The fine and specialty chemicals business: how it really works: Business models and how deals are made

4.1 Custom manufacturing and API-for-generics share plants but are different businesses

In custom manufacturing (CM), which Pollak calls the industry's premier discipline, a drug or crop protection company hands process development, pilot production and industrial manufacture of an active ingredient or precursor to one or a few fine chemical companies. The customer owns the product and usually the process; the relationship runs under an exclusive supply agreement. The customer provides a 'technology package', at its simplest a laboratory recipe plus safety notes, and the supplier scales it up about a million-fold, from 10 g to 10 tonnes. Pollak calls such suppliers contract manufacturing organisations (CMOs).

Toll manufacturing is a variant: the customer, often an agrochemical company, supplies the starting materials free and a full industrial process, so the toller mainly sells reactor time, with revenue typically under half that of CM for the same product.

API-for-generics is a product business. The supplier develops its own process for an off-patent drug substance, files the regulatory dossier and sells to many generic drug makers at market prices, often through agents.

Contract research organisations (CROs) sit upstream. 'Product' CROs do process research and make samples, charging per full-time-equivalent scientist-year (FTE); 'patient' CROs run clinical trials. Firms combining research and manufacturing are called CRAMS or one-stop shops. Pollak is sceptical: over 90% of projects stop at the sample stage, and quick lab work and economical large-scale production are different skills.

Combining CM with generics can lift utilisation, since generic campaigns can fill gaps between rigid exclusive ones. One rule is absolute: never sell a generic company an API you make for an originator.

• Pollak (2011) · pdf pp. 30, 37-42, 168-172, 177

4.2 From first inquiry to commercial supply takes one to two years of project work

A drug offers a fine chemical company four 'windows' over its roughly 20-year patent life. The first, preclinical samples, is CRO territory with slim odds. The second, Phases II and III, is the most attractive: the customer freezes its first-generation route, takes its make-or-buy decision, and launch becomes likely. Drug firms usually give such projects only to approved suppliers, so newcomers must enter at Phase I. The third window opens near peak sales, when a cheaper second-generation process is wanted; the fourth as the patent nears expiry.

The customer selects suppliers in four steps: desk assessment, a request for information, audits of a shortlist, and a request for product. Inside the supplier, a new product committee asks Pollak's three questions: can we sell it, can we make it, will we make money? The customer's technology package should give the reaction scheme, product, quantities, dates, specifications, analytical methods, required reports, safety issues, materials it will supply, and packing instructions.

Pollak's model schedule follows the flowchart. Technical work usually takes 12-24 months and can cost several million dollars. The pilot plant uses vessels of 100-2,500 litres and includes a hazard and operability (HAZOP) study, a structured review of what could go wrong.

For cGMP products the process itself must be validated: designed, qualified and continually verified. Plants are inspected by national authorities and, for the US market, by the FDA, whose findings range from 'no action indicated' to 'official action indicated', which can lead to a warning letter or seizure. For APIs sold to many generic customers, the producer files a drug master file (DMF) with the FDA once; that dossier then supports any generic drug using the API.

• Pollak (2011) · pdf pp. 86-87, 90-91, 94-96, 127, 164-170, 179-180, 269-270

4.3 Prices are built bottom-up for exclusives and top-down for generics

Pollak describes three pricing methods. Research and samples are priced by time, per FTE, often with milestone payments tied to yield or quality targets. Exclusive products at industrial scale are priced 'bottom-up': raw materials (about 35% of price) plus conversion (about 55%) plus a margin. Standard products such as API-for-generics are priced 'top-down': start from the market price and subtract margin, conversion and raw materials; if the answer is negative, costs must fall or there is no business.

The buyer sees the molecule differently. For a drug company the API is a raw material worth under 10% of the medicine's price, which Pollak splits into cost of goods about 20%, R&D 15-20%, marketing and administration 35-40% and profit about 25%. Drug prices follow the daily cost of treatment within a class, not the molecule's cost. Low-dose blood-pressure drugs cost about $1-1.40 a day whatever their sales: beta-blockers $1.00, ACE inhibitors $1.05, calcium antagonists $1.28, 'sartans' $1.40. A pharmacy charged the same $419.95 for 100 Lipitor tablets of 20, 40 or 80 mg. Suppliers have no reference price, and for high-dose, complex molecules the gap between buyer and seller can be wide.

Margins range from not covering fixed costs to markups of 100% or more. Cheap Asian quotes should be compared on total cost of ownership: freight, audits, registration and supply risk. Pollak's trade-off: a $1 billion drug out of stock for 10 days loses about $30 million of sales, against a few percent saved on the API.

Contracts reflect power. CM supply contracts typically run 3-10 years and can involve hundreds of millions of dollars. Around 2000 the market turned from sellers' to buyers'; GlaxoSmithKline told suppliers to commit to cutting every price by 20% over three years.

• Pollak (2011) · pdf pp. 178, 184-189, 191-194, 239

4.4 Patent expiry turns a drug into a price war within a year

A drug is protected by a product patent and often by process patents on how the API is made. Pollak says product patents typically expire 18-20 years after filing and 8-12 years after launch; Kent describes US patents as giving 20 years of exclusive rights. Then anyone may sell a generic: same API, same strength, bioequivalent, under another name and, if a process patent still runs, by a different route.

Prices fall fast. API prices dropped in the first year to 45-75% of their pre-expiry level. For drugs that had sold over $400 million a year they fell to about 30% after five years and 20% after ten; smaller drugs fell less. The money is made in the first months, by whoever is ready on day one: in 2002 Barr shipped 100 million generic Prozac capsules within 48 hours of approval.

Pollak lists six requirements: picking the right molecules among more than 2,000 APIs (avoiding both blockbusters and tiny drugs, preferring those needing a niche technology); being first to file; knowing patent law (180-day US exclusivity for the first challenger, the Roche/Bolar provision allowing development before expiry, European certificates adding about five years); being the low-cost producer; a GMP culture able to produce a successful abbreviated new drug application; and financial muscle, since work starts 5-8 years before expiry.

National patent history shaped who makes generics. Italy introduced product patents only in 1978 and Spain in 1992, making both strongholds of API-for-generics; China and India joined international patent rules in 2005. Pollak's example in the table shows why chasing a blockbuster's expiry is a poor bet.

• Pollak (2011) · pdf pp. 134-135, 171-177, 223, 237-239
• Kent (2012), Ch 11 Chemistry in the Pharmaceutical Industry · pdf p. 405

Check your understanding

A fine chemical company is asked to quote for an exclusive advanced intermediate for a Phase III drug. Which pricing logic applies?

Why do pharma buyers and fine chemical suppliers often disagree sharply on the price of an API for a high-dose drug?

In Pollak's example, why is the API for Femara (US no. 101) a better generic target than Lipitor (no. 1)?