Independent Validation of the MD Anderson Cancer Center Risk Model for Myelodysplastic Syndromes (MDS), and Comparison to the International Prognostic Scoring System (IPSS) and the World Health Organization-Based Prognostic Scoring System (WPSS).

The 1997 IPSS remains the most widely used prognostic scoring system for patients with MDS, and it has important strengths, but several other prognostic systems have been proposed recently to overcome some of the well-recognized IPSS limitations. These systems include a new risk model for patients with MDS and CMML – including previously treated patients and those with secondary (treatment-related) MDS – proposed by Kantarjian and colleagues at the MD Anderson Cancer Center (MDACC) (Cancer 2008; 113:1351), as well as the WPSS, proposed by Malcovati and colleagues in Europe (J Clin Oncol 2007; 25:3503) To date, the MDACC risk model has not been validated in a large external independent cohort or directly compared to the WPSS. We assessed the performance of the MDACC model compared to the WPSS and the IPSS in a 12-year Mayo Clinic MDS/CMML cohort.

We reviewed the medical records of 1,503 adult patients (pts; 984 males; median age 71 years, range 17-98) with MDS (n=1,249) or CMML (n=254) evaluated at Mayo Clinic between January 1996 and December 2007. Pediatric pts (age ≤16 and pts with ≥30% marrow blasts were excluded. IPSS and WPSS scores were calculated, and data collected for MDACC risk model assignment (age, performance score, platelet count, hemoglogbin, marrow blast proportion, white blood cell count, karyotype, and transfusion history). Data were analyzed using Kaplan-Meier survival analysis and Cox proportional hazards regression models.

The overall median survival was 17.8 months (mos) – 13.7 mos for CMML, 19.1 mos for WHO-defined MDS, and 9.6 mos for RAEB-t – similar to the 14.1 mos observed in the test cohort used to derive the MDACC risk model. Follow-up was complete until death in 1,122 (74.7%) pts. Pts excluded from the cohort stratified using IPSS included 122 patients with CMML and leukocytosis (>12 × 10⁹/L), and 176 pts with secondary MDS/CMML (there were 12 pts with secondary CMML with leukocytosis) – total excluded, 286 pts. Patients excluded from the WPSS cohort included 27 pts with RAEB-t, 254 pts with CMML, and 148 pts with secondary MDS – total excluded, 429 pts. Survival by IPSS risk group was 38.9 mos for Low (n=413), 17.9 mos for Int-1 (n=541), 9.6 mos for Int-2 (n=206), and 6.6 mos for High (n=57). Survival by WPSS risk group was 39.0 months for Very Low (n=352), 26.6 months for Low (n=276), 15.9 months for Intermediate (n=185), 11.2 mos for High (n=221), and 4.9 months for Very High (n=40). Survival by MDACC risk group was 45.6 mos for Low (n=426), 20.0 mos for Int-1 (n=552), 12.3 mos for Int-2 (n=325), and 4.9 mos for High (n=200). When the MDACC risk model was used to classify only the 1,074 pts with conditions for which the WPSS has been validated, survival was 51.4 mos for Low (n=336), 21.2 mos for Int-1 (n=411), 13.3 mos for Int-2 (n=213), and 4.6 mos for High (n=114) (p<0.001 for all comparisons). In a multivariable proportional hazards model, all of the MDACC risk model components except WBC>20 × 10⁹/L retained independent prognostic significance.

All 3 systems stratify patients accurately, but the MDACC risk model best identifies the lowest-risk patients, and also classifies the broadest group of patients (i.e., primary and secondary MDS, and primary and secondary CMML with or without leukocytosis). The revised IPSS that is currently in development should include the patient factors accounted for by the MDACC risk model, with the possible exception of leukocytosis.

No relevant conflicts of interest to declare.