In low-risk myelodysplastic syndrome, don’t shoot the STAG2 Free

A core therapeutic objective in low-risk myelodysplastic syndrome (LR-MDS) is to improve anemia and alleviate its impact on quality of life. In this issue of Blood, Raddi et al¹ introduce a new molecular predictive scoring system, ESA-PSS-M, which improves prediction of response to first-line erythropoiesis-stimulating agents (ESAs), a cornerstone of LR-MDS therapy.

Although factors modulating response to ESAs have been delineated (low serum erythropoietin [sEPO] level and transfusion requirement),² ESA resistance remains a persistent and challenging clinical phenomenon. The inception of the ESA-PSS-M is timely. A myriad of novel options for anemia treatment in LR-MDS are being evaluated or have recently been made available, including luspatercept and imetelstat; this broader therapeutic arsenal mandates even finer predictive capacities to select and sequence agents effectively. The stakes are higher than simply personalizing initial management - transfusion independence in response to anemia-directed therapies has been shown to correlate with overall survival.³ Skilled marksmanship in focusing our therapies will therefore be critical.

One of the most significant advances in MDS prognostic evaluation has been the development and validation of the Molecular International Prognostic Scoring System (IPSS-M) score,⁴ which integrates genomic data obtained by next-generation sequencing (NGS) with variables included in previous scoring systems.⁵ In their multicenter retrospective study, Raddi et al, using targeted NGS, evaluated 535 patients with LR-MDS who received ESA therapy. They explored the predictive propensity of IPSS-M score and specific somatic mutations in determining ESA response. Patients were categorized according to World Health Organization 2022 MDS genetic classification,⁶ with ESA response per International Working Group 2018 criteria.⁷ The ESA-PSS-M score generated ultimately leverages 3 broad-gauged and pragmatic variables found to significantly predict ESA response on multivariate analysis: IPSS-M score and conventional predictors (sEPO and transfusion dependence). Validation was subsequently performed in an external cohort of 223 LR-MDS cases.

Comprehensive genomic analysis enabled the authors to make several interesting observations. Considering the entire cohort, no single gene mutation was predictive of ESA response, but the IPSS-M score, which includes and differentially weighs mutations, was predictive of ESA response in univariate analysis. However, when Raddi et al stratified mutational landscape by sex at birth, they documented that males had more total mutations and were enriched in TET2 and SRSF2, as well as X-linked ZRSR2 and STAG2. Focusing on patients in the cohort with sEPO levels <200 U/L, they observed that male nonresponders were enriched in STAG2, BCOR, and RUNX1; ESA response rate (RR) was significantly lower in males harboring the former 2 mutations. In females, conversely, TET2 mutations predicted higher ESA RR.

This approach of exploring MDS genotype, phenotype, and treatment response under a sex-specific lens yielded some of the most notable and innovative findings of this study. The differential prevalence and impact of X-linked MDS-related genes between males and females is not only of interest but of direct clinical relevance. Specifically, for STAG2 mutations, men were more frequently mutated (27 men vs 7 women), and nonsense mutations were more prevalent in ESA nonresponders; in addition, only men harbored frameshift mutations. Comutation of STAG2 with RUNX1 or other X-linked genes BCOR or ZRSR2 in males further predicted nonresponse to ESA. These male-specific patterns are difficult to explain. One potential theory relates to females being mosaic at birth, with an equal proportion of cells harboring the paternal or maternal X-chromosome in the active state. Cells acquiring the STAG2 mutation on the active X can be selected against cells with the wild-type STAG2 gene on the active X. Negative selection of deleterious X-linked alleles in females has been demonstrated in several other conditions, such as Wiskott-Aldrich syndrome and agammaglobulinemia. This can also explain how the most deleterious mutations are less frequent in females. However, this hypothesis is only supported if the growth advantage given to the MDS stem cell is induced by a genetic lesion that is present on the autosome, which is equally expressed in the 2 mosaic populations.

Raddi et al’s findings also corroborate the physiologic role of STAG2 cohesin in erythropoiesis, shown at both the erythroid progenitor commitment and differentiation stages,^8,9 exemplifying how loss of this function results in integrally impaired erythropoiesis, ostensibly translating, at least in part, the resistance observed to ESA.

Ultimately, the predictive model developed by Raddi et al provides a valuable new tool informing treatment selection regarding ESA in LR-MDS. It also serves as a platform for personalized management strategies in the future, encouraging the development and use of corresponding models to predict responses to alternative therapies, such as luspatercept.¹⁰ The crowning challenge will be interconnecting these various models into a cohesive therapeutic stratagem that includes all relevant molecules.

A fundamental question we will then have to address is this: can we potentially overcome resistance to specific agents? And if so, how?

Finally, the referenced study epitomizes the no longer discretional but imperative accounting of sex-driven differences in MDS precision health. In the case of LR-MDS, sex-stratified genomics data revealed the importance of not shooting the (proverbial) STAG2(-mutated males) with ESAs. This lesson in sex-distinct genetic target practice is crucial to apply moving forward in order not to miss our mark on judicious, informed, and precise selection of therapy for MDS.

Conflict-of-interest disclosure: The authors declare no competing financial interests.