Abstract While existing scoring systems provide physicians with methods of assessing prognosis of patients with MDS and help with the choice of appropriate treatment regimens, some patients defined as lower risk still have poor outcomes and tests which can refine current stratifications are of clinical interest. Serum samples, clinical and survival data were available from 106 MDS patients with refractory anemia (RA) (n = 46), RA with excess blasts (n = 29), RA with ring sideroblasts (n = 14), and chronic myelomonocytic leukemia (n = 17). The international prognostic scoring system (IPSS) score distribution 0/0.5/1/>1 was n = 36/23/24/12. Matrix assisted laser desorption/ionization (MALDI) mass spectra were acquired from the samples using the deep MALDI method1, allowing a deep probing of the proteome. The spectra were preprocessed and spectral features defined. The integrated intensities of these features were combined with the survival data using deep learning based machine learning techniques to create classifiers able to stratify patients into groups with better and worse survival. Patients in the good prognosis group had median survival of 52 months, compared with 23 months in the poor prognosis group (hazard ratio = 0.37, p < 0.001). Within the subgroup of patients with refractory anemia (n = 46) it was possible to identify a group of patients with no death events before 30 months, while median survival in the other patients was only 25 months. Multivariate analysis stratified by MDS subtype found that the mass spectral (MS) classification was a significant independent predictor of survival even when adjusted for international prognostic scoring system (IPSS) score and cytogenetic status. MS classification was able to separate low to intermediate risk (IPSS 0-1) and good karyotype subgroups into groups with significantly better and worse survival. If validated in an independent dataset, this serum test would provide complementary information for physicians to consider in addition to existing scoring systems to aid in planning treatment for MDS patients. 1. M Duncan et al, Proceedings of the 61st ASMS Conference on Mass Spectrometry and Allied Topics. Minneapolis, MN June 8-9, 2013 MP 181. Citation Format: Joanna Roder, Judith Löffler-Ragg, Reinhard Stauder, Ulrich Germing, Wolfgang R. Sperr, Peter Valent, Heinrich Roder, Arni Steingrimsson, Heinz Zwierzina. A serum protein test for improved prognostic stratification of patients with myelodysplastic syndrome (MDS). [abstract]. In: Proceedings of the 106th Annual Meeting of the American Association for Cancer Research; 2015 Apr 18-22; Philadelphia, PA. Philadelphia (PA): AACR; Cancer Res 2015;75(15 Suppl):Abstract nr 5304. doi:10.1158/1538-7445.AM2015-5304