If Nostradamus were treated for MDS

The International Prognostic Scoring System (IPSS)¹  and World Health Organization Prognostic Scoring System (WPSS)²  are great tools to distinguish lower- from higher-risk myelodysplastic syndromes in untreated de novo patients, accurately predicting survival and leukemia evolution. But what system would you use to advise a patient embarking on disease-modifying therapy?

The hypomethylating drug azacitidine was the first agent approved (in 2004) by the US Food and Drug Administration specifically for the treatment of myelodysplastic syndromes (MDS).³  Its US label includes all MDS subtypes, and approval was based in part on a combined endpoint of a significant delay in time to transformation to acute myeloid leukemia and death, compared with best supportive care, in a phase 3 Cancer and Leukemia Group B trial.³  Subsequently, in a phase 3 European study, azacitidine significantly improved overall survival in higher-risk MDS patients when compared with conventional care regimens, with a median survival of 24.5 months versus 15 months, respectively (hazard ratio .58, P = .0001).⁴  The improved outcomes were probably due to more narrowly defining the MDS population in whom azacitidine is thought to have more activity (hypermethylation with consequent gene inactivation is more prevalent in higher-risk MDS, compared with lower-risk subtypes)⁵  and to preventing crossover from nonresponding control arm patients to the azacitidine arm, as was allowed in the Cancer and Leukemia Group B study. Patients with higher-risk MDS comprise approximately 25% of newly diagnosed, and 15%-20% of established cases.⁶  It is difficult to determine which of these patients is more likely to respond to azacitidine and live longer as a result.

In this issue of Blood, Itzykson and colleagues report the results from an analysis of 282 higher-risk MDS patients treated with azacitidine in a compassionate use program in 42 centers in France.⁷  As opposed to patients enrolled on many clinical trials, this is a representative MDS patient group, with 22% having refractory anemia with excess blasts in transformation/acute myeloid leukemia; 26% with secondary MDS; 10% having been previously treated with cytarabine; and 32% with an erythropoiesis-stimulating agent. Patients received a median of 6 cycles of therapy, and 17% achieved a complete or partial response with an additional 21% achieving hematologic improvement, according to 2006 International Working Group response criteria⁸  and similar to the European survival study. In addition, the investigators verified the finding from the survival study that patients achieving a hematologic improvement or better response lived longer.

In multivariate analyses, previous treatment with cytarabine, bone marrow blasts more than 15%, and abnormal karyotype were associated with a lower response rate, while complex cytogenetics was associated with a shorter response duration. Median overall survival was 13.5 months, and again in multivariate analyses, higher performance status (Eastern Cooperative Oncology Group score ≥ 2), intermediate and high-risk IPSS cytogenetic risk categories, presence of circulating blasts, and high red blood cell transfusion needs (≥ 4 units/8 weeks) were associated with worse survival. From these 4 variables, a simple scoring system was developed, with each given a value of 1 point (and 1 for intermediate, and 2 for poor-risk cytogenetics), and patients were assigned to 3 risk categories (see table): low (score = 0, median survival not reached), intermediate (score = 1-3, median survival 15 months), and high (score = 4-5, median survival 6.1 months).

Although it is easy to use statistical tests to identify factors associated with worse outcomes in any study, and by chance alone a factor can be isolated if 20 are explored, what makes this analysis special was validation of these factors in a separate cohort of patients treated with azacitidine—those enrolled in the European azacitidine survival study. In this population, the prognostic scoring system continued to accurately distinguish survival among risk groups, with median survival not reached for low risk, 21.4 months for intermediate risk, and 9.3 months for high-risk categories. As would be expected in a clinical trial with restrictive inclusion criteria, survival was higher for each risk category compared with survival in the compassionate use program.

So, how should the azacitidine-specific prognostic scoring system be used in context of broader MDS prognostic scoring systems, such as the IPSS¹  and WPSS² ? These nonspecific MDS prognostic systems were developed based on clinical and pathologic data from largely untreated MDS patients—92% of patients included in the IPSS had received no therapy, 100% in the WPSS—and it would be the rare patient included in one of these systems who received disease-modifying therapy subsequent to inclusion. That being said, the IPSS and WPSS are useful in determining whether a patient is considered higher risk at baseline, and thus more appropriate to receive therapy with a hypomethylating agent such as azacitidine. At that point, the Azacitidine Prognostic Scoring System can be invoked, to identify the group of patients who would be predicted to have favorable survival on azacitidine therapy, and those with comparatively poor survival, in whom alternative therapeutic strategies should be selected. It will be exciting to see how this prognostic schema compares to another schema used in treated patients, developed by the M. D. Anderson group,⁹  and whether it can next be validated in combination regimens that include azacitidine and either a histone deacetylase inhibitor or lenalidomide.