Determination of fitness and therapeutic options in older patients with acute myeloid leukemia

Abstract Treatment of older patients with AML remains challenging. Although age, performance status, and comorbidities are commonly employed to determine fitness for intensive treatment, several studies have demonstrated improved outcomes with treatment in older and classically unfit patients, highlighting the importance of other disease‐related and patient‐related factors that have prognostic value for treatment outcome in AML. However, consistent and objective assessments for fitness are lacking. Multi‐parameter geriatric assessment tools offer more comprehensive evaluation, but are limited by the required resources and lack of standardization and consensus regarding prognostic value. These assessments are particularly important considering the emerging new AML therapies that represent a spectrum of intensities. Patients should therefore be evaluated holistically for fitness to receive a specific treatment, with the aim of providing individualized care, and such definitions of fitness should also consistently be applied to clinical trials. This review will examine evolving criteria for the determination of fitness among AML patients and discuss treatment options for older and/or unfit patients with AML.

lacking. Multi-parameter geriatric assessment tools offer more comprehensive evaluation, but are limited by the required resources and lack of standardization and consensus regarding prognostic value. These assessments are particularly important considering the emerging new AML therapies that represent a spectrum of intensities. Patients should therefore be evaluated holistically for fitness to receive a specific treatment, with the aim of providing individualized care, and such definitions of fitness should also consistently be applied to clinical trials. This review will examine evolving criteria for the determination of fitness among AML patients and discuss treatment options for older and/or unfit patients with AML.

| INTRODUCTION
Acute myeloid leukemia (AML) is the most common adult acute leukemia, accounting for 80% of cases, with an incidence estimated at 3-5 cases per 100 000 persons in the United States. 1 AML is primarily a disease of the elderly, with a median age of 68 years at diagnosis. 2 Historically, the 5-year overall survival (OS) rate for AML was 29%, but declined to 8% among patients aged ≥65 years. 2,3 Older age was historically considered a poor prognostic factor and also the main criterion for determining whether an AML patient could receive intensive therapy. 4 Outcomes among older AML patients treated with conventional induction chemotherapy vary widely, clouding the definition of fitness. The MD Anderson Cancer Center evaluated 446 patients aged ≥70 years who received intensive chemotherapy for AML and found 54% had unfavorable cytogenetics and 31% had a prior malignancy. 5 Response to intensive chemotherapy included complete remission (CR) in 45%, with an 8-week mortality of 36%. Median OS was <6 months for all patients and 13.8 months for those achieving CR. The authors concluded that, despite reasonable CR rates, the OS and 8-week mortality rates did not support intensive chemotherapy for patients aged ≥70 years; however, the study did not include a comparison with non-intensive regimens. 5 Similarly, Vey et al reported a CR rate of 43% among AML patients who were aged ≥75 years, with an early mortality rate of nearly 20% and median OS of 9 months. 6 Results from the Swedish Acute Leukemia Registry update in 2011, which included 998 AML patients aged 70-79 years, indicated lower 8-week mortality rates with intensive chemotherapy vs palliative treatment, but this population had a lower proportion of patients with high-risk cytogenetics (proportion of high-risk cytogenetics in de novo AML: 30%; proportion of high-risk cytogenetics in secondary AML: 40%). 7 In a recent, broader analysis from the Swedish AML Registry, which included 6994 AML patients diagnosed between 1997-2016, OS improved significantly over time in those aged 50-75 years, whereas no improvement was seen in younger (<50 years) or older (>75 years) patients. 8 Overall, 60% of patients received intensive therapy, and patients not receiving intensive therapy had higher early death rates irrespective of age.
As prognostically relevant as age is, there are other factors associated with patient outcomes. Some of these disease-related and patient-related characteristics include cytogenetic risk, history of myelodysplastic syndrome (MDS), and comorbidities. Thus, the therapeutic paradigm for older AML patients has been shifting in recent years, 4,9 with clinicians recognizing the need to assess patients holistically for appropriateness to receive a specific therapy/regimen. However, studies evaluating AML patients vary in design and often have a vague and subjective characterization of fitness for therapy. For example, a recently published study of venetoclax plus a hypomethylating agent (HMA) included patients aged ≥65 years who were ineligible for standard induction chemotherapy, loosely defined as having "various comorbidities, such as age >75 years, cardiac disease or prior anthracycline use, secondary AML, or high probability of treatment-related mortality." 10 These criteria were not defined by objective measures, such as a specific New York Heart Association (NYHA) functional class, ejection fraction, maximum dose of anthracycline, or Chemotherapy Risk Assessment Scale for High-Age Patients (CRASH) scores. Other studies have been more precise in defining ineligibility to receive intensive therapy, such as the randomized study of glasdegib plus low-dose cytarabine (LDAC) vs LDAC alone that specified the following criteria: age ≥75 years, serum creatinine >1.3 mg/dL, severe cardiac disease (left ejection fraction <45%), or Eastern Cooperative Oncology Group (ECOG) performance status (PS) of 2. 11 The goal of this review is to examine and discuss evolving criteria for the determination of fitness among AML patients and evaluate treatment options for older and/or unfit adults with AML. Although there has been less emphasis on age in recent years as a sole determinant of fitness, a retrospective analysis of 968 patients enrolled across five Southwest Oncology Group trials identified frequent correlation between age and other poorprognosis factors. 15 In this early report, published in 2006, older age was associated with a smaller proportion of patients with a PS of 0, relative to younger age. Additionally, the promortion of patients wtih favorable cytogenetics significantly decreased from 17% in younger patients to 4% in patients aged >75 years. There was also a corresponding increase in unfavorable cytogenetics and a higher proportion with multidrug resistance among older patients (57%-62% for ages ≥56 years vs 33% for ages <56 years). Patients with older age and a poor PS had a significantly higher likelihood of 30-day mortality. 15 More recently, Lazarevic et al reported clinical and diagnostic features with a focus on patients aged ≥80 years using data from the Swedish AML registry. 16 Patients aged >85 years had slightly higher WBC counts and blood absolute blast counts, and less elevated lactate dehydrogenase (LDH) levels. Although older patients tended to undergo less morphologic subclassification and genetic evaluation in this study, complex and monosomic karyotypes were more common in this group. Secondary AML was most common in patients aged 70-80 years, but less common in patients aged ≥85 years. 16 These data suggest modest differences in clinical AML subsets across ages 70-100 years and encourage collection of molecular data in these patients, particularly in the context of emerging therapies, many of which may benefit patients with specific AML subtypes (eg, secondary AML) or molecular features. Studies from the German AML Cooperative Group further underlined the significance of molecular data collection to identify subsets of patients who will most likely benefit from intensive induction therapy. In a study by Metzeler et al in AML patients who received intensive induction therapy, the mutational spectrum in older patients (≥60 years) differed from younger patients (<60 years). 17 20 Importantly, improved supportive care, including the use of prophylactic antibiotics and antifungals in older patients with AML, has helped improve the safety of delivering intensive therapy in older patients.
It is worth noting that PS was developed mostly for evaluating patients with solid tumors. In AML, PS usually refers to function prior to onset of AML-related symptoms, as factors such as the manage- Leukemia Registry evaluated the effect of the decision to treat on outcomes. 21 Thus, PS was best in patients aged 40-44 years and declined with increasing age. As PS worsened, the proportion of patients receiving intensive therapy also declined. Thirty-day mortality rates were dependent on age and PS, but older patients with good PS had low early death rates and patients with poor PS had increased early mortality across all ages. Early death was reported for 36% of patients aged 76-89 years with a PS of 3-4 who were given intensive therapy vs 52% of patients who received palliation only (P = .023). While the early mortality rate was higher in patients with impaired PS across age groups, there were some long-term survivors, suggesting intensive therapy may be of benefit for selected patients. 21 Among 57 patients with PS of 3-4 treated with intensive cytarabine-based therapy at MD Anderson Cancer Center, the CR rate was 25% and 8-week mortality was 77%. 5 A multivariate analysis found high 8-week mortality was associated with ECOG PS of 2-4, among other factors.
Together, these studies suggest intensive therapy is superior to lowintensity therapy, and the latter is superior to supportive care alone in older AML patients, and most patients should be considered for treatment.
So, PS is highly linked to age and comorbidities but insufficient alone to accurately assess fitness. Varying degrees of comorbidity, some of which may be optimally managed, in older AML patients highlight the need for better strategies to assess fitness. Thus, more sensitive approaches are needed to better identify candidates for intensive therapy. 4,19

| Comorbidities/medical history
The likelihood of comorbidities increases with age in AML patients and can affect treatment administration and toxicity. 19 heart valve disease (except mitral valve prolapse), and a prior solid tumor. 22 Among 177 AML patients aged >60 years and treated with induction chemotherapy, those with an HCT-CI score ≥3 had an early mortality rate of 29% vs 3% and 11% in patients with scores of 0 and 1-2, respectively (P <.001). 23 However, aging and frailty related to aging are not entirely a function of comorbidities. Patients with several well-managed comorbidities may be reasonably fit and vice versa. Thus, assessment of comorbidities may help better define fitness for intensive therapy, but still does not fully represent the possible outcome and tolerability of treatment for AML patients. 4,19

| Multi-parameter assessment tools
In response to the somewhat overlapping, yet incomplete, influences of age, PS, and comorbidities to define fitness in AML patients, use of geriatric assessment tools and multi-parameter assessments has been considered to provide additional prognostic information. Geriatric assessment tools evaluate multiple health domains to more globally assess patient fitness and may assist in refining risk stratification and personalizing therapy for older AML patients 25 ; however, there is no consensus, yet on the ideal domains to include and how best to incorporate different factors. Table 1 provides an overview of domains considered in geriatric assessments that have been used in AML patients, and Table 2 summarizes multi-parameter assessment tools developed from clinical trials.
A prospective cohort study evaluated the predictive value of geriatric assessments, including measures of cognitive function, depressive symptoms, distress, physical function, and clinical characteristics, for OS in patients with newly diagnosed AML who were aged ≥60 years and received intensive therapy. 26 The OS was associated with cytogenetic risk group, prior MDS, and baseline hemoglobin level, but not with age or ECOG PS. Among geriatric assessment measures, poor cognitive function (Modified Mini-Mental State score <77) and low physical performance (Short Physical Performance Battery score <9) were associated with poor OS and increased the predictive power of the more standard clinical measures by 60%.
Another study examined geriatric and quality-of-life assessments in 195 AML and MDS patients aged ≥60 years. 27 The study measured patient-related factors, including PS, activities of daily living (ADLs), comorbidities, and disease characteristics ( Table 2) In the previously mentioned 2006 study by Kantarjian et al, 20 a prognostic model was built to predict outcomes in older AML patients using various patient-related and disease-related factors to categorize patients into risk groups. Both OS and CR rates were higher in the favorable-risk and intermediate-risk groups relative to the unfavorable-risk group. In a prospective trial of 909 AML patients aged >60 years, prognostic factors that included mutational status were investigated for predictive value on clinical outcomes. 28 A multivariate analysis determined age, karyotype, NPM1 mutation status, WBC count, LDH level, and CD34 expression were independent prognostic indicators of OS, and these factors were assigned relative point values ( Table 2). Based on the total points and a patient's cytogenetic risk, four prognostic profiles were determined: favorable-risk cytogenetics, intermediate-risk cytogenetics with favorable-risk features (score ≤3), intermediate-risk cytogenetics with adverse-risk features (score >3), and high-risk cytogenetics. The OS for these groups was 40%, 30%, 11%, and 3%, respectively.
Results of cytogenetic analysis to determine risk may not be readily available for AML patients who require immediate treatment. Thus, a web-based application was used to calculate risk scores from standard clinical and laboratory variables, such as body temperature, age, hematologic measures, LDH level, and AML subtype, with or without knowledge of cytogenetic and molecular risk. 29 These variables were closely and independently associated with CR and early death, and may assist in making treatment decisions for these patients.
The Geriatric Assessment in Hematology (GAH) scale was designed as a brief evaluation of older patients with hematologic malignancies and consists of eight dimensions of performance, mental status, and health status ( Table 2) that contribute to a score of 0-8. It was validated in 349 patients aged ≥65 years with newly diagnosed hematologic malignancies, including AML. 30 The GAH scale correlated with ECOG PS and KPS, except in the comorbidities domain. Increasing GAH score groups of ≤1, 2-6, and >6 were predictive of survival (P <.001). 31 An abridged geriatric assessment was compared with KPS and the Physical Performance Test in 100 cancer patients aged >70 years, including 14% with hematologic malignancies. 32 The assessment included some domains consistent with GAH (ADLs, affective status, nutritional status, and polypharmacy), but also considered risk for falls, hearing, vision, urinary incontinence, and pain. Note, OS was associated with the abridged geriatric assessment, but not KPS or the Physical Performance Test score.

| THERAPEUTIC APPROACHES IN OLDER AND/OR UNFIT PATIENT POPULATIONS
Several new therapies have been approved for the treatment of adult AML patients in the past few years, substantially changing the treatment paradigm. Although therapies were traditionally classified as  • Get-up and Go Test to neutrophil and platelet recovery may be faster with the traditional 7 + 3 regimen than with decitabine and venetoclax. 33,34 Some newer therapies are specifically indicated for use in older and/or unfit patients, but others may also be appropriate for some older patients depending on their overall fitness, thereby expanding treatment options for older patients while avoiding the toxicities posed by conventional chemotherapy. 25 All patients should thus be assessed for fitness to receive a given therapy or regimen, rather than deemed "fit" or "unfit" overall. Other factors, such as the patient's goals, also need to be included in the treatment decision. The suitability of the treatment administration setting should also be considered; this includes, for example, the ability to administer transfusions for a longer time and to manage septic episodes for therapies with more prolonged myelosuppression. Clinical studies of therapies appropriate for some older AML patients are reviewed below and summarized in Table 3.
In the clinical management of newly diagnosed AML, immediate treatment start is typically recommended due to poor prognosis.
However, more recently, a real-world analysis from the German Study Alliance Leukemia-Acute Myeloid Leukemia registry arrived at a different conclusion. 35

| CPX-351
The drug CPX-351 (Vyxeos; Jazz Pharmaceuticals, Inc., Palo Alto, CA, USA) is a liposomal co-encapsulation of cytarabine and daunorubicin that delivers a synergistic 5:1 molar drug ratio preferentially to leukemia cells vs normal cells in the bone marrow. 62  Among patients who achieved CR + CRi, median time to recovery of neutrophil and platelet counts was longer with CPX-351 vs 7 + 3.
Early mortality rates at 30 and 60 days were lower with CPX-351 vs 7 + 3, although the difference did not reach statistical significance. 33 Although CPX-351 has primarily been studied as intensive chemotherapy, a phase 2 study evaluated lower-intensity doses of CPX-351 in adults with newly diagnosed AML who were considered less fit and had a composite treatment-related mortality score of >13.1 (corresponding to a >13.1% probability of death within 28 days of receiving intensive chemotherapy). 64 Among patients who received CPX-351 32 units/m 2 /dose (n = 38) and 64 units/ m 2 /dose (n = 10), respectively, the CR + CRi rates were 29% and 20%, 12-month OS rates were 17% and 20%, and early mortality rates within 28 days were 29% and 40%. 64 An ongoing clinical trial is also evaluating lower-intensity CPX-351 plus venetoclax in adults considered unfit for intensive therapy (ClinicalTrials.gov Identifier NCT04038437).

| PRACTICAL GUIDANCE AND ADDITIONAL CONSIDERATIONS FOR IDENTIFYING APPROPRIATE THERAPIES FOR OLDER PATIENTS
A significant proportion of older AML patients are not offered chemotherapy because of the perceived lack of efficacy and toxicity of intensive chemotherapy. As a result, prognosis in this patient population remains poor. 12,25,65 While the need for a more holistic approach to determining specific therapies and regimens is recognized, proper objective fitness assessments have traditionally been lacking.
Results of multi-parameter assessments and prognostic models have led to several important findings: (a) disease-related factors of unfavorable cytogenetics and multidrug resistance increase with age and are consistently associated with poorer outcomes; (b) prior MDS, percentage of blasts, WBC count, LDH level, and cytogenetic and molecular aberrations may play a role in defining prognosis; (c) increasing age, poor PS, and comorbidities correlate with poorer AML outcomes, but are insufficient to fully assess patient fitness on their own; (d) patient-related factors, including physical status/frailty, cognitive status, psychologic status, nutritional status, functionality (instrumental ADL), and ability to perform ADLs, may predict OS. These assessments, however, can be time consuming, and additional evidence of their predictive ability for individual therapies/regimens are needed, specifically in older AML patients.
Each patient should be assessed for fitness to receive particular therapies/regimens and other considerations that might inform treatment decisions (eg, molecular targets, patient's treatment goals, logistics), with the aim of providing individualized care. Newer treatments with different safety profiles are generally better tolerated than conventional chemotherapy and may be preferable options for older and unfit patients. In many instances, these drugs have been specifically investigated and/or approved for segments of this patient population.

| CONCLUSIONS
In addition to age, PS, cytogenetic risk category, and AML subtype