BiDil for Heart Failure in Black Patients: Implications of the U.S. Food
and Drug Administration Approval
Kirsten Bi
ins-Domingo, PhD, MD, and Alicia Fernandez, MD
In 2005, the combination of hydralazine hydrochloride and isosor-
ide dinitrate was approved by the U.S. Food and Drug Adminis-
tration (FDA) for treating heart failure in black patients. In depart-
ing from its long history of approving drugs for general clinical
indications without regard to demographic classification, the FDA
cited the need to address racial disparities in health as an important
contributor to their decision. The authors argue that this decision,
although perhaps well-intentioned, was based on flawed scientific
interpretation of trial results that claimed differential drug response
y race and ignored the considerable literature on the cause of
acial disparities in health and health care. Because of its potential
impact on future drug approvals, the FDA’s decision is a setback in
the scientific and policy discourse on medical therapeutics and race
and specifically hinders the efforts aimed at eliminating health and
health care disparities.
Ann Intern Med. 2007;146:52-56. www.annals.org
For author affiliations, see end of text.
In 2005, the U.S. Food and Drug Administration (FDA)approved BiDil (NitroMed, Lexington, Massachusetts), a
combination of 2 generic medications—hydralazine hy-
drochloride and isoso
ide dinitrate—in a single tablet, fo
treating heart failure in black patients. BiDil was approved
for a specific racial group—which is a first for the FDA and
is a clear departure from the FDA’s long history of approv-
ing treatments for clinical conditions regardless of demo-
graphic classification (with the occasional exception of sex).
One rationale for this departure was that the drug approval
was important for addressing racial disparities in health
(1–3). We argue that the FDA decision, although perhaps
well-intentioned, may be a setback to scientific discourse
on therapeutics and may be specifically deleterious to ef-
forts aimed at addressing disparities in health and health
care.
CLINICAL TRIALS AND DRUG APPROVAL
Because many patient characteristics can influence
variability in drug response, randomized, controlled trials
have attempted to include a
oad spectrum of the patient
population, an approach specifically endorsed by the Na-
tional Institutes of Health policy on the inclusion of
women and minorities (4). Including a
oad patient pop-
ulation assures that most information on variations in drug
efficacy is available for patients who are affected by the
same condition. It also allows for investigators to research
explanatory factors that may underlie variability, from sim-
ple descriptions (for example, “drug A is less effective in
older individuals”) to more complex explanations (for ex-
ample, “the lower efficacy of drug A in older individuals is
partially explained by a higher burden of coronary dis-
ease”).
The merits of including
oad patient populations in
andomized, controlled trials are most clearly seen in stud-
ies in which this practice has not occu
ed, such as the
trials of aspirin for the primary prevention of coronary
disease. Although earlier trials—almost exclusively in
men—demonstrated the efficacy of aspirin (5–7), a recent
trial in women did not show this benefit (8). Were the
observed differences because of fundamental physiologic
differences between men and women in their manifestation
of coronary disease and their response to aspirin or because
of the consequences of specific features of the trials (such as
distinct dosing regimens, variations in inclusion criteria, o
advances in other adjunctive therapies)? A trial that had
included adequate numbers of both men and women
would have answered many lingering questions despite data
from well-done trials done in each population separately.
In the past, trials have restricted enrollment to specific
demographic groups. The African American Study of Kid-
ney Disease and Hypertension (AASK) trial (9, 10) is a
ecent prominent example. This randomized, controlled
trial of different antihypertensive treatments for hyperten-
sive nephrosclerosis included only black participants. Black
people are disproportionately affected by hypertension and
kidney disease, and arguments both for efficiency and eq-
uity underlie the decision to explore treatments in this
understudied patient population. However, the AASK in-
vestigators did not claim a race-specific effect of these ther-
apies and reported instead on the general treatment effects
for hypertensive nephrosclerosis (9, 10). A fundamental
consistency in human biological responses is assumed in
this type of trial reporting and in the FDA drug approval
process, which has for years based approval of therapeutics
for the general population on evidence from trials in
mostly white, mostly male patient populations. Drugs are
approved for treating clinical entities, not for treating spe-
cific demographic subgroups. Departures from this logic
should require a compelling scientific argument with clea
evidence of the biological mechanism underlying the dif-
ferential response. Pharmacogenomic advances may, ove
See also:
Print
Related article. . . . . . . . . . . . . . . . . . . . XXXXXXXXXX57
Annals of Internal Medicine In the Balance
52 © 2007 American College of Physicians
time, provide that scientific argument, but such data are
not available in most cases. The recent case of bucindolol
offers some important insights.
SCIENTIFIC IMPLICATIONS OF THE FDA APPROVAL OF
BIDIL
The FDA approval of BiDil was based on the results of
the African American Heart Failure Trial (A-HeFT) (11), a
andomized, controlled trial of BiDil added to standard
therapy for self-identified black patients with heart failure.
The makers of BiDil had initially sought approval for the
drug in the general patient population (not just in black
patients) on the basis of results of earlier trials of the ge-
neric combination (12, 13), but approval was denied be-
cause of statistical concerns that the earlier trials had failed
to account for the multiple end points analyzed (14). In
addition, the earlier trials did not address the important
question of whether the hydralazine hydrochloride–isosor-
ide dinitrate combination confers benefit when added to
standard heart failure therapy (including angiotensin-con-
verting enzyme [ACE] inhibitors and !-blockers). The A-
HeFT evidence was clear—BiDil, when added to standard
therapy, reduced mortality in black patients with advanced
heart failure. Faced with this evidence, the FDA had 2
choices: approve BiDil for the general population—a prac-
tice consistent with years of previous drug applica-
tions—or approve BiDil only in black patients. In choos-
ing the latter, the FDA chose the path with the most
concerning scientific consequences both for heart failure
treatments and for studies of medical therapies.
1. Drug approval for specific groups implies a differential
drug response that has not been rigorously tested.
The FDA approval of BiDil in black patients implies
that the treatment is efficacious only in black people and
not in other racial groups. Although a 1999 post hoc sub-
group analysis of 2 earlier hydralazine hydrochloride–
isoso
ide dinitrate trials in black and white patients
(Veterans Administration Cooperative Vasodilator Heart
Failure Trial [V-HeFT] I [12] and II [13]) suggested that
the response to treatment may differ among races (15), the
evidence is insufficient to base the cu
ent FDA decision.
First, because the V-HeFT II investigators included enala-
pril as an active treatment comparison group, the subgroup
analysis could not determine whether the observed racial
difference was due to an increased response to hydralazine
hydrochloride–isoso
ide dinitrate or a decreased response
to enalapril among black participants. Second, these olde
trials do not provide evidence for differential response to
hydralazine hydrochloride–isoso
ide dinitrate among con-
temporary patients who are receiving other standard heart
failure therapies.
Finally, and perhaps most important, evidence from
the subgroup analysis must be interpreted with the same
caution as with other post hoc analyses of particular sub-
groups enrolled in clinical trials. Although such analyses
are generally performed to determine whether response dif-
fers between groups, observed differences are far more
likely to be in the degree of response (for example, “drug A
is more efficacious in diabetics than nondiabetics”) than in
the type of response (for example, “drug X is efficacious
only in diabetics and not in nondiabetics” XXXXXXXXXXReports of
qualitatively different responses have not been consistently
eplicated and are particularly suspect when hypotheses
have not been prespecified or appropriate adjustments have
not been made for multiple hypothesis testing (16, 17)
(which are both well-documented critiques of the analyses
from V-HeFT I and II [1, 14]). Notably, the V-HeFT
subgroup analysis did not prompt the major heart failure
practice guidelines to recommended using these medica-
tions differently in black and white patients (18, 19).
In studying only black patients, the recent A-HeFT
trial could not address the lack of treatment efficacy in
other racial groups. However, because the approval of Bi-
Dil for black patients implies a lack of treatment efficacy in
other race groups, the FDA decision has important clinical
and scientific implications. Clinically, the use of this effec-
tive medication in other demographic groups is now tech-
nically “off-label” (a paradox for BiDil because the use of
the generic combination as an alternative to standard ther-
apy is approved for all individuals). Scientifically, racial
differences in response to hydralazine hydrochloride–
isoso
ide dinitrate (and perhaps other heart failure medi-
cations) may now be thought of as “proven,” although the
statistical interaction on which such a conclusion is based
has not explicitly been tested. In addition, untested as-
sumptions about differential responses by race generally
lend credence to other untested assumptions, including
heart failure as a different disease entity in black patients.
We should note that plausible biological mechanisms
may explain a differential response of black patients to
hydralazine hydrochloride–isoso
ide nitrate, namely in-
creased levels of nitric oxide in black people that are asso-
ciated with adverse heart failure outcomes and may be re-
duced by hydralazine hydrochloride–isoso
ide nitrate
therapy (20). While such observations certainly merit fur-
ther exploration, the association of a potential mediator of
the differential drug response with a certain race group
should not be the sole basis for drug approval exclusively in
that race group.
Recent data on racial differences in drug response to
another heart failure drug—bucindolol—offer an impor-
tant lesson in this regard. The negative results of the ran-
domized, controlled trial of bucindolol (the Beta-Blocke
Evaluation of Survival Trial [BEST] [21]) are in stark con-
trast to those of several trials of other !-blockers that show
mortality benefits in heart failure (22–24). The BEST au-
thors speculated that the reason for their negative results
may have been the substantial numbers of black partici-
pants in the trial and a decreased response to !-blocke
therapy among them. In fact, subgroup analysis of the
BEST results demonstrated a survival benefit among the
In the BalanceImplications of BiDil for Treating Heart Failure in Black Patients
www.annals.org 2 January 2007 Annals of Internal Medicine Volume 146 • Number 1 53
nonblack participants that was not observed among the
lack participants. Recent genetic evidence suggests a bio-
logical mechanism for this difference because a polymor-
phism in the !1-adrenergic receptor seems to be responsi-
le for drug response. This polymorphism is more
common in white patients (53%) than in black patients
(38%), and when the genotypes are accounted for, the