TO THE EDITOR:

The transcription factor CCAAT/enhancer-binding protein-alpha(CEBPA) is a critical mediator of granulocytic differentiation. Mutations of the CEBPA gene (CEBPAmut) occur in 5%–15% of adult AML patients and in-frame mutations within the bZIP domain of CEBPA (CEBPAbZIP-inf) define a distinct entity associated with favorable prognosis in AML patients when treated by conventional chemotherapy [1, 2]. CEBPA mutations could activate the BCL2 P2 promoter and induce its expression via interaction with nuclear factor-κB (NF-κB) p50 in hematopoietic cell lines and display a markedly hypermethylated profile by multi-omics analysis in primary leukemia cells [3,4,5]. Meanwhile, venetoclax plus hypomethylating agents (VEN + HMA) were efficient in AML patients with specific molecular profiles (such as NPM1, IDH2, etc.) [6, 7]. However, relevant data related to the role of VEN + HMA regimens in CEBPAbZIP-inf AML patients is limited. Therefore, in the current study, we retrospectively analyzed the clinical features, co-mutational spectrum, and prognostic role of CEBPA mutations, particularly CEBPAbZIP-inf mutations, in 996 newly diagnosed adult AML patients inducted with chemo-free regimens or chemotherapy between 2016–2022 at the First Affiliated Hospital of Soochow University in China.

A high frequency of CEBPA mutations (17.8%,177/996) and CEBPAbZIP-inf cases (13.6%,135/996) were identified in our analysis (Fig. S1). CEBPAbZIP-inf patients were diagnosed at a younger age, higher hemoglobin counts, lower platelet counts, more likely to have intermediate cytogenetics (normal karyotype) and less number of co-mutations compared with other CEBPA mutated (CEBPAother-mut) and CEBPAwt AML patients (Table S1 and Fig. S2). CEBPAbZIP-inf mutations exhibited a higher complete remission/complete remission with incomplete cell recovery (CR/CRi) rate (87.6% vs. 64.1% vs. 47.3%, P < 0.001), favorable overall survival (OS) (3-year OS: 91.2% vs. 66.0% vs. 55.3%, P < 0.001) and relapse-free survival (RFS) (3-year RFS: 73.8% vs. 52.0% vs. 55.3%, P < 0.001) compared with CEBPAother-mut and CEBPAwt cases in the matched cohort (Fig. S3) (Table S2 and S3).

Among 130 CEBPAbZIP-inf patients with available information, 116 cases were inducted by 7 + 3 chemotherapy while 14 patients were by VEN + HMA regimens and consolidated by chemotherapy or hematopoietic stem cell transplantation (HSCT). There was no difference in the baseline and genetic characteristics between the 7 + 3 cohort and the VEN + HMA cohort (Table 1). With a median follow-up time of 22 months, among 13 patients evaluable all attained CR/CRi, 46.2%(6/13) relapsed and 3 patients died in the VEN + HMA group (disease progression, n = 1; transplant-related complications, n = 1; pneumonia, n = 1); simultaneously, 113 of 116 patients attained CR/CRi, 22.1%(25/113) relapsed and 6 cases died in the 7 + 3 group. Intriguingly, the VEN + HMA regimens seemed to show similar CR/CRi rates after one cycle of induction therapy (100.0% vs. 86.2%, P = 0.324), lower RFS (1-year RFS: 46.9% vs. 88.9%; P < 0.001) and OS (1-year OS: 84.6% vs. 99.1%; P < 0.001) than 7 + 3 regimens (Fig. 1A, B). When patients were censored at HSCT, worse 1-year RFS and 1-year OS were also observed in the VEN + HMA cohort (41.5% vs. 83.5%, P = 0.003 for RFS; 90.0% vs. 98.4%, P = 0.010 for OS) (Fig. 1C, D). In accordance with these findings, the multivariable analysis demonstrated adverse RFS (HR, 2.72; 95% CI: 1.01–7.30; P = 0.047) and a trend to dismal OS (HR, 5.14; 95% CI: 0.83–31.60; P = 0.078) in VEN + HMA cohort compared to 7 + 3 cohorts (Table S4).

Table 1 Baseline characteristics of patients with CEBPAbZIP-inf mutations in the 7 + 3 and VEN + HMA cohort.
Full size table
Fig. 1: Kaplan-Meier survival curves for overall survival (OS) and relapse-free survival (RFS) in CEBPAbZIP-inf patients.
figure 1

A, B compared survival outcomes among patients inducted with standard 7 + 3 and VEN + HMA regimens; C, D showed survival outcomes in the standard 7 + 3 and VEN + HMA cohorts with censoring at the time of HSCT; E, F illustrated survival by consolidation treatment (HSCT in CR1 or not). HSCT hematopoietic stem cell transplantation, CR1 first complete remission.

Full size image

Totally, 126 CEBPAbZIP-inf cases (96.9%) achieved CR/CRi post-induction and 31 of them (24.6%) relapsed, especially in the VEN + HMA cohort with a CR/CRi rate of 100% and relapse rate of 46.2%, which may be different from primary refractory group or persistent remission group and could be classified into “remission but relapse” group. Thus, it’s rather meaningful to explore high-risk indicators (such as co-mutations) for relapse and treatment regimens (such as consolidation) to reduce the relapse rate. With regard to consolidation therapy, 76 patients received cytarabine-based chemotherapy and 54 patients underwent transplantation in CR1 due to Measurable Residual Disease (MRD) positivity, more PTPN11, or FLT3-ITD mutations. HSCT at CR1 improved RFS (3-year RFS: 95.2% vs. 57.5%; P < 0.001) but not OS (3-year OS: 96.2% vs. 86.8%; P = 0.156) compared with chemotherapy alone in CEBPAbZIP-inf AML patients (Fig. 1E, F), which is consistent with Cox Proportional Hazard Model (HR 0.04, 95% CI: 0.01–0.19; P < 0.001 for RFS in multivariable analysis; HR 0.34; 95% CI: 0.07–1.70; P = 0.184 for OS in univariable analysis) (Table S4).

On the other hand, overlapping gene mutations may affect the clinical outcome of CEBPAbZIP-inf patients. The most frequently co-mutated genes within the CEBPAbZIP-inf group were WT1 (27.7%), GATA2 (26.2%), NRAS (13.8%), FLT3-ITD (13.1%) and RUNX1 (10.8%) (Fig. S2). Notably, KIT mutations had a significantly poor impact on RFS (P < 0.001) and OS (P = 0.002) compared to KIT wild type in CEBPAbZIP-inf group (Fig. S4), and CSF3R mutation showed a worse RFS (P = 0.035) by performing landmark analysis after 10 months (crossover) (Fig. S4). Meanwhile, WT1 and NRAS mutations revealed a tendency toward adverse RFS (P = 0.053, P = 0.067) and OS (P = 0.068, P = 0.069) (Fig. S4). Additionally, MDS-related gene mutations (MRs, including ASXL1, BCOR, EZH2, SF3B1, SRSF2, STAG2, U2AF1, ZRSR2) and GATA2 aberrations had no effect on survival (Fig. S4).

Vazquez previously found that all 4(100%) CEBPA-mutated AML patients achieved CR/CRi but 3(75%) relapsed with a shorter duration of remission (median: 4.65 months) in 19 unfit AML patients treated by venetoclax-based regimens [8]. Similarly, our cohort suggested that VEN + HMA induction exhibited a similar remission rate, worse RFS, and a tendency to poor OS in CEBPAbZIP-inf cohort compared with 7 + 3 groups. It seems the treatment depth or clonal evolution of VEN + HMA needs to be further explored. On the side, as earlier studies reported [9, 10], we also demonstrated HSCT in CR1 significantly improved RFS but not OS compared to chemotherapy alone, possibly resulting from a high remission rate of re-induction by chemotherapy after relapse and high transplantation-related mortality. Therefore, HSCT is not recommended in consolidation therapy at first CR in CEBPAbZIP-inf AML patients. In other words, conventional chemotherapy is preferred in both induction and consolidation courses among CEBPAbZIP-inf AML cases.

The co-occurrence of other genetic mutations has had a controversial prognostic impact in patients with CEBPAmut AML by previous studies, generally, mutations of WT1, CSF3R, KIT, NRAS, and CCS mutations (mutations in chromatic/DNA modifiers (C), cohesion complex (C), and splicing genes (S)) were associated with adverse prognosis while mutations of GATA2 were correlated with favorable outcome [9, 11,12,13,14,15]. Similar to these reports, KIT mutation was significantly related to inferior RFS and OS in Kaplan-Meier methods (RFS, P = 0.002; OS, P < 0.001) but not in multivariable analysis (HR 2.45, 95% CI: 0.63–9.55, P = 0.197 for RFS; HR 2.63, 95% CI: 0.30-23.04, P = 0.381 for OS), a possible explanation maybe the small sample size of KIT mutated patients in CEBPAbZIP-inf patients. Inferior prognosis in terms of RFS was found in WT1(P = 0.053), CSF3R (P = 0.035), and NRAS (P = 0.069) mutated group while a trend of worse OS without statistical significance was observed (WT1, P = 0.068; NRAS, P = 0.067). However, no clinical significance of GATA2 mutation and MRs were found in the present cohort.

Recently, Georgi divided 1010 CEBPAmut adult AML patients into 8 mutational subgroups and refined superior prognosis was associated with in-frame insertions/deletions within bZIP domain of CEBPA (CEBPA bZIPInDel) rather than in-frame mutations within bZIP domain of CEBPA (CEBPAbZIP-inf) by excluding missense mutations within bZIP domain (CEBPA bZIPms) [15]. The present CEBPAbZIP-inf AML cohort was reanalyzed and 14 CEBPA bZIPms patients were excluded, the prognostic value of VEN + HMA induction and HSCT at CR1 consolidation was consistent with previous results (Fig. S5). Whether truncated mutations in the N-terminus and truncated sites influence the clinical outcome of CEBPA-mutated patients were analyzed. No statistical significance of OS (P = 0.490) and RFS (P = 0.412) were found in CEBPA transactivation domain (TAD) frameshift mutated patients and CEBPA other mutated patients (Fig. S6). No differences in OS (P = 0.240) and RFS (P = 0.571) were evaluated in N-terminal truncated mutations spanning the second start codon (Fig. S6).

In conclusion, CEBPAbZIP-inf patients exhibited higher CR rates, improved OS and RFS, and benefited from traditional chemotherapy compared with VEN + HMA induction and transplantation consolidation at CR1 in our preliminary study. Furthermore, CEBPAbZIP-inf patients with WT1 and NRAS mutation adversely affected the RFS in multivariable analysis. Taken the retrospective nature and limited sample size into consideration, future research efforts aimed at validating our results and elucidating the potential molecular mechanisms are warranted to improve therapeutic strategy in specific types of CEBPAmut AML.