You are running an unsupported browser, please upgrade your browser. Close

Cost Modelling Comparison of Static, Suspension and Fixed Bed Bioreactors to Manufacture Commercial Gene Therapy Products



Emmanuelle Cameau, Andreia Pedregal, Clive Glover




Recombinant adeno-associated virus (rAAV) vectors are emerging as the most important virus for in vivo gene therapy. Despite significant advances in development of suspension cell lines and transfection methods, viral vector batches for clinical trials are still largely produced by scaling-out the laboratory methods in static 2D multitray stacks (MT). However, the use of such process in compliance to manufacture commercial material under Good Manufacturing Practice (GMP) conditions leads to very high manufacturing costs in addition to introducing large risks into the process.


To avoid this, cost modeling tools can be used during process development to ensure that manufacturing process being developed will be economically viable for commercial production. This study compares three upstream (USP) methods for production of rAAV using models generated in BioSolve software, a reference for costs analyses in the biopharmaceutical industry.


The traditional MT-based process is compared to suspension and fixed-bed processes based on iCELLis® bioreactor (Pall Biotech). Clinical scale (200 L) and manufacturing scales (800–1000 L) are considered for each model, and downstream process (DSP) and quality control (QC) are included to give realistic idea of cost of goods (CoGs). Cost structure is determined and different scenarios are compared to identify parameters that can leverage CoGs such as quantity of GMP-grade plasmid DNA (pDNA), the essential reagent for transient transfection. The optimization of the iCELLis bioreactor process appears to be a very promising alternative for transient transfection viral vector manufacturing.





flowchart of raav production process using either mt10 suspension bioreactor or icellis bioreactor



  • MT process: based on literature data
  • Suspension process: based on existing process optimized in 200 L extrapolated at 1000 L
  • iCELLis bioreactor process – 2 sets of data:
    • Direct process transposition from benchmark productivity (MT)
    • Optimized process based on experimental data



parameters used for throughput and cost modeling of 3 processes at clinical and manufacturing scales. processes rely on single use technologies for usp media and buffer preparation



BioSolve software generates a comprehensive view of a factory including:


  • Footprint required and capital investment
  • Media, buffers and reagents utilization (pDNA)
  • Process duration and workload
  • Consumables in USP Suspension





Single-Use Bioreactors vs. Multi-Trays


  • 30 to 40% dose cost reduction on USP operating expenses allowed by bioreactors
  • Footprint x2 between clinical and manufacturing scales for MT; stable for bioreactors
    • MT in USP increases OPEX
    • CAPEX can become a NO-GO factor with MT
    • Same factory for clinical and industrial phases with bioreactors


iCELLis Bioreactor vs. Suspension Bioreactor


  • Similar CAPEX, OPEX and footprint
  • Non-optimized: 30% decrease on dose cost / Optimized: up to 44% savings
  • Non-optimized: 14% lesser dose throughput / Optimized: 8% more throughput
    • iCELLis bioreactor process optimization allows CoGs reduction over suspension bioreactor
    • Choice of technology should be based on process development risks, lesser for fixed-bed bioreactor



figure 2 gogs per dose and total doses produced per year with each process configuration figure 3 installed capital process footprint and labor fte usp and dsp for each process





  • Benchmark process modeling reveals significant impact of labor and USP consumables
  • GMP-grade pDNA is a common cost driver in all processes
    • Importance of optimizing pDNA use
  • Number of iCELLis bioreactor batches/year is lower because the bioreactor is used for a longer time during production (7 days for iCELLis bioreactor vs. 6 days for suspension bioreactor), see Figure 1



figure 4 repartition of cogs per category of cost for manufacturing scales. capital charge is spread over 8 years considering 12% interest and 10% future value. other includes utilities maintenance waste management insurances and hse.





  • Suspension and non-optimized iCELLis bioreactors (worst case scenario) are similar in terms of CAPEX/OPEX 
  • iCELLis bioreactor experimental results show decreased pDNA use in transfection compared to suspension[1]
  • Transient transfection in 1000 L suspension bioreactor is very challenging due to the quantity and cost of pDNA
  • Further optimization of rAAV titer can be achieved by optimizing feed medium quantities
  • Process throughput can be increased by pooling 2 bioreactors in DSP
    • iCELLis bioreactor process has high chances to significantly outdo suspension process for transient transfection production of AAV for the scales considered



figure 5 cost of dose for different scenarios at manufacturing scales in % of mt dose cost. worst case: no productivity gain in fixed bed. Realistic case: pDNA reduction with over 20% of cells transfected.





This study emphasises the cost-inefficiency of static system (MT) and the impossibility for this option to provide an affordable large scale manufacturing method for expensive therapies.


Use of high-producing suspension cells in bioreactors allows significant costs reductions in installed capital and labor, but the cost of GMP-grade pDNA still represents up to 40% of total batch cost. At this point, a limit is reached in CoGs reduction that can hardly be overtaken. Transient transfection at 1000 L is challenging due to the quantity and cost of pDNA required.


Use of fixed-bed bioreactor is a viable option for further process optimization as it will allow productivity increase and reduction of pDNA quantities used in transfection. This approach is being followed by several groups that already reported their achievements: Saint Jude Hospital reported rAAV production levels in iCELLis Nano bioreactor (small scale bioreactor) similar to static controls[2].


Further process optimization not explored here is the selection of rAAV-secreting cell line and operation of process in perfusion mode. This approach has been described in literature for AAV production[3] and has already been applied at large scale for adenovirus vectors production taking advantage of iCELLis bioreactor built-in perfusion capacities[4]. 


By their flexibility and potential costs reductions using currently available biological systems for viral vector production, fixed-bed bioreactors play a major role in transient gene therapy production.





[1] Reniers et al., 2016, ESGCT poster

[2] Powers et al., 2015, HGT Methods

[3] Grieger et al. 2015, Mol. Therapy

[4] Lesch et al., 2015, Human Gene Therapy



Download Poster