Placebo Controlled Procedural Trials for Neurological Conditions

Summary

Neurological disease has been a central focus in the ongoing ethical debate over the use of invasive placebo controls, especially sham surgery. The risk to research subjects and necessary use of deception involved in these procedures must be balanced against the methodological need to control for bias and the placebo effect. We review a framework formulated for the ethical assessment of sham surgery in the context of research evaluating novel procedures for neurological conditions. Special issues raised include the growing evidence of expectation and conditioning effects in a number of neurological diseases, the escalating scale of risk from different types of invasive placebo interventions, and the increasing use of crossover designs, which allow a switch from placebo to active intervention without additional procedures.

Introduction

The ethical debate on sham surgery has intensified in the past decade, particularly in response to placebo controlled trials of dopaminergic cell transplantation for Parkinson’s disease.^1–4 Ethical discourse on placebo or sham surgery was initiated in a 1961 commentary by Henry Beecher, who argued for its utility in randomized trials to control for evaluator bias and the placebo effect.⁵ Sham surgery under double-blind conditions exposed mammary artery ligation to be no more effective than a placebo intervention in treating angina.^6,7 In 2002, a sham controlled trial of arthroscopic lavage and debridement for osteoarthritis of the knee similarly discredited an invasive and costly procedure.⁸ Using these trials as examples, we formulated a framework for the ethical assessment of placebo surgery.^9–11 Despite an increase in the acceptance and design of sham-controlled surgical trials among physician-scientists,^12,13 ethical opposition has mounted on issues of scientific necessity and the level of risk posed to patient-subjects.^14,15

Neurological procedure trials have been at the forefront of this discussion primarily for two reasons. First, neurological procedures range from minimally to critically invasive and may involve conditions of a highly emergent nature. Sham controls at the extreme end of this spectrum may pose a risk of death and irreversible sensory, cognitive or motor dysfunction. Second, evidence for the placebo effect and an understanding of its mechanisms are emerging for various neurological conditions including pain, depression and motor disorders, such as Parkinson’s disease.¹⁶ In these conditions, the parts of the nervous system subject to disease may be modulated directly by expectation and/or conditioning. These placebo effects must be controlled for in order to dissociate these processes from those reflecting the efficacy of a given procedure. In addition, blinding patient-subjects to the treatment they receive is necessary to avoid bias in trials evaluating subjective outcomes, which are common in neurological research.

The goal of this article is to elaborate on our ethical framework for sham invasive procedures with a special emphasis placed on highly invasive procedures for neurological conditions. We will review six benchmarks for the justification of sham surgery and use examples from the literature that represent critical points along a spectrum of risk.

Methodological Justification

Commentators opposing the use of sham surgery argue that it violates the principle of nonmaleficence in posing risk to research participants without the prospect of direct benefit.^2,3,24,17 Unlike inert pharmaceutical placebos, invasive sham surgery may involve risks associated with surgical incision, general anesthesia and immunosuppression. This objection is not valid because it conflates the ethics of clinical care with the ethics of clinical research.^9,18 While the aim of clinical care is to maximize the health and well-being of particular patients, the goal of research is to acquire clinically relevant knowledge that will benefit future patients.⁹ In clinical research, there is no requirement to offer participants the prospect of direct benefit or to ensure the absence of risk. Indeed, critical areas of biomedical research, including Phase I toxicity testing and studies of pathophysiology, expose participants to risk without the prospect of direct benefit. This research is necessary in order to translate medical innovation into safe and effective treatments.

Reliable data on the efficacy of novel interventions, whether pharmaceutical or procedural, enables physicians to choose the most effective among treatment options and avoid the risks and unnecessary cost of inferior or ineffective treatments. In clinical trials, the control arm to which a novel intervention is compared is critical to the evaluation of its efficacy. Sham surgery controls for four factors that may lead to false positives: 1) investigator bias during unblinded evaluation, especially when outcomes involve subjective rating scales, 2) patient-subject bias in unblinded reporting, 3) the placebo effects of expectation or conditioning and 4) any additional therapeutic effects due to processes unrelated to the hypothesized mechanism of the active intervention. In many cases, sham surgery is the only way to ensure rigorous control for these confounding factors.

Double-blinding is especially important in neurological trials because outcome measures often involve ratings with a degree of subjectivity, as with pain, depression, motor function, and quality of life.¹⁹ Controlling for the placebo effect associated with invasive neurological procedures is also critical, as expectation may alter neurological outcomes in significant ways. Expectation of treatment alone has been shown to modulate a diversity of physiological processes, including a decrease in blood flow to pain-processing brain areas in healthy subjects subjected to painful stimuli, a decrease in activity in neurons of the subthalamic nucleus correlated with increased motor performance in Parkinson patients, and an increase in striatal dopamine release in Parkinson patients.^20–22 Conditioning may contribute an additional therapeutic effect, as it has been shown that associating a flavored drink with cyclosporin A leads to conditioned immunosuppression in human subjects.²³ Patient-subjects in neurosurgical trials, such as those for Parkinson’s disease, have potentially experienced repeated clinical treatments or research trial interventions.²⁴ These individuals may be conditioned to respond therapeutically to procedures included in a clinical trial, or conversely, they may harbor lowered expectations for any future interventions, however novel. Finally, non-specific effects of a procedure, such as lesioning due to electrode placement for deep brain stimulation in epilepsy, have been shown to have therapeutic effects themselves.²⁵ Therefore, there is often no effective alternative to sham surgery that will effectively blind the evaluation and reporting of neurological symptoms from bias, as well as control for the non-specific effects of the procedure and the neuromodulatory effects of expectation and conditioning.

These methodological considerations are ethically relevant. The risks to which research subjects are exposed cannot be justified unless the research design is scientifically valid and thus has the potential to produce clinically valuable knowledge.²⁶

Ethical Evaluation of Risks and Benefits

Despite sound scientific reasons for sham-controlled trials of neurological procedures, patient-subjects must be protected from excessive risk. In the context of clinical research, risk must be minimized and balanced by the prospect of direct benefit and/or the knowledge to be gained from the study.²⁶ As mentioned above, sham surgery typically exposes subjects to the risk of an invasive procedure without the potential of direct benefit from the active intervention. However, patient-subjects in the sham arm also forgo the potential risks of the active intervention, which may include hemorrhage and/or infection from dural penetration during implantation or lesion operations, dyskinesia from deep pallidal stimulation for Parkinson’s disease, or nausea and/or anorexia from pallidal infusion of glial cell line-derived neurotrophic factor (GDNF) in Parkinson patients.^27–29 Thus, sham surgery should not be assumed a priori to present a less favorable risk-benefit ratio than the active arm. Patient-subjects in the sham and active arms encounter different risk-benefit ratios and these should be assessed and justified independently. The former are exposed to risks of the sham surgery that are balanced not by personal benefit but the importance of clinically useful data, the interpretation of which necessitates a placebo control. The latter are exposed to risks of surgery and active treatment that are balanced by the prospect of direct benefit as well as the clinical value of the data.

Given the absence of direct benefit other than the potential therapeutic benefit of the placebo effect and other nonspecific aspects of the procedure, what level of risk in sham surgery is ethically acceptable? A reasoned principle specifying the upper limit on acceptable risk in clinical research has yet to be effectively formulated, if one exists at all, and we recommend a case-by case process of deliberation and judgment. As a first step, risks of the sham procedure should be compared to the threshold of acceptable risks posed in other areas of research that offer little or no prospect of direct benefit, such as Phase I toxicity testing, or natural history studies involving diagnostic procedures, such as biopsies, lumbar puncture, or bronchoscopy.¹⁰ Next, a careful assessment of whether known and unknown risks from the novel procedure are acceptable should be performed by the investigators and institutional review board, possibly with consultations from patient advocates and members of the public. A useful operational approach to risk assessment proposed by London³⁰ recommends that equal consideration be given to the interests of subjects bearing the burden of research risks and to the interests of the individuals who will eventually receive the benefits of the knowledge to be gained. Below, we review six recent examples from the literature that illustrate the spectrum of risk-benefit considerations in sham controls of neurological trials (See Table 1). Critical factors in the ethical assessment of risk are highlighted.

Table 1.

Escalating Risk in Placebo Controlled Trials for Neurological Procedures

Trial Description	Placebo Control	Risk-Benefit Ratio	Ethically Acceptable?
transcranial magnetic stimulation for epilepsy (29)	placement of coils over skull without stimulation	no risk, no prospect of benefit from active intervention	yes
acupuncture for migraine (32)	fine needle stimulation, no manual stimulation	very low risk, no prospect of benefit from active intervention	yes
deep brain stimulation for primary dystonia(26)	implantation of device, no stimulation possibility of crossover after trial	penetration of dura, device implantation, prospect of direct benefit in crossover to active intervention	yes
cell transplantation for Parkinson’s disease (1)	bilateral full burr holes into skull	no penetration of dura, local anesthesia, no prospect of benefit from active intervention	yes
intracranial drug infusion for Parkinson’s disease (27)	catheter implantation, infusion of saline	penetration of dura, device implantation, continual risk of infection, hemorrhage, and dislodgement during placebo infusion, prospect of direct benefit in crossover to active intervention	?
bypass surgery and laser perforation for aneurysm in arterio-venous malformation (36)	not designed	emergent condition, withholding of intervention is life thretening	no

Examples of Escalating Risk

Transcranial magnetic stimulation (TMS)

A number of studies have tested the efficacy of TMS in treating epileptic seizures, motor function after stroke, and memory after cognitive decline.^31–33 A non-invasive procedure delivering magnetic pulses to the cortex, TMS decreases cortical excitability using high frequency stimulation and increases cortical inhibition using low frequency stimulation. A related technique, transcranial direct stimulation (TDS) applies weak DC current to the cortex and has been tested in sham controlled trials for traumatic spinal cord injury.³⁴. Sham application of both techniques involves placement of coils or electrodes over the skull with either a diversion of magnetic pulses or current away from the skull or a complete absence of stimulation. These procedures are non-invasive and pose very little risk.

Acupuncture

In 2006, Linde et al.³⁵ conducted a randomized, sham controlled trial of acupuncture to prevent migraine headaches in patients suffering from 2 to 8 attacks per month. In the sham arm, patient-subjects received fine needle insertion, but no manual stimulation, at non-acupuncture points. This sham procedure involved minimal invasiveness and the level of risk was minimized. Ethical issues of deception are raised, as subjects were not informed that they had a chance to receive a placebo intervention,³⁶ and administration of the sham required extensive deception as the acupuncturist “pretended” to perform valid procedures. These issues will be discussed later.

Deep brain stimulation (DBS)

Implantation of devices producing high frequency neurostimulation allows for a crossover design in which all patient-subjects receive the activated device. Those randomized to placebo (no stimulation) may be switched to active treatment after the end of a blinded evaluation period. A 2006 sham-controlled crossover trial tested the therapeutic efficacy of pallidal DBS in patient-subjects with primary dystonia.²⁸ All participants received bilateral surgical implantations of a quadripolar electrode device under general anesthesia. Surgery penetrated the dura but the ability to control device settings offered both the active and placebo group a prospect of direct benefit. Patient-subjects in the placebo arm were exposed to the full risks of a complete implantation surgery, which included infection and seroma, but also received the prospect of direct benefit after a 3-month blinded evaluation period. Trials with this design pose a reasonable level of risk to subjects in the sham arm; invasiveness of device implantation is balanced by prospect of direct benefit. The placebo group experiences a delay in receiving a potentially useful treatment, but also avoids any unknown risks of the active treatment, which in the cited case, included dysarthria and dysethesias, until data on efficacy has been collected. An on-off crossover design that allows for a switch from the sham to active intervention without additional procedures permits greater risk than designs that lack this feature, given the prospect of benefit from the active intervention.

Cell transplantation

There have been two published sham-controlled trials of human fetal dopaminergic cell transplantation into the putamen of Parkinson’s patients.^1,37 Placebo surgery involved burr holes into the skull with no penetration of dura. One trial administered local anesthesia alone with no immunosuppression, while the other used general anesthesia and immunosuppressive treatment. In both trials, risk of the sham surgery was minimized by limiting the invasiveness of the procedure to burr holes sufficient to maintain the double-blind and control for the placebo effect of expectation, which has been characterized in biochemical, neurophysiological and behavioral assays in patients with Parkinson’s disease. Non-specific effects of the more invasive injection procedure were not controlled for, but were appropriately judged in this case as unlikely to contribute any therapeutic mechanism alone. This type of trial design involving a moderately invasive sham surgery without penetration of the dura was judged to pose an acceptable level of risk in non-beneficial research by 90.2% of investigator members of the Parkinson’s Study Group.¹² Moreover, the clinical importance of using a randomized, sham-controlled trial to test fetal dopaminergic grafting is underscored by the negative results obtained, which counter earlier positive results from open-label trials.²⁷ The demonstrated lack of efficacy in the tested protocols has driven the field forward to make changes in order to improve this technique.

Intracranial drug infusion

Catheter implantation is a general technique for allowing targeted drug delivery to dysfunctional regions of the brain. Two sham-controlled trials have been conducted to test the efficacy of glial cell derived neurotrophic factor (GDNF) infusion into the cerebral ventricles or putamen of Parkinson’s patients. In these trials, the sham arm involves saline infusions through a surgically implanted catheter in order to control for bias and the placebo effects of expectation.^29,38 Surgical implantation also controlled for nonspecific effects of catheter placement into the putamen or cerebral ventricles. Patient-subjects in the placebo group were allowed to cross over to active GDNF infusion at the end of the trial if indicated. Like the deep brain stimulator device implantation, catheter placement is invasive past the dura. However, maintenance and continual use of the catheter poses additional risks of infection, hemorrhage, and dislodgement. Although all patient-subjects received eventual prospect of direct benefit, the time before cross-over in which the sham group was exposed to these substantial risks without the prospect of direct benefit carries a heavy burden of proof and may exceed a threshold of acceptable risk.

Bypass surgery

Invasive surgeries with critically emergent outcomes are not appropriate for sham-controlled trials. An example is a novel artery ligation and laser perforation procedure to prevent the pending rupture of an aneurysm in patients with arterio-venous malformation.³⁹ This surgery allows for a bypass graft to be ligated without clamping the artery, which imposes the risk of ischemia and subsequent brain damage during ligation. The risks of this invasive and life-threatening surgery are justified by the prospect of direct benefit during an emergent state of a potential aneurysm rupture. The decision to operate is made when the risks of withholding active intervention are judged greater than those of the operation. Because the active intervention would never be tested in a non-emergent context, a sham control, however minimally invasive, would exceed an acceptable level of risk by withholding treatment.

Informed Consent and the Use of Deception

In addition to risk, ethical concern over the use of deception in administering sham procedures has been raised.² Unlike pharmacologic placebos, sham procedures and surgery sometimes require that an investigator actively mislead participants into believing that the active intervention is being performed in order to maintain the blind and control for the placebo effect. In the sham acupuncture trial published by Linde et al.,³⁵ acupuncturists were trained to perform the sham procedure with oral instructions, a videotape and a brochure. In the Freed et al.¹ study of fetal cell transplantation in Parkinson’s patients, subjects receiving sham surgery were awake under local anesthesia and were asked if they were “ready for the implant.”

Active deception on the part of investigators appears to violate clinical research requirements of informed consent and respect for subjects, including the autonomous decision to participate in research.²⁷ Authorized deception, a process by which patient-subjects are informed of the possibility of deception in trial participation and of the risks involved with this deception, is an effective way to maintain informed consent and respect for autonomous decision making while effectively controlling for expectation effects.⁴⁰ Indeed, previously published guidelines for the use of deception include an authorization process in which the use of misleading tactics is disclosed during informed consent.⁴¹ In order to promote informed consent in sham-controlled trials, we recommend that subjects be clearly informed about the placebo arm and that they will have a 50% or less chance of receiving a fake procedure for the purpose of a scientifically valid evaluation of the effectiveness of the real procedure. The arthroscopic knee surgery trial conducted by Moseley et al.⁸ used a similar consent process in which subjects were required to write out a statement referring to the possibility that they might receive a placebo procedure. As an additional safeguard to ensure full understanding of both randomization to placebo and the use of active deception to mask the placebo (or treatment), it may be desirable for a clinician independent of the research team to obtain informed consent. Existing studies have shown no difference in expectation effects between subjects consenting under authorized deception and unauthorized deception, though effects may exist and vary by trial type and could be addressed with further study.⁴² Further guidelines require that the misleading is necessary for the scientific aims of the study, risks and the use of sham controls are specified and debriefing is made at the end of the trials.⁴³

In cases in which potential patient-subjects may lose decisionmaking capacity, we recommend that participation in sham-controlled trials should be limited to those who are capable of giving consent.

Ethical Framework for Justifying the Use of Placebo Procedures and Surgery

Placebo-controlled trials for neurological procedures should be justified using the following benchmarks formulated for sham surgery: 1) the research design addresses a valuable, clinically relevant question, 2) the placebo control is methodologically necessary and no sufficiently rigorous alternative design exists to address the study hypothesis, 3) the risk of the placebo is minimized, 4) the risk of the placebo arm does not exceed a threshold of acceptable research risk, 5) the risk of the placebo in the absence of prospect of direct benefit is justified by the clinical knowledge to be gained, and 6) the deception used to blind the placebo arm is disclosed to and authorized by participants.¹⁰ Differences in ethical considerations of trial design for varying neurological conditions may stem from variations in: 1) evidence for a placebo effect and necessity of a sham control if blinding may be maintained without sham surgery, 2) ability to create expectation while minimizing invasiveness of the placebo, 3) level of risk in the sham surgery, 4) ability to crossover to the active intervention without additional procedures, and 5) level of active deception involved in the sham administration.

Conclusion

Placebo-controlled trials are regarded as the gold standard in testing the efficacy of novel medical interventions, including drugs, devices, procedures, and surgery. In neurological procedures, subjective outcome measures and evidence for the neuromodulatory effects of expectation and conditioning necessitate a control arm that both protects an interpretation of results from evaluator and reporter bias and accounts for improvements due to the placebo effect. For these reasons, many researchers find sham surgical controls an ethically acceptable design when minimally or moderately invasive, such as in the use of burr holes to the skull with no pial penetration in trials of cell transplantation for Parkinson’s disease.^12,13 Invasive interventions that penetrate the dura pose a risk of infection and hemorrhage and are not acceptable when there is no prospect of direct benefit. However, crossover designs in which surgical implantations of a device may be tested using an on-or-off state increase the threshold of acceptable risk because all patient-subjects receive the prospect of direct benefit. A cross-over design may exceed the threshold of acceptable risk if the participants in the placebo arm are exposed to significant and continual risk before crossover to the active intervention. Critically emergent conditions should not be subject to a sham control. In many cases, the level of risk in sham surgery must be balanced against the value of subjecting invasive procedures to rigorous, blinded evaluation. Risky procedures introduced into clinical practice despite being ineffective or costly may cause a comparable or greater harm to patients. We recommend the framework and special considerations described above in the ethical justification of sham surgery in trials for neurological procedures.