Impact of age on the survival of pediatric leukemia: an analysis of 15083 children in the SEER database

BACKGROUND & AIMS Age at diagnosis is a key factor for predicting the prognosis of pediatric leukemia especially regarding the survivorship assessment. In this study, we aimed to assess the impact of this prognostic factor such as age in children with pediatric leukemia. METHODS In this study, Surveillance, Epidemiology, and End Results Program-registered children with leukemia during 1988-2013 were analyzed. All patients were divided into five groups according to the age at the time of diagnosis (<1, 1-4, 5-9, 10-15, >15 years old). Kaplan-Meier and multivariable Cox regression models were used to evaluate leukemia survival outcomes and risk factors. RESULTS There was significant variability in pediatric leukemia survival by age at diagnosis including ALL, AML and CML subtypes. According to the survival curves in each group, survival rate were peaked among children diagnosed at 1–4 years and steadily declined among those diagnosed at older ages in children with ALL. Infants (<1 year) had the lowest survivorship in children with either ALL or AML. However, children (1-4 years) harbored the worst prognosis suffering from CML. A stratified analysis of the effect of age at diagnosis was validated as independent predictors for the prognosis of pediatric leukemia. CONCLUSIONS Age at diagnosis remained to be a crucial determinant of the survival variability of pediatric leukemia patients.


INTRODUCTION
Leukemia is the most common malignant tumor in children worldwide representing up to 30% of all pediatric cancers [1][2][3].For example, approximately 3000 children are diagnosed with leukemia every year in the United States [4,5].Among the major types of leukemia, Acute lymphoid leukemia (ALL) contributes to 76% of all leukemia cases and 43% of all deaths of pediatric leukemia patients [6,7].Acute myeloid leukemia (AML) accounts for more than 30% of the deaths from pediatric leukemia, although it makes up only 15~20% of pediatric leukemia [5,8].Chronic myeloid leukemia(CML) constitutes 2% of all leukemia in children younger than 15 years and 9% of all leukemia in adolescents between 15 and 19 years, with an annual incidence of 1 and 2.2 cases per million in these age groups, respectively [9,10].With the improvement of diagnosis and treatment criteria, impressive advancements in childhood leukemia treatment entail cure rates reaching 90% for ALL and 60%-70% for AML [11].Age is a recognized prognostic factor, with poorer survival for older adults than children, but less attention has been given to the effects of age on prognosis within childhood especially on pediatric leukemia [12][13][14][15].
Prognostic factors identified to date include morphology, immunophenotype, molecular and cytogenetic markers as well as host factors such as age at diagnosis and race.For instance, age at diagnosis has been recognized as an important prognostic factor of both incidence and survival of pediatric ALL [16].Various studies have conducted to pinpoint specific genetic and biological processes occurring in different age groups Clinical Research Paper www.impactjournals.com/oncotarget to account for the prognostic value of age at diagnosis.For example, researchers have identified that the lowest survival among patients diagnosed during infancy, followed by children who are diagnosed between 15 and 19 years of age [17].They also point out that chromosomal rearrangement TEL/AML1, DNA index or BCR/ABL rearrangement may responsible for the poor prognosis at certain age of diagnosis [18][19][20].
To further clarify the issue of age on pediatric leukemia prognosis, Surveillance, Epidemiology, and End Results (SEER) population-based data during 1988-2013 were analyzed in this study enrolled ALL, AML and CML [21].We further employed an independent cohort set including ALL and AML children to analyze the association between the age at diagnosis and prognosis rate.

Clinicopathologic parameters of patients
As presented in Table 1, 15083 children were enrolled as diagnosed with leukemia during the 25-year study period (between 1988 and 2013) in the SEER database.Among this, 8407 (55.7%) were males and 7931 (44.3%) were females.Consistent with disease morbidity, 11624 children were suffering from ALL, 2606 children were diagnosed as AML and 853 patients were CML.According to the age classification, 723 were infants, 6074 children were between 1-4 year-old, 3448 children were between 5-9, 2696 children were between 10-14 and 2142 children were elder than 15.The median follow-up period was 32, 72, 68, 66 and 57 months in each group.

Clinicopathological differences between subgroups
As illustrated in Table 1, significant differences were found between the 5 groups, including sex (more frequent in men, P < 0.001), race (less frequent in Black; P < 0.001), pathologic type (ALL > AML > CML, P < 0.001).However, no difference was found in years of diagnosis (P = 0.187).

Impact of age on pediatric leukemia survival outcomes
The 36-month and 60-month survival rates of pediatric ALL were 19    Multivariate analysis was also performed by the Cox regression model.The following three factors were found to be independent prognostic factors for either ALL, AML or CML (Table 3).
Among the three histological type, no statistical difference were observed with regards to sex (p = 0.46, 0.22, 0.31, respectively) according to multivariate survival analysis.

Evaluating the SEER database outcomes in an independent cohort set
To evaluate the reliability of the SEER results, data for 107 eligible ALL pediatric patients and 125 AML from the NCHNMU were analyzed.Patients' demographic characteristics and pathological features were summarized in Supplemental Table 1.Due to the limitation of morbidity, we could only analyze the follow-up data for ALL and AML at this time.The age group was divided as described before.Consistent with the SEER database, it was interesting to note that infant with ALL harbored the worst survival while the children with 1-4 year-old indicated a better prognosis (Figure 2A).In children with AML, we found that either infant or children with 1-4 year old indicated a poor survival rate (Figure 2B).

DISCUSSION
The prognostic value of age at diagnosis in pediatric leukemia has long been recognized.The present study is conducted to assess the effect of age at diagnosis on the survival patterns of children diagnosed with leukemia including ALL, AML and CML.Our main finding is that there is a significant variation in survival by age at diagnosis, with the worst outcome for children diagnosed in infancy, the best outcome for those diagnosed during the age of 1-4 years in pediatric ALL children.Hossain et al also indicated infants ( < 1 year) had the lowest survivorship.In a multivariable Cox proportional hazard model stratified by year of diagnosis, those diagnosed in age groups 1-4, 5-9, 10-14, and 15-19 years were 82%,75%, 57%, and 32% less likely to die compared to children diagnosed in infancy, respectively among ALL children [22].In children with AML, we found that infant and children within 1-4 year-old indicating a worse prognosis.In addition, we also identified that the worst outcome for children diagnosed as CML during the age of 1-4 years.The differential survival patterns of pediatric leukemia by age at diagnosis persist after accounting for the effects of known prognostic factors: sex, race, receipt of radiation therapy, immunophenotype, and the number of primary tumor sites.These patterns may be partly due to a variety of age-dependent favorable and unfavorable clinical and biological features mentioned in the introduction.
Among solid tumor, age was also considered as a predictor for patients' survival.Various studies have also reported that age plays a paradoxical role on the prognosis of HCC [23].Cho et al. demonstrated that young patients had poorer survival rates than elderly patients due to the a more advanced tumor stage at diagnosis [24].In this study, the poor prognosis of either ALL or AML in infants might be partial due to a poorer immune system, more advanced stage or chromosome abnormality.

CONCLUSIONS
There is a differential survival pattern of pediatric leukemia by age at diagnosis.The detailed biological mechanism and environmental processes occurring at different stages of development may give rise to this association.Future research could focus on identifying these processes and elucidating their mechanisms.

Patient selection
The study included 15083 children who were diagnosed with leukemia between ages 0 and 18  This study was based on public data from the SEER database.The data did not include the use of human subjects or personal identifying information.Thus, no informed consent was required for this part of the study.This study was also in compliance with the Declaration of Helsinki of the World Medical Association and was approved by the ethics committee of Nanjing Children's Hospital Affiliated with Nanjing Medical University (NCHNMU).

Figure 2 :
Figure 2: Survival curves of patients with pediatric leukemia in independent cohort set in different age groups.A. ALL; B. AML.

Table 1 : Characteristics of patients from SEER database by ages Feature Number of patients
.1% and 17.2% in group 1, 26.2% and 24.1% in group 2, 23.2% and 21.1% in group 3, 21.1% and 19.1% in group 4 and 20.2% and 18.8% in group 5, respectively.The overall log-rank test showed that the overall 5-year ALL survival was presented in Figure 1A IQR: Interquartile Range; ALL: Acute lymphoid leukemia; AML: Acute myeloid leukemia; CML: Chronic myeloid leukemia www.impactjournals.com/oncotargetoveralllog-ranktest showed that the overall 5-year AML survival was presented in Figure1B (P <0.001).The survival rate of CML were different with ALL or AML.The 36-month and 60-month were 22.3% and 19.7% in group 1, 17.4% and 14.4% in group 2, 18.9% and 15.1% tendency (men), an early year of diagnosis (1988-2003), black race as well as age distribution (infant for ALL and AML, 1-4 year-old for CML) were regarded as significant risk factors for low 60-month pediatric leukemia followup rate (Table 2).

Table 2 : Univariate survival analysis of patients with leukemia according to various clinicopathological variables.
mo: month; ALL: Acute lymphoid leukemia; AML: Acute myeloid leukemia; CML: Chronic myeloid leukemia

Table 3 : Multivariate Cox model analysis of patients with leukemia according to various clinicopathological variables.
HR: Hazard ratio; CI: confidence interval; ALL: Acute lymphoid leukemia; AML: Acute myeloid leukemia; CML: Chronic myeloid leukemia a: P values were adjusted for years of diagnosis, sex, age, race and age as covariates.
years during 1988-2013, whose information was reported to one of the 17 SEER registries.In detail, 11624 children with ALL, 2606 children with AML and 853 children with CML were enrolled.The SEER Cancer Statistics Review (http://seer.cancer.gov/data/citation.html), a report on the most recent cancer incidence, mortality, survival, prevalence, and lifetime risk statistics, is published annually by the Data Analysis and Interpretation Branch of the National Cancer Institute, (Bethesda, MD, USA).The SEER data contain no identifiers and are publicly available for studies of cancer-based epidemiology and survival analysis.The National Cancer Institute's SEER*Stat software (Surveillance Research Program, National Cancer Institute SEER*Stat software, www.seer.cancer.gov/seerstat)(Version 8.1.5)was used to identify pediatric patients diagnosed with leukemia based on Site recode International Classification of Diseases for Oncology (ICD-O)-3/WHO 2008.Morphology codes for leukemia were expanded to include the following histologies: 9811, 9814, 9826, 9837, 9846, 9861, 9866, 9867, 9875 and 9920.