11.0 STATISTICAL CONSIDERATIONS 
11.1 Analysis of Peripheral Blood Specimens 
Systemic IL-4 levels and the phenotype of circulating peripheral blood 
lymphocytes will be determined on days 0 (pre - vacc ination) , 7, 14, 21, and 28. 
Follow-up multiple comparisons, based on either the paired-t or signed rank 
statistic, will be applied in the interpretation of significant results. 
11.2 Analysis of Local Toxicity 
The incidence of severe, vacc ine - related , local reactions will be recorded. The 
relationship between local toxicity and the number of transferred cells and the 
amount of IL-4 elaborated per site per day will be determined. A formal test for 
increasing dose - response may be based on a scored-rank test with p-values 
obtained by within-patient permutation of dose levels. To insure the validity of 
this approach, and more importantly to eliminate the possibility of positional 
bias, the relative location of vaccinations will be randomized in double-blind 
fashion for each patient. 
11.3 Dose -Response for Mononuclear Cell Infiltration 
Biopsies obtained after primary vaccination on day 14 will provide an opportunity 
to quantify the relationship between the amount of IL-4 delivered and the degree 
of infiltration by mononuclear cells. These data will first be investigated by 
standard repeated measures methods; but, because the description of between- 
patient variability both in level of response and in optimal dose are of 
considerable interest, additional analyses are planned as described below. 
It is thought likely that the dose - response relationship may not be monotonically 
increasing, but rather may re-descend at sufficiently high levels of IL-4. The 
distribution over patients of the maximal response and of the maximizing dose 
will be tabulated. These quantities will be estimated for each patient under an 
assumption of unimodal dose - response curves. Briefly, a pooled-adjacent- 
violators algorithm will be used, but adapted to unimodal rather than monotonic 
curves. A formal test of dose - response may be based on a permutation test whose 
statistic is an aggregation of the deviations between each patient's estimated 
maximum and his or her average level of infiltration. Comparison of IL-4 
vaccinations against the nontransduced control may be based on a similar 
strategy . 
A similar approach may be used to analyze the dose -response relationship for 
other endpoints, including local endothelial activation and cytokine production. 
11.4 Kinetics of T-Cell Infiltration 
Serial biopsies of secondary vaccination sites on days 15, 16, 18, 21, and 28 
will help to define the time course of the local immune response. These data 
will be analyzed in a manner similar to that described in 11.3. 
[510] 
Recombinant DNA Research, Volume 16 
