GEMINI-ACS-1: Rivaroxaban vs ASA added to P2Y12 inhibitor following ACS

Ohman EM, et al. Clinically significant bleeding with low-dose rivaroxaban versus aspirin, in addition to P2Y12 inhibition, in acute coronary syndromes (GEMINI-ACS-1): A double-blind, multicentre, randomised trial. Lancet 2017 (epub ahead of print)

Bottom-line:

• In patients with ACS, low-dose rivaroxaban produces a similar or greater risk of clinically-significant or major bleed compared to ASA (potential NNH 100), with unclear effect on efficacy outcomes.

• Although this trial is insufficient to change practice, it provides preliminary evidence that will guide further trials refining antithrombotic regimens in various ACS subpopulations.

 

Context

• The ATLAS trial demonstrated that adding low-dose rivaroxaban to clopidogrel-based DAPT reduces CV events and death following PCI, offset by an increase in major bleeds
• The PIONEER trial and other studies have investigated various anticoagulation-based regimens in patients with AF who undergo PCI

Patients (n=3037)

• Included:
  • Age > 18 years
  • ACS (unstable angina, NSTEMI or STEMI)
  • If age <55 years, at least 1 of the following:
    • Diabetes
    • Previous MI
  • If unstable angina, at least 1 of the following:
    • TIMI score 4+
    • Ischemic changes on EKG
    • Revascularization for this event
• Key exclusion criteria:
  • Hx of bleeding, ICH or GI bleed within past year
  • CrCl <20 mL/min
  • Need for full-dose anticoagulation
• Typical study patient
  • Age 62 years
  • Male 75%
  • ACS type: unstable angina 11%, NSTEMI 40%, STEMI 49%
  • Procedure for ACS: Cath 94%, PCI 87% (DES 67%, BMS 33%), CABG <1%
  • Randomized median 5.5 days after ACS
  • Previous MI 22%
  • Previous PCI/CABG 20%/4%
  • PAD 5%
  • CV risk factors
    • Smoker 33%
    • HTN 73%
    • Dyslipidemia 56%
    • Diabetes 30%
  • Meds
    • ACEI/ARB 63%
    • Beta-blocker 64%
    • Statins 69%

Interventions

• I: Rivaroxaban 2.5 mg PO BID for a minimum 180 days
• C: ASA 100 mg PO daily for a minimum 180 days
• Median duration of study drug 9.6 months
• Co-interventions:
  • Clopidogrel 75 mg daily (44%) or ticagrelor 90 mg BID (56%) based on investigator preference (i.e. choice of P2Y12 inhibitor not randomized)
    • Prematurely discontinued in 4%

Results @ median 10.7 months

• Primary outcome (TIMI non-CABG clinically significant bleeding): 5% in both groups, hazard ratio (HR) 1.09 (0.80-1.50)
• Results vary based on bleeding definition
  • ISTH major bleeding: Rivaroxaban 2% versus ASA 1% (NNH 100), HR 1.83 (1.01-3.31)
  • BARC 3a or higher: 1% in both groups, HR 1.70 (0.85-3.37)
  • GUSTO life-threatening or severe: <1% in both groups, HR 1.50 (0.25-8.95)
  • TIMI major bleeding: 1% in both groups, HR 1.25 (0.49-3.17)
• Efficacy outcomes
  • Death: Rivaroxaban 1%, ASA 1.5%, HR 0.95 (0.53-1.71)
  • CV death, MI, stroke or definite stent thrombosis: 5% in both groups, HR 1.06 (0.77-1.46)
  • Any stent thrombosis: 1% in both groups, HR 1.06 (0.54-2.11)

Internal validity

• Low risk of bias (including allocation, performance, detection, attrition and reporting biases)
  • Computer-generated randomization
  • Centralized, electronic allocation (allocation concealment)
  • Blinding using identical placebos
  • <0.5% (1 patient) lost to follow-up

Generalizability & applicability

• Excluded patients with an indication for full-dose anticoagulation (e.g. AF, LV thrombus, VTE)
• Rivaroxaban dose is dose found to be efficacious in ACS in ATLAS trial, but only 25% of dose demonstrated to have efficacy in AF & VTE
• Majority used ticagrelor as P2Y12 inhibitor
• Phase 2 trial:
  • Not powered to compare low-dose rivaroxaban to ASA for: Death, major bleed, and efficacy (including MI, stroke or stent thrombosis)
  • Not formally designed as a non-inferiority trial, but primary analysis met wide pre-defined non-inferiority margin (HR <2.0)

Beta-blocker side-effects

Ko DT, et al. Beta-blocker therapy and symptoms of depression, fatigue, and sexual dysfunction. JAMA 2002;288:351-7.

Bottom line: In patients with HF, MI or HTN, beta-blockers increased the risk of fatigue (NNH 34) & sexual dysfunction (NNH 24), particularly erectile dysfunction in men.

Design

• Systematic review and meta-analysis of 15 trials (n= ~35,000) published up to 2001
• Included RCTs HF, MI or HTN with >100 patients & >6 months of follow-up

Results @ 0.5-6 years

• Fatigue: Beta-blockers 33.4% vs placebo 30.4% (NNH 34), relative risk (RR) 1.15 (1.05-1.26)
  • Absolute risk increase 1.8%/year
  • Risk greater for older beta-blockers (e.g. propranolol; RR 1.78) than new beta-blockers (e.g. atenolol, metoprolol; RR 1.06)
• Sexual dysfunction: Beta-blockers 21.6% vs placebo 17.4%, RR 1.10 (0.96-1.25)
  • Impotence in men: RR 1.22 (1.05-1.41)
• Depression: Beta-blockers 20.1% vs placebo 20.5%, relative risk RR 1.12 (0.89-1.41)
• None of the risks differed based on lipid solubility

FOURIER - Evolocumab in patients with CVD

FOURIER Steering Committee and Investigators. Evolocumab and clinical outcomes in patients with cardiovascular disease. NEJM [online ahead of print]

Bottom line:

• In patients with ASCVD & high CV risk (~5%/year), evolocumab reduced the risk of CV events (NNT 67 over 2.2 years, i.e. ~150/year) with a ~2% risk of injection-site reactions & no other safety concerns.

• LDL-lowering with evolocumab did not reduce the risk of death.

• Although FOURIER is a high-quality trial that conclusively demonstrates that lowering LDL with a PCSK9 inhibitor reduces CV risk, it's not clear from this study for whom this reduction is clinically important enough to justify the substantial cost of these drugs.

• EBBINGHAUS substudy: Evolocumab did not differ in placebo in change in cognitive function over 19 months.

 

Context

• There's been substantial controversy around the LDL hypothesis in the past few years
  • For the longest time, statins were the only lipid-lowering agents with robust evidence of efficacy in reducing the risk of CV events
  • In 2013, American guidelines flipped their recommended approach from "treat to LDL target" to "treat to risk, whereas Canadian guidelines maintained the treat-to-target approach (though emphasis shifted further from non-statins such as fibrates & niacin)
  • In 2015, the IMPROVE-IT trial of ezetimibe in post-ACS patients, though underwhelming some with its fairly unimpressive absolute risk reduction in CV events, provided support for the LDL hypothesis by offering another class of agents which reduces CV risk seemingly via LDL reduction
  • PCSK9 inhibitors, potent LDL-lowering agents, were thus positioned as a "third pillar" for the LDL hypothesis
• Previous meta-analyses, based almost exclusively on statin trials, demonstrated a reduction of CV events with lipid-lowering proportional to LDL reduction; relative risk reduction (RRR) per 1-mmol/L LDL reduction (estimates from Cholesterol Treatment Trialists' Collaboration meta-analysis):
  • Death ~10-15%
  • Major vascular events (MI, stroke, coronary revascularization) ~20-25%
    • Non-fatal MI or coronary death ~20-25%
    • Coronary revascularization ~20-25%
    • Stroke ~10-20%
• The open-label OSLER trial, the largest RCT of evolocumab prior to FOURIER, found a statistically significant reduction in CV events (NNT 82 @ 11 months) with the ~60% reduction in LDL with evolocumab over placebo, but was based on a very small number of outcome events

Patients (n=27,564)

• Included:
  1. Age 40-85 y
  2. Atherosclerotic cardiovascular disease (ASCVD), history of any of the following:
     • MI
     • Non-hemorrhagic stroke (TIA does not count)
     • Symptomatic PAD (claudication + ABI <0.85, peripheral artery revascularization or amputation due to atherosclerotic disease)
  3. LDL (most recent) >1.8 mmol/L or non-HDL >2.6 mmol/L after 2+ weeks of stable lipid-lowering therapy (max-tolerated statin, "preferentially" high-intensity with equivalent of atorvastatin 20+ mg/d +/- ezetimibe)
  4. Fasting triglycerides <4.5 mmol/L
  5. Additional risk factors: 1+ major or 2+ minor
     • Major:
       • Age 65+
       • MI/non-hemorrhagic stroke in prior 6 months
       • Multiple MI/non-hemorrhagic stroke or PAD+MI/non-hemorrhagic stroke
       • Current daily smoker
       • Diabetes
     • Minor
       • Hx non-MI-related coronary revascularization
       • Residual CAD 40+% stenosis in 2+ large vessels
       • HDL (most recent) <1.0 mmol/L
       • LDL (most recent) >3.4 mmol/L or non-HDL >4.1 mmol/L
       • Metabolic syndrome
• Key exclusion criteria (list of all 25 in supplemental appendix)
  1. Any hx hemorrhagic stroke
  2. MI/stroke within 4 weeks
  3. Planned/expected cardiac surgery or revascularization within 3 months
  4. HF NYHA III-IV or last-known LVEF <30%
  5. Uncontrolled/recurrent VT
  6. BP >180/110 mm Hg
  7. eGFR <20
  8. Uncontrolled hypo- or hyperthyroid
• Average study patient
  • Age 62.5 y
  • Male 75%
  • North American 17%
  • Type of ASCVD:
    • MI 81% (most recent median 3.4 y before enrolment)
    • Non-hemorrhagic stroke 19% (most recent median 3.2 y before enrolment)
    • PAD 13%
  • CV risk factors
    • Current smoker 28%
    • HTN 80%
    • Diabetes 37%
  • Labs
    • Total cholesterol 4.3 mmol/L
    • HDL 1.1 mmol/L
    • LDL 2.4 mmol/L
    • Triglycerides 1.5 mmol/L
    • Lipoprotein(a) 37 nmol/L
  • Meds
    • Statins: High-intensity 69%, moderate-intensity 30%, low-intensity <1%
    • Ezetimibe 5%
    • Antiplatelet 92%
    • ACEI/ARB 78%
    • Beta-blocker 76%

Interventions

• I: Evolocumab
  • Either 140 mg q2 weeks or 420 mg q1 month (per patient preference)
• C: Placebo

Results @ 2.2 years

• Lipid parameters
  • LDL, median
    • Baseline: 2.4 mmol/L
    • @ week 48: Evolocumab 0.8 mmol/L, placebo  (-1.45 mmol/L vs placebo, ~60% reduction)
      • <0.65 mmol/L: 42% vs 0%
      • <1.0 mmol/L: 67% vs 0.5%
      • <1.8 mmol/L: 87% vs 18%
  • ApoB reduced by 46% with evolocumab
  • Lp(a) reduced by 27% with evolocumab
• Efficacy
  • No difference in death: 3.2% vs 3.1%, hazard ratio (HR) 1.04 (0.91-1.19)
  • Reduction in composite primary outcome (CV death, MI, stroke, hospitalization for unstable angina, coronary revascularization): 9.8% vs 11.3% (NNT 67), HR 0.85 (0.79-0.92)
    • Reduction in Cholesterol Treatment Trialists "major vascular events": 9.2% vs 11.0% (NNT 56), HR 0.83 (0.77-0.90)
    • Reduction in "key" secondary composite efficacy outcome (CV death, MI, stroke): 5.9% vs 7.4% (NNT 67), HR 0.80 (0.73-0.88)
      • CV death: 1.8% vs 1.7%, p=0.62
      • MI: 3.4% vs 4.6% (NNT 84), HR 0.73 (0.65-0.82)
      • Stroke: 1.5% vs 1.9% (NNT 250), HR 0.79 (0.66-0.95)
      • Hospitalization for unstable angina: 1.7% in both groups, p=0.89
      • Coronary revascularization: 5.5% vs 7.0% (NNT 67), HR 0.78 (0.71-0.86)
• Safety:
  • Only statistically-significant difference: Injection-site reaction: 2.1% vs 1.6% (NNH 200)
  • Serious adverse events: ~25% in both groups
  • Permanent drug discontinuation: 12% vs 13%
  • Any adverse event: 77.4% in both groups
  • Specific adverse events
    • Allergic reaction: ~3% in both groups
    • Neurocognitive event: 1.6% vs 1.5%
    • Cataract: 1.7% vs 1.8%
    • ALT/AST >3x ULN: 1.8% in both groups
    • New-onset diabetes (in those without diabetes at baseline): 8.1% vs 7.7%, non-significant relative risk (RR) 1.05
    • Muscle-related event: 5% in both groups
      • CK >5x ULN: 0.7% in both groups
      • Rhabdomyolysis: 0.1% in both groups
  • Development of evolocumab-binding antibodies in evolocumab group: 0.3% (none were neutralizing antibodies)
• Results consistent within all tested subgroups (including q2 week vs monthly dosing); RRR similar regardless of baseline LDL

Generalizability

• Main consideration that should limit over-generalizing the findings of this trial: High cost of evolocumab (Repatha; ~$8000/year in Canada)
  • NNT to prevent 1 CV event/year ~150 in FOURIER
  • Cost of preventing 1 non-fatal CV event with evolocumab (treating 150 patients/year): ~1.2 million
• Population:
  • Patients in this trial were at very high CV risk (~5%/year) owing to requirement of existing ASCVD + additional risk factors. Extrapolating the NNT/year to lower-risk groups:
    • Secondary prevention or high-risk primary prevention (>2%/year): ~300 (cost ~$2.4 million/non-fatal CV event prevented)
    • Intermediate risk (1-2%/year): 450 (cost ~$3.6 million/non-fatal CV event prevented)
    • Low-risk primary prevention (<1%/year): 800-1500 (cost $7-12 million/non-fatal CV event prevented)
  • Authors did not report the % of patients with FH; if similar to other PCSK9 inhibitor trials, it is likely a small minority of enrolled patients (~3-5%). FOURIER doesn't address long-term effectiveness of lowering LDL with PCSK9 inhibitors in FH & its cost-effectiveness
• Co-interventions
  • Only ~70% were on equivalent of atorvastatin 20 mg/d or more. It's not clear how aggressive clinicians were in trying to optimize statins before enrolling their patients into FOURIER
  • Only 5% were receiving ezetimibe at baseline. In many jurisdictions, a trial of ezetimibe to lower LDL is a prerequisite to initiating a PCSK9 inhibitor. Although the reduction in CV events is modest with ezetimibe, wider use likely would have resulted in a lower baseline CV risk & further reduced the absolute benefit (& increased the NNT) from PCSK9 inhibitors

Internal validity

• Low risk of
  • Allocation bias: Allocated centrally by computer
  • Performance bias: Use of matching placebo, allocation concealment maintained
  • Detection bias: Outcomes adjudicated by central committee unaware of treatment assignment & lipid levels
  • Attrition bias: <0.1% lost to follow-up; analyzed by intention-to-treat
  • Selective outcome reporting bias: All outcomes of interest reported

Additional considerations

• How does evolocumab compare to other lipid-lowering agents?
  • Statins: The effect of evolocumab on outcomes is consistent with the RRR/1-mmol/L LDL reduction with statin monotherapy, except for evolocumab's neutral effect on mortality. The 95% CI for death in FOURIER rules out the 10-15% RRR seen with statins in the Cholesterol Treatment Trialists' Collaboration meta-analysis
  • Beyond low/moderate-intensity statin:
    • High-intensity statin: Maximizing statin dose & lowering LDL by an extra 20-25% (e.g. going from atorvastatin 10 mg/d to 80 mg/d) further reduces the relative risk of a major vascular event by 15% (& are cost-effective for doing this). As with evolocumab, high-intensity statins have not demonstrated a lower risk of death versus moderate-intensity statins
    • Other than evolocumab, only ezetimibe has evidence for reducing CV outcomes beyond statin monotherapy. Fibrates, niacin & CETP inhibitors all failed as add-on to statin therapy
• What does FOURIER mean for lipid targets?
  • Although debate will continue regarding the clinical significance of benefits from increasing lipid-lowering intensity, FOURIER & IMPROVE-IT consistently demonstrate that "lower is better" for LDL, with no clear safe lower limit
  • FOURIER inclusion criteria required patients to have LDL >1.8 mmol/L, but patients achieved similar absolute & relative risk reductions with evolocumab regardless of baseline LDL (from 1.8-2.0 mmol/L to >2.8 mmol/L)
  • Despite most patients in the evolocumab group achieving an LDL <1.8 mmol/L (and most reaching <1.0 mmol/L, half the Canadian LDL target), their CV risk remained elevated at ~4-5%/year. This is a reminder that dyslipidemia is but 1 of many CV risk factors & that achieving LDL targets cannot be the only means of reducing an individual's CV risk

EBBINGHAUS substudy

• Subgroup study of FOURIER (n=1974, 7.2% of study population; 1204 analyzed in primary analysis) evaluating the effect of evolocumab vs placebo on cognitive function using the Cambridge Neuropsychological Test Automated Battery (CANTAB) done at baseline, week 24, annually, & then at the end of the trial
• Primary outcome (spatial working member strategy index of executive function, score ranges from 4-28, lower = better)
  • Baseline: 17.8
  • Last follow-up (median 19 months): Evolocumab 17.5, placebo 17.6
  • Change from baseline: -0.21 vs -0.29 (p<0.001 for non-inferiority)
• Secondary outcomes:
  • Working memory (score 0-270, lower = better)
    • Last follow-up: Evolocumab 20.3 vs placebo 20.1
    • Change from baseline: -0.52 vs -0.93 (p=0.36 for difference)
  • Episodic memory (score 0-70, lower = better)
    • Last follow-up: Evolocumab 24.9 vs placebo 23.6
    • Change from baseline: -1.53 in both groups (p=0.49 for difference)
  • Psychomotor speed (median time, lower = better)
    • Baseline: 356.7 vs 355.1
    • Last follow-up: 361.8 vs 355.7
    • Change from baseline: +5.2 vs +0.9 (p=0.06 for difference)
• Subgroup analyses suggested that evolocumab was worse than placebo in patients with baseline LDL <85 mg/dL, but better than placebo in those with baseline LDL >85 mg/dL (p=0.01 for interaction), though this was not supported by subgroup analysis of secondary outcomes

Testosterone CV safety

Bottom-line:

• CV safety of testosterone replacement therapy has not been proven. Existing studies have not been able to convincingly rule-out a clinically important increase in CV events, & some studies demonstrate potential harm.

• Theoretically, testosterone could also possibly reduce the risk of CV events in men with proven testosterone deficiency, though enthusiasm for this should be tempered by lessons learned from estrogen replacement therapy in post-menopausal women.

Context

• "Low-T" (generally defined as late-onset hypogonadism or androgen deficiency syndrome) is a controversial syndrome generally presenting with bothersome symptoms with confirmation of low serum testosterone levels, such as:
  • Fatigue, depression, cognitive impairment, sleep disturbances
  • Sexual dysfunction (low libido, erectile dysfunction)
  • Somatic symptoms (decreased muscle mass & increased body fat, low BMD, anemia, hot flushing, decreased facial/body hair growth)
• Cutoffs for "low" testosterone & indication for treatment differ by guidelines
  • European Association of Urology:
    • Total testosterone <8 nmol/L (230 ng/dL)
    • Free testosterone <225 pmol/L (65 pg/mL)
    • To be accurate, testing must be done between 7-11 am, or <3 h of waking in shift workers
• Although guidelines generally recommend testosterone replacement therapy (TRT) in men with low serum testosterone levels and symptoms, evidence is generally weak, conflicting, & suggests clinically non-important benefits for sexual function, cognition & muscle strength

Best-available evidence for CV safety

Systematic reviews

• Am J Med 2017
  • Search to Aug 2015 of PubMed, MEDLINE, Embase, CENTRAL, clinicaltrials.gov, hand-search
  • Included 34 RCTs, 10 observational studies (n=5451)
  • Risk of bias: Trials generally poorly reported, unclear risk of bias for most domains
  • P (Typical patient characteristics):
    • Men aged 50-60 y
    • Serum testosterone level <300 ng/dL to <480 ng/dL
    • 10/34 RCTs included pre-existing CVD, T2DM or metabolic syndrome
  • I/C: All compared 1 or more testosterone formulations to placebo, except 1 trial that compared TRT injection to patch
  • Outcomes in RCTs at follow-up ranging from 6 weeks to 3 years
    • All-cause death (20 RCTs): Peto odds ratio (OR) 0.88 (0.55-1.41), I^2 8%
    • MI (16 RCTs): Peto OR 0.87 (0.39-1.94), I^2 36%
    • Stroke (9 RCTs): Peto OR 2.17 (0.63-7.54), I^2 30%
  • Observational studies not meta-analyzed due to clinical & statistical heterogeneity. Results were variable:
    • 3 found TRT associated with increased CV risk
    • 5 found TRT associated with no difference or inconsistent effect on CV risk
    • 2 found TRT associated with decreased CV risk
• Older systematic review (BMC Medicine 2013) found an increased CV risk with TRT, particularly in studies NOT sponsored by the pharmaceutical industry
  • 27 RCTs (n=2994)
  • CV events: OR 1.54 (1.09-2.18)
  • Sensitivity analysis (p=0.03 for interaction)
    • CV events in trials funded by pharmaceutical industry: OR 0.89 (0.50-1.60)
    • CV events in trials NOT funded by pharmaceutical industry: OR 2.06 (1.34-3.17)

Recent studies not included in above systematic reviews

• JAMA 2017 RCT suggested accelerated atherosclerotic plaque buildup with TRT
  • Double-blind RCT of 170 men with mean age 71 years, serum testosterone <275 ng/dL & symptoms of low-T
  • Measured only surrogate outcome of plaque volume on coronary CTA: TRT increased total & non-calcified plaque volume vs placebo (+41 mm^3) over 1 year
  • No major adverse CV events occurred in either group over trial duration
• NEJM 2016 RCT primarily evaluating symptomatic improvement with TRT vs placebo was underpowered to evaluate CV safety
  • Major adverse CV events: <2% in both groups
• JAMA Internal Medicine 2017 cohort
  • Administrative cohort of 8808 men with mean age 58 years, diagnosis of androgen deficiency (3%) &/or serum testosterone <300 ng/dL (97%), <2% with prior CVD
  • At median 3.2 years, TRT (Rx filled at any time) versus no TRT (Rx never filled) was associated with a decreased risk of CV events: Hazard ratio (HR) 0.67 (0.62-0.73)
    • Results consistent with follow-up limited to very short-term (90 days), in those <65 or >65 years-old, with or without prior CVD

Main limitations in existing evidence

• Incredibly underpowered for evaluation of clinical CV events (small sample size, short follow-up, patients at low risk of CVD)
• Most trials at high/unclear risk of bias, leading to imprecise & inaccurate estimates of effect
  • In observational studies, a key source of confounding not adequately addressed is impact of socioeconomic status (wealthier patients more likely to experience good outcomes, also more likely to be able to afford relatively-expensive TRT)
• No use of standardized outcome definitions in most trial

ACCOMPLISH - Amlodipine vs HCTZ added to benazepril in HTN

Jamerson K, et al. Benazepril plus amlodipine or hydrochlorothiazide for hypertension in high-risk patients. NEJM 2008;359:2417-28.

Bottom line: 

• When added to an ACE inhibitor, amlodipine reduced the risk of CV events versus HCTZ (NNT 46 over 3 years). Amlodipine produced less dizziness (NNT 20), but far more peripheral edema (NNH 5) than HCTZ.

• Calcium channel blockers are not superior to other thiazides.

 

Patients (n=11,506)

• Included
  • HTN
  • High CV risk, defined as hx of any of the following:
    • Existing CVD: Stroke, ACS/MI, PAD, prior revascularization
    • LVH
    • CKD
    • Diabetes
• Typical study patient
  • Age 68 y
  • Female 40%
  • Prior CVD
    • Stroke 13%
    • MI 24%, unstable angina hospitalization 12%
    • Coronary revascularization 36%
  • Other CV risk factors
    • Smoker 11%
    • LVH 13%
    • eGFR <60 18%, other renal disease 6%
    • Diabetes 60%
    • Dyslipidemia 74%
  • BP 145/80 mm Hg (baseline, treated or untreated)
  • eGFR 79
  • Meds
    • Antiplatelet 65%
    • Statin 68%

Interventions

• All: Benazepril, plus addition of non-ACEI/CCB/thiazide BP-lowering med after maxing out study meds to achieve target BP)
  • Initial: 20 mg daily
  • After 1 month: Increased to 40 mg daily
• Amlodipine
  • Initial: 5 mg daily
  • Increased to max 10 mg daily to achieve BP <140/90 (or <130/80 if diabetes/CKD)
    • 61% receiving this dose @ month 6
• Hydrochlorothiazide (HCTZ)
  • Initial: 12.5 mg daily
  • Increased to max 25 mg daily to achieve BP <140/90 (or <130/80 if diabetes/CKD)
    • 60% received this dose @ month 6

Results @ ~3 years

• Achieved SBP ~132 with amlodipine vs ~133 mm Hg with HCTZ (SBP 0.9 mm Hg lower with amlodipine)
  • Office BP <140/90: 75% vs 72%
• Death: 4.1% vs 4.5%, hazard ratio (HR) 0.90 (0.76-1.07)
• Primary outcome (composite of: CV death, non-fatal MI, stroke, unstable angina hospitalization, coronary revascularization, or resuscitation after sudden cardiac arrest): 9.6% vs 11.8% (NNT 46), HR 0.80 (0.72-0.90)
  • CV death: 1.9% vs 2.3%, HR 0.80 (0.62-1.03)
  • MI: 2.2% vs 2.8%, HR 0.78 (0.62-0.99)
  • Stroke: 1.9% vs 2.3%, HR 0.84 (0.65-1.08)
  • Coronary revascularization: 5.8% vs 6.7%, HR 0.86 (0.74-1.00)
• Study drug discontinuation: 28.8% vs 31.2%
• Specific adverse effects
  • Dizziness: 20.7% vs 25.4% (NNT 22)
  • Hypotension: 2.5% vs 3.6%
  • Peripheral edema: 31.2% vs 13.4% (NNH 5)
  • Hypokalemia: 0.1% vs 0.3%

Generalizability

• Patients in this trial were either "secondary CV prevention" or high-risk "primary CV prevention" patients, so absolute benefit of amlodipine over HCTZ likely lower in a population with lower baseline CV risk
• Interventions
  • HCTZ is arguably the least effective available thiazide diuretic; therefore, this trial cannot be extrapolated to conclude that amlodipine is superior to other thiazides, particularly chlorthalidone or indapamide when added to an ACE inhibitor
  • Non-comparable doses: Amlodipine 5-10 mg represents the "mid to high" dose range for this agent, whereas HCTZ 12.5-25 mg represents the "low to mid" dose range
    • This likely explains the small but significant difference between groups in mean BP & % of patients achieving target BP

Internal validity

• Low risk of allocation, performance & detection bias
  • Central allocation; patients, personnel & outcome adjudicators unaware of allocated treatment
  • However: Given high rate of peripheral edema with amlodipine, likely that a sizable portion of patients and clinicians were unblinded to treatment allocation. The effect of this on risk of performance bias is unclear, but likely minimal
• Low risk of attrition bias
  • Analyzed by intention-to-treat; low loss-to-follow-up (1%) with similar time on treatment in both groups
• Other: Study stopped early due to benefit based on pre-defined stopping rule
  • This is unlikely to have a meaningful impact on the results as primary outcome results based on >1200 events

Other studies

• ALLHAT is the major trial that seems to conflict with ACCOMPLISH:
  • Randomized 33,357 adults (55+ y) with BP 140-180/90-110 + high CV risk (existing atherosclerotic CVD, LVH, T2DM, HDL <0.90 mmol/L, or smoker) to chlorthalidone vs amlodipine vs lisinopril for a mean 4.9 years
  • Achieved SBP: ~134 vs ~135 vs ~136 mm Hg
  • Chlorthalidone vs amlodipine: No difference between any group in primary outcome (fatal coronary event or MI) or individual CV outcomes, except lower lower risk of HF incidence (NNT 40) & HF hospitalization (NNT 53) with chlorthalidone
• A meta-analysis of 7 RCTs comparing diuretics vs CCB (n=51,450), driven largely by ACCOMPLISH & ALLHAT, found no difference in rates of major CV events
• In my opinion, the above studies suggest that HCTZ is inferior to chlorthalidone. No RCT has ever compared effect of HCTZ vs chlorthalidone on CV events, so only low-quality evidence exists evaluating this theory:
  • Indirect comparison by network meta-analysis suggests that HCTZ may indeed be inferior to chlorthalidone for reduction in CV events;
  • Small RCT showed that chlorthalidone produces greater reduction in 24h & nighttime BP (which is more strongly associated with CV events than office BP) vs HCTZ despite similar office BP lowering;
  • A large administrative cohort study found no difference in CV events between chlorthalidone & HCTZ.