KEY POINTS
- The retrospective study included 141 patients with locally advanced rectal cancer receiving long-course neoadjuvant chemoradiotherapy. IMRT delivered 50 Gy in 25 fractions to GTV and 45 Gy in 25 fractions to CTV, with capecitabine 825 mg/m² twice daily on radiation days.
- Pelvic bone marrow was delineated across the ilium, sacrum, pubis, ischium, and acetabular regions. Dose metrics were converted to EQD2 using α/β = 10 Gy.
- Grade 2 or higher hematologic adverse events occurred in 49 of 141 patients (34.8%). Grade ≥2 leukopenia occurred in 22.0%, neutropenia in 16.3%, haemoglobin reduction in 12.8%, and thrombocytopenia in 2.8%.
- Median pelvic marrow DmeanEQD2 was 22.83 Gy, while median V40EQD2 was 11.15%.
- In multivariable analyses, both DmeanEQD2 and V40EQD2 were significantly associated with grade ≥2 hematologic toxicity.
- NTCP modeling estimated a 50% toxicity probability at a mean dose of approximately 24.2 Gy and a V40EQD2 of 13.3%.
- ROC analysis produced an optimal DmeanEQD2 threshold of 24.09 Gy, with AUC 0.78 (95% CI 0.70–0.86), and V40EQD2 threshold of 13.0%, with AUC 0.83 (95% CI 0.76–0.90).
- The authors propose DmeanEQD2 ≤24.09 Gy and V40EQD2 ≤13% as potential marrow-sparing objectives. However, these thresholds were derived from the same retrospective single-centre population and underwent bootstrap internal validation rather than independent external testing.
CLINICAL TAKEAWAY
Pelvic bone marrow may deserve explicit consideration during long-course rectal chemoradiotherapy, particularly when marrow sparing can be achieved without compromising targets or bowel constraints. The proposed 24 Gy mean-dose and 13% V40 thresholds are useful candidates for prospective testing, not established clinical limits.