Heart Failure HF Findings: Are They Real

1
Heart Failure (HF) FindingsAre They Real?
Stanley S. Franklin, MD, FACP, FACC Clinical
Professor of Medicine University of California at
Irvine Associate Medical Director UCI Heart
Disease Prevention Program Irvine, California
2
Presenter Disclosure Information
Stanley S. Franklin, MD, FACP, FACC, FAHA, FASN
DISCLOSURE INFORMATION The following
relationships exist related to this
presentation Speakers bureau for Boehringer
Ingelheim, Bristol-Myers Squibb, Merck.
Consultant for AtCor Medical, Bristol-Myers
Squibb, and Merck
3
HF Objectives

• Characterize HF in ALLHAT by its antecedent risk
  factors and underlying conditions.
• Examine occurrence of HF by treatment groups
  overall, in subgroups, and over time.
• Examine post-HF mortality overall and by
  treatment group.

Davis BH, et al. Circulation 20061132201-10
4
Decision to StopDoxazosin Arm

• Futility of finding a significant difference for
  primary outcome compared to chlorthalidone
• Statistically significant 25 percent higher rate
  of major cardiovascular events, including near
  twofold higher rate of HF (hospitalized, treated
  out-of-hospital, or fatal)

5
Blood Pressure TrialDesign

• Randomized, practice based
• Double-blind (not PROBE)
• Diagnoses assigned by clinic investigators guided
  by protocol-defined diagnostic criteria
• Randomization stratified by clinic
• Exclude h/o symptomatic HF (stage C) and/or
  known LVEF lt35

6
Baseline Characteristics
Davis BH, et al. Circulation 20061132201-10
7
Hospitalized/ Fatal HF by ALLHAT Treatment Group



.1
.08
Chlorthalidone Amlodipine Lisinopril
.06
Cumulative Event Rate
.04
.02
0
0
1
2
3
4
5
6
7
Years
Davis BH, et al. Circulation 20061132201-10
8
HF Before and After 1 Year

• A test of the proportional hazards assumption for
  Cox regression revealed that RRs were not
  constant over time. Therefore, a Cox regression
  that used a time-dependent indicator variable
  (lt1 year versus gt1 year) was utilized.

Davis BH, et al. Circulation 20061132201-10
9
Hospitalized/ Fatal HF by ALLHAT Treatment Group
Within 1 Year and gt1 Year
.1
.02
.08
.06
Cumulative Hosp/Fatal HF Rate
.01
.04
.02
0
0
0
1
2
3
4
5
6
7
0
.5
1

Years to Hosp/Fatal HF
Years to Hosp/Fatal HF
Davis BH, et al. Circulation 20061132201-10

10
Hospitalized/fatal HF in Subgroups - Amlodipine /
Chlorthalidone Relative Risks from Baseline to 1
Year of Follow-up
Favors Amlodipine
Favors Chlorthalidone
Relative Risk (95 CI)
2.22 (1.69 - 2.91)
Total
2.89 (1.62 - 5.17)
Age lt 65
2.06 (1.51 - 2.80)
Age 65
2.12 (1.49 - 3.01)
Non-Black
2.37 (1.55 - 3.63)
Black
2.27 (1.56 - 3.30)
Men
2.17 (1.46 - 3.21)
Women
2.71 (1.83 - 4.02)
Diabetic
1.83 (1.25 - 2.67)
Non-Diabetic
0.50
1
2
3
4
5
6
Davis BH, et al. Circulation 20061132201-10
11
Hospitalized/fatal HF in Subgroups - Amlodipine /
Chlorthalidone Relative Risks After 1 Year of
Follow-up
Favors Amlodipine
Favors Chlorthalidone
Relative Risk (95 CI)

1.22 (1.08 - 1.38)
Total
1.38 (1.10 - 1.73)
Age lt 65
1.17 (1.02 - 1.35)
Age 65
1.20 (1.04 - 1.39)
Non-Black
1.28 (1.03 - 1.58)
Black
1.28 (1.09 - 1.50)
Men
1.16 (0.97 - 1.39)
Women
1.23 (1.04 - 1.46)
Diabetic
1.21 (1.02 - 1.43)
Non-Diabetic
0.50
1
2
3
4
5
6
Davis BH, et al. Circulation 20061132201-10
12
Hospitalized/fatal HF in Subgroups - Lisinopril /
Chlorthalidone Relative Risks from Baseline to 1
Year of Follow-up
Relative Risk (95 CI)
Favors Lisinopril
Favors Chlorthalidone
2.08 (1.58 - 2.74)
Total
2.53 (1.39 - 4.59)
Age lt 65
1.98 (1.45 - 2.70)
Age 65
2.04 (1.43 - 2.90)
Non-Black
2.15 (1.39 - 3.33)
Black
1.80 (1.22 - 2.67)
Men
2.40 (1.63 - 3.54)
Women
1.99 (1.31 - 3.05)
Diabetic
2.16 (1.50 - 3.10)
Non-Diabetic
0.50
1
2
3
4
5
Davis BH, et al. Circulation 20061132201-10
13
Hospitalized/fatal HF in Subgroups - Lisinopril /
Chlorthalidone Relative Risks After 1 Year of
Follow-up
Relative Risk (95 CI)
Favors Lisinopril
Favors Chlorthalidone
0.50
1
2
Davis BH, et al. Circulation 20061132201-10
14
4 Unanswered Questions
1. Can the early divergence of HF curves in the
treatment arms be explained by the preferential
discontinuation of diuretics upon entry into
ALLHAT?
15
Potential Confounders

• Confounders by indication why was the patient
  placed on a specific class of drug prior to
  participation in the study?
• Missing data approximately one third of
  HF cases lacked information on specific drugs
  used prior to entry into ALLHAT

Grimm R, et al J Am Cardiol Coll 200749350A
16
Baseline Characteristics of Participants with HF
within First Year Following Randomization
Grimm R, et al J Am Cardiol Coll 200749350A
17
Validation of Case-Only Analyses

• A technique know as case-only analyses was used
  to examine if there was interaction between prior
  drugs and treatment effects.
• Does any prior meds (yes/no) have the same
  interaction effect with treatment on outcomes in
  a cases and non-cases analysis versus a case
  only analysis ?

Grimm R, et al J Am Cardiol Coll 200749350A
18
Interaction OR between prior use of diuretic and
treatment effects in HF

• Prior use of antihypertensive agents
  39 diuretics
• 37 ACEIs
• 47 CCBs
• Prior use of diuretic on A effect for new HF
  A vs C OR
  1.08 (0.53-2.21, p0.83)
• Prior use of diuretic on L effect for new HF
  L vs C OR 1.33 (0.65-2.74, p0.44)

Grimm R, et al J Am Cardiol Coll 200749350A
19
Summary

• Patients on any prior BP med (vs. none) were at
  higher risk of developing HF.
• No evidence for any statistically significant
  interaction between prior drug type (e.g.,
  diuretic) and treatment effect for HF, overall or
  during the first year
• These findings suggest that the type of BP drug
  at entry is not a major determinant of the HF
  results.

Grimm R, et al J Am Cardiol Coll 200749350A
20
2. How accurate is the diagnosis of HF?
21
Origin of the HF Validation Study

• HF endpoint defined as treated in hospital or
  out-of-hospital or fatal
• A component of combined CVD (CHD, stroke, HF,
  PAD) pre-specified secondary outcome
• Systematic central review of hospitalized HF
  events initiated in 2001, on advice of the DSMB

Einhorn PT, et al. Am Heart J 200715342-53
22
HF Validation Study Objectives

• Evaluate ALLHAT site physician-assigned diagnoses
• Evaluate treatment effects reported in December
  2002 (JAMA. 20022882981-2997)
• Compare RRs of validated HF between randomized
  treatment groups with RRs reported in 2002
• Evaluate incidence of validated HF and examine
  subsequent mortality rates as indicators of
  clinical significance of HF

Einhorn PT, et al. Am Heart J 200715342-53
23

HF Validation Study

• 2850 hospital records for 1987 patients received.
• 2778 records of 1935 patients suitable for
  review.
• Centrally abstracted by cardiology fellow blinded
  to treatment assignment.
• Each record independently reviewed by two
  reviewers.
• For algorithmic criteria (ALLHAT and Framingham),
  diagnoses were assigned by computer.
• Reviewers clinical judgment entered as yes, no,
  dont know.

Einhorn PT, et al. Am Heart J 200715342-53.
24
HF Validation Study ACEI versus diuretic
Definition, Relative Risk and 95 Confidence
Intervals


0.50
1
2
Favors Lisinopril Favors
Chlorthalidone
Pre-specified endpoint of treated in hospital
or as outpatient or fatal
25
Percent agreement with investigator-assigned
diagnosis of HF
Einhorn PT, et al. Am Heart J 200715342-53.
26
HF Outcome VerifiedClinically Significant

• ALLHAT site physician diagnoses confirmed in most
  patients
• Treatment differences based on site physician
  reports corroborated when applying validation
  criteria sets
• RRs approximating these for the HF prespecified
  endpoint
• 6-year incidence rates of validated HF events
  comparable to those of stroke (5.6) and to about
  half of non-fatal MICHD deaths (11.4)
• High mortality rates subsequent to validated
  hospitalized HF (55 at 5 years)

Einhorn PT, et al. Am Heart J 200715342-53.
27
3. How important are the blood pressure
differences in the three treatment arms?
28
BP Results by Treatment Group
Compared to chlorthalidone SBP significantly
higher in the amlodipine group (1 mm Hg) and the
lisinopril group (2 mm Hg, and in blacks 4 mm
Hg)
Compared to chlorthalidone DBP significantly
lower in the amlodipine group (1 mm Hg).
29
BP Differences

• Adjustment for follow-up SBP as time-dependent
  covariates in a Cox regression model only
  slightly modified the relative risks
• Amlodipine/chlorthalidone 2.22 ? 2.16 first year,
  1.22 ? 1.18 after 1 year
• Lisinopril/chlorthalidone 2.08 ? 2.01 first year,
  0.96 ? 0.93 after 1 year

Davis BH, et al. Circulation 20061132201-10
30
Exposure to different rates of BP Reduction

• Early, inadequate blood pressure responses are
  never fully corrected (ALLHAT, Syst-Eur, SCOPE,
  ASCOT, VALUE)
• The comparator was never able to catch up to the
  active drug after short differences in initial BP
  despite attempts to increase therapy.

Benefit-Differences Persists Over Time
Time
31
What we dont and never will know!

• 24 hour blood pressure ?
• Night time blood pressure ?
• Central blood pressure ?

32
4. How can differences in secondary endpoints be
termed significant when primary endpoints are
equal in all three treatment arms?
33
Drug comparisons for HF

• Chlorthalidone vs Amlodipine
  RR 1.35 (95 CI 1.21-1.50,
  plt0.0013)
• and consistent with external data
• Meta-analysis RR 1.30 (1.21-1.47) in favor
  of Diuretics/ß blocker over CCBs for
  preventing HF.
  
  (BPLTT Collaboration Lancet, 20033621527)

Yusuf SY, Circulation 20061132166
34
Drug comparisons for HF
Chlorthalidone vs lisinopril HF RR 1.19
(95 CI 1.07-1.31), plt0.001 ?Pre-specified
endpoint of treated in hospital or as
outpatient or fatal (Einhorn PT, et al. Am
Heart J 200715342-53) and
consistent with external data Network
metaanalysis RR 0.88 (0.80-0.96) plt0.01
in favor of a diuretic over ACEI for preventing
HF. (Psaty BM, et al. JAMA 20032892534-2544)

35
Final Conclusions

• Chlorthalidone was superior to amlodipine in both
  time periods in preventing HF in the aggregate
  and in all subgroups age, race, sex, diabetic
  history.
• Chlorthalidone was superior to lisinopril in
  preventing HF during the first year of treatment
  thereafter, the 2 drugs were equally effective in
  preventing HF.
• The ALLHAT studies confirmed that thiazide-type
  diuretics should be a preferred first-step drug
  treatment for prevention of HF in high-risk
  patients with hypertension and/or post MI.

36

Postscript
What constitutes optimal treatment of ACC/AHA
stage A or B HF to prevent progression to stage
Covert symptomatic HF?
37
Heart Failure Causal Mechanisms
Diastolic Dysfunction
LVH
Obesity Diabetes
Vasan RS and Levy D. Archives Int Med 1996
38
Current ACC/AHA Guidelines Management of HF as
Applied to ALLHAT Patients

• ALLHAT patients were divided between stage A
• and B categories (Stage C patients were
  excluded).
• For the stage A patients (high risk without
  structural
• abnormalities), ACEIs and diuretics are
  recommended
• for treatment of HTN.
• For the stage B patients (structural heart
  disease), ACEIs
• and diuretics are recommended for treatment of
  HTN
• ACEIs and BBs are recommended for post MI,
  LVH,
• and reduced LVEF.
• Therefore, poly-pharmacy will be necessary in
  the majority
• of patients for optimal control of HTN (Stage
  A and B) and
• for treatment of structural heart disease
  (Stage B).

Hunt. et al. Circulation 20051121825-1852
39
JNC-7 Guidelines for HF Treatment

• HF is a compelling indication for the use of
  ACEI. Abundant evidence exists to justify their
  use with all stages of HF.
• Blood pressure targets in HF have not been
  firmly established. In most successful trials SBP
  were lowered to the range of 110-130 mm Hg.
• Therefore, ACEI (or ARB)/diuretic combinations,
  rather than single agents, are necessary in the
  majority of patients for achieving optimal
  control of HTN, preventing and/or reversing
  structural heart damage, and preventing
  progression to overt HF.

JNC- 7 Report. JAMA 20032892560-72.