Abstract 3796

Poster Board III-732

Therapy-related myelodysplastic syndrome (Tx-MDS) is a poorly defined disease entity with in general a poor prognosis. The natural history of Tx-MDS is not well understood. In this analysis, we evaluated patient characteristics in an attempt to define the natural history and create a prognostic model of Tx-MDS. This analysis focuses on patients with Tx-MDS that have been previously treated for other malignancies. We identified a total of 438 patients who had a history of one or more malignancies prior to diagnosis of Tx-MDS evaluated at MD Anderson Cancer Center between 1997 to 2007. Finally, 281 Tx-MDS patients were selected by identifying those that had received prior chemotherapy (CTx) and/or radiation therapy (RTx) with a prior history of other malignancy, and by eliminating non-MDS entities from WHO classification such as RAEB-T. Patient characteristics are as follows: male sex was 165 (58.7%), median age at diagnosis of MDS was 66.2 years (range 13.4-89.4). IPSS was low in n=30 (11%), INT-1 in n=87 (31.0%), INT-2 in n=120 (42.7%) and high in n=35 (12.5%). The most common cytogenetic abnormality was -5 and/or -7 (n=149, 53.1%). Seventy five patients (26.7%) were diploid. Prior cancers included: head and neck (n=7, 2.5%), thyroid (n=3, 1.1%), lung (n=7, 2.5%), breast (n=32, 11.4%), gastrointestinal (n=13, 4.6%), prostate (n=34, 12.1%), other genitourinary or gynecological (n=16, 5.7%), melanoma/skin cancers (n=5, 1.8%), sarcomas (n=8, 2.8%), other solid cancers (n=2, 0.7%), lymphoma (n=102, 36.3%), CML/CLL (n=6, 2.1%), AML/ALL (n=5, 1.8%), multiple myeloma (n=11, 3.9%) and multiple cancers (n=30, 10.7%). Prior Tx was CTx only (n=107, 38.1%), RTx only (n=73, 26.0%) or both CTx and RTx (n=101, 35.9%). A total of 54 patients had received hematopoietic stem cell transplantation (HSCT) (autologous n=52 (18.5%); allogeneic n=2 (0.7%)). Potential prognostic factors were selected by univariate analyses and validated by multivariate analysis. The final prognostic factors were incorported into a prognostic model of Tx-MDS. We also evaluated the impact of specific forms of therapy in outcome in Tx-MDS. Univariate analyses for better survival revealed that presence of hepatomeglay (no hepatomegaly vs. hepatomegaly; Median 10.7 vs. 1.7 months (M); 95% confidence interval [CI] 0-5.8 vs. 8.7-12.7M; p=0.023), chromosome alterations (8+, 20q-, Y-, normal vs. others; median 22.1M vs. 8.2M ; 95% CI 17.0-27.2 vs. 6.9-9.5M; p<0.001), types of MDS by WHO classification (RA, RCMD, MDSu vs. others; median 22.4 vs. 8.8M; 95% CI 10.4-34.4 vs. 7.5-10.1M; p<0.001), time from Tx to MDS (≤6 vs. >6Y; median 13.3 vs. 10.6; 95% CI 5.4-21.2 vs. 6.8-14.4; p=0.027), treatment line(s) of therapy (1 vs. ≥2; median 14.4 vs. 8.6M; 95% CI 10.1-18.8 vs. 5.5-11.7M; p=0.011), serum albumin (≥4 vs. <4g/dL; median 14.4 vs. 9.1M; 95% CI 9.7-19.1 vs. 7.5-10.8m; p=0.005), serum beta2 microglobulin (≤3 vs. >3mg/L; median 12.2 vs. 9.1M; 95% CI 8.2-16.2 vs. 6.3-11.8M; p=0.015), serum creatitine (≤1 vs. >1mg/dL; median 11.9 vs. 10.2M; 95% CI 7.1-16.7 vs. 8.1-12.3M; p=0.061), ECOG performance status (0-1 vs. ≥2; median 11.8 vs. 5.5M; 95% CI 9.7-14.0 vs. 2.4-8.6M; p<0.001). Age (p=0.109), sex (p=0.862), prior Tx (CTx vs. RTx only; p=0.471), prior malignancies (hematological vs. solid cancer; p=0.650), prior lymphoma (p=0.958), prior HSCT (p=0.691) and serum ferritin level (p=0.420) were not significant. When incorporated into the multivariate model, only chromosomal alterations (HR=0.366) and WHO classification (HR=0.339) were significant prognostic factors. Based on this, we divided patients into three prognostic groups: good (n=38; no risk factor; median survival 33.8 months), intermediate (n=108; 1 risk factor; median survival 15.3 months) and poor (n=135; 2 risk factors; median survival 6.8 months). This model could predict time to acute leukemia transformation (events 2 vs. 12 vs. 18; p-value=0.017. In conclusion, we have developed a model that predicts for overall survival and time to acute leukemia transformation in Tx-MDS. This model will serve to develop risk-adapted therapeutic strategies.

Disclosures:

No relevant conflicts of interest to declare.

Author notes

*

Asterisk with author names denotes non-ASH members.

Sign in via your Institution