Validating an analytical method can seem daunting, with a litany of performance characteristics that must be demonstrated to establish that our methods are suitable for their intended purpose. While each performance characteristic must be individually demonstrated, the same sample measurement can produce data for multiple performance characteristics [A]. This approach enhances the efficiency of analytical method validations by reducing the number of samples needed.
Grouping Accuracy and Precision
One of the easiest sample sets to group are those for accuracy and precision (which includes repeatability and intermediate precision). Demonstration of accuracy requires three replicate measurements at three different concentrations, for a total of nine samples (3x3 sample sets) [1,2]. Likewise, repeatability can also be demonstrated by three replicate measurements at three different concentrations [1,2]. If we use the 3x3 sample set for both these performance characteristics, we can derive all the data we need from one set of measurements. We can calculate both the percent recovery and percent relative standard deviation (RSD) of all nine of our samples. A different analyst on a different instrument can measure the same 3x3 sample set and calculate the cumulative percent RSD to determine intermediate precision [1, 2].
Grouping Limit of Quantitation (LOQ) with Accuracy and Precision
We must demonstrate both accuracy and precision at our LOQ, as this is the lowest amount of analyte that can be quantified with suitable accuracy and precision [1, 2]. If we select the LOQ (identified during method development) as one of the concentrations in our 3x3 sample set, we can show three performance characteristics with a single set of samples.
An example of a 3x3 sample set would include one concentration at the LOQ, one at the specification level (100%), and one above the specification level at the upper bound of our range (typically 120 to 150%).
Grouping Linearity with Range
The final set of performance characteristics that can be co-demonstrated is linearity with range.
Linearity represents the ability of a method to elicit results that are directly, or by a well-defined mathematical transformation, proportional to the amount of analyte [1, 2]. The range of a method is the upper and lower bounds of where results are reliably accurate, precise, and linear [1, 2]. Linearity must be demonstrated at no less than five concentrations across the range of the method, and the range of the method should show accuracy, precision, and linearity [1, 2].
We can set up our linearity experiment in such a way that we measure triplicate samples at five concentrations (15 samples total) between the upper and lower bounds of our range, which include concentrations for the 3x3 sample set.
For example, we can make a sample set that includes concentrations at the LOQ (lower bound of the range), 50%, 75%, 100%, and 120% (Upper bound of range) of the specification limit. Note that the LOQ, 100%, and 120% constitute our 3x3 sample set from earlier. From this sample set, we have enough data to calculate recovery, percent RSD, and linearity at all five concentrations, demonstrating the method's range. Therefore, we have grouped accuracy, precision, LOQ, linearity, and range.
With this strategy in mind, let's look at a full analytical method validation sample set. We are trying to validate a cleaning analytical method for the product MinQ. Product MinQ has a maximum allowable carryover (MACO) limit of 10 ppm [B], so that will be our 100% specification level. During method development, we determined that the LOQ for this method was 1 ppm. We must demonstrate accuracy, precision, LOD, LOQ, linearity, specificity, and range to validate this impurity method [A].
Specificity and LOD cannot be paired with any other performance characteristics, so each will need its own sample set. However, we can prepare a single sample set of five concentrations with three replicates of each sample to determine accuracy, precision, LOQ, linearity, and range, from just 15 samples. The sample set has concentrations of 1 ppm (LOQ), 5 ppm (50%), 7 ppm (70%), 10 ppm (100%), and 12 ppm (120%).
To demonstrate the range of the method, the lowest concentration of this sample set is the LOQ, and the highest is the upper limit of the range we wish to validate, in this example, 12 ppm (120%). Since any sample over our 10 ppm limit will lead to re-cleaning, we don't need to demonstrate a range far above that.
We can calculate the percent recovery and percent RSD from the 1 ppm (LOQ), 10 ppm, and 12 ppm samples, effectively demonstrating the accuracy, precision, and LOQ. We can then determine linearity by plotting the LOQ, 5 ppm, 7 ppm, 10 ppm, and 12 ppm measurements on a graph and report the slope, y-intercept, correlation coefficient, and residual sum of squares. We can have another analyst measure the 1 ppm, 10 ppm, and 12 ppm samples once more and derive the cumulative RSD to determine intermediate precision. Since we have already calculated the recovery and precision at the upper (120%) and lower bounds (LOQ), we have determined the range where samples are accurate, precise, and linear.
While analytical method validation is essential, it doesn't have to be overly complex. Maximizing the data extracted from each sample allows us to validate our methods efficiently and effectively.
References
USP. (2017). Chapter <1225> Validation of Compendial Procedures
ICH. (2023). Q2 (R2). Validation of Analytical Procedures
Related Mindful Quality Memos:
Contributors: Alec Fufidio, Joanna Joseph, and Jenna Carlson