The Enduring Vitality of Implausible Claims for Neurostimulation and Psychosurgical Therapies Free

This journal has recently published a position statement on the future for neuromodulation in psychiatric disorders [1]. Just as politicians sometimes have to rise above principles, the unnamed authors on The Psychiatric Neurosurgery Committee and Board of Directors of the ASSFN had to rise above the facts. We searched in vain to find such a committee listed on the ASSFN and WSSFN websites. If absence of author information was an error, several months have elapsed without a correction. The WSSFN president recently noted in their newsletter that such a committee (again, no names) was preparing new guidelines for the performance of psychosurgery in human subjects - an alarming announcement given the discussion that follows.

The present writer participated in years of physician consultations, protocol development, site selection, regulatory submissions, trial conduct, device troubleshooting, adverse-event adjudication, and acted as a member of a the data safety monitoring board for the Reclaim trial (Medtronic, Inc., Minneapolis, Minn., USA). We also closely followed the program that became the Broaden trial via presentations, publications, and communications with investigators (Advanced Neuromodulation Systems/St. Jude Medical, St. Paul, Minn., USA). Substantial contradictions exist between the ASSFN position statement and information that was available to the individuals involved in both trials.

The statement refers to ‘discontinuation of two pivotal trials evaluating the efficacy of deep brain stimulation (DBS) for the treatment of depression'. The Reclaim feasibility trial conducted at five US centers successfully ran to conclusion some years ago; it was not pivotal; it was not discontinued; it did not fail. The results and analyses published online in 2014 revealed no antidepressant efficacy attributable to DBS therapy [2]. To the best of our knowledge, the Broaden trial concluded after a futility analysis in 2013. That analysis apparently was performed upon data acquired after a 2011 expansion of the initial feasibility cohort. Results of the analysis determined - presumably, before the full pivotal cohort had been enrolled - that additional patient enrollments and/or additional efficacy analyses were futile. Therefore, the trial had to stop. Each trial gathered sufficient data over a sufficient duration of follow-up to perform valid, prospectively defined efficacy analyses, as agreed upon by the investigators [https://clinicaltrials.gov/ct2/show/NCT00837486?term=%22reclaim%22+AND+%22depression%22&rank=1, http://www.sjm.com/broaden, http://blogs.scientificamerican.com/cross-check/much-hyped-brain-implant-treatment-for-depression-suffers-setback/].

Although one may debate whether the Broaden trial reached the pivotal N number threshold, the circumstances of how each trial ended do not comport with the core message of the ASSFN statement that: ‘An unintended consequence of the discontinuation of these two trials is the interpretation by the public, industry, regulatory bodies, and health care professionals, that trial discontinuation represents a failure of the therapy, rather than just an unsuccessful trial.' To the contrary, without new rigorously acquired controlled data, it is no longer tenable to claim that DBS at either target effectively treats depression. Despite empirical disconfirmation of efficacy at two different DBS targets, in two different trials, by independent sets of investigators, under two competing industry-sponsored protocols, the ASSFN authors remain ‘… committed to untangling these complex disorders and remain steadfast in our conviction that neuromodulation will ultimately be an important therapeutic modality'. Rational interpretations are dismissed as ‘unintended consequence(s)' or, more pejoratively, as ‘nihilistic'. Now that both trials have generated results that were the opposite of what the authors anticipated, these individuals have distanced themselves from responsibility for and influence on key elements of the trial designs. As one example, the use of the ad hoc diagnosis ‘treatment resistant depression' (not coded within DSM-IV) reflects decisions made by the investigators and sponsors to reproduce clinical features of the patients enrolled in earlier single-center pilot studies. Both studies also followed published guidelines required by the FDA to approve new indications for DBS - which the FDA classifies as a class III (high-risk) medical device [3].

The absence of critical insight in light of new facts - neglecting the possibility that their hypotheses and conclusions may have been mistaken - suggests that the authors' absence of equipoise remains in play. The position statement's authors genuinely expected to find antidepressant benefits from DBS. Afterwards, they blamed disconfirming results on ordinary clinical trial-related factors (well accounted for in the N numbers and statistical plans) and on euphemistically called ‘dilution of therapeutic effect in these relatively small cohorts'. A better-informed appraisal might conclude that some of these same authors' initial efficacy claims based on smaller, unblinded, uncontrolled single-center cohorts were not substantiated. Their call for ‘future trials with more targeted outcomes and the potential for adaptive trial design' sounds like sore losers whining for a do-over. In contrast, unbiased scientists view setbacks as enlightening - not threatening.

No corner of medical inquiry is entirely free of bias, as explored in depth by Chapman et al. [4], Ioannidis et al. [5,6], and others [7]. Therapies for psychiatric disorders and chronic noncancer pain - both of which occupy the intersection of neurological- and psychosocial-behavioral phenomena - appear particularly susceptible to structural biases that remain resistant to rational correction [8]. Such bias phenomena include: misplaced rationality, bias blind spots (illusory immunity to bias), ostrich effects, anchoring bias (focalism), reactive devaluation, false equivalence, illusions of statistical power, experience and truth, the illusion of explanatory depth, pictorial superiority, misattribution (post hoc ergo propter hoc), illusory superiority bias (Dunning-Kruger or Lake Wobegon effects), and irrational escalation [8,9,10,11,12,13].

Operation of cognitive biases has led ineffective practices to become entrenched orthodoxies within stereotactic and functional surgery and neuromodulation. Such practices commenced with efficacy claims for novel applications of implanted devices. Hypotheses regarding purported mechanisms of action, supported by neuroimaging or electrophysiological tests, accompanied (or followed soon after) the initial reports [14,15,16,17]. Although a full discussion of diagnostic confusion and malleability are beyond the scope of this letter, the diagnoses under study may be contentious or nonexistent. Thereafter, investigators and industry sponsors conduct larger phase II studies with the conviction that validation, regulatory approval, and insurance coverage will ensue. The most common outcome is best expressed by the pharma industry cliché that, ‘everything fails in phase-II'. Plainly stated, efficacy vanishes in industry-sponsored studies that require FDA approval and that employ adequate blinding and control groups. However, committed investigators seldom revise their positions. Many continue to advocate and perform device implant procedures for ineffective indications. A few examples that have not stood the test of time, formal investigation, or independent analysis include DBS for pain, cluster headache, obesity, Tourette syndrome, vegetative state, and memory loss; motor cortex stimulation for pain or poststroke paralysis; spinal cord stimulation for angina, limb ischemia, complex regional pain syndrome, or any other indication; occipital nerve stimulation for migraine; transcutaneous nerve stimulation for facial pain; intraspinal (opioid) drug delivery for noncancer pain, and delivery of growth factors, autologous or fetal tissue, or viral vectors for Parkinson's disease. Robust debates and advances in knowledge have characterized the aftermath of cellular and biological trials for Parkinson's disease. In contrast, delayed or nonpublication of trial results, and noncoherent reviews and editorials have made corrections in the pain and psychobehavioral domains of the neuromodulation literature scarce [18,19,20,21,22,23,24].

Patient expressions of desperation also contribute to a well-meaning surgeon's inclination (in spite of disconfirming evidence) to perform marginal procedures of last resort - especially if she perceives the operation or device-based therapy to be relatively safe. Neither rationale holds up to scrutiny for current neuromodulation therapies for psychiatric disorders or noncancer pain. Patient desperation can be infectious, and surgeons have no immunity. Errors of irrational escalation (associated with escalating risks) ensue when an ineffective operation or device implant fails to provide relief for a potentially erroneous (or nonexistent) diagnosis in a desperate patient willing to undergo anything that will help her. Performance of destructive operations in failed DBS for psychiatric indications represents one salient example. With respect to risk, and independent of questionable efficacy, relative safety and minimal invasiveness remain common rationales for the performance of device-based neuromodulation therapies. In the case of depression, investigators wrote that among 20 DBS clinical trial subjects, ‘No significant adverse events were reported during this follow-up, although two patients died by suicide during depressive relapses' [25]. Blaming clinical trial subjects for their own fatal suicidality events is to be regarded with skepticism. In any event, when benefits are nil, calculations of risk versus benefit become a mathematical impossibility with zero in the denominator.

In light of having been proven wrong, the authors have initiated ‘a dialogue with the leadership of the neuromodulation industry'. Engagement with and support from industry arises from company executives having to maintain profitable relationships with influential physician customers. Apart from suspect efficacy claims and how poorly things have turned out for psychiatric indications to-date, industry participants learned that outside of the clinical trial setting, no mechanism exists to satisfy clinically unimproved patients' demands for device programming and troubleshooting. There was no therapy there, and there was no business model there. In contrast, high-implant-volume physicians comprise the crucial market for already approved device-based therapies. We have observed that competing executives feel obliged to remain engaged with key customers. More recently, military funding recently has emerged for device-based human cognitive and psychosurgical investigations from the US Department of Defense Advanced Research Projects Agency (DARPA) under the project name Systems-Based Neurotechnology for Emerging Therapies (SUBNETS) [http://www.nytimes.com/2013/10/25/science/pentagon-agency-to-spend-70-million-on-brain-research.html]. It remains to be seen whether a military-industrial-academic partnership will yield valuable technology breakthroughs (like digital radiography, which arose from espionage satellite technology) or whether biased decision-making will lead to less salubrious results.

The US-FDA has authority to regulate the food, human and veterinary drug, and medical device industries. The FDA does not regulate physician medical practices, and state medical societies, hospitals, and medical or surgical departments have little authority to mandate or proscribe specific therapies. Civil litigation, malpractice claims and/or class action lawsuits occasionally censure physicians or compensate individuals for an injury or loss that already has occurred. Given the absence of practice oversight, committed physicians, including biased or mistaken ones, experience few incentives to alter their habitual practice patterns based on rational analyses or evidence [8].

To paraphrase one investigator (in confidence), ‘DBS for psychosurgical indications is the new neurostimulation for chronic pain.' When we analyzed how implantable device-based pain therapies had zero efficacy, disagreements over the quality of evidence were of limited consequence in public health terms [23,24]. However, as data have emerged that undermine the notion that spinal cord stimulation and intrathecal opioid drug delivery are safe [ [26,27], http://www.wsj.com/articles/SB10001424052702304512504579493843647492538] a subset of implanting physicians have collaborated with industry to generate publications that misrepresent efficacy and underestimate (or deflect attention from) relative risks [28,29]. Readers may not be aware that with the exception of ziconotide (Prialt, Jazz Pharmaceuticals, Dublin, Ireland), and despite pre-1976 legacy status and other historical factors, no FDA-approved neurostimulation or intrathecal drug therapy for noncancer pain has successfully undergone formal clinical trial programs to establish safety and efficacy. In this setting, an explicit aim of recent physician-industry collaborative trials, practice guidelines, and safety-related reviews has been to defend and/or expand reimbursement and physician (i.e., industry customer) practice patterns. Salient examples of trials with straw-man control groups for spinal cord stimulation include the EVIDENCE and PROMISE trials [[30], https://clinicaltrials.gov/ct2/show/NCT01697358?term=NCT01697358&rank=1]. Industry-sponsored activities to defend and promote arguably hazardous and ineffective intrathecal drug delivery practices include a decade-long series of polyanalgesic consensus documents [28,29,30,31,32,33,34]. Despite manufacturers and FDA having information to the contrary, physician-industry collaborative publications that relied upon incomplete or selective data have provided unrealistic assurances of safety and substantially underestimated risk - specifically, in discussions of paralysis related to implantation of surgical (paddle) spinal cord stimulation leads and death during maintenance of intraspinal opioid therapy, respectively [31,32,33,34,35,36,37].

The history of psychosurgery has not reflected well on neurosurgery, revisionist histories notwithstanding [38,39,40,41,42,43,44,45,46,47]. As is the case for noncancer pain, current psychiatric nosology is an unstable shambles, subject to core disputes over pathologization of the normal, and subject to bootstrapping from one hypothetical treatment for one disorder to another [48]. Dissenting clinical or experimental publications are unusual [49]. However, faculty, department chairs, and neurosurgical societies - if they are willing to invest the effort - have a platform from which to advocate rational inquiry and respect for evidence, and to teach the next generation how avoid patient harm that arises from errors, illusions, and fallacies. Without such efforts, the current misplaced enthusiasm for psychosurgery is perched to become something more pernicious than an edgy experiment. Specifically, it's downright creepy to find a practitioner of punitive judicially ordered psychosurgery in a one-party totalitarian state (People's Republic of China) on the ASSFN board. Ironically, health insurance companies, Medicare, Medicaid, Workman's Compensation and other payers may now represent the last line of defense to protect patients from ineffective and potentially dangerous device-based neuromodulation therapies. Under such circumstances, it's fair to note that the interlocking directorates and boards of the Stereotactic and Neuromodulation Societies appear little interested in scientific inquiry and rigorous testing of hypotheses - yet deeply committed to the promotion of parochial surgical and/or medical practices, regardless of efficacy, safety, or lack of either. But it is not too late to change.