AUTHORS
Yael Kusne, MD, PhD, Aubrey Pinteric2, Abhishek Avinash Mangaonkar, MBBS3, Antoine N. Saliba, M.D.4, Terra Lasho, PhD5, Christy Finke, BS6, Tariq Kewan, MD7, Jodi Rindflesch, RN8, Aref Al-Kali, M.D.9, Cecilia Arana Yi, M.D.10, Talha Badar, MD11, James Foran, FRCPC12, Amer M. Zeidan, MD13, Kaaren Reichard, MD14, Robert Hasserjian15, Sanam Loghavi, MD16, Mrinal M. Patnaik, MD, MBBS17
AFFILIATIONS
1 Mayo Clinic Arizona, Phoenix, Arizona, United States
2 Mayo Clinic Arizona, Phoenix, Arizona, United States
3 Mayo Clinic Rochester, Rochester, United States
4 Mayo Clinic Rochester, Rochester, United States
5 Mayo Clinic Rochester, Rochester, United States
6 Mayo Clinic Rochester, Rochester, United States
7 Mayo Clinic Rochester, Rochester, United States
8 Mayo Clinic Rochester, Rochester, United States
9 Mayo Clinic Rochester, Rochester, United States
10 Mayo Clinic Arizona, Phoenix, Arizona, United States
11 Mayo Clinic Florida, Jacksonville, United States
12 Mayo Clinic Florida, Jacksonville, United States
13 Yale University, New Haven, Connecticut, United States
14 Mayo Clinic Rochester, Rochester, United States
15 Mass General, Vineyard Haven, Massachusetts, United States
16 MD Anderson Fountain, Houston, Texas, United States
17 Mayo Clinic Rochester, Rochester, United States
ABSTRACT
Background: The distinction between clonal cytopenia of undetermined significance (CCUS) and myeloid neoplasms (MN) is often subjective, relying on morphological thresholds that are difficult to apply consistently. We conducted a prospective natural history study with independent, blinded morphological review to identify objective markers for MN diagnosis that may help refine current diagnostic criteria.
Methods: Using patients enrolled in our prospective CCUS natural history cohort (2016–2024; NCT02958462), three expert hematopathologists from three US academic medical centers performed blinded bone marrow (BM) review. Reviewers were blinded to the original pathologic diagnosis, molecular profiles, and clinical data, with access only to patient age and a peripheral blood smear, and therefore to the degree of cytopenias. Each BM specimen (n=134, from 116 unique patients: clinically diagnosed CCUS [n=92], MN positive controls [n=17], and normal controls [n=7]) was independently assessed by two hematopathologists, who quantified dysplasia and assigned each case to CCUS, myelodysplastic syndrome/neoplasm (MDS), or chronic myelomonocytic leukemia (CMML) based on integrated bone marrow morphological assessment. For cases in which the two reviewers disagreed on the diagnostic assignment, final adjudication was performed by a second round of review by all three pathologists to reach consensus. Clinical progression among CCUS patients was assessed over longitudinal follow-up as a composite of treatment initiation and/or transfusion dependence; a maximum VAF threshold of ≥21% was derived by receiver-operating-characteristic analysis (Youden's J).
Results: Blinded review reclassified 32 of 92 CCUS patients (35%): 22 as MDS and 10 as CMML. Inter-reviewer disagreements crossing the 10% dysplasia threshold occurred in 26% of megakaryocytic, 20% of granulocytic, and 11% of erythroid assessments. Compared with stable CCUS, cases diagnosed as CMML displayed a distinct morphological phenotype: BM hypercellularity (median cellularity 70% vs 40%, p<0.001), BM monocytosis (60% vs 12%, p=0.002), hypolobated megakaryocytes ≥10% (100% vs 10%, p<0.001), and MPN-like megakaryocytes (50% vs 3%, p<0.001). Megakaryocyte and granulocytic dysplasia ≥10% were strongly associated with both MDS and CMML but did not discriminate between them. In contrast, BM monocytosis and MPN-like megakaryocytes were exclusive to CMML and, on univariable analysis, strongly correlated with spliceosome mutations (OR 12.1, p<0.001 for CMML; non-significant for MDS), particularly SRSF2 (OR 32.2, p<0.001), present in 50% of CMML cases at a median VAF of 46%. TET2 and SRSF2 co-mutation was present in 2 of the 10 CMML cases and in no other patient, corresponding to 100% positive predictive value and 20% sensitivity for CMML. MDS reclassification, by contrast, was morphologically driven, with no individual molecular predictor reaching significance — expected, since SF3B1-mutated, del(5q), and multi-hit TP53 cases are excluded from CCUS by definition. Maximum mutation VAF was a continuous marker of reclassification risk for both MDS and CMML (OR 1.06 per 1% increase, p<0.001) and independently predicted clinical progression among CCUS patients, defined as a composite of treatment initiation, transfusion dependence, or MDS; those with a maximum VAF ≥21% (a threshold identified by ROC analysis) had a 4.5-fold higher risk of composite progression (p=0.025).
Conclusions: In this prospective CCUS cohort, blinded expert review reclassified over one-third of patients as having an MN, while inter-observer variability around the 10% dysplasia threshold highlighted the limitations of morphology-based diagnosis. Notably, several morphological features associated with MN — including hypercellularity, BM monocytosis, and MPN-like megakaryocytes — are not part of current diagnostic criteria. Spliceosome mutations (particularly TET2/SRSF2 co-mutation) and larger clone size (as measured by VAF) reliably identified patients with CMML, whereas MDS remained largely morphology-defined. These findings support integrating molecular features with morphology to improve diagnostic accuracy and refine the classification of clonal cytopenias and myeloid neoplasms.
DISCLOSURES
Yael Kusne, MD, PhD: No relevant conflicts of interest to declare.