Anyone who has set foot in a pharmaceutical plant can’t fail to be impressed by all the engineering that goes into drug manufacturing. Materials flow, automated equipment, all is carefully ordered.
Yet, to an engineer from any other industry, pharma would appear to be out of control.
The variability and wide control specs in drug manufacturing are no secrets. Former FDA Commissioner Mark McClellan remarked on it years ago when he famously cited potato chip and soap manufacturers as being far more advanced.
Consider the fact that active ingredient loading can vary by 10% for the same medication (or by up to 20% for generics, which now account for most of the prescriptions issued in the U.S.), and that problems can result from manufacturing glitches in timed-release formulas.
As expert observers often point out, if paint were produced this way, one would achieve a slightly different tint each time one made a batch. The color swatch would be a joke, and, at stores everywhere, hordes of angry consumers would demand their money back.
In other industries, the stakes are much higher. If petrochemical plants were controlled as pharma plants are today, the earth would be as pockmarked with craters as the moon, from plant explosions, says one consultant.
Pharmaceutical plants don’t explode each time a patient suffers an adverse effect from a drug, but subtle differences in API or other ingredient loading or quality can be a matter of life or death. FDA and industry are just beginning to connect the dots between manufacturing issues and adverse patient responses, particularly in the area of generic drugs.
So, what is the essential philosophical difference between the way other industries control processes, and the way pharma does? Last month, PhD chemist Ali Afnan, formerly with FDA’s PAT team, distilled the answer, which boils down to whether the focus is on product or process parameters. Where other industries look at the physical material attributes of the product throughout production, and use them to set specs for process control, pharma sets controls based on process parameters. So, where the chemical industry would use a material’s moisture content as the measurement to control drying, pharma measures inlet and outlet dryer temperatures, and airflow, and then, when a critical point is reached, stops processes, pulls samples and tests.
Is it any wonder then that most chemical processes are self-validating and self-calibrating, where pharma has needed not only cGMP’s but two rounds of guidance on validation?
Perhaps pharma can’t move to the chem model completely. But it can get closer. The Quality by Design guidance document promotes this way of thinking by getting people to focus on what is critical to product quality, and to connect that to critical process parameters. But today, there are still gaps and disconnects between what QbD espouses and daily workplace realities.
Let’s consider something rather basic: excipients, the source of several major adverse patient responses and quality problems over the years. Today, information about the physical properties of excipients is either not available, or it’s not uniform. For example, density can currently be measured in several different ways. In addition, the way that powders flow depends directly on compression.
Formulation scientists thus have a real problem when trying to decide on an excipient to use with a certain process and product. As Prabir Basu, head of the National Institute for Pharmaceutical Technology and Education (NIPTE) points out, engineers and practitioners in other industries all have databases of physical material properties that they can use to design processes. For generics manufacturers, who may be formulating 200 products at a time, compared with five to 20 for the typical Big Pharma company, this information is critical.
Two years ago, FDA began to fund a NIPTE project to develop a database for properties of pharmaceutical excipients to serve as a tool for formulation scientists. Last month, at Purdue, representatives from industry and academia, including key people from IPEC, the excipient manufacturers’ trade group, met to discuss the progress that has been made with this database. You can view the work in progress here.
The goal will be, not just to compile a list of dry data, but to relate it to product performance data. Projects like this would seem to be essential if the industry is truly to embrace QbD. What do you think?
As NIPTE’s Basu puts it, “When Kennedy made the statement that we’d have a human being on the moon, people thought it was crazy. But unless we set a goal we will never get there.” It seems ridiculous that, at this point, we can send unmanned vehicles to Mars, yet can’t control excipients in our drugs, he says. Eventually, this type of data could also be used for API’s, bringing pharma closer to a product-centric control model.
Sound far-fetched? What do you think? Please write in with any opinions or criticisms.
