From The Permanente Medical Group, Kaiser Foundation Hospital,
Sacramento, Calif.
Correspondence to Howard Slyter, MD, The Permanente Medical Group, Kaiser Foundation Hospital, 2025 Morse Ave, Sacramento, CA 95825. E-mail Howard.Slyter{at}NCAL.KAIPERM.ORG
Summary of Review—A number of completed trials of proposed stroke
treatments raise classical ethical issues in challenging ways. The
combination of acutely ill and vulnerable patients, the use of
potentially toxic drugs, and very short time frames for decision making
and drug administration demand an especially careful evaluation of risk
and benefit, the process of consent, and the permissible treatment of
control patients.
Conclusions—The study of acute stroke treatments may require more
complex safeguards than other neuroclinical trials.
As stroke trials have been designed, reviewed, and implemented, all of
the classic ethical issues-risk, benefit, consent-have required
attention. Alves and Macciocchi5 have recently
reviewed these issues as they apply to neuroclinical trials in general.
Yet many stroke trials, bringing together combinations of acutely
disabled and vulnerable patients, potentially toxic drugs, and very
short time frames for decision making, pose questions that have yet to
receive a thorough airing.
In this article I will discuss two studies that raise challenging
issues of risk-benefit, consent, and the treatment of patients in
control groups. I will suggest that particular attention be paid to the
ethical importance of the timing of treatment and the ability of acute
stroke patients to give valid consent to participation in clinical
research.
Ethical questions arise around both the lowering of BP to facilitate
enrollment and the subsequent treatment of BP in both control and rtPA
patients whose BP exceeded 180 to 185/105–110. If, as I believe,
reducing BP so soon after presentation, and at this
threshold, was contrary to commonly recommended management and carried
at least a theoretical risk of harm (see below), this intervention
breached the venerable maxim "primum non nocere": first, do no
harm. While those who ultimately received rtPA stood to gain something
critical if their risk of hemorrhage was diminished, there was
no proposed medical rationale for treating the control group in this
way. Urgent BP reduction offered the subjects in the placebo arm no
benefit that would make a risk of hypotension worth taking.
Furthermore, exposing control patients to this unconventional and
potentially risky intervention denied them what control patients should
receive (optimal standard treatment), and by doing this created
potential doubts about the study's conclusion that rtPA is superior to
best current medical management.
The issue of if, and when, to treat elevated BP in the setting of acute
stroke has been debated for decades.8 9 Many
experts have advised that even very severe hypertension not be treated
without an additional medical justification,8 10
while proponents of treatment have advised caution and have set targets
at or above the levels treated in this
study.11 12
The rationales for restraint in the setting of acute stroke are many
and well known: (1) the tendency for hypertension to
self-correct,13 (2) the elevation of the
autoregulatory threshold in many stroke patients and the loss of
autoregulation in the ischemic area,14
(3) the need to maintain perfusion of at-risk (penumbral)
tissue,15 (4) reports of patients with
cerebrovascular symptoms that worsened with BP reductions very close to
this study's intervention point,16 17 18 and (5)
evidence that reduction of MAP exceeding 13% to 16% can impair
cerebral perfusion without producing any clinical signs of
deterioration.19 20 Supporting the traditional
caution about blood pressure, two recently published therapeutic
trials21 22 have suggested an association of
worsened outcomes with rather modest BP reductions in acute stroke
patients.
Restricting BP treatment to "nonaggressive" measures in the NINDS
study was clearly intended to safeguard those receiving rtPA while
minimizing the risk of harm to the placebo group. However, it was not
possible to know in any given patient that BP reduction would not
impair cerebral perfusion to some degree and thereby worsen outcome.
Labetalol in the recommended doses has been reported in a few cases to
reduce BP more than the critical 13% to
16%,23 24 and there are many cases reported of
abrupt hypotension with sublingual
nifedipine.25 While those in the rtPA
group might have been protected from any adverse effects of BP
reduction had reperfusion been established, no such benefit would have
been expected for those in the control group.
Whether the treatment of blood pressure actually harmed patients or
skewed the study's conclusions cannot be gauged without knowing, at
the very least, whether many or few were treated and whether their
outcomes were different than those whose BP did not require
intervention. Aside from this methodological question, there are, in my
view, ethical problems with the otherwise unjustified reduction of BP
to recruit experimental subjects and the further treatment of the
control group with this potentially risky intervention without a
compelling medical rationale.
The issue of what risks one may expose control patients to has also
arisen in studies of intra-arterial
thrombolysis. Such treatment has required angiography
to identify an occluded artery, to administer the drug directly into
this vessel, and then to evaluate the effect of the treatment on the
occlusion. While some have suggested that the performance of
angiography and sham treatment for the control arm of such studies
would expose these patients to unconventional risks without
commensurate benefit,26 others have written that
this would be acceptable if consent were
obtained.27 Although this issue seems as yet
unresolved, at least one such study has already been
reported28 and another is in the planning
stages.29
In a multicenter study of safety, tolerability, and
pharmacokinetics, 51 acute stroke patients received dextrorphan up to
48 hours after the onset of stroke.38 39 To be
eligible, patients had to be alert, neurologically stable, and without
a significant psychiatric disorder. Their baseline National Institutes
of Health Stroke Scale scores ranged from 1 to 20.
All subjects had "adverse events." A quarter of the patients
vomited, more than one third experienced dizziness or nausea, and over
50% suffered somnolence, agitation, hallucinations, and/or confusion.
These latter symptoms were especially frequent during 11-hour
continuous infusions of the drug. Seven of 21 patients receiving the
highest loading doses had symptomatic blood pressure drops
(in 6 patients exceeding 50 mm Hg). Another patient became
hypotensive after treatment of hypertension, which was likely a
consequence of drug-induced agitation. At the highest
maintenance infusion rate, two patients became unresponsive and
one became apneic, requiring intubation. Dextrorphan is no longer in
clinical trials because of the sudden hypotension it can produce.
Several features of this study raise ethical concern. First is
the number of patients who experienced significant adverse events
before the study was completed. Dose-escalation studies are generally
designed to carefully titrate doses upward in a limited number of
patients and to stop when dose-limiting events are encountered. For
instance, when new cancer agents are in phase 1 testing, a common
paradigm is to expose only 3 to 6 patients to each successive dose and
to halt the testing if more than 2 subjects experience grade 3
toxicity.40 (Such a careful, step-by-step
approach was used in a pilot study of rtPA in acute
stroke.41 ) In the dextrorphan study, 21 patients
received the maximal loading dose (200 to 260 mg/h IV), and 7 of these
experienced symptomatic hypotension. Even if one accepts
all of the other unpleasant side effects such as vomiting and
hallucinations, the number of patients suffering potentially harmful
hypotension seems excessive.
From both an ethical and a practical perspective, the most challenging
aspect of this trial was the timing of drug administration. NMDA
antagonists had been reported to be effective in animal
models of focal ischemia only when given within 2 to 3 hours of
stroke onset. One study performed in the early 1990s reported a trend
toward worse outcomes with dextrorphan than with placebo
when administered to rabbits as little as 4 hours after stroke
onset.42 Although the therapeutic window for
neuroprotective agents remains uncertain in humans, most experts have
put it far short of 31 hours, the mean interval to the initiation of
treatment in this study.43 44 If this estimate of
the therapeutic window is even close to correct, there could be little
if any expectation that these patients would benefit medically from
receiving dextrorphan. From their perspective, toxicity was the only
likely outcome.
There are two ethical issues raised specifically by the decision to
administer the drug to patients in the acute phase of their strokes but
after the putative therapeutic window had passed. The first concerns
the effect that the delay in administration had on the risk-benefit
ratio for the experimental subjects, and the second concerns the
patients' ability to give an informed consent.
There has been a tremendous effort at public education and
streamlined care delivery to facilitate enrollment of patients in
therapeutic stroke trials as rapidly as possible, since even
small delays in drug administration may deny patients their best chance
for benefit and thus cause the study to miss a drug's advantage over
placebo.7 45 46 47 Those enrolled in the
dextrorphan study may have been patients who arrived too late for phase
2 or 3 studies and were deemed suitable for a study in which the
primary scientific goal (evaluation of pharmacokinetics and toxicity)
was less time sensitive. Unfortunately, such a delay virtually
eliminated the chance that the new drug would help these patients and
left only the risk of side effects, if not injury. In this
circumstance, the physician's duties to do no harm, to advocate for
his patient's best interests, and to avoid risks without commensurate
benefits were superseded by the quest for pharmacologic data.
An evaluation of the risk-benefit ratio was not only incumbent on
the treating physicians but was one of the primary tasks, and the major
ethical duty, of the IRBs where this research was
conducted.48 IRBs are expected to disapprove
research when the risks to patients are judged to be unreasonable in
relation to the anticipated benefits.48 If the
IRBs failed to grasp the critical effect that timing had on the
risk-benefit ratio in acute stroke therapy, they may have been too
willing to tolerate the foreseeable toxicity of dextrorphan for these
patients.
The second ethical issue related to timing concerns the capacity
of these patients to meaningfully consent to participate in such a
study, and more particularly their ability to distinguish between
interventions with real therapeutic potential for themselves and
interventions in which the risks of toxicity were high and the
potential for benefit remote. As I will describe, there are a number of
barriers to consent in stroke patients, even those who are alert, and
the ability to gain consent is especially problematic when
one is conducting phase 1 trials.
Recognizing the difficulties with informed consent in the acute stroke
setting, particularly when patients are cognitively impaired, many
treating physicians and investigators will turn to close relatives to
secure a proxy (or surrogate) consent to treat the patient or to enroll
him or her in a clinical trial.5 Although the
legal status of proxy consent for research is not well defined in many
states,50 California and New York appear to draw
a distinction between therapeutic research, in which proxy consent may
be valid, and "nontherapeutic" research (ie, research providing no
prospect of direct benefit to individual subjects), in which proxy
consent has unclear validity.51 52 Whether
permitted by law or not, the American College of Physicians explicitly
disapproves surrogate consent for nontherapeutic experimentation that
presents more than a minimal risk of harm or
discomfort.53
New federal regulations allow a waiver of consent altogether for
studies in emergency situations in which life is threatened and an
investigational treatment offers the prospect of direct benefit to the
subject.54 These rules appear to remove a barrier
to enrollment in phase 2 and phase 3 stroke
trials55 ; however, once again, they would not
reasonably apply to toxicity trials if there is no prospect of patient
benefit.
In sum, both legal and ethical considerations should severely limit the
enrollment of cognitively impaired stroke patients in toxicity trials,
even if family members can be found who will consent to such
studies.
Finally, there are problems that may arise when the treating physician
and the researcher are the same person or part of the same team. In
this situation the acutely ill patient or his family cannot always
distinguish between standard treatments, therapeutic trials, and
toxicity studies.56 They will usually do what the
doctor recommends. Of course, the duty of the treating physician is to
promote and safeguard the patient's welfare. But given the dual role
of physician and scientist, conflicts may arise between this duty and
the goals of the investigation.57 Because of this
potential for conflict of interest, many commentators have advised that
the roles of treating physician (or team) and researcher be
separated,57 58 especially in phase 1 studies,
where patients routinely mistake what they undergo as receiving
therapy.56 59 60
I believe the NINDS study ventured onto ethically debatable ground when
the decision was made to reduce patients' BP (however carefully) so
they could qualify for the trial. Further, the continued treatment of
hypertension in control patients added an unconventional risk to their
care that was not medically or ethically justified.
The dextrorphan study demonstrates what willing stroke patients
may be asked to endure for the sake of scientific advancement-a toxic
ordeal not redeemed by any likely medical benefit. It is a troubling
picture that leads me to suggest the application of the following very
strict guidelines for the conduct of phase 1 research involving acute
stroke patients:
(1) Because of formidable problems in obtaining informed consent,
dose-escalation/toxicity studies of new stroke drugs should be
undertaken and completed in normal healthy volunteers or in patients
sufficiently beyond the acute phase of illness to minimize
misunderstanding of the nontherapeutic nature of the research.
(2) If dose-escalation studies must be done in acute stroke patients,
several safeguards should be in place: (a) The IRB should consider
these patients as a "vulnerable population" and impose additional
safeguards that in its judgment are appropriate to the
research.61 (b) The IRB should be clearly
informed of the critical importance of the timing of drug
administration so it can properly evaluate the risk/benefit ratio for
potential enrollees. (c) A protocol of carefully monitored dose
escalation should be employed. (d) Only stroke patients with intact
decision-making capacity should be recruited. Proxy consent for
enrollment, especially beyond the likely therapeutic window, should not
be permitted. (e) There should be a physician responsible for the
patient's care who is not a part of the research team.
At the same time, we must remember that at least five large studies of
thrombolytic and neuroprotective agents, involving
thousands of patients, have been stopped when the complications of
experimental treatment appeared
unacceptable.21 62 63 64 65 The experience of patients
in one phase 1 trial has been described earlier. Therefore,
participation in acute stroke research thus far has proved to be a
risky undertaking for many patients.
Those engaged in stroke research are aggressively pursuing the
twin goals of improved care and better outcomes for stroke victims,
which cannot be reached without clinical trials involving thousands of
patients. Those conducting such trials surely view them as
opportunities to offer patients the latest and most hopeful
interventions and are anxious to proceed as quickly as possible.
However, this enthusiasm must be tempered by a candid assessment of the
risks to which patients will be exposed and a sensitivity to the
ethical complexity of studying such an acutely ill and vulnerable
population.
As the philosopher Hans Jonas wrote nearly 30 years ago,
Let us not forget that progress is an optional goal, not an
unconditional commitment, and that its tempo in particular, compulsive
as it may become, has nothing sacred about it. Let us also remember
that a slower progress in the conquest of disease would not threaten
society, grievous as it is to those who have to deplore that their
particular disease be not yet conquered, but that society would indeed
be threatened by the erosion of those moral values whose loss, possibly
caused by too ruthless a pursuit of scientific progress, would make its
most dazzling triumphs not worth having.66
Received July 28, 1997;
revision received November 26, 1997;
accepted November 26, 1997.
© 1998 American Heart Association, Inc.
Comments, Opinions, and Reviews
Ethical Challenges in Stroke Research
Abstract
Top
Abstract
Introduction
Thrombolysis and the Treatment...
The Timing of Neuroprotective...
The Stroke Patient as...
Discussion
Conclusion
References
Background—Ethical issues are a
critical consideration in the design and conduct of clinical
research.
Key Words: clinical trials • ethics, medical • informed consent
Introduction
Top
Abstract
Introduction
Thrombolysis and the Treatment...
The Timing of Neuroprotective...
The Stroke Patient as...
Discussion
Conclusion
References
In the past decade,
clinical research into potential treatments for acute stroke has
dramatically increased. With the support of the NINDS and the National
Stroke Association, the watchword has been "acceleration"-both of
the involvement of patients in studies and of the research enterprise
itself.1 2 Although the translation of animal
experiments into human trials has, in some cases, been criticized as
ill conceived,3 4 this has been a relatively
minor voice in a chorus of enthusiasm.
Thrombolysis and the Treatment of Control
Groups
Top
Abstract
Introduction
Thrombolysis and the Treatment...
The Timing of Neuroprotective...
The Stroke Patient as...
Discussion
Conclusion
References
In December 1995, the groundbreaking NINDS
study6 reported that rtPA can reduce long-term
disability from acute ischemic stroke with an acceptable margin
of safety when given to carefully selected patients within the first 3
hours of the onset of symptoms. Because pilot studies had suggested
that patients with severe hypertension would be at increased risk of
cerebral hemorrhage with rtPA, potential enrollees who were severely hypertensive were treated with
"nonaggressive" measures, such as intravenous boluses
of labetalol and/or nitroglycerin paste, to try to
reduce their BP to an acceptable range-below 185 systolic and
110 diastolic.7 Once patients had
been enrolled and randomized to rtPA or placebo, their blood pressures
were closely monitored, and because it was a double-blind study,
antihypertensive treatment was given to patients in either group if
pressure exceeded the threshold of 180/105. Treatment options included
intravenous sodium nitroprusside, intravenous
labetalol, or oral or sublingual nifedipine. During and
after treatment patients were to be observed for hypotension.
The Timing of Neuroprotective Drugs
Top
Abstract
Introduction
Thrombolysis and the Treatment...
The Timing of Neuroprotective...
The Stroke Patient as...
Discussion
Conclusion
References
The excitotoxicity theory of cell death in acute cerebral
ischemia has focused attention on glutamate as a potent
neurotoxin. The finding that NMDA antagonists could reduce
glutamate-induced cell death in tissue culture led to the testing of a
number of candidate agents in animal models of
stroke.30 31 Although concerns existed about
possible neuronal injury induced by NMDA antagonists, it
appeared time to proceed to human phase 1 trials when several animal
studies reported encouraging results if the drugs were given within 2
to 3 hours after stroke onset.31 32 33 By the early
1990s there were some data on the effects of dextrorphan, and the
closely related dextromethorphan, in humans at therapeutic doses, and
there was substantial reason to anticipate that when given to stroke
patients, there would be psychotomimetic and vestibular side
effects,33 34 35 if not hypotension or respiratory
depression.30 36 Nevertheless, it was hoped that
the neuroprotective benefit would outweigh any short-term, reversible
adverse effects.37
The Stroke Patient as Research Subject
Top
Abstract
Introduction
Thrombolysis and the Treatment...
The Timing of Neuroprotective...
The Stroke Patient as...
Discussion
Conclusion
References
The issue of whether an individual has the capacity to provide
informed consent to treatment is familiar to all who see acutely ill
patients. Such patients are under stress and vulnerable-suddenly ill,
fearful, and likely to trust their caregivers implicitly to do what is
best for them. In the case of stroke patients in particular, aphasia,
decreased alertness, and preexisting deficits may all impair
comprehension. One stroke researcher has written that, on a neurologic
basis alone, most stroke victims are incapable of giving informed
consent.49 When the time pressure imposed by the
short therapeutic windows of many new drugs is added, it is all the
more difficult to secure ethically valid consent for treatment, let
alone for participation an experimental trial.
Discussion
Top
Abstract
Introduction
Thrombolysis and the Treatment...
The Timing of Neuroprotective...
The Stroke Patient as...
Discussion
Conclusion
References
The balancing of ethical and logistic considerations makes the
design and conduct of acute stroke trials more difficult than many
other areas of clinical research. As shown in the NINDS rtPA trial, the
design of an efficient double-blind controlled study may conflict with
strict ethical constraints regarding both potentially risky
interventions to promote patient enrollment and the optimal treatment
of patients assigned to the control group. The dextrorphan study
highlights the more classic conflict between the desire to learn how
these new drugs might be tolerated by stroke patients in general and
the duty to protect the immediate patients/subjects from undue harm. In
all stroke studies, securing the patients' consent is difficult
because the time frame for intervention is so short and the patients
are suddenly ill and may be cognitively impaired. If, as in many
institutions, stroke specialists or teams both provide the care and
conduct the research, a time-saving efficiency is accomplished, but
only through giving physicians the conflicting roles of caregiver and
researcher.
Conclusion
Top
Abstract
Introduction
Thrombolysis and the Treatment...
The Timing of Neuroprotective...
The Stroke Patient as...
Discussion
Conclusion
References
The emergence of many factors-the high prevalence and devastation
of stroke, recent advances in understanding the pathophysiology of
ischemic injury, a cadre of dedicated researchers, major
investments by the pharmaceutical industry-has led to a flowering of
clinical research unprecedented in the area of stroke. While several
trials of thrombolytic and neuroprotective agents have
already been completed, many more are under way or expected in the next
decade. The era of nihilism in stroke is, thankfully, over.
Selected Abbreviations and Acronyms
BP
=
blood pressure
IRB
=
institutional review board
MAP
=
mean arterial pressure
NINDS
=
National Institute of Neurological Disorders and Stroke
NMDA
=
N-methyl-D-aspartate
rtPA
=
recombinant tissue-type plasminogen
activator
Acknowledgments
Howard Barkan, DrPH, reviewed this manuscript and made
helpful suggestions.
Footnotes
The opinions expressed in this article are not necessarily those of the editors or of the American Heart Association.
References
Top
Abstract
Introduction
Thrombolysis and the Treatment...
The Timing of Neuroprotective...
The Stroke Patient as...
Discussion
Conclusion
References
This article has been cited by other articles:
 |
|
 |
|
  J F Meschia and J G Merino Reporting of informed consent and ethics committee approval in genetics studies of stroke J. Med. Ethics, December 1, 2003; 29(6): 371 - 372. [Full Text]
|
|
|
|
 |
|
 B. T. Bateman, P. M. Meyers, H. C. Schumacher, S. Mangla, and J. Pile-Spellman Conducting Stroke Research With an Exception From the Requirement for Informed Consent Stroke, May 1, 2003; 34(5): 1317 - 1323. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
G. W. Albers, J. A. Zivin, and D. W. Choi Ethics in Clinical Trials : Response to Ethical Challenges in Stroke Research Stroke, August 1, 1998; 29(8): 1492 - 1493. [Full Text]
|
|
|
|
|
|
G. W. Albers, J. A. Zivin, and D. W. Choi Ethics in Clinical Trials : Ethical Standards in Phase 1 Trials of Neuroprotective Agents for Stroke Therapy Stroke, August 1, 1998; 29(8): 1493 - 1494. [Full Text]
|
|
|
|
|
|
J. R. Marler and M. D. Walker Ethics in Clinical Trials : Progress ni Acute Stroke Research Stroke, August 1, 1998; 29(8): 1491 - 1492. [Full Text]
|
|
| |||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||