Prognostic Value of Circulating Myeloblasts in Patients with Myelodysplastic Syndromes (MDS): A Proposal for Incorporation in the Revised International Prognostic Scoring System (R-IPSS)

Introduction

The World Health organization (WHO) MDS classification proposes presence of 2-4% peripheral myeloblasts (PB) as criteria for Refractory anemia with excess blasts I (RAEB-I) and 5-19% PB for RAEB-II classification, while 1% PB persistently is recognized as MDS unclassified. The most widely used clinical prognostic models such as IPSS, R-IPSS, and MD Anderson risk model (MDAS) have not incorporated PB as a prognostic variable. We evaluated the prognostic value of PB in a large MDS cohort and provide a proposal to incorporate presence of PB in R-IPSS.

Methods

The MDS database at Moffitt Cancer Center (MCC) was used to identify MDS patients (pts). Pts were classified into two groups based on presence of PB (1% or more) at time of diagnosis as PB-MDS group and those without PB called BM-MDS. Kaplan-Meier curves were used to depict survivals, and the log-rank test was used to compare median overall survival (OS). We collaborated with Genoptix Medical Laboratory to assess the correlation between presence of PB and gene mutations identified by next generation sequencing (21 myeloid gene panel) in a cohort of de-identified MDS pts.

Results

There were 1905 pts included from MCC MDS database among whom 260 pts (14%) had evidence of PB near time of diagnosis. PB-MDS patients were younger, more likely to have trilineage cytopenia, complex karyotype, transfusion dependent and more in therapy related MDS. Among PB-MDS pts 175 (67%) received hypomethylating agent (HMA) compared to 977 (59%) in the BM-MDS group, p 0.017

According to IPSS risk stratification PB-MDS pts were more likely to be classified as higher risk (HR-MDS) 157 (63%) compared to 445 (28%) in the BM-MDS group (p<0.005). Pts with PB-MDS were also HR-MDS by MDAS (p <0.005). The rate of AML transformation was 46% (n=120) compared to 25% (n=418) in PB-MDS and BM-MDS group respectively, p <0.005. Median overall survival (OS) was 46 month (95% CI 42 to 49.6) in the absence of PB compared to 17.5 mo (95% CI 14.9-20) with PB (p < 0.005). Impact on OS was greater in IPSS lower risk MDS patients where median OS was 34 mo in PB-MDS compared to 60 mo in BM-MDS patients (p < 0.005), while in IPSS HR- MDS the median OS was 16.5 in PB-MDS compared to 18 month BM-MDS (p 0.018)

We then examined prognostic discrimination of PB among each R-IPSS category. Median OS for very low risk R-IPSS was 104 mo in absence of PB compared to 37 mo in BM-MDS (p 0.032). Among low risk patients the median OS was 69 mo in absence of PB compared to 40 mo (p 0.07). In the intermediate risk group, median OS was 40 mo in absence of PB compared to 23 mo with PB (p 0.001), whereas median OS in the high risk group was 24 mo without PB compared to 20 mo with PB (p 0.11). Finally, in the very high risk R-IPSS the median OS was 15 mo compared to 13 mo respectively (p 0.44). The presence of PB upgraded pts with very low or low risk R-IPSS to intermediate risk. The outcome of intermediate risk group with PB was similar to the high risk group. In Cox regression analysis the presence of PB was an independent prognostic covariate for OS after adjusting for R-IPSS and age, HR 1.5 (95% CI 1.3-1.8). Among HR-MDS pts treated with HMA (n=470) presence of PB was independent prognostic variable for OS. (HR 1.3, p 0.027)

We next created a R-IPSS+PB risk model where one additional point was awarded for presence of PB and pts categorized into risk groups based on the same lump score suggested by R-IPSS for each risk category. We applied this score for 245 pts with PB where R-IPSS score was known (Table-1). Sixty three pts (26%) were upstaged to high or very high risk group and most pts were upstaged to next risk group.

Among 51 pts in Genoptix Medical Laboratory database with known PB, the rate of at least single gene mutation identification in pts with PB-MDS was 100%, (4 out of 4 with PB) compared to 81% in those without PB (38 out of 47 without PB, 12 of those were single SF3B1 gene mutation in ring sideroblasts MDS subtype). The gene mutations in PB-MDS included U2AF1 gene mutation in 2 pts, SRSF2, TET-2, ASXL-1, and RUNX-1. Two pts had two gene mutations.

Conclusions

Presence of PB in MDS is an adverse independent prognostic variable that refines prognostic discrimination in Low to Intermediate risk R-IPSS groups. Accounting for presence of PB particularly the intermediate risk group, prioritizes disease altering therapeutic strategies.