The Impact of Major Events on Ongoing Noninferiority Trials, With Application to COVID-19.

Abstract-The COVID-19 pandemic has impacted ongoing clinical trials. We consider particular impacts on noninferiority clinical trials, which aim to show that an investigational treatment is not markedly worse than an existing active control with known benefit. Because interpretation of noninferiority trials requires cross-trial validation involving untestable assumptions, it is vital that they be run to very high standards. The COVID-19 pandemic has introduced an unexpected impact on clinical trials, with subjects possibly missing treatment or assessments due to unforeseen intercurrent events. The resulting data must be carefully considered to ensure proper statistical inference. Missing data can often, but not always, be considered missing completely at random (MCAR). We discuss ways to ensure validity of the analyses through study conduct and data analysis, with focus on the hypothetical strategy for constructing estimands. We assess various analytic strategies of analyzing longitudinal binary data with dropouts where outcomes may be MCAR or missing at random (MAR). Simulations show that certain multiple imputation strategies control the Type I error rate and provide additional power over analysis of observed data when data are MCAR or MAR, with weaker assumptions about the missing data mechanism.

A randomized, placebo-controlled trial of cenicriviroc for treatment of nonalcoholic steatohepatitis with fibrosis.

The aim of this study was to evaluate cenicriviroc (CVC), a dual antagonist of CC chemokine receptor types 2 and 5, for treatment of nonalcoholic steatohepatitis (NASH) with liver fibrosis (LF). A randomized, double-blind, multinational phase 2b study enrolled subjects with NASH, a nonalcoholic fatty liver disease activity score (NAS) ≥4, and LF (stages 1-3, NASH Clinical Research Network) at 81 clinical sites. Subjects (N = 289) were randomly assigned CVC 150 mg or placebo. Primary outcome was ≥2-point improvement in NAS and no worsening of fibrosis at year 1. Key secondary outcomes were: resolution of steatohepatitis (SH) and no worsening of fibrosis; improvement in fibrosis by ≥1 stage and no worsening of SH. Biomarkers of inflammation and adverse events were assessed. Full study recruitment was achieved. The primary endpoint of NAS improvement in the intent-to-treat population and resolution of SH was achieved in a similar proportion of subjects on CVC (N = 145) and placebo (N = 144; 16% vs. 19%, P = 0.52 and 8% vs. 6%, P = 0.49, respectively). However, the fibrosis endpoint was met in significantly more subjects on CVC than placebo (20% vs. 10%; P = 0.02). Treatment benefits were greater in those with higher disease activity and fibrosis stage at baseline. Biomarkers of systemic inflammation were reduced with CVC. Safety and tolerability of CVC were comparable to placebo. CONCLUSION: After 1 year of CVC treatment, twice as many subjects achieved improvement in fibrosis and no worsening of SH compared with placebo. Given the urgent need to develop antifibrotic therapies in NASH, these findings warrant phase 3 evaluation. (Hepatology 2018;67:1754-1767).

Multiple comparisons in non-inferiority trials: Reaction to recent regulatory guidance on multiple endpoints in clinical trials.

We consider analysis of active control, non-inferiority clinical trials with multiple primary endpoints for assessing efficacy of an investigational treatment. Many of the issues with multiple endpoints for non-inferiority trials are similar to issues for superiority trials, but there are important differences. Because non-inferiority trials typically make decisions with confidence interval bounds instead of p-values, care must be taken in adjusting for multiple hypotheses. Composite endpoints are more difficult to interpret in non-inferiority trials due to difficulties in indirectly comparing the investigational treatment to placebo on each component. Otherwise many of the methods used in superiority trials (including sequential testing, graphical procedures and gatekeeping procedures) can be applied to non-inferiority trials with a little additional care. We focus on the differences between non-inferiority and superiority trials to provide guidance on application of recent regulatory guidance to non-inferiority trials.

Andexanet Alfa for Acute Major Bleeding Associated with Factor Xa Inhibitors.

BACKGROUND: Andexanet alfa (andexanet) is a recombinant modified human factor Xa decoy protein that has been shown to reverse the inhibition of factor Xa in healthy volunteers. METHODS: In this multicenter, prospective, open-label, single-group study, we evaluated 67 patients who had acute major bleeding within 18 hours after the administration of a factor Xa inhibitor. The patients all received a bolus of andexanet followed by a 2-hour infusion of the drug. Patients were evaluated for changes in measures of anti-factor Xa activity and were assessed for clinical hemostatic efficacy during a 12-hour period. All the patients were subsequently followed for 30 days. The efficacy population of 47 patients had a baseline value for anti-factor Xa activity of at least 75 ng per milliliter (or ≥0.5 IU per milliliter for those receiving enoxaparin) and had confirmed bleeding severity at adjudication. RESULTS: The mean age of the patients was 77 years; most of the patients had substantial cardiovascular disease. Bleeding was predominantly gastrointestinal or intracranial. The mean (±SD) time from emergency department presentation to the administration of the andexanet bolus was 4.8±1.8 hours. After the bolus administration, the median anti-factor Xa activity decreased by 89% (95% confidence interval [CI], 58 to 94) from baseline among patients receiving rivaroxaban and by 93% (95% CI, 87 to 94) among patients receiving apixaban. These levels remained similar during the 2-hour infusion. Four hours after the end of the infusion, there was a relative decrease from baseline of 39% in the measure of anti-factor Xa activity among patients receiving rivaroxaban and of 30% among those receiving apixaban. Twelve hours after the andexanet infusion, clinical hemostasis was adjudicated as excellent or good in 37 of 47 patients in the efficacy analysis (79%; 95% CI, 64 to 89). Thrombotic events occurred in 12 of 67 patients (18%) during the 30-day follow-up. CONCLUSIONS: On the basis of a descriptive preliminary analysis, an initial bolus and subsequent 2-hour infusion of andexanet substantially reduced anti-factor Xa activity in patients with acute major bleeding associated with factor Xa inhibitors, with effective hemostasis occurring in 79%. (Funded by Portola Pharmaceuticals; ANNEXA-4 ClinicalTrials.gov number, NCT02329327 .).

Extended Thromboprophylaxis with Betrixaban in Acutely Ill Medical Patients.

BACKGROUND: Patients with acute medical illnesses are at prolonged risk for venous thrombosis. However, the appropriate duration of thromboprophylaxis remains unknown. METHODS: Patients who were hospitalized for acute medical illnesses were randomly assigned to receive subcutaneous enoxaparin (at a dose of 40 mg once daily) for 10±4 days plus oral betrixaban placebo for 35 to 42 days or subcutaneous enoxaparin placebo for 10±4 days plus oral betrixaban (at a dose of 80 mg once daily) for 35 to 42 days. We performed sequential analyses in three prespecified, progressively inclusive cohorts: patients with an elevated d-dimer level (cohort 1), patients with an elevated d-dimer level or an age of at least 75 years (cohort 2), and all the enrolled patients (overall population cohort). The statistical analysis plan specified that if the between-group difference in any analysis in this sequence was not significant, the other analyses would be considered exploratory. The primary efficacy outcome was a composite of asymptomatic proximal deep-vein thrombosis and symptomatic venous thromboembolism. The principal safety outcome was major bleeding. RESULTS: A total of 7513 patients underwent randomization. In cohort 1, the primary efficacy outcome occurred in 6.9% of patients receiving betrixaban and 8.5% receiving enoxaparin (relative risk in the betrixaban group, 0.81; 95% confidence interval [CI], 0.65 to 1.00; P=0.054). The rates were 5.6% and 7.1%, respectively (relative risk, 0.80; 95% CI, 0.66 to 0.98; P=0.03) in cohort 2 and 5.3% and 7.0% (relative risk, 0.76; 95% CI, 0.63 to 0.92; P=0.006) in the overall population. (The last two analyses were considered to be exploratory owing to the result in cohort 1.) In the overall population, major bleeding occurred in 0.7% of the betrixaban group and 0.6% of the enoxaparin group (relative risk, 1.19; 95% CI, 0.67 to 2.12; P=0.55). CONCLUSIONS: Among acutely ill medical patients with an elevated d-dimer level, there was no significant difference between extended-duration betrixaban and a standard regimen of enoxaparin in the prespecified primary efficacy outcome. However, prespecified exploratory analyses provided evidence suggesting a benefit for betrixaban in the two larger cohorts. (Funded by Portola Pharmaceuticals; APEX ClinicalTrials.gov number, NCT01583218.).

Andexanet Alfa for the Reversal of Factor Xa Inhibitor Activity.

BACKGROUND: Bleeding is a complication of treatment with factor Xa inhibitors, but there are no specific agents for the reversal of the effects of these drugs. Andexanet is designed to reverse the anticoagulant effects of factor Xa inhibitors. METHODS: Healthy older volunteers were given 5 mg of apixaban twice daily or 20 mg of rivaroxaban daily. For each factor Xa inhibitor, a two-part randomized placebo-controlled study was conducted to evaluate andexanet administered as a bolus or as a bolus plus a 2-hour infusion. The primary outcome was the mean percent change in anti-factor Xa activity, which is a measure of factor Xa inhibition by the anticoagulant. RESULTS: Among the apixaban-treated participants, anti-factor Xa activity was reduced by 94% among those who received an andexanet bolus (24 participants), as compared with 21% among those who received placebo (9 participants) (P<0.001), and unbound apixaban concentration was reduced by 9.3 ng per milliliter versus 1.9 ng per milliliter (P<0.001); thrombin generation was fully restored in 100% versus 11% of the participants (P<0.001) within 2 to 5 minutes. Among the rivaroxaban-treated participants, anti-factor Xa activity was reduced by 92% among those who received an andexanet bolus (27 participants), as compared with 18% among those who received placebo (14 participants) (P<0.001), and unbound rivaroxaban concentration was reduced by 23.4 ng per milliliter versus 4.2 ng per milliliter (P<0.001); thrombin generation was fully restored in 96% versus 7% of the participants (P<0.001). These effects were sustained when andexanet was administered as a bolus plus an infusion. In a subgroup of participants, transient increases in levels of d-dimer and prothrombin fragments 1 and 2 were observed, which resolved within 24 to 72 hours. No serious adverse or thrombotic events were reported. CONCLUSIONS: Andexanet reversed the anticoagulant activity of apixaban and rivaroxaban in older healthy participants within minutes after administration and for the duration of infusion, without evidence of clinical toxic effects. (Funded by Portola Pharmaceuticals and others; ANNEXA-A and ANNEXA-R ClinicalTrials.gov numbers, NCT02207725 and NCT02220725.).

Recent Statistical Contributions to Medical Device Development.

Clinical trials in the development of new medical device products are in many ways analogous to clinical trials in the development of new drug or biologic products. However, the differences are important and not always intuitive to a statistician with only experience supporting development of drug and biologic products. In this paper we discuss some of the interesting differences with focus on the statistical innovation that is coming out of the medical device area. We discuss examples of the differences in clinical trial design and effects of these differences on clinical development programs.

Selection of hypothesis weights and ordering when testing multiple hypotheses in clinical trials.

This article discusses the problem of selecting free parameters of multiple testing procedures in confirmatory Phase III clinical trials with multiple objectives, including hypothesis weights and hypothesis ordering. We identify classes of multiple testing procedures that provide different interpretations of these parameters. This includes basic single-step procedures (Bonferroni procedure) that employ fixed hypothesis weights, as well as more powerful stepwise procedures (Holm, fallback, and chain procedures) that reweight the hypotheses during the testing process. We examine the behavior of different classes of multiple testing procedures in problems with unequally weighted hypotheses and a priori ordered hypotheses and provide practical guidelines for the choice of hypothesis weights and hypothesis ordering. The concepts discussed in the article are illustrated using case studies based on clinical trials with multiple endpoints, multiple dose-placebo comparisons, and multiple patient populations.

Multiple testing procedures in two pivotal clinical trials for approval of a new pharmaceutical product.

We consider analysis of two identical pivotal trials with correlated multiple hypotheses evaluated by the fixed-sequence, weighted Holm, or fallback procedure. For approval, at least one hypothesis must be rejected in both studies. Various weights are considered for the fallback and weighted Holm procedure to provide separation in a single study. Evaluation of the procedures as closed tests distinguishes which has the highest power in different situations. No procedure is universally optimal. The best procedure depends on the power to reject the various hypotheses, something that is never known with certainty, and the goals of the analyses. If the most important goal is to demonstrate a difference on the first hypothesis or on all hypotheses, the fixed-sequence procedure performs best. However, the fixed sequence often has the highest chance of obtaining inconsistent results between the two independent studies, which makes it less appealing. The weighted Holm and fallback procedures are very similar, with various weighting schemes providing modest differentiation. The alpha exhaustive version of the fallback procedure often has higher power for some endpoints and lower power for other endpoints compared to the weighted Holm procedure, but the differences are rarely large.

A selective endothelin-receptor antagonist to reduce blood pressure in patients with treatment-resistant hypertension: a randomised, double-blind, placebo-controlled trial.

BACKGROUND: Hypertension cannot always be adequately controlled with available drugs. We investigated the blood-pressure-lowering effects of the new vasodilatory, selective endothelin type A antagonist, darusentan, in patients with treatment-resistant hypertension. METHODS: This randomised, double-blind study was undertaken in 117 sites in North and South America, Europe, New Zealand, and Australia. 379 patients with systolic blood pressure of 140 mm Hg or more (>/=130 mm Hg if patient had diabetes or chronic kidney disease) who were receiving at least three blood-pressure-lowering drugs, including a diuretic, at full or maximum tolerated doses were randomly assigned to 14 weeks' treatment with placebo (n=132) or darusentan 50 mg (n=81), 100 mg (n=81), or 300 mg (n=85) taken once daily. Randomisation was made centrally via an automated telephone system, and patients and all investigators were masked to treatment assignments. The primary endpoints were changes in sitting systolic and diastolic blood pressures. Analysis was by intention to treat. The study is registered with ClinicalTrials.gov, number NCT00330369. FINDINGS: All randomly assigned participants were analysed. The mean reductions in clinic systolic and diastolic blood pressures were 9/5 mm Hg (SD 14/8) with placebo, 17/10 mm Hg (15/9) with darusentan 50 mg, 18/10 mm Hg (16/9) with darusentan 100 mg, and 18/11 mm Hg (18/10) with darusentan 300 mg (p<0.0001 for all effects). The main adverse effects were related to fluid accumulation. Oedema or fluid retention occurred in 67 (27%) patients given darusentan compared with 19 (14%) given placebo. One patient in the placebo group died (sudden cardiac death), and five patients in the three darusentan dose groups combined had cardiac-related serious adverse events. INTERPRETATION: Darusentan provides additional reduction in blood pressure in patients who have not attained their treatment goals with three or more antihypertensive drugs. As with other vasodilatory drugs, fluid management with effective diuretic therapy might be needed. FUNDING: Gilead Sciences.

General multistage gatekeeping procedures.

A general multistage (stepwise) procedure is proposed for dealing with arbitrary gatekeeping problems including parallel and serial gatekeeping. The procedure is very simple to implement since it does not require the application of the closed testing principle and the consequent need to test all nonempty intersections of hypotheses. It is based on the idea of carrying forward the Type I error rate for any rejected hypotheses to test hypotheses in the next ordered family. This requires the use of a so-called separable multiple test procedure (MTP) in the earlier family. The Bonferroni MTP is separable, but other standard MTPs such as Holm, Hochberg, Fallback and Dunnett are not. Their truncated versions are proposed which are separable and more powerful than the Bonferroni MTP. The proposed procedure is illustrated by a clinical trial example.

A note on tree gatekeeping procedures in clinical trials.

Dmitrienko et al. (Statist. Med. 2007; 26:2465-2478) proposed a tree gatekeeping procedure for testing logically related hypotheses in hierarchically ordered families, which uses weighted Bonferroni tests for all intersection hypotheses in a closure method by Marcus et al. (Biometrika 1976; 63:655-660). An algorithm was given to assign weights to the hypotheses for every intersection. The purpose of this note is to show that any weight assignment algorithm that satisfies a set of sufficient conditions can be used in this procedure to guarantee gatekeeping and independence properties. The algorithm used in Dmitrienko et al. (Statist. Med. 2007; 26:2465-2478) may fail to meet one of the conditions, namely monotonicity of weights, which may cause it to violate the gatekeeping property. An example is given to illustrate this phenomenon. A modification of the algorithm is shown to rectify this problem.

Ambrisentan for the treatment of pulmonary arterial hypertension: results of the ambrisentan in pulmonary arterial hypertension, randomized, double-blind, placebo-controlled, multicenter, efficacy (ARIES) study 1 and 2.

BACKGROUND: Ambrisentan is a propanoic acid-based, A-selective endothelin receptor antagonist for the once-daily treatment of pulmonary arterial hypertension. METHODS AND RESULTS: Ambrisentan in Pulmonary Arterial Hypertension, Randomized, Double-Blind, Placebo-Controlled, Multicenter, Efficacy Study 1 and 2 (ARIES-1 and ARIES-2) were concurrent, double-blind, placebo-controlled studies that randomized 202 and 192 patients with pulmonary arterial hypertension, respectively, to placebo or ambrisentan (ARIES-1, 5 or 10 mg; ARIES-2, 2.5 or 5 mg) orally once daily for 12 weeks. The primary end point for each study was change in 6-minute walk distance from baseline to week 12. Clinical worsening, World Health Organization functional class, Short Form-36 Health Survey score, Borg dyspnea score, and B-type natriuretic peptide plasma concentrations also were assessed. In addition, a long-term extension study was performed. The 6-minute walk distance increased in all ambrisentan groups; mean placebo-corrected treatment effects were 31 m (P=0.008) and 51 m (P<0.001) in ARIES-1 for 5 and 10 mg ambrisentan, respectively, and 32 m (P=0.022) and 59 m (P<0.001) in ARIES-2 for 2.5 and 5 mg ambrisentan, respectively. Improvements in time to clinical worsening (ARIES-2), World Health Organization functional class (ARIES-1), Short Form-36 score (ARIES-2), Borg dyspnea score (both studies), and B-type natriuretic peptide (both studies) were observed. No patient treated with ambrisentan developed aminotransferase concentrations >3 times the upper limit of normal. In 280 patients completing 48 weeks of treatment with ambrisentan monotherapy, the improvement from baseline in 6-minute walk at 48 weeks was 39 m. CONCLUSIONS: Ambrisentan improves exercise capacity in patients with pulmonary arterial hypertension. Improvements were observed for several secondary end points in each of the studies, although statistical significance was more variable. Ambrisentan is well tolerated and is associated with a low risk of aminotransferase abnormalities.

Efficacy and safety of darusentan in patients with resistant hypertension: results from a randomized, double-blind, placebo-controlled dose-ranging study.

In this phase 2, randomized, double-blind, placebo-controlled forced dose-titration study, 115 patients with resistant hypertension, receiving background therapy with >/=3 antihypertensive medications including a diuretic at full doses, were randomized 2:1 to increasing doses of darusentan (10, 50, 100, 150, and 300 mg), a selective endothelin receptor antagonist, or matching placebo once daily for 10 weeks. Darusentan treatment decreased mean systolic and diastolic blood pressure levels in a dose-dependent fashion compared with placebo; the largest reductions were observed at week 10 (300-mg dose) (systolic, -11.5+/-3.1 mm Hg [P=.015;] diastolic, -6.3+/-2.0 mm Hg [P=.002]). Darusentan (300 mg) also decreased mean 24-hour, daytime, and nighttime ambulatory blood pressures from baseline to week 10. Darusentan was generally well tolerated; mild to moderate edema and headache were the most common adverse events. This study demonstrates a clinical benefit from a new class of antihypertensive agent in patients classified as resistant by the Seventh Report of the Joint National Committee on Prevention, Detection, Evaluation, and Treatment of High Blood Pressure guidelines.

The role of intention to treat in analysis of noninferiority studies.

In analysing clinical trials designed to show superiority of one treatment compared to another, it is standard to use an intention to treat analytic approach. In active-controlled noninferiority studies, this is not standard, due to concerns that such an analysis will inflate the chance of falsely rejecting the null hypothesis, accepting therapeutic noninferiority when it is not justified. The reasons for using intention to treat (ITT) approaches in superiority studies include a desire to capture all information on study subjects, a need to prevent bias, and assurance that comparative groups are, on average, equivalent in prognostic factors. In this commentary, we argue that these same justifications carry over to noninferiority studies, and that for those and other reasons it should be the preferred analytic approach. We review regulatory guidelines, and propose a number of approaches to minimizing the potential disadvantages of the ITT approach in the noninferiority setting.

Tree-structured gatekeeping tests in clinical trials with hierarchically ordered multiple objectives.

This paper discusses a new class of multiple testing procedures, tree-structured gatekeeping procedures, with clinical trial applications. These procedures arise in clinical trials with hierarchically ordered multiple objectives, for example, in the context of multiple dose-control tests with logical restrictions or analysis of multiple endpoints. The proposed approach is based on the principle of closed testing and generalizes the serial and parallel gatekeeping approaches developed by Westfall and Krishen (J. Statist. Planning Infer. 2001; 99:25-41) and Dmitrienko et al. (Statist. Med. 2003; 22:2387-2400). The proposed testing methodology is illustrated using a clinical trial with multiple endpoints (primary, secondary and tertiary) and multiple objectives (superiority and non-inferiority testing) as well as a dose-finding trial with multiple endpoints.

Randomization as a basis for inference in noninferiority trials.

Noninferiority testing in clinical trials is commonly understood in a Neyman-Pearson framework, and has been discussed in a Bayesian framework as well. In this paper, we discuss noninferiority testing in a Fisherian framework, in which the only assumption necessary for inference is the assumption of randomization of treatments to study subjects. Randomization plays an important role in not only the design but also the analysis of clinical trials, no matter the underlying inferential field. The ability to utilize permutation tests depends on assumptions around exchangeability, and we discuss the possible uses of permutation tests in active control noninferiority analyses. The other practical implications of this paper are admittedly minor but lead to better understanding of the historical and philosophical development of active control noninferiority testing. The conclusion may also frame discussion of other complicated issues in noninferiority testing, such as the role of an intention to treat analysis.

The fallback procedure for evaluating a single family of hypotheses.

In testing multiple hypotheses, control of the familywise error rate is often considered. We develop a procedure called the "fallback procedure" to control the familywise error rate when multiple primary hypotheses are tested. With the fallback procedure, the Type I error rate (alpha) is partitioned among the various hypotheses of interest. Unlike the standard Bonferroni adjustment, however, testing hypotheses proceeds in an order determined a priori. As long as hypotheses are rejected, the Type I error rate can be accumulated, making tests of later hypotheses more powerful than under the Bonferroni procedure. Unlike the fixed sequence test, the fallback test allows consideration of all hypotheses even if one or more hypotheses are not rejected early in the process, thereby avoiding a common concern about the fixed sequence procedure. We develop properties of the fallback procedure, including control of the familywise error rate for an arbitrary number of hypotheses via illustrating the procedure as a closed testing procedure, as well as making the test more powerful via alpha exhaustion. We compare it to other procedures for controlling familywise error rates, finding that the fallback procedure is a viable alternative to the fixed sequence procedure when there is some doubt about the power for the first hypothesis. These results expand on the previously developed properties of the fallback procedure (Wiens, 2003). Several examples are discussed to illustrate the relative advantages of the fallback procedure.

First and subsequent cycle use of pegfilgrastim prevents febrile neutropenia in patients with breast cancer: a multicenter, double-blind, placebo-controlled phase III study.

PURPOSE: We evaluated the efficacy of pegfilgrastim to reduce the incidence of febrile neutropenia associated with docetaxel in breast cancer patients. PATIENTS AND METHODS: Patients were randomly assigned to either placebo or pegfilgrastim 6 mg subcutaneously on day 2 of each 21-day chemotherapy cycle of 100 mg/m(2) docetaxel. The primary end point was the percentage of patients developing febrile neutropenia (defined as body temperature >/= 38.2 degrees C and neutrophil count < 0.5 x 10(9)/L on the same day of the fever or the day after). Secondary end points were incidence of hospitalizations associated with a diagnosis of febrile neutropenia, intravenous (IV) anti-infectives required for febrile neutropenia, and the ability to maintain planned chemotherapy dose on time. Patients with febrile neutropenia were converted to open-label pegfilgrastim in subsequent cycles. RESULTS: Nine hundred twenty-eight patients received placebo (n = 465) or pegfilgrastim (n = 463). Patients receiving pegfilgrastim, compared with patients receiving placebo, had a lower incidence of febrile neutropenia (1% v 17%, respectively; P < .001), febrile neutropenia-related hospitalization (1% v 14%, respectively; P < .001), and use of IV anti-infectives (2% v 10%, respectively; P < .001). The percentage of patients receiving the planned dose on time was similar between patients receiving pegfilgrastim and patients who initially received placebo (80% and 78%, respectively), as would be expected of the study design. Pegfilgrastim was generally well tolerated and safe, and the adverse events reported were typical of this patient population. CONCLUSION: First and subsequent cycle use of pegfilgrastim with a moderately myelosuppressive chemotherapy regimen markedly reduced febrile neutropenia, febrile neutropenia-related hospitalizations, and IV anti-infective use.

Assessing the impact of endpoint shopping on power in confirmatory clinical trials.

When planning a confirmatory study, one of the important aspects is choosing a primary endpoint and evaluating power to show a difference. Data from preliminary studies are generally used for such planning. It is natural to want to use the endpoint from the preliminary study that best differentiates between test drug and control as the primary endpoint in the confirmatory study. However this leads to the possibility of bias in estimation of the effect size in the preliminary study, and, hence, lower than anticipated power in the confirmatory study. In this paper we quantify the impact of such endpoint shopping on the power of confirmatory studies. We find the upper bound on bias and show that for low to moderate correlation it is not very conservative. We derive the asymptotic distribution of the treatment effect and propose a test forequal treatment effects for the data from the preliminary study when endpoints are correlated. We study properties of this test. We propose a strategy to use the data from the preliminary study to plan confirmatory studies with unbiased or conservative estimates of power.