21 January 2008. Can an FDA-approved arthritis drug double as a therapy for Alzheimer disease? Claiming it did in a patient, a case report published in the January 9 Journal of Neuroinflammation generated a wave of media attention that splashed across television, the BBC, newspapers and out to numerous online sources and blogs. The coverage ran the gamut from breathless approval to stern criticism. Some headlines announced “Extraordinary Breakthrough-Alzheimer’s Symptoms Reversed in Minutes,” other outlets reprinted the press release, CNN’s Sanjay Gupta balanced promise and caveats, and a blogger posted his open letter to the study authors in which he charged “This, gentlemen, appears to be a sham.” So what is a reader to make of it? The Alzforum looked behind the initial media accounts to report the story in greater depth. Scientists in the field were intrigued about the approach of inhibiting certain cytokines to treat inflammation in AD, yet also expressed reservations about the way these studies were conducted.
The present paper was submitted to J. Neuroinflammation, an open-access journal published by BioMedCentral, on November 29, 2007. It describes the response of an 81-year-old physician who met diagnostic criteria for probable AD and came to the principal investigator’s medical practice in October of 2007. The patient received weekly injections of etanercept (marketed under the name Enbrel®), a biologic drug that binds and neutralizes tumor necrosis factor α (TNFα). The paper’s authors are Edward Tobinick, a dermatologist and internist who runs a private medical group in Los Angeles, and Hyman Gross, a neurologist in private practice in Santa Monica, California. The paper is freely available and therefore not summarized in detail here. In brief, the authors report that this patient visibly and measurably improved within minutes of receiving drug injections into the back of the neck, and that his improvement lasted throughout a 7-week assessment period during which he received one injection every week for the first 5 weeks. Tobinick and Gross hypothesize that the man’s cognitive improvement might be due to etanercept’s ability to quench the action of excess TNFα at the brain’s synapses.
In an accompanying editorial, Sue Griffin of the University of Arkansas School of Medical Sciences in Little Rock speculates that this drug might be able to counteract “gliotransmission” in the human brain, i.e., the effects on synaptic activity of TNFα released by activated glial cells. In an unusually personal editorial, Griffin recounts her journey to Tobinick’s practice to observe the treatment herself, writing: “…—I was amazed!” Griffin is editor-in-chief of J. Neuroinflammation; the news bureau at her institution issued the press release. Griffin argues for larger, double-blind follow-up studies to explore both etanercept and the delivery method used by Tobinick and Gross. The present paper followed publication in 2006 of an open-label pilot study of 15 patients. This study did not assess patients’ cognition so soon after they had received the injection, but did claim to provide proof-of-concept that 6 months of weekly etanercept administration improved the performance of patients on the MMSE, ADAS-Cog, and SIB assessment tools (Tobinick et al., 2006). In this paper, one patient, an 80-year-old study participant, was reported to have died suddenly and without obvious cause during the last week of the trial.
When asked about this study, seven independent academic investigators who were not involved in it all said that the scientific rationale behind targeting TNFα to dampen inflammation in AD and some other neurological conditions is valid. Case reports have led to the development of new drugs before, and the independent scientists all agreed that formal clinical research should be pursued. At the same time, they expressed unease about the way in which these studies were performed. “The use of etanercept in this fashion is an interesting idea; however, I believe these studies should be done in randomized controlled trials, so the field can know how to interpret the data,” said David Holtzman, who chairs the neurology department at Washington University in St. Louis, Missouri. Some criticized the way the study was publicized. “This is exciting but very preliminary information,” said Cynthia Lemere, associate professor of Neurology at Brigham and Women’s Hospital in Boston. “The appropriate way to pursue it at this stage is to apply to FDA and NIH for support to run a rigorous trial, not to promote it to the general public.”
These comments encompass a range of specific concerns scientists raised. For one, none of the studies reported so far included controls. Controls could include injecting placebo, i.e., sterile saline, in the exact same way that etanercept is injected. This would test whether any response is due to the drug or to something else about the procedure. One reason why this is important is that patients with probable Alzheimer disease have been noted to have strong placebo effects, as occurs in other diseases. “I never fail to be dismayed at the magnitude of placebo effects in Alzheimer disease; they are remarkably large,” said Ben Barres, a professor of neurobiology and leader in glial cell research at Stanford University in Palo Alto, California (see comment below.)
Another reason is that symptoms in Alzheimer disease patients are known to vary widely over time. The long-term trajectory of this progressive disease is invariably downward, but within that slow overall decline, patients show frequent ups and downs in cognitive function. Patients change not only because of their disease but also in response to things around them—stress, the environment, changes in daily routines, Holtzman explained. This zigzag pattern of overall decline renders it nearly impossible to interpret observations of any intervention done without controls, especially in a single case.
If follow-up studies substantiate that the observed effect is due to etanercept, fundamental scientific questions about where and how the treatment works remain to be addressed. Toward that goal, other controls could include peripheral administration of etanercept. Side-by-side comparisons of delivery modes and uptake/metabolism in plasma and cerebrospinal fluid (CSF) would begin to address issues around whether the injection used in this study is the best way to deliver the drug. Scientists interviewed questioned whether etanercept reaches the brain and acts there, or instead reduces TNFα in a more indirect fashion from outside the central nervous system (CNS, see comment by Greg Cole below). According to his papers, Tobinick injected etanercept between the spinous processes of the C6 and 7 vertebrae in the neck in such a way that the needle does not penetrate the CSF space itself but deposits the drug around one of the membranes covering the spinal cord. There are numerous veins in this region, which are proposed to allow some transport into the brain. To enhance this transport, the patients were laid on their backs, with their feet higher than their heads, for 5 minutes after the injection. Tobinick calls this type of injection “perispinal extrathecal” administration. This administration involves a shallower injection than the better known and widely used lumbar puncture farther down the spine, which some physicians consider safer in part because it is less likely to rupture veins.
There is no formal published data yet to show that the perispinal injection does indeed cause brain uptake of TNFα. Scientists in the laboratory of Pat McGeer took up this question and have unpublished results. (A noted researcher at the University of British Columbia, Vancouver, McGeer was one of the pioneers who established the concept of inflammation in neurologic disease and advanced it as a therapeutic target. To date, most controlled trials of anti-inflammatory drugs have been negative, but research is ongoing.) The Canadian scientists imaged the brains of rats injected with labeled etanercept. They were able to detect etanercept in the CSF but not the brain parenchyma, where neurons and synapses are located (see comment below).
In addition to determining whether this special injection is effective at delivering etanercept to the brain’s synapses, one might study whether its inherent risk is necessary, said Murali Doraiswamy, chief of biological psychiatry Duke University Medical Center in Durham, North Carolina. Published evidence from a large randomized, controlled trial of people with psoriasis suggests that etanercept given peripherally (it usually is injected under the skin) can ameliorate depression, mood, and other brain-related symptoms (Tyring et al., 2006). This leads some researchers to wonder whether some small fraction of etanercept—a large molecule generally considered unable to cross the blood-brain barrier—might somehow get across, or act on brain TNFα levels indirectly from the outside. Monkey studies with labeled etanercept would be suitable to answer these questions in preparation for formal clinical trials, said Lemere.
A related question concerns whether people with AD truly have elevated TNFα levels in their CSF. Several studies have reported such findings while some others have not, and some researchers noted having measured elevated CSF TNFα in some AD patients but not in others. As yet, scientists have not come to a general consensus that CSF TNFα is elevated in AD, as they have, for example, for tau or oxidative products called isoprostanes. The relationship between peripheral (i.e., plasma) and central (i.e., CSF) levels of TNFα in AD needs to be sorted out (e.g., see Cole comment below).
Edward Tobinick is a board-certified dermatologist and internist, who has been active in laser hair removal prior to developing an interest in the use of etanercept for CNS indications. The hair removal clinic at which Tobinick is medical director, the Institute of Laser Medicine, is in the same building as the Institute for Neurological Research, the medical group that sees AD patients for etanercept treatment; both are in suites 205-210 at 100 UCLA Medical Plaza. Etanercept is approved for rheumatoid arthritis, juvenile rheumatoid arthritis, ankylosing spondylitis, and the skin disease psoriasis. Tobinick has previously disclosed that he owns stock in Amgen, the owner of etanercept. He holds numerous patents on delivery methods of etanercept for neurologic conditions (e.g., United States Patent 7,214,658).
As of January 2008, the Medical Board of California lists Tobinick as being on probation. Details of the disciplinary action were not available in time for this news article, but the Board’s April 2007 newsletter [.pdf] on page 20 published a paragraph indicating that Tobinick was charged with unprofessional conduct relating to advertising an unproven treatment for back pain. Tobinick published uncontrolled open-label treatment results suggesting that etanercept might serve to ameliorate disc-related pain (Tobinick et al., 2004). Researchers at Johns Hopkins University, Baltimore, Maryland, tested these observations with a double-blind, placebo-controlled dose-response trial. Last summer they reported that etanercept injection was not effective for this indication (Cohen et al., 2007).
A related area of uncertainty pertains to the distinctions between clinical practice and research. Physicians routinely prescribe drugs off-label, and charge for the drug, as part of their medical practice. This is a common practice tolerated by the FDA. In contrast, when the intent is to study a drug for a new indication, i.e., to collect and publish data, it is treated as research, and as such becomes subject to approval by an institutional review board (IRB). IRBs often mandate that the physicians submit their study protocols to the FDA for review. In an interview, an FDA official was unable to say whether an investigational new drug (IND) exemption had been submitted for the study of perispinal etanercept in AD, as IND issuances are confidential. The official pointed out that typically under IRB and IND rules, patients cannot be charged for the treatment given in a research study. Attempts to clarify how the interrelated issues of IRB rules, payment, and treatment versus research were handled in these studies were unsuccessful. According to the published papers, the patient described in this month’s case report was treated as part of clinical practice. The 2006 paper of the pilot study notes that a central institutional review board, a form of IRB often used for multicenter trials, approved the study, and that the families of the study subjects supplied the study drug.
Most media reports presented the work as a “UCLA study.” Tobinick lists a UCLA e-mail address on the manuscript, and the private medical group where he works full-time is located in a building at UCLA’s medical plaza. According to the UCLA media office, Tobinick is a voluntary assistant clinical professor with UCLA’s Division of Dermatology. The office stated that his etanercept study did not go through the UCLA IRB. Greg Cole, associate director of the UCLA Alzheimer Disease Center (ADC), said that the center had nothing to do with Tobinick’s etanercept work. (Cole is known for his research on inflammation and cytokine action in AD. He studies, among other things, curcumin, a spice that has anti-TNFα activity and is being tested in a 12-month Phase 2 clinical trial at UCLA.) The J. Neuroinflammation paper identifies coauthor Gross as being at the Department of Neurology, University of Southern California, School of Medicine. That university’s website does not list Gross as a faculty physician in the neurology department. John Weiner, media representative at USC Health Sciences, said that Gross is not on USC’s regular, paid faculty but has a voluntary appointment. Weiner added that such voluntary associations are similar to “adjunct” appointments at other academic institutions.
Amgen, the maker of etanercept, distanced itself from the study. A statement regarding Tobinick and Gross’s study on the company website noted: “This study was not supported nor endorsed by Amgen. While Amgen and others have long recognized the potential for TNF inhibitors to have an effect on neurological conditions, we have carefully examined this study and believe that at this time there is insufficient scientific data to support the use of a TNF inhibitor as a means of treating Alzheimer's disease.” Reached by telephone, Amgen spokeswoman Sonia Fiorenza confirmed that this statement referred not only to the present paper, but that Amgen had studied etanercept’s potential for treating AD internally, as well. “As common sense would suggest, we have looked very closely at this,” Fiorenza said. According to Amgen’s 2006 Annual Report, etanercept (Enbrel®) sales that year totaled $2.9 billion.
On January 15, this reporter e-mailed Tobinick questions relating to long-term cognitive effects, the precise nature of the injection, IRB approval, patient payment for treatment, total number of patients who responded vs. who did not, and other issues for an author Q&A in the format the Alzforum frequently publishes. Dr. Tobinick requested a phone conversation instead, but during the subsequent call declined to take questions. In a later e-mail, Tobinick noted that he is in discussions with academic medical centers.—Gabrielle Strobel.
Tobinick EL, Gross H. Rapid cognitive improvement in Alzheimer's disease following perispinal etanercept administration. J Neuroinflammation. 2008 Jan 9;5(1):2. Abstract
Griffin WS. Perispinal etanercept: Potential as an Alzheimer therapeutic. J Neuroinflammation. 2008 Jan 10;5(1):3. Abstract