Appearance
25.2 — How We Got Thousands of Medicines
In October 1937, a chemist at a small Tennessee drug company had a problem a customer had asked him to solve. Sulfanilamide, the first drug that reliably cured a streptococcal infection, only came as a tablet or a powder, and children would not take it. He needed to dissolve it in a liquid.
Water would not do it. Alcohol would not do it. Diethylene glycol did it beautifully — a sweet, syrupy solvent that today you would recognise as the main ingredient of antifreeze. He added raspberry flavouring, the company shipped 240 gallons of it across the country, and it killed 105 people, most of them children.
Nobody had broken a law. In 1937 there was no law requiring anyone to test whether a medicine was safe before selling it. There was no law requiring the ingredients to be listed. The only charge the government could bring was that the word "elixir" implied alcohol, and this product contained none — a mislabelling offence.
Thirteen months later the Federal Food, Drug, and Cosmetic Act of 1938 was law, and for the first time a company had to show a medicine was safe before it could sell it. That single sentence is the ancestor of every clinical trial, every submission, and every validated system you will ever work on.
This chapter is the history of how the industry got here, and it is really the history of a series of disasters, each of which added a permanent rule. Learn the disasters and you will never again have to memorise why a rule exists.
Before there was an industry
For almost all of human history, medicine came from plants, and the dose was a guess. Willow bark for pain, foxglove for a failing heart, cinchona bark for malarial fevers, opium poppy for everything. Some of it worked. Nobody knew which part of the plant did the working, or how much of it was in any given batch, so the same remedy could do nothing one week and kill someone the next.
The change came when chemists learned to pull the active substance out of the plant and weigh it. In 1804 a young German apothecary's assistant, Friedrich Sertürner, isolated a crystalline substance from opium and named it morphine after Morpheus, the god of dreams. In 1820 two French pharmacists, Pierre Joseph Pelletier and Joseph Bienaimé Caventou, isolated quinine from cinchona bark.
That is the moment medicine became measurable. A pure compound has a molecular weight, a melting point and a dose. Everything downstream in this Part — a specification, a batch record, a dose-response curve — depends on the idea that the thing in the bottle is a known quantity of a known substance.
The next step was making a molecule that no plant had ever produced. In 1897 a Bayer chemist, Felix Hoffmann, produced a stable form of acetylsalicylic acid; Bayer put it on the market in 1899 as Aspirin. A chemical company had turned itself into a drug company, and the modern pharmaceutical industry started there — in the dye works of Germany and Switzerland, which is why so many of the old names are chemical companies.
The idea that changed everything: the magic bullet
Paul Ehrlich, a German physician, had a thought while staining tissue slides. Some dyes stuck to bacteria and not to human cells. If a dye could pick out a microbe from among human cells, then a poison could be attached to something that picks out a microbe, and it would kill the microbe and leave the person alone. He called it a Zauberkugel — a magic bullet.
His team tested hundreds of arsenic compounds against the organism that causes syphilis. Number 606 worked. Sold from 1910 as Salvarsan, it was the first drug designed on purpose to attack a specific organism, and it was the first real cure for a disease that had been incurable for four centuries.
Ehrlich also gave the industry its working method: make many related compounds, test them all, keep the one that works. Chapter 25.6 shows what that method looks like today when the "many" is two million compounds and a robot does the testing.
Insulin, and the first time a company made a human protein
In 1921 in Toronto, Frederick Banting and Charles Best, working in John Macleod's laboratory with the biochemist James Collip, extracted the pancreatic hormone that controls blood sugar. In January 1922 they gave a purified extract to a fourteen-year-old boy named Leonard Thompson, who was dying of type 1 diabetes. He recovered. Before insulin, that diagnosis in a child meant death within about a year.
Two things about this story matter for your job. The first is that the University of Toronto sold the patent for one dollar and licensed manufacturers to make it, which is the earliest large example of the tension between medicine as a public good and medicine as a product — an argument still running today.
The second is that insulin came from ground-up animal pancreases for the next sixty years. Getting a human protein out of a cow is not a business that scales, and the impurities caused reactions. That constraint is exactly what biotechnology was invented to remove, and Chapter 25.3 picks up the story there.
Penicillin: the accident, and the twelve-year gap
Alexander Fleming returned to his London laboratory in September 1928, looked at a culture plate he had left out, and saw that a mould contaminating it had killed the staphylococci around itself. He identified the mould, named the substance it produced penicillin, published in 1929 — and then largely put it aside, because he could not purify enough of it to be useful.
The gap between that picture and a usable drug is the single most useful lesson in this chapter. Discovery is not development. In 1940 at Oxford, Howard Florey, Ernst Chain and Norman Heatley purified enough penicillin to treat infected mice, then a patient. They still could not make enough. The problem was no longer biology; it was manufacturing.
The solution came from a fermentation laboratory in Peoria, Illinois, and from deep-tank fermentation methods borrowed from industrial chemistry, plus a higher-yielding mould strain famously found on a cantaloupe from a local market. By the invasions of 1944 there was enough penicillin for the wounded of an entire army.
Fleming, Florey and Chain shared the 1945 Nobel Prize. Nobody gave a prize to the fermentation engineers, and this is a pattern you will meet constantly: the science makes the headline, and the process engineering decides whether patients ever get the product. Chapter 25.11 is about exactly that step.
Thalidomide, and the birth of the modern approval system
In 1957 a German company began selling thalidomide as a sedative, and it was soon recommended for morning sickness in pregnancy. It was sold in dozens of countries without prescription. Around 1961 doctors in Germany and Australia connected it to a sudden wave of babies born with severely shortened or absent limbs. Roughly ten thousand children were affected worldwide, and about half survived infancy.
The United States was largely spared, and the reason was one person. Frances Oldham Kelsey, a reviewer newly arrived at the FDA, refused to approve the American application. The company pressed her repeatedly. She kept asking for data she had not been given, particularly on whether the drug crossed the placenta and on reports of nerve damage in long-term users. She simply did not sign, and the disaster reached America only through samples that had been handed out as "investigational" — which at the time required almost no controls.
The Kefauver–Harris Amendments of 1962 followed, and they created the system you work inside today. From that Act came four requirements that had never existed before: a manufacturer must prove the drug is effective, not merely safe, using adequate and well-controlled investigations; investigational drugs may only be given under a controlled application to the FDA; patients must give informed consent to receive an experimental drug; and adverse events must be reported to the agency.
Every one of those is now an industry. Proof of effectiveness became the clinical trial machine of Chapters 25.8 to 25.10. The investigational application became the IND of Chapter 25.7. Informed consent became the ethics and site machinery of Chapter 25.9. Adverse event reporting became pharmacovigilance, Chapter 25.22, which employs tens of thousands of people worldwide and is one of the most commonly outsourced functions of all.
The other kind of disaster: what was done to patients
Safety law came from products that harmed people. Ethics law came from researchers who harmed people, and the two histories are separate.
At the Nuremberg trials after the Second World War, the medical experiments carried out in concentration camps produced the Nuremberg Code of 1947, whose first principle is that voluntary consent of the subject is absolutely essential. The World Medical Association turned that into the Declaration of Helsinki in 1964, still the ethical reference for research involving humans.
Then the United States discovered it had its own case. From 1932 to 1972 the US Public Health Service ran a study in Tuskegee, Alabama, following the natural course of untreated syphilis in about six hundred poor Black men. They were not told they had syphilis. When penicillin became the standard cure in the 1940s, it was withheld from them, and the study continued for a further twenty-five years, ending only when a journalist exposed it.
The National Research Act of 1974 and the Belmont Report of 1979 followed, and they gave American research three governing principles — respect for persons, beneficence, and justice — plus the requirement that an independent committee review any research on humans before it starts. That committee is the Institutional Review Board, or IRB, called an Ethics Committee in most other countries. No trial anywhere in the modern world starts without one, and Chapter 25.9 shows exactly what it approves and what it can stop.
Biotechnology: making the medicine out of living cells
In 1973 Stanley Cohen and Herbert Boyer showed that a piece of DNA could be cut out of one organism, joined into a circular piece of bacterial DNA, and put into a bacterium that would then make the protein that DNA codes for. The mechanism is the one in Chapters 2.2 and 2.3 of this volume; what was new was doing it on purpose across species.
Genentech was founded in 1976 to sell the consequences, and in 1982 human insulin made by bacteria — Humulin — was approved. For the first time a medicine was a human protein produced by an engineered organism rather than a chemical made in a reactor or an extract from an animal.
The second biotechnology wave was antibodies. In 1975 Georges Köhler and César Milstein worked out how to make a line of identical cells producing one identical antibody, and in 1986 the first such monoclonal antibody was approved as a drug. Early ones were mouse proteins and the human immune system attacked them; the fix was to engineer them to look progressively more human, which is why the names tell you what you are getting. A generic name ending in -ximab is part mouse, -zumab is mostly human, and -umab is fully human. Rituximab in 1997 and trastuzumab in 1998 proved these could treat cancer by hitting one specific molecular target.
This split the industry into two technical worlds that persist to this day. A small molecule is a chemical you can draw and synthesise. A biologic is a large protein made by living cells, which cannot be exactly copied, only made similar — the reason Chapter 25.15 has to treat generics and biosimilars as different legal animals.
The last forty years, compressed
A few laws and a few products carried the rest of the story, and each one left a permanent structure behind.
| Year | What happened | What it left behind |
|---|---|---|
| 1983 | Orphan Drug Act | Incentives for rare disease |
| 1984 | Hatch–Waxman Act | The modern generics industry |
| 1987–96 | AZT, then combination HIV therapy | Accelerated approval, patient activism |
| 1992 | Prescription Drug User Fee Act | Industry-funded review, real deadlines |
| 2001 | Imatinib for leukaemia | Targeted therapy proven |
| 2003 | Human Genome Project completed | Genomic target finding |
| 2004 | Vioxx withdrawal | Modern post-market safety powers |
| 2011–14 | Checkpoint inhibitors | Immunotherapy as a class |
| 2017 | First CAR-T cell therapy | Living cells as a product |
| 2020 | mRNA COVID-19 vaccines | Platform manufacturing at speed |
| 2023 | First CRISPR gene-editing therapy | Editing the patient's own genome |
Three of those rows deserve a sentence each, because they explain behaviour you will see at clients.
Hatch–Waxman in 1984 is why cheap medicines exist. It let a generic company win approval by proving its copy behaves the same in the body as the original, instead of repeating the whole clinical programme, and in exchange it gave the original manufacturer extra years of protection. That bargain is the reason roughly nine out of ten American prescriptions are filled with generics while most of the spending goes to brands.
The user fee act of 1992 is why review has deadlines. Companies pay fees, and in return the agency commits to review goals — currently around ten months for a standard new drug application and six months for a priority one, counted from the point the application is accepted. When a client says "our PDUFA date is in March", they mean the date the FDA has committed to decide.
The withdrawal of the painkiller rofecoxib in 2004, after evidence of increased heart attacks, is why post-market safety is now a legal system rather than a courtesy. The 2007 amendments gave the FDA power to demand studies after approval and to require formal risk management programmes. Chapter 25.22 covers the machinery that resulted.
What the history is actually telling you
Look back at the pattern. In every case, a rule you may find annoying is a scar.
Batch records exist because a solvent nobody documented killed a hundred and five people. Efficacy trials exist because a sedative given to pregnant women deformed ten thousand infants. Informed consent and ethics committees exist because researchers, including well-intentioned ones, treated people as material. Audit trails exist because data has been falsified. None of this was invented by bureaucrats in a vacuum, and saying so in a client meeting will earn you more credibility than any amount of technical fluency.
Next: Chapter 25.3, the branches of the industry — what actually separates a pharmaceutical company from a biotechnology company from a device company, and why the difference decides which rulebook applies to your project.