On the use of published radiobiological parameters and the evaluation of NTCP models regarding lung pneumonitis in clinical breast radiotherapy

Abstract

In this study we sought to evaluate and accent the importance of radiobiological parameter selection and implementation to the normal tissue complication probability (NTCP) models. The relative seriality (RS) and the Lyman-Kutcher-Burman (LKB) models were studied. For each model, a minimum and maximum set of radiobiological parameter sets was selected from the overall published sets applied in literature and a theoretical mean parameter set was computed. In order to investigate the potential model weaknesses in NTCP estimation and to point out the correct use of model parameters, these sets were used as input to the RS and the LKB model, estimating radiation induced complications for a group of 36 breast cancer patients treated with radiotherapy. The clinical endpoint examined was Radiation Pneumonitis. Each model was represented by a certain dose-response range when the selected parameter sets were applied. Comparing the models with their ranges, a large area of coincidence was revealed. If the parameter uncertainties (standard deviation) are included in the models, their area of coincidence might be enlarged, constraining even greater their predictive ability. The selection of the proper radiobiological parameter set for a given clinical endpoint is crucial. Published parameter values are not definite but should be accompanied by uncertainties, and one should be very careful when applying them to the NTCP models. Correct selection and proper implementation of published parameters provides a quite accurate fit of the NTCP models to the considered endpoint.

Developing of predictive models for pneumonitis with forward variable selection and LASSO logistic model for breast cancer patients treated with 3D-CRT

“…These models derived from dose distributions in the target/OAR volumes. According to the such models, IV20 [13], equivalent uniform dose (EUD), and the MLD, have been reported as predictive parameters for inducing pneumonitis [1,23]. Dosimetric and clinical factors (as the predictive parameters for OAR complications) in breast cancer patients treated with 3D-CRT have not been assessed and reported based on logistic regression LASSO model in literatures yet.…”

Parameters of the Lyman Model for Calculation of Normal-Tissue Complication Probability: A Systematic Literature Review

“…In any case, predictions using NTCP models need to be interpreted with great caution 41 . While some useful estimations of NTCP may be possible with the Lyman model, several reports have demonstrated its failures.…”

Prescription Value-Based Automatic Optimization of Importance Factors in Inverse Planning

“…NTCP-based physical-biological objective function. NTCP-based optimization has been investigated in inverse treatment planning [42][43][44][45][46] and incorporated into some commercial treatment planning software. 44,47 The NTCP-based objective function was expressed in Equation 13.…”