I recently read the EMA (European Medicines Agency) report on Genentech’s (Roche) attempt to file Perjeta (Pertuzumab) as a Quality–by–Design (QbD) submission. I was excited to see a Biologics product (monoclonal antibody) going through QbD for the first time. I tout their effort. Sadly, FDA rejected it as a full QbD and resulted in a traditional submission. So one down for biologics QbD.
Why was it rejected as a QbD submission?
“Design Space” was not properly characterized.
From page 31 of the report,
“The principles used to define the proposed Design Space were endorsed. However, there were several issues which, taken together, led to a Major Objection at Day 120 and Day 180 of the procedure that precluded the approval of the Design Space. As a consequence, the claimed Design Space was withdrawn.”
To save you time, I summarized the main points and excerpts from the report.
The main information related to QbD is described on Pages 30~32 of the CHMP assessment report, Rev04.12,
1. Missing the design space of ADCC “… was considered as a Major Objection.”
“ADCC (Antibody-Dependent Cell-Mediated Cytotoxicity) in the mode of action of pertuzumab was unclear but could not be totally dismissed according to the data presented (characterisation and non clinical). Assessment of the fucosylation pattern, which can impact ADCC activity, was initially not taken into consideration as part of the control strategy. This was considered as a Major Objection.
In their response, the Applicant provided a justification to support their view that ADCC activity does not add to the efficacy of pertuzumab. However, the Applicant acknowledged that residual risk from ADCC impacting variants cannot be totally excluded.
The Applicant evaluated various attributes in relation to a possible increased or decreased impact on ADCC. Studies clearly demonstrated that afucosylation increases in vitro ADCC and the predominant afucosylated form, G0-F, gives a strong correlation with ADCC. G0-F was added to the list of CQAs. The Attribute Testing Strategy (ATS) RRF tool was applied to G0-F to determine the suitable testing strategy and it was concluded that this attribute needs to be tested as part of the active substance Control System testing. Currently, and until the method is finally validated, the test is part of the process validation protocol. The Applicant will have to apply for a variation to add %G0-F to the active substance release specifications.”
2. Grandfathered specs and ranges “…were not considered acceptable.”
“The CQA-ACs and specifications initially proposed by the Applicant for selected attributes, taking into consideration the experience with other monoclonal antibody products licensed by the Applicant, were not considered acceptable. In their Day 180 responses, the Applicant tightened the proposed limits to be more in line with levels found in clinical batches. The revised limits were considered acceptable.”
3. Without experimental data, criticality was determined. Hence, “a Major Objection was raised.”
“The strategy to conduct a risk assessment of each unit operation in order to estimate the impact of process parameters on CQAs, non-CQAs and KPIs, with the subsequent determination of a severity score to determine the conduct of univariate and multivariate studies, was endorsed.
It was acknowledged that the concept of Impact Ratio is an objective measure of process parameter criticality. However, a Major Objection was raised. It was considered that the calculation of this Impact Ratio could be affected by potential sources of bias related to:
– The adequacy of the statistical multivariate models which could then affect the effect estimates;
– The offset introduction which could lead to inaccurate predictions for scale differences.
Justification was provided for the different possible calculations of an Impact Ratio and the choice of the cut-off values of 0.10 and 0.33. Following narrowing of the CQA-AC for selected attributes at Day 180, with consequential narrowing of the respective CQA-TR, the process characterisation was reviewed and validation data associated with these CQAs and determined that for use in a traditional process description without a Design Space Perjeta claim, no further narrowing of the process parameter ranges is required and no additional parameters were elevated to CPPs.”
4. Lifecycle management means nothing without a design space. “..was removed from the dossier.”
“A Post-Approval Lifecycle Management (PALM) plan was initially proposed to describe how the Applicant would:
– Perform real-time and/or retrospective monitoring of process and product attributes (CPP and KPI excursions, quality attributes in the Control System testing and those classified as CaM attributes);
– Manage changes in attribute criticality;
– Manage changes to CPPs and non-CPPs;
– Update the Control System as necessary based on additional process and product knowledge gained;
– Manage the PALM plan in the Pharmaceutical Quality System.
The principle of implementing such a PALM plan was fully endorsed as part of the QbD approach developed by the Applicant. Following the withdrawal of the Design Space, the PALM plan was removed from the dossier.”
There is still hope though. “CHMP acknowledged the efforts made by the Applicant on the development of a Design Space. There is the possibility to reapply for a Design Space post-approval.”
Why is Design Space so important for QbD and yet so difficult?
I will discuss this in the next article.