Integrating 2D dosimetry and cell survival analysis for predicting local effect in spatially fractionated radiotherapy

Abstract

Background and purpose: Robust methods for analysis and prediction of local cell survival after spatially fractionated radiotherapy (SFRT) in vitro remain limited. We present a methodology integrating spatial dosimetry with colony formation assessment and modelling to improve prediction of SFRT-induced responses.

Patient/material and methods: A549 lung cancer cells were irradiated with 220 kV X-rays in three field patterns: open, striped, and dotted. Colony centroid locations were mapped from scanned images of culture flasks. Dose distributions were measured using radiochromic film dosimetry. Digital images with colony locations and dose maps were divided into 1 mm² quadrats. A Poisson regression model was fitted to colony counts per quadrat, incorporating linear-quadratic (LQ) model parameters α and β. A modified LQ (MLQ) model included an additional interaction between dose and nearest distance to a peak region, with parameter δ.

Results: The methodology was successfully implemented. LQ fitting across all quadrats and patterns yielded α = 0.254 Gy⁻¹ and β = 0.039 Gy⁻², while the MLQ model gave α = 0.249 Gy⁻¹, β = 0.032 Gy⁻², and δ = −0.040 Gy⁻¹ cm⁻¹. Parameter uncertainty was below 0.5%. The MLQ model showed slightly lower fitting errors than the LQ model, indicating improved predictive accuracy.

Interpretation: We introduce a novel analysis pipeline for 2D localization of colonies and SFRT survival modelling in vitro. Findings suggest that distance to peak dose regions significantly influences local SFRT effects. Incorporating this spatial factor via an MLQ model may enhance understanding and prediction of SFRT-induced survival.