Chapter I
Nothing Acts Unless It Binds
Two lines of work converged on the same idea around 1900. John Newport Langley, studying the nerve endings in muscle, found that nicotine stimulates where curare blocks, and that the two interfere with each other rather than with the nerve or the muscle. Something in the tissue must be the point of attachment for both — a "receptive substance", which is where the word receptor comes from.
Paul Ehrlich came from staining. He had spent years finding dyes that mark one tissue, one cell type, one granule and nothing else, and he drew the obvious conclusion: chemical affinity is specific enough to discriminate between parts of a body. If a dye can find one structure, a poison can find one organism. Corpora non agunt nisi fixata — substances do not act unless bound.
Making that into a drug took a systematic search. Ehrlich and Sahachiro Hata worked through hundreds of organic arsenic compounds against the spirochaete of syphilis, looking for one that killed the organism at a dose a rabbit could survive. Compound 606, marketed as Salvarsan in 1910, was it. The drug was difficult and dangerous — a weekly intravenous infusion, for over a year, with serious toxicity — and it was the only effective treatment for a common fatal disease, and it was designed, in the sense that the search was directed by a principle rather than by folklore.
Ehrlich also gave the field its central number. The chemotherapeutic index is the ratio of the dose that harms the host to the dose that harms the target. Everything else about a drug is negotiable; a ratio near one means there is no drug.
Chapter II
From Effects to Measurements
Alfred Joseph Clark turned the subject into a quantitative science in the 1930s by assuming the simplest possible thing: a drug molecule occupies a site, there are finitely many sites, and the effect is proportional to the fraction occupied. That gives the same hyperbolic saturation curve that Michaelis and Menten had derived for enzymes, described under biochemistry. Plotted against the logarithm of concentration it becomes a sigmoid, and the concentration at half-maximal effect measures affinity.
The consequences are practical. Potency and maximum effect become separate quantities, so a drug can be weak but complete, or strong but partial. Antagonism becomes testable: a competitive blocker shifts the agonist's curve to the right without lowering its ceiling, while a non-competitive one lowers the ceiling. And the existence of a receptor can be inferred from the pharmacology before anyone has isolated the protein, which is exactly what James Black did twice.
Black's first target was the receptor through which adrenaline drives the heart. The received approach to angina was to widen the coronary arteries; he argued it would be better to reduce the heart's demand for oxygen, by blocking the signal that raises it. Propranolol followed in 1964 and became one of the most-used drugs in the world. He then inferred, from data alone, that histamine acts on a second class of receptor in the stomach distinct from the one antihistamines block, and set his chemists to build an antagonist for it. Cimetidine made most surgery for peptic ulcer unnecessary.
Meanwhile Gertrude Elion and George Hitchings pursued the other strategy: attack a metabolic step. The sulfonamides, found by screening dyes at Bayer, had turned out to work by mimicking a precursor of folate — a vitamin bacteria must synthesise and humans eat, which is the selectivity in one sentence. Elion and Hitchings applied the logic to nucleic acid metabolism deliberately and produced drugs for leukaemia, gout, transplant rejection and malaria from a single pathway. Their best demonstration of selectivity was acyclovir: the compound is inert until phosphorylated, and the enzyme that phosphorylates it efficiently is one the herpes virus supplies. An uninfected cell does essentially nothing with it.
Chapter III
A Closer Look: Why HIV Takes Three Drugs
The arithmetic that forced combination therapy is worth doing, because it is short and it decided clinical practice.
Measurements in 1995 and 1996 established the scale of HIV replication in an untreated patient: on the order of new virions produced per day. The viral reverse transcriptase makes errors at roughly
Suppose resistance to a particular drug requires one specific base change. The number of virions produced each day already carrying it is about
Three hundred thousand resistant virions per day, before treatment starts. Monotherapy cannot work: the mutant is not created by the drug, it is merely given the field. This is precisely what was observed — viral load fell for weeks to months, then returned as the resistant lineage took over.
Now two drugs, requiring two independent mutations in the same genome. Assuming independence,
Nine doubly resistant virions a day is still certain failure, on a timescale of months rather than weeks.
Three drugs:
or about one triply resistant virion every
That is the whole argument for triple therapy, and in 1996 it worked as the arithmetic said it would: viral loads fell below detection and stayed there, and HIV became a managed chronic infection.
The assumptions deserve to be stated, because each is a known failure mode. Independence is optimistic — recombination between two virions in a co-infected cell can combine resistance mutations in one step, and a single mutation sometimes confers resistance to several drugs in the same class, which is why combinations must mix mechanisms and not just molecules. The calculation also assumes the drugs reach the virus everywhere; sanctuary sites where concentrations are low support replication and therefore evolution. And it assumes doses are taken, which is why adherence, rather than pharmacology, is the main determinant of treatment failure in practice.
The same arithmetic explains tuberculosis, treated with four drugs for six months, and the resistance that follows interrupted courses. It explains why cancer chemotherapy is given in combination — the tumour is also a fast-reproducing population under selection — and why a single targeted agent against a kinase usually buys months before a mutation in the binding pocket appears. Imatinib's unusual durability comes from the leukaemia being driven by one abnormal enzyme on which the cells have become entirely dependent, which is rarer than anyone hoped in 2001.
Chapter IV
The Race That Does Not End
Ehrlich saw the problem in 1907: he reported that trypanosomes exposed to his arsenicals became resistant, and that resistance was inherited. The sulfonamides met resistance within a decade, penicillin within years of mass production, and every class since has followed. The explanation is in the preceding section, applied to bacteria rather than viruses, and it means that resistance is not a sign that a drug was flawed.
What has changed is the supply of replacements. The pathway from a new target to a usable antibiotic against Gram-negative bacteria has produced almost nothing for sixty years, because the chemical properties that let a molecule cross an outer membrane and resist efflux pumps conflict with the properties that make it a good inhibitor. The incentives are also perverse: a drug that should be held in reserve cannot be sold in quantity. The open problem above is the field's clearest case of a difficulty that is simultaneously biochemical, evolutionary and economic, and it is the reason routine surgery, transplantation and cancer treatment — all of which assume infection is treatable — are considered at risk.