CD48 as an additional marker for measurable residual disease assessment in pediatric Acute Myeloid Leukemia. Free

Background

Multidimensional flow cytometry-based measurable residual disease (MFC-MRD) assessment is an essential tool for response evaluation and risk stratification in pediatric acute myeloid leukemia (pedAML). Its accuracy critically depends on the identification of leukemia-associated immunophenotypes (LAIPs) or “different-from-normal” (DfN) patterns that allow reliable distinction between leukemic blasts and regenerating normal hematopoietic cells. Although novel markers like CD11a, CD99, CD371, and CD45RA have improved MFC-MRD, challenges remain when leukemic and normal cells overlap. CD48, a broadly expressed hematopoietic surface antigen, has already been established as a stable and informative MRD marker in T-cell acute lymphoblastic leukemia (T-ALL), due to its consistent downregulation on T-ALL blasts compared to normal T-cells. Conversely, the expression profile and diagnostic utility of CD48 in pedAML has not yet been evaluated. Hence, we investigated CD48 as a potential new MFC-MRD marker in pedAML.

Methods

CD48 was included in the diagnostic MFC panel for newly diagnosed pediatric acute leukemias (n=641) among which 98 were pedAML cases. CD48 was subsequently incorporated into the MRD panel for patients whose blasts were CD48-negative, weakly positive, or partially negative at diagnosis, with identical panels used at follow-up. MRD monitoring employed a 10-color format with two dried antibody cocktail tubes (DuraClone, BECLS): ReAML1 (HLA-DR, CD45, CD15, CD34, CD117, CD33, CD14, CD11b) and ReAML2 (HLA-DR, CD45, CD38, CD371, CD34, CD117, CD33, CD99, CD123, CD45RA). PE and APC channels in ReAML1 were used for patient-specific “drop-in” markers such as CD48. MRD-negative regenerative bone marrow (BM) samples from non-AML patients (n=6) were also analyzed to define baseline CD48 expression in normal hematopoiesis. Associations between CD48 protein and transcript expression with morphological or genetic AML subtypes were also evaluated.

Results

CD48 expression was found absent on normal erythropoietic precursors, including CD34⁺ erythro-myeloid progenitors and erythroblasts, while broadly expressed among other normal BM cell subsets. Leukemic blasts from pedAML patients showed no CD48 expression in 45/98 (45.9%) cases, providing a robust DfN framework for residual disease detection. Differentiation subtype-specific analysis revealed significant heterogeneity in CD48 expression. The highest frequency of CD48 negativity was observed in FAB M6/M7 (15/16, 93.8%), while all FAB M3 cases (n=7) exhibited strong CD48 expression on blasts. CD48 negativity across other FAB subtypes was lowest in FAB M5 (4/20, 20.0%), as compared to M1/M2 (9/20, 45%) and M4 (7/16, 44%). Additionally, CD48 loss was noted in 5 of 12 (41.7%) cases of secondary AML or dominant-myeloid mixed-phenotype acute leukemia (MPAL) and in 3/7 (42.9%) unclassified/M0 AML cases. Two of 3 (66.7%) cases of transient myeloproliferative disorder (TMD) were also negative. Protein-level variability was mirrored in transcriptomic data, with consistently high CD48 transcripts in FAB M3 (PML::RARA), while other FAB/genetic subtypes showed a broad range, including many with low or undetectable levels, especially in FAB M7. Given the strong enrichment of CD48 negativity in FAB M7 — in which CD11a negativity is a known immunophenotypic hallmark — we investigated the additional diagnostic value of CD48 independent of CD11a. To this end, we also included HLA-DR, another frequently downregulated antigen in AML in the analysis. Among all 45 CD48-negative cases, HLA-DR and CD11a remained expressed in 34/45 (75.6%) and 29/45 (64.4%) cases, respectively.

To assess CD48 stability, we analyzed paired diagnosis and follow-up samples from 23 cases. Eleven were MRD-positive at follow-up, focusing on early, clinically relevant timepoints. In 8/11 (72.7%) cases, CD48 expression was unchanged, with negative or partially negative patterns persisting. In 2 cases, only the CD48-negative clone remained, while 1 case showed partial gain of CD48, though blasts remained dim and different from regenerating cells. In all MRD-positive cases, CD48 reliably aided blast identification at follow-up.

Conclusion

Given the immunophenotypic heterogeneity of AML and the limitations of existing markers, we demonstrated CD48 as a critical tool for a more refined and precise MFC-MRD assessment in pedAML.