Prognostic Factors of Pediatric Acute Myeloid Leukemia Patients with t(8;21) (q22;q22): A Single-Center Retrospective Study

This retrospective study aimed to analyze the treatment effect and prognostic factors of pediatric acute myeloid leukemia (AML) patients with t(8;21). A total of 268 newly diagnosed pediatric AML (pAML) enrolled from 1 January 2005 to 31 December 2022 were retrospectively reviewed, and 50 (18.7%) patients harbored t(8;21) translocation. CR rate, OS, EFS, and RFS were assessed by multivariate Logistic and Cox regression models in these patients. Of the 50 patients, 2 patients abandoned treatment during the first induction course. Of the remaining 48 patients who received double-induction therapy and were included in the final analyses, CR1 and CR2 were 75.0% (36/48) and 95.8% (46/48), respectively. The overall three-year OS, EFS, and RFS were 68.4% (95% CI, 55.0–85.1), 64.2% (95% CI, 50.7–81.4), and 65.5% (95% CI, 51.9–82.8), respectively. The presence of loss of sex chromosome (LOS) at diagnosis (n = 21) was associated with a better 3-year OS [87.5% (95% CI, 72.7–100) vs. 52.7% (95% CI, 35.1–79.3), p = 0.0089], 3-year EFS [81.6% (95% CI, 64.7–100) vs. 49.7% (95% CI, 32.4–76.4), p = 0.023], and 3-year RFS [81.6% (95% CI, 64.7–100) vs. 51.7% (95% CI, 33.9–78.9), p = 0.036] than those without LOS (n = 27), and it was also an independent good prognostic factor of OS (HR, 0.08 [95% CI, 0.01–0.48], p = 0.005), EFS (HR, 0.22 [95% CI, 0.05–0.85], p = 0.029), and RFS (HR, 0.21 [95% CI, 0.05–0.90], p = 0.035). However, extramedullary leukemia (EML) featured the independent risk factors of inferior OS (HR, 10.99 [95% CI, 2.08–58.12], p = 0.005), EFS (HR, 4.75 [95% CI, 1.10–20.61], p = 0.037), and RFS (HR, 6.55 [95% CI, 1.40–30.63], p = 0.017) in pediatric individuals with t(8;21) AML. Further analysis of combining LOS with EML indicated that the EML+LOS− subgroup had significantly inferior OS (92.9%, [95% CI, 80.3–100]), EFS (86.2%, [95% CI, 70.0–100]), and RFS (86.2%, [95% CI, 80.3–100]) compared to the other three subgroups (all p < 0.001). LOS and EML are independent prognostic factors of OS, EFS, and RFS with t(8;21) pAML patients. LOS combined with EML may help improve risk stratification.


Introduction
Acute myeloid leukemia (AML) constitutes 20% of pediatric leukemias, with 5.1% of AML patients being diagnosed at an age younger than 20 years [1][2][3].AML1::ETO (also known as RUNX1::RUNX1T1) resulting from t(8;21) (q22;q22) is one of the most frequent genetic aberrations, with a incidence of 20% in newly diagnosed pAML patients [4][5][6][7].Although t(8;21) AML is generally categorized as a favorable subgroup, 20%-35% of patients with this genotype experience relapse, and the outcomes of this genotype are heterogeneous [8][9][10].We conducted a single-center retrospective study of children with this genotype from 1 January 2005 to 31 December 2022 to search for prognostic factors that could potentially be useful to further improve outcomes and update therapy strategies.
The morphologic assessment was founded on May-Grünwald-Giemsa stains, myeloperoxidase reactions, and nonspecific esterases utilizing α-naphthyl acetate following the FAB and WHO classifications [13][14][15].A chromosome G-banding analysis was conducted on all cases according to standard procedures [16,17].The description of karyotypes followed the ISCN [18].Loss of sex chromosomes (LOS) was defined as the presence of a -X/-Y clone in more than 3 of the 20 cells analyzed [19].Complex karyotype (CK) was defined as ≥3 chromosomal aberrations [20].Monosomal karyotype (MK) was defined as ≥2 separate autosomal monosomies or 1 monosomy plus ≥ 1 structural aberrations [21,22].Immunophenotyping was conducted in bone marrow (BM) specimens of all cases following previous descriptions [23,24].Quantification of minimal residual disease (MRD) in the BM (after induction therapy I and II, or other later timepoint) using multiparameter flow cytometry (MFCM) commenced in June 2018.

Molecular Biology Analysis
Molecular studies were performed from June 2018, and the results for 20 patients are available.Mononuclear cells were isolated, and both DNA and mRNA were extracted from BM and peripheral blood samples.Additionally, random-primed cDNA synthesis was conducted according to the protocols [25].As for fusion genes, interphase fluorescence in situ hybridization (FISH) with probes for RUNX1 and RUNX1T1 was conducted using commercially available probes.Investigations for FLT3-ITD, C-KIT/D816, NPM1 (Exon12), and CEBPA were carried out in these patients [26].

Statistical Analysis
Each statistical assessment was performed utilizing R software (version 4.1.0).Outcomes were evaluated by using complete remission (CR), overall survival (OS), event-free survival (EFS), and relapse-free survival (RFS) (Table S1) [27].Categorical variables were compared between groups utilizing the χ2 test and continuous variables by the Mann-Whitney U test.To identify factors associated with CR and prognosis, variables with p < 0.2 [28] in univariate analyses entered a multivariate Logistic regression model and a multivariate Cox regression model, respectively.Results were presented as OR or HR and 95% CI.Survival outcomes were determined utilizing the Kaplan-Meier method, and comparisons between groups were conducted utilizing the logrank test.Statistical importance was denoted by two-sided p < 0.05.

Impact of Clinical and Biological Characteristics on CR Achievement
We assessed the association of clinical and biological features with CR achievement in a logistic regression model.As shown in Table 2, age at diagnosis, WBC, PB blast, FAB, CD19, MK, and Year of diagnosis were significantly linked to CR achievement of those patients.We then included these seven variables in the multivariate logistic analysis and revealed that none of these values remained significantly different for CR achievement.

Impact of Clinical and Biological Characteristics on CR Achievement
We assessed the association of clinical and biological features with CR achievement in a logistic regression model.As shown in Table 2, age at diagnosis, WBC, PB blast, FAB, CD19, MK, and Year of diagnosis were significantly linked to CR achievement of those patients.We then included these seven variables in the multivariate logistic analysis and revealed that none of these values remained significantly different for CR achievement.

Risk Stratification
To provide an estimation of prognostic stratification, we classified individuals into four risk groups based on EML and LOS: the EML+LOS+ (n = 4), EML+LOS− (n = 4), EML−LOS+ (n = 17), and EML−LOS− (n = 23) subgroups.As for other secondary ACAs that accompany t(8;21), one out of four individuals in the EML+LOS− subgroup had trisomy 4, 1 out of 23 individuals in the EML−LOS− group and 2 out of 17 patients in the EML−LOS+ subgroup had del(9q), and no patients had trisomy 8. Kaplan-Meier survival analysis unveiled that t(8;21) pAML patients with EML+LOS− had the most inferior outcome, while those with EML−LOS+ had the best prognosis, showing the following statistically significant survivals: the 3-year OS [0 vs. 92.9%(95% CI, 80.3-100), p < 0.001], the 3-year EFS [0 vs. 86.2%(95% CI, 70.0-100), p < 0.001], and the 3-year and EML+LOS− subgroups was too small, we need to exercise caution when explaining the impact of risk stratification on prognosis.The observation of distinct prognostic effects between the two groups based on risk stratification suggested that EML+LOS− should be differentiated from EML−LOS+ in future research.

Discussion
This study demonstrated that, among a large cohort of 268 pAML individuals, 50 (18.7%)harbored t(8;21) translocation.This finding aligns with the range reported in other studies [29,30].We explored the prognostic factors of t(8;21) pAML patients at a single Chinese children's medical center.Previous studies demonstrated that t(8;21) pAML individuals likely benefit from regimens incorporating high doses of Ara-C during induction therapy [29].In this study, the three-year OS, EFS, and RFS were 68.4% (95% CI, 55.0-85.1),64.2% (95% CI, 50.7-81.4),and 65.5% (95% CI, 51.9-82.8),respectively, which were greater than that in the research by Wu et al. [31], comparable with the research by Che et al. [32], and obviously lower than that reported by a Japanese team [8].Their better therapeutic effect may be attributed to a higher cumulative dose of Ara-C (59.4-78.4g/m 2 ) and adequate supportive care.
There was no difference in CR rate among different induction protocols [33,34].In accordance with these studies, our cohort showed that patients treated with CHFU-AML 2005, CCLG-AML 2015, and CALSIII-AML18 regimens also showed no significant difference in CR rate (p > 0.05).Walter et al. [35] demonstrated that CR is a unique clinical significance factor in trials of de novo AML.Fang et al. [34] observed that AML individuals achieving CR1 showed better prognoses than those without CR1.In our study, however, we did not verify their finding that the clinical and biological features of individuals between CR1 and PR/NR groups were compared, but there was no statistical difference in CR rate among all factors (p > 0.05).Our results were slightly different from previous studies, considering that the sample size included in our study was small and needs to be further expanded in future studies.
In line with prior reports, the prevalence of ACAs was high (72.9%),and the most common cytogenetic aberration was LOS [36][37][38][39][40][41].According to the Cox model, EML and LOS were independent prognostic factors for t(8;21) pAML individuals.Previous studies found that loss of Y chromosome (LOY) was linked to a high relapse risk and shorter OS [42,43].However, some studies demonstrated that LOS had no significant associations with survival probabilities [19,36,44,45].Mitterbauer et al. [46] suggested that LOS was linked to a significantly better EFS outcome.In our study, LOS was correlated with significantly better OS, EFS, and RFS, aligning with the research by Chen et al. [47].
The prognostic values of EML remain controversial across studies.Previous studies found that t(8;21) pAML patients with EML were linked to a low CR rate and poor RFS and OS [48][49][50].However, Bisschop et al. [51] did not find significant associations with survival outcomes.The differences in results between studies may relate to the small-scale cohort, differences in race, and the wide range of case cohorts.In our study, EML was

Discussion
This study demonstrated that, among a large cohort of 268 pAML individuals, 50 (18.7%)harbored t(8;21) translocation.This finding aligns with the range reported in other studies [29,30].We explored the prognostic factors of t(8;21) pAML patients at a single Chinese children's medical center.Previous studies demonstrated that t(8;21) pAML individuals likely benefit from regimens incorporating high doses of Ara-C during induction therapy [29].In this study, the three-year OS, EFS, and RFS were 68.4% (95% CI, 55.0-85.1),64.2% (95% CI, 50.7-81.4),and 65.5% (95% CI, 51.9-82.8),respectively, which were greater than that in the research by Wu et al. [31], comparable with the research by Che et al. [32], and obviously lower than that reported by a Japanese team [8].Their better therapeutic effect may be attributed to a higher cumulative dose of Ara-C (59.4-78.4g/m 2 ) and adequate supportive care.
There was no difference in CR rate among different induction protocols [33,34].In accordance with these studies, our cohort showed that patients treated with CHFU-AML 2005, CCLG-AML 2015, and CALSIII-AML18 regimens also showed no significant difference in CR rate (p > 0.05).Walter et al. [35] demonstrated that CR is a unique clinical significance factor in trials of de novo AML.Fang et al. [34] observed that AML individuals achieving CR1 showed better prognoses than those without CR1.In our study, however, we did not verify their finding that the clinical and biological features of individuals between CR1 and PR/NR groups were compared, but there was no statistical difference in CR rate among all factors (p > 0.05).Our results were slightly different from previous studies, considering that the sample size included in our study was small and needs to be further expanded in future studies.
In line with prior reports, the prevalence of ACAs was high (72.9%),and the most common cytogenetic aberration was LOS [36][37][38][39][40][41].According to the Cox model, EML and LOS were independent prognostic factors for t(8;21) pAML individuals.Previous studies found that loss of Y chromosome (LOY) was linked to a high relapse risk and shorter OS [42,43].However, some studies demonstrated that LOS had no significant associations with survival probabilities [19,36,44,45].Mitterbauer et al. [46] suggested that LOS was linked to a significantly better EFS outcome.In our study, LOS was correlated with significantly better OS, EFS, and RFS, aligning with the research by Chen et al. [47].
The prognostic values of EML remain controversial across studies.Previous studies found that t(8;21) pAML patients with EML were linked to a low CR rate and poor RFS and OS [48][49][50].However, Bisschop et al. [51] did not find significant associations with survival outcomes.The differences in results between studies may relate to the small-scale cohort, differences in race, and the wide range of case cohorts.In our study, EML was identified as an independent prognostic factor that is correlated with worse OS, EFS, and RFS, in line with the findings reported by Stove et al. [49] In addition, some reports claimed that patients with EML involving different sites had different survival outcomes [52,53].In our study, subgroup analysis for the prognosis effect of EML at different sites could not be conducted due to the insufficient number of cases.Furthermore, individuals were categorized into four risk groups based on the two independent prognostic factors, LOS and EML, enabling the refinement of patients with different survival outcomes and suggesting that MRD should be closely monitored, as well as that HSCT should be considered in time among those subgroups with worse outcomes.To enhance the robustness of the findings, a larger cohort involving multiple medical centers is required for further verification of the results.
C-KIT mutations are reported to occur in 12−46% of adult t(8;21) AML patients [52][53][54][55], whereas they account for approximately 17−43% in pediatric t(8;21) patients [29,53,[56][57][58][59].The frequency of C-KIT in this study reached 60.0% (12/20), surpassing previous pediatric series [30,60].This discrepancy could be owing to factors such as the small-sized sample from a single center, as well as the variations in the detection method and sequencing depth and coverage.Although C-KIT mutations mediate an adverse prognostic impact on the prognosis of adults with AML [61,62], the impact of C-KIT on pAML is still inconclusive [57,63].Chen et al. [30] observed that C-KIT mutations are genetic markers linked to inferior prognoses in pAML patients.Another study indicated that C-KIT mutations were strongly correlated with poor outcome in t(8;21) pAML individuals [58].In our study, despite the high frequency of C-KIT in t(8;21) pAML, it did not impact long-term prognosis, in line with previous pediatric reports [29,34].This suggests potential differences in prognostic factors between pediatric and adult individuals diagnosed with this disease.
MRD is a pivotal marker in modern studies of pAML, with a commonly accepted threshold set at 0.1% [64].MRD levels exceeding this threshold are linked to an elevated risk of relapse and unfavorable prognosis.In our study, MRD levels were quantified using MFCM following the first and second induction therapy phases in the CALSIII-AML18 protocol.However, the definitive prognostic value of MRD status derived from our findings may be inconclusive, likely due to the limited sample size.
The primary limitation of our study stems from its retrospective design, which may lead to missing data and an unavoidable bias.The power of the subsequent stratified analysis is constrained by the relatively small size of the subgroups and prospective research with a large-scale sample size is warranted to be practicable.

Conclusions
In conclusion, LOS is prognostically favorable, whereas EML is deemed to be strongly correlated with adverse prognoses in pediatric patients with t(8;21) AML.LOS combined with EML may help improve risk stratification and potentially facilitate the customization of future therapies.

Figure 3 .
Figure 3. Outcomes depending on the loss of sex chromosomes (LOS) in t(8;21) AML.Kaplan-Meier analysis of OS (a), EFS (b) and RFS (c) for patients stratified by LOS.

Figure 3 .
Figure 3. Outcomes depending on the loss of sex chromosomes (LOS) in t(8;21) AML.Kaplan-Meier analysis of OS (a), EFS (b) and RFS (c) for patients stratified by LOS.

Figure 5 .
Figure 5. Kaplan-Meier curves for risk stratification.Kaplan-Meier analysis of OS (a), EFS (b) and RFS (c) for patients stratified by risk levels depending on EML and LOS.

Figure 5 .
Figure 5. Kaplan-Meier curves for risk stratification.Kaplan-Meier analysis of OS (a), EFS (b) and RFS (c) for patients stratified by risk levels depending on EML and LOS.
Figure S2.Probability of survival depending on the C-KIT mutation in t(8;21) AML.Kaplan-Meier survival curves of OS (a), EFS (b) and RFS (c) for patients stratified by C-KIT mutation.Figure S3.Probability of survival depending on the Induction I MRD status in t(8;21) AML.Kaplan-Meier survival curves of OS (a), EFS (b) and RFS (c) for patients stratified by Induction I MRD status. Figure S4.Probability of survival depending on the Induction II MRD status in t(8;21) AML.Kaplan-Meier survival curves of OS (a), EFS (b) and RFS (c) for patients stratified by Induction II MRD status.

Table S1 .
Definitions of end points.Table S2.Comparison of characteristics for patients with t(8;21) AML in differrent consecutive protocols.Table S3.Univariate and multivariate COX analyses of OS in 48 patients with t(8;21) AML.Table S4.Univariate and multivariate COX analyses of RFS in 48 patients with t(8;21) AML.Table S5.Univariate and multivariate COX analyses of overall survival (OS) in 48 patients with t(8;21) AML.Table S6.