Immunity to influenza and correlates of protection

Several excellent studies and reviews have described and discussed immunity to influenza infection and vaccination, and only points pertaining to the evaluation of vaccines are presented here. Although all arms of the immune system are triggered by influenza virus, primary protective immunity in humans is thought to be mediated predominantly by B cell responses and the resultant antibodies. These comprise mainly the airway mucosal immunoglobulin A (IgA) but also the immunoglobulin G (IgG) transduced from the serum to the respiratory tract.⁸ It has been observed that specific B cells appear in blood as early as 7 to 10 d after the onset of symptoms, and neutralizing antibodies directed to the hemagglutinin (HA), and to a lesser extent to the neuraminidase (NA) or the matrix (M) protein, mediate protection against infection and disease.⁸ Indeed, the level of hemagglutination inhibition (HI) antibodies has been used as a surrogate measure of protection against influenza when assessing vaccine immunogenicity and efficacy. However, there is substantial evidence suggesting that broadly neutralizing non-HI antibodies may also play an important role.^8,9

As far as T lymphocyte responses are concerned, several studies have shown that human CD4⁺ and CD8⁺ T cell responses typically cross-react to multiple influenza strains and subtypes, and mainly target internal components of the virus,^10-12 and that both CD4⁺ and CD8⁺ T cells may independently protect humans from infection or disease.^13,14 Evidence from elderly subjects, in whom the antibody response is quantitatively poor compared to adults,¹⁵ suggests that CD4⁺ and CD8⁺ T cell responses may correlate better with protection.^16-18 Studies focusing on identifying early predictors of correlates of protection following infection or vaccination have identified an array of other molecular signatures.¹⁹ While these markers correlate with later induction of functional antibodies, standardized regulatory criteria are yet to be defined for any of the markers or assays for evaluating cell-mediated immunity.^20,21 In any case, while it is intuitive that T cell responses against conserved regions of the genome would be expected to be boosted every year, it is known that prior immunity is generally insufficient to protect against future infections, and that annual seasonal vaccines are required against even the same subtype. These findings argue against extensive correlation between T cell responses and protection. On the other hand, the relative contribution of antibody- versus cell-mediated immunity to protection following vaccination may differ significantly from infection-induced immunity.

For any infectious disease, the correlates of vaccine-mediated protection are likely to be similar to those in subjects who suffer from mild to moderate illness and recover, although the quality, quantity and the breadth of immune response may be much wider in the case of infection due to the widespread replication of the virus in the host. Indeed, adaptive immune memory to influenza, in terms of humoral and poly-functional effector or memory T cell responses, has been found to be similar between those who were vaccinated and those who experienced a mild form of the disease.²² However, the type and extent of immune response depends on the type of vaccine being administered, whether it has been adjuvanted or not, the route of administration, as well as the age of the vaccinee.²³ The inactivated vaccines administered by the intramuscular route induce robust secondary recall IgG, good CD4⁺ T cell responses with the ability to help B cells, and weak CD8⁺ T cell as well as mucosal IgA responses, whereas the intranasally administered live attenuated vaccines (LAV) produce good mucosal IgA, relatively lower IgG, and pronounced and diverse CD4⁺ and CD8⁺ T cell as well as type I interferon (IFN) responses.^24,25 Specific B cell and polyclonal antibody responses are significantly higher with inactivated compared to LAV vaccines, whereas IgA, nucleoprotein-specific and cross-reactive B cell responses are better with LAV.²⁶ Efficacy of inactivated vaccines correlates well with HI antibody titres, which in turn may be predicted by CD4⁺ T cell responses,^27-29 whereas using HI titres as a yardstick may underestimate the efficacy of LAV because of weaker antibody responses.^30-32 On the other hand, LAV efficacy may be better correlated with neutralizing antibody titres, the frequency of IgG⁺ B cells, virus-specific airway IgA, or a combination of these immune parameters.^31-34

Current regulatory guidelines on influenza vaccine immunogenicity and efficacy assessments in humans are based solely on antibody responses. Although it might be intuitive to assess respiratory mucosal IgA, serum IgG levels have been shown to correlate with protection, and form the basis for vaccine evaluation. The most commonly employed serological assay evaluates the ability of antibodies to inhibit hemagglutination of chicken or human type ‘O’ erythrocytes. Another acceptable assay is the single radial hemolysis (SRH) assay, while the virus neutralization assay has recently been put forward as criteria for regulatory acceptance. A hemagglutination inhibition (HI) titer of ≥40 or hemolytic zone area of ≥25 mm² in SRH assay are considered as protective titres. Titres of antibodies are converted to seroconversion (SRC), seroprotection (SRP), and geometric mean titres (GMT). SRC is defined as a pre-vaccination (baseline) HI titer of ≤10 and a post-vaccination titer of ≥40 or at least a four-fold increase if baseline titer is ≥10. SRP is defined by the HI titer of at least 40, and GMT is defined as the arithmetic mean of the logarithm of the last positive dilution of serum. Geometric mean ratio (GMR) is calculated as the fold increase in GMT from baseline to post-vaccination time-point. A vaccine is deemed to meet regulatory criteria when SRC, SRP and GMR are >70%, >40 %, and 2.5 for subjects of 18–60 y of age, and >60%, >30% and 2.0 for the elderly, respectively. The proportion of subjects showing protective titres constitutes the efficacy of the vaccine, whereas effectiveness is an estimate of protection against the occurrence of disease.

Vaccine production using chicken embryos: Eggs rule the roost

Based on a method developed a century ago, influenza vaccines have been produced since the 1940s in embryonated chicken eggs (ECE). High titers can be obtained by growing influenza virus in ECE, and extensive experience in large-scale production has led to the streamlining and automation of a highly standardized process. There is also extensive safety data available as billions of doses of ECE-produced influenza vaccine have been administered to humans. However, besides being labor intensive, ECE-derived influenza vaccines have several drawbacks. First and foremost, the dependency on eggs is a matter of concern. Because 1 to 2 ECE are required for the production of each human dose of influenza vaccine, the method requires the availability of a large number of eggs simultaneously or within a short window of time, necessitating considerable planning (up to a year in advance) to produce sufficient numbers of ECE. In addition, the eggs need to be set synchronously since inoculations need to be carried out 10–12 d after initiation of incubation. More importantly, the eggs need to be from specific pathogen-free flocks or at least certified as ‘clean’ in order to avoid adventitious agents. This could be an issue especially in developing countries since the vaccine manufacturer will have to rely on the quality control of the supplier of the ECE. Secondly, some virus strains, especially the recent H3N2 strains, do not grow well in ECE, and others such as the highly pathogenic avian influenza strains, viz., H5N1, could be lethal to embryos, resulting in low titres.^35-37 Indeed, this is one of the reasons why seasonal influenza vaccine strains are generated through reassortment with a high yielding strain such as A/PR/8/34. Third, occasional breakdown in sterility during downstream processing could lead to rejection of large volume of vaccine bulk, leading to a need to revisit the long process of planning and execution, or to the lack of availability of vaccine when needed. Fourth, influenza viruses appear to mutate more frequently around the receptor binding site and be selected when passaged in ECE, compared to passage in cells cultured in vitro.^38-43 This could potentially affect vaccine efficacy, as has been seen in animal models.^44,45 Moreover, egg-based antigens may elicit weaker immunogenicity compared to cell culture-derived antigens.^46,47 Further, there have been reports of narrower reactivity when ECE-derived antigens are used in immunogenicity assays as compared cell-derived antigens.^46,48 Fifth, despite an extensive purification process, residual allergenicity of egg proteins is a serious concern. Finally, ECE-based vaccine production simply cannot meet global demand in case of widespread epidemics and pandemics, as current production capacities can only deliver about half a billion doses every year. A large increase in ECE production is probably not possible, in part because of eggs being a major human proteinaceous food source.

Vaccine production using cultured cells: ‘tis time to roust

In the last 2 decades, cell-culture based systems have been advocated because of the drawbacks with ECE-derived vaccine technology. Vaccine production in cultured cells has several advantages. First, cell lines can be extensively characterized and stored for future use without the need for repeated full range testing, and cell culture avoids dependency on supply and quality control of a raw material such as ECE. Second, certain viruses grow better in cells, avoiding the down-time required for the generation of high growth reassortants. Alternatively, high growth reassortants can be directly generated in cells.^49,50 Third, a much more standardized and controlled process can be set up to track vaccine manufacturing. Fourth, scalability is better with cell culture than with egg-based production platforms. Fifth, allergies to egg proteins can be avoided. Sixth, virus propagated in mammalian cell systems has been shown to be structurally or antigenically more similar or identical to the field virus as compared to that grown in ECE.^{37,41,42,45,47,51-54} Furthermore, immune responses elicited by mammalian cell-derived vaccines have been shown to be more cross-reactive than responses produced by ECE-derived vaccines,^46,55 although protective efficacy may not be affected.⁴⁶ Finally, the same facilities can be used for the production of other vaccines when not being used for the production of influenza vaccine for extended periods.

Vero cell-derived influenza vaccines: Here we come to boost

Vero (African monkey kidney epithelial) cells have been used for human vaccine production for nearly 40 years, and were the only World Health Organization approved continuous cell line substrates for some time. Beginning with polio, several vaccines have been produced or developed using Vero cells.⁵⁶ One of the advantages of Vero cells is that they do not produce IFN, thus allowing unhindered replication of viruses.⁵⁷ However, Vero cells were only recently shown to be suitable for influenza virus isolation and growth.^54,58 This observation was immediately followed by the adaptation of Vero cells to produce high titres of influenza virus, for vaccine production at a commercial scale,^57,58 and for the generation of reassortant vaccine strains using the reverse genetics technology.⁴⁹ The resultant vaccines were found in animal studies to be safe and immunogenic.^59-65 An A/H5N1 vaccine produced following generation of the reassortant strain in Vero cells could elicit antibodies which could neutralize a broad range of A/H5N1 strains in mice and guinea pigs.⁶² The whole virion killed vaccine also induced subtype-specific T-helper type 1 (Th1) and subtype cross-reactive T-helper type 2 (Th2) responses in mice.⁶² Mice were also protected against at least 2 heterologous A/H5N1 strains belonging to 2 different clades.⁶² Indeed, Vero cell-derived vaccines were better inducers of T cell proliferation and Th1 responses in mice than were ECE-derived vaccines.⁶⁶

Human clinical trials for Vero cell-derived A/H5N1 influenza vaccines showed that doses ranging from 3.75 to 45 μg of hemagglutinin (HA) protein equivalent were well tolerated, induced robust clade- and cross-clade-reactive antibodies in pediatric and adult populations, and that these responses could be boosted 6–24 months later.^67-72 Seroconversion rates ranged from 6% to 86%; the responses were typically higher with a higher dose or following boosting, but lower when adjuvanted with aluminum salts, in older individuals, and against viruses belonging to heterologous clades. Significantly, responses persisted for 12–24 months in a considerable proportion (26–86%) of the subjects.^67,69,71,72 Seroprotective levels of antibodies were achieved in 10–100% of the individuals, generally mirroring the variation observed in seroconversion.^67-72 In general, SRP levels were observed in >70% of the subjects after 2 conventional doses, and the proportion of subjects rose to >90% of the adults and >70% of the elderly following another dose at 6, 12–15 or 24 months following a single or a double dose administered earlier.⁷⁰ GMRs ranging from 1.6 to 32.8 have been observed, again following a pattern similar to SRC and SRP.^67,69,71,72 The vaccine could also induce (a) strong and effective, albeit not boostable, anti-neuraminidase antibodies,⁷³ and (b) clade and cross-clade-reactive CD4⁺ T cell responses in both adults and the elderly, the responses lasting at least 6 months at significant levels in adults.⁷⁴ Importantly, the CD4⁺ T-cell responses cross-reacted with A/H1N1, but not to A/H1N2 or B seasonal influenza strains.⁷⁴ In addition, human serum antibodies protected mice against virus challenge.⁷⁵

A large-scale multi-center trial of Vero cell-derived seasonal influenza vaccine in adults found that SRC rates were 70.4% against A/H1N1, 79.1% against A/H3N2 and 65.7% against influenza B, SRP rates were 88% against A/H1N1, 93.3% against A/H3N2 and 97.1% against influenza B, in a population which showed baseline SRP values of 29.5%, 39.4% and 56.4% against A/H1N1, A/H3N2 and influenza B, respectively.⁷⁶ In the same study, GMR of 11.1, 13.5 and 7.6 were achieved against A/H1N1, A/H3N2 and influenza B, respectively. The protective efficacy was 79%, 50% and 100% by culture positivity assay, and 75.2%, 50% and 60.1% by nucleic acid positivity assay, against A/H1N1, A/H3N2 and influenza B, respectively.⁷⁶ Overall efficacy of Vero cell-derived seasonal influenza vaccines has been reported to be 73–82% against antigenically matched and 68–83% against all strains of influenza viruses.^77,78 With the 2009 pandemic A/H1N1 (H1N1pdm09) strain, higher and comparatively longer lasting protective efficacy was reached, with a single dose in adults and older adolescents, and with 2 doses in young adolescents and children, and these responses were boosted strongly by the A/H1N1 component of the seasonal trivalent vaccine.^76,79

Safety profiles of Vero cell-derived influenza vaccines have shown that they are generally as well tolerated as the ECE-derived vaccines. Total local adverse events (AEs) have ranged from 11% to 29%, typically higher after the first dose than the second dose, as well as when adjuvanted with aluminum salts, and in younger individuals.^67,70,71,79 The most frequently reported local AEs have been pain at injection site (10–43% of the subjects), followed by redness, swelling, induration and ecchymosis.^{67,70,71,79-81} Total systemic AEs have been reported in 18–51% of the subjects,^65,68,69 the most common being headache, fatigue, myalgia and malaise (3–20% of the subjects) in adults as well as the elderly.^70,71,81 A wider variety of events have been reported in children.^79,80 However, the frequency of occurrence of the AEs is not significantly different from subjects administered a placebo or an ECE-derived vaccine.

MDCK cell-derived influenza vaccines: The final joust?

Compared to other cell lines, the MDCK cells present several advantages for influenza vaccine production. First, MDCK cells are the most suitable substrates among cultured cells to obtain primary isolates of influenza viruses.^37,54,82-85 This might be partly because (a) of the inability of canine IFN-induced myxovirus resistance protein 1 (Mx1) to inhibit influenza virus replication, and (b) trypsin used to facilitate virus entry during infection of cells exerts negative effects on IFN-induced antiviral proteins.^86-88 Together, the ineffectiveness of the IFN system may allow the virus to replicate to higher titres. A comparison of various cell lines for supporting the replication of live attenuated influenza viruses showed that MDCK cells are better than Vero, Medical Research Council-5 (MRC-5) human fetal lung fibroblast, Wistar Institute-38 (WI-38) human fetal diploid lung, fetal rhesus lung (FRhL), A549 human lung carcinoma and National Cancer Institute (NCI) H292 human mucoepidermoid bronchiolar carcinoma cells.^89-91 Whereas A549, NCI H292, MRC-5, WI-38 and FRhL cells produced low to moderate titres of virus, both Vero and MDCK cells could support the replication of influenza virus to higher titres.^89,90 However, Vero cells could only produce high virus titres with certain strains and when grown in the presence of serum.⁸⁹ MDCK cells may also be better suited for neutralization assays.⁹² Second, MDCK cells have been shown to be the most suitable for large-scale production of influenza virus.^89,93,94 Head-to-head comparison in laboratory scale bioreactors showed that MDCK cells yielded more virus than did Vero cells.⁹⁴ Third, influenza virus replicates more rapidly in MDCK cells compared to other cell lines,⁹⁵ and can be adapted to produce high titres in MDCK cells in as few as 3 to 10 passages, i.e., in 10–30 days, depending on the strain. This may reduce the lead time for vaccine production.⁹⁶ Fewer passages during adaptation would also reduce the chances of accumulation of mutations around the receptor binding site of the HA protein. In addition, trypsin does not need to be added frequently for virus propagation in MDCK cells, avoiding potential chances of contamination, although trypsin inhibitors have also been reported to be secreted by MDCK cells.⁹⁷ Fourth, the use of MDCK cells may be significantly more advantageous for the production of some influenza B virus vaccines.⁹³ Finally, MDCK cells are refractory to human and mouse prions,⁹⁸ and in vitro data suggests that MDCK cell derived components are not allergenic.^99,100

Extensive literature exists on the adaptation of MDCK cells for scaling up and influenza vaccine production. The cells can be easily adapted to and be grown in serum-free media, and in suspension, as well as on various microcarriers maintained under various bioreactor conditions.^93,101-108 Subclones of MDCK cells adopted to grow in suspension and support robust virus production have also been described,^91,108,109 although adherent MDCK cells appear to support more robust virus production than suspension MDCK cells.¹¹⁰

Influenza vaccines derived from MDCK cells are also safe and immunogenic. Initial studies which compared ECE- and MDCK cell-derived vaccines in Phase I clinical trials demonstrated the comparable safety and immunogenicity of the 2 vaccines in children, healthy adults and the elderly.^111-114 Other studies found that MDCK cell-derived vaccines were at least equivalent, and sometimes better and more efficacious as compared to ECE-derived antigens.^{111,112,114-120} In one instance, it was reported that at risk adult and elderly subjects who did not respond serologically to a previous ECE-derived vaccine responded better when boosted with MDCK cell-derived vaccine as compared to an ECE-based vaccine.¹²¹

Since the early 1990s, reports of more than 20 clinical studies involving greater than 20,000 subjects in over a dozen countries, as well as large-scale immunization programs have further confirmed the safety and immunogenicity of MDCK cell-derived influenza vaccines. As far as safety is concerned, overall AEs have been reported in up to 84% of the subjects,^{112,114,116,117,122-125} with a higher incidence in adults (60–84%) as compared to children (50–60%) and lowest (typically 15–25%, but sometimes up to 50%) in the elderly.^{114,122,123,125} Total local AEs have ranged from 10% to 84%,^{112,114–117,122,124,126} again, typically higher in adults than in children, and lowest in the elderly.^{114,116,118,122} Local AEs are also higher in the case of adjuvanted vaccine formulations as compared to unadjuvanted vaccine.^122,123,126 The most common local AE has been pain at injection site (12–75%), followed by erythema (2–20%), induration (6–15%), swelling (2–15%), and ecchymosis (0–18%).^{112,115,116-120,122,123,125–129} Some investigators have also reported limitation in movement, tenderness and bruising.^114,127 In general, the local reactions are mild, and are not significantly different from subjects administered ECE-derived vaccine or a placebo. Mild to severe reactions requiring medical attention are observed at the most in 25% of the total local AEs, and are usually more frequent in children.^112,114,124

Systemic AEs to MDCK cell-derived influenza vaccines have been found to be lower as compared to local AEs. Total systemic AEs have ranged from 20% to 55%.^{112,114-117,122,126} Similar to local AEs, systemic AEs are also lowest in the elderly.^{114,116,118,122} However, in contrast to the local AEs, systemic AEs are only slightly lower in children as compared to that in adults.^114,122 Adjuvanted preparations typically produce higher local AEs but systemic AEs are either similar or only slightly more as compared to unadjuvanted vaccines.^122,123,126 The commonest systemic AE is headache, being reported in 6.7–32% of the subjects, followed by myalgia (2–30%), fatigue (4–24%), malaise (3–25%), sweating (0–16%), chills (0–14%), and arthralgia (0–15%).^{112,114-120,122,123,125,126,128,129} Other systemic AEs, which are typically observed in less than 10% of the subjects, include nausea, loss of appetite, diarrhea, vomiting, fever and rash. A wider variety of systemic reactions, including sleepiness, inappetence, irritability, and unusual crying have been reported in young children.¹²⁹ None of the systemic AEs are significantly different from those due to ECE-derived vaccine. In addition, the systemic AEs typically disappear following a short symptomatic treatment.

Immunogenicity studies with MDCK cell-derived influenza vaccines in humans have revealed that (a) the SRC rates range from 25% to 100%, (b) the SRP rates are achieved in 70–100% of the subjects, and (c) the GMRs vary widely from as low as 1.89 to as high as 478, depending on the quantity of antigen injected, the frequency and time duration between the doses, the age of the vaccinee and whether the vaccine was formulated with an adjuvant.^{111,112,114-120,122,123,125,127,128} The SRC rates and GMR are typically the best in adults, lower in children and the lowest in the elderly, whereas the SRP rates are similar in children and adults while being slightly lower in the elderly.^{114,116,118,120,122,123,125,128} When comparing between subtypes of the viruses used as part of a trivalent vaccine, overall responses in terms of SRC, SRP and GMR are equivalent against the different influenza A subtypes H3N2 and H1N1, whereas responses to influenza B virus are comparatively weaker.^117,125,128 Some studies have shown that substantial proportion of the subjects retain seroprotective titres for at least 6 months, and in some cases, more than a year,^115,123 and that the dose can be reduced at least by half by using MF59 as an adjuvant.^{122,127,130-132}

Whereas all the above described studies involve killed vaccines, a recent study has shown that an MDCK cell-derived LAV vaccine containing 3 reassortant virus strains does not cause AEs that are significantly different from placebo, but immunogenicity, as determined by serum antibody titres, could only be demonstrated for H3N2, but not for H1N1 or B, influenza viruses.¹³³ However, since no ILI was observed, it is possible that either airway IgA or cell-mediated responses may have been responsible for protection against infection and/or disease.

Other cell lines tested for influenza vaccine production

Several other cell lines including the human embryonic retinal cell line PER.C6, the chick embryo cell line PBS-1, the duck embryo retinal cell line AGE.CR, and the duck embryonic stem cell line EB66 have been explored for producing influenza vaccine. The only cell line from humans is PER.C6, which was derived by transformation of embryonic retinal epithelial cells with human adenovirus type 5 E1 region genes.¹³⁴ The cell line has been characterized extensively to meet regulatory requirements.¹³⁵ These cells can be propagated in suspension and be adopted to serum-free and bioreactor conditions.¹³⁶ Influenza viruses can be propagated in PER.C6 cells, and reassortant vaccine strains can be generated in these cells.^63,136-138 Avian influenza H7N1 vaccine produced in PER.C6 cells has been shown to be well tolerated by humans.¹³⁹ However, immunogenicity studies indicated slightly lower antibody titres, which were enhanced somewhat by aluminum adjuvant.¹³⁹ There was also a correlation with antibody response and antibody secreting cells and interleukin-2 production.¹³⁹ However, poor immunogenicity in humans is a concern with influenza vaccine produced from PER.C6 cells, although more extensive studies are required. The related human embryonic kidney 293 cells, which are also derived through transformation using sheared adenovirus genome containing the E1 region genes,¹⁴⁰ have also been shown to support production of scalable quantities of influenza virus.¹⁴¹

Three avian cell lines have also been developed for the production of influenza vaccines. The best characterized of these are the AGE1.CR and the EB66 cell lines. The AGE1.CR cell line was derived by transforming Muscovy duck embryo retinal cells through stable transfection with human adenovirus type 5 E1A/E1B genes. The cells have been extensively characterized to meet regulatory requirements, and can grow in suspension, although they require the addition of trypsin for influenza virus propagation,^105,142-144 The EB66 cell line was derived as a stable line through a non-chemical non-genetic selection process, and can be adopted to various culture conditions, including suspension culture, bioreactor conditions and to serum-free or chemically defined media, and can grow to high density.¹⁴⁵ Another avian cell line potentially useful for influenza virus production is the immortalized chick embryo cell line PBS-1. This is a non-tumorigenic cell line which has been reported to support the propagation of influenza virus to titres higher than that obtained with Vero, MDCK and primary chick embryo kidney cells.¹⁴⁶ The cells have also been shown to be free of adventitious agents, and are readily adaptable to a variety of culture conditions, including growth in serum-free media and on microcarrier beads.^146,147 Importantly, propagation of influenza viruses in PBS-1 cells does not require the addition of exogenous trypsin, reducing chances of contamination, and simplifying downstream processing.¹⁴⁶ However, pre-clinical development and clinical trial studies are necessary for influenza vaccines produced in any of the avian cell lines.

Another cell line demonstrated to support growth of influenza viruses to titres comparable to those with MDCK cells is the mink lung epithelial cell line Mv1Lu.¹⁴⁸ Cytopathic effects can be observed earlier in these cells, and titres of influenza A viruses are higher in these cells than in MDCK cells.^92,148-150 However, further characterization and clinical studies would be required before they could be adopted as cell substrates for influenza vaccine production.

Issues related to cell culture-derived influenza vaccines

No vaccine production system is without shortcomings and cell-based influenza vaccine platforms are not an exception. Besides being vulnerable to occasional contamination, the cells must be free from additional or specific extraneous agents relevant to the species of origin of the cell line. This is typically not an issue since tests carried out for characterization of cell banks include extensive adventitious agent testing. Another issue is the fact that the reference reagents used for quality control and potency assays were largely produced using ECE-grown virus, and there is evidence that reactivity could be broader when MDCK-grown virus is used.^46,55,92 However, this could be overcome by shifting to producing reagents using cell culture-derived virus. It is also possible that some virus strains may not grow well in certain cell types, and several cell lines or clones may need to be tested for initial expansion of the vaccine virus.³⁷ A particular concern with the use of cell lines is that intellectual property issues may hinder wide utilization of an available system or technology. And finally, manufacturing influenza vaccines in cell culture is typically more expensive than vaccine production using ECEs.

Initial problems with the use of Vero cells in influenza vaccine production stemmed from the fact that virus replication was found to be low.^151,152 It was demonstrated that one of the reasons for this was that Vero cells contain an anti-tryptic activity, and that repeated addition of TPCK-treated trypsin could aid in multicycle replication and generation of high virus yields.¹⁵³ Even so, comparative studies have shown that whereas peak titres could be higher,⁹⁴ the growth of influenza A and B viruses may be slower in Vero cells compared to that in MDCK cells,^48,93-95,101 and that extended passaging may be required to obtain high titer virus suitable for industrial scale production.^154,155 This may be disadvantageous in case of serious epidemics or pandemics, when vaccines need to be available in a short amount of time. In addition, as the whole process could take 1 to 3 months, the virus may accumulate mutations during the passaging process. The trypsin concentration may also need to be optimized for each culture condition as the enzyme may influence adherence of cells which are already under bioreactor shear forces.⁹³ On the other hand, the ratio of genome copies to infectious unit or infectious unit to total virions have been found to be better with influenza A propagated in Vero cells than in MDCK cells.^94,95 The virus produced in Vero cells has also been shown to be more stable at least with A/H1N1, compared to MDCK-derived virus, especially when propagated in serum-free medium.⁹⁴ In addition, reverse genetics systems have also been developed so that high-yielding reassortant vaccine strains can be rapidly generated directly in Vero cells.^49,156,157

A major issue with the use of MDCK cells for the production of influenza vaccines is tumorigenicity. MDCK cells were derived in 1958 from the kidney of an apparently healthy female cocker spaniel dog, and are reported to be reverse transcriptase negative and non-tumorigenic.¹⁵⁸ Since then, multiple independent lines showing varying degrees of tumorigenicity, including apparently non-tumorigenic clones, have been derived by various groups.^{113,159–162} However, several lines of argument have been put forward to dispel concerns about tumorigenicity. First of all, the probability of live cell remaining in the final product is extremely low (10^{−25 to −35}/dose) due to the multiple redundant manufacturing steps which remove live cells.^163-166 Moreover, any live MDCK cell is likely to be rejected by the immune system as it is xenogenic to humans.¹⁵⁸ Indeed, except in completely immunocompromised nude mice, MDCK cells have not been shown to be tumorigenic, and either cell extracts or DNA extracted from cells are not oncogenic even in nude mice.^163,164 In addition, models for mitigation of risks, including those due to adventitious agents, have found that MDCK cells are safe cell substrates, and equivalent or better than ECE or Vero cells for producing vaccines.^165,167-173 In addition to MDCK cells, PER.C6 cells are also potentially tumorigenic, although these risks can be mitigated through the production process.¹⁷⁴ However, concerns about tumorigenicity might be more relevant, even if infinitesimally minor, for LAV vaccines produced in these cells.

Other concerns relating to the use of MDCK cells should also be noted here. One issue is that the use of MDCK cells to isolate viruses from respiratory samples can also yield viruses other than influenza virus, but this may be only a minor problem as these other adventitious viruses could be rapidly eliminated upon passaging.¹⁷³ A couple of studies have also reported that passage of influenza virus in MDCK cells may select antigenically and structurally distinct variants, e.g., those that mirror ECE-derived viruses,¹⁷⁵ or those that show altered fusion and entry characteristics,¹⁷⁶ and these altered viruses may outgrow other, desired phenotypes. Another study has also reported quasispecies consisting of HA variants in Vero and MDCK cells.¹⁵⁴ It is not clear how these would affect vaccine production or overall immunogenicity of the vaccine. On the other hand, certain other mutations have been shown to affect serological assays, and may influence the serological evaluation of the vaccine.^177,178 Other issues with cell culture-derived vaccines include lower levels of neuraminidase in the virion compared to virus grown in ECE,⁴⁶ and differential glycosylation patterns.^{92,152,176,177} Although these may influence immunogenicity in animal models,¹⁷⁹ overwhelming evidence from clinical studies in humans suggests that there is no significant difference in immunogenicity. However, this has not been supported with a detailed characterization of how these differences affect the various components of the immune response.

Despite advancements, the use of the cell culture platform for influenza vaccine production faces commercial challenges. First, each manufacturer will almost certainly have to optimize their cells for growth conditions and to support propagation of influenza virus. In addition, growth of cells in serum- and animal component-free media, and in suspension or in bioreactor conditions may need to be standardized. Second, the virus yield and titer per unit volume are lower and variable as compared to ECE-derived vaccine production, necessitating the use of steps to concentrate the virus. However, this could be circumvented by generating a high yielding reassortment recipient virus. Third, there is a need for extensive testing for adventitious agents, not only for human pathogens but also for other mammalian pathogens, especially those originating from the host from which the cells were derived. In addition, since the virus is highly lytic, host cell-derived proteins and DNA need to be carefully removed to meet regulatory requirements. Fourth, the cost of production per unit dose is high. Fifth, the technology for the production of influenza vaccines using ECEs is unrestricted whereas widespread application of the cell culture platform for influenza vaccine production is hindered by restrictions due to intellectual property rights. Finally, influenza is still not considered to be a serious disease in most of the developing and underdeveloped countries, where diseases such as malaria, tuberculosis, acquired immunodeficiency syndrome, dengue etc. are more important. However, it has to be noted that experience with the production of cell culture-based influenza vaccines is barely 2 decades old compared to 6 decades with the production using ECEs, and it is anticipated that recent investments in new facilities, research and development, and production and downstream processes will increase the practical utility of this vaccine platform.

I thank Dr. Kimberley Goldsmith, Institute of Psychology, Psychiatry and Neuroscience, King's College London, for critical review of the manuscript.