New medicines often begin with big ideas, and Sir Marc Feldmann had one. Born in Poland in December 1944, his family moved to France immediately after the war, and then to Australia when Feldmann was eight. ‘Immigrants have a very strong motivation to work hard and succeed,’ he thinks, having gained his own work ethic from his father who worked long hours as an accountant while studying at night.1 At medical school in Melbourne, Feldmann was bored by rote learning of human anatomy, but thrilled by the uncertainties and emerging ideas he found in scientific papers. It was while doing research for his PhD at the Walter and Eliza Hall Institute in Australia – ‘fuelled by coffee and the music of the Rolling Stones’2 – that he took the first real step on the journey towards his great contribution to immunology, ultimately providing relief from pain to millions of people and spawning what is now a multibillion-dollar industry.
It began with a feeling of dissatisfaction. It had recently been established that an immune response involved many different cells – emphasised by Steinman’s discovery of dendritic cells (Chapter Two) – and, looking down his microscope, Feldmann could see that immune cells were dynamic and moved about.3 By contrast, the research he was doing on these isolated immune cells seemed too reductionist, too far removed from what really happens in the body. ‘Concepts generated in one precise circumstance often do not extrapolate to complex and non-reductionist reality,’ he later wrote.4 Of course, all scientific experiments are reductionist in some way; it would be impossible to conclude much – if anything – from their outcomes without isolating to some degree the particular aspect of the whole whose effects we wish to study. But Feldmann’s point was that he wanted to know what’s happening in the body, across the system, not just what’s going on inside a single type of immune cell. His thoughts turned to the way that different immune cells talk to one another.
To study this, he set up a flask containing two glass tubes, one inside the other, with a porous membrane at the ends of both tubes, and filled the flask with culture broth. The broth could flow freely through the membrane but larger particles, such as cells, could not. With this set-up he could put different types of immune cells in the inner and outer tubes and keep them separate, while allowing them to bathe in the same culture broth. He set up several of these flasks and by comparing what happened in them with ones without separate tubes so that cells moved around and interacted with each other freely, he could assess which kinds of immune reactions required direct contact between cells and which could be triggered by secretions from cells into the liquid. A handful of other scientists were doing similar experiments around the globe, and even though they had almost no understanding about what was in the liquid specifically, they were essentially all studying the effects of cytokines. In 2016, I asked Feldmann what he learnt from these early experiments, and he replied, with a chuckle: ‘We discovered that life is complicated.’5
Feldmann was one of the pioneers present at the first cytokine workshop in 1976, among around forty scientists who gathered in a hotel near the US National Institutes of Health with the aim of establishing a coherent picture of what cytokines did. It was an almost hopeless task at first, because there was no way of isolating different cytokines and therefore establishing if the various effects on each type of immune cell were caused by one or several of them.6 It was only after the cytokine genes had been isolated, allowing the different cytokine proteins to be produced individually, that the effects of each could be studied systematically. This showed that each cytokine had multiple diverse activities, which was a controversial idea at first because it was generally thought at the time that each kind of protein in the body did only one job.7 Many acronyms used to name cytokines had to be abandoned because several of them, it turned out, applied to the same molecule. Eventually, with the tools in hand to properly dissect the cytokine world, the excitement of a gold rush ensued, electrifying (or intoxicating, depending how you view such things) some of the scientists involved with the prospect of money and fame.
Science’s supposed moral purity took a hit in October 1984 at the fourth cytokine workshop. At an exclusive resort in the Bavarian Alps, Philip Auron from Charles Dinarello’s lab at Massachusetts Institute of Technology (MIT) announced that his team had isolated the gene for one of the forms of the cytokine named IL-1.8 One scientist who was in the audience vividly remembers the excitement,9 and another recalls it as ‘the big moment of the workshop’.10 At the start of his speech, the session chair announced that photography was strictly prohibited, a condition that Auron made before he agreed to present his data. During his talk, Auron flashed up the genetic sequence for IL-1, briefly.11 As soon as Auron’s talk ended, someone dashed to the audience microphone and shouted ‘This is not IL-1.’12
The heckler was Christopher Henney, who, in 1981, had co-founded the biotech company Immunex, based in Seattle.13 Henney was approaching forty years old when he started Immunex with colleague Steven Gillis, both then at Seattle’s Fred Hutchinson Cancer Research Center. ‘I couldn’t see myself doing the same thing for twenty-five more years,’ Henney says. ‘Some guys get their hair permed, put gold chains around their neck, and go chase girls. I decided to start a company.’14 At the audience microphone, Henney announced that his company had isolated the IL-1 gene, and that the sequence Auron just showed wasn’t it.15 Auron asked Henney to show what he thought was the correct gene for IL-1. But Henney refused and went back to his seat.
A summary of the workshop, published shortly afterwards, noted that ‘this lack of etiquette was most surprising…especially since Christopher Henney had a long and distinguished university career prior to joining Immunex’.16 Soon after the event, Immunex published the sequences of two forms of the IL-1 gene, alpha and beta, in the journal Nature. One of these was in fact identical to the gene the MIT researchers had found and announced at the workshop.17 The MIT team published a letter in Nature stating that this proved they were right all along and that Immunex had had no reason to cause such outrage at the cytokine workshop.18 But more than pride was at stake. A small biotech company, Cistron, were working with the MIT team. They and Immunex had both filed patent applications around the IL-1 genes. Digging into who did what and when to resolve the patent squabble, it emerged this was not a simple case of normal scientific competition.
Cistron alleged that Immunex had cheated. They claimed that Immunex co-founder Gillis gained information about one of the IL-1 genes when he was sent the MIT team’s paper by Nature for peer review, supposedly a confidential process.19 Nature rejected this paper on account of the peer reviews they received for it. Crucially, the patent application submitted by Immunex included errors in the gene sequence identical to errors found in the draft paper from MIT – something which is highly unlikely to have happened by chance. Cistron alleged that this proved Immunex submitted their patent for IL-1 using the genetic data that they had been sent for peer review. Immunex countered that it was a simple clerical error. And their lawyers built a case claiming that in fact, there were no hard rules dictating confidentiality during the peer review of scientific papers anyway.20
It took twelve years for the situation to be resolved and the final settlement involved Immunex paying $21 million to Cistron.21 Reportedly, both Henney and Gillis helped with the payment personally.22 By that time, Cistron had gone through bankruptcy, and the value of the disputed patents had anyway become limited as high doses of IL-1 proved to be toxic.23 Immunex, on the other hand, had by this time discovered and studied a long list of genes that are important in the immune system and in 2002 the company was purchased by another biotech company, Amgen, for $16 billion.24 Henney and Gillis moved on to be directors of several other biotech companies.
A year before the IL-1 debacle kicked off in the Alps – while Immunex and Dinarello’s team at MIT were racing to isolate cytokine genes – Feldmann was on holiday in a small town dominated by a fifteenth-century castle ruin on the Costa Brava coast of Spain. Relaxed and away from the hubbub, he had an epiphany. He worries that long holidays are often shunned in labs nowadays: ‘Holidays,’ he later wrote, ‘provide not only an opportunity to enjoy family, friends, and the splendour of our planet, but also time to think creatively and strategically.’25 His big idea, which he published later in the Lancet, was about the origin of autoimmune disease.26
He mused whether immune cells might activate each other through their cytokine secretions to such an extent that the activation becomes self-perpetuating, creating a vicious circle that overstimulates the immune system and causes it to harm the body. This was a powerful new idea. Though he had little proof of it being true he put it out there nonetheless, he recalls, in the ‘slightly overconfident mode of the young’.27 Today, it would be almost impossible to publish an idea like this without a wealth of supporting data, especially in a journal as prestigious as the Lancet, but this was a different age for biological science: there were fewer scientists, less competition for space in journals, and editors were probably more open to papers about an idea than they are today. At any rate, ideas can sometimes move us forward even in the absence of evidence, and the most important implication of this one, from a medical point of view at least, was that blocking a cytokine might stop immune cells from driving each other on and thus prevent autoimmune disease.
Feldmann decided to focus on one autoimmune disease in particular, rheumatoid arthritis, a long-term inflammation in joints which causes pain, stiffness and sometimes disability. It affects around one in a hundred people in every country.28 We don’t understand precisely how the problem starts, which probably varies in different people, but the symptoms come about because immune cells accumulate in joints and, over time, cause the destruction of cartilage and bone. To a small extent, rheumatoid arthritis runs in families and forty-six genes have been linked to the disease.29 But if a person with rheumatoid arthritis has an identical twin (who shares the exact same set of genes), there’s still only a one in five chance that their twin will also develop the disease. That’s because there are many non-genetic factors involved, which we don’t understand very well. Drinking lots of coffee (defined as four or more cups a day in one study) correlated with a slight increase in risk, for example, in one analysis.30 This connection is not entirely clear, however, because different studies have come to different conclusions,31 but even if taken as fact, it’s hard to disentangle whether or not this is a direct effect of drinking lots of coffee or if drinking lots of coffee is simply indicative of some other cause. At the time Feldmann set his mind to working on rheumatoid arthritis, the feeling among experts was that this was a very complicated disease, with many factors involved, so no simple treatment, certainly not a drug which targeted one particular molecule, would be likely to help.
After his PhD, Feldmann had moved to London, in part because ‘there was more money for research than in Australia’.32 Here the clinician Sir Ravinder ‘Tiny’ Maini helped Feldmann focus on rheumatoid arthritis. Born in Ludhiana, India in 1937, Maini had moved to Uganda in 1942, where his father became a minister in the British Ugandan government, and then to the UK in 1955.33 He was recommended to Feldmann as a physician open to new ideas. Two days after they first spoke on the phone, Maini was in Feldmann’s office in London, and a long-lasting friendship began. ‘It was a meeting of minds,’ Maini recalls.34 Although friendship isn’t strictly required for a successful collaboration, Feldmann thinks the kind of trust that friendship brings is essential.35 They each brought to the table different experiences and backgrounds – Feldmann was better known as an immunologist while Maini had the relevant clinical expertise – but they had enough in common to be able to communicate with the same jargon. Neither led the other; they became two halves of a team. ‘A union,’ as Maini puts it.36
Choosing to tackle rheumatoid arthritis as opposed to other autoimmune diseases was important because the relevant human tissue was accessible to study. Maini could easily provide needle samples from patients’ joints, whereas the relevant tissues required for the study of other autoimmune diseases were, and still are, very hard to obtain – the brain in multiple sclerosis or the pancreas in diabetes, for example. They decided that their first goal should be to find out which of all the cytokines were made by immune cells accumulated in the inflamed joints of arthritis patients.37 Studying cells and fluid isolated from patients’ joints is what set Feldmann and Maini apart from most other researchers, and it’s what got them on the right path towards finding a way to tackle the disease. They discovered that many cytokines were present but that one – with the unwieldy name ‘tumour necrosis factor alpha’ (often abbreviated to TNF)38 – was especially abundant.39
TNF had been identified in 1975 as a factor released from immune cells that was able to turn tumours black and dead.40 This immediately led to great interest in the cytokine in the hope that the tumour-killer could be used to treat cancer patients, a hope that was dashed when it became clear that the cytokine itself is quite toxic to the body even at doses too weak to be able to impact a tumour. But every cytokine has a multitude of activities and TNF’s ability, at high doses, to kill tumours was not what interested Feldmann and Maini. Rather, they wanted to test what would happen if they blocked the activity of TNF in the inflamed joints of arthritis patients. To do so, they needed an anti-cytokine – which is something that can be produced in the form of an antibody.
Antibodies are secreted by the white blood cells known as B cells and are our body’s ‘magic bullets’, a term coined in the 1890s by German Nobel laureate Paul Ehrlich. They are soluble protein molecules that stick to and neutralise all kinds of germs and other potentially dangerous molecules. Each individual B cell produces an antibody with a uniquely shaped tip, the part of the antibody that sticks to its target molecule, called an antigen, which might be, for example, something on the outer coat of a bacteria or virus. However, antibodies are not designed to bind to germs per se. The shape of each antibody’s tip is created almost randomly by a process of chopping up and rearranging the genes that create the antibody, a remarkable process in its own right. B cells which happen to have made an antibody that could stick to healthy cells and tissues are killed off (or inactivated) so that the only B cells allowed in the bloodstream are the ones that make antibodies that stick to something not normally found in the body. This is the process we first met in Chapter One, and is how these cells are able to distinguish self, components of your body, from non-self, anything that’s not part of you.
In more detail, every B cell also has a version of its own antibody tethered to its surface (the B cell receptor we also met in Chapter One), so that the cell can tell when there is something in the body that its antibody could lock onto. When a B cell does have the right antibody to lock onto something alien and troublesome, the B cell multiplies so that its useful antibody is produced in bulk, ready to neutralise the intruding molecule or germ. With around 10 billion B cells in the average person’s immune system, each of us has the ability to make around 10 billion differently shaped antibodies, each of which is able to recognise something that hasn’t been in the body before, ensuring that antibodies can be produced against virtually any structure alien to the body. This is essential if our immune defence is to tackle germs which the body hasn’t seen before – even germs which have never even existed. Crucially for Feldmann’s and Maini’s purposes, it also means that any animal could make antibodies against a protein found in any other animal. Therefore a mouse immunised with the cytokine TNF could produce antibodies that would lock onto the human cytokine and stop it working – an anti-cytokine.
Precisely such an antibody was made by scientist Jan Vilček at the New York University Medical School. Vilček was born in Czechoslovakia in 1933, to Jewish parents who were proud of their ancestry but not religious. In 1942, his family were granted exemptions from the normal degrading restrictions placed on Jews so that, for example, they were not required to wear yellow Stars of David and could keep their jobs.41 Such exemptions involved hefty administrative fees, possibly bribes, and conversion to Christianity, but officially, they were given on the basis that the country needed people in certain jobs to continue; Vilček’s mother was an ophthalmologist and his father worked in the coal-mining industry, presumably both important enough.42 At age eight, Vilček was sent for protection to live in an orphanage run by Catholic nuns.43 After German troops suppressed an insurrection by the Slovak resistance in 1944, Vilček’s parents were worried that a new hard-line government wouldn’t recognise their exemptions from restrictions on Jews. Vilček left with his mother to shelter for many months among a handful of remote farming peasants, and later in an isolated village.44
After the war, with his family reunited, Vilček went to medical school in communist Czechoslovakia, where ‘a general atmosphere of fear and suspicion permeated life’.45 At this time, the term ‘gene’ was practically banned because Stalin favoured a direct inheritance of traits acquired in one lifespan, and many scientists who disagreed were imprisoned.46 It was in 1957 during his time as a student that he was inspired to study cytokines after hearing a talk by Alick Isaacs, the co-discoverer of interferon, who was on a visit to Czechoslovakia.47 As Vilček spoke English very well, he was selected as Isaacs’ guide and so got to know him a little.48 Later, this proved to be an important personal connection.
After medical school, Vilček joined a research centre dedicated to viruses, and in 1960 published a paper in Nature, providing evidence for the existence of interferon.49 The director of the research institute had wanted Vilček to publish in the local journal Acta Virologica, but – importantly, it would soon turn out – Vilček did not take the advice.
In 1964, Vilček and his wife Marica, an art historian, defected from Czechoslovakia.50 Because he had published in Nature, not Acta Virologica, he had three job offers before he even made it out of Europe to his new home in the USA.51 He picked New York University Medical School, where he has since stayed for his entire career. Vilček later learned that another reason he was offered a faculty post without an interview was that they had received a recommendation letter from Isaacs, whom Vilček had acted as a guide for seven years earlier.
Vilček’s life story is one of amazing achievement following an early life of adversity. With the royalties Vilček received from making an anti-TNF antibody, he set up the Vilček Foundation, which champions the contributions of immigrants to life in the USA. In 2005, he had received so much money from making this antibody that he could donate to New York University Medical School the largest gift any New York healthcare institute ever received: $105 million.52 This bought new professorships, lab refurbishments, a student dormitory, research fellowships, studentships and more. Occasionally – very occasionally – a career in academia can be financially lucrative. Vilček laughs at the thought of it all: ‘Getting rich was not really something that was ever a goal of mine. To be honest, I’m still a little embarrassed by it.’53
To make the antibody, Vilček first had to obtain a sample of the human cytokine, TNF, to be injected into mice. In late 1985, the company Genentech had isolated the gene for TNF and obtained significant quantities of the protein by expressing the gene in bacteria. Vilček was able to obtain samples of it in 1988 because he was collaborating with them on another project. To make the antibody using mice, he then followed a method that had been worked out by César Milstein and Georges Köhler in Cambridge in 1975, an enormously important method for which Milstein and Köhler won a Nobel Prize in 1984.54 First, Vilček immunised mice with Genentech’s TNF protein and, after a few days, isolated B cells from their spleens, knowing that many of these B cells would be producing antibodies against TNF. Outside the animal’s body, B cells can’t survive for long, a few weeks at best when incubated in cell-culture broth, but Vilček used a trick to keep them alive – the Nobel Prize-winning trick that Milstein and Köhler had hit upon55 – which was to fuse the B cells with myeloma tumour cells and create new cells, called hybridomas, which retain the growth traits of a tumour with the antibody-producing abilities of the original B cell. In effect, this creates immortal versions of the mouse B cells. Vilček then isolated each single hybridoma cell, separating them by pipetting a minuscule amount of the liquid suspension that contains the cells into each of the many indentations in a hand-held rectangular plastic dish. Then, the antibody produced by each one could be tested for its ability to block the activity of TNF. The cell found to produce the appropriate antibody was then cultured to produce an almost limitless supply of anti-TNF antibody.
This type of antibody is called a monoclonal antibody as it derives from a single B cell. The process can be used to create a protein shaped to lock onto any molecule of our choosing. As well as being used as medicines, antibodies are used in all kinds of scientific experiments, to mark out particular cells, block the activity of something or switch on the activity of something else, check the level at which something is being secreted, and so on. ‘No single class of reagents stirs our creativity, or propels our goals, our successes, even our dreams, with as much excitement as do monoclonal antibodies,’ as one expert has said.56
Vilček had a long-standing agreement with the then fledgling company Centocor to develop commercial applications from the antibodies made in his lab.57 In return the company paid for some of Vilček’s lab costs,58 including the salary of the postdoctoral researcher, Junming ‘Jimmy’ Le,59 who helped make the anti-TNF antibody.60 It was, however, elsewhere in New York where it became apparent that an antibody against TNF might be medically important. Bruce Beutler, the Nobel laureate we met in Chapter One, who helped discover that toll-like receptors lock onto bacteria, worked, earlier in his career, with Anthony Cerami at the Rockefeller University Hospital, and discovered the mouse version of TNF. In 1985, he found that TNF was one of the cytokines produced in mice during sepsis, a disease caused by an immune response going into overdrive, usually because of a bacterial infection.61 Importantly, Beutler and Cerami found that blocking TNF could protect mice from the symptoms of sepsis.
In humans, sepsis (called septic shock when symptoms include a drop in blood pressure) can kill patients in a matter of hours and, with many cases not being easily treated with antibiotics, the illness accounts for up to tens of billions of dollars in hospital care in the US.62 Centocor had already highlighted treatments for sepsis as an important focus for the company.63 Now, inspired by Beutler’s and Cerami’s work, they wanted to try treating sepsis in humans by blocking TNF.
They couldn’t use Vilček’s anti-TNF antibody in people straight away. Since the antibody was made in a mouse, it had to be modified to more closely match antibodies naturally made in humans. Otherwise, the antibody itself would be seen as something alien in the human body and could trigger an immune reaction. To avoid this, segments of genes for the mouse antibody were combined with human genes to create a new half-mouse half-human antibody.64 Actually about 34% mouse, 66% human, its front end was kept in its mouse form to lock onto the TNF cytokine, and the back end made human. A chimera, like a lion with a head of a goat, made real on a molecular scale.
In 1991, Centocor tested the chimeric antibody in patients with sepsis.65 Although there were no adverse side effects noted, there was no clear therapeutic benefit. What worked in mice didn’t work as well in humans; a common theme in medical research. It looked like the anti-TNF antibody might end up as a scientific tool – perhaps part of a diagnostic test for cytokine levels in blood – but not an actual medicine.66 Then, in early 1991, Feldmann visited Centocor to present his case for trying the antibody in patients with rheumatoid arthritis.
By that time, Feldmann had some evidence to support the idea that TNF was important in rheumatoid arthritis and that blocking its activity might help. Maini’s team had found that the cytokine was present in the right place at the right time for being involved in the symptoms of the disease. In Feldmann’s team, Fionula Brennan (who sadly died young from breast cancer in 2012) had looked at what happened when an anti-TNF antibody was added to cells taken from the diseased joint of patients. The result was a eureka moment. Brennan discovered that when TNF was blocked, other cytokines stopped being produced by the cells.67 She repeated the experiment seven times to be absolutely sure.68 This implied that TNF was at the top of a cascade of events, or the hub of a network, which led to the other inflammatory cytokines being produced. These results flew in the face of the scientific consensus at the time, that no single molecule could be responsible for something as complex as inflammation in rheumatoid arthritis. Also, most scientists thought that the cytokine system involved lots of redundancy, so that if you blocked one component, it wouldn’t make much difference to inflammation overall, because other cytokines would continue to work. The dogma was wrong, Feldmann argued, and blocking this one cytokine, TNF, might stop this autoimmune disease.
Another of Feldmann’s team, Richard Williams, tested the idea in mice. The symptoms, though not the underlying cause, of human arthritis were recreated in mice by immunising the animals with collagen so that an immune reaction develops against this protein, a major component of cartilage, which in turn results in swelling of the animal’s joints. The afflicted mice were then given an injection of anti-TNF antibody and, at high doses, the inflammation was reduced and cartilage at the animal’s joints was spared from damage.69 This showed that mice could be relieved of the symptoms of arthritis by an injection of anti-TNF antibody.70 Still, many at Centocor were sceptical of this working in people – in part because the company’s only rheumatologist had his own ideas about what would work in treating rheumatoid arthritis.71
Importantly, James ‘Jim’ Woody, who had done research for his PhD under Feldmann’s supervision in London, was now the chief scientific officer at Centocor and he liked the idea. In fact, he took up the position at Centocor knowing that this opportunity was on the horizon, and he helped prepare the way by involving Feldmann in other projects at Centocor, so that when Feldmann pitched his idea for treating rheumatoid arthritis, he was already known in the company as a leading academic. Vilček thinks that without Woody there, supporting his former boss, Centocor wouldn’t have tried the anti-TNF antibody in rheumatoid arthritis patients because it was ‘such a long-shot’.72 But Woody was there and Centocor agreed to provide enough of the antibody for a small trial. A personal connection is often what’s needed to make things happen.
Centocor agreed that Feldmann and Maini could conduct a small trial at Charing Cross Hospital in London, with just ten patients at first, then later another ten, all tested without placebo controls. The reason they didn’t use placebo controls was that at this point they thought of this as a scientific experiment, to test if blocking TNF could do anything to help patients, and – though it seems strange in hindsight – they weren’t really thinking that the anti-TNF antibody might itself be used as a medicine.73 Maini recruited patients who had not responded to any other medicine available at the time. Everyone he approached was happy to participate, despite him explaining the risks of an experiment that was possibly dangerous.74
Centocor’s trial in 1991 in sepsis patients had showed that the antibody was at least broadly safe, but Feldmann and Maini were nevertheless cautious and each infusion was started slowly. For the first patient, treated on 28 April 1992, they had a nurse spend the night in the patient’s room. They needn’t have worried. As the antibody was infused, many patients said they felt better immediately. ‘It was a very thrilling time,’ Feldmann recalls, ‘all the patients we treated improved dramatically.’75 Graphs, bar charts and statistical analysis recorded the results – the reduction in swelling and tenderness in patients’ joints was formally significant after two weeks76 – but video footage of patient number eight said it all.
Before treatment, patient eight walks very slowly up and then down some stairs, one step at a time, holding onto the handrail, clearly in pain. Four weeks after treatment, she runs down the same steps as fast as anyone could. At the bottom, she throws her arms up in the air and exclaims ta da! The happiness on her face is a reminder of what this story is all about.
Vilček remembers seeing the video in Centocor’s offices soon after it was recorded.77 Even though this small trial didn’t have any control patients, it was clear to him that such a profound change in the patient’s well-being could not have been brought about by the placebo effect, a consequence of the brain’s anticipation of a positive response. But there was still a huge question to be answered: how long would the benefit last? Everyone involved – patients, clinicians and scientists – knew that the next few months were vitally important. All the patients returned to their normal lives, health improved. A dentist – patient number three – was able to play golf just two weeks after his treatment and later returned to work.78 But unfortunately the benefits were short-lived. Everyone relapsed.
Evidently the antibody wasn’t a cure; but it could relieve symptoms. This meant that the next logical step was to test the benefits of blocking the cytokine repeatedly. Feldmann and Maini obtained ethical permission to re-treat some of the patients, and again, all those tested improved. Still, the results were anecdotal, with no controls; just twenty patients had received the first dose, and after relapse, eight had been re-treated. As a scientific experiment, it was informative, but a medical advance requires a proper clinical trial, randomised and double-blind, where neither clinicians nor patients know who is getting the new treatment and who isn’t.
The results of the first formal trial were unequivocal; anti-TNF antibody improved the health of rheumatoid arthritis patients. Detailed analysis of what was happening in patients’ blood revealed that the antibody was working just as Feldmann and Maini predicted: blocking this one cytokine reduced the production of other inflammatory cytokines, and biopsies showed that fewer immune cells were entering the diseased joints. Feldmann thinks such detailed analysis of what was happening in patients – taking biopsies and analysing 400 ml of blood from everyone – was done because the trial was carried out by academics rather than a company. Most clinical programmes, Feldmann believes, don’t spent the time and money on analysing what happens in patients to such extent, and this is a great loss.79
That’s not to say it all ran smoothly in the hands of the academics. At one point, a freezer defrosted and crucial samples were destroyed; ‘It was physically painful to think of the major scientific opportunities lost,’ Feldmann recalls.80 For the next clinical trial – a full-scale comparison of anti-TNF antibody with existing treatments, a so-called phase III trial – Centocor was keen to get their antibody approved as a medicine as fast as possible, so fewer samples were taken and, Feldmann says, the emphasis on detailed analysis was lost.
The phase III trials proved that anti-TNF antibody was an effective therapy and was better than other treatments available at the time. Experiments in mice revealed that its benefits were enhanced when combined with other drugs that helped dampen immune responses. This led directly to what is commonly prescribed to patients today: anti-TNF antibody usually taken with another drug, methotrexate, which has many effects in the body including, it turns out, dampening T cell immune responses. This was an early example of medicines being used in combination to treat a disease, something that’s far more common today. Poly-pharmacy, Feldmann calls it.81
Centocor’s ambition to treat sepsis was never fulfilled; the disease remains notoriously difficult to tackle, likely because a storm of inflammation builds up so rapidly in the body that it’s especially hard to control. When it became clear that its efforts to treat sepsis weren’t working out, Centocor’s share price collapsed, down from $50 per share to just $6 within a few months in 1992, and the workforce was reduced from around 1,600 to 400.82 Anti-TNF therapy came to the company’s rescue. Centocor’s human-mouse chimeric anti-TNF antibody was marketed as the drug Remicade.83 And as a result, in 1999, the company was bought by Johnson & Johnson for $4.9 billion.84 Vilček recalls that the price seemed high at the time – initially, sales of Remicade were slow because it was such a radical new kind of medicine and physicians did not to rush to use it – but later, $4.9 billion looked like a bargain.85
Feldmann regrets that the blocking of TNF became another British invention commercialised in the US. He had approached UK companies, but they weren’t interested; only the charismatic leaders at Centocor took the chance.86 Many of Centocor’s rivals in the US, including Abbott, Roche and Immunex, went on to develop other drugs that also block TNF. Beutler helped create one of the alternatives: a soluble protein version of the cytokine’s natural receptor. Effectively this acts as a decoy receptor to prevent the cytokine’s engagement with its real receptor on immune cells. Clinical development of Beutler’s drug, led by Immunex, began two years behind Centocor’s but raced ahead to such an extent that in November 1998 it became, in fact, the first anti-TNF medicine approved to treat rheumatoid arthritis in the US, marketed as Enbrel.87 Other US companies made alternative versions of anti-TNF antibody, including a fully human version available from 2002.88 All of these medicines have been very successful. By convention within the pharmaceutical industry and according to, for example, the European Commission, a drug is considered a blockbuster if sales reach $1 billion, which makes Centocor’s anti-TNF antibody a blockbuster many times over. UK-based companies largely missed out on a therapy which, in 2012 alone, made $9.3 billion. This could be viewed as important or trivial, but what is undoubtedly important is that because of drugs that block TNF, far fewer people with rheumatoid arthritis are forced to use a wheelchair.
If rheumatoid arthritis were the only condition that could be treated by blocking TNF, then this therapy would still be a blockbuster, but it proved to be of even wider use. Blocking this cytokine helps stop inflammation in many situations where it is a problem: in the digestive system, as happens in Crohn’s disease and colitis; in skin, as happens in psoriasis; and in the joints of the spine in ankylosing spondylitis. Across the globe, Centocor’s anti-TNF antibody has been used to treat at least 1.8 million people.89
This success did not come about in any easy or linear way; it took a multitude of small steps for Centocor to make the part-mouse part-human antibody, tested by Feldmann and Maini, based on the antibody Vilček made first. Triumph came from imagination and hard work, but also from a web of coincidences, chance events and serendipities. Reflecting on this in his memoir, Vilček quotes E. B. White, the author of children’s books Stuart Little and Charlotte’s Web: ‘No one should come to New York to live unless he is willing to be lucky.’90
All those who directly contributed to the discovery of anti-TNF therapy deserve celebration. Many have, rightly so, won prestigious scientific prizes. In 2013, Vilček received the US National Medal of Technology and Innovation from President Barack Obama, who summarised Vilček’s life in a few moving sentences. Vilček wished his parents had still been alive to hear them.91 In 2003, Feldmann and Maini won the prestigious Albert Lasker Award for Clinical Medical Research and, in 2014, the Canada Gairdner International Award. But those who pursued other cytokines, in other illnesses, were doing important science too. After all, it might well have turned out that blocking another cytokine would help patients with rheumatoid arthritis – in fact, we now know that blocking IL-6 can also help – and it might have worked out that blocking TNF would help treat sepsis, for example. It takes a community to explore all the possibilities.
Indeed, unlike vaccination – discovered long before anybody had any detailed understanding of how it worked – anti-TNF therapy emerged directly from understanding the molecules and cells that make up our immune system; knowledge that was generated by thousands of scientists. We tell the stories of individuals – and perhaps ego propels scientists into action – but no scientist is an island. At some level, this therapy was achieved by a collective scientific mission to understand immunity. Maini is especially proud of this fact; that his work helped show how the detailed molecular science of immunology can be harnessed for medicine.92
Science doesn’t have any endings; discoveries are made, new treatments found, but everything leads to something else. The discovery of anti-TNF therapy was a watershed moment because it introduced a new way to combat disease – manipulating the immune system rather than directly fighting germs with, say, antibiotics, in a way that’s very different to vaccination. Feldmann’s next big idea was to wonder how many other illnesses might be tackled with drugs that block cytokines. Although we do not yet know the extent to which asthma, diabetes, coughs, the common cold and strokes can be tackled by manipulating cytokine levels, these and many other conditions and diseases are all potential targets. Pharmaceutical companies and many academic research labs are on the case, betting that the success of blocking TNF is not a one-off fluke but the dawn of something even bigger.
That’s not to say blocking TNF is a perfect medicine; far from it. There are at least three significant problems with anti-TNF therapy. First, blocking this part of the immune system inevitably weakens our defence against infections.93 Major problems are rare, but for people with latent tuberculosis – showing no sign of illness because their immune system normally controls the infection – there is an increased risk of the disease reactivating when their immune system is compromised by anti-TNF medicine.94
A second problem with anti-TNF therapy is that a significant fraction of patients do not benefit from it: as many as four in ten rheumatoid arthritis patients show little improvement.95 Combinations of drugs can improve response rates, but unfortunately, we currently have no way of knowing in advance who will respond and who won’t. Standard clinical practice is to proceed by trial and error: patients are simply given one of the drugs that block TNF and if no significant improvement is experienced in three months or so, they are switched to a different type of anti-TNF therapy or something else entirely. Maini thinks that one factor which might be important in determining whether or not a patient responds to anti-TNF therapy is how long ago the problem began.96 If the inflammation in a person’s joints has already lasted a very long time, Maini thinks, it probably becomes – in some sense that it is not well understood – more complex and harder to control. Perhaps this is also one reason why some patients respond well to anti-TNF therapy at first, but the drug loses its potency over time.97 The third problem is that blocking TNF is an effective treatment, but not a cure.98 The quest for a cure continues.
Feldmann’s and Maini’s research also had far-reaching consequences because of the type of medicine they used – an antibody. At the time, the potential for antibodies to be used as medicines was not widely recognised because they were – and still are – very expensive to produce. The B cell hybridomas that produce antibodies have to be grown in a broth that contains something like fifty different ingredients. Even in optimal conditions, stirred just right in a bioreactor, each cell produces only a minuscule amount of antibody, which then has to be purified to a standard that’s safe for use as a medicine. A few companies, Centocor included, were built on the premise that antibodies could make money, but even these companies thought that they were most likely to obtain regulatory approval for their use in diagnostic blood tests far more easily than for actual therapies. Centocor’s first antibody product was a test for the hepatitis B virus, for example.99 The therapeutic and commercial success of the anti-TNF antibody showed everyone the true scale of antibodies’ potential as medicines.
Anti-TNF antibody was not actually the first antibody approved as a medicine; that was an antibody sold as Orthoclone, approved in 1985, designed to lock onto the white blood cells known as T cells and eliminate or deactivate them. It was hoped that this antibody could stop an immune reaction from developing in transplant patients, which might otherwise cause a transplanted organ to be rejected. This antibody was approved for use in kidney, heart and liver transplant patients but is no longer used today. The antibody didn’t work well and the side effects were serious; some people developed a potentially life-threatening condition, likely on account of the treatment sometimes triggering T cells to release high levels of cytokines. Ever since Milstein and Köhler learnt how to make antibodies à la carte, it seemed that there had to be a role for antibodies in medicine.100 But in practice, for the nearly two decades it took for anti-TNF to be developed, the pursuit felt like chasing a rainbow.
One of the most important antibodies developed subsequently is rituximab. Instead of blocking a cytokine, this antibody directly targets immune cells, specifically B cells. When it locks onto a protein molecule on the surface of a B cell, that particular B cell is destroyed in one of three ways. First, the antibody itself can cause the B cell to self-destruct. Billions of our body’s cells die this way every day to allow for a healthy turnover of cells in the body; rituximab can simply trigger this same programme of cell death. A second way that the antibody kills B cells is that, while its front end is tethered to a B cell, its back end attracts factors in the blood which then kill the B cell. Alternatively, its back end can be recognised by the immune system’s Natural Killer cells, which flatten up against the B cell and kill it. Again, these last two processes happen as part of our normal immune defence; antibodies usually lock onto germs or infected cells, things which warrant attack in this way. Rituximab essentially causes a person’s own B cells to be detected by the immune system as something to be eliminated.
The loss of B cells in the body that results from this antibody can in turn dampen inflammation in a patient’s joints, so rituximab is prescribed as an alternative medicine for those rheumatoid arthritis patients who do not benefit from anti-TNF therapy.101 However, it was first approved not for the relief of rheumatoid arthritis but in 1997 to treat cancer. It has since been used by over 750,000 cancer patients. At a glance, it seems quite unlikely that a cancer drug could help with rheumatoid arthritis; these maladies have little in common. But an antibody which kills B cells is useful for the types of cancer – chronic lymphocytic leukaemia and non-Hodgkin lymphoma – where it is a B cell that has lost control and become malignant. In fact, this antibody is so important that it features on a list of the world’s most essential medicines, produced by the World Health Organization, each selected because ‘their potential health impact is remarkable’.102
Our detailed knowledge of the way in which antibodies kill has led to improvements in their design: for example, antibodies can be produced with subtly different structures that make them more effective at triggering attack by Natural Killer cells. It has also led to the discovery that some people have a genetic variation which makes their Natural Killer cells less efficient at killing cells coated with antibody. In lymphoma patients, there is evidence that this genetic variation correlates with a less successful response to rituximab (but the issue is controversial as this has been found in some, but not all, studies).103
Not all antibodies which lock onto B cells are equally effective at causing them to die and so an important scientific frontier is to understand why rituximab is so effective. In my own lab, we have used state-of-the-art laser-based microscopes – which cost around half a million pounds each – to make videos of the process by which rituximab binds to a cancer cell and then causes immune cells to attack.104 We discovered that rituximab doesn’t coat a cancerous cell uniformly but tends to gather on one side of the cell, drawing some proteins towards and into the area where the antibody has accumulated and sending others to the opposite side of the cell. In effect, this antibody creates a front and back to the roughly spherical cancer cell, or in the jargon, causes the cancer cell to become polarised. We found that cancer cells which gained this polarity were, for reasons we don’t understand, more easily killed.
Just by watching what happens with the use of a microscope, we can infer that rituximab’s effectiveness as a drug is, in part, attributable to this ability to change the structure of the cancer cell, making it especially susceptible to being killed by immune cells. This implies that when making new antibody-based medicines to kill off a particular type of cell, it would be useful to screen not just for those which stick to the right type of cell but those which trigger equivalent changes in the target cell’s structure as well. This is, however, at the edge of knowledge – a hypothesis rather than a fact – and it’s hard to know for sure how important this is because we can only watch such activity in a lab culture dish. It is alas impossible to watch whether or not antibodies trigger these events inside patients. This was, after all, Feldmann’s point at the outset: we need to know what happens inside the body, where the entire system is at work, not just in isolated cells in a lab dish. The lens, used in a microscope or a telescope, has opened up all kinds of new worlds, in space, in ponds and in us. New technologies that improve how we see, especially how we see inside the human body, will, for a long time to come, play an ever-increasing role in medical research.
The success of the anti-TNF antibody, and rituximab, began a fashion for seeking more antibody-based medicines, but in 2006 momentum was lost. Controversy erupted when a clinical trial testing a different antibody drug, named TGN1412, went horribly wrong. The trial – implemented by a small company that soon became insolvent – used an antibody that was designed to activate the T cell without the usual need for dendritic cells, the alarm cells that Steinman discovered, to detect danger first. The thinking was that these T cells should then attack cancer cells more readily.
In animals, the drug didn’t cause any problems, so it was tested in people – thankfully at a low dose. All six patients given the drug suffered heart, liver and kidney failure and ‘many months of hell’, as one victim put it when interviewed later by the BBC.105 As we’ve seen already, drugs often behave differently in people than they do in mice. In the patients, the drug activated T cells to such a high degree that they began to attack the body’s healthy cells and tissue. The overly active immune cells also released cytokines at such high levels that they became toxic to the body. What happened to the patients in the trial is somewhat like what can happen in sepsis, an overreaction of the immune system caused by an acute bacterial infection. The patients all developed fevers, one developed pneumonia, their blood circulation began to fail and their fingers and toes went black. Thankfully, nobody died, but the clinical trial was a tragedy.
Many scientists profess to know – in hindsight, of course – that switching on a vigorous immune response, circumventing the body’s normal checks and balances, is bound to be a bad idea. An official investigation found that the problems were, however, due to an ‘unpredicted biological action of the drug in humans’.106 Whether or not the disaster was predictable, the fallout was huge and included major changes in how human trials are now approved. For example, a group of patients should never be given a new drug at the same time: an interval between them allows for side effects to become apparent. In this case, patients developed an inflammatory response which could have been picked up within ninety minutes.107 So waiting even this short amount of time could have spared subsequent patients the same trauma. The important lesson for science was finding out – all too dramatically – that tinkering with our immune system is like trying to harness nuclear power: there is great potential but a mistake can be catastrophic.
Ultimately, the discovery of anti-TNF therapy showed us that a detailed knowledge of immunity pays off, not only because it reveals a hidden beauty in how the human body works but because this is an area of science that leads to new medicines. Still, the road to each new medicine is not a highway; it is a narrow lane, uncharted on the satnav and full of blind corners. Driving fast is unsafe. We must map more of the immune system to understand how and why its activity varies, to understand the boundaries it operates safely within and, crucially, how it connects with other body systems – which is where we will turn our attention next.