AT 1pm on October 8, the final curtain came down on Hope and Reay v
British Nuclear Fuels, so ending one of the longest suits in the history
of British civil law. The ‘Sellafield cancers case’, as it was dubbed by
the nation’s headline writers, had all the makings of classic legal drama
– allegations about radioactivity and genetic damage, families of children
with cancer seeking compensation, a large and powerful industry in the dock.
The case also threatened to break new legal ground. For the first time a
British court was being asked to rule on personal injury claims based on
alleged genetic damage from radiation. Victory would throw the doors wide
open for other Sellafield families to sue BNFL.
At issue was whether the cancers suffered by Vivien Hope and Dorothy
Reay – the daughters of lifelong BNFL employees–could be blamed on radiation
damage to their fathers’ sperm at BNFL’s nuclear reprocessing plant at Sellafield.
But as the hearing unfolded it soon became clear that the safety of the
Sellafield plant was not the only thing on trial. Also in the dock was the
competence of science, specifically its ability to judge the risks of radiation
to reproductive cells. Could exposure to relatively low doses cause genetic
defects to be passed on to children through damaged sperm?
Expert witnesses flew in from across the globe and for 90 days scientists
and lawyers grappled with a mass of seemingly conflicting findings. Half
this time was devoted to discussing epidemiological research into the possible
links between radiation exposure to adults and cancers in children. And
in this crucial part of the hearing the court examined more than 15 different
studies. In the end, however, the debate about these studies boiled down
to a duel between two opposing pieces of research.
Advertisement
The plaintiffs’ hopes were pinned largely on a controversial British
study by the late Martin Gardner, professor of Medical Statistics at the
University of Southampton. Three years ago, Gardner claimed to have found
a statistical link between the incidence of leukaemia and non-Hodgkin’s
lymphoma in children born to men employed at Sellafield and the amount of
radiation the men had been exposed to before their children had been conceived.
Pitted against his research in court were results from the largest and most
influential of all studies of the genetic effects of radiation – the huge
follow-up study of survivors of the atomic bomb blasts in Hiroshima and
Nagasaki in 1945. This study, as BNFL’s lawyers eagerly pointed out, found
no excess of cancers or genetic defects in the children of survivors who
had been exposed to radiation.
When the hearing ended, the ruling ran to more than 200 pages. Yet the
essence of the judgment was simple enough: the ‘negative’ findings from
the atomic bomb study were deemed more convincing than Gardner’s ‘positive’
findings. In the judge’s eyes, the atomi-bomb study seemed to convince researchers
who advise governments on radiation risk, while Gardner’s study had provoked
some seemingly valid criticisms from BNFL in court. In particular, since
the link with radiation was based on a relatively small number of cancer
cases, some witnesses for BNFL worried that Gardner and his colleagues had
unwittingly selected a biased set of geographical boundaries and subjects.
‘BNFL cleared’ was the tone of most newspaper headlines. Sadly, the
same cannot yet be said of science and its ability to evaluate the risks
of radiation exposure. Since the case closed a large scale re-examination
of the Sellafield cancer data has partly confirmed the statistical validity
of Gardner’s link between childhood cancers and occupational exposure to
radiation. And the courtroom debate exposed potential complications with
the atomic bomb research which have never before been aired in public.
The genetic studies on the children of atomic bomb survivors began in
the spring of 1948, and were led for many years by Jim Neel and Jack Schull.
Neel, now professor emeritus of human genetics at the University of Michigan
Medical School, Ann Arbor, became acting director of field studies for the
Atomic Bomb Casuality Commission in Hiroshima in 1947. Two years later he
was joined by Schull, now professor of human genetics at the University
of Texas. The Radiation Effects Research Foundation succeeded the ABCC at
Hiroshima . .
From the outset of the case, the results from the ABCC and RERF posed
a major stumbling block to the Hope and Reay families. In more than forty
years of research, Neel, Schull and their colleagues have failed to uncover
any convincing evidence of an excess of harmful genetic mutations in children
whose parents had been exposed to the atomic bombs. This is despite the
fact that they have looked at as many as 75 000 such children, and that
the average exposure to radiation at Hiroshima and Nagasaki was higher than
that for workers employed at Sellafield.
The contrast with Gardner’s results couldn’t be greater. In 1990, he
and his colleagues published a paper in the British Medical Journal claiming
that children born to Sellafield workers who had previously been exposed
to a cumulative dose of more than 100 millisieverts of radiation – twice
the maximum annual dose presently allowed under British law – were eight
times more likely than other children to develop leukaemia or non-Hodgkin’s
lymphoma.
Gardner believed that the discrepancy between his study and the Japanese
data could best be explained either by differences in radiation dose rates
– a sharp blast in the case of the atomic bombs versus chronic exposure
to low levels at Sellafield – or else by differences in the average intervals
between exposure and conception. But the plaintiffs’ lawyers knew this wouldn’t
be enough to convince the court. In addition, they had to try to expose
weaknesses in the ABCC and RERF study, to explain why Neel and his colleagues
did not find an excess of cancers in children of atomic bomb survivors.
Bizarre paradox
Part of their approach was to focus on what seems to be a bizarre paradox
in the Japanese data: namely, that the children born to atomic bomb survivors
appear from these data to be healthier than children of the same age born
in Japan as a whole. And not by a trivial margin either. Stillbirths were
down 85 per cent, deaths from cancer by 20 per cent, overall mortality by
30 per cent and infant mortality (from all causes) by 35 per cent. So counterintuitive
are these findings, argued Eva Alberman, a reproductive epidemiologist at
St Bartholomew’s Hospital in London and a key witness for the plaintiffs,
that they can mean only one thing: the researchers must have inadvertently
failed to identify all the deaths and genetic deformities in children of
survivors who had been exposed to radiation. It was either that, or the
bombs had conferred some beneficial effect – which ‘seems crazy’. This was
robustly denied in court by Neel and Schull, who said the beneficial effect
was either statistically insignificant or the product of terminology differences.
Testifying in court, Alberman said that some of her doubts about the
completeness of the Japanese data sprang from two apparent inconsistencies:
first, that the data suggest the incidence of Down’s syndrome among children
born to survivors was half what would be expected on the basis of statistics
from other countries; and secondly, that they show that Down’s syndrome
was twice as prevalent in the children of survivors classified as ‘not exposed’
as it was in children of ‘exposed’ survivors.
Neel and Schull’s explanation is that, on statistical grounds, the rates
are too unreliable to compare in that way. But Alberman believes that the
most likely explanation is that the ABCC study failed to diagnose all cases
of Down’s syndrome. And if that is the case, she says, perhaps other congenital
conditions had been underdiagnosed. According to the ABCC’s figures, about
1 in 100 babies conceived by atomic bomb survivors between 1948 and 1954
were diagnosed as having a major congenital deformity such as hydrocephalus,
spina bifida and polydactyl. The comparable figure for Britain in 1958 was
1.7 in 100.
In the early years after the bombs, Alberman told the court, most Japanese
women gave birth at home, attended by midwives who were not trained in diagnosing
deformities. Another potential problem might have been deliberate underreporting.
Neel’s opinion about this seems to have shifted with time. In court he said:
‘It rapidly became known throughout Hiroshima and Nagasaki that malformation
was a possible outcome of exposure to the bomb. This did much to remove
the stigma because then you had a causal agent inflicted upon the women.
It was a cause she could accept.’ But in 1958 Neel wrote: ‘The birth of
a malformed child stigmatises the family involved to a greater extent in
Japan than in most Western cultures, and it seemed quite possible that the
physician-interviewer had not established sufficient rapport with the family
to obtain an accurate history.’
So could the ABCC and RERF studies have underestimated the medical problems
suffered by the children of survivors exposed to radiation? Testifying on
behalf of BNFL, Neel and Schull accepted in court that the studies had been
difficult to carry out. But they were adamant that their estimates of the
genetic risks of radiation exposure were accurate. The main reason for their
confidence was that these estimates are based on a comparison of similar
groups of children – those born to people who had been exposed to radiation
from the bombs, and those born to people living in or near Nagasaki and
Hiroshima who had been exposed to little or no radiation.
The reported rates of stillbirths, congenital deformities, infant mortalities
and deaths from cancer are almost identical in these groups of children.
And it is this comparison that matters most, say Neel and Schull. Even if
the children of atomic bomb survivors seem to have been abnormally healthy
compared with children living elsewhere (and this claim was denied in court),
the fact that parental exposure to radiation seems to have had little affect
on the childrens’ health means only one thing in the eyes of the ABCC researchers:
that exposure to radiation failed to damage the reproductive cells of the
atomic bombsurvivors.
To shatter this certainty, the plaintiffs’ lawyers had to try to show
that the diagnosis of medical problems may have been uneven in the ‘exposed’
and ‘unexposed’ groups. Could the American researchers, for example, have
overlooked a disproportionate number of leukaemia cases in the children
of the ‘exposed’ groups? This was unlikely, said Neel, because any difficulties
with diagnosis and underreporting of medical problems would automatically
apply equally to all children in the study.
Yet there was one uncertainty which Neel freely acknowledged in court.
Early in their study, the ABCC researchers had had to abandon their attempts
to gather data on miscarriages in the first five months of pregnancy. It
became clear, said Neel, that it was impossible to monitor these losses.
This meant that the researchers had to make two assumptions when calculating
the genetic risks of radiation exposure: first, that fetuses were lost for
the same reason in ‘exposed mothers’ as in ‘unexposed mothers’; and secondly,
that the rate of loss was the same in all groups.
Because these assumptions are unverifiable Alberman contends that the
ABCC results cannot be used to assess accurately the inheritable risks of
radiation exposure. A high proportion of spontaneous abortions in the first
few months of pregnancy are thought to result from genetic mutations that
affect fetal development, she told the court, and there are at least some
grounds for thinking that such mutations might have been more prevalent
in the reproductive cells of survivors exposed to high doses of radiation.
If Alberman’s concerns are justified, then the ABCC’s estimates of genetic
risk are based on data gathered after the period during which most of the
lethal damage from radiation would have been eliminated.
Rocketing abortions
Another potential source of bias between the ‘exposed’ and ‘unexposed’
groups emerged in court: abortions. All over Japan the abortion rate rocketed
after the war. With thousands of people returning from Japan’s former colonies,
the government decided to make abortion freely available as a form of birth
control. One consequence of this policy was that as many pregnancies were
deliberately terminated in Hiroshima in 1950 as came to term.
What if a disproportionate number of these abortions were of fetuses
with genetic defects, asked the plaintiffs’ lawyers. Might not the high
abortion rate then mask some of the harmful effects of radiation? This was
unlikely, said Neel. The high abortion rate could only have biased the results
if some women chose abortions because they knew they had been exposed and
as a result suspected they might be carrying damaged fetuses. Certainly,
Schull told the court, there was a great fear among exposed survivors that
radiation might cause congenital deformities in children. But there was
no evidence that any women survivors had decided to have an abortion for
these reasons, he said. Rather, the most commonly acknowledged reason for
abortion was poor economic circumstances, according to a study of a small
sample of survivors by Schull. But this failed to satisfy all critics. Why,
asked Alberman, did Neel and Schull’s data from Hiroshima show an increasing
trend in abortion with increasing radiation dose.
It is clear from the ruling that the debate about miscarriages and abortions
failed to convince the judge that the diagnosis of medical problems in the
‘exposed’ and ‘unexposed’ groups of survivors could have been seriously
biased. But this still leaves a paradox unexplained: why did the children
born in both groups appear to be healthier than those born elsewhere in
Japan or in other countries? Why were the rates of infant mortality, stillbirths
and cancers seen in the children of exposed and unexposed survivors so low?
Let’s examine the explanations given by Neel and Schull in court.
*Cancer rates. Death rates from cancers in children of ‘exposed’ and
‘unexposed’ survivors were 20 per cent lower than in people of the same
age in the rest of Japan, according to the ABCC study. The researchers’
explanation? That cancer rates are rarely uniform within a country and it
therefore makes no sense to compare results gathered in specific towns and
cities with national averages.
But in court, cancer epidemiologist Scott Davis and biostatistician
Ken Kopecky, who have both worked at RERF, said that it was reasonable to
presume that the ABCC study had missed some cases of leukaemia, particularly
in the early years of the follow-up when conditions were difficult, and
that this may have contributed to the low cancer rate. Schull conceded that
just 10 misdiagnosed leukaemia cases among children of ‘exposed’ survivors
could have obscured a link between the cancers and inherited effects of
radiation exposure. One potential source of error highlighted in court was
the risk of misdiagnosing leukaemia as respiratory disease: children with
leukaemia are 300 times more vulnerable to pneumonia.
*Stillbirths. The frequency of stillbirths among atomic bomb survivors
was recorded to be six times lower than the number of stillbirths recorded
by the Japanese government for the districts of Hiroshima and Nagasaki as
a whole, and which included individuals who were too far away from the nuclear
explosions to be irradiated.
Neel argues that this gap reflects differences in the definitions of
stillbirth used by the ABCC and the Japanese government. The government
figures are relatively high, he maintains, because unlike those from the
ABCC, they include babies who lived for a few hours and then died – cases
which the ABCC counted as ‘infant deaths’ – and some deliberate abortions.
The ABCC figures, on the other hand, are relatively low because they exclude
miscarriages before the fifth month of pregnancy (the Japanese government’s
miscarriage records start from the third month) and stillborn babies with
major malformations.
But such differences cannot be the whole story, argue critics, who raise
three problems. First, since the ABCC was counting as ‘infant deaths’ cases
which the government termed ‘stillbirths’, one might expect the ABCC’s infant
mortality figures to be higher than those collected by the Japanese government
for Japan as a whole. In fact, the opposite is true. Secondly, the government
stillbirth rates for Hiroshima and Nagasaki were based on medical data for
the surrounding rural areas as well as for the cities themselves (unlike
the ABCC figures, which were largely confined to the cities). This makes
the real gap between the ABCC and government data even greater, for the
official stillbirth rate for cities in Japan was almost twice the rate for
rural areas, according to 1950 figures.
Finally, even without adjusting for urban-rural differences, it is hard
to make the government and ABCC figures tally. Adjusting the ABCC figures
so as to include babies stillborn between 3 and 5 months gestation and those
with major malformations would increase the ABCC figures by about 65 per
cent, while adjusting the government figures so as to exclude deliberate
abortions would produce a 10 per cent reduction in the official stillbirth
rate. The government and ABCC figures would still differ by at least a factor
of three. Alberman’s conclusion about the ABCC stillbirth figures: they
fail to include ‘babies that should have been registered but possibly died
before registration’.
*Infant mortality rates. Infant mortality rates for children of exposed
and unexposed survivors were 35 per cent lower than in Japan as a whole.
In the 1940s and 1950s most infant deaths in Japan were caused by infectious
diseases like TB, cholera and pneumonia, which exacted their biggest tolls
on poor, badly housed families with little access to health care. Neel told
the court that the most likely explanation for the low infant mortality
figures was ‘that living conditions were difficult in Hiroshima and Nagasaki
prefectures, but they were more difficult elsewhere’. If that is true then
things must have been very bad indeed elsewhere. Neel and Schull agreed
in court that there was a shortage of trained medical staff and diagnostic
equipment in Hiroshima and Nagasaki in the first few years after the bombs,
and in a 1990 paper they state: ‘It is not unreasonable to assume that the
aftermath of the bombing worsened the socioeconomic status of many survivors.’
As late as 1968 the Japanese government felt the need to pass a special
law ‘to extend welfare, by providing special allowances, to atom bomb survivors
who suffered injuries and even now experience special hardships’.
Until the recent High Court case, the unusually low rates of stillbirths,
mortality, congenital malformations and cancer reported for the children
of atom bomb survivors had barely been debated by scientists. Why? One reason
is the authority the ABCC study has acquired, worldwide, as a yardstick
for assessing the risks of direct exposure to radiation on non-reproductive
(somatic) cells. Many governments base their rules on occupational exposure
on the Japanese data. If there is nothing wrong with the data gathered from
people who were directly exposed, ask most scientists, why should there
be anything wrong with the data gathered for survivors’ offspring.
But the acceptance of the somatic studies has not been universal. For
three decades, Alice Stewart, an epidemiologist from the University of Birmingham,
has argued that the data underestimate the risk of cancer developing in
children exposed to radiation while still in the womb (see ‘Radiation: how
safe is safe’, New ÐÓ°ÉÔ´´, 15 May). Stewart discovered that children
who had been exposed to X-rays in the womb were 60 per cent more likely
than average to develop cancer. In contrast, the follow-up study of Japanese
children who had been exposed to radiation while in the womb found no increased
risk. Many scientists belittled Stewart’s findings for several years, until
other epidemiological studies began to confirm her findings.
Commenting on the discrepancy between the Japanese studies of womb exposure
and data from Western populations, Nick Day, director of the Medical Research
Council’s Biostatistics Unit, said in court: ‘Instead of the 2 childhood
cancer cases observed in the survivors, you would expect 11.45. So between
11.45 and 2 there is clearly quite a divergence.’
The results on womb exposure show most dramatically how the studies
on the atomic bomb survivors have been allowed to take precedence over other
studies that have produced conflicting results. Now that Stewart’s research
has been confirmed, the RERF researchers are claiming that their study is
consistent with her’s. Is the same likely to happen with the paper by Gardner?
Time will tell.
Sara Downs is a researcher who worked for the lawyers acting for Elizabeth
Reay and Vivien Hope. The views expressed in this article are not necessarily
those of her former employers.
* * *
Cancer theories in limbo
When the story broke on British TV in 1983, the equation seemed worryingly
simple. An unexpected outbreak of leukaemia among children living in Seascale
– a nearby nuclear reprocessing plant where most of the childrens’ fathers
worked. Instinct, buttressed perhaps by B-movie perceptions of the genetic
risks of radiation, suggested there might be a link, that Sellafield workers
who had been exposed to radiation might have unwittingly passed on genetic
mutations to their children. What else in west Cumbria could blamed for
the cancers?
Ten years on, and a deluge of scientific papers later, researchers are
still asking the same question. The enigma of the Seascale ‘leukaemia cluster’
has become one of medical science’s great whodunits, replete with imaginative
theories, tantalising clues and what appear to be sibling crimes in the
shape of leukaemia clusters in other places.
Instinct or no, radiation is no longer the only suspect. Over the past
few years rival epidemiologists have attempted – with mixed success – to
build cases against radon gas, electromagnetic fields from power cables
and chemical emissions. And as part of a quest to uncover possible links
with infectious agents, some researchers have begun to test radical ideas
about how socioeconomic factors might influence the geographical spread
of diseases. If they are right, the underlying culprit may, in classic fashion,
turn out to be one you would probably never suspect: affluence and improved
standards of sanitation.
But the plot is still young and firm bets would be inadvisable. Moreover,
in the case of Seascale the epidemiological evidence linking the leukaemias
to radiation stubbornly refuses to crumble away despite the best efforts
of critics. Last month Britain’s official watchdog on safety in the workplace,
the Health and Safety Executive, issued a report vindicating the statistical
methods of the controversial study at the centre of the debate.
In 1990 the late Martin Gardner and his team at the University of Southampton
reported higher than average rates of leukaemia in children born to men
who had previously been exposed to low levels of radiation while working
at Sellafield. Anti-nuclear campaigners seized on the finding, but scientists
were baffled. Gardner’s study conflicted with those of Japanese A-bomb survivors,
and the idea that irradiating sperm could produce leukaemia-causing mutations
in offspring seemed implausible biologically. With the exception of one
highly controversial study with irradiated mice, animal experiments suggest
that the radiation doses would have had to be a thousand times higher to
produce the necessary genetic mutation rate. And even then there would be
a problem. Radiation is no smart bomb: it cannot home in on specific genes.
So why would the mutations be limited to ones that turn white blood cells
cancerous? Why no excesses of other childhood cancers or congenital malformations
in Seascale?
Moreover, in the pop-ulation as a whole only a fraction of cases of
leukaemia are thought to have an inherited component. So even if low doses
of radiation could produce inheritable mutations in sperm, leukaemia would
be an unexpected outcome, says Roger Cox, of the National Radiological Protection
Board. A more likely legacy, he says, would be commonly inherited cancers
like Wilm’s tumour or retinoblastoma, which can be passed on through a single
defective gene.
Perhaps, then, Gardner’s finding is merely a cruel statistical mirage,
the product of unconscious observer bias acting on a relatively small number
of leukaemia cases in a village where many men worked at the nuclear plant.
The HSE report argues otherwise. After reworking Gardner’s data with new
control subjects, it concludes that the statistical link with radiation
is real for cases where the fathers had started working at the plant before
1965, when guidelines on exposure were less strict. But at the same time,
the HSE finds no evidence of such a link for cases of leukaemia in children
born to Sellafield employees living outside Seascale. The report offers
a two-stranded conclusion: occupational exposure to radiation cannot alone
explain the Seascale leukaemia cluster, but nor can it yet be excluded
as a contributing factor.
This glimmer of a reprieve for the radiation theory comes after a series
of blows which have left it losing ground to its main rival – the idea that
leukaemia clusters are caused indirectly by people moving into settled communities
to build and staff new industries. Such population mixing (the argument
goes) results in people suddenly being challenged by strains of viruses
and other infectious agents to which they have previously acquired little
resistance. And in such circumstances, an abnormally weak immune response
to a common, but as yet unidentified, infectious agent could conceivably
explain higher than average levels of childhood leukaemia.
The strongest advocate of this theory is Leo Kinlen, an epidemiologist
at the University of Oxford. Over the past few years, Kinlen claims to have
identified leukaemia clusters in places where there has been population
mixing – towns with army camps, oil workers or a high density of commuters.
In 1989, researchers reported leukaemia clusters at spots where nuclear
plants had been planned but never built. And more recently, Kinlen claims
to have identified a higher than expected leukaemia rate among children
who moved to west Cumbria but whose fathers were never exposed to radiation.
Critics argue that such studies are plagued by numerical problems. Yet
there are other reasons for taking the theory seriously, says Melvyn Greaves
of the Institute for Cancer Research in London. First, viruses are known
to cause two rare forms of leukaemia in humans. And secondly, there is plenty
of circumstantial evidence linking childhood leukaemia to general socioeconomic
trends. The disease is less prevalent in deprived urban centres than in
affluent villages and suburbs, for example, and its incidence in Britain
has mushroomed since the 1920s.
Paradoxically, says Greaves, leukaemia clusters may be a symptom of
dramatic improvements in sanitation and hygiene. The fact that young children
are no longer exposed to so many ‘immunising’ infectious organisms, he argues,
could make them susceptible to abnormal responses to viruses. Loss of immunity
due to improved living conditions is the conventional explanation for why
poliovirus infections increased among affluent families earlier this century.
‘It may be that immunising levels of infectious organisms are only being
maintained in places with high population density,’ says Greaves.
Like the radiation theory, the population mixing theory lies in epidemiological
limbo – neither proven nor firmly refuted. The HSE report concludes that
it ‘has to be seriously considered even if the precise way it operates is
far from clear’.
An alternative interpretation of the link between childhood leukaemia
and affluence points to car ownership. Children who spend long periods
inside cars will be exposed to higher than average levels of benzene, a
known carcinogen, argues Simon Wolff of University College London.
Other researchers have speculated that irradiation of sperm might produce
mutations that damage the immune systems of offspring, weakening their response
to leukaemia-causing viruses. But until candidate viruses are identified,
critics are likely to remain as sceptical about such ideas as they are about
the population mixing theory.
In the meantime, everyone will continue to puzzle over the statistical
link between the Seascale leukaemias and parental exposure to radiation.
Why does the link disappear for Sellafield employees living elsewhere? ‘There’s
something special about Seascale’ goes the refrain. But what?