Record of Telephone Conversation, April 19, 2013 - Kcentra
April 19, 2013 at 11:30 am EST
Wei Hua MD, PhD, MHS - Lead Epidemiologist
Michael D. Nguyen MD - Medical Officer
Sukhminder Sandhu PhD, MPH, MS – Epidemiologist
Antoinette Mangione MD, Pharm.D., Global Clinical Therapeutic Head Acquired Bleeding, Sr. Director, Clinical Research and Development
Bruce Hug, MD, PhD, Director, Clinical Research and Development
David Desris, Pharm.D., Sr. Manager, Regulatory Affairs
Robert Goldberg-Alberts M.A., Therapeutic Area Lead Statistician, Acquired Bleeding, Clinical Research & Development
Doug Watson, PhD, Director Clinical Epidemiology, Clinical Research & Development
Val G. Romberg, Senior Vice President, Plasma R&D
Purpose: FDA requested a telecon to finalize some questions regarding the PMR Concept
I. Sample size
FDA asked CSL to clarify the current sample size calculation, which was noted as 95% for a confidence interval less than 2.
The sponsor confirmed that it as a one-sided test.
FDA asked how was the background rate calculated?
The sponsor looked at subjects in study 3002, based on several definitions discussed between FDA and CSL, the sponsor felt a 5% rate was reasonable rate among the different definitions.
The FDA recommended a more conservative rate of 3% for plasma subjects. CSL agreed to power the study for a two-sided test at 0.025 on each side and 3% baseline rate, but they noted that as the sample size goes up, length of the study increases.
FDA inquired how the entire study was being powered (i.e. between historical and contemporaneous plasma groups). CSL confirmed the study was being powered using the historical plasma group.
II. Contemporaneous vs. Historical controls –
FDA asked if the sample size is for the contemporaneous vs. historical controls and will the sponsor be any matching with Kcentra subject? The current proposed study is for historical plasma controls. CSL indicated they will need to discuss this issue --------(b)(4)---------. Per the protocol, both Kcentra and contemporaneous plasma subjects will be enrolled. The sponsor will continue the study enrollment of Kcentra subjects and contemporaneous plasma subjects until they have sufficient Kcentra subjects. Therefore, the contemporaneously enrolled subjects will include a required sample size of Kcentra subjects and a number of contemporaneous plasma subjects that could be enrolled at the time that the Kcentra sample size is achieved. Historical plasma subjects will be enrolled in a separate process. Since the primary analysis is based on the historical plasma control, CSL viewed this as satisfactory approach. CSL will compare the Contemporaneous vs. Historical controls to determine if there is a meaningful difference. If not, CSL could potentially combine both groups into a single plasma group to extract additional statistical data from the study. To further clarify why the primary analysis is based on the historical plasma control, CSL expressed concerns that contemporaneous plasma subjects may be systematically and drastically different from Kcentra subjects and (2) that they may not be able get the same number of plasma and Kcentra subjects in the study depending on if use of plasma decreases.
FDA thought the above proposal was reasonable justification for the proposal. The FDA expressed agreement with the Sponsor about channel bias.
FDA asked if at the minimum, CSL will perform a 1:1 comparison between Kcentra vs. Contemporaneous groups. CSL stated no, because they suspect that after Kcentra is on the market, the use of plasma will fall off dramatically. CSL did not want to delay the completion of study waiting to enroll additional plasma patients. FDA agreed with this approach, but was concerned about major changes to the standard of care, which could be a major limitation of using historical plasma control. CSL expressed that they do not anticipate medical practice changed in the past decade that may result in dramatic differences in VKA treatment and reversal of VKA in the setting of acute major bleeding; however, CSL will collect covariant information and monitor markers and trends over time in the historical control group.
FDA asked CSL to clarify why they chose not to match in the study since matching would provide enough subjects in different strata to access underlying risk for TEs and address channeling bias. Sponsor will have medical history of their subjects, however they are reluctant to match because they will have to wait for enrollment of Kcentra subjects to enroll historical plasma subjects.
CSL proposed to take all subjects from the prospective population of the study and perform a matched analysis for the contemporaneous control
III. 365-Day enrollment criteria
FDA wanted CSL to keep the inclusion criterion of “a minimum of 365 days of enrollment in the -(b)(4)- prior to index VKA reversal treatment date”, without changing it to 180 days. CSL agreed and will further discuss options with FDA if enrollment becomes difficult and a study delay is anticipated. One option proposed by CSL in the event recruitment is difficult is making the enrollment criteria less stringent to allow more patients to enter. FDA has concerns with this option and suggested possibility of looking at other sites for additional recruitment. CSL will look into the possibility of broadening the study to other (b)(4) sites and needs to discuss with (b)(4).
IV. 14-Day secondary analysis
FDA wanted the sponsor to consider the risk of TE’s after 14 days of treatment as a secondary analysis as is being done in the PMC. CSL agreed to additional14-days (secondary analyses).
V. NDI vs. SSA DMF
CSL provided further clarification between the NDI and SSA DMF:
The Social Security Administration (SSA) database is a collection of databases from varied sources that tracks death data within 1 to 2 months of death. This database guards against fraudulent payments of benefits. The SSA does not guarantee the veracity of their information. The SSA does not track the cause of death.
National Death Index (NDI) also tracks death data and the cause of death. However these data are not as current as the SSA; the data tends to be about a year and quarter behind. NDA provides linkages to actual copies of death records, such as copies of death certificates.
Sponsor feels it is important to determine the cause of death. CSL proposes to use the NDI as the preferred method of determining the cause of death. In cases in which the Contemporaneousand Kcentra patients are not picked up by the NDI database, then the SSA DMF can be used to obtain death data. In the finalized protocol, CSL will indicate that NDI is the preferred and most accurate indication cause of death and the SSA database will be used to track recent death information.