Walter William Offen

Multiple Co-primary Endpoints: Medical and Statistical Solutions: A Report from the Multiple Endpoints Expert Team of the Pharmaceutical Research and Manufacturers of America

There are quite a few disorders for which regulatory agencies have required a treatment to demonstrate a statistically significant effect on multiple endpoints, each at the one-sided 2.5% level, before accepting the treatments efficacy for the disorders. Depending on the correlation among the endpoints, this requirement could lead to a substantial reduction in the studys power to conclude the efficacy of a treatment. To investigate the prevalence of this requirement and propose possible solutions, a multiple-disciplinary Multiple Endpoints Expert Team sponsored by Pharmaceutical Research and Manufacturers of America was formed in November 2003. The team recognized early that many researchers were not fully aware of the implications of requiring multiple co-primary endpoints. The team proposes possible solutions from both the medical and the statistical perspectives. The optimal solution is to reduce the number of multiple co-primary endpoints. If after careful considerations, multiple co-primary endpoints remain a scientific requirement, the team proposes statistical solutions and encourages that regulatory agencies be receptive to approaches that adopt modest upward adjustments of the nominal significance levels for testing individual endpoints. Finally, the team hopes that this report will draw more attention to the problem of multiple co-primary endpoints and stimulate further research.

Efficacy of xanomeline in Alzheimer disease: cognitive improvement measured using the Computerized Neuropsychological Test Battery (CNTB).

The cognitive efficacy of the M1-selective muscarinic agonist xanomeline in mild-to-moderate Alzheimer disease (AD) was measured using the Computerized Neuropsychological Test Battery (CNTB) and the Alzheimers Disease Assessment Scale-Cognitive Subscale (ADAS-cog) in this 17-center, double-blind, placebo-controlled study. Three hundred forty-three patients were randomly assigned to receive 25, 50, or 75 mg xanomeline tartrate or placebo three times daily (t.i.d.) for 24 weeks, followed by placebo for 4 weeks in a single-blind washout phase. Cognitive function was assessed at randomization and after 4, 8, 12, 24, and 28 weeks. Three hundred nineteen patients were included in an intent-to-treat (ITT) analysis; 209 completers had evaluable data at week 24. ITT analysis showed a significant (p < or = 0.05) dose-response trend and a significant (p < or = 0.05) between-group comparison favoring 75 mg t.i.d. over placebo for the CNTB summary score but not for the ADAS-cog. In the completer analysis, however, the ADAS-cog showed a significant (p < or = 0.05) dose-response trend and between-group comparison, whereas the CNTB Summary Score did not. The ADAS-cog was less sensitive to treatment effects in mildly impaired patients (ADAS-cog < 21) than in moderately impaired patients (ADAS-cog > or = 21), whereas the CNTB was sensitive in the entire study population (mean ADAS-cog = 22.5+/-9.6). Significant (p < or = 0.05) beneficial treatment effects were seen in measures of simple reaction time and delayed verbal recall, which are included in the CNTB but not in the ADAS-cog. During the single-blind placebo washout period, the ADAS-cog score of the placebo group worsened dramatically (change of 2.63 points; p < or = 0.001), whereas the CNTB score remained stable (change of 1.04 points; p=0.694). Thus, the CNTB appears to be more objective than the ADAS-cog.

Intravenous nizatidine kinetics and acid suppression

The effectiveness of intravenous nizatidine in suppressing gastric acid secretion was evaluated by different methods of inducing secretion in two single‐blind studies. In study 1, seven subjects were given single, 20‐min intravenous infusions of nizatidine (6.25, 25, 75, 150, or 250 mg) before modified sham feeding (MSF). Gastric acid suppression after nizatidine was contrasted with that after placebo and MSF. All doses of nizatidine reduced secretion; the 150‐ and 250‐mg doses of nizatidine suppressed secretion for at least 2.5 hr. In study 2, eight subjects received one 5‐min intravenous infusion of placebo, cimetidine (300 mg), or nizatidine (25, 50, 100, or 250 mg) weekly for 6 wk. Secretion was induced by infusing pentagastrin (2 µg/kg/hr) 45 min before the study drug was dosed and for 3.5 hr thereafter. Again, all doses of nizatidine reduced gastric acid, chiefly by decreasing volume. Nizatidine induced a clear dose‐response effect; nizatidine (100 mg) and cimetidine (300 mg) had approximately equal suppressive effects. Nizatidine (100 mg) and cimetidine (300 mg) reduced acid output by 62% and 63% and reduced volume of secretion by 48% and 51% over the 3.5‐hr period. Gastric acid suppression and plasma nizatidine concentrations were directly related. Nizatidine kinetics were linear and proportional to dose. The t½ was 1.3 hr (range 0.7 to 2.1 hr), the volume of distribution was 1.2 ± 0.5 l/kg, and clearance was 0.6 ± 0.2 l/kg/hr. Laboratory abnormalities and side effects were minor in both studies.

Nizatidine versus placebo in gastroesophageal reflux disease

In a randomized, multicenter trial, nizatidine 150 mg or 300 mg, or placebo, was administered twice daily for six weeks to 515 patients with gastroesophageal reflux disease (GERD). Gelusil antacid tablets were taken as needed for pain. Significantly superior rates of endoscopically proven complete healing (normal-appearing mucosa) versus placebo occurred after three weeks with nizatidine 150 mg, and after six weeks with nizatidine 300 mg. Six-week healing rates were 38.5% for nizatidine 300 mg, 41.1% for nizatidine 150 mg, and 25.8% for placebo. The nizatidine 150 mg treatment group had significantly greater improvement in daytime and nighttime heartburn severity after one day of therapy versus placebo. Twice-daily administration of nizatidine 150 mg or 300 mg provides prompt relief from the major symptom of GERD, heartburn, and complete healing of esophagitis is seen in many patients.

A new classification approach for comparing two active treatments when there is no prior projection on which one is better.

We developed a new classification approach in this paper to compare two active treatments. This approach is especially useful when there is no prior judgment on which treatment is better and the traditional hypothesis testing approach is thus not applicable. Our method classifies all the possible outcomes into categories and draws conclusions on the difference in the outcome measurement between two treatment arms according to the location of the confidence interval for the treatment difference in the response variable. This method controls the misclassification rate regardless of the true difference in the response between the two treatment arms. The method was applied to a diabetes clinical trial.

Discussion of “Some Controversial Multiple Testing Problems in Regulatory Applications”

We want to commend the authors on addressing two challenging issues that are of fundamental importance to the developers of innovative pharmaceutical products. These two issues are related, but are of a different nature. Simply stated, the first issue deals with the level of evidence needed to include additional endpoints in a product label. The second issue deals with the level of evidence and thus the required sample size when a new treatment needs to demonstrate effect on multiple coprimary endpoints in order to be acceptable. In what follows, we will comment on the points raised by the authors in these two areas. In addition, we will offer some additional insight on the multiple coprimary endpoints issue. Most of our comments related to Section 3 are outside the scope of the original paper. We include them in this commentary with the hope that these comments will help stimulate additional interest and research on this topic.

Nizatidine versus placebo in active benign gastric ulcer disease: An eight‐week, multicenter, randomized, double‐blind comparison

Study objective: To determine if 150 mg nizatidine twice daily or 300 mg nizatidine at bedtime are similarly effective and to compare each dose with placebo in healing benign gastric ulcers and relieving peptic ulcer symptoms.

Relationship between steady‐state plasma nizatidine concentrations and inhibition of basal and stimulated gastric acid secretion

Steady‐state plasma nizatidine concentrations were related in a linear fashion to nizatidine infusion rate. Infusion rates of 2.5, 10, and 20 mg/hr resulted in mean plasma nizatidine concentrations of 69, 247, and 575 ng/ml. Basal acid secretion was inhibited by 50% and 90% at mean plasma nizatidine concentrations of 60 and 430 ng/ml. Protein‐stimulated acid secretion was inhibited by 50% and 90% at mean plasma nizatidine concentrations of 75 and 490 ng/ml. The mean pH of basal gastric secretions was 1.6 during placebo infusion and 4.6 when the mean plasma nizatidine concentration was 575 ng/ml.

Healing and recurrence of active duodenal ulcer with nizatidine

Nizatidine, a new H2‐receptor antagonist for treatment of duodenal ulcer disease, was evaluated in a unique two‐phase, placebo‐controlled, randomized, double‐blind, multicenter clinical trial. Patients received either 150 mg nizatidine twice daily or placebo for 4 weeks (phase I). If ulcer healing did not occur during phase I, patients were randomly reallocated to receive either 150 mg nizatidine twice daily or placebo for an additional 4 weeks (phase II). Patients with a healed ulcer continued on the same therapy. All patients were endoscoped at week 8. Healing rates at week 2 were 93 of 265 (35%) nizatidine‐treated patients and 55 of 260 (21%) placebo‐treated patients (p < 0.001); at week 4, healing rates were 198 of 259 (76%) nizatidine‐treated patients and 95 of 243 (39%) placebo‐treated patients (p < 0.001). In phase II, ulcer healing occurred in 46 of 86 (53%) nizatidine‐treated patients and in 23 of 90 (26%) placebo‐treated patients (p = 0.002). In patients who had a healed ulcer at previous endoscopies, 18 of 178 (10%) nizatidine‐treated patients and 10 of 81 (12%) placebo‐treated patients had a recurrence of duodenal ulcer. Smokers who had histories of previous ulcers were more likely to have an early recurrence.

A simple stopping rule for declaring treatment ineffectiveness in clinical trials

We consider the problem of stopping a clinical trial before its scheduled termination due to the apparent ineffectiveness of the experimental therapy, as compared with a control. We propose a simple-to-implement, intuitive decision rule based on the unadjusted attained significance levels from any appropriate statistical test. The proposed procedure may be used at any time during the study as an aid to help determine whether the study of an experimental treatment should be terminated early with the conclusion of treatment ineffectiveness. Much of the power of the usual fixed-sample test is retained while maintaining the nominal test size.