To the Editor,

CEBPA-mutated (CEBPAmut) acute myeloid leukemia (AML) is recognized as a distinct entity in the updated WHO and ICC classification [1, 2]. While previously associated with double mutations only, the clinical-biological profile [3] and favorable outcomes [4] were later demonstrated to be related to in-frame mutations in the bZIP domain, occurring in ~90% of cases with bi-allelic mutations [5, 6]. CEBPAmut AML definition is largely overlapping between the WHO 5th edition [2], which includes biallelic or single bZIP CEBPA mutations, and the ICC [1], which only recognizes bZIP mutations as distinct entity defining.

Although correlating with favorable prognosis, a proportion of CEBPAmut patients still experience relapse. Due to the high rate of second complete remission (CR), CEBPAmut patients are usually allocated to allogeneic hematopoietic stem cell transplant (HSCT) in second CR [7]. However, an unexpectedly high incidence of treatment-related mortality (TRM) during CR (around 10%) was reported in two large multicenter studies involving intensively treated CEBPAmut patients [7, 8]. As regards co-mutational profile, no conclusive findings have been reached regarding its prognostic value in this setting [6, 9].

In view of clinical observations indicating persistent cytopenias despite CR achievement, we analyzed a cohort of intensively treated CEBPA-bZIP mutated AML patients with the aim to correlate baseline and treatment characteristics with outcomes and hematological toxicity.

The study was approved by the local institutional review board (protocol number: 2013/0021560). Written informed consent was obtained from all participants in accordance with the Declaration of Helsinki. From 2004 to 2023, 50 patients met the inclusion criteria at study sites (Firenze, Bergamo). The characteristics of this patient series are detailed in Table 1. Treatment details are described in the Supplemental file. The datasets generated and/or analyzed during the current study are available from the corresponding author on reasonable request.

Table 1 Characteristics of patients, overall and according to study site.
Full size table

Six patients were excluded from the analysis of interactions with treatment due to a shift to decitabine (n = 1), primary refractory disease (n = 2), death in aplasia (n = 1), and inability to proceed to consolidation (n = 2). In the first two cycles, corresponding to induction and first consolidation, the remaining 44 patients received no (n = 2, 4.6%), one (n = 14, 31.8%), or two (n = 28, 63.6%) anthracycline (ANTHRAC) -containing cycles. As regards high-dose cytarabine (HDAC), 19 (43.2%), 17 (38.6%), and 8 (18.2%) patients received no, one or two HDAC-containing cycles, respectively.

CR rate was achieved in 45 out of 49 cases (91.8%) after an intensive induction cycle, with a disease-free (DFS) and overall (OS) survival of 70.1% and 80.2% at 5 years, respectively. Forty-one (82.0%) patients had a mutation in the bZIP domain, of whom 35 (70.0%) had insertion/deletion (InDel) mutations, whereas 15 (30.0%) had other mutation types. As regards the co-mutational profile, 25 (50.0%) patients showed at least one mutation and were thus classified as NGS+ (Fig. S1). The most frequently involved genes were GATA2 (15 cases, 62.5% of mutated patients), DNMT3A (5, 20.8%), TET2 (4, 16.0%), and CSF3R (2, 8.3%). NGS status did not show any impact on CR after the first induction cycle (88.0% vs. 96.0% in NGS+ and NGSwt, respectively, P = 0.609). However, NGSwt status was associated with significantly longer DFS (HR = 3.20; 95% CI 0.98–10.41; P = 0.041) (Fig. 1). In multivariate analysis, the prognostic value of NGS maintained its significance (HR = 4.23, P = 0.047), together with WBC (HR = 4.42; P = 0.022), whereas age (HR = 0.96; P = 0.184) and karyotype (HR = 3.88; P = 0.060) were not significant. OS did not differ significantly, although a trend was observed for shorter survival in the NGS+ group (HR = 3.76; 95% CI 0.76–18.70; P = 0.081), that reached significance after censoring at HSCT (P = 0.034; Fig. S3), suggesting a favorable effect exerted by HSCT on NGS+ patients.

Fig. 1: Analysis of survival and time to hematopoietic recovery according to NGS status.
figure 1

Disease-free survival (A) and overall survival (B) according to the presence of mutations by NGS. Box plots illustrate the distribution of time to neutrophil ≥ 0.5 × 109/L (C) and platelet count ≥20 × 109/L (E) according to NGS and number of anthracyclines-containing cycles (0–1 versus 2). Cumulative incidence of recovery for neutrophil ≥0.5 × 109/L (D) and platelet count ≥20 × 109/L (F) for NGS+ patients receiving two cycles with anthracyclines (red curve) and other categories merged (blue curve). The curves of patients with no mutations (NGSwt) are depicted in blue; the curves of patients with at least one mutation (NGS+) are depicted in red. HR: hazard ratio; 95% CI: confidence interval. Box plots and cumulative incidence plots were generated by R software. Boxes represent the interquartile range containing 50% of the cases; the horizontal line marks the median; dots are single cases. KW Kruskal–Wallis test.

Full size image

We explored the impact of mutational status on hematopoietic recovery during CR. Overall, CEBPAmut patients had delayed platelet recovery after consolidation (median 26 days to platelets ≥50 × 109/L) compared to a set of intensively treated patients with intermediate-risk (IR) karyotype (n = 87) from our institutional database (24 days; P = 0.009). No significant differences were observed for neutrophil recovery (Table S3). To rule out any impact from chemotherapy dosage and scheduling, we performed propensity score matching for a number of cycles and cumulative dosages (Tables S3–5) of ARA-C and ANTHRAC. The time to recovery of platelets was confirmed longer for CEBPAmut compared to IR-CEBPAwt patients (P = 0.00018 and 0.0087 for platelets ≥50 × 109/L after matching for number of cycles and cumulative dose, respectively) (Figs. S6, S7).

We then explored the potential effect of NGS on recovery within CEBPAmut patients. NGS+ patients showed slower neutrophil recovery (median 22 days to ≥0.5 × 109/L) compared to NGSwt group (20 days P = 0.011, respectively) (Fig. S8). We observed similar findings for platelets: median days to count ≥50 × 109/L were 29 in NGS+ compared to 25 (P = 0.035) in NGSwt group.

After categorizing patients according to NGS status and the administration of two ANTHRAC-containing cycles (ANTHRAC-2), we observed a significantly delayed hematopoietic recovery in NGS+/ANTHRAC-2 patients (Kruskal–Wallis’s test P = 0.026 and 0.021 for ANC ≥ 0.5 × 109/L and platelet count ≥20 × 109/L, respectively) (Fig. 1). Specifically, NGS+/ANTHRAC-2 patients required significantly longer time to recover from the first consolidation, with 28 days for ANC ≥ 0.5 × 109/L and 33 days for platelet count ≥20 × 109/L, compared to 20 days (P = 0.008) and 18 days (P = 0.009), respectively, for the other categories. Consistent findings were noted for ANC ≥ 1.0 × 109/L and platelet count ≥50 × 109/L (Fig. S10). The categorization of patients according to NGS and delivery of HDAC did not show a significant correlation with recovery.

In an intention-to-treat analysis, as many as 46.1% (6/13) of NGS+/ANTHRAC-2 patients could not complete the planned chemotherapy program due to persistent cytopenia, compared to 18.7% (6/32, P = 0.075) in the other categories. With the limitation of a small number of cases, GATA2mut/ANTHRAC-2 showed particularly delayed neutrophil recovery, with a median of 38 days for ANC ≥ 1.0 × 109/L compared to 23 (P = 0.004) in the other categories (Fig. S13).

Our data align with the ongoing debate on the role of genotype in predicting the prognosis of CEBPAmut AML. Compared to other published experiences, we gathered patients harboring at least one co-mutation in a separate category (NGS+). Additional factors contributing to the variability in outcome may be related to differences in treatment protocols, including policies regarding allocation to HSCT. In our study, we describe previously unreported findings on hematological toxicity, with significantly slower hematopoietic recovery among CEBPAmut patients harboring additional NGS mutations. Despite the limitations owing to the retrospective nature of our study, we can infer that the observed hematological toxicity was the primary cause of reduced adherence to the scheduled treatment plan, which conventionally includes the delivery of four cycles and could not be indeed completed in almost half of CEBPAmut/NGS+ patients. We infer that prolonged cytopenias might have been responsible for the unexpectedly high rate of treatment-related mortality reported in two multicenter trials [7, 8]. Longer aplasia was further exacerbated by higher cumulative doses of anthracyclines in the first two chemotherapy cycles, especially in the GATA2mut group.

Two different patterns have been reported for additional mutations once CEBPAmut AML enters CR: while typical pre-leukemic gene mutations (i.e., DNMT3A, TET2) tend to persist [10], GATA2 mutations detected at initial diagnosis usually become no longer detectable [11]. The latter observation might be interpreted as if CHIP-related mutations, such as GATA2, impair normal hematopoietic recovery after chemotherapy, in analogy to what has been observed in multiple myeloma patients receiving high-dose chemotherapy. Another possibility is that CEBPAmut/GATA2mut cells may extrinsically affect hematopoietic recovery by impairing residual hematopoietic progenitors. Somatic mutations involving the GATA2 gene can induce pleiotropic effects, including qualitative defects in immune effector cells and an increased risk of invasive fungal infections [12, 13]. Extrinsic effects have been observed in other settings, such as the induction of an immunosuppressive tumor microenvironment by mutant-p53-harboring cells, which promotes immune escape.

While anthracyclines are an established standard approach in induction, no clear indications exist on their use in the consolidation phase, although several protocols include their repeated administration alongside intermediate/high ARA-C doses. There is still debate on the role of higher cumulative doses of anthracyclines in specific subsets such as FLT3-ITD mutated AML [14], and 2022 ELN recommendations indicate 3-4 cycles containing intermediate ARA-C dose as the standard approach [15].

Our findings support the incorporation of NGS data for the therapeutic management of the CEBPAmut AML subset. Rather than for prognostic stratification, the presence of additional mutations may suggest the opportunity to limit the cumulative dosage of anthracyclines to reduce hematological toxicity, which could otherwise impair the treatment plan. Moreover, it is surmisable that the use of anti-fungal prophylaxis also in the post-CR phase could help mitigate the risk of infections in this patient category.

Our work has some limitations that should be acknowledged. First, the retrospective design does not allow to exclude selection biases. Additionally, AML risk assessment and treatment allocation, regarding HSCT were likely influenced by the long enrollment period. Similarly, changes in supportive care, including the increasing use of triazole prophylaxis and growth factors, might have impacted compliance with treatment. Data on hematopoietic recovery could be partially affected by the timing of blood count cell assessment, especially after patient hospitalization, although overall adhering to standard clinical procedures.

In conclusion, our study on a set of CEBPA-bZIP mutated AML patients, beyond confirming a high response rate and favorable outcomes, provided novel evidence regarding delayed hematopoietic recovery, particularly in patients carrying additional mutations (especially in the GATA2 gene) who were treated with anthracyclines during the first two cycles. Confirmation of our findings in larger datasets is warranted.