Identification of Attributes Requiring Testing

The mock drug product used for this case study is DP-Y. DP-Y is a generic drug product containing one API formulated as an extended-release tablet. DP-Y has two known process impurities, IP-L and IP-M, and five well-characterized degradation impurities, ID-A, ID-B, ID-C, ID-D, and ID-E. A quality target product profile (QTPP) is shown in Table I, which contains the critical quality attributes (CQAs) for DP-Y based upon pharmaceutical equivalence to the reference product. Note that the CQAs are based upon the drug product, not the procedure. CQAs guide the development of the procedure, and therefore, are not defined as analytical procedure ECs.

Table I Quality Target Product Profile (QTPP) for DP-Y

Some CQAs must be assessed using an analytical procedure The ATP links the CQA of the product to the performance of the related procedure. This linkage is accomplished through performance characteristics and associated acceptance criteria, which are technology-independent requirements based upon CQAs. The ATP is maintained throughout the product lifecycle and can serve as the basis for continual improvement efforts related to the analytical procedure. Elements within the ATP often require regulatory approval for loosening requirements and therefore would not be subject to reduced reporting categorization. Alternatively, tightening specifications or adding additional criteria may, in some instances, be appropriate for reporting relief.

For this case study, the Impurities CQA was selected to develop the ATP (Table II).

Table II Analytical target profile for quantitation of degradant impurities in DP-YTechnology Selection

The ATP informs the selection of an appropriate analytical technology. Although multiple technologies may be capable of achieving the procedure’s intended purpose, other factors such as the availability of instrumentation, institutional expertise, the operating environment (e.g., off-line), and economic considerations may also influence the selection (10). In the current case study, ultra-performance liquid chromatography with ultraviolet detection (UPLC-UV) was selected as the analytical technology based on development work demonstrating UPLC-UV’s ability to separate and quantify all impurities at levels specified in the ATP. Additionally, this technology is economical and well understood, which facilitates the generation of knowledge and understanding about the procedure using a systematic approach.

Preliminary development studies identified approximate conditions and procedure parameters capable of resolving relevant components. This consisted of a 1.7 µm C18 column with a linear mobile phase gradient of 10 mM ammonium formate and acetonitrile. A chromatogram illustrating the separation is shown in Fig. 1.

Fig. 1

UPLC-UV chromatogram and performance requirements (orange)

Risk Assessment

The chromatogram demonstrated resolution of all components and highlighted three critical performance requirements (annotated in Fig. 1 with orange text) that must be attained to avoid procedure failure. These included the resolution between impurities ID-A and ID-B (R1), the resolution between impurities ID-D and ID-E (R2), and the signal-to-noise ratio for the ID-B peak (S/N). Resolution between peaks was needed to demonstrate specificity and to accurately quantitate impurity content, while adequate signal-to-noise was needed to ensure that the quantitation limit (QL) was below the reportable range of ID-B. Inability to obtain these performance requirements would have indicated a failure to adhere to the ATP.

To determine the most probable causes of procedure failure, a risk assessment was conducted using a risk priority number (RPN) strategy. This approach evaluated the impact (I), probability (P), and detectability (D) of each potential cause of failure and assigned each a score of 1–10 based upon the cause’s concordance with that variable. For example, deviation from the intended column temperature was assigned a high impact score (9) because temperature would significantly affect specificity. The probability of occurrence of a temperature change, assessed using an interdisciplinary team of experts and historical data, was deemed low (2). Additionally, temperature change events are largely detectable (3) by monitoring temperature sensors and column pressure. Following score assignments, the RPN was calculated as the product of the three scores. Based upon the RPN and predetermined thresholds, risks were categorized into low (1–15), medium (16–40), or high (41–1000).

Results of the risk assessment using the RPN methodology are presented in Table III. Other methods for assessing risk, consistent with ICH Q9 principles, could also be used (11).

Table III Risk assessment for UPLC-UV analytical procedureIdentification of Parameter Set-Points and Ranges using DoE

Performing a DoE study enhances knowledge of the procedure while also achieving several key objectives of development and validation. For example, DoEs can provide an indication of robustness (12) and can be used to determine the parameter ranges and optimal set points for the analytical procedure (13). In a multi-variate DoE study, multiple experimental parameters are systematically varied and their combined effects on the critical performance requirements are determined. In DoE terminology, the parameters to be tested are the factors, and the critical performance requirements are the responses.

Unlike more traditional robustness approaches which evaluate the variance of a single independent parameter, DoE investigates multi-dimensional space. The design allows the establishment of a method operable design region (MODR) where the analytical procedure consistently meets performance characteristics. By varying multiple parameters simultaneously, enhanced knowledge can be acquired that relates analytical parameters to procedure performance (14). The increased knowledge and understanding generated by DoE studies can, in turn, be used to help justify reduced reporting categories for changes to ECs.

Based upon the results of the risk assessment, parameters determined to pose a high risk to procedure failure – column temperature, injection volume, starting mobile phase composition, gradient slope, and buffer pH – were included into the DoE study. Analytical parameter ranges for the DoE were chosen using knowledge gained from preliminary development studies, and a five-factor DoE model was used to generate contour plots. A five-factor Box-Behnken model was employed using Design Expert (version 13) with a response surface study type (15). Contour plots in Fig. 2 show the relationship between two sets of variables and their effects on the three critical performance requirements. Although only two relationships are displayed for the purposes of this description, the DoE study establishes relationships between all variables tested.

Fig. 2

Relationship between (top) the starting mobile phase composition and the injection volume, and (bottom) mobile phase A pH and column temperature. Each column corresponds to one of the three critical performance requirements (R1, R2, and S/N)

Effects of parameters on procedure performance identified through the DoE can be used to determine the MODR and parameter set points of the procedure. These are provided in Table IV, which presents the parameter ranges corresponding to conditions where each individual critical performance requirement met target criteria. Notably these data do not guarantee performance throughout the region, and validation is still needed.

Table IV MODR and parameter set-points

While only two relationships are displayed above, all parameters within the DoE should be considered when establishing the MODR. The result would be a multi-dimensional space within which all performance characteristics are expected to meet acceptance criteria. The MODR may be determined using a specialized software package rather than manually as performed above, but either approach could be acceptable with justification. Validation of the MODR serves to evaluate the accuracy of the approach and justify the design space region. Set-points identifying the intended operational conditions should be located within the MODR, ideally away from the edges of the region. Validation of the MODR serves to evaluate the accuracy of the approach and justify the design space region.

The Analytical Procedure Control Strategy

The analytical procedure control strategy ensures a procedure’s adherence to performance characteristics described in the ATP. These performance characteristics are based upon a product’s CQAs, meaning that the control strategy plays a crucial role in demonstrating the procedure’s ability to reliably assess CQAs. As knowledge of the procedure increases, confidence in the control strategy to evaluate a procedure’s performance should also increase. This point is critical – enhanced knowledge of an analytical procedure can inform the design of a control strategy which increases assurance that CQAs are appropriately evaluated. The enhanced approach also leads to greater understanding of how changes to the procedure will affect performance and how the impacts of changes can be assessed.

Control strategies include elements such as the procedure description along with parameter set points and/or ranges, materials, instructions for data processing, and the system suitability test (SST). Proposed analytical procedure set-points and the MODR for the current procedure are displayed in Table V. The MODR may also be represented graphically or by using equations, which may lead to a larger portion of the design space being presented (16).

Table V Proposed analytical procedure parameter set points and MODR

The partial analytical procedure description in Table V includes set-points for parameters not included in the DoE, ranges for DoE parameters illustrated in Fig. 2, and ranges for DoE parameters whose data is not shown. An example in the last category includes the gradient slope, determined by the change in mobile phase A percentage versus time during the 3.0 – 15.0 min period. This demonstrates how parameters may gain post-approval flexibility outside of typical regulatory allowances.

The SST often serves as a basis for monitoring the procedure’s performance by verifying selected characteristics within the ATP. In the current study, each of the three critical performance requirements identified early in development are included as SST criteria (Table VI). An additional criterion evaluating precision was included to capture potential points of failure not directly covered by the other SST criteria. Establishing an appropriate level of precision may also allow other acceptance criteria, like the QL, to be derived, and is unique as serving as an evaluation of drift throughout the injection sequence (17, 18).

Table VI System suitability criteriaProcedure Validation

Validation of a procedure containing an MODR may be performed using one of two strategies: the minimal or the optional extent. In the minimal extent strategy, the procedure is validated within a partial, incomplete region of the MODR. This could range from a single set of univariate parameters, most commonly set points, to a larger, yet not fully contained region of the MODR. While this approach requires less initial data, future changes to analytical parameters within the broader MODR may require additional risk assessments and validation work. Accordingly, unrestricted movement throughout the region may not be permitted without additional assessment or validation activities.

Alternatively, the optional extent strategy involves validating the MODR throughout a fully contained region. This approach can provide full flexibility to adjust parameters within a space without the need for additional validation. Because validation at all points in the MODR is not realistic, validation using the extrema and set points can be used to demonstrate acceptability throughout the entire region. Alternatively, other boundary conditions between the extrema and set points can be chosen for validation purposes, with these boundary conditions defining the validated space.

Selection of validation boundary conditions for the MODR can be made and justified using data acquired during development. It is not necessary to include all extrema from each DoE parameter in the validation, as only those combinations that risk not meeting ATP acceptance criteria are needed to demonstrate the region’s acceptability. For example, in the current case study the development data indicated that parameter combinations that include a large injection volume (4 mL) with a high starting mobile phase A percentage (95%) were less likely to result in a failed performance characteristic compared to parameter combinations with the injection volume and the starting mobile phase A percentage near the lower extremes. Thus, combinations that include variations with large injection volumes or high mobile phase A percentages could be excluded from the full validation protocol design based on supporting evidence provided from development data.

Regardless of the validation extent utilized, analytical procedure validation should be conducted in accordance with ICH Q2(R2) (19). As per ICH Q14, data generated for the DoE during development could be repurposed for validation, reducing the need for duplicate studies. Additional validation data can be acquired separately and combined with prior development data in the validation data package.