KEY POINTS
- This two-center German dosimetric study evaluated 20 patients with 22 spinal metastatic target volumes enrolled in a prospective treatment protocol. Twelve targets involved thoracic vertebrae, eight lumbar vertebrae, and two cervical vertebrae. Patients had controlled extraspinal disease, Karnofsky performance status ≥50, stable vertebral anatomy (SINS 0–6), and limited epidural involvement (Bilsky 0–1). All lesions had partial vertebral involvement, predominantly with horseshoe-shaped geometries rather than complete circumferential targets.
- Simultaneous integrated boost (SIB) prescriptions were used for all targets. Most treatments (18/22, 81.8%) delivered 30 Gy to PTVHigh and 25 Gy to PTVLow in five fractions, while two targets received 27/21 Gy in three fractions and two received 24/18 Gy in two fractions. PTVHigh was generated using a 3-mm GTV expansion excluding the thecal sac, while PTVLow incorporated a 2-mm CTV expansion. A 1.5-mm spinal cord PRV was used, with D0.03cc constraints of 17.0, 20.3, and 25.3 Gy for two-, three-, and five-fraction schedules, respectively.
- Three independently generated treatment-plan sets were compared using identical patient contours and target definitions. Clinically used plans were generated with the general-purpose Monaco treatment planning system (TPSgen1), while the same cases were retrospectively optimized using Brainlab Elements Spine SRS (TPSded). A second experienced physicist at another center independently generated additional Monaco plans (TPSgen2) while blinded to the original plans. All systems used Monte Carlo-based dose calculation with a 2-mm dose grid, although optimization engines, beam models, and arc configurations differed.
- Target coverage was comparable between the dedicated and general-purpose systems. No significant differences were observed in PTVLow or PTVHigh D98% and D2%. Median PTVLow D98 differences between general-purpose and dedicated planning were +0.42 Gy for TPSgen1 (p=0.09) and +0.07 Gy for TPSgen2 (p=0.46), while PTVHigh D98 differences were +0.10 Gy (p=0.86) and +0.45 Gy (p=0.06), respectively. Although PTVLow mean dose was higher with the dedicated system, the overlapping high-dose boost region contributed to this metric, and the result does not independently demonstrate superior target coverage.
- The dedicated system consistently reduced near-maximum spinal cord dose. Compared with TPSgen1 and TPSgen2, median spinal cord D2% was lower by 2.66 Gy (p<0.01) and 2.63 Gy (p<0.001), respectively. Spinal cord mean dose was also lower by 1.05 and 0.87 Gy, respectively, while esophageal D2% decreased by 1.39 and 1.22 Gy. Among the 30 evaluated OAR and conformality DVH parameters, eight differed significantly for each general-purpose comparison, with seven favoring dedicated planning against TPSgen1 and all eight favoring dedicated planning against TPSgen2.
- Dose fall-off outside the target was more favorable with dedicated optimization. A common conformality evaluation structure was generated by expanding PTVLow by 2 cm to assess surrounding-tissue exposure. Its mean dose was lower with TPSded by a median of 1.63 Gy versus TPSgen1 and 2.60 Gy versus TPSgen2 (both p<0.001), suggesting steeper dose gradients outside the target. However, one comparison favored general-purpose planning: kidney mean dose was 0.39 Gy lower with TPSgen1 than with TPSded (p<0.01). These differences were primarily dosimetric, with no associated clinical toxicity assessment.
- Dedicated planning required fewer monitor units, but actual delivery efficiency was not measured. Median MU per prescribed Gy was 247.0 with TPSded, compared with 333.2 with TPSgen1 (p<0.001) and 268.3 with TPSgen2 (p=0.03). Dedicated plans consistently used three arcs, while the general-purpose plans used different arc numbers and gantry-angle restrictions. The analysis did not include measured delivery times, experimental end-to-end verification, independent cross-recalculation in a common TPS, or patient-specific treatment-delivery QA. Its small sample and platform-specific comparison limit generalizability.
CLINICAL TAKEAWAY
A dedicated spine SBRT planning system achieved comparable target coverage with improved spinal cord sparing and lower surrounding-tissue dose than two independently generated general-purpose plan sets, supporting the potential value of site-specific optimization for complex SIB treatments. However, these findings remain dosimetric, and differences in planning strategies, beam configurations, and dose calculation implementations may contribute to the observed advantages. Clinical safety, actual delivery efficiency, and reproducibility across other systems require independent validation.