SHEP - Targeting systolic BP <160 mm Hg in patients with isolated systolic hypertension

SHEP Cooperative Research Group. Prevention of stroke by antihypertensive drug treatment in older persons with isolated systolic hypertension. JAMA 1991;265:3255-64.

Bottom line: In patients with isolated systolic HTN with SBP ~170, reducing SBP <160 mm Hg with a chlorthalidone-based regimen reduced the risk of CVD (NNT 19), including individual components such as stroke (NNT 46) over 4.5 years. This benefit was countered by an increase in adverse events, including bothersome adverse events (NNH 14), syncope (NNH 112) and electrolyte abnormalities.

 

Patients (n=4736)

• Included: SBP 160-219 mm Hg & DBP <90 mm Hg
• Excluded: Existing "major CV disease"
• Typical study patient
  • Age 72 y
  • Female 57%
  • Previous CVD: Stroke 1.4%, MI 5%
  • BP 170/77 mm Hg

Interventions

• I: Chlorthalidone 12.5-25 mg/d +/- atenolol 25-50 mg/d to achieve BP goal
  • If baseline SBP 160-179 mm Hg: Goal to reduce BP by >20 mm Hg
  • If baseline SBP 180+ mm Hg: Goal SBP <160 mm Hg
• C: Placebo & attempt to reach same BP goals as above

Results @ mean 4.5 years

• Achieved SBP at year 5: 144 vs 155 mm Hg (mean difference 11 mm Hg)
• Receiving BP meds by year 5: 90% vs 44%
• Death: 9.0% vs 10.2%, relative risk (RR) 0.87 (0.73-1.05)
• Efficacy
  • CV disease (CV death, MI, stroke/TIA, coronary revascularization, aneurysm, endarterectomy): 12.2% vs 17.5% (NNT 19), RR 0.68 (0.58-0.79)
    • Total stroke (primary outcome): 4.1% vs 6.3% (NNT 46), RR 0.63 (0.49-0.82)
    • Non-fatal MI or coronary death: 4.4% vs 5.9% (NNT 72), RR 0.73 (0.57-0.94)
  • "LV failure": 2.0% vs 4.3% (NNT 44), RR 0.46 (0.33-0.65)
• Safety
  • Any adverse event: 91.8% vs 86.4% (NNH 19)
  • Any intolerable adverse event: 28.1% vs 20.8% (NNH 14)
  • Key adverse events (not a comprehensive list)
    • Falls: 12.8% vs 10.4% (NNH 42)
    • Loss of consciousness/passing out: 2.2% vs 1.3% (NNH 112)
    • Na <130: 4.1% vs 1.3% (NNH 36)
    • K <3.2: 3.9% vs 0.8% (NNH 33)

Internal validity

• Low risk of allocation, performance, detection & attrition bias
  • Central allocation
  • Double-blind
  • Protocolized, stepped approach to treatment

SPRINT - Targeting a systolic BP <120 vs <140

The SPRINT Research Group. A randomized trial of intensive versus standard blood-pressure control. NEJM 2015;373:2103-16.

Bottom line:

• In patients with initial SBP 130-180 mm Hg, targeting SBP <120 vs 135-139 mm Hg reduces the risk of death (NNT 84), CV events (NNT 63; primarily from fewer CV deaths & new cases of HF) over 3.3 years.

• Targeting this lower target results in prescribing of 1 additional BP med & an increase in a number of adverse events related to BP meds, including syncope (NNH 91), AKI (NNH 56) & electrolyte abnormalities (NNH 100) versus the "standard" target.

 

Patients (n=9361)

• Included
  • Age >50 y
  • SBP 130-180 mm Hg (upper limit lower with higher # of BP meds pre-enrolment)
  • High CV risk, defined as 1 or more of:
    • Age 75+ y
    • Clinical CVD other than stroke (prior ACS, PCI, CABG, carotid endarterectomy, carotid stenting, PAD with revascularization, stenosis >50% of coronary, carotid or lower-extremity artery, AAA 5+ cm with or without repair)
    • Subclinical CVD (CAC >400 units, ABI <0.9, LVH) in past 2 years
    • Framingham risk score 15% or more over 10 y
    • CKD (stable eGFR 20-59 in past 6 months)
• Key exclusion
  • Prior stroke
  • Diabetes
  • Symptomatic HF with EF <35%
  • CV event or procedure, or hospitalization for unstable angina within 3 months
  • SBP <110 mm Hg after 1 min of standing 
  • Indication for specific BP med that pt is not taking
  • Renal
    • ESRD or eGFR <20
    • Proteinuria defined as: 24h urinary protein excretion >1 g, 24h urinary albumin excretion 600+ mg, spot urine PCR 1+ g/g, spot urine ACR 600+ mg/g, urine dipstick 2+ protein
    • PCKD
    • Glomerulonephritis
  • Any organ transplant
• Typical study patient
  • Age 67.9 y (75+ 28%)
  • Female 36%
  • CV risk criterion met
    • Clinical CVD 17%
    • Subclinical CVD 5%
    • FRS 15%+ 61%
    • Race: Black 30%, Hispanic 10%, white 58%, other 2%
    • BP 140/78 mm Hg
    • eGFR 72
    • Mean FRS 20%
    • Meds
      • Mean # of BP meds 1.8
      • ASA 51%
      • Statin 44%

Interventions

• BP med adjustments based on automated office BP (AOBP) using Omron device based on average of 3 measurements while sitting after 5 min of quiet rest
• I: Intensive therapy (target SBP <120 mm Hg, with secondary target of DBP <90 mm Hg)
  • At randomization, started on combo of 2-3 BP meds including a thiazide, ACEI or ARB, & CCB (or other agents for compelling indication, e.g. beta-blocker post-MI)
  • Mean AOBP SBP during follow-up: 121.5 mm Hg
  • Mean # of BP meds during follow-up: 2.8
• C: Standard therapy (target SBP 135-139 mm Hg, with secondary target of DBP <90 mm Hg)
  • At randomization, converted to SPRINT formulary
  • If SBP <130 x1 visit or <135 x2 visits, BP meds changed to increase SBP back into 135-140 range
  • Mean AOBP SBP during follow-up: 134.6 mm Hg
  • Mean # of BP meds during follow-up: 1.8
• Agents used in each class as part of SPRINT formulary: Thiazide (chlorthalidone 12.5-25 mg), ACEI (lisinopril 5-40 mg), ARB (azilsartan 40-80 mg, losartan 25-100 mg), beta-blocker (atenolol 25-100 mg, metoprolol tartrate 50-100 mg), CCB (amlodipine 2.5-10 mg, diltiazem 120-540 mg), other classes
• The control group required a very tight SBP range (135-139 mm Hg), which may have resulted in frequent medication adjustments & resulting visits to maintain this target (these data have not yet been reported)

Results @ median follow-up 3.3 years

• BP meds used at last visit (full list included in study's supplemental appendix)
  • Thiazide 55% vs 33%
  • ACEI or ARB 77% vs 55%
  • Beta-blocker 41% vs 31%
  • CCB 57% vs 35%
• Mean AOBP SBP difference at year 1: ~15 mm Hg
• Death: 3.3% vs 4.5% (NNT 84), hazard ratio (HR) 0.73 (95% confidence interval 0.60-0.90)
• Serious adverse event: 38.3% vs 37.1%, p=0.25
• Primary outcome (composite of CV death, ACS, stroke, acute decompensated HF): 5.2% vs 6.8% (NNT 63), HR 0.75 (0.64-0.89)
  • CV death: 0.8% vs 1.4%, HR 0.57 (0.38-0.85)
  • HF: 1.3% vs 2.1%, HR 0.62 (0.45-0.84)
  • MI: 2.1% vs 2.5%, HR 0.83 (0.64-1.09)
  • Stroke: 1.3% vs 1.5%, HR 0.89 (0.63-1.25)
• Renal event
  • Prior CKD (eGFR decreased >50%, development of ESRD requiring long-term dialysis or kidney transplant): 1.1% in both groups
  • No prior CKD (eGFR decreased >30% to <60 mL/min/1.73 m2): 3.8% vs 1.1% (NNH 37), HR 3.49 (2.44-5.10)
  • Subgroup analyses demonstrated no significant subgroup differences
• Events leading to ED visit, hospitalization, death, or resulting in persistent disability:
  • Injurious fall: 7.1% in both groups
  • Syncope: 3.5% vs 2.4% (NNH 91)
  • Hypotension: 3.4% vs 2.0% (NNH 72)
  • Bradycardia: 2.2% vs 1.8%
  • Electrolyte abnormality 3.8% vs 2.8% (NNH 100)
  • AKI: 4.4% vs 2.6% (NNH 56)
• Orthostatic hypotension with dizziness: 1.3% vs 1.5% (asymptomatic: 16.6% vs 18.3%)

Generalizability

• Population caveats
  • Specifically excluded patients with diabetes & stroke due to ongoing BP-target trials at time of SPRINT design (ACCORD-BP for diabetes & SPS3 for lacunar stroke)
    • Therefore, results of SPRINT not directly applicable to patients with diabetes or stroke, but must consider SPRINT in interpretation of both ACCORD-BP & SPS3
  • Excluded patients with symptomatic HFrEF; these patients should be treated to achieve maximally-tolerated doses of HFrEF-modifying drugs as previously reviewed
  • >75% of SPRINT patients had no CVD at baseline (primary prevention), so the absolute risk reduction reported in SPRINT mainly applies to this "lower risk" population, & absolute benefits are likely greater for patients at higher baseline CV risk
• Intervention caveats
  • The SPRINT treatment protocols encouraged use of interventions with greatest evidence base, i.e. thiazides recommended as 1st-line, & chlorthalidone (not hydrochlorothiazide) included on formulary
    • Absolute benefit of intensive BP control may therefore vary based on agents used in practice. Clinicians should strive to use these proven agents

Internal validity

• Low risk of allocation, attrition, & selective outcome reporting biases
  • Random sequence generation & allocation concealment
  • Loss-to-follow-up minimal (~3%), analyzed by intention-to-treat
  • Reported (or in process of gathering & publishing) all clinically-important outcomes of interest
• Unclear risk of performance & detection bias
  • Outcome adjudicator blinded, but patients and clinicians aware of allocated treatment group. Overall, likely low-risk due to protocolized intervention for each group & clearly-defined outcome definitions with objective criteria
• Early truncation unlikely to introduce significant bias due to large number of primary outcome events (>500) & valid, pre-defined stopping rules

Additional publications of SPRINT

• Clinicians are reasonably cautious with BP lowering in very elderly patients, particularly in those who are frail. The SPRINT investigators explored the efficacy and safety of intensive SBP lowering in the 28% of patients in SPRINT 75+ years old at baseline
  • In these 2636 patients with a mean age of 80 y, results were largely consistent with those seen in the overall SPRINT population, with greater absolute benefits due to age-related increases in baseline risk:
    • Efficacy: Reduced primary outcome (NNT 27), death (NNT 41), and HF
    • Safety: No increase in serious adverse events (HR 0.99, 0.89-1.11), injurious falls; non-statistically significant increases in syncope (NNH 167), symptomatic orthostatic hypotension (NNH 167), electrolyte abnormalities and AKI
  • Among those aged 75+ years, efficacy outcomes did not differ based on frailty status or gait speed 
  • A further analysis of patients 75+ years old in SPRINT found that intensive versus standard BP lowering had no impact on gait speed or progression to mobility limitation
    • In other words, intensive BP lowering in SPRINT did not inadvertently result in an increase in frailty or disability due to hypotension-related falls, fractures, gait instability or lightheadedness

Other studies

• ACCORD BP (low target SBP in diabetes):
  • Open-label RCT of 4733 type 2 diabetics with baseline SBP 130-180 mm Hg, A1c 7.5% or more & high CV risk (40+ y with CVD or 55+ y with atherosclerosis, albuminuria, LVH, or 2+ other CV risk factors) randomized to target SBP <120 or <140 mm Hg
  • Mean achieved SBP 119 vs 133 mm Hg (mean difference 14 mm Hg) using a mean of 3.4 & 2.1 BP meds, respectively
  • At mean 5 y, no statistically significant difference in death (HR 1.07, 0.85-1.35) or primary outcome (CV death, MI, stroke; HR 0.88, 0.73-1.06), but statistically significant reduction in stroke (NNT 91; HR 0.59, 0.39-0.89)
  • Subgroup analysis by assignment to intensive vs standard glycemic control (stratified at baseline) suggested that target SBP <120 mm Hg beneficial on primary outcome in patients assigned to standard glycemic control (p=0.08 for interaction)
  • Bottom line of ACCORD BP: Does not rule out clinically important benefit of lower SBP target (especially considering interaction with harmful intensive glycemic control of ACCORD)
• SPS3 (low target SBP post-stroke):
  • Open-label RCT of 3020 patients with symptomatic lacunar stroke within past 6 months randomized to target SBP <130 or 130-149 mm Hg
  • Mean achieved SBP 127 vs 138 mm Hg (mean difference 11 mm Hg) using mean 2.4 vs 1.8 BP meds, respectively
  • At mean 3.7 y, no statistically significant difference in stroke (HR 0.81, 0.64-1.03), MI/vascular death (HR 0.84, 0.68-1.04) or serious adverse events (HR 1.53, 0.80-2.93)
  • Bottom line of SPS3: Does not rule out clinically important benefit of lower SBP target
• Conflicting results between low vs standard BP target in other studies of overall HTN population
  • HOT (low target diastolic BP in overall HTN population):
    • Open-label RCT of 18,790 patients with baseline diastolic BP 100-115 mm Hg (baseline BP 170/105 mm Hg) randomized to target DBP <80, <85, or <90
    • Mean achieved SBP 140 vs 141 vs 144 mm Hg (mean SBP difference 4 mm Hg between DBP <80 and <90 groups)
      • Standard BP med step-wise process: Step 1 (felodipine), step 2 (ACEI or beta-blocker), step 3 (increase felodipine), step 4 (increase ACEI/BB), step 5 (add diuretic)
    • At mean 3.8 years, no difference between groups in primary outcome or individual CV outcomes in overall population
    • Investigators performed subgroup analysis in patients with diabetes (without appropriately performing a test for interaction to assess for a subgroup effect of diabetes vs no diabetes)
      • Identified a lower risk of the primary outcome & CV mortality, but not MI or stroke, with a target DBP <80 vs <90 in diabetes
    • Bottom line of HOT: No reduction in CV events with a lower target DBP using the aforementioned 5-step regimen; HOT diabetic subgroup effect is questionable at best
  • Cardio-Sis (target SBP <130 in HTN with 1 other CV risk factor)
    • Open-label RCT of 1111 patients with baseline SBP >150 randomized to target SBP <130 vs 140 mm Hg 
    • Mean achieved SBP 132 vs 136 mm Hg (mean SBP difference 4 mm Hg)
    • At 2 years, reduction in patients with new-onset LVH (primary outcome), as well as a broad CV composite outcome (death, MI, angina, coronary revascularization, stroke/TIA, AF, HF admission; NNT 22, RRR 50%)

Fibrates for CV prevention

References: ACCORD Lipid, FIELD

Bottom-line:

• In patients taking a statin, the addition of a fibrate does not significantly reduce the risk of CV events.

• In patients at intermediate-to-high CV risk who are absolutely not able to take a statin, fibrate therapy reduces the relative risk of non-fatal, but not fatal, CV events by ~20%.

  • In my opinion, ezetimibe & bile-acid sequestrants (& soon possibly PCSK9 inhibitors) should be considered before fibrates in these patients.

Patients

View fullsize

Intervention

• FIELD: Fenofibrate vs matching placebo
  • Fenofibrate given as 200 mg (micronized formulation) PO once daily
  • Non-study lipid-lowering drug: Fenofibrate group 8%, placebo group 17%
  • ~20% in each group discontinued study drug by end of study
• ACCORD Lipid: Fenofibrate + simvastatin vs simvastatin + matching placebo
  • Fenofibrate given as 160 mg PO once daily
  • Average dose of simvastatin 20 mg/d in both groups (open-label, adjusted to lipid targets)
  • Fenofibrate discontinued in 22%, placebo discontinued in 19% by end of study; ~20% in each group discontinued simvastatin by end of study

Results @ ~5 years

View fullsize

Internal validity

• Both trials at low risk of bias (including allocation, performance, detection, & attrition bias)
  • Central randomization
  • Patients, clinicians, investigators, adjudicators all blinded to treatment allocation
  • Loss-to-follow-up <1%
  • Analyzed using intention-to-treat principles

Generalizability

• FIELD represents the effects of fibrate monotherapy in a population with type 2 diabetes at mostly intermediate risk of CV events (estimated ~10-12% over 10 years)
  • Mechanistically, primarily testing the mechanistic effect of lowering triglycerides by ~30% over placebo, as effect on both LDL & HDL modest
• ACCORD Lipid represents the effects of adding a fibrate to a statin in a higher-risk population of patients with type 2 diabetes (estimated ~25% over 10 years without statin)
  • Again, primarily testing the mechanistic effect of lowering triglycerides, as no discernible effect on LDL or HDL

Other fibrate studies

• VA-HIT: In 2531 men with CAD not receiving a statin, gemfibrozil lowered trigs by ~30% more than placebo. At a median 5.1 years, this resulted in a 4.4% absolute risk reduction in MI or coronary death (NNT 23).
• BIP: In 3090 men with CAD not receiving a statin, bezafibrate increased HDL & lowered trigs by ~15% more than placebo. At a mean 6.2 years, this did not result in a statistically significant effect on MI or sudden death.
• Subgroup analyses from these various trials have provided more confusion than clarity. For example, some studies demonstrate an interaction by baseline triglycerides (VA-HIT, BIP), but not others (FIELD, ACCORD Lipid). The ACCORD Lipid study, but not the BIP trial, found a subgroup interaction based on the combination of low HDL and high trigs.
• A systematic review of 18 fibrate trials, including the ones already mentioned here, found results largely consistent with the FIELD trial:
  • No reduction in death or fatal CV events;
  • Reduction in CV events, entirely driven by non-fatal coronary events (i.e. non-fatal MI & coronary revascularization): Relative risk 0.81 (0.75-0.89);
  • 5% relative risk reduction in coronary events per 0.1 mmol/L reduction in triglycerides;
  • Notably, ACCORD Lipid was the only trial included in this systematic review which had routine statin administration.

RAAS inhibitor dose in HFrEF

Reference: Turgeon RD, et al. Higher versus lower doses of ACE inhibitors, angiotensin-2 receptor blockers and beta-blockers in heart failure with reduced ejection fraction: Systematic review and meta-analysis. PLoS ONE 14(2):e0212907

Bottom line:

The available evidence evaluating dose-response of medications in HFrEF have several caveats that require close scrutiny. With that said, pushing the dose of ACE inhibitors and ARBs appears to:

• Further reduces the risk of HF worsening or HF hospitalizations over 4-5 years;

• Increases the risk of side-effects such as lightheadedness, renal dysfunction & hyperkalemia. These are generally predictable, reversible, and manageable with holding or decreasing the ACEI/ARB dose.

The decision to aim for target ACEI/ARB doses must account for the consequences & severity of all of these events in an individual, rather than the relative size of the NNT versus NNHs. For example, HF hospitalizations impair patient quality of life, are costly to the healthcare system, & put patients at risk for numerous complications, such as hospital-acquired infections, VTE, & deconditioning. The severity of ACEI/ARB-related renal dysfunction can range from dialysis-dependent renal failure to a slight creatinine bump that's reversible upon decreasing the ACEI/ARB dose, & most often falls in the latter category.

 

Inclusion criteria from key trials

View fullsize

Baseline patient characteristics from key trials

View fullsize

Interventions in key trials

• ATLAS

  • I: High-dose lisinopril (32.5-35 mg PO once daily)

    • 91% achieved target dose

    • 27% discontinued study drug

  • C: Low-dose lisinopril (2.5-5 mg PO once daily)

    • 31% discontinued study drug

• High Enalapril Dose Study

  • I: High-dose enalapril (30 mg BID - 3x target dose in SOLVD)

    • ~45% achieved target dose by year 1

    • Mean achieved dose ~20 mg BID

  • C: Standard target-dose enalapril (10 mg BID)

    • ~80% achieved target dose by year 1

    • Mean achieved dose ~10 mg BID

  • Both groups started at 2.5 mg BID, uptitrated q1 week to target dose

• NETWORK

  • I-1: Standard target-dose enalapril (10 mg BID)

  • I-2: Half target-dose enalapril (5 mg BID)

  • C: Low-dose enalapril (2.5 mg BID)

  • All started at 2.5 mg BID; standard/half-dose groups doubled after 1 week; standard-dose further doubled 1 week later

• HEAAL

  • I: Losartan 150 mg once daily

    • Initial 50 mg daily, then uptitrated over 3 weeks

    • 94% achieved target dose

  • C: Losartan 50 mg once daily

Results (meta-analysis forest plots available in the open-access article)

Risk of bias

Other issues

• Inadequate follow-up duration

  • The High Enalapril Dose Study and NETWORK were both at substantial risk of a type 2 error, i.e. falsely concluding that higher doses did not result in additional benefits, as they had short follow-up periods of 6-12 months that may not adequately capture the potential benefits of targeting a higher ACEI dose

• Contamination bias: High risk

  • The ATLAS trial reported that ~20% of patients in each group started open-label ACEI therapy, which increased overall "ACEI dose" in the low-dose ACEI group and minimized potential differences in outcomes between the 2 treatment groups

• Run-in phases

  • ATLAS & HEAAL both had active run-in phases to exclude patients prior to randomization who would not tolerate a low-moderate dose of the study drug

    • Safety data from these studies will be lower than seen in an unselected real-world population

Generalizability

• Of the 4 key trials, ATLAS & HEAAL, which both included only HFrEF patients & used “high” doses that were attainable by >90% of patients in the trial, are the most generalizable to a real-world HFrEF population

• The High Enalapril Dose Study enrolled a population that is approximately 10 years younger than the usual mean age for HF trials. This is reflected in the higher proportion of patients who could achieve the 10 mg BID dose (80% versus 50% in SOLVD)

• Background HF medical therapy was seldom reported & none reported device therapy

INVEST - Verapamil- vs atenolol-based HTN treatment in CAD

Pepine CJ, et al. A calcium antagonist vs a non-calcium antagonist hypertension treatment strategy for patients with coronary artery disease. The International Verapamil-Trandolapril Study (INVEST): A randomized controlled trial. JAMA 2003;290:2805-16.

Bottom line:

• Although INVEST technically demonstrated "equivalence" of a HTN regimen based primarily on verapamil SR + trandolapril versus atenolol + hydrochlorothiazide in patients with CAD & HTN, this does not generalize to patients with (1) HF or LV dysfunction or (2) recent MI without contraindication for a beta-blocker.

• Additionally, this likely does not generalize to patients on better evidence-based thiazides (chlorthalidone or indapamide), & possibly not to other beta-blockers.

• Overall, there was no clinically relevant benefit of the verapamil-based regimen in CAD that would justify using this agent over a beta-blocker in this population.

 

Patients (n=22,576)

• Included
  • Age 50+ y
  • CAD, clinically stable, defined as any of:
    • Remote MI >3 months ago;
    • Stenosis of >50% of at least 1 coronary artery on angiography;
    • "Classic" angina
    • Ischemia on 2+ non-invasive investigations (EKG, echo, and/or nuclear imaging)
  • Hypertension (HTN)
• Exclusion
  • HF NYHA functional class IV (I-III could be enrolled)
  • Taking beta-blockers for an MI that occurred <1 year
• Typical study patient
  • Age 66 y (~33% >70 y)
  • Female 52%
  • Inclusion criteria
    • Angina 67%
    • Coronary stenosis on angiogram 39%
    • MI 32%
    • Abnormal stress test 21%
  • PMHx
    • CABG or PCI 27%
    • Stroke 5%
    • HF (NYHA I-III) 5-6%
    • Current smoker 12%
    • Dyslipidemia 56%
    • Diabetes 28%
  • BP 150/86 mm Hg
  • HR 76 bpm
  • Meds
    • Antiplatelet 57%
    • Lipid-lowering 37%
    • Hormone replacement therapy 18% of women

Interventions

• Both groups received treatment for HTN to target office-based BP <140/90 mm Hg (<130/85 mm Hg for patients with diabetes or renal dysfunction)
• I: Verapamil SR-based HTN regimen
  • Step 1: Verapamil SR 240 mg PO once daily (if HF, diabetes, or renal impairment: also add trandolapril 2 mg/d)
  • Step 2: Add trandolapril 2 mg PO daily
  • Step 3: Increase verapamil to 180 mg PO BID & trandolapril to 2 mg PO BID
  • Step 4: Add hydrochlorothiazide 25 mg PO daily
  • Steps 5+: Max doses of verapamil 480 mg/d, trandolapril 8 mg/d, hydrochlorothiazide 100 mg/d, non-beta-blocker non-study antihypertensive
  • At 2 y, 82% on verapamil SR
• C: Atenolol-based HTN regimen
  • tep 1: Atenolol 50 mg PO once daily (if HF, diabetes, or renal impairment: also add trandolapril 2 mg/d)
  • Step 2: Add hydrochlorothiazide 25 mg PO daily
  • Step 3: Increase both atenolol & hydrochlorothiazide to BID
  • Step 4: Add trandolapril 2 mg/d
  • Steps 5+: Max doses of atenolol 200 mg/d, hydrochlorothiazide 100 mg/d, trandolapril 8 mg/d, non-CCB non-study antihypertensive
  • At 2 y, 77% taking atenolol

Results @ mean 2.7 years

• Vitals @ year 2
  • BP reduction of ~19/10 mm Hg in both groups
  • Achieved BP <140/90 mm Hg: 71.7% vs 70.7%
  • Mean resting HR 73 bpm vs 69 bpm
• Primary outcome (death, non-fatal MI, non-fatal stroke): 9.9% vs 10.2%, relative risk (RR) 0.98 (0.90-1.06)
  • Death: 7.75% vs 7.9%
  • Non-fatal MI: 1.3% in both groups
  • Non-fatal stroke: 1.2% vs 1.3%
• Angina:
  • Baseline: 1.5 episodes/week in both groups
  • At 2 years: 1.3 vs 1.6 episodes/week (p=0.02)
• Adverse events:
  • Lightheadedness: 0.4% vs 0.6%
  • Symptomatic bradycardia: 0.7% vs 1.3%
  • Dyspnea: 1% vs 0.7%
  • Wheezing: 0.2% vs 0.4%
• Subgroup analyses:
  • Different effect based on whether patients had HF (possibly ~5% absolute risk reduction with beta-blockers in primary outcome) versus those who did not (no difference)
  • Otherwise, no other baseline characteristics had a significant subgroup effect

Generalizability

• Population widely representative CAD population, with key caveats:
  • First, those with an MI in the last year who were already on a beta-blocker were excluded. Thus, this trial likely selected out patients who may benefit most from beta-blockers.
  • Second, this trial included patients with clinical HF. Notably, although only 5-6% of the population, there appeared to be much worse outcomes in those receiving verapamil subgroup, which is consistent with previous concerns of HF-related mortality due to the negative inotropic effect of non-dihydropyridine CCBs like verapamil. It should therefore be avoided in this population.
  • Third, there was generally poor use of CAD secondary prevention therapies such as ASA & lipid-lowering, & uncharacteristically high use of hormone replacement therapy (55% of enrolled patients were female & 18% of them were receiving HRT despite a confirmed CAD history). This would be expected to increase the underlying risk of primary outcome events in both treatment groups.
• Multiple considerations regarding the intervention arms are required to interpret the results:
  • First, the interventions represent BP-lowering "regimens" focused on using verapamil or atenolol, rather than a true head-to-head comparison of these agents;
  • Second, although it lowers BP, atenolol seems to reduce the risk of CV events less than antihypertensive agents from other classes in patients with HTN. Similarly, hydrochlorothiazide is a weaker antihypertensive agent & generally has less evidence supporting its use than other thiazides, which have a clearer signal for benefit. This combination may have therefore been disadvantaged versus using a beta-blocker + thiazide combination with more robust supporting evidence (e.g. bisoprolol plus chlorthalidone).
  • Third, adherence to the primary drug of the regimen (verapamil & atenolol) as well as to BP goals was fairly poor, with only 75-80% in each group still taking this drug & only ~70% with BP <140/90 mm Hg at year 2.

Internal validity

• Low risk of allocation bias due to appropriate sequence generation & allocation concealment (centralized, automated Internet-based randomization with permuted blocks).
• As this trial was a "prospective, randomized, open-label, blinded-endpoint" (PROBE) trial, it is by default at high risk of performance & detection bias due to knowledge of clinicians & patients of the allocated treatment
  • The risk of performance bias was minimized with a standardized treatment algorithm for HTN & a similarly low proportion of patients (~70%) achieved their BP target. Additionally, few patients had revascularization procedures & there were similar rates of crossover between groups
  • Investigators attempted to minimize the risk of detection bias by adjudication of events by a committee unaware of treatment allocation, however, they could not eliminate this bias as well as if they would have blinded patients and clinicians to allocation to verapamil or atenolol. Given that this was a feasible option with a few minor changes to the study HTN treatment algorithm, it is not possible to give the investigators "a pass" for this.
• Low risk of attrition bias as analyses followed intention-to-treat principles, & loss-to-follow-up was low at ~2-3% in both groups
• This trial was technically designed as an equivalence trial, with an equivalence boundary for the relative risk of 0.83-1.20.