subsequent patients may be entered at the next dose level. DLT is defined as 
NIH Grade III-IV toxicity (Appendix II) in any system that can be attributed to 
treatment or Grade III-IV GVHD (Appendix III) with Neo R gene containing T 
cells detected on biopsy. 
If DLT occurs in 1 of 3 patients at a given dose, an additional 3 patients will be 
evaluated at that level. If none (0) of these 3 patients experience DLT, the TSE 
is escalated for the next cohort of 3 patients. However, if DLT occurs in 1 of 
the additional 3 patients, then the MTD has been exceeded and 3 more patients 
are treated at the previous lower dose level if only 3 patients were treated that 
level. 
If DLT is noted in 2 of 3 patients at a given dose level, the MTD has been 
exceeded and 3 more patients are treated at the next lower level (if only 3 
patients were treated previously at that dose). 
At any time that 2 patients in a dose cohort (of 3-6) patients experience DLT, 
no further patients will be entered at that dose and the study will be halted for 
modification as considered appropriate by local and federal regulatory 
authorities. 
The MTD is defined as that dose level immediately below the dose level at 
which 2 or more patients of a cohort (of 3-6 patients) experience dose limiting 
toxicity. 
9.2 Number of Subjects 
Because of the Phase I design, the exact number of patients required to 
complete this trial cannot be determined. At least three dose levels are planned 
and 3-6 patients will be treated per level. We anticipate enrolling approximately 
10-15 patients over a period of 18-24 months. 
9.3 Design Parameters 
With the escalation plan outlined in section 9.1, the probability of observing 2 
patients with toxicity at a given dose level (conditional on starting a level) is 9 % , 
51% and 83% if the true incidence of toxicity associated with that dose is 10%, 
30% or 50%, respectfully. 
The following table gives estimates of 95% confidence intervals of the true 
incidence toxicity based on the observed incidence and the sequential nature of 
the stopping rule. This table simply emphasizes the lack of precision of the 
phase I design for estimating the true incidence of DLT. 
[ 210 ] 
Recombinant DNA Research, Volume 17 
