David Adamson believes that assisted reproduction is safe, but wants to see more research. Credit: M. HARTFORD

Kelly and Mike were desperate to have a child, and endured nine rounds of traditional in vitro fertilization (IVF) treatment at clinics in their native Australia. Over the course of two years, each attempt ended in failure and heartbreak. Then they were given the chance to participate in research to test a new reproductive technology in the United States. The method, which involves growing IVF embryos in culture for five days to increase the chance that they will implant in the womb, had proved safe and effective in mice and cows. But, as Kelly and Mike were informed, its safety in humans couldn't be guaranteed.

It was a chance they were willing to take. Nine months after their trip to the Colorado Center for Reproductive Medicine in Denver, Kelly gave birth to twins — a girl and boy — who both seem perfectly healthy. Their smiling family portrait is one of scores decorating the walls of the Denver clinic.

Although Kelly and Mike were fortunate to attend an academic centre that makes research into the safety of IVF a top priority, other couples are served by commercial clinics where the bottom line — abetted by a genuine desire on the doctors' part to help patients experience the joys of parenthood — may mean that the potential risks aren't so carefully explained. “There's a commercial aspect on one hand and couples' desperation on the other,” says Robert Winston, a specialist in reproductive medicine at Imperial College, London, who last year wrote a review of the safety of assisted reproduction1. “That's quite a risky combination and not a very good way of conducting responsible medicine.”

Over the past year or so, evidence has begun to emerge that suggests assisted reproduction is associated with increased risks of birth defects, low birth weight, genetic disorders, and maybe even cancer. Although the findings obtained to date are far from conclusive, there is a growing feeling that the regulation of fertility clinics may need to be tightened up, and a more rigorous system for monitoring the health of IVF babies put in place. “There is such a dearth of information,” says Laura Schieve, an epidemiologist at the US Centers for Disease Control and Prevention in Atlanta, Georgia.

A helping hand

Since the first test-tube baby, Louise Brown, was born in Britain in 1978, more than a million children have been brought into the world through assisted reproductive technology. The original IVF technique, which involved mixing eggs and sperm in the lab and then implanting the resulting embryos into the womb, was developed to help women with blocked fallopian tubes. Early follow-up studies of IVF children, conducted in several countries, suggested that the technique was safe2,3,4,5.

Emboldened by this success, reproductive specialists turned to progressively more interventionist techniques to treat infertility. Women have been prescribed drugs to boost ovulation, and eggs and sperm have been manipulated in various ways to ensure that fertilization occurs. One technique in particular — known as intracytoplasmic sperm injection, or ICSI — was quickly taken up because it allowed men who can't produce healthy, mobile sperm to become fathers. Many reproductive techniques were first perfected in animals, but ICSI does not work in cattle or mice. It was discovered in 1991 in a Belgian IVF lab, when a technician tried to place a sperm cell just under the egg's protective coat and accidentally injected it all the way into the egg6. Today, ICSI accounts for nearly half of all assisted reproductive treatments in the United States.

For some observers, the rapid adoption of such an aggressive technique raised nagging concerns. Might the injection damage the egg's machinery for cell division, for instance? And is it wise to bypass natural selection in such a determined way? At the very least, the technique will pass infertility problems down to the next generation. It is also possible that the defects that cause male infertility are just part of a wider spectrum of genetic abnormalities carried by some faulty sperm.

Culturing embryos such as this for five days boosts the chances of a successful pregnancy.

“This was a no-holds-barred, 'you-will-fertilize' kind of thing,” says Jennifer Kurinczuk, an epidemiologist at the University of Leicester, UK, one of several researchers who was prompted by the widespread uptake of ICSI to take another look at the safety of assisted reproduction. Kurinczuk and her colleagues used data from the Western Australia Reproductive Technology Registry to look at the risk of birth defects diagnosed by the age of one. In a study published in March 2002, they found that the incidence increased from about 4% in a control group of naturally conceived infants to about 9% in children conceived either by conventional IVF or by ICSI. Even after adjusting for factors including the mother's age and prior births, the risk of birth defects seemed to be doubled for both assisted-reproduction groups7.

In a study published back-to-back in the same journal, Schieve and her colleagues mined US data and found that infants conceived by assisted reproduction were two-and-a-half times as likely as the general population to have a low birth weight — and this wasn't simply because there were more multiple births among the assisted-reproduction group8. Given that babies with low birth weights are more likely to die in their first 12 months, are more susceptible to infections, and often have respiratory problems, this was clearly a worrying finding.

Defect dispute

These two papers caused a flutter of alarm, but attracted criticism from practitioners of assisted reproduction. The studies, IVF clinicians said, had design flaws, including their retrospective nature. Gleaning data from patients' records can allow biases to creep into the sampling — a prospective study that follows outcomes for a group of patients is regarded as more objective. Kurinczuk's report, in particular, was questioned because her figure for birth defects among the naturally conceived population was about twice that generally accepted as the norm. David Adamson, director of Fertility Physicians of Northern California, a clinic in Palo Alto, finds the results for assisted reproduction hard to believe. “Almost one in ten babies with a defect?” he asks. “It's just not true.”

The disagreement stems in part from differences in the definition of what constitutes a birth defect. Kurinczuk, for instance, uses criteria borrowed from the British Paediatric Association that include problems such as an undescended testicle. Two ongoing prospective studies, one in Belgium9, the other across six European countries, are monitoring ICSI children and have found no evidence of any worrying trends. But when Kurinczuk reanalysed the Belgian data using her criteria, she found more than twice as many birth defects as had previously been reported10.

While the arguments about birth defects rumble on, the debate over the safety of assisted reproduction has now spread to rare disorders affecting genetic imprinting — the process that, early in development, can silence either the copy of a gene inherited from your mother, or that from your father.

Andrew Feinberg of Johns Hopkins University in Baltimore, Maryland, and Michael DeBaun of Washington University in St Louis, Missouri, study children with Beckwith–Wiedemann syndrome, which occurs in 1 in 15,000 births, and results from faulty imprinting. Children with this syndrome are larger at birth than normal, have enlarged organs, and are at increased risk of developing certain cancers. Feinberg and DeBaun were struck by the fact that 4.6% of their registered patients were conceived by assisted reproduction, a high proportion given that less than 1% of the general population is conceived artificially11. DeBaun believes that the similarities to 'large offspring syndrome', sometimes seen in sheep and cattle produced by assisted reproduction12, are too obvious to overlook. In fact, one of the genes implicated in the oversized livestock, IGF2, is also affected by faulty imprinting in Beckwith–Wiedemann syndrome.

An independent British study has since confirmed the association between Beckwith–Wiedemann syndrome and assisted reproduction13, and three cases of an even rarer imprinting disorder, called Angelman syndrome, have been discovered among children conceived by ICSI14,15. Most recently, five Dutch children conceived by assisted reproduction have been diagnosed with retinoblastoma, a childhood cancer of the retina that occurs in just 1 in 17,000 births16. Given this incidence, one would expect to find only one child with the condition in the Netherlands who was conceived artificially.

Catalyst for action

These studies into imprinting disorders and one rare form of cancer are very small. But the fact that these diseases have known genetic causes has led even those who dismiss the warnings about birth defects to accept that they warrant further investigation. “We'd be foolish to think there are no problems,” says Adamson. “These studies absolutely require that we look into it.”

That caution stems in part from a recent experience in which the US Food and Drug Administration (FDA) was forced put the brakes on a new assisted reproductive technique. The procedure, called cytoplasmic transfer, was used to 'rejuvenate' eggs from older mothers. It involved injecting into the eggs a tiny amount of material, including the mitochondria that 'burn' sugars to create cellular energy, from an egg donated by a younger woman. The hope was that the younger woman's more vigorous mitochondria would supplement the rapidly developing embryo's energy needs. Although it seemed to help older women to get pregnant using their own eggs17,18, the procedure also dramatically increased the risk of Turner's syndrome, an X-chromosome defect that can cause miscarriage in early pregnancy. That was enough for the FDA to step in, and in July 2001 it ruled that anyone wanting to pursue the technique would have to seek permission to perform a full clinical trial as if they were investigating a new drug.

Some observers believe that this ought to become standard practice for any new technique of assisted reproduction. And against this background of heightened concern, even techniques that were introduced precisely to improve the safety of IVF are coming under closer scrutiny. The embryo-culturing method that allowed Kelly and Mike to achieve their dream of starting a family is a case in point. It was developed in the hope of ending the practice of transferring several IVF embryos into the womb to maximize the chance that one of them will implant. Transferring several embryos greatly increases the likelihood of multiple pregnancies, which can be extremely dangerous for both mother and fetuses. For this reason, Britain's Human Fertilisation and Embryology Authority (HFEA) allows clinics to transfer only two embryos at a time under normal circumstances. But in some countries, such as Russia, Greece and Spain, four or more embryos are still transferred in many cases.

Cultured approach

At the Colorado Center for Reproductive Medicine, David Gardner and Michelle Lane have shown that it is possible to obtain much higher success rates for implantation if embryos are cultured in the lab for five days instead of the conventional two or three, using a succession of culture media to mimic the changing conditions experienced by a newly fertilized embryo as it drifts towards the uterus19. Encouraged by these results, about one in five US fertility clinics has already adopted the technique, with some using it exclusively. But although Gardner and Lane's goal is to make IVF an efficient technique even when only one embryo is transferred, thus completely eliminating the risk of multiple pregnancies, for the time being the culturing technique is still being used in conjunction with the transfer of two or more embryos.

Some researchers fear that increasing the duration of culture, and exposing embryos to different culture media, might endanger early development. Experiments in mice, for instance, have shown that even subtle changes in the ingredients of culture media can dramatically alter the activity of imprinted genes20. “We are using humans as guinea pigs,” suggests Kelle Moley, who studies pre-implantation mouse embryos at Washington University in St Louis. Because the formulas for human culture media are proprietary and funding for human research is scarce, she warns, studying the effects of culture conditions on the health of IVF babies is nearly impossible. “There is a small number of companies making culture media and their compositions change sporadically,” Moley says.

Gardner and Lane counter that their animal experiments, which encompass more than 6,000 embryos, have yielded no evidence of any problems21. They have also looked at the birth weights of babies who were cultured as pre-implantation embryos for five days, rather than three, and found no differences that would suggest the technique is causing a human equivalent to large offspring syndrome22.

Uncertain future

Despite the recent upsurge of concern, IVF practitioners stress that most babies conceived by assisted reproduction seem to be perfectly healthy — and they argue that all the evidence suggests that infertile couples are willing to accept small risks in order to have their own children. But with Louise Brown having yet to reach her 25th birthday, other experts warn against complacency. It may be extremely unlikely, says Alastair Sutcliffe, a paediatrician at London's Royal Free Hospital, but there is still the possibility that IVF could cause subtle genetic or cellular changes that might manifest as cancer or some other disease only later in life. “It could be a disaster waiting to occur,” he says.

Given the existing worries, and the potential for further unpleasant surprises, some researchers are calling for increased funding for projects to investigate the basic biology of fertilization and implantation, investigations of the effects on eggs and embryos of being manipulated in the lab, and bigger and better epidemiological studies. Done properly, the latter will require the prospective follow-up of many thousands of children, collecting standardized data on parental age and medical history, type of infertility, and myriad details about the specific form of assisted reproduction used. Ideally, the children involved would be monitored at regular intervals throughout their lives.

Whether the money to conduct such mammoth studies will be forthcoming remains unclear. But those charged with regulating the field argue that our current state of relative ignorance shouldn't be allowed to persist. “For the benefit of future patients and offspring, it is hugely important to improve the evidence base,” says Suzi Leather, who chairs Britain's HFEA.