KEY POINTS
- This retrospective single-institution pilot included 51 patients undergoing two thoracic RT courses, providing 102 treatment courses. Median age was 74 years, 34 patients had primary lung tumors, and most treatments were hypofractionated, with a median of 5 fractions per course.
- Investigators quantified image-based radiosensitivity, or iRS, from the slope of post-RT normal-lung CT density change across EQD2 dose bins. Lung receiving ≥10 Gy EQD2 from only one course was classified as single-irradiated, while overlapping lung receiving ≥10 Gy from both courses was analyzed separately as double-irradiated tissue.
- In newly irradiated lung, individual radiosensitivity was partly reproducible across treatment courses: iRS at the first and second RT courses correlated at r=0.34 (p=0.047). No comparable correlation was observed in double-irradiated lung (r=0.17; p=0.33).
- Treatment interval mattered specifically in previously irradiated tissue. An interval <9 months was strongly associated with greater change in iRS in double-irradiated lung (ρ=0.53; p=0.001), while no significant interval effect was seen in single-irradiated regions.
- Prior hypofractionation showed a similar spatial effect. Hypofractionated RT1 was associated with greater iRS change in double-irradiated lung (ρ=0.37; p=0.028), but not in lung receiving radiation for the first time.
- In multivariable analysis, an RT interval <9 months remained the only significant predictor of increased iRS in double-irradiated lung, with a coefficient of 4.50 HU/Gy3 (95% CI 0.30-8.69; p=0.04).
- The spatial analysis also raises a limitation of conventional cumulative DVH metrics. Greater overlap between courses can paradoxically reduce cumulative low-dose metrics such as lung V20 even though the overlapping tissue may be biologically more vulnerable, potentially concealing retreatment risk.
- Clinical validation is still missing. Only two patients developed grade 2 pneumonitis and none had grade ≥3 pneumonitis, so the investigators could not determine whether iRS actually predicts clinically relevant pulmonary toxicity. The retrospective, single-centre design and image-registration uncertainties further limit immediate use.
CLINICAL TAKEAWAY
Thoracic re-irradiation may eventually be personalized using how an individual patient's lung responded to the first RT course rather than relying exclusively on population-based cumulative constraints. The most interesting signal is the heightened response of overlapping lung when retreatment occurred within nine months, but iRS remains an experimental imaging biomarker and should not yet guide dose or recovery assumptions.
SOURCE
International Journal of Radiation Oncology, Biology, Physics