Therapy-related MDS deserve specific diagnostic sub-classification and risk stratification—an approach to classification of patients with t-MDS

Historically, patients with therapy-related myelodysplastic syndromes (t-MDS) were categorized together into the large group of therapy-related myeloid neoplasms, irrespective of blast count and morphological characteristics like dysplasia or cellularity. The World Health Organization (WHO) classification currently groups MDS together with therapy-related acute myeloid leukemia (AML) and t-myelodysplastic/myeloproliferative neoplasms into one subgroup, again irrespective of prognostic or morphological traits. Not categorizing t-MDS as a subgroup of MDS limits proper clinical decision making, as most prognostic tools for MDS exclude patients with t-MDS, with the exception of the MD Anderson Prognostic Scoring System. As such, this classification of t-MDS interferes with research, leading to the common exclusion of the disease from clinical studies and limiting the evolution of therapeutic improvements.

In a study recently published in Leukemia, Andrea Kuendgen and colleagues gathered comparative prognostic data in t-MDS relative to primary MDS (p-MDS), assessed the performance of currently existing tools for classification and prognostication, and finally enhanced the current stratification systems for use in t-MDS.¹

Study design

This study compiled a database of a total of 2,087 patients with MDS from eight different study groups in the US, Germany, Spain, Italy, Austria, and the Netherlands. As a comparison, p-MDS data was used from 4,593 patients in the revised International Prognostic Scoring System (IPSS-R) database, limited to the institutions contributing data to both projects. Exclusion criteria were defined as: age < 16 years, AML-defining cytogenetic abnormalities, peripheral blasts > 19%, normal karyotype, proliferative chronic myelomonocytic leukemia, AML as a primary diagnosis, survival or time to AML < 2 months, and progression of primary disease. Treatment in the MDS phase was not an exclusion criterion.

WHO-2016, IPSS-R, revised WHO classification-based prognostic scoring system (WPSS-R), and French-American-British (FAB) classification calculations were performed, with the WHO classification grouping refractory cytopenia with unilineage dysplasia (RCUD), refractory anemia with ring sideroblasts (RARS), and MDS del(5q) together as per WPSS. To test the ability of these tools to differentiate, the stratified log-rank test and the stratified Dxy coefficient were performed. Dxy represents a concordance coefficient varying between −1 and 1. Zero indicates no monotone discriminative ability, and 1 shows a tool has perfect monotone discrimination with respect to the time in question. Transformation-free survival was used in this study as it more accurately represented disease-related risk.

Results

1,245 of 2,087 t-MDS patients successfully fulfilled all relevant selection criteria. Patient characteristics of t-MDS and p-MDS are summarized in Table 1. In p-MDS, 71% belonged to the large good-risk group within the cytogenetic component of IPSS-R, and approximately half of the patients presented with a normal karyotype, whereas in t-MDS these values were only 37% and 30%, respectively. Compared to the p-MDS group, patients with high-risk (22%) and very high-risk (31%) IPSS-R score were more frequent among t-MDS.

Table 1. Patient characteristics of therapy-related (t-MDS) and primary (p-MDS) myelodysplastic syndromes¹

CMML, chronic monomyelocytic leukemia; FAB, French-American-British classification; IPSS-R, revised International Prognostic Scoring System; MDS, myelodysplastic syndromes; MDS-U, MDS, unclassified; p-MDS, primary MDS; RA, refractory anemia; RAEB, refractory anemia with excess blasts; RAEB-T, RAEB in transformation; RAEB-1, RAEB type 1; RAEB-2, RAEB type 2; RARS, refractory anemia with ring sideroblasts; RCMD, refractory cytopenia with multilineage dysplasia; RCUD, refractory cytopenia with unilineage dysplasia; t-MDS, therapy-related MDS; WHO, World Health Organization classification; WPSS-R, revised WHO classification-based prognostic scoring system. * HMAs, chemotherapy, and/or allogeneic HSC transplant.
Characteristics	t-MDS patients, % n = 1,245	p-MDS patients, % n = 4,593	p value
MDS treatment Treated* Untreated Total with information	63 37 91	0 100 100
Stem cell transplantation (yes)	19	0
Age (years) ≤ 60 > 60 to ≤ 70 > 70 to ≤ 80 > 80 Median, years	28 32 32 8 68	23 27 35 15 70	< 0.001
Gender Male	55	62	< 0.001
FAB RA RARS RAEB RAEB-T CMML Unclassified	40 9 38 7 4 2	37 18 27 7 9 2	< 0.001
WHO RCUD RARS RCMD RAEB-1 RAEB-2 MDS (del5q) MDS-U	17 6 31 22 20 1 3	17 13 29 16 19 4 2	< 0.001
IPSS-R Very low Low Intermediate High Very high	8 21 18 22 31	19 36 19 14 12	< 0.001
Cytogenetic component of IPSS-R Very good Good Intermediate Poor Very poor	2 37 16 15 30	3 71 14 4 8	< 0.001
WPSS-R Very low Low Intermediate High Very high	8 15 21 38 18	22 28 17 24 9	< 0.001
Number of cytogenetic aberrations 0 1 2 3 4 ≥ 5	30 21 12 6 4 27	73 19 4 2 1 2	< 0.001
Therapy for primary disease All chemotherapy All radiation inc. radioiodine Radiation alone Radioactive iodine Chemotherapy alone Radiation and chemotherapy	80 54 19 1 45 35	—	—
Primary diagnosis Hematological Breast Prostate Other solid tumor Non-malignant disease	43 16 10 28 3	—	—

Despite the performance on the scoring systems for t-MDS being inferior to p-MDS, the systems were able to determine risk groups. As shown in Table 2, the prognostic power for the various classification and scoring systems was calculated by stratified Dxy coefficient in t- and p-MDS. The FAB classification could only discriminate two different adjacent risk groups (RA vs RAEB, p < 0.001) in either MDS type. As a surrogate test for the WHO classification, the WPSS categorization was used, which led to a better separation of the combined low-risk group versus the remaining categories in both MDS types.

Prognostic classification systems performed well, with IPSS-R separating five different risk groups for all outcomes tested and WPSS separating five risk groups regarding overall survival, but only four regarding progression-free survival and AML transformation. In addition, when analyzed separately, the prognostic power of the cytogenetic component of IPSS-R was very high, with only the difference between the very low and low-risk group not reaching significance (p = 0.210). This performance was at least equivalent in t-MDS versus p-MDS. Interestingly, all scoring systems showed a greater level of distinction when a sub-sample of treatment-naïve patients with MDS were used in the analysis.

Table 2. Dxy for the different scoring systems and outcomes presented for t- and p-MDS¹

FAB, French-American-British classification; IPSS-R, revised International Prognostic Scoring System; MDS, myelodysplastic syndromes; p-MDS, primary MDS; t-MDS, therapy-related MDS; WHO, World Health Organization classification; WPSS-R, revised WHO-based Prognostic Scoring System.
Score	Transformation-free survival		Overall survival		Time to AML
	t-MDS	p-MDS	t-MDS	p-MDS	t-MDS	p-MDS
FAB	0.19	0.30	0.17	0.28	0.24	0.42
WHO	0.24	0.29	0.19	0.26	0.41	0.44
IPSS-R	0.37	0.41	0.38	0.40	0.36	0.53
WPSS-R	0.35	0.38	0.33	0.36	0.40	0.51
Cytogenetic component of IPSS-R	0.30	0.23	0.32	0.23	0.23	0.28
Number of aberrations	0.29	0.13	0.32	0.13	0.22	0.14
Primary diagnosis	0.05	—	0.05	—	0.03	—

Conclusion

In conclusion, the findings of this study indicated that classification tools established in p-MDS were successful in stratifying subgroups in t-MDS and demonstrated the high prognostic relevance of cytogenetics in t-MDS. These results, from the largest t-MDS database to date, are expected to start a discussion regarding a prospective revision of the WHO classification and motivate clinicians to apply the current tools for risk evaluation and treatment plans for patients with t-MDS into clinical studies.