Direct Oral Anticoagulants: Drug Selection by Means of the SOJA Method

Robert Janknegt1*, Roel Van Kampen2, Niels Boone3 and Renée Vossen4 1PhD, Hospital Pharmacist, Clinical Pharmacologist, Zuyderland Medical Centre, The Netherlands 2MD, PhD, Haematologist, Oncologist, Zuyderland Medical Centre, The Netherlands 3Hospital Pharmacist, Clinical Pharmacologist, Zuyderland Medical Centre, The Netherlands 4Clinical Chemistry Specialist, Zuyderland Medical Centre, The Netherlands


Approved indications
The number of licensed indications is a good measure of the applicability and documentation of the drugs. The fact that a drug is approved for [almost] all indications listed below is, from a formulary point of view, advantageous to another drug, which is approved for only one or two applications.
The percentage of the maximum score for approved indications was obtained as follows:

Indication
Maximum Score (%) Prevention of DVT and pulmonary embolism after orthopaedic surgery 20 Treatment and prevention of venous thromboembolism 25 Prevention of stroke during atrial fibrillation 25 Treatment of acute coronary syndromes 15 Coronary artery disease (CAD) or symptomatic peripheral artery disease (PAD) 15 Table a Number of available formulations A large number of available ready to use doses offer the possibility to give each patient an optimal dosage with minimal manipulation of the product.
This was scored as follows [percentage of the relative weight] Solid oral formulation 80% Liquid or dispersible oral formulation 20% Table b Variability of the area under the curve A therapy may fail because of great differences in bioavailability and incomplete absorption. The score is based on the variability of bioavailability of the drugs after oral administration. This was scored as follows: The score for each product was inversely proportional to the coefficient of variation of the AUC. If the CV was 46%, the drug scored [100-46] 54% for this criterion.

Drug interactions
Interactions play a role only in patients who use other drugs which may interact with DOACs. However, it is a relevant criterion from a formulary point of view.
The score for each drug was dependent on the frequency and severity of observed drug interactions.

Clinical efficacy
The results of clinical studies were taken into account to judge the clinical efficacy of DOACs. The better the clinical efficacy, the higher the score for this medicine. All approved indications were scored separately.
Clinical efficacy was only scored for indications applicable to all DOACs.

Side effects
The extent and the severity of adverse effects is another important selection criterion for drugs. A distinction was made between "minor" side effects, such as gastrointestinal disturbances or skin reactions, occurring in clinical trials and severe or even lifethreatening adverse reactions observed with large scale use of the drugs. The evaluation of the "minor" adverse effects was based on results of double blind comparative clinical studies.

Dosage frequency
The dosage frequency plays an important role in patient compliance. Compliance is not usually a problem in patients taking the drugs once or twice daily, but decreases considerably in the event dosage frequency is higher than twice daily. The method of evaluation of this criterion corresponded with that of all of the other SOJA scores. Prevention of stroke during atrial fibrillation 50% According to frequency detailed above Table d Documentation The score for this criterion was divided over 4 subcriteria: The 5% of the relative weight for this sub criterion was awarded for each randomised comparative study.

Number of patients in these studies
Besides the number of clinical studies, the number of patients that were treated with the drug in question must also be taken into consideration.

Number of available formulations
The following solid oral presentations are available: • Apixaban: 2.5 mg and 5 mg   Table g In general a low variability of the pharmacokinetics is found for most drugs. The variability of dabigatran pharmacokinetics is higher than that of the other medicines.

Pharmacokinetic interactions
Summary of the effect on AUC of individual DOACs:
clopidogrel] monotherapy was permitted and resulted in increased clinically relevant bleeding although with a lower risk of bleeding on edoxaban compared to warfarin.

NSAIDs
Co-administration of naproxen and edoxaban increased bleeding time relative to either medicine alone. Naproxen had no effect on the Cmax and AUC of edoxaban. In clinical studies, co-administration of NSAIDs resulted in increased clinically relevant bleeding. Chronic use of NSAIDs with edoxaban is not recommended. There are no major differences in the incidence and severity of drug interactions between individual DOACs. The extent of drug interactions appears to be a bit smaller for edoxaban.

CYP3A4 and P-gp inhibitors
This drug is awarded 50%, the other medicines are awarded 45%.

Clinical efficacy
The judgement of the relative clinical efficacy should ideally be based on a large number of double-blind direct comparative studies between the DOACs using clinically relevant endpoints. Unfortunately, there are no direct comparative studies between two or more DOACs.
The main results of comparative studies are summarised in tables 1-21. Statistically significance was indicated when applicable.

Apixaban
Four randomized comparative studies were performed between apixaban and enoxaparin. The first study was a dose-finding study using enoxaparin and open-label warfarin as comparators. No relevant differences between the 3 drugs became apparent. The

Atrial fibrillation
Vitamin K antagonists such as warfarin or acenocoumarol are the drugs of choice to prevent stroke in patients with atrial fibrillation. A meta-analysis showed that warfarin was substantially more effective than antiplatelet agents [76].
The key results of all studies are summarized in tables 6-12.
Apixaban Two large-scale comparative studies were performed with apixaban.
The Re-sonate study compared dabigatran to placebo, using similar methodology and endpoints as the study described above.
The primary endpoint was seen significantly less frequent for dabigatran compared to placebo: 0.4% vs 5.6%. The incidence of myocardial coronary syndromes was similar for both drugs: 0.1% [57].

Edoxaban
Two studies were performed with edoxaban by the fact that there are no direct comparative studies between two or more DOACs for any indication.

Orthopaedic surgery
Apixaban was more effective than enoxaparin regarding the primary endpoint in 2 of 4 studies, dabigatran in 1 of 5 studies, edoxaban in 1 of 4 studies and rivaroxaban in all 4 studies.

Atrial fibrillation
Apixaban was more effective than warfarin regarding the primary endpoint in all 2 studies, dabigatran in one study, although this was only valid for the high dose, edoxaban in 0 of 1 study and rivaroxaban in 1 of 2 studies. A recent meta-analysis showed the best result for apixaban 5 mg twice daily [125].
Dabigatran has shown a reduction of the clinical endpoint ischaemic stroke and a reduced mortality compared to warfarin in the Re-Ly study [26].
Apixaban and dabigatran are awarded a higher score than rivaroxaban and edoxaban, because of a documented positive effect on the incidence of ischaemic stroke.

Deep venous thrombosis
All comparative studies were designed to demonstrate non inferiority compared to enoxaparin. This was confirmed in all studies. There are no indications for major differences in clinical efficacy between the DOACs, but direct comparative data are lacking. Edoxaban was assigned a lower score because of limited clinical data.
It should be stated that no direct comparative studies have been performed, which limits the value of [network] meta-analyses to a certain extent.
The overall score for efficacy was calculated as follows: Score:

Bleeding
The main results of comparative studies regarding safety are summarised in tables 1-21 .

Orthopaedic surgery
Four randomized comparative studies were performed between apixaban and enoxaparin. No differences between both medicines became apparent regarding safety, both regarding bleeding and nonbleeding adverse events [26][27][28][29].

Atrial fibrillation
There were no significant differences between apixaban and aspirin in the incidence of bleeding in the Averroes study [44].
Apixaban was associated with a lower incidence of bleeding compared to warfarin in the Aristotle study, the incidence of major bleeding, intracranial bleeding and any bleeding was significantly lower [45]. The rate of major bleeding was 2.1% per year in the apixaban group versus 3.1% in the warfarin group.
Major extracranial hemorrhage was associated with reduced hospitalisation, medical or surgical intervention or transfusion for apixaban compared to warfarin [126].
A history of bleeding at baseline was associated with higher risk of major bleeding [127].

Treatment or prophylaxis of deep venous thrombosis
Apixaban showed a lower incidence of bleeding compared to enoxaparin followed by warfarin in the Amplify study. Both the incidence of major bleeding [0.6% vs 1.8%] and the incidence of On the other hand, apixaban showed a higher incidence of major bleeding compared to enoxaparin in the Adopt trial [54].

Acute coronary syndromes
The Appraise study was a phase 2 dose-finding study comparing apixaban dosages of 2.5 mg twice daily [n = 317], 10 mg once daily  dosage arms were prematurely discontinued because of a high incidence of bleeding. The primary safety endpoint was reached in 3.0% for placebo and in 5.7% and 7.9% for the 2.5 mg bid and 10 mg qd dosages of apixaban, respectively. The difference between the high dose of apixaban and placebo was statistically significant, The details of the Appraise 2 study are summarized in Tables 18-21. The incidence of all forms of bleeding were significantly higher for apixaban compared to placebo, without a favourable effect on the efficacy endpoint. Apixaban resulted in a higher incidence of the primary safety endpoint of major bleeding according to the

Orthopaedic surgery
Five randomized comparative studies were also performed between dabigatran and enoxaparin. Again, no differences between both medicines became apparent regarding safety, both regarding bleeding and non bleeding adverse events [30][31][32][33][34].

Atrial fibrillation
Dabigatran resulted in a significantly lower incidence of bleeding compared to warfarin in the Re-Ly study [46]. On the other hand a higher incidence of gastrointestinal bleeding was found for

Treatment or prophylaxis of deep venous thrombosis
Rivaroxaban showed a lower incidence of major bleeding in compared to enoxaparin followed by warfarin in the Einstein

Discussion on safety
• Two meta-analyses showed a lower incidence of bleeding with apixaban compared to the other medicines.
• Apixaban has the lowest incidence of major bleeding, which is best documented in atrial fibrillation.
• Dabigatran also shows a higher incidence of GI side-effects compared to other DOACs. An advantage of dabigatran is the availability of a reasonably priced antidote.
• This results in a 10% lower score for dabigatran compared to apixaban: 70% • Rivaroxaban has a relatively high incidence of major bleeding in an analysis of studies in atrial fibrillation. The study population of the Rocket A study had the highest baseline CHADS/ CHADS-VASc score and thus the highest baseline HAS-BLED score of all NOAC AF-registration trial populations. High CHADS/CHADS-VASc scores drive embolic events and also high HAS-BLED scores drive bleeding.
Relatively limited data are available for edoxaban. Edoxaban also scores 70%.
The scores for side-effects are as follows

Documentation
The documentation of DOACs is summarised below: Prevention of DVT and pulmonary embolism after orthopaedic surgery Apixaban 80% Dabigatran 70% Edoxaban 70% Rivaroxaban 70%      is potential debate on the correct scoring system with respect to each criterion and individual decisions are highly subjective. This is the case with any method used to quantify properties of drugs.
The SOJA method is intended as a tool for rational drug decision making, enabling clinicians and pharmacists to include all relevant aspects of a certain group of drugs, thereby preventing formulary decisions being based on only one or two criteria. Besides this, possible "hidden criteria" [such as personal financial interest] are excluded from the decision making process. The outcome of this study should be seen as the basis for discussions within formulary committees and not as the absolute truth.

Outcome
Relatively limited differences in score are seen between the DO-ACs [about 10% between the highest and lowest score]. Of course, the present scoring is based on the weights assigned by the authors. The essence of the SOJA method is that users of the method may assign their own relative weight to each selection criterion.
This interactive program is available on the internet: www.tablet.
sojaonline.nl. Other relative weights will of course affect the relative scores for the medicines.
The present score is an overall score for all approved indications. It could be argumented that a different selection should be made for each indication, because of differences in documented efficacy between DOACs for the various applications.
It should be stressed that one single DOAC will not be suitable for all patients. All DOACs have advantages and disadvantages.
The relatively high score for rivaroxaban is caused by a favourable score for the number of approved indications, dosage frequency and documentation.
Apixaban shows the highest score for safety. Dabigatran and rivaroxaban show the highest scores for documentation, whereas edoxaban is given once daily for all approved indications and shows few relevant drug interactions.
Dabigatran does not score well for variability of AUC [a criterion that is usually not very relevant for most physicians]. The documentation of edoxaban [as well as the number of approved indications] is more limited than that of the other DOACs. The incidence of side-effects of rivaroxaban and dabigatran appears to slightly higher than that of apixaban, although direct comparative studies are lacking.

Selection criteria
Of course, other selection criteria could be applied as well. We did not include Contra-indications and Warnings and Precautions in the matrix. There were no relevant differences between the DO-ACs in this respect. Differences in the incidence of bleeding or drug interactions were accounted for in the current selection criteria.
Variability of the AUC is a standard criterion for SOJA. Its relevance for DOACs is unclear. That is why a low weight was assigned tot his criterion. When one considers this criterion to be completely irrelevant,a zero weight can be given to this criterion in the interactive program.
Clinical efficacy and safety are the most important selection criteria for all groups of medicines. Unfortunately these criteria are difficult to score for DOACs because of the lack of direct comparative studies and differences in patient populations, study design and applied endpoints. Meta-analyses and registry data may be of value in the judgement of efficacy and safety. All data sources have specific strenghts and weaknesses.
Acquisition cost was not included as a selection criterion to make the score internationally applicable. The present matrix can be used as a pre selection tool of the most suitable DOACs from a quality point of view. Because prices may different institutions and in different healthcare systems, individual procurement procedures should lead to a selection of the best options.
The acquisition cost of the DOACs included in this analysis is relatively high, especially compared to vitamin K antagonists, such as acenocoumarol and warfarin. INR monitoring costs, as well as patients' discomfort of these determinations must be taken into account regarding vitamin K antagonists. On the other hand, detem- ining INR values also contributes to good compliance in users of vitamin K antagonists, whereas such a check is lacking for DOACs.
This should be taken into consideration before preferring DOACs over vitamin K antagonists.
Of course, other selection criteria could be applied as well.
There are some differences in the suitability of DOACs to be included in Baxter rolls or whether or not specific DOACs can be swallowed irrespective of food intake. We did not include these selection criteria. efficacy and safety. This score should therefore be considered as preliminary. On the other hand, it seems quite unlikely that large scale direct comparative studies with more than 2 DOACs will be published in the near future, so we will have to deal with indirect comparisons.
Because of the lack of direct comparative studies, the results of meta-analyses and registries were also taken into consideration.
These kind of studies also have limitations. The quality of metaanalyses is as good as the quality of the studies which are included.
Patient populations may be quite different for patients treated with the individual DOACs in registry studies.

Formulary choices versus decisions in treatment of individual patients
It should be stated that formulary selection is a different process than decision making in individual patients. Selection criteria like variability in AUC, number of approved indications and documentation are typical criteria that may be relevant from a formulary perspective, but not for the selection of a DOAC in individual patients.
The above described differences in properties of DOACs may lead to drug and dosage choices based on the specific situation of the patient, such as comedication [risk of interactions], bleeding risk, personal preference for once daily or twice daily dosing, renal or hepatic function and individual tolerability.

Conclusions
We found limited differences in the scores of the available DO-