Outcomes of adult patients with type 1 primary refractory acute myeloid leukemia: a single center experience.

Adult patients with newly diagnosed de novo acute myeloid leukemia (AML), who had less than a 50% reduction in blast numbers and with > 15% residual blasts after first cycle of induction chemotherapy, defined as type 1 primary refractory (REF1), have grave prognosis. We retrospectively analyzed the data of 58 patients with REF1 who received salvage treatments with curative intension to evaluate the impact of salvage regimens with regard to response and overall survival (OS). Seventeen patients received intermediate- or high-dose cytarabine (ID/HD Ara-C) based intensive salvage chemotherapy, 36 patients received G-CSF primed less intensive chemotherapy and 5 patients received novel targeted drugs based low intensive therapy. The CR/CRi and MLFS rate was 6/17 and 2/17, 14/36 and 3/36, 3/5 and 0/5, respectively. The median OS for the whole cohort was 20.3 months. Median OS was comparable between the 3 arms. Overall, 42 patients underwent allogeneic hematopoietic stem cell transplantation (allo-HSCT), 14 patients in the intensive arm, 24 patients in the less intensive arm and 4 patients in the low intensive arm. Median survival for allo-HSCT patients was significantly longer than for non-allo-HSCT patients (38.8 months vs. 2.1 months, p < 0.001). In multivariate analysis, achievement of CR/CRi after the salvage regimen were predictive of OS. We conclude that no significant difference in outcome among traditional salvage regimens in patients with REF1. G-CSF primed less intensive chemotherapy could serve as an alternative of ID/HD Ara-C based intensive chemotherapy and allo-HSCT is indispensable for long-term survival.

Keywords: Acute myeloid leukemia; adult; allogeneic stem cell transplantation; primary refractory

The standard induction chemotherapy in adult patients with newly diagnosed acute myeloid leukemia (AML) aged < 60 years has long been '3 + 7', i.e. 3 days of idarubicin and 7 days of cytarabine (IA) [[1]]. However, approximately 30% of patients represent refractory to the initial therapy [[2]]. The 2022 European LeukemiaNet (ELN) characterizes primary refractory AML as the failure to achieve a completer remission (CR) after at least two courses of intensive chemotherapy [[4]]. Such patients are associated with a very poor prognosis, with median overall survival (OS) near 6 months [[4]]. Patients with less than a 50% reduction in blast numbers and with > 15% residual blasts after first cycle of induction chemotherapy, defined as type 1 primary refractory (REF1), have the same dismal prognosis and the utilization of REF1 criteria may permit the early identification of these high-risk patients [[5]]. Moreover, treatment is challenging in this setting as optimal salvage therapy remains undetermined [[6]].

Regimens containing intermediate – or high-dose cytarabine (ID/HD Ara-C) alone or in combination with an anthracycline, fludarabine, are generally considered as the standard option for patients with REF1 [[4], [6], [8]]. However, the CR rates between 30% and 50% is unsatisfactory and many patients with REF1 may not be suitable for the second intensive chemotherapy due to factors such as poor performance status. A less intensive therapy, CAG regimen (aclarubicin, cytarabine, and granulocyte colony-stimulating factor [G-CSF]) or CAG-like regimen, with or without a hypomethylating agent (HMA) is a widely used special chemotherapy regimen for AML in East Asia [[9], [11]]. Previous studies have demonstrated comparable antineoplastic activity with less toxicity in patients with relapsed/refractory (r/r) AML. Besides, HMA-based combination therapy with novel targeted drugs, like venetoclax, FLT3 inhibitors and IDH1/2 inhibitors, have also shown encouraging results as a low intensive therapy in this population in recent years [[7], [12], [14]].

Herein, from a phase III study of induction therapy for newly diagnosed primary AML, we report the efficacy of salvage regimens with different intensity in 58 patients with REF1, mainly intensive chemotherapy with ID/HD Ara-C and less intensive chemotherapy with G-CSF priming, to evaluate the impact of salvage regimens in this patient population with regard to response and overall survival (OS).

Clinical trial NCT02323022 is a prospective, randomized controlled phase III clinical trial to compare the clinical efficacy and safety of IA regimen and IAC regimen (with an additional 5 days of cladribine) as induction chemotherapy for initial diagnosed adult de novo AML patients. Treatment was administered as described previously [[15]]. Patients who withdrew from the trial due to REF1 and received salvage treatments with curative intension were included in this study. The primary objective of this study was to evaluate the efficacy of salvage regimens with different intensity on response and OS in these patients. The study was approved by the Institutional Review Board of our hospital. Informed consent was obtained from all patients in accordance with the principles expressed in the Declaration of Helsinki.

Salvage regimen for individual patient was selected at the discretion of each physician. ID/HD Ara-C based regimens comprised of at least cytarabine 1–2 g/m2 iv daily for 3–5 days. As for the less intensive chemotherapy, the regimen consisted of cytarabine (10 mg/m2/q12 h, subcutaneously, on days 1–14), aclarubicin 20 mg infused intravenously qod for 7 days and G-CSF administered 1 d before chemotherapy (150 μg/m2/day, subcutaneously, on days 0–14). Decitabine or azacitidine, could be used prior to chemotherapy. Low intensive therapy is a standard HMA regimen (decitabine or azacitidine) in combination with venetoclax or sorafenib. No patients received IDH1/2 inhibitors in this study. Prophylactic antibiotics, antifungals, and antiviral agents were administered in accordance with institutional guidelines. G-CSF support in the aplastic period was permitted and determined by doctors' discretion. Allogeneic hematopoietic stem cell transplantation (allo-HSCT) is recommended for patients who achieved therapeutic response and was performed as described previously [[16]].

To assess response to therapy, a bone marrow (BM) aspiration was performed no late than 4 weeks after therapy. Responses were graded per the ELN 2022 guidelines. Overall response rate (ORR) includes complete remission (CR), CR with incomplete blood count recovery (CRi), and morphologic leukemia-free state (MLFS). CR was defined as a normal BM aspirate with ≤ 5% marrow blasts, absolute neutrophil count ≥ 1 × 109/L, platelet count ≥ 100 × 109/L, and no evidence of circulating or extramedullary leukemic blasts. Partial remission (PR) was defined by a reduction of 50% or more leukemia blast cells as long as the absolute blast count was between 5% and 25%. Measurable residual disease (MRD) in BM samples was assessed by multiparameter flow cytometry with a sensitivity of 0.1%. Side effects were assessed according to the Common Toxicity Criteria for Adverse Events version 5.0. OS was measured from start of re-induction until death or censored at last follow-up. Kaplan–Meier method was used to estimate the OS probability. The data cutoff for this study was October 31, 2022. Multivariable Cox regression analyses were used to evaluate the associations between patient characteristics and OS. The data analyses were done using GraphPad Prism 8.

Between January 2015 and December 2020, 58 patients with REF1 were included in this study (Figure 1). There were 36 men and 22 women with a median age of 40 (range 18–58) years. The detailed patient baseline characteristics are shown in Table 1. Seventeen patients (29.3%) received ID/HD Ara-C based intensive salvage chemotherapy, 36 patients (62.1%) received G-CSF primed less intensive chemotherapy and 5 patients (8.6%) received novel targeted drugs based low intensive therapy. Compared with the ID/HD Ara-C based intensive chemotherapy group, patients in the G-CSF primed less intensive chemotherapy group were older, with median ages of 28 and 42 years, respectively (p =.04). The remaining characteristics were similar between the two groups.

DIAGRAM: Figure 1. Flow diagram of the patients.

Table 1. Patient characteristics.

Among 17 patients treated with ID/HD Ara-C based intensive chemotherapy, 6 patients (35.3%) achieved CR/CRi and 2 (5.9%) achieved MLFS, compared with 14 patients (38.9%) who achieved CR/CRi and 3 patients (8.3%) who achieved MLFS in the G-CSF primed less intensive chemotherapy group (p = 0.80). MRD-negativity rate was 35.3% versus 22.2%, respectively (OR, 1.9; 95% CI, 0.5-6.8, p = 0.50). Among 5 patients treated with novel targeted drugs based low intensive therapy, 3 patients achieved CR/CRi, whereas 2 patients were refractory. Six patients without response in the ID/HD Ara-C based intensive chemotherapy group received subsequent reinduction, ORR was achieved in 2 patients and PR was achieved in other 2 patients. Six patients without response in the G-CSF primed less intensive chemotherapy group received subsequent reinduction, ORR was achieved in 2 patients and PR was achieved in other 4 patients. Given the small patient number, the value of mutational genotypes was not detected.

The median OS for the whole cohort was 20.3 months (95% CI, 9.6–31.0) (Figure 2a.). Median OS in the ID/HD Ara-C based intensive chemotherapy arm was 27.8 months (95% CI, 3.8–41.0), compared with 17.7 months in the G-CSF primed less intensive chemotherapy arm (95% CI, 0–30.2; HR, 1.0, 95% CI, 0.5–2.2; p = 0.93) (Figure 2b).

Graph: Figure 2. Kaplan–Meier curves for overall survival. (a) The whole cohort. (b) ID/HD Ara-C based intensive chemotherapy group vs. G-CSF primed less intensive chemotherapy group. (c) ID/HD Ara-C based intensive chemotherapy group vs. G-CSF primed less intensive chemotherapy group, censored at time of allo-HSCT. (d) Patients who received allo-HSCT vs. those not.

Overall, 42 patients underwent allo-HSCT, 14 patients in the intensive chemotherapy arm, 24 in the less intensive chemotherapy arm and 4 patients in the low intensive therapy arm. Seventeen patients (40.5%) received an HLA matched related donor, 3 patients (7.1%) received an unrelated donor and 22 patients (52.4%) received an haploidentical related donor. Thirteen patients (31.0%) had active disease before allo-HSCT. After censoring for allo-HSCT, median OS was 6.4 months in the ID/HD Ara-C based intensive arm compared with 7.7 months in the-CSF primed less intensive arm (HR, 1.8, 95% CI, 0.5–6.6; P = 0.37) (Figure 2c). Median OS for allo-HSCT patients was significantly longer than for non-allo-HSCT patients: 38.8 months (95% CI, 4.9–72.7) versus 2.1 months (95% CI, 0.1–4.1), respectively (p < 0.001) (Figure 2d). Median OS was 33.2 months (95% CI, 12.1–54.3) for patients treated with ID/HD Ara-C based intensive chemotherapy who underwent allo-HSCT and not reached for patients treated with the G-CSF primed less intensive chemotherapy and then underwent allo-HSCT (Figure 3a). The difference between the two arms did not reach statistical significance.

Graph: Figure 3. Kaplan–Meier curves for overall survival of the 42 patients underwent allo-HSCT according to (a) reinduction regimen and (b) disease status after salvage treatment.

Twenty-two patients died after allo-HSCT. Seven patients died of transplant-related causes, 1 patient died of myocardial infarction, 14 patients had relapsed at a median time of 9.6 months (range, 1.2–20.7) after transplantation and ultimately died. In univariate and multivariate analysis, achievement of CR/CRi after the salvage chemotherapy regimen and ECOG score at transplant were predictive of OS (Table 2) (Figure 3b).

Table 2. Univariate and multivariate analysis of clinical parameters on overall survival.

Early death (within 30 days) was 0% in the ID/HD Ara-C based intensive group compared with 5.6% (n = 2) in the G-CSF primed less intensive group, while the 60-day mortality was 5.9% (n = 1) and 8.3% (n = 3), p = 0.75. Myelosuppression was ubiquitous. For patients with CR/CRi, the median time to absolute neutrophil count more than 0.5 × 109/L and platelet levels more than 20 × 109/L was 16 (range 10-20) and 17 (range 9-28) days in the ID/HD Ara-C based intensive group, and 20 (range 8-33) and 16 (range 7–30) days in the G-CSF primed less intensive group, respectively. The incidence of febrile neutropenia and documented infections were 29.4% (5/17) and 52.9% (9/17), 27.8% (10/36) and 50.0% (18/36), respectively.

In this study, we described our experience with salvage regimens with different intensities in 58 de novo AML patients with REF1. Our results show that there was no significant difference in OS among patients assigned to the different salvage regimens and allo-HSCT is indispensable for long-term survival in these high-risk patients.

Failure to respond to induction therapy is a major unfavorable prognostic factor for outcome in AML [[6], [8]]. However, there is no universally agreed criteria used to define induction failure [[5], [17]]. A large-scale retrospective analysis by Ferguson et al. demonstrated that compared to the classic criteria with failure to achieve CR after two courses of induction chemotherapy, using the REF1 criteria allowed early identification of patients whose outcome after one course is very poor, and permitted early intervention [[5]].

For patients with induction failure, the goal of further anti-leukaemic therapy is to achieve remission and subsequently proceed to allo-HSCT, which may offer the best chance of cure [[6], [8]],[18]]. The treatment is extremely challenging for patients with REF1 as the optimal salvage regimen option is unclear. ID/HD Ara-C based intensive salvage chemotherapies, including ID/HD Ara-C alone, MEC, FLAG/FLAG-Ida, are the commonly used regimens, with CR rates between 30% and 50% [[6], [8]]. Due to the significant heterogeneity in the patient's population, it is hard to directly compare these regimens. In general, no salvage therapy showed superiority compared to the others, but HD AraC combined with gemtuzumab ozogamicin is associated with a high probability of achieving a response in one study [[6]]. An alternative to these intensive chemotherapy, concurrent use of G-CSF to sensitize the leukemic cells to the cytotoxicity of chemotherapy with aclarubicin and low-dose but relatively long-term Ara-C, named CAG regimen, is a less intensive chemotherapy, which was first reported by a Japanese group with encouraging results and low toxicity in elderly and advanced AML patients [[9]]. Similar to our study, former pilot and retrospective studies demonstrated comparable, even better efficacy to that of ID/HD Ara-C based intensive chemotherapies [[10]]. Unfortunately, no prospective randomized clinical trials to compare these two treatment modalities have been reported. Meanwhile, it is noteworthy that although early evidence showed chemotherapy given concomitantly with G-CSF had a better CR rate and outcome compared to chemotherapy alone, multicenter randomized trials failed to verify the G-CSF priming benefit [[19], [21]]. On the other hand, as remission rates are unlikely to be further improved with conventional chemotherapy alone, combination chemotherapy with targeted agents, like BCL-2 inhibitors, FLT3 inhibitors and IDH1/2 inhibitors, is the direction for future trials. Besides, it would be important to prospectively identify those most likely to benefit from these salvage treatments.

A sequential transplant approach combining a short course of intensive chemotherapy and a reduced-intensity conditioning allo-HSCT, named FLAMSA protocol, was a slightly different therapeutic strategy [[22]]. Previous studies showed this sequential therapy for patients with r/r AML is associated with a significant antileukemic effect, relatively low toxicity, and 40-60% long-term OS [[22], [24]]. In a randomized controlled phase III ASAP Trial, sequential transplant approach results in similar OS compared to intensive remission induction chemotherapy followed by allo-HSCT [[25]]. According to the intention-to-treat, 3-years OS from randomization was 51.0% and 54.2%, respectively, (p = 0.47). Thirteen patients underwent salvage allo-HSCT with active disease in our study, it is of no surprise that the outcome is much inferior compared to the ASAP trial, as patients with active disease in this study are more refractory. These data emphasize the importance to proceed early to up-front allo-HSCT for securing a favorable outcome in FLAMSA protocol.

Because of the time and logistics required to perform an unrelated donor search, the majority of patients received allografts from either HLA matched siblings or haploidentical related donors in our study, only 3 patients underwent transplantation from an unrelated donor. Compared to HLA-matched HSCT, haplo-HSCT showed comparable long-term survival with similar relapse rate and transplant-related mortality rate. As relapse after allo-HSCT remains the major clinical challenge and some studies showed that grafts from HLA-haploidentical donors may have a superior graft-versus-leukemia effect [[26]], it could be reasonable to accept the haploidentical related donor as the first choice for patients with REF1 in experienced transplant centers.

There are several limitations of this study, mainly attributed to its retrospective nature, the relatively small sample size, the age discrepancy between the two groups and some refractory patients with low blast percentage may be excluded. As most patients achieved remission after first salvage treatment proceeded to allo-HSCT in our study, it is difficult to assess treatment effect of the salvage regimen independent from allo-HSCT. Moreover, only few patients received targeted therapy as first salvage treatment in our study for the reason that these agents were not available at that time in China. As these agents have significantly changed the treatment paradigm in AML [[7]], they appear likely to become the standard of care for patients with REF1 in the future, at least in a subset of patients. Meanwhile, these new emerging therapies will continue to address the question about the choice of salvage treatment as an interim step and direct allo-HSCT in patients with REF1. However, three patients in this study who achieved remission with targeted agents and underwent allo-HSCT eventually relapsed, indicating that long-term leukemia-free survival remains an unresolved problem. Best using these agents, including maintenance therapies post HSCT to decrease the risk of relapse may further benefit patients [[28], [30]].

In conclusion, this retrospective analysis suggests that no significant difference in outcome among traditional salvage regimens in patients with REF1. G-CSF primed less intensive chemotherapy could serve as an alternative of ID/HD Ara-C based intensive chemotherapy and allo-HSCT is critical for long-term disease control.

The data that support the findings of this study are available from the corresponding author upon reasonable request.

All procedures performed in studies involving human participants were in accordance with the ethical standards of the institutional and/or national research committee and with the 1975 Helsinki declaration and its later amendments or comparable ethical standards.

Informed consent was obtained from all individual participants included.

Z.Y.Y., Y.Y., and Y.M.Z. interpreted the data and wrote the paper; J.C., Y.X., S.L.X., H.Y.Q., X.W.T., Y.H., S.N.C. and A.N.S. collected research data and revised the manuscript. D.P.W. and Y.W. designed the research, analyzed the data and edited the manuscript.

No potential conflict of interest was reported by the author(s).