Skip to content
    Real-World Studies

    Nonlinear TET2 Variant Allele Frequency Refines CMML Prognostic Models

    Wang et alBlood neoplasia

    Nonlinear TET2 Variant Allele Frequency Refines CMML Prognostic Models
    In 1,206 patients with chronic myelomonocytic leukemia (CMML) from a UK discovery cohort and two validation cohorts (validation cohort 1, an international cohort; validation cohort 2, a French cohort with CD14+ monocyte sequencing), segmented regression of TET2 variant allele frequency (VAF) distributions identified two reproducible inflection points (discovery cohort: 32% and 55%; validation cohort 1: 33% and 54%; validation cohort 2: 36% and 56%), indicating a consistent subdivision of TET2 mutation burden. Restricted cubic spline Cox regression showed a nonlinear relationship between TET2 VAF and survival, where clinical risk was lowest near a VAF of 39% (the empirical first quartile), while patients with TET2 wild-type status experienced hazards comparable to individuals with high VAF.
    Applying a continuous per-integer TET2 VAF adjustment to existing risk systems—including the CMML-specific Prognostic Scoring System-Molecular (CPSS-Mol), Groupe Français des Myélodysplasies (GFM) model, and Mayo Molecular Model—improved discrimination and group-level stratification for overall survival and acute myeloid leukemia-free survival. The adjusted scores achieved Harrell C-index gains in 16 of 18 comparisons and significant integrated discrimination improvement in all 18 comparisons (P < .01), suggesting that TET2 VAF functions as a continuous prognostic variable that can refine standard CMML risk models.

    Key Takeaways

    • Restricted cubic spline Cox regression revealed a nonlinear relationship between TET2 variant allele frequency (VAF) and survival, with mortality risk lowest near a VAF of 39% and TET2 wild-type patients exhibiting hazards comparable to those with high VAF.
    • Incorporating per-integer TET2 VAF adjustments into CPSS-Mol, GFM, and Mayo Molecular Model scores improved discrimination for overall survival and acute myeloid leukemia-free survival, yielding Harrell C-index gains in 16 of 18 comparisons and significant integrated discrimination improvement across all 18 comparisons (P < .01).