Ambulatory capacity;Bayes regression;Lumbar spinal metastases;Mortality;Principal component analysis;Prognosis;Surgical intervention