Abstract
Background:
Children are more susceptible to radiation-induced damage than adults, but little research has compared the risk of cancer after exposure to radiation during computed tomography (CT) among children at different ages. We aimed to explore the risk of intracranial tumours, leukemia or lymphoma among children, adolescents and young adults (aged < 25 yr) after radiation exposure from CT at or before the age of 18 years.
Methods:
We conducted a nested, population-based case–control study using data from Taiwan’s publicly funded health care system. We identified participants younger than 25 years with newly diagnosed intracranial tumours, leukemia or lymphoma, from Jan. 1, 2000, to Dec. 31, 2013. We assigned 10 non-cancer controls for each case, matching by sex, date of birth and day of entry to the cohort. We considered CT scans received at or before the age of 18 years and 3 or more years before the index date (the date of cancer diagnosis for cases) as exposure. We used conditional logistic regression models and incidence rate ratios (IRRs) to estimate the relationship between risk of these cancers and CT radiation exposure.
Results:
We identified 7807 cases and matched to 78 057 controls. Compared with no exposure, exposure to a single pediatric CT scan did not increase risk of intracranial tumours, leukemia or lymphoma. However, participants exposed to 4 or more CT scans had an elevated incidence (IRR 2.30, 95% confidence interval 1.43–3.71) of one of the cancer outcomes of interest. Receiving 4 or more CT scans at or before 6 years of age was associated with the highest risks of cancer, followed by ages 7–12 years and 13–18 years (p for trend < 0.001).
Interpretation:
Exposure to a single CT scan was not associated with increased risks of subsequent intracranial tumours, leukemia or lymphoma among children; however, we observed increased cancer risks among those with 4 or more CT scans, especially among younger children. Although these cancers are uncommon, the findings of this study underscore the importance of prudent use of CT in the pediatric population.
The use of computed tomography (CT) in children has increased globally over the past several decades.1–3 In North America, the annual incidence of childhood CT scans plateaued in 2006, followed by a slow decline, but the rate of pediatric CT imaging in the United States and Canada was still higher in 2016 than in 2000.3
Although low-dose (< 500 milliGray) radiation does not cause immediate illness in humans, it inflicts damage to the genome and can induce oncogenesis.4–6 Preclinical cell and animal models have been proposed to illustrate the dose–response effect after low-dose radiation.4,5
Clinical studies have shown that exposure from diagnostic CT is associated with increased risk of cancer among both adults and children,7–13 and that children are more vulnerable to radiation than adults.14 However, studies comparing risk of cancer after CT exposure among children at different ages are limited and have reported conflicting conclusions.7–9,12 It remains unclear how CT exposure at different ages in childhood affects the risk of cancer. In addition, studies that focus on risk of cancer among young adults associated with radiation exposure in childhood are lacking. Compared with cancers among older people, cancers among young adults have shown similar manifestations to those in adolescents in terms of the type of cancer, treatment response and prognosis.15–17
Since head CT is the most common type of CT used for children,3 and hematopoietic tissues are the most radiosensitive,18 we sought to investigate whether childhood CT exposure (at or before age 18 yr) was associated with risks of intracranial tumours, leukemia, non-Hodgkin lymphoma and Hodgkin lymphoma among children, adolescents and young adults. We also sought to evaluate whether any incremental increases in risk of these cancers after pediatric CT would last from adolescence to early adulthood.
Methods
Study design and setting
We conducted a population-based, nested case–control study using the National Health Insurance (NHI) Research Database (NHIRD) in Taiwan, to evaluate the association of radiation exposure (by total number of CT scans, cumulative radiation doses and cumulative number of CTs received at different ages) with subsequent risk of intracranial tumour, leukemia, non-Hodgkin lymphoma and Hodgkin lymphoma.
The population of Taiwan is about 23 million, and the NHIRD contains the health records for all NHI beneficiaries.19 All newborns in Taiwan become beneficiaries of the single-payer NHI program at birth, along with foreign nationals who have established a registered domicile for at least 6 months and those with a regular employer.20
We reported this study according to the Strengthening the Reporting of Observational Studies in Epidemiology (STROBE) checklist for case–control studies.21
Data source
The NHIRD contains deidentified health information for NHI beneficiaries, including demographic information, diagnoses and management of each medical visit.19 Under the NHI program, patients receiving therapies for cancer are eligible to be part of the Registry for Catastrophic Illness (a subset of the NHIRD) to reduce copayments.19,22
The NHI stipulates that health care providers attach reports written by board-certified specialists before reimbursement for procedure claim, including imaging;20 therefore, the NHIRD records only procedures that are actually performed. The NHI employs multiple audits to ensure the appropriateness of claims for medical services, and its accuracy and completeness have been shown in several validation studies.19,23
Cases, controls and matching
For our base cohort, we extracted data on all NHIRD beneficiaries who were younger than 25 years from Jan. 1, 2000, to Dec. 31, 2013.
To form our case cohort, we extracted cases of intracranial tumours (grades I–IV under World Health Organization classification),24 leukemia, non-Hodgkin lymphomas or Hodgkin lymphomas that were newly diagnosed during the study period. We used codes from the International Classification of Diseases, Ninth Revision (ICD-9) to identify these cases from the Registry of Catastrophic Illness; a previous study reported that this approach had a positive predictive value of 94%.25 We further used incidence density sampling to randomly select non-cancer controls from the base cohort.26
Before matching, we excluded patients with any malignant disease diagnosed before the study period. We also excluded those with cancer-predisposing conditions that were potential confounders for our exposure of CT imaging.27 For example, children with Down syndrome, who have an increased risk of leukemia, may receive CT scans for cardiac defects, and children with immunodeficiency may be imaged for recurrent infections.28–30
We identified patients with cancer-predisposing conditions using at least 1 inpatient or 3 outpatient claims.31,32 The relevant ICD-9 codes are listed in Appendix 1, Supplementary Table 1 and Table 2, available at www.cmaj.ca/lookup/doi/10.1503/cmaj.221303/tab-related-content. Finally, we excluded patients with missing data (i.e., unreported sex, date of birth or cohort entry date).
We pooled the remaining participants to form an at-risk population. In this at-risk population, when 1 cancer case was diagnosed during the study period (index case), we randomly assigned up to 10 participants free from cancer at the index diagnosis date as controls, to form a risk set with matched sex, date of birth (± 1 yr), and cohort entry date (± 1 yr).26,33 We repeated this process until all cases had been matched. We excluded risk sets when fewer than 5 controls could be assigned to a case. We defined the exposure period as the period from cohort entry to 3 years (the lag period) before the index date (i.e., the index case’s date of cancer diagnosis). The length of follow-up and the exposure period of the index case and the assigned controls were equivalent.
Exposure
Exposure was CT-associated radiation received at or before the age of 18 years, quantified by the cumulative number of CT scans7–11 and the organ-specific cumulative dose.13 Given that some cancer-related symptoms would prompt CT, we considered exposure to include only CT scans performed 3 or more years before the index date. A descriptive analysis from the United Kingdom found that the mean interval between symptom onset and diagnosis of low-grade pediatric brain tumours was 28.2 weeks.34 Other studies have shown that intracranial tumours associated with radiotherapy in childhood occurred 3–5 years after treatment.35,36 An increased incidence of hematologic malignant diseases was observed 2 years after occupational radiation exposure among adults.37 Therefore, we considered a 3-year lag period to be appropriate to minimize reverse causation bias and avoid improperly excluding cases with malignant diseases relating to radiation from CT scans.
Radiation causes direct damage to tissues and the organ-specific radiation dose varies in different CT types.4,5,38–41 Dosage calculation for CT exposure was done by consensus of a pediatric hematologist and oncologist (W.H.W.) and a radiologist at Changhua Christian Hospital. We used the cumulative radiation dose to the brain to assess the risk of intracranial tumour and the cumulative dose to red bone marrow for leukemia and lymphoma. We quantified effective radiation doses to the brain and red bone marrow based on sex, age, CT type and radiation dosimetry data, as reported by Gao and colleagues40 (Appendix 1, Supplemental Table 3). Gao and colleagues40 combined patient-specific CT parameters with various pediatric-sized phantoms (objects that mimic human tissue) to estimate the organ-absorbed radiation dose. As we were unable to find studies of radiation doses for CTs of extremities, we considered the radiation dose from these procedures to be 0. We divided the calculated cumulative organ-specific CT dosage into quintiles to assess the relationship between total radiation dosage and risk of cancers.
We determined the type of each CT scan according to the corresponding ICD-9 code, if not recorded in the NHIRD database. We also extracted data on other high-radiation procedures using claims for cardiac catheterizations and common nuclear medicine procedures.42,43
Statistical analysis
We calculated the incidence rate ratio (IRR) and built conditional logistic regression models to assess the adjusted odds ratios (ORs) between groups with and without CT exposure. As usage of CT scans increased over our investigation period,2 we adjusted the ORs for the calendar year of cohort entry as a linear variable. We also adjusted for family income (as a linear variable) and degree of urbanization of place of residence (as a categorical variable), as these variables are associated with leukemia among children.44–46
We used the 1-sided Cochran–Armitage test to investigate the relationship between age at CT exposure (categorized as ≤ 6, 7–12 and 13–18 yr) and cancer risk. We applied the Fisher exact test for categorical variables.
We completed all statistical analyses using SAS statistical software version 9.4 (SAS Institute).
Sensitivity analyses
We conducted sensitivity analyses using lag periods of 1–4 years and including the total number of other high-radiation procedures as a covariable in the model.42,43
We recomputed the radiation organ dose using data from Kim and colleagues,41 derived from radiological technician surveys and from realistic, normal-stature, pediatric phantoms ranging in age and body size (Appendix 1, Supplementary Methods and Table 4). We also reassessed the relationship between cumulative organ-specific CT dosage (by quintiles) and risk of cancers using these data.
We noted that participants in the top 2% of cumulative organ-specific doses had received 4 or more CT scans. Therefore, we conducted additional sensitivity analyses for this high-exposure group to evaluate the associated cancer risks, initially using doses calculated with data from Gao and colleagues40 and then with data from Kim and colleagues.41
Ethics approval
This study was approved by the Taipei Medical University – Joint Institutional Review Board (no. N201602055).
Results
In total, we initially identified 8055 patients with intracranial tumours, leukemia, non-Hodgkin lymphoma or Hodgkin lymphoma diagnosed during the study period. We excluded 153 patients with malignant diseases diagnosed before the study period, 72 with cancer-predisposing conditions and 19 with missing information (Figure 1). None of the patients with missing information had CT exposure. Demographic characteristics of the remaining 7807 patients and 78 057 matched controls are summarized in Table 1.
Figure 1:
Flow chart of case and control group selection. Note: NHIRD = National Health Insurance Research Database.
Table 1:
Participant characteristics
| Characteristic | Intracranial tumour | Leukemia | Non-Hodgkin lymphoma | Hodgkin lymphoma | ||||
|---|---|---|---|---|---|---|---|---|
| No (%) of patients* n = 2245 |
No. (%) of controls* n = 22 450 |
No (%) of patients* n = 3667 |
No. (%) of controls* n = 36 667 |
No (%) of patients* n = 1331 |
No. (%) of controls* n = 13 300 |
No (%) of patients* n = 564 |
No. (%) of controls* n = 5640 |
|
| Sex | ||||||||
| Male | 1341 (59.7) | 13 410 (59.7) | 2130 (58.1) | 21 300 (58.1) | 836 (62.8) | 8350 (62.8) | 316 (56.0) | 3160 (56.0) |
| Female | 904 (40.3) | 9040 (40.3) | 1537 (41.9) | 15 367 (41.9) | 495 (37.2) | 4950 (37.2) | 248 (44.0) | 2480 (44.0) |
| Age, yr, median (IQR) | ||||||||
| At entry | 5.3 (11.6) | 5.4 (11.5) | 3.8 (11.0) | 3.9 (11.0) | 9.7 (11.5) | 9.5 (11.5) | 11.1 (8.6) | 11.0 (8.6) |
| At first CT scan | 9.0 (9.6) | 7.8 (9.9) | 6.3 (8.4) | 8.5 (10.4) | 9.5 (10.8) | 12.4 (8.9) | 14.8 (6.0) | 11.1 (7.2) |
| Degree of urbanization† | ||||||||
| Metropolitan | 1299 (57.9) | 13 293 (59.2) | 2164 (59.0) | 21 966 (59.9) | 821 (61.7) | 8227 (61.9) | 376 (66.7) | 3436 (60.9) |
| City | 726 (32.3) | 7203 (32.1) | 1159 (31.6) | 11 379 (31) | 385 (28.9) | 3989 (30.0) | 147 (26.1) | 1716 (30.4) |
| Town | 105 (4.7) | 1031 (4.6) | 185 (5.0) | 1734 (4.7) | 64 (4.8) | 562 (4.2) | 25 (4.4) | 252 (4.5) |
| Rural area | 65 (2.9) | 681 (3.0) | 112 (3.1) | 1180 (3.2) | 40 (3.0) | 351 (2.6) | 12 (2.1) | 160 (2.8) |
| Missing | 50 (2.2) | 242 (1.1) | 47 (1.3) | 408 (1.1) | 21 (1.6) | 171 (1.3) | 4 (0.7) | 76 (1.3) |
| Annual family income, percentile | ||||||||
| > 75th | 519 (23.1) | 5693 (25.4) | 969 (26.4) | 9591 (26.2) | 312 (23.4) | 3039 (22.8) | 134 (23.8) | 1182 (21) |
| 50th to 75th | 507 (22.6) | 5610 (25.0) | 837 (22.8) | 9079 (24.8) | 332 (24.9) | 3451 (25.9) | 137 (24.3) | 1487 (26.4) |
| 25th to 49th | 558 (24.9) | 5634 (25.1) | 880 (24) | 9111 (24.8) | 320 (24.0) | 3354 (25.2) | 137 (24.3) | 1444 (25.6) |
| < 25th | 657 (29.3) | 5491 (24.5) | 970 (26.5) | 8847 (24.1) | 364 (27.3) | 3442 (25.9) | 156 (27.7) | 1517 (26.9) |
| Missing | 4 (0.2) | 22 (0.1) | 11 (0.3) | 39 (0.1) | 3 (0.2) | 14 (0.1) | 0 (0.0) | 10 (0.2) |
| Calendar year of cohort entry | ||||||||
| 2000–2004 | 2017 (89.8) | 20 155 (89.8) | 3119 (85.1) | 31 306 (85.4) | 1267 (95.2) | 12 658 (95.2) | 557 (98.8) | 5570 (98.8) |
| 2005–2009 | 195 (8.7) | 1954 (8.7) | 448 (12.2) | 4401 (12) | 52 (3.9) | 527 (4.0) | 6 (1.1) | 60 (1.1) |
| 2010 or later | 33 (1.5) | 341 (1.5) | 100 (2.7) | 960 (2.6) | 12 (0.9) | 115 (0.9) | < 3 | 10 (0.2) |
| Number of CT scans | ||||||||
| 0 | 2207 (98.3) | 22 162 (98.7) | 3631 (99.0) | 36 219 (98.8) | 1311 (98.5) | 13 119 (98.6) | 557 (98.8) | 5524 (97.9) |
| 1–3 | 32 (1.4) | 282 (1.3) | 30 (0.8) | 435 (1.2) | 16 (1.2) | 175 (1.3) | 6 (1.1) | 113 (2.0) |
| ≥ 4 | 6 (0.3) | 6 (0.0) | 6 (0.2) | 13 (0.0) | 4 (0.3) | 6 (0.0) | < 3 | 3 (0.1) |
| Exposure to other high radiation procedures | ||||||||
| Cystography | < 3 (0.0) | 61 (0.3) | 10 (0.3) | 110 (0.3) | 3 (0.2) | 27 (0.2) | < 3 (0.2) | 10 (0.2) |
| Cardiac catheterization | < 3 (0.0) | 18 (0.1) | 5 (0.1) | 27 (0.1) | < 3 (0.1) | 8 (0.1) | < 3 (0.2) | 7 (0.1) |
| Renal scan | 0 (0.0) | 20 (0.1) | < 3 (0.1) | 22 (0.1) | < 3 (0.1) | 6 (0.0) | 0 (0.0) | 6 (0.1) |
| Renography | 0 (0.0) | 11 (0.0) | < 3 (0.1) | 14 (0.0) | < 3 (0.2) | 4 (0.0) | 0 (0.0) | 5 (0.1) |
| Other‡ | 5 (0.2) | 23 (0.1) | < 3 (0.1) | 35 (0.1) | < 3 (0.1) | 17 (0.1) | 0 (0.0) | 4 (0.1) |
Note: CT = computed tomography, IQR = interquartile range.
Unless indicated otherwise. National Health Insurance Research Database policy on protecting confidentiality requires suppression of cells < 3.
Degree of urbanization of townships in Taiwan was stratified into 7 levels according to official statistics.46 In this study, we classified the townships into 4 categories, namely metropolitan (level 1 and 2), city (level 3 and 4), town (level 5 and 6) and rural area (level 7).
Includes positron emission tomography, skeletal scintigraphy, lung perfusion scan, hepatobiliary scintigraphy, Meckel scan and gastric emptying tests.
The proportion of patients exposed to other high-radiation procedures was similar in the case and control groups (Appendix 1, Supplemental Table 5).
Risk of cancer by number of scans
Compared with no exposure, exposure to a single pediatric CT scan did not increase subsequent cancer risk (Table 2). Participants who received 2–3 CT scans had an increased risk of intracranial tumour (adjusted OR 2.36, 95% confidence interval [CI] 1.40–3.97), but not leukemia, non-Hodgkin lymphoma or Hodgkin lymphoma. Participants who received 4 or more CT scans had a higher risk of intracranial tumour (adjusted OR 9.01, 95% CI 2.89–28.11), leukemia (adjusted OR 4.80, 95% CI 1.79–12.84) and non-Hodgkin lymphoma (adjusted OR 6.76, 95% CI 1.91–23.96), but not Hodgkin lymphoma.
Table 2:
Adjusted odds ratios for intracranial tumours, leukemia and lymphomas related to the number of computed tomography scans
| Number of CT scans | Intracranial tumours | Leukemia | Non-Hodgkin lymphoma | Hodgkin lymphoma | ||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|
| No. of patients* | No. of controls* | Adjusted OR† (95% CI) | No. of patients* | No. of controls* | Adjusted OR† (95% CI) | No. of patients* | No. of controls* | Adjusted OR† (95% CI) | No. of patients* | No. of controls* | Adjusted OR† (95% CI) | |
| 0 | 2207 | 22 162 | 1 | 3631 | 36 219 | 1 | 1311 | 13 119 | 1 | 557 | 5524 | 1 |
| 1 | 14 | 205 | 0.69 (0.40–1.20) | 21 | 309 | 0.68 (0.43–1.05) | 10 | 116 | 0.87 (0.45–1.68) | 5 | 71 | 0.70 (0.28–1.74) |
| 2–3 | 18 | 77 | 2.36 (1.40–3.97) | 9 | 126 | 0.71 (0.36–1.40) | 6 | 59 | 1.02 (0.44–2.38) | < 3 | 42 | 0.23 (0.03–1.68) |
| ≥ 4 | 6 | 6 | 9.01 (2.89–28.11) | 6 | 13 | 4.80 (1.79–12.84) | 4 | 6 | 6.76 (1.91–23.96) | < 3 | 3 | 3.61 (0.37–35.08) |
Note: CI = confidence interval, CT = computed tomography, OR = adjusted odds ratio.
National Health Insurance Research Database policy on protecting confidentiality requires suppression of cells < 3.
Adjusted by the year of cohort entry, annual family income (both as linear variables) and urbanization level (as a categorical variable).
The IRR of participants with one of the cancer outcomes of interest receiving 4 or more CT scans was 2.30 (95% CI 1.43–3.71) relative to nonexposed participants (Table 3).
Table 3:
The incidence rate ratio for intracranial tumour, leukemia or lymphomas according to computed tomography exposure status
| No. of CT scans | No. of patients* | No. of controls | Person-years | Incidence rate ratio (95% CI) |
|---|---|---|---|---|
| 0 | 7706 | 77 024 | 462 594.25 | 1.00 |
| 1–3 | 84 | 1005 | 10 569.71 | 0.48 (0.38–0.59) |
| ≥ 4 | 17 | 28 | 443.27 | 2.30 (1.43–3.71) |
Note: CI = confidence interval, CT = computed tomography.
Patients with intracranial tumour, leukemia, non-Hodgkin lymphoma and Hodgkin lymphoma.
Cancer risk by cumulative radiation dose
Participants in the top quintile of cumulative brain radiation dose had a significantly higher risk of intracranial tumour compared with nonexposed participants (adjusted OR 3.61, 95% CI 1.93–6.75) (Table 4). We did not observe this association between the cumulative dose of radiation to red bone marrow and risk of hematologic malignancies.
Table 4:
Adjusted odds ratios for intracranial tumours, leukemia and lymphomas related to the cumulative organ-specific dose from computed tomography scans, estimated using radiation dosimetry from Gao and colleagues40
| Percentile of cumulative organ dose (absolute dose in mGy) | No. of patients* | No. of controls* | Adjusted OR† (95% CI) |
|---|---|---|---|
| Intracranial tumours | |||
| No exposure | 1 | ||
| Brain dose 20th percentile (0.1 to 31.1) | 7 | 58 | 1.19 (0.54–2.62) |
| Brain dose 21st to 40th percentile (31.2 to 32.2) | 3 | 60 | 0.51 (0.16–1.64) |
| Brain dose 41st to 60th percentile (32.3 to 43.0) | < 3 | 55 | 0.36 (0.09–1.48) |
| Brain dose 61st to 80th percentile (43.1 to 64.0) | 7 | 38 | 1.97 (0.87–4.46) |
| Brain dose 81st to 100th percentile (> 64.0) | 14 | 39 | 3.61 (1.93–6.75) |
| Brain dose 99th to 100th percentile (> 139.8)† | 5 | 3 | 15.49 (3.68–65.27) |
| Leukemia | |||
| No exposure | 1 | ||
| RBM dose 20th percentile (0.1 to 9.0) | 17 | 103 | 1.64 (0.98–2.75) |
| RBM dose 21st to 40th percentile (9.1 to 11.2) | 3 | 73 | 0.41 (0.13–1.31) |
| RBM dose 41st to 60th percentile (11.3 to 15.3) | 4 | 96 | 0.42 (0.15–1.14) |
| RBM dose 61st to 80th percentile (15.4 to 20.0) | 5 | 64 | 0.78 (0.31–1.95) |
| RBM dose 81st to 100th percentile (> 20.0) | 7 | 78 | 0.89 (0.41–1.94) |
| RBM dose 99th to 100th percentile (> 51.9)† | 3 | 6 | 5.38 (1.27–22.8) |
| Non-Hodgkin lymphoma | |||
| No exposure | 1 | ||
| RBM dose 20th percentile (0.1 to 9.0) | 6 | 37 | 1.62 (0.68–3.86) |
| RBM dose 21st to 40th percentile (9.1 to 11.2) | 4 | 22 | 1.87 (0.64–5.47) |
| RBM dose 41st to 60th percentile (11.3 to 15.3) | 4 | 33 | 1.23 (0.43–3.5) |
| RBM dose 61st to 80th percentile (15.4 to 20.0) | < 3 | 36 | 0.28 (0.04–2.03) |
| RBM dose 81st to 100th percentile (> 20.0) | 4 | 36 | 1.11 (0.39–3.13) |
| RBM dose 99th to 100th percentile (> 51.9)† | < 3 | 3 | 6.74 (1.12–40.35) |
| Hodgkin lymphoma | |||
| No exposure | 1 | ||
| RBM dose 20th percentile (0.1 to 9.0) | < 3 | 22 | 0.45 (0.06–3.32) |
| RBM dose 21st to 40th percentile (9.1 to 11.2) | 0 | 8 | NA |
| RBM dose 41st to 60th percentile (11.3 to 15.3) | < 3 | 25 | 0.83 (0.2–3.5) |
| RBM dose 61st to 80th percentile (15.4 to 20.0) | 3 | 29 | 1.01 (0.31–3.32) |
| RBM dose 81st to 100th percentile (> 20.0) | 0 | 20 | NA |
| RBM dose 99th to 100th percentile (> 51.9)† | 0 | < 3 | NA |
Note: CI = confidence interval, mGy = milligray, NA = not applicable, OR = odds ratio, RBM = red bone marrow.
National Health Insurance Research Database policy on protecting confidentiality requires suppression of cells < 3.
Adjusted by the year of cohort entry, annual family income (both as linear variables) and urbanization level (as a categorical variable).
Sensitivity analysis.
Cancer risk by age of exposure
Participants who received 4 or more CT scans at or before the age of 6 years had the highest risk of cancer, followed by those aged 7–12 years and those aged 13–18 years (Figure 2). The correlation was statistically significant (p for trend < 0.001).
Figure 2:
Forest plots for adjusted odds ratios (ORs) of intracranial tumours, leukemia, non-Hodgkin lymphoma and Hodgkin lymphoma by cumulative number of computed tomography (CT) scans among children at different ages (p value for trend < 0.001 for ≥ 4 CT exposures by Cochran–Armitage test). *Adjusted by the year of cohort entry, annual family income (both as linear variables) and degree of urbanization (as a categorical variable). Note: CI = confidence interval.
Sensitivity analyses
Using a 1- or 2-year lag period, we found a stronger association between risk of intracranial tumour and increasing CT exposure than in the primary analysis. This association also held with a 4-year lag period (Appendix 1, Supplemental Table 6).
We included the number of high-radiation procedures in our full model and the results remained consistent (Appendix 1, Supplemental Table 7).
The re-estimated organ-absorbed radiation dose using data from Kim and colleagues41 still showed an association between an elevated cumulative dose of radiation and risk of intracranial tumour and of leukemia. Increased cumulative radiation did not significantly increase the risk of non-Hodgkin lymphoma or Hodgkin lymphoma (Appendix 1, Supplemental Table 8).
Using data from Gao and colleagues40 and from Kim and colleagues,41 the highest 2% of cumulative organ-specific doses (> 98th to ≤ 100th percentile) was associated with elevated risk of intracranial tumour, leukemia and non-Hodgkin lymphoma, but not Hodgkin lymphoma (Table 4 and Appendix 1, Supplemental Table 8).
Interpretation
We found that receipt of a single CT scan at or before 18 years of age was not associated with increased risk of intracranial tumours, leukemia, non-Hodgkin lymphoma or Hodgkin lymphoma, to the age of 25 years. However, children who had received 4 or more CT scans at or before 18 years of age had a 2.3-fold increase in the incidence of these cancers compared with those without exposure. It is important to note that these neoplasms are uncommon among children, with an incidence of 15–40 cases per million in Taiwan over a 15-year period from 1996 to 2010.47
The associated risk of cancer we observed was highest among children who had received 4 or more CT scans at or before 6 years of age, followed by those aged 7–12 years and adolescents aged 13–18 years (Figure 2), suggesting that younger children are more vulnerable to radiation than older children. However, this finding should be interpreted cautiously, as the risks may be overestimates because of residual confounding and the low number of participants in high-radiation groups.48
Several studies have evaluated the association between pediatric neoplasms and CT scans among children of different ages, but the results were inconclusive.7–9,12 These studies focused on the age of the first CT exposure rather than the cumulative number of CT scans at different ages.
The positive relationship between the cumulative organ-specific dose of radiation and the risk of intracranial tumour and leukemia that we observed has also been seen in other studies,7,8 but the association between childhood CT radiation and lymphoma is still unclear. For Hodgkin lymphoma, the lack of association we observed across the cumulative number of CT scans was consistent with earlier reports,14,49 except for 1 Australian study.8 For non-Hodgkin lymphoma, Li and colleagues12 did not find an association between 2 or more CT scans and non-Hodgkin lymphoma, in contrast to the association we observed for those with exposure to 4 or more CT scans. This discrepancy may have been owing to the different CT protocols used in various countries.50 Further studies with a standard scanning protocol might produce more conclusive results. It should be noted that the association we observed between non-Hodgkin lymphoma and higher numbers of scans did not hold when the data were reanalyzed by increasing quintiles of radiation exposure to red bone marrow.
When human cells are exposed to low-dose radiation, small DNA breaks are generated and mended by intrinsic DNA repair processes. Therefore, low-level CT exposure appears not to be carcinogenic and animal models have suggested that it may even protect cells from mutagenesis.4,5 However, when cumulative DNA damage stimulated by recurrent radiation exposure exceeds DNA repair abilities, the risk of carcinogenesis rises.4,5 This mechanism explains the dose–response relationship between the cumulative organ-specific dose of radiation and the risk of cancers observed in our study and in previous studies.7–11
The large population of this nationwide cohort is one of the major advantages of this study. In addition, we matched cases and controls on the calendar year of cohort entry and the duration of the exposure period to ensure equal opportunity for CT exposure, avoiding a time-window bias.51 Furthermore, we excluded patients with cancer-predisposing conditions to avoid related confounding.
Our work reinforces the importance of radiation protection strategies, addressed by the International Atomic Energy Agency.52 Unnecessary CT scans should be avoided, and special attention should be paid to patients who require repeated CT scans.52 Parents and pediatric patients should be well informed on risks and benefits before radiological procedures and encouraged to participate in decision-making around imaging.53
Limitations
Given the observational study design, our results should not be interpreted as causal; rather, our study assesses the association between radiation exposure and subsequent risks of cancer.
Despite our efforts to control for potential confounders, residual confounding may still be present. For example, data on some risk factors for cancer — such as smoking, alcohol consumption, obesity (i.e., body mass index) and exposure to pesticides or phthalate-containing medications — were lacking in the NHIRD database.54–56 About 10% of patients with childhood or adult cancers have germline genomic alterations for which we had no data.57–59 We were unable to eliminate the influence of high-radiation procedures completely; however, the effect should be minimal because the misclassification would have been non-differential between cases and controls.
Although we adopted a longer lag period than those defined in most of the previous literature,8–12 we are unable to eliminate the possibility of reverse causality. In addition, more than 10% of participants were born after 2005, which led to a short follow-up time for neoplasm development.
Because of a lack of local data on CT radiation dosimetry, errors may have occurred in our calculation of the organ-absorbed dose of radiation. Furthermore, the precision of CT claim data in the NHIRD has not yet been validated, despite previous publications adopting the same approach.9,12 Similarly, the code definitions we used for each individual cancer type have not been validated in the NHIRD, although this approach has been shown to have a high positive predictive value for all cancers as a group in this database.25 The algorithm we used to identify cancer-predisposing conditions has not been validated but is a common approach used by researchers extracting data from the NHIRD.9
Importantly, the small numbers of participants in the high-radiation groups reduced our ability to detect potential effects of high radiation doses. The wide OR ranges found in these analyses indicate that these results should be interpreted with caution, as the OR may be inflated because of sparse data.48
We excluded 19 (0.2%) cases with missing data, but the impact on our results was likely small because none of them had CT scans.
Conclusion
This study found that exposure to a single CT scan at or before 18 years of age was not associated with the development of subsequent intracranial tumours, leukemia, non-Hodgkin lymphoma and Hodgkin lymphoma during childhood, adolescence and early adulthood. Children who received multiple CT scans had higher risks of intracranial tumours, leukemia and non-Hodgkin lymphoma, but not Hodgkin lymphoma. Younger children appeared vulnerable to cancer risks associated with repeated CT exposure. Although these tumours are uncommon, these results indicate that judicious CT usage and radiation-reducing techniques should be advocated.
Supplementary Material
Footnotes
Competing interests: None declared.
This article has been peer reviewed.
Contributors: Yu-Hsuan Shao conceived and designed the study. All of the authors the acquisition, analysis, and interpretation of data. Wei-Hao Wang drafted the manuscript. All of the authors revised it critically for important intellectual content, gave final approval of the version to be published and agreed to be accountable for all aspects of the work.
Funding: The authors received no specific funding for this work.
Data sharing: Data are available from the Health and Welfare Data Center published by the Ministry of Health and Welfare. Given legal restrictions imposed by the government of Taiwan in relation to the Personal Information Protection Act, data cannot be made publicly available. Requests for data can be sent as a formal proposal to the Health and Welfare Data Center Administration (https://dep.mohw.gov.tw/DOS/cp-5119-59201-113.html).
References
- 1.Pearce MS, Salotti JA, McHugh K, et al. CT scans in young people in Northern England: trends and patterns 1993–2002. Pediatr Radiol 2011;41:832–8. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 2.Hsueh YW, Lin YC, Lin CH, et al. Trends in computed tomography scan uses in Taiwan from 2000 to 2013. J Chin Med Assoc 2019;82:948–56. [DOI] [PubMed] [Google Scholar]
- 3.Smith-Bindman R, Kwan ML, Marlow EC, et al. Trends in use of medical imaging in US Health care systems and in Ontario, Canada, 2000–2016. JAMA 2019; 322:843–56. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 4.Bonner WM. Low-dose radiation: thresholds, bystander effects, and adaptive responses. Proc Natl Acad Sci U S A 2003;100:4973–5. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 5.Mancuso M, Pasquali E, Giardullo P, et al. The radiation bystander effect and its potential implications for human health. Curr Mol Med 2012;12:613–24. [DOI] [PubMed] [Google Scholar]
- 6.Macià MMI, Calduch AL, López EC. Radiobiology of the acute radiation syndrome. Rep Pract Oncol Radiother 2011;16:123–30. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 7.Pearce MS, Salotti JA, Little MP, et al. Radiation exposure from CT scans in childhood and subsequent risk of leukaemia and brain tumours: a retrospective cohort study. Lancet 2012;380:499–505. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 8.Mathews JD, Forsythe AV, Brady Z, et al. Cancer risk in 680 000 people exposed to computed tomography scans in childhood or adolescence: data linkage study of 11 million Australians. BMJ 2013;346:f2360. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 9.Huang WY, Muo CH, Lin CY, et al. Paediatric head CT scan and subsequent risk of malignancy and benign brain tumour: a nation-wide population-based cohort study. Br J Cancer 2014;110:2354–60. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 10.Nikkilä A, Raitanen J, Lohi O, et al. Radiation exposure from computerized tomography and risk of childhood leukemia: Finnish register-based case-control study of childhood leukemia (FRECCLE). Haematologica 2018;103:1873–80. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 11.Meulepas JM, Ronckers CM, Smets A, et al. Radiation exposure from pediatric CT scans and subsequent cancer risk in the Netherlands. J Natl Cancer Inst 2019; 111:256–63. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 12.Li IG, Yang Y-H, Li Y-T, et al. Paediatric computed tomography and subsequent risk of leukaemia, intracranial malignancy and lymphoma: a nationwide population-based cohort study. Sci Rep 2020;10:7759. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 13.Shao YH, Tsai K, Kim S, et al. Exposure to tomographic scans and cancer risks. JNCI Cancer Spectr 2019;4:pkz072. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 14.Hsu WL, Preston DL, Soda M, et al. The incidence of leukemia, lymphoma and multiple myeloma among atomic bomb survivors: 1950–2001. Radiat Res 2013; 179:361–82. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 15.Crombie JL, LaCasce AS. Current considerations in AYA Hodgkin lymphoma. Br J Haematol 2019;184:72–81. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 16.Boissel N, Baruchel A. Acute lymphoblastic leukemia in adolescent and young adults: treat as adults or as children? Blood 2018;132:351–61. [DOI] [PubMed] [Google Scholar]
- 17.Miller KD, Fidler-Benaoudia M, Keegan TH, et al. Cancer statistics for adolescents and young adults, 2020. CA Cancer J Clin 2020;70:443–59. [DOI] [PubMed] [Google Scholar]
- 18.Singh VK, Seed TM. Pharmacological management of ionizing radiation injuries: current and prospective agents and targeted organ systems. Expert Opin Pharmacother 2020;21:317–37. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 19.Hsieh CY, Su CC, Shao SC, et al. Taiwan’s National Health Insurance Research Database: past and future. Clin Epidemiol 2019;11:349–58. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 20.National Health Insurance Act. Taiwan: Ministry of Health and Welfare; amended; 2021. Jan. 20. Available: https://law.moj.gov.tw/ENG/LawClass/LawAll.aspx?pcode=L0060001 (accessed 2023 Mar. 17). [Google Scholar]
- 21.Von Elm E, Altman DG, Egger M, et al. Strengthening the reporting of observational studies in epidemiology (STROBE) statement: guidelines for reporting observational studies. BMJ 2007;335:806–8. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 22.Nan-Ping Y, Yi-Hui L, Chi-Yu C, et al. Comparisons of medical utilizations and categorical diagnoses of emergency visits between the elderly with catastrophic illness certificates and those without. BMC Health Serv Res 2013;13:152. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 23.Lin LY, Warren-Gash C, Smeeth L, et al. Data resource profile: the National Health Insurance Research Database (NHIRD). Epidemiol Health 2018;40:e2018062. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 24.Louis DN, Perry A, Reifenberger G, et al. The 2016 World Health Organization classification of tumors of the central nervous system: a summary. Acta Neuropathol 2016;131:803–20. [DOI] [PubMed] [Google Scholar]
- 25.Kao WH, Hong JH, See LC, et al. Validity of cancer diagnosis in the National Health Insurance database compared with the linked National Cancer Registry in Taiwan. Pharmacoepidemiol Drug Saf 2018;27:1060–6. [DOI] [PubMed] [Google Scholar]
- 26.Richardson DB. An incidence density sampling program for nested case-control analyses. Occup Environ Med 2004;61:e59. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 27.Berrington de Gonzalez A, Salotti JA, McHugh K, et al. Relationship between paediatric CT scans and subsequent risk of leukaemia and brain tumours: assessment of the impact of underlying conditions. Br J Cancer 2016;114:388–94. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 28.Maarschalk-Ellerbroek LJ, de Jong PA, van Montfrans JM, et al. CT screening for pulmonary pathology in common variable immunodeficiency disorders and the correlation with clinical and immunological parameters. J Clin Immunol 2014;34:642–54. [DOI] [PubMed] [Google Scholar]
- 29.Kebudi R, Kiykim A, Sahin MK. Primary immunodeficiency and cancer in children; a review of the literature. Curr Pediatr Rev 2019;15:245–50. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 30.Meulepas JM, Ronckers CM, Merks J, et al. Confounding of the association between radiation exposure from CT scans and risk of leukemia and brain tumors by cancer susceptibility syndromes. J Radiol Prot 2016;36:953–74. [DOI] [PubMed] [Google Scholar]
- 31.Sung SF, Hsieh CY, Lin HJ, et al. Validation of algorithms to identify stroke risk factors in patients with acute ischemic stroke, transient ischemic attack, or intracerebral hemorrhage in an administrative claims database. Int J Cardiol 2016; 215: 277–82. [DOI] [PubMed] [Google Scholar]
- 32.Lin CY, Chien CC, Chen HA, et al. The impact of comorbidity on survival after hemorrhagic stroke among dialysis patients: a nationwide population-based study. BMC Nephrol 2014;15:186. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 33.Ernster VL. Nested case-control studies. Prev Med 1994;23:587–90. [DOI] [PubMed] [Google Scholar]
- 34.Shanmugavadivel D, Liu JF, Murphy L, et al. Accelerating diagnosis for childhood brain tumours: an analysis of the HeadSmart UK population data. Arch Dis Child 2020;105:355–62. [DOI] [PubMed] [Google Scholar]
- 35.Lee CY, Chen YW, Lee YY, et al. Irradiation-induced secondary tumors following pediatric central nervous system tumors: experiences of a single institute in Taiwan (1975–2013). Int J Radiat Oncol Biol Phys 2018;101:1243–52. [DOI] [PubMed] [Google Scholar]
- 36.Chojnacka M, Pedziwiatr K, Skowronska-Gardas A, et al. Second brain tumors following central nervous system radiotherapy in childhood. Br J Radiol 2014;87:20140211. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 37.Cardis E, Vrijheid M, Blettner M, et al. Risk of cancer after low doses of ionising radiation: retrospective cohort study in 15 countries. BMJ 2005;331:77. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 38.Little JB. Radiation carcinogenesis. Carcinogenesis 2000;21:397–404. [DOI] [PubMed] [Google Scholar]
- 39.Barker HE, Paget JTE, Khan AA, et al. The tumour microenvironment after radiotherapy: mechanisms of resistance and recurrence. Nat Rev Cancer 2015;15: 409–25. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 40.Gao Y, Quinn B, Pandit-Taskar N, et al. Patient-specific organ and effective dose estimates in pediatric oncology computed tomography. Phys Med 2018;45: 146–55. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 41.Kim KP, Berrington de González A, Pearce MS, et al. Development of a database of organ doses for paediatric and young adult CT scans in the United Kingdom. Radiat Prot Dosimetry 2012;150:415–26. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 42.Yuan MK, Chang SC, Yuan MC, et al. Pediatric nuclear medicine examinations and subsequent risk of neoplasm: a nationwide population-based cohort study. Front Med (Lausanne) 2021;8:764849. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 43.Wei KC, Lin HY, Hung SK, et al. Leukemia risk after cardiac fluoroscopic interventions stratified by procedure number, exposure latent time, and sex: a nationwide population-based case-control study. Medicine (Baltimore) 2016;95:e2953. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 44.Poole C, Greenland S, Luetters C, et al. Socioeconomic status and childhood leukaemia: a review. Int J Epidemiol 2006;35:370–84. [DOI] [PubMed] [Google Scholar]
- 45.Li CY, Lin RS, Lin CH. Urbanization and childhood leukaemia in Taiwan. Int J Epidemiol 1998;27:587–91. [DOI] [PubMed] [Google Scholar]
- 46.Liu C-Y, Hung Y-T, Chuang Y-L, et al. Incorporating development stratification of Taiwan Townships into sampling design of large-scale health interview survey. J Health Manag 2006;4:1–22. [Google Scholar]
- 47.Liu YL, Lo WC, Chiang CJ, et al. Incidence of cancer in children aged 0–14 years in Taiwan, 1996–2010. Cancer Epidemiol 2015;39:21–8. [DOI] [PubMed] [Google Scholar]
- 48.Greenland S, Schwartzbaum JA, Finkle WD. Problems due to small samples and sparse data in conditional logistic regression analysis. Am J Epidemiol 2000;151:531–9. [DOI] [PubMed] [Google Scholar]
- 49.Berrington de Gonzalez A, Journy N, Lee C, et al. No association between radiation dose from pediatric CT scans and risk of subsequent Hodgkin lymphoma. Cancer Epidemiol Biomarkers Prev 2017;26:804–6. [DOI] [PubMed] [Google Scholar]
- 50.Smith-Bindman R, Wang Y, Chu P, et al. International variation in radiation dose for computed tomography examinations: prospective cohort study. BMJ 2019;364:k4931. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 51.Suissa S, Dell’aniello S, Vahey S, et al. Time-window bias in case-control studies: statins and lung cancer. Epidemiology 2011;22:228–31. [DOI] [PubMed] [Google Scholar]
- 52.Tsapaki V, Balter S, Cousins C, et al. The International Atomic Energy Agency action plan on radiation protection of patients and staff in interventional procedures: achieving change in practice. Phys Med 2018;52:56–64. [DOI] [PubMed] [Google Scholar]
- 53.Kasraie N, Jordan D, Keup C, et al. Optimizing communication with parents on benefits and radiation risks in pediatric imaging. J Am Coll Radiol 2018;15:809–17. [DOI] [PubMed] [Google Scholar]
- 54.Calle EE, Kaaks R. Overweight, obesity and cancer: epidemiological evidence and proposed mechanisms. Nat Rev Cancer 2004;4:579–91. [DOI] [PubMed] [Google Scholar]
- 55.Karalexi MA, Tagkas CF, Markozannes G, et al. Exposure to pesticides and childhood leukemia risk: a systematic review and meta-analysis. Environ Pollut 2021;285:117376. [DOI] [PubMed] [Google Scholar]
- 56.Ahern TP, Spector LG, Damkier P, et al. Medication–associated phthalate exposure and childhood cancer incidence. J Natl Cancer Inst 2022;114:885–94. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 57.Huang KL, Mashl RJ, Wu Y, et al. Pathogenic germline variants in 10 389 adult cancers. Cell 2018;173:355–70.e14. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 58.Zhang J, Walsh MF, Wu G, et al. Germline mutations in predisposition genes in pediatric cancer. N Engl J Med 2015;373:2336–46. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 59.Akhavanfard S, Padmanabhan R, Yehia L, et al. Comprehensive germline genomic profiles of children, adolescents and young adults with solid tumors. Nat Commun 2020;11:2206. [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.