Peripheral blood-based single-cell and genomic profiling of hematopoietic stem and progenitor cells enables accurate diagnosis and risk stratification in myeloproliferative neoplasms

Abstract

Abstract Diagnosing myeloproliferative neoplasms (MPNs) remains challenging due to overlapping clinical and morphological features, and reliance on subjective bone marrow (BM) assessments. Approximately 10–15% of cases lack JAK2, CALR, or MPL mutations (triple-negative), and bone marrow biopsies may be unobtainable or inconclusive. Moreover, molecular data are often underutilized, limiting diagnostic precision and prognostic assessment. We aimed to: (1) evaluate whether transcriptomic profiling of circulating hematopoietic stem and progenitor cells (cHSPCs) using single-cell RNA sequencing (scRNA-seq) can serve as a non-invasive alternative to bone marrow (BM) biopsies for MPN diagnosis; (2) distinguish borderline entities such as essential thrombocythemia (ET) and pre-fibrotic myelofibrosis (pre-MF); and (3) identify blood-based transcriptomic and genomic biomarkers predictive of disease severity and progression. Methods: We performed single-cell RNA sequencing on cHSPCs from 77 MPN patients (11 primary MF, 4 post-PV MF, 5 post-ET MF, 13 ET, 44 PV) and 9 triple-negative erythrocytosis controls, alongside a reference atlas of 123 healthy individuals. We integrated transcriptomic data with targeted myeloid NGS panel covering 52 recurrently mutated genes. Clustering performance was validated on blinded MPN samples. Results: cHSPC transcriptional profiles, based on cHSPC composition and expression signature, accurately distinguish MPN subtypes and reliably separated MF from ET and pre-MF, with blind samples correctly classified (Sensitivity: 92.6% Specificity: 95.4%). MF cases exhibited aberrant cHSPC, including increased megakaryocyte progenitors and hematopoietic stem cell-multilineage pluripotent progenitor (HSC-MPPs). This MF-specific cHSPC composition correlated with clinical severity and transformation to MF in PV/ET cases. MF cHSPC expresses a distinct cell cycle gene signature (M-Phase, S-Phase), compared to PV/ET, suggestive of aberrant proliferation and differentiation dynamics and a novel transcriptional program anchored by the SOCS2 genes expressed variably in patients' cells, with intensity that is associated with JAK2 and CALR variant allele frequencies (VAFs) burden. Interestingly, weaker SOCS2 signature activation is observed in healthy cHSPC, suggestive of underappreciated variability in JAK2 pathway activation in myeloid progenitors. Using the SOCS2 signature and cHSPC composition we prompted re-biopsy and confirmed progression to MF for 3 patients PV/ET patients in our cohort. Conclusion: Our integrated platform—combining cHSPC single-cell transcriptomics with targeted DNA sequencing—provides a minimally invasive, accurate, and reproducible strategy for diagnosing MPNs. It effectively resolves diagnostic ambiguity in borderline cases and enables prediction of disease severity. This approach is currently being prospectively evaluated in a clinical trial. We identified a novel transcriptional signature consistently overexpressed in MPNs, serving as a bona fide marker of myeloproliferation.