Pain severity does not predict setup error during palliative radiotherapy for spinal metastases

In 376 spinal treatment sites, initial setup error was unrelated to pain severity but differed by spinal level and immobilization.

KEY POINTS

  • This secondary analysis of a prospective observational cohort (UMIN000044984) from Aichi Cancer Center, Japan, evaluated the relationship between pain severity and positioning accuracy during palliative radiotherapy for spinal metastases. The final analysis included 376 spinal treatment sites from 270 patients, treated between August 2021 and September 2023. Median age was 64.5 years, and median baseline pain score was 6 on the 0–10 Numerical Rating Scale (NRS). Pain was classified as mild (NRS 0–4; 132 sites, 35.1%), moderate (NRS 5–6; 83 sites, 22.1%) or severe (NRS 7–10; 161 sites, 42.8%).
  • All patients received three-dimensional conformal radiotherapy using conventional palliative fractionation. The most common prescriptions were 20 Gy in five fractions (40.2%), 8 Gy in one fraction (30.6%), 24 Gy in six fractions (16.2%) and 30 Gy in ten fractions (10.1%). Initial positioning was based on skin marks and room lasers, followed by pretreatment CBCT registration to planning CT using bony anatomy. Cervical spine lesions (66 sites, 17.6%) were immobilized with thermoplastic masks, whereas thoracic, lumbar and sacral lesions (310 sites, 82.4%) were treated without dedicated immobilization. Only first-fraction CBCT measurements were analyzed.
  • Baseline pain severity was not significantly associated with three-dimensional setup error. Median vector error was 0.48 cm (IQR 0.34–0.67) for mild pain, 0.54 cm (IQR 0.37–0.85) for moderate pain and 0.57 cm (IQR 0.37–0.77) for severe pain. Differences between groups were not statistically significant (Kruskal–Wallis p=0.134), and all pairwise comparisons remained nonsignificant after Bonferroni correction. Continuous NRS scores showed only a weak, nonsignificant correlation with setup error (Spearman r=0.088; p=0.089). The findings do not support increasing positioning margins solely because a patient reports severe pain.
  • Spinal level and associated immobilization method showed a stronger relationship with setup accuracy. Median three-dimensional vector error was 0.41 cm (IQR 0.30–0.65) for cervical treatments using thermoplastic masks, compared with 0.54 cm (IQR 0.37–0.76) for thoracolumbar and sacral treatments without immobilization (p=0.0037). Across all sites, median three-dimensional setup error was 0.52 cm (IQR 0.36–0.75), with mean error 0.63 ± 0.46 cm. Directional systematic offsets were small, but lateral variability was greater in unimmobilized thoracolumbar/sacral treatments, with a left–right standard deviation of 0.55 cm, compared with 0.33 cm in the cervical group.
  • Estimated setup margins were approximately 1.0 cm for cervical and 1.5 cm for thoracolumbar/sacral treatments. Using the |mean| + 2SD criterion, the largest directional requirement was 0.81 cm in the anterior–posterior direction for cervical treatments and 1.13 cm in the left–right direction for thoracolumbar/sacral treatments. Rounding these values upward to the nearest 0.5 cm produced the proposed isotropic margins of 1.0 and 1.5 cm, respectively. A more conservative |mean| + 3SD approach yielded maximum directional requirements of 1.18 and 1.68 cm. These estimates describe initial skin-mark positioning uncertainty before CBCT correction, not validated CTV-to-PTV margins for routinely image-guided treatment.
  • Other baseline clinical characteristics did not show significant independent associations with setup error. Neither age (p=0.903) nor body mass index (p=0.277) was significantly associated with vector displacement. Opioid use was associated with larger errors in an exploratory unadjusted analysis, but this association was no longer significant after adjustment for baseline pain score and spinal level. The investigators did not evaluate analgesic dose, administration timing, pathological fracture, or performance status, preventing conclusions about the influence of optimized pain control on positioning accuracy.
  • The analysis has important limitations for clinical margin selection. Cervical treatments always used masks, while thoracolumbar/sacral treatments never used dedicated immobilization, making spinal level and immobilization method inseparable. CBCT was obtained only at the first fraction, so random interfraction setup errors and intrafraction motion could not be assessed. Multiple treatment sites from individual patients were analyzed without adjustment for within-patient correlation, while patients unable to tolerate positioning may have been underrepresented. The proposed margins are therefore most relevant to initial setup without image-guided correction and should not be directly applied to contemporary daily-IGRT or spine SBRT workflows.

CLINICAL TAKEAWAY

Baseline pain severity was not significantly associated with initial positioning error during conventional palliative radiotherapy for spinal metastases, suggesting that pain scores alone should not determine setup margins. Cervical treatments using thermoplastic masks showed smaller errors than unimmobilized thoracolumbar/sacral treatments, although the independent effects of anatomy and immobilization could not be distinguished. The proposed 1.0- and 1.5-cm margins are preliminary estimates for uncorrected initial positioning, not validated expansions for daily image-guided treatment.

SOURCE

Technical Innovations & Patient Support in Radiation Oncology

Browse more research Suggest a correction