Next | ToC | Prev

### CHAPTER 40 Conversion Factors:   The Basis of Column "V"

```
```
Part 1.   Introduction

By the term "conversion-factor" in this book, we mean any formula which estimates how many radiation-induced breast-cancers result from a specified amount of population-exposure --- for example, from 10,000 person-rads of exposure. In this book, the term person-rads refers to breast-rads.

The conversion-factors developed in this chapter differ slightly from the conversion-factors published in our preliminary work for this book. We continue to study the problem of conversion-factors;   many other analysts continue to study the problem, too;   and relevant new evidence continues to accumulate. So we expect the conversion-factors to keep improving in the future. Meanwhile, this chapter presents one of several ways to arrive at conversion-factors, in view of what is presently known, and what is presently unknowable.

Part 2 of this chapter begins where Chapter 15 of our 1990 book ended. Otherwise, we would have to repeat most of the earlier book. Readers who are not familiar with the previous book can, nonetheless, follow what we do here with the "starting point" from the earlier work (Gofman 1990).

```
```
Part 2.   Step-by-Step Development of Our Conversion Factors

Our starting point for developing conversion-factors is the study of atomic-bomb survivors. Why don't we use evidence based in the United States, since this book focuses on the breast-cancer problem in the USA? The answer is simple:   Only the A-Bomb Study provides observations for women irradiated at every age, and then followed-up for decades. No comparable study exists for American women.

We have analyzed both the 1950-1982 and the 1950-1985 follow-up evidence from Japan on radiation-induced cancer mortality. The more recent data are not yet available to us from RERF (as of January 1995). Here, we will use the 1950-1982 data, because readers can refer to Gofman 1990 to check exactly how we handled those data in reaching our starting point for this chapter.

We begin by presenting a series of six boxes, which are subsequently explained in the text. [For accessibility, we also include each box at the beginning of its own section here on the web version --ratitor]

BOX 1, "Starting Point." All cancer-types, excluding leukemia.
```    A            B                   C                 D
Age when    Cancer-deaths      Cancer-deaths     Difference
exposed     per 10,000         per 10,000        per 10,000
(years)     initial females    initial females   initial females
as of 1982, in     as of 1982, in    as of 1982
high-dose group    low-dose group    (B minus C)

0-9           150.94             55.41             95.53
10-19          295.23            153.32            141.91
20-34          637.48            451.57            185.91
35-49         1318.68           1077.37            241.31
50+          1246.33           1040.42            205.91
```

BOX 2, Lifetime Factors. All cancer-types, excluding leukemia.
```    A                     D               E               F
exposed   exposed    per 10,000       factor of     cancer-deaths
(avg.)     initial females  increase      per 10,000
as of 1982                     initial females
(years)   (years)    (B minus C)     (1059/Col C)   (D times E)

0-9        4.1        95.53         19.11207        1825.776
10-19      14.9       141.91          6.90712         980.190
20-34      26.6       185.91          2.34515         435.987
35-49      41.7       241.31          none            241.31
50+       58.9       205.91          none            205.91
```

BOX 3, Supra-Linearity Adjustment. All cancer-types, excluding leukemia.
```    A              F                  G                H
exposed     cancer-deaths      linearity       cancer-deaths
(years)     per 10,000            factor       (per 10K females)
(D times E)       (see text)       supra-linearity
(F times G)

0-9          1825.776            2.40           4381.863
10-19          980.190            2.24           2195.625
20-34          435.987            2.18            950.452
35-49          241.31             2.13            513.990
50+           205.91             2.06            424.175
```

BOX 4, Division by Dose-Difference. All cancer-types, excluding leukemia.
```   A           H                I         J          K            L
exposed   cancer-deaths       group:    group:  difference   cancer-deaths
(years)   (per 10,000        Average   Average     in rads   per 10K females
females)           dose in   dose in               with avg dose
supra-linearity                                    (H / K)

0-9       4381.863          97.61      2.91      94.70        46.271
10-19      2195.625          89.20      2.50      86.70        25.324
20-34       950.452          87.38      2.63      84.75        11.215
35-49       513.990          80.87      2.76      78.11         6.580
50+        424.175          71.73      2.51      69.22         6.128
```

```    A               L                  M                N
exposed      cancer-deaths      increase for     cancer-deaths
(years)      per 10K females       medical       per 10K females
with avg dose          rads         with avg dose
(H / K)                             (L times M)

0-9             46.271             2               92.542
10-19            25.324             2               50.649
20-34            11.215             2               22.430
35-49             6.580             2               13.161
50+              6.128             2               12.256
```

BOX 6, Final Adjustments to Reach Conversion-Factors.
```   A            N                "O"              P           Q          R          S
exposed     cancer-deaths     breast-cancer    mortality   ratio of   reduction     factors
(years)   per 10K females    deaths per 10K    ratio for    breast-      factor  for Master
with avg dose  fem with avg dose     breast-     cancer                  Table.
= 1 medical rad   = 1 medical rad       cancer    vulner-  (see text)   Described
(L times M)        (see text)   (see text)    ability                below *.

0-9         92.542           8.8315           3.704        5.67       0.50      92.74
10-19        50.649           4.8335           3.704        5.67       0.75      76.13
20-34      - 22.430           2.1405           3.704        5.67       1         44.95
35-49        13.161           1.2560           3.704        5.67       1         26.38
50+         12.256           1.1696           3.704        5.67       1         24.56
```
* Each conversion factor is the estimated number of radiation-induced excess breast-cancer cases (lifetime incidence) per 10,000 females (USA) from an average breast-dose of one medical rad. Each entry in Column S is the product of Columns "O" x P x Q x R.

BOX 1, "Starting Point." All cancer-types, excluding leukemia.
```    A            B                   C                 D
Age when    Cancer-deaths      Cancer-deaths     Difference
exposed     per 10,000         per 10,000        per 10,000
(years)     initial females    initial females   initial females
as of 1982, in     as of 1982, in    as of 1982
high-dose group    low-dose group    (B minus C)

0-9           150.94             55.41             95.53
10-19          295.23            153.32            141.91
20-34          637.48            451.57            185.91
35-49         1318.68           1077.37            241.31
50+          1246.33           1040.42            205.91
```

- BOX 1, Starting Point

In Box 1 are the findings for cancer mortality for all types of cancer (leukemia excluded) in the female population, divided into low-dose and high-dose groups for five separate age-groups (age in 1945, at the time of the bombings). The observed differences between high-dose and low-dose groups (Column D, Box 1) are the radiation-induced cancer-deaths.

The boxed data come from Tables 15-G, 15-H, 15-I, 15-J, and 15-K of Gofman 1990. We are using the entries from those tables in the T65DR dosimetry. Those very same tables show that the results are negligibly different when the revised DS86 dosimetry is used properly. By "properly," we mean that the cohorts remain constant (no shuffling of cases) while receiving a revised estimate of average dose from the DS86 system.

Our protest against the retroactive shuffling of cohorts, and against the constant retroactive alteration of the DS86 database, is fully presented in Gofman 1990 (especially Chapters 5 and 6). In our opinion, use of the unstable DS86 database (which already has many versions) is scientifically acceptable only as part of a "constant-cohort, dual-dosimetry" analysis in which the T65DR system remains the stable anchor for the follow-up studies.

BOX 2, Lifetime Factors. All cancer-types, excluding leukemia.
```    A                     D               E               F
exposed   exposed    per 10,000       factor of     cancer-deaths
(avg.)     initial females  increase      per 10,000
as of 1982                     initial females
(years)   (years)    (B minus C)     (1059/Col C)   (D times E)

0-9        4.1        95.53         19.11207        1825.776
10-19      14.9       141.91          6.90712         980.190
20-34      26.6       185.91          2.34515         435.987
35-49      41.7       241.31          none            241.31
50+       58.9       205.91          none            205.91
```

The values in Box 1 cover only the period through 1982. We (and others) have to estimate what the observations may be when all the initial participants have been followed for their full lifespans. In 1945, the average age of the 50+ age-group was 58.9 years for females, and average age was 41.7 years for the age-group 35-49 years (Gofman 1990, Table 26-G). In 1982, both of these age-groups have essentially told their lifetime "story," so we can use those results without adjustment. But we must make estimates for the other three age-groups.

How? Tables 15-G through 15-K of Gofman 1990 show that the whole-body dose in the low-dose groups was very similar in all five age-bands (range = 2.50 rems to 2.91 rems). So we will use the approximation that the low-dose women in the three youngest age-groups will "behave" like the two oldest age-groups, with respect to lifetime rate of cancer mortality. When we average the lifetime rates of the two oldest age-groups, we obtain 1,059 cancer deaths per 10,000 initial women.

For the age-group 0-9 years, the value 1,059 represents a 19.112-fold multiplication of the 55.41 rate observed in 1982 (Box 1, Column C). For the age-group 10-19 years, 1,059 represents a 6.907-fold multiplication of the 153.32 rate observed in 1982. And for the age-group 20-34 years, 1,059 represents a 2.345-fold multiplication of the 451.57 rate observed in 1982. These "lifetime factors" are shown in Box 2, Column E. (As usual, we retain all the digits without endowing them with the formal status of "significant figures.")

And what about the lifetime difference between the high-dose and low-dose groups --- the lifetime radiation-induced excess for the three youngest age-groups? For now, we temporarily use the assumption that the difference will also increase by the same "lifetime factors." Later, in Box 6, we make downward adjustments for the two youngest age-groups. Meanwhile, Box 2 applies the "lifetime factors" to obtain the estimated lifetime excess of cancer-deaths (all sites) per 10,000 initial females (Column F).

BOX 3, Supra-Linearity Adjustment. All cancer-types, excluding leukemia.
```    A              F                  G                H
exposed     cancer-deaths      linearity       cancer-deaths
(years)     per 10,000            factor       (per 10K females)
(D times E)       (see text)       supra-linearity
(F times G)

0-9          1825.776            2.40           4381.863
10-19          980.190            2.24           2195.625
20-34          435.987            2.18            950.452
35-49          241.31             2.13            513.990
50+           205.91             2.06            424.175
```

- BOX 3, Adjustment for Supra-Linear Dose-Response

The differences in Box 1 (Column D) come from assuming a linear dose-response. In reality, supra-linearity describes the dose-response among the A-bomb survivors significantly better than linearity, both in T65DR and DS86 constant-cohort analyses (Gofman 1990, Chapters 14 and 29). The shape of the dose-response for 1950-1982, with all ages and both sexes combined, is reproduced here from Figure 14-E of Gofman 1990. The curve has the same shape in the DS86 dosimetry, properly handled with "constant-cohorts."

We notice with interest that the excess relative risk specifically for breast-cancer incidence also shows a supra-linear shape in Thompson 1994 (Figure 3, p.S26, which is a DS86 analysis not anchored to T65DR cohorts).

If it were not for the bend of supra-linearity, the differences between high-dose groups and low-dose groups would have been approximately twice as high as the values presented in Box 1, Column D. Indeed, we evaluated how much higher for each female age-group separately in Gofman 1990 (Chapter 15, Part 6, tabulation of factors). Now we must use those supra-linearity factors to adjust the differences in Column F. This is done in Box 3.

The supra-linearity adjustment explains why we refer to our conversion-factors as "low-dose conversion-factors" (Chapter 39, Part 1). The supra-linearity adjustment bases the difference between high-dose and low-dose groups on the steeper slope which characterizes dose-response in the low-dose region (approximately zero to five rads). Readers have seen (initially in Chapter 8) how we maintain the applicability of these low-dose conversion-factors to exposures which occurred at higher doses, too. We do it by adjusting doses downward, whenever appropriate, by factors derived from the supra-linear curve itself.

BOX 4, Division by Dose-Difference. All cancer-types, excluding leukemia.
```   A           H                I         J          K            L
exposed   cancer-deaths       group:    group:  difference   cancer-deaths
(years)   (per 10,000        Average   Average     in rads   per 10K females
females)           dose in   dose in               with avg dose
supra-linearity                                    (H / K)

0-9       4381.863          97.61      2.91      94.70        46.271
10-19      2195.625          89.20      2.50      86.70        25.324
20-34       950.452          87.38      2.63      84.75        11.215
35-49       513.990          80.87      2.76      78.11         6.580
50+        424.175          71.73      2.51      69.22         6.128
```

- BOX 4, Division by Dose-Difference

So far, we have not considered the difference in dose which produced the difference in cancer-rates. In order to obtain the lifetime excess per rad, we use Box 4 to divide the values in Column H by the dose-differences, for which the data are provided by age-group in Gofman 1990, Tables 15-G through 15-K. This provides lifetime excess cancer-deaths per 10,000 females, from an average dose of one A-bomb rad.

```    A               L                  M                N
exposed      cancer-deaths      increase for     cancer-deaths
(years)      per 10K females       medical       per 10K females
with avg dose          rads         with avg dose
(H / K)                             (L times M)

0-9             46.271             2               92.542
10-19            25.324             2               50.649
20-34            11.215             2               22.430
35-49             6.580             2               13.161
50+              6.128             2               12.256
```

It is widely agreed that the carcinogenic potency of high-energy gamma-rays is less than the potency of lower-energy x-rays, as mentioned in Chapter 3, Part 1. The "ballpark" estimate is that medical x-rays are about two-fold more carcinogenic, per rad, than high-energy gamma-rays (references in Gofman 1990, Chapter 13, Part 4). Thus, if the Hiroshima-Nagasaki exposure had been to x-rays instead of A-bomb gamma-rays, the number of excess cancers per 10,000 women, from an average dose of one medical rad, would have been about twice as great. Box 5 converts Column L from A-bomb rads to medical rads.

BOX 6, Final Adjustments to Reach Conversion-Factors.
```   A            N                "O"              P           Q          R          S
exposed     cancer-deaths     breast-cancer    mortality   ratio of   reduction     factors
(years)   per 10K females    deaths per 10K    ratio for    breast-      factor  for Master
with avg dose  fem with avg dose     breast-     cancer                  Table.
= 1 medical rad   = 1 medical rad       cancer    vulner-  (see text)   Described
(L times M)        (see text)   (see text)    ability                below *.

0-9         92.542           8.8315           3.704        5.67       0.50      92.74
10-19        50.649           4.8335           3.704        5.67       0.75      76.13
20-34      - 22.430           2.1405           3.704        5.67       1         44.95
35-49        13.161           1.2560           3.704        5.67       1         26.38
50+         12.256           1.1696           3.704        5.67       1         24.56
```
* Each conversion factor is the estimated number of radiation-induced excess breast-cancer cases (lifetime incidence) per 10,000 females (USA) from an average breast-dose of one medical rad. Each entry in Column S is the product of Columns "O" x P x Q x R.

Part 3.   Explanation of BOX 6: Transition to Breast-Cancer Incidence, USA

Every step to reach the entries in Column N has an anchor in various real-world observations. And what we have, so far, are the numbers of radiation-induced cancer-deaths (all sites) per 10,000 females, from an average dose of one medical rad, for five different age-groups. What we want are the estimated numbers of radiation-induced breast-cancer cases per 10,000 females, per medical rad, for each age-group. So, we continue to look for anchors in real-world observations to make the next transitions.

- Transition to Breast-Cancer - Column "O"

We know the relationship of the all-cancer death-rate to the breast-cancer death-rate in Japanese females for 1964-1965 (American Cancer Society 1970, pp.22-23). These rates for the entire country are barely influenced by the atomic bombings. The years 1964-1965 are chosen by us because they are approximately the midpoint of the 1950-1982 follow-up study from which we obtain our knowledge of radiation-induced rates. The all-cancer deaths per 100,000 Japanese females were 94.7 deaths per year. The breast-cancer deaths per 100,000 Japanese women were 3.8 deaths per year. The fraction of cancer-deaths which came from breast-cancer was 3.8 / 94.7, or 0.040127 --- four percent. If all cancers were equally radiation-inducible, then we could assume that approximately four percent of the radiation-induced excess cancer-deaths in Column N would be from breast-cancer. But it is not so simple.

We also have the observation that breast-cancer is more inducible by radiation than other cancers. This observation comes from the 1994 Thompson Study of cancer incidence in the A-bomb survivors, 1958-1987. Although the Thompson Study is a DS86 analysis which is not anchored to the T65DR cohorts, it is unlikely that this shortcoming distorts the ratio of relevance here:   The ratio of observed Excess Relative Risk (per sievert) for breast-cancer, over the observed Excess Relative Risk (per sievert) for cancers at all sites --- including breast-cancer. That ratio is 1.59 / 0.63, or 2.524 (from Thomspon 1994, pages S26, S49, S61). Breast-cancer has by far the highest Excess Relative Risk of any cancer reported at page S26.

In the ratio above, the denominator of 0.63 for all sites includes breast-cancer. The all-site value would be lower than 0.63 if it were not elevated by breast-cancer. This means that the radiation-inducibility of breast-cancer must be somewhat more than 2.524-fold higher than the radiation-inducibility of non-breast-cancers. However, the available data do not allow us to adjust the ratio for females alone.

(a) So, we use the 2.524-factor without adjustment, as a reasonable "first approximation" for the relative inducibility of breast-cancer versus non-breast-cancer by ionizing radiation.

(b) Also, we use the vital statistics from above to estimate the ratio of cancer mortality from breast-cancer to cancer mortality from non-breast-sites. The breast-rate (per 100,000 women per year) is 3.8 when all-site-rate is 94.7, so the non-breast-rate is (94.7 minus 3.8), or 90.9. So the ratio of breast-cancer mortality to non-breast-cancer mortality is (3.8 / 90.9), or 0.041804.

Then, for each radiation-induced entry in Column N, what is the estimated number from breast-cancer?

The first entry, 92.542 excess cancer-deaths = (the radiation-induced number from non-breast sites) + (the radiation-induced number from the breast).

Let n = the number of non-breast cancer-deaths produced per 10,000 women from an average dose of one medical rad.

The number of breast-cancer deaths produced per 10,000 per rad will be (n) x (0.041804, which is the breast to non-breast ratio) x (2.524, which is the radiation-inducibility factor). The product = 0.1055n = breast-cancer cases. This term for breast-cancer cases will be used for all five age-groups. Hence, for the age-years 0-9, we treat the all-site value of 92.542 cases (from Column N) as follows:

92.542 radiation-induced cancers = non-breast cases + breast cases = (n) + (0.1055n) = 1.1055n. Therefore:

n = 83.71 = non-breast cases. And:
Breast cases = 0.1055n = 8.83 radiation-induced cases.

Using the same procedure, we obtain values in Column "O" for all five age-groups.

- Transition to Incidence - Column P

How shall we convert the estimates for radiation-induced breast-cancer mortality to incidence? Again, we have real-world observations from the study of A-bomb survivors. For the 1958-1987 period, the reported ratio of breast-cancer mortality to incidence is 0.27 (Mabuchi 1994, Table 4, page S10). This is, of course, an incidence to mortality ratio of (1 / 0.27), or 3.704. We do not think this ratio is likely to be distorted by a DS86 analysis, even though the Mabuchi analysis is not anchored to the constant T65DR cohorts. (Our comments about T65DR and DS86 occur with Box 1, above.) Mabuchi et al also report that the mortality/incidence relationship for solid tumors is observed to be very nearly the same for exposed and non-exposed participants in the study (Mabuchi 1994, Table 5, p.S10).

We will use the assumption that if radiation-induced cancer-mortality (at any site) goes up by 5 cases (for example), the radiation-induced incidence at the same site goes up by (5 cases) times (the ratio of incidence over mortality). Thus, to convert Column "O" to breast-cancer incidence, we multiply its entries by 3.704 (displayed in Column P).

- Transition to the USA - Column Q

For reasons which are not yet understood --- and which badly need to be understood --- the rates of specific cancers often vary enormously from country to country. In Japan, the so-called spontaneous rate of breast-cancer is very much lower than in the United States. We will evaluate the reported US/Japan rates for 1964-1965, approximately the midpoint of the A-Bomb Study follow-up. For the USA, breast-cancer deaths per 100,000 females per year were 21.55;   for Japan, they were 3.80 (from American Cancer Society 1970, p.23). The vulnerability to breast-cancer for women in the United States was 5.67-fold greater than for women in Japan. In view of this difference, we are making the assumption, tentatively, that the number of radiation-induced cases per 10,000 females, from an average dose of one medical rad, is about 5.67-fold higher in the USA than among A-bomb survivors. So, 5.67 is entered into Column Q as the US/Japan ratio.

Up until this point, our conversion-factors are based on the number of excess cases per rad derived from the Japanese experience. Use of the ratio of 5.67 in Column Q acknowledges the real-world difference in "background" rates of breast-cancer in the USA versus Japan. Use of this factor with our method is equivalent to saying that the percent increase in the background rate, per rad of radiation exposure --- the Excess Relative Risk --- is approximately the same from one population to another, regardless of differences in their background rates. If the percent increase per rad is the same in the USA and Japan, then one rad of breast-exposure induces more cases of breast-cancer in the USA than in Japan.

The concept that ionizing radiation induces different numbers of cancers per rad, in one country versus another, is consistent with the concepts that --- within a single country and a single study, such as the A-Bomb Study --- the lifetime number of cancers induced per rad differs by site (for example, breast versus bone) and differs by age at the time of exposure. In other words:   The number of radiation-induced cases per rad is not fixed by the number of genetic lesions induced. The number of radiation-induced cancers per rad differs with the circumstances.

Circumstances can differ within a population, and also from one country to another. If ionizing radiation acts in concert with some other factors (internal or external), present in the USA but less present in Japan, the interaction could well cause a six-fold disparity in the vulnerability for breast-cancer per medical rad. We do not suggest that a rad in the USA induces six times as many genetic lesions as it causes in Japan. We (and others) suggest only that the impact of carcinogenic lesions is affected by co-factors --- and vice versa. Whatever combined factors cause the breast-cancer mortality rate in the mid-1960s to be 5.67 times higher in the USA than in Japan, these combined factors also cause radiation-induced genetic lesions from medical irradiation to have a greater effect in the USA than in Japan.

At this time, we think the 5.67 ratio of background breast-cancer mortality-rates (USA / Japan, in the mid-1960s) is the most reasonable approximation to use in our conversion-factors. As additional evidence develops, it may point to other ways to handle country-to-country "transport" of observations. Meanwhile, we keep an open mind toward a variety of old and new hypotheses about the actual process of radiation carcinogenesis, including interaction between radiation and other factors.

A Closer Look at Recent Cancer-Rates, USA vs. Japan

It is interesting to compare cancer rates between the USA and Japan. By no means is the far higher U.S. rate for breast-cancer a general phenomenon, as shown by the female and male tabulations below. For stomach cancer, the Japanese rate is far higher than that for the United States. And overall, the Japan cancer rates are not very different from those in the United States (see "All" in the tabulations).

All data are death rates, 1988-1991 per 100,000 population, age adjusted to the WHO world standard population. (From ACS 1994, at page 28.)

```  Cancer    Japan     U.S.      Ratio    Cancer  Japan   U.S.     Ratio
Site  Females  Females  U.S/Japan      Site  Males  Males  U.S/Japan

Breast     6.3     22.4        3.6   Prostate    3.8   16.8     4.4

Lung     8.0     24.7        3.1       Lung   30.1   57.1     1.9

Oral     0.6      1.3        2.2       Oral    2.3    3.7     1.6

All    76.7    110.6        1.4   Leukemia    4.3    6.3     1.5

Uterine                                Colon &

Cervix     1.8      2.6        1.4     Rectum   15.1   16.7     1.1

Leukemia     2.8      3.8        1.4        All  150.0  164.0     1.1

Colon &                                Stomach   34.9    5.2     0.1

Rectum     9.7     11.4        1.2

Uterine

Other     2.4      2.6        1.1

Stomach    15.5      2.3        0.1
```

Comparison of the Extremes

```Cancer Sites       Females        Cancer Sites          Males
Compared                          Compared

Breast USA/Japan      3.56        Prostate USA/Japan     4.42
Stomach USA/Japan     0.15        Stomach USA/Japan      0.15
Ratio: A Factor of   23.73        Ratio: A Factor of    29.47
Quite A Range of Relative Rates   Quite A Range of Relative Rates
```

We do not think any single carcinogen, acting alone, explains these vastly different relative rates.

A Suggestion That Will Not Compute

In the effort to deny the idea of a much greater U.S. vulnerability to breast-cancer induction per medical rad (Column Q), it has been suggested that the observed relative rate in the mid-1960s of 5.67 is explained by a possible difference in x-ray dose. If there was a dose-difference (with higher annual average dose in the USA), then it would help explain the relative rate in the mid-1960s --- and it would have an additional implication worth exploring. We will explore the suggestion, below, that a dose-difference fully explains the 5.67 relative rate for breast-cancer mortality.

The proposal is thus entertained, that female persons in the United States had been getting much more x-radiation than female persons in Japan and that this accounts for the relative rate of 5.67 in the mid-1960s. The relative rate of 5.67 means, of course, an excess relative rate of (5.67 - 1.00), or 4.67. If we were to blame the USA-Japan difference in rates totally upon the difference in radiation dosage up to that time, we would be saying that about 4.67 out of 5.67 breast-cancer deaths in the United States at that time must have been x-radiation-induced --- that is 82% of all breast-cancers were radiation-induced. So, the suggestion leads to the absurdity of trying to disprove the hypothesis of radiation-causation with a proposition which would confirm it.

A Suggestion of an Opposite Nature:   Denial of Co-Action

We must consider another possible explanation for the 5.67 relative rate of breast-cancer mortality. We can suppose that non-radiation causes of breast-cancer exist in the USA which do not exist in Japan, and that they are not co-active with each other or with radiation. They cause breast-cancer without involving radiation-induced lesions at all. When radiation is added to such a "scenario," we would suppose that the effect per rad would be the same in the USA and Japan.

For simplification, we can assume that the amount of radiation exposure was the same in the USA and Japan before the mid-1960s. So, the relative rate of radiation-induced breast-cancer mortality would be 1.0. Then the excess relative rate, from all other causes combined, would be 4.67 in the mid-1960s. This would mean that (4.67 / 5.67), or 82.4 % of all breast-cancer mortality in the USA was due to non-radiation causes (internal and external) --- each acting alone, and present in the USA but less present in Japan. This is a possibility, neither provable nor disprovable by current knowledge.

We think (but can not prove) that it would be a mistake to deny the impact of co-action between cancer-causing agents. We are very comfortable with the concept that a rad of medical radiation has a greater impact in the USA than in Japan, due to interaction with other factors.

- "Lift-Off" Reduction-Factor - Column R

For the two youngest age-groups in the A-Bomb Study, the lifespan follow-up is far from complete. In 1995, the 0-9 year-olds reach an average age of about 54 years;   the 10-19 year-olds reach about 64.5 years of age, on the average (calculated from Gofman 1990, Table 26-E). We and others necessarily have to make assumptions about what the full lifetime follow-up will show, as already noted with respect to Box 2.

We know that the observed number of cancers in the low-dose column (Box 1, Column C) will increase after 1982, as the two youngest age-groups advance in age. We used the observations of their older colleagues to estimate what would happen to the number of cases in the youngest low-dose groups. And then we applied the same factor of increase (Box 2, Column E) to the radiation-induced difference between high-dose and low-dose groups (Column D). This was equivalent to assuming that the total number of cases in the high-dose group (Box 1, Column B) and the low-dose group (Box 1, Column C) would increase by the same factor. For example:   (150.94 x 19.11207) - (55.41 x 19.11207) = (2884.7758) - (1059) = 1825.776, which is the first value shown in Box 2, Column F, for "Lifetime excess."

However, we have shown in Chapter 3, Part 4, why this treatment may overestimate the lifetime difference for the youngest age-group, and possibly for the 10-19 year-olds too.

Chapter 3 explored the "lift-off" phenomenon! By this, we mean the temporary inflation of a disparity between two curves, if they "lift off" at different years or at different slopes from a baseline rate which was near zero. The inflated difference is temporary, and becomes diluted by subsequent observations at older ages.

Because there are more "subsequent observations" to come in the 0-9 age-group than in the 10-19 age-group, we adjust the lifetime difference in Column N more severely downward for the 0-9 age-group than for the 10-19 age-group. In Column R, we are going to slash the lifetime excess in half for the 0-9 year-olds, with a reduction-factor of 0.5;   for the 10-19 year-olds, we use a reduction-factor of 0.75;   we make no adjustment for the 20-34 year-olds. Only additional follow-up data will tell whether these "lift-off" factors are too high or too low. Meanwhile, in view of the "lift-off" phenomenon, we feel that a correction of some sort is indicated.

- Conversion Factors - Column S

In Box 6, Column S provides the conversion-factors used in our Master Table. Each entry is the estimated number of radiation-induced excess breast-cancer cases (lifetime incidence) per 10,000 females (USA) from an average breast-dose of one medical rad. The entries are the product of Columns "O" x P x Q x R.

### # # # # #

```
```
 - "Doesn't Everything Cause Cancer If the Dose Is High Enough?           This question is asked at page 8 of the National Cancer Institute's 1990 booklet, and answered as follows:           "No. High doses of many chemicals are toxic, but they will not cause tumors. Other forms of toxicity, such as loss of hair or weight, various organ malfunctions, or even death, should not be confused with carcinogenesis." And:           "In one study, 120 pesticides and industrial chemicals were tested at the highest doses the mice could tolerate and survive. The mice were exposed for two years. These chemicals were not randomly selected, but were chosen because they were suspected of carcinogenicity. However, only 11 of these chemicals caused cancer in the test animals."

 - "What Happens When People Are Exposed to Several Carcinogens at the Same Time?"           This question is asked at page 10 of the National Cancer Institute's 1990 booklet, and answered as follows:           "The resulting cancer rate may be higher than would be predicted by adding the risks from each carcinogen alone. Cigarette smoking and asbestos exposure, for example, each cause cancer. But asbestos workers who smoke are subject to a cancer risk that is far higher than would be expected by adding the risk from smoking to the risk from asbestos." And:           "Animal studies have shown similar jumps in cancer rates from multiple exposures. This effect does not occur in animals with all carcinogens, and sometimes two carcinogens will somehow interact to give reduced rates of cancer. Although it is conceivable that this type of reduction could occur with human exposures, it is not something people can count on."

Source:   National Cancer Institute (USA), "Everything Doesn't Cause Cancer," 12-page booklet, March 1990 (NIH Publication 90-2039).

Next | ToC | Prev
back to PBC | CNR | radiation | rat haus | Index | Search