Experimental design

From HemOnc.org - A Hematology Oncology Wiki
Jump to navigation Jump to search

The purpose of this page is to explain details relevant to experimental trial design, as well as to explain our notations for such designs. It will be expanded over time.

Experimental arm design

By definition, the experimental arm of a trial does something different from the comparator/baseline arm. In hematology/oncology trials, what that difference is falls into four general categories:

  • Escalations: arms where one or more of the four following types of escalation are undertaken:
    • New drug: eg control arm is A and experimental arm is A + B
    • More cycles: eg control arm is A x 3 and experimental arm is A x 6
    • Higher dose: eg control arm is A 300 mg and experimental arm is A 600 mg
    • Higher intensity: eg control arm is A every 3 weeks and experimental arm is A every 2 weeks
  • De-escalations: the converse of escalations
  • In-class switches: different drugs that are within the same general class (e.g., chemotherapy, immunotherapy). Two common flavors of this:
    • A versus B
    • A + B versus A + C, where A is the "backbone"
  • Out-of-class switches: different drugs that are within different classes (eg, control is chemotherapy, experimental is immunotherapy)

These classes are not always mutually exclusive. For example, the substitution of standard chemotherapy for high-dose chemotherapy with autologous hematopoietic stem cell transplant in the treatment of breast cancer (e.g., Schmid et al. 2005) is an example of an in-class escalation design.

Labeling

Experimental arms are labeled using the following schema:

  • Escalation arms: (E-esc)
  • De-escalation arms: (E-de-esc)
  • In-class switch arms: (E-switch-ic)
  • Out-of-class switch arms (E-switch-ooc)

Registration trials

These represent a special subset of clinical trials that were used to make an approval or labeling decision by the FDA or similar agencies. We have labeled all trials that are explicitly mentioned in regulatory labels with (E-RT) if the experimental arm of an RCT was used to justify approval or a labeling change, and as (RT) if a non-randomized trial was used for such justification.

Putting it together

For trials that are registration trials and RCTs, the label is combined. For example an escalation arm would be labeled as (E-RT-esc).

Placebos

Complex multipart regimens

Some trial designs are much more complicated than the examples described above. In these cases, it can be more difficult to determine the type of difference between the comparator and experimental arms. See the dedicated page for more details.

Endpoints

Of equal importance to the trial design is the choice of endpoint(s). Traditionally, the sample size of a trial is calculated such that there is adequate power to detect one or more primary endpoints (also called primary objective, main endpoint, major endpoint). All other measured endpoints are considered secondary. See the response to treatment page for more details about endpoints.

Superiority designs

Non-inferiority and equivalence designs