Abstract
Linking birth records and cancer registry data from California, we conducted a population-based study with 23,419 cases and 87,593 matched controls born during 1978–2009 to investigate the relationship of parental age to risk of pediatric cancer. Compared with children born to mothers aged 20–24 years, those born to mothers in older age groups had a 13%–36% higher risk of pediatric cancer; the odds ratio for each 5-year increase in maternal age was 1.06 (95% confidence interval (CI): 1.04, 1.09). For cancer diagnosed in children in age groups 0–14 years and 15–19 years, the odds ratios for each 5-year increase in maternal age were 1.05 (95% CI: 1.02, 1.07) and 1.14 (95% CI: 1.09, 1.19), respectively. Having an older father also conferred an increased risk, with an odds ratio for each 5-year increase of 1.03 (95% CI: 1.02, 1.05) for cancer diagnosed at ages 0–19 years and 1.03 (95% CI: 1.02, 1.05) for cancer diagnosed at ages 0–14 years. While advancing maternal age increased risk of leukemia and central nervous system tumors, older paternal age was not associated with risk of either type. Both maternal and paternal older ages were associated with risk of lymphoma. In this large, population-based record-linkage study, advancing parental age, especially advancing maternal age, was associated with higher pediatric cancer risk, with variations across types of cancer.
Keywords: adolescent, case-control study, children, parental age, pediatric cancer
Approximately 16,000 new cases of pediatric cancer (in persons aged 0–19 years) are diagnosed in the United States annually (1), with few well-established risk factors, including certain genetic conditions, ionizing radiation, prior chemotherapy, male predominance, and white race (2–4).
In the United States, the age at which individuals give birth to children has been rising. The average age of first-time mothers was 21.4 years in 1970 (5), 24.9 years in 2000 (5), and 26.3 years in 2014 (6). The average age of fathers at the birth of their first child increased from 25.3 years during 1987–1988 to 29.4 years during 2006–2010 (7). Over the period of 1975–2012, the incidence of pediatric cancer rose at an annual rate of 0.6% (8). Although similar temporal trends in parental age and pediatric cancer incidence do not necessarily suggest an association between the two, it is plausible that parental age plays a role in the etiology of pediatric cancer. Genomic sequencing studies in parent-offspring trios have found a higher number of de novo mutations in the offspring of older parents (9–11), and accumulation of germ-cell mutations may increase cancer risk (12). In addition, older parental age correlated with decreased DNA methylation level in newborns; many of the loci where methylation changes occurred have been linked to oncogenesis (13).
The potential relationship of parental age to risk of pediatric cancer has been examined previously, but the findings are inconsistent (14–21). Although 2 of these studies included over 10,000 cases (14, 15), most existing studies had limited or moderate sample sizes. To elucidate the etiologic role of parental age, we conducted a population- and record-based case-control study with an unprecedented sample size and a low likelihood of bias, evaluating all pediatric cancer as a group as well as by age at diagnosis and types of cancer (e.g., leukemia, central nervous system (CNS) tumors). The racial/ethnic diversity of our study population also enabled us to assess the relationship of parental age to pediatric cancer risk according to race/ethnicity, which had not been evaluated in previous studies.
METHODS
Study population
We linked statewide birth records maintained by the California Department of Public Health for the years of 1978–2009 to cancer diagnosis data for the years of 1988–2011 from the California Cancer Registry (CCR). Beginning in January 1988, California law has required that all new cancer cases diagnosed in state residents be reported to the CCR. CCR data meet all standards of the Surveillance, Epidemiology, and End Results Program and National Program of Cancer Registries for quality, timeliness, and completeness (22). Included in this analysis were children born in California and diagnosed with cancer at the ages of 0–19 years. Children who were born in California during the same period and did not have cancer cases reported to CCR were considered potential controls. For each cancer case, up to 4 controls were randomly selected from the pool of potential controls and matched to the case on birth year and month, sex, and race/ethnicity (non-Hispanic white, non-Hispanic black, Hispanic/Latino, Asian/Pacific Islander, other). The study protocol was approved by the institutional review boards at the California Health and Human Services Agency, University of California (Berkeley and San Francisco), and Yale University.
From the 24,734 cases that were identified, we excluded cases: 1) whose mothers resided outside California at the time of delivery (out of concern that these children might not have been reported to the CCR had they developed pediatric cancer) (n = 18); 2) who were reported as having Down syndrome, an established, strong risk factor for childhood leukemia (23) (n = 85); 3) whose birth records did not include information on maternal or paternal age, the primary exposure of interest (n = 1,094); and 4) who had missing data on other variables that might reflect a less than optimal quality of birth record, including birth weight (n = 10), birth order (n = 58), mode of delivery (vaginal versus cesarean, n = 37), plurality (single or multiple birth, n = 1), or mother's country of birth (n = 11). The final number of cases after exclusions was 23,420. The same exclusion criteria were applied to the 93,667 matched control subjects selected for the 23,420 cases, and 87,593 controls remained after exclusions. One case without matched controls was further excluded. The final sample consisted of 23,419 cases and 87,593 individually matched controls. Each case had at least 1 matched control, and about 77% of cases had 4 matched controls.
Variables of interest
Both maternal and paternal ages (in years) were abstracted from birth records and categorized into 5-year age groups (<20 years, 20–24 years, 25–29 years, 30–34 years, 35–39 years, ≥40 years). To account for potential confounding by other infant and parental characteristics, we also abstracted data on birth weight, length of gestation, birth order, mode of delivery, plurality, maternal country of birth, maternal smoking during pregnancy, maternal history of miscarriage or stillbirth (i.e., fetal loss <20 or ≥20 weeks of gestation), maternal education, and paternal education.
Statistical analysis
Mean values and standard deviations were calculated for parental age. Pearson's χ2 test was used to compare characteristics between cases and controls. Odds ratios and 95% confidence intervals were obtained from conditional logistic regression models. Maternal and paternal ages were modeled as categories (5-year age groups), using the age group of 20–24 years as a referent. Parental ages as continuous variables were used to assess possible trends. Both maternal and paternal ages were included in the models simultaneously. All factors/characteristics listed in Table 1 were considered potential confounding variables in the initial model, but only covariates that had a P value of less than 0.05 after using the SAS (SAS Institute, Inc., Cary, North Carolina) stepwise function were retained in the final model. These included birth weight (grams: <2,500, 2,500–2,999, 3,000–3,499, 3,500–3,999, ≥4,000), length of gestation (weeks: 22–36, 37–41, 42–44, or unknown), birth order (first, second, third or higher), maternal country of birth (United States, foreign countries), maternal smoking during pregnancy (no, yes, or unknown), and maternal and paternal education (up to 8th grade, 9–12th grade, at least some college, or unknown).
Table 1.
Characteristics of the Study Population, California, 1978–2009
| Characteristic | Case Group (n = 23,419) | Control Group (n = 87,593) | P Value | ||
|---|---|---|---|---|---|
| No. of Individuals | %a | No. of Individuals | %a | ||
| Maternal age, years | <0.01 | ||||
| <20 | 2,084 | 8.9 | 8,780 | 10.0 | |
| 20–24 | 5,434 | 23.2 | 22,405 | 25.6 | |
| 25–29 | 7,070 | 30.2 | 25,660 | 29.3 | |
| 30–34 | 5,632 | 24.0 | 19,915 | 22.7 | |
| 35–39 | 2,578 | 11.0 | 8,985 | 10.3 | |
| ≥40 | 621 | 2.7 | 1,848 | 2.1 | |
| Paternal age, years | <0.01 | ||||
| <20 | 817 | 3.5 | 3,655 | 4.2 | |
| 20–24 | 4,094 | 17.5 | 16,800 | 19.2 | |
| 25–29 | 6,454 | 27.6 | 24,512 | 28.0 | |
| 30–34 | 6,180 | 26.4 | 22,254 | 25.4 | |
| 35–39 | 3,692 | 15.8 | 13,119 | 15.0 | |
| ≥40 | 2,182 | 9.3 | 7,253 | 8.3 | |
| Sex | 0.83 | ||||
| Female | 10,557 | 45.1 | 39,418 | 45.0 | |
| Male | 12,862 | 54.9 | 48,175 | 55.0 | |
| Race/ethnicity | 0.40 | ||||
| Non-Hispanic white | 9,851 | 42.1 | 37,325 | 42.6 | |
| Non-Hispanic black | 1,448 | 6.2 | 5,213 | 6.0 | |
| Hispanic | 9,743 | 41.6 | 36,116 | 41.2 | |
| Asian/Pacific Islander | 2,107 | 9.0 | 7,965 | 9.1 | |
| Other | 270 | 1.2 | 974 | 1.1 | |
| Birth weight, g | <0.01 | ||||
| <2,500 | 1,313 | 5.6 | 5,067 | 5.8 | |
| 2,500–2,999 | 8,316 | 35.5 | 32,173 | 36.7 | |
| 3,000–3,499 | 3,261 | 13.9 | 13,151 | 15.0 | |
| 3,500–3,999 | 7,436 | 31.8 | 27,071 | 30.9 | |
| ≥4,000 | 3,093 | 13.2 | 10,131 | 11.6 | |
| Gestational age, weeks | 0.24 | ||||
| 22–36 | 2,251 | 9.6 | 8,085 | 9.2 | |
| 37–41 | 17,585 | 75.1 | 65,891 | 75.2 | |
| 42–44 | 2,118 | 9.0 | 7,946 | 9.1 | |
| Unknown | 1,465 | 6.3 | 5,671 | 6.5 | |
| Birth order | 0.37 | ||||
| First | 9,233 | 39.4 | 34,836 | 39.8 | |
| Second | 7,596 | 32.4 | 27,991 | 32.0 | |
| Third or higher | 6,590 | 28.1 | 24,766 | 28.3 | |
| Mode of delivery | <0.01 | ||||
| Vaginal | 17,735 | 75.7 | 67,401 | 76.9 | |
| Cesarean | 5,684 | 24.3 | 20,192 | 23.1 | |
| Plurality | 0.41 | ||||
| Singleton | 22,874 | 97.7 | 85,473 | 97.6 | |
| Multiple | 545 | 2.3 | 2,120 | 2.4 | |
| Maternal country of birth | <0.01 | ||||
| United States | 14,988 | 64.0 | 54,269 | 62.0 | |
| Foreign | 8,431 | 36.0 | 33,324 | 38.0 | |
| Maternal smoking during pregnancy | <0.01 | ||||
| No | 14,663 | 62.6 | 54,188 | 61.9 | |
| Yes | 239 | 1.0 | 1,077 | 1.2 | |
| Unknown | 8,517 | 36.4 | 32,328 | 36.9 | |
| Maternal history of miscarriage | 0.13 | ||||
| No | 19,451 | 83.1 | 73,129 | 83.5 | |
| Yes | 3,944 | 16.8 | 14,389 | 16.4 | |
| Unknown | 24 | 0.1 | 75 | 0.1 | |
| Maternal history of stillbirth | 0.40 | ||||
| No | 23,026 | 98.3 | 86,180 | 98.4 | |
| Yes | 359 | 1.5 | 1,277 | 1.5 | |
| Unknown | 34 | 0.1 | 136 | 0.2 | |
| Maternal education | <0.01 | ||||
| Up to 8th grade | 1,574 | 6.7 | 7,107 | 8.1 | |
| 9th–12th grade | 7,141 | 30.5 | 26,412 | 30.2 | |
| At least some college | 6,820 | 29.1 | 23,963 | 27.4 | |
| Unknown | 7,884 | 33.7 | 30,111 | 34.4 | |
| Paternal education | <0.01 | ||||
| Up to 8th grade | 1,624 | 6.9 | 7,123 | 8.1 | |
| 9th–12th grade | 6,867 | 29.3 | 25,509 | 29.1 | |
| At least some college | 6,786 | 29.0 | 23,785 | 27.2 | |
| Unknown | 8,142 | 34.8 | 31,176 | 35.6 | |
a Percentages may not add up to 100.0% due to rounding.
In the United States, cancers in the pediatric population have traditionally been categorized in 2 different age groups: 0–14 years and 0–19 years. We conducted the analyses separately for age groups 0–14 years and 15–19 years so our results could be compared with those from previous studies, most of which focused on the younger age group of 0–14 years. Additionally, we conducted analyses in children diagnosed with cancer at the age of 0–5 years (early childhood), because a pooled analysis by Johnson et al. (14) reported a stronger association between advancing maternal age and cancer risk in younger children.
Because the California population is diverse in terms of race and ethnicity, we also conducted subgroup analyses for different racial/ethnic groups, including non-Hispanic white, non-Hispanic black, Hispanic/Latino, and Asian/Pacific Islander. The likelihood ratio test was used to assess whether a statistical interaction existed between race/ethnicity and parental age.
In addition to analyzing pediatric cancer as a group, we conducted separate analyses for 11 major diagnostic groups in the International Classification of Childhood Cancer, Third Edition (24), as well as major subtypes of the 3 most common diagnostic groups: leukemia, lymphoma, and CNS tumors.
For different subgroup analyses, the covariates selected by the SAS (SAS Institute, Inc.) stepwise function were not always the same. We compared the odds ratios for parental age, the primary exposure of interest, from the model that included covariates selected for the analysis of overall pediatric cancer and the model that included covariates selected for any specific subgroup analysis. Because the results were essentially the same, we used the set of covariates selected for overall pediatric cancer for subgroup analyses as well. SAS, version 9.4 (SAS Institute, Inc.), was used for all analyses, and all tests were 2-sided, with a type I error of 5%.
RESULTS
Among 23,419 cases, 10,396 (44.4%), 7,778 (33.2%), and 5,245 (22.4%) were diagnosed at the age of 0–5, 6–14, and 15–19 years, respectively. Compared with controls, cases appeared to have higher birth weight (P < 0.01), be more likely to be delivered by cesarean section (P < 0.01), have US-born mothers (P < 0.01), and have parents with higher education (P < 0.01) (Table 1). Among those with known smoking information, mothers of controls were more likely to have smoked during pregnancy than those of cases (P < 0.01). No differences were observed with regard to length of gestation, birth order, plurality, and maternal history of miscarriage or stillbirth (Table 1).
Pearson's correlation coefficient between maternal and paternal age was 0.74. The mean age of case mothers at delivery (27.6 (standard deviation, 6.0) years) was slightly older than that of control mothers (27.2 (standard deviation, 5.9) years; P < 0.01), and more cases than controls were delivered by mothers who were at least 30 years old (37.7% of cases versus 35.1% of controls; P < 0.01) (Table 1). Similarly, the mean age for fathers of cases (30.3 (standard deviation, 6.8) years) was higher than that of controls (29.8 (standard deviation, 6.8) years; P < 0.01). As shown in Figure 1, both maternal and paternal ages became more advanced over the study period of 1978–2009, and the parents of cases were consistently older than parents of controls across all birth years.
Figure 1.
Mean parental age at delivery, according to year of birth, and pediatric cancer incidence data, California, 1978–2009. Pediatric cancer incidence data were obtained from the California Cancer Registry.
After adjusting for multiple covariates including paternal age, compared with children whose mother were aged 20–24 years at delivery, those born to mothers in the older age groups had a 13%–36% increased risk of pediatric cancer (all P < 0.05), and the odds ratio for each 5-year increase in maternal age was 1.06 (95% confidence interval (CI): 1.04, 1.09; P-trend < 0.01) (Figure 2A, Web Table 1, available at https://academic.oup.com/aje). Similarly, having an older father also conferred an increased risk of pediatric cancer (while adjusting for maternal age and other covariates), but the magnitude of association was smaller than that observed with maternal age; the highest odds ratio was 1.15 (95% CI: 1.07, 1.24) for the paternal age group of ≥40 years. The odds ratio for each 5-year increase in paternal age was 1.03 (95% CI: 1.02, 1.05; P-trend < 0.01) (Figure 2B).
Figure 2.
Adjusted odds ratios (ORs) and 95% confidence intervals (CIs) for parental age and risk of pediatric cancer according to age at diagnosis, California, 1978–2009. A) Maternal age at delivery. B) paternal age at delivery.
Advancing maternal age was associated with pediatric cancer diagnosed at different ages, and the odds ratios for each 5-year increase in maternal age were 1.05 (95% CI: 1.02, 1.07; P-trend < 0.01) for cancer diagnosed in the age group 0–14 years and 1.14 (95% CI: 1.09,1.19; P-trend < 0.01) for cancer diagnosed in the age group 15–19 years (Figure 2A). Advancing paternal age also increased the risk of cancer diagnosed both at the age of 0–14 years (per 5-year increase in paternal age, odds ratio (OR) = 1.03, 95% CI: 1.02, 1.05; P-trend < 0.01) and 15–19 years (per 5-year increase in paternal age, OR = 1.04, 95% CI: 1.00, 1.07; P-trend = 0.03). As for cancer diagnosed in early childhood (age 0–5 years), results regarding parental age were generally comparable to those observed in the diagnostic age group of 0–14 years (Figure 2).
Mean parental ages at delivery differed by racial/ethnic groups, with Asian/Pacific Islander parents having the oldest mean ages at delivery of their children (Table 2). Across all groups, mean ages of cases’ parents were older than those of controls. Pearson's correlation coefficient between maternal and paternal age ranged from 0.69 in Asians/Pacific Islanders to 0.75 in non-Hispanic whites. Among non-Hispanic white children, having an older mother or father elevated the risk of pediatric cancer, with a magnitude of nearly 5% increase in pediatric cancer risk per 5-year increase in parental age (P-trend < 0.01) (Web Table 2). Among non-Hispanic black mothers, being in the age group ≥40 years had a higher likelihood of having offspring with pediatric cancer, but few mothers were in this age group (29 case mothers and 57 control mothers). For Hispanic children, older maternal age also increased their cancer risk; older paternal age did not appear to influence disease risk, although there was indication that having a young father, under the age of 20 years, might confer protection against pediatric cancer. Among Asians/Pacific Islanders, maternal age did not influence pediatric cancer risk. Compared with Asian/Pacific Islander fathers aged 20–24 years, those older than 30 years had a higher likelihood of having offspring with pediatric cancer. The likelihood ratio test for statistical interaction between race/ethnicity and parental age was not significant (P = 0.19 and 0.12 for interaction with maternal and paternal age, respectively).
Table 2.
Mean Values and Standard Deviations for Parental Ages at Delivery According to Race/Ethnicity, California, 1978–2009
| Race/Ethnicity | Maternal Age, years Mean (SD) | Paternal Age, years Mean (SD) | ||
|---|---|---|---|---|
| Case Group | Control Group | Case Group | Control Group | |
| Non-Hispanic white | 28.7 (5.6) | 28.2 (5.7) | 31.4 (6.5) | 30.8 (6.5) |
| Non-Hispanic black | 25.8 (6.1) | 25.4 (5.9) | 29.0 (7.8) | 28.5 (7.4) |
| Hispanic | 26.3 (6) | 25.9 (5.9) | 28.7 (6.7) | 28.4 (6.7) |
| Asian/Pacific Islander | 30.1 (5.3) | 29.6 (5.5) | 33.4 (6.4) | 32.7 (6.5) |
| Other | 28.0 (5.9) | 27.1 (6.1) | 30.7 (6.6) | 29.8 (7.3) |
Abbreviation: SD, standard deviation.
When assessing the association between parental age and risk of specific types of cancer, we found that each 5-year increase in maternal age elevated the risk of pediatric leukemia overall (OR = 1.07, 95% CI: 1.03, 1.11) and acute lymphoblastic leukemia (OR = 1.06, 95% CI: 1.02, 1.10) but not acute myeloid leukemia (Table 3). Similarly, older maternal age also increased the risk of CNS tumors overall (OR = 1.08, 95% CI: 1.03, 1.13), ependymomas and choroid plexus tumor (OR = 1.26, 95% CI: 1.10, 1.45), astrocytomas (OR = 1.09, 95% CI: 1.01, 1.16), renal tumors (OR = 1.15, 95% CI: 1.04, 1.26), and hepatic tumors (OR = 1.19, 95% CI: 1. 01, 1.40). On the other hand, we did not find statistical evidence to reject a null association between paternal age and the risk of leukemia, CNS tumors, renal tumors, or hepatic tumors in our study. Older maternal and paternal ages were both associated with risk of pediatric lymphoma overall and Hodgkin lymphoma. Only older paternal age increased risk of non-Hodgkin lymphoma. No associations were observed between parental age and other types of cancer (Table 3).
Table 3.
Adjusted Odds Ratios and 95% Confidence Intervals for Parental Age and Risk of Pediatric Cancers Among Children Aged 0–19 Years, California, 1978–2009
| Cancer Type and Parental Age Group, years | Maternal Age | Paternal Age | ||||||||
|---|---|---|---|---|---|---|---|---|---|---|
| No. of Cases | No. of Controls | ORa | 95% CI | P Value | No. of Cases | No. of Controls | ORa | 95% CI | P Value | |
| Leukemias, myeloproliferative diseases, and myelodysplastic diseases | ||||||||||
| All | ||||||||||
| <20 | 673 | 2,754 | 0.97 | 0.87, 1.09 | 0.62 | 283 | 1,165 | 0.98 | 0.84, 1.14 | 0.79 |
| 20–24 | 1,689 | 6,809 | 1.00 | Referent | 1,264 | 5,138 | 1.00 | Referent | ||
| 25–29 | 2,082 | 7,750 | 1.08 | 1.00, 1.17 | 0.06 | 1,988 | 7,503 | 1.06 | 0.97, 1.16 | 0.20 |
| 30–34 | 1,682 | 6,038 | 1.13 | 1.02, 1.24 | 0.02 | 1,819 | 6,714 | 1.06 | 0.96, 1.17 | 0.28 |
| 35–39 | 829 | 2,869 | 1.17 | 1.04, 1.32 | 0.01 | 1,145 | 4,020 | 1.09 | 0.97, 1.22 | 0.15 |
| ≥40 | 225 | 603 | 1.53 | 1.26, 1.84 | <0.01 | 681 | 2,283 | 1.10 | 0.96, 1.26 | 0.17 |
| Per 5-year increase in age | 1.07 | 1.03, 1.11 | <0.01 | 1.02 | 0.99, 1.05 | 0.16 | ||||
| Acute lymphoblastic leukemia | ||||||||||
| <20 | 531 | 2,174 | 0.99 | 0.87, 1.13 | 0.93 | 223 | 917 | 0.99 | 0.83, 1.18 | 0.88 |
| 20–24 | 1,293 | 5,340 | 1.00 | Referent | 980 | 4,043 | 1.00 | Referent | ||
| 25–29 | 1,666 | 6,063 | 1.13 | 1.03, 1.24 | <0.01 | 1,550 | 5,844 | 1.07 | 0.97, 1.18 | 0.20 |
| 30–34 | 1,334 | 4,728 | 1.16 | 1.04, 1.30 | <0.01 | 1,455 | 5,299 | 1.08 | 0.96, 1.21 | 0.21 |
| 35–39 | 657 | 2,309 | 1.17 | 1.02, 1.34 | 0.02 | 898 | 3,203 | 1.09 | 0.95, 1.24 | 0.22 |
| ≥40 | 164 | 477 | 1.40 | 1.13, 1.74 | <0.01 | 539 | 1,785 | 1.16 | 0.99, 1.35 | 0.06 |
| Per 5-year increase in age | 1.06 | 1.02, 1.10 | <0.01 | 1.03 | 0.99, 1.06 | 0.10 | ||||
| Acute myeloid leukemia | ||||||||||
| <20 | 108 | 410 | 1.01 | 0.75, 1.34 | 0.97 | 43 | 179 | 0.89 | 0.59, 1.34 | 0.58 |
| 20–24 | 292 | 1,070 | 1.00 | Referent | 207 | 798 | 1.00 | Referent | ||
| 25–29 | 297 | 1,219 | 0.90 | 0.73, 1.10 | 0.29 | 325 | 1,202 | 1.09 | 0.87, 1.36 | 0.45 |
| 30–34 | 248 | 932 | 0.99 | 0.78, 1.26 | 0.93 | 260 | 1,003 | 1.03 | 0.79, 1.33 | 0.84 |
| 35–39 | 122 | 399 | 1.16 | 0.85, 1.59 | 0.36 | 170 | 592 | 1.07 | 0.79, 1.44 | 0.67 |
| ≥40 | 39 | 94 | 1.61 | 1.00, 2.58 | 0.05 | 101 | 350 | 0.97 | 0.68, 1.39 | 0.89 |
| Per 5-year increase in age | 1.08 | 0.99, 1.18 | 0.09 | 1.00 | 0.93, 1.08 | 0.95 | ||||
| Lymphomas and reticuloendothelial neoplasms | ||||||||||
| All | ||||||||||
| <20 | 231 | 1,096 | 0.96 | 0.79, 1.16 | 0.65 | 92 | 426 | 1.05 | 0.80, 1.38 | 0.72 |
| 20–24 | 615 | 2,767 | 1.00 | Referent | 446 | 2,139 | 1.00 | Referent | ||
| 25–29 | 924 | 3,082 | 1.26 | 1.11, 1.44 | <0.01 | 790 | 2,962 | 1.18 | 1.02, 1.36 | 0.03 |
| 30–34 | 671 | 2,302 | 1.18 | 1.01, 1.38 | 0.03 | 747 | 2,641 | 1.24 | 1.06, 1.46 | <0.01 |
| 35–39 | 260 | 950 | 1.07 | 0.87, 1.31 | 0.54 | 436 | 1,432 | 1.41 | 1.17, 1.70 | <0.01 |
| ≥40 | 59 | 168 | 1.37 | 0.97, 1.94 | 0.07 | 249 | 765 | 1.51 | 1.21, 1.88 | <0.01 |
| Per 5-year increase in age | 1.07 | 1.01, 1.13 | 0.02 | 1.08 | 1.03, 1.13 | <0.01 | ||||
| Hodgkin lymphoma | ||||||||||
| <20 | 101 | 534 | 0.85 | 0.64, 1.14 | 0.28 | 46 | 208 | 1.32 | 0.89, 1.96 | 0.16 |
| 20–24 | 276 | 1,281 | 1.00 | Referent | 188 | 1,012 | 1.00 | Referent | ||
| 25–29 | 423 | 1,431 | 1.31 | 1.09, 1.59 | <0.01 | 382 | 1,356 | 1.34 | 1.08, 1.66 | <0.01 |
| 30–34 | 311 | 1,007 | 1.36 | 1.08, 1.71 | <0.01 | 322 | 1,223 | 1.21 | 0.95, 1.55 | 0.12 |
| 35–39 | 119 | 422 | 1.17 | 0.87, 1.58 | 0.31 | 216 | 610 | 1.72 | 1.30, 2.27 | <0.01 |
| ≥40 | 28 | 65 | 1.94 | 1.15, 3.27 | 0.01 | 104 | 331 | 1.42 | 1.01, 1.98 | 0.04 |
| Per 5-year increase in age | 1.15 | 1.06, 1.25 | <0.01 | 1.07 | 1.00, 1.15 | 0.04 | ||||
| Non-Hodgkin lymphomas (except Burkitt lymphoma) | ||||||||||
| <20 | 92 | 395 | 1.04 | 0.77, 1.42 | 0.79 | 37 | 153 | 1.00 | 0.65, 1.55 | 1.00 |
| 20–24 | 221 | 1,003 | 1.00 | Referent | 166 | 767 | 1.00 | Referent | ||
| 25–29 | 342 | 1,084 | 1.33 | 1.07, 1.65 | <0.01 | 269 | 1,048 | 1.08 | 0.85, 1.38 | 0.53 |
| 30–34 | 214 | 815 | 1.04 | 0.80, 1.35 | 0.76 | 280 | 931 | 1.33 | 1.02, 1.74 | 0.04 |
| 35–39 | 86 | 318 | 1.03 | 0.72, 1.46 | 0.88 | 133 | 514 | 1.21 | 0.87, 1.67 | 0.25 |
| ≥40 | 21 | 59 | 1.24 | 0.69, 2.25 | 0.47 | 91 | 261 | 1.67 | 1.15, 2.42 | <0.01 |
| Per 5-year increase in age | 1.03 | 0.93, 1.13 | 0.57 | 1.10 | 1.02, 1.19 | 0.01 | ||||
| Central nervous system and miscellaneous intracranial and intraspinal neoplasms | ||||||||||
| All | ||||||||||
| <20 | 407 | 1,647 | 1.08 | 0.93, 1.25 | 0.33 | 158 | 681 | 0.91 | 0.74, 1.12 | 0.39 |
| 20–24 | 978 | 4,285 | 1.00 | Referent | 758 | 3,113 | 1.00 | Referent | ||
| 25–29 | 1,408 | 5,038 | 1.24 | 1.11, 1.37 | <0.01 | 1,231 | 4,768 | 1.00 | 0.89, 1.13 | 0.96 |
| 30–34 | 1,100 | 3,986 | 1.22 | 1.08, 1.38 | <0.01 | 1,261 | 4,468 | 1.06 | 0.93, 1.20 | 0.40 |
| 35–39 | 556 | 1,801 | 1.39 | 1.20, 1.62 | <0.01 | 742 | 2,654 | 1.01 | 0.87, 1.17 | 0.92 |
| ≥40 | 133 | 397 | 1.51 | 1.19, 1.92 | <0.01 | 432 | 1,470 | 1.03 | 0.86, 1.22 | 0.76 |
| Per 5-year increase in age | 1.08 | 1.03, 1.13 | <0.01 | 1.02 | 0.98, 1.05 | 0.32 | ||||
| Ependymomas and choroid plexus tumor | ||||||||||
| <20 | 40 | 135 | 1.64 | 0.99, 2.72 | 0.06 | 20 | 71 | 1.14 | 0.60, 2.14 | 0.69 |
| 20–24 | 78 | 429 | 1.00 | Referent | 57 | 285 | 1.00 | Referent | ||
| 25–29 | 125 | 488 | 1.41 | 0.99, 2.02 | 0.06 | 122 | 451 | 1.27 | 0.85, 1.91 | 0.24 |
| 30–34 | 120 | 382 | 1.93 | 1.29, 2.89 | <0.01 | 126 | 423 | 1.17 | 0.75, 1.83 | 0.48 |
| 35–39 | 74 | 204 | 2.57 | 1.60, 4.12 | <0.01 | 82 | 291 | 0.95 | 0.57, 1.55 | 0.82 |
| ≥40 | 19 | 39 | 3.71 | 1.77, 7.76 | <0.01 | 49 | 156 | 0.97 | 0.56, 1.69 | 0.92 |
| Per 5-year increase in age | 1.26 | 1.10, 1.45 | <0.01 | 0.99 | 0.89, 1.11 | 0.88 | ||||
| Astrocytomas | ||||||||||
| <20 | 157 | 714 | 0.92 | 0.73, 1.16 | 0.47 | 59 | 296 | 0.82 | 0.59, 1.14 | 0.24 |
| 20–24 | 433 | 1,800 | 1.00 | Referent | 333 | 1,302 | 1.00 | Referent | ||
| 25–29 | 609 | 2,166 | 1.20 | 1.02, 1.40 | 0.03 | 531 | 2,056 | 0.93 | 0.78, 1.10 | 0.39 |
| 30–34 | 484 | 1,745 | 1.16 | 0.96, 1.40 | 0.12 | 531 | 1,937 | 0.94 | 0.77, 1.14 | 0.52 |
| 35–39 | 215 | 737 | 1.25 | 0.98, 1.58 | 0.07 | 332 | 1,090 | 1.02 | 0.81, 1.28 | 0.85 |
| ≥40 | 52 | 173 | 1.38 | 0.95, 2.01 | 0.09 | 164 | 654 | 0.82 | 0.62, 1.07 | 0.15 |
| Per 5-year increase in age | 1.09 | 1.01, 1.16 | 0.02 | 1.00 | 0.95, 1.06 | 0.99 | ||||
| Intracranial and intraspinal embryonal tumors | ||||||||||
| <20 | 102 | 383 | 0.98 | 0.71, 1.34 | 0.88 | 47 | 164 | 1.06 | 0.71, 1.59 | 0.76 |
| 20–24 | 230 | 916 | 1.00 | Referent | 172 | 667 | 1.00 | Referent | ||
| 25–29 | 313 | 1,095 | 1.18 | 0.95, 1.47 | 0.13 | 270 | 1,027 | 0.99 | 0.78, 1.27 | 0.95 |
| 30–34 | 229 | 869 | 1.09 | 0.84, 1.41 | 0.54 | 287 | 1,007 | 1.09 | 0.83, 1.44 | 0.52 |
| 35–39 | 103 | 422 | 1.01 | 0.72, 1.40 | 0.97 | 136 | 615 | 0.89 | 0.64, 1.23 | 0.48 |
| ≥40 | 25 | 82 | 1.17 | 0.69, 2.01 | 0.56 | 90 | 287 | 1.29 | 0.88, 1.89 | 0.19 |
| Per 5-year increase in age | 1.02 | 0.93, 1.12 | 0.70 | 1.03 | 0.96, 1.12 | 0.41 | ||||
| Other gliomas | ||||||||||
| <20 | 55 | 190 | 1.51 | 1.00, 2.28 | 0.05 | 16 | 76 | 0.71 | 0.38, 1.33 | 0.29 |
| 20–24 | 115 | 568 | 1.00 | Referent | 104 | 398 | 1.00 | Referent | ||
| 25–29 | 167 | 638 | 1.28 | 0.94, 1.73 | 0.11 | 130 | 602 | 0.82 | 0.59, 1.14 | 0.23 |
| 30–34 | 133 | 461 | 1.27 | 0.89, 1.81 | 0.18 | 154 | 555 | 0.97 | 0.67, 1.39 | 0.86 |
| 35–39 | 81 | 223 | 1.49 | 0.97, 2.28 | 0.07 | 105 | 329 | 1.09 | 0.72, 1.64 | 0.70 |
| ≥40 | 22 | 56 | 1.57 | 0.83, 2.95 | 0.16 | 64 | 176 | 1.14 | 0.71, 1.85 | 0.59 |
| Per 5-year increase in age | 1.03 | 0.91, 1.16 | 0.69 | 1.05 | 0.95, 1.16 | 0.30 | ||||
| Other specified intracranial and intraspinal neoplasms | ||||||||||
| <20 | 48 | 212 | 1.05 | 0.68, 1.62 | 0.82 | 16 | 72 | 1.05 | 0.55, 2.02 | 0.88 |
| 20–24 | 113 | 533 | 1.00 | Referent | 86 | 425 | 1.00 | Referent | ||
| 25–29 | 182 | 603 | 1.30 | 0.97, 1.75 | 0.08 | 166 | 591 | 1.32 | 0.95, 1.84 | 0.10 |
| 30–34 | 122 | 488 | 1.12 | 0.79, 1.59 | 0.53 | 144 | 505 | 1.36 | 0.94, 1.98 | 0.10 |
| 35–39 | 78 | 198 | 1.83 | 1.19, 2.81 | <0.01 | 84 | 302 | 1.28 | 0.83, 1.97 | 0.27 |
| ≥40 | 15 | 41 | 1.65 | 0.82, 3.33 | 0.16 | 62 | 180 | 1.48 | 0.92, 2.39 | 0.11 |
| Per 5-year increase in age | 1.11 | 0.98, 1.26 | 0.09 | 1.05 | 0.95, 1.16 | 0.32 | ||||
| Neuroblastoma and other peripheral nervous cell tumors | ||||||||||
| <20 | 111 | 380 | 1.35 | 1.01, 1.81 | 0.04 | 39 | 191 | 0.62 | 0.41, 0.94 | 0.03 |
| 20–24 | 253 | 1,021 | 1.00 | Referent | 196 | 735 | 1.00 | Referent | ||
| 25–29 | 359 | 1,309 | 1.06 | 0.87, 1.30 | 0.55 | 309 | 1,183 | 1.05 | 0.84, 1.32 | 0.66 |
| 30–34 | 335 | 1,181 | 1.05 | 0.84, 1.33 | 0.66 | 349 | 1,216 | 1.21 | 0.94, 1.56 | 0.13 |
| 35–39 | 139 | 610 | 0.84 | 0.63, 1.12 | 0.24 | 210 | 823 | 1.14 | 0.86, 1.50 | 0.38 |
| ≥40 | 36 | 116 | 1.17 | 0.74, 1.84 | 0.49 | 130 | 469 | 1.28 | 0.93, 1.76 | 0.13 |
| Per 5-year increase in age | 0.98 | 0.90, 1.06 | 0.61 | 1.07 | 1.00, 1.14 | 0.05 | ||||
| Retinoblastoma | ||||||||||
| <20 | 57 | 232 | 0.88 | 0.60, 1.31 | 0.54 | 22 | 86 | 0.92 | 0.53, 1.60 | 0.78 |
| 20–24 | 139 | 513 | 1.00 | Referent | 106 | 368 | 1.00 | Referent | ||
| 25–29 | 177 | 613 | 1.04 | 0.78, 1.39 | 0.77 | 139 | 606 | 0.77 | 0.56, 1.07 | 0.12 |
| 30–34 | 131 | 501 | 0.85 | 0.61, 1.19 | 0.35 | 154 | 547 | 0.98 | 0.68, 1.41 | 0.92 |
| 35–39 | 64 | 261 | 0.74 | 0.48, 1.13 | 0.16 | 98 | 366 | 1.01 | 0.67, 1.52 | 0.96 |
| ≥40 | 22 | 71 | 0.88 | 0.48, 1.62 | 0.68 | 71 | 218 | 1.26 | 0.79, 2.00 | 0.34 |
| Per 5-year increase in age | 0.99 | 0.88, 1.11 | 0.83 | 1.07 | 0.97, 1.18 | 0.15 | ||||
| Renal tumors | ||||||||||
| <20 | 83 | 349 | 1.10 | 0.79, 1.52 | 0.59 | 34 | 156 | 0.88 | 0.56, 1.38 | 0.58 |
| 20–24 | 204 | 955 | 1.00 | Referent | 161 | 731 | 1.00 | Referent | ||
| 25–29 | 288 | 1,043 | 1.36 | 1.09, 1.69 | <0.01 | 282 | 969 | 1.24 | 0.97, 1.58 | 0.08 |
| 30–34 | 294 | 868 | 1.74 | 1.35, 2.25 | <0.01 | 249 | 940 | 1.02 | 0.77, 1.34 | 0.91 |
| 35–39 | 112 | 407 | 1.40 | 1.01, 1.94 | 0.05 | 174 | 555 | 1.22 | 0.89, 1.67 | 0.23 |
| ≥40 | 25 | 90 | 1.46 | 0.86, 2.49 | 0.16 | 106 | 361 | 1.13 | 0.79, 1.62 | 0.49 |
| Per 5-year increase in age | 1.15 | 1.04, 1.26 | <0.01 | 1.00 | 0.93, 1.08 | 0.97 | ||||
| Hepatic tumors | ||||||||||
| <20 | 38 | 118 | 1.80 | 1.06, 3.05 | 0.03 | 12 | 44 | 0.86 | 0.40, 1.85 | 0.70 |
| 20–24 | 48 | 269 | 1.00 | Referent | 48 | 209 | 1.00 | Referent | ||
| 25–29 | 86 | 344 | 1.49 | 0.95, 2.33 | 0.09 | 83 | 309 | 1.10 | 0.69, 1.76 | 0.69 |
| 30–34 | 93 | 305 | 2.04 | 1.23, 3.39 | <0.01 | 89 | 326 | 0.97 | 0.58, 1.64 | 0.92 |
| 35–39 | 50 | 146 | 2.35 | 1.27, 4.36 | <0.01 | 50 | 211 | 0.70 | 0.38, 1.29 | 0.25 |
| ≥40 | 12 | 33 | 2.22 | 0.91, 5.38 | 0.08 | 45 | 116 | 1.09 | 0.57, 2.09 | 0.78 |
| Per 5-year increase in age | 1.19 | 1.01, 1.40 | 0.04 | 1.03 | 0.91, 1.17 | 0.66 | ||||
| Malignant bone tumors | ||||||||||
| <20 | 91 | 416 | 0.83 | 0.62, 1.12 | 0.23 | 31 | 169 | 0.74 | 0.48, 1.16 | 0.19 |
| 20–24 | 297 | 1,066 | 1.00 | Referent | 219 | 822 | 1.00 | Referent | ||
| 25–29 | 298 | 1,143 | 0.91 | 0.74, 1.12 | 0.38 | 286 | 1,094 | 1.01 | 0.81, 1.27 | 0.93 |
| 30–34 | 222 | 846 | 0.93 | 0.72, 1.19 | 0.54 | 263 | 911 | 1.13 | 0.87, 1.47 | 0.35 |
| 35–39 | 96 | 306 | 1.16 | 0.84, 1.62 | 0.37 | 141 | 568 | 0.93 | 0.68, 1.26 | 0.63 |
| ≥40 | 16 | 66 | 0.85 | 0.45, 1.62 | 0.62 | 80 | 279 | 1.10 | 0.76, 1.60 | 0.60 |
| Per 5-year increase in age | 1.02 | 0.93, 1.13 | 0.63 | 1.00 | 0.93, 1.08 | 0.94 | ||||
| Soft tissue and other extraosseous sarcomas | ||||||||||
| <20 | 134 | 571 | 0.97 | 0.76, 1.26 | 0.84 | 52 | 251 | 0.78 | 0.55, 1.12 | 0.18 |
| 20–24 | 373 | 1,452 | 1.00 | Referent | 279 | 1,102 | 1.00 | Referent | ||
| 25–29 | 435 | 1,643 | 1.03 | 0.86, 1.23 | 0.76 | 411 | 1,558 | 1.06 | 0.88, 1.28 | 0.56 |
| 30–34 | 367 | 1,259 | 1.13 | 0.92, 1.39 | 0.25 | 405 | 1,422 | 1.11 | 0.90, 1.38 | 0.34 |
| 35–39 | 154 | 560 | 1.06 | 0.81, 1.39 | 0.67 | 200 | 797 | 1.00 | 0.78, 1.29 | 0.99 |
| ≥40 | 25 | 91 | 1.01 | 0.61, 1.67 | 0.96 | 141 | 446 | 1.28 | 0.95, 1.73 | 0.10 |
| Per 5-year increase in age | 1.03 | 0.95, 1.11 | 0.48 | 1.05 | 0.99, 1.12 | 0.09 | ||||
| Germ-cell tumors, trophoblastic tumors, and neoplasms of gonads | ||||||||||
| <20 | 149 | 585 | 0.96 | 0.76, 1.22 | 0.75 | 48 | 223 | 0.76 | 0.53, 1.10 | 0.15 |
| 20–24 | 389 | 1,475 | 1.00 | Referent | 315 | 1,124 | 1.00 | Referent | ||
| 25–29 | 418 | 1,609 | 1.03 | 0.86, 1.23 | 0.75 | 396 | 1,555 | 0.92 | 0.76, 1.11 | 0.39 |
| 30–34 | 325 | 1,168 | 1.11 | 0.89, 1.37 | 0.36 | 365 | 1,371 | 0.94 | 0.75, 1.16 | 0.55 |
| 35–39 | 136 | 471 | 1.13 | 0.85, 1.50 | 0.40 | 208 | 774 | 0.94 | 0.72, 1.21 | 0.62 |
| ≥40 | 33 | 105 | 1.16 | 0.73, 1.85 | 0.52 | 118 | 366 | 1.13 | 0.83, 1.53 | 0.45 |
| Per 5-year increase in age | 1.07 | 0.99, 1.15 | 0.10 | 1.04 | 0.98, 1.11 | 0.18 | ||||
| Other malignant epithelial neoplasms and malignant melanomas | ||||||||||
| <20 | 98 | 569 | 0.73 | 0.56, 0.96 | 0.03 | 43 | 235 | 0.99 | 0.67, 1.45 | 0.94 |
| 20–24 | 402 | 1,598 | 1.00 | Referent | 268 | 1,179 | 1.00 | Referent | ||
| 25–29 | 533 | 1,889 | 1.03 | 0.87, 1.21 | 0.77 | 480 | 1,830 | 1.07 | 0.89, 1.28 | 0.50 |
| 30–34 | 378 | 1,340 | 0.96 | 0.79, 1.18 | 0.72 | 432 | 1,532 | 1.13 | 0.91, 1.40 | 0.26 |
| 35–39 | 162 | 534 | 1.02 | 0.78, 1.32 | 0.89 | 265 | 820 | 1.33 | 1.04, 1.70 | 0.02 |
| ≥40 | 31 | 99 | 1.14 | 0.71, 1.82 | 0.58 | 116 | 433 | 1.07 | 0.79, 1.44 | 0.67 |
| Per 5-year increase in age | 1.06 | 0.98, 1.14 | 0.16 | 1.05 | 0.99, 1.12 | 0.11 | ||||
Abbreviations: CI, confidence interval; OR, odds ratio.
a Odds ratios and 95% confidence intervals were derived from multivariable conditional logistic regression models that included maternal and paternal ages, birth weight (in grams: <2,500, 2,500–2,999, 3,000–3,499, 3,500–3,999, ≥4,000), length of gestation (in weeks: 22–36, 37–41, 42–44, or unknown), birth order (first, second, third or higher), maternal country of birth (United States, foreign countries), maternal smoking during pregnancy (no, yes, or unknown), and maternal and paternal education (up to 8th grade, 9–12th grade, at least some college, or unknown).
DISCUSSION
In this largest-to-date, population-based record-linkage study that is less prone to bias, advancing parental age, especially advancing maternal age, was associated with increased risk of pediatric cancer. Compared with previous studies that focused on the age group of 0–14 years, our study also included cases diagnosed at the ages of 15–19 years; in this group, there was a larger magnitude of association between maternal age and cancer risk. We also observed variations across different cancer sites.
Previously, a pooled analysis of population-based record-linkage studies reported that each 5-year increase in maternal age raised risks of overall childhood cancer, leukemia, and CNS tumors by 7%–8% (14). The pooled analysis also found that, for childhood cancer overall, the risk associated with each 5-year increase in maternal age was higher among children diagnosed at ages 0–4 years (OR = 1.08) than among those diagnosed in age group 5–9 years (OR = 1.05) or 10–14 years (OR = 1.04). In our study, the comparable odds ratios were 1.05, 1.05, and 1.03, respectively. In a Swedish study, investigators observed a nearly 50% higher risk of childhood leukemia, but not brain tumors, among the children of mothers aged over 35 years (25). Emerson et al. (16) reported that having a mother over 35 years of age was associated with an increased risk of astrocytoma (but not overall brain tumor or other subtypes of CNS tumors) in children ≤10 years of age. Similar to our study, in a large case-control study in Great Britain, investigators observed an association of maternal age with childhood acute lymphoblastic leukemia but not acute myeloid leukemia (15). Using a study design similar to ours, Oksuzyan et al. (19) observed a slight but not significant increased risk (OR = 1.14, 95% CI: 0.88, 1.49) of childhood leukemia among mothers aged over 35 years in a California-based study that included 5,788 cases born in 1986–2007 and diagnosed in 1988–2008. Our study included subjects who were born as early as 1978 (7,180 leukemia cases) and matched controls by race/ethnicity. Hispanic ethnicity accounts for more than 30% of California population (26) and nearly 50% of pediatric leukemia cases in our study. Hispanic mothers are younger at first delivery than mothers from other racial/ethnic groups (6). In the study by Oksuzyan et al. (19), it was possible for a Hispanic case to have a non-Hispanic control, which may have narrowed the apparent difference in maternal age between cases and controls when compared with our study. Many other studies did not observe associations between maternal age and childhood leukemia (17, 21, 27–30) or CNS tumors (21, 31, 32). Because the magnitude of association we observed between maternal age and pediatric cancer risk was not substantial, moderate sample size and limited statistical power from many existing studies could have contributed to inconsistent findings in the literature. Null findings, especially those from studies with relatively small sample sizes, probably should not be interpreted as providing evidence against an etiological role of advancing maternal age.
In our study, older paternal age also increased pediatric cancer risk, but no association was observed for most specific types of cancer. Because maternal and paternal ages were highly correlated, we also fitted models with paternal age alone, in which advancing paternal age increased the risk of leukemia overall, acute lymphoblastic leukemia, CNS tumors overall, ependymomas and choroid plexus tumor, and other specified intracranial and intraspinal neoplasms (detailed results not presented). This indicated that the role of paternal age may not be independent of maternal age, which is consistent with the finding of a large pooled analysis (14). A few previous studies did not observe an association between older paternal age and childhood leukemia risk (17, 25, 27, 28), while others did (15, 18–20). As for CNS tumors, a positive association with advancing paternal age was observed in 2 Swedish studies (21, 25) but not in other studies (31–33). In addition to possible differences in sample size, whether maternal age was accounted for in the analysis of paternal age might have contributed to inconsistent findings.
The mechanism(s) through which parental age influences pediatric cancer risk are uncertain. Adkins et al. (13) reported that advancing parental age was correlated with decreased DNA methylation level in newborns; the association with maternal age was stronger and involved more sites. In addition, many of the loci where methylation changes occurred have been linked to oncogenesis (13). In addition, maternal age is correlated with aneuploidy in the offspring (34–36), including Down syndrome, and children with such birth defects have a higher risk of developing childhood cancers (37–40). It is also possible that advancing maternal age could be a marker for other factors, such as contaminant levels in maternal blood (41) and breastmilk (42) and age-related changes in hormonal levels during pregnancy (43, 44). Furthermore, genomic sequencing studies revealed a higher frequency of de novo mutations in the offspring as parents (especially fathers) aged (9–11). A majority of paternal age-related germline mutations have also been reported as somatic oncogenic mutations for various tumor types (45). Recently, Mills et al. (46) reported a weak paternal age association for retinoblastoma resulting from de novo germline mutations. Overall, accumulation of mutations may increase cancer risk (12, 45, 46).
Strengths of the current study include an unprecedented sample size and improved statistical power, population-based ascertainment of cases and controls that minimizes selection bias, reliance on data from preexisting birth and cancer registry records (therefore minimizing information bias), adjustment for multiple covariates to reduce confounding, and the racial/ethnic diversity of the study population. In addition, this is a single-site study with uniform criteria for subject selection and data abstraction, therefore bypassing challenges inherent in pooled analyses (14) that stem from heterogeneity across individual studies. Furthermore, we included many adolescents aged 15–19 years, who are understudied and underserved in cancer research (47, 48).
The record-linkage design also gave rise to some important limitations. First, we were restricted to the use of existing data without verification. While errors in existing records are definitely possible, most birth variables included in our analysis are considered accurate (49), and any deviation from truth is unlikely to be differential by case/control status. Second, data on parental education were missing for a rather large proportion of the subjects because the data elements were not always included in California birth records. To address this issue, we conducted sensitivity analyses by excluding subjects with unknown covariate information and the primary results remained the same (detailed data not presented). However, there could still be residual confounding due to missing information on parental education or other markers of socioeconomic status (e.g., household income). Third, some controls could have moved out of California and developed pediatric cancer elsewhere, or they could have developed pediatric cancer within California but were not captured by the CCR. Under the extreme assumption that all controls were lost to follow-up, we would expect about 138 cases of cancer to arise from the 87,593 controls with 800,101 person-years of follow-up, at an incidence rate of 17.3 per 100,000 person-years (50). Relative to the 23,419 cases included in our study, it is unlikely that 138 cases would have biased our results to a measurable degree. Fourth, we were unable to ascertain cancer cases in children who were born in 1978–1987 and diagnosed with cancer over the same time period, most of whom would have been very young at cancer diagnosis. Because the control subjects were matched to cases on year of birth, and we conducted separate analyses for cases diagnosed at 0–5 years, the association is likely on precision (i.e., statistical power) and not on internal validity. Fifth, in this study we were able to include only adolescents who were born before 1996. As parental age has continued to rise, further study with adolescents born more recently is needed. Last, we were unable to account for other putative risk factors of pediatric cancer, such as ionizing radiation, although we did adjust for many covariates.
In summary, in this large population-based study, older parental age, especially older maternal age, increased the risk of pediatric cancer, with variations across different ages of diagnosis, and specific cancer sites. Future studies are needed to clarify the underlying biological mechanisms, given the growing number of children born to older parents in the United States and worldwide.
Supplementary Material
ACKNOWLEDGMENTS
Author affiliations: Department of Chronic Disease Epidemiology, Yale School of Public Health, New Haven, Connecticut (Rong Wang, Andrew T. DeWan, Xiaomei Ma); Yale Comprehensive Cancer Center, Yale School of Medicine, New Haven, Connecticut (Rong Wang, Andrew T. DeWan, Xiaomei Ma); Division of Epidemiology, School of Public Health, University of California at Berkeley, Berkeley, California (Catherine Metayer, Libby Morimoto, Alice Kang); Department of Epidemiology and Biostatistics, University of California at San Francisco, San Francisco, California (Joseph L. Wiemels); and Genetic Disease Screening Program, California Department of Public Health, Richmond, California (Juan Yang).
This work was supported by research grants from the National Institutes of Health (R01 CA155461, R01 CA175737, and P01 ES018172) and the US Environmental Protection Agency (RD83451101). The collection of cancer incidence data used in this study was supported by the California Department of Public Health as part of the statewide cancer reporting program mandated by California Health and Safety Code Section 103885; the National Cancer Institute's Surveillance, Epidemiology, and End Results Program (contract HHSN261201000140C awarded to the Cancer Prevention Institute of California, contract HHSN261201000035C awarded to the University of Southern California, and contract HHSN261201000034C awarded to the Public Health Institute); and the Centers for Disease Control and Prevention's National Program of Cancer Registries (agreement U58DP003862-01 awarded to the California Department of Public Health).
The ideas and opinions expressed herein are those of the author(s), and endorsement by the State of California, Department of Public Health, the National Cancer Institute, and the Centers for Disease Control and Prevention or their Contractors and Subcontractors is neither intended nor should be inferred.
Conflict of interest: none declared.
REFERENCES
- 1. Ward E, DeSantis C, Robbins A, et al. Childhood and adolescent cancer statistics, 2014. CA Cancer J Clin. 2014;64(2):83–103. [DOI] [PubMed] [Google Scholar]
- 2. Stiller CA. Epidemiology and genetics of childhood cancer. Oncogene. 2004;23(38):6429–6444. [DOI] [PubMed] [Google Scholar]
- 3. Spector LG, Pankratz N, Marcotte EL. Genetic and nongenetic risk factors for childhood cancer. Pediatr Clin North Am. 2015;62(1):11–25. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 4. Stiller C. Epidemiology of cancer in adolescents. Med Pediatr Oncol. 2002;39(3):149–155. [DOI] [PubMed] [Google Scholar]
- 5. Mathews TJ, Hamilton BE. Mean age of mother, 1970–2000. Natl Vital Stat Rep. 2002;51(1):1–13. [PubMed] [Google Scholar]
- 6. Mathews TJ, Hamilton BE. Mean age of mothers is on the rise: United States, 2000–2014. NCHS Data Brief. 2016;(232):1–8. [PubMed] [Google Scholar]
- 7. Stykes J. Fatherhood in the US: Men's Age at First Birth, 1987–2010 (FP-11-04). Bowling Green, OH: National Center for Family and Marriage Research; 2011. https://www.bgsu.edu/content/dam/BGSU/college-of-arts-and-sciences/NCFMR/documents/FP/FP-11-04.pdf Accessed November18, 2016. [Google Scholar]
- 8. Howlader N, Noone AM, Krapcho M, et al. SEER Cancer Statistics Review, 1975–2012, National Cancer Institute. Bethesda, MD: Based on November 2014 SEER data submission. http://seer.cancer.gov/csr/1975_2012/ Published April2015. Accessed September27, 2016. [Google Scholar]
- 9. Kong A, Frigge ML, Masson G, et al. Rate of de novo mutations and the importance of father's age to disease risk. Nature. 2012;488(7412):471–475. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 10. Wong WS, Solomon BD, Bodian DL, et al. New observations on maternal age effect on germline de novo mutations. Nat Commun. 2016;7:10486. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 11. Rahbari R, Wuster A, Lindsay SJ, et al. Timing, rates and spectra of human germline mutation. Nat Genet. 2016;48(2):126–133. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 12. Wilcox AJ, Sandler DP, Everson RB. Using father's age to explore the role of germ cell mutation as a cause of human cancer. Int J Epidemiol. 1988;17(2):469–471. [DOI] [PubMed] [Google Scholar]
- 13. Adkins RM, Thomas F, Tylavsky FA, et al. Parental ages and levels of DNA methylation in the newborn are correlated. BMC Med Genet. 2011;12:47. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 14. Johnson KJ, Carozza SE, Chow EJ, et al. Parental age and risk of childhood cancer: a pooled analysis. Epidemiology. 2009;20(4):475–483. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 15. Dockerty JD, Draper G, Vincent T, et al. Case-control study of parental age, parity and socioeconomic level in relation to childhood cancers. Int J Epidemiol. 2001;30(6):1428–1437. [DOI] [PubMed] [Google Scholar]
- 16. Emerson JC, Malone KE, Daling JR, et al. Childhood brain tumor risk in relation to birth characteristics. J Clin Epidemiol. 1991;44(11):1159–1166. [DOI] [PubMed] [Google Scholar]
- 17. Feller M, Adam M, Zwahlen M, et al. Family characteristics as risk factors for childhood acute lymphoblastic leukemia: a population-based case-control study. PLoS One. 2010;5(10):pii:e13156. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 18. Kaye SA, Robison LL, Smithson WA, et al. Maternal reproductive history and birth characteristics in childhood acute lymphoblastic leukemia. Cancer. 1991;68(6):1351–1355. [DOI] [PubMed] [Google Scholar]
- 19. Oksuzyan S, Crespi CM, Cockburn M, et al. Birth weight and other perinatal characteristics and childhood leukemia in California. Cancer Epidemiol. 2012;36(6):e359–e365. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 20. Ou SX, Han D, Severson RK, et al. Birth characteristics, maternal reproductive history, hormone use during pregnancy, and risk of childhood acute lymphocytic leukemia by immunophenotype (United States). Cancer Causes Control. 2002;13(1):15–25. [DOI] [PubMed] [Google Scholar]
- 21. Yip BH, Pawitan Y, Czene K. Parental age and risk of childhood cancers: a population-based cohort study from Sweden. Int J Epidemiol. 2006;35(6):1495–1503. [DOI] [PubMed] [Google Scholar]
- 22. California Cancer Registry, California Department of Public Health About data from the CCR. http://www.ccrcal.org/Data_and_Statistics/About_Data_from_the_CCR.shtml Accessed September27, 2016.
- 23. Cnattingius S, Zack MM, Ekbom A, et al. Prenatal and neonatal risk factors for childhood lymphatic leukemia. J Natl Cancer Inst. 1995;87(12):908–914. [DOI] [PubMed] [Google Scholar]
- 24. Steliarova-Foucher E, Stiller C, Lacour B, et al. International Classification of Childhood Cancer, third edition. Cancer. 2005;103(7):1457–1467. [DOI] [PubMed] [Google Scholar]
- 25. Hemminki K, Kyyronen P, Vaittinen P. Parental age as a risk factor of childhood leukemia and brain cancer in offspring. Epidemiology. 1999;10(3):271–275. [PubMed] [Google Scholar]
- 26. Ennis SR, Ríos-Vargas M, Albert NG. The Hispanic Population: 2010. Washington, DC: US Census Bureau; 2011. (2010 Census Briefs, no. C2010BR-04). http://www.census.gov/prod/cen2010/briefs/c2010br-04.pdf Assessed September27, 2016. [Google Scholar]
- 27. Hjalgrim LL, Rostgaard K, Hjalgrim H, et al. Birth weight and risk for childhood leukemia in Denmark, Sweden, Norway, and Iceland. J Natl Cancer Inst. 2004;96(20):1549–1556. [DOI] [PubMed] [Google Scholar]
- 28. Larfors G, Hallbook H, Simonsson B. Parental age, family size, and offspring's risk of childhood and adult acute leukemia. Cancer Epidemiol Biomarkers Prev. 2012;21(7):1185–1190. [DOI] [PubMed] [Google Scholar]
- 29. Machiela MJ, Zhou W, Sampson JN, et al. Characterization of large structural genetic mosaicism in human autosomes. Am J Hum Genet. 2015;96(3):487–497. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 30. Zack M, Adami HO, Ericson A. Maternal and perinatal risk factors for childhood leukemia. Cancer Res. 1991;51(14):3696–3701. [PubMed] [Google Scholar]
- 31. Mallol-Mesnard N, Menegaux F, Lacour B, et al. Birth characteristics and childhood malignant central nervous sytem tumors: the ESCALE study (French Society for Childhood Cancer). Cancer Detect Prev. 2008;32(1):79–86. [DOI] [PubMed] [Google Scholar]
- 32. Von Behren J, Reynolds P. Birth characteristics and brain cancers in young children. Int J Epidemiol. 2003;32(2):248–256. [DOI] [PubMed] [Google Scholar]
- 33. Heuch JM, Heuch I, Akslen LA, et al. Risk of primary childhood brain tumors related to birth characteristics: a Norwegian prospective study. Int J Cancer. 1998;77(4):498–503. [DOI] [PubMed] [Google Scholar]
- 34. Hook EB, Cross PK, Schreinemachers DM. Chromosomal abnormality rates at amniocentesis and in live-born infants. JAMA. 1983;249(15):2034–2038. [PubMed] [Google Scholar]
- 35. Hook EB. Rates of chromosome abnormalities at different maternal ages. Obstet Gynecol. 1981;58(3):282–285. [PubMed] [Google Scholar]
- 36. Ferguson-Smith MA, Yates JR. Maternal age specific rates for chromosome aberrations and factors influencing them: report of a collaborative European study on 52 965 amniocenteses. Prenat Diagn. 1984;4 Spec No:5–44. [DOI] [PubMed] [Google Scholar]
- 37. Miller RW. Relation between cancer and congenital defects in man. N Engl J Med. 1966;275(2):87–93. [DOI] [PubMed] [Google Scholar]
- 38. Agha MM, Williams JI, Marrett L, et al. Congenital abnormalities and childhood cancer. Cancer. 2005;103(9):1939–1948. [DOI] [PubMed] [Google Scholar]
- 39. Botto LD, Flood T, Little J, et al. Cancer risk in children and adolescents with birth defects: a population-based cohort study. PLoS One. 2013;8(7):e69077. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 40. Carozza SE, Langlois PH, Miller EA, et al. Are children with birth defects at higher risk of childhood cancers. Am J Epidemiol. 2012;175(12):1217–1224. [DOI] [PubMed] [Google Scholar]
- 41. Curren MS, Liang CL, Davis K, et al. Assessing determinants of maternal blood concentrations for persistent organic pollutants and metals in the eastern and western Canadian Arctic. Sci Total Environ. 2015;527–528:150–158. [DOI] [PubMed] [Google Scholar]
- 42. Harris CA, Woolridge MW, Hay AW. Factors affecting the transfer of organochlorine pesticide residues to breastmilk. Chemosphere. 2001;43(2):243–256. [DOI] [PubMed] [Google Scholar]
- 43. Troisi R, Potischman N, Roberts J, et al. Associations of maternal and umbilical cord hormone concentrations with maternal, gestational and neonatal factors (United States). Cancer Causes Control. 2003;14(4):347–355. [DOI] [PubMed] [Google Scholar]
- 44. Troisi R, Hoover RN, Thadhani R, et al. Maternal, prenatal and perinatal characteristics and first trimester maternal serum hormone concentrations. Br J Cancer. 2008;99(7):1161–1164. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 45. Goriely A, Wilkie AO. Paternal age effect mutations and selfish spermatogonial selection: causes and consequences for human disease. Am J Hum Genet. 2012;90(2):175–200. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 46. Mills MB, Hudgins L, Balise RR, et al. Mutation risk associated with paternal and maternal age in a cohort of retinoblastoma survivors. Hum Genet. 2012;131(7):1115–1122. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 47. Bleyer A, Barr R, Hayes-Lattin B, et al. The distinctive biology of cancer in adolescents and young adults. Nat Rev Cancer. 2008;8(4):288–298. [DOI] [PubMed] [Google Scholar]
- 48. Keegan TH, Ries LA, Barr RD, et al. Comparison of cancer survival trends in the United States of adolescents and young adults with those in children and older adults. Cancer. 2016;122(7):1009–1016. [DOI] [PubMed] [Google Scholar]
- 49. Northam S, Knapp TR. The reliability and validity of birth certificates. J Obstet Gynecol Neonatal Nurs. 2006;35(1):3–12. [DOI] [PubMed] [Google Scholar]
- 50. Kohler BA, Sherman RL, Howlader N, et al. Annual report to the nation on the status of cancer, 1975–2011, featuring incidence of breast cancer subtypes by race/ethnicity, poverty, and state. J Natl Cancer Inst. 2015;107(6):djv048. [DOI] [PMC free article] [PubMed] [Google Scholar]
Associated Data
This section collects any data citations, data availability statements, or supplementary materials included in this article.