Validation Of The Revised International Prognostic Scoring System (IPSS-R) In 1000 Newly Diagnosed MDS Patients With Low- and Intermediate-1 Risk MDS In The European Leukemianet MDS (EUMDS) Registry

The EUMDS registry was initiated to provide an overview of the real-world demographics, diagnostics and disease-management of MDS. The first patient was registered in December 2007 and today 16 countries and 131 centers are participating. In 1997 the International Prognostic Scoring System (IPSS) was developed to predict clinical outcomes for patients with MDS and is still the most widely used prognostic scoring system. Recently, the IPSS has been revised (IPSS-R).

To validate the prognostic discrimination of the IPSS and IPSS-R in the first 1000 newly diagnosed lower risk MDS patients.

The median age of the population at diagnosis was 75 years (range 19-95). WHO 2001 classification was RCMD (35%), RARS (18%), RA (18%), RAEB-1 (12%), RCMD-RS (7%), 5q- syndrome (6%), MDS-U (3%) and RAEB-2 (0.4%). Within the first two years of follow-up 57% of the patients received MDS specific treatment: 48% received erythropoiesis stimulating agents (ESA), 11% granulocyte colony-stimulating factor (G-CSF), 51% received at least one red blood cell transfusion and 8% iron chelation therapy.

IPSS risk score was Low in 49%, Intermediate-1 in 45% (0.5=32%, 1=13%) and unknown in 6% (no cytogenetic analysis) of the patients. IPSS-R risk score was Very Low in 25%, Low in 44%, Intermediate in 16%, High/Very high in 4% of the patients, and 10% unknown (Figure 1; Table 1). 77% of IPSS karyotypes were Good, 15% Intermediate, 1% Poor and unknown in 6%. 7% of the IPSS-R cytogenetic groups were Very good, 72% Good, 11% Intermediate, 1% were each Poor or Very poor and 8% unknown.

Figure 1

View largeDownload slide

Distribution of IPSS & IPSS-R subgroups

Figure 1

View largeDownload slide

Distribution of IPSS & IPSS-R subgroups

Close modal

Overall survival (OS) and disease progression (DP) (high risk MDS/Leukaemia) were both evaluated. Median follow-up time was 2.1 years (range 0 - 4.9 years). The mortality rate in patients with IPSS Low was 23% and 38% among those with an Intermediate IPSS (HR 2.09, 95%CI: 1.64-2.66; Figure 2A). The mortality rate in patients according to the IPSS-R was Very Low (21%; HR 0.77, 95%CI: 0.55-1.07), Low (26%; HR 1), Intermediate (51%; HR 2.53, 95%CI: 1.90-3.38) and High/Very high (65%; HR 4.47, 95%CI: 2.94-6.78) (Figure 2B).

Figure 2A

View largeDownload slide

Overall Survival

Figure 2A

View largeDownload slide

Overall Survival

Close modal

The prognostic discrimination of the scoring systems was assessed using the Akaike Information Criterion (AIC) from univariate proportional hazards models; the lower the AIC value the more informative the prognostic scoring system. The AIC of the IPSS and IPSS-R models were 3198.77 and 3154.48, respectively for OS and 1323.16 and 1274.19, respectively for DP (Table 1). A similar assessment of the components of the IPSS and IPSS-R scores revealed comparable fits to both OS and DP, regardless of which component was considered (Table 2).

IPSS and IPSS-R both predict OS and DP very well. IPSS-R was slightly superior in evaluating the clinical outcome, but it identified a subgroup (4.5% of all patients) of High and Very High-risk patients with a very poor prognosis, and another subgroup of good prognosis patients (IPSS-R Very Low) within the IPSS INT-1 cohort (13.6% of IPSS INT-1). Both scoring systems appear to be more strongly associated with predicting the risk of developing DP than OS. This observation may be due to the average high age at diagnosis of MDS reflecting the effect of competing causes of death associated with high age.

Guerci-Bresler:Novartis: Honoraria; BMS: Honoraria; Celgene: Honoraria; Amgen: Honoraria.