KEY POINTS
- This pooled analysis included 68 patients with locally advanced cervical cancer from single-institution EMBRACE-II and HYPOCx-iRex cohorts. Patients received 45 Gy in 25 fractions without marrow sparing (EMBRACE-II, n=28), 45 Gy in 25 fractions with marrow sparing (CVRT, n=19), or 44 Gy in 20 fractions with marrow sparing (HYPO, n=21), alongside concurrent chemoradiotherapy and image-guided brachytherapy.
- Two bone marrow delineation approaches were evaluated. CT-based marrow (BM_CT) included pelvic and lumbosacral bony structures within the treatment region, whereas MRI-based marrow (BM_MRI) used iso- to hypointense T1/T2 marrow as a surrogate for active hematopoietic tissue; the femora were excluded from the primary marrow volumes.
- Acute grade ≥2 composite hematologic toxicity, excluding lymphopenia, occurred in 79% of EMBRACE-II, 68% of CVRT and 67% of HYPO patients (p=0.62). Grade ≥3 toxicity occurred in 18%, 11% and 24%, respectively (p=0.61), indicating no statistically significant difference between groups despite lower planned marrow exposure in the sparing cohorts.
- Across the complete population, CT-based marrow dose showed significant associations with grade ≥3 hematologic toxicity. Probit-derived ED10 values were approximately V20Gy 74%, V25Gy 480 cc, V30Gy 360 cc and V35Gy 235 cc, where ED10 represented modeled dose-volume levels associated with a 10% toxicity probability.
- In the 44 Gy/20-fraction subgroup, MRI-defined active marrow metrics were specifically associated with grade ≥2 hematologic toxicity, with ED10 values of V25Gy 39% and mean dose 24 Gy. The authors proposed more practically achievable hypofractionation objectives of BM_MRI V25Gy <45% and Dmean <25 Gy.
- For broader clinical planning, the authors proposed CT-based marrow constraints of V20Gy <70%, V25Gy <450 cc, V30Gy <300 cc and V35Gy <200 cc. These were derived from the observed dose–toxicity relationships and suggested for incorporation into future prospective protocols, not validated as universal clinical safety limits.
- Severe lymphopenia remained essentially universal despite marrow sparing: 67 of 68 patients developed grade ≥3 lymphopenia. The small pooled sample, heterogeneous fractionation, lack of PET-confirmed active marrow, and multiple exploratory dose-volume comparisons limit confidence in the proposed thresholds and their ability to reduce clinical hematologic events.
CLINICAL TAKEAWAY
CT-based pelvic marrow constraints may offer a practical planning strategy for conventional and hypofractionated cervical cancer chemoradiotherapy, while MRI-derived active marrow metrics may be particularly relevant for the 44 Gy/20-fraction schedule. However, the proposed thresholds should be treated as research planning objectives rather than established constraints, especially because the observed between-group differences in hematologic toxicity were not statistically significant.