Prognosis and Outcomes in MDS-MPN Unclassifiable: Single Institution Experience of a Rare Disorder

Introduction: The 2008 World Health Organization (WHO) classification has recognized a unique overlap category that has features of proliferation found in myeloproliferative neoplasms (MPN) and also dysplasia found in myelodysplastic syndrome (MDS). The least well characterized of the 4 MDS/MPN overlap diseases is a rare entity known as MDS/MPN Unclassifiable (MDS/MPN-U), comprising <5% of myeloid disorders. Furthermore, given the rarity of this disorder, there is no validated risk stratification scoring system, although there are several commonly used prognostic models for MDS, including the International Prognostic Scoring System (IPSS), the Revised International Prognostic Scoring System (IPSS-R), and the M.D. Anderson Cancer Center model (MDAS). The objectives of this study were to evaluate the natural history of this very uncommon diagnosis and to determine which of the current scoring symptoms used for MDS best discriminates outcomes.

Methods: The Moffitt Cancer Center database of over 3000 MDS patients was used to identify patients with MDS/MPN-U and to subsequently perform a comprehensive chart/pathology review. We then applied IPSS, IPSS-R, and the MDAS scores to each patient in order to compare differences in overall survival (OS) amongst different risk groups within each scoring system. Finally, we compared outcomes in the MDS/MPN-U group with a large number of matched MDS cases from within our database, using the MDAS. Descriptive statistical analyses were utilized. Chi square analysis and t- test were performed to compare categorical and continuous variables. Akaike information criteria (AIC) were used to assess the relative goodness of fit of the models. All data was analyzed using SPSS version 21.0 statistical software.

Results: Forty three patients were identified with MDS/MPN-U and were pathologically confirmed to meet WHO criteria. Median age was 71 years (range 55 - 91) and the M:F = 26.17. Median baseline laboratory parameters: WBC 11.2 x 103/dL (range 0.9 - 84.8); Hb 9.7 g/dL (range 5.8-14.4); platelets 137 x 103/uL. Table 1 summarizes risk stratification per current validated MDS scoring systems. The majority of patients had lower risk disease by all the models. Forty of 42 (95%) patients evaluable for prognostic scoring were classified as low/Int-1 by IPSS. However, 11 out of the 40 pts (28%) classified as lower risk by IPSS were upgraded to Int-2 or high risk by MDAS. Twenty-two patients received hypomethylating agents (HMA) as first line treatment after supportive care. Per IWG 2006, 8 of 22, (36%) had complete response, partial remission, or hematologic improvement, 7 (32%) had stable disease, and 6 (27%) had progressive disease. The median OS for all MDS/MPN-U patients was 33 months (95% Confidence Interval 22 - 45). Within each MDS scoring system, statistically significant survival differences were detected between risk stages (table 1). The IPSS-R did not improve the IPSS prognostic value. Patients categorized as lower-risk (low/Int-1) by MDAS had superior survival compared to IPSS. Lastly, we compared outcomes between the 43 MDS/MPN-U patients and 1117 IPSS low/Int-1 matched controls within the MDS database. Median overall survival was inferior in MDS/MPN-U vs. MDS (33.4 mo vs. 57 mo, p = 0.005). In addition, using the MDAS, stage-by-stage, survival was significantly worse in the MDS/MPN-U group.

Conclusions: MDS/MPN-U appears to have a variable disease course but with generally poor outcomes, even amongst lower-risk patients classified by MDS scoring systems, and despite a moderate rate of response to treatment. Matched comparisons indicate inferior outcomes compared with similarly staged MDS patients. The MDAS may offer increased discriminatory capacity for determining prognosis based on disease stage. Further work with a larger patient population and cross comparisons to other MDS/MPN diseases will assist further understanding of this rare disorder. Integration of somatic mutations data may compliment the clinical models.