KEY POINTS
- This retrospective single-center study from Beth Israel Deaconess Medical Center, Boston, evaluated 20 consecutive patients with high-risk prostate adenocarcinoma treated with MRI-guided intraprostatic boost SBRT between April 2023 and May 2024. All patients had multiparametric MRI and PSMA-PET/CT, with 13 receiving a two-fraction SBRT boost and seven receiving five-fraction SBRT. All tumors were ISUP grade group 4–5, and three patients (15%) had cN1 disease.
- Two focal boost strategies were evaluated. Seven patients received prostate SBRT prescribing 36.25 Gy in five fractions, while 13 received 19 Gy in two fractions to the prostate and proximal seminal vesicles followed by 45 Gy pelvic irradiation. All received androgen deprivation therapy. MRI-defined dominant intraprostatic lesions were used for clinical CyberKnife planning, while PSMA-PET lesions were retrospectively segmented using SUV thresholds and manual refinement.
- PSMA-PET identified significantly larger intraprostatic lesions than MRI. Median lesion volume was 2.17 cm³ with PSMA-PET versus 1.45 cm³ with MRI (p=0.04), an absolute difference of 0.72 cm³. Lesion volumes were positively correlated (Spearman ρ=0.741; p<0.01), but two patients had no identifiable PET-defined nodule despite MRI-visible disease.
- Spatial agreement between MRI and PSMA-PET was limited. Among the 18 patients with lesions identifiable on both modalities, median Dice similarity was 0.31 (range 0–0.60), with a median centroid distance of 7.36 mm (range 2.88–33.06 mm). Seven patients (35% of the full cohort; 7/18 evaluable cases) met the predefined discordance criteria of Dice <0.2 and centroid displacement >3 mm. Four patients had no volumetric overlap between the two delineated lesions.
- MRI-based boost planning delivered significantly lower D95% to PET-defined targets. In the five-fraction cohort, median D95% was 45.3 Gy for MRI-defined lesions versus 41.6 Gy for PET-defined lesions, a difference of 3.7 Gy (p<0.001). In the two-fraction cohort, corresponding values were 23.4 versus 21.7 Gy, a difference of 1.7 Gy (p<0.001). These comparisons were performed using the original clinical plans, without reoptimization to cover the PET-defined volumes.
- The dosimetric discrepancy persisted in anatomically discordant cases. MRI-defined nodules received a median 125% of the intended dose (IQR 124–126%), compared with 115% (IQR 111–116%) for PET-defined nodules. Although this represents reduced relative dose escalation to PET-defined disease, it does not demonstrate that these lesions received insufficient therapeutic dose or that clinically significant tumor was missed.
- The study did not establish which imaging modality more accurately identified cancer. No whole-mount histopathological reference, clinical recurrence analysis, or toxicity comparison was available. PSMA-PET was retrospectively registered using pelvic soft-tissue anatomy, whereas planning MRI was registered using fiducials, introducing differences in registration uncertainty. The small cohort, pre-ADT PET acquisition, imaging interval, and subjective segmentation further limit generalizability.
CLINICAL TAKEAWAY
MRI and PSMA-PET can identify substantially different intraprostatic boost volumes in high-risk prostate cancer, with MRI-only focal SBRT planning delivering lower relative dose coverage to PET-defined lesions. These findings support investigating combined MRI–PSMA-PET target delineation, particularly for spatially discordant disease. However, without pathological validation or clinical outcomes, the study cannot establish that PET-guided boosting improves tumor control or should replace MRI-based planning.