Empty and Full AAV Capsids — Analytics and Specification at the Center of Every Review
No single analytical question in AAV CMC generates more CBER comments than the full/empty capsid ratio. And yet most early-stage packages treat it as a secondary characterization attribute.
On this pageArticle overview
No single analytical question in AAV CMC generates more CBER comments than the full/empty capsid ratio. And yet most early-stage packages treat it as a secondary characterization attribute.
That mismatch is not a minor presentation problem. It is a regulatory posture problem that triggers IND complete response letters, delays Phase 1 approval timelines, and forces chemistry, manufacturing, and controls resubmissions at the exact moment a program is trying to move forward. The full/empty capsid ratio is not a curiosity about your production biology. It is a release specification — and CBER reviews it accordingly.
This article addresses what that distinction means in practice, why the analytical package you build today determines the specification you can defend tomorrow, and how XGene’s AAV Capsid Population Analytics Protocol structures that work for CMC submission readiness.
WHY THE FULL/EMPTY RATIO IS A RELEASE SPECIFICATION, NOT A CHARACTERIZATION ATTRIBUTE
The intellectual starting point for most early gene therapy teams is understandable: you demonstrate that your manufacturing process produces AAV, you show that vg/mL is within range, you run a capsid titer by ELISA or ddPCR, and you attach a cryo-EM image showing recognizable icosahedral particles. The full/empty ratio, if it appears at all, appears in the characterization section as a descriptive data point. It is presented as what your process tends to produce, not as what your process is required to produce to release a lot.
CBER does not accept this framing for clinical material.
The FDA’s AAV Characterization Guidance (draft, 2021) is explicit that capsid content — defined as the proportion of capsids containing the intended genome — must be assessed by validated analytical methods and linked to specifications that support lot release. The Hemophilia guidance (2020) reinforces this in the context of clinical BLA submissions by establishing that the immunogenic burden of empty capsids carries meaningful patient risk: empty capsids present the full complement of capsid protein epitopes on the surface without delivering therapeutic benefit. At fixed vg/kg dose, a lot that is 60% full capsids delivers approximately 40% more capsid protein than a lot that is 100% full. That is not a manufacturing nuance. That is a clinical risk that CBER has chosen to address through specification, not just characterization.
The CBER Office of Therapeutic Products (OTP) presentations at ASGCT spanning 2022 through 2024 have consistently communicated the agency’s expectation: for clinical BLA lots, the full capsid fraction should be no less than 90% as measured by a validated, release-qualified analytical method. That benchmark — greater than or equal to 90% full capsids — functions operationally as a release specification, not as a characterization target. The difference matters because a specification requires a validated method with defined acceptance criteria, a qualification study, and documented justification for where the limit is set. A characterization target requires none of those things, which is precisely why characterization-framed packages generate CBER comments at this attribute: they present data that is scientifically interesting but regulatorily insufficient.
ICH Q6B provides the foundational framework here. Biological product specifications are meant to be linked to analytical procedures that are qualified or validated, and specification limits are meant to be set with reference to the clinical and functional data that justify them. For the full/empty ratio, this means the 90% threshold must be grounded in evidence — either from your own program’s correlation of capsid content with in vivo potency, or from published and cited literature that establishes the functional consequence of deviating below that threshold. Stating the specification without that justification is, in CBER’s view, an incomplete submission.
The regulatory logic is not difficult to follow once you internalize the agency’s perspective on immunogenicity risk. Empty capsids have been documented to contribute to anti-capsid immune responses in both preclinical animal models and in human clinical trials. Programs administering high vg/kg doses — which is the norm in many liver-directed and CNS-directed AAV programs today — carry an immunogenic burden that scales directly with the number of capsid proteins the immune system encounters, regardless of whether those proteins are paired with therapeutic genome. A drug substance lot that does not meet a validated full capsid specification is therefore not just a quality problem. It is a safety characterization problem, and CBER reviews it through that lens.
The practical implication for early-stage sponsors is that the full/empty ratio must appear in the drug substance specification table with a defined limit, a defined method, and a defined justification — not in a characterization appendix with a note that additional data will be provided at Phase 2. That approach will generate a comment. The comment will ask for the validated method and the specification limit with clinical justification. The response will require work that should have been done before the initial IND submission.
XGene’s position, grounded in the regulatory pattern we observe across CBER-reviewed AAV programs, is that the full/empty capsid ratio should be treated as a primary critical quality attribute from the first IND submission forward. Primary means it appears in the specification table, not the characterization section. Primary means the assay used to measure it is qualified or validated for its intended purpose at that regulatory stage. And primary means the lot release decision for clinical material is contingent on meeting the defined limit — because that is exactly how CBER will evaluate it.
AUC-SV AND ORTHOGONAL METHODS: THE ANALYTICAL PACKAGE CBER EXPECTS
Accepting that the full/empty ratio is a release specification immediately raises the method question, and the method question is where most programs encounter their first serious analytical gap. Not because validated methods do not exist — they do — but because the methods that are easiest to run at early stage are not the methods that CBER treats as fit for release qualification, and the regulatory gold standard requires both instrumentation and expertise that many early-stage CDMOs do not routinely provide.
Analytical ultracentrifugation with sedimentation velocity analysis — AUC-SV — is the method CBER treats as the primary reference method for full/empty capsid determination. The physical basis is straightforward: empty and full AAV capsids sediment at different rates in a centrifugal field because they differ in mass and buoyant density. For AAV2, the empty capsid sediments at approximately 60 Svedberg units and the full capsid at approximately 110 Svedberg units. For AAV5, those values are approximately 73S and 115S respectively. For AAV8, the empty capsid sediments at approximately 65S and the full capsid at approximately 112S. These sedimentation coefficient differences are large enough that the two populations are resolved clearly in the c(s) distribution generated by SEDFIT software, which is the standard analysis tool for AUC-SV data in the gene therapy field.
The quantitative output from AUC-SV is the percentage of full capsids calculated as the area under the full peak divided by the combined area under the full and empty peaks, multiplied by 100. This calculation is direct, model-based, and highly reproducible when instrument conditions — rotor speed, temperature, wavelength, loading concentration — are controlled within validated ranges. The sensitivity limitation of AUC-SV is relevant: the method performs optimally at capsid concentrations that are accessible for drug substance lots but may be challenging for drug product when dilution is required for formulation. Method development must address this.
CBER’s expectation, stated consistently across Office of Therapeutic Products (OTP) presentations at ASGCT 2022 through 2024, is that at least two orthogonal methods are used to characterize the full/empty capsid population. AUC-SV as the primary method should be complemented by at least one independent analytical approach. The scientifically strongest candidates for that orthogonal role are cryo-electron microscopy, charge-detection mass spectrometry, and ion-exchange HPLC.
Cryo-EM provides direct visualization of capsid populations and can distinguish empty, partial, and full capsids on the basis of the electron density inside the capsid shell. It is not a routine QC method — the data acquisition and image analysis workflow is too resource-intensive and expert-dependent for lot release — but it is the most structurally definitive characterization tool available and carries significant weight in a CMC package when presented alongside AUC-SV data. Cryo-EM and AUC-SV used together on the same lots constitutes a strong orthogonal argument for the validity of your quantitative release method.
Charge-detection mass spectrometry, or CDMS, is an emerging single-particle method that directly measures the mass of individual capsids. Because empty, partial, and full AAV capsids differ in mass by the contribution of the packaged genome — approximately 2.4 MDa for a full-length ssAAV genome — CDMS can differentiate these populations without relying on sedimentation behavior. CDMS is particularly valuable for identifying partial capsid populations that AUC-SV may not resolve cleanly, and it is gaining traction in the regulatory literature as a high-information characterization method. It is not yet a standard release method at most contract manufacturers, but its inclusion in a characterization package strengthens the overall analytical argument.
Ion-exchange HPLC — IEX-HPLC at analytical scale — offers a practical middle ground between the depth of cryo-EM or CDMS and the simplicity of a single-detector release assay. Full capsids carry higher negative charge than empty capsids because the packaged genome contributes phosphate groups to the overall charge density of the particle. This charge difference is the same physical property that drives enrichment by anion exchange chromatography, and it is resolvable at analytical scale with IEX-HPLC methods that have been published and applied across multiple serotypes. IEX-HPLC can function as both a characterization method and a release-supporting QC method, and it is more accessible than AUC-SV at CDMOs that have not invested in analytical ultracentrifugation infrastructure.
The comparison between these methods should be presented formally in your CMC characterization section: AUC-SV and at least one orthogonal method run on the same lots, with concordance data that demonstrates the methods agree within defined limits. CBER reviewers use this comparison to assess whether your primary release method is measuring what you claim it is measuring. A package that presents only one method will typically receive a comment requesting orthogonal data before clinical lot release is accepted.
The method qualification study for AUC-SV as a release assay must address accuracy, precision, linearity, range, and robustness within the framework that ICH Q6B and your Phase-appropriate validation expectations require. At IND Phase 1, full validation is not the expectation — but the method must be qualified with defined acceptance criteria and a documented rationale for why it is fit for its intended purpose at this regulatory stage. CBER has been clear that “we used AUC-SV because our CDMO has the instrument” is not a sufficient method qualification argument. The argument must be analytical, not logistical.
ENRICHMENT STRATEGIES AND SPECIFICATION SETTING: THE MANUFACTURING-ANALYTICS CONNECTION
The analytical strategy for the full/empty capsid ratio does not stand alone. It is directly connected to the manufacturing strategy for enrichment, and both must be documented in the CMC package in a way that shows CBER the complete loop: your process produces a defined capsid population, your enrichment step achieves a specified full capsid fraction, your release method validates that specification, and your specification limit is justified by functional and clinical evidence. That loop must close. If it does not — if, for example, your enrichment step is described in manufacturing but its performance is not linked to the specification limit, or if the specification limit is stated without functional justification — CBER will ask for the missing connection.
The dominant enrichment strategy for AAV drug substance today is anion exchange chromatography. The physical basis is the same charge difference that makes IEX-HPLC work analytically: full capsids, carrying packaged genomic DNA with its high density of phosphate groups, bind more tightly to AEX resin than empty capsids at equivalent ionic conditions. A salt gradient elution separates the two populations, with empty capsids eluting at lower ionic strength and full capsids at higher. Well-designed AEX processes using resins such as CIMmultus QA or equivalent convective interaction media can routinely achieve full capsid fractions greater than 95% in the drug substance pool, which provides meaningful clearance margin above the 90% specification threshold and supports lot-to-lot consistency.
The CMC description of your AEX enrichment step must document not just the operating parameters — resin, load challenge, linear flow rate, gradient profile, pool criteria — but also the characterization data establishing what the step achieves in terms of full capsid enrichment across the manufacturing scale you intend to use clinically. Process development lots, process qualification lots, and clinical lots should all be represented in this data set, with AUC-SV measurements at both the pre-AEX and post-AEX steps showing the enrichment delta. CBER reviewers use this data to assess whether the enrichment step is controlling the attribute or merely observing it.
CsCl equilibrium density gradient ultracentrifugation is the historical reference method for AAV purification and separation characterization. In a CsCl gradient, empty capsids band at a density of approximately 1.32 grams per milliliter and full capsids at approximately 1.41 grams per milliliter. This large density difference makes CsCl gradients excellent research tools for characterizing capsid populations and for generating reference material-grade full capsid preparations. However, CsCl is not a GMP-compatible purification approach — cesium chloride is toxic, gradient ultracentrifugation is difficult to scale and validate, and the process is not consistent with current GMP manufacturing expectations for clinical gene therapy products. Its role in your CMC package is therefore as a research and characterization tool, not as a manufacturing step, and it should be described accordingly if it was used to characterize your capsid populations during development.
Iodixanol step gradient centrifugation provides partial operational separation of empty and full capsids under GMP-compatible conditions and is used in some manufacturing platforms as an intermediate purification step before AEX polishing. The degree of enrichment achievable with iodixanol gradients is typically less than AEX and more variable between lots, which means it is not sufficient as a standalone enrichment strategy if the target specification is 90% full capsids. It functions best as a platform step that reduces the empty capsid burden entering the AEX column, improving AEX performance and consistency.
Specification setting for the full/empty ratio must be grounded in data, and CBER’s expectation under ICH Q6B is that the limit is justified by reference to clinical and functional evidence, not set by what your process happens to produce. The practical workflow for specification justification involves correlating full capsid percentage across a panel of lots — which may include process development lots, comparability lots, and clinical lots — with in vivo potency readouts from relevant animal models or, where available, early human clinical data. If full capsid percentage below 80% is associated with reduced transduction efficiency or altered immunogenicity profiles in your NHP or rodent model, that correlation becomes the functional argument for setting the specification at or above 90%. Sponsors who cannot provide this correlation at BLA submission will receive a request for it. Building the study into your development plan from Phase 1 forward is materially more efficient than generating it retrospectively under regulatory pressure.
The manufacturing-analytics connection is also relevant for the partial capsid population, which neither AUC-SV nor IEX-HPLC always resolves cleanly. Partially packaged capsids — those containing genome fragments shorter than full-length — sediment between the empty and full peaks in AUC-SV and may not be assigned clearly to either population in a standard c(s) analysis. CDMS addresses this most directly by measuring actual capsid mass, and CBER reviewers have increasingly asked about how partial capsids are accounted for in the full/empty ratio calculation. Your method description must clarify how partial capsids are defined, how they are measured, and whether they are included in the full fraction, the empty fraction, or counted separately. Ambiguity on this point will generate a reviewer question.
The specification for full capsid content, validated by AUC-SV as the primary release method and supported by at least one orthogonal characterization method, justified by functional data linking capsid content to potency, and connected to a documented manufacturing enrichment strategy — that is the complete picture CBER expects to see in a clinical BLA CMC package. The programs that build this picture from IND Phase 1 forward, treating it as a primary CQA from the start, consistently experience fewer CMC comments and faster review cycles at the stages that matter most.
THE XGENE AAV CAPSID POPULATION ANALYTICS PROTOCOL
Method Selection, Comparison, and Specification-Setting Guide for AAV Full/Empty Capsid Characterization
The XGene AAV Capsid Population Analytics Protocol maps each available analytical method to its regulatory acceptance status, sensitivity limitations, and role as either a characterization or release method in the CMC package.
AUC-SV (Analytical Ultracentrifugation — Sedimentation Velocity) Regulatory Acceptance Status: CBER-recognized primary reference method Role in CMC Package: Release assay (qualified/validated); primary quantitative method Sensitivity Limitation: Concentration-dependent; requires sufficient absorbance signal; partial capsids not always resolved Key Output: c(s) distribution via SEDFIT; % full = area(full) / [area(full) + area(empty)] x 100 Serotype Reference Values: AAV2 empty ~60S / full ~110S; AAV5 empty ~73S / full ~115S; AAV8 empty ~65S / full ~112S
Cryo-EM (Cryo-Electron Microscopy) Regulatory Acceptance Status: Accepted characterization method; not a routine release method Role in CMC Package: Orthogonal characterization; structural confirmation; supports AUC-SV concordance argument Sensitivity Limitation: Expert-dependent; resource-intensive; not scalable for QC release Key Output: Direct visual count of empty / partial / full capsids from micrograph analysis
CDMS (Charge-Detection Mass Spectrometry) Regulatory Acceptance Status: Emerging; gaining regulatory literature presence; not yet standard release method Role in CMC Package: Orthogonal characterization; differentiates empty / partial / full by direct mass measurement Sensitivity Limitation: Specialized instrumentation; limited CDMO availability Key Output: Single-capsid mass histogram; identifies partial genome-containing capsids
IEX-HPLC (Ion-Exchange High-Performance Liquid Chromatography) Regulatory Acceptance Status: Accepted analytical-scale method; increasingly used as QC/release-supporting method Role in CMC Package: Orthogonal release-supporting method; accessible QC alternative at CDMOs without AUC Sensitivity Limitation: Resolution of partial capsids variable; method development required for each serotype Key Output: Chromatographic peak area ratio; % full by peak integration
CsCl Equilibrium Density Gradient Regulatory Acceptance Status: Historical reference; not GMP-compatible Role in CMC Package: Research and characterization use only; reference material preparation Sensitivity Limitation: Not scalable; cesium chloride toxicity; not appropriate as manufacturing step Key Output: Visual band separation; density assignment (empty ~1.32 g/mL; full ~1.41 g/mL)
Iodixanol Step Gradient Regulatory Acceptance Status: GMP-compatible; used as intermediate purification step Role in CMC Package: Manufacturing (partial enrichment); not a standalone specification-qualifying step Sensitivity Limitation: Incomplete separation; lot-to-lot variability in enrichment efficiency Key Output: Operational partial separation; reduces empty capsid burden prior to AEX polishing
AEX Chromatography (Anion Exchange — Preparative Scale) Regulatory Acceptance Status: GMP-compatible; primary manufacturing enrichment strategy Role in CMC Package: Manufacturing step documentation; enrichment delta data required Performance Expectation: >95% full capsids achievable with optimized gradient; CIMmultus QA or equivalent resins documented Key Output: Pre-AEX and post-AEX AUC-SV data showing enrichment delta across development and clinical lots
CBER Minimum Package Expectation: AUC-SV as primary validated release method + at least one orthogonal characterization method (cryo-EM, CDMS, or IEX-HPLC) with concordance data on same lots; specification limit ≥90% full capsids with functional justification; enrichment strategy with performance characterization data connected to specification.
Primary regulatory references
- https://www.fda.gov/regulatory-information/search-fda-guidance-documents/q2r2-validation-analytical-procedures
- https://www.fda.gov/regulatory-information/search-fda-guidance-documents/q14-analytical-procedure-development
- https://www.fda.gov/vaccines-blood-biologics/cellular-gene-therapy-products/establishment-office-therapeutic-products
- https://www.fda.gov/regulatory-information/search-fda-guidance-documents/chemistry-manufacturing-and-controls-flexibilities-developing-human-cellular-and-gene-therapy