KEY POINTS
- This single-center retrospective analysis included 201 patients with IDH-wildtype glioblastoma treated with standard chemoradiation between 2012 and 2022 who subsequently developed radiologically confirmed recurrence. Median time from the final RT fraction to progression was 6.8 months.
- The investigators introduced RD80: the isotropic expansion from the original GTV needed to contain 80% of the subsequently recurrent tumor volume. The median RD80 was just 1.2 mm (IQR 0–6.0 mm).
- Most recurrence remained very close to the original disease. 80% of patients had an RD80 ≤10.4 mm, meaning a 10.4-mm expansion would have contained at least 80% of the eventual recurrence volume for four of every five patients.
- Conventional pattern-of-failure classification produced a similar picture: 175 of 201 recurrences (87.1%) were in-field, 6 (3.0%) marginal, and 20 (10.0%) distant relative to the clinical 95% isodose distribution.
- Recurrence distance was not uniform across tumors. Smaller GTV volume was strongly associated with greater recurrence distance (β −0.70, p=3.9×10⁻⁷), while involvement of the subventricular zone (β 0.53, p=0.019) and subgranular zone (β 0.51, p=0.032) predicted more distant recurrence.
- These variables are not yet good enough for personalized margins. The final model explained only 12% of the variation in recurrence distance, showing that GTV size and neurogenic-zone involvement capture only a small part of the biological and spatial behavior of glioblastoma.
- Importantly, 34% of the cohort had originally been treated with a 20-mm CTV margin rather than 15 mm. Margin size and fractionation did not significantly alter the model in sensitivity analyses, but the study did not actually replan or treat patients with a 10-mm margin; therefore, it cannot directly establish the safety of reducing current clinical margins.
CLINICAL TAKEAWAY
The geometry is provocative: most recurrent glioblastoma remained extremely close to the original GTV, providing a rationale for prospective trials of smaller CTV margins and potentially less normal-brain irradiation. It does not justify switching routinely to a 10-mm margin, because recurrence-distance analysis cannot show what would have happened had those patients actually been treated that way.