Testimony of Peter Lurie, M.D., M.P.H., and Elizabeth Barbehenn, Ph.D.
Before FDA Peripheral and Central Nervous System Drugs
Advisory Committee Meeting
Public Citizen’s Health Research Group opposes the granting of the new indication (the dementia associated with Parkinson’s Disease) to the cholinesterase inhibitor rivastigmine (Exelon). Minimum criteria for approval of a drug should include: a. the disease to be treated is clearly defined and can be clinically evaluated as distinct; b. the drug has a clinically meaningful benefit; c. the benefit has been demonstrated in well-designed and -conducted studies; and d. the finding has been replicated. Not one of these criteria has been met.
There is No Clear Evidence that the Dementia Associated with Parkinson’s Disease is a Distinct Clinical Entity
In order to justify this second indication for rivastigmine (in addition to the treatment of Alzheimer’s Disease, for which it is already approved), there should be clear evidence that the dementia associated with Parkinson’s Disease is a specific disease or a clinical syndrome. Yet no such evidence exists. The EXPRESS study (the only randomized, controlled trial put forth by the sponsor) is said to have relied upon the diagnostic criteria for the dementia associated with Parkinson’s Disease from the American Psychiatric Association’s Diagnostic and Statistical Manual IV (DSM-IV). But DSM IV does not even provide a basis for making this diagnosis. With masterful circularity, it states: “The essential feature of Dementia Due to Parkinson’s Disease is the presence of dementia that is judged to be of direct pathophysiological consequence of Parkinson’s disease.” A recent Practice Parameter from the American Academyof Neurology puts it bluntly: “DSM-IV criteria for dementia have not been validated in [Parkinson’s Disease].”
DSM-IV goes on to describe some aspects of this dementia (none unique to Parkinson’s Disease), but, as noted by the FDA Medical Officer, patients in the EXPRESS trial “were enrolled based on their having dementia, but without the more distinctive cognitive deficits described in DSM-IV.” Indeed, the Medical Officer even questioned “how distinct the patients in the EXPRESS trial were from those enrolled in the pre-approval Alzheimer’s Disease trials.” It is noteworthy that the clinical courses in the placebo groups in the EXPRESS trial and in rivastigmine’s trials for Alzheimer’s Disease were similar, further calling into question the uniqueness of this alleged clinical entity.
Finally, the EXPRESS trial did not even require baseline MRIs or CT Scans to rule out other potential causes of dementia, even though both Alzheimer’s Disease and the dementia associated with Parkinson’s Disease are in part diagnoses of exclusion and such scans are standard practice in Alzheimer’s Disease studies.
The sponsor will point to supposedly characteristic pathological findings (e.g. Lewy bodies, degeneration of the medial substantia nigra), but there is tremendous overlap between the pathological findings in Alzheimer’s Disease and diffuse Lewy body disease, and those in the dementia associated with Parkinson’s Disease, and there is no study that shows that the clinical features said to be more common in Parkinson’s dementia (as opposed to Alzheimer’s dementia) actually correlate with the pathological findings. Moreover, the positive predictive values for both the clinical and pathological findings have not been defined.
The Medical Officer concludes: “Unless the efficacy of rivastigmine … is judged to be mechanistically distinct from its established effects in Alzheimer’s Disease, the grant of a separate claim for the treatment of mild to moderate dementia associated with Parkinson’s Disease may not be justified.”
A recent issue of PLOS Medicine defines disease-mongering as “the selling of sickness that widens the boundaries of illness and grows the markets for those who sell and deliver treatments.” That is an apt description of this attempt to find another disease for rivastigmine to treat.
The Effects Demonstrated in the EXPRESS Trial are Modest
Depending on which statistical technique is used, the benefit of rivastigmine in the EXPRESS trial was three points on the 70-point ADAS-cog scale, commensurate with the improvement one might observe over six months in a patient with Alzheimer’s Disease. This is a decidedly modest benefit when the limitations of the study are taken into account.
First, the ADAS-cog is, according to the FDA, “not particularly useful for evaluating executive function,” even though this one of the most prominent deficits in the dementia associated with Parkinson’s Disease. Second, the sponsor has provided no evidence that the ADAS-cog has content validity in this condition. Third, a survey asked Canadian geriatricians and neurologists what minimum change in the Mini-Mental Status Examination they viewed as clinically significant in Alzheimer’s Disease patients. They identified a value approximately twice what the FDA accepts. Finally, even when the ADAS-cog showed a statistical benefit with another Alzheimer’s Disease drug, patients and caregivers did not observe such benefits.
There is an aphorism in statistics that “a difference, to be a difference, must make a difference.” Given all the limitations described here, it is not clear that the statistical findings observed in this study have much clinical relevance.
The Drop-Out Rate is High and May Explain Rivastigmine’s Observed Efficacy
One problem with the study design that deserves particular attention, in part because it afflicts all clinical trials of cholinesterase inhibitors for dementia,, is that of attrition bias. The loss to follow-up rate in the rivastigmine arm was considerably higher than in the placebo arm (27% vs. 18%), a difference largely attributable to the adverse effect profile of rivastigmine. Seventeen percent of rivastigmine patients dropped out due to adverse effects, compared with 8% among placebo patients, and rivastigmine-treated patients were also more likely to withdraw informed consent (6% vs. 1%). If those suffering adverse effects or withdrawing consent were also less likely to have derived benefit from rivastigmine, the disproportionate loss of rivastigmine-treated patients likely creates a bias in favor of rivastigmine.
It is true that there were efforts to correct for the missing data using two techniques: 1. intention to treat with last observation carried forward (i.e., if the patient did not continue on medication, the last measurement on medication was considered his or her final outcome) and 2. intention to treat with retrieved drop-outs (i.e., if the patient did not continue on medication but continued to be evaluated, these data were used instead of their last evaluation on study medication). Neither of these adjustments meaningfully altered the findings.
However, these two adjustment methods are among the least sophisticated currently available. Other techniques include regression-based imputation (regression models based on observed values generate the missing values), proper multiple imputation (regression models create more than one imputed data set and thus provide variability within and between imputations) andhot deck imputation (missing values are based on observations from patients with similar characteristics).,,, The FDA and the sponsor should explain why these techniques were not utilized. Attrition bias is particularly important when the observed effects of treatment are as modest as is the case here.
The Sponsor has Failed to Replicate its Findings
To date, five drugs have been approved for the treatment of Alzheimer’s Disease. In each case, at least two randomized, placebo-controlled trials were submitted to support the application. As the FDA asserts, “… if dementia associated with Parkinson’s Disease is indeed a condition that is distinct from Alzheimer’s Disease, it would seem appropriate to require that the results of the study by replicated.” Such replication is particularly important when the impact of treatment is modest and subject to bias, as is the case here. There is no reason to stray from the FDA’s current requirement for replication.
As noted above, the adverse effects of rivastigmine were substantial and frequently led to study withdrawal. Excesses in incidence of 18% for nausea, 15% for vomiting, 6% for tremor, 5% for dizziness and 3% for both diarrhea and anorexia were attributable to rivastigmine. Of note, several of these could be considered worsening of Parkinson’s Disease itself.
In sum, we are left with a single trial of a product of debatable efficacy for a condition that may not exist as a unique entity. The quest for this new indication is itself mired in self-contradiction and should leave this committee with no choice but to reject the drug. For, if rivastigmine is similarly effective in similarly designed trials (similar entry criteria, same assessment scales, same interpretive problems) for both Alzheimer’s Dementia and the dementia associated with Parkinson’s Disease, one might well conclude that the disease processes are not clinically distinguishable and that a separate indication for Parkinson’s is not justified. On the other hand, if a separate indication is warranted, why has the sponsor not submitted two clinical trials, as was done for all the Alzheimer’s Disease drugs? As rivastigmine would be the first drug to be approved for the dementia associated with Parkinson’s Disease, the answers to these two questions are quite simple and are to be found not in science, but in marketing: product differentiation and market segmentation.
 Miyasaki JM, Shannon K, Voon V, et al. Practice Parameter: evaluation and treatment of depression, psychosis, and dementia in Parkinson disease (an evidence-based review): report of the Quality Standards Subcommittee of the American Academy of Neurology. Neurology 2006;66:996-1002.
 Moynihan R, Henry D. The Fight against Disease Mongering: Generating Knowledge for Action. PLoS Medicine2006;3:e191.
 Stern RG, Mohs RC, Davidson M, et al. A longitudinal study of Alzheimer’s Disease: measurement, rate, and predictors of cognitive deterioration. American Journal of Psychiatry 1994;151:390-6.
 Burback D, Molnar FJ, St. John P, Man-Son-Hing M. Key methodological features of randomized controlled trials of Alzheimer’s Disease therapy: minimal clinically important difference, sample size and trial duration. Dementia and Geriatric Cognitive Disorders 1999;10:534-40.
 Rockwood K, Fay S, Song X, MacKnight C, Gorman M. Attainment of treatment goals by people with Alzheimer’s disease receiving galantamine: a randomized controlled trial. Canadian Medical Association Journal 2006;174:1099-105.
 Hills R, Gray R, Stowe R, Bentham P. Drop-out bias undermines findings of improved functionality with cholinesterase inhibitors. Neurobiology of Aging 2002;23(suppl. 1):S89 (abstract 33).
 Kaduszkiewicz H, Zimmermann T, Beck-Bornholdt H-P, van den Bussche H. Cholinesterase inhibitors for patients with Alzheimer’s disease: systematic review of randomised clinical trials. British Medical Journal 2005;331:321-7.
 Press DZ. Parkinson’s Disease dementia – a first step? New England Journal of Medicine 2004;351:2547-9.
 Health Economics Resource Center. Finding and Using Healthcare Data. U.S. Department of Veterans Affairs. Available at: http://www.herc.research.med.va.gov/resources/faq_i10.asp (downloaded May 17, 2006)
 Shih WJ. Problems in dealing with missing data and informative censoring in clinical trials. Current Controlled Trials in Cardiovascular Medicine 2002;3:4.
 Rubin DM. Multiple Imputation for Nonresponse in Surveys. New York:Wiley, 1987.
 Schafer JL. Analysis of Incomplete Multivariate Data. New York: Chapmanand Hall, 1997.