The Pill That Could · The Compound

The Compound, part 2

This was the consensus that the small group of chemists in Basel, in the late 1980s, was about to start arguing with.

* * *

Ciba-Geigy in 1988 was a Swiss pharmaceutical company of moderate ambition and considerable history. It was the result of a 1970 merger between two older Basel firms, Ciba and J. R. Geigy, both of which had begun in the nineteenth century making industrial dyes and had, like several other Basel chemical companies, eventually moved into pharmaceuticals on the back of the chemistry expertise. The company had a respectable but not spectacular drug portfolio. Its oncology pipeline, by the standards of the larger players, was thin.

Inside the company there was a small research effort, run by a physician-scientist named Alex Matter, whose remit was to develop new cancer drugs. Matter had come to Ciba-Geigy from Schering-Plough; he was a clinically trained oncologist who had moved into the research side, with strong opinions about what kinds of drugs would actually help patients and a willingness to make bets on hypotheses that other people thought were premature. He had decided, around the time he took the job, that targeting kinases looked like the most interesting frontier in cancer drug discovery, and that Ciba-Geigy should be working on it whether or not the rest of the industry was.

He needed people. He hired Nick Lydon, a Scottish biochemist who had been working at Schering-Plough on tyrosine kinases, and brought him to Basel as the head of a small biology effort. He also brought on a chemist named Jürg Zimmermann, who would do the actual work of designing and making the compounds. Both were young; both had reputations as careful, patient workers; both believed, as Matter did, that the selectivity problem was an engineering problem and not a fundamental obstacle.

Their first target was not, in fact, BCR-ABL. It was protein kinase C — PKC — a serine/threonine kinase (one of the more conventional kinds, phosphorylating serine and threonine rather than tyrosine) that had been linked to several diseases, including some cancers. The early work was a screening campaign: Zimmermann's chemists made or purchased a library of small molecules, the biology team tested each one against PKC in a test-tube assay, and the molecules that inhibited the enzyme were identified as starting points.

Several scaffolds came out of the screen. One of them was a class of compounds called 2-phenylaminopyrimidines — a chemical structure consisting of a six-membered ring with two nitrogens (the pyrimidine), with another ring (the phenyl) attached through a nitrogen-containing linker (the amino bridge). It was a flexible scaffold; you could decorate it with various chemical groups in various positions, and Zimmermann began making variations.

Some of these variations inhibited PKC, as intended. Some of them, on follow-up, also turned out to inhibit other kinases. This was, in the standard interpretation, a problem — a 'dirty' compound that hits more than its intended target is harder to develop into a drug, because it's harder to predict what it will do in a body. The conventional response was to optimize toward selectivity for the intended target.

Lydon had a different idea. If the compounds were hitting other kinases, that information was worth keeping. Tyrosine kinases were emerging as cancer-relevant targets across the board, and a series that hit several of them was, from the right angle, a starting point for several different programs rather than a contamination problem in one. In particular, some of Zimmermann's 2-phenylaminopyrimidines were inhibiting tyrosine kinases that Lydon was personally interested in: PDGFR, the platelet-derived growth factor receptor, which was thought to be involved in glioblastoma and certain other tumors; and the closely related kinase c-KIT, which would later turn out to drive a different family of cancers.

And ABL.

Lydon had been a tyrosine kinase biochemist at Schering-Plough. He had read the same BCR-ABL literature that Druker had read. He knew exactly what it would mean if you had a selective inhibitor of ABL.

* * *

The optimization was incremental. Zimmermann would make a new variation, the biology team would test it against a panel of kinases, the data would come back, and Zimmermann would make another variation. The goal was to find the modification that pushed potency higher against ABL while pushing potency lower against everything else.

There were many variations. The 2-phenylaminopyrimidine series was eventually elaborated into hundreds of compounds, each with slightly different chemical decorations. Most of them were not impressive. Some of them were a little better than the last batch. Occasionally a single change — a methyl group added in this position, an amine swapped for that one — would produce a compound noticeably more selective than its predecessors.

Around 1992 or 1993 the team converged on a molecule that, in their assays, inhibited ABL with respectable potency and inhibited only a small handful of other kinases meaningfully. It was given a Ciba-Geigy code number: CGP 57148. (The salt form actually used in animal and clinical work was the methanesulfonate salt, sometimes written CGP 57148B. For our purposes, the code number is the molecule.) Its known activity was: ABL, BCR-ABL, PDGFR, c-KIT. Its known non-activity was: most of the other kinases the team had been able to test. The off-target hits were, by the standards of the field, an unusually short list.

By the standards of pharmaceutical drug discovery, however, CGP 57148 was a problematic compound. It was not, in 1993, considered a viable drug candidate by most of the people inside Ciba-Geigy who were paid to make those calls. The reasons were partly chemical — its physical properties, in some respects, were not ideal — and partly strategic. Even if it worked, the company asked, what would it work on? CML was uncommon. The class — selective kinase inhibitor — had no approved precedent. The compound looked more like a research tool than a marketable medicine.

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