EffTox

The EffTox dose-finding method seeks the optimal dose of an investigational agent, balancing the hope of efficacy and the risk of toxicity. This method has numerous advantages over traditional toxicity-only dose-finding methods.

• EffTox seeks what the patients seek. Patients do not enter a trial hoping to be treated at a dose with a specified probability of toxicity. Patients enter a clinical trial wanting to optimize their chance of response and minimize their chance of toxicity.
• EffTox makes utilities explicit. The trade-off between efficacy and toxicity is always present in dose-finding trials, although it is usually implicit. For example, the CRM method assumes that the dose with probability of toxicity closest to an arbitrary target provides the best balance of risk of toxicity and hope for response. By contrast, the design of an EffTox trial begins by asking investigators for the relative desirability of pairs of probabilities of efficacy and toxicity.
• EffTox is suitable for biologic agent trials. Traditional dose-finding methods in oncology have been developed to test cytotoxic agents and therefore assume that the probability of efficacy increases monotonically with dose. However, the probability of efficacy for a biologic agent may increase to a maximum and then decline. The EffTox method has the flexibility to explore non-monotone response curves. (See Scenario 3 of the Pentostatin example in [1].) Furthermore, a biologic agent may be effective at a dosage with little probability of toxicity. If the EffTox method finds a dose which is safe and effective, it will not seek higher doses, unlike a toxicity-based method.
• EffTox can combine phase I and phase II. Combined phase I/II trial designs make decisions based on more data and are more efficient to run. The traditional approach uses only toxicity data from phase I patients and only efficacy data from phase II patients, whereas EffTox uses efficacy and toxicity data from all patients. A single phase I/II trial is less effort to organize than separate phase I and II trials. Also, a combined phase I/II trial avoids the loss of momentum between the end of phase I and the beginning of phase II.

See the EffTox user's guide available here included in the download distribution.

Note: Versions of EffTox up through 2.8 used inverse-quadratic contours to determine efficacy-toxicity trade-offs as described in [1]. Starting with version 2.9, we will switch to the L^p norm trade-off functions first described in [2]. The new trade-off functions are described in more detail in [3].

Clift Norris developed the user interface using C# and the Microsoft .NET framework version 1.1.

John Cook implemented the simulation kernel using Visual C++.

References

[1] Peter F. Thall and John D. Cook. Dose-Finding Based on Efficacy-Toxicity Trade-Offs. Biometrics 60, pp. 684-693, September 2004.

[2] Peter F. Thall and John D. Cook. Adaptive dose-finding based on efficacy-toxicity trade-offs Encyclopedia of Biopharmaceutical Statistics, 2nd Edition. Chein-Chung Chow editor. 2006

[3] John D. Cook Efficacy-toxicity trade-offs based on L^p norms Technical Report UTMDABTR-003-06.

[4] Peter F. Thall, John D. Cook, and Eli Estey. Adaptive dose selection using efficacy-toxicity trade-offs: illustrations and practical considerations. Journal of Biopharmaceutical Statistics Vol 16, No. 5, pp. 623-638, 2006.