Purpose: To analyze a prognostic score index for patients with brain metast
ases submitted to stereotactic radiosurgery (the Score Index for Radiosurge
ry in Brain Metastases [SIR]).
Methods and Materials: Actuarial survival of 65 brain metastases patients t
reated with radiosurgery between July 1993 and December 1997 was retrospect
ively analyzed. Prognostic factors included age, Karnofsky performance stat
us (KPS), extracranial disease status, number of brain lesions, largest bra
in lesion volume, lesions site, and receiving or not whole brain irradiatio
n. The SIR was obtained through summation of the previously noted first fiv
e prognostic factors. Kaplan-Meier actuarial survival curves for all progno
stic factors, SIR, and recursive partitioning analysis (RPA) (RTOG prognost
ic score) were calculated. Survival curves of subsets were compared by log-
rank test. Application of the Cox model was utilized to identify any correl
ation between prognostic factors, prognostic scores, and survival.
Results: Median overall survival from radiosurgery was 6.8 months. Utilizin
g univariate analysis, extracranial disease status, KPS, number of brain le
sions, largest brain lesion volume, RPA, and SIR were significantly correla
ted with prognosis. Median survival for the RPA classes 1, 2, and 3 was 20.
19 months, 7.75 months, and 3.38 months respectively (p = 0.0131). Median s
urvival for patients, grouped under SIR from 1 to 3, 4 to 7, and 8 to 10, w
as 2.91 months, 7.00 months, and 31.38 months respectively (p = 0.0001). Us
ing the Cox model, extracranial disease status and KPS demonstrated signifi
cant correlation with prognosis (p = 0.0001 and 0.0004 respectively). Multi
variate analysis also demonstrated significance for SIR and RPA when tested
individually (p = 0.0001 and 0.0040 respectively). Applying the Cox Model
to both SIR and RPA, only SIR reached independent significance (p = 0.0004)
.
Conclusions: Systemic disease status, KPS, SIR, and RPA are reliable progno
stic factors for patients with brain metastases submitted to radiosurgery.
Applying SIR and RPA classifications to our patients' data, SIR demonstrate
d better accuracy in predicting prognosis. SIR should be further tested wit
h larger patient accrual and for all patients with brain metastases subject
ed or not to stereotactic radiosurgery. (C) 2000 Elsevier Science Inc.