Clinical Factors Associated with Radionecrosis Following Stereotactic Radiosurgery in the Era of Modern Systemic Therapy

Abstract

<h3>Purpose/Objective(s)</h3> Stereotactic radiosurgery (SRS) is commonly utilized in the treatment of brain metastases; however, radiation necrosis (RN) is a dose-limiting toxicity. Systemic therapy has greatly improved in recent years with the advent of immunotherapy and targeted therapy leading to increased overall survival. Here, we examined the impact of clinical factors on development of RN after SRS for brain metastasis in the era of modern systemic therapy. <h3>Materials/Methods</h3> We retrospectively reviewed patients treated with Linac-based SRS at a single institution from 2015-2020. Overall survival was calculated using the Kaplan-Meier method. Univariable (UVA) and multivariable (MVA) analysis was performed using Cox proportional hazards modeling. Covariates included age, sex, race, ethnicity, ECOG, primary cancer site, neurologic deficits at diagnosis, seizures, number/size of tumors, benign vs. malignant disease, prior systemic therapy, post-op vs. intact SRS, whole brain radiotherapy before or after SRS, edema at diagnosis, receipt and timing of immunotherapy/targeted therapy. Concurrent therapy was defined as within 30 days of SRS. <h3>Results</h3> 412 patients treated with SRS to 950 lesions were identified at a median follow-up of 8 months. Median age was 63 (interquartile range [IQR] 54-70) with 90% ECOG 0-1. At presentation, 53% had neurologic deficits and 7.6% had seizures. 82% were treated for malignant disease. 30% received post-op SRS. Prior to SRS, treatment included 1-9 lines of systemic therapy (58%), immunotherapy (24%), and WBRT (7.4%). SRS was given for 1-5 lesions in 91% of patients. Median tumor diameter was 2.0 cm (IQR 1.2-2.8). Median OS was 35 months (95% confidence interval [CI] 21-50); 1- and 2-year OS were 68% (95% CI 75-84) and 54% (95% CI 63-73). 17% developed radiographic RN with a median time to RN of 10 months (95% CI 9-13). Of 68 patients who developed RN, those with clinical symptoms (65%) were successfully treated with steroids (65%), laser interstitial thermal therapy (47%), bevacizumab (10%), and craniotomy (21%). On UVA, factors predictive of time to RN included post-op SRS (HR 1.84, 95% CI 1.12-3.00, p=0.015), number of treated lesions (HR 1.11, 95% CI 1.03-1.21, p=0.011), size of largest lesion (HR 1.46, 95% CI 1.17-1.81, p=0.001), malignant tumor (HR 4.17, 95% CI 1.49-11.11, p=0.006), prior systemic therapy (HR 1.7, 95% CI 1.1-2.86, p=0.046), and vasogenic edema at diagnosis (HR 1.97, 95% CI 1.19-3.26, p=0.008). On MVA, for every 1 cm increase in diameter of the largest lesion, the likelihood of developing RN was increased 1.45-fold (95% CI 1.12-1.88, p=0.005). Timing of systemic therapy did not impact RN rate. <h3>Conclusion</h3> Our data demonstrate that RN occurs in ∼17% of patients at a median of 10 months following SRS and can be successfully managed with medical and surgical therapies. The strongest predictor of RN was the size of the treated tumor. Our findings highlight the need for continued clinical surveillance as improved systemic therapy leads to increased long-term survival.