E Malinen1, S Freykowski, J Kummermehr. 1. Institute of Radiation Biology, GSF--National Research Centre for Environment and Health, Neuherberg, Germany.
Abstract
PURPOSE: To develop a method to analyze regrowth delay times, including cured tumors. METHODS AND MATERIALS: Regrowth delay measured in the in vitro squamous cell carcinoma megacolony system following fractionated irradiation was analyzed using two different approaches. A conventional regrowth delay analysis based on means or medians of dose groups was contrasted with a one-step procedure using the inverse of individual regrowth delay times, where cured tumors are represented by an infinite delay time. Two data sets with different proportions of cured tumors were compared in this way. In addition, for both approaches, confidence limits were also generated by the bootstrap technique. RESULTS: The parameter estimated was the alpha/beta ratio. The inverse analysis yielded similar results as the conventional analysis, but with much smaller confidence limits. When the dataset was stripped by stepwise removal of lower dose groups with no cures, the estimates of the alpha/beta ratio remained definitely more robust in inverse analysis. Results of the bootstrap procedure confirmed the accuracy of the calculated confidence intervals. CONCLUSION: Regrowth delay analysis using reciprocal regrowth delay times is a useful tool, especially when the data shows large variations, or a substantial fraction of the dataset falls into the curative dose region.
PURPOSE: To develop a method to analyze regrowth delay times, including cured tumors. METHODS AND MATERIALS: Regrowth delay measured in the in vitro squamous cell carcinoma megacolony system following fractionated irradiation was analyzed using two different approaches. A conventional regrowth delay analysis based on means or medians of dose groups was contrasted with a one-step procedure using the inverse of individual regrowth delay times, where cured tumors are represented by an infinite delay time. Two data sets with different proportions of cured tumors were compared in this way. In addition, for both approaches, confidence limits were also generated by the bootstrap technique. RESULTS: The parameter estimated was the alpha/beta ratio. The inverse analysis yielded similar results as the conventional analysis, but with much smaller confidence limits. When the dataset was stripped by stepwise removal of lower dose groups with no cures, the estimates of the alpha/beta ratio remained definitely more robust in inverse analysis. Results of the bootstrap procedure confirmed the accuracy of the calculated confidence intervals. CONCLUSION: Regrowth delay analysis using reciprocal regrowth delay times is a useful tool, especially when the data shows large variations, or a substantial fraction of the dataset falls into the curative dose region.