XGene CMC IntelligenceXGene Intelligence

AAV Vector Characterization — The CQA Framework for CBER Analytical Review

SpecificationsAnalytical MethodsImpurity ControlCAPA / QMSBiologics

An AAV drug substance section is not complete because it has a titer and a purity profile. It is complete when it demonstrates understanding of the full CQA space and…

By Khaled Aamer, PhD · Founder, XGene LLC Aug 22, 2026 16 min read

On this pageArticle overview

An AAV drug substance section is not complete because it has a titer and a purity profile. It is complete when it demonstrates understanding of the full CQA space and the analytical methods capable of resolving each attribute.

That sentence is not rhetorical. CBER reviewers are trained to look for gaps in analytical coverage, and the most consequential gaps in AAV Module 3 packages are almost never the methods that are missing entirely. They are the methods that exist on paper — listed in a specification, assigned an acceptance criterion — but whose scientific rationale, validation status, or clinical batch data cannot bear the weight of what regulators actually need to conclude about product quality. Understanding how CBER structures its evaluation of an AAV analytical package requires understanding the CQA framework that underlies it, because that framework is the map reviewers carry into every BLA or IND review.

The AAV CQA Framework: What CBER Expects Beyond Titer and Purity

The regulatory basis for AAV critical quality attributes is a composite of several guidance documents that must be read in concert. ICH Q6B establishes the foundational principle that specifications for biotechnological and biological products must be grounded in an understanding of the relationship between product quality attributes and clinical safety and efficacy. FDA’s 2021 draft guidance on AAV serotype-specific considerations from CBER translates that principle into vector-specific expectations, making explicit that the analytical characterization package must address capsid identity, genome content, capsid integrity, process-related impurities, potency, and safety attributes as distinct categories, each requiring independent analytical resolution. The EMA/CAT guideline on quality aspects for ATMPs adds a European regulatory dimension that sponsors targeting global development cannot ignore. Taken together, these documents define not a checklist but a framework — a logic of control in which each CQA is demonstrated to be controlled by a method that is fit for purpose and specified at a limit that can be scientifically defended.

The first category of CQAs is identity. For an AAV vector, identity confirmation operates at two levels: serotype identity and transgene identity. Serotype identity is typically confirmed by a capsid-specific ELISA — the Progen PRATV system and AAV2-specific kits are the most widely used commercial platforms and measure total capsid protein including VP1, VP2, and VP3. PCR-based confirmation targeting serotype-specific ITR sequences provides complementary genomic identity confirmation. CBER expects both levels of identity to be addressed. A specification that lists only one — the capsid ELISA without transgene sequence confirmation, or a PCR test without capsid protein characterization — will generate a deficiency. Identity is not a single assay category; it is a multi-attribute confirmation that must be orthogonally covered.

Genomic titer is arguably the most technically evolved of the primary CQA categories, and the shift in CBER’s expectation over the last five years is visible to anyone who has reviewed modern AAV INDs alongside those filed a decade ago. Droplet digital PCR has displaced quantitative PCR as the preferred platform for genomic titer determination, and the scientific rationale is straightforward: ddPCR produces absolute quantification without dependence on a standard curve, eliminating the inter-laboratory and inter-run variability that plagued qPCR-based titer comparisons. The primer and probe design for AAV genomic titer by ddPCR targets the inverted terminal repeat sequences — ITR-targeting assays are preferred precisely because they are serotype-agnostic within the ITR family and interrogate the sequence elements most critical to vector function. Vector genome concentration is reported in vg/mL, and method suitability must be demonstrated across the working range of the intended lot release titer.

Capsid titer by ELISA — measuring total VP1, VP2, and VP3 — provides the denominator for several of the most clinically consequential derived metrics. The ratio of vector genomes to infectious units — the VG:IU ratio — is an indirect indicator of manufacturing process quality. Well-manufactured lots from optimized processes consistently achieve VG:IU values below 1,000, meaning fewer than 1,000 genome-containing particles are required to produce one infectious unit in a transduction assay. Lots with VG:IU ratios significantly above that threshold signal process heterogeneity and are a leading indicator of elevated empty capsid burden and suboptimal manufacturing conditions. VG:IU is not always explicitly specified in early-phase INDs, but CBER reviewers will note its absence in the characterization data package and it frequently appears as a Phase 2/3 readiness discussion item.

Physical characterization of particle size and aggregation state is addressed by dynamic light scattering. DLS-measured Z-average diameter for AAV particles falls in the 20–25 nm range when the preparation is free of significant aggregation, and a polydispersity index below 0.2 is the accepted threshold for a monodisperse preparation suitable for clinical use. Elevated PDI values indicate the presence of aggregated or fragmented species and must be correlated with SEC-MALS and TEM data to characterize the nature of the heterogeneity. SEC-MALS provides absolute molecular weight determination of the major peak species and is an orthogonal identity and integrity tool: for AAV2, the assembled capsid consisting of VP1, VP2, and VP3 produces a major peak molecular weight in the range of approximately 3.7 to 5.0 MDa depending on serotype, and deviation from the expected Mw signals capsid assembly defects or the presence of aberrant species.

SDS-PAGE with silver staining provides resolution of the individual capsid proteins and is the primary method for assessing VP subunit stoichiometry. For AAV2, VP1 migrates at approximately 87 kDa, VP2 at 72 kDa, and VP3 at 62 kDa. The expected ratio — approximately 1:1:10 for VP1:VP2:VP3 — reflects the natural assembly preference of the capsid and is functionally relevant because VP1 contains the phospholipase A2 domain essential for endosomal escape during cell transduction. Deviation from expected stoichiometry, particularly VP1 underrepresentation, has functional consequences and must be addressed in the specification. Densitometric analysis of silver-stained gels is not a precise quantitative method, but it provides sufficient resolution to detect gross stoichiometric abnormalities and is expected in the characterization section even where more quantitative methods are also employed.

Safety attributes complete the CQA picture. The specification must include sterility, endotoxin (LAL or rFC method), mycoplasma, appearance (visual inspection), osmolality, and pH. These are table-stakes attributes that CBER reviewers expect to find fully validated with tight, clinically justified acceptance criteria. Residual Benzonase — introduced during manufacturing as a nuclease to digest free nucleic acids — must be controlled by a sensitive immunoassay with a limit that accounts for the systemic exposure calculation at the intended clinical dose. Replication-competent AAV testing, addressed in FDA’s 2020 guidance on retroviral vector-based gene therapy products and extended by analogy to AAV by CBER practice, is a safety attribute that must be present in the specification with a validated assay and a “not detected” acceptance criterion. Absence of RCA testing from an AAV IND filing is a predictable deficiency item.

The structural logic of an AAV CQA framework, when properly built, maps each attribute to its regulatory basis, its characterization method, its release method, its specification limit, and the data justifying that limit. CBER reviewers do not evaluate methods in isolation. They evaluate the coherence of the complete analytical control package — whether the methods are orthogonal where orthogonality is required, whether the acceptance criteria are grounded in clinical batch experience or functional data, and whether the development history demonstrates that the sponsor understands which attributes drive product performance. A submission that presents a titer and a purity profile without this underlying architecture is not an incomplete filing. It is a filing that signals to reviewers that the sponsor’s analytical development program has not yet reached the maturity required for clinical use.

The XGene AAV CQA Evidence Architecture is built on exactly this logic: a systematic mapping of each CQA to its regulatory basis, characterization method, specification rationale, and gap-to-license assessment, deployable from pre-IND through Phase 2/3 readiness reviews. The remainder of this article addresses the two CQAs that require the greatest development investment — full/empty capsid ratio and potency — and then the process-related impurity control package for HCP, HCD, and residual Benzonase.

Full/Empty Capsid Ratio and Potency: The Two CQAs That Require the Most Development Investment

Full/empty capsid ratio and potency are not simply two more items on the CQA list. They are the attributes that most consistently distinguish sponsors with a mature analytical development program from those who have not yet made that transition. The analytical investment required to establish and validate both is substantial, and the specification limits for each carry disproportionate regulatory weight.

The full/empty capsid ratio is determined by analytical ultracentrifugation — sedimentation velocity AUC-SV remains the gold-standard method recognized by CBER. The physical basis is straightforward: genome-containing (full) capsids sediment at a higher S-value than genome-empty capsids because they carry approximately 4.7 kb of single-stranded DNA inside a protein shell of fixed dimensions. For AAV5, full capsids sediment at approximately 115S versus approximately 73S for empty capsids. For AAV2, full capsids sediment at approximately 110S and empty capsids at approximately 60S. These S-value differences provide clean resolution of the two populations in a well-executed sedimentation velocity experiment, and the fractional area under each peak yields the percent full capsid content.

CBER’s expectation for clinical drug substance lots is a full capsid content of at least 90%. That figure is not arbitrary. It derives from the understanding that empty capsids — which retain full immunogenic potential while contributing no therapeutic genome copies — increase the total capsid dose required to deliver a given number of vector genomes, amplifying the inflammatory and immunogenic burden without contributing to efficacy. For AAV serotypes with known capsid immunogenicity profiles — and essentially all clinically deployed serotypes have documented immunogenicity — empty capsid burden is a direct variable in the safety calculation. A specification that sets the full capsid limit below 90% without clinical data or functional data justifying the lower limit will generate a deficiency request for the scientific rationale.

The primary alternative methods for full/empty determination each have recognized limitations in the CBER context. Transmission electron microscopy with negative staining provides direct visualization of capsid content — a well-executed TEM dataset with at least 100 particles per count is considered informative — but is labor-intensive, subject to sample preparation artifacts, and not easily automated for lot release. Ion exchange chromatography exploits the charge difference between full and empty capsids and can be developed into a higher-throughput analytical format, but method development is serotype-specific and the resolution between full and empty peaks must be carefully validated. Charge detection mass spectrometry is an emerging technology with genuine promise for accurate mass-based full/empty resolution, but it has not yet achieved the regulatory acceptance footprint of AUC-SV and is best positioned as a characterization tool rather than a release method in current filings.

The orthogonal method strategy for full/empty — AUC-SV as the primary release method, supplemented by TEM in the characterization package and IEX as a potential in-process tool — represents the current CBER-aligned approach. Sponsors relying exclusively on TEM or IEX for lot release without AUC-SV data in the characterization package will face questions about method adequacy.

Potency is the second high-investment CQA, and its regulatory treatment in AAV development is shaped directly by FDA’s 2011 guidance on potency tests for cellular and gene therapy products from CBER, which articulates the expectation that potency assays must measure the relevant biological activity of the product and that a biologically meaningful potency assay must be established prior to Phase 3 initiation. For early-phase INDs, CBER has historically accepted a tiered approach: a transduction efficiency assay — typically measuring transgene expression in an appropriate cell line by ELISA or activity assay — serves as the primary potency surrogate in Phase 1, with the expectation that a more clinically relevant potency measure will be developed and qualified by Phase 2/3. The VG:IU ratio functions as a process performance indicator but does not fulfill the biological potency requirement independently.

The cell-based potency assay must be designed with clinical relevance in mind. For a liver-directed AAV8 or AAV5 vector expressing a secreted protein, an in vitro transduction assay in a hepatocyte cell line measuring protein production or enzymatic activity is a defensible potency model. For CNS-directed vectors, the choice of cell line and the biological readout require more careful justification. The assay must be shown to be specific — responsive to the intended product activity — and must have a defined and justified acceptance criterion. Potency specifications expressed only as “within X-fold of reference standard” without definition of the reference standard itself, its qualification, and its assigned potency value are analytically incomplete.

USP <1047> provides additional context for the characterization of gene therapy vector preparations, and its framework for potency is consistent with the CBER guidance — biological activity measurement is the expectation, with titer-based surrogates acknowledged as interim tools during development. Sponsors who have not yet begun cell-based potency assay development at IND filing must include in their CMC development plan a specific milestone commitment for potency assay qualification, including the target Phase 2 entry readiness date and the interim control strategy — typically a combination of genomic titer, VG:IU, and transduction assay — that will manage the potency CQA until the validated biological assay is available.

Process-related impurities for AAV vectors divide into two principal categories: host cell protein and host cell DNA, both originating from the production system — whether HEK293-based triple transfection, baculovirus/Sf9, or HSV-based platforms — and process reagent residuals, of which Benzonase is the most consistently scrutinized. The analytical control strategy for each must be grounded in a process understanding that connects the purification unit operations to the anticipated impurity profile, and the acceptance criteria must be justified by safety data rather than by process capability alone.

Host cell protein quantification by ELISA is the standard release method. The most common platform for HEK293-derived AAV is a HEK293-specific HCP ELISA, and the critical analytical qualification question is assay coverage — whether the antibody population in the ELISA reagent recognizes the full diversity of HEK293 HCP species present in the AAV preparation at the stage of lot release. Two-dimensional differential gel electrophoresis or mass spectrometry-based HCP profiling in the characterization section supports the ELISA coverage claim and demonstrates that the lot release ELISA is not blind to a major HCP subset. HCP acceptance criteria in AAV DS specifications vary across the field, but values in the range of 1–10 ng HCP per 1×10^13 vg are commonly cited in peer-reviewed literature and regulatory filings. The limit must be justified by a per-dose HCP mass exposure calculation at the maximum intended clinical dose.

Host cell DNA control is addressed by qPCR or a validated DNA binding assay. The regulatory expectation — drawn from ICH Q6B and elaborated in CBER practice for gene therapy vectors — is that residual HCD should be reduced to levels where the per-dose DNA mass is below a threshold consistent with safety. Residual DNA limits of 10 ng per dose are commonly referenced in the gene therapy field, derived from legacy WHO recommendations for biological products, though CBER has not formally established a universal numeric threshold for AAV. The per-dose calculation must accompany any HCD acceptance criterion, and the calculation must use the maximum intended clinical dose, not a mid-range estimate.

Benzonase — endonuclease added during downstream processing to digest free nucleic acids and reduce HCD burden — must itself be controlled as a process reagent residual. A validated sandwich ELISA for Benzonase quantification is the standard approach, with commercial kits available. The acceptance criterion must be justified by a per-dose residual exposure calculation benchmarked against the known safety profile of the reagent. A specification that carries “NMT X ng/mL” for Benzonase without the accompanying dose-based justification — showing total Benzonase mass at the maximum dose and the safety margin relative to the established NOAEL — is analytically incomplete in the CBER context.

Endotoxin, sterility, mycoplasma, appearance, osmolality, and pH complete the safety attribute section. These attributes are well-established across biologics and do not require the same level of scientific development investment as full/empty or potency, but they must be fully validated, and their acceptance criteria must be clinically justified. Osmolality and pH limits must be consistent with the intended route of administration and the formulation design space. Appearance specifications must be sufficiently specific to detect visible particulates and significant color changes while avoiding over-specification that would reject lots presenting acceptable product-related heterogeneity.

The complete process-related impurity control package — HCP, HCD, Benzonase, endotoxin, and sterility — is evaluated by CBER reviewers not only for the individual adequacy of each assay but for the internal coherence of the overall control strategy. A purification process that is well characterized, with documented log-reduction values for each impurity across each unit operation, provides the manufacturing process understanding context in which the release specification limits make scientific sense. Without that context, even individually adequate methods and limits can appear arbitrary to reviewers.

The XGene AAV CQA Evidence Architecture

XGene AAV CQA Evidence Architecture Systematic CQA-to-Control Mapping for Pre-IND Through Phase 2/3 Readiness

CQA | Regulatory Basis | Characterization Method | Release Method | Specification Rationale | Gap-to-License Assessment

Mycoplasma | 21 CFR 610.30; | Culture + indicator cell | Per 21 CFR 610.30 | Not detected | Required pre-Phase 1 | USP <63> | methods | | |

Gap-to-License Risk Tiers (XGene Assessment): HIGH: Full/empty ratio (AUC-SV validation + ≥90% specification justified by batch data); Potency (cell-based assay development and QRS qualification for Phase 3) MEDIUM: HCP (assay coverage qualification + per-dose safety justification); HCD (per-dose calculation at max clinical dose); Benzonase (NOAEL-based limit justification) LOW-MATURE: Identity, genomic titer by ddPCR, endotoxin, sterility, mycoplasma, appearance, osmolality, pH — when methods are validated and specifications are clinically grounded

AAV Vector Characterization — The CQA Framework for CBER Analytical Review

The AAV CQA Framework: What CBER Expects Beyond Titer and Purity

Full/Empty Capsid Ratio and Potency: The Two CQAs That Require the Most Development Investment

The XGene AAV CQA Evidence Architecture

XGene AAV CQA Evidence Architecture Systematic CQA-to-Control Mapping for Pre-IND Through Phase 2/3 Readiness

CQA | Regulatory Basis | Characterization Method | Release Method | Specification Rationale | Gap-to-License Assessment

Mycoplasma | 21 CFR 610.30; | Culture + indicator cell | Per 21 CFR 610.30 | Not detected | Required pre-Phase 1 | USP <63> | methods | | |

Primary regulatory references

Find the right evidence, insight, or service.

The AAV CQA Framework: What CBER Expects Beyond Titer and Purity

Full/Empty Capsid Ratio and Potency: The Two CQAs That Require the Most Development Investment

HCP, HCD, and Safety Attributes: The Process-Related Impurity Control Package

The XGene AAV CQA Evidence Architecture

XGene AAV CQA Evidence Architecture Systematic CQA-to-Control Mapping for Pre-IND Through Phase 2/3 Readiness

CQA | Regulatory Basis | Characterization Method | Release Method | Specification Rationale | Gap-to-License Assessment

Mycoplasma | 21 CFR 610.30; | Culture + indicator cell | Per 21 CFR 610.30 | Not detected | Required pre-Phase 1 | USP <63> | methods | | |

Primary regulatory references