ESTRO 2024 - Abstract Book

S3688

Physics - Dose prediction, optimisation and applications of photon and electron planning

ESTRO 2024

2652

Poster Discussion

Improving consistency for re-irradiation dose constraints

Joep Stroom, Sandra Vieira, Carlo Greco

Champalimaud Foundation, Radiation Oncology, Lisbon, Portugal

Purpose/Objective:

Re-irradiation (reRT) of cancer patients is occurring more frequently and is expected to grow. OAR dose constraints for reRT are generally obtained from published reRT experiences often yielding cumulative EQD2 Dmax constraints for particular fractionation combinations. However, publications are few, heterogeneous, limited to Dmax (no dose-volume constraints), and recovery factors are generally lacking. In this paper we aim to improve the use of single course radiotherapy (scRT) constraints, which are better validated and include DV constraints, for reRT. We propose 1) to improve the consistency in dose constraints for different fractionation schemes, and 2) include a factor that allows for differences between scRT constraints and reRT constraints even when recovery is zero.

Material/Methods:

Firstly, in reRT the fractionation of the second course (nf2) frequently deviates from the first course (nf1). Delivered doses in nf1 must then be converted to doses in nf2 in order to determine the remaining dose tolerance of the OAR, usually via the LQ model with known α/β ratio (generally 2-3Gy). However, this only works if the dose constraints themselves fit the same model, i.e., if a dose constraint is met exactly in nf1, then, after conversion, it should also meet the dose constraint in nf2 exactly in order to be consistent. If not, large under- or overdoses can occur. We therefore compared the recent Timmerman constraints (2021), which for many OARs have the same critical volume for nf=1-30 fractions, with the LQ model. We also tried to improve the LQ fit by optimizing the α/β ratio. Secondly, to use scRT dose constraints (Dcs,scRT) in reRT scenarios, not only the recovery factor R must be known, but we introduce a reRT change-factor C. This factor represents the change in constraint without recovery, ie, due to changes in accepted NTCP (more risk can be taken for reRT) or in treatment accuracy (due to a different technique). This can be summarized as:

Dcs,reRT = C * Dcs,scRT * (1+R*P),

(1)

with P the fraction of Dcs,scRT delivered in the first course. In previous work we derived the relation between true dose constraints Dtrue (threshold dose above which toxicity will occur) and published dose constraints Dp (also used in treatment planning) as:

Dtrue = Dp +Φ -1 (1-NTCPp)×SD(Dp),

(2)

with Φ -1 the inverse cumulative normal distribution and SD(Dp) the variation in the published dose during the dose-effect study. This relation assumes that the x% highest doses cause the x% NTCP. In a different (reRT) situation with new NTCPn and/or SD(Dn), and with fixed Dtrue, it can be shown that