Myelodysplastic syndromes (MDS) occur mainly in older persons, who are likely to be affected with extra-hematological comorbidities. Recent findings suggest that proper assessment of comorbidities is useful to predict the outcome of MDS patients receiving allogeneic transplantation. However, the results obtained in this highly selected subset of patients cannot be applied to the whole MDS population. In this study, we evaluated the impact of extra-hematological comorbidities on the natural history of MDS with the aim of developing a specific prognostic index. The patients comprised a “learning cohort”, in which we defined the set of variables to be included in the prognostic model and their weighted scores, and a “validation cohort”, in which we confirmed the prognostic value of the scoring system. The learning cohort included 840 MDS patients diagnosed at the Fondazione IRCCS Policlinico San Matteo, Pavia, Italy, between 1992 and 2006, while the validation cohort consisted of 504 patients seen at the Heinrich-Heine-University Hospital, Duesseldorf, Germany, between 1982 and 2006. All cases were classified according to the WHO criteria. Patients who underwent allogeneic transplantation or intensive chemotherapy were censored at the time of the procedure. One or more comorbidities were present in 455 (54%) patients in the learning cohort at the time of diagnosis: the older the age, the higher the prevalence. Cardiac disease was the most frequent extra-hematological morbidity (25% of patients) and the main cause (63%) of non-leukemic death (NLD). In a Cox multivariable analysis with time-dependent covariates, the onset of a comorbidity significantly affected the risk of NLD (HR=2.29, P<.001) and worsened overall survival (OS, HR=1.51, P=.01). Patients who developed RBC transfusion-dependency had a significantly higher risk of NLD (HR=4.31 P=<.001), cardiac disease and death (HR 4.16 and HR 4.88, respectively; P=<.001). In this group, serum ferritin levels were significantly associated with the risk of cardiac disease and death (P=.001). The onset of cardiac, liver, renal, pulmonary disease and solid tumor were found to independently affect the risk of NLD in a multivariable Cox regression (HR from 3.57 to 1.97; P values from <0.001 to 0.04). Based on these results, we developed a dynamic prognostic model (MDS-specific comorbidity index, MDS-CI) for predicting the effect on NLD and OS of comorbidities either present at the time of diagnosis or occurring during the follow-up. Risk scores for each comorbidity were estimated from the regression coefficients (Table 1).Table 1. MDS-specific comorbidity index (MDS-CI)ComorbidityScoreCardiac disease2Moderate-to-severe hepatic disease1Severe pulmonary disease1Renal disease1Solid tumor1Risk groups: Low (score 0), Intermediate (score 1–2), High (score >2).MDS-CI allowed us to identify 3 groups of patients with different probability of NLD and OS (P<.001). The prognostic value of the MDS-CI was then evaluated in the validation cohort of patients, where the 3 MDS-CI risk groups showed significantly different probabilities of NLD (P<.001) and OS (P=.005), with a 2-year risk of NLD since diagnosis of 24%, 42% and 61% in the low, intermediate and high risk group, respectively. In summary, extra-hematological comorbidities significantly worsen the natural history of MDS patients, specifically increasing the risk of NLD. In particular, transfusion-dependency and secondary iron overload are associated with a higher risk of cardiac complications. The MDS-CI improves our ability to stratify the outcome of MDS patients and may be a useful tool for clinical decision-making. An accurate evaluation of extra-hematological comorbidity should be part of the prognostic assessment of patients with MDS.