The southern Chinese are nearly a hundred times more likely to develop nasopharyngeal cancer than Europeans, and researchers believe that a combination of genetic predisposition and chemicals in the salted fish consumed by the southern Chinese makes EBV especially likely to turn epithelial cells malignant. Because it is almost impossible to infect cultured epithelial cells with EBV, however, establishing how the virus causes nasopharyngeal cancer has been difficult.

Researchers led by Nancy Raab-Traub of the University of North Carolina at Chapel Hill have now solved this problem by transferring EBV genes directly into cultured epithelial cells. They focused on a gene for a protein called latent membrane protein 1 (LMP1). This gene becomes embedded in the outer membrane of infected cells and can interfere with the molecular pathways that communicate messages around a cell.

As soon as the genetically engineered cells started producing LMP1, they also began to make unusually large amounts of two human proteins: a cell surface molecule called epidermal growth factor receptor (EGFR) and a protein found in the interior of the cell called A20 (Journal of Virology, vol 69, p 4930).

These results, says Raab-Traub, suggest that EBV triggers cancer by a two-pronged asault. When EGFR binds to the corresponding growth factor, it promotes cell division. So if a cell carries large quantities of EGFR on its surface, it is more likely to pick up the signal to divide, and may begin to proliferate uncontrollably. The A20 protein, meanwhile, is known to block a controlled cell suicide mechanism called apoptosis, which is used to eliminate cells that divide abnormally.

The researchers now need to work out exactly how LMP1 reprograms epithelial cells to produce excessive amounts of these two proteins. One clue comes from work on white blood cells. In parts of Africa, EBV triggers a cancer of the white blood cells called Burkitt’s lymphoma. George Mosialos, Eliot Kieff and their colleagues at Harvard University have shown that the part of the LMP1 molecule which dangles from an infected white blood cell’s outer membrane into its interior interferes with the molecular pathway that transmits signals from the receptor that binds to a molecule called tumour necrosis factor. TNF promotes normal growth and division of white blood cells, but when LMP1 hijacks the pathway, the cells divide continuously and do not differentiate into specialised subtypes.

This interference, Kieff believes, underpins EBV’s ability to induce a variety of cancers. If so, it may be possible to treat EBV-triggered cancers by blocking the hijacked pathway. “If you could interfere with the activation of the TNF growth signalling pathway, presumably you could stop the signals from LMP1 at their root.”

The lake-beds that fill eastern Ethiopia’s Afar Basin are home to a remarkable collection of fossils from ancient elephant species. Jon Kalb, a palaeontologist at the University of Texas at Austin, has compiled a history of these elephantoids that stretches back 10 million years. It reveals a pattern of progressive migrations along the lakes, as tectonic activity shifted their position.

The remarkable fossil record is the result of the Afar’s geological environment, says Kalb. The region lies at the junction of the East African, African and Arabian tectonic plates. Lakes and rivers formed as rifts between the plates widened and deepened. As the lakes moved, following the plate boundaries, so did the animals that lived and died along their shores. Abundant sediments that washed into the rifts buried and preserved the fossils (see Diagram).

The elephantoid fossil record, combined with data on palaeoclimate and sea level, indicates when the large mammals may have crossed one of several land bridges out of Africa, into Asia, and back again, reports Kalb (Palaeogeography, Palaeoclimatology, Palaeoecology, vol 114, p 357).

Although they originated in Africa about 45 million years ago, the ancestors of modern elephants made several intercontinental treks during periods of low sea level. The earliest migration out of Africa was some 22 million years ago. Then, between 11.5 and 7 million years ago, elephantoids migrated back to Africa in two or three waves. Mammoths and the ancestors of the present-day Asian elephant probably evolved in Africa between 4 and 5 million years ago, and began migrating back into Eurasia two million years after that.

As landscape and climate changed, so did the animals. Eighteen distinct types of elephantoid fossil found in the Afar region chart the evolutionary history of today’s great mammals. For example, the most primitive elephants had four tusks that could slash through dense greenery, and teeth adapted to chewing soft vegetation. Those that evolved in the past 5 to 6 million years had two tusks and higher-crowned, blunt-ridged teeth adapted to chewing the coarse foliage of a harsher, drier climate.

“If Kalb can show that the animals are evolving, and moving around at times of sea level change, then he’s demonstrated that global climate change may have a direct effect upon animal migrations and evolution,” says Ray Bernor, a palaeontologist at Howard University in Washington DC. He has seen a similar link between lower sea levels and new species arising in primitive horses that migrated out of Africa.

The elephantoid fossils and lake deposits can also provide a road map for those hunting the remains of our human ancestors. The oldest hominid fossil is 4.4 million years old, and was found by anthropologist Tim White in 1993 at a site in Ethiopia that Kalb had predicted would be prime hominid territory.

The predecessors of humans and of elephants often inhabited the same lush lake shores and river banks. The 10-million-year elephantoid record may lead the way to our even earlier ancestors, perhaps those that existed directly before and after hominids diverged from the great apes.

Experimental drugs known as protease inhibitors, which block the enzyme,
are being tested on people with HIV. ��è����Ƶs hope that these new drugs will
be more effective than existing ones such as zidovudine (AZT). But there is
already clinical evidence that the virus can develop resistance to protease
inhibitors, and researchers are hoping that a deeper understanding of the
enzyme will allow them to engineer resistance-proof drugs.

Protease helps HIV to assemble new copies of itself by cutting to length
some important structural proteins that brace the viral core. The new viral
particles, called virions, look like a bull’s eye. But when protease is
defective, the release of virions slows to a trickle, and abnormal, doughnut-
shaped virions leave the cell already dead.

Although the structure and function of protease is known, scientists have
not understood where and when the enzyme acts in HIV’s life cycle: whether at
the membrane of an infected cell as virions bud from the surface, or
immediately after virions leave the cell.

Now Andrew Kaplan, a virologist at the University of California, Los
Angeles, has answered this question. He found that protease acts at the
membrane during budding, and blocking its activity sabotages the budding
process. His results are reported in the October issue of the Journal of
Virology.

Pinpointing where and when protease works helps scientists to understand
exactly how HIV is assembled from its core components, and how those
components interact. “The more we understand virus assembly the more likely we
can develop therapies to interfere,” Kaplan says.

As virions bud from an infected cell, they pinch off a section of the
cell’s membrane, and wrap themselves in it. Embedded within this “stolen”
membrane coat are about 2000 polypeptide strands, the precursors of the
structural proteins that make up the core of the virus.

Protease cleaves these precursors just as virions bud from the cell,
enabling the streamlined virus to pour out. About half the virions leave the
cell within 90 minutes. However, if protease activity is blocked, it takes 13
hours for half the virions to leave. More importantly, the particles that do
leave the cell are not able to infect other cells in cultures. “Just a tiny
perturbation in protease activity results in profound losses of infectivity,”
Kaplan says.

Apart from protease inhibitors, the only licensed drugs against HIV, such
as AZT target reverse transcriptase, an enzyme that allows the virus to
replicate in cells by building DNA from viral RNA. But such drugs do not
affect mutant viruses which have acquired resistance to them.

People with HIV who have been treated with protease inhibitors in clinical
trials begin by responding well to the drug, but they stop improving when HIV
develops resistance to it, says Kaplan. He believes that the enzyme should be
more vulnerable to drugs than reverse transcriptase because the function of
protease is less complicated and its structure much better understood than
reverse transcriptase.

Roger Levesque and his team at Laval University in Sainte-Foy, Quebec,
found the bacteria in the Agassiz ice sheet on Ellesmere Island. Because
the bacteria were frozen at different times, they provide a freeze-frame
view of bacterial evolution.

Martin Handfield, a member of the Quebec team, recovered microorganisms
from a 130-metre-long core drilled from the top of the ice sheet to the
bedrock. The core yielded six yeast-like organisms, four Gram-negative bacteria
and one organism yet to be identified. The microorganisms were probably
carried by winds to the ice sheet, where they were gradually buried.

Handfield recovered the frozen microorganisms from depths of 109 metres
and 120 metres, where the mean temperature is about -20 degree C. They had
survived the millennia in a state of suspended animation, neither metabolising
nor reproducing.

The organisms were close to the centre of the core. Handfield found
them after first sterilising the core’s outer layers with ultraviolet to
prevent contamination with modern microorganisms. He then melted the ice
core and filtered the meltwater.

When the microorganisms were warmed, a small fraction revived. Handfield
believes they probably survived by associating themselves with protective
particles, such as pollen grains or tiny dirt pockets.

Brian Duval, a microbiologist at the University of Massachusetts, Amherst,
believes the survival of the microorganisms shows that planets such as Mars,
which has water ice and carbon dioxide ice at its north pole, may also harbour
life. ‘If we were to go to Mars we would probably take a drill,’ says Duval,
who has worked with the NASA-Ames Research Center, California, in the study
of possible life beyond Earth.

But Levesque is more concerned with what the microorganisms will reveal
about the evolution of bacteria and their resistance to antibiotics. ‘The
key question is where antibiotic-resistant genes come from,’ says Levesque.
Part of the answer will become clear if these ancient bacteria contain antibiotic
genes or plasmids, small circular pieces of DNA that are separate from bacterial
chromosomes and which often code for antibiotic resistance in bacteria
living today.

According to Stuart Levy of the Center for Adaptation Genetics and Drug
Resistance at Tufts University Medical School in Boston, scientists have
found plasmids in collections of bacteria from the first part of the century
– prior to the antibiotic era. ‘We know plasmids existed before, but we
don’t know how long ago,’ says Levy. If the ancient plasmids are entirely
different from plasmids found today, it could mean that plasmids come and
go, or that they change dramatically over time. However, if the plasmids
closely resemble those of today, it would mean that they have been conserved
throughout bacterial evolution.

So far, in the Gram-negative bacteria, Handfield has found some plasmids
that resemble those in bacterial strains alive today. He has begun sequencing
their DNA, looking for genes that encode beta-lactamases, enzymes that inactivate
antimicrobial agents and are responsible for antibiotic resistance to penicillin.

Although viable microorganisms have been retrieved in the past from
Antarctic and Greenland ice cores, their DNA has not been analysed. The
Agassiz core presents an opportunity to view 100 000 years of bacterial
evolution in a core less than 150 metres deep. While this time span represents
just 1500 generations of human evolution, it represents up to 2.6 billion
generations of bacterial evolution.

Bill Evans, a researcher at the Laboratory of Ornithology at Cornell
University in Ithaca, New York, has been trying to solve this problem. He
has spent the past decade eavesdropping on songbirds that migrate at night,
listening in as they fly from their northern breeding grounds to the tropics
where they spend the winter.

With the aid of sensitive microphones and specially designed computer
software, he has distinguished the characteristic night calls of more than
sixty species. Armed with this information, he and his colleagues plan a
census that will automatically record the movements of most of America’s
migrating songbirds, and show the extent to which these species are declining.

Evans began by taping the calls of birds migrating over Florida, Texas
and Alabama. To match sound with species, he compared the night calls with
known flight paths of particular species, and in many cases with known daytime
calls. Eventually, he was able to assign notes as short as a twentieth of
a second to individual species.

‘During the course of one spring night in Florida, you can hear a call
every second all night long,’ says Evans. He records these calls and feeds
them into a computer programmed to extract bird calls from background noise.
The computer generates a spectrograph for each call (see Graphs), showing
its nuances of frequency and duration. By comparing these ‘call signatures’
with those made by various species, Evans can identify which ones are passing
overhead. He can also count how many birds of each species have flown by.

As a first step in the census, Evans set up an array of seven listening
posts on rooftops in New York state, mainly in the countryside. Each station
can pick up sounds from birds as high as 1000 metres over an area of 7 square
kilometres.

Three seasons of monitoring have revealed that songbirds migrate in
broad fronts up to several hundred kilometres wide. Each front is made up
of many species, but at a particular place and time – depending on the weather
– the density of each species is remarkably consistent. The weather is
important, says Evans, because on nights with low cloud, when the stars
are blocked, the birds pack together and call more frequently in order
to stay in touch.

Evans suggests that because the patterns are so consistent, monitoring
nocturnal flight calls offers a good way to check on how populations change
over time. For example, in each of the past three years, Evans estimates
that about 100 000 Swainson’s thrushes passed over his line of monitoring
stations, which stretches about halfway across New York state.

Ornithologists predict that in the next quarter of a century, the number
of these thrushes will fall by between 25 and 50 per cent because of deforestation
in its wintering grounds in Central America. The sound census should register
any change in the size of the population much sooner than any other census
technique.

These other census methods rely on volunteers counting birds on land.
The breeding grounds in the boreal forests and wintering grounds in the
tropics are generally inaccessible, so census takers estimate the numbers
en route, counting the number of night-migrating species at rest during
the day. Sam Drogy, an ornithologist with the National Biological Survey
in Washington DC says this can be misleading because the roosting birds
may be the ‘losers’ that are unable to keep up with the flock.

But Evans’s approach has its own drawbacks, says Drogy. For one thing,
identifying birds from their spectrographs is time-consuming. Researchers
at the Bioacoustics Research Program at Cornell have recognised this and
are developing software that will count the number of birds of each species
automatically, eliminating the need to check every spectrograph manually.

Evans’s technique can only count birds that fly at 1000 metres or less;
some travel at five times that altitude, and so are missed. The technique
also assumes that a constant proportion of each species calls as they fly.
‘This is unknown, and may be unknowable,’ says Drogy. Evans admits that
he does not know what proportion of birds call as they fly, or if some call
too quietly to be picked up on the microphones.

To fill this gap in understanding, Evans plans to team up with the electricity
company Niagara Mohawk Power, which supplies power throughout New York state.
The company wants to know where it can safely put up new power lines, and
will track the next migration through the state using radar. Although radar
cannot distinguish between species, it can give a picture of the mass of
birds, says Evans. By comparing the radar data with the number of calling
birds, he hopes to estimate how many are callers.

Evans’s first spectrographs of nocturnal flight calls have just appeared
in The Wilson Bulletin, an American ornithological journal. From the cacophony
of a Florida night, the computer picked out a migrating Bicknell’s thrush,
a bird that is now in decline. Previously, the species had flown through
Florida incognito, confused with the common grey-cheeked thrush.

The incident was clearly a breach of conduct and of faith, but it quickly
acquired a political dimension that distorted its significance. Suddenly,
Democratic Congressman John Dingell had an issue tailored to enhance his
reputation for exposing scientific misconduct, while simultaneously allowing
him to appear as a champion of women’s health. A Congressional hearing called
by Dingell raised the decibel level and added to the confusion. In the ensuing
political frenzy, one of the world’s leading clinical trials in breast cancer
research was sent crashing to the ground.

But was this debacle necessary? Did the misconduct merit the sweeping
overhaul of clinical research that is now under way in the US? Will the
increased scrutiny that is planned help or hinder research? And could the
repercussions spread beyond the US to affect similar studies in Europe?

Breach of trust

The incident of fraud has been recounted many times (This Week, 2 April).
It was initially discovered in a pivotal study of breast cancer treatment,
one which established that lumpectomy – the surgical removal of the cancerous
lump – is as effective as the removal of the whole breast in treating certain
kinds of tumour. About 2100 women took part in this study; 354 of them were
enrolled by Roger Poisson, a physician at the St Luc Hospital in Montreal.
Of these 354, Poisson had altered data on six to make them appear eligible
for the trial. He also altered information regarding another 93 patients
taking part in 14 different clinical trials of breast cancer treatments.

The studies were part of the National Surgical Adjuvant Breast and Bowel
Project, a giant undertaking that involves 484 medical centres throughout
North America. The NSABP is one of nine cooperative groups of researchers
in the US investigating cancer. All of them are funded by the government’s
National Institutes of Health (NIH) in Bethesda, Maryland.

The falsification of data has been condemned as unethical by other doctors
in the study. ‘It was a violation of trust that all of us put in each other,’
says Daniel Hayes, an NSABP oncologist at the Dana-Farber Cancer Institute
in New York. Most scientists also believe that such fraud in clinical research
is very rare.

In fact, the fraud did not alter the overall conclusions of the trial.
After the falsifications had been publicly disclosed, the NIH commissioned
an independent reanalysis of the affected studies excluding Poisson’s data.
It concluded that the original findings remain solid. Because the study
was so big, the removal of the falsified data did not affect its overall
power to reach its conclusions, says Gregory Curt, clinical director of
the NIH’s National Cancer Institute. The NSABP’s findings are also backed
by other trials.

Public alarm

But the episode nevertheless raised widespread public alarm. It led
critics to ask whether undetected fraud or even simple error in large clinical
trials could bias results, leading to the adoption of harmful or ineffective
treatments. Susan Ellenberg, former president of the Society for Clinical
Research, says: ‘The public has gotten the impression that they are in grave
danger of being totally misled by medical research and that fraud is rife
in clinical trials.’

Jill Lea Sigal, a 32-year-old woman with breast cancer, told Dingell’s
committee that there was ‘a crisis of confidence and credibility’ over the
work of the NCI. ‘Think about the agony of uncertainty that I, and thousands
of others, are enduring.’

All this has profoundly affected the NIH, not least because Dingell
chairs the committee that authorises its funding. It has also forced the
resignation of Bernard Fisher, the scientist from the University of Pittsburgh
who had headed the NSABP since its birth.

After a hearing convened by Dingell on 13 April the NIH jumped to attention,
initiating a massive overhaul of its systems for monitoring large-scale
clinical trials conducted at a number of dispersed centres. Most controversial
of its actions is the decision to organise surprise audits by NCI officials
– dubbed ‘data police’ by their critics – at participating hospitals.

The changes are aimed at ensuring the good quality of data and the safety
of trials, but they have alarmed some scientists on both sides of the Atlantic.
Gordon McVie, the new president of the European Organization for the Treatment
of Cancer (EORTC) says the NIH is going overboard. He reviewed an early
draft of its new guidelines. ‘It will kill clinical research if carried
out that way,’ he says. ‘They accuse, by inference, all clinicians of being
thieves and bandits.’

Others believe that the affair has been blown out of all proportion.
They say that Poisson’s falsifications of data, however reprehensible, could
not have damaged the trials themselves, for statistical and technical reasons.
‘The errors were of a type that could not in principle have produced any
bias,’ says Richard Peto, director of the Clinical Trials Services Unit
in Oxford, one of the largest coordinating bodies for trials worldwide.
‘There was never any public health issue involved.’

The aim of a clinical trial is to find out whether a new treatment is
better than an existing one. Patients are entered into a trial only if,
in their doctor’s judgment and based on the best medical knowledge available,
it is impossible to be certain which treatment would be best for them. In
a process called randomisation, each patient is then assigned arbitrarily
to one of the treatment groups. Chance alone determines which treatment
the patient receives; neither doctor nor patient have any say in this decision.
Randomising ensures that the differences in outcome between the two groups
reflect the effect of the treatments, not of other hidden factors.

So a clinical trial is a comparison between groups of people randomly
assigned to different treatments. Poisson’s falsifications would not have
affected that comparison, says Peto, because they were made before the patients
were randomised. On a small scale, women whose data had been falsified at
this stage would not affect the outcome because they would have been randomised
and evenly distributed between the groups. By contrast, changes made after
the patients had been randomised – such as switching large numbers from
one group to another – could affect the results, says Peto.

The Poisson affair may not have directly affected public health, but
it has nevertheless raised wider questions about big trials conducted in
a large number of dispersed centres. The advantage of such giant trials
is their statistical power: the more patients involved, the more powerful
the result. Rory Collins, co-director with Peto of the Clinical Trials Service
Unit in Oxford, says large trials can detect small but important differences
in survival between groups of patients, which can give doctors important
information about the best treatment to use and ultimately benefit tens
of thousands.

But the drawback of such trials, say critics, is that some of the participating
doctors may lack sophistication and training in scientific method. As a
result, they may make innocent errors. Earlier this month, the Society for
Clinical Trials set up a committee to educate doctors about of trials, says
Sylvan Green, the society’s new president.

Different rules

While all large-scale trials face some of these problems, the NSABP
is unique, and the NIH’s assessment of what went wrong in the Poisson case
is based on that unique character. The NSABP is perhaps the most important
cooperative trials organisation. Ultimately, it not only demonstrated that
lumpectomies are safe and effective, but also pioneered the use of chemotherapies
and hormonal therapies that delay the recurrence of tumours and improve
survival in women with breast cancer.

The NSABP’s strength, say scientists, has been its ability to involve
large numbers of busy surgeons in trials. But until the overhaul, its rules
differed from the other cooperative groups, says Curt. There was no independent
review of its records, and it was difficult to track mistakes. The project’s
method of payment was also unique in that the money given to an institution
was based on the number of patients enrolled. Curt warns that this could
have encouraged doctors to falsify data to make more people eligible for
trials. By 1993, the NSABP had even stopped following some of its own rules,
suspending all treatment trial audits. Fisher says a rapid expansion in
the number of patients involved, combined with a lack of resources, was
to blame.

The NIH, however, failed to detect this lapse in auditing. And even
after receiving reports of fraud, it let the situation drift for three
years. Now it has taken action. It has doubled its budget for auditing
to $3.3 million over the next year, forced Fisher out, stopped recruitment
to the NSABP until the group is reorganised, and taken possession of computer
data files. And it is to implement visits to sites by monitoring officials
that are to include ‘an element of unpredictability or surprise’.

But critics fear that over-regulation will drive doctors away, leading
ultimately to smaller trials. Edward Lee, an oncologist at the University
of Maryland Cancer Center, described a recent audit by the NCI at his centre.
It involved researchers pulling 100 patients’ charts out of the files with
just three days’ notice. A ‘chart’, however, can run to eight volumes and
several thousand pages. To prepare for an audit, someone must go through
each chart to highlight information. With the increased emphasis on auditing,
‘the economics just got an order of magnitude more expensive’, says Lee.

Researchers in Europe are equally concerned. Collins in Oxford fears
the American reforms will ripple across from the US, contributing to a number
of changes that will make trials more expensive, cum-bersome and smaller.

It is likely that the NIH’s continued scrutiny of the NSABP’s records
will unveil more problems. Auditors are now focusing on patient records
at the Memorial Cancer Research Foundation in Los Angeles, where some patients
were apparently enrolled in NSABP trials without their proper consent.

As further irregularities come to light, it is important to define
the nature and extent of the problems. In a political climate clouded with
suspicion of science, medicine and government, there is a danger that the
NIH, Congress and the public will react to each case without considering
its broader context.

That context remains the question of how best to hasten the cure and
prevention of all cancer. More than a million people will be diagnosed with
cancer this year in the US: less than 3 per cent of them will enter clinical
trials.

‘The alternative to randomising and getting some knowledge is treating
patients outside a trial, haphazardly, and getting no knowledge,’ says Collins.
If large-scale clinical trials prove politically impossible, that would
be a tragedy.

Bernice Wuethrich is a science writer based in Washington DC.

Most archaeological evidence suggests that wild oxen were first domesticated
in southwest Turkey between 8000 and 10 000 years ago. When these long-horned,
flat-backed cattle joined domestic sheep and goats, they provided not only
food but a source of power for haulage, ploughing and transport – accelerating
the shift from hunting and gathering to farming and herding.

According to the interpretation based primarily on archaeological evidence,
the first domestic cattle moved with their owners east towards India, and
southwest into Africa in waves of migration and invasion. After several
millennia, farmers bred a humped-back variety, adapted to survive in dry
climates.

But the archaeological record is patchy, and it is often difficult to
tell the difference between bones from wild and domestic cattle, or humped
and non-humped breeds. So students of the problem turned to mitochondrial
DNA (mtDNA), a form of DNA passed only down the maternal line, for independent
information about domestication.

Daniel Bradley and his colleagues at Trinity College, Dublin, compared
the mtDNA of cattle from six European, four African and three Indian breeds
(Proceedings of the National Academy of Sciences, vol 91). They expected
to find two distinct types of mtDNA, one from the humped or zebu cattle,
and one from the non-humped or taurine cattle. Instead, they found that
the mtDNA analysis threw up a geographical distinction between animals from
India on the one side, and those from Europe and Africa on the other. The
African breeds, however, included both zebu and taurine cattle.

Surprisingly, the difference between the two groups was so large that
it would have taken at least 200 000 years for the mutations to accumulate,
says Bradley. The two types could not possibly have descended from the same
common ancestor just 10 000 years ago. The most logical explanation for
the deep genetic difference is that two separate domestications occurred,
from different races of wild ox, Bradley contends. Domestication in southwest
Turkey led to the non-humped taurine, while a second domestication, at about
the same time and possibly in the Indus valley, led to the humped zebu cattle.

At first glance, however, Africa appears to complicate this tidy picture.
The study included three African zebu breeds and one African taurine breed.
Yet there was no distinction between the mtDNA of African zebu and taurine
types: all had mtDNA similar to that of European taurine cattle.

Bradley argues that Africa’s history of invasions and the effects of
the continent’s famines and disease on animal husbandry can explain the
absence of zebu mtDNA in the African humped animals. Arab invaders who crossed
the Red Sea around AD 670 brought with them a few choice humped cattle.
But in all likelihood, the invaders imported only select bulls to crossbreed
with existing African taurine cows. So the flow of zebu genes into the continent
was a male mediated process – dissemination through insemination, says Bradley.

Both males and females have mitochondrial DNA, but as its name suggests,
it resides not in cells’ nuclei but in energy-producing organelles called
mitochondria, which are in the cytoplasm. While mtDNA occurs in the cytoplasm
of eggs, it is not found in sperm, which contain only nuclear DNA. So while
the zebu bulls’ nuclear DNA was passed on to their African sons and daughters,
and had a big impact on their appearance and physiology, the ancestral zebu
bulls’ mtDNA was never passed on to African cattle. If there were any zebu
cows introduced to Africa, past famines and epidemics that have culled African
stocks may have wiped out mtDNA variability and eliminated any remnants
of exotic zebu mtDNA.

The study augments recent archaeological evidence suggesting that India
was a centre for domestication, and it leaves open the possibility that
ancient Africans tamed wild oxen before the first imports arrived. The genetic
similarity between European and African breeds could be explained if wild
progenitors in both regions were closely related, says Bradley.

Revelations that a doctor in Canada falsified data about patients taking part in a huge trial of breast cancer treatments have prompted debate about how widespread such cases might be, and whether it is possible to ‘police’ studies involving thousands of researchers and tens of thousands of patients.

Last month, the US National Institutes of Health in Bethesda, Maryland, announced that Roger Poisson, a doctor at the St Luc Hospital in Montreal, had faked data in a series of studies. The results of one of the studies, published in 1989, showed that removing a cancerous lump from the breast was as effective as removing the whole breast. It led to an almost immediate change in medical practice.

Poisson’s patients made up 16 per cent of the group on which this particular study focused, and there were initially fears that his false data had invalidated the results. After reanalysing the results without Poisson’s data, the NIH says that the results are still valid.

The breast cancer studies were part of the National Surgical Adjuvant Breast and Bowel Project (NSABP), which involves 5000 doctors at 484 institutions in North America. The programme has treated some 60 000 patients since 1958. The studies were designed to identify the best treatments for a given group of patients – menopausal women suffering from early breast cancer, for example.

Poisson joined the NSABP study in 1975. According to investigators from the Department of Health and Human Services’ Office of Research Integrity (ORI), he began to falsify data no later than 1977. The deception went undetected until 1990. Since then, the NSABP, the National Cancer Institute, the Food and Drug Administration and the ORI have reviewed Poisson’s files.

They found that he had altered the records of at least 99 of the 1511 patients he had enrolled in 14 of the studies. In most cases he made it appear as if patients fulfilled certain eligibility criteria when they did not. In one case he disguised the fact that a patient had cancer in both breasts, rather than just one, and then withheld radiation treatment from one breast. In another, he concealed a patient’s history of congestive heart failure and then entered her into a study that used a drug that is potentially toxic to the heart. He also entered a patient with a relatively advanced cancer into a study designed for patients with earlier stages of cancer.

Poisson routinely altered patients’ ages, the dates of their biopsies and surgery, and data on the size of their tumours and the levels of hormone receptors in the tumour cells. Women who have large numbers of these receptors may respond positively to treatment with the drug tamoxifen. Those with fewer such receptors are likely to respond better to chemotherapy. By altering data on patients’ receptors, Poisson was able to enter dozens of ineligible women into tamoxifen studies.

After the ORI issued a report on its findings late in 1992, Poisson admitted that he had faked his data. In a letter to the ORI, he said that the alterations were ‘white lies’, and that the criteria on which women were judged eligible for a study were often an obstacle to enrolling patients. ‘I believed I understood the reasons behind the rules and I felt that the rules per se were made to be understood and not necessarily blindly followed,’ he said. ‘It always distressed me to see a good and willing patient not being registered just because a small technical detail was missing.’

Investigators concluded that Poisson had shown a ‘disrespect for the truth’ and ‘deluded himself that he was complying with the ‘spirit’ of the protocol’. He was barred from receiving federal research funds, and forced to withdraw from the breast cancer studies.

‘Fudging on eligibility criteria is the worst thing to do,’ says Jeffrey Abrams, a cancer specialist at the NCI. If enough patients are wrongly enrolled, a study can be invalidated. Patients might also be put at risk by investigators ‘so eager to enrol them they may not think enough about their rights and welfare’, says Judith Swazey, an expert on medical ethics at the Acadia Institute in Bar Harbor, Maine.

There are several reasons why a researcher might fake this sort of data. A certain amount of prestige is attached to working as an investigator in a large and important study, and doctors taking part in trials are listed as contributors when the results are published, which can help their career or be useful when applying for funds.

Researchers who falsify their data may believe that what they are doing is best for their patients or for medical progress. Those based in small community hospitals may be poorly trained in research. And there is no formal procedure for accrediting institutions that carry out studies with human subjects – although there is for institutions doing similar work on animals.

However, reports of deliberate deceit in clinical studies are rare, says Lyle Bivens, director of the ORI. He recalls only one similar case.

Swazey notes that there is a distinction between deliberate deception and sloppy research or unintentional error. But it is difficult to assess the frequency of either, she says, because there are no accurate data on them. ‘The reality is that fraud is not an epidemic problem, but there are instances that never get reported or investigated,’ she adds.

The task of investigating any improprieties in clinical trials falls largely to two organisations: the NIH’s Office for the Protection from Research Risks (OPRR) and the ORI. The OPRR investigates complaints that concern the rights of patients enrolled in research programmes. These include the right to know the potential benefits and risks of treatments and alternative therapies. The OPRR has only two full-time staff monitoring projects at a thousand institutions, so it relies heavily on review boards at each institution to monitor their own projects.

The OPRR has not investigated the fraud committed by Poisson and its possible impact on his patients. Its director, Gary Ellis, says that the case falls within the remit of the ORI, which monitors scientific misconduct, including the falsification of data. The Poisson case is one of around eighty being investigated by the ORI. Most of these concern basic research rather than clinical studies. At the ORI, 13 investigators follow up about five hundred complaints filed each year.

Since the discovery of Poisson’s fraud, the ORI has reanalysed the results of the breast cancer studies and found that the conclusions still hold, although they are statistically less powerful. The results of additional checks by three statisticians will be published shortly. But researchers and investigators are concerned that the discovery of just one case of fraud might undermine the trust between doctors and patients, and fuel suspicion of clinical research.

‘If I were a woman with breast cancer, I’d feel a lot more comfortable if some random audits of the NSABP’s raw data were done,’ says Swazey. There are no plans to do this.

The finding reopens the debate on what level of risk assessment is needed
before introducing transgenic plants into commercial production. Since 1987,
some 125 field tests have been carried out on transgenic plants with viral
genes, and the US Department of Agriculture is now considering whether it
should permit the marketing of the first such plant, a transgenic squash
developed by Upjohn. The company had not specifically looked for viral recombination
in its squash plants.

Worldwide, viruses are second only to fungi as plant pathogens; the
resulting crop losses amount to millions of dollars. Transgenic plants are
designed both to protect against those losses and as a means of reducing
the amount of pesticides sprayed to kill the insects that carry viruses.

Allison and Greene genetically engineered plants of Nicotiana benthamiana,
a distant relative of tobacco, so that each of their cells expressed a segment
of RNA from the cowpea chlorotic mottle virus (CCMV). This segment of RNA
consisted of a portion of the virus’s capsid gene, which forms the viral
coat protein and enables the virus to move from cell to cell.

The researchers then infected the transgenic plants with the mottle
virus from which they had deleted the portion of the capsid gene that was
now present in the plant cells. This ensured that the virus could spread
throughout the plant only by recombining with the viral RNA in the plant
cells.

Of the 125 plants in the study 4 became infected, indicating that the
disabled virus and the viral fragment had combined to form a complete virus.
Allison then isolated the virus from the infected plants, sequenced its
RNA – and found that in the course of recombination subtle variations had
occurred, producing four different viral variants. The researchers describe
their results in the journal Science (vol 263, p 1423).

Subtle changes in capsid RNA influence other factors important to the
spread of disease: they determine what type of plants the virus can infect,
which insects can carry the virus, and the severity of the disease. So in
the course of recombination all of these qualities could be altered – perhaps
allowing an whitefly to carry a virus normally transported by aphids, or
allowing a melon to be infected by a virus that normally afflicts corn.

However, some scientists say the findings do not significantly alter
what was previously known about the risks posed by transgenic plants. ‘The
study shows that if a plant supplies a necessary gene for a virus . . .
the virus can take advantage of the situation. I don’t think it shows that
transgenic plants will present a risk in the environment,’ says Bryce Falk,
a plant pathologist at the University of California at Davis. Falk also
points out that viral recombination occurs spontaneously in nature, regardless
of human engineering.

Other scientists say the study demonstrates the need for better assessment
of the risks before transgenic plants are grown commercially. The Union
of Concerned ��è����Ƶs has called for a moratorium on the commercialisation
of such crops until such time as the federal government introduces a stronger
risk assessment programme.