Cellular S-value of beta emitter radionuclide's determined using Geant4 Monte Carlo toolbox, comparison to MIRD S-values

Spatial dose distribution around the radionuclides sources is required for optimized treatment planning in radioimmunotherapy. At present, the main source of data for cellular dosimetry is the s-values provided by MIRD. However, the MIRD s-values have been calculated based on analytical formula in which no electrons straggling is taken to account. In this study, we used Geant4-DNA Monte Carlo toolbox to calculate s-values and the results were compared to the corresponding MIRD data. Similar to MIRD cell model, two concentric spheres representing the cell and its nucleus were used as the geometry of simulation. The cells were assumed to be made of water. Cellular s-values were calculated for three beta emitter radionuclides 131I, 90Y and 177Lu that are widely used in radioimmunotherapy. Few lines of code in C++ were added into Geant4-DNA codes to automatically calculate the s-values and transfer data into excel files. The differences between two series of data were analyzed using Pearson’s correlation and Bland-Altman curves. We observed high correlation (R2>0.99) between two series of data for self-absorption; however, the agreement was very weak and Wilcoxon signed rank test showed significant difference (p-value<0.001). In cross-absorption, Bland-Altman analysis showed a considerable bias between MIRD s-values and corresponding Geant4-DNA data. The percent differences between the data were -79% to +67%. Results of the comparison show a reflection of systematic error rather than statistical fluctuation. The inconsistency is most probably associated with the neglecting of straggling and δ-ray transport in MIRD analytical method.