W. Butler, Renee R Stewart, G. Merrick
PURPOSE To re-evaluate prostate implant dosimetry using American Association of Physicists in Medicine Task Group 43 parameters and the radiobiologic approach of American Association of Physicists in Medicine Task Group 137. METHODS AND MATERIALS Among 1473 consecutive patients implanted with iodine-125 or palladium-103 sources before March 2006, there have been 55 biochemical failures. The dosimetric quality parameter, D(90), was updated according to the radionuclide and dosimetric era of the implant. For each patient, biologically equivalent dose (BED) and tumor control probability (TCP) were calculated from the updated implant D(90) plus any external beam dose using recommended indices and equations. RESULTS There was no significant difference in BED between biochemical failures and nonfailures, 148±27Gy and 145±24Gy, respectively (p=0.352). TCP was 0.90±0.26 for biochemical failures and 0.93±0.21 for nonfailures (p=0.414). Cox regression analysis found that neither BED nor TCP predicted for biochemical control either for the entire population or within each radionuclide-dependent dosimetric era. The only overall predictors of biochemical control were dosimetric era, Gleason score, and percent positive biopsies. Improvements in dosimetric quality over the first 300 patients were evident, but dosimetric era remained a better predictor of biochemical outcome than implant sequence number. CONCLUSION In a large prostate implant population, dosimetric and derived radiobiologic parameters did not predict for failure. Apparently, too few patients had total BEDs below the level necessary for optimum biochemical control. A learning curve extended over hundreds of patients before plateauing but changes in seed characterization parameters also had a pronounced effect.