Analytical Validation of Automated Bone Scan Index (BSI) as an Imaging Biomarker in Patients with Metastatic Prostate Cancer (mPC)

Objectives To assess linearity, precision, reproducibility and inter-observer variability of automated BSI in quantitative analysis of bone scan (BS) of patients (Pts) with mPC.

Methods To analytically validate BSI, we performed a series of simulation and clinical studies. Simulation study - The SIMIND Monte Carlo program and the XCAT phantom was used to simulate BS with pre-defined tumor burdens. To assess linearity, 50 phantoms were simulated with tumor burden ranging from low to high disease confluence, Phantom-BSI ranged from 0.10 to 13.0 BSI. To assess the precision, the second group of 50 phantoms was divided into 5 subgroups, each containing 10 phantoms at 0.5, 1.0, 3.0, 5.0 and 10.0 BSI, respectively. Clinical Study -To assess reproducibility, 52 Pts underwent two bone scans within four hours of single 600 MBq ^99mTc MDP injection. Finally, 173 mPC Pts and their two treatment follow-up BS were analyzed to compare the inter-observer variability of BSI reads against the standard manual subjective interpretation of change in BS. The automated BSI was calculated using the software EXINI bone^BSI. Pearson correlation, Cohen’s Kappa (K) measurement and descriptive statistics were employed to evaluate the results.

Results Pearson correlation to evaluate linearity of BSI, in the given range from 0.10 to 13.0, was observed to be 0.995 (N=50, 95% CI 0.99 - 0.99%, p <0.0001). The mean coefficient of variation indicating precision of BSI at each of the pre-defined tumor burden was observed to be less than 20%. The mean BSI difference between the two repeat BS of 52 Pts was 0.007 with confidence limit of +/- 0.20. Compared to the manual subjective assessment of change in BS of 173 Pts (K 0.72, p <0.0001), the inter-observer agreement in BSI was found to be more consistent (K 0.95, p <0.0001).

Conclusions Automated BSI is a consistent platform which can standardize the quantitative analysis of BS in multi-institutional clinical studies.