Below are the first 10 and last 10 pages of uncorrected machine-read text (when available) of this chapter, followed by the top 30 algorithmically extracted key phrases from the chapter as a whole.
Intended to provide our own search engines and external engines with highly rich, chapter-representative searchable text on the opening pages of each chapter.
Because it is UNCORRECTED material, please consider the following text as a useful but insufficient proxy for the authoritative book pages.
Do not use for reproduction, copying, pasting, or reading; exclusively for search engines.
OCR for page 184
Chapter XIII
THE AUTOPSY FINDINGS
ONE facet of the attempt to characterize the
kinds and frequencies of abnormal terminations
occurring to exposed and non-exposed persons
in Hiroshima and Nagasaki was as ambitious
an infant necropsy program as space and per-
sonnel would permit. The performance of
infant autopsies began in Hiroshima in 1948
although it was not until May, 1949 that a con-
centrated effort was made to obtain the bodies
of as many stillborn registered infants or regis-
tered infants dying during the first six days of
life as was possible. Autopsies were not per-
formed in Nagasaki until somewhat later (Octo-
ber, 1949), due to lack of personnel. The pro-
gram in Nagasaki never attained the standard
of exhaustiveness which we felt necessary to
permit valid inferences. The reasons for this are
somewhat complex but revolve largely around
the fact that both the ABCC and Dr. I. Hayashi
of the Nagasaki University Medical School were
attempting to conduct an infant autopsy pro-
gram. The efforts of Dr. Hayashi were moti-
vated both by the requirements of an active
teaching program in anatomy and an interest
in a possible relationship between irradiation
and malformation. Under the circumstances ob-
taining in Nagasaki, we could not assure our-
selves that the infants whom we obtained for
necropsy represented a random sample. Ac-
cordingly, the data obtained in Nagasaki will
be presented later only with a view towards
supplementing the data of Dr. Hayashi (see
Hayashi, 1955; Sevitt, 1955 ~ .
The genetic argument which would lead us to
expect differences between exposure cells in
the frequency with which major congenital
anomalies would be detected at autopsy is the
same, in all major particulars, as the argument
advanced in Section 8.3.
13.! The randomness of the Hiroshima an-
topsies.- In the interval 1948-1953 approxi-
mately 750 infant cadavers were collected for
i84
autopsy in Hiroshima. Of this number, 717
were actually autopsied and the remainder were
rejected because of advanced autolysis. Among
the 717 infants autopsied only 431 represented
registered pregnancies. The 286 unregistered
infants who were autopsied have been excluded
from the data to be presented in this chapter
because they are known to be a non-random
sample of all unregistered pregnancies. During
the interval, March, 1948-May, 1949, only a
few autopsies were performed on registered in-
fants, the majority of these involving "special
interest" material, and certainly non-random.
In May of 1949, the autopsy program was
greatly expanded. We have felt it wise to ex-
clude from consideration all autopsy results
obtained prior to this expansion of the program,
as well as autopsy material not meeting the
restrictions placed on all data and described in
Section 6.4. The actual analysis is thus confined
to 406 autopsies conducted on registered in-
fants between May, 1949, and December, 1953.
Tables 13.1 through 13.4 present an analysis
of the randomness of this autopsy sample.
From Table 13.1 we note that: (1) the in-
fants coming to autopsy were distributed by
exposure in the same fashion as those registered
infants not coming to autopsy; (2) a signifi-
cantly higher proportion of males than females
was obtained; and (3) in all, some 62 per cent
of the infants who were stillborn or who died
in the first six days of life were examined at
necropsy. In view of the absence of an interac-
tion between sex and exposure and also the
absence of a significant difference in the ex-
posure distribution of autopsied and non-autop-
sied infants, it would appear that valid infer-
ences could be drawn from these data. However,
this body of data is no less affected by con-
comitant variation than those portions of the
total data presented in preceding chapters. Ac-
cordingly, it is of some interest to determine
OCR for page 185
The Autopsy Findings
TAl3LE 13.1
A. The randomness of the distribution of autopsied
infants by sex of infant and parental exposure
Mother Father
Neither ex- ex
parent
ex
~^x posed
Male{xb 1 90
Female
Total:
Ox
156 98
91 60
334 201
211 127
posed; posed;
father mother
not not
103 23
67 18
24
13
47
31
26 304
10 174
65 647
37 406
a Total number of infants stillborn or dying during
the first 6 days of life.
b Number of infants stillborn or dying during the
first 6 days of life who were autopsied.
B. Chi-square analysis of the frequency of autopsy
by sex of infant and parental exposure
Source%2
Total 13.6767
Sex-exposure interaction. 6.2765
Sex (unadjusted) 7.4599
Exposure (adjusted for
sex) 6.1008
50.OI
40.0
MALf OR'vlAT ION
(PER CENT)
30.0
20.01
10.0|
185
DF P
7 .05-.10
3 .05-.10
1 .01-.001
6 .30-.50
1
whether the autopsied infants differ from the
non-autopsied infants in such sources of ex
traneous variation Al, the age of the mother at
the birth of the infant, the economic status of
Both the parents, and the presence of a positive
parents serology on the part of the mother at the time
posed Total of delivery.
39 343 Let us consider first the evidence with regard
27 232 to maternal age. In Figure 13.1, the frequency
of malformation among autopsied infants is
distributed by maternal age. Mother's age is,
we see, a rather important factor in the fre
quency of malformation among infants coming
to autopsy. Krooth's (1955) index of absolute
difference which may be used to evaluate the
significance of these data is, in this instance,
10.61 per cent (variance 1.1770). Maternal
age differences could, then, be an important
source of bias if maternal ages are dissimilarly
distributed among exposure cells. From Table
13.2, we note (a) no evidence of interaction
between the sex, autopsy, or exposure categories,
(b) no effect of sex of infant nor the occurrence
of an autopsy on the age of the mother, but
/
/
/
/
/
/
~0 ~
1 1 . 1 1 -A 1
45 20 25 30 35 no 45
MOTHER S AGE
FIGURE 13.1 The distribution of the frequency of autopsied infants with major malformation, relative to
all autopsied infants, by age of mother with parity ignored.
OCR for page 186
86
Genetic Efects of
(c) a significant difference in mothers' ages
between exposure cells. This latter effect is
principally characterized by a pronounced in-
crease in mean maternal age when both parents
were exposed as opposed to those cases where
only one or neither parent was exposed. The
Atomic Bombs Chapter XIII
is not significant. The apparent trend is not
unexpected in view of the fact that the ABCC
assumed all expenses in the cremation, etc. of
the infant cadaver. It seems quite probable that
this would be a consideration of some moment
to parents in the lower economic strata. The
TABLE 13.2 MEAN MATERNAT AGE OF INFANTS STILLBORN OR DYING DURING THE FIRST 6 DAYS OF LIFE
BY SEX OF INFANT, PARENTAL EXPOSURE, AND THE OCCURRENCE OF AUTOPSY
Class of Neither parent Mother exposed; Father exposed, Both parents
infant exposed father not mother not exposed Total
Autopsied
r 27.16 26.91 26.39 28.96 27.24
Males ~ (120) (67) (18) (27) (232)
r 26.70 26.90 27.54 31.70 27.12
Females ~ (91) (60) (13) (10) (174)
Not autopsied
Total
r 26.02 26.08 25.60 28.42 26.28
Males ~ (58) (36) (5) (12) (111)
~ 27.26 27.84 28.18 28.44 27.65
Females ~ (65) (38) (11) (16) (~30)
r 26.79 26.62 26.22 28.79 26.93
Males ~ (178) (103) (23) (39) (343)
r 26.94 27.27 27.83 29.69 27.35
Females ~ (156) (98) (24) (26) (304)
Analysis of variance
SourceSSDF MS F
Sex 41.872 1 41.872 1.68
Autopsy 11.715 1 11.715 2.12
Exposures 312.946 3 104.315 4.20* *
Interactions 188.875 10 18.888 1.32
Between 537.415 15 35.828 1.44
Within 15,707.445 631 24.893
Total 16,244.860 646
TABLE 13.3 THE DISTRIBUTION OF ECONOMIC STATUS AMONG A{rrOPSIED AND NON-AUTOPSTED INFANTS
-
Infant Very poor
Autopsied 5
Not autopsied 2
Total 7
Poor
56
18
74
Economic status
A_ ~
Average Well-to-do RichUnclassifiable Total
325 20 406
210 11 241
535 31 647
%2 = 6.474 DF = 3 P = .05-.10
difference in maternal age distributions will
lead chiefly to a bias which results in an increase
in the frequency of congenital malformations
when both parents are exposed relative to the
case when neither or only one parent is exposed.
With respect to the distribution of economic
status we note that the autopsied infants appear
more frequently to represent the lower economic
groups (see Table 13.3) although this trend
rise in frequency among the well-to-do parents,
if real, is quite possibly correlated with higher
levels of education and an appreciation of the
information to be gained from a postmortem
examination. In view of the absence of a signifi-
cant difference in economic classes it seems un-
likely that the autopsy data with regard to mal-
formation are grossly biased by parental eco-
nomic status.
OCR for page 187
The Autopsy Findings
In Table 13.4 is given the distribution of
positive serologies by parental exposure and the
occurrence of an autopsy. It is apparent from
the analysis of these data that the autopsied
infants differ somewhat from those not autop-
sied in the frequency with which the mother of
the infant has a positive postpartum serology.
The reason for this is not apparent. However,
there is no evidence of an exposure effect with
respect to serology, nor is the autopsy-exposure
interaction significant. Accordingly, we condude
that there is no evidence of significant ex
187
exposure than in Nagasaki, and as a conse-
quence, an irradiation effect should be more
pronounced in Hiroshima than in Nagasaki
(other factors being equal, of course).
2. The sampling scheme underlying those in-
fants coming to autopsy in Hiroshima is reason-
ably well known.
3. The data from Hiroshima are confined to
infants where the age range is known; the in-
fants were all of no less than 21 weeks' gesta-
tion and were either stillborn or died in the first
six days following birth.
TABLE 13.4 THE DISTRIBUTION OF POSITIVE SEROLOGIES BY PARENTAL EXPOSURE AMONG AUTOPSIED AND
NON-AUTOPSIED INFANTS
(Serology is scored as the number of infants whose mother's serology was positive or negative.)
NeitherMother Father Both
parentexposed; exposed; parents
Infant exposedfather not mother not exposed Total
Autopsied
Positive 11 6 0 2 19
Negative 200 121 31 35 387
Total 211 127 31 37 406
Not autopsied
Positive 15 4 2 1 22
Negative 108 70 14 27 219
Total 123 74 16 28 241
Total
Positive 26 10 2 3 41
Negative 308 191 45 62 606
Total 334 201 47 65 647
Chi-square analysis a
^ ~
Source %8 DF P
Total 9.5635 5 .05-.10
Autopsy-exposure interaction 3.8209 2 .10-.20
Autopsy 5.0427 1 .02-.05
Exposure (unadjusted) 2.4682 2 .20-.30
a Father exposed and both parents exposed pooled.
traneous variation with respect to the frequency
of possibly syphilitic mothers among the
mothers of autopsied infants.
13.2 The data. Two bodies of data exist
pertinent to the occurrence of major malforma-
tions among infants born to exposed and non-
exposed parents and coming to autopsy in Hiro-
shima and Nagasaki. The first body of data to
be presented here is the data obtained in Hiro-
shima under the auspices of the Atomic Bomb
Casualty Commission. We present these data
first for the following reasons:
1. Proportionately more "exposed" individuals
in Hiroshima received significant amounts of
In Table 13.5 are presented the data with
regard to the frequency of congenital malforma-
tions among infants autopsied in Hiroshima,
distributed by parental exposure without ad-
justment for differences in maternal age. Four
comparisons have been made, namely, the
equality of the frequency of congenital mal-
formations (a) among all four exposure cells,
(b) between infants born to non-exposed par-
ents and those born to parents one or both of
whom were exposed, (c) between infants born
to non-exposed parents and to exposed parents
one or both of whom were exposed and in ex-
posure classes 4 or 5, and (d) between infants
OCR for page 188
188
Genetic Effects of Atomic Bombs Chapter XIII
born to exposed parents with exposures less cent of the parents received inappreciable
than 4 with infants born to parents in exposure amounts of irradiation (exposure class 2~. It
classes 4 or 5. In no one of these instances does seems doubtful, therefore, that the autopsy data
there emerge a significant difference. In view are sufficiently extensive to detect small irradia-
of the effect of maternal age, adjustment for the lion effects such as a priori considerations would
age differences among exposure groups would lead us to expect.
TABLE 13.5 THE DISTRIBUTION BY PARENTAL EXPOSURE OF INFANTS BORN IN HIROSHIMA AND FOUND TO
BE GROSSLY ABNORMAL AT AUTOPSY
(a) Distribution of abnormal infants by parental exposure when exposure is defined as "present in city at the
time of the bombing"
Neither parent Mother only Father only Both parents
Class of infant exposed exposed exposed exposed Total
Normal 174 107 24 28 333
Abnormal 37 20 7 9 73
(17.5~) (15.8%) (22.6~o) (24.3~7o)
Total 211 127 31 37 406
(b)Distribution of abnormal infants by parental
exposure when exposure is defined in terms of the
categories given in Chapter IV
Both parents
One or both exposed but
parents in ex- neither in ex
Class of posure categories posure categories
Infant 4 or 5 4 or 5
Normal 33 25
Abnormal 5 (13.2%) 8 (24.2%)
Total 38 33
(c) Analysis of data in Tables 13.5a and 13.5b
Contrast %2
All exposure cells in 13.5a 1.912
Neither parent vs. one or
both parents exposed in
13.5a 0.059
Neither parent (13.5a) vs.
One or both parents ex
posed as 4 or 5 (l3.5b) . 0.440
All exposure cells in 13.5b 1.451
DF P
3 .50-.70
1 .80-.90
1 .50-.70
1 .10-.20
TABLE 13.6 THE DISTRIBUTION BY EXPOSURE CLASS OF THE EXPOSED PARENTS GIVEN IN TABLE 13.5
Father not Mother not more heavily
exposed; exposed; Exposed
mother in father In parent In
class shown class shown class shown
Exposure in first in first in first
class column column column Total
2 72 (56.7%) a 14 (45.2~o) 18 (48.6%`J 104 (53.3%)
3 38 13 7 58
4 4 1 4 9
5 13 3 8 24
Total 127 31 37 195
a Percentage of total.
lead to differences between exposure categories
deviating in a greater fashion from what is to
be expected under an hypothesis of radiation-
induced genetic changes. Obviously the sensi-
tivity, as it were, of the comparisons given in
Table 13.5 is a function of the number of
parents who received significant amounts of
irradiation. The distribution of the exposed
parents by exposure category is given in Table
13.6, from which we note that about 53.6 per
A second body of autopsy data bearing on the
problem of radiation-induced genetic changes
in the malformation rate is the data collected by
Professor Ichiro Hayashi of the Nagasaki Uni-
versity Medical School. We feel compelled to
present and discuss these data in some detail
because of the recent grossly over-simplified,
biased presentation of these data by Sevitt
(1955) . Professor Hayashi has kindly furnished
us with a copy of a preliminary report on these
OCR for page 189
The Autopsy Fir~dir~gs
data given as an address at a meeting of the
Japanese Atomic Bomb Investigation Group and
members of the Japan Science Council, held at
the Nagasaki Medical School on 3 October
1955. At the outset of a discussion of Hayashi's
data, it should be pointed out that his data differ
in two important respects from the data we have
presented from Hiroshima, namely, (1) the fact
that the age range among the infants coming to
autopsy is not clearly defined (they are described
as fetuses or newborn infants), and (2) the
definition of exposure. Hayashi's exposed cases
are defined as those infants who were born to
parents one or both of whom lived Wilshire 10
kilometers of ground zero at the time of the
bombing (August 9, 1945~. It should be noted
that the definition specifies lived within, which
does not necessarily imply present within.
Hayashi asserts that the location of these parents
with regard to the hypocenter is still under in-
vestigation. If we assume, however, that all of
the individuals who lived within were in fact
present within 10 kilometers of the hypocenter
at the time of the bombing, Hayashi's definition
of exposure is still, in our estimation, a most
unfortunate one for a variety of reasons. The
most important of these reasons is that 10 km.
extends some 7 km. beyond the area of signifi-
cant irradiation (5r or higher) except possibly
in the "fallout" sector (principally the Nishi-
yama reservoir area). A definition as all en-
compassing as this can only lead to a dilution
of radiation effects, if such exist, because of
the inclusion of large numbers of essentially
unexposed persons. An additional disadvantage
of Hayashi's definition which is worth men-
tioning is that a radius of 10 km. extends far
beyond the limits of the city of Nagasaki (par-
ticularly to the north, east, and west) and hence
creates difficulty in any attempt to determine
the completeness and/or randomness of sam-
pling. While we believe there is ample justifica-
tion for a conservative definition of exposure,
Hayashi's definition would appear overly con-
servative.
It is extremely difficult to appraise the real
significance of Hayashi's data. Among the fac-
tors which make it difficult is, as Professor
Hayashi has pointed out, the lack of information
regarding (a) the randomness of the sample,
(b) the degree of exposure of the parents, and
(c) other etiological agents, such as maternal
age, of importance in congenital malformations.
189
Some measure of the importance and effect of
these factors can be gleaned from the corpus
of data collected by the Commission. With re-
gard to maternal age, we have commented re-
peatedly on both its importance and the fact
that maternal ages are so distributed among the
terminations occurring in Hiroshima and Naga-
saki as to bias upwards an estimate of the effect
of irradiation.
With respect to the degree of exposure most
probably experienced by the parents in Hayashi's
series, the following observations are pertinent:
1. All dosage estimates, both Japanese and
American, of the bombs detonated over Hiro
TABLE 13.7 HAYASH! s DATA ON CONGENITAL
AsNoRMAt~T~Es IN RELATION To
EXPOSURE OF PARENTS
(After Sevitt, 195 5 ~
Category
One or both parents ex
posed to A-bomb497 a
Both parents exposed 149
Father exposed 80
Mother exposed 259
Neither parent exposed363
Both parents unknown
exposure; information
not available27
%2 (2 X 4~= 10.718 DF = 3
%2 (2 X 2~= 9.062 DF= 1
a Includes 9 children "mother exposed, father un-
known."
Number Number %
of mat- mal-
infants formed formed
92 18.9
27 18.1
18 22.5
47 18.1
40 1 1.0
10
P < 0.02
P < 0.01
shima and Nagasaki indicate that the instan-
taneous irradiation at 3,000 meters was less than
5r and consisted of gamma irradiation alone.
2. The total irradiation received by an indi-
vidual is the sum of the instantaneous irradia-
tion and the residual irradiation to which he or
she may have been exposed. The major com-
ponent of residual irradiation arises from the
so-called "fallout" following the bombing. The
area of fallout in Nagasaki is reasonably well
defined, and is, as has been indicated, east of
ground zero in the region of the Nishiyama
reservoir.] Save for a very small area immedi-
ately adjacent to the reservoir, the estimated
residual irradiation from one hour to infinity
~ The fallout area could be described as a band of
about 3,000 meters' width (on its north-south axis)
extending eastward from a point 1,500 meters from
the hypocenter.
OCR for page 190
190
Genetic Effects of
following the bombing is a dose probably less
than 6r, and possibly less than 2r. The area
encompassed by the fallout is in the main rela-
tively sparsely populated, and consists princi-
pally of hills and rice paddies. It is a safe con-
clusion that relatively few persons residing
outside 3 km. from ground zero received ex-
posures of the order of magnitude of 5r from
the fallout.
3. In the interval 1949-1953, some 363 regis-
tered infants came to autopsy in Nagasaki under
the auspices of the ABCC. The distribution by
exposure of these infants is compared with
those autopsied by Hayashi in Table 13.8. The
two distributions are not significantly different
and by inspection appear to be very nearly
identical one with the other. In Table 13.9, we
have given the distribution of exposed parents
Atomic Bombs Chapter XIII
Hiroshima data, derived from a group whose
average exposure was greater, is also 18.5 per
cent (36 infants out of 195~. Be that as it may,
the more important consideration is that Haya-
shi's control group has a frequency of con-
genital malformation of only 11 per cent (40
out of 363), whereas the Hiroshima control
group has a frequency of 17.5 per cent (37 out
of 211~. This is a significant difference, yet the
evidence from physical examinations presented
in Tables 8.6 and 8.7 suggests no difference in
the malformation rates between these two cities.
(It need hardly be pointed out that if Hayashi's
exposed group is contrasted with the control
group from Hiroshima no significant difference
emerges.) Further evidence which suggests that
Hayashi's control is low stems from the fol-
lowing considerations: (1) Among registered
TABLE 13.S A COMPARISON OF THE EXPOSURE DISTRIBUTION OF HAYASH! S DATA AND THAT OF THE
ATOMIC BOMB CASUALTY COMMISSION COLLECTED IN NAGASAKI
Neither parent Mother exposed; Father exposed; Both parents
Source exposed father not mother not exposedTotals
Hayashi 363 (42.66%) 259 (30.43%) 80 (9.40%) 149 (17.51%) 851
ABCC 155 (~.70%) 115 (31.68~o) 34 (9.37%) 59 (16.25~o) 363
Total 518 374 114 208 1,214
%2 = 0.364
P _ .9-.95
8 See text for differences between Hayashi's and ABCC's exposure definitions.
in our series in terms of the categories of ex
posure defined in Chapter IV. We note that the
large majority of exposed parents (87.5 per
cent) are in exposure category 2 wherein the
average exposure did not exceed 5r. Since
Hayashi's definition of exposure must encom
pass proportionately more individuals outside
the 3 km. zone than are presently residing in
Nagasaki (and hence would enter our study)
we can only conclude that an even greater
number of Hayashi's exposed cases are infants
born to parents whose exposure was insignifi
cant. In summary, then, it would appear that
the great bulk of Hayashi's exposed parents
must have received exposures of less than 5r. a Percentage of total.
The representativeness of Hayashi's series is,
as has been indicated, a matter of conjecture;
however, certain observations which we shall
now discuss suggest that the series may be
biased. We note that Hayashi's data (see Table
13.7 ~ reveal a frequency of congenital mal-
formation of 18.5 per cent among the exposed
group (92 out of 497 infants) whereas the
TABLE 13.9 DISTRIBUTION BY EXPOSURE CLASS OF
EXPOSED PARENTS WHOSE INFANTS CAME TO
AUTOPSY AT ABCC IN NAGASAKI
Father Mother not
not exposed; exposed;
mother in father in
Expo- class shown crass shown
sure in first in first
class column column
2 101 (87.8~%) a 33 (97.1%)
..... 12 1
2 2
Total 115 34 59
Both parents
exposed;
more heavily
exposed
parent in
class shown
in first
column
48 (81.4~o)
9
infants who were stillborn or who died in the
first six days of life and were born to non-
exposed parents in Nagasaki (both parents ex
posure category 1), no less than 9.4 per cent
(42 out of 444) had clearly visible major mal-
formations. (2) Experience in Hiroshima sug-
gests that for every eight infants coming to
OCR for page 191
The Autopsy Fir~dir~gs
autopsy on whom a prior clinical diagnosis of
major malformation has been made (verifiable
at autopsy, of course), six cases will come to
autopsy and a major malformation will be found
on which no prior clinical diagnosis exists. In
short, among infants coming to autopsy, clinical
examination will reveal only 58 per cent of the
infants with major malformations. These two
observations taken conjointly suggest that a
more reasonable figure for Hayashi's control
would be no less than 16 per cent.
From the preceding considerations we are led
to the conclusion that the more obvious explana-
tion for Hayashi's Endings lies either in ex-
traneous variation or in non-representativeness
of the control, rather than in irradiation (for
191
example, a maternal age effect similar to the
one noted in Figure 13.1 could easily produce
a spurious radiation effect in a larger series of
cases) .
13.3 Summary.- The data with regard to
the frequency of congenital malformations
among the pregnancy terminations to exposed
and non-exposed persons in Hiroshima afford
no evidence of a significant irradiation effect,
although the small differences observed are in
the direction of genetic expectation. Reasons
are advanced for believing that the "irradiation
effect" reported by Sevitt ( 195 5 ) from Haya-
shi's data in Nagasaki is more likely due to
extraneous variation (or sampling biases) be-
tween the exposure cells than to irradiation.
Representative terms from entire chapter:
parental exposure
