Less is More in API Process Development

June 22, 2005
API manufacturers in India and China have grasped the need for simplicity, while U.S. process chemists continue along their costly and complex path. Girish Malhotra shares recommendations for improved API process development.

Not only unvarying quality standards, but also profitability and elegance should drive pharmaceutical process development. Any new manufacturing process should be as profitable and simple as possible, and anything used in that process should be easy to use and execute. These rules should hold whether a manufacturing process is batch, semi-continuous or continuous.

So why does the U.S. pharmaceutical industry persist in using complex manufacturing processes to make active pharmaceutical ingredients (APIs)? Consider patents involving API process chemistry: typically, each intermediate must be isolated, a cumbersome and costly process.

In a process involving multiple intermediates, more than one solvent is usually required, and reactions must be carried out for 24 to 48 hours or even longer. In some cases, this is unavoidable, but it always introduces complexity and the need for additional steps — for example, brine washing to facilitate phase separation, the use of sulfates for drying, vacuum stripping of solvents. All the while, reaction progress is measured by HPLC, NMR or TLC — fine techniques all, but expensive and time consuming. This approach is fine for laboratory synthesis but can’t be the rule for commercial manufacturing — at least, not for any process that is to be commercially viable in today’s market.

Out of necessity, API manufacturers in India and China have grasped the need for simplicity, and we’ve seen the results: generic therapies that cost a small fraction of the price for a nongeneric U.S. drug. While API manufacturing in the U.S. continues along this costly path, new API manufacturers in India and China are nimbly developing process chemistries that also lend themselves easily to analysis via process analytical technology (PAT) and to more advanced process control.

What will change this picture? The answer is simple: Less complex laboratory synthesis processes that can be scaled up easily once drug efficacy has been demonstrated. Simple processes also allow simple process control methods.

A review of pharmaceutical chemistry patents shows a “Rube Goldberg” approach, and a large number of unnecessarily complex processes. Typically, three to five steps are required to prepare each intermediate. Reactants are added over time, extending reaction time. The result: an extended time batch process.

There is nothing wrong with the chemistries themselves, but they stand in the way of process modernization and translate into higher costs to the consumer.

The following questions should guide API development efforts, from their very earliest stages, to ensure a final process that is as elegant, cost-efficient, and “controllable” as possible:

  1. Can each reaction step be completed in minimum time? If not, how can time and costs be reduced? This is a very challenging question, but it must be addressed thoroughly, from the very first step.

  2. How will reaction kinetics affect the total reaction time? Kinetics must be evaluated carefully since data will be critical to optimizing the commercial process.

  3. Are you selecting the best solvents for the process? Using solvents that offer a maximum density difference between organic and aqueous phases can eliminate the need for brine washing.

  4. Will the intermediate require isolation?

  5. Can the same solvent be used throughout the process? This is the ideal situation, but if it’s not possible, can the total number of different solvents be minimized? This will have repercussions both for solvent recovery and disposal.

  6. Are you replicating commercial conditions in the development process? Can each reaction step be translated easily to an executable unit operation? Lab glassware and configurations are excellent for synthesis but don’t represent reality.

  7. Is the process such that it delivers quality product rather than quality is achieved by testing the product?

Of course, laboratory methods can’t fully simulate plant conditions. Sophisticated control methods aren’t practical for use in the lab. Nevertheless, there are steps that chemists and engineers can take to apply the best available technology and methods.

As an example, consider amine diazotization, which is typically conducted at low temperatures with ice — a safe but inefficient method that leads to disposal issues. This reaction could instead be conducted safely at room temperature.

Let us analyze each alternative: first, the traditional method, in which an amine is added to water, and sodium nitrite and a mineral acid are added to complete the reaction at temperatures of 0° C or lower. The ice and excess water are needed to control this exothermic reaction.

But consider the historical reasons for using this process. When this reaction was first developed, ice and excess water were the best and cheapest heat sink available. The plants practicing these technologies were usually located near rivers where plenty of free cold water was available. The effluent water after use was treated with the best available technology and sent back to the river.

Now let’s consider the alternative: conducting this reaction at room temperature. This approach could be safely taken if one follows the reaction chemistry along with physical chemistry. A back mix type reactor with stoichiometry controls can be used to control the exotherm and conduct the reaction safely. Although an alternative is possible, most recent patents still describe the older method.

But sometimes the older method can be the most economical. Consider the choice between chemical reduction, typically used to convert a nitro to an amino group, and catalytic hydrogenation. Chemical reduction leads to byproducts that must be treated before disposal, and it’s generally a messy process. Catalytic hydrogenation is much cleaner and is now the method of choice in developed countries.

However, in developing nations, where catalytic hydrogenation was too expensive, manufacturers have instead optimized chemical reduction—and they’ve recovered, purified and sold byproducts, leading to processes that are more economical.

Let’s review some recent API process patents, and the potential cost-reduction opportunities for each:

U.S. Patent # 6,037,483

This patent describes preparation of 3-Bromomethyl-3-methyloxetane. The process is complex: Preparation of oxetane is completed over two days and involves isolation of an intermediate and use of a solvent. A close examination of chemistry and reaction kinetics suggest that it is possible to conduct the reaction in about three hours without the use of the solvent and isolation of the intermediate. The yield is over 85% and productivity much higher compared to the process described. If the process as described in the patent is commercialized, it will definitely be expensive.

U.S. Patent #6,245,913

This is the synthetic procedure for 5-methoxy-2-[(4-methoxy-3,5-dimethyl-2-pyridinyl)-methylthio]-IH-benzimidazole hydrochloride and its conversion to omeprazole.

This patent has simplified the process from previous chemistries, but, still, 11 major steps are needed to produce the product. There are isolation and purification steps in each of the eleven steps. Organic chemists and engineers are very creative in minimizing and/or eliminating isolation. Their creativity has to be tested as it will lead to a simpler process that can be controlled to minimize reaction time. Different solvents are used in the process, adding to the overall complexity. I believe that there are simplification opportunities for each intermediate product on commercial scale. These opportunities come from reaction time and work up procedures. Is the isolation of the intermediates necessary? My estimate is that the overall process yield from the starting raw material is about 20-25%. This presents a significant opportunity to increase yield, which can add substantial savings to the bottom line.

U.S. Patent #6,867,306

The present invention discusses a novel process for the synthesis of [R-(R*,R*)]-2-(4-fluorophenyl)-B,D-dihydroxy-5-(1-methylethyl)-3-phenyl-4- [(phenylamino)carbonyl]-1H-pyrrole-1-heptanoic acid hemi calcium, atorvastatin.

This patent relates to statin synthesis and claims to be “a new, improved, economical and commercially feasible method.” The patent has nine major steps. Within each step, there are four to ten steps. These add to the processing complexity and cGMP becomes an operating challenge. It is my conjecture that the process as described if commercialized would take about one week to make a batch.

Compare these economics to those of the Chinese fermentation process. It is necessary to simplify the process to minimal steps so that isolation, if necessary, is easy and the product cost is significantly less than the comparable product made anywhere in the world. [Reference: UBS Investment Research October 8, 2004]

U.S. Patent # 6,835,848

The present invention relates to a new and industrially advantageous one-pot process for the preparation of alkyl 3-cyclopropyl amino-2-[2,4-dibromo-3-(difluoromethoxy) benzoyl]-2-propenoates of in which R represents methyl or ethyl, which are valuable intermediates for the production of highly active antibacterial quinolone medicaments.

The title of the patent calls for “One Pot” Synthesis. However, six steps are carried out in that pot over two days. Although they are all simple steps and could lead to a manufacturing-friendly process, it could be possible to shorten the time required significantly, and even move from batch to continuous production. The process chemistry as it is described cannot be completed in one pot unless that pot is used repeatedly. It needs simplification to have simple process.

U.S. Patent # 6,331,638

A process for the preparation of the esters of 1,8-disubstituted-1,3,4,9-tetrahydropyrano (3,4-b)-indole-1-acetic acid is delineated.

Although this seems simple, synthesis seems simple but takes about ten to twelve hours to complete the reaction. Batch process time of 1-2 hours or a continuous process would make the total process economics extremely promising. I believe such time reduction may be possible for this process. These may be possible by a careful review of the reaction stoichiometry and kinetics.

Reaction completion and purity are measured using NMR, HPLC, TLC or other sophisticated analytical methods. These prolong the cycle time. For commercial operation, simple test methods that give the desired test result would be helpful.

To reiterate, anyone embarking on API process development needs to ask some hard-hitting questions as early as possible in the process, and to ensure that they are answered: In a nutshell: “Is this process as inexpensive, profitable, safe and environmentally friendly as it can be?”

Development specialists need to put themselves in operators’ shoes: Ask yourselves: “If I had to operate the process in a manufacturing environment, how would I feel? Is this really the best that I can do?”

This approach will simplify process control, via PAT, eliminate rework, slash inventories, and result in “right first time” product quality.

As in art, in process development, less is more. In today’s global economy, U.S. scientists and engineers are competing with the best chemists and engineers of every country in the world, whose work is already guided by this principle.

Cost efficient methods will allow manufacturers to embrace more advanced process control, facilitating cGMP compliance and, perhaps most importantly, reducing product costs to the ultimate consumer.

About the Author

Girish Malhotra | President