Statistical aspects of the TNK-S2B trial of tenecteplase versus alteplase in acute ischemic stroke: an efficient, dose-adaptive, seamless phase II/III design.
Bottom Line:
Phase III incorporated two co-primary hypotheses, allowing for a treatment effect at either end of the trichotomized Rankin scale.Inflation from the dose selection was more than offset by the one-half continuity correction in the test statistics.Inflation from repeated interim analyses was more than offset by the reduction from the clinical stopping rules for futility at the first interim analysis.
View Article:
PubMed Central - PubMed
Affiliation: Department of Biostatistics, Mailman School of Public Health, Columbia University, New York, NY 10025, USA.
ABSTRACT
Show MeSH
Background: TNK-S2B, an innovative, randomized, seamless phase II/III trial of tenecteplase versus rt-PA for acute ischemic stroke, terminated for slow enrollment before regulatory approval of use of phase II patients in phase III. Purpose: (1) To review the trial design and comprehensive type I error rate simulations and (2) to discuss issues raised during regulatory review, to facilitate future approval of similar designs. Methods: In phase II, an early (24-h) outcome and adaptive sequential procedure selected one of three tenecteplase doses for phase III comparison with rt-PA. Decision rules comparing this dose to rt-PA would cause stopping for futility at phase II end, or continuation to phase III. Phase III incorporated two co-primary hypotheses, allowing for a treatment effect at either end of the trichotomized Rankin scale. Assuming no early termination, four interim analyses and one final analysis of 1908 patients provided an experiment-wise type I error rate of <0.05. Results: Over 1,000 distribution scenarios, each involving 40,000 replications, the maximum type I error in phase III was 0.038. Inflation from the dose selection was more than offset by the one-half continuity correction in the test statistics. Inflation from repeated interim analyses was more than offset by the reduction from the clinical stopping rules for futility at the first interim analysis. Limitations: Design complexity and evolving regulatory requirements lengthened the review process. Conclusions: (1) The design was innovative and efficient. Per protocol, type I error was well controlled for the co-primary phase III hypothesis tests, and experiment-wise. (2a) Time must be allowed for communications with regulatory reviewers from first design stages. (2b) Adequate type I error control must be demonstrated. (2c) Greater clarity is needed on (i) whether this includes demonstration of type I error control if the protocol is violated and (ii) whether simulations of type I error control are acceptable. (2d) Regulatory agency concerns that protocols for futility stopping may not be followed may be allayed by submitting interim analysis results to them as these analyses occur. Related in: MedlinePlus |
Related In:
Results -
Collection
License getmorefigures.php?uid=PMC3198122&req=5
Mentions: The simulation of both rapid-response outcomes and 3-month Rankin outcomes requires specification of the joint distribution of two trichotomous random variables, i.e., a 3 × 3 table of joint probabilities, for each treatment arm. We call such a set of four 3 × 3 tables a distribution scheme; 1000 different distribution schemes were used in the simulation studies. Generation of these distribution schemes is described in detail in the Appendix. In the simulation studies, our goal was to produce a set of distributions that would cover a portion of the parameter space that was of direct clinical interest to the TNK-S2B trial, and that the portion covered was sufficiently broad as to represent type I error rates accurately over the entire theoretical parameter space. Let the three-category rapid response be denoted by X, taking values of 0 for ICH, 1 for neither MNI nor ICH, and 2 for MNI. Let the trichotomized 3-month Rankin scale for the clinical outcome be denoted by Y, taking values of 0 for poor outcome, 1 for neither poor nor good outcome, and 2 for good outcome. Let the doses of tenecteplase be labeled A, B, and C, corresponding to 0.1, 0.25, and 0.40 mg/kg, respectively, and let dose D refer to rt-PA. Let T denote any of these four treatment arms. It is most convenient to determine the joint distribution of (X, Y) by first specifying P[X/T], and then specifying P[Y/X, T]. Once these three-component vectors of probabilities are determined, random realizations of the pair (X, Y) can be produced by generating a trinomial response X following P[X/T], and then by generating another trinomial response Y following P[Y/X, T]. In the simulation studies, 10 different marginal conditional distributions for X given T were selected at random in a manner described in the Appendix; see Figure 2(a) for visualizing these distributions. For each one of these, 10 different marginal distributions for Y given T (identical for each T under the hypotheses) were generated randomly (Figure 2(b)); and for each of the 10 × 10 = 100 pairs of marginal distributions for X and Y, 10 different conditional distributions for Y given X and T were generated randomly, subject to the marginalization constraint that the weighted average of P[Y/X, T] using weights P[X/T] agree with the given distribution P[Y/T], together with a clinical monotonicity constraint that P[Y = poor/X, T] is the greatest when X = ICH and least when X = MNI, and similarly, P[Y = good/X, T] is the least when X = ICH and the greatest when X = MNI. Thus, 10 × 10 × 10 = 1000 different distribution schemes were employed for the simulations. Figure 2(c) contains a graphical representation of the complete list of distributions, using barycentric coordinates, displaying a uniform distribution of T over the lower portion of the triangle. Figure 2(d) shows a visual effect of the clinical monotonicity constraint. As stated above, we simulated 40,000 trial replications for each distribution scheme. Each replication generated random samples of up to 150 pairs of outcomes (X, Y) for each member of a quadruplet or matched set for the selection stage, and additional (X, Y) pairs to make up the total sample size of 1908 clinical outcomes for the phase III trial. Figure 2 |
View Article: PubMed Central - PubMed
Affiliation: Department of Biostatistics, Mailman School of Public Health, Columbia University, New York, NY 10025, USA.
Background: TNK-S2B, an innovative, randomized, seamless phase II/III trial of tenecteplase versus rt-PA for acute ischemic stroke, terminated for slow enrollment before regulatory approval of use of phase II patients in phase III.
Purpose: (1) To review the trial design and comprehensive type I error rate simulations and (2) to discuss issues raised during regulatory review, to facilitate future approval of similar designs.
Methods: In phase II, an early (24-h) outcome and adaptive sequential procedure selected one of three tenecteplase doses for phase III comparison with rt-PA. Decision rules comparing this dose to rt-PA would cause stopping for futility at phase II end, or continuation to phase III. Phase III incorporated two co-primary hypotheses, allowing for a treatment effect at either end of the trichotomized Rankin scale. Assuming no early termination, four interim analyses and one final analysis of 1908 patients provided an experiment-wise type I error rate of <0.05.
Results: Over 1,000 distribution scenarios, each involving 40,000 replications, the maximum type I error in phase III was 0.038. Inflation from the dose selection was more than offset by the one-half continuity correction in the test statistics. Inflation from repeated interim analyses was more than offset by the reduction from the clinical stopping rules for futility at the first interim analysis.
Limitations: Design complexity and evolving regulatory requirements lengthened the review process.
Conclusions: (1) The design was innovative and efficient. Per protocol, type I error was well controlled for the co-primary phase III hypothesis tests, and experiment-wise. (2a) Time must be allowed for communications with regulatory reviewers from first design stages. (2b) Adequate type I error control must be demonstrated. (2c) Greater clarity is needed on (i) whether this includes demonstration of type I error control if the protocol is violated and (ii) whether simulations of type I error control are acceptable. (2d) Regulatory agency concerns that protocols for futility stopping may not be followed may be allayed by submitting interim analysis results to them as these analyses occur.