Association of Radiation Dose With Prevalence of Thyroid Nodules Among Atomic Bomb Survivors Exposed in Childhood (2007-2011)

Importance  Few studies have evaluated the association of radiation dose with thyroid nodules among adults exposed to radiation in childhood.

Objective  To evaluate radiation dose responses on the prevalence of thyroid nodules in atomic bomb survivors exposed in childhood.

Design, Setting, and Participants  This survey study investigated 3087 Hiroshima and Nagasaki atomic bomb survivors who were younger than 10 years at exposure and participated in the thyroid study of the Adult Health Study at the Radiation Effects Research Foundation. Thyroid examinations including thyroid ultrasonography were conducted between October 2007 and October 2011, and solid nodules underwent fine-needle aspiration biopsy. Data from 2668 participants (86.4% of the total participants; mean age, 68.2 years; 1213 men; and 1455 women) with known atomic bomb thyroid radiation doses (mean dose, 0.182 Gy; median dose, 0.018 Gy; dose range, 0-4.040 Gy) were analyzed.

Main Outcomes and Measures  The prevalence of all thyroid nodules having a diameter of 10 mm or more (consisting of solid nodules [malignant and benign] and cysts), prevalence of small thyroid nodules that were less than 10 mm in diameter detected by ultrasonography, and atomic bomb radiation dose-responses.

Results  Thyroid nodules with a diameter of 10 mm or more were identified in 470 participants (17.6%): solid nodules (427 cases [16.0%]), malignant tumors (47 cases [1.8%]), benign nodules (186 cases [7.0%]), and cysts (49 cases [1.8%]), and all were significantly associated with thyroid radiation dose. Excess odds ratios per gray unit were 1.65 (95% CI, 0.89-2.64) for all nodules, 1.72 (95% CI, 0.93-2.75) for solid nodules, 4.40 (95% CI, 1.75-9.97) for malignant tumors, 2.07 (95% CI, 1.16-3.39) for benign nodules, and 1.11 (95% CI, 0.15-3.12) for cysts. The interaction between age at exposure and the dose was significant for the prevalence of all nodules (P = .003) and solid nodules (P < .001), indicating that dose effects were significantly higher with earlier childhood exposure. No interactions were seen for sex, family history of thyroid disease, antithyroid antibodies, or seaweed intake. No dose-response relationships were observed for small (<10-mm diameter) thyroid nodules.

Conclusions and Relevance  Radiation effects on thyroid nodules exist in atomic bomb survivors 62 to 66 years after their exposure in childhood. However, radiation exposure is not associated with small thyroid nodules.

A high risk of thyroid cancer has been reported^1-6 among people who were exposed to ionizing radiation in childhood. The long-term consequences of radiation exposure in children are an area of public health concern. A prospective study⁴ in Ukraine after the Chernobyl nuclear reactor accident suggested an increased risk of adult-onset thyroid cancer. A study^6,7 of atomic bomb survivors reported that an increased risk of thyroid cancer associated with childhood exposure has persisted for more than 50 years after exposure. Such studies focusing on adults who were exposed in childhood are useful for predicting the long-term effects of childhood radiation exposure.

Thyroid ultrasonography has been used for the detection of thyroid cancer for children after the accidents at the Chernobyl and Fukushima, Japan, nuclear power plants. In general, recent advances in ultrasonographic technology have shown thyroid solid nodules and cysts to be common, detected in 17% to 67% of adults⁸ and 57.5% of children.^9,10 Although several studies^11-16 evaluating thyroid nodules detected by ultrasonography have been conducted among radiation-exposed populations, dose responses based on individual thyroid radiation dose have rarely been investigated among adults who were exposed in childhood.^17-19 Moreover, although most thyroid nodules detected by ultrasonography were small (<10 mm or <15 mm in diameter),^9,10 no information has been available regarding whether radiation affects the development of small thyroid nodules. It also remains unclear whether other clinical factors, such as age, sex, iodine intake, family history of thyroid disease,²⁰ and presence of autoimmune thyroiditis,²¹ contribute to an association between thyroid nodules and radiation exposure.

We had previously performed a study¹⁹ between 2000 and 2003 on 3185 atomic bomb survivors exposed at all ages to evaluate radiation effects on the thyroid. However, we could not fully examine the effects of radiation exposure in childhood because of the relatively small sample of 709 participants exposed when they were younger than 10 years.

To evaluate atomic bomb radiation effects on the thyroid long after childhood exposure, we conducted a comprehensive survey regarding thyroid disease in 2668 atomic bomb survivors 62 to 66 years after radiation exposure when they were younger than 10 years. The present study investigated radiation dose-response relationships for the prevalence of thyroid nodules and modifying factors in this population.

This study was reviewed and approved by the Human Investigation Committee of the Radiation Effects Research Foundation. Written informed consent was obtained from all participants prior to their inclusion in the study. The participants did not receive financial compensation.

The Adult Health Study (AHS) is a clinical program established in 1958 by the Radiation Effects Research Foundation, formerly the Atomic Bomb Casualty Commission, as a subset of the Life Span Study cohort²² to examine the late effects of atomic bomb exposure. The AHS biennial health examinations present clinical information complementary to data from the death and tumor registries.^23,24 In 2007, a total of 1961 individuals of the Life Span Study cohort who were younger than 10 years at exposure and lived in or near Hiroshima or Nagasaki were added to the AHS cohort to focus on studies of radiation effects as a childhood exposure.²⁵

A total of 3102 AHS cohort members who were younger than 10 years at exposure visited the Radiation Effects Research Foundation for biennial health examinations between October 1, 2007, and October 19, 2011, with no knowledge of the thyroid disease study and were asked to participate in that study at the time of the examination. A total of 3087 participants (99.5%) agreed to and completed the thyroid examination. We excluded 419 participants (367 who were exposed in utero and 52 with unknown radiation doses according to the Dosimetry System 2002²⁶), leaving 2668 participants for analysis (86.4% of the total participants; 714 of the original AHS cohort participants and 1954 of the added AHS cohort participants). Table 1 reports the characteristics of the study participants. The Dosimetry System 2002²⁶ was used in estimating thyroid radiation doses. eTable 1 in the Supplement indicates the number of participants classified by thyroid radiation dose and age at exposure.

We used the same examination and laboratory methods as applied in a previous study.¹⁹ Participants visited the Hiroshima and Nagasaki Laboratories for biennial clinical examinations. A trained nurse recorded information using a questionnaire on detailed current and past thyroid disease and thyroid medication, family history of any thyroid disease among first- or second-degree relatives, and the amount of seaweed intake (daily or not daily). A blood sample was obtained, and antithyroid peroxidase antibody and antithyroglobulin antibody levels were measured by enzyme-linked immunosorbent assay (Medical & Biological Laboratories). Participants were classified as being positive for antithyroid antibodies if the serum concentration of either antithyroid peroxidase antibody or antithyroglobulin antibody was 10 IU/mL or greater.

All participants underwent thyroid ultrasonography (Logiq 500 Pro; GE Yokogawa Medical Systems) by certified ultrasonographers using 11-MHz probes to detect solid nodules and cysts that were 5 mm or larger in maximum diameter. The ultrasonographers were trained at the outset of the study to ensure the uniformity of the procedures. All recorded films were reviewed by radiologists.

Participants with solid nodules that were 10 mm or larger in diameter were referred to hospitals in Hiroshima and Nagasaki, and ultrasonography-guided fine-needle aspiration biopsy was performed upon participants’ agreement to determine cytologic diagnoses. All diagnoses of thyroid nodules were made by investigators (M.I. and N.S.) blinded to thyroid radiation doses.

Participants with nodules that were 10 mm or more in diameter were classified as thyroid nodule cases because persons having smaller nodules have an excellent prognosis,⁸ and most nodules smaller than 10 mm are not considered as candidates for biopsy.²⁷ Participants with a history of thyroid nodule surgery and histologic confirmation of the presence of the nodule also were classified as thyroid nodule cases regardless of ultrasonographic findings at the examination. These classification criteria are consistent with those used in our previous studies.^19,28,29 Thyroid nodules were divided into solid nodules and cysts. Solid nodules were further divided into malignant tumor, benign nodule, and nodules not otherwise specified (NOS) according to results of cytologic or histologic examination. If thyroid surgery was performed, pathology reports or data provided from the Hiroshima and Nagasaki tumor and tissue registries³⁰ were reviewed. Detailed descriptions of diagnostic criteria have been reported.¹⁹

To separately evaluate the effects of radiation on smaller thyroid nodules, we identified participants with thyroid nodules of less than 10 mm (5- to 9-mm diameter). Participants with thyroid nodules that were 10 mm or more in diameter or who had a history of thyroid nodule surgery were not included in these cases. Thyroid nodules with a diameter of less than 10 mm were divided into categories of solid nodules and cysts.

We used the same statistical method used in previous publications^19,29 to analyze dose responses for thyroid nodules. We defined p as thyroid nodule prevalence that could depend on the city (Hiroshima or Nagasaki), sex, age at exposure, and radiation dose. For the analysis of dose responses, we assumed the following linear excess odds ratio (EOR) model^19,31 in dose for prevalence p,

where BGM is the log-linear background model, or the log-linear odds model at 0 Gy, in terms of city (0 for Hiroshima, 1 for Nagasaki), sex (0 for male, 1 for female), age at exposure (in years) − 5, and their second-order interactions; and EM is the log-linear effect modification in terms of the main effects of city, sex, and age at exposure. Therefore, the linear EOR models the OR between the 0-Gy group and the exposed group at dose = d Gy as linear in dose, with a log-linear effect modifier EM term possibly depending on the participant’s city, sex, and age at exposure. Radiation dose d is the Dosimetry System 2002–weighted thyroid dose in grays, with an assigned relative biological effectiveness for neutrons = 10 as the sum of the γ thyroid dose and 10 times the neutron thyroid dose. These thyroid doses were truncated at 4 Gy and adjusted for a 35% dose measurement error,^32,33 which reduces radiation risk-estimation bias. This adjustment for the radiation dose estimation error is the same method of analysis used in the previous report.¹⁹ Parameter β is a radiation dose-response regression variable per gray. Because the EOR, which is approximately the excess relative risk when the prevalence is small, is linear in radiation dose, the EOR per gray can be written as β at the covariate value of the log-linear EM term = 0, or EM term = 1 at the specified covariates in the EM term. For the binary data of thyroid nodules of individuals in the cohort, the GMBO binary regression program in Epicure, version 1.81, software³¹ was used to obtain maximum likelihood estimates of the variables. To ascertain the linear dose-response model, the variable in the linear quadratic EOR model, RR = 1 + EM • (β • d + γ • d²), was tested for γ = 0 of the curvature of the dose response. We conducted a complete data analysis, which would not affect the regression parameter estimates owing to the small number of missing data (52 dose-unknown participants) compared with the complete data (2668 participants).

As the best-model selection criterion, we used the Akaike information criterion (AIC) model selection,^34,35 defined as deviance of the fit plus 2 times the number of variables used in the fit. Using the minimum AIC model selection procedure, we selected the best model for each thyroid nodule under the hierarchical rule that if a 2-factor interaction term is included in the model, the 2 main effects are included as well. The best model selection chosen with the AIC is the best model in terms of prediction. Model selection was carried out separately for the BGM and EM parts of the analysis. Model selection for the log-linear BGM was made in fitting all the submodels in which we chose the hierarchical BGM mentioned above in terms of the AIC, and model selection for the log-linear EM term was then made. Finally, we evaluated whether the obtained model offered the best AIC (ie, attaining the minimum AIC value). The linear EOR model used in the present analysis gave a better or equal fit to the data compared with the usual linear logistic model in terms of AIC model selection. Because small numbers of outliers, if any, do not affect the regression parameter estimates owing to the large number of data for analysis, we did not make regression diagnostics in each analysis. The attributable fraction due to radiation for each thyroid nodule was calculated, and the 95% Wald CI calculated from the binomial distribution with the probability of an attributable fraction was constructed.

We performed analyses for selected characteristics associated with the background prevalence of all thyroid nodules in 2622 participants, excluding 46 individuals with missing family history (n = 33) and/or seaweed intake information (n = 19). The family history, antithyroid antibodies, and seaweed intake indicators were individually included in the best BGM, and we checked whether these indicators were significant effect modifiers when including the EM term individually.

When we constructed 95% CIs by dose group or by age-at-exposure group, the dose category cut points were defined as 0.005, 0.1, 0.5, 1.0, and 2.0 Gy (to convert to rad, multiply by 100), with participants exposed to doses from 0 to 0.005 Gy serving as a reference group. Age-at-exposure category cut points were defined as 3 and 6 years. For known-dose participants, the mean age at exposure was 4.1 years and the mean age at examination was 68.2 years, a difference of approximately 64 years. The 95% CIs were constructed using likelihood ratios, and the P values were 2-sided based on χ² likelihood ratios.

Thyroid nodules were identified in 470 participants (17.6%). Detailed prevalence data are presented in Table 2. Sixty-three participants had undergone thyroid surgery before this study: 37 for a malignant tumor, 24 for a benign nodule, and 8 for nodules NOS, including 6 participants who had both a malignant tumor and a benign nodule. Among 407 nodule cases without a history of thyroid nodule surgery, 329 cases had a single nodule. Most malignant tumors (42 cases) were papillary carcinoma. Three cases of follicular carcinoma, 2 cases of poorly differentiated carcinoma, and no cases of anaplastic or medullary carcinoma were detected. Among the benign nodule group, 13 and 15 cases were histologically confirmed as follicular adenoma and adenomatous goiter, respectively, and 164 cases had cytologically diagnosed benign nodules (6 cases had 2 types of benign nodules). Overall, among 427 participants with solid nodules, 259 individuals underwent aspiration biopsy or surgery at least once, and we obtained cytologic or histologic results on 239 participants. No thyroid nodules produced toxic effects.

The risks associated with the background rates of all thyroid nodules demonstrated a significant female predominance (P < .001) (eTable 2 in the Supplement). Participants with a family history of any thyroid disease had a higher prevalence (P = .003). However, age at exposure, city, presence of antithyroid antibodies, and seaweed intake did not significantly affect the prevalence of thyroid nodules. We also evaluated the background prevalence between 2 groups of study participants: the original AHS participants and the AHS participants included in 2007. The prevalence of all thyroid nodules was not significantly affected by membership in either group (P > .50).

The prevalence of all nodules, solid nodules, malignant tumors, benign nodules, nodules NOS, and cysts was significantly associated with the thyroid radiation dose. The EORs per gray were 1.65 (95% CI, 0.89-2.64) for all nodules, 1.72 (95% CI, 0.93-2.75) for solid nodules, 4.40 (95% CI, 1.75-9.97) for malignant tumors, 2.07 (95% CI, 1.16-3.39) for benign nodules, 0.68 (95% CI, 0.16-1.53) for nodules NOS, and 1.11 (95% CI, 0.15-3.12) for cysts (Table 2 and Figure). Dose responses for all categories of thyroid nodules were demonstrated to be consistent with monotone linearity by tests of quadratic terms in dose. Attributable fractions (95% Wald CIs) for all nodules, solid nodules, malignant tumors, benign nodules, nodules NOS, and cysts were 7.1% (95% CI, 4.8%-9.4%), 9.3% (95% CI, 6.6%-12.0%), 42.1% (95% CI, 28.0%-56.2%), 19.0% (95% CI, 13.3%-24.7%), 7.7% (95% CI, 4.2%-11.2%), and 14.9% (95% CI, 4.9%-24.9%), respectively.

Among the 2668 participants analyzed, 23 individuals had a history of radiotherapy to the neck for treatment of disorders other than thyroid disease. We further analyzed dose responses among 2645 participants after excluding these 23 participants and obtained similar EORs per gray: 1.70 (95% CI, 0.93-2.71) for all nodules, 1.77 (95% CI, 0.96-2.82) for solid nodules, 4.86 (1.95-11.16) for malignant tumors, 2.14 (95% CI, 1.20-3.49) for benign nodules, 0.67 (95% CI, 0.16-1.54) for nodules NOS, and 1.15 (95% CI, 0.16-3.21) for cysts.

The interaction between age at exposure and dose was significant for the prevalence of all nodules, solid nodules, and nodules NOS, indicating that dose effects were significantly higher in individuals exposed in infancy (Table 3) compared with older children. The EORs per gray for all nodules, solid nodules, and nodules NOS changed by 0.79-fold (95% CI, 0.65-0.92), 0.81-fold (95% CI, 0.67-0.95), and 0.75-fold (95% CI, 0.52-1.00) per 1-year increase in age at exposure, respectively. No interaction between sex and dose or city and dose was seen in the prevalence of any categories of thyroid nodules (all P > .17).

We performed further analyses to evaluate the effects of family history of any thyroid disease, presence of antithyroid antibodies, or seaweed intake on EOR per gray. The EORs per gray did not vary significantly by any of these factors in terms of the prevalence of any category of thyroid nodules (all P > .15), except that family history for benign nodules and presence of antithyroid antibodies for nodules NOS were suggestive as effect modifiers (all P = .07). We also evaluated an EM of 2 groups of study participants: the original AHS participants and the AHS participants included in 2007. The EORs per gray were not significantly affected by membership in either group (all P > .50).

We further analyzed the prevalence of small thyroid nodules (5-9 mm in diameter). The prevalence of thyroid nodules that were less than 10 mm in diameter was higher than that of nodules that were 10 mm or more in diameter (Table 4). In contrast to the larger thyroid nodules (Table 2 and Figure), for nodules that were less than 10 mm in diameter, the prevalence of all nodules, solid nodules, and cysts was not associated with the thyroid radiation dose (EORs per gray, −0.08, −0.09, and −0.09, respectively) (Table 4).

To our knowledge, this is the first report from the comprehensive thyroid nodule screening study of Hiroshima and Nagasaki atomic bomb survivors to focus on individuals exposed when they were younger than 10 years. We observed significant radiation effects on thyroid nodules 62 to 66 years after radiation exposure. Risk estimates (EORs per gray) for atomic bomb survivors exposed in childhood (aged 0 to <10 years) in the present study were higher than for those exposed at all ages (0-40 years) in our previous study,¹⁹ particularly for malignant tumors (4.40 and 1.95, respectively). This finding reflects a high risk of malignant tumor development in atomic bomb survivors exposed in childhood. The EOR per gray of benign nodules was also increased compared with that in the previous study,¹⁹ although the increase was not as large (2.07 in the present study vs 1.53 in the previous study) compared with that for the malignant tumor. Compared with risk estimates of ultrasonography-detected thyroid nodules in other populations exposed in childhood, the estimated risk in the present study participants was higher than that from the Hanford nuclear site (estimated slope per gray for benign nodules, −0.008; 249 cases per 3440 participants [7.2%] aged 1-17 years at exposure)¹⁷ and the Semipalatinsk test site (EOR per gray for nodules, 0.74; 916 cases per 2994 participants [30.6%] aged 0-21 years at exposure).¹⁸ The discrepancy might relate to ages at exposure, types of radiation exposure (acute vs protracted), or estimation errors of thyroid doses.

Although our results indicate a linear dose-response relationship in thyroid nodules, careful interpretation is necessary for the radiation risk at low doses. There is uncertainty regarding the radiation risk at low doses among atomic bomb survivors because we have insufficient information about residual radiation and medical radiation exposure as well as sociodemographic and lifestyle factors.³⁶

Among people in the present study who were exposed to radiation when they were younger than 10 years, a younger age at exposure was associated with a higher risk of thyroid nodules. This result indicates that the thyroids of younger children might be more sensitive to radiation. Although some molecular studies have been conducted for thyroid cancer in atomic bomb survivors,^37,38 the mechanisms of radiation effects on thyroid cancer and benign nodules in people exposed in childhood remain unclear.

Genetic factors are considered important in the development of thyroid nodules, particularly thyroid cancer.^39-42 The present study observed a significantly higher risk of thyroid nodules when family members had a history of thyroid disease, but this family history did not affect radiation risk estimates, which are inconsistent with those of an earlier study.⁴³ A limitation of the present study is that a family history of thyroid disease may be associated not only with genetic factors but also with atomic bomb radiation exposure in family members.

Iodine intake may represent a dietary factor affecting the development of thyroid nodules and cancer^44-46 and radiation-related thyroid cancer,^5,47 but the evidence remains inconclusive. Because urinary iodine excretion varies in accordance with the quantity and frequency of seaweed consumption in Japan,⁴⁸ we asked participants about their seaweed intake as a surrogate marker of iodine intake. No influence of seaweed intake on the prevalence of thyroid nodules or radiation risk estimates was seen. Additional studies using urinary iodine concentration over time are needed.

Because of recent advances in the quality of ultrasonographic imaging, small thyroid nodules (<10 mm) are being detected more frequently than are nodules that are 10 mm or larger in adults⁸ as well as children.^9,10 In the present study of adults 6 decades after radiation exposure, the prevalence of small thyroid nodules was not associated with the thyroid radiation dose among atomic bomb survivors, indicating that there is a discrepancy in radiation dose responses between nodules that are 10 mm or more and those that are less than 10 mm. Thyroid nodules detected by ultrasonography usually consist of heterogeneous benign lesions, such as adenoma, hyperplastic nodule, cyst, multinodular goiter, or nodular changes of autoimmune thyroid disease,⁴⁹ although some microcarcinomas have been reported.⁵⁰ Considering the high prevalence of small thyroid nodules in general populations that have not been exposed to radiation,^8,9 we speculate that a large fraction of small thyroid nodules detected in this study likely represented nonprogressive lesions, such as nonneoplastic lesions. In general, 2 steps are considered in the natural history of multinodular goiter: nodule formation and nodule growth.⁵¹ Radiation exposure may affect nodule growth rather than nodule formation.

The major strengths of this study were the availability of estimated thyroid radiation doses and nearly complete ascertainment of thyroid nodule cases, with few false-negative cases using standardized ultrasonographic examination and systematic tumor registry mechanisms. The statistical power in our cohort was sufficient because the observed EORs per gray for solid nodules (1.72) and cysts (1.11) were far larger than the calculated EORs per gray (approximately 0.2 and 0.7 per Gy, respectively) detectable with a statistical power of 90% at a 5% significance level with a variance inflation factor⁵² of 10%. A key limitation of the study was the cross-sectional design. A survival bias also exists in this study.⁵³ We believe that a possible bias owing to motivation did not affect the overall conclusion because the participation rate was high (99.5%).

The present study, conducted 62 to 66 years after the Hiroshima and Nagasaki atomic bombings, revealed that radiation effects on thyroid nodules exist in atomic bomb survivors exposed when they were younger than 10 years. Radiation effects were significantly higher in those exposed at a younger age, but other significant modifiers were not identified. Radiation exposure was not associated with small thyroid nodules.

Accepted for Publication: October 17, 2014.

Published Online: December 29, 2014. doi:10.1001/jamainternmed.2014.6692

Correction: This article was corrected on January 6, 2020, to add a second location for one institution, which altered the affiliation for 4 authors, and to correct the location of another institution in the Author Affiliations section.

Corresponding Author: Misa Imaizumi, MD, PhD, Department of Clinical Studies, Radiation Effects Research Foundation, 1-8-6 Nakagawa, Nagasaki 850-0013, Japan (misaima@rerf.or.jp).

Author Contributions: Dr Imaizumi had full access to all the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis.

Study concept and design: Imaizumi, Fujiwara, Akahoshi.

Acquisition, analysis, or interpretation of data: All authors.

Drafting of the manuscript: Imaizumi.

Critical revision of the manuscript for important intellectual content: All authors.

Statistical analysis: Nakashima.

Administrative, technical, or material support: Ohishi, Neriishi, Yamada, Tatsukawa, Takahashi, Usa, Kawakami, Hida.

Study supervision: Fujiwara, Ando.

Conflict of Interest Disclosures: None reported.

Funding/Support: This study was supported by the Radiation Effects Research Foundation research protocols 2-99 and 3-07. The Radiation Effects Research Foundation, Hiroshima and Nagasaki, Japan, is a public interest foundation funded by the Japanese Ministry of Health, Labour, and Welfare and the US Department of Energy.

Role of the Funder/Sponsor: The funders had no role in the design and conduct of the study; collection, management, analysis, and interpretation of the data; preparation, review, or approval of the manuscript; and decision to submit the manuscript for publication.

Disclaimer: The views of the authors do not necessarily reflect those of the Japanese and US governments.

Additional Contributions: We thank the people working on the Hiroshima and Nagasaki Tumor Registry and Tissue Registry. Shinichiro Ichimaru, PhD, and Sachiyo Funamoto, BSc, provided data preparation and statistical assistance, and Tomohiro Ikeda, BA, provided general assistance (Radiation Effects Research Foundation). There was no financial compensation for the services.