The Pill That Could · A Different Cancer

A Different Cancer, part 3

About a quarter of breast cancers had this. Slamon had identified, in his blots, the molecular driver of an entire subset of human breast cancer.

* * *

The paper appeared in Science in January 1987. It had Slamon as the first author and a team of UCLA collaborators. The title was “Human breast cancer: correlation of relapse and survival with amplification of the HER-2/neu oncogene.” The figure that everyone remembered was a Kaplan-Meier curve — the standard graph oncologists use to display survival over time — in which the patients with HER2-amplified tumors fell off, sharply, from the patients without the amplification. The amplified group had a roughly threefold higher risk of dying. The amplification was, by some statistical analyses Slamon's group ran, a stronger predictor of outcome than tumor size or lymph-node status, the two variables that had been the workhorses of breast cancer prognosis up to that point.

The paper was, in retrospect, a marker. It established that breast cancer was molecularly heterogeneous — that there was a subset of breast cancer driven by a specific genetic event, that the subset behaved differently from the rest, and that the responsible gene was one whose protein product was a known kind of receptor with a known kind of activity. Everything about it pointed at a drug.

The drug, however, would not be the same kind of drug that imatinib turned out to be.

* * *

There was, in the autumn of 1986, when Slamon's blots were going up on the lightbox in Los Angeles, a small biotechnology company in South San Francisco called Genentech.

Genentech in 1986 was about ten years old. It had been founded in 1976 by a venture capitalist named Robert Swanson and a biochemist named Herbert Boyer, the latter being one of the inventors of the recombinant DNA techniques that had made it possible, in the early 1970s, to take a piece of DNA from one organism and insert it into another — the techniques that, in chapter three, made it possible to express BCR-ABL in mouse marrow cells. Genentech had been founded around the bet that recombinant DNA could be used to make human proteins in bacteria, and that the proteins could then be used as drugs. The company's first product had been recombinant human insulin, which it licensed to Eli Lilly in 1978; the second had been recombinant human growth hormone; subsequent products had included tissue plasminogen activator, used to dissolve blood clots in stroke and heart attack.

By 1986 Genentech had begun to look at cancer. Its strategic bet, in this area, was that monoclonal antibodies could be used as targeted cancer therapies.

A monoclonal antibody is a specific kind of protein. Antibodies are the molecules the immune system uses to recognize foreign invaders — viruses, bacteria, anything the body has decided is not itself. Each antibody binds, with very high specificity, to a particular molecular feature on the surface of its target. The immune system makes a vast variety of antibodies, each with a slightly different binding specificity, so that for almost any molecular feature you can imagine, somewhere in the body there is a B cell making an antibody against it. ‘Monoclonal’ means that the antibody comes from a single clone of B cells — all identical, all making the same antibody — so that what you have is one specific antibody molecule, made in industrial quantity, with one specific binding specificity.

The technology to make monoclonal antibodies had been worked out in 1975 by César Milstein and Georges Köhler at the Medical Research Council laboratory in Cambridge, England. They had figured out how to fuse a B cell making a particular antibody with a tumor cell that grew indefinitely in culture, producing a hybrid — a hybridoma — that combined the B cell's antibody-making machinery with the tumor cell's immortality. You could grow hybridomas in tanks. You could harvest the antibody from the supernatant. You could, in principle, make as much of any antibody you wanted, as long as you had the hybridoma.

Throughout the late 1970s and early 1980s, monoclonal antibodies had been deployed as research tools — in fact, the antibodies Brian Druker used to detect phosphotyrosine were monoclonals — with great success. As drugs, however, they had been largely a disappointment. The trouble was that the antibodies were made in mice. A mouse antibody, given to a person, was a foreign protein. The human immune system recognized it as foreign and made antibodies against it. After a few doses, the patient's body had cleared the mouse antibody from circulation faster than the antibody could do whatever therapeutic work it was supposed to do. The problem was called HAMA — human anti-mouse antibody — response, and it had killed several promising therapeutic-antibody programs in the early 1980s.

Genentech, by 1986, thought it had a way around the HAMA problem. The technique was called humanization. You took a mouse antibody with the binding specificity you wanted; you identified the small portions of the antibody that were doing the actual binding (the complementarity-determining regions, technically); and you grafted those portions onto the framework of a human antibody, using molecular biology techniques. The result was an antibody that bound the same target the mouse antibody had bound but looked, to the human immune system, mostly like a human antibody. The HAMA response was much weaker. The drug could be given for longer.

All 13 chapters of The Pill That Could · Library · All courses