Intensive chemotherapy vs venetoclax/hypomethylating agents for patients aged 60 to 75 years with favorable, NPM1-mutated AML Open Access

Nucleophosmin 1 (NPM1) mutations are detected in one-third of adult patients with acute myeloid leukemia (AML). In the absence of a concomitant FLT3 internal tandem duplication (FLT3-ITD) mutation, NPM1-mutated (NPM1m) AML is considered favorable risk due to its responsiveness to intensive chemotherapy (IC).^1,2 Patients with NPM1m AML who achieve complete remission (CR) without measurable residual disease (MRD) after 2 cycles of IC may be cured with continued chemotherapy alone.³ IC is therefore the preferred first-line therapy for NPM1m/FLT3-ITD wildtype (wt) AML in both the National Comprehensive Cancer Network and European LeukemiaNet 2022 (ELN 2022) guidelines (National Comprehensive Cancer Network and ELN guidelines).⁴ 

The combination of venetoclax (VEN) and a hypomethylating agent (HMA) is an alternative to IC that is indicated for patients aged 75 years or older or who are deemed ineligible for IC.⁵ VEN/HMA is still quite effective in this patient population with a median overall survival (OS) of 14.7 months in the original VIALE-A study; long-term follow-up analysis reported a 2-year OS of 38%.⁶ A drawback of the regimen is that the potential for cure with VEN/HMA alone is not well established and most patients will require allogeneic hematopoietic cell transplant (alloHCT) if the goal of therapy is cure.

Although IC remains the standard of care for fit older patients with NPM1m AML, its use is associated with increased toxicity with advancing age. Older patients who are deemed unfit are routinely treated with VEN/HMA instead,^7,8 but the assessment of fitness is subjective and especially challenging in the setting of acute leukemia. In patients with favorable-risk NPM1m AML, the potential for cure without the need for alloHCT makes the initial treatment decision particularly consequential. A clearer understanding of the clinical outcomes associated with each treatment approach would aid decision-making in this patient population. We, therefore, sought to evaluate survival and alloHCT outcomes in a cohort of patients aged 60 to 75 years with newly diagnosed, favorable-risk NPM1m AML, including patients who received both IC and VEN/HMA.

This is a retrospective cohort study using electronic medical record (EMR) data from Johns Hopkins Hospital in Baltimore, MD. The data set included all patients with a new diagnosis of AML who received induction therapy at Johns Hopkins Hospital between 12 August 2013 and 24 May 2024 (Figure 1). Patients were included in the analysis cohort if they had a new diagnosis of AML with an NPM1 mutation in the absence of a concomitant FLT3-ITD mutation or adverse cytogenetics as described in the ELN 2022 guidelines.² 

At the time of diagnosis, detection of NPM1 mutations and other secondary mutations was performed by polymerase chain reaction and next-generation sequencing. Cytogenetic abnormalities were collected via metaphase karyotyping and fluorescent in-situ hybridization at the time of diagnosis. Additionally, patient-specific characteristics were captured including age, reported sex, and reported race. Per institutional practice, Eastern Cooperative Oncology Group (ECOG) functional status scores were documented by the patient’s oncologist at time of diagnosis. Based on past medical history coded by International Classification of Diseases, Tenth Revision scores and medical problems listed in the presenting history of present illness and discharge summary, the Charlson Comorbidity Index (CCI) was calculated.⁹ Patients who received intensive therapy (IC; either conventional 7+3 or liposomal cytarabine/daunorubicin [CPX-351]) or VEN/HMA could have received other therapies such as gemtuzumab ozogamicin for cytoreduction or targeted inhibitors, including FLT3 inhibitors, per the US Food and Drug Administration approval; 10 patients in the IC and zero patients in the VEN/HMA cohort received FLT3 inhibitors for FLT3-tyrosine kinase domain comutations, and 1 patient in each cohort received an IDH2 inhibitor. Supplemental Table 1 lists the induction chemotherapy regimens by intensity.

Patients who underwent alloHCT all received nonmyeloablative conditioning consisting of intravenous cyclophosphamide at a dose of 14.5 mg/kg per day administered on days −6 and −5, IV fludarabine at 30 mg/m² daily from days −6 to −2, and total body irradiation (200 cGy or 400 cGy) delivered on day −1. Graft-versus-host disease prophylaxis consisted of posttransplant cyclophosphamide at 50 mg/m² on days 3 and 4, mycophenolate mofetil administered from days 5 to 35, and either tacrolimus or sirolimus from day 5 until day 60, 90, or 180.

The primary outcome was OS compared between treatment groups using Cox proportional hazard models. Secondary outcomes were rates of remission, rates of consolidative alloHCT, OS between treatment groups when stratified by alloHCT status, and alloHCT outcomes. OS was measured from the day of diagnosis to the day of death from any cause.

All statistical analyses were performed using R, version 4.4.1,¹⁰ and a 2-sided P value of <.05 was considered statistically significant for all analyses. Descriptive statistics were used to summarize the characteristics of the study cohort. All continuous variables were non-normally distributed based on tests of skewness; therefore, medians and interquartile ranges are reported; categorical variables were reported as counts and frequencies. Fisher exact tests and Wilcoxon rank sum tests were used for comparisons of continuous and categorical variables respectively. OS was analyzed using the Kaplan-Meier method and the log method provided the 95% confidence intervals (CIs).

Cox proportional hazard models were used to estimate the effect of covariates on OS in univariable and multivariable analyses. To determine which variables to include in the multivariate analysis, all factors listed in Table 1 were analyzed as an independent variable for OS using the Kaplan-Meier method. Variables with a P value <.1 in their univariate analysis were included for consideration in the multivariable analysis; PTPN11 mutation status was excluded despite a P value <.1 given its low frequency in the cohort.

CR and CR with incomplete hematologic recovery (CRi) were defined in ELN 2022 guidelines. Only patients who received an alloHCT in first remission were included in the transplant data. A competing event analysis was performed to compare rates of nonrelapse mortality (NRM) and relapse for patients who received an alloHCT; patients were stratified by the induction chemotherapy regimen (IC vs VEN/HMA) they had received.

This study was approved by the Johns Hopkins Institutional Review Board (IRB00390712).

Table 1 illustrates the baseline characteristics of the study. Two-hundred forty-five patients with a new diagnosis of NPM1m AML were identified. One hundred ten patients had intermediate or adverse risk disease as defined by the ELN 2022 guidelines and were excluded. Of the 135 patients remaining, 77 patients were excluded for age <60 years or >75 years, and 3 patients were excluded for receiving a non-IC regimen without VEN. Fifty-five patients remained, of which 36 (65%) received first-line IC and 19 (35%) received first-line VEN/HMA. Patients who received VEN/HMA were on average older than patients who received IC (69.6 years vs 66.1 years, P = .005). The study population was majority White (73%) with a slight female predominance (51%). Most patients (81%) in the study had an ECOG functional status score of 0 or 1 at the time of diagnosis. Although there was no statistical difference in functional status between patients who received VEN/HMA and IC (P = .3), the VEN/HMA group had double the prevalence of patients with ECOG score of 2 or 3 (28% vs 14%). Patients who received IC had a lower median CCI score compared to those who received VEN/HMA (4.5 vs 6.2, P = .003). Patients newly diagnosed with NPM1m AML before full approval of VEN/HMA were 2 times more likely to have received IC (81% vs 37%, P = .001).

White blood cell (WBC) count at diagnosis was similar between the 2 groups (74 × 10³/μL vs 47 × 10³/μL, P =.12). One-third of the patients had extramedullary disease; of the patients with extramedullary disease, 61% had central nervous system leukemia, 22% had leukemia cutis, 11% had soft tissue involvement, and 1 patient (6%) presented with both central nervous system leukemia and leukemia cutis. There was no difference in rates of secondary AML between the 2 groups (11% vs 16%, P = .7). The most common comutations in this population were FLT3-tyrosine kinase domain, DNMT3A, IDH2, and IDH1. Although there are no statistical differences between co-occurring mutations between groups, all patients with an NRAS mutation received IC, whereas the VEN/HMA group had over double the prevalence of IDH2 co-mutation (42% vs 19%, P = .07).

There were no differences in rates of response between patients who received IC and VEN/HMA for induction chemotherapy. In the IC group, 81% and 8.3% of patients achieved CR and CRi, respectively; only 13% of patients did not achieve a response. In the VEN/HMA group, 74% and 11% of patients achieved CR and CRi respectively; 16% of patients did not achieve a response (P = .80). Of the 32 patients who received IC and achieved a CR or CRi, 29 patients (91%) had MRD testing by flow cytometry after cycle 1 of therapy, and none were positive. Of the 16 patients who received VEN/HMA and achieved a CR or CRi, 16 patients (100%) had MRD testing after cycle 1 of therapy, and only 1 (6.3%) was positive.

Between patients treated with IC and VEN/HMA, there was no statistical difference in OS (Figure 2A). The median survival for the IC group was 6.2 years (95% CI, 3.26 to not reached) and the median survival for the VEN/HMA group was 4.9 years (95% CI, 1.1 to not reached). The only factors associated (P <.10) with OS were WBC at diagnosis, presence of DNMT3A mutation, presence of PTPN11 mutation, and CCI score (supplemental Table 2). In a multivariate analysis including WBC at diagnosis, DNMT3A mutation status, CCI score and treatment regimen (Table 2), only WBC at diagnosis was associated with OS (P = .04); there was no difference between VEN/HMA and IC in OS (P = .69). Sixty-day survival in the IC and VEN/HMA cohorts were 97.2% and 100%, respectively. PTN11 mutation was excluded from the multivariable analysis given the small number of patients with the mutation in this cohort.

Twenty-five patients who received IC and 7 patients who received VEN/HMA underwent alloHCT in first remission. Supplemental Table 3 shows the baseline characteristics stratified by chemotherapy regimen and alloHCT status. Most donors were haploidentical in the intense (74%) and less intense (67%) groups (P = .23). Despite similar rates of remission between the 2 groups, patients who received IC were more likely to receive a consolidative alloHCT in first remission compared to those who received VEN/HMA (69% vs 37%, P = .020). There was no statistically significant difference in time from induction therapy to alloHCT between those who received intensive induction compared to VEN/HMA (mean 162 days vs 207 days, P = .32).

In a univariate analysis, alloHCT in first remission was associated with improved OS (hazard ratio, 0.30; 95% CI, 0.12-0.74). Given the disproportionate number of patients in the IC group who underwent alloHCT, we evaluated if rates of alloHCT confounded survival. We therefore performed survival analyses between the IC and VEN/HMA groups stratifying for alloHCT. For patients who received alloHCT in first remission, there was no difference in OS (Figure 2B; P = .49) between IC and VEN/HMA. Likewise, for patients who did not receive alloHCT, there was no difference in OS (Figure 2C; P = .86).

We performed a competing events analysis for NRM and relapse to further examine outcomes of patients who underwent alloHCT (Figure 2D). Among patients who received alloHCT, there was no difference in rates of relapse (P = .09) and NRM (P = .89) between IC and VEN/HMA. Out of the 7 patients who received an alloHCT after VEN/HMA induction, no patient relapsed and 1 died of NRM (14%). Of the 25 patients who received an alloHCT after IC induction, 12 relapsed (48%) and 3 died of NRM (12%). After 18 months post-alloHCT, there was no further NRM in the cohort with follow-up up to over 6 years.

In this retrospective analysis of patients aged 60 to 75 years with newly diagnosed, favorable-risk NPM1m AML, there was no difference in OS between patients who received IC compared to those who received VEN/HMA. This study contributes to a growing body of literature examining the comparative efficacy of IC to VEN/HMA in patients with AML. One large, real-world analysis of patients from a nationwide electronic health record–derived database found better OS for older patients with AML who received 7+3 compared to those who received VEN/HMA, though there were notable differences in baseline characteristics between the 2 cohorts. After adjusting for covariates, however, there was no difference in outcomes between treatment arms.¹¹ In another study by the same research group, OS was similar for patients who received CPX-351 compared to those who received VEN/HMA.¹² 

NPM1m AML is a unique subtype in that it is sensitive to both IC and VEN/HMA. Two prior studies have examined this question specifically in the NPM1m AML population. In a single-center cohort of older patients with NPM1m AML, patients aged 65 years and older had better OS with VEN/HMA compared to IC, even after limiting the analysis to FLT3-ITDwt patients.¹³ A more recent multicenter retrospective study of patients with NPM1m AML who were aged 60 years and older found no statistically significant difference in OS between those who received IC and VEN/HMA in the 60 to 75 years age subset. However, after stratifying by FLT3-ITD mutation status, IC led to better OS.¹⁴ 

Our analysis differs from these prior studies in its particular focus on patients with favorable-risk NPM1m AML in the 60 to 75 years age range, a key demographic in which patients are at higher risk of IC-related toxicity but are not routinely considered for VEN/HMA based on age alone. Despite differences in the median age and the burden of medical comorbidities between patients in the 2 treatment arms of our cohort, outcomes were similar before and after adjusting for these and other covariates.

The results of this analysis are directly relevant to the bedside of a newly diagnosed patient with NPM1m/FLT3-ITDwt AML for whom fitness is difficult to ascertain. Our results suggest that either regimen is appropriate in this setting and that the more important factor in determining patient outcomes is the choice of whether to pursue alloHCT in first remission, which appears to benefit this patient population, regardless of induction intensity. Patients who underwent alloHCT after VEN/HMA had equivalent OS to those who received IC. In fact, patients in the VEN/HMA group who underwent alloHCT had a remarkably low rate of relapse relative to IC (P = .09), suggesting that VEN/HMA led to pre-alloHCT remissions that were as deep as those achieved by IC. Although a larger percentage of patients in the IC group underwent consolidative alloHCT, we suspect this was due to differences in alloHCT eligibility between the 2 groups, as reflected in the difference in CCIs.

The similar outcomes between treatment arms in this cohort may be impacted by the treatment setting. First, this is a single-center analysis that included patients treated at an academic center with extensive experience administering IC and where patients are routinely referred for alloHCT in first remission. As a result, a high percentage of patients underwent alloHCT in first remission (69% vs 37% for the IC and VEN/HMA groups, respectively) and outcomes may be different in settings where alloHCT is less common or unavailable.

This study has several limitations. First, the study is limited by its small sample size, particularly in the comparison of alloHCT outcomes. Only 7 patients in the VEN/HMA group underwent alloHCT, making it difficult to detect differences in outcomes given the small sample. Furthermore, as a nonrandomized study, treatment choice and clinical outcomes were likely influenced by confounders, despite our efforts to adjust for patient and disease-related factors. For example, although the difference in WBC between treatment groups did not reach statistical significance, the IC group had nearly double the median WBC compared to the VEN/HMA group. Elevated WBC was associated with increased mortality in the multivariate analysis, suggesting the possibility that the IC group represented a higher-risk cohort. Residual confounding by disease burden may therefore have influenced the observed outcomes. Treatment selection was also likely influenced by additional genetic characteristics. A total of 42% of patients in the VEN/HMA group had a concomitant IDH2 mutation, which is associated with particular responsiveness to VEN/HMA, whereas none of the VEN/HMA treated patients had an NRAS or KRAS mutation (the latter not reported due to the low incidence in this cohort), mutations which confers resistance to VEN-based therapy.¹⁵ Finally, patients who received VEN/HMA were treated more recently such that improvements in supportive care and disease monitoring may have contributed to improved outcomes in the VEN/HMA group compared to the IC patients.

Ultimately, prospective, randomized trials comparing IC to VEN/HMA in specific genetic subsets are needed. Ideally, these studies will capture other outcomes that are often absent from retrospective data sets, such as MRD testing as well as patient-reported outcomes. Several such studies are already underway (www.ClinicalTrials.gov identifiers: NCT05904106, NCT04801797, and NCT05554406). Along with the data presented in this analysis, the results of these studies will provide a better foundation for understanding the comparative effectiveness of existing treatment options, which is critical to inform the direction of future advances.

Contribution: A.Z., V.P.S.K., and A.A. designed research, collected data, performed statistical analysis, and wrote the manuscript; and J.W., M.L., I.G., A.D., G.G., L.G., W.D., T.K., T.J., G.P., R.J., and B.D.S. interpreted data and wrote the manuscript.

Conflict-of-interest disclosure: J.W. received funding from Amgen, Consumer Value Store, Servier, and Syndax. M.L. reports consultancy/honoraria from Astellas, Daiichi-Sankyo, Novartis, Syndax, GlaxoSmithKline, and Takeda. I.G. reports research funding from Incyte, Amgen, Merck, and Kura Oncology; and served on the advisory boards of In8Bio, Syndax, and Astellas. A.D. received funding from Bristol Myers Squibb, Novartis, Takeda, Pfizer, Agios, Keros, Geron, Astellas, Shattuck Labs, Kura, AbbVie, and Servier. T.J. receives institutional research support from Cell Therapeutic, Inc. Biopharma, Kartos Therapeutics Inc, Incyte, Bristol Myers Squibb, TScan Therapeutics, and Karyopharm Therapeutics; and reports advisory board participation with Bristol Myers Squibb, Incyte, AbbVie, CTI, Kite, Cogent Biosciences, Blueprint Medicine, Telios pharma, Protagonist Therapeutics, Galapagos, TScan Therapeutics, Karyopharm, MorphoSys, and In8Bio. B.D.S. reports consultancy for Servier. A.A. recieved funding from Astellas. The remaining authors declare no competing financial interests.

Correspondence: Alexander Ambinder, The Sidney Kimmel Comprehensive Cancer Center, 1650 Orleans St, CRB1 Rm 2M88, Baltimore, MD 21287; email: aambind1@jhmi.edu.