Acute myeloid leukemia (AML) is the most common acute leukemia in adults, with a median age at diagnosis of 69 years and a five-year overall survival (OS) of only 23%—dropping to 10% in patients aged ≥65 [1]. For patients with an estimated relapse risk over 35–40%, consolidation with allogeneic stem cell transplant (alloHCT) is standard of care [2]. Although there is no upper limit of age cutoff, historically, older adults have often been excluded from alloHCT due to concerns regarding treatment-related toxicity (TRM) and non-relapse mortality (NRM). While multiple retrospective and matched analyses support improved progression-free survival and OS with alloHCT [3,4,5,6], other studies have raised concerns about its benefit in older patients, citing limited OS gains and negative impacts on quality of life [7,8,9].

Over the past decade, the use of alloHCT in North America has steadily grown. This increase reflects greater donor availability, advances in transplant techniques, and wider eligibility across different age and risk groups. Earlier studies have highlighted significant underutilization of alloHCT in North America, especially in Canada [10,11,12]. From 2012 to 2015, the alloHCT rate in Ontario was 14.1 per million annually, compared to 26.7 in New York [11]. A Canada-wide analysis reported that only 27% of AML patients received alloHCT, with rates declining sharply after age 50 and particularly low in Ontario [12]. In response to patients dying while waiting for alloHCT, Ontario heavily invested in expanding transplant infrastructure starting in 2016 [13]. This expansion provided a unique opportunity to assess how increased transplant access affected survival outcomes in AML, especially among older adults.

We conducted a retrospective population-based cohort analysis using administrative health databases held at ICES in Ontario to examine alloHCT utilization and impact on survival among adults with AML treated between 2010 and 2022 (as described in Supplementary Methods; Tables S1-S4). These databases were linked using unique encoded identifiers and analyzed at ICES. All methods were performed in accordance with the relevant guidelines and regulations. Approval was obtained from our local ethics committee (HIREB#18438). We identified 8672 patients diagnosed with AML in Ontario between 2010 and 2022. After excluding pediatric cases, those without complete demographic data, and patients not treated within one year of diagnosis or likely misclassified, our final cohort included 5213 adults who received active AML therapy (Fig. S1). The median age at diagnosis was 67 years (IQR 57–75) and 57% were male (Tables S5; S6). Over the study period, 4120 (79%) patients received chemotherapy-only and 1,093 (21%) patients underwent alloHCT. The majority (62%) received low-intensity treatment, whereas 38% received intensive chemotherapy.

AlloHCT utilization increased from 12% in 2010 to 25% in 2022, with the greatest increase occurring after 2017 (Fig. S2; Table S7). Utilization increased across all age groups: 38% to 66% among patients aged 18–39, 20% to 54% in those 40–59, 9% to 36% for 60–69, and from 0% to 5% in patients ≥70. The time from diagnosis to transplant shortened, from a median of 6.5 months (IQR 4.7–8.7) in 2010–2014 to 4.6 months (IQR 3.4–6.0) in 2015–2018 and beyond.

Multivariate cox regression analysis identified age, comorbidity burden, rural residence, and lower income as significant predictors of reduced transplant utilization (Table 1). Compared to patients aged 40–59, the odds of receiving alloHCT were lower for those aged 60–69 (OR 0.5; 95% CI: 0.4–0.6) and markedly lower for those ≥70 (OR 0.04; 95% CI: 0.03–0.06). Higher Charlson Comorbidity Index (CCI ≥ 2) and rural status (OR 0.5; 95% CI: 0.5–0.8) also reduced the likelihood of alloHCT. Patients from a lower-income quintile were less likely to receive alloHCT (OR 0.6; 95% CI: 0.4–0.8). The Ontario Marginalization Index (ON-Marg), a multidimensional index of marginalization that includes dimensions such as racialized and newcomer populations, as well as age and labour force characteristics, was predictive of transplant when analyzed on a univariate basis, but did not independently predict access after adjusting for income and geography. Receipt of intensive chemotherapy (OR 2.0; 95% CI: 1.7–2.3) and more recent diagnosis (OR 3.3 for 2019–2022 vs. 2010–2014) were associated with increased transplant utilization. When we stratified by transplant era (2010–2016 vs. 2017–2022), we found that older age, comorbidity burden, and rural residence were consistently associated with lower transplant receipt, while income-related disparities evident in 2010–2016 were no longer significant in 2017–2022 (Table 1).

Table 1 Factors influencing receipt of hematopoietic stem cell transplant in patients with acute myeloid leukemia.
Full size table

To evaluate survival outcomes, we performed a 150-day landmark analysis, excluding patients who died and analyzing patients based on the treatment received until the landmark date. The resulting cohort (N = 3636) included 561 patients who received alloHCT and 3,075 who received chemotherapy-only. Median OS was 1.3 years for the total cohort, with 6.8 years for alloHCT recipients and 1.2 years for chemotherapy-only patients (P < 0.0001; Fig. 1). Two-year OS was 62% with alloHCT versus 39% with chemotherapy-only. AlloHCT was associated with improved OS regardless of chemotherapy intensity (Fig. 1B, C). For those treated with intensive chemotherapy, median OS improved from 2.3 years to not reached (NR) with transplant. Similarly, among patients receiving low-intensity therapy, median OS improved from 0.9 years to 3.2 years with transplant.

Fig. 1: Overall survival of patients with acute myeloid leukemia based on receipt of allogeneic hematopoietic stem cell transplant.
figure 1

This figure demonstrates the overall survival (OS) in years for the total cohort of patients who received either an allogeneic stem cell transplant (alloHCT) or chemotherapy-only (A) and when further stratified according to intensive (B) or non-intensive (C) chemotherapy. An adjusted multivariate analysis was done to assess the factors that affect OS in the total cohort (D), those who received alloHCT (E), and those who received chemotherapy only (F).

Full size image

We then evaluated whether the benefit of alloHCT persisted when stratifying by age (Fig. S3). In the 18–39 cohort, two-year OS was similar for chemotherapy-only and alloHCT groups (66% vs. 68%; P = 0.33), while in patients aged 40–59 and 60–69, two-year OS improved from 54% to 64% (P = 0.006) and from 38% to 60% (P < 0.0001), respectively. In the ≥70 group, the difference was 23% versus 41% (P = 0.04); the borderline significance in this group likely results from limited sample size (N = 55) and a relatively shorter observation period in the transplant population of this age. Notably, even within this group, a plateau in survival was observed.

Multivariable regression confirmed that alloHCT independently improved OS (HR 0.7), even when accounting for age, sex, comorbidity, and chemotherapy intensity (Fig. 1; Table S8). Older age (HR 1.5 for 60–69; HR 2.1 for ≥70), male sex (HR 1.1), and comorbidity burden (HR 1.4 for CCI ≥ 2) were associated with worse OS, while intensive chemotherapy (HR 0.7) and younger age (HR 0.7) were protective. Notably, year of diagnosis and socioeconomic factors (income quintline; rural status) did not significantly impact survival. The benefit of alloHCT remained consistent across sensitivity analyses using alternative landmark times (90–270 days) and in a time-varying covariate analysis, which further reduced potential selection bias (Tables S9; S10). Within the alloHCT group, more recent transplant years were associated with better outcomes (HR 0.5 for 2019–2022 vs. 2010–2014), suggesting improvements in transplant-related mortality and morbidity over time (Table S11). In contrast, OS in the chemotherapy-only group remained stable over time (HR 1.03 for 2019–2022 vs. 2010–2014; Table S12).

Expanding on this, we examined trends in survival over time in patients ≤65, given the lower rates of transplant above this age (Fig. S4). As a whole, this patient cohort had improved median OS from 1.4 years in 2010–2014 to 2.5 years in 2019–2022. For alloHCT recipients, median OS increased from 2.4 years to NR with two-year OS from 56% to 69%, whereas chemotherapy-only outcomes remained unchanged, with a two-year OS of 35% to 37%.

Our results confirm that alloHCT utilization in Ontario has expanded substantially, particularly after 2017. While this reflects, in part, system-level changes in Ontario, it is consistent with trends observed in other legislations [10, 14]. We found that older patients, those with comorbidities, or those with rural residence and lower income, remained less likely to undergo transplant, which aligns with prior studies [3, 15, 16], and is consistent with data from the U.S. showing lower alloHCT access in marginalized and lower-income populations [14, 16, 17]. Encouragingly, some socioeconomic disparities appeared to be narrowing with broader transplant access in recent years. The growing availability of donors, improved supportive care, and reduced-toxicity protocols have likely enabled more patients to undergo transplant successfully. However, given that rural residence continues to limit access, this highlights the need for targeted strategies to reduce these barriers. Despite some progress, alloHCT remains underutilized—especially in older adults. In our study, transplant rates in patients ≥70 years remained below 5% even in 2022, consistent with national data showing steep declines in alloHCT use beyond age 60 [10, 15, 18].

Across age groups and chemotherapy intensities, alloHCT significantly improved OS. Among patients 40–69 years old, a group enriched for intermediate- and high-risk AML, transplant provided the most striking gains. The benefit persisted even in multivariable and time-varying models and was especially strong in recent years, where the relative risk of death dropped by 47% in 2019–2022 compared to earlier time periods. This temporal trend aligns with broader improvements in transplant protocols, including use of post-transplant cyclophosphamide, better donor selection, and more efficient conditioning regimens. The apparent lack of benefit in the 18–39 cohort likely reflects differences in underlying disease biology and treatment intent, as younger patients with favourable-risk AML are appropriately treated with chemotherapy alone. Our study lacked molecular and cytogenetic data to refine this analysis, highlighting a limitation of using administrative datasets.

In conclusion, our study demonstrates that alloHCT is associated with survival gains across age groups, regardless of induction intensity, which have been increasing in recent years. The positive trajectory in survival after alloHSCT further supports continued efforts to broaden eligibility and reduce disparities. In the current era of growing transplant feasibility and efficacy, ensuring equitable access to this curative therapy remains a critical priority for AML care.