From QTPP to CQAs: A Practical Framework for CMC Teams
In our earlier piece on the QTPP as the foundation of every sound CMC strategy, we walked through what belongs in a Quality Target Product Profile and how its content varies by modality. This piece picks up the next step in that chain: how a QTPP helps define a program’s Critical Quality Attribute.
In practice, this is one of the more commonly mishandled steps in early CMC development. Teams often approach CQA identification as a checkbox exercise, running every measurable attribute through a generic risk matrix until a list falls out the other end. Others skip the QTPP link almost entirely and inherit a CQA list from a platform product or a prior program. Both produce a control strategy that is hard to defend when a reviewer asks why one attribute made the list and a similar one did not.
Start With the QTPP, Not the Assay List
The QTPP is the target, not a wish list, and it is not an inventory of what the analytical team already knows how to measure. A QTPP built from familiar assays produces a CQA list biased toward known attributes and blind to attributes that may matter clinically but have no current analytical method in place yet. Getting the QTPP right, in the sense described in the earlier article, is what keeps this next step grounded. If the QTPP is defined from the target patient population, the intended clinical use, and the mechanism of action, then the CQA exercise has something real to test each attribute against, specific to the program. If it was defined from the existing assay panel, the exercise is rigged from the start.
The Two Questions That Actually Determine Criticality
ICH Q8 and Q9 frame CQA identification as a risk-based exercise, and the underlying logic reduces to two questions, asked independently for each quality attribute:
Severity. If this attribute drifts from its target, what is the impact on safety or efficacy?
Uncertainty. How well do we currently understand the relationship between this attribute and the clinical outcome?
An attribute is critical when severity is high, regardless of how well the current process happens to control it. This is the distinction teams most often get backwards. Criticality is a property of an attribute’s clinical impact. It is not a property of how easy or difficult that attribute is to control, and it is not evidence pulled from a clean batch history. A process that has never shown drift in a given attribute has not demonstrated that the attribute is non-critical. It has demonstrated that the process is, so far, capable. Those are two different claims, and confusing them is the single most common error in this exercise.
CQA Identification Across Modalities
The severity and uncertainty logic is the same across modalities, but what it gets applied to is not. The QTPP elements we described by modality in the earlier article map to a different set of likely CQAs in each case.
Small molecules: QTPP elements typically include dosage form, strength, dissolution profile, impurity limits, polymorphic form, and stability. Of these, assay and related substances are almost always critical given their direct link to potency and safety. Dissolution rate becomes critical when bioavailability depends on it, which is common for oral solid dosage forms. Polymorphic form becomes critical when a given form drives solubility or stability outcomes that would result in an out-of-specification (OOS) before the end of shelf life.
Biologics (monoclonal antibodies and related formats): QTPP elements include potency, glycosylation profile, charge variant distribution, aggregation limits, and immunogenicity risk. Among the product-related attributes, aggregate level is very often critical because of its direct immunogenicity implication, independent of how rarely a given process has produced an out-of-range result. Potency or bioactivity is critical by definition. Charge variant distribution and glycosylation pattern become critical when they are mechanistically linked to activity, effector function, or half-life, which is a molecule-specific determination rather than a default assumption. Process-related impurities, including host cell protein (HCP) and residual host cell DNA, are typically critical given their direct safety implications, independent of how consistently the current process has cleared them.
Antibody-drug conjugates: Building on what we described previously about conjugation-specific QTPP elements, average drug-to-antibody ratio (DAR), DAR distribution, and free (unconjugated) cytotoxic drug level are typically critical, since each has a direct and fairly well-characterized link to both efficacy and toxicity. Conjugation site consistency becomes critical when site heterogeneity is shown to affect stability or activity for the specific conjugate. The same logic extends to other conjugate formats, such as radioconjugates or oligonucleotide conjugates, substituting the modality-specific attributes that matter for that payload.
Cell and gene therapies: This is where the severity and uncertainty distinction matters most. Replication-competent virus, for viral vector products, is critical on severity grounds alone. It does not matter that a process has never generated a positive result across every batch to date. The consequence of an undetected positive is severe enough that the attribute is critical from the first risk assessment, not after enough batches accumulate to justify it. Vector genome titer, full-to-empty capsid ratio, and potency defined through transduction efficiency or functional transgene expression round out the typical CQA list for vector-based products. For cell therapies, viable cell dose, phenotype and CAR expression, and absence of replication-competent virus carry the same severity-first logic.
Where Teams Get This Wrong in Practice
A mistake teams commonly make is treating a clean batch history as evidence that an attribute isn’t critical. A clean history is evidence of good process control, and it says nothing about whether the attribute matters clinically. The two get confused more often than they should.
Take a program that generates a substantial batch history for a biologic drug substance, with aggregate level showing no meaningful drift across any batch in that history. The proposal, reasonably enough on its face, may be to leave aggregate level off the CQA list on the strength of that record. But applying the severity and uncertainty questions independently of the batch history tells a different story: aggregate level carries a well-established immunogenicity risk for this class of molecules, which is a severity argument, not a process-history argument. The attribute stays on the CQA list, not because the process is suspected of instability, but because the consequence of drift, however unlikely, is severe enough to warrant control regardless of how the process has behaved so far.
That’s the pattern worth watching for in any program, in any modality: a clean batch history is evidence of good control, not evidence of low criticality. Mistaking one for the other is how attributes that should stay on the CQA list quietly fall off it.
From CQA List to Control Strategy
The CQA list is not the deliverable. It is the input to specifications, in-process controls, the validation package, and any comparability protocol that will accompany a future change such as process, site, or presentation. A CQA list that can be traced back to a specific QTPP element, and from there to a specific severity and uncertainty judgment, gives reviewers a coherent story: patient need → quality target → critical attribute → control point. A CQA list assembled from a generic checklist or inherited from a prior program does not tell that story, even when the underlying science happens to be sound, because the traceability is not there to show it.

Final Thoughts
A QTPP that is not tied to clinical relevance, and a CQA list that is not tied to severity independent of current process control, are two versions of the same underlying mistake: designing to what is convenient to measure or control, rather than to what actually matters to the patient. The fix in both cases is the same discipline applied at a different step, working from the clinical and quality target outward, rather than from the assay panel or the batch history backward.
| At Spark CMC Consulting Corp, we work with early- and late-stage biotech companies to build CQA lists and control strategies that are scientifically grounded, traceable to the QTPP, and structured to support NDA and BLA submissions. If you are building your CMC strategy or preparing for your next regulatory milestone, let’s talk. |