KEY POINTS
- This study analyzed the radiotherapy quality assurance program of ROAM/EORTC-1308 (ISRCTN71502099), an international randomized phase III trial comparing postoperative 60 Gy in 30 fractions versus observation after gross total resection of WHO grade 2 atypical meningioma. The parent trial reported five-year disease-free survival of 79.9% with RT versus 64.3% with observation; the present publication separately evaluates protocol compliance and QA, rather than reanalyzing treatment efficacy. A total of 58 centers participated across the UK, continental Europe, Australia, and New Zealand.
- The trial required standardized target delineation based on both preoperative and postoperative MRI. The GTV included the postoperative resection cavity, clearly involved dural extensions, and hyperostotic bone, informed by the original tumor extent. The CTV used a 5-mm margin, extended to 10 mm along the dura and into brain where invasion was present, while the standard CTV-to-PTV margin was 5 mm, reducible to 3 mm with high-precision techniques. All trial irradiation used photons; proton therapy was not permitted.
- Pretrial benchmark credentialing identified relatively few cases requiring formal correction. Of 56 centers completing benchmark exercises, 52 (93%) had their initial target outlines approved, with 4 centers (7%) requiring contour resubmission. For treatment planning, 45 centers (80%) were approved initially, while 11 (20%) required resubmission. Nevertheless, minor deviations were frequent, including missing preoperative GTV outlines, inconsistent margin construction, incomplete optic pathway contours, and inadequately delineated distant organs at risk.
- Prospective individual case review revealed substantially more problems than benchmark credentialing. Of 78 patients randomized to radiotherapy, 66 received treatment across 28 centers, and 65 sets of treatment data were available for detailed QA analysis. 31/65 contour submissions (48%) required resubmission, including three cases requiring a second revision, while 12/65 treatment plans (18%) required resubmission. Investigators identified 28 major contouring variations and 14 major planning variations, indicating that initial institutional accreditation did not reliably predict subsequent protocol compliance.
- Inadequate dural coverage was the most frequent major contouring deviation, affecting 18/65 reviewed cases (28%). The main problems involved failure to include the medial, superior, or inferior extent of involved dura and insufficient CTV extension along dural attachments. Other major issues included inadequate GTV delineation (4 cases; 6%) and inappropriate inclusion of normal brain within the CTV (6 cases; 9%). Failure to submit or use the preoperative GTV was documented in 20 cases (31%), potentially obscuring disease extension that was no longer visible after surgery.
- Planning-related failures were less frequent but clinically relevant. Four plans (6%) had inadequate PTV coverage at the mandatory D98 threshold, and four (6%) had incorrect CTV-to-PTV margins, including inappropriate use of 3-mm margins or manual PTV cropping. Two plans (3%) exceeded mandatory OAR constraints without acceptable justification. Additional issues included poor conformity of the 95% isodose to the PTV and uncorrected auto-segmentation errors involving the optic apparatus or brainstem, emphasizing that automated OAR contours still required expert verification.
- Prospective QA enabled deviations to be corrected before treatment in most cases, with contour reviews generally returned within one week. However, four patients began RT before their outlines were submitted, and those plans were subsequently reviewed and accepted retrospectively. An independent post-trial reassessment of a random 10% sample found no disagreement with the original QA acceptance decisions. Experience also appeared relevant: 13/28 treating centers (46%) treated only one patient, while the two highest-accruing centers, responsible for approximately 24% of RT cases, had a lower resubmission rate (25% versus 48% overall).
CLINICAL TAKEAWAY
The ROAM/EORTC-1308 QA analysis demonstrates that pretrial credentialing alone cannot ensure consistent radiotherapy quality for uncommon CNS tumors, particularly when postoperative target delineation requires reconstruction of the original tumor extent and involved dura. Prospective patient-level contour and plan review can identify and correct potentially important deviations, strengthening confidence that outcomes from a multicenter randomized trial reflect the intended treatment protocol rather than uncontrolled variation in radiotherapy delivery.