KEY POINTS
- The analysis included 27 MRI–ultrasound-guided biopsy cores from 15 patients with intermediate-risk prostate cancer treated with HDR brachytherapy monotherapy. Patients received 27 Gy in two fractions of 13.5 Gy, and biopsies were obtained during the second implantation after the first fraction.
- Each biopsy was reconstructed as connected 1-mm voxels and subjected to 10,000 Monte Carlo perturbations. Localization uncertainty was modelled using an isotropic three-dimensional Gaussian translation with σ = 1.25 mm, plus an axial offset representing possible tissue deficit within the biopsy needle.
- Nominal core-averaged dose was 29.0 ± 15.2 Gy, compared with a Monte Carlo median of 24.3 ± 7.5 Gy. At the voxel level, the mean absolute difference between the nominal assignment and an individual perturbed realization was 7.6 Gy, with a 95th percentile of 33.0 Gy.
- Nominal assignments were mildly biased upward: the probability that the nominal dose exceeded a localization-perturbed dose was 0.58. Discrepancies increased strongly with higher nominal dose and steeper local dose gradient, with Pearson correlations reaching approximately 0.89–0.95 and 0.73–0.83, respectively.
- Most biopsy DVH classifications were uncertain under perturbation. For D50% ≥27 Gy, only 1 of 27 biopsies was a confident pass, while 17 of 27 were borderline; for D2% ≥32 Gy, 19 of 26 biopsies were borderline.
- Logistic models estimated that a 95% probability of satisfying the illustrative biopsy thresholds required nominal margins of approximately 6.3 Gy for D98%, 20.8 Gy for D50%, 48.9 Gy for D2%, and 48.9 percentage points for V150%. These values were explicitly presented as robustness buffers, not recommended focal-boost increments.
- Mean absolute dose differences along individual biopsy cores approached approximately 10 Gy at separations of 1–1.5 cm, suggesting that dose becomes substantially decorrelated across the length of a typical core. The model did not account for anisotropic error, needle bending, patient-specific deformation, or inter-fraction anatomical change.
CLINICAL TAKEAWAY
A single nominal dose should not be treated as a reliable representation of the radiation exposure received by a small prostate biopsy core in a steep HDR dose gradient. Monte Carlo-propagated dose distributions may strengthen biopsy-based radiobiology and quality-assurance studies, but the reported margins are not clinical planning constraints.