Due to the high thrombotic risk associated with this infection, systematic antithrombotic prophylaxis should be administered and to date, parenteral administration of a heparin preparation is the preferred procedure in the hospital setting.
Choice of anticoagulant
LMWH is recommended as first-line therapy for the prophylaxis of VTE in hospitalized COVID-19 patients [
15,
20,
117,
118]. Direct oral anticoagulants (DOACs) or vitamin K antagonists, are not recommended because of the risk of drug interactions, among others with some antiviral drugs, the expected broad fluctuation in plasma concentrations (for DOACs), especially in patients at higher risk of rapid clinical deterioration, and because of the late onset of anticoagulation and higher risk of bleeding with VKA [
119]. Heparin, and especially the long heparin chains, was reported to also exhibit anti-inflammatory activity in addition to the anticoagulant effect, by binding and neutralizing inflammatory cytokines and acute phase proteins, while potentially exerting a protective effect on endothelium [
120]. It also interferes with neutrophils recruitment into tissue and impairs neutrophil function by inhibiting the activity of the neutrophil protease’s human leukocyte elastase and cathepsin G, which can promote inflammation [
121,
122]. It has also been hypothesized that heparin may hinder the interaction between the virus and the host cell through non-specific ionic bond, and thus may contribute to decrease the rate of infected cells [
120,
123]. However, we do not yet know precisely to what extent heparin is clinically effective in this infection.
Compared to UFH, LMWH has better bioavailability after subcutaneous administration and longer duration of action, allowing daily administration in one or two injections. No regular laboratory monitoring is necessary for treatment with LMWH because of the predictable anticoagulant activity after administration of doses adjusted to body weight on the one hand, and the lack of formal association demonstrated between laboratory tests results and clinical efficacy or complications on the other [
124]. However, a single anti-Xa activity measurement can be proposed in case of administration of high or intermediate doses in a patient with moderate renal impairment, with a BMI < 18 kg/m
2 or > 30 kg/m
2, or during pregnancy, mainly to rule out drug accumulation [
125]. In the event of increased doses in patients at enhanced thrombotic risk, measurement of anti-Xa activity is also recommended 4 h after the third dose to exclude accumulation. The anti-Xa value must be based on a calibration curve with the specific heparin type and interpretation of test result must also be made for the same compound; e.g. with therapeutic doses of enoxaparin, the mean peak concentration observed is 1.2 IU/mL. This measurement can be repeated for example in case of renal function impairment.
UFH is only recommended in case of severe renal failure (creatinine clearance according to Cockcroft-Gault equation < 30 mL/min), extracorporeal membrane oxygenation (ECMO) [
15] or significant bleeding risk (shorter half-life than LMWHs, easier neutralization by protamine) [
126]. The risk of heparin-induced thrombocytopenia (HIT) is also much higher with UFH [
127]. Regular laboratory monitoring during anticoagulation is necessary because of the high inter- and intra-individual variability in the anticoagulant response [
125].
Laboratory monitoring of unfractionated heparin treatment
Adjustment is thought to be required because of high inter and intra-individual variability. Historically, adjustment of UFH dosage was based on APTT. The measurement is performed 4 to 6 h after initiation and any dose change, and once a day at least, to reach a target APTT ratio between 1.5 and 2.5. This therapeutic target dates from work in 1972 and has never been confirmed in large clinical studies [
128]. Since then, the number of reagents used for the APTT has exponentially increased, the sensitivity of which is very different both to heparin and biological interference (including several proteins of the acute phase) [
129‐
131]. Therefore, calculation of the APTT ratio corresponding to an anti-Xa activity between 0.3 and 0.7 IU/mL would be recommended for each analyzer and each new batch of reagents. This calculation should best be performed with plasma samples from patients treated with UFH because the use of plasma spiked with UFH gives less relevant results due to the influence of the metabolism of heparin in vivo and its bioavailability [
132]. For some reagents, the relationship between heparin levels and APTT is linear, but this can change in case of significant inflammation [
133]. Most importantly, APTT is very dependent on pre-analytical conditions. Among others, platelet factor 4 (PF4) released by activated platelets during inadequate sampling procedure or prolonged delay before centrifugation can neutralize part of the heparin, leading to a risk of underestimating its activity [
134].
Several biological parameters can cause a prolongation of the APTT (e.g. high CRP, presence of lupus anticoagulants, coagulation factors deficiency, high plasma concentration in FDPs which oppose the polymerization of fibrin) or its shortening (e.g. increased plasma concentration of acute phase proteins such as FVIII and fibrinogen) [
63,
135,
136]. The influence of these parameters will also depend on assay methodology, the type of reagents, and has a high inter-individual and intra-individual (i.e., during hospitalization) variability [
137,
138]. For these reasons, the GFHT advises against the use of APTT for monitoring treatment with UFH [
133]. Moreover, the use of APTT is problematic when this test is prolonged prior to initiation of UFH treatment (for example, in case of lupus anticoagulants or coagulation factors deficiency); anti-Xa target should also consider the etiology of this prolongation when clinically relevant (i.e. defects with a bleeding risk).
In COVID-19, the increased plasma concentrations of fibrinogen and FVIII can cause a shortening of APTT (observed in 16% of affected patients) [
55], and this may lead to underestimating the anticoagulant effect of heparin. This situation can contribute to excess heparin dosing, enhancing the bleeding risk, and underlines the importance of collecting a basal APTT value before anticoagulation. This can be problematic in some settings, e.g. in the ICU, where patients are frequently transferred with anticoagulation already started at intermediate or therapeutic doses [
13]. Conversely, APTT prolongation may be related to the transient increased levels of antiphospholipid antibodies, a situation encountered during viral infections [
139], including COVID-19 [
10,
140‐
143], or when CRP is high [
138], thus leading to a risk of heparin underdosage. Lupus anticoagulant screening can also be falsely positive in the presence of variables prolonging clotting time of tests used for its screening (for example, elevated CRP or presence of anticoagulants, among which heparin) [
143‐
146]. The APTT may also be prolonged in the presence of DIC (which is relatively rare in COVID-19). In this situation, its measurement by means of an optical system may become uninterpretable: an immediate and gradual decrease in light transmittance can be observed even before clot formation due to the presence of a complex between CRP and very low density lipoproteins in the presence of calcium [
147], thus rendering the measurement unreliable [
148]. Therefore, mechanical methods are advisable in this circumstance, where measurement anti-Xa activity may even be the best choice.
Thus, heparin monitoring with aPTT may be challenging in COVID-19 patients due to the hyper-inflammatory status of the patient. Indeed, the high fibrinogen and factor VIII levels, the interference of CRP (depending on the reagents used) and also the potential presence of antiphospholipid antibodies may affect the aPTT. Therefore, anti-Xa activity seems more suitable to monitor UFH treatment in these patients and more generally in ICU patients for the very same reasons. However, there are several caveats here as well. First, FXa is not the essential target of UFH. Its inhibition is studied under very artificial conditions: in the fluid phase (and not within the prothrombinase complex formed on a phospholipid surface) and in a calcium-depleted medium. The in vivo inhibitory activity of UFH is indeed three times stronger towards FIIa than FXa [
149]. This difference is further artificially increased in vitro by use of low calcium concentrations in the assay mixture: the anti-Xa activity was halved under these conditions, compared to physiological concentrations of calcium, but the effect of low calcium on anti-IIa activity is more limited [
149]. Having mentioned that, a good correlation exists in vitro between anti-Xa and anti-IIa activities, thus enabling the use of the former test to assess the effect of heparin therapy. The anti-Xa assay consists of measuring in vitro the residual activity on a chromogenic substrate specific for FXa added to citrated plasma. Compared to APTT, this test has the advantage of being less vulnerable to biological interference (possible interference of free hemoglobin and bilirubin in case of significant elevation [
150]) and less dependent on pre-analytical conditions - with the notable exception of PF4 released in vitro by platelets. Even if the validity of anti-Xa activity of UFH in the presence of an important inflammatory syndrome has not been formally established (or for that matter under any circumstances), this measure will be less impacted in this context, particularly if the reagents contain antithrombin (AT). However, it is not advisable to use kits with exogenous AT to avoid overestimation of anticoagulant activity in case of AT deficiency, with risk of heparin underdosing. The plasma concentration of AT can indeed decrease in case of sepsis [
151]. Only few data are available on the sensitivity of kits without exogenous AT to changes in plasma AT concentration [
152]. Some reagents also contain dextran sulfate, which will displace heparin from its non-specific binding (including PF4). Particularly, the influence of PF4 released by activated platelets is therefore minimized, which is favorable for limiting the impact of pre-analytical conditions on test result, but problematic when the concentration of PF4 is actually high in vivo. Unlike a ‘global’ test like APTT, the measurement of anti-Xa activity is insensitive to fluctuations in the underlying hemostatic state (for example, coagulation factor defect following hemorrhage or DIC), which should prompt an adjustment in therapeutic targets to the clinical context and the possible identification of such defects. Finally, significant variability in heparin sensitivity has been reported between the different commercial kits [
153‐
156].
The therapeutic range in terms of anti-Xa activity is considered between 0.3 and 0.7 IU/mL [
125]. These values are derived from the work of 1972 that was previously mentioned. It was hence inferred from the APTT target to be reached for secondary prevention of VTE. The bottom line is that it lacks validation for the same reasons [
157]. The application and interpretation of these tests in a hyperinflammatory context also raises important questions. Given the very high thrombotic risk described or suspected in some subgroups of patients, the GIHP suggested narrowing the target of anti-Xa activity in the upper zone, to 0.5–0.7 IU/mL, for those at the highest thromboembolic risk [
15]. This increase in dosage, not supported by objective data, remains debated [
20‐
23], whilst ongoing prospective, multicenter clinical trials aim at addressing this question. In addition to the selection of patients who could potentially benefit from increased doses of anticoagulants, the question of treatment duration is a major one. The resolution of the inflammatory syndrome should be accompanied by a reduction in the thrombotic risk, thus exposing the patient to excessive anticoagulation and risk of bleeding when higher doses of heparin are maintained. However, no firm recommendation on duration and intensity of COVID-19 anticoagulation can be made at this time.
In patients receiving UFH, a laboratory resistance to the anticoagulant effect of heparin, arbitrarily defined by failure to reach the therapeutic target despite the administration of doses > 1.5 times usual doses, which are about 400 to 600 U/kg/24 h [
158,
159], is frequently observed in COVID-19 [
13,
160], and adds to the clinical resistance previously outlined (occurrence of thrombotic events under well-conducted drug thromboprophylaxis). The hyperinflammatory context could also explain part of this observation. Indeed, UFH is able to non-specifically bind several acute phase proteins as well as activated endothelial cells and platelets, thus limiting its anticoagulant activity [
161,
162]. The administration of an initial bolus of UFH is therefore needed to saturate non-specific fixation [
163]. The increased plasma concentration of fibrinogen and FVIII will also contribute to generate heparin resistance when the effect is monitored with the APTT, which is less likely to be observed when anti-Xa activity is used. Acquired AT deficiency by consumption or production defects (negative protein of the acute phase) could also contribute to the heparin resistance observed in some COVID-19 patients [
151], especially those most seriously affected [
10,
49], but in most patients it does not justify the prescription of AT concentrates. To the best of our knowledge there is one single prospective interventional study on UFH monitoring in case of laboratory heparin resistance. In this study, the utilization of anti-Xa activity instead of aPTT permitted to avoid UFH dosage escalation with similar clinical outcomes [
164]. Whether this holds true for COVID-19 patients remains to be established. When heparin resistance is suspected based on APTT values, UFH shall be administered according to the anti-Xa activity [
165]. The advantages and limitations of APTT vs anti-Xa for UFH monitoring are summarized in Table
3.
Table 3
Main advantages and limitations of APTT and anti-Xa activity for UFH treatment laboratory monitoring
APTT | - largely available, low cost - sensitive to clinically relevant changes of coagulation (coagulation proteins increases or deficiencies) | - numerous interferences; optical methods can be unreliable in case of DIC |
- heparin sensitivity is highly reagent dependent |
- APTT prolongation target needs to be established for each new batch of reagents |
- APTT before UFH needed |
- clinically irrelevant changes, or of dubious clinical relevance |
APTT prolonged with: |
• presence of antiphospholipid antibodies (viral infections) |
• high CRP (depending on the reagent) |
• high plasma levels in FDPs |
• preanalytical (e.g., heparin/EDTA contamination, under-filling, delayed centrifugation, hypertriglyceridemia, hyperbilirubinemia) |
APTT shortened with: |
• preanalytical (e.g., prolonged venous stasis, vigorous mixing, coagulation of the sample, PF4) |
• high factor VIII levels |
Anti-Xa activity | - less vulnerable to biological interferences - no requirement for measurement before UFH administration | - preanalytical interferences: PF4*; free hemoglobin and bilirubin if significant elevation |
- insensitive to fluctuations in the underlying coagulation state (i.e., coagulation factor increases or defects), of potential clinical relevance |
- AT deficiency (e.g. in sepsis): risk of heparin underdosing with kits containing exogenous AT; sensitivity to endogenous AT not evaluated with kits that do not contain exogenous AT |
- variability in reagents composition (AT, dextran…) |
- therapeutic range poorly defined |
- not validated in hyperinflammatory states |
- less available, more expensive |
Diagnosis of heparin-induced thrombocytopenia (HIT)
A final aspect of heparin monitoring is screening for HIT. A platelet count should be performed before administering the first injection of heparin, if possible, or as soon as possible thereafter. In the COVID-19 setting, it is reasonable to monitor the platelet count regularly between the 4th and the 14th day following the initiation of heparin therapy (once or twice a week in case of LMWH treatment, two to three times a week during UFH treatment), then once a week until the end of the first month of therapy. The development of thrombocytopenia (< 100 × 10
9/L) or the rapid decrease in platelet count (especially if ≥50% in less than 24 h) should then suggest the diagnosis of HIT [
166]. However, especially in the presence of acute inflammation and infection, other etiologies may explain a decrease in platelet count. Therefore, a systematic evaluation of clinical probability of diagnosis allows better identification of patients in whom the occurrence of this complication must be suspected, and for whom laboratory work-up for HIT antibodies is indicated. This evaluation is generally performed with the 4Ts’ score, much studied [
167,
168] [
159], despite its limitations, particularly in more complex situations such as those encountered in ICU (no consensus on the drugs responsible for thrombocytopenia, many other causes of thrombocytopenia, very high negative predictive value but not absolute (e.g. thrombosis in the absence of thrombocytopenia), insufficient data on platelet count, weak agreement in the determination of the 4th criteria (other causes of thrombocytopenia)) [
166,
169,
170].
In case of strong suspicion, or as soon as the antibodies are identified, treatment with heparin should be stopped and replaced by a direct thrombin inhibitor (DTI; argatroban, bivalirudin) or by danaparoid sodium. Of note, the presence of a DTI can lead to underestimation of plasma fibrinogen concentrations by inhibition of the thrombin present in Clauss’ reagent [
171]. This interference will vary depending on the thrombin concentration used in the reagent [
172,
173]. To a lesser extent, interference may also exist in the presence of high concentration of UFH (0.6 to 2.0 IU/mL, depending on the reagent), which would exceed the neutralization capacities of the reagent used, or in the presence of high concentration of FDPs (> 100–130 μg/mL) [
173].