Data Analysis
Plan for SFDR2002:
Safety and Efficacy of Inerdiel versus Actiniel and Placebo
Title: A Randomized, Double-Blind,
Active and Placebo-Controlled, Parallel-Groups 12-Week Trial Evaluating the
Safety and Efficacy of Inerdiel 50mcg BID versus Actiniel 100mcg and Placebo in Adult Subjects with moderate
mentarthric impairment.
This
document describes in detail the statistical methods that will be used to
summarize and analyze the background, efficacy, and safety data from SFDR2002. All decisions regarding analysis, defined in
this document, have been made prior to any review of the study data.
Mentarthric
impairment is a condition that affects a small percentage of people. It occurs when rheumatoid arthritis, coupled
with shizophrenia-like tendencies, combine to produce the illusion of bloating
and abdominal discomfort. Actiniel, a
psychoactive substance, is the only treatment so far that has been approved for
mentarthria; it works by relieving perceived abdominal discomfort. The sponsors wish to seek approval for
Inerdiel, another psychoactive substance, which relieves symptoms through
relief of both arthritis and schizophrenia, thereby relieving the perceived
abdominal discomfort as a by-product.
The
objective of this trial is to demonstrate superior efficacy and comparable
safety of Inerdiel compared Actiniel in subjects with moderate mentarthic
impairment.
The
primary endpoints chosen to show the efficacy of the products are
·
Patient-reported
symptoms at each weekly visit, ending with week 12. A single measure on a 1,2,3,4,5 Likert scale response will be
obtained at each visit.
·
Acting
physician-reported evaluation of signs and symptoms at each weekly visit,
ending with week 12, assessed blindly from the patient reports.
Secondary
endpoints are
·
Change
from baseline in each of the 36 measurements used in the SF-36 quality of life
questionnaire.
·
Survival
in study (measuring patient drop-out rate)
Safety
measures will include adverse events, clinical laboratory results, 12-lead
ECGs, and vital signs.
This
is a randomized, double-blind, placebo and active-controlled parallel-group
trial. The study will be conducted on
an outpatient basis.
Qualified
subjects will be randomly assigned to one of three treatment groups for 12
weeks of treatment. The three possible
treatment groups are:
i. Inerdiel 50mcg BID
ii. Actiniel 100mcg BID
iii. Vehicle Placebo
Double-blind
treatment (Visit 2, Treatment Day 1) will be started 14 ± 2 days after Visit 1
and continue for 12 weeks. Subjects
will return to the clinic at 1-week intervals for 4 weeks followed by
visits at 2-week intervals, for a total of 10 visits (for a total of 84 (±2) days of study treatment).
All
analyses outlined in this document will be carried out once the study database
has been authorized by Data Management staff, Clinical Statistics personnel
have had an opportunity to inspect and approve the authorized database, and the
treatment details have been attached to the subject data. Protocol deviations will be identified
before the treatment details are attached to the subject data.
As
the study database is being authorized, Clinical Statistics personnel will
create a SAS dataset of treatment details (subject number, treatment number and
treatment description). This database,
once validated, will become part of the reporting database for this study. Treatment numbers (variable TMT) will serve
as the link between a subject’s treatment details and study data.
Sample
size calculations assume a significance level of 0.05 and are based on
two-sided t-tests. The primary
treatment comparison is the Inerdiel versus Placebo comparison.
Analysis of data from previous studies indicates that 100 patients per
treatment group should provide more than 90% power to detect an average
difference of .5 on the Likert scale, so 300 subjects will be randomized to the
three groups, 100 subjects per group.
The
power calculation described above does not consider multiple comparisons.
The
population analyzed will be the intent-to-treat population consist of all
randomized patients. Analyses based on
the Intent-to-Treat population will include all available data for these patients. This population will form the basis for all
safety, background, and demographic summarizes, as well as for the analysis of
all efficacy measures and of health outcomes data.
No
interim analyses are planned for this study.
Two-sided
statistical tests will be used throughout the analyses; p-values of a=0.05 or less will be
considered to be statistically significant unless specified otherwise. Symbols will be used to indicate significant
model effects. In general, actual
values will be analyzed at baseline, and change from baseline will be analyzed
at all on-treatment time points.
Prior
to the authorization of the study database by Clinical Data Management
personnel the data will be reviewed and any discrepancies or unclear data will
be queried with the investigational sites using Data Clarification Forms
(DCFs).
All
statistical tests of efficacy measures, demographic/background characteristics,
and cardiovascular safety measures (QTc intervals, heart rate, and Holter
monitoring results) will be adjusted for investigator cluster. Investigator clusters will involve the
combining of investigators from geographically similar regions. This combining will be necessary when the
number of patients recruited by individual investigators is small.
Parametric
analyses will be based on models that include terms for treatment and cluster;
nonparametric analyses will be stratified by investigator cluster. In addition, analysis of covariance (ANCOVA)
models assessing weekly results will include baseline value as a covariate and
terms for treatment and investigator cluster.
Subgroup
analysis will be included in this study using age, ethnic, genotype and gender
subgroups will be included for all endpoints.
· Premature discontinuation and missing data
For
any subject who withdraws from the study prematurely, all available data up to
the time of discontinuation will be included in the Intent-to-Treat analysis.
Because of the large number of investigators participating
in this study, and the possibility of small numbers of patients randomized at
some investigational sites, groups or clusters of investigators will be used
for statistical analyses instead of individual investigators. Treatment-by-cluster interaction terms will
be assessed for statistical significance.
Potential
treatment interactions with cluster will be assessed individually in separate
models.
The
three treatment comparisons of primary interest are the Inerdiel 50mcg BID
versus Actiniel 100mcg BID, the Inerdiel 50mcg BID vs placebo, and the Actiniel
100mcg BID versus placebo. No
adjustments for multiple comparisons will be considered.
As
described in the protocol, a window of +2 days exists for each scheduled
Visit. Any data collected outside of a
defined assessment window, or multiple measurements collected within the same
assessment window will be defined as protocol violations. For data collected outside the assessment
window, the data will be used in the ITT analysis. For multiple assessments, the last valid value will be used.
The
number of patients randomized, both overall and by investigator, will be
summarized by treatment group. The
number of patients who complete the 12-week treatment period, as well as the
number who withdraw early from the study and the specific reasons for
withdrawal, will be summarized by treatment group. Reasons for withdrawal due to worsening mentarthria will also be
summarized by treatment group.
Summaries
by treatment group will also be produced for screening failures (demography,
reason for failure) and for run-in drop-outs (demography, AEs, reasons for withdrawal
before randomization).
A summary table and a listing for individual patients will describe the protocol deviations for the clinical report. Deviations will also be categorized as related to inclusion criteria, exclusion criteria, continuation criteria, or other sources.
Compliance
will be calculated as the total estimated number of doses used over the known
time interval (based on treatment start and stop dates and clinic visit and
diary dates) divided by the total expected use during that time period,
multiplied by 100.
The
number of patients will be summarized by investigator and treatment group. Patient demographic characteristics will be
summarized by treatment group using summary statistics appropriate to the
nature of the data. Means, standard
deviations, minima, medians, and maxima will be used to summarize age, height,
and weight; frequency distributions will be used for gender, ethnic origin, age
group, and child-bearing potential.
Inferential
analyses of demographic characteristics (e.g., analysis of variance F-tests for
age, height, and weight, and Cochran-Mantel-Haenszel tests for gender, age
category, ethnic origin, tobacco use, child-bearing potential, and methods of
contraception) will be used to assess the comparability of the treatment
groups. These tests will correct for
investigator cluster.
In
all analyses described below, the null hypothesis being tested is that of no
treatment difference. Alternative
hypotheses assert the superiority over placebo in the cases of placebo
comparisons, or superiority of Inerdiel over Actiniel in the case of the active comparison. All statistical tests use a
two-sided significance level of a=0.05 unless indicated otherwise.
· Treatment comparisons
Statistical
analyses data at each endpoint will be based on a change in response scores
from baseline. Treatment groups will be
compared using an ANCOVA model with baseline response as the covariate and
terms for treatment and investigator cluster.
· Clinically meaningful differences between treatment groups
A
difference of 0.5 or more between treatment groups at endpoint in the mean
change from baseline will be considered to be a clinically meaningful
difference.
· Within-treatment group changes
In addition, changes from baseline at endpoint (either Treatment Week 12 or Discontinuation Visit) will be assessed within each treatment group for each domain using t-tests. Within-treatment changes from baseline will be examined for descriptive purposes in order to gauge the magnitude of absolute change for each treatment group. For this study, a within-treatment change from baseline will be considered to be clinically meaningful if it is both statistically significant and is 0.5 or greater.
The
extent of exposure, calculated as the number of days between start of treatment
and end of treatment (i.e., treatment stop date minus treatment start date +
1), will be summarized by treatment group.
Adverse
events (AEs) will be summarized and grouped by body system and by AE within
body system. Results will be displayed
in order of decreasing frequency, both across body systems and within each body
system. Drug-related AEs, serious AEs,
most commonly occurring (defined as an incidence of >3% in any treatment
group) AEs, and those AEs leading to withdrawal, will be summarized in a
similar manner. The incidences of the
most commonly occurring AEs will be compared across treatment groups using
two-sided Fisher’s exact tests. A by-patient listing of all AEs will also be
provided.
These
will be investigated thoroughly and presented by patient and in summary
form. One-sided Fisher's exact tests
will be used to compare incidences.
For example, drug‑drug interactions.
Vital
signs, ECG results, heart rate and cardiac abnormalities, and physical
examination findings at the end of the study will be summarized by treatment
group and by visit where appropriate.
ECG
results will include summaries of heart rate, QT intervals, and an assessment
by an independent cardiologist of clinically significant abnormalities. The change from baseline in QTcs and heart
rate and the actual values at screening and at baseline will be compared across
treatment groups using ANOVA F-tests based on models which include terms for
treatment and investigator cluster.
If genotyping is included in this study, include a description of the planned analyses.
Genotyping data using 20 pre-defined single nucleotype polymorphisms (SNPs) will be obtained for each patient. Subgroup analyses to detect significant treatment effects and significant interactions will be performed for various genotype subgroups, using ANOVA methods.