|Home | About | Journals | Submit | Contact Us | Français|
Radiation exposure is an established cause of clinical thyroid cancer, but little is known about radiation effects on papillary microcarcinoma (PMC) of the thyroid, a relatively common subclinical thyroid malignancy. Because the incidence of these small thyroid cancers has been increasing, it is important to better understand them and their relationship to radiation.
PMCs were identified in a subset of 7659 members of the Life Span Study of atomic-bomb survivors who had archived autopsy or surgical materials. We conducted a pathology review of these specimens and evaluated the histological features of the tumors and the association between PMCs and thyroid radiation dose.
From 1958 to1995, 458 PMCs were detected among 313 study subjects. The majority of cancers exhibited pathologic features of papillary thyroid cancers. Overall, 81% of the PMCs were of the sclerosing variant and 91% were nonencapsulated, psammoma bodies occurred in 13% and calcification was observed in 23%. Over 95% had papillary or papillary-follicular architecture and most displayed nuclear overlap, clear nuclei, and nuclear grooves. Several of these features increased with increasing tumor size, but no association was found with radiation dose. A significant radiation-dose response was found for the prevalence of PMCs (estimated excess odds ratio/Gy=0.57; 95% CI: 0.01-1.55), with the excess risk observed primarily among females.
Low-to-moderate doses of ionizing radiation appears to increase the risk of thyroid PMCs, even when exposure occurs during adulthood.
In parallel with the growing use of thyroid ultrasound, there has been an increase in the diagnosis of thyroid cancers, especially papillary microcarcinomas(PMC), in Japan(1) and other developed countries(2-6). PMC is defined by the World Health Organization as a tumor of 1 cm or less in diameter that is discovered incidentally(7). PMCs are frequently clinically indolent and are often detected incidentally during surgery for benign thyroid disease or at autopsy, but there have been numerous reports of local lymph node invasion and distant metastases(8-10). The inconsistent findings have resulted in controversy regarding treatment and the precise relationship of PMC with clinical papillary carcinoma of the thyroid(11).
Radiation is a known cause of thyroid cancer and papillary carcinoma is the principal thyroid cancer cell type linked to radiation, especially following childhood exposure(12). The radiation-related risk of clinical thyroid cancer and risk modification by age and other factors have been well characterized(12, 13). Little is known, however, about the association between radiation exposure and PMC.
This study was undertaken to characterize histological features of PMCs, with special emphasis on their relation to tumor size, and to assess the relationship of PMC and radiation exposure among a subgroup of members of the Life Span Study (LSS) of atomic-bomb survivors using DS02 thyroid organ dose estimates(14). The study is part of a broader effort by pathologists to ascertain and histologically verify benign and malignant thyroid tumors in the LSS.
The LSS cohort, described in detail elsewhere(15), consists of 93,000 atomic-bomb survivors from Hiroshima and Nagasaki and 27,000 city residents who were not present in the cities at the time of the bombings. PMCs were ascertained from 1958 through 1995 among the subcohort of 7,659 LSS members with autopsy or surgical pathology records available at Radiation Effects Research Foundation (RERF) or had tissue samples in the Hiroshima and Nagasaki tissue registries (Table 1). To identify other cancers occurring in these subjects, we linked the study file to the Hiroshima and Nagasaki tumor registries(16). We restricted the study period to 1958 through 1995 because the tumor registries were established in 1958 and there was a marked decline in number of autopsies carried out after 1995 (only 20 per year).
The primary source for ascertaining PMCs was the autopsy program conducted in the LSS cohort by the Atomic Bomb Casualty Commission (ABCC) and its successor the RERF. This program was most active from 1958 through 1988, during which time extensive efforts were made to conduct autopsies on as many deceased members of the LSS as possible. Autopsy rates were the highest in the 1960s, ranging from to 30 to 40%, but the rates declined thereafter and by the late 1970s autopsies were conducted on only 10% of deceased cohort members(17). Before 1961, autopsies were performed on deceased subjects referred by local physicians and hospitals, and the selection for autopsy reflected ABCC's research interests, especially malignancies. In 1961, the autopsy program was revised and expanded to obtain a more representative distribution of causes of deaths using a community-based network for prompt identification of LSS cohort member deaths based on information from numerous sources, including major hospitals, city registration offices, local funeral homes and crematories(18, 19).
Between 1958 and 1988, 4,718 deceased LSS subjects were autopsied at ABCC/RERF by staff pathologists following the standardized “complete” autopsy protocol in which all the major organs including the thyroid were examined. In complete autopsies, the entire thyroid gland was removed, inspected and palpated. When a tumor (or nodule) was suspected, sections were made from each nodule for further microscopic examination. When there were no nodules, slides representing the tissue structure were made from each of the lobes and the isthmus.
PMCs were also identified from the ABCC/RERF surgical-pathology program, which was active from the late 1950s through the late 1960s. The tissue registries, started in 1973, are a repository of tissue specimens and blocks sent from local hospitals and pathology laboratories. These registries were used for identifying thyroid microcarcinomas.
First, trained tumor registry staff examined all autopsy and surgical pathology diagnoses for LSS members and identified diagnoses of “latent”, “occult”, or “microcarcinoma” of the thyroid. Two study pathologists (Y.H. and N.T.) then independently reviewed all thyroid tissue slides and pathology records available for these potential cases. When the two pathologists did not agree on histological features or diagnosis, they met to review the materials and reach a consensus diagnosis. Using the WHO classification, PMCs were defined as thyroid carcinomas 1 cm or less in diameter(7). Multifocal (synchronous) PMCs in the same person were treated separately for analyses of patho-morphologic features of PMC. For the radiation risk analysis, only the largest tumor was considered. No new slides were cut for detecting PMCs, however; when multiple slides were available, they were all reviewed. The number of slides reviewed per person ranged from one to 35 with a mean of 5.3 and a median of 3.
DS02(14) weighted thyroid doses in Gy were defined as the gamma dose estimate plus ten times the neutron dose estimate. These estimates incorporate an adjustment to reduce risk estimation bias arising from random errors in individual dose estimates. This adjustment was computed using the methods described in(21) with the assumption of a coefficient of variation of 35% for individual estimates.
Descriptive analyses of tumor characteristics were based on the 7659 LSS members in the study population and the PMCs identified between 1958 and 1995 from any data source. Trend tests were performed to evaluate the relationship between histopathological features and tumor size.
Analyses of radiation dose response were restricted to a subset of the study population: LSS members who had complete autopsies performed at ABCC/RERF between 1958 and 1988, were in the city at the time of the bombs, had DS02 thyroid dose estimates, and did not have a cancer reported from the tumor registries before death. Only complete autopsies were included in the analysis because the thyroid gland was not always thoroughly examined in partial autopsies. Most analyses were limited to first primary PMCs because second or later primary tumors may be intrathyroidal metastases from larger, clinically evident thyroid carcinomas; may have resulted from cancer treatment given for an earlier cancer; and may reflect biased ascertainment since individuals with cancer were more likely to have been autopsied than others.
Prevalence odds ratios (ORs) were estimated for PMCs. Categorical descriptions of the dose response were based on thyroid dose categories of 5 – 99, 100 – 499, and 500 + mGy relative to the risk for subjects with thyroid dose estimates of <5 mGy. The odds of PMC occurrence was described using logistic regression models with effects of gender and age at death. Adjusting for age at exposure, calendar time, and city had no significant effect on the rates and did not markedly affect the radiation risk estimates. The excess OR per Gy (EOR/Gy) was estimated assuming a linear dose-response relationship. Gender, age at exposure, attained age, and when applicable prior history of malignancy, were considered as potential effect modifiers. All 95% confidence intervals were based on likelihood ratio bounds for the parameter estimates. The risk analyses were performed using the EPICURE software(22).
During the study period, 458 PMCs occurred among 313 subjects (Table 1). The number of PMCs per subject varied from 1 to 9; 238 subjects(76%) had single tumors, 59 subjects(19%) had 2 or 3 tumors, and the remaining 16 subjects(5%) had five or more tumors. The majority (378 or 82%) of the PMCs were identified from autopsies conducted between 1958 and 1988; 300 were diagnosed from complete autopsies and 78 from partial autopsies. In addition, 13 PMCs were identified from partial autopsies done outside ABCC/RERF after 1988. The ABCC/RERF surgical program during 1958-72 and the Hiroshima and Nagasaki tissue registries after 1973 contributed 67 additional PMCs.
Table 2 presents histological features of the PMCs and demographic information by tumor size. In the upper panel, we present the data for all 458 PMCs. To explore the possible role of intra-subject correlations in subjects with multiple tumors, we also present the 313 subjects who had a PMC (cases), with the largest tumor selected as the representative PMC (lower panel).
About 24% of the 458 PMCs were ≤1mm, 52% were 1.1-5 mm, and another 24% were 5.1-10 mm in diameter. The majority of PMCs exhibited classical pathologic features of papillary thyroid carcinoma. Overall, 81% of the microcarcinomas were of the sclerosing variant and only 9% were encapsulated. Since these were independent effects, the joint effects are the product of the two effects, e.g. 73% were nonencapsulated sclerosing tumors and 8% were encapsulated sclerosing tumors. The majority of PMCs had papillary architecture. Only 16 PMCs(3%) were follicular variants without any apparent papillary structure, although nuclear findings characteristic of papillary carcinoma, such as ground-glass appearance, remarkable nuclear grooves, and frequent nuclear inclusions often were recognizable in these cancers. Psammoma bodies occurred in 13% of the papillary microcarcinomas and calcification was observed in 23%.
Several pathological features of microcarcinomas were associated with tumor size. As tumor size increased, the proportion of encapsulated and sclerosing tumors increased (p-trend = 0.02 and <0.001, respectively). While fairly rare, the frequencies of psammoma bodies and calcification also rose with increasing tumor size (p-trend = <0.001 for each variable). Among nuclear features examined, about 95% of all tumors showed evidence of nuclear overlapping and clear nuclei, whereas 89% showed evidence of nuclear groove. The frequency distributions did not vary with tumor size. In contrast, only 47% of the PMC had nuclear inclusion bodies, but the frequency increased with increasing tumor size (p-trend <0.001). The distribution of tumor size differed by gender with females having larger tumors than males, so that the F:M ratio increased from 1.3 for the smallest tumors to 3.2 for the largest tumors. No significant difference in tumor size was observed by age at death.
As seen in the lower panel of Table 2, the results for the 313 subjects with independent PMCs were very similar to those for all PMCs, but the difference in the tumor size distribution between male and females was not statistically significant.
When we evaluated the dependence of number of tumors, morphologic features or tumor size on radiation dose (Table 3), we found no evidence of significant relationships (P=0.4, 0.6 and >0.5, respectively).
As would be expected, the number of PMCs detected per person tended to increase with number of slides examined. There was a significant association (R2 = 0.16, P < 0.001) between thyroid dose and the number of slides examined. This correlation arose largely because the three cases with the largest number of slides examined (20, 21, and 25 slides with 7, 3, and 1 PMC, respectively) had three of the five highest doses among the cases (2.5 Gy, 3.4 Gy, and 1.7 Gy). When these three extreme cases were removed there was no significant correlation (R2=0.08, P=0.4) between thyroid dose and the number of slides. Although this association raises the possibility of biased selection, we feel this is unlikely since the pathologists performing the autopsies were unaware of the subjects' distance from the hypocenter or dose estimate and had no particular interest in thyroid tissue.
We restricted the radiation risk analyses to complete autopsies performed at ABCC/RERF during 1958-1988. Among the 4718 LSS members, 196 had 300 PMCs(Table 1). There were 149 PMC cases out of the 3523 subjects who were in the city at the time of the bombs with thyroid dose estimates. We excluded subjects with a prior cancer, leaving 2372 subjects among whom 106 (68 females and 38 males) had a PMC detected at autopsy resulting in a prevalence of 4.5%.
The prevalence by city and other demographic variables (Table 4) showed a slightly higher occurrence in Hiroshima(4.6%) than Nagasaki(3.9%) and among women(5.5%) than men (3.4%). Because the autopsied subjects were deceased the majority(84%) were older than 60 years. Since complete autopsies at ABCC/RERF were carried out during the 30-year period of 1958 through 1988, subjects tended to be exposed to the bombs at relatively old ages and few(9%) were exposed <30 years. The prevalence of PMC by age at exposure ranged from 2.1% to 6.1% and by age at death from 3.0% to 5.4 %, with no apparent age trends. While the number of autopsies decreased dramatically overtime, PMC prevalence was the same during the first two decades of the autopsy program.
Table 5 presents estimated ORs, adjusted for city, gender, age at exposure, age at death and calendar time, for first primary PMC by thyroid dose. The OR estimates increased significantly with increasing dose. The estimated EOR/Gy was 0.57 (95% CI: 0.01, 1.55). While the 106 cases considered in this analysis had no prior history of malignant tumor or other malignancies diagnosed only at autopsy, one case had a diagnosis of a benign thyroid disease prior to death. If the presence of the benign thyroid tumor was considered (though unlikely) to have increased a chance of discovering the PMC, then omitting this case (thyroid dose=3.2 Gy) reduced the EOR/Gy estimate to 0.45 (95% CI -0.09-1.38, p=0.11). Results for all 149 PMC did not differ from those for the 106 first primaries. Using the same adjustment variables and including a history of prior malignancy as an effect modifier, the EOR/Gy was 0.59 (95% CI 0.02–1.58).
Because of the small number of cases and the limited age-at-exposure range, analyses of effect modifiers were not very powerful and there was no evidence of age at exposure (P=0.4) or attained age (P>0.5) effects on the EOR. Significant gender effects were noted and the fit of the linear dose-response model was improved significantly when the EOR was allowed to depend on gender (p=0.01). There was no indication of a radiation effect among men (EOR/Gy= -0.3; 95% CI upper bound1.4, p=0.23), while that for women was quite large (EOR/Gy=2.4; 95% CI 0.3-4.5, p=0.002). Under the model that assumed no gender difference the estimated numbers of radiation-associated PMC's was 2.2 for men and 2.6 for women. When separate effects were estimated, the corresponding numbers of excess cases were -3.2 and 10.2. The dose category-specific EOR estimates for women were 1.00; 0.22 (95% CI -0.38-1.4); 1.1 (95% CI 0.02-3.3), and 1.4 (95% CI 0.03-4.4).
About 75% of PMC were non-encapsulated sclerosing tumors. Non-encapsulated sclerosing tumors were also the most common type (44%), though less frequent than ours, in another autopsy study in Japan(23). The prevalence of non-encapsulated tumors decreased as tumor size increased, whereas the frequencies of sclerosing tumors increased with increasing tumor size, as did frequencies of psammoma bodies, calcification and nuclear inclusion bodies. In a Brazilian study, the smallest papillary carcinomas presented most frequently as non-encapsulated and non-sclerosing tumors(24).
Among the atomic-bomb survivors, thyroid cancer was one of earliest solid cancers linked to radiation exposure(25-30). Studies of clinically undetected, or “occult”, carcinoma of the thyroid were first conducted by Sampson et al., who ascertained carcinomas with a diameter of <15 mm in serial sections of the thyroid from over 3,000 consecutive autopsies conducted in the LSS cohort during 1957-1968(31). A radiation effect was suggested for the smallest papillary carcinomas (<1 mm in diameter)(32). Because this was a special study and could possibly bias the results, we did not include these slides in the present study. The frequency of “occult” carcinoma found by Sampson et al.(31) was about 17%. This is much higher than that of around 3-6% (or about 4%) in the present study, and seems to reflect the serial sectioning methods used by Sampson et al.(32). In a later study of autopsies performed between 1951 and 1985 in Hiroshima LSS subjects, occult carcinoma (<15 mm in diameter) was found, without serial section, in about 3% of the subjects.(33) In that study, a significant increase in radiation-related odd ratio was found for occult carcinoma of the thyroid. In neither of these studies, however, were the analyses restricted to first primary tumors, as they were in the present study.
We found no significant relationship between the number, size or morphologic features of PMCs and radiation dose. In evaluating the radiation effects on PMC, we restricted the analyses to first primary PMCs, minimizing the possible effects of intra-thyroid metastases from larger papillary carcinomas of the thyroid or other organs, potential ascertainment bias in cancer patients and possible treatment effects for previous cancers. This strictly defined study endpoint was possible because of the availability of data from complete autopsies conducted at RERF using the same protocol. A limitation of this study was that the number of PMCs detected at autopsy was related to the number of the thyroid gland slices made, but we believe that the confounding effect of this association was diminished by using only one PMC per person. Since a standard autopsy protocol was used for the complete autopsies conducted at ABCC/RERF, there will not be a bias in the dose-response, however, the absolute number of PMCs is dependent on the autopsy method and therefore it is difficult to compare the prevalence of PMCs in one study population to another.
While young age at exposure is an important effect modifier for the radiation-related risk of thyroid cancer, there may also be some risk associated with adult exposure. In an analysis of thyroid cancer clinically diagnosed between 1958 and 1998 in the LSS cohort the ERR/Gy was estimated to be 0.6 for survivors exposed at age 30(16). This estimate is comparable to the excess EOR/Gy of 0.57 for PMC estimated from this study. In a new analysis of all survivors who were at least twenty years old at the bombings, the ERR/Gy was 0.70 (95% CI 0.20-1.5) for females and was -0.25 (95% CI <0-0.35) for males(34). A large gender difference was also observed in our study of autopsy detected PMCs. Radiation-related risks of thyroid cancer have also been observed among adult patients treated with radiation for benign conditions of the cervical spine(35) and for cancer. Studies of second thyroid cancers among adult cancer patients receiving radiotherapy have generally been limited by methodological problems, e.g. small number of second cancers in the radiotherapy group, incomplete information regarding all treatments, lack of thyroid doses, and comparisons made to the general population rather than between irradiated and non-irradiated patients. Evaluations of breast cancer patients have been somewhat more informative since they often have addressed the issue of the potential role of radiotherapy in the development of second primary thyroid cancers. Huang et al.(36) reported an elevated risk of thyroid cancer in a subgroup of breast cancer patients who were thought to have received particularly high doses to the thyroid, however, this relationship was not found in other studies of breast cancer patients(37-39).
To our knowledge, this is among the largest radiation-exposed populations in which non-clinically detected tumors were identified, yet the number of first primary PMCs available for study of radiation effects was relatively small. The radiation dose response, though significant, was still susceptible to assumptions on case eligibility. Our findings of a higher EOR for women with no indication of a radiation effect for men and of a statistically significant radiation dose response for PMC, even when exposure occurred during adulthood, are provocative and should be investigated in a larger study.
In summary, the PMCs were primarily non-encapsulated sclerosing tumors and the majority exhibited classical pathologic features of papillary thyroid cancer. The female-to-male ratio of PMC increased with increasing tumor size so that for tumors >5 mm, the ratio reached three-fold which is similar to what is found for clinically detected papillary thyroid cancer. Adult exposure to low to moderate doses of ionizing radiation appears to increase the risk of thyroid PMCs detected at autopsy.
We would like to thank the pathologists in Hiroshima and Nagasaki who supported the pathology review and Sir Dillwyn Williams, University of Cambridge, and Dr. Atsuhiko Sakamoto, Kyorin Medical School, Tokyo who collaborated with us in reviewing sampled slides for this study. The Radiation Effects Research Foundation is a private nonprofit foundation funded by the Japanese Ministry of Health, Labour and Welfare and the US Department of Energy, the latter in part through the National Academy of Sciences. The publication was supported by RERF Research Protocol RP 6-91. This research was partially supported by the US National Cancer Institute intramural research program and contract number NO1-CP-31012-66.
Sources of support: RERF Research Protocol RP 6-91; NCI contact number NO1-CP-31012-66; Intramural Research Program of the NIH, NCI, Division of Cancer Epidemiology and Genetics.