Topic: 16. Myeloproliferative neoplasms - Clinical Background: RUX was the first JAK1/2 inhibitor approved for MF patients (pts), with 70% of pts discontinuing within 5 yrs. Recently, the Response to RUX After 6 Months (RR6) model was developed for the early identification of RUX-treated MF pts with worse overall survival (OS). Aims: To investigate predictors of early Rux discontinuation and death while on therapy. Methods: The “RUX-MF” retrospective study includes 889 MF pts receiving RUX in real-life. This analysis focused on 410 pts including: 172 pts (19.3% of the total 889 cohort) who were alive on RUX after ≥5 yrs from RUX start (long-term RUX, LTR); 115 (12.9%) who were alive off RUX after ≥5 yrs (short-term RUX, STR); 123 (13.8%) who died while on RUX after <5 yrs (early death on RUX, EDR). Results: At baseline, LTR pts were more frequently affected by secondary MF (SMF) (57% vs 42.6%, p=0.02), had a lower incidence of thrombocytopenia (PLT<100x109/L, 3.5% vs 10%, p=0.03), anemia (Hb <10 g/dL, 20.4% vs 38.3%, p<0.001) and transfusion request (10.6% vs 22.8%, p=0.003) compared to STR pts. LTR had shorter time between MF diagnosis and RUX start (0.8 vs 2 yrs, p=0.01). ERD pts were older (73.7 vs 65.9/65.4 yrs, p<0.001), more frequently males (68.3% vs 50%/52.2%, p=0.005), with high DIPSS/MYSEC-PM (21.3% vs 4.6%/4.3%, p<0.001), WBC count (13.4 vs 11.2/10.4 x109/l, p=0.02) and transfusion requests (34.8% vs 10%/22.8%, p=0.003) compared to LTR/STR pts. Rates of IWG-MRT defined spleen response (SR) were comparable across LTR and STR at 3 and 6 mos (p=0.07 and p=0.29), while EDR pts had significantly lower rates of SR (18.4% vs 35.2%/24.2%, p=0.009 at 3 mos and 17.5% vs 39.5%/32.6% at 6 mos, p=0.001). Rates of treatment-emergent anemia and thrombocytopenia were comparable in LTR, STR and EDR pts. At last contact, 285 (69.5%) pts had discontinued RUX, 14 (8.7%) had a leukemic transformation and 226 (55.1%) had died. Compared to LTR, STR pts discontinued more frequently due to adverse events (42% vs 11.3% p<0.001) and loss/lack of response (44% vs 14.5%, p<0.001). Cumulative incidence of leukemic transformation with death as a competing event was similar in LTR and STR pts (p=0.08). OS was significantly longer in LTR patients (median survival not reached vs 6.5 yrs, p=0.002) (Fig.A). In multivariate analysis, PLT<100x109/L (p=0.04), Hb<10 g/dl (p=0.02), PMF (p=0.002), no SR at 3 mos (p=0.04) and RUX starting dose <10 mg BID (p=0.005) remained associated to higher probability of STR. Assigning 1 point to each statistically significant variable, a prognostic model for STR (STR-PM) was built and 3 groups were identified: low (score 0-1, 48.1% of the pts); intermediate (score 2, 36.2% of the pts); and high risk (score ≥3, 15.7% of the pts), with a 5-yr STR probability of 37.6%, 58.2% and 90.4%, respectively, p=0.001 (Fig.B). Summary/Conclusion: Long-term RUX therapy results in improved OS compared to earlier discontinuation and its probability is predicted by the STR-PM including RUX dose ≥10 mg BID, no cytopenia, SMF and SR at 3 mos. These results suggest RUX dose optimization and early evaluation of SR; cytopenic pts may be at higher risk of STR and considered for alternative front-line therapies. Overall, the STR-PM may help to identify those patients less likely to receive prolonged RUX monotherapy and, together with the RR6 score, guide differentiated treatment strategies. Also, early death on RUX was associated to high DIPSS/MYSEC risk and anemia; pts with these features may require a personalized approach beyond RUX monotherapy. Figure. OS adjusted for delayed entry according to STR/LTR phenotype (A). Probability of early RUX discontinuation based on the STR-PM (B)Keywords: Ruxolitinib, Prognosis, Myelofibrosis, Long-term follow-up