Impact of <scp> <i>CSF3R</i> </scp> Mutations on Leukemic Transformation in Severe Congenital Neutropenia: A Retrospective Analysis From the Italian Neutropenia Registry

Patients with severe congenital neutropenia (SCN) have an intrinsic risk of developing myelodysplasia and/or acute leukemia (MDS/AML), with transformation rates varying between 10% and 30% across different registries [1-4]. However, leukemic transformation in SCN remains poorly understood. Acquired gain-of-function CSF3R variants generate a truncated G-CSF receptor and induce sustained and dysregulated proliferative signaling [5]. These mutations are reported in up to 40% of SCN patients and are implicated in leukemogenesis [6-9]. However, CSF3R mutations alone are usually insufficient to drive malignant transformation, as longitudinal sequencing studies have shown long-term persistence of CSF3R-mutant clones or even their loss during leukemic progression. These observations support a multistep, cooperative model of leukemogenesis, in which additional somatic variants—such as RUNX1 or TET2—together with chronic inflammatory stress within the bone marrow microenvironment synergistically promote leukemic transformation [10-12]. Nowadays, highly sensitive sequencing technologies enable the detection of small somatic CSF3R clones, but the interpretation of their biological and clinical relevance is challenging. In this study, we longitudinally evaluated the dynamics of CSF3R clones in an Italian cohort of SCN patients and examined their association with progression to MDS/AML, aiming to better understand their role in leukemic transformation and optimize surveillance. We screened SCN patients followed from 1985 to 2025 at 12 centers in Italy and enrolled in the Italian Neutropenia Registry held at the IRCCS Gaslini in Genoa, Italy. Patients were considered eligible if they had a diagnosis of SCN according to international guidelines [10], were treated with continuous G-CSF, and had at least one assessment of somatic CSF3R variants during the follow-up period. Data were retrieved from the Italian Neutropenia Registry and, whenever necessary, collected from clinical charts. All patients were assessed for somatic CSF3R variants at the Genetics Laboratory of the Meyer Children's Hospital IRCCS (Florence), as reported in Supporting Information: Supplementary Methods. Outcomes included progression to MDS/AML, hematopoietic stem cell transplantation (HSCT), and survival. The study was conducted in accordance with the 1964 principles of the Declaration of Helsinki and its later amendments [13]; written informed consent was collected from each patient or their legal guardian. Sixty-five patients underwent longitudinal screening for CSF3R variants; of them, eight were excluded because they did not meet the eligibility criteria (neutropenia other than SCN, and lack of available data). The baseline characteristics of the 57 included patients (females n = 29, 51%) are summarized in Table 1. The median age at SCN diagnosis was 12 months (IQR 3–65) and the median follow-up 7 years (IQR 1.9–10.7). Thirty-five patients (61%) harbored ELANE variants, while 10 (18%) had no detectable germline variants (i.e., gene-orphan). The remaining 12 patients carried variants in SRP54 (n = 3, 5%), SLC37A4 (n = 3, 5%), HAX1 (n = 2, 4%), DADA2 (n = 2, 5%), TAZ (n = 1, 2%), and FNIP1 (n = 1, 2%) [14]. All patients were on G-CSF treatment, with a median age at the G-CSF initiation of 12 months (IQR 3–51). Somatic CSF3R variants were found in 18 patients (32%), with a longitudinal follow-up of at least two samples tested. Among them, 13 carried variants in ELANE, two in HAX1, one in SRP54, and two were gene-orphan. Age, sex, germline mutations, and G-CSF exposure did not differ between patients with or without CSF3R variants (Table 1). Sixteen different CSF3R clones were detected during follow-up, with c.2390C>T being the most frequent (n = 7, 39%; Table S1 and Figure S1A). Nine patients carried a single clone, while the remaining nine developed ≥ 2 clones over time (Figures 1 and S1A). The number of clones was not associated with evolution to MDS/AML. We also analyzed co-occurrence among CSF3R clones and found that the c.2215C>T and c.2384C>T clones were those most frequently associated (Figure S1B). Longitudinal analyses revealed marked variability in clone size overtime, with the largest expansion observed between 5 and 25 years of age. Only few patients were followed after 25 years of age and some clones persisted over a long follow-up period in this subgroup (Figure 1). While most CSF3R clones remained below 10% across all determinations, clones up to > 20% (i.e., c.2227C>T, c.2245C>T, c.2384C>T, c.2221C>T, and c.2396C>T) were detected in few patients. Notably, in four out of six patients who developed MDS/AML, the transformation was preceded by the expansion of a single CSF3R clone (Figure 1). The disappearance of CSF3R-mutant clones was also frequently observed, occurring in all patients harboring the c.2384C>T (n = 4), c.2414C>T (n = 4), c.2433del (n = 1), c.2396C>T (n = 4), c.2212A>C (n = 1), and c.2256 T>G (n = 1) variants. The characteristics of CSF3R clones were also compared between patients with and without leukemic transformation (Table S1). Patients who developed MDS/AML seemed to develop earlier and more frequently multiple CSF3R clones, although these differences were not statistically significant. Analysis of individual clones identified a significant association between the c.2215C>T clone and leukemic transformation (p = 0.025). Moreover, the CSF3R clones were significantly larger in patients who evolved to MDS/AML than in those without malignant transformation (30.3% vs. 3.3%, p = 0.009). Overall, seven patients (12%) developed MDS/AML at a median age of 17 years (IQR 8–19). Among them, five harbored ELANE variants and two HAX1 variants; six (86%) had CSF3R clones. Fourteen patients (25%) underwent HSCT at a median age of 3 years (IQR 2–12), with HSCT occurring at a significantly older age in patients with CSF3R clones (15 vs 3 years, p = 0.022). During follow-up, six patients (11%) died because of infections (n = 1), transplant-related mortality (n = 3), evolution to MDS/AML (n = 1), or sudden death (n = 1). The estimated probability of survival in our cohort was 77% (95% CI 45–91) at 50 years (Figure 2A) and was not different between patients with and without CSF3R clones (p = 0.694; Figure 2B). The estimated MDS/AML-free survival probability of the whole cohort was 80% (95% CI 60–91) at 50 years (Figure 2C). The frequency of leukemic transformation was significantly higher in patients with CSF3R clones (33% vs. 3%, p = 0.004; Table 1), reflected by a significantly inferior probability of evolution-free survival (log rank p = 0.013; Figure 2D) of this subgroup. The difference was further accentuated in patients who developed clones > 10% (log rank p = 0.009; Figure 2E) and > 20% (log rank p = 0.002; Figure 2F). Notably, patients who developed MDS/AML in the period immediately preceding clonal evolution exhibited poor control of infectious episodes and required an increase in G-CSF dosing of at least 10%–15% compared with baseline. No RUNX1 mutations were detected in the four transformed patients who were tested. The regular monitoring of clonal hematopoiesis through panels including the most frequently involved variants was not available for all patients who developed leukemic transformation. Among the three patients analyzed using a myeloid panel including nearly 50 genes, ASXL1, SETBP1, IDH2, and SF3B1 were the most frequent pathogenic variants detected shortly before transformation, together with CSF3R. A subgroup analysis of patients with ELANE germline mutations (n = 35, Table S2) showed no significant differences between those with and without progression to MDS/AML in terms of the predicted leukemogenic risk of ELANE variants [15], duration of G-CSF therapy, and need for G-CSF dose escalation (Table S3). The CSF3R clones were significantly larger in patients who progressed to MDS/AML than in those without malignant transformation (31% vs. 8%, p = 0.007), and the risk of evolution was significantly higher in patients with clonal expansions exceeding 10% (p = 0.003) and 20% (p = 0.002). In this study, we analyzed a large cohort of 57 SCN patients followed longitudinally with regular screening for CSF3R variants. We characterized the clonal dynamics and clinical significance of CSF3R variants and assessed their predictive value for leukemic transformation. Overall, CSF3R clones were detected in 33% of the whole cohort, where the prevalence of leukemic transformation was 12%. All the identified variants were truncating mutations, leading to loss of normal protein function. While clonal heterogeneity and multiplicity reflected the known stochastic behavior of CSF3R variants [7] and were not predictive of transformation, the presence of CSF3R mutant clones was significantly associated with progression to MDS/AML, with a corresponding increase in the estimated risk of transformation. Notably, whereas c.2211delC, c.2336delG, c.2173C>T, and c.2212A>C variants were not observed in patients who progressed to MDS/AML, the c.2215C>T clone was significantly associated with leukemogenesis in our cohort. More importantly, the estimated probability of leukemic transformation increased progressively with the clone size, with a striking significance for clones exceeding 10% and 20%, a finding confirmed also in the ELANE-mutated subgroup. This study has limitations, including its retrospective design, the heterogeneity of techniques and follow-up procedures, incomplete data on cooperating somatic mutations such as RUNX1, and the reduced cohort size that precluded multivariable analysis. Nevertheless, our findings underscore the clinical importance of clone size as a potential surrogate marker of transformation and reinforce the value of periodic monitoring for CSF3R variants in patients with SCN, without the need to routinely apply ultrasensitive techniques. The appearance of a CSF3R clone exceeding 10% or evidence of clonal expansion over time should prompt intensified clinical surveillance and may help guide timely therapeutic decision-making, in accordance with the indications of the International Guidelines (Figure S2) [16]. The authors have nothing to report. The authors have nothing to report. The study was conducted in accordance with the 1964 principles of the Declaration of Helsinki and its later amendments. Written informed consent was collected from each patient or their legal guardian. The authors declare no conflicts of interest. The data that support the findings of this study are available from the corresponding author upon reasonable request. Data S1: Supplementary methods. Figure S1: Characteristics of the CSF3R clones. (A) Representation of CSF3R clones detected during the follow-up period in the patients included in the study; the weight of the lines connecting patients to individual clones is proportional to the maximum size of the individual clones identified during follow-up. (B) Representation of correlation between CSF3R clones during the follow-up period in the patients included in the study; the weight of the lines connecting clones is proportional to the strength of their correlation (based on the Jaccard index). Figure S2: Proposed surveillance algorithm in SCN. Proposal of surveillance protocol based on longitudinal CSF3R clonal monitoring. *The panel should include at least RUNX1, TP53, TET2, ASXL1, DNMT3A. Given the comparable sensitivity of clone detection on bone marrow and peripheral blood, the latter is acceptable for routine testing. Nevertheless, karyotype examination of the bone marrow is strongly indicated in SCN according to international guidelines. AML, acute myeloid leukemia; BM, bone marrow; G-CSF, granulocyte-colony stimulating factor; HSCT, hematopoietic stem cell transplantation; MDS, myelodysplastic syndrome. Table S1: Characteristics of the CSF3R clones in patients with and without progression to MDS/AML. Table S2: ELANE germline mutations detected in the 35 patients included in the study. Table S3: Comparison between patients with and without evolution to MDS/AML in the subgroup of patients with ELANE germline mutations. Please note: The publisher is not responsible for the content or functionality of any supporting information supplied by the authors. Any queries (other than missing content) should be directed to the corresponding author for the article.