At the IFPAC 2010 annual meeting in Baltimore, I spoke with many vendors and FDA-types on the subject of continuous processing. Everyone, even seasoned experts, seems to have a different way of defining the term, and skeptics still doubt that continuous batches can be executed well and affordably, and gain regulatory approval, in our industry.
Efforts are underway to apply continuous processing to small-molecule APIs, although production of biological actives is likely to remain a batch process for some time.
But where continuous processing could have the greatest impact is on the physical steps that account for the majority of time and effort spent on any pharmaceutical process: weighing, mixing (blending), granulating, drying, compressing, and coating. Since each of these steps is independent and carried out in a different container, it always seemed logical to move materials from one location to another over the course of days or weeks.
But it was engineering and regulatory limitations that drove us to take this approach. In the past, there were virtually no on-line measurements, apart from weighing and metal-finding, and there was no impetus to design the necessary equipment because the industry was already realizing solid profits.
Besides, cGMPs lent themselves to batches (i.e., “minimum of three production lots to submit an NDA”), and even HPLC was discouraged as late as the early 1970’s, so other “different” technologies would certainly be shunned. In addition, there were fewer inspectors (who only needed a “degree in science”), which wasn’t a recipe for innovation in the field.
Times have changed. FDA already published its PAT guidance and 21st century GMP’s to advance the concept, but the Agency’s leaders continue to support continuous processing, and to encourage its use.
With the concept of a process signature and the idea that, under PAT/QbD “each batch is a validation batch,” the old idea of three process batches takes on new meaning. Perhaps, a company need only show that a continuous process (over several times the volume of what was considered a “lot”) matches the process signature of a “single batch” that meets all bioavailability, content uniformity, and other parameters to be deemed equivalent.
The biggest hang-up that has been expressed is, “how do we define a batch in a continuous process?” The definition hinges on a process signature and time. All batches, great and small, will have to conform to a process signature, be produced within a design space, and any single dose, at any point in the process, be able to meet the product specifications.
As far as designation, simply place a time code within the batch code. This way, since we have continuous measurements stored electronically, we can match the product signature to the period of time in question for any portion of any lot.
Mechanically, the hardware is available to weigh large amounts of material with great precision and newer processing equipment should be designed to run for many days without need of “cooling off” or cleaning. One immediate time saving will be on cleaning and validation: when you make the equivalent of 50 of today’s batches non-stop, there are 49 fewer clean-ups and cleaning validations needed to be run. Another saving will be in the number of batch records maintained; there might well be only three or four lots a month, where formerly there were 30 or 40.
All this is predicated upon the implementation of true QbD, of course. There will need to be subtle and continuous “corrections” throughout a run to adjust for differences in raw materials and intermediate steps. These need to be accounted for even in single, discrete batches if QbD is to be achieved, so a continuous accounting for these variances shouldn’t be much of a stretch for a process engineering team.
There were some very nice weighing/measuring devices on exhibit at IFPAC as well as small, faster and, yes, cheaper measuring devices for on-line analysis. In addition, there are some very impressive software packages for computation and storage of all these data points over a many-day run. We also are in an age where storage is pennies per gigabyte, where formerly it was dollars.
As the regulatory and technical obstacles to continuous processing are eliminated, there’s almost no reason why we can’t expect to see FDA or EMEA approve the first continuous processes, very soon.
