The urgent debate in universities and pharmaceutical companies alike is about how to get the best out of the knowledge we’ve accumulated, how to translate revelations in our understanding of genetics and disease into actual medical benefit. Many of our best medicines so far have been vaccinations, but development of a vaccine for HIV, for example, has proved to be a long and bumpy ride ever since the US health secretary suggested in 1984 that it would take a couple more years. The discovery of HIV Controllers is encouraging for vaccine development, because these people show us that immune responses at least have the potential to control HIV in the right circumstances. If we could get other HLA types to be as potent as, say, HLA-B*57, then more people might join those who inherited superpowers.
There’s no shortage of scientists trying to translate our knowledge into practical outcomes; conferences about HIV nowadays gather about 20,000 professionals and 2,000 journalists. Such meetings are not unlike Star Trek conventions; the passion is the same and the heroes are equally revered. Both are ignited by imagination and wonder. But the paramount fact distinguishing scientists from their science-fiction counterparts is that they are also driven by an important real cause: to create new medicines.
Some – Nobel laureate Rolf Zinkernagel is one – think the key issue in getting to new medicines is to perform experiments in which everything is as close as possible to being physiologically right: using animals, real viruses and doses that would occur naturally.1 Others – such as Ron Germain, a leading scientist at the NIH – agrees that this is important but also advocates other approaches, such as computer simulations of immune responses.2 The difficulty is that it’s relatively easy to do something new; very hard to do something important, because, as Einstein put it, ‘Not everything that can be counted counts.’ My view is that, since the very essence of discovery is that nobody predicted it, who’s to know what’s best to do next?
In truth, many of the drugs that we use today were found by chance – or at least serendipitously. The discovery of antibiotics is a well-known example: on 28 September 1928, Alexander Fleming noticed that a fungus had contaminated one of his experiments and killed off the bacteria he was studying. A more recent example is Viagra, developed as a drug for high blood pressure and then later discovered to be of use in preventing erectile dysfunction. It has proved difficult, and all too rare, to systematically translate our knowledge into direct medical benefit. Sport commentators call something ‘academic’ when it’s not important, but this issue is anything but ‘academic’; our well-being and even our survival depend on us choosing the right path to new ways of conquering disease. So are there radically different approaches we could take?
Eric Schadt is one leading scientist who says there is. Renowned for turning up to meetings in shorts whatever the weather or formality of the occasion, he argues that molecular biology has been great for uncovering individual genes important for human traits but that we’ve largely failed to fulfil the medical promises that have been touted – and it’s enough already. He argues that the main problem is that we haven’t adequately tackled the complexity of genes and disease.
In general, many genes – not one – contribute to a disease risk or human trait. As we’ve seen, compatibility genes influence our susceptibility and resistance to all manner of diseases, but they don’t fully protect against, or absolutely cause, any one. There are exceptions such as Huntington’s disease, which is caused by a single genetic variant, but by and large things are more complicated than one gene causing one disease or trait. Studies of how frequently twins share an illness are one way in which we can estimate the total effect that genes have. And in comparison to the sum of the individual genes known to be important, we can account for only about 10 per cent of the total genetic risk for many human traits and diseases.3 So, Schadt says, something big is missing in our understanding of genes and disease.
Genome-wide associations studies have worked well in identifying many important genes. But even these huge studies – scanning the genes of thousands of people – are not perfect. Things that aren’t easily picked up include rare genetic variants and modifications to our DNA made after birth (so-called epigenetic changes) and differences we can sometimes have in the number of copies of a given gene. But most important of all is that genes interact – the status of one influences another, and so on – like computer, social and financial networks. And variation in groups of genes is hard to analyse – only the effect of individual genes is easily studied.
So, Schadt, and others like him, suggests that a seismic shift in our approach is needed because most diseases involve interactions between constellations of genes. And things are even more complex than you might think at first because the interactions between genes are affected by diet, age, gender, exposure to toxins and so on.4 Schadt’s close colleague Stephen Friend, a paediatric oncologist, says it plainly: ‘Traditional human disease research models are now archaic. The academic [grant] process is choked by favourite gene efforts that result primarily in “impactful” journal articles . . . And the patients? They’re getting more and more frustrated.’5 Yesteryear’s revolution in biology was the Human Genome Project; Schadt wants to deliver the next one.
There’s always a personal story behind the approach a scientist takes. Schadt’s almost anarchic attitude – and his ability to weather a storm – was undoubtedly shaped by a series of fights early in his life.6 His Christian parents brought him up, with his six siblings, with the attitude that secular education was worthless. Going to college was frowned upon, and Schadt joined the air force. But an accident while rappelling down a rockface left him with poor mobility in his shoulder, and he was told his role in the military would have to change. Having done well in various aptitude tests, in 1986, aged nineteen, he went to college after all. Physical exertion had been Schadt’s release, but once he was at college, ideas and academic challenges became a new source of freedom. His religious upbringing helped focus his mind on big issues – how things are connected, underlying principles – and he was drawn to studying maths and philosophical logic. But by pursuing a college education his father considered he had become possessed by the devil; he told his son that he should never return home.7
Estranged from his family, Schadt had to fight to stay away from the military – after realizing that he had, in fact, been overwhelmingly depressed there – because they were paying for his education and expected him to return. Eventually he got a PhD from UCLA in 1999, and by that time he had already begun working at the pharmaceutical giant Roche. It was a time when large-scale analysis of genes was fairly new, and Roche was using a specific process for analysing genetic data which used machines purchased from another company, Affymetrix. Schadt wasn’t happy that Affymetrix wouldn’t let anyone see the computer codes being used to analyse the genetic data. It meant that he couldn’t fiddle with it to tailor it to his specific needs. So he wrote his own software, which won him fame within Roche. But he grew tired of corporate meetings and decided to move to a small start-up company based in Seattle. Then, in November 1999 – a week before Schadt left Roche – things got ugly.
The first trace of trouble came when Schadt tried to remotely access his office computer but couldn’t. He called his wife and asked if she could go to his office and check it. Maybe it had been turned off by mistake? She went in and found everything in his office had disappeared. Someone had taken all his stuff. Worse, someone had told the president of Roche that Schadt had directly based his own-written software on the code from the other company, Affymetrix, which was illegal and potentially a huge problem for Roche. Schadt knew he was innocent, but one of his lawyers told him, quite plainly, that the truth is irrelevant: plenty of innocent people go to jail.
On Christmas Eve 1999, Schadt was shopping for presents with his two young kids when his wife phoned to say someone had just telephoned her and needed to talk urgently. Schadt had never heard of the person so he didn’t call back and continued shopping with his kids. But the mysterious caller phoned again, this time clarifying that he was from the FBI. Schadt panicked. Why was the FBI after him? Or was this someone pretending to be from the FBI? Tired and scared, he even thought that Affymetrix might have hired a hit man to kill him.
When he got home, it was indeed FBI agents who were waiting for him in a black car outside his house. Schadt was accused of taking the allegedly illegal computer code to the new start-up company.8 His life became unbearable for months.9 He saw endless numbers of lawyers. With one, Schadt spent three days going through all the details of the computer codes. The lawyer seemed to take it all in. But at the end, the lawyer politely asked whether algorithms were like logarithms. They’re nothing like the same – though the words do rhyme.
Eventually, Schadt met lawyers who could understand computer code, and ultimately it was concluded that he had written his code independently. He became a hero to the academic community, because he had written the computer code simply to do better science and he took on a huge corporation to do so. And then, out of the blue, Schadt’s parents called and said he was welcome home again. They had been to a Christian conference where one of the speakers struck a chord with them, leading to them to see a counsellor. They now openly celebrated their son’s success.10
Schadt’s battles – with his family, the military, Affymetrix and the US judiciary – prepared him for the real fight of his life: to establish a new approach in medical research. Whether intentionally or not, Schadt’s shorts outfit reminds you of his focus. He’s got no time to waste on piffle like thinking about what to wear. Yet, when Schadt visited me on one occasion in 2008, he e-mailed ahead to check if wearing shorts would be OK, or whether he’d need other clothes – perhaps at the evening dinner. Like many successful revolutionaries, he can play at being conventional when needed – he had trained in the air force, after all.
The problem with translating our knowledge into medicine, Schadt and his colleagues argue, is that our effort has just been too simplistic: we’ve naively focused on finding a causative gene or protein and then a drug to fix it. Perhaps this approach is an inevitable consequence of our brain having evolved to think in terms of one thing leading to the next, or perhaps it’s because ‘one disease, one gene, one cure’ is a straightforward plan readily sold to investors.11 But our pipeline of new drug development is pretty blocked, because, without mastering the complexity of the system, Schadt argues, it’s almost impossible to know what effect a drug will have on something as intricate as the human body.
The side-effects of any new drug, for example, usually only become apparent during a clinical trial because they are so hard to predict in advance. In fact, it’s been estimated that about 90 per cent of drugs fail to reach the marketplace because of unexpected side-effects, a direct consequence of the complexity and inter-connectedness of human biology. The thought that most life processes don’t work in a straightforward linear way drives most scientists to despair. But here’s Schadt’s punchline: breathe it all in, embrace complexity, and let’s just establish a whole new way of doing things.
Of course, Schadt’s advocacy that genes interact in complex ways is not a new idea in itself. In his 1959 BBC Radio lectures, Peter Medawar had said that ‘the forms of heredity that can be seen to obey fairly simple rules are not a representative sample of heredity as a whole’.12 But what Schadt and his like-minded peers are really doing that’s new is to establish a way to tackle the complexity; the time is ripe, they say, to navigate us through the fog of genetic interactions and reach medically useful ideas. Science has reduced humanity to a list of genes and components; now to figure out how those elements give rise to the beast itself.
Schadt, and his kindred spirits, argue that what’s needed is to reconstruct the underlying networks of interactions between genes causal to traits associated with disease. He thinks that to do this we can sequence DNA, check the functions of cells, measure disease markers – such as levels of sugar and insulin in our blood – and in short obtain an enormous set of information for a huge number of people, to work out which set of genes influences each disease or human trait.
The problem is that, even with just ten genes, the number of possible interactions between them is about 1018 (a one followed by eighteen zeroes – or a billion billion). And, of course, we don’t each have ten genes; we have 25,000. The raw data alone from the sequence of genes in 1,000 individuals amounts to over 1012 bytes.13 Clearly, computing must take centre stage for analysing all the information. The wild-eyed anarchist in a lab coat mixing cells and chemicals is a poor reflection of what’s needed for this kind of science. Here, there needs to be multi-disciplinary multi-talented teams that include scientists clicking at computer screens and thinking about abstract algorithms seemingly far removed from the biological processes they’re studying. Astronomy, climate change and particle physics have all embraced computationally intensive science long ago; the approach must now be used for studying human health.14 But before we wholeheartedly nod in agreement, let us recall an allegory with an appropriate cautionary message – a parable from Argentine storyteller Jorge Luis Borges.
There was once an empire where the art of map-making enjoyed the highest reverence. Old maps seemed insufficient, and the Cartographers Guild set out to attain a description of the empire that was truly perfect. But this ultimate ambition – a point-by-point rendition of all the land – only produced a map the same size as the empire itself. The work of the greatest minds culminated in an exact description of the land – which turned out to be entirely useless. The perfect map was left discarded, and subsequent generations gave less importance to the art of cartography. ‘In the Deserts of the West, still today,’ Borges writes, ‘there are Tattered Ruins of that Map, inhabited by Animals and Beggars; in all the Land there is no other Relic of the Disciplines of Geography.’15
Similarly, a complete and exact simulation of the human body in different states of health and disease may not be a useful ambition – it would be as complex and impenetrable as the human body itself. Schadt and his colleagues aren’t concerned, because they aren’t seeking to rebuild a human. Their argument is just that we currently have things mapped out too simply and that figuring out the interactions between genes is a depth of detail necessary for developing new drugs. In 2001, the pharmaceutical giant Merck thought similarly and spent $620 million on buying the start-up where Schadt worked. Schadt’s plan was to combine conventional laboratory experiments with high-level computational analysis. This would have been difficult in an academic environment but in Merck he could use one of the fastest supercomputers available in the drug industry.
Work in Schadt’s biological laboratory determined the extent to which different genes are turned on and off in mice in different states of health and disease. His computer team then used this information to establish which genes are linked together by assessing which genes increased and decreased their activity in concert.16 They then compared how networks of gene activity altered in mice that had traits associated with obesity, diabetes or hardening of arteries. This complex and iterative process – combining conventional genetic analysis with high-level computation – allowed Schadt and his colleagues to calculate the probabilities of genetic connections and establish causal relationships for disease-associated traits.
With this, they reported the discovery of new genetic networks and sub-networks, with links, edges and hubs – terms more commonly used in describing electrical circuitry. They uncovered a connected group of genes that caused a range of traits such as levels of insulin, glucose, fat and cholesterol – each one being a contributing factor to poor health or disease. They could identify the most important individual genes or proteins that act as nodes or hubs in each genetic circuit. And with that, Merck had new leads for drug development.
Then, in 2008, not long after Schadt’s success was clear, Merck announced that it needed to cut 7,200 staff.17 It was a time when many drug giants were struggling to come up with new products – new drugs based on Schadt’s analysis were still only in a developmental stage – and Merck’s income was down a third in the quarter leading up to the announcement of job losses. Earlier that year, Merck also paid $58 million in settlement against problems with their blockbuster drug for treating arthritis, Vioxx, which allegedly carried an increased risk of heart problems. They were accused of promoting the drug irresponsibly and reportedly had to set aside $4 billion to deal with future claims.18 As part of the job losses and restructuring, the Seattle site where Schadt was based closed down, and some of its staff moved to research facilities on the East Coast. Schadt and his close colleague Friend didn’t want to move – they left Merck and took their big vision with them. But their superfast computers had to stay behind.
Computing technology advances at a breakneck pace: no news there. But biological data is expanding at breakneck pace too, and Schadt’s datasets easily outstrip the 1,048,576-row and 16,384-column limit for spreadsheets in Microsoft’s Excel.19 Schadt and Friend couldn’t continue without high-level computing. Fortunately, however, they left Merck just as new possibilities were becoming available with companies like Amazon, Google and Microsoft offering on-demand access to their computing infrastructure. These companies provided the relatively bargain-priced way forward they needed.
Schadt continued his mission by establishing – with his colleague Friend – a non-profit organization, Sage Bionetworks, using philanthropic and other funds. Their aim is to now create a common repository of information about people’s genes, metabolism and disease from which biological networks can be established and shared by all, and from which drugs can be developed and tested more quickly. The big idea is that, because there is inherent variability in the human population, if only we collect information about enough people then our genetics and traits can be analysed – similar to how Schadt analysed mice – to discover the networks underlying different human diseases.20 They suggest that this approach implies that we have to abandon many textbook descriptions of disease – because genetic networks rather than single genes are important – dispense with academic cycles of grants and papers, and replace recognition of individual egos with celebrations of teamwork.
The idea that solving complex problems requires teamwork, as well as individual heroes, is a theme throughout the story of compatibility genes: think back to how the early heroes – Dausset, van Rood, Payne and their colleagues – came together in a formative series of international workshops to begin our understanding of human compatibility genes. That early collaboration – the workshops that began in 1964 – triumphed in discovering diversity in human genes. Now, researchers like Schadt argue that a new level of international teamwork is needed to join the parts back together again – to understand how networks of genes give rise to complex human biology.
Nobody is sure how far analysis of biological complexity will take us; as we reach an ever-higher view over our constituent molecules, what will we learn? Is it possible that this approach will help us understand aspects of humanity previously only approached by philosophy or religious belief? The UK’s Chief Rabbi, Jonathan Sacks, like many other religious leaders, suggests that science can only put everything together up to a point. He says that humans can do two quite different things. ‘One is the ability to break things down into their constituent parts and see how they mesh and interact. The other is the ability to join things together so that they tell a story, and to join people together so that they form relationships. The best example of the first is science, of the second, religion.’21
But as we tackle complexity head-on, à la Schadt and like-minded biologists, an analysis of networks of genes and traits may allow us to ‘join things together’ and establish a more complete picture of humanity. Then, depending on how successful science is at putting things together as we move through the twenty-first century, religion might gain in popularity or atheism might spread ceaselessly. These next few steps that scientists take may have sweeping social implications. But for Schadt new medicine is the urgent issue – and how to tackle disease is a debate that’s just as fierce as science and religion; because complex analysis of all our variations in genes, traits and diseases isn’t the only option on the table.
Others argue that instead we should focus on finding the rare genes that underlie extreme versions of a disease. Advocates say this will better indicate where new drugs should be targeted, reasoning that a rare genetic variation that has a big effect (in the few people who have it) can indicate which interactions are important for that disease, even in people without the rare mutation. A powerful example of how a rare human variation has a huge impact comes from a group of people who are very lucky in their resistance to HIV – in a different way from the HIV Controllers.
In the early 1980s, before screening blood for HIV was routine, some haemophiliacs were inadvertently transfused with blood containing the virus. Many died. But there were rare cases in which haemophiliacs had been given HIV-positive blood on several occasions, yet they didn’t get AIDS. The tragic mistake of giving people infected blood revealed a new genetic superpower in a few of them.
The mutation that provides this superpower was revealed through some basic research about the virus: specifically, how the virus enters the body’s cells. The virus – much smaller than a human cell – first attaches itself to proteins at the surface of human cells before entering them. In 1996, several teams independently reported that one of the proteins that HIV latched on to at the surface of our cells was one called CCR5. Almost immediately afterwards it was discovered that some people have a mutant form of the CCR5 gene in which a small piece of the DNA is missing. People who have inherited two copies of this mutant gene (one from each parent) can’t make the normal CCR5 protein. And cells from these people can’t be infected with the common form of HIV-1, because it uses the CCR5 protein to latch on to for entering human cells.22 In 1996, a study following people at risk from HIV found that people with two copies of the mutant gene didn’t become sick with AIDS.23 The new superpower – which allowed some haemophiliacs to resist HIV infection – was in having two copies of a mutant CCR5 gene.24
To be clear about how these different results fit together, the study of HIV Controllers set out to discover what gave individuals an ability to stay free from AIDS for some time after being infected with HIV. In that case, compatibility genes were the biggest factor – these genes influence how well you can resist illness after infection with the virus. The studies of haemophiliacs, on the other hand, revealed that a mutant form of CCR5 is able to protect against infection in the first place – by stopping the virus getting into cells.
The example of CCR5 shows how discovery of a rare mutation does indeed expose a vulnerable aspect of the virus – in this case, its entry into cells. The mutation really is very rare – only about 1 per cent of Europeans have this loss of CCR5 and very few, if any, Africans or Asians do. But the rare mutation reveals a vulnerable aspect of how the virus infects all humans. As a result of this, a range of medicines is based on attacking the docking process by which the virus gets into cells. Drugs have been developed to act as decoys for HIV docking sites or to directly block the proteins used by the virus to latch on to our cells. In future, it may even be possible to manipulate a person’s genetics and give them a non-functional version of CCR5. It is – in principle – feasible to treat a person with AIDS by isolating their stem cells, shutting off the CCR5 gene in those cells and then giving them back to the person. This could give anyone the same resistance to HIV as naturally found in those few special haemophiliacs.
There’s one individual whose story is probably the most dramatic example you’ll ever come across of how one’s luck can change.25 His story is published with the patient remaining anonymous – as required – and it could happen to anybody. This story of this patient began with the unhappy news that he had been infected with HIV. While being treated with standard antiretroviral therapeutic drugs, he was then diagnosed, at age forty, as having cancer as well (specifically, acute myeloid leukaemia). To treat his cancer, the patient had chemotherapy and then needed a stem-cell transplant to replace his cancerous cells. But instead of transplanting cells from any donor with appropriately matched compatibility genes, the team of medical doctors and scientists took the chance to transplant stem cells from an individual with the genetic variant known to give protection against HIV – that is, from an individual with two copies of the mutant CCR5 gene. Almost unbelievably – but fitting with everything we’ve discussed – the patient went from having both AIDS and cancer to having neither. With one clever stem-cell transfusion, the patient survived two fatal conditions.
So, what’s the best route to understanding disease – or path to new medicine: looking for rare genetic variants like mutant CCR5 or screening for common genetic differences like beneficial compatibility genes? The answer: as we’ve seen – is that both are important. And that’s a big reason why funding agencies and philanthropists get pulled in all directions, leaving finances thinly stretched; again, who’s to know what’s best to do next?
Our genetic complexity also seeds a completely different route to better medicine. Just as we each respond differently to any given infection, we may also respond differently to any given medicine. So, doctors can embrace our individuality – and exploit human complexity – to provide drugs that are personalized to our own genetic needs. And, given that compatibility genes are especially diverse in the human population, it seems pertinent to explore the possibility that these genes could be used to predict how well a specific drug will work or perhaps how bad any side-effects might be. And again a proven example is at hand in the example of HIV – this time in the use of a drug against HIV, abacavir.
Abacavir is potent against HIV but 2–8 per cent of patients react badly to it. A reaction can cause a rash, fever, stomach ache and breathing problems. Treatment has to then stop, because things only get worse and can become life-threatening. Dramatically, in 2002, two teams independently found that having a particular compatibility gene is a strong indicator for who will react badly.26 Others later confirmed this and the culprit was B*57; in fact, a particular version of B*57. As we’ve mentioned, different class I compatibility genes are denoted as A, B and C, and versions of each are designated a number such as in B*57. But there are some variants that are very similar to each other. These are formally denoted using additional numbers as B*57:01, B*57:02, B*57:03 and so on. These ‘sub-types’ weren’t known in the early days of compatibility gene research – they’re only easily distinguishable by modern genetic analysis. Importantly here, it turned out that having B*57:01 – and not, say, very similar genes B*57:02 or B*57:03 – correlated with who would react badly to abacavir.
A double-blind, randomized clinical trial tested whether or not screening patients for having B*57:01 would work in predicting who would have bad side-effects. Patients were divided into two groups. In one group everyone was given the drug apart from those patients with B*57:01, and in the other group, absolutely everyone was given the drug as was the normal clinical practice. The result was that the number of people who reacted badly to the drug fell considerably in the screened group.27 So information about a person’s HLA type can be used to reduce the frequency in which a drug has bad side-effects.28
Most people would probably accept a doctor’s request of a genetic test to ensure a drug won’t give them bad side-effects. But fundamentally, why does having HLA-B*57:01 correlate with having side-effects from the HIV drug? Again, it comes down to the ability of HLA proteins to present samples of what’s being made inside cells – as peptides – to be checked by immune cells. Cells with B*57:01 and treated with abacavir can potently switch on immune cells, T cells specifically, to kill.29 Nobody knows how this happens. Perhaps the presence of the drug creates something that gets detected by the immune system as non-self. The drug could do this by triggering a new peptide to be made by cells or by modifying a peptide already made (perhaps by sticking to it).30 Whatever the precise details are, immune cells activated by B*57:01 cause the damaging side-effects of abacavir.
It is striking that the very same variant of HLA found to be protective against HIV – B*57 – is also the one predictive of side-effects in patients given the HIV drug. Although it seems an unlikely coincidence, it may be just that. It’s simply not known whether or not the importance of B*57 in sensitivity to abacavir and in control of HIV is linked in any way beyond the fact that B*57 presents peptides to T cells in both cases. HIV itself plays no role in the side-effects of abacavir; the drug can trigger an immune reaction irrespective of HIV infection.31 Also, only one sub-type of B*57 triggers the adverse reaction to abacavir, B*57:01, whereas other HLA types can also confer strong protection against HIV, such as B*27 or B*14, or the very closely related variant B*57:03.
The connection between HLA types and side-effects from abacavir explains why people of African or Asian descent almost never react badly to the drug. HLA types aren’t evenly distributed among us; we don’t just acquire them randomly from the pool of all those possible. Our HLA types are influenced by the diseases our ancestors were exposed to and by the HLA types of people that first populated different parts of the planet. The grand story of human migration and the colonization of our planet is written into the diversity of our HLA genes. And, as a result, Africans or Asians don’t often have B*57:01; so they usually have no problem taking abacavir.
It’s fascinating to explore this more. Around 150,000 years ago, all modern humans lived in Africa.32 Genetic evidence for this came during a flurry of activity in the 1980s and ’90s from analysing DNA stored in mitochondria, the energy-making parts of cells, and in the male-specific Y chromosome. These two parts of our genome are exceptional in that they are not inherited from both of our parents – only one. The Y chromosome is passed from fathers to sons and mitochondrial DNA is inherited from mothers. So variations in these parts of our DNA are passed on in a relatively simple manner and are easiest to analyse as markers of our ancestry. In 1987, the idea that humans originated in Africa gained genetic backing when an analysis of mitochondrial DNA from many people revealed that they all originated from one woman in Africa, who lived about 200,000 years ago.33 More recent complex analyses of our genomes have confirmed that humans populated the world by coming out of Africa, and together with archaeology and anthropology, we have sophisticated models for ancient human migration.34
It can still be debated exactly where people exited Africa from, if they exited only once, and where humans successfully spread to first, but it’s generally accepted that around 100,000 years ago humans successfully left Africa. About 50,000 years ago some reached Europe, and others entered the Americas at least 20,000 years ago. Important to the story of compatibility genes, relatively small groups of founders successfully populated each new territory. It’s been suggested, for example, that just hundreds or a few thousand modern humans made it across the Red Sea from East Africa to settle in Yemen and Saudi Arabia.35 Natural selection then acted on the founders’ gene pool as they adapted to the local environment – especially changes in climate, the available food and different types of infections.
Genetic analysis has even indicated that some of the variation in our compatibility genes probably came from us breeding with prehistoric, archaic, humans – Neanderthals and Denisovans.36 It is possible that the introduction of HLA variants into the modern human population through interbreeding was important in helping increase resistance to local infections. All this together – from human migration to interbreeding – underlies the current geographical structure of our genetic inheritance.
An outcome of this is that Africa retains the greatest genetic diversity in compatibility genes. Within the continent, there’s a close correlation between these genes and the different languages spoken – because human migration within Africa over the last 15,000 years has had a significant impact on both linguistics and genetics.37 Then across the world, diversity in our compatibility genes roughly correlates with distance away from Africa, becoming less variable in populations the further away from Africa – because migration to each new territory is seeded by a founding group of people.38
In some populations, HLA diversity is particularly limited; among the indigenous people of the Americas, for example, probably because they originated from particularly small founder groups. Brand new versions of HLA-B have relatively recently appeared in some. For example, distinct variants of HLA-B have been found in the Kaingang and Guarani tribes of southern Brazil, and the Waorani people of Ecuador.39 These versions of HLA-B are likely to be advantageous in fighting particular infections and their impact is perhaps especially important in a population that otherwise has relatively limited variation in HLA. Other rare compatibility genes are found in populations now living separately but related by a common ancestry, such as B*48, which occurs at relatively high frequency amongst Eskimos and other North American Indians, but rarely elsewhere.40
There’s also a geographical structure to the combinations of compatibility genes we each have. That is, many HLA genes occur together more often than would be expected from their individual frequencies. For example, the pairs of genes A*01 with B*08, and A*03 with B*07, both occur more frequently amongst Caucasians than would be expected by chance. A map of HLA types and their combinations in Europe marks out a boundary which corresponds to where the Alps are. Almost certainly this reflects how these mountains have been a barrier to gene flow during early stages of peopling the region.41
Overall, the current worldwide map of HLA types is an outcome of natural selection during our battles with infections and the pathways of human migration during the peopling of the world. The implication is that, broadly, our geographic heritage correlates with our susceptibility and resistance to various diseases – and even our response to some drugs.
Many places today are, of course, cosmopolitan, and a database used to keep information on people for potential transplant matches in the UK can reveal precisely how diverse we are.42 Amazingly, 268,000 people from across the UK are represented by 119,000 different combinations of compatibility genes. This is actually just an underestimation of the true level of our diversity, because the level of precision at which each gene was classified for this analysis was low. Even so, 84,000 combinations of compatibility genes were each represented by just one individual. Single compatibility genes can occur at relatively high frequency, but our complete set of genes is evidently personal: In this analysis, around one in three people were uniquely defined by their set of compatibility genes.
Although it remains to be seen exactly how important our individual compatibility genes may be in predicting the appropriate medicine for different diseases, it does seem unlikely that HIV, abacavir and HLA-B*57:01 will be a rare example. More cases like this are likely to emerge as we continue to probe the complexity of diseases. Compatibility genes may also correlate with our responses to vaccines, such as those commonly used for ’flu, polio, measles and rubella.43 This makes a lot of sense for our differences in class II genes, which encode the HLA proteins found on the specialized immune cells that play a role in initiating the kind of immune response key to successful vaccination. Population-specific vaccines may prove especially effective, though, in truth, we don’t yet understand what determines the length of time we remain immune after an infection or a vaccination.
Of course, compatibility genes are not the only genes that can influence our response to drugs. Rituximab, for example, a drug used to treat a certain type of cancer – lymphoma – works best for people with a particular version of another gene in our immune system.44 In fact, cancer is a prime candidate for treatments to be tailored to an individual’s genes. Myeloma, for example, is a cancer of immune cells in the bone marrow – and it is currently incurable. Life expectancies have increased over the last decade, thanks to new drugs, but not everyone responds to everything equally well. Across the world there are differences in treatments – the timing, combinations and doses of different drugs used in combination with stem-cell transplants all vary considerably. Much is decided on an ad hoc basis. Like many cancer treatments, personalized regimens are the norm – but the decisions are based on local expertise, far from being standardized.
The genetic make-up of myelomas from different people has been analysed and no single genetic factor causes this cancer.45 In fact, each myeloma cell has, on average, tens of genetic differences from the patient’s normal cells. Some mutations are common, including some involved in processes targeted by drugs already available. But a new opportunity has come from genetic analysis of myeloma; once again, instead of being defeated by the complexity, we could exploit it. Patients could have their cancer cells analysed genetically so that the appropriate treatment could be selected without wasting time testing drugs by trial and error – avoiding side-effects from drugs that can’t help. In this way, even though we don’t understand fully how the mutations combine to make a cell cancerous, we can just exploit the fact that it’s complicated to improve treatment.
It’s just an idea: it remains to be clinically proven. But there are many clinicians and scientists advocating that such personalized (or stratified) medicine should become a reality in the very near future. Technology for analysing genes currently improves about fourfold each year, and, if that continues, we’ll have sequenced the genomes of a million different people by 2016. That will be enough data to validate all kinds of specific genetic diagnostics. But there’s still a problem. One reason why this is not guaranteed to work out is that there can be great variety in the tumour cells themselves – even in a single patient. Even within one person, individual cancer cells can vary in their drug sensitivity. This situation is similar to what we saw with HIV: a highly variable enemy is more difficult to attack.
Viruses and tumours do something else that’s a problem – they actively thwart our defences. One way they do this is by trying to prevent our compatibility genes from working. It’s a battle; our immune system fights back by checking if compatibility genes have been interfered with. This involves a whole other function of our compatibility genes, a different way of looking at the immune system, and another set of immune genes that are also extremely diverse among us – in fact, they are probably only second to compatibility genes in how variable they are from person to person. This next piece of the canvas – hard-won by new heroes – reveals how compatibility genes can work in the exact opposite way to what we’ve discussed so far. Breathe it all in, embrace the complexity.