Data on the impact of age in acute myeloid leukemia (AML) patients <30 years treated in pediatric and adult trials are scarce.
Data on the impact of age in acute myeloid leukemia (AML) patients <30 years treated in pediatric and adult trials are scarce.
In all, 891 patients <18 years were treated in the pediatric trials AML-BFM 93/98 and 290 adolescents and young adults (>16 to <30 years) in the AMLCG 92/99 and AMLSG HD93/98A trials. Treatment schedules and dose intensities were comparable.
Initial features and risk factors differed considerably between infants (<2 years) and older age groups and only slightly between children (2 to <13), adolescents (13 to <21) and young adults (21 to <30). Treatment results were most favorable in children (5-year event free survival [EFS]: 54% ± 3%), slightly inferior in adolescents (46% ± 4%, P = .03), and unfavorable in young adults (28% ± 5%, P = .0001). Excluding patients with favorable karyotypes, the results were similar in infants and children (EFS: 44% ± 4% and 46% ± 3%, respectively) and inferior in adolescents (35% ± 4%) and young adults (23% ± 4%). There was an increased, age-related percentage and inferior outcome in patients with >5% bone marrow blasts after induction. EFS was especially poor in young adults, with blasts >5%. The blast count after induction was of no prognostic value in patients with favorable karyotypes, but a significant risk factor in patients with other cytogenetics.
Biologic data differed mainly between infants and older age groups. When comparing the same age groups, outcome was similar between the trial groups, which differed from reports concerning acute lymphoblastic leukemia. However, the prognosis decreased after childhood independent of other risk factors. This indicates that even in the younger cohorts increasing age may be an additional unfavorable factor. Cancer 2008. © 2007 American Cancer Society.
Acute myeloid leukemia (AML) represents 15% to 20% of all childhood leukemias, about 33% of adolescent, and about 50% of adult leukemias. After a peak during the first 2 years of life the subsequent low annual incidence of AML slowly increases after 9 years of age (incidence rate 5/1 million in 5–9-year-olds, during adolescence 9/1 million in 15–19-year-olds).1 In general the biological features besides age of pediatric and adult AML appear to be similar, but the differences have not been reviewed systematically. Treatment results in AML have improved during the last 20 years for all age groups; however, outcome decreases slowly with advancing age even when risk factors are considered.2 In contrast to children and adults, data on biological features and outcome are scarce in the adolescent age group, which represents only a small group, both in pediatric and adult trials. Treatment protocols designed for children and adults often differ in various aspects from each other, and there are no data elucidating which kind of therapy could be particularly appropriate for young adults. To evaluate the impact of age we analyzed the clinical and biological features, treatment modalities, and outcome within different age groups below 30 years by using a common dataset from 1 pediatric and 2 adult AML trial groups.
Entry criteria were: age 0 to 30 years, newly diagnosed AML, treated in the pediatric trials AML-BFM 93/98 (n = 891, aged <18 years, diagnosed January, 1993, to June, 2003) as well as adolescents and young adults (>16 to <30 years) of studies AMLCG 92/99 (n = 182, diagnosed November, 1992 to June, 2003) and AMLSG HD 93/98A (n = 108, diagnosed July, 1993 to November, 2004).
Written informed consent was obtained at study entry. Patients with AML and Down syndrome, myelosarcoma (<30% blasts), acute promyelocytic leukemia, secondary AML, or myelodysplastic syndrome were excluded. The FAB classification was used for the initial diagnosis of AML.3–5
Treatment schedules and dose intensities of studies AML-BFM 93/98, AMLCG 92/99, and AMLSG HD93/98A were mostly comparable (Fig. 1).
For more details of the treatment protocols, see Creutzig et al., 2001, 2005, 2006 (AML-BFM 93/98),6–8 Büchner et al., 2003 and 2006 (AMLCG 92),9, 10 Schlenk et al., 2003 and 2004 (AMLSG HD93),11, 12 and Schlenk et al., 2006 (AMLSG HD98A).13 The main differences concerned central nervous system (CNS) irradiation, intrathecal therapy (only given in AML-BFM 93/98), and maintenance (1 year in AML-BFM 93/98 and 3 years in AMLCG 92/99) as well as risk-adapted introduction of autologous and allogeneic blood stem cell transplantation in the AMLSG trials. The percentage of patients transplanted from an allogeneic donor in first complete remission (CR) was similar in all trials and also similar in the defined age groups (see below, 7%–11%).
Four age groups were defined: infants, < 2 years; children, 2 to <13 years (because there were significant differences in biologic parameters between these age groups14); adolescents, 13 to <21 years; and young adults, 21 to <30 years.
A common dataset including the initial clinical and outcome data was the basis for the analysis (Table 1). As risk groups were defined according to different parameters in the trials, risk criteria based on cytogenetics that are widely accepted were used for stratified and multivariate analysis (favorable cytogenetics: t(8;21) and inv(16); unfavorable cytogenetics: complex karyotypes, defined as ≥3 aberrations15 or monosomy 7; intermediate cytogenetics: normal and other karyotypes).
|Age, y||<2||2– <13||13– <21||21– <30||P|
|No. of patients||222||463||276||220|
|Sex m/w, %||51/49||56/44||54/46||51/49||.51|
|Performance status grade 3/4, n (%)*||42 (20)||41 (9)||18 (7)||7 (4)||<.0001|
|WBC median, range/μl||21,150||18,800||13,900||16,300||.026|
|WBC > 100,000/μl, n (%)*||55 (25)||79 (17)||56 (20)||29 (13)||.011|
|CNS pos., n (%)*||38 (19)||35 (8)||21 (11)||n.d.||.0002|
|Extramedullary organ involvement, n (%)*||86 (39)||97 (21)||2 (23)||30 (15)||<.0001|
|M0, n (%)||13 (6)||24 (5)||14 (5)||13 (6)|
|M1, n (%)||11 (5)||65 (14)||56 (20)||43 (20)|
|M2, n (%)||18 (8)||168 (36)||84 (30)||63 (30)|
|M4, n (%)||41 (19)||99 (21)||69 (25)||50 (24)||<.0001|
|M5, n (%)||94 (42)||73 (16)||42 (15)||32 (15)|
|M6, n (%)||4 (2)||16 (4)||9 (3)||7 (3)|
|M7, n (%)||40 (18)||16 (4)||1 (0)||4 (2)|
|t(8;21), n (%)*||1 (0.6)||69 (20)||31 (15)||15 (10)||<.0001|
|inv(16), n (%)*||6 (4)||42 (12)||21 (10)||20 (13)||.026|
|11q23, n (%)*||53 (34)||45 (13)||20 (10)||12 (8)||< .0001|
|Complex karyotypes, n (%)*||20 (13)||10 (3)||10 (5)||13 (8)||.0002|
CR was defined according to the CALGB criteria.16 Early death (ED) patients were those dying within the first 6 weeks of treatment. Response after induction was evaluated by blast count in the bone marrow (BM) ≤/> 5%. EFS was calculated from the date of diagnosis to last follow-up or first event (ED, nonresponse, recurrence, second malignancy, or death from any cause). Patients who did not attain CR were considered failures at time zero. Survival was calculated from the date of diagnosis to the date of death of any cause or last follow-up.
When frequencies were sufficiently large, a χ2 statistic was used. Probabilities of survival were estimated using the Kaplan-Meier method with standard errors according to Greenwood and were compared with the log-rank test. Cumulative incidence functions of recurrence and death in CCR were constructed by the method of Kalbfleisch and Prentice and compared with Gray's test. Computations were performed using SAS (v. 9.1, Cary, NC).
Between November 1992 and 2004 a total of 1181 patients were enrolled; AML-BFM 93/98 trials: 891 patients ≤18 years, AMLCG 92/99 and AMLSG HD93/98A trials: 182 and 108 patients, respectively, in the age group chosen for this analysis (>16 to <30 years). The median follow-up was 5.4 years.17
Comparison of initial clinical, morphologic, and cytogenetic data of the patients included in the 6 trials was possible for most features (see Table 1); however, there were reduced data on CNS involvement in the adult studies because lumbar punctures were performed only in case of suspected CNS involvement. Initial data in studies AMLCG 92/99 and AMLSG HD93/98A were similar (data not shown).
Initial clinical data according to age groups are given in Table 1. Infants showed higher rates of grade 3/4 performance status and higher white blood cell count compared with all other age groups. Initial CNS involvement was more frequent in infants (19%) than in older children (8%) and adolescents (11%). In addition, FAB distribution was quite different in infants who predominantly (134/222 [60%]) presented with FAB subtypes M5 or M7, compared with 168/959 (18%) in the older age groups (P [χ2] < .0001). However, apart from a trend toward increasing frequency of FAB M1 and decreasing frequency of FAB M7 types, there was no significant difference in FAB distribution in children (2 to <13 years) and patients 13 to <30 years old.
The favorable karyotypes [t(8;21), inv16] were rarely seen in infants (7/155 = 4.5%) as compared with older patients (P [χ2] < .0001); most frequently in the 2 to <13-year-olds (111/345 = 32%), and less frequent in adolescents (52/207 = 25%) and young adults (35/158 = 22%, P [χ2] .03). In contrast, 11q23 aberrations were frequent in infants (53/155 = 34%) and less frequent in the other age groups (77/710 = 11%).
In summary, the initial features and risk factors were significantly different when comparing infants with older patients (<30 years) but only slightly different when comparing the age groups 2 to <13, 13 to <21, and 21 to <30 years.
Data on BM blast cell reduction after start of induction were available on more than 85% of the patients. There was an increasing percentage of patients with >5% BM blasts with increasing age (P [χ2] < .0001, Table 2).
|Age (years)||Total patients (n)*||Patients (n) %||5-year EFS % (SE)||P-value (logrank)|
|≤5% blasts||>5% blasts||≤5% blasts||>5% blasts|
|<2||179||156 (87%)||23 (13%)||46 (4)||52 (10)||.79|
|2– <13||419||331 (79%)||88 (21%)||62 (3)||33 (5)||<.0001|
|13– <21||236||177 (75%)||59 (25%)||51 (4)||32 (6)||.0002|
|21– <30||187||114 (61%)||73 (39%)||38 (5)||13 (4)||<.0001|
|Total patients (n)*||1,021||778||243||53 (2)||29 (3)||<.0001|
To compare data from similar age groups in pediatric and adult trials, the results for overlapping and nonoverlapping age groups within the trial groups (pediatric/adult) are given separately in Table 3a. Results for EFS and survival were comparable in the overlapping age groups (>15 to <21 years) in all trials. In a Cox regression model including cytogenetic groups (favorable/unfavorable) and blast count after induction (≤/>5%) the risk ratio (RR) for the factor ‘trial group’ was 0.92 (confidence interval [CI] 0.79–1.05). Therefore, all further results are presented only according to age groups.
|a)Trial-related overlapping and non-overlapping age groups|
|Trial/Age (years)||AML BFM <15 y||AML BFM ≥15 y||AML SG/CG <21 y||AML SG/CG ≥21–30 y|
|Total of patients (n)||796||95||70||220|
|ED < day 15 n (%)||27 (3.4)||2 (2.1)||1 (1.4)||9 (4.1)|
|ED ≥ day 15–42 n (%)||15 (1.9)||0||1 (1.4)||5 (2.3)|
|Nonresponse n (%)||77 (9.7)||13 (13.7)||13 (18.6)||47 (21.4)|
|Death in CCR cumulative incidence, % (SE)||3 (1)||12 (4)#||1 (2)||6 (3)|
|Complete remission n (%)||677 (85.1)||80 (84.2)||55 (78.6)||159 (72.3)|
|Relapse cumulative incidence, % (SE)||32 (2)||29 (7)||34 (8)||38 (6)|
|5-year EFS, % (SE)||49 (2)||43 (5)||43 (6)||28 (3)|
|5-year Survival, % (SE)||60 (2)||50 (5)||58 (6)||42 (4)|
|p(Gray): early death/non-response = < .001; relapse = .178; death in CCR = < .001.|
|#p(Gray) = .03; BFM < 15 y vs. AML SG/CG < 21 y|
|b)Cross-trial age groups|
|Age (years)||<2||2– <13||13– <21||21– <30|
|Total of patients (n)||222||463||276||220|
|<day 15 n (%)||11 (5.0)||10 (2.2)||9 (3.3)||9 (4.1)|
|>day 15–42 n (%)||5 (2.2)||7 (1.5)||4 (1.4)||5 (2.3)|
|Nonresponse n (%)||31 (14.0)||36 (7.8)||36 (13.0)||47 (21.4)|
|Death in CCR cumulative incidence, % (SE)||2 (1)||4 (1)||5 (2)||6 (3)|
|Complete remission n (%)||175 (78.8)||410 (88.6)||227 (82.2)||159 (72.3)|
|Relapse cumulative incidence, % (SE)||35 (5)||30 (3)||32 (4)||38 (6)|
|5-year EFS, % (SE)||42 (3)||53 (2)||45 (3)||28 (3)|
|5-year survival, % (SE)||57 (3)||62 (2)||56 (3)||42 (4)|
|p(Gray): early death/non-response = < .001; relapse = .098; death in CCR = < .093|
The CR rate was highest in children (89%) and lowest in infants and patients older than age 21 (<2 years: 79%, 13 to <21: 82%, 21 to <30: 72%). This was mainly due to a low rate of nonresponders in the cohort of 2 to <13-year-old children. Five-year EFS results were also most favorable in children (53% ± 2%), slightly inferior in infants (42% ± 3%, Plogrank = .0002) and adolescents (45% ± 3%, Plogrank = .007) and unfavorable in young adults (28% ± 3%, Plogrank < .0001, Fig. 2A). Excluding patients with favorable karyotypes, results were similar in infants and children (EFS = 44% ± 4% and 46% ± 3%, respectively) and inferior in adolescents (35% ± 4%) and young adults (23% ± 4%). Differences in 5-year survival were smaller between the age groups and mainly seen between young adults and the younger cohorts (Fig. 2B).
The ED rate was low in all age groups, but slightly higher in infants, who often presented initially with hyperleukocytosis, a risk factor for ED due to cerebral hemorrhage.18 The main reason for a better outcome in 2 to <13-year-old children was a higher CR and lower recurrence rate compared with other age groups.
There was a trend for higher treatment related mortality in remission (TRM) which increased with age (PGray = .09, Table 3b).
Results according to favorable cytogenetics were similar in the age groups <21 years but inferior in young adults (Plogrank = .04, Fig. 3A). In patients without favorable cytogenetics there was a trend for inferior results after childhood (Plogrank = .002).
Excluding infants, EFS in patients with > 5% BM blasts was inferior compared with patients with ≤5% blasts after induction (Table 2). In patients with favorable karyotypes the results were independent from the BM blasts cell count after induction (EFS: ≤5% blasts 72% ± 3% and >5% blasts 65% ± 10%).
In addition, patients with intermediate or unfavorable cytogenetics showed similar results in case of ≤5% BM blasts (EFS 48% ± 3% and 42% ± 7%, Fig. 3B).
Presenting with >5% BM blasts after induction was an adverse factor in patients with intermediate and unfavorable cytogenetics (EFS 21% ± 4% and 10% ± 5%, Fig. 3C).
When combining cytogenetics and data on BM blasts after induction it was possible to clearly separate 3 risk groups (Fig. 4).
In a Cox regression model including cytogenetic groups (favorable/intermediate/unfavorable), blast count after induction (≤/>5%) and age groups (<2/≥21 years), favorable cytogenetics (RR .43, CI 0.32–0.58), blast count after induction >5% (RR 1.98, CI 1.58–2.47), and age ≥21 (RR 1.65, CI 1.29–2.12) remained of prognostic significance. When analyzing only patients below 21 years of age no increased risk for a specific age group could be found.
The results presented here are based on original data of AML patients aged <30 from 6 trials performed in pediatric and adult study groups. The aim was to assess the possible effect of the biological factor “age” and the clinical factor “treatment group” (pediatric/adult trial) on outcome of patients with AML.
In contrast to data on elderly AML patients, biological parameters in different age groups of young AML patients are found rarely in the literature. Jeha et al.19 reported the influence of vascular endothelial growth factor (VEGF) in pediatric and adult patients. Unlike in adults, VEGF and VEGF-R2 levels in pediatric AML patients did not correlate with survival. Similarly, and likewise contrary to adults, expression of the multidrug resistance gene (MRD1) failed to define a poor-prognostic group in childhood AML.20 However, the prognostic value of cytogenetics is well established in all age groups.21–23
The biologic data differ considerably between infants and older age groups and only slightly between children, adolescents, and young adults. The distribution of cytogenetic aberrations in infants is different from that in older patients. Infants showed nearly no favorable aberrations, but frequently 11q23 aberrations and complex karyotypes, which indicates some similarities with old AML patients (>60 years).24 Conversely, the percentage of favorable cytogenetics [t(8;21) and inv(16)] was highest in children (32%) and decreased in the older age groups (<25%). Interestingly, only t(8;21) was less frequently seen in adolescents and young adults than in the 2 to <13-year-old children, but not inv(16).
Survival data for the specific age groups of adolescents and young adults are scarce. Population-based data from England and Wales showed a 5-year survival of 46% for 15–29-year-old AML patients (period 1989–1994). For patients of this age group treated in the Medical Research Council (MRC) trials AML9 and AML10, 5-year survival was 55% during the same period.25 Currently, 5-year survival in children enrolled in clinical trials is in the range of 45% to 60%.26 In the AML-BFM studies, survival results improved from 49% (period 1987–1992) to 60% (period 1993–1998).7 For adults <50 years of age treated in the AMLCG trials the overall survival was ≈50% without significant differences between the different age decades.27
There are several reports comparing pediatric and adult treatment of acute lymphoblastic leukemia (ALL) in adolescents that demonstrate a better outcome with the more intensive pediatric protocols.28, 29 The same was reported after comparing adolescents with AML treated on CCG 2891 with intensive timing (1989–1995) and patients of the same age group treated at the M. D. Anderson Cancer Center on relatively less aggressive adult protocols (1980–2000). Patient characteristics were similar; however, 5-year survival in the CCG-trial was 51% compared with 32% in the adult trial.30
As outcome data strongly depend on treatment, it is important to evaluate whether different age groups have been treated similarly. Regarding induction and consolidation the treatment schedules and dosages of the pediatric trials AML-BFM 93/98 and the adult AMLCG and AMLSG trials were similar (Fig. 1).9, 11 Accordingly, the overlapping age groups treated in pediatric and adult trials showed similar outcome. The CR rate was highest in the age group 2 to <13 and lower in infants and patients of older age. Also, long-term treatment results were most favorable in 2 to <13-year-old children (5-year EFS: 54% ± 3%), slightly inferior in adolescents (46% ± 4%, P = .03), and unfavorable in young adults (28% ± 5%, P = .0001). The same decline in results with increasing age was seen when risk factors were considered.
Results on BM blasts after induction showed 1) an age-related increased percentage of patients with >5% blasts and 2) inferior outcome in all groups of patients with >5% blasts except in infants. EFS was especially poor in young adults with >5% blasts after induction (Table 2). Kern et al.31 reported that early blast clearance by remission induction therapy was a major independent prognostic factor for achievement of CR and long-term outcome in adult AML. They found no correlation between outcome and blast clearance within the cohorts with favorable karyotypes, a finding that was similar in our analysis: the blast count after induction was only a significant risk factor in patients with intermediate or unfavorable cytogenetics.
When comparing the same age groups, differences in outcome were minimal between the trial groups; however, survival and EFS decreased after childhood irrespective of other risk factors. Schoch et al.27 analyzed the effect of age and cytogenetics on clinical outcome in adult patients (>16 years). They found that both age and cytogenetics were independent prognostic parameters in AML, but up to the age of 49 years age had no major impact on prognosis, whereas the karyotype did. Our analysis showed that especially the age group of 21 to 30 years has an unfavorable outcome that could not be explained by different therapy intensities; other factors may play a role, eg, varying therapy compliance with more or less rigid adherence to the protocol or varying practice of CNS therapy.
A similar observation of an adverse age effect was reported by Razzouk et al.32 when comparing patients < 10 years and 10 to 21 years. They concluded that younger children benefit more than older children from newer intensive therapies.
Register data show a trend to lower survival rates after the age of 30 when comparing them to the younger age groups: 44% (0–15 years), 42% (15–29 years), and 32% (30–44 years), time period 1993–1998.1
Data from trials including children, adolescents, and young adults indicate only small differences between these age groups: The CCG-2891 trial included patients < 22 years. Five-year survival in 0 to 16 and 16–21-year-old patients treated at intensive timing chemotherapy was 49% and 51%, respectively.30 The MRC trial AML10 (1988–1995) included AML patients < 35 years on the same treatment regimen. They achieved high CR rates for children < 15 years (91%) and young adults aged 15–34 (85%). Survival at 5 years was 53% and 60% in children (after ADE and DAT induction, respectively) and 46% and 47% for the age groups from 15 to 24 and 25 to 34.33 These data do not show the same decrease in the young adult group that we found in our analysis. However, similar unfavorable outcome data for young adults have been reported in more than 1000 Japanese AML patients aged <29 (period 1986–1999) and treated in a variety of institutions and protocols. Seven-year EFS was 34% and 32% in the age groups 10 to 15 and 15 to 19, respectively, and decreased to 26% in the 20–29-year-old adults.34
Generally there were similar results in children and adolescents treated in the same pediatric trial up to 21 years of age (CCG study), a tendency to inferior outcome after childhood when the same treatment protocol was applied in pediatric and adult units (MRC trial), and with various protocols in the Japanese analysis.
Most difficult in the management of adolescents is the indispensable psychosocial care. The needs of adolescents differ from those of young children and arise with the conventional problems that are found in this age group, eg, need of autonomy and independence, social development, sexual maturation, education, and employment.35 In view of the specific needs of adolescents, it is recommended to treat these patients in special units whenever possible.25
There is also a big difference for children and adolescents to access treatment in clinical trials, which might influence prognosis.36, 37 In the Nordic countries and in Germany more than 90% of children <15 years with AML are treated within clinical trials38, 39; in the UK this applies to 67%40 and in the US to >60%. However, regarding the 15–20-year-olds or older with cancer, the number of patients enrolled in clinical trials is much lower.36, 37
The data of our pediatric and adult AML trials indicate that biological factors are similar in children, adolescents, and young adults. Currently, 5-year survival is in the range of 50% to 60% for patients below age 21, but inferior in the age group 21 to 30 years (in the range of 40%). This trend for better survival rates in the younger age groups can be at least partly explained by the finding that age is 1 of the independent risk factors.
The authors thank all participating hospitals of the AML-BFM, AMLCG, and AMLSG study groups