Innovations in phase 1 trial design: where do we go next?

In this issue of Clinical Cancer Research, Dees et al. (1) describe the use of PGDE (2) in a Phase 1 trial of CI-958. PGDE was proposed (2, 3) as a more informative and efficient alternative to the standard design for Phase 1 studies (“modified Fibonacci”). Compared with the modified Fibonacci design, the authors estimate that the use of PGDE in this trial required 15–18 fewer patients. All of these additional patients would have been entered at dose levels far below the MTD. The authors state that this trial was the most successful use of PGDE ever reported. The principal reason for the efficiency of PGDE in this case was the 170-fold range between the starting dose, 5.2 mg/m, and the MTD, 875 mg/m. This report adds to the collection of positive experiences with PGDE as an innovation in Phase 1 design. However, despite encouraging reports from Europe (4), Japan (5), and the United States (6), the success of PGDE has been more academic than practical. As noted by Dees et al., PGDE requires real-time pharmacokinetic monitoring, and this element has been considered a drawback by many investigators. Consequently, PGDE has failed to be widely accepted, and has generally faded from regular use. PGDE is not the only alternative to the modified Fibonacci escalation procedure, and other strategies don’t require plasma concentration monitoring. Dees et al. mention both the accelerated titration and continual reassessment methods, and experiences have been recently surveyed for a wide range of trial designs (7). These methods have improved Phase 1 trials in a number of important ways. It’s particularly noteworthy that fewer patients are exposed to completely inactive doses because of two innovations: (a) reducing the cohort size at early stages from three patients to one; and (b) encouraging intrapatient escalation of doses. The loss of real-time pharmacokinetics information translates into missed (or delayed) opportunities for early identification of interspecies differences, active and/or toxic metabolites, as well as preliminary hypotheses regarding dosage adjustment for impaired organ function. For example, during a Phase 1 study from the same institution for penclomedine (8), a metabolite was identified that was less toxic and more active than the parent molecule.