ASA vs rivaroxaban for extension of VTE prophylaxis after hip/knee arthroplasty

Anderson DR, et al. Aspirin or rivaroxaban for VTE prophylaxis after hip or knee arthroplasty. NEJM 2018;378:699-707.

Bottom line: In patients who underwent elective hip or knee replacement with rivaroxaban 10 mg/d up until POD #5, switching to ASA 81 mg/d on POD #6 is no worse than continuing rivaroxaban 10 mg/d for preventing VTE, with similar rates of bleeding. 

Patients (n=3427)

  • Setting: 15 Canadian university-affiliated hospitals from Jan 2013 to April 2016
  • Included: Elective unilateral hip (THA) or knee arthroplasty (TKA), primary or revision
  • Excluded:
    • Hip or lower limb fracture in previous 3 months
    • Metastatic cancer
    • Did NOT exclude patients who received ASA pre-op (they could continue open-label ASA <100 mg/d in addition to blinded study drug)
  • Baseline characteristics:
    • Age 63 y
    • Male 48%
    • Joint operated: Hip (53%), knee (47%)
    • VTE risk factors: Prior VTE 2-3%, previous surgery ~3%, cancer 2-3%, smoker ~9%
    • BMI 31
    • Post-op mechanical compression: IPCs (50%), graduated stockings (42%), both (8%)
    • Hospital LOS 3.5 days
    • Peri-op tranexamic acid 54.5%
    • Pre-op ASA 25%

Intervention & control

  • All patients received rivaroxaban 10 mg once daily on POD #0 (starting >6h after wound closure) or POD #1, & was continued until POD #5
  • Intervention: Switched to ASA 81 mg daily on POD #6
  • Control: Continued Rivaroxaban 10 mg once daily
  • Duration of study drug in both groups:
    • Hips x30 days (total 35 days of VTE prophylaxis)
    • Knees x9 days (total 14 days of VTE prophylaxis) 

Results @ 90 days

  • Primary efficacy outcome (symptomatic DVT or PE): ASA 0.6% vs rivaroxaban 0.7%
    • Absolute risk difference -0.06% (upper end of 95% confidence interval [CI]: +0.55%), p<0.001 for non-inferiority
  • Mortality: 0.06% (1 death from PE) vs 0%
  • Primary safety outcome (major or clinically-relevant non-major bleed): 1.3% vs 1.0%
    • All consisted of overt bleeding at the surgical site; most occurred between POD #6-16
  • Major bleed: 0.5% vs 0.3%

Generalizability (external validity)

  • Widely applicable to patients who underwent elective THA/TKA (whether they received DOAC or LMWH from POD #0-5)

Risk of bias: Low (high internal validity)

  • Low risk of allocation, performance, detection or attrition bias
    • Random sequence generation: using permuted-block design, stratified by THA or TKA, centre, & use of open-label ASA
    • Allocation concealment: patients & personnel (central pharmacy staff aware)
    • Blinding of patients & personnel: ASA & rivaroxaban administered in identical-appearing opaque gelatin capsules
    • Blinding of outcome assessors: Independent adjudication committee unaware of treatment allocation
    • ITT analysis
    • 1 patient lost to follow-up of VTE, but vital status known (alive @ 90 days)
  • Non-inferiority trial design
    • Non-inferiority margin set at <1.0% absolute risk difference justified based on survey of Canadian thromboembolism specialists & orthopedic surgeons (reference not provided)
    • ITT analysis used for primary analysis; on-treatment analysis consistent with ITT analysis:
      • Primary outcome: 0.4% vs 0.4%
      • Bleeding: 0.9% vs 0.75%
    • No safety advantage for ASA demonstrated in this trial, however, non-inferiority design is justifiable based on substantially-lower cost of ASA (<$5 for 1 month) compared to rivaroxaban ($3/day for 10 mg-tab).

SMART-DATE - 6 vs 12 months of DAPT after PCI for ACS

Hahn JY, et al. 6-month versus 12-month or longer dual antiplatelet therapy after percutaneous coronary intervention in patients with acute coronary syndrome (SMART-DATE): a randomised, open-label, non-inferiority trial. Lancet 2018 [epub]

Bottom line: In patients with ACS who underwent PCI, shorter DAPT (~6 months) increased the risk of recurrent MI (NNH 100) without any clear reduction in bleeding versus a standard DAPT duration of 12+ months.

Patients (n=2712)

  • Setting: South Korea from Sept 2012 to Dec 2015
  • Included:
    • ACS (STEMI, NSTEMI or unstable angina)
    • At least 1 lesions in a native coronary vessel with stenosis >50% amenable to PCI with stents
  • Excluded: Contraindication to antiplatelets, drug-eluting stent coating, or contrast media; Active bleeding, major bleeding within 3 months, or major surgery within 2 months; History of bleeding diathesis or known coagulopathy; Planned elective surgical procedure within next 12 months
  • Baseline characteristics:
    • Age 62 y
    • Women 25%
    • ACS subtype: STEMI 38%, NSTEMI 31%, unstable angina 31%
    • Prior MI 2%, previous revasc 5%
    • Angiography: Multivessel CAD 45%, left main 2%, LAD 59%, bifurcation lesion 9%
    • Other PMHx: Smoker 39%, HTN 49%, diabetes 27%

Intervention & control

  • Intervention: Short DAPT duration (x6 months)
    • Median DAPT duration 6.1 months
    • Adherence to study protocol: 74%
    • Used clopidogrel as P2Y12 inhibitor: 80%
  • Control: Standard DAPT duration (x12+ months)
    • Median DAPT duration 17.7 months
    • Adherence to study protocol: 96%
    • Used clopidogrel as P2Y12 inhibitor: 82%
  • Both groups: DAPT consisted of ASA + P2Y12 inhibitor at standard doses

Results (from PCI to 18 months after PCI)

Efficacy

  • 1o outcome (composite of all-cause mortality, MI, stroke): DAPTx6months 4.7%, x12+ months 4.2%, hazard ratio (HR) 1.13 (0.79-1.62)
    • Absolute risk difference 0.5% (upper limit 1.8% < 2.0% non-inferiority margin, p=0.03 for non-inferiority)
    • Post-hoc landmark analysis from month 6 to 18: HR 1.69 (0.97-2.94)
  • Death: 2.6% vs 2.9%, HR 0.90 (0.57-1.42)
  • MI: 1.8% vs 0.8%, HR 2.41 (1.15-5.05) - number needed to harm (NNH) 100
    • Post-hoc landmark analysis from month 6 to 18: HR 5.06 (1.46-17.47)
  • Stroke: 0.8% vs 0.9%
  • Stent thrombosis: 1.1% vs 0.7%

Safety

  • BARC 2-5 bleeding: 2.7% vs 3.9%, HR 0.69 (0.45-1.05)
  • Major bleeding: 0.5% vs 0.8%, HR 0.60 (0.22-1.65)

Generalizability

  • This trial enrolled a generally representative group of Asian patients with ACS undergoing PCI predominantly treated with DAPT consisting of ASA + clopidogrel
  • Study investigators were hesitant to comply with the study protocol; 26% of patients in the 6-month group continued to receive DAPT beyond 6 months
  • Although the authors claimed to randomize at the time of PCI rather than 6 months later in order to avoid "selection bias, resulting in enrolment of low-risk patients," it's likely that any "high-risk patients" would not have been considered & therefore not enrolled into this trial.

Risk of bias

  • Low risk of allocation bias
    • Web-based randomization;
    • Computer-generated block randomization;
    • Stratified by (1) enrolment site, (2) ACS subtype, (3) diabetes (y/n), (4) type of P2Y12 inhibitor after prasugrel/ticagrelor became available, & randomized to 1 of 3 drug-eluting stents (eluting everolimus [Xience Prime], zotarolimus [Resolute Integrity], or biolimus [BioMatrix Flex]).
  • High risk of performance & detection bias
    • Randomization occurred immediately after PCI rather than at month 6, allowing for differences in performance & collection of outcomes prior to patients receiving allocated treatment (stopping or continuing DAPT at 6 months);
    • Participants & personnel not blinded to treatment allocation;
    • 26% of patients in 6-month group did not comply with the study intervention of stopping DAPT at month 6.
  • Low risk of attrition bias
    • 2.5% loss-to-follow-up (3.0% vs 1.9%);
    • ITT analysis.
  • Non-inferiority design not clearly justified
    • Wide non-inferiority margin of 2.0% absolute risk difference justified based on feasibility (assuming 4.5% incidence of 1o outcome @ 18 months in 12-month group);
    • Non-inferiority criteria based on above wide non-inferiority margin met for primary outcome, however, MI - key component of this composite - was clinically and statistically significantly higher in the short DAPT group;
    • Inclusion of period from PCI to month 6, when both groups were intended to receive DAPT, biases the results toward non-inferiority. The post-hoc landmark analysis evaluating outcomes from month 6 to 18 demonstrates a greater difference between groups, supporting the conclusion that DAPT x6 months is inferior to 12+ months;
    • Similar results between ITT & per-protocol analysis, although with the above limitations.

TREAT - Ticagrelor vs clopidogrel after fibrinolytic therapy for STEMI

Berwanger O, et al. Ticagrelor vs clopidogrel after fibrinolytic therapy in patients with ST-elevation myocardial infarction: a randomized clinical trial. JAMA Cardiol 2018

Bottom line:

  • In patients <75 y/o who received fibrinolytic therapy plus a loading dose of clopidogrel for STEMI & already not actively bleeding from this, switching to ticagrelor with a loading dose 8-24h after administration of the fibrinolytic agent does not appear to increase the risk of major, fatal or intracranial bleeding at 30 days versus continuing with clopidogrel.

  • Follow-up in TREAT is ongoing; 12-month outcomes as well as meta-analysis with PLATO STEMI subgroup will likely be reported in 2019.

Patients (n=3799)

  • Setting: 10 countries including Canada, enrolled from Nov 2015 to Nov 2017
  • Included:
    • 18-75 y/o
    • STEMI presenting <24h after symptom onset
    • Received fibrinolytic therapy
  • Excluded:
    • Contraindication to clopidogrel, ticagrelor or fibrinolysis
    • Use of oral anticoagulant therapy
    • Increased risk of bradycardia (not further defined)
    • Concomitant use of a strong CYP3A4 inhibitor/inducer
    • Dialysis-dependent
    • Clinically-important anemia or thrombocytopenia, or active bleeding (therefore excluding those who bled early with the initial fibrinolytic+antiplatelet regimen)
  • Baseline characteristics:
    • Age 59 y
    • Female 23%
    • White 57%, Asian 33%
    • STEMI type: Anterior only (~34%), inferior only (~31%), LBBB only (1%)
    • Killip class 2-4: 8-9%
    • PMHx: MI (8-9%), stroke (4-5%), PCI (5-6%), CABG (<1%)
    • Meds (baseline + in-hospital):
      • ASA 99%
      • Anticoagulant: Heparin (40%), LMWH (69%), fondaparinux (4%), bivalirudin (~1%), warfarin (<1%)
      • ACEI ~60%, ARB ~10%
      • Beta-blocker 74%
      • Statin 93%
      • PPI ~55%
    • PCI during initial ACS hospitalization: ~56% (DES ~34%)

Intervention & control

  • Intervention: Ticagrelor 180 mg PO x1, then 90 mg BID (plan to continue 12+ months per standard ACS management)
    • Mean 90% of doses taken within 30 days
  • Control: Clopidogrel 300-600 mg PO x1, then 75 mg daily (plan to continue 12+ months per standard ACS management)
    • Mean 91% of doses taken within 30 days
  • Co-interventions:
    • Clopidogrel dose administered before randomization: >300 mg (3%), 300 mg (87%), none or <300 mg (10%)
    • All received ASA 75-100 mg daily unless intolerant
    • Fibrinolytic selection: TNK ~40%, alteplase ~20%, reteplase, 17%, other 23%
    • Median 2.6 h from symptom onset to fibrinolytic administration
    • Median 11.4 h from fibrinolytic administration to randomization
    • PCI during hospitalization in 56%, drug-eluting stent in 34%

Results @ 30 days (follow-up up to 12 months ongoing)

Bleeding

  • 1o outcome (major bleed, TIMI definition): Ticagrelor 0.7%, clopidogrel 0.7%
    • (95% confidence interval for absolute risk difference -0.5% to +0.6%, p<0.001 for non-inferiority based on a non-inferiority margin of 1.0%)
  • Other bleeding definitions:
    • Intracranial hemorrhage: 0.4% vs 0.4% (95% CI -0.35% to +0.45%)
    • Fatal bleeding: 0.2% vs 0.1% (95% CI -0.2% to +0.3%)
    • BARC type 3-5: 1.2% vs 1.4%
    • PLATO, major: 1.2% vs 1.4%
    • TIMI
      • Major non-CABG: 0.7% vs 0.6%, p=0.57
      • Requiring medical attention: 2.0% vs 1.3%, p=0.06
      • Minor: 0.4% vs 0.6%, p=0.47
  • TIMI major bleed based on time from fibrinolytic administration to administration of study antiplatelet
    • <4h: 1.5% vs 1.2%
    • 4-8h: 0.8% vs 1.2%
    • 8-16h: 0.5% vs 0.3%
    • 16+h: 0.5% vs 0.2%

Other safety outcomes

  • Dyspnea: 13.9% vs 7.6% (NNH 16)
  • Premature discontinuation of study drug <30 days: 8.1% vs 6.4%
    • Due to dyspnea: 1.2% vs 0%
    • Due to patient refusal: 1.5% vs 0.3%

Efficacy

  • Death from any cause: 2.6% in both groups (HR 0.99, 0.66-1.47)
  • Vascular death, MI, stroke: 4.0% vs 4.3% (hazard ratio [HR] 0.91, 95% CI 0.67-1.25)
    • MI: 1.0% vs 1.3% (HR 0.79, 0.44-1.42)

Generalizability

  • TREAT enrolled a representative population of STEMI patients who received fibrinolytic therapy.
  • Note that the logistics of this trial are key to interpretation and application. Fibrinolytic therapy is generally reserved for patients who cannot get to a PCI-capable hospital in a reasonable timeframe, & is therefore administered in the pre-hospital or community hospital setting. This trial was undertaken at academic sites, and therefore generally enrolled patients & administered the study therapy hours after administering a fibrinolytic, with 90% having already received a clopidogrel load.
    • This is therefore NOT a trial comparing SIMULTANEOUS fibrinolysis + ticagrelor vs fibrinolysis + clopidogrel, but rather a trial comparing a switch from clopidogrel to ticagrelor within 24h of administering a fibrinolytic. With a median time from fibrinolytic administration to study P2Y12 administration of 11.4h, the fibrinolytic was long-gone by the time they entered the study (e.g. half-life <30 min for 3 most-commonly-used fibrinolytics)
    • Therefore the most direct application of these results would be to administer a loading dose of clopidogrel 300 mg PO with the fibrinolytic, & then switch to ticagrelor by starting with a loading dose 8-24h later (ensuring that the fibrinolytic is eliminated, & therefore pharmacodynamic interaction & bleed risk is minimized)

Internal validity

  • Low risk of allocation bias due to use of an automated web-based system in permuted blocks of 4 stratified according to site
  • Unclear risk of performance and detection bias due to open-label design with blinded adjudication of outcomes (though low risk for important "hard" endpoints of death, major, fatal and intracranial bleeds)
  • Low risk of attrition bias due to very low loss-to-follow-up (0.1-0.2%) & use of ITT analysis
  • Non-inferiority design was appropriate, design decisions were well-justified & conclusion of non-inferiority is reasonable based on threshold set & consistency in analyses
    • Justified based on potential long-term benefits of ticagrelor as observed in the PLATO trial (i.e. to 12 months after ACS)
    • Non-inferiority margin for major bleed set at an absolute risk increase of 1.0%, which the authors justify empirically based on thresholds used in other non-inferiority RCTs
    • Analyzed 3 separate definitions of "major bleed" using both ITT & per-protocol analyses, which were all nearly identical & consistent

AVERROES - Apixaban versus ASA in patients with AF not suitable for warfarin

Connolly SJ, et al. Apixaban in patients with atrial fibrillation. N Engl J Med 2011;364:806-17.

Bottom line:

  • In patients with non-valvular AF, apixaban is more effective at reducing stroke risk than ASA (relative risk reduction 55%; NNT ~46/year), with a small increase in minor bleeding (NNH 84/year) but no significant increase in major bleeding;

  • This favorable benefit/risk profile of apixaban over ASA was present even in patients with a CHADS2 score of 0-1.

Patients

  • Included:
    • 50+ y/o
    • AF (documented within 6 months before enrolment or by 12-lead EKG)
    • At least 1 stroke risk factor (any CHADS2 criteria or PAD):
      • HF (NYHA class 2-4 symptoms or LVEF 35% or less), HTN, Age 75+, diabetes (on treatment), prior stroke/TIA, or documented PAD
    • Not receiving a warfarin because previously demonstrated to be "unsuitable" or expected to be unsuitable
  • Excluded:
    • Additional indication for anticoagulation other than AF
    • Serious bleeding within 6 months
    • High risk of bleeding (eg active peptic ulcer, plt <100, Hb <100 g/L, stroke within 10 days, blood dyscrasias)
    • Serum creatinine >221 umol/L or CrCl <25 mL/min
  • Baseline characteristics:
    • Age 70, male 59%
    • AF type: Paroxysmal (27%), persistent (21%), permanent (52%)
    • CHADS2 mean 2 (0 or 1 in 36%)
    • Stroke risk factors: Clinical HF (40%), LVEF <35% (5%), HTN (86%), diabetes (20%), prior stroke/TIA (14%)
    • Most common reason warfarin was "unsuitable" (multiple reasons in 51%):
      • Unable to measure INR frequently enough 43%
      • Patient refused to take warfarin 38% (the only reason in 15%)
      • CHADS2=1 so warfarin not recommended by physician 21%
      • Unable to keep INR therapeutic 17%
      • Unsure if patient can adhere to instructions to take warfarin 16%

Intervention & control

  • Intervention: Apixaban 5 mg PO BID
    • Decreased to 2.5 mg BID if 2/3 of: Age 80+ y, wt <60 kg, SCr >132 umol/L (occurred in 6%)
  • Control: ASA 81-324 mg/d (64% on 81 mg/d)

Results @ mean 1.1 years

  • Efficacy
    • Primary outcome (any stroke or systemic embolism): Apixaban 1.8% versus ASA 4.0%, hazard ratio (HR) 0.45 (0.32-0.62), NNT=46
    • Death from any cause: 4.0% vs 5.0%, HR 0.79 (0.62-1.02)
    • CV hospitalization: 13.1% vs 16.3%, HR 0.79 (0.69-0.91), NNT=32
  • Safety
    • Major bleed (overt bleed with Hb decrease 20+ g/L over 24h, transfusion 2+ units of RBCs, or bleeding at a critical site [e.g. brain, eyes, pericardium, retroperitoneum): 1.6% vs 1.4%, HR 1.13 (0.74-1.75)
      • Intracranial: 0.4% vs 0.5%, HR 0.85 (0.38-1.90)
      • Extracranial: 1.2% vs 1.0%, HR 1.23 (0.74-2.05)
    • Minor bleed: 6.7% vs 5.5%, HR 1.24 (1.00-1.53), NNH=84
    • Serious adverse events: 22% vs 27%, NNT=20
  • Subgroup analysis by baseline CHADS2 score demonstrated consistent relative risk reductions with apixaban over ASA regardless of score, with higher-risk patients deriving greater ABSOLUTE reductions in stroke (NNT=143/year for CHADS2=0 to 1, NNT=23/year for CHADS2=3+)

Generalizability

  • Representative sample of elderly patients with AF & a wide spectrum of stroke risk who had a difficult time maintaining INRs in the therapeutic range, going to the lab for INR monitoring, or who were expected not to do well with warfarin based on clinical judgement. Results were similar regardless of the reason for being unsuitable for warfarin.

Internal validity

  • Low risk of allocation, performance, detection, attrition, selective reporting bias
    • Central, computerized, automated randomization
    • Double-dummy blinding
    • Blinded outcome adjudication
    • No patients lost to follow-up
    • All relevant & important outcomes reported
  • Trial stopped early after 1st interim analysis for efficacy based on 104 events between groups