BACKGROUND: At the University of California-San Francisco (UCSF), a labeling index (LI) based on in vivo administration of bromodeoxyuridine (BUdR) was routinely determined in tumors from patients with infiltrating gliomas for the 10-year period, 1984-1994. Considered an indicator of tumor proliferative potential, it was anticipated that the BUdR LI would aid in determining patient prognosis. The data have now matured, and a final report on the extensive UCSF experience is appropriate. METHODS: Patients were grouped separately based on tumor grade and whether LI was determined at the time of primary diagnosis or at recurrence. To assure data quality and consistent follow-up, only adult patients entered on clinical trials were included. Patients with primary Grade 2 gliomas could not be included because most of these patients did not enter clinical trials. Each patient LI was categorized as low (lowest quartile), moderate (between the 25th and 75th percentiles), or high (highest quartile). Statistical analyses included log rank and multivariate proportional hazards models. RESULTS: One hundred ninety patients with primary tumors (69 Grade 3, 121 Grade 4) and 149 patients with recurrent tumors (24 Grade 2, 37 Grade 3, and 88 Grade 4) were included. LI score was predictive of survival (P < or = 0.02 for univariate or multivariate analyses) in all but the primary Grade 4 patients. For that group, it was a significant predictor (P = 0.03) only among those surviving > or = 1.5 years. CONCLUSIONS: BUdR LI improved the predictability of long term patient outcome both initially and at the time of recurrence. Further research on measures of tumor proliferative potential as markers of outcome is warranted.
BACKGROUND: At the University of California-San Francisco (UCSF), a labeling index (LI) based on in vivo administration of bromodeoxyuridine (BUdR) was routinely determined in tumors from patients with infiltrating gliomas for the 10-year period, 1984-1994. Considered an indicator of tumor proliferative potential, it was anticipated that the BUdR LI would aid in determining patient prognosis. The data have now matured, and a final report on the extensive UCSF experience is appropriate. METHODS:Patients were grouped separately based on tumor grade and whether LI was determined at the time of primary diagnosis or at recurrence. To assure data quality and consistent follow-up, only adult patients entered on clinical trials were included. Patients with primary Grade 2 gliomas could not be included because most of these patients did not enter clinical trials. Each patient LI was categorized as low (lowest quartile), moderate (between the 25th and 75th percentiles), or high (highest quartile). Statistical analyses included log rank and multivariate proportional hazards models. RESULTS: One hundred ninety patients with primary tumors (69 Grade 3, 121 Grade 4) and 149 patients with recurrent tumors (24 Grade 2, 37 Grade 3, and 88 Grade 4) were included. LI score was predictive of survival (P < or = 0.02 for univariate or multivariate analyses) in all but the primary Grade 4 patients. For that group, it was a significant predictor (P = 0.03) only among those surviving > or = 1.5 years. CONCLUSIONS: BUdR LI improved the predictability of long term patient outcome both initially and at the time of recurrence. Further research on measures of tumor proliferative potential as markers of outcome is warranted.