Palliative re-irradiation of spinal metastases. An independent validation of prognostic tools in terms of overall survival

Due to the constantly improving oncological therapy options, the survival time of patients with advanced tumour diseases is increasing. As a result, after palliative initial irradiation of spinal metastases, local recurrences appear more frequently, which were not clinically apparent in the past due to the patients' demise. Their radiooncological therapy, spinal re-irradiation, has been extensively investigated for possible side effects in the past decades and is considered sufficiently established. However, it is unclear how accurately existing prognostic methods can predict the survival of patients with an indication for palliative re-irradiation of recurrent vertebral metastases. For this reason, seven prognostic methods were identified, analysed and tested in the present study to determine which of these models could most accurately predict the remaining lifetime of the patient population considered. After applying the exclusion criteria, 44 cases of spinal re-irradiation were included in this retrospective analysis. The literature search identified five prognostic models for patients with spinal metastases. In addition, two further models were included in the study. On the one hand, the prognosis procedure developed for general re-irradiation according to Nieder, and on the other hand, the prognosis procedure established in palliative medicine according to Chow. The prognostic models were validated using the statistical methods PSEP, C-index and D-index. In addition, all parameters of the methods were individually tested for significance in the univariate analysis. The prognostic model according to Nieder showed the greatest predictive accuracy in all statistical tests. It also includes all parameters that correlated significantly with survival in the present patient collective in the univariate analysis (Karnofsky index, liver metastases, steroid therapy). Other factors, in particular the type of primary tumour, the number of bone metastases or the presence of pleural effusions, did not show any significant association with survival time after indication for re-radiation in the present patient group. Against this background, the Nieder method was modified to exclude the parameter pleural effusion. This showed a slightly improved numerical precision with regard to the estimation of the quality of life compared to the original Nieder method. However, this analysis is to be regarded as purely exploratory, as an internal validation was not carried out due to the small patient group and an external validation is still pending. Both Nieder models, the original and the modified, accurately classified long-term survivors with late relapses in the cohort studied into one of the best risk groups. Taking all results together, it can be concluded that the use of spine-specific prognostic tools often assigned patients to an incorrect prognostic group. Better results could be achieved with the general palliative instrument according to Chow et al. as well as the instrument developed specifically for re-irradiation by Nieder et al. In particular, Nieder's method, which was ultimately externally validated in this series, seems to be well suited for prognosis assessment before re-irradiation and can support physician and patient in the therapy decision.