The standard treatment for acute myeloid leukemia (AML) consists of induction chemotherapy, followed by consolidation chemotherapy ± allogeneic hematopoietic stem cell transplantation (HSCT). The median age of AML is 68 to 70 years. Most older patients are unfit for such intensive treatments, owing to co-morbidities and frailty. For such patients, palliative chemotherapy with hypomethylating agents (HMA) or low-dose cytarabine (LDAC) have usually been offered.

There has been increasing interest in real-world data in AML which reflect treatments and outcomes in unselected patient populations. Recent real-world registry data from Europe have reported that a significant proportion of older patients do not receive any anti-leukemic therapy at the time of diagnosis, other than best supportive case (BSC).^1-3 However, there has been a paucity of reported analysis on why such patients do not receive treatment. In order to address this, we sought to analyze treatment patterns in newly diagnosed AML patients in Alberta, Canada, to ascertain the frequency of patients not receiving anti-leukemic therapy and the factors related to this.

The University of Alberta Hospital (UAH), located in Edmonton, is the sole leukemia treatment center for the northern two-thirds of Alberta (area approximately 450 000 km²), with a catchment population of approximately 2 million. The Alberta Cancer Registry collects comprehensive lists of all cancer patients by diagnosis. From this registry, we compiled a complete list of all patients aged ≥18 years with a diagnosis of AML according to International Classification of Diseases codes over a 4-year period, from January 2013 to December 2016. Patients with acute promyelocytic leukemia (APL) were excluded. From this list, we identified all patients who were seen at UAH, or who were diagnosed at a hospital within the catchment region regardless of whether they were referred. We then cross-referenced this list with our existing leukemia database, and performed retrospective chart reviews to obtain any missing demographic, diagnostic, treatment and outcome data. For those patients who only received BSC, we then reviewed the available consultations, clinic notes and discharge summaries, in an effort to determine the reasons why anti-leukemic therapy was not administered. Prior approval was obtained from the provincial Cancer Research Ethics Board.

Treatment protocols were relatively uniform and adhered to by all treating leukemia physicians at the center. Patients deemed to be fit for intensive therapy (IT) usually received induction with a 7 + 3 regimen consisting of cytarabine plus idarubicin. Those achieving CR then received up to three cycles of consolidation chemotherapy with high-dose cytarabine; patients with intermediate or adverse risk disease by European LeukemiaNet (ELN) 2010 criteria⁴ were referred for HSCT in first CR.

Patients deemed to be unfit by the treating physician were offered non-intensive therapy (NIT), consisting of either azacitidine, LDAC or a clinical trial (usually either of these drugs combined with an investigational agent). Azacitidine was given at standard doses either at UAH or a community cancer center; LDAC could be given at home, either by self-administration or by a Home Care nurse. The no therapy group (NT) included those who received only BSC ± hydroxyurea for count control.

The OS was defined as the time from AML diagnosis to date of death or last follow-up. The patient's place of residence was defined as urban if located within the metropolitan Edmonton region (population approximately 1.3 million), and rural if located outside of this region (approximately. 0.7 million). As this was a retrospective analysis, precise data with respect to performance status were not available. To compare proportion and means for categorical and continuous variables, respectively, we used chi-square test and one-way ANOVA and reported corresponding two-sided P values. Non-parametric one-way Wilcoxon statistical test was used to compare the medians between groups. Kaplan-Meier survival curve was generated to compare the OS between treatment groups. Log-rank test with Sidak correction for multiple comparisons was reported. All analyses were performed by SAS software, SAS Institute Inc. 2013. SAS 9.4 (Cary, NC, USA) at a significance level of 0.05.

The baseline demographics and laboratory data are summarized in Table 1. Of the 316 patients, 159 (50.3%) received 1-2 cycles of induction chemotherapy. Of these, 140 (88%) achieved CR and 42% of these underwent HSCT in first CR. Seventy-eight patients (24.7%) received NIT; this consisted of azacitidine ± other agent in 49 cases, LDAC ± other agent in 25 and guadecitabine in four. In this group, 25 patients (32%) were enrolled in a clinical trial. Of the patients who received NIT, 22 (28%) achieved CR or CRi. The remaining 25% of patients were in the NT group. No patients in either the NIT or NT group underwent HSCT.

As shown in Figure S1, the proportion of patients in both the NIT and NT groups increased with increasing decade; the NT group accounted for 36% of patients age 70 to 79 and 59% of patients in the 80+ age group. The NIT and NT groups did not differ significantly with respect to age, gender, ELN or cytogenetic risk group (Table 1). There was a higher proportion of patients in the NT group presenting with high white blood count (WBC) compared to the NIT group, although this was not statistically significant.

We next analyzed the major reasons identified for patients not receiving anti-leukemic therapy. There were three categories (Table S1): (1) patients who died rapidly due to severe complications of their leukemia (eg, intracranial hemorrhage, multi-organ failure, sepsis) before any treatment could be instituted, usually within days (38% of patients); (2) patients with severe premorbid co-morbidities precluding treatment, who were mostly residents of long-term care facilities with Eastern Cooperative Oncology Group (ECOG) Performance Status 4 (19%); (3) patients who were offered treatment but declined, or for whom treatment options were limited (43%). The latter group included four patients who transformed from MDS while on azacitidine. The NT categories according to age group is shown in Figure S1. Category three comprised 16% of all patients age 60 and over, 19% of those age 70+, and 30% of those age 80+.

Patients in category one were more likely to present with high WBC (Table S1). When compared with the NIT group, those in category three tended to be slightly older but did not differ significantly with respect to baseline WBC or general biologic features of their disease. The proportion of rural patients in category three was higher than in the NIT group (56% vs 36%), although this difference was only of borderline significance (P = .06).

The OS of the three treatment groups is shown in Figure S1. The median OS of the IT, NIT and NT groups was 29.2, 7.5, and 0.82 months, respectively (P < .001 by log-rank). The median OS of the 34 patients in the NT group that declined treatment (category 3) was 1.6 months.

Although the proportion of patients in the NT group was lower than those in the older Swedish registry¹ data, it was comparable to both the Spanish PETHEMA² and Netherlands registry³ data. The PETHEMA study evaluated patients over age 60 treated in an earlier time period (2007-2013), and found that 33% of those for whom treatment data were available received BSC. As Alberta has one of the lowest median ages in the Western world (36.3 years in 2016), it is possible that the proportion of NT patients may be higher in jurisdictions with older populations. This proportion is even higher in the relapsed/refractory setting; we recently reported that 44% of such patients received BSC alone.⁵

Treatment decisions in older AML patients are complex processes, which take a multitude of factors into account. These include the availability and efficacy of treatment options, co-morbidities, accessibility to the treatments, anticipated treatment-related toxicities, past experiences, the patient's treatment goals (eg, extending life vs quality of life) and the physician's recommendations. It is often difficult to determine retrospectively the exact reasons for a patient's decision to decline treatment, as this includes consideration of many of these factors, and these are often not clearly delineated in clinical notes. The low expected response rates with HMA or LDAC as single agents may have been a factor in some patients. The advent of newer, more effective combinations, such as azacitidine + venetoclax,⁶ could influence some patients to opt for treatment. We also found a higher frequency of rural-based patients in NT category three (56% vs 36% for NIT). This suggests that the proximity and accessibility to leukemia treatment centers may be an important factor in many cases. For such rural-based patients, having to travel to a cancer clinic seven times per month for azacitidine presents a daunting challenge. This may also be an issue for some frail elderly patients living within urban regions.

For such patients, the availability of an effective oral regimen would potentially make treatment more accessible for many patients. The advent of highly bioavailable oral HMA formulations, including CC-486⁷ and ASTX727,⁸ and targeted oral FLT3 or IDH inhibitors, would provide potentially attractive options which should be further explored in the frontline setting, alone and in combination. It will be interesting to see whether the frequency of patients opting for no treatment declines in the coming years, as more effective and oral-based non-intensive treatments become widely available.

CONFLICT OF INTEREST

J.M.B. received honoraria from Celgene, Pfizer, Astellas, Jazz and Roche. The remaining authors declare no competing financial interests.

AUTHOR CONTRIBUTIONS

J.M.B designed the study, analyzed the data and wrote the manuscript; D.H. compiled and analyzed the data; M.K. performed the statistical analysis; N.Z, E.L., L.B., M.H., J.P., M.T., C.W., T.V.S., A.B. and L.S. edited the manuscript.