The fine and specialty chemicals business: how it really works: Products, customers and the chemist's toolbox

2.1 Four layers of product serve a few large, risk-averse customers

Pollak sorts fine chemicals by the value they carry. Building blocks, or standard catalogue products, sit at the bottom: acetoacetates, alkyl halides, picolines, often made by large chemical firms extending their own chains, such as methanol to acetic acid to diketene to acetoacetates. Many cost under $5-10/kg and behave like commodities. Next come nonregulated intermediates, then advanced or 'regulated' intermediates, the last steps before the drug substance, made under cGMP. At the top is the API.

Two chemical features dominate. Most modern actives are built round nitrogen-containing rings (N-heterocycles such as pyridines, imidazoles and tetrazoles): four of the top five drugs and five of the top ten agrochemicals. And about half of APIs are single enantiomers. An enantiomer is one of two mirror-image forms of a molecule, like left and right hands; often only one works, so only that one may be made.

The customers are concentrated. Pollak counts fewer than 20 'big pharma' firms with sales above $5 billion, 50-100 'medium pharma' firms, mostly generics, and 4,000-5,000 'virtual' venture-funded firms with no plants. Agrochemicals come second: the top ten held almost 95% of a $44 billion pesticide market in 2008, about 70% of it generics. Animal health, $19.2 billion in 2008, is third.

The drug customer's economics drive everything. Pollak gives $1.4 billion of development cost per new drug and 5,000-10,000 compounds made for each launch. Kent adds that only 1 in 10,000-20,000 candidates reaches human testing, over 80% of those fail, and 20-25% pass Phases I-III. A supplier attached to a drug early is usually attached to a failure.

• Pollak (2011) · pdf pp. 48-49, 128-129, 136-141, 177-178, 244-245
• Kent (2012), Ch 11 Chemistry in the Pharmaceutical Industry · pdf pp. 401, 406

2.2 Most of the chemistry runs in standard kit; the edge lies in a few hard reactions

A drug synthesis is a chain of reactions, each turning one intermediate into the next. Pollak's examples run from two steps for paracetamol (partial hydrogenation of nitrobenzene, then acetylation) to seven for the ulcer drug omeprazole.

A trade-show directory lists more than 150 reaction types offered by fine chemical firms, and the 430 companies surveyed each claimed about 30. The commonest are amination (adding nitrogen), condensation (joining molecules with loss of water), esterification, Friedel-Crafts alkylation, Grignard reactions (carbon-carbon bonds made with organomagnesium reagents), halogenation, hydrogenation and reduction. Pollak's verdict: mastering these is no competitive advantage, because nearly all run in standard multipurpose plants, and add-ons such as ozone or phosgene generators can be bought off the shelf. Exceptions are organometallic reactions at down to -100 °C, needing cryogenic units cooled by liquid nitrogen, and dangerous gases (phosgene, hydrogen cyanide, diazomethane) that risk-averse drug firms prefer to hand out.

Chirality is a real skill. Either both mirror forms are made and separated, by crystallisation with a chiral helper or by chromatography such as simulated moving bed (SMB) units, or only the wanted form is made with a chiral catalyst. In Kent's aprepitant example, one chiral reduction (91% enantiomeric excess, over 98% yield) sets all three chiral centres, for an overall yield of 54%.

Microreactors, continuous channels with excellent heat transfer, are the first new reactor concept since Perkin's stirred tank of 1857. Experts quoted by Pollak think 70% of reactions could run in them but only 10-15% would pay; DSM's cGMP line for naproxcinod has 16 tonnes a year of capacity. The barrier is economic: a new line is hard to justify beside an idle, depreciated batch reactor.

• Pollak (2011) · pdf pp. 54-60, 91, 254
• Kent (2012), Ch 11 Chemistry in the Pharmaceutical Industry · pdf pp. 403-404, 425-427

2.3 Biotechnology cuts steps and cost, but big molecules are a different business

Biotechnology enters fine chemicals in three ways. Biocatalysis uses enzymes or whole cells as catalysts within a chemical route. Enzymes work near room temperature and neutral pH in water and are highly selective, so they can skip the 'protection and deprotection' steps a chemist uses to stop other parts of a molecule reacting. Immobilised enzymes can be filtered out and reused in ordinary plant. About two thirds of chiral products made at large scale already use biocatalysis.

The cost effects in Pollak's cases are large. An enzymatic route to the key intermediate of AstraZeneca's Crestor cut its cost by 70%. DSM's enzymatic process for the heart drug diltiazem cut cost of goods by 40%, with ten times the throughput. Enzymes shortened the steroid dexamethasone synthesis from 28 steps to 15. A classical separation of mirror forms with a chiral salt can never yield more than 50% of the wanted form; a selective enzyme can approach full yield.

Fermentation, in which microbes turn sugar into product, replaced a six- to eight-step route to vitamin B2 (riboflavin) with one step, cutting waste by 95% and cost by about half. Because productivity per litre is low, fermenters are large (above 250 m3) and dedicated.

Mammalian cell culture, used for proteins and antibodies, is a different industry. The cells grow slowly, contaminate easily and give a few grams per litre after about 20 days; a 10,000-litre fermenter yields only a few kilograms of an antibody. Only a handful of fine chemical firms, notably Lonza and Boehringer-Ingelheim, run such plants. The table compares the two technologies.

• Pollak (2011) · pdf pp. 60-70, 215

See it in the table

Check your understanding

According to Pollak, why does skill in common reactions such as Grignard, hydrogenation or esterification give little competitive advantage?

A chemical resolution of a racemic intermediate with a chiral salt is replaced by an enzymatic step. What is the main economic gain?

Why can mammalian cell culture plants not simply share multipurpose chemical plants?