Beyond DoE: Mechanistic Modeling, Viral Surrogates and ICH Q5A(R2) for Next-Gen Virus Clearance

Design of Experiments is the natural first answer to the limitations of worst-case virus clearance testing. Structured designs resolve individual parameter effects and reveal interactions, which is exactly what a condensed worst-case study cannot do.

But DoE runs into a hard constraint in this particular domain. Designed experiments need runs, and virus clearance runs are severely rationed by cost, by material demand, by BSL-2 laboratory access and by limited assay capacity. A design that would be routine in upstream development is often simply unaffordable when every run requires an infectious virus spike in an external facility.

The answer is not to abandon DoE. It is to surround it with methods that reduce how much of it must be executed with infectious virus - and a revised regulatory framework that permits exactly that.

ICH Q5A(R2): the door that opened

Recent revisions to ICH Q5A reshaped what is acceptable in virus clearance strategy, moving decisively toward science-driven, risk-based validation supported by prior knowledge and genuine process understanding.

Two changes matter most in practice. Regulators now accept alternative approaches - notably the use of CHO-derived endogenous virus-like particles as surrogates for infectious model viruses. And the revised guidance explicitly recognizes prior knowledge as a legitimate basis for parameter classification, rather than requiring every claim to be re-demonstrated from scratch.

Together these convert virus clearance from a compliance exercise executed at a fixed point into a knowledge-building activity that runs across development.

Viral surrogates: making experimentation affordable

Surrogates are the practical enabler. Because CHO-derived virus-like particles do not require the containment and specialized facilities of infectious model viruses, they can be used far more freely - which means designed experiments become feasible at a scale that would be impossible otherwise.

The pattern that emerges is a division of labour. Surrogate-spiked DoE studies map the design space and identify which parameters influence clearance. Confirmatory infectious-virus studies then validate the conclusions at the conditions that matter. Expensive runs are spent on confirmation rather than exploration.

Mechanistic modeling: predicting before you pipette

Mechanistic models describe the underlying physics and chemistry of a unit operation rather than fitting a statistical surface to observations. For anion exchange chromatography - a workhorse virus removal step - models can describe virus removal behaviour directly, and the approach has been developed in the peer-reviewed literature.

The practical value is threefold:

Simulation replaces some experimentation. Process variations can be explored in silico, so laboratory work concentrates on genuinely open questions.

Deviations become assessable. When a commercial batch drifts outside its normal operating range, a validated mechanistic model can support risk evaluation of that deviation - a far stronger position than arguing from a worst-case study that did not probe the relevant condition.

Developability improves. Modeling early in development identifies whether a process design will deliver adequate clearance before substantial investment is committed.

Prior knowledge, made structured

The third pillar is the least technically glamorous and often the most valuable: systematic use of historical and platform data.

Prior knowledge only functions as evidence if it is structured. A parameter can be classified using platform data if that data is contextualized, comparable and retrievable - not if it sits in a report from a program three years ago that no one can locate. This is why organizations that invest in internal databases of clearance behaviour compound an advantage: each program starts from a stronger evidence base than the last.

Prior-knowledge strategies also include referencing structured industry standards - ASTM E2888, E3042 and E3259-22 - to justify proportionate claims and parameter criticality under defined conditions.

The hybrid phase, and what it demands

None of this eliminates infectious-virus studies. The field is in a prolonged hybrid phase where classical studies coexist with model-enabled approaches, and that phase will last years.

What changes is the balance. Programs equipped with surrogates, models and structured platform data spend their expensive infectious-virus runs confirming a well-formed hypothesis. Programs without them spend the same runs discovering what they should have hypothesized - and remain fully dependent on external laboratory scheduling, with all the timeline risk that carries.

Where the lifecycle logic lands

A point worth making explicit: control strategies legitimately differ between clinical and commercial stages. When clinical set-point studies and commercial worst-case studies produce comparable log reduction factors - for instance both below the limit of quantification - collective down-rating of tested parameters is justified, provided no relevant process changes occurred between phases.

That is not a loosening of standards. It is the lifecycle principle applied honestly: evidence accumulated at one stage should inform, rather than be discarded at, the next.

Getting started

For organizations building these capabilities, a sensible sequence is: conduct a structured gap analysis to find where modeling and data-driven tools would add measurable value; pilot the methods on selected projects to build internal expertise and demonstrate benefit; build up structured, contextualized platform data; and invest in the training and change management that embed new methods in existing workflows.

The destination - digital twins and fully model-based clearance claims - is still ahead. The steps toward it are available now, and each one reduces both cost and timeline risk on the way.

---

Rentschler Biopharma's downstream process science team applies mechanistic modeling and surrogate-based studies to virus clearance. To explore a model-supported strategy, contact our Business Development team.