Volume 121, Issue 9 p. 2604-2635
REVIEW
Open Access

Vaccine process technology—A decade of progress

Barry Buckland

Corresponding Author

Barry Buckland

National Institute for Innovation in Manufacturing Biopharmaceuticals, University of Delaware, Newark, Delaware, USA

Correspondence Barry Buckland, National Institute for Innovation in Manufacturing Biopharmaceuticals (NIIMBL), 590 Avenue 1743, Newark, DE 19713, USA.

Email: [email protected]

Search for more papers by this author
Gautam Sanyal

Gautam Sanyal

Vaccine Analytics, LLC, Kendall Park, New Jersey, USA

Search for more papers by this author
Todd Ranheim

Todd Ranheim

Advanced Analytics Core, Resilience, Chapel Hill, North Carolina, USA

Search for more papers by this author
David Pollard

David Pollard

Sartorius, Corporate Research, Marlborough, Massachusetts, USA

Search for more papers by this author
Jim A. Searles

Jim A. Searles

Independent Consultant, St. Louis, Missouri, USA

Search for more papers by this author
Sue Behrens

Sue Behrens

Engineering and Biopharmaceutical Processing, Keck Graduate Institute, Claremont, California, USA

Search for more papers by this author
Stefanie Pluschkell

Stefanie Pluschkell

National Institute for Innovation in Manufacturing Biopharmaceuticals, University of Delaware, Newark, Delaware, USA

Search for more papers by this author
Jessica Josefsberg

Jessica Josefsberg

Merck & Co., Inc., Process Research & Development, Rahway, New Jersey, USA

Search for more papers by this author
Christopher J. Roberts

Christopher J. Roberts

National Institute for Innovation in Manufacturing Biopharmaceuticals, University of Delaware, Newark, Delaware, USA

Search for more papers by this author
First published: 06 May 2024
Citations: 3

Permissions have been obtained to reproduce material from other sources, with appropriate attribution provided in the figure legends.

Abstract

In the past decade, new approaches to the discovery and development of vaccines have transformed the field. Advances during the COVID-19 pandemic allowed the production of billions of vaccine doses per year using novel platforms such as messenger RNA and viral vectors. Improvements in the analytical toolbox, equipment, and bioprocess technology have made it possible to achieve both unprecedented speed in vaccine development and scale of vaccine manufacturing. Macromolecular structure-function characterization technologies, combined with improved modeling and data analysis, enable quantitative evaluation of vaccine formulations at single-particle resolution and guided design of vaccine drug substances and drug products. These advances play a major role in precise assessment of critical quality attributes of vaccines delivered by newer platforms. Innovations in label-free and immunoassay technologies aid in the characterization of antigenic sites and the development of robust in vitro potency assays. These methods, along with molecular techniques such as next-generation sequencing, will accelerate characterization and release of vaccines delivered by all platforms. Process analytical technologies for real-time monitoring and optimization of process steps enable the implementation of quality-by-design principles and faster release of vaccine products. In the next decade, the field of vaccine discovery and development will continue to advance, bringing together new technologies, methods, and platforms to improve human health.

1 INTRODUCTION

In 2012, the review “Vaccine Process Technology” was published in Biotechnology and Bioengineering (Josefsberg & Buckland, 2012) and focused on the various technologies used to manufacture vaccines for the prevention of a variety of infectious diseases in humans. In the past decade, the field has advanced dramatically, in part due to the massive increase in resources devoted to developing a safe and effective COVID-19 vaccine. A surprising result of this intense effort was the introduction and prioritization of new vaccine platforms, specifically messenger RNA (mRNA) and replication-deficient adenovirus, over more traditional approaches, such as using protein antigens as vaccines.

Over the past 10 years, there have been tremendous advances in DNA- and RNA-containing vaccines, which do not themselves contain the antigens of interest, but instead carry instructions for the recipient's own cells to produce the antigens. Viral-vectored and mRNA lipid nanoparticle (LNP) vaccines fall within this category. Ten new viral-vectored vaccines and two mRNA-LNP vaccines have been approved since 2012 (Table 1). Many other important vaccines have also been approved since the 2012 review. Flucelvax®, the first seasonal influenza vaccine manufactured using cell culture technology, was introduced to the United States in 2012 (Novartis, now Seqiris). The Flucelvax process uses suspension MDCK cell culture as the substrate for virus propagation instead of chicken eggs. The vaccine was first approved in Europe in 2007 as OptaFlu. A year later, the U.S. Food and Drug Administration (FDA) approved Flublok®, an insect culture recombinant flu vaccine produced using a baculovirus expression system (Protein Sciences Corporation, acquired by Sanofi in 2017). The vaccine was changed from trivalent to quadrivalent in 2016 and was approved by the European Medicines Agency (EMA) in 2020. Flublok® utilizes recombinant expression of hemagglutinin (HA) proteins exactly matching the selected strains, avoiding potential problems of virus adaptation to egg or cell culture causing antigenic drift (Buckland, 2015).

Table 1. Viral vectored vaccines (McCann et al., 2022).
First approval Vaccine Target pathogen Encoded antigen Developer Vector class Vector
2010 Australia ChimeriVax-JE (Imojev) Japanese encephalitis Viral envelope (prM and E) of JE strain SA14-14-2 Sanofi Pasteur Flaviviruses YF 17D
2015 Mexico CYD-TDV (Dengvaxia) Dengue prM and E genes of DENV 1–4 Sanofi Pasteur Flaviviruses YF 17D
2015 Russia GamEvac-Combi Ebola virus Both glycoproteins Gamaleya Research Institute (Russia) Heterologous regimens VSV/Ad5
2017 China Ad5-EBOV Ebola virus Zaire strain (Makona) of glycoprotein CanSino Biologics (China) Adenoviruses Ad5
2017 DRC VSV-EBOV (rVSV-ZEBOV, Ervebo) Ebola virus Zaire strain (Kikwit 1995) of glycoprotein Merck USA Rhabdoviruses VSV
2020 United Kingdom ChAdOx1- nCoV-19 (Covishield, Vaxzevria) SARS-CoV-2 Spike protein with tissue plasminogen leader sequence Univ Oxford/AstraZeneca Adenoviruses ChAdOx1
2020 Russia Gam-COVID-Vac (Sputnik V) SARS-CoV-2 Both spike proteins Gamaleya Research Institute (Russia) Adenoviruses Ad5/Ad26
2020 EMA

Ad26. ZEBOV (Zabdeno)

MVA-BN-Filo (Mvabea)

Ebola virus

Ad26—Zaire strain

MVA—glycoproteins from the Zaire Ebola virus, Sudan virus, and Marburg virus, and the nucleoprotein from the Tai Forest virus

Janssen Pharma Heterologous regimens Ad26/MVA
2021 FDA, EMA, WHO Ad26.CoV SARS-CoV-2 Prefusion-stabilized spike protein Janssen Pharma Adenoviruses Ad26
2021 China Ad5-nCoV (Convidecia) SARS-CoV-2 Spike protein CanSino Biologics (China) Adenoviruses Ad5
2022 Russia Sputnik light SARS-CoV-2 Spike protein Gamaleya Research Institute (Russia) Adenoviruses Ad26
  • Abbreviations: EMA, European Medicines Agency; FDA, Food and Drug Administration; MVA, modified vaccinia Ankara; WHO, World Health Organization.

In 2017, FDA licensed Shingrix, a herpes zoster (HZ, shingles) vaccine from GSK, for use in adults 50 years of age and older. The vaccine is a recombinant glycoprotein antigen produced in Chinese hamster ovary (CHO) cells and lyophilized, and a separate vial of AS01B adjuvant liquid suspension is used for reconstitution. The vaccine suspension contains 50 μg of recombinant varicella zoster virus (VZV) glycoprotein E with enhanced immunogenicity from two adjuvants: the liposome-based AS01B adjuvant system containing 50 μg of 3-O-desacyl-4′-monophosphoryl lipid A (MPL) and 50 μg of Quillaja saponaria Molina, fraction 21 (Shingrix [Package insert], 2023). Also in 2017, Dynavax received FDA approval of the Heplisav-B hepatitis B vaccine with novel cytosine phosphoguanosine (CpG) 1018® adjuvant and recombinant protein subunit antigen. The product was approved in Europe in 2019. CpG 1018® adjuvant is a modified-backbone oligonucleotide that mimics bacterial and viral DNA to stimulate the innate immune system (Shirota & Klinman, 2017). In 2019, the FDA approved JYNNEOS® (Bavarian Nordic A/S), a smallpox and monkeypox vaccine, for adults 18 years of age and older determined to be at high risk for infection. JYNNEOS® is a live vaccine produced from the MVA-BN strain, an attenuated, non-replicating orthopoxvirus. MVA-BN is grown in primary CEF cells suspended in a serum-free medium containing no material of direct animal origin, then harvested, purified, and concentrated by several tangential flow filtration (TFF) steps including benzonase digestion. To increase the vaccine supply, in August of 2022, FDA issued an Emergency Use Authorization (EUA) of JYNNEOS® for subcutaneous (s.c.) administration (0.5 mL) to individuals under 18 years of age and for intradermal (i.d.) administration (0.1 mL) to individuals 18 years of age and older.

In 2020, World Health Organization (WHO) issued an Emergency Use Listing (EUL) for nOPV2 as part of the evolving strategy for global polio eradication (World Health Organization, 2020b). The vaccine uses live attenuated poliomyelitis virus type 2 of a modified Sabin strain (nOPV2 strain S2/cre5/S15domV/rec1/hifi3) prepared in Vero cells derived from African green monkey kidney. To reduce the incidence of vaccine-derived polio, the vaccine has increased genetic stability and decreased recombination rate achieved with five modifications in the parental genome affecting domain V, the cre element, and RNA-dependent RNA polymerase (Polio Global Eradication Initiative, 2021). Also in 2020, WHO prequalified the first Sabin Inactivated Polio Vaccine (sIPV, injected, LG Chem) (World Health Organization, 2020a). The vaccine is produced using Vero cells on microcarriers, and production does not use wild-type virus, reducing the risk of reversion mutation (Suarez-Zuluaga et al., 2022). In addition, the cell growth medium was changed from serum-containing medium to animal component-free (ACF) medium, a diafiltration step was added before virus inactivation, and recombinant trypsin is used. Most recently, on September 13, 2022, the United States joined a list of 30 countries that meet WHO criteria for circulating vaccine-derived poliovirus (cVDPV) (Centers for Disease Control and Prevention, 2022; Polio Global Eradication Initiative, 2023).

Shortly after the COVID-19 pandemic began in early 2020, the first of more than 200 global vaccine candidates entered clinical trials (McGill University Interdisciplinary Initiative in Infection and Immunity [MI4], 2022). By December of that year, regulatory agencies were granting the first EUAs. By March 2023, more than 13 billion COVID-19 vaccine doses had been administered (Holder, 2023). It is estimated that in 2021 alone, more than 14 million lives were saved by the new vaccines (Watson et al., 2022). Many factors contributed to the accelerated vaccine development, including pre-investment in new modalities (mRNA and replicating deficient adenovirus) and massive government funding for clinical studies. Approximately 40% of the doses were the previously established whole virus inactivated type, and the remainder was split evenly between two types: mRNA-LNPs and adenovirus vectors (World Health Organization, 2023) (Table 2). From a chemistry, manufacturing, and control (CMC) perspective, significant advances were also made in vaccine manufacturing and analytics in response to the pandemic to shorten the timeline from proving efficacy to making vaccine doses broadly available—supporting scaling of production from millions to billions of doses per year. Changes in the scale and efficiency of manufacturing required sophisticated analytical characterization capabilities.

Table 2. Top 10 COVID-19 vaccines (from World Health Organization as of November 1, 2022).
Vaccine Total doses Modality Adjuvant Cells Stability
Sinovac—CoronaVac 6,184,279,673 Whole inactivated adjuvanted Aluminum Vero Refrigerated
Pfizer BioNTech—Comirnaty 6,160,304,332 ss mRNA LNP No −60 to −90°C then 2–8°C 10 weeks
AstraZeneca—Vaxzevria & SII—Covishield 5,470,000,991 Adenovirus dsDNA No Fetal kidney Refrigerated
Beijing CNBG—BBIBP-CorV Sinopharm: Covilo 3,524,063,903 Whole inactivated adjuvanted Aluminum Vero Refrigerated
Moderna—Spikevax 1,826,052,508 ss mRNA LNP No −15 to −50°C then 2–8°C 30 days
Janssen—Ad26.COV 2-S 918,763,386 Adenovirus No Fetal kidney Frozen then 2–8°C
Gamaleya Sputnik V—Gam-Covid-Vac 679,249,657 Two different adenovirus (prime, boost) No Fetal kidney Frozen liquid & refrigerated lyophilized
CanSino—Convidecia Ad5-nCoV-S 307,445,112 Adenovirus, IM prime, inhaled boost) No Fetal kidney Refrigerated
Bharat—Covaxin 165,156,000 Whole inactivated adjuvanted Aluminum Vero Refrigerated
Shifa—COVIran Barakat 140,471,658 Whole inactivated adjuvanted Aluminum Refrigerated
  • Abbreviations: ds, double stranded; IM, intramuscular; LNP, lipid nanoparticle; ss, single stranded.

In this review, we describe the technological advances that have been made in the last 10 years to allow the development and manufacturing of novel vaccines for prevention of a variety of human infectious diseases. Some of these advances, especially in mRNA and viral vector delivery, also apply to the emerging area of therapeutic cancer vaccines. The scope of this review covers the entire vaccine manufacturing process, from cell culture and cell-free systems to formulation, purification, stability, analytics, and changes in the regulatory landscape.

2 CELL CULTURE

Cell culture has been increasingly used to express antigens for vaccine manufacturing (Figure 1). Viral expression can be achieved in either adherent cell lines or suspension-adapted cell lines.

Details are in the caption following the image
Summary schematic of cell-based viral vaccine manufacturing. Image from Kamen and Cervera (2023), reprinted with permission from Taylor and Francis Group LLC US.

2.1 Adherent cell lines

Historically, the choice of cell line for vaccine manufacturing has been limited due to concerns about safety issues. Over time, more choices have become available, especially the use of continuous cell lines; this progress has been summarized recently (Knott et al., 2023). One example of an adherent cell line-based vaccine is a recombinant vesicular stomatitis virus (rVSV)-based Ebola vaccine, rVSVΔG-ZEBOV-GP (V920, trade name ERVEBO®) (Blue, 2016), which was initially constructed and characterized as an anti-bioterrorism agent (Coller et al., 2022). The rVSV strain Indiana contains a deletion of the VSV envelope glycoprotein replaced with the Zaire Ebola Virus (ZEBOV) Kikwit 1995 strain surface glycoprotein, and it is produced in Vero cells (Merck & Co., 2022). Recent innovations also include the development of a variety of single-use bioreactors (SUBs) with very large surface per unit volumes; for example, the bioreactor described by Dohogne and colleagues (2019). The fixed bed reactor described by Drugmand et al. in 2020 combines intensified continuous and automated single use bioprocessing approaches. The example described for polio vaccine links together upstream processing, downstream processing and inactivation.

2.2 Suspension culture

Several vaccine cell culture applications involve suspension culture. For making proteins, insect cell culture (Sf9) with baculovirus expression has been used successfully; for example, in producing HA protein at scale for an FDA-licensed influenza vaccine (Buckland et al., 2014). Alternatively, host cells for virus production, such as Vero cells or PER.C6 cells, can be adapted to suspension culture. If cell culture can be adapted, cells can usually be grown in a classical stirred tank bioreactor using well-established methods. Many of the technology improvements for CHO cell culture that have been developed for manufacturing antibodies (Handlogten et al., 2023; Jagschies et al., 2018) can be applied to vaccine manufacturing (Young et al., 2022). Recently, there has been a transition in the industry from stainless steel bioreactors at the 10,000–20,000 L scale to SUBs at scales of around 3000 L. Watterson et al. (2021) described a SARS-CoV-2 Sclamp vaccine candidate expressed using CHO cells that is compatible with large-scale commercial manufacture and is stable at 2–8°C. When formulated with MF59 adjuvant, it elicits neutralizing antibodies and T-cell responses and provides protection in animal challenge models. The truncated glycoprotein E of VZV expressed in CHO cells is the basis of the very effective Shingrix vaccine (Heineman et al., 2019). The cell culture technology used for vaccines can be very similar to that used for manufacturing antibodies at large scale.

Investigators at Oxford University led efforts to develop effective vaccines during the COVID-19 pandemic (Lane, 2020) using a platform based on replicating deficient adenovirus for delivery of the spike protein. The scale-up of this platform was described at an ECI Conference in 2022 (Engineering Conferences International, 2022). The platform, involving expression of a replicating-deficient adenovirus for delivery of the coronavirus spike protein using suspension cell culture, was scalable and successfully transferred to multiple manufacturing sites by Astra Zeneca to enable the supply of over a billion doses of COVID-19 vaccine in about 18 months.

3 MICROBIAL PRODUCTION

Microbial fermentation platforms have traditionally been used to make major categories of vaccines, and they remain of critical importance. The surface L1 protein on the human papilloma virus (HPV) is expressed by Saccharomyces cerevisiae yeast and assembled into virus-like particles (VLPs) to create the Gardasil vaccine (Nooraei et al., 2021). The hepatitis B virus (HBV) surface antigen is expressed by either S. cerevisiae or Pichia pastoris (Spice et al., 2020). DNA plasmid-based vaccine candidates are typically made by Escherichia coli bacterial fermentation (Liu, 2019). The pneumococcal conjugate vaccines consist of capsular surface polysaccharides (CSPs) expressed by specific Streptococcus pneumoniae strains conjugated to diphtheria CRM197 protein (Javed & Mandal, 2021). Some important vaccines that started as live-attenuated virus (LAV) vaccines, such as HZ vaccine, were eventually replaced by a single recombinant VZV glycoprotein with an adjuvant system. The past decade has seen progress in the development of alternative cultures capable of high production levels, such as the Dyadic platform based on fungal culture of Thermothelomyces heterothallica (known as C1). The C1-based system can develop stable strains in weeks that express immunologically active antigens (Keresztes et al., 2022; Ramot et al., 2022); in general, yields are higher compared to other well-established systems. The C1-based system is easily scalable to industrial volumes and is cost-effective, using standard microbial fermentation reactors and inexpensive media. A COVID-19 vaccine candidate that uses C1 expression technology to make the key receptor binding domain (RBD) component of the spike protein is now being evaluated in the clinic (Dyadic, 2023). Researchers at MIT developed a P. pastoris-based platform that is fully automated and could be used to make a great variety of proteins including vaccine antigens. This yeast was chosen because it can be quickly grown to high cell density with acceptable protein expression (Crowell et al., 2018). An application of similar technology for the production of a trivalent vaccine candidate for rotavirus is underway using a highly automated bioreactor skid that includes end-to-end capability (Dalvie et al., 2021).

For protein antigens, the technical challenge is often to manufacture reproducibly with the key epitopes exposed so that they are recognized by the immune system. This is especially true for VLP-based vaccines, such as the HPV vaccines, where sophisticated analytical capabilities are critical for success. For mRNA vaccines, Escherichia coli fermentation remains the primary method for producing the plasmid DNA (pDNA) template required for in vitro transcription (IVT) into mRNA, such as in the case of the SARS-CoV-2 mRNA vaccines Comirnaty® by Pfizer-BioNTech and Spikevax® by Moderna (Gote et al., 2023). Cell-free synthesis of pDNA is making strides (see Cell-Free Synthesis section) but is still not as cost-efficient as microbial production due to the high cost of the enzymes and nucleotides/nucleosides. The pDNA serves as raw material in the IVT reaction and the overall amount required is significantly less than for DNA vaccines, as many copies of mRNA can be made from a single linearized pDNA template. The sequencing for pDNA intended for use in IVT often requires large numbers of repeat elements (e.g., poly(A) tail), which informs E. coli cell strain selection. Many industry workhorse strains were originally optimized for high pDNA yields rather than pDNA fidelity, but now a variety of strains are available that are capable of correctly replicating larger plasmids with repeat elements (Mairhofer & Lara, 2014; Selas Castineiras et al., 2018). A novel method for making conjugated vaccines in E. coli has recently been described, via insertion of key enzymes into bacteria (Kay et al., 2022). This approach sidesteps the requirement to make the protein and polysaccharide separately, followed by purification, chemical conjugation, and repeat purification (Abouelhadid et al., 2023).

Many innovations have been made to fermentation equipment since the days of manufacturing scales of 500–5000 L in stainless steel pressurized vessels (Gomez & Robinson, 2018), which are becoming increasingly complex to automate, clean, validate, and operate. Due to the high costs of construction, only a limited number of large-scale manufacturing sites exist in developed countries. A major advance in both microbial and cell culture in the past decade has been the transition to SUBs. Heavy investments have been made in disposable SUBs, allowing the manufacture of microbial cultures at smaller scales by eliminating cleaning and sterilizing and increasing flexibility. Many process conditions common to microbial cultures, such as pressure, high air flow rates, solvents, and robust temperature control, limit the upper scale of SUBs for vaccine manufacturing to around 3000 L.

For purification, single-use options have been established for the complete workflow—from depth filtration, tangential flow for concentration and diafiltration, and single-use disposable columns for chromatography (discussed in the Purification section). Vaccine manufacturers are working to establish second source suppliers of single-use systems to avoid supply chain concerns. Industry initiatives, such as those by NIIMBL (Erickson et al., 2021), are establishing flexible single-use systems that have standardized, supplier agnostic assemblies and connections. End users and technology providers are also working together to provide sustainable recycling and circular economy strategies for single-use assemblies.

4 CELL-FREE SYNTHESIS

IVT of vaccines, including protein and RNA vaccines and VLPs, has become established as an elegant and simple approach to accelerate both vaccine process development and manufacturing. Cell-free IVT begins with a linearized DNA template used to transcribe the replicon RNA under the control of T7 RNA polymerase enzyme (Figure 2). The IVT reaction is complete in a few hours to generate a product, such as RNA, at 3–5 mg/mL. For RNA vaccines, the 5′ end of the RNA replicon (Figure 3) can be capped by supplementing the IVT reaction with m7GppG cap structure analogs and keeping a high molar ratio of the cap molecule relative to the first nucleotide (GTP) of the RNA sequence (Whitley et al., 2022). Capping has been commercialized as co-transcription capping using CleanCap Reagent (Vaidyanathan et al., 2018). Alternatively, mRNA can be capped in a second enzymatic reaction using capping enzyme (VCC) and a methyl donor as a substrate. VCC capping efficiency is higher (100%) compared to 60%–80% with the cap analog, but co-transcription capping is a faster process as it does not require a second enzymatic reaction step. At the end of IVT, DNAase I enzyme is added to degrade the DNA template, followed by EDTA to quench DNAase activity, stabilize the RNA product, and mitigate RNA precipitation. The removal of impurities including enzymes, residual NTPs, the DNA template, and aberrant mRNAs (double-stranded [ds]RNA and truncated RNA fragments) is critically important, as RNA purity has been shown to impact translation efficiency and modify the immunostimulatory response of the vaccine in vivo (Rosa et al., 2021). For example, a 10- to 1000-fold increase in protein expression has been shown after the removal of double-stranded RNA (dsRNA) by reversed-phase high-performance liquid chromatography (RP-HPLC) (Karikó et al., 2011). TFF has been used for broad purification of RNA after IVT, achieving purities of 90%–99% and yields of 90%–95% (Berlanda Scorza et al., 2014). TFF is also complemented by chromatography for the removal of dsRNA, as discussed in the Purification section. For RNA, scale-up can be enabled by single-use compatible RNAase-free systems such as microplates, tubes, and bioreactors that include rocking motion or stirred tanks as shown in Figure 2.

Details are in the caption following the image
Cell-free IVT reaction for RNA production. Typical reactions start at 2–5 mL for optimization screening (2–10 mg), 10–15 mL IVT for animal studies (25–50 mg), 5–20 L for clinical studies (10–50 g), and up to 50–200 L for commercial manufacturing (125–500 g). IVT, in vitro transcription.
Details are in the caption following the image
mRNA structure via in vitro transcription (IVT). Image adapted from Linares-Fernández et al. (2020). mRNA, messenger RNA.

The simplified RNA manufacturing process enables fast vaccine development; for example, from DNA template to first in-human trial was achieved in 42 days for the SARS-CoV-2 vaccine (Kis et al., 2021). The time to commercial launch can also be reduced to less than 2 years for IVT RNA vaccine candidates. Commercial facilities can be 2–3 orders of magnitude smaller and constructed in half the time and at one-twentieth of the upfront capital cost (Kis et al., 2021). One billion doses can be manufactured in a low-cost $20 million facility (Kis et al., 2021). Further reduction in facility costs can be achieved through the use of closed processing, which allows production in lower-grade clean room facilities. While the elimination of higher-grade clean room requirements saves both energy and capital costs, reagent costs dominate the economics of the IVT process, contributing up to 96% of the manufacturing cost of goods (COGs; Table 3). Material and reagent supply for Good Manufacturing Practice (GMP) is a limiting factor (Rosa et al., 2021), particularly with the added complexity of ensuring that enzymatic reagents are ACF and GMP grade. Methods to minimize reagent use are significantly important; approaches include post-IVT capping and integrated continuous processing, where reaction rates can be accelerated and reagent reuse implemented. Automated RNA production systems are also being developed, where the IVT reaction is integrated with the purification steps under robotic operations in a single enclosed system (Castillo et al., 2023). Similarly, microfluidic systems have integrated RNA to LNP production for personalized vaccine approaches (Nath, 2022) (Figure 4). Today, benchtop systems can be used to simplify the discovery construct workflow, as oligomers are assembled to either full-length DNA or mRNA sequences at a 10-µg scale, without the need for a pDNA template (Gill et al., 2021; Venter et al., 2021). In the future, benchtop nucleotide manufacturing systems may be possible.

Table 3. Techno-economic assessment of mRNA IVT production to produce 1 billion doses.a
mRNA mRNA mRNA sa-mRNA sa-mRNA
µg/dose 100 30 12 10 1
Total reaction volume (L) 20,000 6000 2400 200 20
IVT reaction volume (L) 50–200 (250 actual) 50–200 (75 actual) 50–200 (30 actual) <50 (5 actual) <50 (1 actual)
Annual throughput (kg) 100 30 12 10 1
COGs ($/dose) 1–3.6 0.2–1.14 0.1–0.45 0.05–0.2 0.02–0.08
COGs ($ billions/year) 1–3.6 0.2–1.2 0.1–0.45 0.05–0.2 0.02–0.04
Material costs (%) 90–96 80–93 70–92 50–70 35–60
  • Abbreviations: COG, cost of goods; IVT, in vitro transcription; mRNA, messenger RNA; sa-mRNA, self-amplifying mRNA.
  • a Assumptions: Reaction output of 5 g/L RNA IVT from one bioreactor with 65% yield and new facility at 80% utilization with 12-month campaign duration. Costs do not include delivery encapsulation. COGs show a range to cover the types of capping and enzyme sourcing supply approaches. Analysis supported by the Biopharm Services Ltd Biosolve Process Cost Analysis Tool.
Details are in the caption following the image
Lipid nanoparticle assembly workflow. Image adapted from Guevara et al. (2020).

While the design structure of the single-stranded RNA is well established, there continues to be expansion of new RNA therapeutic modalities. Self-amplifying mRNA (sa-mRNA), with a high copy number and subsequent high antigen expression, is effective at up to a 60-fold lower dose than single-stranded RNA (Ljungberg & Liljeström, 2015; Vogel et al., 2018). Trans-amplifying mRNA (taRNA) has also been established in animal studies, where one mRNA template encodes the replicase while separate templates amplify multiple target mRNAs, encoding different proteins simultaneously (Beissert et al., 2020). Covalently closed circular RNA (circRNA) has shown feasibility in animal models for continuous translation of proteins (Bai et al., 2022; Chen et al., 2023). CircRNA is generated from IVT linear RNA and can be circularized by a number of approaches such asT4 RNA ligase or intron splicing (Bai et al., 2022). This stable closed loop structure provides protection from cleavage by exoribonucleases, does not require 5′ capping or 3′ poly(A) tails, and has the potential to be stable at room temperature (Bai et al., 2022). It is anticipated that guided in silico molecule design with mechanistic modeling and artificial intelligence (AI) tools will continue to improve not only the molecule design for optimizing half-life, stability, and potency (Mukhopadhyay & Hart, 2021) but also improve CMC attributes including impurity reduction. Recent work shows further yield and purity improvements from reducing dsRNA by lowering the IVT Mg2+ level, running faster reactions at 50°C with a thermostable T7 RNA polymerase, and applying high-salt transcription strategies (Linares-Fernández et al., 2020; MalagodaPathiranage et al., 2023). These efforts to expand the process design space can be supported by expanding automated screening tools to enable statistical design of experiments with integrated workflows of IVT, purification, and process analytics (Skok et al., 2022).

Beyond RNA, cell-free manufacturing has been demonstrated for various vaccine types including subunits, conjugates, VLPs, and membrane-augmented vaccines (Hu & Kamat, 2023; Maharjan & Park, 2023; Rodríguez-Limas et al., 2013). These examples primarily use lysate systems from E. coli, although a range of sources are used, including CHO, wheat germ, and tobacco. Commercially available kits provide fast methods to create material for discovery and process development and scale-up to support clinical manufacturing has been demonstrated (Groff et al., 2022). A recent study applied cell-free expression technology to successfully produce a next-generation 24-valent pneumococcal conjugate vaccine equivalent to the current standard of care (Fairman et al., 2021). The cell-free platform expressed an enhanced carrier CRM protein sequence that contained multiple non-natural amino acids. Their location on the outside of the of the primary T-cell epitope enabled access for site-specific covalent conjugation and the potential for improved serotype coverage. The wider adoption of cell-free vaccine production remains limited due to the challenges of posttranslational modifications, immunogenicity concerns with bacterial-sourced systems, disulfide folding complexity, and the cost of reagents for scale-up. It is anticipated these will improve in the next 10 years with advances in the application of mammalian-based lysate systems and continuous processing systems.

5 PURIFICATION

The purification of vaccines during the last decade has continued to transition away from complicated workflows towards more standardized and intensified approaches. Examples are summarized in Figure 5. This focus continues to drive down costs and dose while improving throughput and yields. Incremental improvements have been made to cell harvesting with centrifugation, depth filtration, cell lysis, and filtration as summarized in recent reviews (Besnard et al., 2016; Zydney, 2021). Improvement in membrane design has facilitated a transition from centrifugation to depth filtration, or a combination of both. The expansion of available higher MWCO membranes has supported concentration, buffer exchange, and impurity reduction (host cell proteins [HCPs]) for larger viruses. Continued improvements in single-pass TFF has enabled intensified and continuous workflows. Sterile filtration of viral products continues to improve with advancements in membrane designs (Zydney, 2021).

Details are in the caption following the image
Vaccine production processes. Process types, including traditional process flows, are compared to intensified/integrated future approaches.

The widening of affinity capture chromatography options provides a key advantage for streamlined and standardized workflows (Zhao et al., 2019). For viral processing, the availability of affinity capture provides a significant advantage in reducing the number of unit operations, as shown in Figure 5. Advances in de novo engineered ligand design has led to high specificity and tunable affinity ligands for viral vaccines, with robust impurity clearance at mild elution conditions (Łącki & Riske, 2020). Commercially available sources, such as those derived from camelid single-chain domains, have become established for a range of adeno-associated virus (AAV) serotypes such as Capture select AAV9. Improvements to ligand design continue to grow, such as the short base (5–15 amino acids) ligands, using combinatorial chemistry approaches (Camperi et al., 2014) and molecular imprinted in silico designed polymers for fast ligand particle generation (Bhalla et al., 2022). For RNA, the deoxythymidine (oligo dT) ligand has become established as the affinity capture approach that binds to the 3′ poly(A) tail. Impurities including dsRNA can be efficiently removed by combining OligodT and polishing IEX (Gagnon et al., 2020a). Advances in affinity chromatography enable wider adoption of efficiency improvements in multicolumn processing (Pollard et al., 2016). For example, a sixfold increase in productivity was shown for two-column counter-current chromatography with size-exclusion chromatography (SEC) for adenovirus, achieving 86% recovery and 90% and 89% clearance of DNA and HCP, respectively (Nestola et al., 2014). A similar intensified approach was applied by Fischer et al. (2018) to the purification of influenza A virus. A quaternary amine (QA) anion exchange monolith with simulated moving bed chromatography achieved a virus yield at 89% with >98% DNA clearance.

The transition from traditional resin bead-based separations to convective formats of membranes and monoliths (Figure 6) continues to gain wider adoption. The diffusional limitations of resin beads, where large virus particles struggle to access the ligands within the beads, is overcome by the advantages of the larger pore size and higher flow rates from convective transport membranes, including microporous hydrogel nanofibers (Zhao et al., 2019) and monolith formats (Sbaizero et al., 2018). These have been successfully applied across a range of vaccine products, with typical recoveries of 60%–90% and minimal binding of impurities, including 99% removal of HCP and residual DNA (resDNA) and 5-log endotoxin reduction (Zhao et al., 2019). Impurities removed include viruses (Pato et al., 2019; Sviben et al., 2017), VLPs (Lima et al., 2019), bacteriophages, extracellular vesicles (Gagnon et al., 2020b), pDNA (Almeida et al., 2020), and RNA (Gagnon et al., 2020a). The efficiency of monolith processing was recently demonstrated with IEX monolith purification of rubella virus suspension in 1 h compared to >10 h with resin bead (Forcic et al., 2011).

Details are in the caption following the image
The expanding format for chromatography. (Left panel) Resin beads. Functional ligands are internalized inside the pores of resin media; this format is dominated by slow diffusive mass transfer creating a slow or lower resolution process with significant size exclusion effect for large molecules. Turbulent flow occurs between the resin with vortex flow and zones of elevated shear stress and lower yields. (Center panel) Membrane technology with functional groups immobilized at the membrane surface. The wide pore size permits high flow rate and minimizes size exclusion effects. (Right panel) The monolithic format supports interconnected channels with functionalized macropores or both macro and mesopores. The large diameter channels provide convective mass transport and enable flow-independent resolution and capacity that support rapid processing. The laminar flow provides low shear stress and higher yields for biologics and vaccines. Image adapted from Zhao et al. (2019), reprinted with permission from Elsevier.

The pursuit of intensified processes and increasing the upstream expression titer has renewed interest in non-chromatographic approaches such as precipitation, aqueous two-phase extraction (ATPE) and crystallization. Chromatography-free processes have the potential to reduce the COGs by 20%–30% and be more amenable to continuous processing. The use of statistical design of experiments reduces the complex burden of optimizing the conditions to maximize target stability and integrity with pH, temperature, conductivity, and mixture composition. Examples of typical polyethylene glycol (PEG) precipitation include hepatitis A virus recovery of 85% with a 65% reduction of HCP achieved using 5% PEG with 0.5 M NaCl. The lysate is then treated with nuclease to prevent co-precipitation of nucleic acid/virus aggregates (Hagen et al., 1996). mRNA precipitation has been combined with TFF to capture the precipitate and impurities removed by diafiltration, then resolubilized (DeRosa et al., 2021). It is anticipated that continuous vaccine precipitation can be achieved using plug flow or coiled flow inversion reactors (Mittal et al., 2022). The intensification of ATPE, with improved upstream titers, involves mixing two solutions of a polymer and kosmotropic salt (phosphate citrate, sulfate), and can overcome the historically large volumes of overly dilute extraction systems (Leong et al., 2022). A number of recent examples have shown feasibility with PEG/salt compositions to isolate a range of vaccine products including viruses, VLPs, and extracellular vesicles, in the upper phase of the extraction (Leong et al., 2022), achieving 70%–99% recovery from crude lysates. Turpeinen et al. (2021) showed 99% HIV VLP recovery and 73% removal of DNA with a 45% PEG and 30% citrate mixture in a continuous mode using simple helical mixers, where such approaches should reduce large-volume buffer usage and processing time.

Although vaccine crystallization is in its infancy as a purification tool, a number of recent advances have shown compelling opportunities. For example, crystallization of a nonreplicating rotavirus vaccine candidate (VP8 subunit proteins fused to the P2 epitope of tetanus toxin) has been demonstrated (Hong et al., 2021). Screening crystal conditions were evaluated in a hanging drop vapor diffusion system, then scaled up using evaporative technology. The subsequent antibody binding profile was found to be similar to pre-crystallization. Technology has also expanded to tunable crystal size control with continuous slug flow crystallization (Mozdzierz et al., 2021). These technologies open the possibility that vaccine crystallization can be used in continuous production. It will be interesting to see how far technology development can drive the impurity clearance from the target molecule, or whether crystallization will need to be combined with chromatography or either precipitation or two-phase extraction.

In the next 10 years, it is anticipated that processes will become further intensified and integrated to smaller footprints, thus providing lower cost processes for easier global deployment. Improved technology could lead to one-step chromatography and continuous processing with single-pass tangential flow filtration and sustainable practices of expensive reagent recycling, which is particularly important for RNA production. Technologies to support non-chromatographic processes, including crystallization, are expected to enter into preclinical development. All of these activities will be supported by mechanistic modeling for AI-guided process development (Canova et al., 2023; Hossienizadeh et al., 2021) and advanced control strategies to support autonomous operations.

6 DRUG PRODUCT (DP) PROCESSING

Most approved vaccines are delivered by intramuscular (IM), s.c., or i.d. injection, with a few approved oral and nasal vaccines. In addition, development programs are underway for vaccines delivered to the lungs via inhalation and to the intradermal space using microneedle patches. Sterile injectable DP manufacturing includes compounding (formulation), sterilizing filtration (in most cases), filling vials or pre-filled syringes, visual inspection, labeling, and packaging (i.e., fill-finish). Some vaccines are freeze-dried (lyophilized) in the final container after filling. Conjugated vaccines require conjugation of antigen to carrier protein. Finally, process steps may be required for incorporation of adjuvants for some vaccines (Josefsberg & Buckland, 2012). Many injectable vaccines, such as those containing live attenuated viruses or aluminum gel adjuvant, are not amenable to sterile filtration, so upstream steps are conducted aseptically to ensure sterility.

Many of the vaccines developed for COVID-19 included innovations in DP manufacturing. Key among these was the use of mRNA-LNPs for Comirnaty® by Pfizer-BioNTech and Spikevax® by Moderna. A global network of fill-finish sites was required to build capabilities for new unit operations of LNP formation and TFF. LNP formation entails controlled pumping of an aqueous mRNA solution and an ethanol solution of four lipids into a micro-mixing apparatus as illustrated in Figure 4 (Thorn et al., 2022). The just-formed mRNA-containing LNPs have very limited physical stability, so the LNP formation step is immediately followed by solvent and buffer exchange using TFF to remove the ethanol and adjust the pH (Ramachandran et al., 2022).

Although the COVID-19 mRNA-LNP vaccines demonstrated remarkable effectiveness (Chi et al., 2022), a major limitation is the requirement for frozen storage and shipping, which increases cost and complexity and hinders wider distribution. The Pfizer-BioNTech vaccine currently has a shelf life of 18 months but must be maintained at −90°C to −60°C (emc, 2023a2023b) whereas the Moderna vaccine shelf life is 9 months at −50°C to −15°C (emc, 2023a2023b). A recent review summarizes the areas of focus for improving stability of mRNA-LNP vaccines, including optimizing the lipids that make up the nanoparticles, shortening the mRNA chain length, increasing the GC content, and using drying technologies such as lyophilization, among others (Blenke et al., 2023).

Lyophilization of mRNA-LNP vaccines is feasible and could improve thermal stability (Ai et al., 2023). While future mRNA-LNP products may be lyophilized, the current mRNA-LNP COVID-19 vaccines are not, as global lyophilization capacity is limited and there was insufficient capacity to manufacture billions of doses at the time of the pandemic. Emerging technologies such as spray freeze-drying and aseptic spray drying may, in the future, be viable options (Searles & Ohtake, 2021). Under certain conditions, a single spray freeze-dryer or spray dryer can match the throughput of many lyophilizers (see Section 6.4).

The response to COVID-19 also included major advances in recombinant viral-vectored vaccines such as AstraZeneca's Vaxzevria® (manufactured by Serum Institute of India as Covishield®), Covilo® by Sinopharm, Ad26.COV 2-S from Janssen, Sputnik V from Gamaleya, and CanSino's Convidecia (Mendonça et al., 2021). In the last 10 years, four viral-vectored vaccines were approved for Ebola and one for Dengue (McCann et al., 2022) (Table 1). The viral-vectored vaccines use conventional fill-finish technology, including a mix of refrigerated liquid, frozen to −20°C, and lyophilization.

Many novel vaccine DP formulations and delivery methods are under development. Some will require new GMP manufacturing steps, such as microneedles (D'Amico et al., 2021) and nanoparticles of many types (Bezbaruah et al., 2022).

6.1 Recombinant human albumin

The use of any animal-derived ingredients in manufacturing or formulation requires extensive precautions against contamination with pathogenic agents. Albumin and gelatin have long been used as formulation components to stabilize some vaccines (Volkin et al., 1999). For many years, the source of albumin was pooled human volunteer donor plasma. In 2006, Merck & Co. Inc. obtained approvals for the use of Recombumin® recombinant human albumin expressed in S. cerevisiae for upstream manufacturing of MMR®II.

In 2019, Merck & Co. Inc. obtained EMA (European Commission, 2020) and FDA (US Food and Drug Administration, 2019) approvals for ERBEVO® Zaire ebolavirus vaccine that uses recombinant human albumin as a DP formulation excipient (Merck & Co., 2022). The albumin is rice-derived Exbumin® by InVitria (Cell Culture Dish, 2021). The vaccine requires storage and transport at −80°C to −60°C with up to 14 days at 2–8°C (Merck & Co., 2022). It is expected that recombinant albumins will be used for more vaccines moving forward.

6.2 Adjuvants

Adjuvants are critical to the effectiveness of the vaccines that include them and are their own specialty area of development and manufacturing. Many of the adjuvants in use today are held as intellectual property (IP). Examples include Merck Aluminum Adjuvant (Merck & Co., Inc), the GSK Adjuvant System, MF59® (Novartis), Matrix-M (Novavax), and several owned by Access to Advanced Health Institute in Seattle, WA.

The most common adjuvant types among vaccines currently licensed by FDA and EMA are insoluble aluminum salt suspensions (amorphous aluminum hydroxyphosphate sulfate, aluminum hydroxide, aluminum phosphate, and potassium aluminum sulfate) (Facciolà et al., 2022). This class has been in use for almost 100 years. They have base particle sizes in the tens of nanometers and form loosely connected networks that measure 1–20 µm depending upon the method of measurement and shear conditions (HogenEsch et al., 2018). Vaccines with aluminum adjuvants are not frozen, as it causes an increase in adjuvant particle size; however, with sufficient stabilizer such as sucrose and/or with sufficiently rapid freezing, these vaccines can be lyophilized or spray freeze-dried to improve stability (Preston & Randolph, 2021). An additional issue is the potential for the aluminum particles to form a sediment that is difficult to resuspend (Langford et al., 2020).

Other adjuvants in commercial vaccines include a chemically modified lipid portion of Salmonella minnesota lipopolysaccharide (MPL, MPLA), synthetic oligonucleotides containing CpG motifs to mimic bacterial DNA (e.g., CpG 1018®), squalene, and saponins (Facciolà et al., 2022). These are used in various combinations, and in some cases adsorbed to aluminum hydroxide or as part of emulsions or nanoparticles. A common theme across all the adjuvants is particle size, an attribute with immunological consequences that can be difficult to manage through manufacturing and quality control.

6.3 Stabilization

During product formulation and process development, degradation pathways are identified to assess the impact of changes in the manufacturing process and product composition on stability. One of the most important chemical degradation pathways for biologics is hydrolysis, which often results in opening peptide bonds. Other chemical degradation pathways include dehydration and oxidation. Physical changes include aggregation and denaturation—or misfolding in the case of proteins—of subunits or the primary modality of the vaccine. These molecular changes result in product quality changes that lead to a loss of function, off-target immunogenicity, or other adverse effects. A single change in the molecule can render it completely ineffective, depending on where the structure is affected. On the other hand, some changes will not have any impact because they are not in an area related to product activity. Due to the complexity of the antigen and adjuvants used in vaccine products, it is difficult to understand or predict all the possible changes and their impact; in general, it is important to minimize all degradation.

Factors that control or accelerate degradation must be controlled and monitored in the vaccine product. These include product pH, oxygen levels, and water activity (e.g., in solid-state formulations), which contribute to hydrolysis and oxidation. Elevated temperature typically accelerates degradation across all pathways. Certain buffer species, ionic strength, and trace metals can impact the stability of a product. Finally, exposure to light, freeze-thaw, and exposure to shear or agitation that introduces turnover at air-liquid or solid-liquid interfaces can cause product changes.

Product stability can be ensured by adding different types of stabilizers, which are identified during formulation studies. The final composition depends on the type of product, degradation risks, and route of administration. Stabilizers include cryoprotectants, oxygen scavengers, and proteins; examples are gelatin, PEG, and specific buffers to manage pH. The particular type and amount of a given stabilizer can be informed by current understanding of the underlying mechanism(s) of degradation. A review of the details of the range of different mechanisms and formulation strategies used to address those issues is beyond the scope of the present review, but the interested reader is directed to other reviews that focus on formulation design and choice of stabilization approach (e.g., different drying techniques) (Kumru et al., 2014; Preston & Randolph, 2021).

Once the product composition has been defined, environmental controls are necessary throughout manufacture and distribution. Reduced temperatures, whether refrigeration or freezing, can protect product quality. Since many degradation pathways can be limited by reducing the presence of water and oxygen, products can be dried through freeze drying (lyophilization) or spray drying and introducing an inert gas into the headspace. Historically, lyophilization has been the major avenue of stabilization of vaccine products and, along with other forms of drying, will be important for future vaccine products as well (Chen et al., 2021; Degobert & Aydin, 2021; Preston & Randolph, 2021) (see Section 6.4).

Table 4 summarizes the different platforms for vaccine products and the types of formats that are used, along with long-term storage conditions. Some liquid products cannot be frozen without impacting product quality; this is particularly important for vaccine products with aluminum adjuvants. Frozen conditions can protect products that would otherwise be solution/liquid-based, but results in complex manufacturing and supply chain challenges. Dried products can be stored at higher temperatures, thereby minimizing quality risks, but present the need for a reconstitution step and/or production of a diluent unless the final dosing form is solid state.

Table 4. Vaccine product formats and storage conditions.
Vaccine type Product format Long-term storage conditions
Liquid Frozen Dry Ambient 2–8°C −20°C −80°C
Live-attenuated virus X X X
Inactivated virus/bacteria X X
VLP X X
Subunit—protein, polysaccharide X X
Viral vector X X X X X X
DNA X X
mRNA/LNP X ? ? X X
  • Abbreviations: LNP, lipid nanoparticle; mRNA, messenger RNA; VLP, virus-like particle.

6.4 Drying trends

Dried vaccine product is packaged in a variety of containers, which determines the operations selected for final dry product processing. The majority of dried vaccine products have been commercialized in vials, which has increased global capacity. Filling machines are very fast, flexible for different sizes of vials, and installed globally. In addition, vial filling equipment can be shared among several liquid products, increasing utilization and the financial return for the high capital investment in aseptic filling facilities. Lyophilization equipment has been designed to optimize production of single or multidose units in glass vials. For lyophilized products, however, a custom diluent may be needed, which doubles the filling requirement. The need for additional components and second manufacturing operations increases costs. At patient administration, two sets of materials introduce extra steps and increase the risk of error in the clinic. Dual-chamber options have been developed for convenience of reconstitution. Although these products eliminate transfer steps in the clinic, additional manufacturing steps are required to lyophilize in one chamber and fill diluent in the second chamber.

Vaccines are typically administered parenterally, so the final stages of manufacturing must be done under aseptic conditions. Containers must be washed and sterilized in line or purchased ready-to-use (at increased cost). For both liquid and lyophilized products, filling and stoppering are necessary. Products to be lyophilized are only partially stoppered then transported aseptically to the cabinets where freezing, drying, gas overlay, and full stoppering occur (Figure 7). Systems have been designed to allow direct transfer from filling lines to the cabinets while maintaining aseptic conditions. For dual-chamber syringes, differences in the process require specific equipment. Figure 8 illustrates the operations needed to complete the product manufacture. Since the equipment is dedicated to this type of product, utilization can be more difficult to manage. The cost for dual-chamber products is increased by the number of components, number of steps, and the reduced capacity of the equipment.

Details are in the caption following the image
Process operations for liquid and lyophilized products. Image adapted with permisssion from ATS Scientific Products (2023).
Details are in the caption following the image
Process operations for dual chamber systems. Image from Werk et al. (2016), reprinted with permission from Elsevier.

6.4.1 Advanced product drying technologies

Advances in manufacturing technologies for dry vaccine product must include capabilities for the new vaccine formats discussed above while focusing on increased productivity and improved quality assurance. Reducing the duration of the drying operation provides a significant opportunity for increased productivity. The major limit for drying cycle time is the rate of moisture removal, which must often be reduced to maintain a low product temperature to prevent “melt-back” or “collapse.” Collapse is a product defect where structure of the product cake is lost, which can impact stability and product activity. Both heat and mass transfer must be controlled to ensure final product quality. Current lyophilization cabinets use heat transfer fluid flowing inside the shelves to freeze the product and drive vaporization. For both freezing and heating, energy must be transferred across the shelf through the bottom of the glass vial and reach the full volume of the product, which can be inefficient. Inhomogeneities throughout the cabinet due to heat and mass transfer differences from the walls, between the shelves, and vial location on the shelf drive further inefficiencies.

Improvements in heat transfer, such as with the novel methods discussed below, can significantly shorten cycle times. Alternatively, a drying environment with very low partial pressure of water can drive evaporation of moisture from the product without heating the product directly (Gerde et al., 2010). Continuous production systems improve productivity by increasing equipment utilization. This principle can be applied to drying of all product types, including vials and dual chamber devices. Some of these technologies allow drying in situ, while others generate a dried product that must be distributed into containers at the proper dosage. Production of a bulk dried product provides opportunities for development of novel formats beyond the typical vial, with improved stability. Continuous lyophilization methods for vial products that leverage existing principles and vial filling capacity would be the simplest to implement (Pisano et al. 20172019).

Spray drying, which results in a dried powder product of spherical particles, is the focus of many advanced drying processes (Adali et al., 2020; Breit & DuBose, 2013; GEA Niro, 2023; Gerde et al., 2010; IMA, 2016; Kanojia et al., 2017; Lowe et al., 2018). For dried bulk technologies, filling into unit dose containers requires powder filling, which can be challenging for an amorphous powder. In addition to improved flowability of spray-dried products, the risks of generating fine particulates and contamination of external vial surface are reduced. The number of process steps remains constant, although the sequence of aseptic operations is reversed (Figure 9).

Details are in the caption following the image
Traditional lyophilization compared to bulk drying processes.

Any process improvements must meet aseptic requirements to provide appropriate product quality. The equipment must be sterilized and operate in a highly controlled environment to maintain the highest levels of sterility assurance; current trends for new installations include implementation of closed systems wherever possible. All inputs must be sterilized at the point of use through sterile filtration or aseptically introduced as pre-sterilized materials. The room classification is determined based on the equipment design and interventions required.

6.4.2 In situ drying process alternatives

In situ drying in final containers such as vials and dual-chamber syringes is currently most common for commercial products. Improvements in traditional lyophilization have been investigated for all steps in the cycle. The freezing step is critical to allowing a proper porous macroscopic structure and enhanced mass transfer for vapor removal. Developments have focused on improved control of nucleation and increasing surface area for drying steps (Arsiccio et al., 2020; Assegehegn et al., 2019). Primary and secondary drying steps can be improved by focusing on the delivery of energy, which is limited by the heat transfer through shelves to the glass vial to drive evaporation in current lyophilization cabinets. Continuous processing could lead to improvements in productivity as well as consistency.

Foam drying

Foam drying does not require freezing and leverages the evaporative cooling during drying to maintain product temperatures (Ambros et al., 2019; Hensley et al., 2021; Kubbutat et al., 2020; Lyu et al., 2022; Smith et al., 2015; Uzri, 2013). The bubbles that form provide membrane-like surfaces and a high surface area for drying; process development focuses on controlling the size of the bubbles and the thickness of the membrane to ensure a rapid and efficient drying process. Foam drying technologies have been demonstrated for preservation of vaccines as well as biopharmaceuticals, microbial samples, and agricultural products (Lovalenti & Truong-Le, 2020).

Microwave vacuum drying (MVD)

Microwave energy can be used to replace heating fluid in the shelves, with application in traditional lyophilization cabinets that could allow continued use of current equipment. Rapidly cycling water molecules in a microwave field results in heating at or near the molecular level and evaporation of water without significant heating of sensitive products (Mohsen, 2018; Stefanidis et al., 2014). In addition, the energy penetrates immediately across the entire system without generating localized heat and mass transfer effects observed in traditional lyophilization. Frequency control of the microwave field provides a tool for optimization of the system during process development for individual products (Abdelraheem et al., 2022; Sickert et al., 2023).

MVD technology has been demonstrated for vaccines and other biopharmaceuticals (Bhambhani et al., 2021; Gitter et al., 2019; Härdter et al., 2023; Kubbutat, 2021). Bhambhani et al. (2021) investigated MVD using EnWave pilot scale systems, where the product is frozen offline in vials and transferred to a FreezeREV® dryer (Enwave, 2023) (Figure 10a). Sublimation occurs due to the energy provided by the magnetrons generating the microwaves in the drying chamber. A dry ice condenser is used to condense the water vapor, and product vials are manually stoppered. Cycle times for MVD were 87% shorter than for lyophilization. Results from the early pilot-scale study showed equivalent moisture content and activity as traditional lyophilization (Bhambhani et al., 2021).

Details are in the caption following the image
Alternative drying processes. (a) Schematic of the EnWave FreezeREV Pilot Scale. Image reprinted from Mohammadi et al. (2020). (b) Continuous lyophilization for vial products. Image reprinted with permission from Capozzi et al. (2019), WIPO PCT patent no. WO 2018/204484 A1. (c) Laminar PACE system. Image reprinted with permission from Ziccum AB (2023a). (d) IMA Life Lynfinity Continuous Spray Freeze Dryer. Reprinted with permission from Sharma et al. (2021) under the terms of the Creative Commons CC BY License (http://creativecommons.org/licenses/by/4.0/).

Continuous processing

A continuous process for in situ drying proposed by Pisano and colleagues (Capozzi et al., 2019; Pisano et al., 2017) is shown in Figure 10b. Vials are moved through modules, and specially designed transfer modules are required between each step to maintain the appropriate pressure, temperature, and gas composition. This system was designed as an opportunity to produce dry product in a rapid and continuous manner using traditional and familiar process operations. The complexity of operations for moving each vial through the stages of lyophilization has led to a focus on producing bulk dried powder.

Bulk drying process alternatives

While blending and milling are often required to achieve particle size targets for small molecules or dry powder inhalable products, these extra steps are not common for injectables. Excipients are incorporated into the spray dryer feed and materials are able to be transferred directly for powder filling.

As illustrated in Figure 9, bulk drying processes reverse the order of filling and drying relative to in situ drying processes. This requires fully aseptic powder or pellet filling capability (“dry filling”). Powder filling has existed for some time to fill sterile antibiotic powders, and although challenging, the process is enabled by improved flowability of spray-dried materials. Pellets can also be filled using variations of the powder filling equipment. An advantage of bulk drying is that bulk sterile powder or pellets can be filled into almost any type of container, including conventional vials and dual-chamber syringes, as well as novel reconstitution-injection systems that are not amenable to in situ drying (Searles & Ohtake, 2021). Use of bulk drying and dry filling could dramatically reduce the cost of manufacturing dual-chamber syringes. Several other approaches have been developed to address the concerns and challenges of dry filling.

Spray drying

Aseptic spray drying is now available and allows for continuous drying and powder production. An example is the Aseptic SD™ system (GEA Niro, (2023). In spray drying, a liquid solution containing the product flows through a nozzle that disperses the product into fine particles. One of the advantages of spray drying is the ability to control particle size by tuning the spray nozzle. Sterile gas must be provided as heat transfer medium for product drying; an inert gas such as N2 or Ar may be used to limit oxidative degradation. In the traditional mode of spray drying, hot gas flows through the system to drive evaporation. The gas temperature can be up to 120°C, which could cause degradation of sensitive biologic products. However, evaporative cooling occurs rapidly in the system to reduce the product temperature, and the heat exposure is very short. These systems have been applied successfully to biological production.

Another system uses a mesh nebulizer to create spherical droplets in a dry nitrogen laminar flow (Figure 10c). In addition, dry nitrogen is provided in counter-current flow along a membrane to remove moisture from the gas. The system runs at ambient temperatures to limit heat stress on sensitive products. Proof-of-concept of the technology has been demonstrated for mRNA-LNP products with excellent results for maintaining encapsulation efficiency, particle size, and distribution, and in vitro and in vivo mRNA activity (Gidner, 2024; Ziccum, 2023b).

Spray freeze drying

Otherwise known as a form of “bulk freeze drying,” spray freeze drying entails droplet freezing in a tower containing cold gas, followed by vacuum sublimation in a drying chamber. Two main competitors currently exist in this space and have been the subject of a number of recent reviews (Adali et al., 2020; Farinha et al., 2023; Langford et al., 2018). The first technology is based on droplet freezing in a tower followed by vacuum drying of the frozen pellets in a unique rotating drum (Luy et al., 2023). The technology has been successfully scaled up with product yields of up to 97% (Luy et al., 2023). The second technology is a continuous spray freeze dryer, shown in Figure 10d (IMA, 2016). Droplets are generated from the spray nozzle and frozen as they drop through a liquid nitrogen-cooled column, which leads to the formation of small diameter spherical particles. The frozen spheres are moved by conveyors through different levels of pressure and temperature to achieve a low-moisture product. The final product is filled directly into an appropriate container, which is sealed in an aseptic manner to be used as a feed for a powder filling system.

7 ADVANCES IN CHARACTERIZATION OF VACCINES

Recent introduction of new vaccine technology platforms (i.e., mRNA and viral vectored vaccines) has prompted rapid advances in analytical technologies to support vaccine development and quality analysis.

7.1 Biophysical and analytical technologies

Innovative improvements in the resolution and sensitivity of existing biophysical and analytical technologies are allowing more stringent characterization of purity and heterogeneity of vaccines, including inactivated or attenuated viruses, recombinant protein subunits and VLPs, and polysaccharide conjugates. These advancements have contributed to higher reliability in quantitative evaluation of critical quality attributes (CQAs) of vaccines and greater confidence in lot release.

The use of mass spectrometry (MS) for vaccine development has been reviewed previously (Sharma et al., 2020), with specific examples of applications to recombinant protein antigens and glycoconjugates. Capillary electrophoresis (CE) coupled with time-of-flight (TOF)-MS has been used for characterization of Mycobacterium tuberculosis antigens and their glycoconjugates (Tengattini et al., 2016). Charge detection MS (CDMS), in which mass/charge and charge are detected simultaneously, has been used to quantify purity and heterogeneity in multiple classes of vaccines and multivalent vaccines (Miller et al., 2021).

The remarkable success of mRNA vaccine technology has prompted improvements in resolution and sensitivity as well as innovative applications of analytical methods such as capillary gel electrophoresis (CGE) and ion-paired RP-HPLC (IP-RP-HPLC) to evaluate integrity and heterogeneity of vaccine constructs (Azarani, 2001; Lu et al., 2020; Raffaele et al., 2022; Yang & Chang, 2012). Furthermore, analysis of multivalent vaccines with adequate resolution has been made more feasible. A combination of IP-RP-HPLC and MS was employed to detect and characterize adducts of lipid fragments with mRNA nucleosides in mRNA-LNP formulations of an mRNA COVID-19 vaccine (Packer et al., 2021). The formation of these adducts was shown to reduce vaccine potency as a function of storage time and temperature.

dsRNA is a reactogenic impurity in mRNA products and its level must be kept to as close to zero as possible. While analytical and immunochemical assays are available, improved sensitivity and throughput would facilitate process and product monitoring. Towards this end, a microfluidics-adopted CE method has been developed that utilizes differential fluorescence intensities of two different fluorophores for single-stranded RNA and dsRNA (Coll De Peña et al., 2023).

Improved resolution and data analysis for dynamic and multiangle light scattering (DLS and MALS) techniques have allowed precise determinations of size heterogeneity and aggregation of protein antigens, viruses and viral vectors, and LNPs. Size binning of mRNA-LNPs has allowed researchers to determine particle size dependence of in vivo immunogenicity of vaccine formulations (Hassett et al., 2021). A recent innovation, a light scattering-based mass photometry technique, can determine the sizes of single particles and, thereby, heterogeneity in an ensemble of particles (Young et al., 2018). This technology has been used to study interactions between SARS-CoV-2 spike protein and ACE2 receptor (Burnap & Struwe, 2022).

New single-particle detection methods allow observation of formulated vaccine products such as mRNA-LNPs in solution and offer a dynamic view as opposed to static electron microscopy (EM) images. A recent innovation in imaging single particles in solution is convex lens-induced confinement or CLIC (Kamanzi et al., 2021). CLIC is a promising technology for visualization and quantitative analysis of mRNA-LNP structure, sizes, and heterogeneity of mRNA loading in mRNA vaccine candidates. Parallel advancements in light scattering technologies have made single-particle size analysis possible, as opposed to traditional average ensemble measurements (Guerra et al., 2019). Improvements in resolution as well as simulation and machine learning (ML)-based data analysis techniques have played key roles (Hussain et al., 2020).

The recently developed tunable resistive pulse sensor (TRPS) method is being used for single-particle analysis, including size distribution and zeta potential, of biological nanoparticles such as LNPs (Vogel et al., 2017). The impact of higher precision in size binning has been demonstrated by testing mRNA-LNPs of different sizes on in vivo immunogenicity (Hassett et al., 2021). Small (<80 nm) mRNA-LNPs appeared to elicit lower immunogenicity in mice compared with larger (100–140 nm) mRNA-LNPs, although the impact on immunogenicity in humans is unknown. While SEC-MALS has been an established technology for two decades, field flow fractionation (FFF)-MALS offers the advantage of size-based separation without requiring interaction with a solid matrix. This technology was used to correlate aggregation-dependent changes in potency of a subunit fusion protein-based respiratory syncytial virus (RSV) vaccine candidate and for LNPs containing encapsulated mRNA (Djagbare et al., 2018; Mildner et al., 2021).

BioSAXS was recently introduced as a combination of small angle X-ray scattering (SAXS) and SEC-HPLC. Similar to DLS, SAXS is a solution scattering technique used for low-resolution structural characterization and modeling of proteins and protein complexes (Mertens & Svergun, 2010; Mertens & Svergun, 2017). BioSAXS has been used to determine radius of gyration for a sa-mRNA vaccine (IMP-1), with high agreement with hydrodynamic radius determination by DLS (Myatt et al., 2023).

The remarkable improvement in resolution of cryo-EM has made it possible to visualize mRNA-filled LNPs and determine the numbers of mRNA constructs carried by each nanoparticle (Kloczewiak et al., 2022). These measurements have revealed heterogeneity in mRNA loading into LNPs and the presence of empty LNPs in a given population (Brader et al., 2021; Kloczewiak et al., 2022). Furthermore, in some mRNA-LNP particles, mRNA was observed to form bleb-shaped structures that were detached from the lipids. High-resolution cryo-EM has also been used to determine structures of wild type and a mutant S-protein trimer to support development of a subunit COVID-19 vaccine (Ma et al., 2021).

Technological innovations in label-free assays have contributed to several aspects of vaccine development ranging from antigen selection to functional characterization of DS and formulated DPs. For example, bio-layer interferometry (BLI) has offered additional advantages over surface plasmon resonance (SPR), though both techniques offer real-time measurements of association and dissociation kinetics between macromolecules and ligands (Murali et al., 2022). BLI was used to evaluate binding kinetics and affinities of the SARS-CoV-2 S-protein RBD trimer with the ACE2 receptor and with a neutralizing human monoclonal antibody (Vogel et al., 2021).

7.2 Immunoassays

In addition to continued use of the popular enzyme-linked immunosorbent assay (ELISA) technique, a few other technologies have been developed with a view towards improving sensitivity of detection, precision, speed, and throughput. Examples of these platforms include AlphaLisa, VaxArray, and Luminex, which offer relative ease of multiplexing (Bandiera et al., 2019). A comprehensive comparison of ELISA and Luminex immunoassays for detection and quantification of SARS-CoV-2 antibodies has been reported (Santano et al., 2021). Multiplexed VaxArray has been used for characterization of measles and rubella vaccine (Gillis et al., 2021) and influenza vaccines (Byrne-Nash et al., 2018). VaxArray has recently been applied to simultaneously and rapidly detect and quantify 23 polysaccharide serotypes commonly included in pneumoconjugate vaccines (Hu et al., 2022). In an innovative variation of antigen-antibody based immunoassays, VaxArray has been used to identify and characterize mRNA constructs employing complementary oligonucleotides as capture reagents (Gao et al., 2022). AlphaLisa is a chemiluminescence-based high-throughput immunoassay (Beaudet et al., 2008) that has been used to detect the SARS-CoV-2 nucleocapsid protein (Gorshkov et al., 2020). These immunoassays can be used in cell cultures, for example, to evaluate protein expression after transfection of cells with mRNA vaccines. As the vaccine function of mRNA, pDNA, and viral vector-carried transgenes depends on intracellular expression of the encoded protein antigens, immunoassays provide a valuable quantitative characterization for these classes of vaccines as well.

The newer immunoassay technologies described above are increasingly being used as diagnostic tools for detection of antibodies in sera induced by viral antigens. They also offer a platform for comparing immune response levels induced by infection and by prophylactic vaccination in uninfected populations. Antibodies conjugated with fluorophores of high quantum yield have facilitated sensitive detection and visualization of these functional events in cell-based assays such as flow cytometry, which utilizes a combination of laser-induced light scattering and fluorescence detection (McKinnon, 2018). Advances in fluorescence microscopy have also allowed visualization of translation of mRNA sequences into their corresponding proteins (Chao et al., 2012).

7.3 Sequencing and PCR technologies

Advances in sequencing technology over the past decade have had a major impact on quality testing of vaccines in several ways. The significantly higher accuracy of next-generation sequencing (NGS) over Sanger sequencing has made NGS an invaluable tool for testing the identity and genetic stability of nucleic acid-based vaccines (Dhiman et al., 2009; Luciani et al., 2012). NGS was used as the lot release identity test for one of the approved COVID-19 mRNA vaccines (European Medicines Agency, 2021a).

One regulatory requirement for releasing vaccine lots is the absence of adventitious viruses with the challenge that these may be unknown or emergent. A primary source of adventitious virus is the cell substrates used in production, including mammalian, insect, and bacterial cell lines (Khan et al., 2018). Master and working cell banks (MCB, WCB) should be demonstrated to be free of such adventitious viral agents (AVA). Historically, in vivo (in five species of animals) and cell-based in vitro infectivity tests have been used and accepted to ensure viral safety in vaccines. However, in the past decade, substantial momentum has been building to replace in vivo tests with NGS. Many of these initiatives have been driven by regulatory agencies in Europe and the United States, and by WHO in partnership with the biopharmaceutical industry and expert advisors (European Medicines Agency, 2022; US Food and Drug Administration, 2023; World Health Organization and Expert Committee on Biological Standardization, 2020). Recent advances in the sensitivity and breadth of virus detection by NGS, improvements in bioinformatics and reference databases, the need for speedy release of vaccine lots as highlighted by the COVID-19 pandemic, and the push to reduce the use of animals have collectively driven this progress (Charlebois et al., 2020; European Medicines Agency, 2022; Khan, 2021). A recommended list of reference standard viruses and a stable cell line for validating NGS have been published by WHO and FDA, respectively (US Food and Drug Administration, 2023; World Health Organization and Expert Committee on Biological Standardization, 2020).

In the past decade, significant progress has been made towards replacement of the monkey neurovirulence test with NGS, as a safety test required in the manufacturing process of live attenuated oral polio virus vaccine and for viruses other than polio that can exhibit neurovirulence (Neverov & Chumakov, 2010; World Health Organization, 2019). Another in vitro method, mutational analysis by PCR and restriction enzyme cleavage (MAPREC), is capable of detecting and quantifying frequencies of revertant mutants that are primarily responsible for neurovirulence (Rubin, 2011; World Health Organization and Expert Committee on Biological Standardization, 1996). Comparative analysis of NGS with MAPREC has produced identical results. However, the capacity of NGS for whole-genome sequencing and detection of low-frequency mutations has made it a more attractive tool for ensuring viral safety and consistency in manufacturing of LAV vaccines (Charlton et al., 2020; Neverov & Chumakov, 2010). Validation of NGS as a lot release safety test for a novel type 2 oral polio vaccine has been reported (Konz et al., 2021). As well as the cost and timing advantages, NGS can replace traditional testing based on animals with obvious ethical benefits.

The introduction of droplet digital PCR (ddPCR) has been a highly significant advancement in PCR technology over the past decade (Kojabad et al., 2021; Pinheiro et al., 2012). ddPCR is a high-throughput and quantitative method that can be applied to accurately determine genome copy numbers of DNA and RNA viruses and vaccines. The ddPCR assay has the additional advantage of not needing a standard curve or reference standards for quantification. This technology is expected to have wide applications in identity and genetic stability testing for different classes of vaccines including pDNA, mRNA, viral vectored, and LAV vaccines. ddPCR was validated for quantitation of RNA copy numbers in development of a live attenuated Rift Valley Fever (an RNA virus) vaccine and to detect a reversion mutant (Lokugamage & Ikegami, 2017).

While advances in analytical characterization have led to high purity and structural integrity of vaccine antigens, host-dependent biological factors often make it difficult to predict if the desired immune response maybe elicited in vivo. Nevertheless, high-resolution structural analysis, complemented by immunoassays and mutational scanning, has been used to identify antigenic sites that are critical to generation of virus-neutralizing antibodies (Francino-Urdaniz & Whitehead, 2021; Zhao et al., 2015). Such correlations are more challenging when the primary mechanism of vaccine efficacy is cell-mediated immunity (CMI). However, a recent example of MS-aided identification of HLA-presented T-cell epitopes and design of a CMI-eliciting mRNA vaccine has been reported (Arieta et al., 2023).

8 ANALYTICAL RELEASE ASSAYS

Analytical methods are a critical foundation piece for vaccines and require continuous monitoring and updating to provide the best information to guide the development and manufacture of these products. Analytical quality by design (QbD) outlines a path for the development of robust analytical methods and several recent references discuss the topic in detail (Junker et al., 2015; Schweitzer, 2010; Verch et al., 2022). The concepts codified in analytical QbD are based on guidance documents including ICH Q14 (ICH and European Medicines Agency, 2022), which describes enhanced approaches for the development and lifecycle maintenance of robust methods. Applying the approaches outlined in those guidance documents to conventional methods can deliver significant gains, but there are inherent limitations when applying them to older technologies.

The speed at which the SARS-CoV-2 vaccines were developed and commercialized is encouraging for future pandemic responses, but more must be done to shorten the development time for all new prophylactic vaccines. Vaccine manufacturing processes are difficult to control and the end products resist facile definition and characterization. Investing in new and emerging analytical technologies will help programs generate critical information earlier to avoid stalling in later phases of clinical development and will lead to better product and process understanding. It is important to develop and establish precedents for the use of newer technologies that will minimize risk and allow for their widespread adoption. Within the analytical QbD schema, during the development of the analytical target profile (ATP), introduction of novel technologies should be given thoughtful consideration. Selected advances in analytical technologies for both product-independent and product-specific methods are discussed below.

8.1 Product-independent methods

Safety testing on vaccine DSs include microbiological testing such as sterility, bioburden, endotoxin, mycoplasma, and adventitious agent testing; many of which are described in detail by Pharmacopeial authorities (US Pharmacopeia [USP], European Pharmacopoeia [PhEur], Japanese Pharmacopoeia). These methods have the advantage of being well established and understood but can be very laborious and time consuming, in addition to raising ethical concerns about animal use. Recent guidance (European Commission, 2022; US Food and Drug Administration, 2020) has encouraged consideration of alternative approaches to traditional methods. From the FDA, “Analytical procedures different than those outlined in the USP, FDA guidance, or Code of Federal Regulations (CFR) may be acceptable under IND if sponsors provide adequate information on test specificity, sensitivity, and robustness.” Similarly, Annex 1 states, “The adoption of suitable alternative monitoring systems such as rapid methods should be considered by manufacturers to expedite the detection of microbiological contamination issues and to reduce the risk to product.”

The incorporation of new, rapid microbiological methods is gaining momentum, due in part to the response to the COVID-19 pandemic, as well as the needs of cell and gene therapy products, many of which have very short shelf lives. The limitations of the traditional growth-based sterility assessment methods due to sample size and growth conditions are well known. The BacT/Alert 3D system has joined the compendial 14-day growth-based sterility test as a validated sterility method (Jimenez et al., 2012; Paris, 2020); other tests, some based on growth and others not, are in various stages of development. PCR- and ELISA-based detection methods are now widely accepted for the detection of mycoplasma (Morris et al., 2021) and methods based on NGS are also being developed, as described in the Advances in Characterization of Vaccines section.

Adventitious agent testing requires methods that can detect a broad range of possible contaminants. Newer technologies such as laser force cytology are being explored to identify cytopathic effects and improve classical cell-based assays (Hayes et al., 2022). NGS technology holds significant promise not only for detection of microbial contaminants but also to replace the battery of in vivo, in vitro, and PCR tests that are currently used for detection of AVAs (Khan et al., 20182020).

8.2 Product-specific methods

In addition to compendial release assays, vaccines require testing for CQAs such as identity, strength, purity, and potency that are specific to the vaccine being developed. Methods that have been traditionally used for these attributes suffer from limitations that emerging technologies can, in many cases, mitigate.

Vaccine potency methods come in a wide variety of types to meet the expectation that they will provide information about the desired immune response in patients (Verch et al., 2018). In vivo potency assays are fraught with variability, irreproducibility, and long turn-around times inherent to animal-based assays. For this reason, as well as in considerations of animal welfare and cost, development of in vitro assays has become centrally important in characterization and lot release of vaccines. Understanding the mechanism of action (MOA) of the vaccine is still the key to developing the most relevant potency assay, but there are many more tools available to the development scientist than ever before. In cases where the exact nature of a protective immune response is not known, newer technologies are providing richer and more precise information about how a vaccine fundamentally works. In certain cases, the interaction of antigen with adjuvant to create the desired immune response still cannot be assessed by current methods and knowledge (Goyal et al., 2022).

The plaque assay has been a virological workhorse for live virus vaccines for decades, but scientific and technical advances made in recent years can be used as enhancements to and/or replacements for virus quantification methods like the traditional plaque and TCID50 methods. Enhancements in resolving power and quantitative accuracy of fluorescence imaging technology have contributed to replacement of traditional assays with automated foci counting (Rush et al., 2018). Visual identification of plaques or cytopathic effect (CPE) in a cellular monolayer can be replaced by PCR (Ranheim et al., 2006), laser force cytology (McCracken et al., 2022), and cellular electrical impedance, to name a few alternatives. In cases where the presence of an adjuvant inhibits the use of these technologies, microphysiological systems show promise as model systems for correlating vaccine potency to in vivo efficacy. Using these systems may also provide a way to evaluate older vaccines and the impact of process changes without the need to use pre-clinical animal models or clinical trials.

Protein subunit and VLP vaccines typically rely on antigenicity assays for in vitro potency. Assays based on technologies such as AlphaLisa, Fluorescence Energy Transfer (FRET), Luminex, and electrochemiluminescence offer advantages in speed, sensitivity, dynamic range, data density and multiplex capability compared to a standard ELISA method. In addition to protein-based vaccines, immunoassays are also being used for vaccines developed with newer technologies such as mRNA where the ultimate antigen is a protein. Examples include the human rabies vaccine, for which envelope glycoprotein-based ELISA and time-resolved immunofluorescence have been developed (Lin et al., 2017) and the whole cell pertussis vaccine ELISA, which has shown to be a promising replacement for the currently used in vivo Kendrick test (Viviani et al., 2022). Flow cytometry, presumably using fluorescently labeled anti-spike protein antibody, was employed for potency assays of the first approved COVID-19 mRNA vaccine (European Medicines Agency, 2021a2021b).

In the absence of specific immunoassays for potency measurements, structural and functional assays that correlate with potency may be developed as surrogates (Sanyal, 2022). Regulatory openness to this concept was indicated by a WHO Expert Committee for Biological Standardization report published as a guideline for mRNA vaccine development in 2021. Indeed, such a surrogate assay was accepted for characterization of vaccine potency of a bivalent mRNA COVID-19 vaccine (European Medicines Agency, 2021b).

As an example of advancement of an older technology for purity and integrity analysis, CE has evolved as a superior alternative to SDS-PAGE and immunoblotting techniques, providing greatly improved process and product understanding (Geurink et al., 2022). Similarly, MS shows promise to improve on the traditional ELISA-based methods for the detection of HCPs (Huang et al., 2021).

The development and release testing of the SARS CoV-2 vaccines have paved the way for future pandemic vaccines and also provided approaches that promise to reduce the development timeline for prophylactic vaccines.

9 ANALYTICS FOR PROCESS MONITORING

The recent COVID-19 pandemic has highlighted the need for real-time or rapid release of vaccine lots to minimize preventable deaths and suffering. Furthermore, failure of a lot to meet release specifications at the end of GMP manufacturing of a DS or DP can be extremely costly. Therefore, monitoring of the process through every step with in-line or at-line testing is important for ensuring high quality and speedy release of the final product. Every new vaccine in development typically has a well-defined quality target product profile (QTPP). The concept of bioprocess optimization to reach a target QTPP has been described with supporting analytical data for an mRNA DS process (van de Berg et al., 2021). This basic principle is relevant to implementation of QbD principles in developing processes for all biopharmaceuticals across all platforms.

Manufacturing process steps should be monitored for preservation of antigenic epitopes that are required for potency. The improved ability to measure potency in the presence of cell substrates has made this possible in many cases, thereby averting process failures and allowing optimization. For example, at-line monitoring of defective viral particles by “flow virometry,” combined with laser force cytology-based potency measurements, has been used to improve an LAV vaccine manufacturing process (McCracken et al., 2022; Ricci et al., 2021). Coupled with capacitance tools and high-content imaging to monitor cell density and peak infection in real time, these process analytical technology (PAT) tools may collectively offer potency assay surrogates for LAV vaccines (Yi et al., 2022).

The benefits of integration of PAT into bioprocesses have been reviewed recently (Gerzon et al., 2022). Raman, nuclear magnetic resonance (NMR), near and mid-infrared (IR) spectroscopy, and DLS are being increasingly used as PAT tools, as these biophysical measurements provide structural information, which relates to function (Graf et al., 2022; Liu et al., 2021). NMR spectroscopy has been valuable in optimizing the quality of capsular polysaccharides in purification process development for vaccine antigens against S. pneumoniae (Lee et al., 2020). NMR has been used to monitor tertiary structure of polysaccharides in the process of developing conjugate vaccines against Haemophilus influenza b and meningococcal subgroup A (Beri et al., 2019).

DLS and other light scattering techniques have been used for years in monitoring aggregation and polydispersity of subunit and VLP vaccines, and to optimize purification and formulation conditions to prevent aggregation-induced loss of potency (Djagbare et al., 2018; Shi et al., 2005). More recently, DLS has been used to select relatively monodisperse populations of optimally sized LNPs for mRNA-LNP vaccines. This information is helping to control microfluidics mixing parameters, especially flow rates, in the development of mRNA DP processes (Roces et al., 2020). PAT will continue to fill an important role as QbD-driven continuous manufacturing processes for mRNA DS and DP are further developed and optimized (Schmidt et al., 2022).

10 VACCINE MANUFACTURING REGULATORY LANDSCAPE

To encourage innovation in vaccine manufacturing, the U.S. Department of Health and Human Services released the Vaccines National Strategic Plan 2021–2025 (US Department of Health and Human Services, 2021) on January 19, 2021, which was updated from the 2010 National Vaccine Plan (discussed in Josefsberg & Buckland, 2012). In the 2021–2025 Vaccine Plan, several themes remain at the forefront, including vaccine supply and the pandemic response, improved quality testing procedures, global regulatory harmonization, and involvement of developing-country manufacturers in vaccine supply. The 2021–2025 Vaccine Plan has five primary goals, the first two of which involve improving vaccine production and safety: (1) Foster innovation in vaccine development, manufacturing, and related technologies; and (2) Maintain the highest possible levels of vaccine safety. The first goal is discussed throughout this review, but the second is driven by the evolution of the government regulatory approval process and innovations in industry process control systems, mainly current GMPs and QbD.

The regulatory approval process exists to ensure patient safety. The rapid response to the COVID-19 pandemic set a new standard for the speed in which the regulatory process could be completed (11 months) and left many wondering if we can reduce time-to-market for all vaccine candidates while maintaining the same assurance of patient safety (Agrawal et al., 2021). The accelerated timelines for the COVID-19 vaccines were driven by the level of industry and government investment risk (billions of dollars), global sharing of the genome sequence, previous coronavirus research, decades of development work on mRNA vaccines, regulatory acceptance of proceeding with drug development on risk, seamless clinical trials, and rapid review by regulatory agencies. Some of the factors could not be replicated in a non-crisis situation, but others could shorten the current discovery-to-market vaccine development timeline, which is 10–15 years (Anderson, 2022).

Pre-pandemic, some regulatory tools existed to lend agility, such as EUA by the FDA and Conditional Marketing Authorizations (CMA) by the EMA, as well as fast-track designations by the FDA and rolling data submissions by the EMA (Kalinke et al., 2022). During the pandemic, adjustments of standard practices accelerated approvals, including reliance on other agency inspections and reviews (reliance pathways), dynamic/rolling/real-time reviews, additional agency guidance, increased acceptance of digitalization and e-labeling technologies, reduced hard-copy documents, and loosened restrictions on individual batch release testing and local clinical data (Patel et al., 2023). Reliance pathways are a risk-based approach to regulatory submissions and involve recognizing the data, assessments, and/or decisions of other authorities and institutions. While this has always occurred at some level, the pandemic encouraged sharing of data to accelerate vaccine development. The International Coalition of Medicines Regulatory Authorities (ICMRA) COVID-19 Working Group created a list of all COVID-19 master protocols to inform policy discussions and collected knowledge regarding regulatory agilities (International Coalition of Medicines Regulatory Authorities, 2022). ICMRA is also collaborating with the International Council for Harmonisation (ICH), the International Pharmaceutical Regulators Programme (IPRP), and the Pharmaceutical Inspection Co-operation Scheme (PIC/S) to develop a Pharmaceutical Quality Knowledge Management (PQKM) capability (International Coalition of Medicines Regulatory Authorities, 2023). Collaborations among public and private entities were also created, including the Access to COVID-19 Tools (ACT) Accelerator to “accelerate development, production, and equitable access to COVID-19 tests, treatments, and vaccines” (World Health Organization, 2022) and the COVID-19 Evidence Accelerator (EA) to share Real-World Evidence (RWE) (Grossmann et al., 2022). Some authorities are also using reliance pathways to abridge the evaluation process and speed up the review process in their countries. The acceptance of Certificates of Pharmaceutical Product (CPP) by some national regulatory authorities has reduced timelines, while an EMA pilot project called “OPEN” enables international agency participation in their scientific evaluation process (confidentiality arrangements required), which promotes transparency and the tackling of common challenges (Chisholm & Critchley, 2022).

Risk-based approaches, such as reliance pathways, can also be used to accelerate post-approval changes (PACs). Post-approval change management protocols (PACMPs) shared between manufacturers and regulatory agencies allow rapid implementation of changes, while grace periods and/or waivers allow PACs to be rapidly approved and accessible to patients (IFPMA, 2022). Reliance pathways are particularly important in low-to-middle-income countries (LMIC), where agencies might not have the maturity and capacity to deal with health emergencies.

A major paradigm shift in the review process is the concept of breaking down submissions into smaller packages for segmented review, rather than submitting the complete dossier at the end of a campaign. The data can be presented upon agreed milestones or in real-time. In some cases, these intermittent data packages can be reviewed simultaneously to the next step being initiated, sometimes referred to as a phased or rolling regulatory review. One example is the European Federation of Pharmaceutical Industries and Associations (EFPIA) and their concept of Dynamic Regulatory Assessment (DRA) that involves submission of data packets as they become available, well in advance of the full Marketing Authorization Application (MAA) (Herrero-Martinez et al., 2022; Todd et al., 2023).

During the pandemic, the EMA and FDA offered additional guidance and continuous dialogs with manufacturers as a method of accelerating both development and review timelines. Examples include the EMA's PRIority MEdicines (PRIME) scheme, which issued supplemental Q&A documents discussing new guidance and early development plans, and the target product profiles and detail requirements that were provided by the FDA and EMA. This was information that was not provided before the pandemic (Kalinke et al., 2022).

Digitalization increased continuity of regulatory activities and improved interconnectivity of all stakeholders during the COVID-19 pandemic, especially due to virtual meetings and online platforms. The use of e-documents, e-signatures, and e-submissions advanced sustainability through a decreased need for travel and paper. The use of structured datasets that are both human and machine readable is rapidly advancing, which enables efficiencies from automation and cloud-based platforms (Khalil et al., 2023). Generative AI (Gen AI) will also inevitably play a role in the regulatory approval process. A recent McKinsey & Company report proposes that Gen AI-enabled intelligence engines will significantly improve efficiency in writing submission documents and responding to Health Authority Queries (HAQ) by predicting potential HAQ patterns, quickly drafting responses, and informing submission strategies that could reduce HAQ follow-ups (Viswa et al., 2024). The report also suggests that Gen AI can serve as a major submission content writer, especially for drafting clinical-study reports based on the protocols, data, and statistical analysis plans, as well as for creating tables and figures. Medical writers would then be freed up to concentrate on sections that require more complex clinical interpretations, improving collaborations across teams, reducing costs, and limiting quality issues.

Harmonization of the types of data and the methods of sharing will be critical to improve timelines. Cloud-based systems have the potential to streamline parallel agency review and reliance. Forums such as ICH and ICMRA can help facilitate this and work with stakeholders to avoid the emergence of fragmented or conflicting systems (Stewart et al., 2021). Additional agilities used during the pandemic could applied to future vaccine programs, including innovative clinical trial strategies (e.g., platform approaches to simultaneously test multiple candidates in the clinic), using historical data for predictive analyses, leveraging real-world data or placebo/standard-of-care data from other trials, and virtual inspections, which have gained acceptance and can reduce timelines due to delays associated with scheduling and travel (Stewart et al., 2021). Other options include loosening restrictions on individual batch release testing (e.g., exemption of duplicative lot release/local test requirements), exemption of requirements for local clinical data, and exemptions on country-specific labeling language/artwork/details, including acceptance of QR codes on packs (e-labeling) in place of physical leaflets. Each of these agilities may not be possible in every case; for example, the use of e-labeling in LMICs may not be feasible due to a lack of technology. Overall, these types of transformations in the regulatory mindset have the potential to revolutionize historical ways of working, which will improve efficiency and timelines without compromising quality, safety, and efficacy.

In addition to the regulatory innovations and agilities implemented over the last decade, manufacturer process control strategies have advanced. As discussed at the beginning of this section, the first goal of the 2021–2025 Vaccine Plan is to foster innovation in vaccine development, manufacturing, and related technologies, which can be accomplished through two main pathways. At a high level, CMC and cGMP are both process control strategies (US Food and Drug Administration, 2022) with the same goal of ensuring that the product consistently meets the established quality requirements, but the difference is in their scopes. CMC information is specific to a product, whereas cGMP is a high-level framework for the entire operation of a manufacturer. CMC requires meticulous attention to detail regarding product quality and involves defining CQAs and their associated critical process parameters (CPPs). Once these are identified, a set of activities can be defined for process and product specifications to ensure batch-to-batch consistency. The CMC plan will demonstrate total control over product components, equipment, manufacturing conditions, records, and personnel through every stage of the product lifecycle. On the other hand, cGMP is a high-level set of quality assurance activities and guidelines that define product planning and development, as well as facility design and operation. A cGMP plan includes maintenance of documentation, equipment, training, and facilities, of which traceability is an essential component. CMC and cGMP systems often intersect and are not mutually exclusive, with both being critical to the success of a product. Synchronizing CMC and cGMP plans can be complex and the approach may vary from product to product and manufacturer to manufacturer.

The CMC and cGMP space has benefitted from innovations over the last decade, for example, AI and ML, both of which can automate tasks such as data analysis, process optimization, and quality control (Castellanos et al., 2023). Big data analytics are powerful tools that can be used to identify trends and improve decision-making. 3D printing gives unprecedented ability to quickly create prototypes of products and equipment, speeding up the development process. Even technologies such as virtual reality (VR) and augmented reality (AR) are being used to train staff and simulate processes. One often overlooked advancement is cybersecurity, which is critical in protecting IP and data from cyberattacks.

Improvements in planning and execution have reduced timelines for technology transfer to emerging manufacturers in developing countries, which contributes to the global availability and affordability of important vaccines, such as the Hib conjugate vaccine (Hamidi et al., 2014). Before the pandemic, a typical vaccine technology transfer took 27–29 months, so reducing this timeline was critical to supplying COVID-19 vaccines rapidly to a global population (O'Sullivan et al., 2020). The control-tower approach to technology transfer, which involves a defined sequence of events and meetings that ensures effective communication, especially rapid escalation of issues to leadership, is becoming more popular. Other agile ways of working include nonhierarchical efficient decision making, early risk-tolerant investment, fast capacity ramp-up, flexibility to pivot upon negative outcomes, ensuring a high level of expertise among interdisciplinary team members, using a common vocabulary between functions and companies, and envisioning quality requirements as enablers rather than barriers. Collaboration and partnerships between industry, organizations, regulatory agencies, and government political organizations have become imperative (Fu et al., 2022). For international technology transfer to be successful, existing capabilities must be mobilized, new ones built, manufacturing and regulatory processes coordinated, and resources allocated—activities which need to be enabled politically (Fonseca et al., 2023).

Another change in mindset in CMC and cGMP is the concept of centralized versus distributed models of manufacturing. Traditionally, vaccines are manufactured in a centralized location, which is economically driven due to the high cost of tech transfer to additional sites (Sell et al., 2021). Distributed manufacturing, on the other hand, produces products closer to the end user, which is useful during a pandemic. The ability to establish a process at multiple sites has become more feasible due to process platforms, but the extensive regulatory requirements to approve each site remains a hurdle, as well as for local testing and quality assurance. Although distributed manufacturing cannot offer economies of scale similar to centralized production, advances in flexible manufacturing technologies and regulatory processes could make the approach more feasible in the future (Gomez & Robinson, 2018).

QbD involves creation of a design space that is a multidimensional combination of CPPs and CQAs that have been demonstrated to provide assurance of quality (Calcott 2013; Oliveira, 2022; Yu et al., 2014). An iterative approach to creating a bioprocess QbD space (Figure 11) was developed for an RNA vaccine synthesis bioreactor (van de Berg et al., 2021). In the case of a pandemic where a rapid response is critical, the existence of a disease-agnostic platform process with an established QTPP-driven design space will enable a team to generate material as soon as the raw materials are available. In theory, process changes within a robust design space do not require additional regulatory submissions, creating processes with more flexibility. The QbD framework can be used to support both the development and the operation of production processes and can follow an iterative development cycle, which will provide continuous improvement throughout the lifecycle (Haas et al., 2014; Schlindwein & Gibson, 2018).

Details are in the caption following the image
In-process measurements of CQAs can drive optimization of critical process parameters (CPP) and the normal operating range (NOR) for a process resulting in desired yield and quality target product profile (QTPP) for mRNA drug substances (DS). Image reproduced without changes from van de Berg et al. (2021) under the terms of the Creative Commons CC BY License (http://creativecommons.org/licenses/by/4.0/).

11 CONCLUSION

Significant advances in vaccine development capabilities have been established in the past several years, reflected in the accelerated response to the COVID-19 pandemic. mRNA and replicating-deficient adenovirus have been firmly established as major new vaccine platforms that can be developed relatively quickly and scaled to make billions of doses at a reasonable cost. Progress exceeded all expectations with the rapid development of safe and effective vaccines against COVID-19. Other vaccine platforms such as protein antigens and LAV vaccines continue to be important and related technology improvements have been highlighted in this review. We anticipate further spectacular progress in vaccine development based on advances in biology and understanding of disease.

Vaccine manufacturing innovations in continuous manufacturing, single-use processes, cell-free technologies, process analytic technologies, and release assays are reducing costs and improving efficiencies. Ongoing innovations are also replacing legacy lot release tests to remove the need for animal testing. The detailed summary of advances in vaccine analytics in this review is deliberate; this area has seen an enormous technological leap over the past decade that has impacted every aspect of vaccine manufacturing, resulting in improvements in consistency, productivity, product quality, speed of product testing, and scale of operation (Buckland et al., 2022).

In recent years, significant efforts have been made in designing and developing vaccines that aim to elicit T-cell mediated immune responses. This emerging area promises to deliver effective prophylactic and therapeutic vaccines against infectious diseases such as HPV, HIV, and cancer (Cohen et al., 2023; Kartikasari et al., 2018; Nathan et al., 2021; Panagioti et al., 2018; Yang et al., 2017). Efforts are also underway to deliver mRNA vaccines that offer T-cell based immunity against broad subtypes of influenza and SARS-CoV-2, and that prolong duration of protection compared to current mRNA-based COVID-19 vaccines (Arieta et al., 2023; van de Ven et al., 2022). Peptide-based T-cell targeted COVID-19 vaccines are also in early clinical development (ClinicalTrials.gov2023) and analytical methods and assays to support development and immunogenicity assessment are progressing (Altosole et al., 2023; Cimen Bozkus et al., 2021; Phetsouphanh et al., 2015).

The range of opportunities to advance the field has further increased as vaccines and vaccination evolve beyond prophylactic applications to therapeutics, made possible by rapid and detailed analysis of target diseases. We predict that progress will continue at a rapid pace into the next decade, resulting in new vaccines that are made rapidly available worldwide at a reasonable cost, with innovations in design, technology, and analytics.

ACKNOWLEDGMENTS

Stacey Tobin, PhD, provided editorial support in the preparation of this manuscript, with compensation from NIIMBL. Preparation of this article was supported by funding from the U.S. Department of Commerce, National Institute of Standards and Technology, through Cooperative Agreements 70NANB17H002, 70NANB21H086, and 70NANB21H085. This publication is based on research funded in part by the Bill & Melinda gates Foundation under grant #INV-038807. The findings and conclusions contained within are those of the authors and do not necessarily reflect positions or policies of the Bill & Melinda Gates Foundation. 

    CONFLICT OF INTEREST STATEMENT

    Barry Buckland is a board member and owns shares in Dyadic. Gautam Sanyal is a consultant and subject matter expert for the Bill & Melinda Gates Foundation. Steffi Pluschkell owns stock in Pfizer Inc. All other authors declare no conflicts of interest.

    DATA AVAILABILITY STATEMENT

    Data sharing not applicable to this article as no datasets were generated or analyzed during the current study.