FDA Review of Viral Clearance and Adventitious Agent Control in Monoclonal Antibody BLAs Under ICH Q5A

How FDA reviews viral clearance and adventitious agent control in monoclonal antibody BLAs under ICH Q5A

FDA does not accept a negative final-product test as proof that a monoclonal antibody (mAb) is virally safe. Its guidance treats viral safety as three controls working together: selecting and testing cell lines and raw materials, testing at the right production stages, and showing that the purification process removes or inactivates endogenous and adventitious viruses ⁹³. Public BLA review packages and Complete Response Letters (CRLs) show reviewers applying that framework consistently. The issues they raise most often fall into a few groups. Scale-down models must be representative and run under worst-case conditions. Retrovirus safety factors must be calculated conservatively. Clearance must be re-evaluated after process changes. Resin lifetime claims must be supported. Raw-material and cell-bank provenance must be documented.

Key takeaways

The clearest public viral-safety approvability issue for a mAb in this set was in the April 2022 CRL for toripalimab-tpzi. Products made for non-US markets on equipment or materials shared with US toripalimab had not been tested for adventitious viruses at the unprocessed bulk stage ⁷⁹⁸⁰⁸¹.
Reviewers repeatedly asked sponsors to show that small-scale viral clearance models matched commercial operating conditions and represented a worst case (Yervoy, raxibacumab, Trogarzo, Besponsa, Darzalex) ¹⁵⁹¹⁶⁰³⁰⁹³¹⁰⁴⁷⁰³³³.
When a process changed, FDA asked for clearance to be reassessed instead of accepting the earlier validation (Cyramza process C, Darzalex post-change process) ¹⁶²³³³.
Some issues were handled after approval. Stelara carried postmarketing commitments for end-of-life resin viral removal studies ¹²⁴³⁶⁵. Ilaris carried a postmarketing requirement to replace a working cell bank made with non-US-licensed human serum albumin ²³²⁴⁴³.
Most mAb reviews ended with the standard conclusion that the product was free of endogenous and adventitious infectious agents to a degree that meets FDA-recommended parameters. This held even where information requests had been raised during review (Cyramza, Darzalex, Cosentyx) ³⁸⁰³³⁷⁵⁰⁶.

The regulatory framework reviewers apply

ICH Q5A(R2): clearance validation expectations

ICH Q5A(R2) calls for controlled, documented studies that identify effective removal or inactivation steps and estimate overall virus reduction. Virus is typically spiked into unprocessed bulk or process intermediates, and inactivation steps should be sampled over time to build an inactivation curve ²³⁸. Several design points come up again and again in FDA review comments:

Scale-down qualification. The model must represent commercial manufacture, including comparable yield and purity. Operating conditions such as chromatography contact time, buffers, resin, pH and temperature should be representative. Unavoidable deviations should be discussed, and worst-case conditions should be considered to test whether clearance holds up against material and process variability ²³⁹.
Virus selection. Model viruses should resemble potential contaminants and cover a range of physicochemical properties. Non-enveloped viruses deserve particular attention when testing general process robustness ²⁴⁰²⁴¹.
Step-wise assessment. Each step likely to clear virus should be assessed on its own, with titers measured before and after, and with reproducibility and relevant process parameters evaluated ²⁴²²⁴³.

Reduction factor arithmetic and retrovirus safety margins

Q5A defines a step's log10 reduction factor from virus load (titer multiplied by volume) before and after the step. The overall factor is the sum of the individual step factors. An individual factor should exceed 1 log10 to be used in that sum, because titrations are imprecise ⁵²². Reductions below 1 log10 are considered negligible. Adding factors from steps with similar mechanisms, from repeated near-identical procedures, or from steps that reuse the same or similar buffers or columns requires justification ⁵²³⁵²⁴⁵²⁵.

For endogenous retroviruses, estimated particles per dose is calculated as starting virus concentration multiplied by the volume of starting material per dose, divided by the clearance factor ⁵²⁶. The process should remove substantially more virus than is estimated in a single-dose equivalent of unprocessed bulk ⁵²⁷. For the CHO-cell Case B scenario, Q5A sets out a per-dose acceptability criterion for estimated retrovirus-like particles, provided testing does not identify infectious retroviruses ⁵²⁸; FDA reviews commonly describe the target as fewer than one particle per million doses.

What Q5A(R2) adds

The R2 revision introduces several provisions that sponsors now cite in mAb dossiers:

Platform and prior knowledge. Clearance data from similar products on an established platform may support a claim for the same step. The sponsor must justify comparability of the step and the intermediates, understand the clearance mechanism, and keep the claim conservative. If the package is not sufficient, product-specific studies are needed ²¹²².
Resin reuse. Resin lifetime and influential parameters should be defined, and reuse should be supported by clearance studies. For Protein A capture, product-specific end-of-life resin studies may not be needed where prior knowledge shows resin age does not affect clearance. Extending that position to other chromatography modes requires equivalent prior knowledge and detailed justification ²³²⁴.
Virus filtration. Throughput, pressure and pressure or flow interruptions can be critical parameters. If prior knowledge is used to claim parvovirus removal, at least one confirmatory product-specific parvovirus run under worst-case conditions is expected ²⁵²⁶²⁷.
Continuous manufacturing. The viral-safety principles used for batch processing still apply. Sponsors also need a risk assessment of connected operations, a justified testing frequency, and a sampling strategy suited to longer culture durations ⁸²⁸²⁹.
Molecular methods. Non-targeted next-generation sequencing (NGS) may replace in vivo assays and supplement or replace in vitro cell-culture assays without a head-to-head comparison, provided it is shown fit for purpose. Positive molecular findings should be investigated for association with infectious virus ³⁰³¹.

FDA's 1997 Points to Consider for mAbs

FDA's 1997 monoclonal antibody Points to Consider recommends the following cell bank testing ⁸⁸¹¹². Q5A(R2) updates parts of this menu (for example, NGS may replace some in vivo assays), so the list is not the current unchanged battery:

adventitious virus and retrovirus testing of the master cell bank (MCB) and end-of-production cells
species-specific virus testing of the MCB
sterility and mycoplasma testing of the MCB, the working cell bank (WCB) and end-of-production cells

It calls for lot-to-lot testing of unprocessed bulk for mycoplasma, adventitious and species-specific viruses, and retrovirus. Routine in vitro testing uses three indicator cell lines ¹⁰¹¹¹³. Generic or modular clearance studies are allowed, but within limits: applying them requires bulk viral-load data from at least three lots, a changed clearance module must be studied independently, and CBER should be consulted beforehand ¹¹⁶.

Recurring review themes and deficiencies in mAb BLAs

1. Representativeness of scale-down models and worst-case conditions

This is the most common viral-clearance request in mAb reviews.

Product (BLA)	Reviewer request or concern	Outcome in the record
Yervoy (ipilimumab), BLA 125377	Detailed scale-down operating conditions compared with commercial manufacturing, to show that the models represented the full-scale process ¹⁵⁹	Resolution not stated in the excerpt ¹³⁴
Raxibacumab, BLA 125349	Rationale for how certain clearance studies were run; chromatography flow rates and contact times ¹⁶⁰	Sponsor pointed to a linked report. FDA's view of the response is not shown ¹⁶¹
Trogarzo (ibalizumab-uiyk)	Validation plan "generally acceptable," but FDA asked how the conditions represented a worst case and said the justification would be a review issue in the BLA ³⁸⁸³¹⁰	Product quality assessment found the process robust for adventitious agent inactivation and removal, and approvable ³⁹¹
Besponsa (inotuzumab ozogamicin, antibody-drug conjugate)	The BLA lacked a viral clearance validation report. FDA requested the full report and evidence that the study process was representative ⁴⁷⁰	Reports and justification provided and found acceptable ⁴⁷¹
Darzalex (daratumumab)	Viral clearance study requested for the post-change process ³³³	Four model viruses, scaled-down models under worst-case conditions, and cytotoxicity, interference and PCR-recovery studies. Reviewer found the virus selection appropriate ³³⁴³³⁵³¹²

2. Conservative retrovirus safety factor calculations

For Trogarzo, FDA told the sponsor to calculate the virus safety factor from the highest retrovirus-like particle concentration found across the tested GMP and/or PPQ lots, and to calculate estimated particles per dose under Appendix 5 of ICH Q5A(R1). FDA also said it expected the commercial process to achieve higher retrovirus clearance. The numerical target is redacted ³⁸⁹. The point behind the request is that the safety margin should rest on the worst observed particle load, not an average.

3. Re-evaluating clearance after process changes

Before process C material was used in clinical trials of Cyramza (ramucirumab), FDA asked ImClone to reassess viral clearance, including retroviral load, and to explain whether the existing small-scale validation studies still supported process C ¹⁶²³⁷⁷. FDA later asked for information on viral clearance steps in connection with manufacturing-site changes ³⁷⁸. The final product quality review recommended approval ³⁸⁰. Darzalex followed the same pattern: process changes that could affect clearance led to a new study of the post-change process ³³³.

4. Resin lifetime and end-of-life clearance

At approval, Stelara (ustekinumab) carried postmarketing commitments for two studies ¹²⁴:

reduced-scale end-of-life viral removal studies for a redacted resin, run under conditions that reflect manufacturing scale
manufacturing-scale end-of-life validation of chromatography resins, assessing yield, chromatographic profile and impurities

The review template gives September 2010 as the target date for the viral removal final report ³⁶⁵. An approval-letter excerpt lists September 2011 ⁴⁸³. A June 2025 FDA letter states that all postmarketing requirements and commitments acknowledged in the 2009 approval letter were completed ³⁶⁷. Q5A(R2) now says end-of-life studies may not be needed for Protein A where prior knowledge supports it, while other resin types still need justification ²³²⁴.

5. Process parameter ranges must stay within validated clearance conditions

In a March 2018 CRL to Samsung Bioepis (BLA 761066), FDA said an upper process-parameter range was not supported by the viral clearance data. FDA asked that the parameter be classified as key and the range tightened. The letter lists this as an additional comment, not an approvability issue ⁸⁵. BLA 761066 is Eticovo (etanercept-ykro), and its review records the same request to tighten an operating-range upper limit that the clearance data did not support, which the reviewer later considered adequately addressed ³¹¹. Etanercept is an Fc-fusion protein rather than a mAb, but the expectation applies directly to antibody processes: proposed operating ranges should not go beyond the conditions under which clearance was demonstrated.

6. Cell bank qualification and 9 CFR compliance

For Darzalex, reviewers asked the applicant to confirm that a master cell bank contamination test met 9 CFR requirements. They also asked for a risk evaluation of a redacted material whose stage of use was not described in the manufacturing or viral safety sections ²³⁰²³¹. For Cosentyx (secukinumab), FDA questioned whether the information on how the production cell line was generated gave enough assurance about the MCB, and asked for data or additional testing on an MCB vial. FDA also could not open the viral test reports and asked Novartis to confirm that bovine testing, like the porcine testing, was done under 9 CFR ⁵⁰⁴. Novartis's substantial response resolved these issues, and the quality review listed no deficiencies ⁵⁰⁵⁵⁰⁶.

In several other mAb and biosimilar reviews, cell bank adventitious agent testing was judged adequate without further comment, including Riabni (rituximab-arrx), Entyvio SC and Tepezza ²²⁵²²⁶²²⁷.

7. Raw materials, animal-derived components and TSE risk

Reviewers routinely list viral contamination as a raw-material risk in their assessments. Examples include secukinumab, fremanezumab, adalimumab-aqvh, bevacizumab-maly, rituximab-abbs and inebilizumab-cdon. These entries record a risk, not a control failure ⁶²⁶³⁶⁴⁶⁵⁶⁶⁶⁷.

Cosentyx. FDA asked for an assessment covering animal-sourced materials, raw-material testing, whether the assays could detect viruses introduced through those materials, and whether wild-type viruses might grow less readily than laboratory stocks ⁵⁰³.
Ilaris (canakinumab). The cell banks had been made with human serum albumin from non-US-licensed sources. After consulting FDA TSE experts, reviewers judged the transmission risk extremely low and not a bar to approval. They did not ask for a new MCB, because they judged the risk from changing the product to be greater than the TSE risk ²³². FDA did impose a postmarketing requirement for a new WCB made with US-licensed HSA. The protocol was to preserve the validated cell-generation time and be submitted as a prior-approval supplement ⁴⁴³⁴⁴⁴.
Enspryng (satralizumab). The review credits an animal-origin-free process with minimizing viral risk, while still identifying raw materials and personnel as possible contamination sources ⁷¹.

8. Unprocessed bulk testing and cross-contamination

In most mAb and biosimilar facility assessments, unprocessed bulk mycoplasma and in vitro adventitious virus testing were recorded as acceptable or approvable without findings. Examples include Ilumya, Kanjinti, Avsola, Riabni, Truxima and Herzuma ²⁷³²⁷⁵²⁷⁷²⁷⁸²⁷⁹²⁸⁰.

The toripalimab-tpzi CRL (Coherus, BLA 761240, April 29, 2022) is the exception. FDA found that unprocessed bulk from products made for non-US markets, on equipment or materials shared with US toripalimab, had not been tested for adventitious viruses. FDA treated this as a cross-contamination risk and as a virus control strategy inconsistent with ICH Q5A ³⁹²⁵⁴¹⁵⁴². FDA required the following:

unprocessed bulk virus testing of every batch of every product sharing product-contact equipment or materials ⁵⁴³
replacement and qualification of higher-risk multi-use materials, and virus-inactivating sterilization of shared equipment where feasible, with supporting justification ⁵⁴⁴
US drug product made only from drug substance manufactured after these measures. FDA said earlier batches should not be distributed for human use in the US ⁵⁴⁵
a full rationale showing the revised strategy met ICH Q5A ⁵⁴⁶

The lesson for multi-product and multi-market facilities is that FDA's adventitious agent review extends beyond the US product to anything that shares product-contact surfaces with it.

9. Incomplete clearance packages at action

Two other CRLs and resubmissions show FDA's attitude to clearance data that is still pending:

A May 2018 CRL to Evolus for BLA 761085, Jeuveau (prabotulinumtoxinA-xvfs), a botulinum toxin rather than a mAb, noted that viral clearance evaluations were still ongoing and asked for the results at resubmission ⁸⁴.
The Trodelvy (sacituzumab govitecan, an antibody-drug conjugate) resubmission included additional viral clearance data and an updated clearance section. This followed broader findings that the commercial process was not adequately validated ¹⁶⁴⁵¹⁵⁵¹⁷.

What non-mAb reviews show about reviewer scrutiny

Some of the most detailed viral clearance criticism in the public record comes from non-antibody biologics. It shows how reviewers think about the same Q5A principles:

Fabrazyme (agalsidase beta). The reviewer found that resin-cleaning studies did not show that viruses built up on a column would be removed or inactivated ¹²⁷. The reviewer judged modeled retrovirus clearance insufficient ¹²⁸ and said the virus panel left out highly resistant virus families that should have been included ⁴²⁸.
Retacrit (epoetin alfa-epbx). FDA asked for a risk assessment of a 14-day in vitro assay on unprocessed bulk used in place of a 28-day assay with amplification, because slowly replicating agents might be missed ²³⁴.
Zenpep (pancrelipase). Reviewers criticized duplicate samples used in place of independent inactivation experiments, and asked for data supporting the toxicity and interference dilutions ³¹³.

Practical implications for mAb sponsors

Document scale-down equivalence in the BLA itself, using parameter-by-parameter comparisons such as flow rate, contact time, load and pH. State explicitly why each spiking run represents a worst case ¹⁵⁹¹⁶⁰³¹⁰.
Base the retrovirus safety factor on the highest particle count observed across GMP and PPQ lots, and show particles-per-dose calculations under Q5A ³⁸⁹⁵²⁶.
Treat any change to a clearance-relevant step as a trigger to reassess clearance, and set proposed process ranges within validated conditions ¹⁶²³³³⁸⁵.
Justify resin lifetime claims. For non-Protein A media, plan end-of-life clearance data or a documented prior-knowledge justification ²³²⁴¹²⁴.
Trace every animal- or human-derived material used in cell bank creation and manufacture. Confirm 9 CFR compliance for bovine and porcine testing ⁵⁰⁴²³².
In shared facilities, extend unprocessed bulk adventitious virus testing and equipment controls to every product that shares product-contact surfaces with the US product ⁵⁴³⁵⁴⁴.

Redaction limits how far the public record goes. Most reviews withhold log reduction values, model virus results and safety factor targets, so the published record shows what FDA asked for more clearly than the numerical thresholds it applied ³⁸⁹¹²⁸¹⁶⁸.