Data were collected from five institutions: The First Affiliated Hospital of Soochow University, Nanfang Hospital of Southern Medical University, General Hospital of PLA, Nanjing Drum Tower Hospital and Ruijin Hospital Affiliated to Shanghai Jiao Tong University School of Medicine. During January 2016 and February 2021, 103 de novo ETP-ALL/LBL patients 15 years of age or older were included in this retrospective study ( Figure 1 ). The diagnosis of ETP-ALL/LBL was based on the 2016 WHO criterion (10). A cutoff of < 20% BM blasts was used to define LBL. The median follow-up time was 18 months (0.5–60 months). The last follow-up time was April 2021. This study was approved by the institutional review board at each institution.

For patients with ETP, flow cytometry immunophenotyping (FCI) defined the absence of CD1a/CD8, weak expression of CD5 (<75% positive lymphoblasts), and the presence of one or more myeloid or stem cell markers (CD117, CD34, HLA-DR, CD13, CD33, CD11b, or CD65) on at least 25% of lymphoblasts. Minor residual disease (MRD) analysis was performed using flow cytometry detection. We defined 0.01% (1*10^-4) as the threshold to distinguish MRD-positive from MRD-negative patients: MRD≥0.01% (1*10^-4) of nucleated cells (NC) was defined as positive, and MRD<0.01% (1*10^-4) of nucleated cells (NC) was defined as negative (11).

Genomic DNA was extracted from BM-derived mononuclear cells at diagnosis. The Ion S5 system (Personal Genome Machine, Thermo Fisher, Grand Island, NY, USA) was used to evaluate the panel of 51 common variant gene targets in hematologic malignancies, including ASXL1, ASXL2, BCOR, BCORL1, BIRC3, BRAF, CALR, CBL, CEBPA, C-KIT, CSF3R,CSMD1, DNMT3A, ETNK1, ETV6, EZH2, FBXW7, FLT3,GATA2, IDH1, IDH2, IL7R, JAK1, JAK2, JAK3, KRAS, MPL,MYD88, NOTCH1, NPM1, NRAS, PAX5, PDGFRA, PDGFRB,PHF6, PI6, PIGA, PTEN, PTPN11, RUNX1, SETBP1, SETD2,SF3B1, SH2B3, SRSF2, STAG2, TET2, TP53, U2AF1, WT1 and ZRSR2.The coverage rate of the NGS expander was 98.03%, the depth of the average sequence was 2 500×, and the sequencing depth of more than 95% target regions was 2 000×.

The choice of the treatment regimen was determined by the individual physician based upon CALLG2008 protocol (a protocol developed by the Chinese Acute Lymphoblastic Leukemia Cooperative Group for ALL) and augmented MDACC Hyper-CVAD. Ninety-seven patients accepted induction therapy, including a VDCP-based regime (vindesine 4 mg/d, intravenous infusion, D1, D8, D15, D22, daunorubicin 40 mg/(m2/d), intravenous infusion, D1, D8, D15, D22, cyclophosphamide 1 mg/(m2/d), intravenous infusion, D1, D15, prednisone 1 mg/(m2/d), oral, D1–14, D15–28 (2/3 dose) with or without L-asparaginase 6000IU/m2/d, intramuscular, D11, D14, D17, D20, D23, D26), augmented MDACC hyper-CVAD A protocol (cyclophosphamide, 300 mg/(m2/q12 h), intravenous infusion, D1–3, vindesine, 3mg/m2/d, intravenous infusion, D4, D11, doxorubicin 50 mg/m2, intravenous infusion, D4, dexamethasone 40 mg/d, intravenous infusion, D1–4, D11–14), VDCP-based regime combined with chidamide (chidamide 30 mg, oral, twice per week, vindesine 4 mg/d, intravenous infusion, D1, D8, D15, D22, daunorubicin 40 mg/(m2/d), intravenous infusion, D1, D8, D15, D22, cyclophosphamide 1 mg/(m2/d), intravenous infusion, D1, D15, Prednisone 1 mg/(m2/d), oral, D1–14, D15–28 (2/3 dose) with or without L-asparaginase 6000IU/m2/d, intramuscular, D11, D14, D17, D20, D23, D26) and a DAC combined with AAG priming regimen [decitabine 20mg/m2/d, intravenous infusion, d1-d3, aclarubicin 10mg/d, intravenous infusion, d1-d8, cytarabine10mg/m2/q12h, subcutaneous injection, d1-d14, granulocyte colony-stimulating factor (G-CSF), 50-300μg/d, subcutaneous injection, (when WBC counts are less than 20×10^9/L)]. Patients who were unfit for intensive chemotherapy received dose-reduced chemotherapy regimens. CNS prophylaxis proceeded as needed.

Patients who experienced primary refractory disease or relapse received a reinduction regimen (VDCP-based regimen, augmented MDACC protocol, or a DAC combined with AAG priming regimen (decitabine 20mg/m2/d, intravenous infusion, d1-d3, aclarubicin 10mg/d, intravenous infusion, d1-d8, cytarabine10mg/m2/q12h, subcutaneous injection, d1-d14, granulocyte colony-stimulating factor (G-CSF), 50-300μg/d, subcutaneous injection, (when WBC counts are less than 20×10^9/L)).

After achieving CR, consolidation therapy was carried out, including based on high-dose methotrexate, high-dose cytarabine, L-asparaginase or pegaspargase, and cyclophosphamide. The choice of different consolidation treatment schemes was based on the choice of each center. Allo-SCT was performed in patients with disease status and potential donors.

Among the patients who chose allo-SCT, all received a myeloablative conditioning regimen. Hematopoietic stem cell transplantation included HLA-matched donors and HLA-mismatched donors. A BU/CY-based conditioning regime was as follows: Ara-c 2 g/m2/12h on days -9 to -8, Bu 0.8 mg/kg/6h from day -7 to day -5, cyclophosphamide 1.8 g/m2/d from day -4 to -3, and semustine 250 mg/m2/d on day -10. The TBI/Cy regime consisted of TBI (12 Gy on days -8 to -6), cyclophosphamide 1.8g/m2/d from day -4 to approximately day -3, and semustine 250mg/m2/d on day -9. GVHD prophylaxis consisted of continuous cyclosporine infusion at 3 mg/kg/day starting on day −10, mycophenolate mofetil 1.0 g/day from days −10 to +30, short-term methotrexate administered on days +1, +3, +6, and +11 at dosages of 15, 10, 10, and 10 mg/m2, respectively, and rabbit antithymocyte globulin (ATG) 2.5 mg/kg/d (on days −5 to −2) was administered in the human leukocyte antigen (HLA)-haploidentical related donor (haplo-RD) and unrelated donor groups. The detailed procedure of allo-SCT was reported previously (12, 13).

The measured outcomes were complete remission (CR), partial response (PR), no response (NR), overall survival (OS) and relapse-free survival (RFS). CR was defined as <5% bone marrow blasts, no blasts in blood, absolute neutrophil counts (ANCs) >10⁹/L, platelets >10⁹/L and no extramedullary leukemia. PR was defined as bone marrow blast counts >5% but ≤20% and/or persistent evidence of extramedullary disease and a reduction of total lymphoma volume of >50%, no evidence of progression and no new lesions for >3 weeks. NR was defined as bone marrow blast counts >20%, blasts in peripheral blood, failure of hematological recovery, or persistent evidence of extramedullary disease not meeting PR criteria. OS was measured as the time from the date of diagnosis to the date of death or last follow-up. RFS was defined as the time from first CR to relapse, censoring at death in CR or last follow-up. When only allo-HSCT patients were analyzed, the outcome was calculated from the date of transplantation. Refractory disease was defined as refractory to primary therapy or to salvage with intensive combination chemotherapy, first relapse with first remission duration < 12 months, second or greater relapse, or relapse at any time after allo-SCT (14). Relapsed disease was defined as reappearance of blasts in the blood or bone marrow (>5%) or in any extramedullary site after CR. Acute GVHD was graded on the Glucksberg scale from grade I-IV.

Continuous variables were analyzed by the Mann–Whitney U-test, and categorical parameters were analyzed by Pearson chi-squared test or Fisher’s exact test. OS and RFS were estimated by the Kaplan–Meier method and log-rank test. A landmark analysis for survival was performed to mitigate the effect of early death prior to allo-SCT (15). The landmark time was 6 months (median time from first chemotherapy to allo-SCT). The cumulative incidence of GVHD was estimated with competing risk methods. Relapse was considered a competing risk for GVHD. A P value < 0.05 was considered statistically significant. All statistical analyses were performed using SPSS Version 26 and R Version 4.4.1.

The characteristics of all 103 patients are shown in Table 1 . There were 73 males and 30 females, and the median age was 29 years (range, 15–70 years). The numbers of patients with ETP-ALL and ETP-LBL were 98 and 5, respectively. The median follow-up time was 18 months (range 0.5–60 months). The median WBC count was 9.09 (0.22–643.0) ×10⁹/L, the median PLT count was 109 (10–413) ×10⁹/L, and the median HGB level was 103.5 (41–172) g/L. The WBC count of 14 patients was ≥100×109/L at initial diagnosis. The LDH of 41 patients was ≥245 U/L. Among the 103 patients, 45 cases (43.7%) had splenomegaly, 49 cases (47.6%) had peripheral lymph node enlargement, and 24 cases (23.3%) had mediastinal masses. The median percentage of bone marrow primitive cells was 80% (4.5% ~ 99%). The karyotypes of 17 patients were complex. 79 patients with treatment were younger than 39 years, they tended to have more splenomegaly and mediastinal masses in clinical manifestations than older adult ( Supplementary Table 1 ).

After a median follow-up of 18 months (range, 0.5–60 months), the 2-year OS and RFS rates were 54% and 57.7%, respectively ( Figure 2A ). Patients who presented with ETP-ALL had a median survival time of 38 months compared with 28 months for ETP-LBL. However, this was not statistically significant (P=0.357). The 2-year OS of patients was 77% of allo-SCT at CR1 and 25% of chemotherapy alone (p < 0.001, HR: 0.2833, 95%CI 0.1126 to 0.7130) ( Figure 2B ).

Next-generation sequencing was performed on 94 patients to detect 51 gene mutations, and the results are shown in Figure 3 , with a median of 2 (0–6) gene mutations. NOTCH1 (35.1%, 33 cases) was the most frequently mutated gene, followed by JAK3 (16.0%, 15 cases), PHF6 (13.80%, 13 cases) and EZH2 (11.70%, 11 cases). Thirty patients did not have gene mutations. Mutated genes were grouped into several signaling pathway. The mutations involved in regulating cytokine and RAS signaling (51.06%), hematopoietic development (30.90%), methylation (26.6%), and histone modification (20%) were found frequently. There was no statistically significant difference in OS between NOTCH1 mutation and NOTCH1 wild-type patients (P=0.299, HR:1.386, 95%CI 0.7151 to 2.686). The most common fusion gene was SET-CAN (7/94). Six out of 7 patients with SET-CAN in our cohort achieved CR after the first induction treatment, and 5 of them successfully experienced allo-SCT in CR1. During the follow-up, 4 patients were still in relapse-free survival. Two patients died of relapse and aGVHD. One patient was lost to follow up.

Ninety-seven patients had data regarding the first course of induction therapy. In this group of patients, 61.9% (60/97) received VDCP-based regimens, 17.5% (17/97) received hyper-CVAD A, 5.2% (5/97) received VDCP-based combined with chidamide regimens, 6.2% (6/97) received DAC combined with AAG priming regimen, and 9.3% (9/97) received other induction regimens. Complete remission (CR) was obtained in 74.2% (72/97) of patients after induction therapy, while CR after the first induction therapy was only 44.3% (43/97). A total of 3.1% (3/97) of patients died during the first induction therapy. Besides, in this study, we compared the differences in response among four different induction chemotherapy regimens (VDCP-based, Hyper-CVAD-A, VDCP-based combined with chidamide and DAC combined with AAG priming regimen) ( Supplementary Table 2 ). Sixty patients received VDCP-based chemotherapy, including 30 patients with CR, 30 patients with NR. The CR rate was 50% (30/60). Seventeen patients received the hyper-CVAD A regimen, 4 patients received CR, 13 patients received NR, with a CR rate of 24% (4/17). Five patients received VDCP-based combined with chidamide regimen chemotherapy, 3 patients achieved CR, and 2 patients achieved NR, with a CR rate of 60% (3/5). Six patients received DAC combined with AAG priming regimen, 3 patients received CR, 3 patients received NR, with a CR rate of 50% (3/6). Among these patients, a pairwise comparison was made between the four groups. After adjustment by Bonferroni, there was no significant difference between the other groups (P > 0.125).

A total of 12 patients received a decitabine combined with AAG priming regimen in this study, including 6 who were treatment-naïve and 6 who were relapsed/refractory. Among the 6 relapsed/refractory patients, 5 patients were primary refractory, and 1 patient relapsed after transplantation. After one course of reinduction, 3 (3/6) patients achieved CR, and 3 (3/6) patients remained NR. At subsequent follow-up, all 3 patients who received CR experienced recurrence, 1 relapsed during chemotherapy and 2 relapsed after allo-SCT. Moreover, 3 died as a result of disease progression, and 2 died from severe infection during the follow-up. The median survival of the 6 patients was 13.5 months (6–36 months), and only 1 patient remained in remission within the follow-up period.

A total of 58 patients received allo-HSCT, including 46 CR1 patients, 2 CR2 patients and 10 NR patients. The median interval from CR to transplantation was 3.5 (range, 1-12) months. A total of 13.0% (6/46) of the patients experienced relapse after allo-HSCT at CR1, while 50% (1/2) of the patients experienced relapse after allo-HSCT at CR2. Patients who experienced transplantation in CR had better OS than transplantation in NR (P=0.029, HR: 0.3625, 95%CI 0.1005 to 1.308), while the RFS was no statistical significance. (P=0.078, HR:0.4431, 95%CI 0.1340 to 1.465) ( Figure 4A ). There were 33 MRD negative and 15 MRD positive patients in the 48 patients who transplanted in CR, and patients who were MRD negative at transplantation had better OS and RFS than those who were MRD positive (OS: P= 0.032, HR: 0.3306, 95%CI 0.09829 to 1.112; RFS: P=0.0035, HR: 0.2492, 95%CI 0.07993 to 0.7770) ( Figure 4B ). Additionally, 17 patients had HLA-matched donors, and 41 patients had HLA-mismatched donors. There was no statistical significance in the OS and RFS of HLA-matched and HLA-mismatched patients (OS: P=0.629, HR: 0.7996, 95%CI 0.3144 to 2.034, RFS: P=0.972, HR: 0.9844, 95%CI 0.3982 to 2.433) ( Figure 4C ). Forty patients received a BU/CY-based conditioning regimen, 18 patients received a TBI/CY-based conditioning regimen, and these two cohorts had no difference in OS and RFS (OS: P=0.570, HR:1.340, 95%CI 0.5125 to 3.503, P=0.770, HR:1.150, 95%CI 0.4577 to 2.887). Ten patients received salvage hematopoietic stem cell transplantation, and all received CR after transplantation. However, 3 patients died of recurrence, 1 patient died of GVHD, 2 patients died of severe infection, and only 2 patients were still in relapse-free survival. The median times for neutrophil engraftment and platelet recovery were 11 days (range, 7–28 days) and 14 days (range, 7–25 days), respectively. Two patients died of serious infection before reaching platelet recovery. The cumulative incidence of II-IV aGVHD was 12% (7/58). Notably, we observed that patients with MRD negative after 1 course of consolidation therapy had better OS and RFS (OS: P=0.014, HR: 0.1180, 95%CI 0.03414 to 0.4078; RFS: P=0.022, HR: 0.2058, 95%CI 0.06637 to 0.6382) ( Supplementary Table 3 ). The main transplantation characteristics of the patients are listed in Supplementary Table 4 . The comparison clinical data of allo-SCT and chemo alone are listed in Supplementary Table 5 .

Prognostic factors affecting OS in 97 patients with ETP-ALL/LBL were analyzed by univariate Kaplan–Meier analysis and a Cox multivariate model analysis, the results are shown in Table 2 . Age, gender, white blood counts, hemoglobin counts, platelet counts, LDH concentration, BM blasts, cytogenetics, the number of mutations, NOTCH1 mutation, presence of mediastinal involvement at diagnosis, disease status after the first course induction and allo-SCT were included in univariate and multivariable analyses. Age >39 years old, CR after the first course of induction therapy and allo-SCT were prognostic factors affecting OS in patients with ETP-ALL/LBL. A multivariate analysis showed that whether CR was received after the first course of induction (P < 0.001, HR: 0.24, 95% CI 0.12 to 0.46) and whether allo-SCT was performed (P < 0.001, HR: 0.25, 95% CI 0.13 to 0.48) were independent prognostic factors for ETP-ALL/LBL.

Univariate and multivariate analyses were performed for the survival of ETP in 58 patients who received allo-SCT, including age, sex, MRD status at transplantation, transplantation type and conditioning regimen ( Table 3 ). Patients with MRD negative at transplantation and transplantation at CR had better OS, and those with negative MRD at transplantation had better RFS. A multivariate analysis showed that MRD negative at transplantation was independent prognostic factors of OS and RFS for ETP-ALL/LBL with allo-SCT (OS: P= 0.042, HR: 3.30, 95% CI 1.05 to 10.40; RFS: P=0.006, HR: 4.57, 95%1.53 to 13.65).