A fresh preprint reveals the results of combining three trials to evaluate therapeutic anticoagulation among severe COVID patients (defined as patients requiring high flow nasal cannula, noninvasive ventilation, intubation, vasopressors, or inotropes). Patients were randomized to therapeutic anticoagulation versus thromboprophylaxis per usual care guidelines. REMAP-CAP also simultaneously randomized some patients to receive aspirin as well. Inclusion required admission to the ICU within <48 hours (REMAP-CAP) or admission to the hospital within <72 hours (ACTIV-4a, ATTACC).
This is a combination of three open-label, pragmatic, adaptive, multiplatform trials. Now, even within a single open-label, pragmatic, adaptive RCT, things can get a bit messy. What will happen when we combine three of these trials together? Brace yourself.
Patient characteristics are shown below. Groups were generally well matched. The D-dimer values were peculiarly low for sick COVID patients (with median levels of only ~850 ng/ml). The majority of patients had a D-dimer value which was below twice-normal.
The treatments that patients received are shown below. This is, frankly, a mess. The control group is almost evenly split between low-dose thromboprophylaxis versus intermediate-dose thromboprophylaxis. Meanwhile, only ~78% of the patients in the therapeutic anticoagulation group received therapeutic dose anticoagulation.
To make matters even murkier, therapeutic anticoagulation was achieved by a mixture of low-molecular-weight heparin and intravenous unfractionated heparin. These interventions may seem to be interchangeable, but they are not. Patients with COVID commonly display heparin resistance. Treatment with weight-based low-molecular-weight heparin will therefore tend to underdose the heparin. Alternatively, an unfractionated heparin infusion which is titrated against coagulation laboratories will be uptitrated to effect, thereby achieving full therapeutic anticoagulation (which often requires the administration of prodigious doses of heparin).
So, in summary, patients in the study received roughly eight different interventions:
- Low-dose thromboprophylaxis
- Low-dose thromboprophylaxis + aspirin
- Intermediate-dose thromboprophylaxis
- Intermediate-dose thromboprophylaxis + aspirin
- Therapeutic low-molecular-weight heparin
- Therapeutic low-molecular-weight heparin + aspirin
- Therapeutic heparin infusion
- Therapeutic heparin infusion + aspirin
Heparin is a high-risk medication. The exact dose and route of administration matter. How the drug is combined with other anticoagulants (e.g., aspirin) also matters. As such, each of the above eight regimens is different. None of these eight regimens is necessarily interchangeable with another. So, when we scatter eight different interventions in a hodgepodge fashion across two different groups, what we have is chaos.
This actually gets even more problematic if we drill down on exactly how low-molecular-weight heparin was being dosed (shown above). For example, let’s imagine a patient who is 5 feet tall and weighs 94 kg (with a body mass index of 41). If this patient receives intermediate-dose enoxaparin, they could receive up to 60 mg enoxaparin BID. This is not much lower than a full therapeutic dose of enoxaparin (94 mg BID in this patient). So, the intermediate-dose enoxaparin may come surprisingly close to a full therapeutic dose of enoxaparin. If we also consider the effects of variable renal function, it becomes likely that patients in both groups will have overlapping levels of anticoagulation (i.e., anti-Xa activity levels – which weren’t generally measured).
The results are shown above. They may be summarized roughly as follows:
- There was no difference in the number of organ support-free days (the primary endpoint).
- There were no differences in mortality.
- There was a lower risk of major thrombotic events in the full anticoagulation group (5.7% vs. 10.3%; p=0.012).
- There was a nonsignificant, small trend towards more bleeding in the full anticoagulation group (3.1% vs. 2.4%; p=0.6)
It’s not clear how to best interpret this data. Poor protocol adherence and overlap between heparin regimens used in the two groups reduced the statistical power, impairing the ability to demonstrate any effect on global outcomes (e.g., mortality, organ support-free days). Nonetheless, there is a signal that higher doses of heparin reduced the incidence of major thrombotic events. And perhaps most intriguing, there was no signal of harm – full therapeutic anticoagulation caused no statistical difference in the rate of bleeding!
Any interpretation of these results should also take into account that the average D-dimer values were rather low (~850 ng/mL). Consistent with these low D-dimer values, the rates of thrombosis were low in the control group (only 10%, lower than many other ICU series of COVID patients). Lack of a major heparin effect among patients with a relatively low level of coagulation activation should come as little surprise. This finding fails to exclude a greater benefit from therapeutic anticoagulation among patients with markedly elevated D-dimer levels (e.g., >2,000 ng/ml).
Ultimately, exactly how to interpret these findings is a subjective matter of judgement – one which I will leave to you. Overall, I don’t believe that this study has any singular, definitive conclusion. This should be a reminder that neither a press release, nor a tweet, nor the full manuscript of a study will necessarily provide a simple clinical answer!
- REMAP-CAP/ACTIV-4a/ATTACC is a conglomerate of three open-label, adaptive, multicenter RCTs evaluating therapeutic anticoagulation among COVID patients requiring vasoactive medications, high flow nasal cannula, noninvasive ventilation, or intubation.
- There was a lot of overlap between heparin regimens used in the two groups. The control group was mostly split between low-dose and intermediate-dose anticoagulation. Meanwhile, only ~78% of patients in the therapeutic group received full therapeutic dose anticoagulation.
- The median D-dimer level was relatively low, at ~850 ng/mL. Most patients had a D-dimer which was below twice the upper limit of the assay! Based on these low D-dimer levels, these are not patients where full therapeutic anticoagulation would be expected to be most beneficial.
- There were no differences in the primary endpoint (organ support-free days) or mortality.
- Full therapeutic anticoagulation did reduce the rate of major thrombotic events, without causing any statistically significant difference in the rate of hemorrhage.
- This conglomerate study is messy, eluding any simple conclusion. Is the heparin vial half empty, or half full? Despite what all the headlines will say, this remains extremely murky.