key: cord-0966443-kr6t80j5
authors: Kis, Z.; Tak, K.; Ibrahim, D.; Papathanasiou, M. M.; Chachuat, B.; Shah, N.; Kontoravdi, C.
title: Pandemic-response adenoviral vector and RNA vaccine manufacturing
date: 2021-09-02
journal: nan
DOI: 10.1101/2021.08.20.21262370
sha: 5db0ff6a9282d112eb97376d3225c62a3c1d0864
doc_id: 966443
cord_uid: kr6t80j5

Rapid global COVID-19 pandemic response by mass vaccination is currently limited by the rate of vaccine manufacturing. This study presents a techno-economic feasibility assessment and comparison of three vaccine production platform technologies deployed during the COVID-19 pandemic: (1) adenovirus-vectored (AVV) vaccines, (2) messenger RNA (mRNA) vaccines, and (3) the newer self-amplifying RNA (saRNA) vaccines. Besides assessing the baseline performance of the production process, the impact of key design and operational uncertainties on the productivity and cost performance of these vaccine platforms were also evaluated using variance-based global sensitivity analysis. Cost and resource requirement projections were also computed for manufacturing multi-billion vaccine doses for covering the current global demand shortage and for providing annual booster immunizations. This model-based assessment provides key insights to policymakers and vaccine manufacturers for risk analysis, asset utilisation, directions for future technology improvements and future epidemic/pandemic preparedness, given the disease-agnostic nature of these vaccine production platforms.

The COVID-19 pandemic, caused by the SARS-CoV-2 virus, created an unprecedented demand for rapid, large-scale vaccine deployment that the world is struggling to meet. This urgency and scale of immunization against a new disease poses enormous challenges on the entire vaccine deployment pipeline [1] [2] [3] [4] . This pipeline has the following main parts: 1) pre-clinical development and testing, 2) clinical development and testing, 3) production process development, scale-up and technology transfer for the manufacturing of the vaccine active ingredient (drug substance, DS), 4) sourcing of raw materials and consumables for manufacturing both the drug substance and the final packaged vaccine product filled into glass vials or other containers (fill-to-finish processes, f2f), 5) drug substance production under current Good Manufacturing Practices (cGMP), 6) fill-to-finish processes under cGMP, 7) vaccine distribution and 8) vaccine administration to the population [1] [2] [3] [4] .

COVID-19 vaccination programmes around the world are currently limited by the number of vaccine doses that can be manufactured [5] . To pre-emptively address this challenge and reduce deployment timelines, setting up vaccine manufacturing for pandemic-response production started "at risk" before the safety and efficacy of the vaccines was confirmed in clinical trials [4, 6] . This has the following challenges: (a) uncertainty in the DS amount per dose and number of doses per person, which are determined during clinical trials, (b) production processes have to be developed, optimised and scaled up [1] . Additionally, new vaccine designs may be required to tackle new virus variants. Finally, manufacturing needs to be low-cost to enable mass immunization, including in lowand middle-income countries [1, 7] .

Herein, we review the manufacturing processes for adenovirus-vectored (AVV), messenger RNA (mRNA), and the newer self-amplifying RNA (saRNA) vaccines. These vaccines contain genetic instructions, in the form of DNA for the AVV vaccine and RNA in case of the mRNA and saRNA vaccines, based on which the cells of the human body produce the vaccine antigen, such as the spike protein of the SARS-CoV-2 virus [8 -13] . Because only the genetic instruction and not the antigen is produced, the vaccine production processes serve as platform technologies. A platform technology implies that once validated and established at production scale, the same production processes can produce a wide range of different vaccines and vaccine candidates against both known and currently unknown, future, pathogens. The AVV and mRNA vaccine platforms have matured in terms of technology development and have high technology readiness levels. On the other hand, the saRNA vaccine platform is currently in clinical development and has a low technology readiness level.

We then analyse key uncertainties andtheir impact on COVID-19 vaccine production, as well as quantifying the production process scales, timescale and manufacturing resources required for producing 1 billion COVID-19 vaccines per year. These estimates can serve as a basis for calculating the requirements to produce vaccines for the global demand. Following drug substance production, we further evaluate three fill-to-finish technologies with respect to their pandemic-response manufacturing performance: conventional fill-to-finish in 5-dose or 10-dose vials, blow-fill-seal in single-dose vials, and the new 200-dose bag Intact™ Modular Filler [1, [14] [15] [16] .

All the models included in this study are representative of industrially-relevant cGMP COVID-19 vaccine manufacturing processes [17, 18] . The technologies used for COVID-19 vaccine production and their productivity in terms of number of vaccine doses produced per unit time and unit scale of the process varies tremendously among the three technologies. In order to compare them on the same basis, the following key performance indicators (KPIs) have been used: (i) annual production amounts expressed in doses per year, and (ii) and cost per dose expressed in USD per dose. The values of these KPIs and the degree by which these are impacted by the uncertainties was also assessed using global sensitivity analysis [19] [20] [21] . The rate at which batches can be produced and the number of doses produced per batch was also evaluated, as completing production batches in a short space of time would yield a steadier vaccine supply which can be advantageous in an emergency response situation compared to a more fragmented supply characteristic to production batches that require a longer time to complete. This study can inform policy makers and vaccine manufacturers on how to improve manufacturing and asset utilisation against COVID-19 and its variants, but also against future outbreaks due to the disease-agnostic nature of these vaccine production platforms [1, 22] .

The AVV, mRNA and saRNA DS production processes (primary manufacturing) are described in the SI document. Fill-finish processes are described in Table S1 . Likely baseline scenarios are presented in Table S2 , and one-factor-at-a-time uncertainty analysis in Figure S1 shows the impact of scale, titre and DS amount per dose on annual production amounts and production costs.

In reality, these input parameters may vary simultaneously. Thus, we have carried out variancebased stochastic global sensitivity analysis [23] [24] [25] to evaluate how input uncertainty propagates to outputs and aportion it individual inputs and their interactions [23] [24] [25] , as illustrated in Figure 1 .

Graphical illustration of the computational framework for uncertainty quantification. The aim of this approach is to evaluate process performance under uncertainty and variability resulting from both the design and operation of the new vaccine production platform technologies. The uncertainty is propagated from the inputs via the model to the outputs. In addition, the sensitivity of the model output key performance indicators (KPIs) is attributed to the individual inputs to determine the degree to which individual inputs impact the output KPIs. Model inputs include the scale of the production process, batch failure rate, titre/yield in the production bioreactor, cost of labour, drug substance amount per dose and cost of quality control. Model outputs include capital investment cost requirements, operating costs, number of batches produced per year, amount of drug substance produced per batch, amount of drug substance produced per year, number of doses produced per batch, number of doses produced per year and the production cost per dose.

The uncertain input variables and corresponding ranges for the three platforms are listed below in Table 1 and with additional explanation on the factors influencing these ranges in Table S4 of the SI document. The probability of the uncertainty distribution was set to triangular when the probable value was considered to have a high probability, and set to uniform when the input values in the range were considered to have equal or similar probability of occurring. For each platform technology 10,000 production process models were simulated using quasi-randomly generated combinations of model input variables within the ranges shown below in Table 1 . Sensitivity and uncertainty analyses were not performed for the fill-to-finish process as these are well-established technologies (relevant values are shown in Table S1 ). * Uncertainty distribution was considered either triangular, with highest probability for the central value, or uniform, with equal probability for all the values in the range.

** These are the values which were used in the baseline scenarios. For parameters which have a triangular distribution, the central value also corresponds to the mode. For uniform distributions the central value has the same frequency/abundance as the other values in the range.

*** The changes in recovery rate or losses in the downstream purification were modelled by changing the product titre in the bioreactor, as the ranges of changes in product titre also account for variations in amount of drug substance produced per batch, which could be due to changes in downstream purification losses.

Each of the scatter plots in Figure 2 show the impact of one input parameter onto one output parameter. The global sensitivity analysis results for AVV show that the annual production amounts are mostly influenced by scale, followed by titre and then AVV amount per dose as shown by the height of the bars in Figure 2G . The AVV cost per dose is mostly impacted by uncertainty in scale, titre, and AVV amount per dose, however the contribution of these inputs is of comparable magnitude as shown in Figure 2H and Figure 2D -F. The annual production amounts for both mRNA and saRNA depend mostly on the RNA amount per dose followed by the production scale, as shown in Figure 2O&I -J and Figure 2W&Q -S for mRNA and saRNA, respectively. The cost per dose for both mRNA and saRNA is predominantly influenced by the RNA amount per dose and to a relatively lesser extent by the production titres and the price of the CleanCap 5' cap analogue, as shown in Figure  2P&L -N and Figure 2X&T -V for mRNA and saRNA, respectively. The differences between the magnitude of the St and Si indicate that all process models have predominantly separable or additive characteristics with regards to the impact of the inputs on the outputs. The input parameters which have the highest impact on the outputs align with the width of the ranges of the input parameters. The width of these ranges is associated with the specifics of the platform technologies and the corresponding technology readiness levels. is the author/funder, who has granted medRxiv a license to display the preprint in (which was not certified by peer review) preprint

The copyright holder for this this version posted September 2, 2021. ; https://doi.org/10.1101/2021.08.20.21262370 doi: medRxiv preprint Figure 2 . The impact of co-variation of the input parameters on the annual production amounts and cost per dose of AVV, mRNA and saRNA drug substance production captured by stochastic global sensitivity analysis using quasi-random sampling based on Sobol sequences [23] [24] [25] . A-H (top two rows), I-P (middle two rows), Q-X (bottom two row) shows AVV, mRNA and saRNA vaccine drug substance production performance, respectively. A-F. Scatter plots showing the random co-variation of AVV drug substance annual production amounts in function of production scale, titre and AVV amount per dose. G-H. The main-effect (aka. 1st-order effect) Sobol indexes (Si) and total-order effect Sobol indexes (St) plotted in function of the seven input parameters shown on the x-axis for AVV drug substance production global sensitivity analysis. I-N. Scatter plots showing the random covariation of mRNA drug substance annual production amounts in function of production scale, titre and mRNA amount per dose. O-P. The Si and St plotted in function of the seven input parameters shown on the x-axis for mRNA drug substance production global sensitivity analysis. Q-V. Scatter plots showing the random co-variation of saRNA drug substance annual production amounts in function of production scale, titre and saRNA amount per dose. W-X. Si and St plotted in function of the seven input parameters shown on the x-axis for saRNA drug substance production global sensitivity analysis. Dots clustered around a narrower region on the Y-axis indicate that the respective input parameter explains most of the variance of the output KPI. On the contrary, dots spread out over a wider region on the Y-axis indicate that the respective input parameter explains little or no variance of the output KPI. Large Si and St values account for a strong impact of respective inputs shown on the X-axis on the output KPIs on the Y-axis, and low Si and St values indicate a weaker dependence of the KPIs on the respective inputs. is the author/funder, who has granted medRxiv a license to display the preprint in (which was not certified by peer review) preprint

The copyright holder for this this version posted September 2, 2021. ; https://doi.org/10.1101/2021.08.20.21262370 doi: medRxiv preprint

The productivity of the three platform technologies was compared using variance-based global sensitivity analysis presented above and based on the model input parameter ranges described in Table 1 . For this comparison, only DS production was modelled and it was assumed that production takes place in one facility and one production line per facility for each of the three vaccine platform technologies. In addition, production processes are assumed to be fully developed, validated and implemented at production scale. The required times (excluding quality control testing) and productivity together with their uncertainty distributions for producing 1 billion doses of DS are shown in Figure 3 . For this, inside the violin plots, box-and-whisker plots also show the minimum (0th percentile or Q0) and maximum (100th percentile or Q4) values using the extremities of the whiskers and the box plots show the interquartile ranges delimited by the 25th percentile (first quartile or Q1) and the 75th percentile (third quartile or Q3). The median is shown by a white dot inside the box plot, and the outliers are outside the whiskers, thus beyond the minimum and maximum values. The width of violin plots represents the probability distributions.

As shown in Figure 3A , the mRNA platform is likely to require the longest time to produce a unit of 1 billion vaccine doses. More specifically, an mRNA facility with a single production line would require a 2.19 −0.9 +1.91 years to produce 1 billion vaccine DS doses. The 2.19 −0.9 +1.91 years representation indicates a median of 2.19, the +1.91 superscript is the difference between the 75 th percentile and the 50 th percentile (which is the mean) and the -0.9 subscript is the difference between the 25 th percentile and the 50 th percentile. The saRNA platform is more likely to be faster, being capable of producing 1 billion DS doses in 0.083 −0.041 +0.084 years in a facility with a single production line. However, depending on the uncertainty realisation the saRNA platform may be surpassed by the AVV platform which can produce 1 billion doses in 0.96 −0.39 +0.76 years in a facility with a single production line.

The AVV platform relies on cell-based production which introduces more biological variability which in turn might lead to higher failure rates compared to the mRNA and saRNA platforms, which might further reduce the productivity of the AVV platform. The productivity of the three vaccine platforms, expressed in million doses produced per year per unit scale of the production process (represented by the bioreactor working volume), is shown in Figure 3B . Therein, the violin plots show the productivity ranges on the horizontal axis and the vertical width of the violin plots shows the probability density distribution. The mRNA vaccine production process is at 26.4 −5.6 +10.2 million doses per L of bioreactor working volume per year. As above, the median is shown by the base value, 26.4, the superscript shows the top quartile (75 th percentile minus median) and the subscript shows the bottom quartile (25 th percentile minus median) values. The productivity of the mRNA platform is two orders of magnitude higher than that of the AVV platform which is at 158 −43 +55 thousands. The saRNA platform at 731 −275 +601 million doses per L of bioreactor working volume per year is one to two orders of magnitude more productive than the mRNA platform and four orders of magnitude more productive than the AVV platform. The productivity difference between the RNA and AVV platforms is due to the highly concentrated, cell-free enzymatic reaction mix based production of the mRNA and saRNA vaccines. The productivity difference between mRNA and saRNA vaccines is due to the substantially lower amount of RNA per dose of the saRNA vaccines.

At the beginning of the production campaign, the start of the fill-to-finish is delayed by the time required to produce and quality test the first DS batch. During pandemic-response manufacturing, the DS can be produced and stockpiled in parallel to carrying out the clinical trials [45] , thus the start of fill-to-finish will be even more delayed relative to the start of DS production. The DS production bottlenecks depend on the specific vaccine platform technology. In case of AVV-based vaccines, the bottleneck is caused by the time needed to culture mammalian cells to reach the required amounts in the production bioreactor. In case of mRNA and saRNA vaccine DS production, the bottleneck is in is the author/funder, who has granted medRxiv a license to display the preprint in (which was not certified by peer review) preprint

The copyright holder for this this version posted September 2, 2021. ; https://doi.org/10.1101/2021.08.20.21262370 doi: medRxiv preprint the microfluidics LNP formulation unit operation. The LNP formulation bottleneck can be removed by increasing the size (scaling up) or the number of parallel equipment (scaling out) for the formulation unit operation. However, mRNA vaccine production is most effectively enhanced by reducing the mRNA amount per vaccine dose. The relationship between annual production amount and amount per dose is multiplicative inverse, as shown in Figure S1H and 2K.

Besides bottlenecks in the actual production processes, additional waiting times can be expected for the completion of certain QC tests especially in case of new platform technologies such as the RNA platform. A potential solution for this would be the use of a Quality by Design (QbD) framework to streamline quality assurance by building quality assurance into the design and operation of the production process, which is currently limited by suitable process analytical technology (PAT) [46] . is the author/funder, who has granted medRxiv a license to display the preprint in (which was not certified by peer review) preprint

The copyright holder for this this version posted September 2, 2021. Figure S3 of the SI document.

Additionally, the following three different fill-to-finish technologies (secondary manufacturing) processes were also assessed: conventional filling into 10-dose or 5-dose vials, blow-fill-seal into single-dose vials, and the new 200-dose bag Intact Modular Filler system [1, [14] [15] [16] . For details see SI document and Table S1 . The overall production bottleneck, when considering both DS production and fill-to-finish, depends on the combination of the specific technologies. For example, in case of filling AVV vaccines in 10-dose vials, the bottleneck will be in the DS production when one AVV DS production line with baseline characteristics (cf . Table S2 ) at the common 2000 L bioreactor working volume scale is coupled to a single 10-dose vial filling line which fills at 400 doses per minute at 60% overall equipment effectiveness (OEE). However, depending on the uncertainty realization (e.g. when larger DS production scales are also considered), the DS production rate might surpass the fillto-finish rate, as shown by the global sensitivity analysis results presented in Figure 3A compared to the values shown in Table S1 . The baseline mRNA vaccine production rate (cf . Table S2 ) is also slower then filling into 5-dose vials. If an mRNA vaccine production line is coupled with a 10-dose vial (Moderna vaccine) filling line which fills at 400 vials per minute, the overall production bottleneck would again be at the DS production stage. On the other hand, in case of saRNA DS vaccine production coupled with fill-to-finish into 5-dose vials at the 400 vials/minute rate, the overall production bottleneck is at the fill-to-finish stage.

Considering the above-presented uncertainties, violin plots have been generated to estimate the resource and production capacity requirements for producing a unit of 1 billion doses of vaccine drug substance per year, cf. Figure 4 . These CapEx, OpEx, production scales and number of batch values as well as their uncertainty distributions were obtained from the variance-based global sensitivity analysis presented in section 2.1 using linear scaling and the model input parameter ranges described in Table 1 . Linear scaling was used because these production processes and raw material supplies were modelled at large scale, and this would scale approximately linearly when scaling out for meeting the global pandemic demand. The computed CapEx and OpEx requirements for producing 1 billion doses of vaccine drug substance and drug product per year are shown in Figure 4A and Figure 4B , respectively. For this, it was assumed that AVV vaccines are filled into 10dose vials, and that both mRNA and saRNA vaccines are filled into 5-dose vials, based on the Oxford/AstraZeneca AVV and the Pfizer/BioNTech mRNA vaccine, respectively. The production scales is the author/funder, who has granted medRxiv a license to display the preprint in (which was not certified by peer review) preprint

The copyright holder for this this version posted September 2, 2021. ; https://doi.org/10.1101/2021.08.20.21262370 doi: medRxiv preprint required to produce a unit of 1 billion doses of AVV, mRNA and saRNA vaccine drug substance per year are shown in Figure 4C . The number of AVV, mRNA and saRNA production batches required to produce 1 billion doses of vaccine drug substance per year are shown in Figure 4D .

Out of the three vaccine drug substance production platform technologies, the mammalian cellbased AVV platform is predicted to have the highest CapEx with a median of ≈340 million USD, with a top quartile (75 th percentile of ≈410 million USD minus the median) of +70 million USD and with a lower quartile (25 th [7, 47] . The CapEx, OpEx, production scale and batch number ranges and values were presented in Figure 4 on a 1 billion dose annual throughput basis, however, by linear extrapolation the CapEx, OpEx, production scale and batch number ranges and values can be approximated for producing vaccines for the global population. Therefore, the CapEx, OpEx, production scale and batch number values for meeting a c. 15 billion dose annual production target will be 15-fold higher than the values presented in Figure 4 .

The fill-to-finish costs is additive to the DS production costs. By filling the AVV vaccine in 10-dose vials and by filling mRNA and saRNA vaccines in 5-dose vials, the production cost per dose, including both DS production and f2f, is 0.54, 2.39 and 0.39 USD per dose for AVV, mRNA and saRNA, respectively, cf. Table S1 and S2. For saRNA f2f cost per dose is the major cost contributor, for AVV the DS production and fill-to-finish costs are similar, whereas for mRNA the DS cost per dose accounts for most of the total production costs. The DS amount per vaccine dose could in principle decrease, not only for mRNA vaccines, but for all three platform technologies when developing second and third generation vaccines. is the author/funder, who has granted medRxiv a license to display the preprint in (which was not certified by peer review) preprint Operating costs (OpEx) required to produce 1 billion doses per year of AVV, mRNA and saRNA vaccine drug substance (DS) and drug product. It was assumed that AVV vaccine is filled into 10-dose vials, whereas the mRNA and saRNA vaccine is filled into 5-dose vials. B. Capital costs (CapEx) required to produce 1 billion doses per year of the vaccine drug substance and drug product using the three platform technologies. AVV vaccine fill-to-finish was modelled based on 10dose vials, whereas the mRNA and saRNA vaccine fill-to-finish was modelled based on 5-dose vials. C. Production process scales required to produce 1 billion doses of DS per year using the AVV, mRNA and saRNA vaccine production platforms. The scale of the production process is represented by the working volume in the bioreactor and the entire process is scaled based on the mass balances proportionally to the bioreactor working volume. D. Number of batches required to produce 1 billion doses of DS per year using the AVV, mRNA and saRNA vaccine production platforms. The bottom 5% and top 5% of all values were excluded from all violin plots in order to obtain a better visualization of the region of interest around the box plot. The equivalent violin plots showing all the data (including the top 5% and bottom 5%) are presented in Figure S4 of the SI document, which also shows the CapEx and OpEx for drug product manufacturing by fill-to-finish.

The global demand for COVID-19 vaccines is estimated at 15.6 billion doses, assuming a 2-dose regimen for the global population. The most optimistic estimate for current annual production . CC-BY-ND 4.0 International license It is made available under a perpetuity.

is the author/funder, who has granted medRxiv a license to display the preprint in (which was not certified by peer review) preprint

The copyright holder for this this version posted September 2, 2021. ; https://doi.org/10.1101/2021.08.20.21262370 doi: medRxiv preprint capacity is 10 billion doses [48] . Therefore, there is a shortfall of at least 5.6 billion doses, without accounting for potential immunization demand caused by vaccine-escaping variants [49] . The resource requirements for meeting this shortfall computed based on the global sensitivity analysis results are illustrated in Figure S5 . Therein, the violin plots show the ranges and probability distributions of the CapEx, OpEx, production scales and number of batches required to produce the 5.6 billion doses shortfall within a year using the three vaccine production platform technologies. For the risk analysis, worst-case scenarios can be defined at the maximum resource requirement values, illustrated by the maximum (100th percentile or Q4) top whisker. As shown by the probability distribution, there is a very small chance for this worse-case scenario to materialize based on this analysis. However, even in this worst-case scenario the benefits of establishing new production capacity based on all three platform technologies outweigh the costs by several orders of magnitude when considering the mortality, healthcare burden of the COVID-19 pandemic and economic decline. It is estimated that the pandemic has cost the global economy over 10 trillion USD [50] , and the UN projects that the COVID-19 pandemic will reduce the global economy by a further 8.5 trillion USD over a 2-year period [51] . These substantial detrimental impacts of the COVID-19 pandemic can be avoided by comparatively small stimuli in the form of capital and operating costs, ranging between several hundred million to a few billion USD, as shown in Figure S5A and S5B, respectively. Based on these expenses the total drug substance production capacity shown in Figure S5C can be built to produce the number of batches ( Figure S5D ) required to meet the current shortfall. However, it is worth noting that such investments have to be made ideally in advance, or as soon as possible, considering the years-long timescale required to build such manufacturing capacity [1] . If this is built based on platform technologies, such as the RNA and AVV platforms, the resulting facilities could be used for producing a wide range of vaccines over their lifetime. Besides the financial resources, key raw materials (e.g. 5' cap analogues or capping enzymes, cationic lipids and pseudouridine triphosphate), expertise and facilities for mass-producing mRNA vaccines and consumables for producing all vaccines are also in limited supply. An analysis of the material, consumables, labour and facility requirements for mass-producing mRNA vaccines for pandemic response has been carried out [47] .

A large share of the COVID-19 vaccine shortfall described above is likely to be met by adapting or repurposing manufacturing facilities that were used to manufacture other vaccines and biopharmaceuticals pre-pandemic. However, the healthcare impact of not sustaining routine childhood immunisations can outweigh that of the COVID-19 pandemic, especially in Africa [52, 53] . Thus, it is crucial to manufacture and supply lifesaving vaccines against all vaccine-preventable diseases and to minimize the disruption in manufacturing of non-COVID-19 vaccines caused by the manufacturing of COVID-19 vaccines. This is even more important in the likely scenario of needing periodic, e.g. annual, COVID-19 booster doses to immunize adults at risk of severe COVID-19 and frontline workers in the foreseeable future [54] . If such one-dose booster vaccinations are to be required every year, building dedicated Covid-19 vaccine production facilities is a viable option. These facilities can be designed around platform technologies, enabling the production of vaccines or candidates against future SARS-CoV-2 variants, new coronavirus strains, or other emerging viruses [1, 22] . The global population vulnerable to COVID-19, including people aged over 60 and adults with underlying medical conditions is around 2.2 billion [48] . In this scenario, ≈0.2 billion frontline personnel might also be required to be immunized to prevent the spread of the disease [48] . Therefore, in the single-dose boost scenario ≈2.4 billion COVID-19 vaccine doses would be required annually. The resource requirements for producing the annual boost vaccinations are illustrated in Figure S6 . Therein, the violin plots show the ranges and probability distributions of the CapEx, OpEx, is the author/funder, who has granted medRxiv a license to display the preprint in (which was not certified by peer review) preprint

The copyright holder for this this version posted September 2, 2021. ; https://doi.org/10.1101/2021.08.20.21262370 doi: medRxiv preprint production scales and number of batches required to produce the 2.4 billion boost doses per year using the three vaccine production platform technologies.

Additionally, by investing in dedicated COVID-19 vaccine production facilities for supplying the annual COVID-19 boost vaccination the severe healthcare impact of other vaccine-preventable diseases can also be minimized by reinstating the production scale of other vaccines and biopharmaceuticals. Since vaccinations started in late 2020 or early 2021, administration of annual booster shots would potentially begin by the end of 2021. Using these platform, booster COVID-19 vaccines can be produced relatively quickly, as can production of new vaccines against emerging variants (which would still require clinical trials). The rate of mass-producing the 2.4 billion booster vaccinations will depend mostly on the platform technologies used, available production capacity and amount of drug substance per dose, which can be antigen specific. A feasible option would be to combine the annual COVID-19 booster dose with the annual influenza vaccine into a multivalent vaccine. For this, both vaccines can be produced using the same platform technologies. In this case, in contrast to current approaches, the influenza vaccine could be produced on demand for the strain in circulation, without the need of forecasting the 3-4 most prevalent influenza strains more than 6 months ahead of the start of the vaccination programme [55, 56] . However, manufacturing the DS for multivalent vaccines comes with similar complexities and costs as a manufacturing of several monovalent vaccines. Since currently the manufacturing of vaccines represents the bottleneck for global immunization programs the need to develop and manufacture vaccines against a new variant would further limit immunization rates because more vaccine DS doses are required to immunize a person. Importantly, these platform technologies can speed up both the development of candidate vaccines and the manufacturing of regulatory-approved vaccines against a wide range of viral diseases, including currently known and currently unknown, future diseases.

The modeling of AVV, mRNA and saRNA drug substance production as well as drug product fill-tofinish was carried out using SuperPro Designer (Version 11, Build 2) by Intelligen, Inc. Further details are available in the SI document.

Information regarding mRNA and saRNA vaccine production processes and costs was obtained from the scientific literature [38, [57] [58] [59] [60] [61] [62] [63] , patent databases [26] [27] [28] [29] 64] , from GMP grade material suppliers [31, 65, 66] and from experts [31, 67, 68] . Information regarding mRNA drug substance amount per dose was based on clinical trial databases [41, 42, 69, 70] and the scientific literature [71] . For saRNA vaccines the drug substance amount per dose was obtained from the clinical trial registry [72] . Information on AVV vaccine production was obtained from the scientific literature [73, 74] and consultation with experts [75] . The AVV vaccine production process was modelled based on the manufacturing of the replication-deficient chimpanzee adenovirus-vectored (ChAdOx1) vaccine which was co-developed by Oxford University and AstraZeneca plc. Information on AVV drug substance amount per dose was based on clinical trial databases [39, 40, 43, 44, [76] [77] [78] . Information on fill-to-finish technologies was obtained from the literature [15, [79] [80] [81] , equipment suppliers [14, 16, 82, 83] and industry experts [84, 85] . Additional production process data for all drug substance and drug product manufacturing processes were retreived from the equipment, materials, utilities and cost databases in SuperPro Designer [86] . The annualized CapEx is included in the OpEx. The is the author/funder, who has granted medRxiv a license to display the preprint in (which was not certified by peer review) preprint

The copyright holder for this this version posted September 2, 2021. ; https://doi.org/10.1101/2021.08.20.21262370 doi: medRxiv preprint CapEx value is also presented individually, in order to illustrate the financial requirements for building new facilities.

Variance-based stochastic global sensitivity analyses were conducted using SobolGSA Version 3.1.1 [87] under MatLab R2020a. 10,000 quasi-random scenarios were generated using Sobol sequences [23] [24] [25] 88] according to the process parameter ranges and distributions in Table 1 then passed to SuperPro Designer for evaluating the techno-economic KPIs in each scenario. A metamodel was generated in SobolGSA using the random-sampling high dimensional model representation (RS-HDMR) [89, 90] based on which the main-effect and total-effect Sobol indices were derived [20] . A further 1,250 uncertainty scenarios were simulated in SuperPro Designer to test the predictions of the metamodel. The link between SobolGSA and SuperPro Designer was enabled by a Component Object Model (COM) interface in MS-Excel VBA available from from MS-Office 365 Enterprise. Further details are available in the SI document. Data processing and visualization/plotting is also described in the SI document.

Data is available from: https://github.com/ZKis-ZK/RNA_AVV_vaccine_production-cost_modelling_global_sensitivity_analysis

In this computational modelling study, the COVID-19 pandemic-response manufacturing performance of the AVV, mRNA and saRNA vaccine platforms has been assessed using technoeconomic modelling and variance-based global sensitivity analysis. The three vaccine production processes have been presented and key design and operation uncertainties and variations were assessed in terms of their impact on the productivity and resource-intensity performance indicators of these processes. It was established that variations in both the annual productivity and cost per dose of AVV vaccines can be primarily attributed to variations in the scale and titre/yield of the production process. On the other hand, the variations in the annual productivity and cost per dose of mRNA and saRNA vaccines can be primarily attributed to variations in the RNA amount per dose. The saRNA platform is likely to be the fastest to meet the global demand, followed by the AVV and then by the mRNA platform. The results of these assessments depend on the specific characteristics, such as the production scale, at which these platform technologies are implemented, and these characteristics are fundamentally different among these platform technologies alongside the drug substance amount per dose and the time required to produce batches. The performance of the AVV platform can be improved by increasing the yield in the production bioreactor. Decreasing the RNA amount per dose, would improve the production rates and volumes of mRNA and saRNA vaccines and by increasing their (thermo)stability would improve their usability across the globe, including in low-and middle-income countries. To meet the current global projected vaccine shortfall of 5.6 billion doses and annual booster vaccination production, investments ranging from hundreds of millions to a few billion USD would be required, which is very small in comparison to the mortality, healthcare and economic cost of the COVID-19 pandemic, estimated at over 10 trillion USD. Overall, this model-based assessment can inform policymakers and vaccine manufacturers for risk assessment, on how to improve manufacturing and asset utilisation against COVID-19 and its variants, but also against future outbreaks due to the disease-agnostic nature of these vaccine production platforms. The platform technology-based dedicated COVID-19 vaccine production . CC-BY-ND 4.0 International license It is made available under a perpetuity.

is the author/funder, who has granted medRxiv a license to display the preprint in (which was not certified by peer review) preprint

The copyright holder for this this version posted September 2, 2021. ; https://doi.org/10.1101/2021.08.20.21262370 doi: medRxiv preprint would prevent the reduction of the manufacturing throughput of other, non-COVID-19 vaccine and therapeutics, and would allow rapid-response vaccine production against a wide a range of viral targets. These platform technologies will enable faster vaccine development and production for overcoming future epidemics and pandemics, especially if surge manufacturing capacity is maintained during times without outbreaks.

Rapid development and deployment of highvolume vaccines for pandemic response

Williams 3rd RO. The COVID-19 Vaccine Race: Challenges and Opportunities in Vaccine Formulation

COVID-19 vaccines: challenges and promises of trials, manufacturing and allocation of doses. Futur Drug Discov

COVID-19 vaccine development during pandemic: gap analysis, opportunities, and impact on future emerging infectious disease development strategies

Looking beyond COVID-19 vaccine phase 3 trials

SARS-CoV-2 vaccines in development

Resources, Production Scales and Time Required for Producing RNA Vaccines for the Global Pandemic Demand

An mRNA Vaccine against SARS-CoV-2 -Preliminary Report

Current Status of COVID-19 (Pre)Clinical Vaccine Development. Angew Chemie Int Ed

Safety and Immunogenicity of Two RNA-Based Covid-19 Vaccine Candidates

COVID-19 vaccine BNT162b1 elicits human antibody and TH1 T cell responses

Self-amplifying RNA SARS-CoV-2 lipid nanoparticle vaccine candidate induces high neutralizing antibody titers in mice

Safety and immunogenicity of the ChAdOx1 nCoV-19 vaccine against SARS-CoV-2: a preliminary report of a phase 1/2, single-blind, randomised controlled trial

Cost of goods sold and total cost of delivery for oral and parenteral vaccine packaging formats

Blow-Fill-Seal Solutions

Rommelag Kunststoff-Maschinen Vertriebsgesellschaft mbH

World Health Organization. Annex 2 -WHO good manufacturing practices for pharmaceutical

Sobol' sensitivity analysis of a complex environmental model. Environ Model Softw

Variance based sensitivity analysis of model output. Design and estimator for the total sensitivity index

The identification of model effective dimensions using global sensitivity analysis. Reliab Eng Syst Saf

Emerging Technologies for Low-Cost, Rapid Vaccine Manufacture

On sensitivity estimation for nonlinear mathematical models

Sensitivity Estimates for Nonlinear Mathematical Models. Mathematical Modeling and Computational experiment

Construction and Comparison of High-Dimensional Sobol' Generators. Wilmott

Manufacturing methods for production of RNA transcripts

Methods and means for enhancing RNA production

USA: United States Patent and Trademark Office; US20170114378A1

Methods for purification of messenger rna

A method for producing and purifying rna, comprising at least one step of tangential flow filtration

SuperPro Designer User Guide -A Comprehensive Simulation Tool for the Design

Telephone conversation with representatives from TriLink

) 33. National Institute of Allergy and Infectious Diseases (NIAID). Safety and Immunogenicity Study of 2019-nCoV Vaccine (mRNA-1273) for Prophylaxis SARS CoV-2 Infection (COVID-19

Medicines & Healthcare products Regulatory Agency. Information for UK recipients on COVID 19 Vaccine AstraZeneca

UK Government Digital Service

Medicines & Healthcare products Regulatory Agency. Package leaflet: Information for the recipient COVID-19 mRNA Vaccine BNT162b2 concentrate for solution for injection

UK Government Digital Service

Fact sheet for healthcare providers administering -Vaccine (vaccination providers) -Emergency Use Authorization (EUA) of the Moderna Covid-19 Vaccine To Prevent Coronavirus Disease

Self-Amplifying RNA Vaccines Give Equivalent Protection against Influenza to mRNA Vaccines but at Much Lower Doses

Nonviral delivery of self-amplifying RNA vaccines

Investigating a Vaccine Against COVID-19

A Study of a Candidate COVID-19 Vaccine (COV001)

A Phase 3, Randomized, Stratified, Observer-Blind, Placebo-Controlled Study to Evaluate the Efficacy, Safety, and Immunogenicity of mRNA-1273 SARS-CoV-2 Vaccine in Adults Aged 18 Years and Older

A Phase 2a, Randomized, Observer-Blind, Placebo Controlled, Dose-Confirmation Study to Evaluate the Safety, Reactogenicity, and Immunogenicity of mRNA-1273 SARS-COV-2 Vaccine in Adults Aged 18 Years and Older

An Adaptive Phase I/II Randomized Placebo-controlled Trial to Determine Safety, Immunogenicity and Efficacy of Non-replicating ChAdOx1 SARS-CoV-2 Vaccine in South African Adults Living Without HIV; and Safety and Immunogenicity in Adults Living With HIV

Controlled, Phase III Study to Determine the Safety, Efficacy, and Immunogenicity of the Non-Replicating ChAdOx1 nCoV-19 Vaccine

Challenges in ensuring global access to COVID-19 vaccines: production, affordability, allocation, and deployment

How to Make Enough Vaccine for the World in One Year

Global, regional, and national estimates of target population sizes for covid-19 vaccination: descriptive study

SARS-CoV-2 variants and ending the COVID-19 pandemic

Unprecedented: the toxic synergism of Covid-19 and climate change

United Nations, Department of Economic and Social Affairs. COVID-19 to slash global economic output by $8.5 trillion over next two years

Vaccines and routine immunization strategies during the COVID-19 pandemic

Routine childhood immunisation during the COVID-19 pandemic in Africa: a benefit-risk analysis of health benefits versus excess risk of SARS-CoV-2 infection

Covid-19: Booster dose will be needed in autumn to avoid winter surge, says government adviser

Considerations of strategies to provide influenza vaccine year round

Models for predicting the evolution of influenza to inform vaccine strain selection

Developing mRNA-vaccine technologies

Self-replicating alphavirus RNA vaccines. Expert Rev Vaccines

Self-Amplifying RNA Vaccines Give Equivalent Protection against Influenza to mRNA Vaccines but at Much Lower Doses

mRNA vaccines -a new era in vaccinology

Chapter Seven -Self-Amplifying mRNA Vaccines

Optimization of Lipid Nanoparticles for Intramuscular Administration of mRNA Vaccines

RNA purification methods

CustomBiotech Catalog 15th Edition BioPharma

Quality and security that empower breakthroughs -Raw materials for mRNA therapeutics

Telephone and email correspondence with representatives from Roche Diagnostics GmbH

Telephone and email correspondence with biopharmaceutical manufacturing experts from the Centre for Process Innovation Limited

COVID-19: A Phase 2a, Partially Observer-blind, Multicenter, Controlled, Doseconfirmation Clinical Trial to Evaluate the Safety, Reactogenicity and Immunogenicity of the Investigational SARS-CoV-2 mRNA Vaccine CVnCoV in Adults >60 Years of Age and 18 to

A Phase 1, Partially Blind, Placebo-controlled, Dose-escalation, First-in-human, Clinical Trial to Evaluate the Safety, Reactogenicity and Immunogenicity After 1 and 2 Doses of the Investigational SARS-CoV-2 mRNA Vaccine CVnCoV Administered Intramuscularl

Phase I/II study of COVID-19 RNA vaccine BNT162b1 in adults

Clinical trial to assess the safety of a coronavirus vaccine in healthy men and women

Simian adenovirus vector production for early-phase clinical trials: A simple method applicable to multiple serotypes and using entirely disposable product-contact components

Production and Purification of Adenovirus Vectors for Gene Therapy

Manufacturing a chimpanzee adenovirus-vectored SARS-CoV-2 vaccine to meet global needs

Double-blind, Placebo-controlled Multicentre Study in Participants Aged 18 Years or Older to Determine the Safety and Immunogenicity of AZD1222, a Non-replicating ChAdOx1 Vector Vaccine, for the Prevention of COVID-19

A Phase III Open-label Study in Adults to Determine the Safety and Immunogenicity of AZD1222, a Non-replicating ChAdOx1 Vector Vaccine, for the Prevention of COVID-19

Double-blind, Placebo-controlled Multicenter Study in Adults to Determine the Safety, Efficacy, and Immunogenicity of AZD1222, a Non-replicating ChAdOx1 Vector Vaccine, for the Prevention of COVID-19

Advantages of single-use technology for vaccine fill-finish operations

Bosch introduces new vial filling and closing machine

Bosch Introduces Vial Filler -MLF 5088 offers an output of 400 vials per minute with IPC

Contract Pharma Magazine

Rommelag bottelpack bp460 -Aseptic filling in ampoules

Aseptic blow-fill-seal fill/finish technology and vaccines

Brazil: DCVMN

Email and teleconference correspondence with experts from MEDInstill

Email and telephone correspondence with Tim Kram from Rommelag USA

SuperPro Designer User Guide -A Comprehensive Simulation Tool for the Design

SobolGSA Software

Algorithm 659: Implementing Sobol's Quasirandom Sequence Generator

SobolHDMR: a general-purpose modeling software