KEY POINTS
- The retrospective study analyzed 150 fractions from 30 prostate cancer patients receiving five-fraction stereotactic body radiotherapy on a 0.35-T MRIdian system. Prescriptions were 36.25 Gy in 16 patients, 37.5 Gy in 12, and 40 Gy in two; planning margins were 3 mm posteriorly, 5 mm laterally and 4 mm elsewhere.
- Online dose re-optimization was actually performed in 90/150 fractions (60%), with a median of three adapted fractions per patient. Use increased from 40% at fraction 1 to 70% at fraction 5 (p=0.02).
- Reference plans achieved PTV V100% ≥95% in 148/150 fractions (98.7%). Without adaptation, median PTV V100% was 3.7 percentage points lower (p<0.01) and 65.3% of fractions fell below the 95% coverage criterion; every retrospectively or clinically adapted plan remained above 95%.
- The adaptive advantage increased across treatment: non-adapted plans showed PTV V100% approximately 1.6–5.5 percentage points lower than adapted plans across fractions, with a significant difference between fractions 1 and 5.
- Target volume progressively increased despite relatively stable position. Median PTV enlargement was about 1% during the first two fractions and 2.6–3.6% after fraction 3 (p<0.05), while Dice similarity remained ≥0.85 and mean distance-to-agreement did not change significantly.
- PTV expansion correlated directly with loss of target coverage when adaptation was omitted, with Pearson coefficients ranging from −0.65 to −0.37 across fractions. After re-optimization, that relationship was largely abolished (r −0.29 to 0.19).
- Several bladder and rectal parameters worsened without adaptation, but median values still remained inside predefined constraints. The findings therefore suggest that waiting for an organ-at-risk constraint violation alone may miss substantial degradation in target coverage.
CLINICAL TAKEAWAY
In five-fraction magnetic resonance-guided prostate treatment, target-volume expansion may be a more useful trigger for adaptation than waiting for an organ-at-risk limit to fail. A PTV V100% approaching or falling below 95% appears a practical warning signal, but the threshold requires validation outside this single-centre workflow.