Prognostic factors for allogeneic haematopoietic cell transplantation outcomes in primary refractory acute myeloid leukaemia (2013–2022): A retrospective study by the adult acute myeloid leukaemia working group of the Japanese Society for Transplantation and Cellular Therapy

Summary

Primary refractory acute myeloid leukaemia (AML) remains a major clinical challenge, with poor outcomes despite salvage chemotherapy. Allogeneic haematopoietic cell transplantation (HCT) continues to be a potentially curative option for these patients. However, recent data on outcomes and prognostic factors specific to primary refractory AML remain limited. We conducted a retrospective analysis of 2600 adult patients with primary refractory AML who underwent their first allogeneic HCT between 2013 and 2022, using data from the Japanese national registry. The 3‐year overall survival (OS) and leukaemia‐free survival (LFS) rates were 28.5% and 24.4% respectively. The multivariate analysis identified older age (≥50 years), poor performance status (≥2), adverse cytogenetics, extramedullary disease at diagnosis and higher peripheral blood blast count at HCT (≥10%) as significant risk factors for worse OS and LFS. The cumulative incidences of relapse and non‐relapse mortality at 3 years were 49.8% and 25.8% respectively. Based on the five significant risk factors, we developed a scoring system that effectively stratified patients into distinct prognostic groups for OS and LFS. This nationwide analysis demonstrated that allogeneic HCT offers the potential for long‐term survival in adult patients with primary refractory AML. The proposed risk‐scoring system may support clinical decision‐making and patient counselling.

Three‐year overall survival (OS) and leukaemia‐free survival (LFS) rates were 28.5% and 24.4% in allogeneic haematopoietic cell transplantation (HCT) for primary refractory acute myeloid leukaemia from 2013 to 2022 in Japan. Age ≥ 50 years, performance status ≥2, adverse cytogenetics, extramedullary disease at diagnosis and peripheral blood blasts at HCT ≥10% were significant factors for stratifying OS and LFS.

INTRODUCTION

The initial goal of treating acute myeloid leukaemia (AML) is to achieve complete remission (CR), which is achieved in approximately 70%–80% of patients who have received intensive chemotherapy. ¹ , ² , ³ However, the remaining patients fail to achieve CR and require salvage chemotherapy, and ultimately, long‐term survival is difficult to achieve with chemotherapy alone. Allogeneic haematopoietic cell transplantation (HCT) is the only curative option for patients with primary refractory AML who fail to achieve CR after induction or salvage chemotherapy. ⁴ , ⁵ , ⁶ , ⁷ Although patient and disease characteristics differ between those receiving allogeneic HCT for primary refractory versus relapse refractory AML, most studies group primary and relapse refractory patients together under the term relapsed and refractory (R/R) AML. ⁸ , ⁹ , ¹⁰ , ¹¹ , ¹² , ¹³ , ¹⁴ , ¹⁵ , ¹⁶ , ¹⁷ , ¹⁸ , ¹⁹ , ²⁰ Data on outcomes and prognostic factors following allogeneic HCT for primary refractory adult AML are limited. ²¹ , ²² , ²³ Recent advances in allogeneic HCT for R/R AML, including extended graft availability, improved conditioning regimens and additional supportive care, may improve posttransplant outcomes. ²⁴ , ²⁵ , ²⁶ To better understand the outcomes and prognostic factors for adult primary refractory AML treated with allogeneic HCT, we performed a retrospective analysis using a nationwide Japanese database.

MATERIALS AND METHODS

Data collection

This retrospective analysis was conducted by the Adult AML Working Group of the Japanese Society for Transplantation and Cellular Therapy (JSTCT). Data were extracted from the Transplant Registry Unified Management Program dataset of the Japanese Data Center for Hematopoietic Cell Transplantation (JDCHCT) and JSTCT. ²⁷ , ²⁸ Eligibility criteria for this study included patients aged 16 years or older with de novo or secondary AML who underwent a first allogeneic HCT during primary induction failure between 2013 and 2022 in Japan. Ultimately, 2600 patients meeting these criteria were enrolled. This study was approved by the Adult AML Working Group of the JSTCT and by the institutional review board of the Institute of Medical Science, The University of Tokyo (2024‐43‐0912), where the study was conducted.

Study objectives

The primary objective was to identify prognostic factors affecting overall survival (OS) following allogeneic HCT in adult primary refractory AML. The secondary objectives were to investigate leukaemia‐free survival (LFS), relapse and non‐relapse mortality (NRM) following allogeneic HCT in adult primary refractory AML.

Definitions

Primary induction failure was defined as failure to achieve CR after at least one course of induction chemotherapy. CR was determined by the presence of fewer than 5% bone marrow (BM) blasts. OS (the inverse of overall mortality) was defined as the time from HCT to death from any cause, and LFS was defined as the time from HCT to relapse or death from any cause. Surviving patients were censored at the time of the last observation. Relapse was defined as morphologic evidence of AML, and NRM was defined as death during remission. The HCT‐specific comorbidity index (HCT‐CI), ²⁹ Eastern Cooperative Oncology Group performance status (PS), ³⁰ cytogenetic risk, ³¹ and intensity of conditioning regimen ³² were classified according to published criteria. The number of HLA disparities between recipient and donor was based on low‐resolution HLA‐A, ‐B and ‐DR matching.

Statistical analysis

The Kaplan–Meier method was used to estimate the unadjusted probability of OS and LFS, which were compared using the log‐rank test. Cumulative incidence estimates were used to calculate the unadjusted cumulative incidence of relapse and NRM, which were compared using Gray's test. The Cox proportional hazards regression model for overall mortality (1‐OS) and treatment failure (1‐LFS) and the Fine and Gray proportional hazards model for relapse and NRM were used to estimate hazard ratios (HR) with 95% confidence intervals (95%CI) in the multivariate analysis. Multivariate analyses included the following variables: recipient age at HCT (16–39 vs. 40–49 vs. 50–59 vs. ≥60 years), PS (0–1 vs. 2–4), HCT‐CI (0–2 vs. ≥3), recipient cytomegalovirus serostatus (negative vs. positive vs. not evaluable), conditioning regimen (myeloablative conditioning [MAC] vs. reduced‐intensity conditioning [RIC]), graft‐versus‐host disease (GVHD) prophylaxis (calcineurin inhibitors and methotrexate [CI + MTX] vs. others), use of antithymocyte globulin (ATG) (without ATG vs. with ATG), donor type (related vs. haploidentical vs. unrelated vs. cord blood), HLA disparities (match vs. mismatch), sex incompatibility (female donor to male recipient vs. others), ABO incompatibility (major mismatch vs. others), year of HCT (2013–2017 vs. 2018–2022), prior history of myelodysplastic syndrome (MDS)/myeloproliferative neoplasm (MPN) (absence vs. presence), cytogenetic risk (adverse vs. other than adverse), white blood cell (WBC) count at diagnosis (<3 vs. 3–20 vs. ≥20 × 10⁹/L), extramedullary disease at diagnosis (absence vs. presence), BM blasts at HCT (< 20 vs. 20–40 vs. ≥40%), peripheral blood blasts at HCT (absence vs. >0–10 vs. ≥10%) and duration from diagnosis to HCT (<3 vs. ≥3 months).

The adjusted HRs from the multivariate analysis were used to develop a scoring system. In accordance with a previously published method, HRs rounded to one decimal place were categorized and converted into integer weights of 0, 1, 2 and 3, corresponding to HR ranges of ≤1.2, 1.3–2.0, 2.1–3.0 and ≥3.1 respectively. ¹⁶ Patients were classified into different risk groups based on OS and LFS, using a scoring method that was established as the sum of these integer weights.

Statistical analyses were performed using EZR version 1.68 (Saitama Medical Center, Jichi Medical University), ³³ a graphical user interface for R 4.4.0 software (R Foundation for Statistical Computing). Two‐sided P‐values are reported, with p < 0.05 considered significant.

RESULTS

Characteristics of patients and allogeneic HCT

Table 1 summarizes the characteristics of patients and allogeneic HCTs. The median recipient age at HCT was 57 years (interquartile range [IQR], 46–63). Approximately two‐thirds of recipients were male (63.9%). A total of 482 patients (18.6%) had a PS of 2–4, and 586 patients (22.9%) had an HCT‐CI ≥3. MAC regimens were administered to 65.7% of patients, while 58.4% received CI + MTX‐based GVHD prophylaxis. Regarding donor type, 382 patients received HLA‐matched related HCT, 102 received HLA‐mismatched related HCT, 350 received HLA‐haploidentical related HCT, 316 received HLA‐matched unrelated HCT, 195 received HLA‐mismatched unrelated HCT and 1189 received unrelated cord blood transplantation (CBT). Notably, 66.7% of the transplants were HLA‐mismatched (Table 1).

TABLE 1.

Characteristics of patients and allogeneic transplantations.

Variables	Value
Recipient age at HCT, median (IQR) years	57 (46–63)
Recipient age category
16–39 years	390 (15.0%)
40–49 years	452 (17.4%)
50–59 years	706 (27.2%)
≥60 years	1052 (40.5%)
Recipient gender
Female	939 (36.1%)
Male	1660 (63.9%)
ECOG PS at HCT
0–1	2115 (81.4%)
2–4	482 (18.6%)
HCT‐CI
0–2	1977 (77.1%)
≥3	586 (22.9%)
Recipient CMV serostatus
Negative	327 (12.6%)
Positive	2160 (83.1%)
Not evaluable	113 (4.3%)
Conditioning regimen
MAC	1708 (65.7%)
RIC	892 (34.3%)
GVHD prophylaxis
CI + MTX	1519 (58.4%)
CI + MMF	767 (29.5%)
Others	314 (12.1%)
ATG
No use of ATG	2237 (86.0%)
Use of ATG	363 (14.0%)
Donor
Matched related	382 (15.1%)
Mismatched related	102 (4.0%)
Haploidentical related	350 (13.8%)
Matched unrelated	316 (12.5%)
Mismatched unrelated	195 (7.7%)
Unrelated cord blood	1189 (46.9%)
HLA disparity
Match	865 (33.3%)
Mismatch	1733 (66.7%)
Sex incompatibility
Female donor to male recipient	660 (25.9%)
Others	1892 (74.1%)
ABO incompatibility
Match/minor mismatch	1736 (66.9%)
Major mismatch	860 (33.1%)
Year of HCT
2013–2017	1322 (50.8%)
2018–2022	1278 (49.2%)
Disease type
De novo	2219 (85.4%)
Secondary	380 (14.6%)
Prior history of MDS/MPN
Absence	1898 (73.0%)
Presence	702 (27.0%)
Cytogenetic risk
Favourable	58 (2.2%)
Intermediate	1307 (50.3%)
Adverse	1127 (43.3%)
Unevaluable	108 (4.2%)
WBC count at diagnosis, median (IQR) × 109/L	4.97 (1.97–25.81)
WBC count category
<3 × 109/L	957 (37.6%)
3–20 × 109/L	861 (33.8%)
≥20 × 109/L	727 (28.6%)
MPO‐positive blasts
<50%	1102 (59.2%)
≥50%	758 (40.8%)
Extramedullary manifestation at diagnosis
Absence	2375 (91.3%)
Presence	225 (8.7%)
Bone marrow blasts at HCT, median (IQR) %	17.2 (6.8–40.7)
Bone marrow blasts category
<20%	1178 (53.2%)
20%–40%	467 (21.1%)
≥40%	570 (25.7%)
Peripheral blood blasts at HCT, median (IQR) %	2.0 (0–15.0)
Peripheral blood blasts category
Absence	885 (34.1%)
>0%–10%	869 (33.5%)
≥10%	840 (32.4%)
Duration from diagnosis to HCT category
<3	708 (27.2%)
≥3 months	1892 (72.8%)

Abbreviations: ATG, antithymocyte globulin; CI, calcineurin inhibitor; CMV, cytomegalovirus; ECOG PS, eastern cooperative oncology group performance status; GVHD, graft‐versus‐host disease; HCT, haematopoietic cell transplantation; HCT‐CI, haematopoietic cell transplantation‐specific comorbidity index; HLA, human leukocyte antigen; IQR, interquartile range; MAC, myeloablative conditioning; MDS, myelodysplastic syndrome; MMF, mycophenolate mofetil; MPN, myeloproliferative neoplasm; MPO, myeloperoxidase; MTX, methotrexate; RIC, reduced‐intensity conditioning; WBC, white blood cell.

Regarding disease characteristics, 702 (27.0%) patients had a prior history of MDS or MPN, 1127 (43.3%) had adverse cytogenetics, and 225 (8.7%) had extramedullary disease at diagnosis. The median WBC count at diagnosis was 4.97 × 10⁹/L (IQR, 1.97–25.81 × 10⁹/L). The median proportions of BM and peripheral blood (PB) blasts at HCT were 17.2% (IQR, 6.8%–40.7%) and 2.0% (IQR, 0–15.0%) respectively (Table 1).

Regarding the duration from diagnosis to HCT, 708 (27.2%) patients received HCT within 3 months of diagnosis (Table 1). The median follow‐up period for survivors was 40 months (IQR, 16–72 months).

OS and LFS

In the entire cohort, the probability of OS at 3 years was 28.5% (95%CI, 26.6–30.3%) (Figure 1A). The univariate analysis identified several factors significantly associated with poorer OS: older age at HCT, poor PS, higher HCT‐CI, RIC regimen, secondary AML, adverse cytogenetic risk, intermediate WBC count at diagnosis, extramedullary disease at diagnosis, higher proportion of BM blasts at HCT, and higher proportion of PB blasts at HCT (Figure 2; Figure S1).

FIGURE 1.

Probability of overall survival (OS) (A) and leukaemia‐free survival (LFS) (B), and cumulative incidence of relapse (C) and non‐relapse mortality (NRM) (D) after allogeneic haematopoietic cell transplantation (HCT) for primary refractory acute myeloid leukaemia (AML).

FIGURE 2.

Probability of overall survival (OS) after allogeneic haematopoietic cell transplantation (HCT) for primary refractory acute myeloid leukaemia (AML) according to recipient age at HCT (A), performance status (PS) (B), HCT‐specific comorbidity index (HCT‐CI) (C), conditioning regimen (D), graft‐versus‐host disease (GVHD) prophylaxis (E), donor type (F), cytogenetic risk (G), white blood cell (WBC) count at diagnosis (H), extramedullary disease at diagnosis (I), bone marrow (BM) blasts at HCT (J), peripheral blood (PB) blasts at HCT (K) or duration from diagnosis to HCT (L).

In the multivariate analysis, significant factors for higher mortality were aged 50–59 years (HR: 1.55, 95%CI: 1.29–1.86), age ≥ 60 years (HR: 1.92, 95%CI: 1.60–2.31), PS 2–4 (HR: 1.59, 95%CI: 1.39–1.83), HCT‐CI ≥3 (HR: 1.16, 95%CI: 1.02–1.31), adverse cytogenetics (HR: 1.72, 95%CI: 1.55–1.98), WBC 3–20 × 10⁹/L at diagnosis (HR: 1.18, 95%CI: 1.04–1.34), extramedullary disease at diagnosis (HR: 1.43, 95%CI: 1.18–1.74), BM blasts at HCT ≥40% (HR: 1.25, 95%CI: 1.09–1.44), PB blasts at HCT >0–10% (HR: 1.16, 95%CI: 1.02–1.33), PB blasts at HCT ≥10% (HR: 1.42, 95%CI: 1.22–1.65) and duration from diagnosis to HCT ≥3 months (HR: 1.16, 95%CI: 1.03–1.32) (Table S1).

The probability of LFS at 3 years was 24.4% (95%CI, 22.7%–26.2%) (Figure 1B). The univariate analysis identified several factors significantly associated with LFS: older age at HCT, poor PS, higher HCT‐CI, conditioning regimen, secondary AML, cytogenetic risk, WBC at diagnosis, extramedullary disease at diagnosis, BM blasts at HCT and PB blasts at HCT (Figure 3; Figure S1).

FIGURE 3.

Probability of leukaemia‐free survival (LFS) after allogeneic haematopoietic cell transplantation (HCT) for primary refractory acute myeloid leukaemia (AML) according to recipient age at HCT (A), PS (B), HCT‐CI (C), conditioning regimen (D), GVHD prophylaxis (E), donor type (F), cytogenetic risk (G), WBC count at diagnosis (H), extramedullary disease at diagnosis (I), BM blasts at HCT (J), PB blasts at HCT (K) or duration from diagnosis to HCT (L).

In the multivariate analysis, significant factors for higher treatment failure were age 50–59 years (HR: 1.36, 95%CI: 1.14–1.63), age ≥ 60 years (HR: 1.52, 95%CI: 1.27–1.81), PS 2–4 (HR: 1.59, 95%CI: 1.39–1.83), adverse cytogenetics (HR: 1.64, 95%CI: 1.47–1.82), WBC 3–20 × 10⁹/L at diagnosis (HR: 1.16, 95%CI: 1.03–1.31), extramedullary disease at diagnosis (HR: 1.43, 95%CI: 1.18–1.72), BM blasts at HCT ≥40% (HR: 1.16, 95%CI: 1.01–1.34), PB blasts at HCT >0–10% (HR: 1.19, 95%CI: 1.05–1.36), PB blasts at HCT ≥10% (HR: 1.43, 95%CI: 1.23–1.66) and duration from diagnosis to HCT ≥3 months (HR: 1.16, 95%CI: 1.03–1.30) (Table S1).

Relapse and NRM

In the entire cohort, the cumulative incidence of relapse at 3 years was 49.8% (95%CI, 47.8%–51.8%) (Figure 1C). The univariate analysis identified several factors significantly associated with relapse: PS, GVHD prophylaxis, donor type, cytogenetic risk, WBC count at diagnosis, extramedullary disease at diagnosis, BM blasts at HCT and PB blasts at HCT (Figure 4).

FIGURE 4.

Cumulative incidence of relapse after allogeneic haematopoietic cell transplantation (HCT) for primary refractory acute myeloid leukaemia (AML) according to recipient age at HCT (A), PS (B), HCT‐CI (C), conditioning regimen (D), GVHD prophylaxis (E), donor type (F), cytogenetic risk (G), WBC count at diagnosis (H), extramedullary disease at diagnosis (I), BM blasts at HCT (J), PB blasts at HCT (K) or duration from diagnosis to HCT (L).

In the multivariate analysis, significant factors for a higher risk of relapse were HLA disparity (HR: 0.75, 95%CI: 0.61–0.92), adverse cytogenetics (HR: 1.77, 95%CI: 1.57–2.01), WBC 3–20 × 10⁹/L at diagnosis (HR: 1.21, 95%CI: 1.05–1.40), WBC ≥20 × 10⁹/L at diagnosis (HR: 1.19, 95%CI: 1.00–1.40), extramedullary disease at diagnosis (HR: 1.54, 95%CI: 1.25–1.90), PB blasts at HCT >0–10% (HR: 1.27, 95%CI: 1.09–1.48), PB blasts at HCT ≥10% (HR: 1.44, 95%CI: 1.21–1.72) and duration from diagnosis to HCT ≥3 months (HR: 1.15, 95%CI: 1.00–1.33) (Table S2).

The cumulative incidence of NRM at 3 years was 25.8% (95%CI, 24.0–27.5%) (Figure 1D). The univariate analysis found that age at HCT, HCT‐CI, GVHD prophylaxis, donor type, secondary AML and extramedullary disease at diagnosis were significantly associated with NRM (Figure 5; Figure S1).

FIGURE 5.

Cumulative incidence of NRM after allogeneic haematopoietic cell transplantation (HCT) for primary refractory acute myeloid leukaemia (AML) according to recipient age at HCT (A), PS (B), HCT‐CI (C), conditioning regimen (D), GVHD prophylaxis (E), donor type (F), cytogenetic risk (G), WBC count at diagnosis (H), extramedullary disease at diagnosis (I), BM blasts at HCT (J), PB blasts at HCT (K) or duration from diagnosis to HCT (L).

In the multivariate analysis, significant factors for higher NRM were age 40–49 years (HR: 1.51, 95%CI: 1.05–2.17), age 50–59 years (HR: 1.95, 95%CI: 1.40–2.72), age ≥ 60 years (HR: 2.55, 95%CI: 1.84–3.53), HCT‐CI ≥ 3 (HR: 1.41, 95%CI: 1.16–1.71), HLA disparity (HR: 1.42, 95%CI: 1.06–1.91) and adverse cytogenetics (HR: 0.81, 95%CI: 0.67–0.98) (Table S2).

Impact of conditioning regimen on outcomes

Next, we investigated whether the conditioning regimen affects OS, LFS, relapse or NRM after allogeneic HCT for primary refractory AML. First, the total body irradiation (TBI)‐based MAC regimen showed slightly better OS, LFS and lower NRM compared to the busulfan (BU)‐based MAC regimen (Figure S2). Second, the standard MAC regimen, which is defined as TBI/cyclophosphamide (CY), BU/CY or BU/fludarabine (Flu), was not inferior to the intensified MAC regimen, which is defined as the addition of chemotherapeutic drugs to the standard MAC regimen in terms of OS and LFS (Figure S2). Finally, the Flu/BU‐based RIC regimen had more relapse rates but lower NRM, OS and LFS compared to the Flu/melphalan (MEL)‐based RIC regimen (Figure S2).

Impact of FLT3‐ITD, NPM1 and TP53 on outcomes

We also investigated the impact of FLT3‐ITD, NPM1 or TP53 on OS, LFS, relapse or NRM after allogeneic HCT for primary refractory AML. Univariate analysis showed no significant difference in the risks of OS, LFS, relapse and NRM based on the presence or absence of FLT3‐ITD or NPM1 mutation, except that patients with FLT3‐ITD had a higher risk of NRM compared to those without FLT3‐ITD (Figure S3). The probabilities of OS, LFS and the cumulative incidences of relapse and NRM at 3 years were 13.4%, 12.3%, 66.2% and 21.5% for patients with TP53 mutation (Figure S3).

Scoring system for OS and LFS

We further investigated the impact of a scoring system on OS and LFS following allogeneic HCT for primary refractory AML. The scoring system, based on five factors for OS (age ≥ 50 years, PS ≥2, adverse cytogenetics, extramedullary disease at diagnosis and PB blasts at HCT ≥10%) (Table S1), effectively stratified OS. Patients with scores of 0, 1, 2 and ≥3 had 3‐year OS probabilities of 54.6%, 38.1%, 23.4% and 7.1% respectively (Figure 6A).

FIGURE 6.

Probability of overall survival (OS) (A) and leukaemia‐free survival (LFS) (B) after allogeneic haematopoietic cell transplantation (HCT) for primary refractory acute myeloid leukaemia (AML) according to risk score.

Similarly, the scoring system could stratify LFS. Patients with scores of 0, 1, 2 and ≥3 had 3‐year LFS probabilities of 45.8%, 33.5%, 19.3% and 6.5% respectively (Figure 6B).

DISCUSSION

In this registry‐based retrospective analysis of allogeneic HCT for adult primary refractory AML, we evaluated the clinical outcomes of 2600 adult AML patients who received their first allogeneic HCT between 2013 and 2022 in Japan. The probability of OS and LFS at 3 years was 28.5% and 24.4% respectively. The cumulative incidence of relapse and NRM at 3 years was 49.8% and 25.8% respectively. These results were consistent with previous reports on the results of primary refractory AML following allogeneic HCT. ²¹ , ²² , ²³ We also found that age at HCT, PS, cytogenetics, extramedullary disease and PB blasts at HCT were significant prognostic factors for OS and LFS.

Risk factors for primary refractory AML following induction chemotherapy originally included advanced age, poor PS, adverse cytogenetics, high WBC count at diagnosis, delayed blast clearance in PB and BM and the presence of certain genetic mutations such as TP53. ³⁴ , ³⁵ , ³⁶ , ³⁷ Some of these factors, such as advanced age, poor PS and adverse cytogenetics, were also prognostic factors for survival following allogeneic HCT for primary refractory AML in our study. Therefore, potential factors for remission induction failure in AML may also be relevant adverse factors for allogeneic HCT for primary refractory AML. Among factors associated with patients, advanced age and poor PS were associated with poor survival following allogeneic HCT for AML, irrespective of disease status at HCT. ⁸ , ⁹ , ¹⁴ , ¹⁵ , ¹⁶ , ¹⁹ , ²⁰ , ²⁵ , ³⁸ , ³⁹ , ⁴⁰ Advanced age is more likely to be associated with adverse cytogenetics, therapy‐related AML and a prior history of MDS/MPN, ⁴¹ , ⁴² , ⁴³ which could be associated with poor prognosis in AML. However, our data showed that advanced age was a risk factor only for NRM, but not relapse, following allogeneic HCT for primary refractory AML. Among disease characteristics, adverse cytogenetics ⁸ , ⁹ , ¹⁰ , ¹⁴ , ¹⁵ , ¹⁶ , ¹⁹ , ²⁰ , ²² , ²⁵ and higher PB ⁸ , ¹⁴ , ¹⁶ , ¹⁹ , ²⁰ , ²³ and BM blasts ¹¹ , ¹² , ¹⁴ , ¹⁷ , ²¹ , ²² at HCT are important prognostic factors for survival following allogeneic HCT in R/R AML. Indeed, adverse cytogenetics cannot be modified by therapeutic intervention before allogeneic HCT, whereas PB or BM blasts may be reduced by pretransplant treatment or by determining the optimal timing for initiation of the conditioning regimen prior to allogeneic HCT. PB blasts have been shown to be more strongly associated with unfavourable prognosis than BM blasts following allogeneic HCT for primary refractory AML, both in our study and in several previous reports. ⁸ , ¹⁴ , ¹⁶ , ¹⁹ , ²⁰ , ²³ This is likely because PB assessments are performed more frequently and more effectively capture blast variability than BM assessments conducted during the pretransplant evaluation, just prior to initiation of the conditioning regimen. Collectively, our data suggest that reducing PB blasts to <10% at the time of HCT may provide a practical guideline for pretransplant treatment prior to allogeneic HCT in primary refractory AML, when possible.

Our study showed that transplantation factors, such as donor type, conditioning regimen or GVHD prophylaxis, were not associated with OS or LFS. Several previous studies have shown that donor type significantly affects the survival rate following allogeneic HCT for R/R AML. ⁸ , ⁹ , ¹⁵ , ¹⁶ , ¹⁷ , ¹⁸ , ⁴⁴ Although unrelated donors ¹⁶ or donors other than HLA‐identical siblings ⁸ , ⁹ , ¹⁷ are adverse factors for survival in allogeneic HCT for R/R AML, our study demonstrated that haploidentical related donors, unrelated adult donors and cord blood were not inferior to matched related donors in terms of OS, LFS, relapse and NRM after allogeneic HCT for primary refractory AML. Indeed, only 15.1% of patients received allogeneic HCT from a matched related donor, and over half of the patients received allogeneic HCT from an HLA‐mismatched alternative donor, such as unrelated cord blood and haploidentical donors in our cohort. The quick availability of these alternative donors is useful for allogeneic HCT for primary refractory AML. In connection with donor type, our data showed that HLA mismatch was associated with a lower risk of relapse, but a higher risk of NRM. Graft‐versus‐leukaemia (GVL) effects could be enhanced by HLA‐mismatched HCT, ⁴⁵ , ⁴⁶ , ⁴⁷ which may contribute to the stronger GVL effects observed after HLA‐mismatched HCT. Therefore, the stronger GVL effect with HLA mismatch could be an attractive concept for primary refractory AML. Collectively, allogeneic HCT could offer a promising curative option for primary refractory AML, irrespective of donor type, conditioning regimen and GVHD prophylaxis.

Several prognostic scoring systems have been reported for predicting survival following allogeneic HCT in R/R AML. ⁸ , ⁹ , ¹⁶ , ²¹ Although several factors, such as poor PS, ⁸ , ¹⁶ adverse cytogenetics ⁸ , ¹⁶ and PB blasts, ⁸ , ¹⁶ were consistent with our data, our data identified that extramedullary disease was also included in the prognostic scoring system for allogeneic HCT for primary refractory AML. Importantly, donor type, which has been reported as one of the major factors in prognostic scoring systems in previous studies, ⁸ , ¹⁶ did not impact OS and LFS in our cohort. Interestingly, around half of the donor types were unrelated CBT in our cohort, which could also affect the impact of duration from diagnosis to HCT on outcomes after allogeneic HCT for primary refractory AML. Therefore, the prognostic scoring system in our study should be validated using an external cohort.

Our study had several limitations. First, this is a retrospective study using Japanese registry data, and several data points were unavailable, such as genetic mutation profiles of AML, types of induction or salvage chemotherapy, number of induction chemotherapy cycles before allogeneic HCT, dynamics of absolute PB and BM blast counts before allogeneic HCT and maintenance therapy following allogeneic HCT. All these findings could affect the outcomes after allogeneic HCT for primary refractory AML. ⁴⁸ , ⁴⁹ Second, our study was certainly subject to selection bias because it only included patients who received allogeneic HCT for primary refractory AML. Therefore, prospective studies are needed to evaluate the relative benefits of allogeneic HCT over conventional and novel agents without HCT. ⁵⁰ Despite these limitations, the strength of this study was the large number of adult patients who received allogeneic HCT for primary refractory AML, which provided a valuable opportunity to evaluate the prognostic factors of allogeneic HCT for this group.

In summary, our study revealed that the 3‐year OS and LFS rates were 28.5% and 24.4% after allogeneic HCT for primary refractory AML, which remains a dismal prognosis. Older age (≥50 years), poor PS (≥2), adverse cytogenetics, extramedullary disease at diagnosis, and higher PB blasts at HCT (≥10%) are significant risk factors for worse OS and LFS, and a scoring system using these five factors could help stratify OS and LFS. However, donor type, conditioning regimen and GVHD prophylaxis were not associated with OS or LFS. The proposed risk‐scoring system may aid in clinical decision‐making and patient counselling.

AUTHOR CONTRIBUTIONS

Takaaki Konuma designed the research, analysed the data, performed the statistical analysis and wrote the first draft of the manuscript. Yoshimitsu Shimomura, Shohei Mizuno, Satoshi Yamasaki, Shunsuke Yui, Yoshinobu Kanda and Masamitsu Yanada contributed to the critical review of the manuscript. All the other authors contributed to data collection. All authors approved the final version.

FUNDING INFORMATION

This work was supported in part by the Practical Research Project for Allergic Diseases and Immunology (Research Technology of Medical Transplantation) from Japan Agency for Medical Research and Development, AMED under grant 18ek0510023h0002.

CONFLICT OF INTEREST STATEMENT

Shohei Mizuno received research funding from Hayashikane Sangyo and a lecture fee from Johnson & Johnson. The other authors declare no competing interests.

ETHICS STATEMENT

This study was approved by the Adult AML Working Group of the JSTCT and by the institutional review board of the Institute of Medical Science, The University of Tokyo (2024‐43‐0912), where the study was conducted.

PATIENTS CONSENT STATEMENT

Written informed consent was obtained before registration for the TRUMP.

Supporting information

Figure S1.

Figure S2.

Figure S3.

Table S1.

Table S2.

Konuma T, Shimomura Y, Mizuno S, Yamasaki S, Yui S, Uchida N, et al. Prognostic factors for allogeneic haematopoietic cell transplantation outcomes in primary refractory acute myeloid leukaemia (2013–2022): A retrospective study by the adult acute myeloid leukaemia working group of the Japanese Society for Transplantation and Cellular Therapy. Br J Haematol. 2025;207(2): 484–497. 10.1111/bjh.20208

DATA AVAILABILITY STATEMENT

The data of this study are not publicly available due to ethical restrictions that it exceeds the scope of the recipient/donor's consent for research use in the registry. Data may be available from the corresponding author upon reasonable request and with permission of the JSTCT/JDCHCT.