OR WAIT null SECS
In 1980, Atwood and Goldstein of PerkinElmer published one of the earliest studies on lot-to-lot selectivity variability across 24 lots of a commercial high-performance liquid chromatography (HPLC) packing material (1). In mid-August of 2020, I was asked by a major column manufacturer to participate in a focus group on a proposed new quality control (QC) procedure for their LC columns. I guess, even after 40 years, we have yet to find a perfect solution to the problem of demonstrating and achieving acceptable column precision.
During the conversation, I made an off-hand comment that most end users probably can’t remember the last time they looked at the paperwork inside a column box. Indeed, many vendors are moving toward minimal packaging and inserts, whether motivated by cost-saving measures or environmental concerns. As long as the information is easily retrievable, either from the internet or embedded in the electronic data chips leashed to some vendors’ columns, this seems like a positive change. I still highly recommend that every user read the column care guides on a new column model. Knowing the difference between a pH 7 and pH 8 upper limit (or 45 vs. 60 °C) can save you headaches, and the guides usually provide the best storage and rescue procedures.
Bad Column or Bad Method?
Aside from column care, most vendors provide a Certificate of Analysis (CoA) with some test data, and perhaps with some acceptance criteria. These CoAs are where I think the points of contention really start. My colleagues and I have horror stories (spanning multiple column manufacturers) of separations that fail during column lot-to-lot robustness studies or upon transfer to another lab that purchased its own column. Of course, the columns all passed their manufacturer’s QC tests. Was the root cause a poor method (for example, robustness issues, leveraging minor column interactions, and so on), poor execution of the method (those pesky humans), or insufficient vendor tests? Honestly, each of these has been a factor at least once in my history, but I believe improved column testing could help prevent many of these instances.
Column tests can span from the very simple (plates or tailing factor on a single component), to what is most commonly seen (a few components with a plate, selectivity, and tailing factor assessment), to much more comprehensive probing of the actual chromatographic interactions, such as those described by the National Institute of Standards and Technology (NIST) and Waters (2,3). The more thorough testing is usually performed at the batch level, and not as an individual column release test, because the latter are more resource-intensive and move from a QC test to true column characterization. A recent review of the many possibilities of executing this type of evaluation demonstrates there is no “one size fits all” test, even within the relatively constrained realm of reversed-phase stationary phases (4).
In my industry (pharmaceuticals), there is the expectation that our drug manufacture process is fully characterized and controlled; yet the tools we use to assess our processes, such as chromatography columns, are not held to the same standards. The nightmare scenario for me is not being able to release life-saving medicines because a new column failed to perform consistently with previous lots. If I had a magic wand, every column I receive would have every chemical interaction identified and tightly controlled by the manufacturer. I had even hoped to use characterization tests, such as the hydrophobic subtraction model (5), to assess column manufacture precision (6). The reality is that the execution of this type of rigorous testing on every column, at present, would be cost-prohibitive to the industry.
The Answer Is...
Kudos to those manufacturers that perform more than the easy, simple column tests. Laurels to those who listened to Atwood and Goldstein’s advice “...to deliberately choose test compounds that expose variations” (7). Accolades to those who put realistically appropriate specifications on those tests. Please keep working toward “the perfect test.” In pharmaceuticals, we’re counting on you, the column makers, to deliver the same product, day-in and day-out, for the life cycle of our products.
To give an answer to the question I posed as the title of this blog: No, the paper in the box is irrelevant. Having the best tests possible on those CoAs, however, is absolutely critical to us, the end users.
(1) J.G. Atwood and J. Goldstein, J. Chromatogr. Sci. 18, 650–654 (1980).
(2) L.C. Sander and S.A. Wise, J. Sep. Sci. 26, 283–294 (2003).
(3) U.D. Neue, E. Serowik, P. Iraneta, B.A. Alden, and T.H. Walter, J. Chromatogr. A 849, 87–100 (1999).
(4) P. Žuvela, M. Skoczylas, J. Jay Liu, T. Baçzek, R. Kaliszan, M.W. Wong, and B. Buszewski, Chem. Rev. 119, 3674–3729 (2019).
(5) L.R. Snyder, J.W. Dolan, D.H. Marchand and P.W. Carr, Adv. Chromatogr. 50, 297–376 (2015).
(6) J.G. Shackman, J. Chromatogr. A 1475, 116–118 (2016).
(7) J.G. Atwood and J. Goldstein, Anal. Chem. 53, 1046–1046 (1981).
Jonathan Shackman is a senior principal scientist in the Chemical Process Development department at Bristol Myers Squibb (BMS) and is based in New Jersey.