HERDOO2 rule for stopping anticoagulation in unprovoked VTE

Clinical background

Deciding how long anticoagulation should continue after a first unprovoked venous thromboembolism (VTE) is one of the most difficult trade-offs in thrombosis medicine. The risk of recurrence after treatment is stopped is substantial: up to 10% during the first year and about 30% at eight years [1]. At the same time, continued anticoagulation reduces the risk of recurrence by 80 to 90%, but at the price of an annual risk of major bleeding of about 1.3% in patients who have already completed initial treatment without a bleeding complication [1]. The International Society on Thrombosis and Haemostasis (ISTH) has proposed that it is safe to stop anticoagulation if the risk of recurrence is below 5% during the first year, with an upper confidence limit of no more than 8% [1].

The HERDOO2 rule was developed to identify the subgroup of patients whose risk of recurrence is low enough for stopping treatment to be defensible. A central finding in the derivation work was that no subgroup of men could be identified as low risk, which is why the rule was restricted to women [1].

Applying the HERDOO2 rule

The rule consists of four dichotomous variables, each worth one point:

$$\text{HERDOO2} = \text{Post-thrombotic signs} + \text{D-dimer} + \text{BMI} + \text{Age}$$

where:

• Post-thrombotic signs: hyperpigmentation, oedema or redness in either leg (yes = 1, no = 0)
• D-dimer: $\geq 250,\mu\text{g/L}$ measured during ongoing anticoagulation (yes = 1, no = 0)
• BMI: $\geq 30,\text{kg/m}^2$ (yes = 1, no = 0)
• Age: $\geq 65$ years (yes = 1, no = 0)

A score of 0 to 1 classifies the patient as low risk; a score of 2 or more classifies her as high risk.

The rule was derived within the REVERSE I study and validated prospectively in REVERSE II, a multinational cohort study at 44 secondary and tertiary care centres in seven countries [1]. In total, 2,785 participants with a first unprovoked VTE (proximal deep vein thrombosis or pulmonary embolism) who had completed 5 to 12 months of anticoagulation were included. Of these, 1,213 were women, of whom 631 (51.3%) were classified as low risk by HERDOO2. The index VTE was defined as unprovoked in the absence of major provoking factors (leg fracture, plaster cast, immobilisation >3 days, major surgery within 3 months, malignancy within 5 years). VTE associated with weak risk factors such as travel, oestrogen or minor surgery counted as unprovoked [1].

The D-dimer was measured with the VIDAS assay (bioMérieux) at a threshold of 250 µg/L, which is half the usual diagnostic exclusion threshold. Sampling was performed during ongoing anticoagulation: for vitamin K antagonists at any time, and for direct oral anticoagulants (DOACs) at peak effect (dabigatran 4 to 8 hours, apixaban/rivaroxaban/edoxaban 2 to 6 hours after the dose) [1].

Interpretation in practice

In REVERSE II the low-risk group met the ISTH criterion: the risk of recurrence was 3.0% per patient-year with an upper confidence limit of 4.8%, well below the 5% threshold [1]. Among high-risk patients who nonetheless stopped treatment, the risk of recurrence was 8.1% per patient-year, compared with 1.6% among those who continued anticoagulation [1].

It is important to note that the rule is a rule for weighing risks, not an absolute rule. The decision should be made in dialogue with the patient, taking into account bleeding risk, patient preference and practical factors such as adherence.

Validation and performance

REVERSE II is the only prospective validation of HERDOO2 and is at the same time the study underpinning its clinical use [1]. The study was designed as a cohort management study: low-risk women stopped anticoagulation per protocol, while management of high-risk patients and of men was left to the treating clinicians. All recurrences were adjudicated independently and blindly by two physicians [1].

Two systematic reviews have examined prognostic models for VTE recurrence. Ensor et al. (2016) identified three models (HERDOO2, the Vienna prediction model and DASH) and found that HERDOO2 had several methodological weaknesses in its derivation, and that all the models were judged to be at at least moderate risk of bias and to require independent external validation [2]. De Winter et al. (2021) found that models for VTE recurrence generally performed weakly to moderately in external validations, with c-statistics between 0.39 and 0.66 (one study reported 0.83), and that all but three studies were judged to be at high risk of bias. The review noted, however, that impact studies suggest HERDOO2 and the Vienna model can identify low-risk patients, but concluded that the results do not support routine use in clinical practice for weighing recurrence risk against bleeding risk [3].

One critical implementation issue concerns the D-dimer assay. Rodger et al. (2018) tested four alternative commercial assays (Innovance, HemosIL, Tina-quant and Liatest) against VIDAS at corresponding thresholds. Concordance was poor for all of them, with kappa values between 0.30 and 0.38, and the conclusion was that these assays should not be used in the HERDOO2 rule because of unacceptable misclassification of women into the high- and low-risk groups [4]. This means that in practice the rule is tied to one specific D-dimer method.

Limitations

HERDOO2 applies only to women with a first unprovoked VTE. Men have a higher risk of recurrence and no low-risk subgroup could be identified; they are therefore managed as high-risk patients regardless of the score [1].

The rule does not apply after provoked VTE or when a major risk factor persists. The following patients were excluded from REVERSE II and the rule should not be applied to them: patients with malignancy, known high-risk thrombophilia (protein S, protein C or antithrombin deficiency, antiphospholipid syndrome, homozygous factor V Leiden or prothrombin gene mutation), a mechanical heart valve, atrial fibrillation, a vena cava filter, pregnancy-associated VTE, and patients with planned oestrogen treatment [1].

The absence of independent external validation is an important limitation. REVERSE II was conducted by the same research group that derived the rule, and no other group has prospectively validated HERDOO2 in a separate cohort. The systematic reviews point out that this constitutes a methodological weakness [2, 3].

The dependence on the D-dimer is a practical bottleneck. Since only the VIDAS assay at 250 µg/L has been validated, and other assays have shown poor concordance [4], the rule cannot be applied in laboratories using other methods without risking misclassification. This limits its dissemination, particularly outside larger hospitals.

Finally, the rule rests on a single D-dimer measurement during ongoing anticoagulation. If the patient has already stopped treatment at the time of assessment, the D-dimer cannot be interpreted according to the rule, since the threshold has been validated for sampling during active anticoagulation.

References

1. Rodger MA, Le Gal G, Anderson DR et al. Validating the HERDOO2 rule to guide treatment duration for women with unprovoked venous thrombosis: multinational prospective cohort management study. BMJ 2017;356:j1065. PMID: 28314711
2. Ensor J, Riley RD, Moore D et al. Systematic review of prognostic models for recurrent venous thromboembolism (VTE) post-treatment of first unprovoked VTE. BMJ Open 2016;6:e011190. PMID: 27154483
3. de Winter MA, van Es N, Büller HR et al. Prediction models for recurrence and bleeding in patients with venous thromboembolism: A systematic review and critical appraisal. Thromb Res 2021;199:85-96. PMID: 33485094
4. Rodger MA, Le Gal G, Langlois NJ et al. "HERDOO2" clinical decision rule to guide duration of anticoagulation in women with unprovoked venous thromboembolism. Can I use any d-Dimer? Thromb Res 2018;169:82-86. PMID: 30031290