Formulation Studies of an Adsorbed Conjugate Vaccine

ABSTRACT

NeisVac-C is a polysaccharide-protein conjugate vaccine for use against Neisseria meningitidis serogroup C infection. The Phase 1 clinical formulation consisted of physiological saline, thimerosal, and aluminum hydroxide. In an attempt to maintain pH for the commercial vaccine, phosphate buffered saline (PBS) replaced saline in a subsequent clinical formulation. In addition, thimerosal was eliminated and the vaccine was filled in single-dose syringes. Before the clinical trial began, the amount of group C meningococcal polysaccharide tetanus toxoid (GCMP-TT) conjugate adsorbed to aluminum hydroxide (percent adsorption) dropped from >90% to about 80% in 1–4 months. The PBS formulation was determined to be unacceptable and new lots were produced with the saline formulation. These lots were used for Phase 2–3 clinical trials. The saline formulation continued to be used for commercial product. The long-term stability data for both the PBS and saline formulations are presented in this article.

NeisVac-C was introduced in the United Kingdom in 2000. Currently, it is licensed in 35 countries. The vaccine is used to prevent the invasive disease caused by N. meningitidis serogroup C. The active ingredient of the vaccine is a polysaccharide–protein conjugate. Each dose contains 10 μg of De-O-acetylated group C meningococcal polysaccharide (GCMP) conjugated to 10–20 μg of tetanus toxoid protein (TT) and adsorbed to 0.5 mg of aluminum in saline. This report summarizes the formulation studies performed during product development and demonstrates the importance of formulation in vaccine stability.

(Baxter Healthcare)

Materials and Methods

The antigen component of the vaccine, De-O-acetylated GCMP, was purified from culture supernatant of N. meningitidis serogroup C, strain C11. The carrier protein, tetanus toxoid, was obtained from Statens Serum Institut, Denmark. The conjugation of GCMP to TT used reductive amination technology.¹ The adjuvant, aluminum hydroxide, was obtained from Brenntag Biosector, Denmark.

Percent adsorption by GCMP ELISA (method A): In a formulated sample, the aluminum hydroxide phase was removed by centrifugation and a competitive ELISA measured GCMP in the supernatant. In the competition assay, sample and serum containing antibody against the analyte were incubated together in a coated plate. The 96-well microtiter plates were coated with GCMP–human serum albumin conjugate and blocked with 2% casein. The plates were incubated with rabbit anti-GCMP–HSA antiserum, which was mixed with the supernatant from the formulation sample. Goat anti-rabbit IgG (H+L)–horseradish peroxidase conjugate was used as the probing antibody with TMB (3,3', 5,5'-tetramethylbenzidine) as the substrate. The color development was inversely proportional to the amount of GCMP in the sample. The parent lot of conjugate was used as the standard. Calculations were based on the maximum optical density (OD_max), measured using wells containing no GCMP, and the OD from the standard. Figure 1 shows a typical standard curve. Percent adsorption was calculated as 100% – percent in the supernatant.

Figure 1. Percent adsorption by competitive GCMP ELISA: standard curve

Percent adsorption by resorcinol-HCl method (method B): This method measured GCMP monomer (sialic acid) by a colorimetric assay using N-acetylneuramic acid (sialic acid) as a standard.² GCMP in the aluminum hydroxide phase (precipitate) and in the supernatant were separated by centrifugation. The amount of GCMP in the precipitate and in the supernatant and the whole formulated sample were measured separately. The interference from aluminum hydroxide was minimal because OD was measured by extracting the color substances into the organic phase while aluminum hydroxide was left in the aqueous phase and discarded. The percent adsorption was determined directly by dividing the amount of GCMP in the precipitate by the total GCMP.

The protein content in the adsorbed supernatant was determined by the Bradford method using human immunoglobulin G (Pierce) as a standard.³

Potency of the formulated vaccine was evaluated using Swiss Webster mice (Harlan, Indianapolis, IN). Mice, 10 in each group, were injected with 0.2 mL of the diluted vaccine sample at 10 μg/mL, a negative control, or a positive control, on days 0, 14, and 28. The animals were exsanguinated on day 38, and sera were collected and stored frozen until they were analyzed. The immune response of the murine serum was determined by serum bactericidal antibody titer, which reflected functional antibodies.⁴ The assay used baby rabbit complement (Pel-Freeze Biologicals, Rogers, AR) and strain C11 of N. meningitidis. Test samples consisted of serum pools prepared from equal aliquots of individual mouse sera in each group.

Results

The Phase 1 formulation consisted of 20 μg/mL GCMP–TT (μg refers to GCMP) in 150 mM NaCl (saline), 0.01 % thimerosal and 1 mg/mL aluminum (added as aluminum hydroxide), and filled in single-dose vials with a deliverable dose of 0.5 mL. The 3 mL vial was Type 1 Flint glass and the closure was 13 mm chlorobutyl rubber stoppers. After four years of storage at 5 °C, the pH of the vaccine rose from 5.5 to 7.3. The slow rise in pH was caused by gradual leaching of alkaline from the glass vials.

Phosphate buffered saline (PBS, 10 mM sodium phosphate and 150 mM NaCl, pH 7.4) was chosen for the Phase 2–3 clinical formulation because of pH stability concerns in the unbuffered vaccine.

The adsorption kinetics were studied at small scale and GCMP ELISA (method A) was used to measure percent adsorption. Adsorption of GCMP–TT to aluminum hydroxide gel in saline reached completion instantaneously (>99.9% adsorbed). Adsorption in PBS went slowly and in an hour only 98% was adsorbed. For clinical production, we formulated in PBS for one hour and used a tentative specification of ≥90% for percent adsorption. The formulated vaccine was filled into 1-mL syringes with a 0.5-mL deliverable dose. The syringe barrels were silicon lubricated European and US Pharmacopeia type 1 borosilicate glass (Becton Dickinson). The closures were siliconized bromobutyl plunger stoppers. Medical grade DC 360 silicone helped stopper placement and movement and was consistent with US standards. Tip caps (chlorobutyl natural rubber blend) maintained sterility of the closure. Thimerosal preservative was not necessary for single-dose fills and was eliminated. The formulation consisted of 20 μg/mL GCMP–TT and 1 mg/mL aluminum in PBS.

Two lots (49801 and 49802) were produced for Phase 2–3 trials and placed on stability studies. The stability indicating methods included percent adsorption, pH, potency, sterility, general safety, and appearance.

After completing formulation, the percent adsorption samples were taken from the formulated bulk and determined to be 97% and 98% (method A) for lots 49801 and 49802, respectively. Both lots passed specification.

Before the start of clinical trials, percent adsorption dropped for both lots as determined by GCMP ELISA. Because of the decrease in percent adsorption, the supernatant contained detectable amounts of GCMP and protein that allowed us to confirm the percent adsorption by chemical assays measuring sialic acid and protein, respectively (Table 1).^2,3

Table 1. Desorption of conjugate vaccine from the aluminum hydroxide adjuvant in PBS

The PBS formulation was determined to be unstable and unacceptable. New lots were formulated in saline and used for clinical trials. The same formulation has been used for commercial product.

Table 2. Percent adsorption data (method B) for development lots

For commercial product, a resorcinol method (method B) was used to determine percent adsorption. For the 12 development lots examined, the results ranged from 90% to 102% (Table 2); this range is consistent with the variability of the sialic acid assay. Statistical analyses with 99% (one-sided) confidence limits suggested a specification of greater than or equal to 87%. Although the supernatant contained minimal amounts of GCMP, the recovery varied from 90% to 102% and the percent adsorption also varied from 90% to 102%. The change of percent adsorption specification reflected a change in assay method and not a decrease in percent adsorption.

Commercial lots were filled in a Baxter facility in Vienna, Austria, and three lots were placed on long-term stability study.

Figure 2a. Percent adsorption for GCMP–TT in PBS

Figure 2b. pH of GCMP–TT in PBS

Figure 2c. Potency of GCMP–TT in PBS

Figure 2d. Percent adsorption for GCMP–TT in saline

Figure 2e. pH of GCMP–TT in saline

Figure 2f. Potency of GCMP–TT in saline

The stability data for PBS and saline formulations are presented in Figure 2. The time points for each study are summarized in Table 3. The specifications for saline and PBS formulations are presented in Table 4. The saline formulation provided stable adsorption with some drift in pH. The PBS formulation provided a stable pH, but, the percent adsorption dropped initially and stayed low. The vaccine remained potent in animals with both formulations.

Table 3. Summary of the time points for the long-term stability studies at 2–8 °C (data shown in Figure 2)

Discussion

Desorption of antigen from aluminum hydroxide adsorbed vaccine by phosphate has been reported.⁵ Rinella, et al., reported desorption of negatively charged ovalbumin from aluminum hydroxide adjuvant by the addition of phosphate. Phosphate anions were adsorbed to aluminum hydroxide and lowered the isoelectric point of the adjuvant. This in turn decreased the electrostatic interaction with negatively charged antigen. The extent of desorption depended on the ionic strength, the phosphate concentration, and the age of the vaccine. As the antigen adjuvant complex aged, the ability of phosphate to desorb the antigen decreased. It was believed that the antigen might undergo conformational changes to optimize its interaction with adjuvant. Our data supported this hypothesis. The desorption of GCMP–TT from aluminum hydroxide occurred only in the first few months and leveled off during the two-year storage.

Table 4. Specifications established for clinical and commercial lots

Iyer, et al., treated aluminum hydroxide adjuvant with phosphate ion, which resulted in ligand exchange of phosphate for surface hydroxyl groups.⁶ This ligand exchange decreased the isoelectric point of the adjuvant. In addition, the adsorptive capacity and adsorptive coefficient for a negatively charged antigen were reduced.

The antigen in the current study, GCMP–TT, is negatively charged because of functional groups on the sialic acid monomer. The lowering of percent adsorption in PBS was caused by the ligand exchange of phosphate for hydroxyl groups on the aluminum and decreasing electrostatic interaction between aluminum and GCMP–TT. In the adsorbed vaccine formulation, phosphate buffer should be avoided. If buffering capacity was required to maintain the stability of the vaccine, we would recommend using Tris buffer instead of phosphate buffer.

Acknowledgement

The authors would like to thank Professor Stanley Hem for his interpretation of the PBS adsorption data.

SHWU-MAAN LEE, PHD, is a technical director, BOB KRUSE, PHD, is aresearch scientist, and CHRIS DONALDSON is a research associate, all at Baxter Healthcare, Beltsville, MD, 301.419.8587, lees4@baxter.com

References

1. Jennings HJ, Lugowski C. Immunochemistry of groups A, B, and C meningococcal polysaccharide-tetanus toxoid conjugates. J Immunol. 1981;127:1011–8.

2. Svennerholm L, Quantitative estimation of sialic acids. II. A colorimetric resorcinol-hydrochloric acid method. Biochimica et Biophysica Acta. 1957;24:604–11.

3. Bradford M. A rapid and sensitive method for the quantitation of microgram quantities of protein utilising the principle of protein-dye binding. Anal Biochem. 1976;72:248–54.

4. Michon F, Huang CH, Farley EK, Hronowski L, Di J, Fusco PC. Structure activity studies on group C meningococcal polysaccharide-protein conjugate vaccines: effect of O-acetylation on the nature of the protective epitope. Dev Biol. 2000;103:151–60.

5. Rinella JV Jr, White JL, Hem SL. Effect of anions on model aluminum-adjuvant-containing vaccines. J Colloid Interface Sci. 1995;172:121–30.

6. Iyer S, HogenEsch H, Hem SL. Effect of the degree of phosphate substitution in aluminum hydroxide adjuvant on the adsorption of phosphorylated proteins. Pharm Dev Tech. 2003;8(1):81–6.