Consecutive patients, diagnosed with mAA between January 2019 and December 2022, were enrolled. Bone marrow aspirate specimens for flow cytometric immunophenotyping were obtained from 56 patients before they received immunosuppressive therapy. Ten healthy pediatric stem cell donors served as normal controls. Ethical approval for this study was obtained from the Medical Ethics Committee of Wuhan Children's Hospital, Tongji Medical College, Huazhong University of Science and Technology.

The diagnosis of acquired AA was established according to the published criteria (16). Patients were considered to have mAA if they had ≥2 of the following: peripheral blood neutrophils <1.0 × 10⁹/L, platelets <50 × 10⁹/L, and reticulocytes <60 × 10⁹/L. sAA was defined as bone marrow cellularity <25% and at least two of the following: peripheral blood neutrophils <0.5 × 10⁹/L, platelets <20 × 10⁹/L, and reticulocytes <20 × 10⁹/L.

The mAA patients received cyclosporine (5–6 mg/kg per day, serum level of 200–300 ng/ml) with or without the addition of granulocyte colony stimulating factor for patients with peripheral blood neutrophils <0.2 × 10⁹/L.

The hematologic response to cyclosporine was evaluated at 3 and 6 months after treatment initiation. A complete response (CR) was defined as achieving peripheral blood neutrophils >1.5 × 10⁹/L, platelets >100 × 10⁹/L, and hemoglobin >100 g/L. A partial response (PR) was defined as achieving peripheral blood neutrophils >1.0 × 10⁹/L, platelets >50 × 10⁹/L, and hemoglobin >80 g/L. Patients with no PR, no CR, or requirement for blood transfusion were deemed as having no response. Relapse was defined by conversion to no response from PR or CR and/or the requirement of blood product transfusion (17).

Bone marrow aspirate specimens were obtained at the time of first diagnosis of AA and before treatment initiation, where 3 × 10⁵ nucleated cells per sample were analyzed. Immunophenotyping was performed by multiparameter flow cytometry (BD FACS Canto II system, BD Biosciences, San Jose, CA, USA). The flow panels included the markers CD2, CD3, CD4, CD5, CD7, CD8, CD10, CD11b, CD13, CD14, CD16, CD19, CD20, CD33, CD34, CD38, CD56, CD64, CD71, CD117, HLA-DR, TCRαβ, TCRγδ, and CD45. CD marker expression of >10% was deemed positive (18). Immature/mature granulocytes and monocytes were identified using a CD45/sideward scatter (SSC) backbone strategy and lineage-defined gating, as previously reported (19, 20). Briefly, granulocytes were identified as SSC^int/highCD45^dim/brightCD33^dim, and monocytes were identified as SSC^intCD45^dim/brightCD64⁺. Granulocytes and monocytes were also identified by differentiation antigen CD14 and CD16, respectively. Aberrant expression was determined on the basis of deviation from normal patterns (expression intensity at least 0.5 decades disparate from the corresponding normal cells from healthy pediatric stem cell donors). The percentage of the CD56⁺ population in myelomonocytes was assessed in a graded regimen: +, positive in 10%–30%; ++, positive in 31%–60%; +++, positive in >60%; negative, positive in <10%. The mean fluorescence intensities of CD56⁺ myelomonocytes were recorded.

All calculations were performed using SPSS 20 (IBM, Armonk, NY, USA). Continuous variables with non-normal distribution are expressed as the median and range, and the Mann–Whitney U-test was used for statistical analysis. Categorical variables are presented as frequencies and proportions (%) and analyzed by the χ² test (or Fisher's exact test for expected cell frequencies <5). The associations between clinical variables with response to cyclosporine were calculated by a multivariate logistic regression analysis. The Kaplan–Meier curve demonstrated the cumulative incidence of the response to cyclosporine and time to event, and the differences were evaluated by the log-rank test. P < 0.05 was considered statistically significant.

Of the 56 mAA patients enrolled, 29 were male and 27 were female (male:female ratio, 1.07:1). The median age was 4.7 years (range: 0.1–15.7 years). Table 1 presents the patients' characteristics. The median time between diagnosis and treatment was 7 days (range: 0–35 days). At mAA diagnosis, 14 patients (25%) presented with infection and the median C-reactive protein (CRP) level was 2.11 mg/L (range: 0.10–118.9 mg/L). Among the 56 patients, 24 had CD56⁺ myelomonocytes and were classified as CD56⁺ patients. The remaining 32 patients without CD56⁺ myelomonocytes were classified as CD56⁻ patients. There were no significant differences between the CD56⁺ and CD56⁻ patients in the gender and age distributions, duration between diagnosis and treatment, presence of infection, CRP concentration, and platelet count. However, the CD56⁺ patients had higher neutrophil counts and hemoglobin levels than the CD56⁻ patients at the time of diagnosis.

Analysis of the patients' bone marrow immunophenotype data revealed that all cases had normal expression of lymphocyte subset cell surface markers. The lymphocytic antigens involved in this study included: the T cell markers CD2, CD3, CD4, CD5, CD7, and CD8; B cell markers CD19 and CD20; and the NK cell markers CD16 and CD56. Immature/mature granulocytes and monocytes exhibited normal myelomonocytic markers: CD4, CD11b, CD13, CD14, CD16, CD33, CD64, and HLA-DR.

Twenty-four patients (24/56, 43%) had CD56⁺ myelomonocytes [three cases (13%) with CD56⁺ granulocytes, seven cases (29%) with CD56⁺ monocytes, and 14 cases (58%) with both CD56⁺ granulocytes and CD56⁺ monocytes]. CD56 was expressed frequently on CD16^dimCD33^dim immature granulocytes (15/24, 62.5%) and CD14^brightCD64⁺ mature monocytes (18/24, 75%). Aberrant CD56 expression was also observed on CD16^brightCD33^dim mature granulocytes (5/24, 20.8%) and CD14^dimCD64⁺ immature monocytes (3/24, 12.5%), but were infrequent. Table 2 details the immunophenotyping results. Figure 1 depicts representative CD56 dot plot distributions of the CD56⁺ and CD56⁻ patients on flow cytometry. Figure 2 shows the profile of mean fluorescence intensity of CD56⁺ myelomonocytes.

The proportions of monocytes, granulocytes, and lymphocytes in nucleated cells were assessed by flow cytometry. Compared with the CD56⁻ patients, the CD56⁺ patients had increased percentages of monocytes (median, 6.5% vs. 4.1%) and granulocytes (median: 42.4% vs. 17.8%; Figures 3A,B) and a higher proportion of mature monocytes (Figures 3C,D). The CD56⁺ patients had a higher percentage of immature granulocytes than the CD56⁻ patients (median: 16.2% vs. 4.3%) (Figure 3E).

Paralleling the increased immature granulocytes, the CD56⁺ patients had an increased percentage of mature granulocytes compared with the CD56⁻ patients (median: 17.6% vs. 6.3%; Figure 3F). Likely partly due to the increased myeloid cells, the CD56⁺ patients had a strongly reduced proportion of lymphocytes compared with the CD56⁻ patients (median: 23.9% vs. 59.6%), which included reduced B cells, CD4⁺ T cells, and CD8⁺ T cells (Figures 3G–J). Unsurprisingly, the CD56⁺ patients had a lower lymphoid/myeloid ratio than the CD56⁻ patients (median: 0.4 vs. 2.7) (Figure 3K). Interestingly, the CD56⁺ patients had an increased percentage of CD56^bright NK cells compared with the CD56⁻ patients (median: 9.7% vs. 5.8%, Figure 3L).

In China, cyclosporine is more frequently used for pediatric AA as it costs less and produces fewer adverse effects than anti-thymocyte globulin (21, 22). We evaluated the hematologic response of pediatric mAA patients to cyclosporine therapy at 3 and 6 months after treatment initiation. The 3-month response rate to cyclosporine did not differ significantly between the two groups (33% vs. 16%, P = 0.200, Figure 4A). At 6 months, 83% of the CD56⁺ patients achieved PR or CR. However, only 28% of the CD56⁻ patients responded to cyclosporine (P < 0.001, Figure 4B). Kaplan-Meier analysis demonstrated that the cumulative response rate of the CD56⁺ patients was significantly higher than that of the CD56⁻ patients (P < 0.001, Figure 4C). We also found that the cumulative response rates had no difference between patients with CD56⁺ mature monocytes and CD56⁺ immature monocytes (Supplementary Figure S1). Multivariate analysis of laboratory variables at the time of diagnosis revealed that age (older or younger than 4.7 years), sex (male or female), presence or absence of infection, peripheral platelet (more or less than 20 × 10³/µl), and reticulocyte count (more or less than 20 × 10³/µl) could not be deemed as predictors. Only two parameters (peripheral neutrophils >500/µl and abnormal CD56 expression) were significant factors associated with good response to cyclosporine therapy at 6 months (Table 3). Subsequently, we analyzed the patients' prognoses after cyclosporine therapy. The median follow-up time was 20.5 months (range: 10.2–40.4 months). In contrast to the response rate, the probability of evolution into sAA or relapse at 3 years tended to be higher in CD56⁻ patients (22%) than in CD56⁺ patients (8%) (Figure 4D).