The demarcation between younger and older acute myeloid leukemia patients: A pooled analysis of 3 prospective studies
Masamitsu Yanada MD,
Fujita Health University School of Medicine, Toyoake, Japan
Corresponding author: Masamitsu Yanada, MD, Department of Hematology, Fujita Health University School of Medicine, 1-98 Dengakugakubo, Kutsukake-cho, Toyoake, 470-1192, Japan; Fax: (011) +81-562-95-0016; firstname.lastname@example.org
Contemporary treatment protocols for adult acute myeloid leukemia (AML) are age-specific, and older patients are generally treated less intensively than younger patients. However, it remains uncertain whether older but fit patients with AML really need to have their treatment attenuated.
To evaluate the contribution of age to outcome for patients with AML receiving intensive chemotherapy, data were analyzed for 2276 patients aged less than 65 years who were treated uniformly, regardless of age, in 3 consecutive prospective studies conducted by the Japan Adult Leukemia Study Group.
A substantial drop in overall survival (OS) between patients aged 40 to 49 years and 50 to 64 years led to a focus on 2 comparisons: 1) age < 50 versus ≥ 50 years; and 2) age 50 to 54 versus 55 to 59 versus 60 to 64 years. OS was significantly better for patients aged < 50 years than that for those aged ≥ 50 years (49.6% and 37.0% at 5 years; P < .001); older patients were more susceptible to relapse, but not to early death or nonrelapse mortality. The significant differences in OS between these 2 age groups were equally seen for patients with favorable, intermediate, and adverse cytogenetics (P < .001 each). Outcomes for those aged 50 to 54, 55 to 59, and 60 to 64 years were similar, with 5-year OS rates of 38.2%, 35.1%, and 38.0%, respectively (P = .934), and no differences in early death or nonrelapse mortality were observed among these age groups.
Age is among the most important prognostic factors in acute myeloid leukemia (AML).[1-5] Increasing age in AML is associated with a higher frequency of unfavorable biological characteristics such as adverse cytogenetics, preceding myelodysplastic syndrome (MDS), and expression of the multidrug resistance phenotype, all of which are involved in intrinsic resistance to chemotherapy.[6-9] In addition to the disease biology, patient-related factors such as poor general condition and significant comorbidities also contribute to inferior outcomes for older patients.[8, 10, 11] Because of such distinct biological and clinical features, contemporary treatment protocols for adult AML are age-specific and are typically divided into those for younger and older patients, with older patients treated less intensively than younger patients. For this purpose, age 55 or 60 years is generally used as the demarcation between these 2 groups[1, 2]; however, this cutoff age is quite arbitrary, and it remains uncertain whether patients over such age limits really need to have their treatment attenuated.
For the recent prospective AML studies conducted by the Japan Adult Leukemia Study Group (JALSG), age less than 65 years was used as the eligibility criterion, with dose modifications not having been adopted according to age. This situation provides a welcome opportunity to evaluate the contribution of age to outcome for patients with AML treated with uniform intensive chemotherapy. For the study reported here, we integrated data for 2276 patients entered into 3 consecutive prospective studies between 1995 and 2005 for a comparison of patient characteristics and treatment outcomes among different age groups.
MATERIALS AND METHODS
All patients were subjects of one of the three phase 3 studies conducted by the JALSG, that is, the AML95 (from 1995-1997), AML97 (from 1997-2001),[13, 14] and AML201 (from 2001-2005) studies.[15, 16] All of these studies adopted the same eligibility criteria: newly diagnosed AML (acute promyelocytic leukemia excluded), age 15 to 64, an Eastern Cooperative Oncology Group performance status 0 to 3, adequate functioning of the liver (serum bilirubin level < 2.0 mg/L), kidneys (serum creatinine level < 2.0 mg/dL), lungs (PaO2 ≥ 60 Torr or SpO2 ≥ 93%), and heart (no significant abnormalities on electrocardiograms and echocardiograms). Patients with AML secondary to MDS or cytotoxic treatment were not eligible for enrollment. Written informed consent was obtained from all patients prior to registration. Each protocol was reviewed and approved by the institutional review boards of the participating centers, and was conducted in accordance with the Declaration of Helsinki.
The treatment schedule for each study is described in detail elsewhere.[12-16] The AML95 study compared a fixed schedule (ie, “3+7”) and an individualized schedule (up to “4+10” depending on the bone marrow findings on day 8) for induction therapy with idarubicin and cytarabine. Postremission therapy consisted of 3 courses of consolidation therapy including behenoyl cytarabine and 12 months of maintenance therapy. The AML97 study adopted the 3+7 induction therapy with idarubicin and cytarabine for all patients.[13, 14] After achieving complete remission (CR), patients were randomized to receive 3 or 4 consolidation courses that included standard-dose cytarabine, followed by 12 months of maintenance therapy only for the 3 courses. Those with a human leukocyte antigen (HLA)-identical sibling donor were assigned to allogeneic hematopoietic cell transplantation (HCT) if they were younger than 50 years and at intermediate or poor risk, as determined with a scoring system which took into account cytogenetics, white blood cell count, and other factors. The AML201 study compared idarubicin (12 mg/m2 for 3 days) and daunorubicin (50 mg/m2 for 5 days) both combined with cytarabine for induction therapy.[14, 15] Patients in CR were randomly assigned to either 4 consolidation courses with standard-dose cytarabine or 3 courses with high-dose cytarabine. Allogeneic HCT was offered to patients aged 50 or younger if they presented with intermediate or adverse cytogenetics and had an HLA-identical sibling donor. In principle, doses were not modified according to age for any protocol. The single exception was for high-dose cytarabine in the AML201 study, in which reduction of the cytarabine dose from 2 g/m2 to 1.5 g/m2 was allowed for patients aged 60 years or older.
Karyotypes were classified as favorable, intermediate, or adverse, in line with the revised UK Medical Research Council (MRC) criteria. Monosomal karyotype was defined according to the criteria developed by Breems et al.
CR was defined as the presence of all of the following: < 5% of blasts in bone marrow, no leukemic blasts in peripheral blood or extramedullary sites, and recovery of peripheral blood counts. Early death was defined as death from any cause occurring within 30 days after the start of induction therapy. Overall survival (OS) was defined as the time from the start of treatment to death or last visit, and relapse-free survival as the time from CR to relapse, death or last visit. Patients undergoing allogeneic HCT were not censored at the time of transplantation unless indicated.
Distributions of patient characteristics between and among groups were compared by using the chi-square test for categorical variables. Differences in continuous variables were compared by means of the Wilcoxon rank-sum test for distribution between 2 groups, and the Kruskal-Wallis test for distribution among 3 groups. The probabilities of OS and relapse-free survival were estimated by using the Kaplan-Meier method, with differences between groups qualified with the log-rank test. First, we examined OS by dividing patients into 4 age groups: 15-29, 30-39, 40-49, and 50-64 years. This provisional analysis disclosed a substantial drop in OS between patients aged 40-49 and 50-64 years (Fig. 1A). This finding led us to focus on 2 comparisons for subsequent analyses: 1) age < 50 versus ≥ 50 years; and 2) age 50 to 54 versus 55 to 59 versus 60 to 64 years. Relapse and nonrelapse mortality were considered as competing risk events for each other, and the probabilities of relapse and nonrelapse mortality were estimated by using the cumulative incidence functions, with differences between groups qualified by the Gray test. The Cox proportional hazards regression model was used for multivariate analysis, and a hazard ratio (HR) was calculated in conjunction with a 95% confidence interval (CI). All statistical analyses were performed by using Stata version 12.0 software (StataCorp, College Station, Tex).
A total of 2276 patients (430 from AML95, 789 from AML97, and 1057 from AML201) were analyzed for this study, with a median follow-up of surviving patients of 4.2 years (range, 0.0-8.0 years). Table 1 shows baseline characteristics of the patients according to age groups (age < 50, 50-54, 55-59, and 60-64 years). There was no significant relationship between performance status and age. The distribution of cytogenetic risk differed modestly but significantly for patients aged < 50 and ≥ 50 (P < .001), but the difference was not significant for those aged 50 to 54, 54 to 59, and 60 to 64 years (P = .577). Comparison of patients aged < 50 and ≥ 50 years showed that t(8;21) and inv(16)/t(16;16) occurred more frequently in younger patients (P < .001 and P = .043, respectively), whereas the frequencies of add(5q)/del(5q)/-5, add(7q)/del(7q)/-7, complex karyotype, and monosomal karyotype were higher for older patients (P = .002, P = .021, P < .001 and P < .001, respectively). None of these cytogenetic aberrations, however, showed significant differences in distribution among those aged 50 to 54, 54 to 59, and 60 to 64 years.
Table 1. Patient Characteristics
Age <50 y N = 1339
Age 50–54 y N = 334
Age 55–59 y N = 322
Age 60–64 y N = 281
White blood cell count, ×109/L
Hemoglobin level, g/dL
Platelet count, x109/L
Bone marrow blasts, %
Peripheral blood blasts, %
Specific cytogenetic aberrations
Complete Remission and Early Death
Rates of CR and early death are summarized in Table 2. Patients younger than 50 years tended to show higher CR rates than those aged 50 or older, but the difference failed to reach statistical significance (P = .078), whereas there was no difference in the CR rates among those aged 50 to 54, 54 to 59, and 60 to 64 years (P = .829). Early death within 30 days after the start of induction therapy occurred in 1.8%, 1.5%, 2.5%, and 3.2% of patients aged < 50, 50 to 54, 55 to 59, and 60 to 64 years, respectively, with no significant difference between those aged < 50 and ≥ 50 (P = .367), or among those aged 50 to 54, 54 to 59, and 60 to 64 (P = .356). The rates of death within 60 days were 2.7%, 5.1%, 5.0%, and 6.1% for the respective age groups (P = .003 for age < 50 and ≥ 50, and P = .813 for age 50 to 54, 54 to 59, and 60 to 64 years).
Table 2. Remission Induction Results and Outcomes at 5 Years by Age Group
Cumulative incidences of relapse and nonrelapse mortality for the 1788 patients who attained CR are shown in Table 2. Patients 50 years of age or older were more likely to experience relapse than were those younger than 50 (P = .008), whereas there was no difference in relapse rates among those aged 50-54, 55-59, and 60-64 years (P = .196). The nonrelapse mortality rates did not differ significantly between patients aged < 50 and ≥ 50 years (P = .695), or among those aged 50 to 54, 54 to 59, and 60 to 64 years (P = .388).
Figure 1A compares OS for patients divided into 4 age groups: 15 to 29, 30 to 39, 40 to 49, and 50 to 64 years. As mentioned above, this result prompted us to first postulate a distinction between patients younger and older than 50 years. OS was significantly better for patients aged < 50 years than that for those aged ≥ 50 years (49.6% and 37.0% at 5 years, P < .001). Among those aged ≥ 50, however, increasing age did not seem to affect OS, because survival curves for patients aged 50 to 54, 55 to 59, and 60 to 64 years were superimposed, with 5-year OS rates of 38.2%, 35.1%, and 38.0%, respectively (P = .934; Fig. 1B).
To evaluate whether the difference in OS between those aged < 50 and ≥ 50 years depends on cytogenetic risk, comparisons between these 2 groups were made within each cytogenetic risk group. This analysis showed that the intergroup difference was significant for favorable (P < .001; Fig. 2A), intermediate (P < .001; Fig. 2B), and adverse cytogenetic risk (P < .001, Fig. 2C). The effect of age on OS remained significant in a multivariate analysis adjusting for other covariates (Table 3). Allogeneic HCT was performed for 687 (51%) of the patients aged < 50 years, 101 (30%) of those aged 50 to 54 years, 58 (18%) of those aged 55 to 59 years, and 19 (7%) of those aged 60 to 64 years. Censoring the findings for these patients at the time of allogeneic HCT did not alter the main results; Kaplan-Meier survival curves with censoring of patients undergoing allogeneic HCT are shown for those aged < 50 and ≥ 50 years in Fig. 3.
Table 3. Multivariate Analysis of Risk Factors for Overall Survival
Finally, we examined whether lack of significant interaction between age and OS in patients aged 50 to 64 years remains after adjusting for other potentially confounding factors. When a multivariate analysis was undertaken for these older patients by including the covariates listed in Table 3, age group had no impact on OS (HR = 0.93; 95% CI = 0.76-1.13, for patients aged 55-59 years; HR = 0.93; 95% CI = 0.76-1.14, for patients aged 60-64 years; both with reference to those aged 50-54 years).
To investigate how increasing age affects outcomes for patients with newly diagnosed AML, we analyzed data for 2276 patients 15 to 64 years of age who were treated uniformly, regardless of age, in 3 consecutive prospective AML studies by JALSG. This large-scale retrospective analysis yielded several relevant findings: 1) age 50 was a significant dividing point for outcomes; 2) patients aged 50 to 64 years were more susceptible to relapse, but not to early death or nonrelapse mortality than those younger than 50 years; and 3) outcomes did not differ among patients aged 50 to 54, 55 to 59, and 60 to 64 years.
Why were the survival rates for patients 50 years of age or older in our study significantly inferior to those of patients younger than 50? Our data indicate that worse outcomes for older patients resulted from higher relapse rates. In AML, it has been well established that cytogenetic findings at diagnosis are associated with the risk of relapse.[17-20] Comparison of the frequencies of distinct cytogenetic aberrations showed that younger patients were more likely to exhibit favorable cytogenetics such as t(8;21) and inv(16)/t(16;16), whereas older patients were more likely to show adverse cytogenetics such as abnormalities of chromosome 5 or 7, complex karyotype, and monosomal karyotype. However, such a difference in the distribution of cytogenetics alone could not have accounted for the difference in outcomes between patients aged < 50 and ≥ 50 years observed in this study, because the significant differences in OS between these 2 age groups were seen for all cytogenetic risk groups.
Moreover, it could be expected that allogeneic HCT would result in more favorable outcomes for younger patients. However, although the proportion of patients who had undergone allogeneic HCT was indeed higher among younger than older patients, censoring the findings obtained at the time of allogeneic HCT produced no major changes in the study results. Therefore, it seems that neither cytogenetics nor allogeneic HCT can explain why older patients suffered relapse more frequently than younger patients. Secondary AML could not have been the reason, either, because our study cohort consisted of only patients with de novo AML. Other mechanisms that had not been studied here, such as molecular profiles, may play a significant role in differences in outcomes for younger and older patients.[21-24]
The analytic results for data of patients 50 years of age or older in our study also provide insights into the treatment of older patients with AML. Patients aged 50 to 54, 55 to 59, and 60 to 64 years had similar long-term survival, and no differences in early death or nonrelapse mortality were observed among them. This finding calls into question whether older patients really need to be treated differently. Recently, Lowenberg et al compared the effect of a doubled dose of daunorubicin of 90 mg/m2 with that of a conventional dose of 45 mg/m2 in the context of the 3+7 regimen for patients aged 60 years or older. Although no difference in outcome was observed overall, patients between 60 and 65 years of age significantly benefited from the doubled dose of daunorubicin. Taking these results into account, older and fit patients, especially those under the age of 65 years, may still benefit from intensified chemotherapy.
When interpreting our data, we should bear in mind that our study cohort consisted exclusively of newly diagnosed AML patients under the age of 65 years who were entered into phase 3 studies. In addition, our cohort did not include patients with AML secondary to MDS or cytotoxic treatment. Secondary AML accounts for approximately 35% of the whole AML population,[26, 27] and the frequency is even higher among older patients.[3, 5] Our results therefore might not be applicable to the general AML population. This limitation may well be partly complemented by the findings of several previous studies. Buchner et al analyzed data of 2776 patients with de novo AML with no upper age enrolled onto 2 prospective studies by the German AML Cooperative Group. In that study, OS for patients older than 60 years was only half that of younger patients, and this difference was attributable to less frequent CR and more frequent relapse in older patients. It seems likely that inclusion of elderly patients might have contributed to a larger prognostic difference between younger and older patients. By using the combined data of 5 AML studies conducted by the Southwest Oncology Group, Appelbaum et al evaluated effect of age on outcomes. Their study included not only patients with de novo AML but also those with secondary AML, with no upper age limit employed in trials for older patients. CR rates and OS were shown to worsen with advanced age, and this held true even if patients aged 56 to 65, 66 to 75, and older than 75 years were compared. It is conceivable that discrepant results among these studies, including ours, could be a reflection of differences in analyzed patient population. Through an entirely different approach, Juliusson et al evaluated the effect of age on outcomes for AML by using data for 2767 unselected patients with AML who were consecutively enrolled in the Swedish Acute Leukemia Registry. They showed that intensive chemotherapy was associated with improved survival even for elderly patients, although it should be remembered that patients in that study were treated heterogeneously, and the choice of treatment must have been dependent on known and unknown confounding factors. Our study, in contrast, is advantageous in that the study population consisted of patients who were treated homogeneously regardless of age.
To summarize, we analyzed data for a large number of patients with AML aged 15 to 64 years who were treated uniformly in the context of clinical studies, and could not determine a specific age limit over which attenuation of treatment intensity is advisable. Our results justify the use of intensive chemotherapy without dose attenuation toward older but fit AML patients at least up to the age of 64.
This work was supported in part by a Grant-in-Aid for Cancer Research from the Ministry of Health, Labour and Welfare of Japan (Clinical Cancer Research 23-004); and the National Cancer Center Research and Development Fund (23-A-23).
CONFLICT OF INTEREST DISCLOSURE
Dr. Kobayashi has received grants from Ohtsuka, Celgene, and CMIC, and has received payment for lectures from Nippon Shinyaku, Bristol, Novartis. Dr. Ohno has been a consultant for Nippon Shinyaku KK and Novartis Japan, and has received payment for lectures from Nippon Shinkayu KK, Kyowa Hakko Kirin KK, Chugai Pharmaceutical KK, and Bristol-Myers Squibb Japan. All other authors made no disclosure.