Abstract
Abstract Introduction: Epidemiologic observations have suggested a protective effect of soybeans against a number of epithelial cancers including oral malignancies and by inference precursor lesions such as leukoplakia. Several compounds in soybeans have shown activity in preclinical models; we have focused our studies on BBI (Bowman-Birk inhibitor), which is active against the protease chymotrypsin. Our phase I trial demonstrated a very low toxicity profile and a 31% response rate in a 1-month nonrandomized study that was associated with favorable modulation of protease activity and neu oncogene in exfoliated buccal mucosal cells (EBMC). Methods: An intent-to-treat(ITT) randomized placebo (Quaker mass harina, a corn flour)-controlled, double-blind clinical trial of a soybean concentrate (C) of BBI (600 C.I. units) was performed in a multinstitutional investigation (7 sites). The study duration was 6 months and included pre/interim/postevaluation of lesions sizes and pre/post photographic assessments and oral mucosa biopsies(with post central pathology review) of the involved area(s). Intermediate biomarkers (IBM) included serial measurements of EBMC neu protein (ng/mg) and protease (Delrfu/min/ug protein) and serum neu protein (ng/ml). 325 patients underwent preliminary screening and 148 per protocol eligible were enrolled. Of these, 132 were randomized and 105 completed 6 months on study. All data on lesion sizes, photo judgments, and pathology indications of degree of abnormality or change in abnormality were entered into SAS datasets and subjected to 100% verification against the crf forms by the statistician. The several IBM measurements were converted from the original Microsoft Excel sheets into SAS data sets, and subject to spot checks against the original spreadsheets. Similarly, host-factor information from the questionnaires was spot checked against the original records. In all cases, the primary, per-protocol analyses was ITT. The per-protocol, intent-to-treat cohort, and all other categorizations of study participants will also be described with appropriate descriptive summary measures. Results: The ITT data set is composed of all those with valid, two-dimensional measurements on all lesions observed at both the randomization and 6-month visit. 89 evaluable patients met these criteria: 43 in the treatment and 46 in the placebo group. For the BBIC group, the mean relative percent change in total lesion area was −20.6% and for the placebo group −17.1%. Clinical responses for the 89 patients were: four showed a complete response (4.5%), 22 showed a partial response (25%), 53 showed stable disease (60%), and 10 showed disease progression (11.2%). For the drug group the CR+PR(>50% change) was 27.91% and for placebo group 30.43%. Neither the lesion size nor response comparisons demonstrated differences between the two groups that were significantly different (p>0.05). Photos of the same lesion at baseline and at the 6-month exam were available for 91 participants. Five qualified reviewers made judgments of the degree of change in abnormality on a seven-point scale, blinded to study arm and timepoint of photos. For mean comparison scores, 1 was substantial improvement over time, 4 indicated no change and 7 meaning much worse decline over time. Preliminary assessments of 77% of the patients having pre/post photos indicates that there were no significant differences between the placebo and treatment groups. Conclusion: BBIC is not effective as a chemoprevention agent for the management of oral leukoplakia. Central pathology review by two reviewers is near completion, but is unlikely to affect this conclusion. Final measurements of the three biomarkers should be available by the time of presentation and subanalysis will be presented for the two groups and for the patients who seemed to have had a clinical response. Citation Information: Cancer Prev Res 2010;3(12 Suppl):CN02-05.