Abstract

Abstract 3852

Background:

Azanucleosides (AZN) and allogeneic stem cell transplantation (ASCT) are the only disease altering treatment modalities for patients with myelodysplastic syndrome (MDS). Azacitidine (AZA) demonstrated an overall survival (OS) advantage in higher risk MDS. The goal of treatment in patients with lower risk MDS is alleviation of symptomatic cytopenias. A key question remains the ability to identify the subset of lower risk MDS patients with poor risk features who may experience an OS benefit from AZN treatment. We hypothesized that recently proposed prognostic models such as the Global MD Anderson risk model (MDAS), revised International Prognostic Score (R-IPSS), and the Lower risk MD Anderson model (LR-MDAS), may identify patient subsets originally classified as lower risk by IPSS who have higher disease risk features and will benefit from AZA treatment.

Methods:

Patients classified as lower risk MDS by IPSS were identified retrospectively from the Moffitt Cancer Center (MCC) MDS database. The primary objective was to examine the utility of risk models identifying patients who benefit from AZA. The primary end point was OS. Patients were treated with AZA based on clinical judgment and primarily for management of cytopenias. The MDAS, R-IPSS, and LR-MDAS scores were recorded and calculated as previously described. The Kaplan–Meier method was used to estimate median OS. Log rank test was used to compare Kaplan–Meier survival estimates between the groups.

Results:

The MCC MDS database captured 608 MDS patients classified as low/int-1 risk by IPSS. The median age was 69 years, 2/3 were males, RCMD was the most common WHO subtype (38%), and 3 % had a poor risk karyotype. Among the 608 patients, 252 were treated with AZA. Based on IPSS, 162 patients were low risk and 56 received AZA, with a median OS of 86 mo (95%CI 64–107) vs. 94 mo (95%CI 66–121, p=0.9) for those 106 patients who did not. Among the 444 patients classified as int-1 risk, 192 were treated with AZA with a median OS of 45 mo (95%CI 37–53) vs. 33 mo (95%CI 32–54) for AZA-untreated (p=0.6).

Based on MDAS, only 478 patients were lower risk, among whom 177 received AZA treatment with a median OS of 59 mo (95% CI 48–70), vs. 72 mo (95%CI 55–88) (p=0.24) in the 301 patients who did not receive AZA. The MDAS upstaged 162 patients (27%) into int-2/high risk categories, 71 of whom received AZA with a median OS of 27 mo (95%CI 21–34) compared to 17 mo (95% CI 14–21) (p=0.003) in the 55 patients did not receive AZA treatment.

Characterization of LR-MDAS was available for 322 patients, 38 of whom stratified as category 1. Twelve of these patients were treated with AZA and 26 patients were not with a median OS that was not reached for both groups (p=0.25). Category 2 included 134 patients, 41 of whom received AZA with a median OS of 69 mo (95%CI 53–85) compared to 77 mo (95% CI 48–106) (p=0.22) in the 93 patients who did not receive AZA treatment. Finally, 150 patients were characterized as category 3, 87 of whom received AZA with a median OS of 46 mo (95% CI 39–53) vs. 34 mo (95%CI 8–61) (p=.85) for the 63 who did not.

By R-IPSS, 539 patients were included in the very low/low/intermediate risk prognostic categories, including 212 patients treated with AZA and 327 who were not with corresponding median OS of 53 mo (95%CI 46–60) and 58 mo (95%CI 45–71) (p=0.16), respectively. The R-IPSS upstaged 67 patients (11%) to high or very high risk among whom 36 received AZA with a median OS of 56 mo (95%CI 18–93) vs. 23 mo (95% CI 17–29) (p=0.16) in the 31 patients who did not.

Conclusion:

The new proposed risk models identify a subset of patients with higher risk features originally stratified as lower risk by IPSS. AZA treatment yielded an OS advantage in patients upstaged to int-2 or high risk by MDAS, suggesting that disease modifying treatments effectively extend OS in such patients. A larger sample size is needed to determine the utility of LR-MDAS or R-IPSS.

Table-1
ModelAZAMedian OS (mo)P
IPSS n=608 Low (n=162) Yes (n=56) 86 0.9 
No (n=106) 94 
Int-1 (n=444) Yes (n=252) 45 0.6 
No (n=192) 43 
MDAS n=608 Low/int-1 (n=478) Yes (n= 177) 59 0.24 
No (n=301) 72 
Int-2/high (n=162) Yes (n=71) 27 0.003 
No (n=55) 17 
LR-MDAS n=322 Category1 (n=38) Yes (n=12) NR 0.25 
No (n=26) NR 
Category 2 (n=134) Yes (n=41) 69 0.22 
No (n=93) 77 
Category 3 (n=150) Yes (n= 87) 46 0.85 
No (n= 63) 34 
R-IPSS n=606 Very low/low/int (n=539) Yes (n=212) 53 0.16 
No (n= 327) 58 
High/very high (n= 67) Yes (n= 36) 56 0.16 
No (n=31) 23 
ModelAZAMedian OS (mo)P
IPSS n=608 Low (n=162) Yes (n=56) 86 0.9 
No (n=106) 94 
Int-1 (n=444) Yes (n=252) 45 0.6 
No (n=192) 43 
MDAS n=608 Low/int-1 (n=478) Yes (n= 177) 59 0.24 
No (n=301) 72 
Int-2/high (n=162) Yes (n=71) 27 0.003 
No (n=55) 17 
LR-MDAS n=322 Category1 (n=38) Yes (n=12) NR 0.25 
No (n=26) NR 
Category 2 (n=134) Yes (n=41) 69 0.22 
No (n=93) 77 
Category 3 (n=150) Yes (n= 87) 46 0.85 
No (n= 63) 34 
R-IPSS n=606 Very low/low/int (n=539) Yes (n=212) 53 0.16 
No (n= 327) 58 
High/very high (n= 67) Yes (n= 36) 56 0.16 
No (n=31) 23 
Disclosures:

Komrokji:Celgene: Speakers Bureau. List:Celgene: Consultancy.

Author notes

*

Asterisk with author names denotes non-ASH members.