Hygiene Conundrum

Many companies have been on the receiving end of FDA warning letters – and issues around cleaning arise time and time again. Failure to adequately clean equipment, failure to validate the cleaning verification method, failure to follow cleaning procedures... A quick reminder of the regulations: the cleanliness of non-dedicated equipment should be verified before subsequent release for use in the manufacture of intermediates and APIs, and at product change over to prevent cross-contamination. The cleaning procedure must be validated according to FDA requirements (1), and companies must employ cleaning steps that are reproducible and effective.

When validating the analytical methods used for cleaning verification, the industry standard is to use stainless steel coupons with the same surface and finish as the manufacturing equipment. The sample of interest is spiked onto the stainless coupon at an amount determined by the maximum allowable carry-over (MACO) limit and recovered by swabbing or wiping the surface of the coupon. Spiked residues may include API, precursors of the drug substances, by-products, degradation products, cleaning agents, and so on (2). The purpose of the cleaning verification is to prove that the residual contaminant is below its MACO limit.

Sounds straightforward, right? But in reality cleaning verification has always been (and perhaps always will be) a challenge, partly because of the analytical techniques used to determine trace analysis components. At such levels, several sources may contribute to variability, such as change in the sample composition, purity of reagents used, presence of microorganisms, matrix effect, adsorption, drug product degradation, or evaporation of the components of interest (3). Analysts are usually satisfied with a “cautious but good enough” approach rather than deep insight into understanding the sources of variability. In our recent study, recovery varied from excellent (97% ± 1%) to failing (50% ± 1%). The variability in results was observed at different spike levels (up to 23 percent difference between 50 and 150 percent cleaning limit, all other conditions being the same), at various ratios of API/excipient, among different analysts, and even for the same analyst on different days. We made a number of attempts to improve recovery by modifying the experimental conditions (swabbing technique, organic/aqueous ratio in diluent, the spiking solvent, and so on), but no change eliminated the observed variability.

Lack of understanding in variable recovery results is common, but we were not satisfied. We decided to investigate whether the coupon surface itself was a source of variability. The coupons were initially cleaned by rinsing and wiping the surface a few times with water and methanol to get rid of any residual deposit, but this did not help with improving recovery numbers. So the approach was modified to include fully immersing and sonicating the coupons in CIP100 and CIP200 solutions (a basic and acidic commercial cleaning agent, respectively) (4). By applying this cleaning procedure to the previously used coupons that failed the cleaning acceptance criteria, multiple analysts were able to obtain consistent recoveries from day-to-day for different APIs, and API/excipient ratios at various spike levels. The approach was successfully applied for cleaning verification of small molecules (<1000 Da) as well as large biomolecules (up to 50,000 Da).

So what does this all mean? Our study showed that the lack of a well-defined procedure and cleaning agents for cleaning the coupon surface was the major contributor to low and variable recoveries. The low and variable recovery obtained from the uncleaned surface of the coupon may be caused by deposition of residual material on the surface, a change in the oxidation state of the surface of the coupon (enthalpic interactions with the sample are introduced), or both.

Many companies no doubt experience variability in cleaning, which can be frustrating. It is worth taking the time to better investigate variable recovery rates – it will likely result in cleaning and cleaning validation being far less stressful exercises.