Hammond et al. showed that response rate at 2 years was 70% in ITP patients treated with rituximab after unsuccessful splenectomy [135]. Wang et al. have recently demonstrated that a positive ANA test is associated with a better initial response but with an unfavourable long-term outcome in ITP patients treated with rituximab [136].

Rituximab should not be prescribed to patients with evidence of an active or previous HBV infection due to the risk of fulminant hepatitis, and other treatment options should be considered [129]. An increased tendency to minor infections after rituximab therapy has been reported. On the other hand, progressive multifocal leukoencephalopathy seem to be rare [137]. Taken together, due to the lower efficacy and higher complications compared with TPO-RAs [138], rituximab should be avoided as first line therapy and used only if there is high evidence for remission [4].

Fostamatinib is an orally available spleen tyrosine kinase (Syk) inhibitor. Syk-dependent phagocytosis of FcγR-bound platelets plays a role in the pathophysiology of ITP, and fostamatinib inhibits antibody-mediated destruction of platelets [139]. Pooled analyses of two randomized controlled trials demonstrated a response within 12 weeks in 43% of the patients compared to 14% of those receiving placebo [140]. In addition, a sustained platelet count ≥<sup>50</sup> × <sup>10</sup>9/L for up to 24 weeks was observed in 18% of refractory ITP patients compared to 2% of those receiving placebo [140]. In the open label extension study with the patients who had a stable response, 21 (78%) patients had maintained the response for 1 year and 15 (56%) for 2 years [141]. In a post-hoc analysis of the phase 3 and open-label extension study, Boccia et al. observed a higher platelet response rate (≥<sup>50</sup> × <sup>10</sup>9/L) (78% vs. 48%) and lower bleeding events (28% vs. 45%) when fostamatinib was used as a second

line therapy as compared to its use as a third-or-later-line of therapy [142]. The recommended initial dose is 100 mg twice daily, and the dose can be increased to 150 mg twice daily, if platelet count has not increased to at least 50 × 109 /L after 4 weeks of therapy. Most common adverse reactions are diarrhoea, hypertension and nausea. A monthly monitoring for hepatotoxicity and neutropenia is recommended [143]. Long-term studies are needed to better understand the efficacy and safety profile of fostamatinib in patients with chronic ITP.

#### 5.3.3. Splenectomy

Spleen is the main site of the autoantibody production and platelet destruction. Splenectomy is long regarded as the gold standard therapy for ITP patients who are unresponsive to corticosteroids [144]. Compared to other treatment options, splenectomy has a higher sustainable response rate [4]. However, with the introduction of new medicaments, splenectomy has lost its place in the treatment of ITP [75,145].

Splenectomy achieves a high rate of durable remissions in 60 to 70% of the patients [146]. The need for the third-line treatment is significantly lower in patients who have undergone splenectomy (20%) compared to patients treated with second-line therapy (39–44%) [147]. Vianelli et al. reported a relapse free survival in 67% of the patients for up to 20 years after splenectomy [148]. However, due to the surgical risks and potential longterm complications, splenectomy is usually reserved to chronic ITP patients who failed to respond to standard medical therapies or when therapies are contraindicated [50,144].

Furthermore, the lack of reliable predictors of response to splenectomy hinders the selection of the patients who will benefit from splenectomy [146]. Revealing the main site of platelet sequestration can help to predict the success of splenectomy. Autologous platelet scanning can be used to detect the site of platelet sequestration, but it is technically challenging and not widely available [149]. Knowledge of desialylation capacity of the anti-platelet autoantibodies might also be helpful to detect Fc-independent clearance of platelets in the liver [22].

Complications associated with splenectomy are post-operative bleeding, infection with encapsulated bacteria, sepsis as well as thromboembolic events in venous and arterial circulation (i.e., coronary artery disease, stroke and chronic thromboembolic pulmonary hypertension) [144]. In a retrospective analysis of medical records, among second line treatments, splenectomy had the highest frequency of deep vein thrombosis and pulmonary embolism [147]. Compared to open surgery, laparoscopic splenectomy has a lower rate of postoperative mortality and morbidity and a shorter hospitalization [146,150]. Moreover, the immediate as well as the persistent risks of venous thromboembolism have been shown to be higher among patients with ITP who have undergone splenectomy as compared those who have not [151,152].

Patient's age must also be taken into consideration during the selection process for splenectomy. Maria et al. showed that patients age at the time of the surgery predicted the response in children [153]. Older children show a better outcome after splenectomy. Recently, Kwiatkowska et al. showed that age (<41 years) together with (BMI < 24.3 kg/m2) and preoperative platelet count (≥<sup>97</sup> × 103 mm3) are independent prognostic factors for ITP remission after splenectomy [154]. Geriatric patients are prone to surgical complications and an increased relapse has been reported in ITP patients over 60 years [155,156]. Therefore, splenectomy should be implemented as a last resort in elderly patients. Last but not the least; splenectomy should not be performed in the first 12 to 24 months after ITP diagnosis because of the chances of spontaneous remission or disease stabilization [11].

#### *5.4. New Drugs under Investigation*

Rozanolixizumab is anti-neonatal Fc receptor (FcRn) antibody that reduces plasma IgG levels. In a recent phase 2 study, >50% patients with persistent/chronic primary ITP achieved clinically relevant platelet responses (≥<sup>50</sup> × 109/L) by day 8 after a single injection of rozanolixizumab at a dose of 15 and 20 mg/kg [157]. Treatment related mild-tomoderate adverse events have been seen in 15 of 66 (21%) patients, and no serious infections have been reported. A phase 3 study is currently recruiting participants (NCT04224688).

Bortezomib, a proteosom inhibitor, induces apoptosis of long-lived autoreactive plasmocytes and reduces secretion of anti-platelet antibodies. In murine models of ITP, bortezomib eliminated long-lived plasmocytes and alleviated thrombocytopenia [158]. Beckman et al. used bortezomib to treat a 63-year-old female patient who had severe thrombocytopenia and bleeding episodes despite the utilization of several treatments including splenectomy [159]. The patient received bortezomib injections in addition to other treatments, and platelet count increased rapidly after the initiation of bortezomib. The results of the ongoing clinical trials (NCT03443570, NCT04083014) will help us to better define, if any, the role of bortezomib in ITP.

Efgartigimod is an Fc fragment that blocks FcRn. In a recent study, patients with a platelet count <30 × <sup>10</sup>9/L despite treatment received four weekly intravenous injections of either placebo or efgartigimod, at a dose of 5 mg/kg or 10 mg/kg [160]. Antiplatelet antibody levels reduced 40% or more in 8/12 (66.7%) patients treated with efgartigimod at 5 mg/kg and in 7/10 (70.0%) patients treated with efgartigimod at 10 mg/kg. A platelet response >50 × 109/L on 2 occasions has been achieved in 46.2% of the patients on efgartigimod as compared to 25% on placebo [160]. A Phase 3 Study investigating the safety and efficacy of efgartigimod at a dose of 10 mg/kg is ongoing (NCT04225156).

Decitabine is an inhibitor of DNA methylation and used in the treatment of myelodysplastic syndrome. Considering the possible role of DNA-methylation in the aetiology of ITP [161], decitabine seems to be a potential treatment option. Low dose decitabine promotes megakaryocyte maturation and platelet production in patients with myelodysplastic syndrome and ITP [162,163]. In a prospective open label study, Zhou et. al. showed that an overall response rate of 51% with a median initial response time of 28 days in ITP patients [164]. The sustained response rates at 6, 12 and 18 months were 44.44% (20/45), 31.11% (14/45) and 20.0% (9/45), respectively [164].

#### **6. Conclusions**

In recent years, ITP guidelines have been updated in the context of improved understanding of the pathophysiology of ITP and evidence supporting newly introduced treatments. Despite recent developments, the expected increase in the success rate of treatments has not been achieved yet. A substantial number of patients either do not respond at all or respond only transiently to many treatment interventions. The use of different treatment regimens targeting different key points in the pathophysiology of the disease may increase the success rate. In addition, the development of patient-specific testing methods, which can predict treatment success, may assist in avoiding complications, wasted time and associated costs from unnecessary treatments.

The management of ITP during ongoing 2019 coronavirus disease (COVID-19) pandemic has emerged as an additional challenge for clinicians. COVID-19, caused by severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2), is known to be associated with increased coagulopathy and thrombotic complications [165]. Current data are insufficient to make evidence-based recommendations related to the ITP management. Pavord et al. published a series of recommendations based on expert opinion on the management of ITP during the COVID-19 pandemic [166]. They drew attention to a possible further increased risk of thrombosis in patients with COVID-19 from ITP or its treatment (particularly with TPO-RA). Mahevas et al. reported in a case series that COVID-19-associated ITP can lead to profound thrombocytopenia and severe bleeding manifestations but has a favourable outcome in most cases [167]. More studies are needed to make evidence-based decisions on managing ITP during the pandemic.

Current guidelines state that patient preferences should be prioritized when choosing a treatment regimen. Important factors that determine patient preferences include treatment efficacy and the potential for complications. Efficacy and safety data from post-marketing studies of new treatments will be helpful in this regard. In addition, randomized controlled trials comparing existing treatments not only in terms of treatment response or safety but also in terms of their impact on the health-related quality of life of patients with ITP are needed.

**Author Contributions:** A.S. conducted the literature search, created the figures and wrote the sections on pathophysiology and diagnosis of ITP. G.U. conducted the literature search and wrote the sections on clinical manifestations and treatment of ITP. T.B. designed the original layout and edited the manuscript. All authors have read and agreed to the published version of the manuscript.

**Funding:** This work was supported by grants from the German Research Foundation, the Herzstiftung (BA5158/4), Günther Landbeck Foundation and German Red Cross to T.B.

**Institutional Review Board Statement:** Not applicable.

**Informed Consent Statement:** Not applicable.

**Data Availability Statement:** Not applicable.

**Acknowledgments:** We thank Karina Althaus for insightful suggestions.

**Conflicts of Interest:** The authors declare no conflict of interest.

#### **References**


### *Review* **Thrombocytopenia in Virus Infections**

**Matthijs Raadsen 1, Justin Du Toit 2, Thomas Langerak 1, Bas van Bussel 3,4, Eric van Gorp 1,5 and Marco Goeijenbier 1,5,\***


**Abstract:** Thrombocytopenia, which signifies a low platelet count usually below 150 <sup>×</sup> <sup>10</sup>9/L, is a common finding following or during many viral infections. In clinical medicine, mild thrombocytopenia, combined with lymphopenia in a patient with signs and symptoms of an infectious disease, raises the suspicion of a viral infection. This phenomenon is classically attributed to platelet consumption due to inflammation-induced coagulation, sequestration from the circulation by phagocytosis and hypersplenism, and impaired platelet production due to defective megakaryopoiesis or cytokine-induced myelosuppression. All these mechanisms, while plausible and supported by substantial evidence, regard platelets as passive bystanders during viral infection. However, platelets are increasingly recognized as active players in the (antiviral) immune response and have been shown to interact with cells of the innate and adaptive immune system as well as directly with viruses. These findings can be of interest both for understanding the pathogenesis of viral infectious diseases and predicting outcome. In this review, we will summarize and discuss the literature currently available on various mechanisms within the relationship between thrombocytopenia and virus infections.

**Keywords:** virus infection; thrombocytopenia; thrombocytopathy; aggregation; HIV; SARS-CoV-2; hantavirus; coronavirus; influenza

#### **1. Introduction**

In patients presenting to care with signs or symptoms of infectious disease, a full blood count is part of a routine diagnostic evaluation. Mild thrombocytopenia, often combined with lymphocytopenia is typical of most acute viral infections, but neither are sufficiently sensitive nor specific to reliably distinguish viral from bacterial or parasitic pathogens. Except for viral hemorrhagic fevers and rare cases of severe disseminated viral infections, virus-induced thrombocytopenia does not lead to significant bleeding, rarely requires platelet transfusions, and is therefore easily dismissed as clinically irrelevant. However, when the relationship between platelets and viral infection is studied more closely and in larger study populations, important findings emerge which shed light on previously unrecognized aspects of viral diseases. The incidence of thromboembolic complications is elevated in individuals during and after influenza virus infection, for example, a relation which may not be apparent to physicians diagnosing and treating influenza-like illness [1]. Platelet counts during peak symptomatic disease have also been found to be a marker of disease severity in certain viral infections, [2–19] or can serve as a first clue towards diagnosing chronic viral infections [20–23]. These phenomena are

**Citation:** Raadsen, M.; Du Toit, J.; Langerak, T.; van Bussel, B.; van Gorp, E.; Goeijenbier, M. Thrombocytopenia in Virus Infections. *J. Clin. Med.* **2021**, *10*, 877. https:// doi.org/10.3390/jcm10040877

Academic Editor: Hugo ten Cate

Received: 7 January 2021 Accepted: 17 February 2021 Published: 20 February 2021

**Publisher's Note:** MDPI stays neutral with regard to jurisdictional claims in published maps and institutional affiliations.

**Copyright:** © 2021 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (https:// creativecommons.org/licenses/by/ 4.0/).

typically not explained by changes in platelet quantity, but rather by the effects of viral infections on platelet function.

Platelets are small, anucleate cells that circulate in the blood for approximately 7 to 10 days after being formed. Their main physiological role is hemostasis, forming blood clots (thrombi) to safeguard vascular integrity. Platelets originate from megakaryocytes, which are giant polyploid cells residing in the bone marrow that have themselves formed from hematopoietic stem cells. Megakaryocytes develop proplatelets that bud off numerous platelets into the blood stream, after endoplasmic maturation [24]. In individuals with normal bone marrow function, platelets circulate at levels between 150 to 450 × 109/L [25]. Megakaryopoiesis is stimulated by a number of cytokines, with Stromal Derived Factor 1 (SDF-1), Granulocyte-Monocyte Colony Stimulating Factor (GM-CSF) Interleukins (IL-3, IL-6, and IL-11) Fibroblast Growth factor 4 (FGF-4) and thrombopoietin (TPO) being the most important [26]. Whereas TPO plays a crucial role in maintenance of hematopoietic stem cells [27], most of these cytokines are proinflammatory and induce rapid maturation and activation of leukocytes, as well as stimulating megakaryopoiesis, which illustrates how platelet production is affected by inflammatory processes.

Conversely, platelets also affect the inflammatory response to viral infection and can even internalize several viruses directly. In response to infection, platelets interact with leukocytes and vascular endothelial cells before activating and secreting soluble prothrombotic and inflammatory mediators stored within granules [28]. Despite not having a nucleus, platelets do contain some RNA and maintain a limited ability for protein translation, enabling some regulation of this response [29], but also potentially supporting replication of some RNA viruses [30].

In this review, all of the literature on the relationship between platelets and viral infectious diseases published between 2010 and late 2020 has been systematically assessed and summarized in a narrative format, classified per virus category.

#### **2. Search Strategy**

The online database from the national Center of biotechnology information, Pubmed, was queried on August 3rd 2020 using the search term available in the supplemental information section included with this paper. This search was repeated on December 21st 2020, in order to include the latest publications on the currently pandemic Severe Acute Respiratory Syndrome Coronavirus 2 (SARS-CoV-2).

Additional filters were applied to restrict results to original research papers published during the previous 10 years, related to humans and with full text available. Individual papers were assessed by the first author, based on title and abstract for relevance, originality and quality and sorted based on virus species or virus family. The findings of the papers that are relevant for the subject matter of this review were presented in a narrative format within the following subgroups: "General Topics", "Arboviruses", "Blood Borne Viruses", "Rodent Borne Viruses", "Gastrointestinal Viruses", "Herpesviruses" and "Respiratory tract Viruses". Additional references were added to provide more context when necessary.

#### **3. Results**

The search yielded 413 papers. The results of the author classification process are summarized in Figure 1.

**Figure 1.** Flow diagram of PubMed search results.

*3.1. Platelets and Viral Infections: General Principles*

An overview of the mechanisms that contribute to thrombocytopenia in the main viral infections discussed in this review, including references, can be found in Table 1.

#### 3.1.1. Aggregation

Platelet agglutination or adhesion to leukocytes is often found in patients with systemic inflammatory diseases, including viral infections. Standard automated hematology analyzers are often unable to accurately detect leukocyte-bound platelets or platelet aggregates, leading to a false finding of a reduced platelet count (pseudo-thrombocytopenia). This can also be caused by drawing blood into tubes containing EDTA, the most common anticoagulant used for complete blood counts. If pseudo-thrombocytopenia is suspected, performing a manual peripheral blood smear or repeating the platelet count in using different anticoagulants, can help avoid unnecessary diagnostic procedures or transfusions [31]. Isolated thrombocytopenia should prompt investigation into chronic viral infections, such as hepatitis B, C and HIV, whereas leukocyte abnormalities and rise in other infection biomarkers, raises suspicions of an acute viral illness [32].

#### 3.1.2. Impaired Hematopoiesis

One of the most common accepted etiologies underlying virus-induced thrombocytopenia is the one where viruses can directly infect bone marrow stromal cells and hematopoietic stem cells leading to defective hematopoiesis and thrombocytopenia [33]. Furthermore, decreased platelet production can be the result of changed cytokine profiles, during infection, leading to lower TPO production in the liver, reducing megakaryopoiesis [34,35]. Finally, viruses can infect and replicate in megakaryocytes while other viruses modulate megakaryocyte function or decrease the expression of Myeloproliferative Leukemia Protein (c-MPL), the receptor for TPO, leading megakaryocyte destruction and subsequent lowered platelet production [36–38].

#### 3.1.3. Sequestration and Intravascular Destruction

Platelet destruction can occur via direct interaction of platelets with viruses. This interaction occurs via a range of receptors including Toll-Like Receptors (TLRs), integrins (GPIIb/IIIa) and c-type lectins (CLEC), that interact with the different viruses leading to platelet activation, degranulation and clearance in liver and spleen [39–41]. Upon viral infection the host defense generally induces a systemic inflammatory response, which leads to platelet activation and subsequent clearance [42]. Furthermore, platelets can bind to neutrophils, forming platelet-neutrophil aggregates, which in turn triggers the phagocytosis of platelets [43,44]. Additionally, many viruses can activate the coagulation system by induction of tissue factor leading to thrombin generation and platelet activation with subsequent platelet clearance via protease activating receptor (PAR) signaling. PARs are present on platelets, leucocytes and endothelial cells which modulate the innate immune responses [45]. Platelets can bind immunoglobulins (attached to viruses) via Fc-gamma-RII receptors leading to platelet activation, aggregation and clearance [46,47], while immunoglobulins produced by B-lymphocytes that target viruses can cross-react with platelet surface integrins (GPIIb/IIIa or GPIb-IX-V) leading to immune thrombocytopenia (ITP) [48]. In depth, more virus specific mechanisms will be discussed in the specific virus groups.

#### 3.1.4. Platelet Expression of Pattern Recognition Receptors (PRR)

PRRs such as TLRs and CLECs, [49] or messenger ribonucleic acids (mRNAs) can identify pathogen associated molecular patterns (PAMPs) from viruses and many are expressed by platelets [44]. This direct interaction of a virus or its genome with PRR can lead to platelet activation and subsequent release of chemokines. This enhances endothelial cell signaling, leucocyte migration and direct interaction and activation of leucocytes [50]. These complex interactions may have both an immune protective mechanism, [44] or be injurious to the host.

#### 3.1.5. Platelets Can Induce Inflammation and Secrete Anti-Microbial Proteins

Activated platelets undergo degranulation and release numerous inflammatory mediators, cytokines and chemokines stored in granules. Three types of granules exist: α granules, dense granules and lysosomal granules. These granules contain different molecules that can exert pro-thrombotic and immune effects leading to direct and indirect interactions with different pro-inflammatory immune cells, causing a local or systemic inflammatory milieu [51]. In addition, the α-granules can secrete platelet microbicidal peptides (PMPs) that have direct anti-viral effects, for example, it has previously been shown that synthetic PMPs have strong viricidal effects against vaccinia virus [52].

#### 3.1.6. Platelets Act as Antigen Presenting Cells (APCs)

APCs require MHC-I molecules to present antigen to CD8+ T cells. There is evidence now that platelets and megakaryocytes contain all the MHC-I and co-stimulatory molecules necessary for antigen presentation including the entire proteome [53,54]. In addition, it has been shown that platelets can successfully present ovalbumin and malarial antigens to activate CD8+ T cells [55].

#### *3.2. Arboviruses*

Arthropod borne viruses (arboviruses) are viruses that are transmitted to humans through arthropods, mainly mosquitoes and ticks. Some of these arboviruses, especially flaviviruses, potentially cause hemorrhagic fever. In this section, we will discuss the most prevalent arboviruses as well as potentially emerging arboviruses.

#### 3.2.1. Dengue Virus

With an estimated 390 million annual infections, dengue virus (DENV) is the most prevalent arbovirus worldwide [56]. Thrombocytopenia is a hallmark of severe DENV infection and platelet levels are lower in DENV infected patients compared to other febrile illnesses [57]. Severe thrombocytopenia <20 × 109/L occurs frequently in hospitalized dengue patients, and is associated with prolonged admission, plasma leakage and the presence of clinical warning signs [58,59]. Severe DENV infections with severe thrombocytopenia, hemorrhage and plasma leakage, occur more often in secondary DENV infection compared to primary DENV infection, due to antibody-dependent enhancement [60–62]. Below, the many mechanisms for thrombocytopenia in DENV disease that have been identified will be discussed.

Platelets can directly be activated by circulating DENV particles and by immune factors released during the acute phase of DENV infection. This activation induces upregulation of platelet adhesion molecules on the surface of vascular endothelial cells, causing more indirect platelet activation [63,64]. Activated platelets undergo degranulation, attach themselves to the vascular wall and form thrombi, effectively removing them from the circulation, which results in thrombocytopenia. Despite this activation, platelets from dengue patients are hyporesponsive to procoagulant stimuli in aggregometry assays, which is likely the result of exhaustion [65]. This illustrates thrombocytopenia and platelet dysfunction go hand in hand during DENV infection. The DENV NS1 protein interacts with both TLR 4 and TLR 2, expressed by platelets, leading to platelet activation, aggregation, adherence to endothelial cells and phagocytosis by macrophages [66]. DENV can also activate platelets via CLEC2, that in turn stimulate macrophages and neutrophils via CLEC5A and TLR 2. Activated neutrophils subsequently form Neutrophil Extracellular Traps (NETs), providing a scaffold for prothrombotic factors, such as platelets, red blood cells and molecules involved in both the intrinsic and extrinsic coagulation pathways. NETs also activate platelets via TLR-4, creating a positive feedback loop between NETosis and platelet activation. These mechanisms contribute to thrombocytopenia, bleeding, vascular leakage and lethality in mouse models of dengue hemorrhagic shock syndrome [66,67]. The formation of platelet-monocyte aggregates in DENV infection has also been demonstrated and correlates with thrombocytopenia and clinical signs of vascular leakage [68,69]. These platelets

alter the monocyte's innate immune response, by inducing production of IL-10 [68] and inhibiting the production of interferon α [70]. Serotonin enhances DENV-mediated platelet activation and is released by perivascular mast cells in a mouse model of DENV infection, leading to thrombocytopenia, suggesting serotonin-blocking drugs might be beneficial [71]. Platelet activating factor (PAF) is another inducer of platelet aggregation and vascular leakage elevated in plasma of acute dengue patients. A recent small phase II randomized placebo-controlled clinical trial showed that Rupatadine, a licensed antihistamine with PAF receptor blocking activity, is safe in acute dengue patients, but not clinically beneficial, with no effect on nadir platelet counts [72].

In response to DENV infection, the immune system also depletes platelets through direct cytotoxic effects. This occurs through complement factor C3 binding to the platelet surface and opsonization of platelets by DENV NS1 specific IgG, leading to subsequent phagocytosis by macrophages [73,74]. The observation that patients with severe DENV infection have increased afucosylated IgG1 antibodies and that these antibodies can crossreact with platelet antigens, further supports the notion that platelet-NS1 cross-reactive antibodies contribute to the depletion of platelets during severe DENV infection [75]. Infusion of anti-NS1 IgG cannot elicit thrombocytopenia in mouse models. However, it does enhance thrombocytopenia when administered during DENV infection. This could indicate that platelet binding and destruction is mediated by IgG bound NS1 dimers, rather than a direct interaction between cross-reactive IgG and platelets. This could have implications for vaccine design [76,77]. Platelets upregulate their expression of Human Leukocyte Antigen (HLA) class I molecules in the presence of DENV, suggesting a possible role in viral antigen presentation and T cell mediated cytotoxicity as a mechanism for thrombocytopenia [64].

A third mechanism is direct infection of platelets and megakaryocytes by DENV, leading to viral replication, cell lysis and impaired production of platelets. DENV infects platelets by binding DC-SIGN (CD209) and heparan sulfate proteoglycan (HSP) on the platelet surface and indirectly via Fc receptor FcΥR2A (CD32), after binding of antibodies to the virus particle [30]. The presence of DENV specific antibodies is not required for platelet infection [78]. Especially during the viremic stage and in those patients with more severe dengue, less DC-SIGN and FcΥR2A expression is detected on platelets when compared to patients suffering from other febrile illnesses. The relation between this phenomenon with thrombocytopenia has not been explored however and it is unclear whether the presence of DENV particles in the blood could interfere with the binding of the detection antibodies. There could also be a survival advantage of platelets with low DC-SIGN and FCΥR2A expression [79]. DENV productively infects a megakaryoblast-like cell line, using Glycoprotein Ib (GPIb) and was found circulating megakaryocyte-like cells in Rhesus Macaques [80]. This suggests that DENV can infect mature megakaryocytes and impair platelet production and survival through replication, although this has not been confirmed in vivo [81].

#### 3.2.2. Chikungunya virus (CHIKV)

Despite sharing many clinical characteristics with DENV, thrombocytopenia has only been occasionally described in CHIKV infected patients [82]. In a cross-sectional study of arbovirus infections in Pakistan, thrombocytopenia was observed in 18% of CHIKV infections, compared to 74% in DENV. Furthermore, thrombocytosis was significantly associated with CHIKV infection (OR 2.2) [83].

#### 3.2.3. Crimean Congo Hemorrhagic Fever (CCHF)

CCHF is a tick-borne virus from the Bunyaviridae family that is associated with pronounced thrombocytopenia, accompanied by bleeding complications [84]. CCHF cases with bleeding have lower platelet counts and slightly raised platelet distribution width compared to those who do not bleed [85]. A lower platelet lymphocyte ratio (PLR) upon presentation to hospital care is predictive of adverse clinical outcomes in CCHF patients [6].

#### 3.2.4. Japanese Encephalitis Virus (JEV)

This vaccine preventable flavivirus has a wide distribution in eastern Asia. Although central nervous system infection has a substantial mortality rate of 20–30%, most cases are either asymptomatic or mild, without clinically overt signs of encephalitis [86]. A recent prospective study performed in hospitalized patients with dengue-negative febrile illness in Indonesia found that 6% had serological evidence of recent JEV infection. Thrombocytopenia was common during the acute phase of illness in these non-encephalitic cases (69%) but did not occur as frequently as dengue cases from the same study (92%) [87].

#### 3.2.5. Severe Fever with Thrombocytopenia Syndrome (SFTS)

SFTS is an emerging infectious disease caused by a recently discovered Bunyavirus that, as the name suggests, is associated with profound thrombocytopenia [88]. DC-SIGN, which is expressed by human platelets, was identified as a receptor for Bunya viruses, but no studies have shown virus entry or replication of SFTS in platelets [89]. A recent detailed study in Chinese SFTS patients revealed a severity-dependent depletion of the essential amino acid arginine due to arginase released by granulocytic Myeloid-derived suppressor cells (gMDSC's), which are recruited during the acute stage of the disease. The resulting lack of both arginine and its metabolite, nitric oxide, are believed to disinhibit platelet activation, leading to platelet aggregation, destruction and thrombocytopenia. A randomized, noncontrolled clinical trial was performed were arginine supplementation with best supportive care was compared with supportive care alone. Whilst not demonstrating a statistically significant survival benefit, platelet counts returned to normal more rapidly in the treatment arm [90].

#### 3.2.6. Tick-Borne Encephalitis Virus (TBEV)

TBEV is a flavivirus that is transmitted through bites of ticks and is prevalent in Europe and northeastern Asia [91]. After initial viremia, the central nervous system may become infected, which can result in severe neurological damage and occasionally death. Like most other viruses that can enter the bloodstream, TBEV causes thrombocytopenia, albeit generally mild and not typically associated with bleeding. A case–control study in patients with suspected and confirmed central nervous system infections, reported a decreased mean platelet count in TBE cases versus neuroborreliosis cases. However, the mean platelet count (173.8 × <sup>10</sup>9/L) remained above the lower limit of normal in the TBE group. In addition, platelet counts correlated positively with concentrations in serum of IL-23, a cytokine secreted by dendritic cells, which is believed to stimulate essential hostdefense mechanisms against viruses [92]. This could indicate a role for IL-23 in stimulating megakaryopoiesis during TBEV infection.

#### 3.2.7. Viral Hemorrhagic Fevers (VHF)

The extreme containment precautions required to study the most highly pathogenic VHF's, such as Ebolavirus, Marburgvirus and Lassavirus, makes studying the interactions of live virus with human platelets an expensive and labor intensive endeavor. Only a small number of laboratories worldwide are equipped with biosafety level 4 facilities and only two are located on the African continent, where most VHF infections occur [93]. Lymphocytic Choriomeningitis Virus (LCMV) is an arenavirus, which has sporadically caused severe Lassavirus-like illness in humans. Its reservoir host is mice, who only develop very mild disease. Experimental infections of mice with LCMV, after platelet depletion treatment, however, result in a more severe VHF with uncontrolled viral replication and dissemination, similar to that observed in humans. The platelet-depleted mice had impaired LCMV specific CD8+ T Cell responses. Severe disease and mortality only occurred in mice whose platelet reduction treatment was initiated shortly before LCMV infection. This may suggest that the innate antiviral response against LCMV requires platelets. However, Interferon α and β production appeared to be unaffected by platelet depletion treatment, indicating this part of the innate immune response had remained intact. Further examination of the spleen in LCMV infected mice who underwent platelet depletion demonstrated extensive disruption of the splenic architecture and cellular necrosis, which could be the common mechanism of both the defective innate and T cell responses. These experiments provided novel insights in the role of platelets in controlling VHF through protection of splenic vascular integrity and the importance of this organ in mounting sufficient cellular and innate immune responses to eliminate the virus [94].

#### 3.2.8. West Nile Virus (WNV)

West Nile Virus (WNV) is a neurotropic flavivirus that mainly causes asymptomatic infections, while severe complications such as meningitis or encephalitis can occur, but are rare. WNV infections with hemorrhagic complications have sporadically been reported but severe thrombocytopenia and hemorrhage typically do not occur during WNV infection [95]. A case of WNV transmission through a platelet transfusion unit from a donor that tested negative on whole blood on the day of transfusion has been described, raising the possibility that the virus may concentrate in platelets, although the platelet unit itself was never tested [96].

#### 3.2.9. Yellow Fever Virus (YFV)

Yellow fever virus (YFV) is a flavivirus that can be transmitted through Aedes mosquitoes and other vectors and is endemic in Africa and the Americas but not in Asia [97]. Even though there is an effective vaccine against YFV, annually several tens of thousands of persons die of YFV infection in Africa [98]. Disseminated yellow fever is characterized by hepatitis and hepatic failure, with resulting thrombocytopenia, deficiencies in plasma coagulation factors, prolonged activated Partial Thromboplastin Time (aPTT) and an elevated International Normalized Ratio (INR) [99,100]. A Brazilian retrospective cohort study of patients suffering severe YFV infections, showed severe hemorrhagic complications, mostly from the gastrointestinal tract. Despite this, only mild thrombocytopenia was present, with a median platelet count of 74 × 109/L, indicating platelet consumption as a result of hemorrhage, rather than thrombocytopenia-induced bleeding [100]. This is supported by the presence of ischemic and hemorrhagic microvascular pathology upon fundoscopic examination of hospitalized yellow fever patients, which correlate with the degree of thrombocytopenia present, disease severity and markers for renal and hepatic disease [101].

#### 3.2.10. Zika Virus (ZIKV)

Like DENV, ZIKV is a flavivirus that is mainly transmitted through Aedes mosquitoes. In 2015/2016, there was a large outbreak of ZIKV in the Americas, which led to an increase in the incidence of Guillain-Barré syndrome in adults and congenital abnormalities in newborns [102]. Seroprevalence studies indicate that, depending on the country and location, up to 60% of inhabitants got infected with ZIKV during this outbreak [103–105]. A large prospective cohort study of ZIKV infected patients in Puerto Rico identified thrombocytopenia (defined in this study as a platelet count below 100 × 109/L) in 1.2% of confirmed cases. Only 25% of those had platelet counts below 20 × 109/L and 16% had etiologies other than ZIKV infection [106]. Platelet counts below 150 × 109/L are present in a small minority of ZIKV infections, although profound thrombocytopenia has been described in fulminant cases, associated with bleeding, liver failure and other coagulation disorders [107,108]. Most ZIKV patients have platelet counts between 150 and <sup>300</sup> × <sup>10</sup>9/L [109–111]. Thrombocytopenia on first diagnosis was significantly associated with higher odds of hospitalization in a case series study of U.S. veterans with laboratoryconfirmed or presumptive ZIKV infection (OR 6.4, 95% CI 4.0–10.1) [3]. Suspected diagnosis of de novo Immune Thrombocytopenic Purpura (ITP) has been reported in patients presenting with severe thrombocytopenia during ZIKV convalescence, who responded to treatment with corticosteroids, intravenous immunoglobulin (IVIG), or both [112–114]. Exacerbation of pre-existing ITP during the acute phase of ZIKV disease has also been

described [115]. ZIKV infects monocytes, macrophages and dendritic cells, but is unable to propagate in megakaryocyte-differentiated human hematopoietic stem cells and platelets [116]. Thrombocytopenia presents more often in DENV infections compared to ZIKV and can be used to distinguish the two otherwise similar diseases to a limited extent [117–119].

#### *3.3. Blood-Borne Viruses*

#### 3.3.1. Hepatitis B and C (HBV and HBC)

Chronic hepatitis B and C virus infections are associated with hepatic cirrhosis, portal hypertension, liver failure and hepatocellular carcinoma (HCC). Thrombocytopenia is a predictor of such adverse outcomes, especially when combined in clinical scoring systems with other parameters such as age, gamma-glutamyl transpeptidase, Alanine and Aspartate aminotransferase [7–10,13–15,17,19,120–122] However, in individuals that have already developed HCC, thrombocytosis carries a negative prognosis rather than thrombocytopenia. Elevated platelet counts, when combined in a ratio with lymphocyte counts (PLR), is a biomarker of malignant inflammation and elevation is predictive of a poor prognosis in patients with HCC [123,124]. However, this marker does not appear to have much added value when tumor size and histological parameters are known or in patients with either no cirrhosis of very severe cirrhosis [11,125].

A likely explanation for the lower platelet counts observed in patients with hepatic cirrhosis, including those with a non-viral etiology, is portal hypertension leading to increased sequestration of platelets and red blood cells in the liver and spleen [126].

Another factor more specific to HBV and HCV infection could be increased platelet activation and consumption due to chronic inflammatory processes caused by these viruses. Mean platelet volume (MPV), a biomarker of platelet activation, was found to be increased in patients with chronic hepatitis B infection, compared to controls with no documented HBV infection [127]. Chronic HBV infection is also associated with a low platelet response to clopidogrel, which could be due to platelet activation by the virus or changes in drug metabolism caused by liver disease [128].

HBV and hepatitis Delta (HDV) co-infection also appears to lower platelet counts compared to HBV mono-infection, independently of the severity of liver disease, though the mechanism is unknown [129].

In hepatitis C infection, intravascular destruction of platelets is likely also mediated by anti-platelet antibodies and ITP related to HCV has been reported in the literature [130]. In HCV infected patients, platelet auto-antibodies were more commonly identified as a cause in mild (platelet count 126–149 × 109/L) thrombocytopenia, whereas decreased megakaryopoiesis was a more prominent contributing factor in more severely (platelet count < 100 × <sup>10</sup>9/L) thrombocytopenic patients [131].

Since both hepatitis B and C are blood-borne pathogens, a direct interaction between these viruses and platelets is possible. Hepatitis C virus RNA appears more stable when incubated with platelets than with platelet-free plasma, although there is no evidence of replication [132]. Instead, platelets may shield the virus from complement, interferons, cytokines, antibodies or other antiviral factors present in serum, perhaps through internalization. There have been reported cases of HCV infected patients with much higher viral loads in the platelet compartment compared to plasma [130]. One study found that, while HCV viral load was higher in plasma compared to platelets in patients before interferon/Ribavirin treatment, some patients had detectable RNA in platelets after treatment, but not in serum, resulting in relapse [133]. This could indicate that platelets and their megakaryocyte precursors can provide a site for therapeutic drug and immune evasion, either perhaps due to diminished response to interferons or a relative impermeability for nucleoside analogues or both, while extrahepatic replication is of minor importance during untreated HCV infection. The therapeutic relevance of this finding today is questionable however, given the high curation rate that can be achieved using the novel direct antiviral

agents that have become available. The effect of direct interactions between HCV and platelets on platelet lifespan remains to be established.

#### 3.3.2. Human Immunodeficiency Virus (HIV)

Thrombocytopenia is a common finding in HIV infection and HIV testing is part of a routine evaluation for unexplained thrombocytopenia. Screening individuals with thrombocytopenia and other so-called indicator conditions, has been shown to be a costeffective and more efficient compared to universal screening [20–22]. HIV as the underlying cause for otherwise unexplained thrombocytopenia is frequently missed [23]. Decline of platelet count over time in HIV patients has been associated with development of dementia and reduced gray matter volume on MRI scans, although univariate and multivariate analyses were not entirely consistent [134]. Several mechanisms for thrombocytopenia in HIV infected individuals have been proposed and these will be discussed in more detail below.

Chronically infected HIV patients, even when adequately suppressed with combination Anti-Retroviral Therapy (c-ART), have a substantially increased risk of developing cardiovascular disease and deep venous thrombosis compared to HIV negative individuals. Platelet activation is believed to be a major driving factor behind this phenomenon. Untreated HIV positive individuals have elevated platelet activation markers in plasma compared to matched healthy controls, which positively correlate with viral load and negatively with CD4+ T cell counts. In addition, platelets from HIV patients express more oxidative stress-related proteins [135]. Studies investigating platelet morphology (MPV, PDW) in HIV patients, report contradictory results [136,137]. Platelets can bind HIV-1 virus through DC-SIGN and serve as a carrier, protecting virions in the circulation from antiviral factors through formation of RBC-platelet aggregates [138]. Platelets may also become activated after binding to HIV and release platelet factor 4 (PF4/CXCL4), from their α granules, which has been shown to inhibit cell attachment and entry of several HIV-1 strains through binding of its main envelope glycoprotein, GP-120 [139]. PF4 and other α-granule cytokines such as CCL5 likely also contribute to a chronic proinflammatory state in HIV patients. Interestingly, HIV patients on c-ART who become infected with DENV, were found to have milder disease and their platelets released less proinflammatory PF4 and CCL5 from α granules, likely resulting from previous HIV-associated depletion [140].

Very recently, platelets from long-term virologically suppressed HIV-infected individuals were shown to contain infectious HIV-1, demonstrating its ability to infect platelets and remain viable. Both viral RNA and proviral DNA of HIV was detected in bone marrow megakaryocytes from these same individuals, indicating platelets may already be infected upon formation. Platelets phagocytosis by macrophages was shown to lead to productive infection in this cell type [141]. The effects of HIV-1 infection on the activation status or lifespan of platelets were not investigated, meaning the role this mechanism plays in thrombocytopenia in people living with HIV remains unclear.

Another factor contributing to the prothrombotic state of HIV patients is the induction of platelet-monocyte aggregates (PMA) by the virus, which causes mutual activation of these two cell types [142,143]. Monocytes expressing CD16 (nonclassical and intermediate monocytes) appear to be primarily involved in the formation of PMA [144]. Electron microscopy (EM) studies of PMA in HIV infected individuals show a maximum of 4 simultaneously attached platelets per monocyte [145]. However, a single monocyte is likely capable of transiently interacting with many more platelets during its lifetime, leading to platelet activation and cumulative loss over time. The formation of PMA is mediated by binding of P-selectin on the platelet surface with P-Selectin Glycoprotein Ligand-1 (PSGL-1) on the monocyte surface. This leads to monocyte activation, including secretion of proinflammatory cytokines and Tissue Factor (TF), which initiates the extrinsic coagulation pathway. Monocytes may be able to downregulate PSGL-1 expression in response to platelet binding, but this has not been definitively established [143]. HIV transactivating factor (Tat) interacts with platelet integrin β3 and CCL3, resulting in the secretion of soluble

CD40 ligand (sCD40L), a platelet activator [146]. Subsequently, activated platelets initiate the formation of PMA [144,145]. The CD40L released, also increases B cell activation and secretion of immunoglobulins in vitro and in vivo in mouse models, which has been associated with ITP in non-HIV infected individuals and could play a similar role in HIV positives [147]. Platelets isolated from HIV patients secrete significantly more sCD40L in response to stimulation compared to healthy controls. The same effect was seen for other proinflammatory molecules, such as CCL5 and P-selectin expression. Patients who are not on c-ART can also have diminished responses however, which is attributed to exhaustion [148]. Overall, this evidence suggests that HIV infection sensitizes platelets to procoagulant and proinflammatory stimuli, which in response release more procoagulant and proinflammatory factors and induce monocytes to do the same. This creates a positive feedback loop underlying chronic inflammation and risk of thromboembolic complication. Although initiation of c-ART normalizes both platelet and monocyte activation markers in plasma within 12 weeks [149], these studies using isolated platelets and monocytes demonstrate that virological suppression seems unable to completely disrupt this inflammation-coagulation cycle present in people living with HIV.

#### *3.4. Rodent-Borne Viruses*

#### Hantavirus

Hemorrhagic Fever and Renal Syndrome *(HFRS)* caused by Eurasian hantaviruses is associated with severity-dependent thrombocytopenia. The clinical characteristics of the disease include both hemorrhagic and thrombotic complications. This suggests that platelet consumption plays a significant role in the pathogenesis of thrombocytopenia. A longitudinal study of 35 hospitalized HFRS patients with Puumala virus (PUUV) infection revealed a biphasic pattern in platelet counts over time. This started with marked thrombocytopenia upon first presentation to the clinic, followed by a rise to just below the upper limit of normal approximately 2 weeks after symptom onset, and subsequent normalization of platelet counts. Elevated plasma TPO levels, reticulated platelet counts and MPV (each peaking at the platelet count nadir) suggest a compensatory myeloproliferative response to an acute loss of platelets within the circulation during early infection. Ex vivo expression of platelet surface activation markers was higher during the acute disease stage compared to follow-up, although the opposite was found for in vitro stimulation assays, indicating platelet exhaustion. Finally, patients who had signs of Disseminated Intravascular Coagulation (DIC) or thrombosis during the disease had higher plasma platelet activation markers, such as soluble P-selectin and soluble Glycoprotein IV (GP IV) compared to those who had not [150]. Another longitudinal study measured Von Willebrand Factor, Fibrinogen, fibronectin and A Disintegrin Additionally, Metalloproteinase with a ThromboSpondin type 1 domain (ADAMTS13) concentrations in plasma of PUUV infected patients during the acute and recovery phase. These factors where indeed elevated during acute disease when compared to recovery. Thrombocytopenia was present in 15 out of the 19 patients studied, but the exact numbers and how they correlate with the markers measured in plasma is not described [151]. Furthermore, during the acute phase of PUUV infection platelet aggregation appears impaired, especially when induced with thrombin, when tested on impedance aggregometry. Platelet adhesive mechanisms on collagen are intact, despite thrombocytopenia, while thrombopoiesis is active [152]. Potential mechanisms explaining the decrease in platelet count based on In vitro data include decreased production due to bone marrow invasion and megakaryocyte infection [153] and binding of platelets to infected endothelial cells [154].

#### *3.5. Gastrointestinal Tract Viruses*

#### 3.5.1. Enteroviruses

Since the near-eradication of poliomyelitis, severe enterovirus infections have become rare, typically causing mild cold-like illness in children and hand-foot and mouth disease. Coxsackievirus infections occasionally cause myocarditis in adults, and Enterovirus 68

has recently been implicated in episodes of acute flaccid myelitis in children. Thrombocytopenia has been described in neonatal cases of coxsackievirus B3 (CoxV B3) infections, both in Japan, with one case being related to secondary Hemophagocytic Lymphohistiocytosis (HLH) triggered by CoxV B13 [155,156]. A cross-al study of neonatal cases with severe enteroviral infection in Japan demonstrated significantly decreased platelet and WBC counts in Human Parechovirus 3 infections compared to RSV infected controls. In addition, the Human Parachovirus 3 infected patients showed elevated plasma ferritin and lactate dehydrogenase (LDH) levels, both compared to the RSV controls and infants with other enterovirus infections. This may suggest an HLH-like illness secondary to this viral infection [157]. Human platelets express the Coxsackie-Adeno Receptor (CAR), aCoxVB3 cell entry receptor. Whether this mechanism plays a role remains unclear, in particular as the virus appears to be unable to replicate in platelets in vitro. However, platelets become activated after incubation with CoxVB3, increasing their expression of P-selectin and showing signs of apoptosis (i.e., increased phospatidylserine (PS) expression) [158].

#### 3.5.2. Rotavirus (RotV)

Viral gastroenteritis is a major cause of child mortality in the developing world and a significant burden on the healthcare system in developed countries [159]. Outbreaks of norovirus and rotavirus frequently occur in childcare institutions and nursing homes, where those most vulnerable to dehydration are affected. These infections are typically limited to the gastrointestinal epithelium and rarely cause severe systemic disease with involvement of multiple organs or severe inflammation. It is therefore not surprising that thrombocytopenia is not a dominant clinical feature of viral gastroenteritis. Mean platelet counts in children with rotavirus gastroenteritis are reported within the normal range and not diverging from children with other viral causes of gastroenteritis [160]. It is important to recognize both concentration and dilution effects of intravascular fluid shifts when assessing platelet counts or other blood cell counts in patients with conditions such as severe gastroenteritis or other critical illnesses.

#### *3.6. Herpesviruses*

#### 3.6.1. Cytomegalovirus (CMV)

CMV is present in nearly all humans and remains latent for life after primary infection, which is usually asymptomatic in immunocompetent hosts. In contrast, severe immune suppression can lead to reactivation in later life leading most notably to CMV-mediated enterocolitis, hepatitis, retitinis or encephalitis [161]. However, very little is known about platelet counts during primary infection in immunocompetent hosts, because CMV infection comes to the attention of clinicians only when these rare complications develop. In vitro, CMV has been shown to interact with platelets through TLR-2. However, rather than directly resulting in platelet aggregation, these activated platelets produce a proinflammatory response, form aggregates with leukocytes and increase their adhesion to human vascular endothelial cells (HUVEC). Thus, in this model of CMV infection, platelets act as an intermediary between the virus and circulating immune cells [39].

#### 3.6.2. Epstein Barr Virus (EBV)

Like all other human herpesviruses, infection with EBV occurs in the majority of the population at an early age. This leads frequently to a mild but sometimes protracted viral symptomatic episode called infectious mononucleosis, which is followed by lifelong latency. Occasionally reactivation occurs in immunocompromised hosts, most commonly in organ transplant recipients. Recently, primary EBV infection has been associated with a variety of auto-immune diseases, whereas latent EBV infection and reactivation plays a role in the pathogenesis of Hodgkin's lymphoma and B and T cell lymphoma's, through poorly understood mechanisms. EBV infection is also a well described trigger for secondary Hemophagocytic Lymphohistiocytosis (HLH) [162].

In the case of EBV reactivation, the relation with lymphoproliferative disorders is important to keep in mind when evaluating a patient with thrombocytopenia, especially when other cell-lineages are involved.

In cases of primary EBV infection, thrombocytopenia and hemolytic anemia are occasionally also found and have been associated with the presence of platelet and erythrocyte auto-antibodies. Typical of primary EBV infection is the production of heterophile antibodies by naïve B cells that have become infected with latent-phase EBV [162]. Some of these antibodies may be autoreactive and bind to platelets, leading to their destruction. Due to EBV's restriction to human hosts, well established animal models to study viral-platelet interactions in vivo do not exist. However, experimental infections with the related murine gammaherpesvirus 68 (γHV68) produce a mononucleosis-like illness in mice. This shows a significantly reduced platelet count during the early latent replication phase (nadir 17 days post infection). In this model, thrombocytopenia was found to be the result of antibodies induced by the infection and depended on viral latency, supporting the notion that polyclonal antibodies produced by latently infected B cells include autoantibodies against platelets [163]. This mechanism appears to be unique to EBV infection and is separate from auto-antibodies induced by other viral infection, which is believed to be the result of molecular mimicry between viral and self-antigens.

#### 3.6.3. Human Herpesvirus 6 (HHV-6)

HHV-6 causes a near universal childhood illness, exanthema subitum, before entering its latent stage. Reactivation is rare, and generally only occurs during profound immunosuppression, such as during allogenic hematopoietic stem cell transplantation. In this population, HHV-6 reactivation (defined as a positive PCR on blood samples) was significantly associated with delayed platelet engraftment and the development of graft versus host disease (GVHD) [164].

#### 3.6.4. Varicella Zostervirus (VZV)

Primary VZV infection almost universally presents itself as a self-limiting childhood illness, with more significant sequalae emerging later in life, ranging from common herpes zoster to rare cases of severe disseminated disease. The latter is typically only found in immunocompromised hosts, although not exclusively [165]. Typical Herpes Zoster manifests itself as a vesicular cutaneous eruption restricted to one dermatome and is most frequently seen in the elderly and patients who received chemotherapy for solid or hematological malignancies. Herpes Zoster can have long-term sequalae, such as post-herpetic neuralgia, but does not cause systemic disease and patients usually have normal platelet counts [166]. Profound thrombocytopenia has been described in reports of disseminated VZV infection, combined with DIC, hemorrhaging, ischemic strokes, ileus, abdominal pain, hepatitis, meningoencephalitis and vasculitis [167–170]. The vasculitis is believed to be caused by VZV infection of the arterial walls themselves and can be found in arteries in various organs, including smaller cerebral arteries, where it is associated with stroke [171]. Case series describing VZV related strokes report elevated platelet activation markers, such as PF-4 and β-thromboglobulin levels in some patients [172]. Splenomegaly with associated hypersplenism is a common feature of systemic herpesvirus infections, which contributes to thrombocytopenia and sometimes leads to splenic rupture [173]. The differential diagnosis of thrombocytopenia during a VZV infection is broad, because of comorbidity-related immune suppression, and includes immune thrombocytopenia, drug induced thrombocytopenia [174] and bone marrow dysfunction, particularly if other lineages are affected. First presentation or relapse of ITP has been reported during primary VZV infection in adulthood [175,176]. A platelet count <200 × 109/L was found to be predictive of a poor outcome in patients suffering from Ramsey Hunt Syndrome [2]. A large prospective cohort study identified thrombocytopenia as an independent risk factor for ICU admission in hospitalized children with VZV infection, although the rate of underlying

hematological comorbidities and bacterial coinfections was high, suggesting VZV was not the sole cause of this phenomenon [177].

#### *3.7. Respiratory Tract Infections*

A platelet count close to the lower reference limit is a common finding in more severe viral respiratory tract infections [178]. Data on platelet counts in mild respiratory infections are scarce, possibly due to the fact that these cases typically do not present to care and blood counts are rarely performed. Interestingly, the literature reports thrombocytosis in infants hospitalized with respiratory tract infections, especially RSV and rhinoviruses, with platelet counts decreasing with age [179–181]. Platelet counts in patients with acute exacerbations of heart failure who tested positive for respiratory viruses by PCR, did not differ significantly from those who tested negative [182]. The lungs have recently been found to host resident megakaryocytes, which contribute to platelet production [183]. Investigating the interactions between respiratory viruses and platelets could be key to understanding the high rate of thromboembolic complications that arise during viral acute respiratory distress syndrome (ARDS).

#### 3.7.1. Adenoviruses (Adv)

Among viruses causing mild upper respiratory tract infection, adenoviruses appear to be most studied in relation to platelets. Coxsackie and Adenovirus receptor (CAR) is the receptor adenoviruses use for cellular attachment. Expression of this receptor has been reported in healthy human platelets, albeit at a very low frequency (3.5%) [184]. In vitro studies where platelet rich plasma was incubated with very high concentrations of human adenovirus 3 and 5, showed a moderate increase in platelet aggregation and platelet activation marker expression, with uptake of adenovirus 5 by platelets demonstrated using EM [185,186]. Since natural infection in humans is unlikely to expose platelets to the high viral titers used in these incubation experiments, the clinical relevance for this finding is mostly related to the potential future use of adenoviruses for gene therapy purposes. Indeed, cancer patients experimentally treated intravenously with oncolytic adenovirus where serially sampled to determine relative abundance of viral DNA in various blood cell populations. Although very little platelet-associated virus was found in vivo, in vitro experiments where whole blood was incubated with the studied adenoviruses revealed a large proportion of virus bound to platelets. Given the thrombocytopenia observed during adenovirus-based treatments, this discrepancy could be the result of a survival disadvantage of adenovirus bound platelets in the circulation [187].

#### 3.7.2. Influenza Virus (IAV/IBV)

Influenza virus infection is associated with a severity-dependent thrombocytopenia. Pediatric outpatients with confirmed IAV or IBV infection showed slightly, though significantly, lower mean platelet counts compared to asymptomatic controls. Children with influenza-like illness who were IAV and IBV PCR negative had platelet counts in between the confirmed positive and healthy groups, and platelet counts could not reliably distinguish between influenza positive and negative children [188]. In adults, severe influenza infection is accompanied by an increased risk of pulmonary thromboembolisms and cardiovascular events (Sellers, 2017 #1) suggesting platelet activation occurs during infection. Whole-blood transcriptome studies have found gene expression signatures in patients during H1N1 infection that are associated with a poor response to antiplatelet agents. Conversely, patients undergoing coronary catheterization that had a gene expression signature associated with viral infection, where more likely to have a confirmed myocardial infarction compared to those that did not express this signature [189]. Pathogenic H3N2 and H1N1 strains are capable of infecting pulmonary vascular endothelial cells, which increases platelet adhesion to both infected and nearby uninfected cells through interaction between endothelial fibronectin and platelet integrins [190]. Various influenza A strains cause thrombocytopenia in experimentally infected ferrets, with highly pathogenic strains

(H5N1) showing a stronger decrease compared to moderate (H1N1) or mildly pathogenic (H3N2) strains. In addition, these viruses are capable of directly infecting platelets in vivo through binding of sialic acids on glycans on their cell surface. EM imaging has demonstrated the ability of platelets to phagocytose influenza virus particles. This infection of platelets results in their activation, aggregation and subsequent clearance from the circulation. Interestingly, desialylation of platelet glycans by viral neuraminidase is hypothesized to reduce the lifespan of affected platelets through increased hepatic clearance [191]. Influenza virus can also interact with platelets through TLR7, which leads to the formation of platelet-neutrophil aggregates and neutrophil NETosis, through complement (C3) secreted by the platelets [192]. Immune-complexes of antibodies against influenza virus are also capable of activating platelets through an interaction with the Fc-ΥIIA receptor present on the platelet surface, leading to thrombocytopenia in a humanized mouse model. These findings, combined with reported influenza vaccine induced ITP, point to a link between influenza virus-specific adaptive immunity and thrombocytopenia [193–195].

#### 3.7.3. Measles Virus (MV)

Likely the most contagious virus known to affect humans, this virus first infects the respiratory tract and subsequently spreads to lymphoid organs, infecting lymphocytes, including memory B and T cells [196]. A highly effective vaccine has been available for several decades, yet immunization programs have not been able to reach sufficient coverage to eradicate the disease, leading to sporadic outbreaks [197]. Studies published in the past 10 years describing natural infection in adults report mild leukocytopenia and thrombocytopenia as a frequent finding, occasionally with minor bleeding complications, but no thromboembolisms [198–200]. A link between Subacute Sclerosing Panencephalitis (SSPE), a late complication of Measles caused by persistence of MV in the brain, and ITP has been proposed, based on the co-occurrence of both extremely rare diseases in 3 pediatric cases [201]. This is further supported by an increased incidence of ITP after MV vaccination, where platelet binding anti-MV (and anti-rubella) IgG and IgM was demonstrated [202].

#### 3.7.4. Parvovirus B19 (PVB-19)

While best known as a mild, self-limiting childhood illness (fifth disease), PVB-19 can occasionally cause more severe disease, especially during pregnancy, resulting in hydrops fetalis. Due to its tropism for erythroid progenitor cells and megakaryocytes, fetal PVB-19 infection causes severe anemia and thrombocytopenia, requiring Intrauterine Transfusion (IUT) of platelets and erythrocytes in some cases [203,204]. While most severe and best described in fetal infections, PVB-19 can also cause thrombocytopenia, anemia, leukopenia or pancytopenia in children and adults [205–208]. A retrospective cohort study reports PVB19 infection in children undergoing chemo- and radiation therapy for non-hematological malignancies increases the risk of thrombocytopenia and transfusion of blood products [209]. A similar study comparing malignant and nonmalignant hematological disease in a pediatric population found that PVB-19 DNA positivity was not associated with a higher risk of transfusion, but the number of platelet transfusion units administered per patient was over 3-fold higher in PVB-19 DNA positive patients [210]. PVB-19 is also able to infect myocardial tissue, leading to clinical myocarditis and dilated cardiomyopathy. This raises the question whether PVB-19 is also capable of infecting vascular endothelium and cause vasculitis and platelet adhesion to infected vessel walls. A case–control study exploring differences in microparticle (MP) profiles in the peripheral circulation of patients with myocarditis caused by PVB-19 versus other causes, found significant increases in apoptotic endothelial, platelet and leukocyte-derived MPs in PVB-19 mediated disease. This suggests that, in addition to impaired hematopoiesis, PVB also causes platelet destruction and vascular damage [211]. In vitro studies suggest PVB-19 nonstructural protein 1 (NS1) causes endothelial activation, upregulation of adhesion molecules and an increase in platelet and monocyte binding [212].

#### 3.7.5. Respiratory Syncytial Virus

Severe respiratory infections with RSV occur mainly in children, the immunocompromised and those with underlying pulmonary disease [213]. In contrast to other respiratory infections, thrombocytosis rather than thrombocytopenia appears to be a common phenomenon found during acute RSV disease [181]. In vitro experiments demonstrate a reduction of monocyte RSV infection when platelets are added to the culture, possibly by binding and internalization of RSV. Platelets increase surface P selectin expression in the process, but why this would lead to thrombocytosis rather than thrombocytopenia is unclear [214].

#### 3.7.6. SARS Coronavirus 2 (SARS-CoV-2, COVID-19)

The literature cited in this part of the review was updated shortly before submission to include the high volume of scientific work that has been published on this virus, which has caused a pandemic of severe pneumonia of historical proportions. Besides bilateral pneumonia, critical COVID-19 cases are characterized by multi-organ disease [215], and a remarkably high incidence of pulmonary embolisms [216,217]. Several mechanisms involving hypercoagulability and inflammation interact resulting in thrombotic phenomena both in the microvasculature and in the larger, mostly pulmonary blood vessels [218].

In fact, upon autopsy these embolisms were found to be mainly composed of platelets, fibrinogen and neutrophils [219–221]. Another typical finding during autopsy of deceased COVID-19 patients is the presence of widespread microvascular thrombosis in both pulmonary and extrapulmonary vessels, including in patients without true thromboembolisms, indicating a systemic prothrombotic state [220].

A low to low-normal platelet count is present during peak symptomatic illness, with increased MPV and PDW, and expression of surface activation markers [222,223]. However, one study identified subpopulations of platelets with a downregulated phenotype, which were highly enriched in severe, but not in intermediate COVID-19 cases, suggesting exhaustion of circulating platelets [220]. The total platelet population from these patients still showed hyperresponsiveness to procoagulant stimuli in vitro, likely driven by a hyperactive minority that was also present. This hyperresponsiveness was also found in other studies [224]. Some clinical studies report that thrombocytopenia is associated with increased mortality [4,5], whereas other do not [215,225]. This discrepancy might depend on disease severity, comorbidities or the type of care provided. For example, a welldefined cohort of mechanically ventilated critically ill patients showed that daily platelet concentrations were not associated with intensive care unit survival [215]. However, this observation does not exclude a role for platelet (dys)function in immunothrombosis. Other coagulation-related markers during acute illness show strongly elevated levels of D-dimers and fibrinogen degradation product (FDP), normal to slightly prolonged PT, APTT, elevated plasma viscosity and coagulability and normal to mildly increased INR [5,12,222,223]. Whether these markers have diagnostic or prognostic value requires investigation and might differ along the course of infection depending on disease severity, comorbidities and type of care provided. Platelet counts appear to rise slowly over the course of the disease, which coincides with a sharp peak in IL-6, suggesting this cytokine may play a role in the thrombopoietic response [220]. Plasma TPO levels are elevated in severe COVID-19 patients, but gene expression of its receptor, c-MPL is decreased, suggesting desensitization of the bone marrow as an additional mechanism for thrombocytopenia in COVID-19. When thrombocytopenia is present, it is often accompanied by relative deficiencies in other myeloid and lymphoid cell lineages [226,227], which could indicate bone marrow displacement caused by a proliferative response to hyperinflammation, either as a toxic effect of cytokines to progenitor cells in the bone marrow or a result of homing to inflamed tissues and extravasation.

Considerable work within a relatively short timespan has been done unraveling the mechanisms through which SARS-CoV-2 infection causes platelet activation. One study shows that platelet activation in severe COVID-19 is associated with detectable viral RNA in blood. Furthermore, the viral Spike protein enhanced platelet activation, aggregation, thrombus formation and degranulation in vitro and in a mouse model. This effect was only seen when the full Spike protein or its ACE-2 binding S1 subdomain were used, not the S2 domain. This suggests that ACE-2 signaling mediates this platelet activation. Further analysis of intracellular messaging points towards involvement of the MAPK signaling pathway. The same study confirmed expression of ACE-2 in human platelets using immunofluorescence [223]. However, another study did not detect any ACE-2 mRNA or protein expression in COVID-19 patients by RNA-seq, qPCR or Western blot [224]. To date, no study has demonstrated SARS-CoV-2 internalization by platelets.

It is also clear that platelets influence the host immune response to SARS-CoV-2. Platelet gene expression profiles in severe and critically ill COVID-19 patients showed shared pathways with sepsis and Influenza H1N1 infection. These show related antigen presentation and immune regulation, including differential expression of interferon-induced transmembrane protein 3, which has antiviral properties [224]. The formation of plateletleukocyte aggregates was also found, with neutrophil-platelet aggregates correlating with the severity of lung injury and leading to the formation of NETs [220,221,224,228,229]. Similar to observations in HIV and DENV, platelet-monocyte complexes are formed in severe COVID-19 patients via platelet P-selectin, which results in overexpression of Tissue Factor on the monocyte surface, the key initiator of the extrinsic coagulation pathway [230].

As with many other viral infections, reports have been published of cases of ITP associated with COVID-19 infection, including one case of Evans syndrome [231]. Another case report illustrates the importance of performing a peripheral blood smear in COVID-19 patients with severe thrombocytopenia to exclude EDTA dependent pseudothrombocytopenia [232].

#### **4. Conclusions**

The topic of thrombocytopenia in viral infectious diseases has been actively studied for many decades, with the last 10 years yielding many new insights. A scientific field combining the disciplines of virology, hematology and increasingly immunology is revealing a complex system of interactions between various viruses, the coagulation cascade and the innate and adaptive immune system. Increasingly, platelets are regarded as part of the immune system, in addition to being capable of forming blood clots. The rapidly changing world of viruses ensures that this field is constantly forced to adapt to new outbreaks and is therefore equally dynamic. The current COVID-19 pandemic has brought platelet-virus interactions to the forefront, with many publications addressing this topic being available within a year after the SARS-CoV-2 virus first emerged.

The absence of research on the "classical" hemorrhagic fevers, such as Ebola, Lassa and Marburgvirus, has been notable however, despite two large outbreaks of Ebola occurring in the last decade. The high level of biological containment required to safely study these viruses, combined with the extremely resource-limited settings in which these outbreaks occurred make doing research into these viral diseases extremely challenging. Nonetheless, significant progress has been made in preventive and therapeutic interventions for Ebolavirus, with the successful trials of several vaccines, [233] antiviral drugs and monoclonal antibodies [234].



DENV, another viral infection disproportionately affecting people in resource limited settings, was the virus we found most publications about in relation to platelets in the past decade. This is not surprising, given the considerable role platelets play in the pathophysiology of severe disease and the enormous public health burden associated with the virus. Despite the considerable knowledge gained, this has so far not been translated into clinically effective interventions. However, we did find several studies investigating therapeutics aimed at modifying platelet function in DENV infection, which will hopefully bear fruit in the coming decade.

Looking in detail at the interactions between viral infections and platelets revealed several common pathways connecting inflammation and platelet activation, which has been termed "Immunothrombosis". This is a term which has not yet been clearly defined as a clinical or pathological entity and has some overlapping features with DIC, with the main clinical difference being the absence of significant bleeding. COVID-19 may provide us the opportunity to increase our overall understanding of thrombocytopenia in viral infections and perhaps to study a new dimension of immunothrombosis which could be translated to other viral infections. It is especially important to gain more understanding about which interventions could aid in reducing the morbidity and mortality related to immunothrombosis. As platelet dysfunction is often accompanied by an increased risk of both bleeding and thrombosis, approaching this issue with conventional anticoagulants often involves having to choose the lesser of two evils. Immunomodulation therapy is a rapidly evolving field, with many newly available therapeutics, most of which have not yet been trialed in viral infectious diseases. This approach warrants further study, but here caution is also advisable, given the possibility that some mechanisms involved in immunothrombosis are required in the antiviral response in the host.

**Author Contributions:** Conceptualization: M.R., M.G., E.v.G. Writing: M.R., J.D.T., M.G., T.L. Search strategy & literature review: MPR. Editing & expert contribution: B.v.B., J.D.T., E.v.G., M.G. All authors have read and agreed to the published version of the manuscript.

**Funding:** This research received no external funding.

**Data Availability Statement:** Data available in a publicly accessible repository. The data presented in this study are available via the national Center of biotechnology information.

**Conflicts of Interest:** The authors declare no conflict of interest.

#### **Abbreviations**




#### **References**

