Pharmaceuticals and fermentation: from molecule to medicine: The pharmaceutical value chain

1.1 Five stages turn a molecule into a medicine

The chain starts in discovery. Medicinal chemists design and make candidate molecules, usually small organic compounds of molecular weight about 200 to 600, and biologists test them against a target such as an enzyme or receptor. Kent notes that only one in 10,000 to 20,000 compounds made as candidates ever reaches testing in humans, and more than 80% of those then fail.

Development follows: animal safety studies, then three phases of clinical trials, then a filing to a regulator such as the US Food and Drug Administration (FDA). While trials run, process chemists redesign the laboratory route so that it can be run in tonnes. The medicinal chemist rarely worries about the cost or toxicity of reagents; the process chemist must, and often invents an entirely new route.

The third stage is making the active pharmaceutical ingredient (API), the molecule that does the work. Pollak calls this primary manufacturing. It is fine chemistry: multistep batch synthesis or fermentation, run under strict quality rules. Most of this course module is about this stage.

The fourth stage, secondary manufacturing, is formulation. The API is mixed with excipients, inactive ingredients that give bulk, stability, colour and taste, and made into a tablet, capsule, liquid or injection. The fifth is packaging and labelling, after which the product is sold as a specialty: a branded drug or a generic.

The economic point is that value is added very unevenly along this chain. A kilogram of API may be worth a few thousand dollars, while the same kilogram in tablets in a pharmacy is worth tens of thousands. The tiles "Small-molecule APIs", "Pharma starting materials and intermediates" and "Pharmaceutical excipients" sit at different points on this chain.

• Kent (2012), Ch 11 Chemistry in the Pharmaceutical Industry · pdf pp. 401-404
• Pollak (2011), Fine Chemicals · pdf pp. 27-29

1.2 Discovery and trials, not chemistry, set the price of a new drug

A new drug is expensive mainly because most candidates fail, and the cost of the failures is carried by the few that succeed. Kent describes the sequence. Phase I trials in healthy volunteers check safety and dose. Phase II, in up to several hundred patients over 6 months to 2 years, tests whether the drug works; only about one third pass. Phase III, in several hundred to several thousand patients over 1 to more than 3 years, confirms effect and long-term safety. Only 20 to 25% of drugs that enter Phase I come through all three phases. The FDA's standard review then takes 10 months.

The cost of this has risen sharply over the decades that the sources span. Shreve, writing in 1984, gives more than seven years and $8 to $10 million per drug. Dryden, in 1997, gives an average of $250 million and 12 years in the United States. Kent, in 2012, puts it above $800 million, including the cost of failed projects, over 10 or more years.

Patents are what make this investment rational. A composition-of-matter patent gives the holder exclusive rights to the molecule for 20 years from filing. But filing happens early, long before trials end, so much of that life is used up before launch. Shreve's 1980 example is stark: of 12 new chemical entities approved that year, the longest remaining patent life was 11 years 10 months, and one had none left.

For a manufacturer of APIs this matters in two ways. While a drug is patented, its maker cares most about reliable supply and quality, not the last dollar of API cost. Once the patent expires, generic competition turns the API into a price-driven product, and process cost becomes everything.

• Kent (2012), Ch 11 Chemistry in the Pharmaceutical Industry · pdf pp. 401, 405-406
• Shreve (1984), Ch 40 Pharmaceutical Industry · pdf p. 814
• Dryden (1997), IVH Pharmaceutical Industry · pdf pp. 566-567
• Kent (2012), Ch 30 Industrial Biotechnology: Discovery to Delivery · pdf p. 1180

1.3 Small tonnage, high value: the shape of the industry

Pharmaceutical manufacturing is small in tonnes and large in value. Shreve reports that US production of bulk medicinal chemicals was 65,000 t in 1964 and 110,000 t in 1980, with 1980 bulk sales of $1,153 million. That is about $10 per kg on average, and the figure excludes the tablets, capsules and ampoules made from those bulk chemicals.

Shreve also observes that pharmaceuticals use a greater variety of complicated steps than almost any other process industry, and place more emphasis on purity than any sector except nuclear materials. Life processes, meaning fermentation and cell culture, make a large share of the products: antibiotics, vitamins, hormones and biologicals. Dryden adds that by the 1990s more than half of US pharmaceutical production was by biochemical rather than synthetic routes.

India's industry, as Dryden describes it in the mid-1990s, was large in number of units and small in money. There were over 15,000 drug manufacturing units, mostly small, with about 250 larger units producing nearly 70% of output. Turnover was about Rs 5,700 crore in 1994, against Rs 20,000 crore for iron and steel. India made 1.6% of world pharmaceutical output with 15% of world population. Antibiotics were more than half the value of drugs sold in India.

Two features stand out. First, Indian law then recognised only process patents, not product patents. Indian companies could legally invent a new route to a drug patented abroad and sell it at a fraction of the originator's price. Second, Indian drug prices were well below world prices, the reverse of bulk chemicals, where Indian prices were above world levels.

• Shreve (1984), Ch 40 Pharmaceutical Industry · pdf pp. 814, 817-818
• Dryden (1997), IVH Pharmaceutical Industry · pdf pp. 564-569

See it in the table

Check your understanding

Using Pollak's Lipitor figures, roughly what share of the pharmacy price of the tablets is the API?

Why did Indian companies in the 1990s make drugs that were still patented abroad?

According to Kent, what share of drugs entering Phase I clinical trials come through all three phases?