Keywords

These keywords were added by machine and not by the authors. This process is experimental and the keywords may be updated as the learning algorithm improves.

1 Introduction

Over the last decade, advances in subunit vaccine technology, achieved in many cases with recombinant DNA techniques, have created a dramatic increase in demand for vaccine adjuvants which can help to ellicit protective responses from subunit antigens which, in general, are poorly immunogenic when administered in the absence of an adjuvant (1). The purpose of this chapter is to summarize our extensive experience with the oil-in-water emulsion adjuvant MF59 and to emphasize to the reader that MF59 is no longer a research formulation, but a functional commercial adjuvant. Here we provide information on good manufacturing processes (GMP), and methods of characterization for postproduction release and demonstration of long-term stability of MF59, as well as representative data to demonstrate the consistency of the product. We also provide information on the in vitro and in vivo performance of MF59 in combination with various vaccine antigens, which we have gathered during the decade of development, which has resulted in formulation of MF59 in the Fluad® vaccine, that marks the first European approval of a nonalum adjuvant for human use.

1.1 Mechanistic Approaches to Adjuvant Design

A variety of approaches have been utilized in searching for adjuvants that would be useful for clinical application to human disease prevention. Both alum, the principal adjuvant licensed for use in conjunction with human vaccines, and polylactide-coglycolide (PLG) microspheres have been reported to provide a depot that release antigen over an extended period of time (2). A number of molecular immunostimulators including muramyl peptides (3), lipid A derivatives (4), saponins (5), and bacterial toxins (6) have been used in combination with subunit antigens for both systemic and mucosal immunization. Many of the adjuvant formulations recently under development, including syntex adjuvant formulation (SAF) (8) and the SmithKline-Beecham (SKB) emulsions (9), liposomes (10), immunostimulating complex (ISCOMS) (11), and virus-like particles (VLPs) (Dupuis, M., personal communication), function as delivery systems that enhance transport of either antigen or molecular immunostimulators to antigen presenting cells. The emulsion adjuvant MF59 does not contain any known immunostimulatory molecules nor has any association between antigen and the emulsion droplets been shown to be critical for adjuvant activity. The administration of this particulate immunostimulator has been shown to result in the recruitment of antigen-presenting cells (APCs) to the site of injection (12), and to increased uptake of soluble antigen by the APCs (13).

1.2 Adjuvant Potency

The success of any vaccine/adjuvant formulation is dependent upon fulfillment of several requirements. The most important of these are potency, tolerable reactogenicity, and pharmaceutical feasibility. In order to be protective, the antigen/adjuvant formulation must have a specific potency for generation of the appropriate immune function. It is useful to classify adjuvants in terms of their performance regarding the correlation of immunity historically established for a variety of disease models. The generation of neutralizing antibody, present in serum or at mucosal surfaces, has frequently been correlated with protection (1). The presence of specific antibody subtypes and the demonstration

of antibody-dependent cellular cytotoxicity are likely to be associated with protection in some cases. Generation of cellular immunity, most particularly, cytotoxic T-lymphocytes (CTLs), has been thought to be contributory to protection in other cases. Finally, generation of specific cytokine profiles or generation of other soluble factors, including chemokines, has been considered to be potentially critical for protection in yet other systems. MF59 has been demonstrated to generate antibody titers determined by both enzyme-linked immunosorbent assay (ELISA) and neutralization assays, which are significantly greater than those obtained with aluminum salts for an extensive list of subunit antigens. Recognizing that cytokine profiles obtained after immunization are dependent on a number of variables, we would characterize MF59 as a Th-2 directing adjuvant. Finally, because CTLs have been obtained with some antigens by the very active subcutaneous (sc) route in mice, we do not characterize MF59 as a potent adjuvant for CTL generation.

1.3 Adjuvant Reactogenicity

Limiting reactogenicity of adjuvanted vaccines to a tolerable level for widespread administration to humans has been a critical problem in transition from animal models to the clinic. The first priority in formulation of MF59 was to ensure safety, thus a very conservative formulation based upon low-risk components has been utilized. Clinical testing with both influenza and herpes simplex virus (HSV) vaccines in more than 18,000 subjects have demonstrated minimal reactogenicity of these formulations and the adjuvant has been approved for both commercial use and for further testing in both infants and pregnant women.

1.4 Pharmaceutical Feasibility

The advanced state of development of MF59 offers the potential user significant advantages. MF59s manufactured under GMP conditions at a scale commensurate with commercial use as part of an adjuvanted influenza vaccine (Fluad). Extensive characterization of raw materials, development of a reproducible manufacturing process and derivation of suitable conditions for longterm stability have been achieved during the product development cycle. MF59 has shown excellent compatibility with a variety of subunit antigens, all of which have been formulated by a simple mixing of the antigen with the adjuvant. This simple approach allows final formulation to be performed in the clinic and reduces the need for extensive stability studies on early-phase candidate vaccines. On selected vaccines, storage stability in the presence of MF59 has also been established demonstrating the feasibility of long-term compatibility of antigens with MF59 on a case-by-case basis.

2 Materials

The second generation MF59 emulsion with enhanced stability characteristics has been designated as MF59C.1 and consists of the following components.

  1. 1.

    Squalene (2,6,10,15,19,23-hexamethyl-2,6,10,14,18,22-tetracosahexaene).

  2. 2.

    Polysorbate 80.

  3. 3.

    Sorbitan trioleate.

  4. 4.

    Trisodium citrate dihydrate.

  5. 5.

    Citric acid monohydrate.

  6. 6.

    Water-for-injection (WFI).

3 Methods

3.1 Manufacturing Process for 50l-Scale Production of MF59

Methods for laboratory-scale production of MF59 previously described (14), allow preparation of as little as 10 mL of emulsion with the Microfluidizer 110S. Here we describe the manufacturing process for sterile clinical grade MF59C.1 having defined release specifications and demonstrated long-term stability. The 50l scale manufacturing process for MF59C.1 is shown in Fig. 1 . Briefly, polysorbate 80 is dissolved in WFI and combined with aqueous sodium citrate-citric acid buffer solution. Separately, sorbitan trioleate is dissolved in squalene. These two solutions are combined together and processed in an inline homogenizer to yield a coarse emulsion. The coarse emulsion is fed into a microfluidizer, where it is further processed to obtain a stable submicron emulsion. The coarse emulsion is passed through the interaction chamber of the microfluidizer repeatedly until the desired particle size is obtained. The bulk emulsion is filtered through a 0.22-µm filter under nitrogen to remove large droplets, yielding MF59C.1 adjuvant emulsion bulk that is filled into glass bottles. For vaccine antigens that have demonstrated long-term stability in the presence of MF59 for shelf storage, the antigen and MF59 are combined and sterile-filtered through a 0.22-µm membrane. The combined “single-vial” vaccine is filled into single-dose containers. For vaccine antigens, where longterm stability has not been demonstrated, the adjuvant is supplied as a separate vial. In such cases, the MF59 bulk is filter-sterilized, filled, and packaged in final single-dose vials.

Fig. 1.
figure 1

MF59 manufacturing process.

Full size image

3.2 In Process Assays for MF59C. 1

The manufacturing process yields an MF59C.1 product in a reproducible and consistent manner. Figure 2 presents the combined particle size data from seven representative lots to demonstrate the efficient and repeatable reduction of mean particle size of MF59C.1 during each pass of the microfluidizer in the process. In addition to the mean particle size, we monitor the number of large particles, i.e., particles>1.2 µm in size, per milliliter of the adjuvant emulsion. Emulsions are thermodynamically unstable systems and are subject to flocculation (reversible aggregation of oil particles) and coalescence (irreversible aggregation of oil particles to form large particles and eventual, irrevocable separation of oil, and aqueous phases) during storage. For manufacturing of stable emulsions, a key objective is to keep the number of large particles down to a minimum because large particles act as nucleation sites for further aggregation during storage potentially leading to phase separation. Figure 3 shows data from seven representative lots of MF59C.1 for reduction in the number of large particles during the manufacturing process. After the microfluidization step, filtration of the emulsion through a 0.22-µm membrane removes 99.5% of particles>1.2 µm in size. The bulk emulsion contains less than 0.1% of total particles that are>1.2 µm.

Fig. 2.
figure 2

Mean particle size profiles of MF59 during microfluidization.

Full size image
Fig. 3.
figure 3

Number of ≥1.2 µm particles in MF59 bulk.

Full size image

3.3 Release Assays for MF59C.1

MF59C.1 is a well-defined emulsion produced to preestablished release specifications. The emulsion bulk and final single-dose adjuvant are analyzed using a battery of assays in accordance with Chiron’s standard operating procedures. Key assays include visual appearance, pH, mean particle size, and number of large particles per milliliter for quality, squalene, polysorbate 80, and sorbitan trioleate concentrations by high-performance liquid chromatography (HPLC) procedures for content and, endotoxin and bioburden content for safety. Visually, MF59C.1 is a milky white, homogenous liquid that is free of extraneous particles. Under stress conditions, such as prolonged exposure to high temperatures or freezing, large oil globules are formed. Storage of MF59C.1 under such conditions must, therefore, be avoided. MF59C.1 is buffered with the sodium citrate-citric acid buffer to pH 6.5. As during the process, mean particle size of MF59C.1 and the number of large particles per milliliter are important quality parameters for the bulk and final adjuvant emulsions. The mean particle size of MF59C.1 is approximately 150 nm and an upper specification of 1×107 particles ≥1.2 µm is observed. The quantities of squalene, polysorbate 80, and sorbitan trioleate must be within an acceptable range around the nominal concentration values of 39, 4.7, and 4.7 mg/mL, respectively.

3.4 Stability Assays for MF59C. 1

MF59C.1 is stable at 2-8°C for three years when stored in glass bottles protected from direct light. Physically, the emulsion is stable except for slight flocculation seen after it is placed at 2-8°C for a few weeks. Studies have demonstrated that MF59C.1 in its original particle size profile is obtained by inverting the closed container a few times. The pH, mean particle size, and squalene concentrations of MF59C.1 remain unchanged from initial values throughout the three-year shelf-life at 2-8°C. Finally, the number of large particles per milliliter of the adjuvant remains below its upper specification during the shelf-life period. Stability data from representative MF59C.1 lots are displayed in Figs. 4 and 5.

Fig. 4.
figure 4

Long-term particle size stability of the three representative MF59C.1 bulk lots at 2-8°C.

Full size image
Fig. 5.
figure 5

Long-term stability of squalene in three representative MF59C.1 bulk lots at 2-8°C.

Full size image

3.5 Final Vaccine Formulation

3.5.1 Vaccine Formats

A variety of antigens including HSV-2 gB/gD, HBV, HIV gp120 and/or p24, CMV gB, and influenza hemagglutinin (HA) have been formulated with the adjuvant emulsion MF59C.1 in one of two formats: either in single containers or in separate vials (admixed prior to administration and therefore referred to as dual vial vaccines). The antigens tested fall into three distinct classes of proteins:

  1. 1.

    soluble antigens: either monomeric low-molecular-weight type (e.g., HSV2 gD, 43 kDa) or high-molecular-weight aggregate (e.g., CMV gB 800 kDa aggregates);

  2. 2.

    hydrophobic: integral membrane protein with an intrinsic ability to self-associate forming protein micelles or to associate with lipid bilayers or emulsions (e.g., influenza HA);

  3. 3.

    particulate antigens: the protein is naturally associated to form defined structures optionally with lipids embedded as an integral part of the structure [e.g., hepatitis B surface antigen (HBsAg)].

    Antigens of all three types have been successfully formulated into singlevial vaccines after sufficient formulation optimization. These single-container vaccines showed different stability behaviors depending on the nature of the antigen and the susceptibility of the antigen to interact with the adjuvant. Table 1 summarizes much of the stability data obtained for MF59 vaccine formulations at Chiron Corporation. Only two of the antigens, HIV gp120 and CMV gB, were not suitable for single-vial formulation. HIV gp120 has been shown to undergo time-dependent conformational changes in single-container formulations. Such changes are evident by a loss in CD4 binding, a measurement, considered an in vitro surrogate potency assay. At this point, it is not clear if the polysorbate 80 in the adjuvant bound to the V3 loop of gp120 or the adjuvant (or oxidized MF59) triggered reactions occurred in a functional domain of the antigen to cause a loss in activity. The conformational changes observed could account for the changes in the immunological responses to the singlecontainer vaccines. In order to meet rapid timelines, dual-vial formulations were developed as candidate vaccines for advanced clinical trials. Recombinant CMV gB has shown the propensity to exist as an equilibrium mixture of various high molecular weight aggregates. Single-vial formulations of CMV gB with MF59 have shown a change in the composition of the antigen presumably because of the surfactant-induced disaggregation of the antigen. In addition, incubations of the single-container vaccines at>25°C led to pronounced crosslinking of the adjuvant with the antigen. A major portion of this reaction appeared to be unrelated to disulfide crosslinking of the protein. Mechanistically, oxidized surfactant and/or squalene may have generated reactive species such as malondialdehyde capable of generating nonreducible covalent crosslinks with the protein. In view of these experimental results, dual-vial formulations were further developed for clinical evaluations of this vaccine. In all other cases, single-vial formulations stable for several years were derived. For early phase testing where formulation optimization is impractical, antigens for the dual-vial vaccines are provided either frozen (especially for early clinical trials) or as a liquid at 2-8°C. Antigens are typically combined with the adjuvant, MF59C.1, by gentle mechanical mixing to provide single-container vaccines. The stabilities of the dual-container vaccines were governed by the intrinsic storage stabilities of the antigens and MF59 and short-term compatabilities of the vaccines generated upon mixing. In general, the antigen formulations were stable to storage for at least two years at 2-8°C. As discussed earlier, the buffered adjuvant was also quite stable to storage and the vaccines generated from these components were stable for at least for one workday, i.e., eight hours at ambient temperatures that facilitated administrations in clinical studies.

Table 1 Summary of Formulation Experience with MF59
Full size table

3.6 The Use of MF59 as a Vaccine Adjuvant

The MF59 emulsion adjuvant was developed with the objective of generating a broad spectrum of recombinant vaccines for human use. The specific aim was to elicit neutralizing titers in humans significantly greater than those obtainable with the alum adjuvants in common usage. An extensive body of preclinical efficacy data has been obtained based on ELISA titers obtained with a variety of subunit antigens in a spectrum of animal model systems. A summary of this data is shown in Table 2. Data are presented as the ratio of serum ELISA titers obtained with MF59 to that obtained with alum formulated vaccine. (In one case (*), where alum vaccines have been shown to be ineffective, the ratio of titers obtained with MF59 to that obtained with antigen alone is presented.) The principal conclusion to be drawn from this data is that MF59 is a significantly more potent adjuvant than the aluminum salts for most of the antigens tested in a variety of animal models. The enhancement of titer typically falls in the range from 5 to 40X. Most of the antigens used for development of MF59 were soluble recombinant truncates of viral surface glycoproteins (HIV gp120, HSV gD2, CMV gB, HCV E2). Significant activity has been demonstrated with glycoconjugate antigens (Hib, MenC) (15). MF59 has shown dramatic effects with two particulate antigens influenza HA (14) and (HBsAg) (16). No systematic trends have yet been established for antigenic characteristics that determine the degree of efficacy of MF59, though some very poorly soluble antigens have not shown good titers in this system (data not shown). Data in Table 2 also demonstrate that MF59 is effective across a spectrum of animal models typically used for preclinical testing. So far, no species tested has been unresponsive to MF59, which should be an excellent candidate for formulation of a variety of veterinary vaccines.

Table 2 Summary of Preclinical Experience with MF59 a
Full size table

Because MF59 is clearly effective for generation of serum antibody titers and CD4 T-cell response (data not shown), these responses are only protective against a subset of the pathogens for which novel or improved vaccines would be of utility. Two attributes of a number of vaccines formulated with MF59 may be of significance in design of additional adjuvanted vaccines. The cytokine balance associated with immunization has been shown to be important to protection in several instances. Figure 6 shows typical serum cytokine data obtained at either 3 or 12 h after a third intramuscular (im) immunization with MF59/HIV gp120 or gp120 with several Quil A-containing formulations. Immunizations with formulations containing MF59 result in greater serum concentration of interluken-5 (IL-5) and smaller concentrations of interferon (IFN-γ) than those obtained with Quil A containing systems. While we are aware that the Th-1/Th-2 balance represented by the IFN-γ/IL-5 ratio obtained upon immunization is a multivariate function, depending upon antigen identity and dose, adjuvant identity, genetics of the animal immunized and route of administration, we would characterize MF59 as a Th-2 directed adjuvant. The ratio of immunoglobulin (Ig) isotypes (IgG1/IgG2a), which frequently correlates with cytokine ratio, is a significant factor in complement-mediated cytotoxicity. Immunization with both HIV and HSV antigens in combination with MF59 results in high ratios of IgG1/IgG2a that are consistent with our characterization of MF59 as a Th-2 adjuvant. In contrast, we have shown that it possible to generate CTL with several antigens when formulated with MF59 Table 3 ). However, the mouse system we have used, base of the tail sc administration of antigen to mice, is the most permissive system for generation of CTL in routine use. So far, attempts to extend the data set into the somewhat more restrictive im mouse mode have not been successful. Limited work in primates (data not shown) has not shown reproducible production of CTL activity. We conclude that the MF59C formulation is most appropriate for application where antibody is the desired end point (e.g., antihormone therapies) or Th-2 cytokines and CD4 T cells are advantageous (e.g., Helicobacter pylori) including situations where CTL are not essential or may be dangerous.

Fig. 6.
figure 6

In vivo cytokine response to HIV envelope proteins and adjuvants.

Full size image
Table 3 Cytotoxic T-cell Activity Induced in Mice with Whole Recombinant Proteins Combined with MF59
Full size table

3.7 Assurance of Clinical Safety

The most critical concern in the development of postalum adjuvants has been the demonstration of safety for the large populations who receive the vaccine. Adjuvants, by definition, increase immunological responsiveness, which may result in immune reactivity to epitopes other than those necessary for protection as well as giving rise to side effects associated with generation of the response. The MF59 formulation has been restricted to components that are not individually immunostimulatory. The principal component, squalene, is a naturally occurring intermediate in cholesterol synthesis that is widely distributed in nature and is the primary component of shark liver oil. The MF59 emulsion is immunostimulatory and the safety of each vaccine formulated with the adjuvant must be demonstrated under controlled clinical conditions. Chiron has tested MF59 under such controlled clinical conditions with an extensive set of well-characterized antigens in more than 18,000 subjects Table 4 ). These clinical cohorts have been screened and carefully monitored postvaccination. The results have been published in detail for trials with HIV (16), HSV (17,19), CMV (20), and influenza (21–22 antigen-containing MF59 vaccines. The MF59 vaccines tested have been shown to be generally safe and immunogenic. Chiron’s influenza vaccine, Fluad, was approved by the Italian regulatory authorities in 1997. Since its introduction, more than 500,000 doses of Fluad have been commercially distributed. In accordance with current regulatory practices, Chiron continues to monitor safety data. The safety of Fluad, as demonstrated in a Phase IV trial, was excellent and similar to that of a licensed influenza vaccine that did not contain any adjuvant.

Table 4 Summary of Subjects Receiving MF59 and Vaccine Antigens through March 1999 in Chiron-Sponsored Programs
Full size table

4 Notes

  1. 1.

    In situations where high antibody titers are desired in combination with Th-1 cytokines and higher IgG2a/IgG1 ratios, we have shown that combinations of MF59 with PLG microspheres are useful. Development of MF59-based formulations, which generate Th-1 cytokine profiles and CTL, is ongoing.

  2. 2.

    Fluorescently labeled MF59 (3,3’-dioctadecylindocarbcyanine (CM DiI) has been successfully used to track adjuvant in vivo (13). The adjuvant is taken up by Mac 1+antigen-presenting cells in the muscle where it induces enhanced local uptake of antigen. Labeled cells have been shown to migrate to the draining lymph node where they express the markers associated with active dendritic cells.

  3. 3.

    Chiron Corporation routinely makes limited volume samples of clinical grade MF59 available for evaluation of preclinical vaccine formulations. In order to obtain MF59 for testing, it is necessary to submit a brief written summary of the proposed experiments and to agree to the terms of a standard materials transfer agreement, which will be transmitted upon receipt of the proposal summary. The process is typically accomplished within four weeks. We reserve the right to impose additional terms in situations where Chiron Corporation has commercial interest. For further information or materials, please contact the first author, G. Ott.