Establishment and validation of a novel survival prediction scoring algorithm for patients with non-small-cell lung cancer spinal metastasis.

Abstract

This study was to develop an algorithm capable of predicting the survival of patients with NSCLC spinal metastasis for individualized therapy. We identified 176 consecutive patients with NSCLC spinal metastasis between 2006 and 2017. Twenty-four features, including age, gender, smoking, KPS, paralysis, histological subtype, tumor stage, surgery, EGFR status, CEA, CA125, CA19-9, NSE, SCC, CYFRA21-1, calcium, AKP, albumin, the number of spinal, extra-spinal bone and visceral metastasis, time to metastasis, pathological fracture, and primary or secondary metastasis, were retrospectively analyzed. Features associated with survival in the multivariate analyses were included in a scoring model, which was prospectively validated in another 63 patients (NCT03363685). The median follow-up period was 12.00months (interquartile range 6.00-23.40months). One hundred forty-seven patients died during follow-up, with a median survival of 13.6months being observed. Multivariate analysis revealed that the following features were associated with survival: age, smoking, CA125, SCC, KPS, and EGFR status. A scoring system based on these features was created to stratify patients into low-risk (0-3), intermediate-risk (4-6) and high-risk (7-10) groups, whose estimated median survival times 29.10, 10.40 and 3.90months, respectively. The Harrell's c-index was 0.72. Model validation supported this model's validity and reproducibility. In patients with NSCLC spinal metastasis, survival was associated with age, smoking, CA125, SCC, KPS, and EGFR status. A validated scoring system based on these features was devised that can predict the survival times of those patients. This scoring system provides a basis for applying the NOMS framework and for facilitating individual treatment.