Superior outcomes with Argatroban for heparin-induced thrombocytopenia: a Bayesian network meta-analysis

Background Argatroban, lepirudin, desirudin, bivalirudin, and danaparoid are commonly used to manage heparin-induced thrombocytopenia related complications. However, the most suitable drug for this condition still remains controversial. Aim of the review This Bayesian network meta-analysis study compared the most common anticoagulant drugs used in the management of heparin-induced thrombocytopenia. Method All clinical trials comparing two or more anticoagulant therapies for suspected or confirmed heparin-induced thrombocytopenia were considered for inclusion. Studies concerning the use of heparins or oral anticoagulants were not considered. Data concerning hospitalisation length, thromboembolic, major, and minor haemorrhagic events, and mortality rate were collected. The network analyses were made through the STATA routine for Bayesian hierarchical random-effects model analysis with standardised mean difference (SMD) and log odd ratio (LOR) effect measures. Results Data from a total of 4338 patients were analysed. The overall mean age was 62.31 ± 6.6 years old. Hospitalization length was considerably shorter in favour of the argatroban group (SMD: − 1.70). Argatroban evidenced the lowest rate of major (LOR: − 1.51) and minor (LOR: − 0.57) haemorrhagic events. Argatroban demonstrated the lowest rate of thromboembolic events (LOR: 0.62), and mortality rate (LOR: − 1.16). Conclusion Argatroban performed better overall for selected patients with HIT. Argatroban demonstrated the shortest hospitalization, and lowest rate of haemorrhages, thromboembolisms, and mortality compared to bivalirudin, lepirudin, desirudin, and danaparoid.


Impacts on practice
• In patients with heparin-induced thrombocytopenia, rapid discontinuation of heparin and transition to a different anticoagulant therapy are essential. • Several drugs have been proposed for the management of heparin-induced thrombocytopenia; however, the most effective drug remains controversial. • Anticoagulant therapy in patients with heparin-induced thrombocytopenia reduces thrombotic and hemorrhagic risks. • While the most effective drug remains controversial, this meta-analysis suggest argatroban may be superior to the other used drugs for the management of heparin induced thrombocytopenia.

Introduction
Heparin-induced thrombocytopenia (HIT) is an immunomediated disorder that occurs in up to 5% of patients receiving therapeutic doses of heparin [1,2]. In selected patients, heparin exposure induces the formation of IgG-PF4-heparin complex [3] which can promote platelet activation and aggregation [4]. Typically, HIT occurs within ten days after the start of heparin administration [1]. A second exposure to heparin before antibodies have disappeared can result in a rapid onset of HIT [5]. Of note, the incidence of HIT is higher in patients receiving unfractionated heparin than in those exposed to low molecular weight heparin [6]. Low molecular weight heparin induces a smaller complex with PF4 compared to unfractionated heparin [7]. HIT occurs mainly in surgical patients, especially after cardiac or orthopaedic surgery [8]. The diagnosis of HIT is both clinical and serological [9]. ELISA assay and serotonin release assay have high sensitivity but low specificity for HIT [9]. The diagnosis is confirmed by the presence of "HIT antibodies" and any of the following events: (1) unexplained fall in platelet count (< 30% to < 50%); (2) thrombosis; (3) skin lesions at the heparin injection site; (4) acute systemic (anaphylactic) reactions [10]. Up to 55% of patients present deep venous thrombosis, while arterial thromboembolisms are uncommon [11]. Cessation of heparin may not be sufficient to prevent thromboembolic complications; thus, anticoagulant therapy must be promptly initiated [12,13]. Argatroban and lepirudin are direct thrombin inhibitors (DTI) approved by the Food and Drug Administration (FDA) for the treatment of HIT [14]. Bivalirudin, desirudin, and danaparoid are also commonly used for the management of HIT. Bivalirudin is a synthetic analogue of hirudins indicated for patients with HIT undergoing percutaneous coronary interventions (PCI) [15,16]. Desirudin is another DTI, belonging to the hirudins family, which is commonly administered in major orthopaedic procedures [17][18][19]; although danaparoid is a low molecular heparinoid, it also has been commonly administered to prevent complications of HIT [20,21].

Search strategy
This Bayesian network meta-analysis was conducted according to the PRISMA extension statement for reporting of systematic reviews incorporating network meta-analyses of health care interventions [41]. The PICO protocol guided the initial search:

Eligibility criteria
All clinical trials investigating parenteral anticoagulant therapies for suspected or proven HIT were considered for inclusion. Studies in English, Italian, German, French, and Spanish, according to the authors language capabilities, were included. Studies of level of evidence I to III, according to the Oxford Centre of Evidenced-Based Medicine [42], were considered. Reviews, comments, editorials, opinions, reports, and data from registries were not considered. Animal and cadaveric studies were not included. Studies using other drugs rather than parenteral anticoagulants were not included. Trials comparing anticoagulants with heparin were not included, nor were studies that evaluated the effectiveness of these compounds in other thrombocytopenia conditions (e.g. neoplastic, idiopathic, haemolytic). Articles performing studies on oral anticoagulants (e.g. vitamin K antagonists, direct oral anticoagulants) were not included.
Only clinical trials reporting quantitative data under the outcomes of interest were considered for inclusion. Disagreements between the reviewers were debated and solved by a third author (SH).

Outcomes of interest
Data extraction was performed by two independent reviewers (GC, FM). Generalities of the studies were collected (author and year, journal, study design, eligibility criteria). Patient baseline was extracted, along with the name, dose, therapeutic and maintenance doses for each drug. The length of hospitalization was retrieved. Data on the rate of major and minor haemorrhagic events, thromboembolic complications, and mortality were collected.

Methodological quality assessment
The methodological quality assessment was performed by one reviewer (GC). For this purpose, the bias summary tool of the Review Manager Software (The Nordic Cochrane Collaboration, Copenhagen) was used. The following biases were analysed: selection, detection, attrition, reporting, and other sources of bias.

Statistical analysis
The statistical analyses were performed by the senior author (FM). For baseline comparability, the IBM SPSS software was used. Comparability was assessed through the Analysis of Variance (ANOVA), with P > 0.1 considered satisfactory. The network analyses were made through the STATA/MP software (Stata Corporation, College Station, Texas, USA). The analyses were performed through the Stata routine for Bayesian hierarchical random-effects model analysis. Continuous variables were analysed through the inverse variance method, with the standardized mean difference (SMD) effect measure. Binary data was analysed through the Mantel-Haenszel method, with the Log Odd Ratio (LOR) effect measure. Edge, interval, and funnel plots were performed and analysed. The overall transitivity, consistency, and heterogeneity, as well as the size of the treatment effect of interest within-study variance, were evaluated. The overall inconsistency was evaluated through the equation for global linearity via the Wald test. If P Wald values > 0.05, the null hypothesis could not be rejected, and the consistency assumption could be accepted at the overall level of each treatment. Confidence and percentile intervals (CI, PrI, respectively) were each set at 95%.

Search result
The literature search resulted in 836 articles. Of them, 252 were duplicates. An additional 530 articles were excluded: nature of the study (N = 207), non-clinical studies (N = 153), use of other anticoagulants (N = 84), use of adjuvant(s) (N = 72), language limitations (N = 12), uncertain results (N = 2). Another 21 articles were rejected as quantitative data under the outcomes of interests were missing. Ultimately, 33 articles were included: 4 randomized clinical trials, 16 prospective, and 13 retrospective clinical studies (Fig. 1).

Methodological quality assessment
The Cochrane bias of summary tool evidenced some limitations of the present study. First, the overall retrospective nature of the included studies, along with the overall lacking of blinding methods led to increased selection and detection biases. Indeed, only 12% (4 or 33 studies) were RCTs, and only two of these were blinded. The overall risks of attrition, reporting, and unknown sources of bias were acceptable. In conclusion, the risk of bias of the present work was fair-moderate. The Cochrane bias of summary tool is shown in Fig. 2.

Patient demographics
Data from 4338 patients were retrieved. A total of 1846 patients were analysed in the argatroban group, 318 in the bilivarudin group, 973 in the lepidurin group, 68 in the desirudin group, and 126 in the danaparoid group. 53.92% (2339 of 4338 patients) were men. The overall mean age was 62.31 ± 6.6 years old. The ANOVA test found good comparability concerning age (P = 0.07). The generalities and demographics of the included studies are shown in Table 1.

Discussion
According to the main findings of this Bayesian network meta-analysis, argatroban performed better than bivalirudin, lepirudin, desirudin, danaparoid for selected patients with HIT. Argatroban demonstrated the lowest rate of haemorrhages, thromboembolisms, and mortality, along with the shortest length of the hospitalization, compared to the other drugs of interest.
HIT is an immune mediated reaction triggered by the exposure to unfractionated heparin or low molecular weight heparin. Clinical manifestation is typically seen in 5 to 10 days after the start of heparin therapy [5,43]. The most severe complications of HIT are thromboembolic events, and several studies have reported that heparin cessation may not be enough to prevent thrombosis: another anticoagulation method is recommended [44,45]. Hirsh et al. [4] investigated the treatment of HIT with argatroban, lepirudin, and danaparoid. Despite the use of another anticoagulation method, the mortality in HIT is up to 22%, while the risk of a new thromboembolic event is up to 18%. In the present study, the efficacy and safety of five different anticoagulants were analysed: four belonging to the DTI family (argatroban, lepirudin, desirudin, bivalirudin) and one low molecular weight heparinoid (danaparoid). Argatroban is a univalent DTI that binds thrombin, inhibiting its action. Its good tolerability and short half-life make argatroban an attractive option for the management of HIT complications [46]. However, given its hepatic metabolism, argatroban is not recommended for patients with hepatic dysfunction [47], unless meticulous dose adjustments and aPTT monitoring are undertaken [44]. Conversely, danaparoid has longer halflife than argatroban, and good efficacy in maintaining stable anticoagulation [48]. Moreover, despite danaparoid being a low-molecular-weight heparinoid, its cross-reactivity with HIT antibodies is rare [48]. Hirudins (lepirudin, desirudin) are bivalent DTI, that bind with a high affinity and specificity to two distinct sites of thrombin [49]. However, give its renal metabolism, lepirudin must be administered with caution in nephropathic patients. Furthermore, immunogenicity has been associated with lepirudin consumption [50,51]. In a meta-analysis by Greinacher et al. [43], lepirudin demonstrated lower mortality rate and thromboembolic events compared to the control group. Treschan et al. [52] compared argatroban versus lepirudin on 28 patients requiring renal replacement therapy. Patients randomized to lepirudin experienced a greater incidence of haemorrhagic complications [52]. These results were confirmed by the present study, which found a greater rate of haemorrhagic events following administration of lepirudin compared to argatroban and danaparoid. In a retrospective analysis comparing danaparoid (126 patients) versus lepirudin (175 patients) [53], interestingly, danaparoid was associated with a greater risk of new thromboembolic events in patients without thrombosis on admission. Although bivalirudin is mainly metabolized in the liver, nephropathic patients require dose adjustment [38,54,55]. Bivalirudin and argatroban were compared in a retrospective study which found that both anticoagulants reached the aPTT goal within six hours, and evidenced similarity between the two compounds [37]. Lastly, a meta-analysis, demonstrated similar efficacy and safety between argatroban, lepirudin, and bivalirudin [56].
This network meta-analysis has certain limitations. The retrospective nature of most of the included studies increased the risk of selection bias, reducing the reliability of the conclusions of the present study. The current literature lacks high-quality studies; therefore, additional studies should tackle this limitation. Furthermore, the low number of included studies and related patients represents another limitation. The presence of HIT-related complications on admission were not analysed in the present study, along with patient comorbidities (e.g. patients with liver and/or renal dysfunctions). Given the lack of quantitative data, a subgroup analysis according to the initial and maintenance doses could not be performed. Parenteral anticoagulants are frequently chosen for the management of HIT [22,57].    While many options for anticoagulant therapy are available, the present study focused on the effects of parenteral anticoagulation. Further analyses concerning the effects of direct oral anticoagulants and vitamin K antagonists for HIT are required. Lastly, fondaparinux is a synthetic heparin polysaccharide, [58] which has not been considered in this study. Given the lack of quantitative data, fondaparinux was not included in the network comparisons. Considering these limitations, data from the present Bayesian network metaanalysis must be interpreted with caution.

Conflict of interest
The authors declare that they have no conflicts of interest.
Open Access This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if changes were made. The images or other third party material in this article are included in the article's Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article's Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visit http:// creat iveco mmons. org/ licen ses/ by/4. 0/.  Fig. 6 Edge, funnel, and interval plots of the network comparison: mortality rate