This case study considers the production of an ISS and ISE for an orphan drug. The fact that the compound
was an orphan drug introduced some unique problems, as the number of patients was lower than in most
integrated summaries. As well as the common problems identified in this article, below I present a selection of the specific issues associated with this particular ISS and ISE. Due to close collaboration between the clinical team and the statistician, identification of all these issues occurred during the writing of the statistical analysis plans. This enabled identification and explanation of the appropriate methods prior to starting the analyses.
|The primary endpoint was time to beginning of
relief of symptom. However, in one pivotal study
the definition used the time of first identification
of relief of symptoms, whilst in the second study
persistence of the symptoms was also required.
|The primary analysis for the ISE was defined as
time to beginning of relief of symptoms with
persistence to avoid any potential for multiple
primary endpoints in the ISE.
|The drug was developed to treat patients who
had an underlying genetic disease that resulted
in frequent attacks. Studies included healthy
volunteers, asymptomatic patients and patients
suffering from an attack. Trials that included
healthy volunteers and asymptomatic patients
were primarily uncontrolled. However, including
patients from these trials in the ISS had the
potential to bias the denominator when assessing
the group of patients at risk.
|Two separate sets of safety summaries were
produced from a single derived database. One
set for symptomatic patients and the other set for
asymptomatic patients and healthy volunteers.
|Due to the excellent efficacy and difficulty in
recruiting high patient numbers, the number
of patients treated in a Randomized Controlled
Trial (RCT) was very low (N = 38 treated with the
marketed drug, of which only 12 had been treated
at the recommended marketing dose).
|After discussions with the regulatory authorities,
results from the open-label extensions were also
included in the ISE and ISS. The RCT trials were
used to produce the key outputs discussed but use
was made of supporting evidence from outputs
incorporating the RCT and open-label information.
|Patients suffered multiple attacks, and sometimes
received multiple treatments in the open-label
extension phases. The regulatory authorities raised
concerns that the efficacy of the drug may reduce
when taken multiple times.
|Statistical methodologies were employed that
allowed within patient comparisons of the time
to event primary endpoint. In addition, patients
treated for more than 5 attacks were summarized
using by patient figures to show that the efficacy
did not decrease with multiple attacks.
|The regulatory authorities raised similar
concerns over the safety impact of treating patients
|AEs and other safety endpoints were summarized
by attack. These were used in conjunction with
immunological results, to show that there was no
evidence of effects due to multiple dosing.
|Prior to treatment, the severity of the patient’s
attack was assessed to decide whether the subject
was eligible. However, the criteria for assessing
the attack’s eligibility were different in the early
studies compared to the pivotal studies. In
addition, the primary endpoint used a VAS that
assessed severity of symptoms, but the VAS in the
early phase studies assessed the change in severity
of symptoms since baseline.
|Based on the substantial differences in the early
studies, and as these studies were not randomized
controlled studies, they were included in the safety
database but not the efficacy database.
|Receipt of certain doses only occurred in a single
trial. If trial was included as a covariate in the
analyses this was confounded with treatment,
meaning that the treatment effect could not be
|For each analysis, careful consideration and
adjustment of the covariates allowed the
answering of the specific questions the analysis
was trying to address.
|There were a large number of different treatment
regimens, so presenting by regimen made
producing combined results difficult.
|Treatment doses were grouped into ‘Very low
doses’ (known to be inefficacious), ‘Low doses’,
‘Marketing dose’ or ‘Higher than the marketing
dose’. As well as allowing sensible grouping
of treatment groups, this approach helped in
justifying the choice of the requested marketing
dose by showing that efficacy did not increase
in higher doses. The safety of the dose was also
justified by showing that safety in the higher dose
was comparable to lower doses and placebo.
The below example timelines are designed to show how front loading the planning and production of the ISS and ISE can enable the production of the outputs to be significantly reduced post unblinding of the pivotal studies. The timelines cover up until the production of final outputs but clearly the reduction in timelines to this stage enables a reduction to the final submission.
Some points to highlight in this plan: