CyberKnife radiosurgery for brainstem metastases: Management and outcomes and a review of the literature

• This is the first paper to use recursive partitioning analysis (RPA) for brainstem metastasis (BSM) outcomes after CyberKnife radiosurgery (CKRS).
• Nine previous publications analyzed BSM patients using RPA but were mainly after Gamma Knife radiosurgery (GKS).
• 5 median single session equivalent dose (SSED), 13 less than SSED and 17 greater than SSED-treated BSM had LC in 4 (80%), 10 (77%) and 14 (82%).
• 54 RPA Class II/III BSM patients had a median OS of 5 months and for each Class, 8 and 2 months respectively.
• Univariate analysis showed that Karnofsky Performance Status, RPA Class and extracranial systemic disease-control predicted OS.

To our knowledge this paper is the first to use recursive partitioning analysis (RPA) for brainstem metastasis (BSM) patient outcomes, after CyberKnife radiosurgery (CKRS; Accuray, Sunnyvale, CA, USA); nine similar previous publications used mainly Gamma Knife radiosurgery (Elekta AB, Stockholm, Sweden). Retrospective chart reviews from 2006–2013 of 949 CKRS-treated brain metastasis patients showed 54 BSM patients (5.7%): 35 RPA Class II (65%) and 19 Class III (35%). There were 30 women (56%) and 24 men (44%). The median age was 59 years (range 36–80) and median follow-up was 5 months (range 1–52). Twenty-three patients (43%) had lung carcinoma BSM and 12 (22%) had breast cancer BSM. Fifty-four RPA Class II and III BSM patients had a median overall survival (OS) of 5 months, and for each Class 8 and 2 months, respectively. Of 36 RPA Class II and III patients with available symptoms (n = 31) and findings (n = 33), improvement/stability occurred in the majority for symptoms (86%) and findings (92%). Of 35 cases, 28 (80%) achieved BSM local control (LC); 13/14 with breast histology (93%) and 10/13 with lung histology (77%). All six RPA Class II and III patients with controlled extracranial systemic disease (ESD) experienced LC. Median tumor volume was 0.14 cm3; of 34 RPA Class II and III cases, 26 LC patients had a 0,13 cm3 median tumor volume while it was 0.27 cm3 in the eight local failures. Of 35 cases, single session equivalent dosages less than the median (n = 13), at the 17.9 Gy median (n = 5) and greater than the median (n = 17) had BSM LC in 10 (77%), four (80%) and 14 cases (82%), respectively. Univariate analysis showed Karnofsky Performance Score, RPA Class and ESD-control predicted OS. CKRS is useful for RPA Class II and III BSM patients with effective clinical and local BSM control.