KEY POINTS
- STEP (NCT04503772) was a prospective, open-label, single-arm phase II trial conducted at six hospitals in France between February 2021 and April 2025. Among 73 patients screened, 67 were eligible, and 63 completed both preoperative stereotactic radiotherapy (preopSRT) and surgical resection, forming the modified intention-to-treat population for local control and toxicity analysis. Eligible patients had Karnofsky performance status ≥70, up to four brain metastases, and at least one resectable lesion measuring ≤5 cm. Median age was 62 years, mean maximum lesion diameter was 32.5 mm, and 28.6% of treated lesions exceeded 4 cm.
- Treatment consisted of three fractions of 7.7 Gy prescribed to the encompassing 70% isodose, corresponding to 23.1 Gy at the PTV periphery and 33 Gy at the isocenter. The GTV was delineated on contrast-enhanced MRI with a 2-mm isotropic PTV margin, and fractions were intended to be delivered on alternate days. Median preoperative RT duration was five calendar days, followed by surgery after a median of two days. Resection occurred within the specified three-day interval in 59/63 patients (93.6%), although approximately 22% received at least two fractions on consecutive days because of scheduling constraints.
- The primary endpoint of six-month local control was achieved. No local recurrence occurred during the first six months, yielding 100% local control (95% CI 95–100%) on centrally reviewed MRI, exceeding the predefined efficacy benchmark of 86%. Kaplan–Meier local control remained 98.2% at nine months and 96.4% at 12 months (95% CI 87–99%). Gross total resection was achieved in 60/63 patients (95.2%). These outcomes are encouraging, but the single-arm design prevents direct comparison with contemporary postoperative SRT.
- The one-year cumulative incidence of index-lesion radiation necrosis was 7.0% (95% CI 2–16%), accounting for death as a competing risk and meeting the key secondary endpoint. Four patients developed radiologically defined RN at 4.7, 10.6, 11.6, and 11.7 months after treatment. Three events were asymptomatic grade 1, and one was symptomatic grade 2. The corresponding Kaplan–Meier estimate was 8.1%, illustrating the effect of the statistical method. Importantly, no leptomeningeal disease was detected during the reported follow-up, although the limited sample size and observation period cannot exclude later events.
- Distant intracranial progression remained frequent despite excellent index-lesion control. New or progressive brain metastases outside the resected site occurred in 22/63 patients (35%), at a median interval of six months after treatment. Most patients received additional SRT or systemic treatment changes, while one underwent whole-brain radiotherapy. Estimated 12-month overall survival was 81.2% (95% CI 70–89%), with median survival not reached. The one-year probability of remaining free from the composite endpoint of local progression, radiation necrosis, or leptomeningeal disease was 89.8% when death was treated as a competing event.
- Postoperative cerebral complications occurred in 4/63 patients (6.3%), comprising two surgical site infections, one acute hydrocephalus, and one wound complication; all four were successfully managed. Additional serious events included one grade 4 bowel obstruction and one fatal pulmonary embolism, attributed to surgery and/or underlying disease rather than specifically to preoperative radiotherapy. Of the four eligible patients who did not complete both treatments, three did not receive preopSRT because of progression detected on planning MRI, while one deteriorated after irradiation, did not undergo surgery, and died from disease progression. These events highlight the importance of patient selection and the need for reliable coordination between RT and surgery.
- Neurocognitive function and quality of life remained broadly stable during the 12-month protocol follow-up. Mean Mini-Mental State Examination scores remained approximately 28 points at baseline and 12 months (p=0.8), while the EORTC QLQ-C30 summary score remained around 75 points (p=0.6). Clinically meaningful cognitive deterioration of at least three MMSE points occurred in 11.9% at 12 months. Nevertheless, median follow-up was only 12.7 months, and the study lacked a randomized postoperative SRT comparator, limiting conclusions about long-term neurocognitive protection, delayed radionecrosis, and comparative effectiveness.
CLINICAL TAKEAWAY
Three-fraction preoperative stereotactic radiotherapy followed by surgery achieved excellent early local control with a low observed incidence of radiation necrosis in selected patients with resectable brain metastases. The absence of leptomeningeal disease and preservation of neurocognitive outcomes are encouraging, particularly for larger lesions, but the nonrandomized design and short follow-up prevent conclusions about superiority over postoperative SRT. Ongoing randomized trials are needed before preoperative treatment can replace established postoperative strategies.