Well over a decade ago I was sitting in a weekly project meeting, eagerly awaiting my workload for the week. I was a keen bioananalyst, only a couple of years into my role, learning to produce good quality pharmacokinetic profiles for phase I clinical studies. I routinely analyzed 100 samples a day, predominantly protein precipitation or liquid/liquid extraction followed by LC-MS/MS analysis for target drug concentrations. Back then, life was easy -- validate assays, analyze samples, report and move on.
“We have another ISR issue”
It was then that the Study Director raised her head and said, “We have another ISR issue.” Our heads all dropped. Another weekend of unplanned analysis was ahead of us due to Incurred Sample Reproducibility (ISR) failure.
ISR had recently been introduced into our lab and has since been causing repeated headaches. Previously successful assays were beginning to fail this new criterion.
For those unfamiliar with the concept, ISR was introduced as guidance in many of the world’s regulatory bodies. Rightly so, the bioanalytical community had embraced this new guidance as repeat analysis of incurred samples promised to give more confidence in their reported concentrations. Practically, it involved re-analyzing a percentage of the study samples to demonstrate that the original results were valid, within QC-like acceptance criteria. The headache-inducer was that not all reanalyses agreed with the original result. This threw the validity of all the sample results into question.
But the assay was validated! We proved accuracy, precision, and storage stability. We showed good recovery from our sample preparation and minimal matrix effects from control plasma. A full investigation was required for each ISR failure leading to lengthy troubleshooting and ultimately partial method re-validation. Improvements made in sample preparation and LC gradients were reassessed. Samples were then reanalyzed and ISR looked good, but why didn’t we catch this in the validation?
The Unpredictable Composition of Incurred Samples
There was a set of practical assumptions used in developing methods back then. We assumed the control plasma used in validation was a good representation of ‘real’ study samples. In truth, a perfect surrogate is impossible to obtain.
Despite our best efforts to add variables into the control plasma used for validation and batch acceptance samples (QCs) such as multiple individual sources, lipemic and haemolysed plasma assessments, the composition of patient samples can vary depending on several factors and, as such, contain other components which can interfere with the analysis and cause inaccurate results. The age, ethnicity, diet and lifestyle of the patient all adds to the diversity of each sample. Did the patient take recreational drugs? Was the patient on co-medication? Even the method of drug administration could potentially add extra complexity to the sample. All of these unresolved matrix components could lead to variability in obtained results, for example through ion suppression.
Flash forward a few years -- these days it’s very rare that surprises of this nature occur. We are now more aware of the effects of these parameters and take steps early in method development to mitigate against such variables.
Tools for Robustness
Being aware of potential issues that may be encountered, as well as measurable issues currently faced, is the sign of an experienced bioanalyst. The primary reason ISR failure is less prevalent today is that we are more aware of the variables, and fortunately, industry has developed a vast array of tools to ensure greater selectivity towards the analytes of interest and chromatographic stability to ensure this is highly repeatable for each analysis:
Robust LC platform of Thermo Scientific Vanquish™ UHPLC systems, adds both speed of analysis and reproducible retention times, to give further assurance that any interferences will remain resolved from the peaks of interest in our chromatogram.
Proper implementation of these tools and techniques mitigate against unforeseen matrix issues, adding to the robustness of the method. Whatever the purpose of my analysis, the last thing I ever want to do is re-analyze samples. I like my weekends too much!
Watch how sample preparation can help with your analysis here: