Unveiling prognostic subtypes in myelofibrosis through routine blood counts: A population-based analysis of the cytopenic and proliferative phenotypes in andalusia, Spain

Abstract Background Most published myelofibrosis (MF) data come from clinical trials or hospital registries, mainly from large centers, with follow-up starting at treatment. Data from small centers are often missing, as data is manually entered by investigators, with non-systematic digitization errors, resulting in limited real-world evidence from unselected populations. Using the Andalusian Public Health (SAS) Population Database, we conducted one of the largest population-based MF cohorts to date, with longitudinal follow-up from diagnosis and up to 10 years of observation. Our goal was to evaluate the prognostic significance of the cytopenic vs. proliferative phenotypes using routine laboratory parameters at diagnosis. Methods This retrospective cohort included adults (≥18 years) diagnosed with primary or secondary MF between September 1, 2017, and December 31, 2022, with ≥1 year of follow-up and continuous inclusion in the SAS population registry. The cytopenic phenotype was defined by Hb <10 g/dL and/or platelets <100×10⁹/L within 12 months prior to diagnosis; patients without cytopenia were classified as proliferative. Exclusion criteria included concurrent neoplasms (except PV/ET), splenectomy, irradiation, HIV, or major data gaps. We used Kaplan-Meier and Aalen-Johansen estimators to assess the overall survival (OS) and cumulative incidence of death or leukemic transformation. Results The cohort included 1629 patients, with a balanced distribution of primary (50.7%) and secondary (49.3%) MF. The proliferative group included 1157 patients, with an average age of 68.6 years (standard deviation, SD, 15.2), whereas the cytopenic group consisted of 472 individuals with an average age of 72.8 years (SD 13.2). Males accounted for 51.6% of the proliferative group and 60.8% of the cytopenic group, with statistical significance (p=0.001). Regarding anemia, 72.9% of the proliferative group had no anemia, whereas only 7.0% of the cytopenic group were non-anemic. Severe anemia was observed in 19.7% of the cytopenic patients, while none was reported in the proliferative group. Platelet counts also differed significantly, with the proliferative group having a mean of 466.9x109/L (SD 309.2) compared to 246.6x109/L (SD 285.1) in the cytopenic group (p<0.001). Median follow-up was 2.6 [1.5-4.5] years. Cytopenic patients had significantly inferior OS compared with proliferative cases, HR= 3.41 (95% IC [2.86-4.05], p<0.001) and progression to acute myeloid leukemia (AML), HR=2.61 (95% IC [1.78-3.83], p<0.001), with survival curves diverging early. Stratification by the number of cytopenias (none, one, or both) revealed a stepwise worsening of prognosis. Competing risk analysis confirmed a higher cumulative incidence of death and AML transformation. These differences were evident despite not accounting for cytogenetics or splenomegaly. Conclusions This is one of the largest real-world MF cohorts with follow-up from diagnosis, including all patients regardless of diagnosis/treatment site, with no data loss or systematic entry bias, ensuring high data reliability and robustness. Our findings confirm the existence of two distinct clinical phenotypes: proliferative and cytopenic MF. The latter is associated with significantly poorer outcomes and likely harbors worse cytogenetic and molecular features, as previously described. However, we demonstrated that a basic blood test, such as hemogram, which includes specific thresholds of hemoglobin <10 g/dL and platelets <100×10⁹/L can effectively predict the outcome at diagnosis, regardless of spleen size or the results of more advanced tests. These data support the integration of phenotype-based risk stratification into routine practice and early therapeutic decision-making.