Prognostic Significance of Cell-Free DNA Derived 5-Hydroxymethylcytosine Signatures in Newly Diagnosed Multiple Myeloma.

PURPOSE: While survival outcomes in multiple myeloma (MM) have improved with contemporary combination therapies, predicting disease trajectories for individual patients at diagnosis remains a significant challenge. We investigate the prognostic value of a noninvasive biomarker-cell-free DNA (cfDNA)‑derived 5-hydroxymethylcytosine (5hmC) signature-in newly diagnosed MM, aiming to improve risk stratification at diagnosis.

MATERIALS AND

METHODS

In this prospective cohort study, 321 patients with newly diagnosed MM were enrolled between 2010 and 2017, with follow-up through 2022. We profiled genome-wide 5hmC modifications (gene bodies) in cfDNA collected at diagnosis. We applied elastic net regularization to a Cox proportional hazards model to identify 5hmC signatures associated with overall survival (OS) and progression-free survival (PFS). A weighted prognostic score (wp-score), based on 18 key 5hmC-modified genes, was developed through machine learning and validated in the validation set, controlling for clinical prognostic factors.

RESULTS

During a median follow-up of 70.5 months, 127 deaths occurred. The cfDNA 5hmC at diagnosis reflected the 5hmC in bone marrow‑derived tumor cells and differed across clinical subgroups. The wp-score showed strong prognostic ability for OS (hazard ratio [HR], 2.9 [95% CI, 1.7 to 4.9]; P < .0001) and PFS (HR, 1.8 [95% CI, 1.3 to 2.5] P = .00014) in the validation set, controlling for known prognostic factors, including stage, lactate dehydrogenase levels, and treatment type. The wp-score calculated at diagnosis remained associated with OS and PFS at 24-, 48-, and 72-month follow-up periods.

CONCLUSION

This prospective study suggests that cfDNA-derived 5hmC signatures at diagnosis provide independent prognostic information for OS and PFS in MM. These findings support further investigation and independent validation of noninvasive epigenomic biomarkers within contemporary MM risk stratification frameworks.