Appendix B
Copy of the Memorandum from Dr. Charles Land to Dr. Richard Klausner
DATE: 
September 23, 1997 
TO: 
Dr. Richard Klausner, Director, NCI 
FROM: 
Charles Land, Ph.D., Health Statistician, DCEG/EBP/REB 
THROUGH: 
Director, DCEG 
SUBJECT: 
Calculation of lifetime thyroid cancer risk for an average thyroid dose of 0.02 Gy from I131 in fallout 
My calculations of the thyroid cancer risk that might be associated with exposure to the American public to ^{131}I fallout from the Nevada Test Site resulted in an estimated range of 7,500 to 75,000 excess thyroid cancers during the lifetime of those exposed before 20 years of age. This range of estimates may be compared with about 400,000 expected, according to current SEER rates, among this segment of the US population. Thus, the estimated excess is between 2% and 19% of what might be expected in the absence of exposure. The calculations were based on a published, pooled analysis of thyroid cancer risk data from 5 cohort studies of populations exposed during childhood to medical x ray, or to gamma ray from the atomic bombings of Hiroshima and Nagasaki (Ron et al., 1995). They also incorporate various assumptions about the relative biological effectiveness (RBE) of ^{131}I compared to x ray or gamma ray. Significant excess risk was assumed to occur only following exposure before 20 years of age, in accordance with the epidemiological literature. A linear does response was assumed, and the dosespecific excess relative risk, which was assumed to decrease sharply with increasing age at exposure, was also assumed to remain constant over the lifetime of the exposed population.
The calculations (see attached Excel spreadsheet)
Column 1 identifies the exposure ages considered. The first year of life was treated separately and older ages were grouped: 14, 59, 1014, and 1519. Exposure at ages older than 20 was ignored because there is little or no evidence of an excess cancer risk associated with exposure in adult life even to gamma and xray irradiation. Columns 2 and 3 give the estimated number of persons in the 1952 population of the US, by age and sex, as interpolated from 1950 and 1960 census numbers. The total number exposed at ages 019 also includes persons born in 1953, 1954, etc., but the entry into the population of newborn persons is largely compensated by the loss of persons reaching age 20 in the same years. With a linear doseresponse model and lifetime excess risk, error introduced by acting as if the population 019 years of age in 1952 received all the dose that was actually received by those who were 019 years old during any part of the aboveground testing period is relatively unimportant.
Column 4 gives agespecific average thyroid doses in rad corresponding to the assumed average dose of 2 rad (0.02 Gy), based on information provided by André Bouville (this is why the first year of life was separated from the next four). As you know, thyroid doses to children are larger than those for adults because of smaller gland size, higher milk intake, and higher metabolism.
Column 5 gives the agespecific, linear doseresponse coefficients for x ray and gamma ray, derived from Ron et al. (1995). Their overall coefficient for excess relative risk (ERR) at 1 rad was 0.077. They also did analyses suggesting that the ERR decreases by a factor of 2 for each successive 5year interval of age at exposure, over the range 014 years of age. I derived the values in column 5 from the Ron et al. analysis, and extended the 2fold reduction rule to 1519 years at exposure. In each subsection, the agespecific coefficients have been multiplied by the specified RBE value.
Columns 6 and 7 are the estimated lifetime excess thyroid cancer rates for males and females, computed by multiplying the product of columns 4 and 5 by 0.25% for males and 0.64% for females, respectively; these percentages are the SEER (19731992) report's estimated lifetime thyroid cancer rates for men and women. The 19731994 SEER volume is now out, and gives 0.27% for males and 0.66% for females. Use of the new values would increase the total by about 4%.
Columns 8 and 9 were obtained by multiplying columns 2 and 3 by columns 6 and 7, respectively, and column 10 is the sum of columns 8 and 9. One implication of column 10 is that 75% of all the excess risk is estimated to result from exposure during the first 5 years of life.
The calculations are repeated for RBE values of 1.0, 0.66, 0.3, and 0.1.
Sources of uncertainty
NCRP report No. 80, ''Induction of Thyroid Cancer by Ionizing Radiation," 1985, gave a range of 0.1 to 1.0 for the RBE of thyroid dose from ingested or inhaled I131 compared to gamma ray or x ray, based on experimental studies. The report recommended 0.3 for radiation protection purposes, as the highest credible value. The NCRP report also stated that the RBE of ^{131}I relative to x ray may be lower at high doses and dose rates, and higher (nearer to x ray in effectiveness) at low dose and dose rates. Thus, Walinder (1972, summarized in the NCRP report) obtained an RBE of 0.1 using ^{131}I thyroid doses in the range 220011,000 rad whereas Lee et al. (1982) found near equivalence using dose groups at 80, 330 and 850 rad. Laird (1987) conducted parallel and combined analyses of 6 cohorts of children exposed to external radiation and one exposed to ^{131}I, and reevaluated experimental data from the large study of Lee et al. (1982) specifically designed to investigate the RBE of ^{131}I. Her RBE estimate was 0.66 with 95% confidence limits 0.143.15 (however, there is no support that I know of for an RBE greater than 1). The RBE value at low doses remains a contentious issue.
The range of estimates does not take into account statistical uncertainty about the Ron coefficients or statistical and subjective uncertainty about the estimated average dose. The Ron estimate of ERR_{1Gy} = 7.7 had 95% confidence limits 2.128.7, corresponding to a geometric standard deviation (GSD) of about 1.95. The average dose estimated by NCI, 2 rad, was assigned a GSD of 3, and therefore the product of that dose and the estimated ERR at 1 rad has a GSD of 3.6 (calculated as the exponential of the square root of the sum of squares of the natural logarithms of 1.95 and 3). Approximate 95% confidence limits for the number of excess cases are obtained by dividing and multiplying by 12.4 (= 3.6^{1.96}). Thus, for example, ignoring all other possible sources of error, an estimate of 49,000 lifetime excess cases (corresponding to RBE = 0.66) would have confidence limits 4,000608,000.
According to the model used for the estimates, ERR is constant over time following exposure, and about one third of the total excess lifetime risk among men in the exposed population, and about half among women, should already have taken place. It is possible, however, that the actual excess relative risk per unit dose may decline over time following exposure, most of which occurred over 40 years ago. Ron et al. found significant variation by time following exposure, but did not find a statistically significant trend. At the present time there are few data on radiationrelated thyroid cancer risk 40 or more years following exposure during childhood, and therefore little basis for a discussion of the question.
References
Laird NM. Thyroid cancer risk from exposure to ionizing radiation: a case study in the comparative potency model. Risk Analysis 1987; 7: 299309.
Lee W, Chiaccierini RP, Schlein B, Telles NC. Thyroid tumors following I131 or localized x irradiation to the thyroid and the pituitary glands in rats. Radiation Research 1982; 92: 307319.
NCRP Report No. 80. Induction of thyroid cancer by ionizing radiation. National Council on Radiation Protection and Measurements, Bethesda, 1985.
Ron E, Lubin JH, Shore RE, Mabuchi K, Modan B, Pottern L, Schneider AB, Tucker MA, Boice JD Jr. Thyroid cancer after exposure to external radiation: a pooled analysis of seven studies. Radiat Res 1995; 141: 25977.
Walinder G. Late effects of irradiation on the thyroid gland of CBA mice. I. Irradiation of adult mice. Acta Radiol Ther Phys Biol 1972; 11: 433.
Age at exposure 
1952 population count 
Estimated average thyroid dose (rad) 
ERR at 1 rad (Ron et al. 1995) 
Excess rate (lifetime) 
Lifetime excess thyroid cancers 

Males 
Females 
Males 
Females 
Males 
Females 
Total 





RBE = 1 





0 
1,757,800 
1,698,600 
10.3 
0.098 
0.002524 
0.00646 
4,435.80 
10,973.20 
15,409.00 
14 
7,171,000 
6,931,200 
6.7 
0.098 
0.001642 
0.004202 
11,771.20 
29,126.60 
40,897.80 
59 
7,174,043 
6,929,430 
4.5 
0.049 
0.000551 
0.001411 
3,954.70 
9,778.80 
13,733.50 
1014 
6,235,357 
6,023,970 
2.8 
0.0245 
0.000172 
0.000439 
1,069.50 
2,644.80 
3,714.30 
1519 
5,916,664 
5,715,114 
1.8 
0.01225 
5.51e05 
0.000141 
326.2 
806.5 
1,132.70 
Total 
28,255,864 
27,298,314 




21,557 
53,330 
74,887 




RBE = 0.66 





0 
1,757,800 
1,698,600 
10.3 
0.06468 
0.001666 
0.004264 
2,927.60 
7,242.30 
10,170.00 
14 
7,171,000 
6,931,200 
6.7 
0.06468 
0.001083 
0.002773 
7,769.00 
19,223.50 
26,992.50 
59 
7,174,043 
6,929,430 
4.5 
0.03234 
0.000364 
0.000931 
2,610.10 
6,454.00 
9,064.10 
1014 
6,236,357 
6,023,970 
2.8 
0.01617 
0.000113 
0.00029 
705.9 
1,745.50 
2,451.40 
1519 
5,916,664 
5,715,114 
1.8 
0.008085 
3.64e05 
9.31e05 
215.3 
532.3 
747.6 
Total 
28,255,864 
27,298,314 




14,228 
35,198 
49,426 




RBE = 0.3 





0 
1,757,800 
1,698,600 
10.3 
0.0294 
0.000757 
0.001938 
1,330.70 
3,292.00 
4,622.70 
14 
7,171,000 
6,931,200 
6.7 
0.0294 
0.000492 
0.001261 
3,531.40 
8,738.00 
12,269.30 
59 
7,174,043 
6,929,430 
4.5 
0.0147 
0.000165 
0.000423 
1,186.40 
2,933.6 
4,120.10 
1014 
6,236,357 
6,023,970 
2.8 
0.00735 
5.15e05 
0.000132 
320.9 
793.4 
1,114.30 
1519 
5,915,664 
5,715,114 
1.8 
0.003675 
1.65e05 
4.23e05 
97.8 
242 
339.8 
Total 
28,255,864 
27,298,314 




6,467 
15,999 
22,466 




RBE = 0.1 





0 
1,757,800 
1,698,600 
10.3 
0.0098 
0.000252 
0.000646 
443.6 
1,097.30 
1,540.90 
14 
7,171,000 
6,931,200 
6.7 
0.0098 
0.000164 
0.00042 
1,177.10 
2,912.70 
4,089.80 
59 
7,174,043 
6,929,430 
4.5 
0.0049 
5.51e05 
0.000141 
395.5 
977.9 
1,373.40 
1014 
6,236,357 
6,023,970 
2.8 
0.00245 
1.72e05 
4.39e05 
107 
264.5 
371.4 
1519 
5,916,664 
5,715,114 
1.8 
0.001225 
5.51e06 
1.41e05 
32.6 
80.7 
113.3 
Total 
28,255,864 
27,298,314 




2,156 
5,333 
7,489 
Written amendment to 23 September memo, provided by Charles Land to NAS committee on 19 December, 1997
Calculation of the Estimated Lifetime Risk of RadiationRelated Thyroid Cancer in the U.S. Population from NTS Fallout

Thyroid cancer risk associated with gammaray and xray exposure, from studies of the HiroshimaNagasaki survivors and of various medicallyexposed populations, is well quantified. Findings are summarized in a pooled analysis of seven studies (Ron et al., Radiation Research 1995; 141:259277).

The evidence for a radiationrelated risk is strong for childhood exposure, and weak or nonexistent for adult exposure.

Dosespecific excess risk decreases with increasing age at exposure. At ages 59, it is about half that associated with exposure at ages 04, and at 1014 it is about half that at 59.

For any given exposure age, excess risk appears to be proportional to thyroid dose (linear dose response).

Ron et al. estimated an excess relative risk (ERR) of 7.7 per Gy, or 0.077 per rad, for childhood exposure at ages younger than 15.


The average (caseweighted) exposure age in the pooled data was a little over 4 1/2 years. By linear interpolation between the midpoints of the first two intervals, and extension of the observed reduction in ERR with increasing age at exposure, the following agespecific coefficients were inferred:
Age at exposure 
ERR at 1 rad 
04 
0.098 
59 
0.049 
1014 
0.0245 
>20 
negligible 

Although there was evidence of variation radiationrelated relative risk over time following exposure, there was no evidence of a trend. Accordingly, ERR was assumed to remain constant over the remainder of life.

Data on risk associated with thyroid exposure from ingested or inhaled ^{131}I suggest that there is a risk, but precise doseresponse estimates are not available. Accordingly, it is reasonable to use the coefficients developed from data on xray and gammaray exposure, with an appropriate value for the relative biological effectiveness of ^{131}I compared to gamma rays or x rays.

NCRP report No. 80, "Induction of Thyroid Cancer by Ionizing Radiation," 1985, gave a range of 0.1 to 1.0 for the RBE, based on experimental studies. The report recommended 0.3 for radiation protection purposes, as the highest credible value. The NCRP report also stated that the RBE of


^{131}I relative to x ray may be lower at high doses and dose rates, and higher (nearer to x ray in effectiveness) at low doses and dose rates.

Thus, Walinder (1972, summarized in the NCRP report) obtained an RBE of 0.1 using ^{131}I thyroid doses in the range 220011,000 rad, whereas

Lee et al. (1982) found near equivalence using dose groups at 80, 330, and 850 rad.

Laird (1987) conducted parallel and combined analyses of 6 cohorts of children exposed to external radiation and one exposed to ^{131}I, and reevaluated experimental data from the large study by Lee et al. (1982) specifically designed to investigate the RBE of ^{131}I. Her RBE estimate was 0.66 with 95% confidence limits 0.143.15 (however, there is no support that I know of for an RBE greater than 1).

The RBE value at low doses remains a contentious issue.

In the calculations for NCI, RBE values of 1, 0.66, 0.33, and 0.1 were assumed.

In addition to being more sensitive to the carcinogenic effects of ionizing radiation, the thyroid glands of children receive higher doses from ingested or inhaled ^{131}I than do the glands of adults, because of smaller gland size, higher intake of milk, and higher metabolism. Using conversion factors obtained from Dr. Bouville, the estimated average thyroid dose of 2 rad to the U.S. population from Nevada Test Site fallout was converted to the following values for children:
Exposure Age 
Estimated Average Dose 
<1 
10.3 
14 
6.7 
59 
4.5 
1014 
2.8 
1519 
1.8 

Lifetime cumulative thyroid cancer incidence rates of 0.25% for males and 0.64% for females, respectively, were assumed, based on the SEER report for 19731992. The 19731994 SEER volume is now out, and gives 0.27% for males and 0.66% for females. Use of the new values would increase the total by about 4%.

For simplicity of calculation, it was assumed that the U.S. population in 1952 received the total thyroid dose from NTS fallout in that year, instead of spread out over 12 years. This simplification was possible because, using a linear doseresponse model, lifetime radiationrelated thyroid cancer risk is proportional to summed collective dose, in personrads, over exposure ages weighted by agespecific risk coefficient.

For each single year of age (column 1 in the spreadsheet), the sexspecific estimated numbers of lifetime excess thyroid cancer cases in the US due to NTS fallout (columns 8 and 9) were obtained as the product of:

the number of male or female persons in the 1952 US population (columns 2 and 3)

the agespecific estimated average cumulative thyroid dose over the entire period of aboveground testing (column 4)

the agespecific linear doseresponse coefficient (ERR at 1 rad) for x ray and gamma ray (column 5), times the assumed RBE for ^{131}I

the cumulative lifetime thyroid cancer risk for men or women (0.25% or 0.64%), as appropriate.

The age and sexspecific totals were summed over sexes (column 10) and ages. The sums are given below columns 810 in each table.

Besides uncertainty about the RBE, there is also statistical uncertainty about the risk coefficients, and subjective and statistical uncertainty about the average doses used. The combined uncertainty is substantial. For example:

95% confidence limits (2.128.7) for the Ron estimate of ERR_{1Gy} = 7.7 correspond approximately to a lognormal model geometric standard deviation (GSD) of about 1.95.

The uncertainty of average dose estimated by the NCI, 2 rad, was stated to be between 1 and 4, i.e., a factor of 2 in each direction. This corresponds approximately to 95% confidence limits and thus to a GSD of about 1.4.

Therefore, the product of that dose and the estimated ERR at 1 rad has a GSD of 2.1 (calculated as the exponential of the square root of the sum of squares of the natural logarithms of 1.95 and 1.4).

Approximate 95% confidence limits for the number of excess cases can be obtained by dividing and multiplying by 4.3 (=2.1^{1.96}). Thus, for example, ignoring all other possible sources of error, an estimate of 49,000 lifetime excess cases (corresponding to a fixed RBE of 0.66, which here is assumed to be known without error) might be given with uncertainty 11,300212,000.
