Variables associated with the development of complications from radiosurgery of intracranial tumors.

Between 5/21/86 and 11/1/89, we treated 64 recurrent or inoperable intracranial tumors in 60 patients (40 primary, 24 metastatic) with stereotactic radiosurgery using a modified 6 MeV linear accelerator at the Joint Center for Radiation Therapy. Patients were followed until death or 1/1/90. The median follow-up was 8 months (2-43 months). Fourteen patients experienced complications from 12 hours to 7 months (median 3 months, but only two patients more than 4 months) following radiosurgery. To determine variables related to complication, we calculated integral dose-volume histograms for 61/64 lesions and the surrounding CT-defined normal tissue. We excluded 16 lesions in 15 patients for follow-up less than 4 months (12 patients) or insufficient treatment information (3 patients). The variables for which higher values were associated with significantly more toxicity in a univariate score test were: a) tumor dose inhomogeneity (p less than 0.00001), b) maximum tumor dose (p = 0.00002), c) number of isocenters (p = 0.00002), d) maximum normal tissue dose (p = 0.00005) and e) tumor volume (p = 0.0001). These variables were all highly correlated with tumor dose inhomogeneity (coefficients of rank correlation 0.75-0.81). Tumor dose inhomogeneity had a much higher loglikelihood in a logistic model than any other single variable and a higher loglikelihood than any other two variables combined. None of the 21 patients with metastatic lesions experienced a complication. When we excluded the metastatic lesions, the above five variables remained significant in univariate tests. The mean tumor dose, number of treatment arcs, total degrees of arc, tumor location, previous radiotherapy, tumor geometry, pretreatment performance status, collimator size, and age were not significantly associated with toxicity. We conclude that radiosurgery of intracranial tumors is associated with a low risk of complications for lesions less than 10cc treated with a single isocenter to maximum tumor doses less than 25 Gy with tumor dose inhomogeneity less than 10 Gy, but that treatment of larger lesions will require new treatment strategies which reduce the tumor dose inhomogeneity associated with multiple isocenter treatments.