Bioequivalence and Bioavailability Forum • TSD procedure

TSD procedure [Two-Stage / GS Designs]

posted by Helmut – Vienna, Austria, 2014-01-04 15:41 (4545 d 02:54 ago) – Posting: # 12138
Views: 9,590

Hi Tina,

IMHO that’s a matter of taste. As an appetizer I can give you my procedure:

Make sure that the clinical database is reviewed, locked, and can be reopened for adding data of the second stage. The last point is important. I once had to deal with a CRO telling me that they cannot open the database again and offered me preliminary data. That’s not acceptable.
Make sure that you get final results from the analytical site. It’s not necessary to have an analytical report, but the data have to be reviewed and released by the QAU.
Merge the data (administrations times, actual sampling times, concentrations, randomization) and perform NCA/biostats of the first stage.
Based on the outcome (and following the decision tree of the chosen TSD method):
- If BE is shown for all PK metrics, report it to the sponsor (formal written communication). The study will be stopped and all reports (clinical, analytical, statistical, and integrated) shall be prepared.
- If not, continue:
Perform the sample size estimation (n_total) based on the highest CV of relevant PK metrics.
Calculate the preliminary sample size of the second stage n₂ = n_total – n₁.
Based on the drop-out rate observed in stage 1, increase n₂ accordingly in order to maintain the desired power.¹
The published methods were validated for balanced studies only. In order to prevent questions from nitpicking assessors, consider adjusting for imbalance in the second stage.¹ The wording I use in my protocols is:
“If an unequal number of subjects in the respective sequences drop out in the first stage resulting in an imbalanced design, in the optional second stage subjects will be unequally randomized to sequences in order to counteract imbalance.”
Communicate the sample size with the sponsor in writing.² The second stage will be initiated. Wait for the data. Merge them with the first data set. Business as usual.

All of the above should be unambiguously stated in the clinical study protocol and the SAP.

Example:
RT TR subjects / sequence
12 12 24 n₁ dosed
11 10 21 n₁ eligible
14% drop-out rate
30 n_total (based on CV_max)
9 preliminary n₂ (n_total – n₁)
11 final n₂ (adjusted for drop-out rate)
5 6 11 randomized/dosed in stage 2 (imbalanced to keep overall balance)
4 5 9 expected eligible subjects in stage 2
15 15 30 expected pooled data set (hopefully balanced)
CVs of relevant PK metrics and calculated sample size of the second stage. IMHO no need for a full report. An e-mail requesting acknowledgment is sufficient.

—
Dif-tor heh smusma 🖖🏼 Довге життя Україна!
Helmut Schütz

The quality of responses received is directly proportional to the quality of the question asked. 🚮
Science Quotes

Complete thread:

Interim analysis report generation Tina 2014-01-02 09:33 [Two-Stage / GS Designs]
- TSD? Helmut 2014-01-02 10:26
  - TSD? Tina 2014-01-03 17:41
    - TSD procedureHelmut 2014-01-04 14:41
    - R-code for balance Helmut 2014-01-04 22:36
      - R-code for balance Tina 2014-01-06 11:15
    - Fixed sample size? Helmut 2014-01-06 13:52