FUS::ERG/ t(16;21)(p11;q22) and RUNX1::CBFA2T3/ t(16;21)(q24;q22) are two rare non-random chromosome translocations involving chromosomes 16 and 21 and appear in less than 1% of acute myeloid leukemia (AML) cases.¹ Prior research reported that FUS::ERG was characterized as an adverse-risk subtype with a high relapse rate and mortality, while the occurrence of RUNX1::CBFA2T3 was generally believed to be associated with chemoradiotherapy after the diagnosis of primary malignancies.^{2, 3}

Previous studies on t(16;21) rearrangement focused on pediatric cases or were confined to case reports, thus not fully elucidating these subtypes in adults. Consequently, this study aims to detail the clinical characteristics and prognosis of adult AML patients with FUS::ERG and RUNX1::CBFA2T3 fusion genes.

To construct the cohort, we sourced patients from the Mitelman Database of Chromosome Aberrations and Gene Fusions in Cancer (https://mitelmandatabase.isb-cgc.org/) and studies in PubMed and the China National Knowledge Infrastructure from 1985 to 2023. We excluded individuals under 18, those with inconsistent diagnoses, lacking clinical data, or untreated. Our institution's patients with FUS::ERG and RUNX1::CBFA2T3 were included, resulting in a cohort of 111 adult AML patients with FUS::ERG and 20 with RUNX1::CBFA2T3 (Figure S1). Their baseline characteristics were compared with those of the 669-scale reference cohort (Figure 1A), which comprised adult AML patients (excluding t(16;21) cases and acute promyelocytic leukemia) admitted to our institution from 2018 to 2021.

Compared to the reference cohort, AML patients with FUS::ERG exhibited a younger age of onset (34 years vs. 59 years), a decrease in platelet (PLT) count (30 × 10⁹/L vs. 51 × 10⁹/L), and higher bone marrow (BM) blasts (80% vs. 60%). Patients with RUNX1::CBFA2T3 had a similar age of onset and comparable hematological results to the reference cohort. The FUS::ERG group showed higher M1 FAB subtype incidence (24.3% vs. 4.5%) and lower M2 (19.8% vs. 39.8%) and M5 (33.3% vs. 43.5%) incidences compared to the reference cohort. Conversely, the RUNX1::CBFA2T3 group had no dominant FAB subtype, with most cases (65%) being M1/M2 and a lower M5 incidence (10.0%). Additionally, AML-M7 and acute basophilic leukemia were identified in the FUS::ERG cohort. Both FUS::ERG (7.2%) and RUNX1::CBFA2T3 (15.0%) groups showed more cases with indeterminate morphological features than the reference cohort (1.3%). In FUS::ERG patients, 11.1% (7/63) had eosinophilia, and 44.4% (28/63) showed vacuolation or hemophagocytosis in BM biopsies, while only one RUNX1::CBFA2T3 patient had peripheral eosinophilia.

Significantly, both FUS::ERG (trisomy 8: 16.4% vs. 9.0%; complex karyotype: 37.8% vs. 11.6%) and RUNX1::CBFA2T3 patients (trisomy 8: 40.0% vs. 9.0%; complex karyotype: 35.0% vs. 11.6%) showed increased trisomy 8 and complex karyotypes compared to the reference cohort. Trisomy 10, a distinct feature in FUS::ERG-positive patients, was more prevalent than in the reference cohort (9.1% vs. 0.3%), while monosomy prevalence remained consistent across groups.

Additionally, 70% of the patients with RUNX1::CBFA2T3 were diagnosed with secondary AML (s-AML), which was significantly different from the percentage of those with FUS::ERG (2.7%) and the reference cohort (7.9%). Two patients with FUS::ERG evolved from myelodysplastic syndrome, and one had thyroid adenocarcinoma. For s-AML patients with RUNX1::CBFA2T3 (Table S1), 42.9% (6/14) had a history of hematologic malignancies, and 50% (7/14) had prior solid tumors. All patients received chemotherapy once, and some underwent radiotherapy or autogenic hematopoietic stem cell transplantation (HSCT). The median interval from the diagnosis of the primary malignancy to AML was 31 months (range, 4–76 months).

Although patients who received intensive or reduced intensive chemotherapy in the three groups did not exhibit differences, a large number of t(16;21) patients were absent from the detailed treatment protocols (Table S2). Among those with FUS::ERG, 85 (84.2%) achieved complete remission (CR), which was comparable to that of the reference cohort (80.6%). Seventy-four patients (73.3%) achieved CR after induction chemotherapy (33.7% after cycle 1, 15.8% after cycle 2, 3.0% after more than two cycles, and 20.8% without detailed information), and six patients (5.4%) were refractory to the initial therapy but achieved CR after HSCT. Moreover, 75% of patients with RUNX1::CBFA2T3 achieved CR, with 50% reaching CR after the first chemotherapy course. HSCT frequencies were similar in both t(16;21) subgroups (36.2% vs. 35.0%), yet the FUS::ERG group showed a higher frequency than the reference cohort (26.6%), with a significantly higher relapse rate (78.6%) compared to both the reference cohort (44.7%) and the RUNX1::CBFA2T3 group (33.3%). However, patients with RUNX1::CBFA2T3 showed no significant difference from the reference cohort.

For long-term outcomes, patients with FUS::ERG had shorter median overall survival (OS) (13 vs. 35 months, p < .001) and disease-free survival (DFS) (7.5 vs. 20.5 months, p < .001) compared to the reference cohort, with a 3-year OS of 11.4% (95% confidence interval [CI]: 5.9–22.3) and a 3-year DFS of 9.5% (95% CI: 4.3–21.3). Median OS for ELN 2022 F-risk versus I-risk versus FUS::ERG was not reached (NR) versus 33.6 versus 13.0 months (p < .001), with FUS::ERG showing even shorter OS than A-risk (13.0 vs. 16.8 months, p = .015). Additionally, median DFS for ELN 2022 F-risk versus I-risk versus A-risk versus FUS::ERG was NR versus 19.8 versus 13.3 versus 7.5 months (p < .001) (Figure 1B–E). In contrast, patients with RUNX1::CBFA2T3 had a median OS of 18.7 months and a median DFS of 16 months, with a 3-year OS and 3-year DFS of 12.9% (95% CI: 2.4–68.7) and 20.4% (95% CI: 4.3–96.6), respectively. It also presented a relatively unfavorable OS trend and indistinctive DFS when compared with the whole reference cohort and all risk-dependent subgroups (Figure 1B–E). Moreover, the RUNX1::CBFA2T3 group had a better median DFS than did the FUS::ERG group (16 vs. 7.5 months, p = .016), but the median OS was comparable (18.7 vs. 13 months, p = .219) (Figure 1B–E).

For patients with FUS::ERG, univariable and multivariable analysis revealed that age >60 years (hazard ratio [HR] = 3.39, 95% CI: 1.37–8.38, p = .008), white blood cell count >100 × 10⁹/L (HR = 2.69, 95% CI: 1.21–6.01, p = .016) and monosomy (HR = 2.63, 95% CI: 1.24–5.57, p = .011) were independent risk factors for OS. Moreover, monosomy was related to poor DFS (HR = 3.72, 95% CI: 1.69–8.20, p = .001), whereas receiving HSCT could significantly improve both OS (HR = 0.21, 95% CI: 0.11–0.40, p < .001) and DFS (HR = 0.31, 95% CI: 0.16–0.57, p < .001) (Figure 1F). To better understand the protective effect of transplant timing, we analyzed chemotherapy versus HSCT in FUS::ERG patients (Table S3). The transplant group, compared to the chemotherapy-only group, showed no baseline characteristic differences but exhibited improved 1-year DFS rates (42.7% vs. 14.3%, p < .001) and 3-year OS rates (27.5% vs. 2.4%, p < .001), with median DFS and OS extended by 4.4 and 9 months, respectively (Figure 1G,H). However, 72.7% of patients undergoing HSCT experienced recurrence, with 63.2% dying, comparable to the chemotherapy group (relapse: 80.4%; death: 76.2%). Additionally, the limited number of patients with RUNX1::CBFA2T3 precluded prognostic analysis in this subgroup. We explored the impact of a high s-AML proportion, finding no significant OS or DFS difference (Figure S2).

To expand the sample size and overcome the insufficient cases brought about by the lower incidence rate in the adult group, we adopted a different methodology from pediatric AML, utilizing several massive medical databases and integrating the case data from our institution to establish the study cohort.¹ Consistent with previous research, we observed that FUS::ERG was independently classified into an adverse-risk subtype with young onset age, high frequency of M1 type, and high recurrence rate.³ However, a reduction in PLT count, high BM blasts, and vacuolization occurred in approximately 50% of patients; nonetheless, no correlation between these features and poor prognosis was found. Trisomy 10 was found to be a unique karyotype feature of FUS::ERG patients, and the significantly higher incidence of trisomy 8 and complex karyotype increased the cytogenetic risk and exerted an unfavorable impact on its OS.⁴ Moreover, RUNX1 and PTPN11 were reported as the most frequent mutations in FUS::ERG patients, which commonly function as signal transduction factors in hematopoiesis and are associated with reduced CR rates and inferior RFS and OS.^4-6 Among three patients from our institution, we found one carried RUNX1 and PTPN11 mutations, and he did not respond to intensive chemotherapy in the first cycle.

RUNX1::CBFA2T3 can also be considered an adverse-risk subgroup in adults. Nearly 70% of adult patients were classified into therapy-related AML (t-AML), and most of them had long-term exposure to topoisomerase inhibitors and alkylating agents.² Moreover, we firstly found that EP300, TTN, TET2, IDH1, DNMT3A and CSF1R gene mutations and fusion genes KMT2A::MLLT4 co-occurrence with RUNX1::CBFA2T3. In comparison with pediatric patients, we found FUS::ERG performed consistent clinical characteristics among children, whereas RUNX1::CBFA2T3 had a reverse long-term survival without any recurrence in the pediatric cohort.¹ We thought the difference in RUNX1::CBFA2T3 may be related to a higher prevalence of t-AML in adult patients (70%) than in pediatric patients (22%) and insufficient treatment in adult patients, as well as a chance effect due to the small size of patients.

For patients with FUS::ERG, HSCT's potent protective effect was unreported in pediatric cases, yet it significantly benefited adult FUS::ERG patients in our study, despite not altering the high recurrence and mortality rates post-HSCT. While HSCT's protective impact on RUNX1::CBFA2T3 patients was unproven, and its effect on FUS::ERG patients appeared limited, opting for HSCT proved advantageous when relapse probability exceeded 40% or fell into the adverse-risk category.^{4, 6} The challenges have turned into post-HSCT management on how to prevent recurrence.

Nevertheless, it is important to acknowledge this study's limitations. The retrospective design meant reliance on medical database data, introducing accuracy and completeness constraints. Despite efforts to increase the sample size, patients with RUNX1::CBFA2T3 were still limited, potentially biasing our analysis. Thus, future multicenter prospective studies are needed to better identify clinical features and prognostic factors for these adult AML subtypes. In conclusion, our study first summarizes and compares the clinical characteristics of FUS::ERG and RUNX1::CBFA2T3, defining them as adverse risk factors in adults. HSCT was recommended for both, especially the FUS::ERG group, though solutions for post-transplant relapse are crucial.