My understanding is for blending of batches you typically need to be certain that all the batches that you’re going to blend will all meet the quality criteria that are appropriate for that material. And then you’re allowed to blend them. You can’t, if you will, blend away a certain level of impurity by averaging. With pooling of chromatography fractions though, the fractions aren’t equal to begin with; they’re not supposed to be equal. They are different as they are different cuts of the material as it comes off of the column, and the fractions that meet the criteria are allowed to be pooled but they would all have to be tested. The central issue is reproducibly meeting prospectively defined cutting criteria to fully meet the purpose of that particular step where it’s a polishing step at the end or a capture step at the beginning or things like this. Often in development, we’ll take fractions, characterize the process and never take fractions again, but do in-line cutting based on OD or on other criteria, and the central issue here is setting prospectively defined cutting criteria that allow you to reproducibly meet the purpose of that step and then to continue to do that over and over and over in a consistently monitored setting. Again, sometimes you take fractions, sometimes you don’t. Sometimes you cut by OD based on concentrations, sometimes there’s a different co-eluting chromaphor. That is really the essence of reproducibly controlled chromatography. It is important to establish the relationship between the OD profile and whatever else you’re actually going to be measuring during processing and product quality. That is usually figured out in development, and if not, then you usually end up collecting a bunch of fractions in manufacturing and trying to figure out how to pool them. If you’ve been able to establish these relationships, then it’s usually much easier to use OD or some very simple surrogate to give you a product that will consistently meet quality attributes.
