Abstract Introduction: The median progression free survival (PFS) and overall survival (OS) of multiple myeloma (MM) patients have been prolonged due to novel agents combined with ASCT but the median OS in MM is still 7-8 years. Thus, the feasibility of new combinations and dosing of available agents must be investigated. The first proteasome inhibitor (PI), bortezomib (B), combined with elotuzumab and dexamethasone (d) showed superiority to Bd with PFS of 9.7 vs. 6.9 months, respectively, without excessive toxicity (Jakubowiak et al. Blood 2016;127:2833-40). In our study we investigate the safety, feasibility and initial efficacy of a second generation PI carfilzomib (K), SLAMF7 antibody elotuzumab (E) and dexamethasone (D) combination (KED) in relapsed or refractory MM (RRMM) patients. Patients and methods: Forty RRMM patients after 1-3 prior lines will be included in this phase 2 study after written informed consent. The primary endpoint is overall response rate (ORR). In patients achieving at least very good partial remission (VGPR) the quality of response will be assessed with high-sensitivity multicolour flow cytometry (MFC) according to the 8-colour EuroFlow protocol. Carfilzomib is given once weekly 20 mg/m2 on D1C1 and thereafter 70 mg/m2 in 28 day cycles on days 1, 8 and 15 in cycles 1-8 and on days 1 and 15 thereafter combined with weekly elotuzumab 10 mg/kg on days 1, 8 and 15 in cycles 1-2, thereafter on days 1 and 15; dexamethasone 40 mg on days 1, 8, 15 and 22 on cycles 1-8, thereafter on days 1 and 15. Treatment will continue until progression or excessive toxicity. Carfilzomib dose was 20/56 mg/m2 for the first two cycles for the first five patients to evaluate the safety. Additionally, patient samples collected prior to treatment will be comprehensively profiled by whole exome and RNA sequencing and evaluated for ex vivo response to the agents. Together, the study addresses clinical response, ex vivo-in vivo translation, identifies molecular biomarkers for the KED combination and facilitates precision guided clinical trials for RRMM. Results: By the end of July 2018, 11 patients have been enrolled. Median number of prior lines was 2 (1-3). Seven IgG-κ, two IgA-κ and one kappa and lambda light-chain patient are included. After a median of 6 (2-12) cycles ORR is 91% with 3 patients in VGPR (median MFC-MRD of 0,002%), 7 patients in PR, one in MR. Initial molecular characterization highlighted diverse subclonal backgrounds among the treated patients (Figure 1), but interestingly, the best responding VGPR patients displayed mutations to RAS genes in the dominant clones (NRAS 828, 2662; KRAS 733). At least PR was achieved after a median of 1 (1-4) cycle. Three patients have progressed and one patient withdrawn due to suspected thrombotic microangiopathy with manifestation of convulsions and pulmonary embolism. We noticed only one grade 2 infusion reaction after premedication. One patient developed autoimmune hemolytic anemia (AIHA), without red cell antibodies, suspected to be related to elotuzumab. She recovered with steroids and elotuzumab discontinuation and continued Kd without reappearance of AIHA. Another patient had grade 3 liver transaminase elevation but was able to continue treatment with dose reduction of carfilzomib and dexamethasone. Conclusion: To the best of our knowledge this is the first study evaluating the carfilzomib, elotuzumab and dexamethasone combination in RRMM with comprehensive molecular annotations. Among this small group of patients we noticed two unexpected severe adverse events; atypical AIHA and suspected thrombotic microangiopathy. AIHA should be excluded if unexpected anemia will appear during elotuzumab treatment. Preliminary results of this KED combination showed efficacy in patients with clonal RAS mutations and ORR of 91% after the median of two prior treatment lines using weekly carfilzomib dose of 70 mg/m2. Figure 1: Subclonal diversity in patient derived plasma cells. Clonal and subclonal fractions were evaluated by assessing peaks of variant allele frequencies of somatic mutations present in the myeloma plasma cells using kernel density estimation. For each plot, the fraction of cells carrying the mutation is represented on the x axis (1 = 100% of cells), and the probability density on the y axis. The mutated cell fraction (x axis) was calculated by adjusting mutant allele burden by copy number of the mutated loci. Disclosures Silvennoinen: Takeda: Honoraria, Research Funding; Celgene: Honoraria, Research Funding; BMS: Honoraria, Research Funding; Amgen: Honoraria, Research Funding. Tsallos:Novartis: Research Funding. Anttila:Amgen: Membership on an entity's Board of Directors or advisory committees; Janssen: Membership on an entity's Board of Directors or advisory committees; Takeda: Consultancy, Membership on an entity's Board of Directors or advisory committees. Räsänen:Amgen: Honoraria. Luoma:Amgen: Honoraria. Jantunen:Amgen: Honoraria; Genzyme/Sanofi: Honoraria; Takeda: Honoraria. Heckman:Celgene: Research Funding; Orion Pharma: Research Funding; Novartis: Research Funding.