Report of an international working group to standardize response criteria for myelodysplastic syndromes

Bruce D. Cheson, John M. Bennett, Hagop Kantarjian, Antonio Pinto, Charles A. Schiffer, Stephen D. Nimer, Bob Löwenberg, Miroslav Beran, Theo M. de Witte, Richard M. Stone, Moshe Mittelman, Guillermo F. Sanz, Pierre W. Wijermans, Steven Gore and Peter L. Greenberg


Standardized criteria for assessing response are essential to ensure comparability among clinical trials for patients with myelodysplastic syndromes (MDS). An international working group of experienced clinicians involved in the management of patients with MDS reviewed currently used response definitions and developed a uniform set of guidelines for future clinical trials in MDS. The MDS differ from many other hematologic malignancies in their chronicity and the morbidity and mortality caused by chronic cytopenias, often without disease progression to acute myeloid leukemia. Whereas response rates may be an important endpoint for phase 2 studies of new agents and may assist regulatory agencies in their evaluation and approval processes, an important goal of clinical trials in MDS should be to prolong patient survival. Therefore, these response criteria reflected 2 sets of goals in MDS: altering the natural history of the disease and alleviating disease-related complications with improved quality of life. It is anticipated that the recommendations presented will require modification as more is learned about the molecular biology and genetics of these disorders. Until then, it is hoped these guidelines will serve to improve communication among investigators and to ensure comparability among clinical trials.


The myelodysplastic syndromes (MDS) are a heterogeneous group of hematopoietic disorders characterized in most patients by peripheral blood cytopenia with hypercellular bone marrow and dysplasia of the cellular elements.1-3 MDS may develop after exposure to toxins such as benzene, chemotherapy drugs, or high doses of radiation, though its etiology is unknown in more than 80% of patients.

MDS have historically been referred to as oligoblastic leukemia, refractory anemia, smoldering acute leukemia, or preleukemia. In 1982, the French-American-British (FAB) group presented a classification, modified in 1985, that is the most widely used.1 2Recently, a World Health Organization (WHO) steering committee proposed modifications to the MDS subtypes. The major changes were a decrease, from 30% blasts to 20% blasts, in the threshold for diagnosing acute myeloid leukemia (AML) and the creation of a category of MDS/myeloproliferative disorders to include patients with chronic myelomonocytic leukemia.4 However, these recommendations have not yet been uniformly adopted, and the distinction between AML and MDS reflects the pace of the disease, the biologic differences in neoplastic cells, and the number of bone marrow blasts. Thus, because of its known clinical usefulness, we have elected to retain the FAB classification of MDS.

Prognostic factors

MDS are heterogeneous with regard to clinical characteristics, cytologic and pathologic features, and cytogenetics. Even within morphologic subtypes, there are differences in outcome. Therefore, effective and prospective stratification of patients for clinical studies is critical for designing trials and for evaluating and clarifying outcomes of treatments.

Recently an international working group of experienced clinicians developed a prognostic scheme that has been widely adopted.5 Critical prognostic factors regarding survival and the potential for evolution to AML, which were identified and included in the International Prognostic Scoring System (IPSS), included bone marrow cytogenetics, percentage of bone marrow blasts, and number of cytopenias; age and gender were also important for predicting survival in a multivariate analysis. Within the IPSS, patients were categorized according to these features into relatively low risk (IPSS Low or Intermediate-1) and relatively high risk (IPSS Intermediate-2 or High) subgroups for risk-based treatment options.

Therapeutic goals

The MDS differ from many other hematologic malignancies in their chronicity and in the morbidity and mortality caused by chronic cytopenias, often without disease progression to AML.6 As such, alleviation of disease-related complications and improved quality of life (QOL) are important goals of therapy. Thus, improvements in cytopenias (ie, clinically meaningful hematologic improvement) and their associated complications should be objectively measured and evaluated. In addition, some of the newer classes of drugs are more likely to be cytostatic than cytotoxic, or they may induce cellular differentiation; therefore, time to disease progression may be the primary endpoint rather than the response rate. Responses, then, should reflect the goals of the treatments.

Various strategies have been used to treat patients with MDS, often with varying objectives. Low-intensity therapies, defined as treatments capable of permitting predominantly outpatient management (eg, cytokines, certain biologic response modifiers, immunosuppressive therapy), are often directed at patients with low risk MDS (IPSS Low and Intermediate-1). The goals of many of these low-intensity therapies are improvement in blood counts, disease palliation, and enhanced quality of life.7 Such treatments are not necessarily associated with improved survival or progression-free survival. Thus, hematologic improvement is appropriate for measuring responses to this type of treatment.

The aim of other low-intensity therapies (low-dose chemotherapy, 5-azacytidine; low-dose cytarabine) and high-intensity therapies (aggressive anti-leukemic chemotherapy, stem cell transplantation) is to induce hematologic responses (complete remission [CR] or partial remission [PR]) and to alter the natural history of the disease (prolonging survival, progression-free survival). Cytogenetic responses may be helpful to determine the degree to which the natural history of the disease may be altered.

There are no curative therapies for patients with MDS other than bone marrow transplantation, which is successful in only a subset of treated patients.8-11 As a result, numerous therapies have been and are being evaluated to improve the outlook for these patients. However, the published results of clinical trials are difficult to interpret for a number of reasons, including patient selection bias, sample size, and inconsistent response criteria. For example, hematologic improvement has been used to indicate variable levels of increase in one hematologic lineage in some studies or multilineage improvement in others.12-14

Several new agents are used in clinical trials for MDS, including 5-azacitidine, decitabine, amifostine, topotecan, and others.12 13 15-18 In addition to defining comparable patient groups among studies and collecting complete prognostic factor information, the availability of uniform response criteria would improve analysis of clinical trials in MDS.

An approach to standardizing response criteria

To resolve the problems resulting from variability in definitions of the quality and quantity of response in MDS, a group of international investigators with expertise in MDS convened to establish standardized response criteria for clinical trials involving patients with MDS. Some of the response categories (hematologic improvements) are more relevant to therapies designed with palliative intent, whereas others (complete remission, partial remission) are relevant to the goals of treatments directed at altering the natural history of the disease. The proposal for evaluating response is shown in Table1 and indicates 4 levels of response criteria based on the intent of specific therapies: hematologic response, altering the natural history of the disease, cytogenetic response, and quality of life. Although data are lacking regarding the correlation between cytogenetic response and clinical outcome in MDS, based on the valuable role of this parameter in the management and prognosis of patients with chronic myelogenous leukemia,19 20 we have included this category to facilitate prospective evaluation.

Table 1.

Measurement of response/treatment effect in MDS

Table 2.

Definitions of endpoints for clinical trials in MDS

Similarly, whereas only limited data exist regarding the value of quality of life instruments in assessing treatment outcomes in MDS, such methods of evaluation have been valuable for patient assessment in other neoplasms.21 22 QOL studies are most valuable in randomized trials, not only with placebo controls but also when comparing 2 approaches. QOL studies may also be useful in single-arm trials to assess interactions among disease-related symptoms, treatment-related toxicity, and disease response. These aims should be defined prospectively. We anticipate that the use of this criterion in appropriate clinical trials, using instruments such as the WHO Performance Score or the FACT Questionnaire,22 will provide valuable insights into patients' physical, functional, emotional, and social status.

Also critical for evaluating and clarifying outcomes of treatments and for designing clinical trials is to stratify patients effectively and prospectively using risk-based criteria for patient entry and evaluation (FAB subtype or, perhaps more important, IPSS risk group5). For example, comparison of survival or progression-free survival for patients at IPSS low risk would differ from those at IPSS high risk and should be indicated as such. Such stratification could include age, performance status, and prognostic risk category, as suggested by the National Comprehensive Cancer Network Panel on MDS.5 7


The goals of clinical research in MDS are to prolong the survival of patients and to improve their quality of life. In phase 2 trials, in which the activity of a new agent may be the most important objective, response rates are valuable and may provide support for approval by regulatory agencies. However, in some clinical settings, incremental increases in response rates have not translated into prolongation of time to treatment failure or survival. In MDS, added objectives for patients with this often indolent, chronic illness are to reduce morbidity associated with cytopenias (diminish transfusion requirements, infections, or bleeding) and to improve quality of life. In this regard, we believe that the use of risk-based prognostic stratification and clinically relevant response criteria would be valuable for evaluating clinical outcomes and for trial design in MDS.

It will be important to apply these guidelines prospectively in large trials and to critically assess their validity and usefulness. Uniform criteria will help to determine the impact of a specific clinical outcome on patient survival or improved quality of life. The IPSS should be used as the primary stratification factor. In addition, studies should prospectively assess whether there is a difference in outcome for patients from 0 to 6 months, 6 to 12 months, and longer than 12 months from diagnosis of MDS. We have also set an arbitrary threshold for minor cytogenetic response at 50% normal metaphases; future studies must better correlate cytogenetic response with survival. We anticipate that these recommendations may require modification as more is learned about the molecular biology and genetics of these disorders. We hope that these guidelines will serve to improve communication among investigators and to ensure comparability among clinical trials.


  • Bruce D. Cheson, National Cancer Institute, Executive Plaza North, Room 741, Bethesda, MD 20892; e-mail:chesonb{at}

  • The publication costs of this article were defrayed in part by page charge payment. Therefore, and solely to indicate this fact, this article is hereby marked “advertisement” in accordance with 18 U.S.C. section 1734.

  • Submitted January 24, 2000.
  • Accepted August 8, 2000.


View Abstract