1. Endpoint Comparison for Osteoporosis Assessment in Cancer Control Studies (N02C1 and N03CC) A. C. Dueck 1, P. J. Atherton 2, H. Liu 2, S. L. Hines 3, C. L. Loprinzi 2, E. A. Perez 3, A. D. Tan 2, K. Burger 2, X. Zhao 2, B. Diekmann 2, J. A. Sloan 2 1 Mayo Clinic, Scottsdale, AZ; 2 Mayo Clinic, Rochester, MN; 3 Mayo Clinic, Jacksonville, FL Abstract Background: Methodological issues hamper efficacy assessment of agents to prevent osteoporosis in cancer survivors. The diagnosis of osteoporosis can vary depending upon which bone mineral density (BMD) site is used, which machine is used for the assessment, and which set of normative values are applied. This analysis compared different endpoints for the assessment of efficacy in 2 cancer control studies. Methods: Data from 2 NCCTG phase 3 cancer control studies were employed involving 216 (N02C1) and 558 (N03CC) patients each comparing 2 treatments for the prevention of osteoporosis. BMD data were collected from spine (primary endpoint), femoral neck, and total hip. Endpoints included raw BMD score (RawBMD); raw machine-based T-score (TRaw); sample-standardized T-score (TSamp); reference population standardized T-score (TRef); and Z-score percentile corresponding to TRef (TPerc). For each of the 5 endpoints, summary statistics were computed and the analytical comparison of treatment arms was carried out via Student’s t-test, Wilcoxon rank-sum, and baseline BMD-adjusted ANCOVA using change from baseline (CB) and percentage change of baseline (%CB) at 1 year. Each study was analyzed separately. Results: Results differed by BMD site and whether CB or %CB was used. Among the 5 endpoints, TRaw, TSamp, and TRef had 11, 8, and 7 statistically significant results out of 36 tests, respectively. TPerc and RawBMD produced the most statistically significant results, 13 and 14, respectively. Correlations among the 5 endpoints at baseline ranged from.79 to >.99. Conclusions: Treatment comparisons differed across BMD site and endpoint used. Transforming via sample statistics provided similar results to transforming via reference or machine norms. RawBMD and the new approach (TPerc) may be more sensitive to change with TPerc having the added benefit of being readily interpretable as a percentage.Background Which bone mineral density (BMD) statistic is the “best” one to use? In our studies we have a variable amount of data per person, ranging from a single value to a series of BMD measures for various sites (spine femoral neck, total hip). There are numerous possible entities that could be used as the basis for analysis: Raw BMD = the raw BMD values T Raw = the observed t-score (presumably using a machine-based reference sample) T Samp = the transformed t-score using the sample mean and sample standard deviation T Ref = the transformed t-score using the mean and standard deviation from a reference population T Perc = the z-score percentile corresponding to the t-score for the reference population (new idea) Discussion SELECTING THE PROPER ENDPOINT IS IMPORTANT! SPINE (L TOTAL) appears most sensitive among the anatomical sites. CHANGE FROM BASELINE appears more sensitive than % change from baseline. RAW BMD and T PERC appear most sensitive among the five endpoints. T PERC has the added benefit of being readily interpretable as a percentage. Number of Significant Tests for Different BMD locations and estimates Summary Statistics -- Spine BMD Change from Baseline Methods Data for two clinical trials were used: N02C1 1 : A Phase III Randomized, Placebo- Controlled, Double-Blind Trial of Risedronate for Prevention of Bone Loss in Premenopausal Women Undergoing Chemotherapy for Primary Breast Carcinoma N03CC 2 : A Randomized, Controlled, Open- Label Trial of Empiric Prophylatic vs. Delayed Use of Zoledronic Acid for Prevention of Bone Loss in Post-menopausal Women with Breast Cancer Initiating Therapy with Letrozole After Tamoxifen 1)For each of the five study endpoints, change from baseline and percentage change from baseline were calculated using all available data for eligible patients for three anatomical sites: spine (L Total), femoral neck (Femur Neck), and total hip (Femur Total). 2)Summary statistics were calculated. 3)Student’s t-test, Wilcoxon rank-sum test, and analysis of covariance (baseline BMD value used as a covariate) were used to compare treatment arms for each study endpoint. 4)All analyses were completed separately for each clinical trial. References 1)N02C1 ASCO ref 2)N03CC ASCO ref