The Food and Drug Administration will ask its Peripheral and Central Nervous System Drugs Advisory Committee to weigh in Oct. 17 on whether a clinical trial with a randomized-start design, if appropriately designed and conducted, can detect a drug’s disease-modifying effect in patients with Parkinson’s disease.
Teva employed a randomized-start design in the Azilect (rasagiline mesylate) ADAGIO trial, its basis for seeking an indication as therapy to delay clinical disease progression in patients with Parkinson’s. The drug was approved in 2006 to treat the signs and symptoms of the disease.
FDA reviewers are skeptical that the trial demonstrated the efficacy of Azilect for the new indication, according to agency briefing documents.
But advisory committee support for the trial design could reassure industry and patient advocates that there is a way to demonstrate a medicine’s ability to delay disease progression, for which there currently is no biological marker.
Ralph D’Agostino, Ph.D., Boston University, noted that the design has "great promise" in a Sept. 24, 2009, editorial that accompanied the ADAGIO publication in the New England Journal of Medicine. Dr. D’Agostino is one of four biostatisticians FDA has appointed as a temporary voting member on the advisory committee to help address the statistical issues raised by agency reviewers.
The Trial Design
The randomized-start design is conducted in two steps. Patients are randomized to treatment or placebo for an appropriate duration, and then patients receiving placebo are switched to the active drug and those on the active drug continue treatment.
The interpretive principle is that a drug has a disease-modification effect if those in the treatment arm during the first phase continue to show greater treatment benefit than those who only initiate treatment in the second phase. The basis for this is that "early (longer) treatment provides a persistent benefit that cannot be achieved if treatment is delayed, presumably due to an effect on the underlying pathology that cannot be ‘re-captured,’" the FDA explains.
The briefing documents note that dropouts between the first and second phase of the trial can introduce complexity into analyses and interpretation of the data. Also, those entering the second phase "will no longer be randomized groups, making statistical comparisons treacherous."
ADAGIO was a multicenter, double-blind study in patients with early Parkinson’s. The treatment arms received either 1 mg/day or 2 mg/day of Azilect. There were two placebo arms in the first phase. In the second, the placebo arms were switched to either the 1-mg or the 2-mg dose. Both phases lasted 36 weeks.
Effects were measured based on changes in Unified Parkinson’s Disease Rating Scale scores. Analyses were conducted to determine: superiority of treatment compared with placebo at the end of 36 weeks; superiority at the end of 72 weeks for early start therapy compared with delayed start treatment; and noninferiority of early start to delayed start in the rate of change in the UPDRS score between weeks 48 and 72.
The last measurement was to ensure that there continued to be an "absolute" difference between the two groups at the end of the study and the levels of their improvement were not converging.
Per protocol, the analyses were to be based on all data from all groups combined. Teva, however, calculated the change from baseline to week 72 for each dose contrast, using data only from that dose.
These analyses found that patients who received 2 mg/day in the first phase received no statistically significantly greater benefit compared to those who initiated therapy in the second phase.
The company’s analysis found statistically significant differences in benefit between the two groups receiving 1 mg/day of Azilect. Analysis per protocol produces a P value of .0506, which, the agency points out, would not be considered significant.
Even if the company’s analysis were accepted for the 1-mg dose, the FDA says, there is no obvious biological explanation for why the 2-mg dose also was not beneficial at week 72. That finding "calls into question any ostensibly positive findings that might be the basis for a disease modifying claim for rasagiline."
Agency Issues With Analyses and Results
FDA reviewers identified a number of issues of concern with the analyses and results from ADAGIO and TEMPO, and will ask the panel to discuss them. TEMPO was the trial used to support Azilect’s original approval and is being offered as supporting evidence for the current supplemental application.
These issues are: sponsor-conducted analyses that differed from those specified in the protocol; a differential response in men and women and the baseline differences in early and delayed women starters in ADAGIO; potentially significant baseline differences in UPDRS scores between early-start and delayed-start patients in datasets used for analyses at 72 weeks, and potential biases introduced because these include nonrandomized groups; the nonlinearity of slopes that presumably are related to varying early effects of treatment; and re-analyses of slopes without early data suggest parallel slopes in the first phase for drug and placebo.