A Novel TCP and NTCP Modeling Framework Incorporating Inter-Fractional Deviations and Biological Effectiveness for...

High-dose hyper-fractionated radiotherapy has been recently included in NCCN guidelines for limited-stage small cell lung cancer (LS-SCLC), which clinical outcomes are dependent on actually delivered biologically effective dose (BED). However, most tumor control probability (TCP) and normal tissue complication probability (NTCP) models rely on planed physical dose without considering inter-fractional deviations and BED of fractionization. This study aims to develop TCP and NTCP models for LS-SCLC incorporating CBCT-based inter-fractional anatomies and hyper-fractional radiobiological dose.

Inter-fractional dose was accumulated based on deformable image registration between CBCT and planning CT (pCT). Both TCP and NTCP were formulated within a BED-based framework, using dose-volume parameters calculated from planned dose and CBCT-based accumulated dose for comparison. The LQ model with "4R" effects and Gompertzian tumor growth (LQRG) and Lyman-Kutcher-Burman (LKB) models were used for TCP and NTCP modeling. Based on the receiver operating characteristic (ROC) curve (AUC), negative log-likelihood (NLL), and Akaike information criterion (AIC), model performances were assessed for local progression-free survival (LPFS), radiation pneumonitis and radiation esophagitis.

Compared with pCT dose, CBCT dose showed superior discrimination of normal tissue toxicity. For grade ≥2 radiation pneumonitis, the CBCT-based NTCP model achieved significantly higher AUC (0.67 vs. 0.59) and lower NLL (62.39 vs. 66.13) than that of the pCT-based mode. Superiority of CBCT-based NTCP model was also observed for grade ≥3 radiation esophagitis, i.e., AUC (0.62 vs. 0.58) and NLL (55.55 vs. 56.64). Using the observed LPFS as reference, marginally improved TCP modeling was observed using CBCT dose than that of pCT, where the AUC values were 0.80 vs. 0.79 (GTV 45 Gy) and 0.77 vs. 0.75 (GTV 54 Gy).

By incorporating inter-fractional deviation and biological effectiveness, the improved accuracy of TCP and NTCP models provides preliminary evidence supporting the treatment planning and toxicity management for LS-SCLC patients treated with the hyper-fractionated radiotherapy, which needs further clinical validation.