OSMR takes the leap: from NULISA to ELISA Free

In this issue of Blood, Reville et al have identified oncostatin M receptor (OSMR) as a serum-based biomarker predicting acute myeloid leukemia (AML) outcomes, offering a clinically feasible complement to current genomics-driven risk models.¹ 

AML remains one of the most heterogeneous hematologic malignancies, and accurate risk stratification continues to be a moving target. Although genomics has revolutionized prognostic models, these approaches are costly and time-consuming. Serum-based biomarkers could complement genomic information and offer broader clinical applicability, but robust candidates have remained elusive.

Reville et al report the largest serum proteomic screen in AML to date, identifying OSMR as a novel and reproducible biomarker of clinical outcome. Using the nucleic acid–linked immuno-sandwich assay (NULISA),² an ultrasensitive platform that combines antibody detection with DNA barcoding and polymerase chain reaction amplification, the authors profiled 251 proteins in serum samples from 543 patients newly diagnosed with AML. The assay achieves femtomolar sensitivity, enabling detection of proteins well below the limits of conventional enzyme-linked immunosorbent assay (ELISA).

The authors derived an 8-protein classifier, termed the Leukemia Inflammatory Risk Score (LIRS), which was independently validated in internal and external cohorts. Among these candidates, OSMR emerged as the most robust predictor of outcome. Importantly, OSMR levels could also be measured using standard ELISA, making the biomarker immediately translatable to clinical laboratories, even in settings without access to high-sensitivity platforms like NULISA.

This technical feasibility is a central strength of the study. Most prognostic proteins identified were present in concentrations too low for ELISA detection, limiting their near-term clinical relevance. In contrast, OSMR bridges the gap between discovery and implementation. Its detection by conventional methods accelerates potential integration into existing diagnostic workflows and supports risk stratification at scale.

The cohort used was large, well annotated, and diverse in terms of age and treatment intensity, enhancing generalizability. Although mainly from a single center, the validation steps were rigorous. The prognostic value of OSMR and the LIRS model held independently of the current European LeukemiaNet classifications for intensive-treatment³ and nonintensive-treatment⁴ patients. This supports the notion that serum biomarkers can complement, and in some cases refine, genetically driven stratification.

Biologically, OSMR has been implicated in several solid tumors, including glioblastoma and squamous cell carcinoma, where it promotes immune evasion and stemness.^5-7 Its role in AML is less defined but may involve interactions within the microenvironment. The finding that OSMR is associated with adverse outcomes in AML hints at broader immunobiologic functions and invites further mechanistic exploration.

Clinically, OSMR mirrors the precedent of other serum markers such as β2-microglobulin in chronic lymphocytic leukemia or lactate dehydrogenase in aggressive lymphomas: simple, reproducible, and informative. Although genomics remains essential for therapeutic decisions (eg, FLT3 or IDH1/2 mutations), serum biomarkers may enhance prognostication and dynamic monitoring.

Naturally, limitations remain. The external validation cohort was relatively small, and interlaboratory reproducibility needs to be demonstrated. Moreover, the biological function of OSMR in AML pathogenesis requires further elucidation. Still, the authors acknowledge these caveats and offer a clear road map for future validation and functional studies.

In conclusion, this study marks a potential turning point for serum-based prognostic markers in AML, shifting focus beyond genomics and toward multiparametric risk models. The identification of OSMR adds a new dimension to individualized risk assessment. Future prospective trials should evaluate the integration of OSMR into treatment algorithms, particularly in combination with existing genomic classifiers.

Conflict-of-interest disclosure: T.H. has received research funding from Roche; serves on the advisory boards of Servier and Jazz Pharmaceuticals; and has received travel support from Jazz Pharmaceuticals.