Recursive Partitioning Analysis Index Is Predictive for Overall Survival in Patients Undergoing Spine Stereotactic Body Radiation Therapy for Spinal Metastases

Purpose

To generate a prognostic index using recursive partitioning analysis (RPA) for patients undergoing spine stereotactic body radiation therapy (sSBRT) for spinal metastases (sMet).

Methods & Materials

From an institutional review board-approved database, 174 patients were treated for sMet with sSBRT between February 2006 and August 2009. Median dose was 14 Gy (range, 8–24 Gy), typically in a single fraction (range, 1–5). Kaplan-Meier analysis was performed to detect any correlation between survival and histology. Histologies were divided into favorable (breast and prostate), radioresistant (renal cell, melanoma and sarcoma), and other (all other histologies). RPA was performed to identify any association of the following variables with overall survival (OS) following sSBRT: histology, gender, age, Karnofsky performance status (KPS), control of primary, extraosseous metastases, time from primary diagnosis (TPD), dose of sSBRT (≤14 Gy vs. >14 Gy), extent of spine disease (epidural only, bone and epidural, bone only), upfront or salvage treatment, presence of paraspinal extension, and previous surgery.

Results

Median follow-up was 8.9 months. Median OS time from sSBRT was 10.7 months. Median OS intervals for favorable histologies were 14 months, 11.2 months for radioresistant histologies, and 7.3 months for other histologies (p = 0.02). RPA analysis resulted in three classes (p < 0.0001). Class 1 was defined as TPD of >30 months and KPS of >70; Class 2 was TPD of >30 months and KPS of ≤70 or a TPD of ≤30 months and age <70 years old; Class 3 was TPD of ≤30 months and age ≥70 years old. Median OS was 21.1 months for Class 1 (n = 59), 8.7 months for Class 2 (n = 104), and 2.4 months for Class 3 (n = 11).

Conclusion

sSBRT patients treated for sMet have a wide variability in OS. We developed an RPA classification system that is predictive of OS. While many patients are treated for palliation of pain or to avoid symptomatic progression, this index may be used to predict which patients may benefit most from sSBRT.