This article reviews the history of Cleaning Validation Acceptance Limits for Active Pharmaceutical Ingredients (APIs) and identifies where the currently used industry limits came from.
Cleaning Validation for the 21st Century: Acceptance Limits for Active Pharmaceutical Ingredients (APIs): Part I by Andy Walsh
art I of this article reviews the history of Cleaning Validation Acceptance Limits for Active Pharmaceutical Ingredients (APIs) and identifies the origins of currently used industry limits. The current approaches to setting acceptance limits will be analyzed and some of the problems and weaknesses of these approaches will be discussed.
Early Ideas on Cleaning Validation Acceptance Limits
In the early 1980s, most companies were just beginning to grapple with the FDA’s shift to validation. This activity was of course more focused on Process Validation, but some companies took a wider view and began looking at cleaning processes as well. In 1984, Samuel Harder published an article, “The Validation of Cleaning Procedures,” and discussed many aspects of what he saw would be required to validate a cleaning process.1 Most, if not all, of the ideas he expressed can still be seen in practice in cleaning validation today. Concerning the setting of Acceptance Limits, Harder wrote that limits “...must be practical and achievable by a reasonable cleaning procedure... ...must be verifiable by analytical methodology existing in the company... ...and must be safe and acceptable and in line with residual limits set for various substances in foods.” (Note: The phrase “Practical, Achievable, and 74 PHARMACEUTICAL ENGINEERING July/August 2011
Verifiable” began to appear in many guidelines and literature subsequently.) Harder goes on to reference 21 CFR 193 “Tolerances for Pesticides in Food Administered by the Environmental Protection Agency” and showed a table of limits for a variety of hazardous pesticides and herbicides. These limits ranged from a low of 0.5 ppm for Diquat to a high of 200 ppm for hydrogen cyanide. Harder pointed out that the amount of drug products ingested by an individual is much lower than the amount of food ingested. Therefore, he suggested that Acceptance Limits for drug substances comparable to those used for pesticides would be reasonable. In 1988, the FDA had its first major experience with cross contamination traceable to inadequate cleaning and cleaning validation.2 A supplier of the API Cholestyramine Resin USP had to recall the product, due to contamination with low levels of intermediates and degradants from the production of agricultural pesticides. This cross-contamination was believed to have occurred from the use of drums in the manufacture of the API that had been used to recover solvents from the manufacture of agricultural pesticides at another location. The drums were not properly cleaned leading to agricultural pesticides entering the API manufacturing process. This incident raised the FDA’s awareness to the potential for cross contamination from unvalidated cleaning processes and heightened the concern in the industry that validation would become required for cleaning as well as for process. In 1989, Doug Mendenhall of Abbott Labora-
Acceptance limits for APIs
tories published a chapter “Cleaning Validation” in Drug Development and Industrial Pharmacy.3 Mendenhall expanded upon the ideas presented by Harder adding ideas, such as using a matrix approach, testing for cleaning agents, placebo batches, and most interestingly, pointed out the potential use of visual inspection. In addition, for Acceptance Limits, he suggested to “...establish in collaboration with tox and medical authorities an effect threshold” or alternatively, to “... superimpose an appropriate safety factor, e.g., 10X, or 100X.” Although not cited as such, this suggestion was probably connected to a much older article by Lehmann and Fitzhugh on the use of a...