The Utility Of Newer Risk Models In Predicting Outcomes Of Patients (pts) With Higher-Risk (HR) Myelodysplastic Syndromes (MDS) Treated With Azactidine (aza)

Background While aza is the only drug shown to prolong survival in MDS, only half of aza-treated pts achieve objective responses (10-20% complete remission), and 4-6 months might be needed before a response is seen. Therefore the ability to select pts with high and low likelihoods of benefit from aza therapy is a clinical and a research priority. No clinical or laboratory parameter consistently predicts response or survival with aza therapy. A French prognostic scoring system (FPSS) was proposed to predict survival among aza-treated pts with HR-MDS. We sought to compare the relative prognostic discriminatory power of the FPSS with that of the revised IPSS (IPSS-R) and the global MD Anderson prognostic scoring system (MDAPSS) in a large cohort of aza-treated pts with IPSS HR-MDS. Methods The MDS database at Moffitt Cancer Center (MCC) was used to identify patients with HR-MDS (International Prognostic Scoring System [IPSS] intermediate-2 [INT-2] and high-risk) who received aza therapy. Kaplan-Meier curves were used to depict survivals, and the log-rank test was used to compare median overall survival (OS). Akaike information criteria (AIC) were used to assess the relative goodness of fit of the models. Results We identified 259 patients HR-MDS (74.1% with INT-2 and 25.9% with high IPSS) treated with aza at the MCC. The median duration of follow up since diagnosis was 53 months (M) (95% confidence interval [CI], 49-59 M). The median number of aza cycles was 5 (range 1-72), with 75% of pts receiving 4 cycles of therapy or more. The median time from diagnosis to aza initiation was 1.5 M. The median OS for the entire cohort was 19 M (95%CI, 16-22 M), 64% of pts were males, 91% were white, 80% were older than 60 years, and 25% had therapy-related MDS. For the IPSS, the median OS was 23.5 M (CI, 18.4-28.6 M) for INT-2 and 15.9 M (CI, 13.6-18.2 M) for high-risk group (P=0.004). For the FPSS, the median OS was 29.5 M (CI, 14.1-44.9 M) for low risk (LR), 21.1 M (CI, 16.4-26.0 M) for intermediate risk (IR), and 14.1 M (CI, 8.7-19.5 M) for HR (P=0.001). For the MDAPSS the median OS was not reached (NR) for low, 53.5 M (CI, 37.5-69.4 M) for INT-1, 29.5 M (CI, 18.3-40.6 M) for INT-2, and 16.1 M (CI, 14.5-17.7 M) for high risk groups (P<0.0001). For the IPSS-R, the median OS 40.6 M (CI, 22.8-58.4 M) for low, 44.7 M (CI, 25.0-64.4 M) for INT, 23.5 M (CI, 20.3-26.8 M) for high, and 15.2 M (CI, 13.7-16.8 M) for very-high risk groups (P<0.0001). Scores generated using AIC to assess the relative goodness of fit (lower is better) were 1453 (IPSS-R), 1480 (MDAPSS), 1512 (FPSS), and 1522 (IPSS). The IPSS could not refine any of the other models. All three newer models refined IPSS INT-2 but not IPSS high-risk. The IPSS did not refine any of the newer models. The MDAPSS refined the FPSS LR and IR and the IPSS-R high and very-high groups. The FPSS refined IPSS-R very high risk group but not the not the MDAPSS. The R-IPSS did not refine the MDAPSS or the FPSS. Response rates were not statically significantly different within the prognostic groups in any of the scoring systems. Conclusions The IPSS-R, MDAPSS, and the FPSS all functioned well to separate aza-treated pts with IPSS HR-MDS into prognostic groups with different survivals, but the IPSS-R and MDAPSS appear superior to the FPSS. None of the prognostic systems predicted response to aza therapy. HR-MDS patients with poor projected survival with aza therapy might be considered for experimental approaches. Disclosures: Off Label Use: entinostat for MDS. Lancet:Celgene: Research Funding. List:Celgene: Research Funding. Komrokji:Celgene: Research Funding, Speakers Bureau.