Chernobyl-related thyroid cancer: what evidence for role of short-lived iodines?

Over 500 cases of thyroid cancer were diagnosed in Belarus between 1986 and 1995 among persons exposed as children (under 15 years of age) to radioactive contamination from the Chernobyl nuclear accident. There is little doubt that radioactive iodine isotopes emitted during the nuclear explosion and subsequent fire were instrumental in causing malignancy in this particular organ. Comparison of the observed geographic distribution of Chernobyl-associated thyroid cancer incidence rates by districts with contamination maps of radioactive fallout shows a better fit for estimated 131I contamination than for 137Cs. Because 131I used for medical purposes had not been considered carcinogenic in humans in the past, and in view of the unusually short latency period between exposure and clinical manifestation of cancer, it is suspected that not only 131I but also energy-rich shorter-lived radioiodines may have played a role in post-Chernobyl thyroid carcinogenesis. Measurements of iodine isotopes are not available, but reconstruction of geographic distributions and estimations of radioactive fallout based on meteorological observations immediately following the accident could provide a basis for comparison with the distribution of thyroid cancer cases. In this paper, data from the Epidemiological Cancer Register for Belarus will be used to show geographic and time trends of thyroid cancer incidence rates in the period from 1986 to 1995 among persons who were exposed as children, and these will be compared with the estimated contamination by radioiodines. Tentative conclusions are drawn from the available evidence and further research requirements discussed.


Introduction
First suspicions of an increased incidence of thyroid cancer in children arose in Ukraine, where three well examined cases had occurred close to Chernobyl in 1990 (1). A year later, Belarusian scientists reported an increase of cases of thyroid cancer in children as well (2). However, scientists in the West remained skeptical, and a number of possible artifacts were brought forward (3)(4)(5). These included active case finding and previous underreporting. A main argument was the unexpected geographical distribution: There were only a few cases of thyroid cancer in Mogilev Oblast, some parts of which are known for their high 137Cs radioactive contamination. This argument later lost its validity, as it appeared that contamination with radioactive iodine did not strongly parallel cesium deposition because of changes in meteorological conditions, with winds in different directions at different altitudes and continuing release of radioactivity over 10 days, which resulted in a complex dispersion pattern (6,7). However, radioactive iodine is more likely to induce thyroid cancer than cesium because the latter is much less attracted or not attracted at all to the thyroid gland.
This paper examines to what extent a relationship between radioactive exposure and thyroid cancer incidence in individuals who were children in 1986 can be studied with available epidemiologic data and draws conclusions about the direction of future research.

Materials and Methods
Basic data included all cancer cases reported to the Belarus Centre for Medical Technology (Minsk, Belarus) and transmitted until 1995 to the State Research Institute of Oncology and Medical Radiology (Lesnoy/Minsk, Belarus), where they were processed in cooperation with the Institute of Social and Preventive Medicine of the University of Bern, Bern, Switzerland. Although the geographical analysis was intended to cover only individuals under 15 years of age in 1986, the first steps in data processing also included those born from 1963 on. This large number of cases was selected to obtain a larger basis for assessing the age distribution on the one hand and reliability of testing for multiple reporting on the other. This was accomplished using a probabilistic linkage method (8) followed by manual control, and revealed 48 of 857 cases (5.6%) to be duplicates. The resulting database contained 809 cases. In 18 of the cases cancer was diagnosed before 1986 and therefore could not be related to the Chernobyl accident. Thus, the final database consisted of 791 cases of thyroid cancer diagnosed between 1 January 1986 and 31 December 1995 in individuals born in 1963 or later.
Census data of the Belarusian population of 1989 were available in DBase 3 format. They included figures for all age groups in steps of 5 years up to age 80 and older for both genders. Separate figures for urban and rural populations as well as sum values allowed quality assessment through crosschecking. There were no inconsistencies.
Time trends and age distribution were analyzed over all cases. Geographic analysis including counting of cases by district and calculation of incidence rates included only patients less than 15 years of age in 1986. The number of individuals under age 15 at the 1989 population count was taken as a midperiod estimate for 1986 to 1995 and multiplied by 10 (10 years of observation time). Because 72% of all districts had only three cases or less, age and age-sex stratification were not done; thus, the rates are crude rates.
The data were processed on an IBM-PC clone computer. Data management and programming of analysis software were done in the FoxPro program V.

Results and Discussion
The oldest cancer patient was 32 at the time of diagnosis. Because the registered data did not reflect the month of birth, 30 June was taken as the effective date for age calculation. Figure 1 shows the age distribution over all individuals with thyroid cancer born after 1962 and diagnosed between 1986 and 1995. As the Belarus population is distributed quite equally over the five age groups up to 24 years (19-22% contribution), an age-adjusted analysis of the age distribution would reveal an almost equal pattern. The small number of cases with 1987 or later as the year of birth suggests that the thyroids of individuals not yet born at the time of the accident will not be affected by cancer related to the Chernobyl accident. In addition, the increase in cases with birth dates closer to the time of the accident suggests an increase in sensitivity of the thyroid to radiation with decreasing age. The number of cases born in 1985 and 1986 is smaller than that of those born in 1983 and 1984. Because the accident occurred at the end of April 1986, two-thirds of the children born in 1986 or one-third of those born in 1985 to 1986 were born after the accident, thus reducing the numbers exposed. If all had been exposed, this would have led to about 130 cases rather than the observed 87 cases shown in Figure 1. This estimate should be verified when data on exact dates of birth become available.
According to the age distribution, there is again an increase in cancer incidence in To avoid an artifact attributable to th normal incidence increases with age ii adults, further analysis was restricted t4 individuals who were less than 15 year old in 1986. Also, the seven children bori in 1987 or later were not included in th following data analysis. It is understoo4 that 10 years after the accident, childrei who were older than age 4 in 1986 wil not be counted if they developed a cance in 1995, as long as the investigation remain limited to cancer in children Therefore, analyses of a cohort of individ uals who were children at the time of th accident give a more realistic picture o the situation.
The data in Table 1 may differ fron the numbers of cases published earlie (9)(10)(11). In addition to the fact that man numbers are related to age at diagnosis, dii ferences in data sources and other method ological issues may also have an impact Table 1 reflects major efforts to eliminat overreporting. The automatic probabilisti linkage and subsequent human contro determine if two similar records are botl counted or considered duplicates; in doubt ful cases the decision remains arbitrary Nevertheless, the number of cases pub lished to date is of about the same magni tude. Table 1 suggests that a steady increas in the number of new cases will occur ove time; the next years will either confirm a refute this suggestion. At present, there ar 1987 or 1988- 1984-1982-1980-1978-1971-1974-1972-1970-1988-1966-1984later 1985 1983 1981 1979 1977 1975 1973 1971 1909 1967 1985 1963 Date of birth  Ls not enough data to extrapolate into the I. future and predict future trends. 1-In addition to the distribution of the e number of childhood thyroid cancer cases f over time, the spatial distribution is of major interest. The numbers of cases n detected between 1986 and 1995 by disr tricts vary considerably. Gomel, including y the city of Gomel, with 95 cases, and f-Minsk, including the city of Minsk, with I-54 cases, have the largest numbers of cases. t. If these numbers are related to population :e figures by district, the calculated incidence ic density in this period reaches as high as 1 I 1.72/10,000 person-years. As the age group h 0 to 4 years, which has the most cases, varies t-only ± 10.6%, and the gender ratio (63% of females) is relatively balanced, it is acceptable to compare crude rates.
i- Figure 2 shows the incidence density by ;e district. Bragin to the east of Chernobyl .r (1.72/10,000 person-years), followed by )r Narovlya (1.68/10,000 person-years), and *e Hoiniki (1.28/10,000 person-years) have the highest incidence density (1986)(1987)(1988)(1989)(1990)(1991)(1992)(1993)(1994)(1995) for children younger than 15 years in 1986. The vicinity of these three districts to Chernobyl suggests a relation between thyroid cancer in children and the Chernobyl accident. This is supported by a comparison with published estimations of 131I contamination as of 10 May 1986 (shown in Figure 3). The similarity of the geographical distributions of 131I and of thyroid cancer incidence densities of persons exposed as children in 1986 is remarkable. In particular, besides the increase in the region of Gomel, an increase both in disease incidence and estimated contamination with 131I west of Chernobyl in an area over 300 km from the accident site is striking. The distribution of the 18 cases was excluded because they occurred before the Chernobyl accident and gave no indication of a particular preaccident pattern that could affect the interpretation of results.  Observations on meteorological developments immediately following the accident as reported by the Lawrence Livermore National Laboratory (12,13) give additional support to a link between radioiodine contamination and the observed increase in thyroid cancer incidence. In addition to identifying 1311 as a possible carcinogenic agent, they may even be interpreted as suggesting a particular role for short-lived radioiodines. The basis of this consideration is that radioiodines that are volatile contributed substantially to the radioactive material ejected during the first phase of the nuclear accident (7). In the first hours after the explosion, the wind transported the radioactive plume in a northwesterly direction (12,13), where short-lived radioiodines may have been deposited in areas affected by rainfall. By 30 April 1986, by which time most of the short-lived radioiodines should have been dispersed, the wind changed to the southeast and the northeast, which led to contamination by the longer-lived 131I in these regions. Rather than relating disease incidence densities to only contamination data, correlations between the geographical distribution of disease frequencies and meteorological indicators for different days could therefore add some useful information about the role of short-lived radioiodines.
As far as field measurements of environmental contamination are concerned, they can be used only when obtained within the first few days after the accident because of the short half-life of radioiodines; therefore, few data from actual measurements are available for that time. As a substitute, determination of concentrations of longlived 1291 in soil specimens and other methods of dose reconstruction are being investigated (14). Because this is a very time-consuming and cost-intensive task, coverage of the whole of Belarus will not be achieved soon. We therefore propose to carefully take the incidence density of children's thyroid cancer into consideration when deciding on locations for soil samples for reconstruction of soil contamination by short-lived radioiodines.
In all these cases, rather than using purely visual methods, comparison of geographical distributions should be done by formal statistical testing. The needed methodology is currently being developed.