Anti-TNF for treating rheumatoid arthritis

In the early 1980s, Feldmann, a medically trained immunologist from Australia, was working at University College, London, and had begun studying a different autoimmune condition called Graves’ disease, which causes the thyroid to become overactive. In this illness, particular non-immune thyroid cells take on unusual properties, acquiring molecules (such as HLA class II proteins) that allow them to stimulate an immune reaction that leads to inflammation. Scientists knew that these same molecules cropped up inappropriately in other autoimmune diseases as well, and that cytokines stimulate production of them.

Feldmann proposed that cytokines stir up autoimmune disease. He decided to test this idea; as a first step, he would check whether tissue afflicted with an autoimmune illness contained excess cytokines. He chose rheumatoid arthritis, because it would provide the opportunity to study tissue at the height of inflammation, which was not possible in Graves’ disease: Doctors routinely remove diseased tissue from rheumatoid arthritis patients to relieve symptoms.

A mutual acquaintance recommended that Feldmann meet Maini, a leading rheumatologist and researcher at the Arthritis Research Campaign’s Kennedy Institute of Rheumatology at Imperial College in London. With one foot in the lab and one foot in the clinic, Maini had good access to the human tissue that would be needed for the investigations. The notion that deranged cytokine behavior fostered rheumatoid arthritis had snagged his attention as well, and the two scientists teamed up, hoping eventually to harness the chaos.

By the early 1980s, scientists were developing new research tools that could identify individual cytokines — an important advance because these chemical messengers often show up in groups. The ability to catalog which cytokines were present — and to foil them separately — would prove crucial to Feldmann’s and Maini’s analysis.

The two researchers, as well as others, showed that the joints of people with rheumatoid arthritis teem with proinflammatory cytokines. Furthermore, Feldmann and Maini found that the diseased joint cells themselves produced these cytokines in an uncontrolled fashion. When the researchers grew the cells from afflicted knee joint tissue in culture dishes, the mixture churned out cytokines continuously for six days; during a healthy inflammatory response, the signaling molecules appear only briefly. The observation that arthritic cells gush cytokines implied that the normal means of tempering the immune response had gone awry.

The identity of the cytokines hinted at the underpinnings of specific disease features. For example, one of the cytokines in the joints recruits immune cells to sites of tissue injury, an observation that could explain the local inflammation; another activates antibody-producing cells, and might thus spur the output of antibodies that ravage the body’s own tissues. Despite these potential windows into the molecular mechanisms of the disease, the results vexed scientists. Many of the cytokines present in the joints perform overlapping duties. A treatment that disarmed one cytokine could prod another to work overtime, experts reasoned, and wouldn’t quell the disease.

Maini and Feldmann, however, suspected that a single cytokine might act as a fire alarm to jolt the entire system into action. Studies had shown that a cytokine called IL-1 causes joint damage in animal tissue. Furthermore, mice suffering from a condition that mimics some aspects of rheumatoid arthritis produced IL-1 in their inflamed joints. These observations pointed to IL-1 as a rheumatoid arthritis suspect. The researchers wanted to know whether they could find a cytokine that would kick-start the disease — and in particular, IL-1 production.

As a first step, Feldmann and Maini tested whether several cytokines emitted by joint cells in culture dishes influenced the manufacture of IL-1. To accomplish this task, they added antibodies that knocked the cytokines out of commission. Most antibodies had no effect, but one that foiled TNF quashed IL-1 output. Feldmann and Maini then conducted the converse experiment, adding instead an antibody that disabled IL-1; TNF quantities remained unchanged. Additional work showed that the anti-TNF antibodies also squelched the manufacture of other proinflammatory cytokines. These results and those from other groups indicated that TNF played a key role in governing the creation of cytokines and other inflammatory molecules: Blocking this single molecule could apparently turn down the entire inflammatory response. To move toward their goal of treating humans, the researchers wanted to test whether similar events occurred in mice with arthritis.

Injecting collagen — a component of connective tissue, which includes cartilage and bone — into a particular strain of susceptible mice sparks a condition that shares some key features with rheumatoid arthritis. In particular, a comparable immune response erupts in the joint and a related type of tissue damage develops. Using collagen, the researchers induced rheumatoid arthritis-like symptoms in mice and then injected anti-TNF antibody. This treatment reduced swelling and joint destruction. Several other groups — in New York City, Geneva, and Athens — independently generated similar results, supporting the idea that an excess of TNF can incite the whole disease.

This success in animals gave the researchers confidence to attempt therapy in patients. But the antibody they had used in the rodent studies wouldn’t suffice because it inactivated mouse, but not human, TNF. The researchers had antibody that bound human TNF, but it had been made in mice and would likely cause dangerous side effects as the human immune system mounted a response against it; furthermore, this immune response would expel the antibody. Feldmann and Maini wanted instead a molecule that bound human TNF and contained a large portion of a human antibody. A person’s body would treat the resulting antibody more normally and wouldn’t reject it.

In attempts to fight sepsis, a number of companies had developed such chimeric human-mouse antibodies as well as other compounds that knocked TNF out of commission. However, at the time (early 1990s), the conventional wisdom in the pharmaceutical industry was that therapeutic antibodies wouldn’t thwart chronic diseases. Humans wouldn’t tolerate antibodies that were part human and part mouse over the long term, the thinking went; the immune system would notice the portion from mouse and attack it, limiting the benefit and inducing dangerous side effects. Furthermore, many experts thought that disrupting the cytokine system would spur it to reorganize; inflammation would flare up again and patients would wind up back where they started. Finally, given the number of cytokines in diseased joints, scientists were still skeptical that hobbling a single cytokine would relieve symptoms. Although the strategy had triumphed in mice, rodent and human physiology differs and many apparently promising therapies had failed to transfer from animals to people. Even Feldmann and Maini didn’t think that an antibody-based therapy would prove ideal in the long run. Producing these molecules is very costly and thus, therapeutic antibodies — even if they worked — wouldn’t flourish in the marketplace.

However, if Feldmann and Maini could remedy disease in humans by blocking their target molecule, they figured they could eventually develop a less expensive drug that would mimic the antibody. Therefore, they were eager to know whether their novel scheme would work.

Fortunately, one of Feldmann’s former visiting scientists named James Woody had taken a position as research director at a company called Centocor Inc. in Malvern, Pennsylvania, which had created a chimeric anti-TNF antibody for use against sepsis. Woody was amenable to Feldmann’s and Maini’s idea, and Centocor agreed to provide material for a preliminary test in humans.

In 1992, Feldmann and Maini, working together at the Kennedy Institute by this time, performed the first clinical trial of anti-TNF therapy for rheumatoid arthritis. Because no one knew whether it would trigger harmful side effects in patients who were already ill, the researchers recruited only patients who had not responded to other therapies and had no other treatment options. They injected the antibody, called cA2 at the time — subsequently known as infliximab, and now Remicade — into the bloodstream.

Within a few hours, the patients reported dramatic symptomatic relief; they felt more energetic and their joints had loosened up. Within several weeks, previously incapacitated people were playing golf and climbing stairs. Maini and Feldmann were thrilled — but they worried about the placebo effect — a phenomenon in which people feel better even when they’re receiving a fake drug. Objective measures, however, indicated that the disease was retreating. Quantities of a blood-borne protein that accompanies inflammation dove and joint swelling subsided.

Between 6 and 12 weeks after finishing the two-week treatment, symptoms recurred. The researchers re-administered the antibody to eight of the 20 patients, which again considerably benefited them. Eventually disease returned, indicating that blocking TNF temporarily did not permanently cure individuals with resistant disease. However, the antibody seemed safe and the results justified further trials. Feldmann and Maini presented their findings at a meeting in 1992, drawing the attention of other companies, which started brushing up their own anti-TNF agents for possible use in combating rheumatoid arthritis.

The striking initial success persuaded Centocor to support the first multi-center clinical trial of TNF blockade in four European cities. To generate formal proof that the treatment helps patients, the researchers set up a rigorous experiment in which two groups of patients would receive the active drug in different doses and another would receive a placebo; neither doctors nor patients would know who was in which group. The researchers injected the agent of choice once into 72 people and then tracked disease over a period of four weeks. Again, joints became less tender and swollen, and amounts of the blood-borne inflammatory marker plummeted.

These results heartened Feldmann and Maini, but rheumatoid arthritis is a long-term disease, and their patients eventually relapsed. To test whether re-treatment might further fend off the disease, they conducted another study in which they administered five infusions over a period of 14 weeks and observed patients for six months. This tactic kept the disease on hold, indicating not only that people could tolerate the antibody over the long term, but also that it continued to exert a therapeutic effect. In this study, the researchers also assessed whether another drug enhanced the anti-TNF regimen. Feldmann and Maini knew from their work in animals that depleting a particular class of immune cells amplified the effects of the anti-TNF antibody. A compound called methotrexate, the most commonly used anti-rheumatic drug, hinders the activity of these same immune cells, so the researchers administered methotrexate along with the antibody to half of the patients. Their predictions panned out: Methotrexate magnified the positive effects of anti-TNF, and combination therapy is currently used for the majority of rheumatoid arthritis patients who undergo long-term anti-TNF therapy.

The treatment was clearly performing well by every yardstick they had used so far, but Maini and Feldmann, joined by Peter Lipsky in Dallas and several other North American and European groups, wanted to measure structural joint damage per se, which they could track over a long period of time with X-rays. A year of treatment with anti-TNF antibodies and methotrexate every eight weeks arrested cartilage and bone devastation in about half of 428 patients. Not only did joint destruction stall in these individuals, but the results hinted that the body managed to repair previous injury. These results gave Centocor sufficient results to apply for FDA approval of their antibody. Even after two years on the combination regimen, the patients continued to do well.

Although Centocor was the first company to sign on to anti-TNF antibody trials for rheumatoid arthritis, Feldmann and Maini’s 1992 disclosure of their results prompted other companies to compete vigorously and test their own anti-TNF agents. Wyeth/Immunex’s etanercept (Enbrel) was approved in November 1998, a year before Centocor’s drug was approved for use in combination with methotrexate for rheumatoid arthritis. Three drugs that restrain TNF — Remicade, Enbrel, and Abbott’s adalimumab (Humira) — are now licensed in the United States and in Europe. Patients inject themselves under the skin from twice a week (Enbrel) to once every two weeks (Humira), or come to the clinic every eight weeks for an infusion into a vein (Remicade). All of these agents pack an effective punch; for many patients, they are continuing to keep the disease in check even after five years of treatment. Several more TNF blockers have been tested and appear promising in clinical studies, but have not yet been approved.

The clinical trials established that the treatment helps many people — but they also revealed a gap. Approximately 60 percent of the individuals studied responded to anti-TNF therapy. Because these patients included only those with severe disease whose illness defied other therapies, that statistic is impressive — but it leaves room for improvement. In an attempt to help people who derive no benefit from treatment and to better understand the molecular mechanism of the disease, Feldmann and Maini have dug deeper into how anti-TNF therapy works. As predicted, amounts of particular cytokines in the blood and joints ramp down after treatment and, presumably as a consequence, inflammatory cells stop gravitating toward the ailing joints. Furthermore, new blood vessels — which normally nourish the congregating immune cells — no longer form. Together, these observations suggest that the treatment scrambles signals that would otherwise draw troublemaking cells to the joints and minimizes their ability to gather. Additional experiments hint at how the anti-TNF agents deter — and possibly even heal — joint damage. The treatment, for example, depletes a particular type of tissue-destroying enzyme. Such information might provide information that could goad researchers toward the development of alternative treatments for patients with recalcitrant disease.

The success of anti-TNF-based therapy for rheumatoid arthritis led to clinical trials to assess these agents’ ability to curb other chronic illnesses in which cytokines become unruly. The first was Crohn’s disease, an inflammatory condition of the bowel. Anti-TNF antibody ameliorated the illness, and Remicade is now approved for the treatment of severe Crohn’s disease. More recently, scientists have conducted successful trials of agents that foil TNF in other chronic diseases of excess inflammation, and the drugs are now licensed to treat juvenile rheumatoid arthritis, ankylosing spondylitis, and psoriatic arthritis.

All in all, therapies that muzzle TNF have benefited approximately 400,000 people, approximately 70 percent of whom suffer from rheumatoid arthritis. By pursuing their vision, Feldmann and Maini turned a speculative hypothesis into a potent discovery that will aid myriad patients for years to come.

by Evelyn Strauss

Key publications of Marc Feldmann and Sir Ravinder Maini

Brennan, F.M., Chantry, D., Jackson, A., Maini, R.N., Feldmann, M. (1989). Inhibitory effect of TNF alpha antibodies on synovial cell interleukin-1 production in rheumatoid arthritis. Lancet. 2, 244–247.

Chu C.Q., Field, M., Feldmann, M., and Maini, R.N. (1991). Localization of tumor necrosis factor alpha in synovial tissues and at the cartilage-pannus junction in patients with rheumatoid arthritis. Arthritis Rheum. 34, 1125–1132.

Williams, R.O., Feldmann, M., and Maini, R.N. (1992). Anti-tumor necrosis factor ameliorates joint disease in murine collagen-induced arthritis. Proc. Nat. Acad. Sci. USA. 89, 9784–9788.

Elliott, M.J., Maini, R.N., Feldmann, M., Kalden, J.R., Antoni, C., Smolen, J.S., Leeb, B., Breedveld, F.C., Macfarlane, J.D., Bijl, H., and Woody, J.N. (1994). Randomised double-blind comparison of chimeric monoclonal antibody to tumour necrosis factor alpha versus placebo in rheumatoid arthritis. Lancet. 344, 1105–1110.

Maini, R.N., Breedveld, F.C., Kalden, J.R., Smolen, J.S., Davis, D., Macfarlane, J.D., Antoni, C., Leeb, B., Elliott, M.J., Woody, J.N., Schaible, and T.F. Feldmann, M. (1998). Therapeutic efficacy of multiple intravenous infusions of anti-tumor necrosis factor alpha monoclonal antibody combined with low-dose weekly methotrexate in rheumatoid arthritis. Arthritis Rheum. 41, 1552–1563.

Feldmann, M., Maini, R.N. (2001). Anti-TNF-alpha therapy of rheumatoid arthritis: what have we learned? Annu. Rev. Immunol. 19, 163–196.