According to the Surveillance, Epidemiology, and End Results (SEER) Program, the 5-year relative survival for acute myeloid leukemia (AML) is 33.4%, based on patients diagnosed between 2016 and 2022. In 2026, an estimated 22,720 new AML cases and 11,500 deaths are expected in the United States.
That figure describes AML across a population but says little about an individual patient. AML comprises biologically distinct subtypes with very different treatment sensitivity, relapse risk, and potential for cure. Acute promyelocytic leukemia treated with modern differentiation therapy can have 5-year survival above 90%, some genetically adverse subtypes, by contrast, still have median survival measured in months. Between these extremes are leukemias whose prognosis changes considerably with treatment and molecular evolution.
Prognosis in modern AML is not a survival percentage assigned once at diagnosis. Genetics sets the initial risk, treatment determines which targets are exploited, MRD shows the depth of response, and remission or relapse revises the estimate again.
Remission, Survival and Cure Are Not the Same Endpoint
AML survival discussions frequently mix endpoints. SEER‘s 33.4% figure is relative survival: it compares survival among people diagnosed with AML with expected survival in a comparable general population. It is not a cure rate, and doesn’t mean every newly diagnosed patient has a 33.4% chance of surviving five years. It also covers a period in which FLT3, IDH1/IDH2 inhibitors, venetoclax combinations, and oral azacitidine maintenance began entering practice.
Overall survival counts survival regardless of relapse. Relapse-free survival starts after remission and measures time without recurrence or death. Event-free survival can include failure to achieve remission, relapse, and death. Relative survival, as reported by SEER, compares observed survival with expected survival in the general population. None is identical to a formal cure rate.
The same caution applies to medians: a median overall survival of 12 months means half the study population had died by about that time, not that every patient survives about a year, and it says nothing about the shape of the long-term curve.
You can also read:
Age, AML Biology, and Survival
Age is among the strongest clinical correlates of AML survival. Older patients are more likely to have comorbidities, impaired organ reserve, and frailty that limit treatment intensity or transplantation, and their AML is biologically different: myelodysplasia-related mutations, antecedent clonal hematopoiesis or myelodysplastic syndrome, TP53 abnormalities, and adverse cytogenetic patterns all become more common with age.
A large validation of the European LeukemiaNet (ELN) 2022 classification in 1,570 intensively treated adults shows both effects. Among patients aged 60 or younger, 5-year overall survival was 63% for favorable-, 42% for intermediate-, and 24% for adverse-risk AML. Among patients older than 60, it was 22%, 15%, and 3.5%.
Part of the survival gap between younger and older patients therefore reflects a different distribution of AML subtypes, not fitness alone. An older patient with treatment-sensitive molecular disease can have a very different outlook from a same-age patient with TP53-mutated complex-karyotype AML.
ELN Risk Groups
For patients treated with intensive chemotherapy, ELN 2022 sorts AML into favorable-, intermediate-, and adverse-risk categories based principally on cytogenetic and molecular abnormalities.
In the 1,570-patient validation cohort, complete remission rates were 87.3%, 76.6%, and 49.2% for favorable-, intermediate-, and adverse-risk AML. Five-year overall survival was 53%, 32%, and 13%. An independent validation produced 5-year overall survival estimates of 54.6%, 34.2%, and 14.8%. The categories still contain substantial heterogeneity.
APL: When the Driver Is Exceptionally Druggable
APL is driven by PML::RARA, usually generated by t(15;17). The fusion disrupts retinoic-acid receptor signaling and blocks normal promyelocytic differentiation, and all-trans retinoic acid (ATRA) and arsenic trioxide (ATO) target that differentiation block rather than relying on conventional cytotoxicity.
Long-term follow-up of 146 patients treated with ATRA plus ATO, with gemtuzumab ozogamicin in selected higher-risk patients, showed 5-year overall survival of 93.1%, event-free survival of 92.4%, and disease-free survival of 93.6%. The main remaining vulnerability is early mortality from the characteristic coagulopathy and hemorrhage, which makes rapid recognition and immediate treatment crucial.
You can also read:
Core-Binding Factor AML: Favorable Does Not Mean Low Risk
Core-binding factor (CBF) AML includes RUNX1::RUNX1T1, usually t(8;21), and CBFB::MYH11, usually inv(16) or t(16;16). Both disrupt transcriptional programs required for normal hematopoietic differentiation and are classified as favorable risk.
Favorable describes an average treatment response, not uniform curability. Relapse still occurs, and molecular response after treatment separates patients far more effectively than the fusion alone: a meta-analysis of 13 studies found MRD-negative CBF-AML was associated with substantially better recurrence-free and overall survival than persistent molecular disease. KIT and FLT3 alterations have been associated with poorer outcomes in contemporary analyses, though their effects depend on fusion subtype, treatment, and MRD kinetics.
NPM1-Mutated AML: When Response Becomes More Informative Than Baseline Risk
NPM1 mutation establishes a distinct leukemia biology and, in the appropriate genetic context, places a patient in favorable-risk disease. Mutant NPM1 transcripts can also be tracked, so residual leukemia can be measured far below the threshold of conventional morphology.
In AML17 and AML19, post-induction NPM1 MRD strongly separated subsequent outcomes, and it identified patients for whom transplantation had very different implications. Among MRD-positive patients, allogeneic transplantation in first remission was associated with 3-year overall survival of 61% versus 24% without transplantation. Among MRD-negative patients, 3-year survival was 79% with transplantation and 82% without, with no demonstrated survival benefit from transplant in that analysis.
Even among patients with concurrent FLT3-ITD, transplant benefit remained concentrated in the MRD-positive population: 3-year survival was 45% versus 18% with and without transplant among MRD-positive patients, compared with 83% versus 76% among those who were MRD-negative. NPM1 mutation has also become a direct therapeutic target through menin inhibition. In 2025, revumenib and ziftomenib received FDA approval for relapsed or refractory disease.
CEBPA-Mutated AML: The Importance of Mutation Location
The older concept that biallelic CEBPA mutation marks favorable AML has been replaced by recognition that in-frame mutations affecting the basic leucine zipper (bZIP) region identify the favorable biological group, whether the alteration is monoallelic or biallelic.
In an AML Study Group analysis of 528 patients with CEBPA-mutated AML, 5-year overall survival across the cohort was 55%. Patients with bZIP in-frame insertion/deletion mutations had significantly better event-free and overall survival than patients with other CEBPA mutation types.
FLT3-ITD Prognosis in the Targeted Therapy Era
FLT3-ITD constitutively activates proliferative and survival signaling and historically marked AML with increased relapse risk. FLT3 is also therapeutically targetable, so the marker’s prognostic effect and its value for treatment selection are separate questions.
ELN 2022 no longer uses the previous FLT3-ITD allelic-ratio threshold. In the absence of another risk-determining abnormality, FLT3-ITD AML is placed in the intermediate group regardless of allelic ratio or NPM1 status, a change that reflects the limits of using a historical prognostic marker independently of modern FLT3-directed therapy.
Midostaurin brought FLT3 inhibition into frontline intensive treatment, quizartinib further changed frontline therapy for FLT3-ITD AML, and gilteritinib provides targeted treatment in relapsed or refractory FLT3-mutated disease. “FLT3 mutation means poor survival” is therefore an incomplete statement: prognosis now depends on which alteration is present, whether FLT3-directed therapy is used, depth of molecular response, co-mutations, and whether transplantation is performed.
IDH-Mutated AML: An Actionable Target
IDH1 and IDH2 mutations alter cellular metabolism through production of the oncometabolite 2-hydroxyglutarate, disrupting epigenetic regulation and differentiation. Their prognostic effect depends on coexisting genetics and treatment, but their therapeutic importance has grown because mutant IDH can be directly inhibited.
In IDH1-mutated AML unsuitable for intensive chemotherapy, azacitidine plus ivosidenib produced median overall survival of 24.0 months versus 7.9 months with azacitidine plus placebo, with longer follow-up, median survival in the ivosidenib arm reached 29.3 months. In IDH2-mutated AML, enasidenib plus azacitidine produced an overall response rate of 74% versus 36% with azacitidine alone, but with similar median overall survival.
KMT2A-Rearranged AML: The Impact of Coexisting Mutations
ELN 2022 classifies t(9;11)/KMT2A::MLLT3 as intermediate risk and most other KMT2A rearrangements as adverse. KMT2A fusions drive an abnormal HOX/MEIS transcriptional program, but fusion partner alone does not fully explain outcome.
A HARMONY analysis of 205 adults with KMT2A-rearranged AML found a median overall survival of 1.4 years, with no significant survival difference between t(9;11) and other KMT2A rearrangements. Instead, age above 60 years, secondary AML, DNMT3A mutation, and KRAS mutation independently predicted poorer survival among intensively treated patients.
Among younger patients with de novo disease, KRAS or TP53 mutations reduced the complete remission rate from 86% to 50% and median overall survival from 30 to 7 months. Allogeneic transplantation in first remission was associated with improved survival.
Menin inhibition has also introduced a targeted option for KMT2A-rearranged AML. In the AUGMENT-101 trial, revumenib produced an overall response rate of 63.2% and a complete remission or complete remission with partial hematologic recovery rate of 22.8% in heavily pretreated patients with relapsed or refractory KMT2A-rearranged acute leukemia.
You can also read:
DEK::NUP214 AML: Can Treatment Strategy Overcome Part of Adverse Genetic Risk?
AML with t(6;9)/DEK::NUP214 is rare, frequently coexists with FLT3-ITD, and is classified as adverse risk. Historical chemotherapy outcomes were marked by substantial relapse, while contemporary transplantation data show improved survival in selected patients.
A 2026 European Society for Blood and Marrow Transplantation analysis included 544 patients with t(6;9) AML who underwent allogeneic transplantation. At two years, overall survival was 65.7%. Among 431 patients transplanted in first complete remission, overall survival was 71.7%, leukemia-free survival 65.8%, and relapse incidence 18.2%. These numbers describe a selected population.
MECOM-Rearranged AML: Why Outcomes Remain Poor
AML with MECOM rearrangement, including inv(3)(q21.3q26.2), t(3;3)(q21.3;q26.2), and other 3q26 rearrangements, is rare and remains one of the most difficult AML subtypes. MECOM/EVI1 dysregulation alters transcriptional control, stem-cell self-renewal, and differentiation, but no targeted strategy has yet transformed outcomes.
Published survival varies considerably by cohort and by transplantation. In a 2025 MD Anderson analysis, patients with variant MECOM rearrangements who underwent allogeneic transplantation had a 3-year overall survival of 71%, compared with 11% among those who did not. This comparison was based on a small, selected subgroup.
Myelodysplasia-Related AML: One Risk Category, Several Diseases
Modern AML classification has moved away from identifying “secondary AML” solely by a previous diagnosis of myelodysplastic syndrome. Mutations in genes such as ASXL1, BCOR, EZH2, RUNX1, SF3B1, SRSF2, STAG2, U2AF1, and ZRSR2 can mark myelodysplasia-related biology even when no antecedent MDS was recognized.
ELN 2022 assigns these abnormalities to adverse risk in the appropriate genetic context, but they don’t share a single prognosis. In the large ELN validation, some subgroups had 5-year survival closer to 30%, whereas U2AF1-mutated disease had markedly poorer survival in that cohort. Treatment context changes the picture: under less-intensive therapy, AML with myelodysplasia-related mutations without FLT3-ITD, NRAS, KRAS, or TP53 alterations had a reported median survival of approximately 23 months, compared with ~13 months when activating signaling mutations were present.
Complex and Monosomal Karyotype
Molecular sequencing has not made conventional cytogenetics obsolete. Complex and monosomal karyotypes reflect broad chromosomal instability, remain adverse features, and frequently overlap with TP53-altered disease.
That overlap complicates interpretation of abnormalities such as −5/del(5q), −7, or 17p, whose prognostic meaning can’t always be separated from the surrounding genomic architecture. An isolated chromosomal abnormality, a highly complex karyotype, and TP53-mutated complex-karyotype AML are not biologically interchangeable, so cytogenetics and sequencing need to be interpreted together.
TP53-Mutated AML: Where Modern Therapy Still Falls Short
TP53 normally coordinates DNA-damage responses, cell-cycle arrest, and apoptosis. Its disruption lets genetically damaged cells survive and is strongly associated with complex cytogenetics, treatment resistance, and rapid relapse.
The adverse effect persists across approaches. In the ELN 2022 validation cohort, TP53-mutated AML had a 5-year overall survival of only 3.3%. In analyses underlying ELN 2024, TP53-mutated AML treated with HMA-based approaches had median overall survival of approximately 5-8 months.
You can also read:
Prognostic vs Predictive Biomarkers and Treatment Context
Subtype comparisons show two different functions of molecular testing. A prognostic biomarker describes the expected disease course; a predictive biomarker identifies patients more or less likely to benefit from a particular treatment. TP53 remains mainly a powerful adverse prognostic marker, when PML::RARA predicts extraordinary sensitivity to differentiation therapy.
ELN 2017 and 2022 were developed primarily from patients receiving intensive chemotherapy and were not designed to predict outcomes in older or less-fit patients treated with HMA-based regimens. When applied to HMA/venetoclax-treated populations, their prognostic discrimination was suboptimal.
ELN therefore proposed a separate 2024 genetic-risk classification for patients receiving less-intensive therapy. The model reorganizes risk according to how specific genetic profiles behave under these treatments. Survival estimates for rare AML subtypes often come from retrospective registries, transplant cohorts, or small institutional studies conducted across different treatment periods.
MRD Makes Prognosis Dynamic
Two patients can begin with the same favorable-risk subtype and emerge with very different relapse probabilities. Morphologic complete remission rests on conventional disease assessment and hematologic recovery, MRD-negative remission reaches a deeper level of detection. Neither proves every relapse-capable cell is gone, but relapse probability differs substantially with response depth.
Not every mutation works for this purpose. Mutations associated with clonal hematopoiesis can persist after successful AML therapy without representing residual leukemia the way NPM1 or a leukemia-specific fusion does, so MRD interpretation depends on what is measured, how sensitively, and when.
You can also read:
Stem Cell Transplant Does Not Have Uniform Survival Benefit
Allogeneic transplantation is generally recommended in CR1 for adverse-risk AML and considered for intermediate-risk disease. It reduces relapse through conditioning and a donor-derived graft-versus-leukemia effect, but also introduces non-relapse mortality and substantial acute and chronic toxicity. Whether it improves survival depends on the relapse risk without transplant set against the treatment-related risk with it.
Prognostic Factors After Relapse
AML can return because treatment doesn’t always eradicate every leukemia-producing clone. Resistant cells may survive below conventional detection, and the disease can undergo clonal evolution under therapeutic pressure, so relapsed AML may differ molecularly from the leukemia originally diagnosed.
Prognosis after relapse depends on factors unknown at diagnosis: duration of first remission, molecular profile at relapse, prior exposure to targeted therapies, previous transplantation, ability to achieve another remission, and whether a subsequent transplant or targeted treatment is feasible.
That’s why molecular profiling should be repeated at relapse: a targetable abnormality may emerge, disappear, or change in clonal dominance. Relapsed/refractory AML still has poor long-term survival overall, but a single relapse survival percentage obscures the differences between the patients.
AML Survival Statistics and Their Limitations
No single source provides the true survival rate for every patient. Population registries offer large numbers and mature follow-up but represent earlier treatment eras. Molecularly characterized trials reflect contemporary therapy and detailed genomics but include selected patients with shorter follow-up.
Transplant registries can address rare high-risk subtypes, but only among patients who reached transplantation. SEER suits population-level questions, ELN validation cohorts suit intensively treated, genetically stratified AML. For an individual patient, prognosis is a conditional estimate that gets recalculated as disease biology, treatment exposure, depth of response, and durability of remission become known.
You can also read:
FAQ
Can AML prognosis change after treatment begins?
Yes. Risk at diagnosis is based largely on leukemia genetics and clinical factors, but prognosis becomes more precise once treatment response is known. MRD status, duration of remission, transplantation, and molecular changes at relapse can substantially alter the original estimate.
Does favorable-risk AML mean the leukemia is likely to be cured?
Not necessarily. “Favorable risk” describes better average outcomes compared with intermediate- or adverse-risk AML; it does not guarantee cure. Patients within the same favorable-risk group can have different outcomes, particularly according to MRD response and additional molecular abnormalities.
Can AML genetics change between diagnosis and relapse?
Yes. AML can undergo clonal evolution during treatment. Some clones disappear while resistant clones survive or expand, and new molecular abnormalities may become detectable at relapse. This is one reason molecular testing is often repeated when AML returns.
Why can two patients with the same AML mutation have different outcomes?
A single mutation rarely determines prognosis by itself. Coexisting mutations, cytogenetics, age, treatment received, MRD response, transplantation, and subsequent clonal evolution can all modify outcome. Even genetically defined AML subtypes can therefore contain substantial prognostic heterogeneity.
Can an adverse AML mutation also be a useful treatment target?
Yes. A genetic alteration can carry prognostic information while simultaneously creating a treatment opportunity. FLT3 and IDH mutations are important examples because they can guide the use of targeted inhibitors. NPM1 mutations and KMT2A rearrangements can also identify leukemias susceptible to menin inhibition. Pasted markdown
Why do AML survival rates differ between studies?
Studies may measure different endpoints and include very different populations. Population registries, clinical trials, molecular cohorts, and transplant registries differ in patient selection, treatment period, follow-up, age, disease biology, and therapies received. Their survival percentages therefore should not be interpreted as directly interchangeable. Pasted markdown
Are AML survival rates improving with targeted therapies?
For several molecular subtypes, targeted therapies have improved outcomes or expanded effective treatment options. FLT3 and IDH inhibitors are established examples, while menin inhibitors have recently added targeted therapy for NPM1-mutated and KMT2A-rearranged disease. However, long-term population statistics take years to capture the full effect of newer treatments.
