Cell density measurement

The following case study describes how in-line permittivity was utilized in a viral expression animal culture to estimate VCD in-situ and identify optimal time for viral infection using a continuous predictive model.

The Role of Critical Time Points in Biopharma

With the adoption of Process Analytical Technologies (PAT) and Good Manufacturing Practice (GMP) guidelines, there is an industry wide push towards data-driven characterization of bioprocess operations to increase processing consistency, reduce waste, and improve overall product quality. In upstream cell culture operations, in-line sensors, at-line sampling systems, and off-line sampling analytic techniques are all utilized to effectively characterize and control all critical process and environmental variables encompassing cell culture operations.

Historically, process control strategies are critical to initiating or extending culture productivity. These strategies include temperature shift growth control, bolus feed additions, and infection timings. Process control often relies upon either a fixed time for execution or pre-established Key Performance Indicators (KPIs) or Critical Process Parameters (CPPs) to determine an appropriate time of action.

Fixed-timing strategies are effective and simple to implement but cannot account for batch-to-batch variation inherently present in complex biological systems. A common example of process control in cell culture based on a fixed time strategy is staging bolus feed additions of a predefined formulation at defined time points within a culture. These additions effectively extend a culture life cycle but often result in inefficient use of feed materials, such as an over addition of expensive media components to avoid detrimental growth impacts. Fixed-time feed strategies also do not account for variations in growth metrics between production runs, leading to an un-optimized feed to growth profile.

Control strategies based around process KPIs such as viable cell density (VCD) or in-situ glucose concentrations can be advantageous at mitigating biological variation and improving control effectiveness. In these cases, evaluation of KPIs and CPPs typically relies on discrete sampling techniques to extrapolate process control decisions, such as using off-line VCD to predict the culture time when a target VCD will be reached. While possible with off-line VCD sampling, limited data points and inherent measurement variations impact the precision and usefulness of off-line-derived models for determining process actions.

For some off-line KPIs there exist continuous inline sensors that provide either identical or analogous signals to current off-line measurements. In-line permittivity provides one such sensor, generating a continuous signal which correlates with off-line viable cell density.1 In-line permittivity provides a data-driven avenue of improving predictive model robustness as well as enabling real-time action and automation from the signal. Application of in-line permittivity for process calculations and bolus feed control are discussed in the Integral of Viable Cell Density in Biopharma paper. This in-line permittivity control can be applied to a wide variety of cell lines and applications, including the gene therapy field.

cell culture predicting timeframe measurement

Discover more with our Predicting Actionable Timeframes in Cell Culture Using In-line Cell Density Measurement Article.

Download

Case study: Optimizing Infection Timing with Incyte

Recombinant adeno-associated viruses (rAAVs) are a growing class of therapeutics in the gene therapy space, owing to versatility in targeting both dividing and non-dividing cells, longevity of transgene expression, and minimal adverse patient effects in clinical trials.2

The insect baculovirus expression vector system is one ideal method for scalable rAAV production due to high volumetric productivities and the absence of mammalian-derived products.3 While run traditionally as a batch style process, the robustness of an SF9 infection culture is dependent upon the critical time points of infection and harvest. Infection timing is important for consistent cell-specific productivities and overall titer, as the cell density at time of infection directly influences the efficacy of infection and subsequent viral productivities. Too early an infection and overall cell concentration is detrimentally impacted, while too late an infection results in a lower efficacy of infection per cell. Harvest is also an important time point in SF9 baculovirus processes, as harvesting too late leads to excessive cell death and release of proteases which complicates downstream processing and impacts product quality. The chosen harvest time should maximize viral product concentration while minimizing the presence of cellular material released due to apoptosis.

In the following case study, an rAAV vector was expressed via SF9 infection using an in-line permittivity sensor to monitor and predict optimal time of infection.4 Previously, the optimal time of infection was determined by using viable cell counts from an off-line viable cell counter to estimate when a target VCD is reached within the culture. Due to the limited amount of off-line VCD data points able to be feasibly generated early in the culture, predictive models developed with this data set suffered from limited predictive capacity and inability to compensate for an incorrect cell count unless consistent sampling was performed close to estimated time of target VCD (Figure 1). In-line permittivity measurement was proposed as a continuous solution to improve predictive model robustness, while minimizing the need for manual intervention as cultures approach the target VCD.

Predicting-Actionable-Timeframes-GRAPH-

Figure 1. Predicting time to infection using a linear model from limited offline VCD samples.

In order to utilize permittivity to estimate VCD continuously in-situ, a linear correlation was first established between off-line VCD and permittivity during the growth phase of replicate SF9 runs. This generated an optimal permittivity-equivalent VCD measurement spanning the exponential growth phase of the desired SF9 cultures. The continuous permittivity measurement was then input into a linear predictive model, which utilized the slope of the growth curve to estimate the time post seeding when the optimal permittivity value would be reached. The inclusion of a continuously measured signal into the predictive model provided significant improvements in terms of automation and precision compared to using off-line data inputs.

The implementation of a time dependent linear weighting scheme allowed for the inclusion of all permittivity data into the linear prediction model. Heavier weighting of newly measured data effectively captured the change in culture growth rate during the non-linear transition from lag phase to exponential growth phase. The improved predictive power of the linear model provided estimated infection times well in advance of the action time point without any additional sampling required.

The in-line permittivity-predicted model was then tested in ten different batches with various seeding densities to gauge predictive robustness in the presence of different starting densities and expected culture growth characteristics. The permittivity model converged within an hour of the actual time of target VCD over 24 hours in advance for all seeding densities, allowing sufficient time to prepare the baculovirus infection for all batches (Figure 2). The predictive model facilitates reproducible and precise infection densities, calculations for various parallel runs performed in advance, and preemptive scheduling of labor and resources to maximize operation efficiency.

https://assets-sensors.hamiltoncompany.com/File-Uploads/Predicting-Actionable-Timeframes-GRAPH-2.jpg?v=1715358345

Figure 2. Predicting the time of infection ~24 hours before action is required with a continuous permittivity signal.

Extending Prediction Strategies

While this case study demonstrates the effectiveness of a continuous permittivity measurement for predictive analytics in an infection process, the idea of in-line measurements for predicting process decisions extends to many other applications. A predicted time to target VCD can be used to determine other critical time points such as peak exponential growth within a culture or to optimize bolus feed additions.

Additionally, predictive models can be used with other in-line measurements of cell density, such as total cell density via NIR optical density (OD) measurement. Off-line OD, typically measured with OD600 UV/Vis spectrophotometry, is used in many bacterial applications as a straightforward method of monitoring growth throughout the duration of a fermentation. In-line OD measurement with a Hamilton Dencytee sensor can provide a correlated signal to off-line OD. Predictive modeling using this signal can determine critical time points such as time to perform a bacterial induction or when a fermentation will achieve maximum OD. Using in-line measurement for predictive control, coupled with the prospect of automating process actions at a predicted time point is an essential technology for improved biopharma process control and reproducibility.

References

  1. The Integral of Viable Cell Density in Biopharma.
  2. Wang, D.; Tai, P. W. L.; Gao, G. Adeno-Associated Virus Vector as a Platform for Gene Therapy Delivery. Nat. Rev. Drug Discov. 2019, 18 (5), 358–378. https://doi.org/10.1038/s41573-019-0012-9.
  3. Kotin, R. M. Large-Scale Recombinant Adeno-Associated Virus Production. Hum. Mol. Genet. 2011, 20 (R1), R2–R6. https://doi.org/10.1093/hmg/ddr141.
  4. Pais, D. A. M.; Brown, C.; Neuman, A.; Mathur, K.; Isidro, I. A.; Alves, P. M.; Slade, P. G. Dielectric Spectroscopy to Improve the Production of RAAV Used in Gene Therapy. Processes 2020, 8 (11), 1456. https://doi.org/10.3390/pr8111456.
fade out

Download this Article in PDF Format

https://assets-sensors.hamiltoncompany.com/File-Uploads/Download-PDF.jpg?v=1715358373

Download the Predicting Actionable Timeframes in Cell Culture Using Inline Cell Density Measurement Article

Fill out the form below to download a PDF format of this article.