Wilcock GK, Black SE, Hendrix SB, Zavitz KH, Swabb EA, Laughlin MA, .
Efficacy and safety of tarenflurbil in mild to moderate Alzheimer's disease: a randomised phase II trial.
Lancet Neurol. 2008 Jun;7(6):483-93.
Please login to recommend the paper.
To make a comment you must login or register.
The publication of the tarenflurbil Phase 2 study for mild to moderate AD is welcome and instructive, as this is one of the first putatively disease-modifying compounds to have undergone peer review and publication. Unlike the Phase 2 studies for Alzhemed and LY450139, this study emphasized clinical measures rather than biological markers for demonstration of potential efficacy. If it had proceeded as planned, there would have been sufficient power to explore three primary outcome measures (ADAS-cog, CDR-SB and ADCS-ADL) across two doses of drug. Although the authors planned in advance to focus on the high (800 mg BID) dose in the milder (MMSE 20-26) patients if there was an interaction between baseline score and treatment effect, this pre-specification did not save them from the interpretive problems produced by an underpowered, subgroup analysis.
It is very difficult to determine exactly how many mild and moderate subjects are involved in the study at each time point, and to determine what factors might have led to lost subjects (and introduced bias into the results) at any given time point. Journal space limitations and the complexity of this study are likely reasons that the information is not clearly spelled out. But the exercise of trying to keep track of the mild and moderate subjects as they make their way through 12 months of double-blind and an additional 12 months of open-label helps to highlight the difficulty of doing such a study.
Although 210 patients were randomized, 207 actually took drug, and 189 are included in the ITT population (subjects from two sites were excluded due to irregularities). Of the 189 in the ITT population, 130 were evidently mild, and only 62 of these mild subjects (i.e., 48 percent) completed the one-year double-blind study. This low completion rate allows for the possibility that the more rapidly progressing patients dropped out. This could amplify the treatment signal, or decrease it, depending on the distribution of rapid progressors across the drug versus placebo groups. Since we do not know the rates of decline for placebo patients by severity, it is difficult to hypothesize about the effects of this substantial number of dropouts on the findings, and we would only be speculating anyway. A differential rate of change in placebo patients in the low-dose group is alluded to when referring to specific outcome measures, and if placebo decline varied by outcome measure it likely varied by other subgroups as well. Only 44 percent of the low-dose group and 44 percent of the placebo patients completed the 12-month double-blind study.
The 12-month open-label extension is further plagued by incompleteness. Only patients from Canada were approved to go forward (seemingly 107 of the 160 who completed the double-blind), and only 80 percent of the Canadian subjects agreed to do so. Of the 86 subjects who entered the open-label, 14 were lost due to irregularities and not included in the ITT analysis population. The number who completed the open-label extension is not indicated. The slopes analysis is particularly problematic for this phase of the study.
The experience represented by this publication highlights how difficult it is to try to do everything in one Phase 2 study: identify at least one effective dose, demonstrate a dose response, define the target population, demonstrate a signal of efficacy, and gather evidence of “disease modification.” To really understand tarenflurbil, and to design the best possible test for it in Phase 3, it would have been nice to follow this study with several other small, Phase 2 studies, such as one study using 800 mg BID as well as a higher dose (if justified by the preclinical and PK data), and, of course, a replication study to show a differential effect in mild versus moderate patients.
The data generated by the current study would suggest that tarenflurbil will not improve patients over baseline, but that separation between the treated and placebo groups might be evident by six or nine months, allowing shorter studies. Unfortunately, a ticking time clock on patient life and the necessity for long Phase 2 studies for compounds that are not expected to improve patients over their baselines have forced more difficult decisions regarding how and when to move on to Phase 3. Fortunately for patients, the Phase 3 studies of tarenflurbil will start to become available soon, because of the decision Myriad made to move forward. The Phase 3 studies will have to reverse the negative finding on ADAS-cog seen in Phase 2, and will have to replicate a signal on the global and/or ADL. If they do not, we will be left hoping for clearer results than we have from Phase 2. If the studies are successful, we will be left wishing that moderate patients had been included or definitively shown to be unable to respond to the treatment.
We would like to thank Dr. Doody for her balanced commentary on our paper about the Phase 2 trial of Tarenflurbil. It was a successful study at this level in that it prepared the way for the Phase 3 studies that have followed.
Number of Patients COMPLETED out of Number Treated (Safety Population)Within Each Phase
400 mg BID 800 mg BID Placebo
0-12 m 31/40 (78%) 41/50 (82%) 41/50 (82%)
12-24 m 17/17 (100%) 22/26 (85%) 15/20 (75%)
0-12 m 19/31(61%) 16/20 (80%) 12/16 (75%)
12-24 m 4/10 (40%) 4/6 (67%) 5/6 (83%)
Gordon Wilcock on behalf of the authors.
Space limitations imposed by the journal did indeed lead to constraints that made it difficult to present some of the data as clearly as we would have wished, and this was compounded by the complexity of the trial design. We apologize for the difficulty in determining exactly how many mild and moderate subjects are involved in the study at each time point and would like to clarify that the dropout rate was approximately 20 percent in the first 12 months of the study, not 50 percent. The table above may help clarify this.