An Analysis of Prognostic Markers and the Performance of Scoring Systems in 1837 Patients with Therapy-Related Myelodysplastic Syndrome - a Study of the International Working Group (IWG-PM) for Myelodysplastic Syndromes (MDS)

Background: The International Prognostic Scoring System (IPSS) for MDS has recently been revised (IPSS-R). However both scoring systems were developed after exclusion of therapy-related cases and data on its usefulness in treatment-related MDS (tMDS) is limited.

Aims and Methods: We analyzed 1837 pts from Spanish, German, Swiss, Austrian, US, Italian, and Dutch centers diagnosed 1975-2015. Complete data to calculate the IPSS/-R was available in 1511 pts. The impact of prognostic features was analyzed by uni- and multivariable models and estimated by a measure of concordance for censored data (Dxy).

Results: Median age was 68 years. 1% of pts had 5q-syndrome, 13% RCUD, 4% RARS, 27% RCMD/-RS, 18% RAEB 1, 18% RAEB 2, 4% CMML 1, 2% CMML 2, 3% MDS-U, and 7% AML (RAEB-T) according to WHO-classification. Regarding cytogenetics 38% exhibited good, 14% intermediate, and 48% poor-risk according to IPSS, and 2% very good, 36% good, 17% intermediate, 15% poor, and 31% very poor according to IPSS-R. Prognostic risk groups were 12% IPSS low, 34% int 1, 36% int 2, and 18% high, while the IPSS-R was very low in 8%, low in 20%, intermediate in 17%, high in 23%, and very high in 32%.

The most frequent primary diseases were NHL 28%, breast cancer 16%, myeloma 6%, prostate cancer 6%, Hodgkins disease 5%, and 4% gastrointestinal tumors. Patients received chemotherapy in 75% and radiotherapy in 47%. Regarding chemotherapeutic drugs, most pts received combination regimens containing alkylating agents in 65%, topoisomerase inhibitors in 44%, antitubulin agents in 26%, and antimetabolites in 26%.

Median follow-up from MDS diagnosis was 59 months, median survival 16 months. Since a disease altering treatment is, at least in higher risk disease, which is overrepresented in tMDS, standard of care, we decided to analyze treated as well as untreated pts to avoid a selection bias. This included stem cell transplantation in 16% with a median survival of 24 months.

Features with influence on survival and time to AML in univariable analysis included FAB, WHO, IPSS, IPSS-R, cytogenetics, hb, platelets, marrow and peripheral blasts, ferritin, LDH, fibrosis, ß2-microglobulin, and use of alkylating agents for the treatment of primary disease. For hemoglobin, platelets, LDH, fibrosis, and ß2-microglobulin the influence was stronger on survival. Year of diagnosis, age, gender, neutrophil count, WBC, use of chemo or radiotherapy as well as other chemotherapeutic agents had no marked influence on both outcomes.

According to our results, both the IPSS (Dxy 0.29 for survival, 0.32 for AML) and IPSS-R (Dxy 0.34, 0.32 for AML) perform moderately in tMDS, but not as well as in primary MDS (pMDS). Therefore, existing prognostic models need to be adjusted to tMDS. However, this appears to be not without difficulties. The scores tested, as well as most prognostic variables themselves perform inferior compared to pMDS. It becomes even more complicated since tMDS in itself is even more heterogeneous than pMDS. Scores and variables perform differently depending on the primary disease or therapy. The IPSS/-R and its variables perform for example better in pts with solid tumors compared to hematologic diseases or in pts who have received radio- instead of chemotherapy, but also in pts after prostate compared to breast cancer.

In addition to the integration of further variables, new cutoffs, or the weighting of existing variables, we are currently testing the possibility of separate score versions for different tMDS subgroups. Separate score versions for survival and time to AML would also give differing weights to most features. Hemoglobin, platelets and cytogenetics would get more weight for survival, while marrow blasts would be more important regarding AML.

Conclusion: In contrast to early descriptions of tMDS, with poor risk cytogenetics in the vast majority of pts and a uniformly poor prognosis, surprisingly we find good risk karyotypes in a relatively large number of pts. Although, poor risk cytogenetics are still overrepresented, this indicates, different types of tMDS exist. Our analysis shows that many variables exhibit prognostic influence in tMDS and the IPSS or preferably IPSS-R can be applied in these pts. However, the prognostic power of both scores is inferior compared to pMDS, making an optimized tMDS score reasonable. Currently data from further IWG centers is integrated in our database and further analyses are performed to propose a tMDS specific score.

Figure 1.

View largeDownload slide

Figure 1.

View largeDownload slide

Close modal