Assessment of the in vitro radiosensitivity of pancreatic cancer cells lines using two cell survival models

To assess the radiosensitivity of different pancreatic cancer (PC) cell lines and examine their differences regarding their sensitivity to radiation, and to quantify the inter-study variability. The parameters for the Linear Quadratic (LQ) model and a Nonlinear log-logistic regression (LL) model were derived using the data from clonogenic assays reported in the literature. In this study, the cell survival data from 108 clonogenic assays for four PC cell lines: Panc-1, BxPc-3, MiaPaCa-2 and AsPc-1, were fitted to the LQ model and the LL model performing bootstrapping. The lognormal, the generalized extreme value and the normal distributions were parametrized to describe the distribution of model parameters. The best representation was used to determine the median value of the parameters and their interquartile (IQR) ranges for each cell line. For the LQ model, no differences in the radiosensitivity between the four cell lines were found. A median α of 0.14 Gy -1 [IQR 0.08-0.25], β of 0.04 Gy -2 [IQR 0.03-0.06] and α/β of 3.2 Gy [IQR 1.6-6.7] were obtained. For the LL model, MiaPaCa-2 showed slightly higher radiosensitivity compared to BxPc-3 and AsPc-1 cell lines. The median value for all PC cell lines of φ2 was 2.44 Gy [IQR 1.99-3.09] and φ3 was 2.54 ]. The LQ model remains a robust method to assess radiosensitivity of cells, but due to its susceptibility to experimental heterogeneity, caution must be taken during the selection of a proper α/β value. The LL model can offer a more reliable method than the LQ model to capture radiobiological differences between cells' response to radiation. There is need for standardization of the clonogenic assays procedure, and a more thorough report of the setup used.