1. Introduction
The incidence of thyroid cancer appears to be increasing worldwide [
1]. In the USA, for example, there was an increase from 3.6 per 100,000 in 1973 to 8.7 per 100,000 in 2002,
i.e., a 2.4-fold increase [
2]. In China, the incidence of thyroid cancer was also increasing in recent years, as the annual incidence has increased by 14.51% for females during 2003–2007 [
3]. In Tianjin (China), the thyroid cancer incidence for females increased from 1.3 per 100,000 in 1981 to 4.2 per 100,000 in 2001 [
4]. In Beijing, a total number of 862 thyroid cancer cases were found in 2006–2007, in comparison with 258 cases in 1998–1999, making it one of the fastest growing cancer types in the last 10 years [
5]. In addition to improvements in thyroid cancer detection, this increase may reflect changes in environmental factors, including an increase in medical exposure to ionizing radiation from diagnostic imaging, especially during childhood, which is one of the few established risk factors for thyroid cancer and attributed in large part to the growing use of CT [
6,
7] . The frequency of CT utilization in pediatric patients has increased exponentially because of the sensitivity of CT scanner images, which is ten-fold higher than that of a conventional X-ray device. CT can also be performed within seconds, so there is no need to sedate very young patients [
8]. However, the effective doses of radiation from CT scans may be 5–20 times higher than those from routine conventional radiology [
9]. Moreover, the smaller size of children means that the effective dose from a CT scan is usually higher than that received by adults [
10]. Children as the most radiosensitive subgroup, the lifetime risk of solid cancer induced by radiation exposure is 2 or 3 times higher than general population [
11]. It is also well known that children have a longer life expectancy than adults, which may mean that cancer has a greater opportunity to occur and develop.
There have been several previous studies of the radiation dose received in the thyroid during CT scans and estimates have been made of the cancer risk [
12,
13,
14], with the risk estimation being based on the model risk projection, although the number of study samples was very small in some studies. Recently, Pearce
et al. [
15] found that there was a significant association between the estimated radiation doses resulted from CT scans to red bone marrow and brain in childhood and subsequent incidence of leukemia and brain tumors. Moreover, Mathews
et al. [
16] calculated that after accounting for age, sex and year of birth, the overall cancer incidence for exposed from CT scan was 24% higher than for unexposed people, which was statistically significant. These two large record-linkage epidemiology studies proved an increased cancer risk in the future associated with patients receiving CT scans during childhood.
In China, the CT was introduced into the mainland China in 1972, and quickly experienced a sharp increase by 418 CT scanners per year during 1999–2009 [
17,
18]. CT scans are very common for pediatric patients, especially for head CT [
19], but there are few studies concerning the frequency of CT scans for pediatric patients in China, let alone the cancer risk. Consequently, this study was based on one general hospital in China to estimate the radiation dose in the thyroid attributable to the commonly used CT scans and the thyroid cancer risk in pediatric patients.
4. Discussion
In this study, we estimated the radiation doses received by pediatric patients with the three most commonly used different CT protocols at one hospital by abstracting the scanning parameters from the DICOM headers and we estimated LAR of thyroid cancer incidence using the BEIR VII model based on the Chinese population. The dose and risk estimates provide valuable information, which warn us to be more concerned about potential cancer risk from CT scans for pediatric patients and radiation protection. Compared with the baseline lifetime thyroid cancer incidence for newborn boys and girls in China which are 180 per 100,000 and 580 per 100,000, respectively [
29], the higher LAR of thyroid cancer incidence induced by chest CT were 7.8 per 100,000 for boys and 55.5 per 100,000 for girls, were rather lower. Meanwhile, the most exposed body part was head (nearly 64%), next was abdomen (19%), and the third was chest (11%) in this study, which was similar with the result of a survey from Switzerland [
30]. The total frequency of CT protocols that may cause high radiation dose for thyroid was very large, so even with a rather lower LAR, we cannot neglect the potential thyroid cancer risk from CT scans.
The radiation doses received in the thyroid varied with different CT protocols. The data showed clearly that the thyroid dose with chest CT scans was higher than that with paranasal sinus or head CT scans, and the results were similar to the published reports. In UK, based on a series of hybrid computational human phantoms coupled with Monte Carlo radiation transport, the thyroid dose from head and chest scans were 2.6 m
Gy and 26.7 m
Gy, respectively, for a newborn baby, 0.5 m
Gy and 21.8 m
Gy for 1-year-old children, 0.6 m
Gy and 17.7 m
Gy for 5-year-old children, and 0.8 m
Gy and 20.3 m
Gy for 10-year-old children [
7]. In Hong Kong, Feng
et al. used a 5-year-old anthropomorphic phantom to measure levels of 2.52 m
Gy for head CT scans and 3.40 m
Gy for chest CT scans [
20]. Journy
et al. applied CT-expo software to estimate the thyroid dose was 8 m
Gy with chest CT scans for children aged <1 year and 7 m
Gy for children older than 1 year but younger than 10 years [
31]. Through the comparison with these estimated by different methodology, the thyroid dose caused head CT scans was similar in different studies, but for chest CT scans, the radiation dose was relatively lower in this study. That may because the effect of z-axis overscanning which is defined as extra-exposure along the
z-axis extends beyond the planned scan region was not to be considered in this study, although
z-axis overscanning may contribute thyroid organ dose particularly in pediatric chest CT (spiral mode). In the previous report [
32], the percentage in the effective dose values between axial and helical scans (z overscanning was taken into account) were up to 43% for the head-neck CT, 70% for the chest CT. Therefore, the estimation of thyroid dose from chest and paranasal sinus CT scans was conservative in this study. Additionally, another factor that underestimated the thyroid dose was the automatic exposure control (AEC) which was used by the CT scanner. We selected the average of tube current for each patient, but there was reported that estimating thyroid doses using the average tube current rotation time product values (CTDI
vol) would have underestimated thyroid doses by 44% from CT scan on neck regions [
14], so the influence of AEC for the radiation organ dose should be considered in the future research. The abstracted parameters were no difference among different age groups in this hospital, although many researches stressed that optimizing the parameters of CT scans was one of important factors to radiation dose reduction [
33,
34], especially for pediatric patients. This indicated that physicians should be trained to reduce the parameters for pediatric patients, thus decreased radiation dose.
Using the projected risk model from the BEIR VII report, we employed the Chinese demographic and cancer incidence data to estimate the LAR of thyroid cancer incidence. In this study, for a 5-year-old child who underwent head CT once, the LAR was 8.8 per 100,000 for girls and 1.3 per 100,000 for boys; for chest CT, the LAR was 22.4 per 100,000 for girls and 3.2 per 100,000 for boys. For each protocol, the LAR of thyroid cancer incidence for girls was several times higher than that for boys, which is mainly due to the higher baseline cancer incidence in female (about 3 times) and the higher radiation-related risk in this model (about 2 times). Making the comparison with the study from Hong Kong, Feng
et al. [
20] reported that the LAR of thyroid cancer incidence was 4 per 100,000 in 5-year-old boys for each chest CT scan and 21 per 100,000 for girls, as well as 3 per 100,000 for boys from each head CT scan and 15 per 100,000 for girls, we also found that LAR estimated for Chinese girls was higher than Hong Kong girls, while for boys was lower than Hong Kong, which is mainly due to the differences in life table and cancer statistics. The LAR of thyroid cancer incidence decreased with the growth of exposed age in this study, as same with the report that the gender-averaged LAR incidence of thyroid cancer from chest CT scans, from birth to age 100 per 100,000 children, was 46 for 1-month-old children, 43 for 1-year-old children, 31 for 5-year-old children, and 20 for 10-year-old children [
31].
There were some limitations in this study. First, this study was only based on one single general hospital in China and one CT scanner, although this situation was common in most hospitals in China, and the more comprehensive study including multi-institution and more CT scanners should be considered in the future that will be more scientific and convincing. Second, the dose estimation used ImPACT software which was based on the bodies of European patients, not from Chinese ones, which may cause some uncertainty, to some extent. Third, there were also uncertainties in estimating LAR by using the method in the BEIR VII report. For thyroid cancer, the multiplicative model was used, reflecting the international basis of the Ron study [
35], so the ERR between these populations and Chinese were same. We made no correction for the value of dose and dose rate effectiveness factor (DDREF). Land
et al. [
36] considered the DDREF and the relative biological effectiveness factor appropriate to medical X-ray should have opposite and approximately equal effects on risk. The International Commission on Radiological Protection [
37] recommended a DDREF of 2 and the effectiveness per unit absorbed dose of standard X-rays may be about twice that of high-energy photons. While, BEIR VII report recommends the value of 1.5 and the value of relative biological effectiveness (RBE) is still an open question, as the general reference value is 2 or 3. Thus, the LAR of thyroid cancer incidence for patients may be underestimated by at least 30% in our study, if 1.5 for DDREF and 2 for RBE are appropriate.