Page 121 - Predicting survival in patients with spinal bone metastasesL

Chapter IV, 1043 patients who were treated for symptomatic SBM were evaluated. Overall median survival of the entire cohort was short, with just 4.8 months (95%- CI 4.3-5.4). Patients were stratified according to the clinical profile of their primary tumor: favorable (e.g. breast), moderate (e.g. prostate) or unfavorable (e.g. lung), after which all risk factors that might be associated with survival were evaluated with a multivariate analysis. It was shown that survival in patients with SBM from a favorable clinical profile was influenced by the presence of visceral metastases and brain metastases, as well as the performance status. In patients with a moderate or unfavorable clinical profile, only performance status was associated with survival. Based on these results a flowchart was created, stratifying patients into four categories (A-D) with significantly different survival. Patients in category A had a median overall survival of 31.2 months (95%CI 25.2-37.3), compared to 15.4 months (95%CI 11.9-18.2) for patients in category B, 4.8 months (95%CI 4.1-5.4) for patients in category C and 1.6 months (95%CI 1.4-1.9) for patients in category D. Predictive accuracy of the model was assessed by using Harrell’s C-statistic,12 which was equal to 0.71, indicating reasonable discriminative accuracy. A C-index equal to one means that the model has perfect discrimination and a C-index equal to 0.5 means that the model predicts just as well as flipping a coin.
It has been shown that the primary tumor causing the SBM is the strongest predictor of survival. Patients with symptomatic SBM from breast cancer tend to live longer than patients with SBM from, for instance, lung cancer. This is reflected in all predictive models, where breast cancer is considered a favorable prognostic factor, compared to other primary cancers.
Based on estrogen receptor status (ER), progesterone receptor status (PR) and human epidermal growth factor receptor 2 status (HER2), several molecular phenotypes of breast cancer can be distinguished.13,14 It is known that these phenotypes are associated with survival when measured from the time of diagnosis of the primary tumor, however, it is unclear whether this effect remains when the disease has progressed up to the point of symptomatic SBM. If so, the manner in which our predictive model stratifies patients needs to be adjusted, distinguishing between different breast cancer phenotypes. In Chapter V we examined this effect by combining all breast cancer patients from our database, with all breast cancer patients from two external databases from Graz, Austria and Aarhus, Denmark. Based on the ER, PR and HER2 expression, tumors were further subdivided into
GENERAL DISCUSSION
IX 119