Clone stability in bioprocessing: A deep dive into current clone stability trials

20 March 2021

Valitacell have partnered on clone stability trials with many biopharmaceutical companies. In this post we review what a typical clone stability trial looks like, suggest how trials might be improved, and outline future trends.

by Dr Eligio F. Iannetti & Dr Paul D. Dobson

Clone stability trials are a crucial component of the Cell Line Development (CLD) process. Assessment of clone stability is necessary to demonstrate to regulators that the cell produces a consistently high-quality product, and to help production managers guard against product loss. Given the notoriously high genetic instability of Chinese hamster ovary (CHO) cells, which dominate mammalian protein production, this is a formidable challenge.


CHO cells are infamously genetically unstable, from small point mutations right up to changes as large as altered chromosome structure and even chromosome number. While genetic alterations can bubble up to disrupt the production phenotype via any number of host cell mechanisms, it has been noted that depleted (lost or silenced) transgene copy number is a particularly significant driver of production performance loss. This appears to be related to random integration into unstable genomic loci that are by their nature apt to accommodate transgenes, but that very same nature also makes them likely to relinquish those transgenes readily. This is a major opportunity for improvement. With the arrival of CRISPR-mediated precision genetic modification, transgenes can now be targeted with exquisite precision into more stable genomic loci from which they will not be so readily lost. Such a strategy has already been shown to enhance stability in prolonged culture (1). While random integration likely will prevail across CLD for some time, the transgene stability enabled by targeted integration through precision molecular biology makes the rise of this strategy seem inevitable. The impact could be profound, but it is no panacea: the CHO genetic background will remain highly plastic and, as we discuss below, new classes of unnatural therapeutic proteins are already generating novel clone stability challenges.


Although a loss in titer is of course a concern for production managers, the ICH Q5D guidelines (2) mandate stability trials primarily to secure consistent product quality (the logic being that a consistent cell should yield a consistent product). The Q5D guidelines only loosely sketch out the broad principles of a stability trial but they do not closely mandate a specific experimental design. Nevertheless, industry has converged upon a widely accepted trial design based upon monitoring titer change throughout prolonged subculture. Generally, this is applied once CLD has reduced clone numbers down into the low tens of candidates, but even so it remains a major laboratory undertaking.

Typically, a trial will last for as long as a production run (~60 generations), although across Valitacell’s wide experience as a trial partner we have seen trials as short as 40 and as long as 200 generations. Titer loss, or in some companies titer change (wherein an increase is also considered evidence of undesirable change), of not more than 30% is taken as evidence of clone stability. Most companies only take measurements at the start and end of trials, although we have been fortunate to work with companies that also take intermediate measurements.

Most trials are conducted in shake flasks or scale-down reactors, with very few of our partners assessing stability in sizeable (>500mL) reactors. The lack of correlation between small bench experiments and production volumes is well known, but as a trial is intended to quantify cell stability rather than simulate what it will do in production, we do not consider this a major shortcoming.

There is a notable lack of biological replicates in most stability trials, principally due to workload. This cannot sit well with any scientist, but practically it is difficult to know how to address this as mandating replicates could make the entire approach unworkable. If trials persist (a question we address below), then thought must be given to automation.


Valitacell have been a partner in clone stability trials with many companies, ranging from small CMOs to some of the world’s biggest biopharmaceutical manufacturers. While each company’s trial designs have their idiosyncrasies, curiously, the metrics and thresholds by which results are judged vary remarkably little. This is not a good thing in our opinion.

Consider stability trial duration, which varies across companies and even within a company by project. Clearly there is more opportunity for a clone to change in a long trial than in a short one, yet the 30% titer loss threshold is applied almost universally. Analyses should account for trial length.

Perhaps more concerning is the reliance upon percentage titer change. To be clear, this is the change in titer from start to end expressed as a percentage of the initial titer. It is a measure of both initial value and change. Owing to this, it can sometimes be a little difficult to interpret.

·      If clone A has an initial titer of 6g/L and loses 2g/L over the trial, it has a percentage titer change of -33%. At the -30% titer loss threshold, it is considered unstable.

·      If clone B has an initial titer of 7g/L and also loses 2g/L over the trial, it has a percentage titer change of -28.5%. It is considered stable.

For both clones, titer drops by the same amount but, solely because they started from different initial titers, their stability calls are different. This matters in practice. In our collaborations Valitacell have encountered many real-world clones with similar rates of titer change but different initial titers, leading to different stability calls.

We are not suggesting to disregard percentage titer change as it is valuable to consider the rate of change in context. Clearly losing 1g/L from an initial titer of 2g/L (-50%) is much more serious than losing the same from a 10 g/L start (-10%). The general point is that lumped statistics like percentage change discard useful information. To maximise our understanding of how the cell has changed, Valitacell’s current practice is to look at both initial values and changes per generation.


The ICH Q5D challenge is to demonstrate clone stability as this is seen as guarantor of product quality. Yet trials focus mainly upon titer. Does a stable titer imply stability of the underlying cell? Emphatically, it does not.

We have analysed stability trial data from hundreds of clones by dissecting titer into rate of change per generation of specific productivity (qP) and rate of change per generation of the integral of viable cell density (IVCD, a measure of growth). We identified slow-growing, high producing clones that, over the trial, became fast-growing, low producers, and vice versa. Often these changes ran contrary to each other, so their net effect on titer was negligible. At worst, we found a clone that dropped qP by 70% and gained 117% IVCD, but barely changed titer.

What we hardly ever found was a clone that maintained its titer by maintaining IVCD and qP. Underneath titer, cells in prolonged culture are almost always changing at a deeper level. We might call this scenario titer stability, but it certainly is not clone stability.

The lesson is clear: if a stability trial looks only at a high-level metric like titer it will miss profound underlying cellular changes, not only to IVCD and qP but also beneath these attributes. Omics methods could provide the missing deep cellular readouts but will undoubtedly prove challenging to integrate into the already labour-intensive and costly stability trial process. In our next post in this series we will explain how we have used our ChemStress® platform to provide these insights in a cost-effective manner.


Despite the trial design and analysis limitations above, manufacturers are typically able to produce sufficient quantities of safe, high-quality biologics. Largely this is because the sector has vast experience manufacturing major classes of drugs like monoclonal antibodies (mAbs), so we are well positioned to anticipate pitfalls when making novel mAbs with similar structures. Parental cells are available that consistently give rise to good mAb-producers, media has been tailored for mAb production, and the process engineering methods needed to iron out common mAb issues are well established. We also have robust quality control methods to catch product quality deviations.

Consequently, many biopharmaceutical manufacturers will claim their routine mAbs can be made safely without trials. If trials do hold value for manufacturers, it is a way to avoid titer loss. Layer in targeted transgene integration to mitigate against titer loss, and the question is then whether clone stability trials are necessary for well-established product classes like mAbs or not? In short, if cellular instability is not a major issue that limits mAb titer or quality, and there is no obvious link between stability of titer and stability of product quality, do manufacturers need to suffer the time and expense of full stability trials?


Which biologic categories cannot be produced with tried-and-tested methods? Valitacell have worked with several manufacturers trying to produce difficult-to-express (DtE) proteins. Unusual proteins place unusual burdens upon the cell. Cellular stresses induced by DtEs are often enhanced in large scale bioreactors of production setting. Many DtE manufacturers find themselves challenged by profound production instability caused by cellular adaptation in response to adverse or varying environmental changes (bioreactor stress conditions). In this scenario, stability trials really do need to identify fundamentally stable, stress-tolerant clones. For DtE proteins there is a profound need to understand what makes a production cell tick and how robust these mechanisms are likely to be at scale. Apparent stability tells us none of this can be inferred simply from the stability of high-level metrics like titer. This can only happen with tools to look deeper into cell function stability.


Clone stability trials are a regulatory requirement that place a great burden upon companies. Running a stability trial is very time-consuming and costly. The absence of biological replicates and less-than-ideal trial analyses in current use could be improved, but to-date this has not proven to be a barrier to the safe and economic production of major biologics classes as we have the wisdom and methods in place to minimise errors and catch mistakes.

The main stability challenge lies in DtE proteins that impose unusual stresses upon the production cell. We do not have the wealth of experience with DtEs that we have for major biologics classes, so stability trials really must identify stable clones that are resistant to production stresses and stand a good chance of maintaining this throughout culture in production.

 The improvements to stability trial analyses we have developed should cut through the data fog to give more insightful results, but to understand clone stability in depth we will need to assay fundamental cellular functions throughout trials.

 The next post in this mini-series on clone stability will describe how we use our ChemStress® technology to profile cell functionality and flag apparent stability to identify deeply stable clones.


1.     Grav, L.M.; Sergeeva, D.; Lee, J.S.; de Mas, I.M.; Lewis, N.E.; Andersen, M.R.; Nielsen, L.K.; Lee, G.M.; Kildegaard, H.F. Minimizing clonal variation during mammalian cell line engineering for improved systems biology data generation. ACS Synth. Biol. 2018, 7, 2148–2159.

2.     ICH Q5D Derivation and characterisation of cell substrates used for production of biotechnological/biological products