Adult AML with NUP98 rearrangements should be stratified into adverse-risk group

TO THE EDITOR,

NUP98, a nuclear pore component, forms chimeric proteins through rearrangements (NUP98-r) involving its N-terminal domain and the C-terminal regions of partner genes. Over 30 NUP98 fusion partners have been identified in hematologic malignancies, particularly acute myeloid leukemia (AML) [1]. While the 2022 WHO [2] and ICC [3] classifications recognize NUP98-r AML, its risk stratification remains undefined in the 2022 ELN guidelines [4]. Only one study has suggested a trend toward categorizing NUP98::HOXA9 AML as adverse-risk, but the data did not reach statistical significance [5]. Venetoclax (VEN), a BCL-2 inhibitor, has transformed AML treatment, especially for chemotherapy-ineligible patients [6]. Combining venetoclax with hypomethylating agents (VEN + HMA) has shown promise in improving response rates, but its efficacy in NUP98-r AML patients is not yet fully understood. To address this, we conducted a retrospective study analyzing 65 NUP98-r AML patients (aged 16-65 years) treated with either intensive chemotherapy (IC) or VEN + HMA (Fig. S1). Additionally, outcomes of NUP98-r AML patients receiving IC were compared to 731 non NUP98-r AML patients stratified according to ELN-2022.

A total of 65 newly diagnosed AML patients aged 16-65 years with NUP98-r were enrolled, including 29 (44.6%) with NUP98::HOXA9, 20 (30.8%) with NUP98::NSD1, and 16 (24.6%) with other NUP98::V fusions (Fig. 1A, B). Baseline patient characteristics are summarized in Table 1. No significant differences were observed in median age, sex distribution, white blood cell counts, hemoglobin levels, or platelet counts at diagnosis across the groups. However, bone marrow blast percentages at diagnosis were significantly higher in NUP98::NSD1 group compared to the other two subgroups (median: 63.25% vs. 44.50% and 38.00%; p = 0.033). A big difference was observed in the distribution of FAB classifications, most NUP98::HOXA9 patients were M2 subtype, whereas most NUP98::NSD1 patients were M4/M5 subtype (M2: 82.8% vs. 20.0%; M4/M5: 10.3% vs. 80.0%; p < 0.001). Patients with NUP98::HOXA9 frequently harbored chromosome 11 abnormalities and co-mutations in WT1 (34.5%), NRAS (24.1%), and KRAS (24.1%). In contrast, NUP98::NSD1 patients predominantly exhibited normal karyotype and concomitant FLT3-ITD mutations (65.0%) (Table 1 and Fig. 1C).

Fig. 1. Baseline of NUP98-r AML and response to intensive chemotherapy and Venetoclax plus hypomethylating agents.

A Circos plot representing different fusion partner genes of NUP98-X translocations in adult AML patients. B Pie charts depicting the partition of different NUP98 fusions in our cohort. C Baseline characteristics landscape of 65 adult AML with NUP98 rearrangements patients. D Complete remission rate per different chemotherapy in NUP98 rearrangement and subtypes. E Complete remission rate per different NUP98 partners in same induction regimen. F Swimming plot for 21 adult NUP98-r AML receiving venetoclax plus hypomethylating agents. CR complete remission, PR partial response, NR no response, HSCT hematopoietic stem cell transplantation.

Table 1.

Baseline characteristics for 65 NUP98 rearrangement AML patients receiving IC or VEN + HMA.

		Different fusion proteins
Characteristic	Total n = 65	NUP98::HOXA9 n = 29	NUP98::NSD1 n = 20	Others$ n = 16	p value
Age, years, Median [IQR]	40.00 [30.00, 50.00]	40.00 [30.00, 45.00]	44.00 [38.50, 50.25]	33.50 [27.00, 47.25]	0.439
Gender, n (%)					0.284
Female	30 (46.2)	11 (37.9)	9 (45.0)	10 (62.5)
Male	35 (53.8)	18 (62.1)	11 (55.0)	6 (37.5)
Hematological parameters
WBC, ×109/L, Median [IQR]	46.50 [7.12, 101.53]	39.58 [6.27, 83.83]	73.57 [10.02, 139.68]	38.70 [12.81, 97.25]	0.374
PLT,×109/L, Median [IQR]	47.00 [30.00, 60.00]	39.00 [26.50, 55.00]	51.00 [43.25, 57.75]	44.50 [29.75, 78.75]	0.117
Hb, g/L, Median [IQR]	80.00 [69.00, 97.00]	75.00 [67.50, 88.00]	81.00 [70.75, 96.00]	91.00 [77.55, 103.75]	0.228
Bone marrow blasts, %, Median [IQR]	49.25 [28.38, 69.38]	44.50 [25.50, 57.75]	63.25 [47.88, 77.38]	38.00 [20.75, 57.75]	0.033
FAB Category, n (%)					<0.001
M0	1 (1.5)	1 (3.4)	0 (0.0)	0 (0.0)
M1	2 (3.1)	1 (3.4)	0 (0.0)	1 (6.2)
M2	38 (58.5)	24 (82.8)	4 (20.0)	10 (62.5)
M3	0 (0.0)	0 (0.0)	0 (0.0)	0 (0.0)
M4	11 (16.9)	3 (10.3)	5 (25.0)	3 (18.8)
M5	13 (20.0)	0 (0.0)	11 (55.0)	2 (12.5)
M6	0 (0.0)	0 (0.0)	0 (0.0)	0 (0.0)
M7	0 (0.0)	0 (0.0)	0 (0.0)	0 (0.0)
Chromosomal aberration, n (%)
Normal karyotype	19 (29.2)	2 (6.9)	13 (65.0)	4 (25.0)	<0.001
t(7;11)	23 (35.4)	21 (72.4)	0 (0.0)	2 (12.5)	<0.001
Abnormality11	9 (13.8)	4 (13.8)	0 (0.0)	5 (31.2)	0.026
Trisomy8	5 (7.7)	2 (6.9)	3 (15.0)	0 (0.0)	0.239
Complex karyotype	7 (10.8)	7 (24.1)	0 (0.0)	0 (0.0)	0.008
Mutation, n (%)
FLT3-ITD	22 (33.8)	5 (17.2)	13 (65.0)	4 (25.0)	0.002
WT1	19 (29.2)	10 (34.5)	4 (20.0)	5 (31.2)	0.537
NRAS	12 (18.5)	7 (24.1)	1 (5.0)	4 (25.0)	0.175
KRAS	12 (18.5)	7 (24.1)	1 (5.0)	4 (25.0)	0.175
TET2	11 (16.9)	5 (17.2)	3 (15.0)	3 (18.8)	0.955
RUNX1	8 (12.3)	2 (6.9)	2 (10.0)	4 (25.0)	0.195
PTPN11	7 (10.8)	3 (10.3)	1 (5.0)	3 (18.8)	0.415
CEBPA	6 (9.2)	2 (6.9)	4 (20.0)	0 (0.0)	0.101
GATA2	4 (6.2)	2 (6.9)	0 (0.0)	2 (12.5)	0.293
NPM1	2 (3.1)	0 (0.0)	1 (5.0)	1 (6.2)	0.426

WBC white blood cell, PLT platelet, Hb hemoglobin. Others^$ represent a group of patients with NUP98:V; V refers HOXA11, HOXA13, HOXC13, HOXD12, PRRX2, LNP1, TOP1, MLLT10, KDM5A and AC087441.2.

The overview and prognosis of IC and VEN + HMA treatments in our cohort are summarized in Table S1. Among 65 NUP98-r AML patients, 44 (67.7%) received IC regimens as initial induction and 21 (32.3%) were treated with VEN + HMA. No significant difference was observed in complete remission (CR) rates between the two regimens (IC: 52.3% vs VEN + HMA: 57.1%, p = 0.794, Fig. 1D and Table S2). Notably, the NUP98-r genetic subtypes showed differential responses. In the NUP98::HOXA9 cohort (n = 29), all 100% (5/5) of patients treated with VEN + HMA achieved CR, compared to 62.5% (15/24) in the IC group (p = 0.153, Fig. 1D and Table S2). For both NUP98::NSD1 (n = 20) and NUP98::V fusions (n = 16), CR/CRi rates showed no significant differences between two regimens (both p > 0.05). Patients with NUP98::NSD1 demonstrated markedly inferior response to IC compared to both NUP98::HOXA9 (11.1% vs 62.5%, p = 0.017, Fig. 1E and Table S3) and NUP98::V fusions (11.1% vs 63.6%, p = 0.028, Fig. 1E and Table S3).

Notably, the NUP98-rearranged subgroups demonstrated distinct responses to VEN + HMA therapy, with the NUP98::HOXA9 subgroup achieving a 100% CR rate compared to only 36.4% in the NUP98::NSD1 subgroup (p = 0.034, Fig. 1E). This differential response correlates with their characteristic FAB classifications. NUP98::HOXA9 predominantly presents as FAB-M2 subtype (82.8%), while NUP98::NSD1 is mainly associated with monocytic (FAB-M4/M5) leukemias (80.0%). Previous studies have documented that BCL-2 expression progressively decreases during myeloid progenitor differentiation and maturation, with elevated BCL-2 levels observed in M2 leukemic cells versus lower expression in monocytic leukemias [7]. Multiple researches have revealed primary resistance to venetoclax in leukemias with prominent monocytic features, which aligns well with our current results [8, 9].

For the launch of venetoclax in China in December 2020, the maximum follow-up duration was 48 months. Due to the relatively low number of NUP98-r AML patients treated with VEN + HMA, survival curves were not generated. Disease progression for each of the 21 patients receiving VEN + HMA as initial induction therapy, categorized by partner genes, is detailed in Fig. 1F. 4 of 5 NUP98::HOXA9 patients were alive at the end of follow-up, and 1 died from relapse. 2 NUP98::HOXA9 patients performing allo-HSCT were alive. In the NUP98::NSD1 cohort, 4 (36.4%) achieved CR after initial VEN + HMA induction, 7 (63.6%) finally achieved CR, but 4 (57.1%) have experienced recurrence. 2 of 10 (20.0%) who have undergone transplantation died from recurrence. In the NUP98::V group, 3 (60.0%) achieved CR after initial VEN + HMA induction, 4 (80.0%) ultimately reached CR, and 2 (50.0%) had recurrence. 1 of 2 (50.0%) transplant recipients died from recurrence.

Collectively, our findings demonstrate that the VEN + HMA regimen achieves a higher CR rate than traditional IC (100.0% vs. 62.5%) in adult AML patients with NUP98::HOXA9. Consistent with our results, a study by Jie Tian et al. also reported an excellent response to VEN + HMA in this subtype (2/2, 100% CR) [10]. While VEN + HMA shows promising efficacy in initial induction therapy in NUP98::HOXA9 AML, larger-scale retrospective analyses or prospective studies are warranted to further validate the efficacy of venetoclax in this subtype.

For other NUP98-r subtypes, particularly NUP98::NSD1, the VEN + HMA regimen did not improve CR rates compared to IC. The KMT2A complex is a key factor regulating the pathogenesis of NUP98 fusion genes. Targeting this pathway with menin inhibitors (e.g., revumenib) could be a rational therapeutic strategy. Indeed, preclinical studies have shown the efficacy in NUP98::NSD1 leukemia models, supporting further exploration of this approach [11, 12].

Otherwise, we identified FLT3-ITD co-mutations in 22 of 65 (33.8%) NUP98-r patients, with marked subtype-specific differences: only 5 of 29 (17.2%) NUP98-HOXA9, 13 of 20 (65.0%) NUP98-NSD1 (Table 1 and Fig. 1C). Of the 22 FLT3-ITD mut patients, 8 received TKIs treatment (sorafenib or gilteritinib), including 7 with NUP98-NSD1 and 1 with NUP98-HOXA9. TKIs were not used during initial induction in our cohort. Among 4 patients treated after induction failure, 1 (25%) achieved CR with sorafenib. Of 2 patients treated at relapse, 1 (50%) achieved CR with gilteritinib. FLT3-ITD is highly prevalent in NUP98::NSD1 AML, in our cohort TKIs treatment was not used in initial induction and showed limited efficacy. Perhaps TKIs therapy may provide greater benefits to patients during the initial induction phase, which may require more clinical trial data to validate.

Furthermore, we compared 44 adult NUP98-rearranged AML patients treated with IC to 731 non NUP98-r AML patients with the same initial induction regimen stratified by ELN-2022 risk criteria. Baseline demographics, genetic profiles, therapeutic responses, and outcomes are presented in Table S4. Baseline characteristics and detailed subgroup distributions of non NUP98-r group based on ELN-2022 are shown in Table S5, 6. The NUP98-r cohort was younger than adverse-risk patients (median years: 38 vs. 42, p = 0.007) and showed distinct hematologic features: markedly higher WBC counts (48.39 vs. 9.60–25.93 × 10⁹/L, p < 0.001) but lower bone marrow blasts (48% vs. 54–62%, p < 0.001) compared to all risk groups.

NUP98-r patients achieved comparable CR rates to the adverse-risk group (p > 0.05, Fig. S2). The overview and prognosis of IC treatments in different NUP98 fusions are summarized in Table S7. Survival outcomes in the ELN-2022 Risk group and subgroup are shown in Fig. S3. Survival analysis revealed NUP98-r AML exhibited significantly inferior 5-year overall survival (OS: 53.3% vs. 71.3%) and relapse-free survival (RFS: 28.6% vs. 57.2%) compared to the intermediate-risk group (median OS: not reached vs. not reached, p = 0.033, Fig. 2A; median RFS: not reached vs. 19.5 months, p = 0.022, Fig. 2B), showing no statistical differences in OS and RFS when compared to the adverse-risk group (OS: p = 0.462, Fig. 2A; RFS: p = 0.257, Fig. 2B).

Fig. 2. NUP98 rearrangement AML versus control groups receiving IC and the impact of allo-HSCT.

A, B Survival outcomes among control groups and NUP98-r AML. C, D Survival outcomes among control groups and NUP98::HOXA9 AML. E Overall survival among HSCT in CR, salvage HSCT and no HSCT in NUP98-r. F Overall survival between HSCT or not in NUP98-r. G Overall survival between HSCT after CR1 or not in NUP98-r AML. H Overall Survival between HSCT in PR/NR or not in NUP98-r AML. HSCT hematopoietic stem cell transplantation, CR1 first complete remission, HSCT hematopoietic stem cell transplantation, PR partial remission, NR no response.

For NUP98::HOXA9 patients, survival outcomes were markedly worse than intermediate-risk group, both in 5-year OS (37.5% vs. 71.3%) and RFS (17.8% vs. 57.2%) diverging substantially (median OS: 20 months vs. not reached, p = 0.001, Fig. 2C; median RFS: 14 months vs. not reached, p = 0.002, Fig. 2D).

Among NUP98-r patients who receiving IC in first induction chemotherapy, those undergoing allogeneic hematopoietic stem cell transplantation (allo-HSCT) (Fig. 2E) demonstrated significantly superior overall survival compared to non-transplanted counterparts (median OS: not reached vs. 13 months; p < 0.001, Fig. 2F). NUP98-r patients who underwent allo-HSCT in first complete remission (CR1) showed markedly better OS than those receiving consolidation chemotherapy alone (OS: p = 0.011, Fig. 2G). In relapsed/refractory NUP98-rearranged patients, salvage transplantation was associated with a marked improvement in OS (p = 0.009, Fig. 2H). The impact of HSCT in different times for control groups and compared to NUP98-r are shown in Fig. S4, 5. Cox regression analyses further identified allo-HSCT as the sole independent prognostic factor for OS in NUP98-r AML treated with IC (univariate HR: 2.682, p < 0.001; multivariate HR: 3.458, p = 0.002; Fig. S6). No other clinical variables showing no significant associations in both two analysis (p > 0.05).

Under IC induction therapy, NUP98-r AML patients exhibited significantly inferior 5-year OS and RFS rates compared to the intermediate-risk group. Allo-HSCT significantly improve outcomes in these patients. Based on these findings, we propose classifying NUP98-r AML into the adverse-risk group. The optimized ELN-2022 risk stratification system, which incorporates NUP98 rearrangements, demonstrated statistically significant differences in OS and RFS (both p < 0.001, Fig. S7).

Our findings regarding the risk stratification of NUP98::HOXA9 AML are consistent with reported by Min-Yen Lo et al. [5]. However, our study expands the sample size and incorporates the impact of allo-HSCT. Moreover, all results are statistically significant. In 2018, Kaito Harada et al. published a study examining the differences in post-transplant outcomes between patients with AML carrying the t(7;11)(p15;p15) translocation and those with intermediate- or high-risk AML, as defined by the standards provided in the National Comprehensive Cancer Network (NCCN) AML guidelines, Version 1 [13]. They found that patients with t(7;11)(p15;p15) who underwent HSCT in CR1 had superior outcomes compared to those with intermediate- or high-risk AML. Among patients with t(7;11)(p15;p15), HSCT in CR1 yielded better outcomes than those in CR2 or in high-risk categories. These findings differ from our results, which may be due to differences in control group classification criteria. Additionally, their study did not mention any variations in pre-transplant induction regimens, and the patient cohort included individuals treated from 1986 to 2014, which could also contribute to the differing outcomes.

Our study demonstrates that NUP98-r AML patients have poor outcomes regardless of chemotherapy regimen (IC or VEN + HMA), with allo-HSCT being the only intervention that significantly improves survival. Both HSCT in CR1 and salvage transplantation provided meaningful survival benefits. Currently, allo-HSCT remains the optimal therapeutic strategy for NUP98-r AML. For patients who fail to achieve remission, salvage transplantation may still offer survival benefits. Furthermore, efforts should be made to maximize remission rates through venetoclax-based regimens, menin inhibitors, and others, as achieving first complete remission prior to allo-HSCT, which is strongly associated with optimal survival outcomes for NUP98-r patients.

In conclusion, under IC induction therapy, AML with NUP98 rearrangements demonstrated significantly inferior OS and RFS, comparable with the adverse-risk group. Our study suggests that NUP98-r AML should be classified into the adverse-risk group. Compared to traditional intensive chemotherapy, venetoclax with hypomethylating agents has shown efficacy in initial induction therapy in adult AML patients with NUP98::HOXA9. However, for NUP98::NSD1, neither VEN nor IC demonstrates satisfactory efficacy. Currently, allo-HSCT remains the primary approach to improve prognosis for NUP98-r AML. Further exploration of targeted therapies is still needed.

Author contributions

QRW, SNC and HPD were the principal investigators and designed this study. WSXH performed most of materials collection and statistical analysis. QRW performed the experiments, materials collection and data interpretation. MW, JDX and QW performed RNA-Seq and NGS and their analysis. WSXH, QRW, SNC and HPD wrote and revised the manuscript.

Competing interests

The authors declare no competing interests.

Supplementary information

The online version contains supplementary material available at 10.1038/s41408-025-01325-6.