Importance of MR technique for stereotactic radiosurgery.

We investigated how frequently the imaging procedure we use immediately prior to radiosurgery--triple-dose gadolinium-enhanced MR performed with the patient immobilized in a nonrelocatable head frame and 1-mm-thick MPRAGE (magnetization-prepared rapid gradient echo) images (SRS3xGado)-identifies previously unrecognized cerebral metastases in patients initially imaged by conventional MR with single-dose gadolinium (1xGado). Between July 1998 and July 2000, the diagnoses established for 47 patients who underwent radio-surgical procedures for treatment of cerebral metastases at The Gamma Knife Center of New York University were based initially on the 1xGado protocol. In July 1998, we began using SRS3xGado as our routine imaging protocol in preparation for targeting lesions for radio-surgery, using triple-dose gadolinium and acquisition of contiguous 1-mm Tl-weighted axial images. Because our SRS3xGado scans sometimes unexpectedly revealed additional metastases, we sought to learn how frequently the initial 1xGado scans would underestimate the number of metastases. We therefore reviewed the number of brain metastases identified on the SRS3xGado studies and compared the results to the number found by the 1xGado protocol, which had initially identified the brain metastases. Additional metastases, ranging from 1 to 23 lesions per patient, were identified on the SRS3xGado scan in 23 of 47 patients (49%). In 57% of the 23 patients, only one additional lesion was identified. The mean time interval between the 1xGado and the SRS3xGado scans was 20.6 days (range, 4-83 days), and the number of additional lesions detected and the time interval between two scans were negatively correlated (-0.11). The number of lesions detected on the SRS3xGado was associated only with the number of lesions on the 1xGado and not with any other patient or tumor pretreatment characteristics such as age, gender, largest tumor volume on the 1xGado, or number of days between the 1xGado and the SRS3xGado or prior surgery. The identification of additional lesions with SRS3xGado MR may have implications for patients who are treated with stereotactic radiosurgery alone (without whole-brain irradiation) with single-dose gadolinium imaging, in that unidentified lesions may go untreated. As a result of these findings we continue to use and advocate SRS3xGado scans for radiosurgery.

Long-term outcomes after staged-volume stereotactic radiosurgery for large arteriovenous malformations.

BACKGROUND: Stereotactic radiosurgery is an effective treatment modality for small arteriovenous malformations (AVMs) of the brain. For larger AVMs, the treatment dose is often lowered to reduce potential complications, but this decreases the likelihood of cure. One strategy is to divide large AVMs into smaller anatomic volumes and treat each volume separately. OBJECTIVE: To prospectively assess the long-term efficacy and complications associated with staged-volume radiosurgical treatment of large, symptomatic AVMs. METHODS: Eighteen patients with AVMs larger than 15 mL underwent prospective staged-volume radiosurgery over a 13-year period. The median AVM volume was 22.9 mL (range, 15.7-50 mL). Separate anatomic volumes were irradiated at 3- to 9-month intervals (median volume, 10.9 mL; range, 5.3-13.4 mL; median marginal dose, 15 Gy; range, 15-17 Gy). The AVM was divided into 2 volumes in 10 patients, 3 volumes in 5 patients, and 4 volumes in 3 patients. Seven patients underwent retreatment for residual disease. RESULTS: Actuarial rates of complete angiographic occlusion were 29% and 89% at 5 and 10 years. Five patients (27.8%) had a hemorrhage after radiosurgery. Kaplan-Meier analysis of cumulative hemorrhage rates after treatment were 12%, 18%, 31%, and 31% at 2, 3, 5, and 10 years, respectively. One patient died after a hemorrhage (5.6%). CONCLUSION: Staged-volume radiosurgery for AVMs larger than 15 mL is a viable treatment strategy. The long-term occlusion rate is high, whereas the radiation-related complication rate is low. Hemorrhage during the lag period remains the greatest source of morbidity and mortality.

Results of surgical resection for progression of brain metastases previously treated by gamma knife radiosurgery.

OBJECTIVE: To determine treatment outcome after surgical resection for progressive brain metastases after gamma knife radiosurgery (GKR) and to explore the role of dynamic contrast agent-enhanced perfusion magnetic resonance imaging (MRI) and proton spectroscopic MRI studies (MRS/P) in predicting pathological findings. METHODS: Between 1997 and 2002, 32 patients underwent surgical resection for suspected progression of brain metastases from a cohort of 245 patients with brain metastases treated with GKR. Postradiosurgery MRI surveillance was performed at 6 and 12 weeks, and then every 12 weeks after GKR. In some cases, additional MRI scanning with spectroscopy or perfusion (MRS/P) was used to aid differentiation of radiation change from tumor progression. The decision to perform neurosurgical resection was based on MRI or clinical evidence of lesion progression among patients with a Karnofsky performance score of 60 or more and absent or stable systemic disease. RESULTS: Thirteen percent (32 out of 245) of patients and 6% (38 out of 611) of lesions required surgical resection after GKR. The median time from GKR to surgical resection was 8.6 months (range, 1.7-27.1 mo). The 6-, 12-, and 24-month actuarial survival from time of GKR was 97, 78, and 47% for the resected patients and 65, 40, and 19% for the nonresected patients (P < 0.0001). The two-year survival rate of patients requiring two resections after GKR was 100% compared with 39% for patients undergoing one resection (P = 0.02). The median survival of resected patients was 27.2 months (range, 7.0-72.5 mo) from the diagnosis of brain metastases, 19.9 months (range, 5.0-60.7 mo) from GKR, and 8.9 months (range, 0.2-53.1 mo) from surgical resection. Tumor was found in 90% of resected specimens and necrosis alone in 10%. MRS/P studies were performed in 15 resected patients. Overall, MRS/P predicted tumor in 11 lesions, confirmed pathologically in nine lesions, and necrosis alone was found in two. The MRS/P predicted necrosis alone in three, whereas pathology revealed viable tumor in two and necrosis in one lesion. CONCLUSION: Surgical intervention of progressive brain metastases after GKR in selected patients leads to a meaningful improvement in survival rates. Further studies are necessary to determine the role of MRS/P in the postradiosurgery surveillance of brain metastases.