Dual Use of the METSSS Model Predicting Survival After Palliative Radiotherapy: An Exploratory Analysis

Introduction: The recently published METSSS model, which was developed for prediction of survival after palliative radiotherapy, includes age, sex, cancer type, localization of distant metastases, Charlson-Deyo comorbidity score and radiotherapy site. Its ability to predict other relevant endpoints has not been studied yet. Therefore, this exploratory study analyzed the endpoints “unplanned termination of radiotherapy” and “treatment in the last 30 days of life” in the METSSS-defined risk groups (low/medium/high). Methods: The risk group was assigned in the METSSS online calculator for our patient cohort with non-hematological malignancies treated between 2009 and 2014 during the first course of treatment (resembling details of the original METSSS study). All patients were treated with classical palliative dose/fractionation regimes such as five fractions of 4 Gy, 10 fractions of 3 Gy or 13 fractions of 3 Gy. No stereotactic high-dose radiation was utilized. Given that single-fraction radiotherapy cannot be discontinued, patients treated with 8 Gy x1 for uncomplicated painful bone metastases were excluded. Both completed and discontinued multi-fraction radiotherapy courses (at least two fractions intended) were included. Results: The study included 290 patients, 19 of whom failed to complete their prescribed course of palliative radiotherapy (7%). Thirty-nine (13%) were irradiated in the last 30 days of life. Only one patient was classified as low-risk according to the METSSS model (medium: 15, high: 274). Only Eastern Cooperative Oncology Group (ECOG) performance status (PS) was significantly associated with incomplete treatment. All 16 patients with low/medium METSSS risk scores completed their prescribed course of radiotherapy, compared to the 93% completion rate in the high-risk group, p=0.41. With regard to treatment in the last 30 days of life, ECOG PS, metastases to brain, liver and lung, and the number of prescribed fractions were statistically significant. One patient with a low/medium METSSS risk score was treated in the last 30 days of life (6%), compared to 14% in the high-risk group, p=0.49. Conclusion: Unexpected imbalances in the METSSS risk group size resulted in lower statistical power than anticipated. Patients with low/medium METSSS risk scores performed numerically better. However, other predictive factors, especially ECOG PS, which is not part of the METSSS model, maybe more relevant. Further efforts towards the application of the model beyond its original objective cannot be recommended.