Irradiated tumor volume as a predictor of local recurrence and radionecrosis in lung cancer with brain metastases treated with stereotactic radiosurgery.

Abstract

2024 Background: Stereotactic radiosurgery (SRS) is a standard local treatment for brain metastases (BM), but it may result in local recurrence (LR) or radionecrosis (RN). This study evaluates irradiated tumor volume as a predictor of LR and RN in SRS-treated lung cancer patients with BM. Methods: We retrospectively analyzed 431 lung cancer patients with BM who underwent SRS at Karolinska University Hospital, Sweden, (2009-2020), encompassing all-comers from the Stockholm region. Associations among irradiated tumor volume and risks of RN, symptomatic RN, as well as LR at 6 and 12 months, were assessed using Cox regression models. Furthermore, we evaluated the diagnostic performance of Methionine PET-CT in differentiating RN from LR. Results: 40 patients (9.2%) developed asymptomatic RN, 37 (8.3%) symptomatic RN, and 67 (15.5%) LR. Larger tumor volumes significantly increased the risks of RN and LR. At 6 months, a tumor volume of 4.75 cm³ was associated with an RN risk reaching the upper limit of 20%. By 12 months, substantially smaller volumes, such as 1.13 cm³, were related to same risk levels. Symptomatic RN followed a similar trend, with a volume of 13.58 cm³ presenting a risk of up to 20% at 6 months, while at 12 months, considerably smaller volumes, such as 3.8 cm³, corresponded to a symptomatic RN risk as high as 40%. 20% risk of LR was observed with volumes of 5.66 cm³ and 3.28 cm³ at 6 and 12 months, respectively. The sensitivity and specificity of Methionine PET-CT are 0.909 and 0.600 when MRI was considered the gold standard. Conclusions: Larger irradiated tumor volumes were positively correlated with an increased risk of both RN and LR. At 12 months post-SRS, smaller tumor volumes were associated with higher RN and LR risks in comparison with 6 months. Methionine PET-CT, when used alongside MRI, did not demonstrate a clear advantage in differentiating LR from RN.