This is a ready-to-use reference matrix that sets a default comparability testing tier from two inputs: how severe a clinical consequence would be if a quality attribute shifted, and how well current methods would actually detect that shift. It is a standing lookup tool, not a per-change score sheet: apply it after the change-specific change-to-CQA risk assessment (assessment-comparability-change-cqa) has scored your specific change against each attribute, to translate that score into a testing depth everyone in the program applies the same way. Replace every <<FILL: ...>> placeholder. A filled specimen follows.
Document control header
| Field | Entry |
|---|---|
| Document title | Risk-Based Comparability Testing Tier Matrix, <<FILL: product / program>> |
| Document number | <<FILL: MAT-ID, e.g. MAT-COMP-TIER-001>> |
| Version / effective date | <<FILL>> |
| Linked change-to-CQA risk assessment(s) | <<FILL: assessment ID(s) this matrix is applied against>> |
| Owner | <<FILL: Analytical Development / CMC lead>> |
1. Purpose
Provide a single, pre-agreed rule that converts an attribute’s severity and detectability into a default comparability testing tier, so that two reviewers looking at the same attribute reach the same testing scope without re-arguing it for every change. This matrix does not replace the per-change risk assessment; it is the lookup table that assessment resolves to.
2. How this differs from the change-to-CQA risk assessment
The change-to-CQA risk assessment scores a specific proposed change against each attribute (impact, likelihood, detectability) to size that one comparability exercise. This matrix is broader and more stable: for any attribute, given its severity if it shifts and how detectable a shift would be with current methods, independent of which specific change is being made, it sets the default testing tier that attribute normally sits in. Use the change-specific assessment to confirm severity and detectability for the change at hand, then read the tier from this matrix.
3. Definitions
- Severity: the clinical consequence if the attribute shifted meaningfully, independent of whether it is likely to shift for a given change.
- Detectability: how well the currently available methods, as validated or qualified today, would actually catch a meaningful shift in this attribute.
| Score | Severity | Detectability |
|---|---|---|
| High | Drives safety or efficacy (for example potency, immunogenicity-related glycans, a key process-related impurity, vector identity) | Methods are validated, orthogonal, and demonstrably stability-indicating |
| Medium | Quality-relevant, indirect link to clinical effect | One qualified method, limited orthogonality |
| Low | Minor or well-characterized, no plausible clinical link | Non-discriminating, unqualified, or no current method |
4. The tier matrix
| Detectability \ Severity | Low severity | Medium severity | High severity |
|---|---|---|---|
| High detectability | Tier 1 | Tier 1 | Tier 2 |
| Medium detectability | Tier 1 | Tier 2 | Tier 3 |
| Low detectability | Tier 2 | Tier 3 | Tier 3 |
Note the asymmetry: low detectability pulls an attribute toward a higher tier even at lower severity, because an undetectable shift cannot be ruled out by data, only assumed away. High severity with low detectability always lands in Tier 3, never averaged down.
5. What each tier requires
| Tier | Methods | Lot count | Stability | Bridging default |
|---|---|---|---|---|
| Tier 1 | Existing release methods against the registered specification; no new orthogonal method required | Available lots, no dedicated campaign | Not required beyond routine stability | None expected |
| Tier 2 | Release methods plus the pre-change quality-range comparison; add one orthogonal or characterization method | <<FILL: pre-specified minimum, e.g. 3 to 6 pre- and post-change lots>> | Side-by-side accelerated stability recommended | Quality data expected to suffice; document the rationale |
| Tier 3 | Full orthogonal panel, forced degradation, extended characterization, statistical equivalence testing where lot count allows | <<FILL: extended minimum>>, engineering runs where commercial lots are scarce | Full side-by-side real-time and accelerated stability required | Nonclinical or clinical bridging likely; engage Regulatory Affairs early |
6. Attribute application table
| Attribute | Severity | Detectability | Tier (from section 4) | Notes |
|---|---|---|---|---|
<<FILL>> | <<FILL: High/Medium/Low>> | <<FILL: High/Medium/Low>> | <<FILL>> | <<FILL>> |
<<FILL>> | <<FILL>> | <<FILL>> | <<FILL>> | <<FILL>> |
7. Acceptance criteria
- Every attribute in the comparability program has a severity and detectability rating with a stated basis, not an unexplained tier assignment.
- The tier assigned matches the matrix in section 4; any deviation from the matrix result is documented and approved by Quality Assurance with a stated rationale.
- Detectability ratings are re-confirmed when a method’s validation or qualification status changes, not left static after the method improves or degrades.
- High-severity, low-detectability attributes are never assigned below Tier 3.
8. References
ICH Q5E, Comparability of Biotechnological/Biological Products Subject to Changes in Their Manufacturing Process. ICH Q9(R1), Quality Risk Management (severity and detectability framing). ICH Q6B, Specifications: Test Procedures and Acceptance Criteria for Biotechnological/Biological Products.
Confirm the current version of each reference before issue.
9. Revision history
| Version | Date | Author | Summary of change |
|---|---|---|---|
<<FILL: 1.0>> | <<FILL: date>> | <<FILL: author>> | Initial matrix. |
10. Approvals
| Role | Name | Signature | Date |
|---|---|---|---|
| Author (Analytical Development) | <<FILL>> | ||
| Statistics | <<FILL>> | ||
| Quality Assurance | <<FILL>> |
Filled specimen
Program: monoclonal antibody drug substance, standing attribute tiering used across the comparability program.
| Attribute | Severity | Detectability | Tier | Notes |
|---|---|---|---|---|
| Host cell protein (HCP) | High | Medium (single ELISA, limited orthogonality) | Tier 3 | Add orthogonal LC-MS HCP method per section 5 |
| Potency (relative, cell-based) | High | High (validated, stability-indicating) | Tier 2 | Existing validated assay covers detectability; still add quality-range comparison |
| Aggregate (%HMW by SEC) | High | High (validated, stability-indicating) | Tier 2 | Quality-range comparison; existing lots sufficient |
| Charge variants | Medium | High (validated icIEF) | Tier 1 | Registered specification comparison sufficient |
| Subvisible particles | Medium | Medium (qualified, single method) | Tier 2 | Add orthogonal imaging method if program scales to Tier 3 attributes generally |
Reading: HCP lands in Tier 3 even though it is a routine impurity assay, because a single ELISA has limited orthogonal coverage against a high-severity attribute; the matrix forces the extra method rather than letting a familiar assay pass by default. Potency and aggregate sit in Tier 2 because their methods are already validated and stability-indicating, so the residual risk is only that quality-range data have not yet been generated for the specific change, not that the method itself is weak. Charge variants, medium severity with a strong method, need no more than the specification comparison already in routine use.
Common inspection findings this matrix prevents
- A high-severity attribute tested only against the registered specification because the assigned tier was never checked against a consistent rule.
- Two reviewers on the same program assigning different testing depth to the same attribute with no documented basis for the difference.
- A method’s detectability rating left unchanged in the matrix after the method was improved (or degraded), so the tier no longer reflects current capability.
- A Tier 3 attribute (high severity, low detectability) tested at Tier 1 depth because the change itself looked minor.
How to adapt
- Set your document number and link the change-to-CQA risk assessment(s) this matrix supports.
- Populate section 6 with your product’s real attribute list; for CGT add vector identity, genome titer, capsid content, infectious titer, and replication-competent virus.
- Set the lot-count minimums in section 5 to your program’s statistical basis, not the placeholder ranges.
- Re-review the detectability column whenever a method’s validation status changes.