The Prognostic Impact of the Revised IPSS Cytogenetic Scoring System for Patients with MDS Allows a Risk-Stratification for Patients with MDS-Derived Acute Myeloid Leukemia

Abstract 1715

Besides cytopenias and the medullary blast count, cytogenetic risk groups (good vs. intermediate vs. poor) according to IPSS are of main prognostic relevance for overall survival (OS) in patients with myelodysplastic syndrome (MDS). Recently, the revised IPSS (rIPSS) was introduced involving 5 (very good vs. good vs. intermediate vs. poor vs. very poor) instead of 3 cytogenetic risk groups, which better predict disease progression to MDS-derived acute myeloid leukemia (mdsAML) and OS of MDS patients receiving supportive care only. We analyzed the impact of the rIPSS-based cytogenetic scoring systems on the outcome of patients with AML undergoing intensive chemotherapy within the AML96, AML2003, and AML60+ trials of the Study Alliance Leukemia (SAL). This was done in an intention to compare its general prognostic influence as well as between patients with mdsAML and those with a de novo disease (dnAML).

A total of 258 patients (median age 63 years, range 24 – 82) with mdsAML were identified and 258 patients with dnAML were matched with regards to age, gender, clinical trial, induction and consolidation therapy, respectively. Distributions of the cytogenetics in both groups according to MRC, IPSS and rIPSS score are shown in Table 1. Expectedly, the MRC cytogenetic scoring system revealed a stratification into two risk groups for patients with mdsAML with intermediate (3-year OS 27%) and adverse (3-year OS 10%), p=.004, and stratification into three groups for dnAML with favorable (3-year OS 50%), intermediate (3-year OS 32%) and adverse (3-year OS 10%), p=.001. When using the new rIPSS, this allowed a stratification of mdsAML patients with a 3-year OS of 28% for good+intermediate, 12% for poor, and 2% for very poor, p<.001, compared to 28% for good, 22% for intermediate, and 7% for poor risk cytogenetics according to the IPSS, p=.002. Importantly, the rIPSS allowed for a refined subdivision of patients within the poor and very poor group. By applying the rIPSS in dnAML patients we observed a 3-year OS of 34% for good+intermediate, 22% for poor, and 11% for very poor, p<.001, compared to 37% for good, 23% for intermediate, and 19% for poor risk cytogenetics according to the IPSS, p=.028.

In conclusion, the rIPSS and IPSS-based classifications are feasible for prognostic risk stratification of patients with both dnAML and mdsAML. Interestingly, the rIPSS-based good and intermediate risk groups do not separate patients in both groups sufficiently. Furthermore, the rIPSS as compared to the current MRC-based cytogenetic scoring system allowed for a more concise distribution of mdsAML patients with the detection of a very poor (rIPSS) risk group with a dismal outcome.