A Sex-Informed Approach to Improve Prognostication and Personalized Decision-Making Process in Myelodysplastic Syndromes. a European Study of 11.878 Patients

Introduction

Sex represents a major source of diversity among patients in terms of pathophysiology, clinical presentation, prognosis and response to therapy, and therefore sex (gender)-informed medicine is becoming a new paradigm to refine clinical decision making process in different human diseases. Myelodysplastic syndromes (MDS) are heterogeneous disease characterized by ineffective hematopoiesis and risk of leukemic evolution. We aimed to study clinical effect of sex in MDS as a basis to improve patient prognostication and personalized treatment.

Material and Methods

We analysed three MDS populations from disease-specific registries (Italian registry n=3015, Dusseldorf registry, n=1185 and Spanish registry, n=7678).

Results

We first analysed the association of sex with clinical and biological disease-specific features. These analyses were conducted on Italian MDS cohort. Considering WHO categories, female patients showed higher prevalence of single lineage dysplasia and MDS with del(5q), while males were characterized by higher frequency of multilineage dysplasia and excess blasts (P<.001). Considering cytogenetic abnormalities, an increased frequency of del(5q) in female patients was observed (P<.001).

We analysed mutations in 47 MDS-associated genes. Males presented higher prevalence of mutations with respect to females (82.2% vs. 76.2%, P<.001), with higher prevalence of co-mutations. Considering specific gene pathways, splicing factors mutations were reported more frequently in males vs. females (66% vs. 56%, P<.001). Focusing on MDS with ring sideroblasts (MDS-RS), we noticed a significant different distribution of splicing factor mutations according to sex: SF3B1 was mutated in large majority of female patients (95% of mutated cases), while SRSF2 and ZRSR2 mutations accounts for a significant proportion (26%) of mutated cases in males. In addition, males showed higher prevalence of mutations in DNA methylation genes (P<.001), while tumor suppressor genes (TP53) were more frequently mutated in female patients. Clustering analysis showed specific co-mutation patterns in splicing and DNA methylation genes according to sex.

Overall survival was significantly worse for male vs. female patients (P<.001). In a multivariable analysis including age, neutrophils, hemoglobin and platelet counts, percentage of marrow blasts and cytogenetics as covariate, sex showed an independent effect on probability of survival (HR for female vs. male patients 0.56, P=.012). The prognostic effect of sex was observed in very-low, low and intermediate risk category according to IPSS-R (Table 1). The independent prognostic effect of sex was confirmed in two independent populations (Dusseldorf and Spanish registries).

Competitive risk analysis showed higher prevalence of non-leukemic vs. leukemic deaths (P<.001) in patients with early disease stage. In these patients, we observed a higher prevalence of cardiac comorbidity/deaths in males vs. females (P<.001 and P=.005, respectively). Moreover, the risk of non-leukemic death was higher in males vs. females especially when hemoglobin levels were <10 g/dl. These results suggest that in MDS patients, sex may capture prognostic information on the detrimental interactions between anemia and cardiac comorbidity.

As a final step we aimed to integrate the prognostic value of sex into currently available prognostic systems (IPSS-R). We used random effects Cox proportional-hazards multistate modelling for developing an innovative personalized prognostic model (“Sex and age-adjusted IPSS-R”, IPSS-RAS). All the three study populations were included (n=11.878). The predicted and observed outcomes correlate well in internal cross-validation. IPSS-RAS substantially improved predictive accuracy of original IPSS-R (concordance 0.68 vs. 0.62), especially in patients with early disease stage. Interestingly, demographic factors (age and sex) accounted for >30% of whole prognostic power.

Conclusion

In MDS, sex captures additional prognostic information at individual patient level, possibly reflecting a different molecular background and, most importantly, a sex-specific interaction between disease-related factors and comorbidity. Our results strengthen the rationale to include sex in personalized prognostic assessment in these diseases.