P2049 BRAIN METASTASES FROM BREAST CANCER TREATED WITH
LINAC- RADIOSURGERY: PREDICTING FACTORS FOR DISTANT
RELAPSE
Brain metastases from breast cancer treated with Linac- Radiosurgery: predicting factors for distant relapse
Ch. BOSKOS, M.LEKEHAL, M. DUYME*, I. LAMPROGLOU,
G. BOISSERIE, Ph. CORNU, J.J. MAZERON, Ch. A. VALERY
Services de Neurochirurgie et de radiothérapie, Hôpital de la Pitié-Salpêtrière, (*)Unité de biostatistiques, épidémiologie et recherche clinique, EA 2415, Université de Montpellier, France
Introduction: For years, neurological control including distal control of brain metastases, hs
been achieved by whole brain radiation therapy (WBRT). Iterative Stereotactic Radiosurgery
(SRS) is now increasingly proposed in order to reach this goal while minimizing side-effects.
Prolonged survival in patients treated with SRS for brain metastases from breast cancer, is
favourable for observation of parameters involved with intracranial distal control.
Material and Methods : SRS was performed with a 6 MV VARIAN Clinac 2100C and a
BrainLab M3 micromultileaf collimator. Delivered doses were 16 Gy at the reference isodose
(6.6-17) and 20Gy at the isocenter (18-24).
112patients were treated with median delay 3.5 years after the primary diagnosis. Mean age
was 54 years (26-82) and mean Karnofsky index was 90 (60-100). Primary disease was
controlled in 87.5% of the cases. Theer were extra-CNS lesions in 78% of the cases.
According RPA patients were categorized: 14% RPA group I, 84% RPA group II and 2% group
III.
According SIR, patients were categorized: SIR 8-10 in 35% of the cases, SIR 4-7 in 65% of the
cases. Under previous surgery was 12% and under previous WBRT 50% of the cases. Mean
number of metastases was 1.89, median lesion volume was 1.5cc(0.008-44). RECIST criteria
were used to precise tumor response.
Results: Median survival time was 536 days. Median delay before distal relapse was 518
days. In a monovariate initialanalysis, age (p0.016), delay primary-brain met(0.05) and
number of lessions (p:0.004)were significant. Presence of WBRT was not significant (p:0.9).
In a multivariate Cox’s regression analysis using backward elimination approach, the best
model (chi2=23.2)predicting distal relapse included 5 parameters: age, primary control,
previous surgery, delay between primary and brain metastases, and number of metastases,
the latest being significant (p=0.007). A ROC analysis looking for a potential cut-off showed
that two lesions was the number above whom the probability of displaying new lesions was
increasing significantly [sensitivity(95%CI)=0.91 / specificity(95%CI)=0.25].
Conclusion: Initial number of lesions seems to be the only significant predictive parameter of
distal relapse. Patients with more than two lesions having significantly shorter delay before
relapse. Interestingly, WBRT does not influence delay before relapse.