- Email: [email protected]
27Al and 31P NMR spectroscopy method development to quantify aluminum phosphate in adjuvanted vaccine formulations
Journal of Pharmaceutical and Biomedical Analysis 159 (2018) 166–172
Contents lists available at ScienceDirect
Journal of Pharmaceutical and Biomedical Analysis journal homepage: www.elsevier.com/locate/jpba
27
Al and 31 P NMR spectroscopy method development to quantify aluminum phosphate in adjuvanted vaccine formulations Rahima Khatun a,1 , Howard N. Hunter b,1 , Yi Sheng a , Bruce W. Carpick c , Marina D. Kirkitadze c,∗ a
Departments of Biology and Chemistry, York University, 4700 Keele Street, Toronto, Ontario, Canada York University, 4700 Keele Street, Toronto, Ontario, Canada c Analytical Research and Development, Sanofi Pasteur Ltd., 1755 Steeles Avenue West, Toronto, Ontario, Canada b
a r t i c l e
i n f o
Article history: Received 23 February 2018 Received in revised form 12 June 2018 Accepted 16 June 2018 Available online 21 June 2018 Keywords: Quantitative qNMR Aluminum Phosphate Adjuvant Vaccine
a b s t r a c t A novel qNMR method is described for the quantitative determination of total aluminum and phosphate in aluminum phosphate (AlPO4 ) adjuvanted vaccine samples using solution 27 Al and 31 P nuclear magnetic resonance (NMR) spectroscopy. External standard calibrations of AlPO4 solutions established excellent linearity in the range of 15–40 × 10−3 M and additional studies determined the level of detection for both nuclei. A commercialized combination vaccine product (Quadracel® ), along with several individual adsorbed antigen components used in the vaccine were employed as model systems for method development. The developed method is also capable of quantitating the free phosphate (i.e. the fraction not bound to AlPO4 particles) in adjuvanted vaccines. This study is the first demonstration of a solution NMR method that is suitable for measuring total aluminum, and free and total phosphate concentrations in vaccine formulations consisting of antigen(s) adsorbed to aluminum adjuvant, in a single analytical workflow. © 2018 The Authors. Published by Elsevier B.V. This is an open access article under the CC BY license (http://creativecommons.org/licenses/by/4.0/).
1. Introduction Adjuvants are substances added to vaccine antigens in order to enhance the immune response. Aluminum salts such as aluminum hydroxide (AlOOH) or aluminum phosphate (AlPO4 ) are the most prevalent adjuvants in commercial vaccines due to their excellent safety profiles and ability to induce robust immunity against pathogens [1]. Although the precise mechanism of action for aluminum adjuvants is not known [2,3], the physical interaction of the adjuvant with the vaccine antigens, typically termed adsorption, may play a role in immunogenicity. Antigen-adjuvant adsorption can occur through different mechanisms, such as electrostatic interactions or ligand exchange [4]. Regardless of the mechanism, the interaction of the antigen with the adjuvant surface in the vaccine formulation is thought to lead to an enhanced immune response, and is furthermore a potential marker of vaccine quality and consistency. On a molecular level, it has been shown that the aluminum and phosphate content of AlPO4 are important for the surface charge and can therefore have an impact on the antigen adsorption capac-
∗ Corresponding author. E-mail address: Marina.Kirkitadze@sanofi.com (M.D. Kirkitadze). 1 Authors contributed equally.
ity [5]. Furthermore, the colloidal stability of the AlPO4 adjuvant suspension can be affected by the components of vaccine matrix, e.g., by the excess of free phosphate [5]. Therefore, it is important to measure both free and total phosphate in vaccine formulations to ensure that surface charge remains stable and, in turn, does not alter absorption capacity of protein antigens to AlPO4 adjuvant [6]. As part of a batch release of vaccine, typically only the aluminum content is measured and this quantitation is completed using a colorimetric method involving the chelation with EDTA, followed by back titration with cupric sulfate [7]. Unfortunately, the European Pharmacopoeia monographs describe two different methods for assessing both free and bound phosphate [8,9]. Clearly, it would be quite convenient and advantageous to have a single, rapid method to measure these three characteristic components of adjuvant content during development, optimization, and monitoring of the vaccine formulation process [5]. A method which can quantitate all three components in vaccines, with minimal sample preparation, offers the potential to streamline the manufacturing process development and optimization for aluminum-adjuvanted vaccines. Advancements in technology and expanding NMR applications have continued to demonstrate the utility of NMR in the pharmaceutical industry. These applications range from the design of innovative drugs to the characterization of polymers, natural products, and metabolites [10]. Quantitative NMR (qNMR) continues to
https://doi.org/10.1016/j.jpba.2018.06.025 0731-7085/© 2018 The Authors. Published by Elsevier B.V. This is an open access article under the CC BY license (http://creativecommons.org/licenses/by/4.0/).
R. Khatun et al. / Journal of Pharmaceutical and Biomedical Analysis 159 (2018) 166–172
gain popularity as a method for rapid, analytical determination of drug metabolites, capsular polysaccharides and residuals in complex solutions, including vaccines [11–14]. The utility of NMR for such application relates to its unique ability to identify and quantify various nuclei in solution with minimal interference from the surrounding solvent matrix, making it well suited for monitoring complex samples and manufacturing processes. Most often, qNMR analysis has focused on the measurement of protons (1 H), however, other nuclei such as 31 P and 19 F have been utilized for quantitative metabolic studies [15–17]. The favorable 100% natural abundance of the 27 Al and 31 P isotopes are an initial indicator that Al and P are ideal targets for qNMR, since both nuclei have a high receptivity similar to that of a proton. However, 27 Al is a quadrupolar nucleus (spin 5/2) and the non-insignificant quadrupole moment of aluminum can dominate nuclear T1 and T2 relaxation resulting in extremely broad, or more often, unobservable resonances in the 27 Al spectrum [18,19]. The effect of the relaxation contribution from the electric field gradient about the 27 Al nucleus can be minimized by a coordinating environment for aluminum which results in higher symmetry. This coordination leads to longer T1 and T2 relaxation resulting in narrower, and more importantly, quantifiable NMR resonances [20]. Under these conditions, quantitation of aluminum by NMR has been shown by Maki et al to be feasible [21]. In a typical vaccine formulation, the high molecular weight of the active component (such as a protein antigen) far exceeds the size limit for the utility of high resolution NMR in solution. Under these conditions, spin diffusion becomes the predominant method of relaxation for any NMR active nucleus associated with the protein, and any NMR spectra obtained would be devoid of sharp resonances. Therefore, the detection of aluminum and phosphorous can only take place through release of phosphate and aluminum ions from the AlPO4 adjuvant suspension solid phase into the solution. As the pH is lowered, phosphate and aluminum ions are released into the solution and phosphate establishes equilibria between its PO4 3− , HPO4 2- , H2 PO4 - and H3 PO4 forms. In turn, the phosphate ions can combine with aluminum, increasing the complexity of the aqueous matrix. However, upon further acidification, the solution matrix can be shifted to smaller and more symmetrical aluminum-phosphate-water complexes as determined by Mortlock et al [22,23]. Although the 27 Al and 31 P spectra may show multiple resonances attributable to an ensemble of aluminum and phosphate complexes in solution, the total 27 Al or 31 P present can be correlated to the sum of the area under all 27 Al or 31 P resonances. The total area of these resonances can therefore be used to calculate the 27 Al or 31 P content. Here, for the first time, we present a novel qNMR method for the analysis of total aluminum and phosphorous in formulated vaccine products. The method was applied to a commercialized adjuvanted combination vaccine product, Quadracel® , along with several individual adsorbed antigen drug substances used to manufacture the combination vaccine. This method requires minimal sample preparation, provides rapid, quantitative analysis, and is amenable to routine monitoring of vaccine adjuvant and formulation processes.
2. Materials and methods 2.1. Materials Aluminum nitrate (Al(NO3 )3 ·9H2 O) (solid), 36% hydrochloric acid and sodium hydroxide solutions were purchased from Caledon Laboratories Ltd. Georgetown, Canada. Sodium phosphate monobasic dehydrate (NaH2 PO4 ) and deuterium oxide were purchased from Sigma-Aldrich (Oakville, Ontario and Milwaukee, WI). Quadracel® , an AlPO4 -adjuvanted combination vaccine product, consisting of adsorbed Diphtheria and Tetanus toxoids,
167
adsorbed acellular Pertussis antigens (Pertussis toxoid, Filamentous Haemagglutinin, Pertactin, and Fimbriae), and inactivated Poliomyelitis vaccine, along with individual adsorbed Diphtheria toxoid, Tetanus toxoid, and acellular Pertussis adsorbed antigen drug substances, were produced by Sanofi Pasteur Limited (Toronto, Ontario). 2.2. Sample preparation Calibration solutions were prepared by combining appropriate amounts of Al(NO3 )3 ·9H2 O and NaH2 PO4 stock solutions and adjusting the pH to either 7.0 or 0.79 to 0.81 with either HCl or NaOH solutions. Additional water and deuterium oxide were added to create a final volume of 2 mL of solution consisting of the desired concentration of 27 Al and 31 P along with 10% deuterium oxide for field-frequency lock. Exactly 0.6 mL of this solution was placed in an NMR tube for measurement. Quadracel vaccine analysis involved the combination of a 1.6 mL sample of vaccine with 177 L D2 O (10%). A 0.6 mL aliquot from this solution was placed in a NMR tube for 31 P NMR acquisition. Following data collection, the test sample was then returned to the stock solution and the pH was adjusted to 0.79-0.81 using concentrated HCl. A 0.6 mL aliquot of this solution was placed in a NMR tube and both the 27 Al and 31 P NMR spectra were acquired. A similar procedure was used, beginning with smaller initial volumes, for the other vaccine samples tested. All data were corrected for dilution factors. One of the Quadracel samples was spiked with 5 mM AlPO4 following initial analysis. The 27 Al and 31 P spectra were re-acquired and the increase in the resonance area of the spiked sample was compared to the original sample peak area to confirm the concentration of total 27 Al and total 31 P. 2.3. NMR measurements All NMR measurements were performed at 21 ◦ C on a DRX 600 NMR spectrometer operating with XWINNMR software version 3.5 (Bruker BioSpin Ltd., Karlsruhe, Germany) at 600.00 MHz using a 5 mm BBO probe. The T1 relaxation values for both 27 Al and 31 P nuclei were determined using the inversion-recovery method and the subsequent 1D spectra were acquired with 5 times the longest T1 value measured. The 27 Al spectra were acquired with 64 K data points, 128 scans, a 1 s relaxation delay and processed with 2 Hz line broadening function. At neutral pH the 31 P spectra were acquired with 64 K data points, 64 scans, and a 35 s relaxation delay. The 31 P spectra of acidified samples were acquired with 64 K data points, 128 scans, and a 25 s relaxation delay. A constant receiver gain value was used for each nucleus and was initially determined using the most concentrated calibration sample. The 27 Al and 31 P spectra were externally referenced to Al(NO3 )3 and 85% H3 PO4 respectively. All spectra were processed using TopSpin version 3.5.b.91 pl7 software (Bruker BioSpin Ltd., Karlsruhe, Germany) with 2 Hz line broadening function and baseline correction using a polynomial fit. For each of the spectra within a grouping (i.e. neutral pH phosphate, total aluminum or total phosphate), peak areas were standardized by integrating one of the calibration samples and then using the lastscale feature of the integration module of the software to permit comparison. With this feature peak areas from an initial data set are stored with a fixed digitally assigned value which can be recalled and directly applied to any subsequent data set. 3. Results and discussion Previous NMR studies showed that various aluminumphosphate complexes are formed in aqueous solution containing aluminum and phosphorous and that narrow resonance lines and the chemical shifts strictly depend on the pH of the solution
168
R. Khatun et al. / Journal of Pharmaceutical and Biomedical Analysis 159 (2018) 166–172
Fig. 1. 1D 31 P NMR spectrum of Quadracel® vaccine sample with 10% D2 O at neutral pH. The spectrum was acquired at 242.88 MHz with 64 scans, 35 s relaxation delay and externally referenced to 85% H3 PO4 .
[21–23]. Furthermore, 31 P NMR resonances of these aluminumphosphate complexes were successfully assigned to P atom at various pH values, while the detection and assignment of 27 Al resonances were only possible at low pH. It therefore seemed reasonable that for total concentration of aluminum and phosphate, the summation of the area under all of the resonances in the respective spectra could be evaluated. As 31 P NMR resonances for free phosphate in solution can readily be assigned at pH > 6.0 [22,23], a two-step analysis of vaccine samples was followed whereby the phosphate content was measured at neutral pH to assess free phosphate followed by adjusting the pH to acidic values to permit the measurement of total aluminum and total phosphate. The 31 P spectrum from Quadracel® vaccine at pH 7.0 shows a single sharp resonance in solution at 2.4 ppm representing free (non-complexed) phosphate-related species (Fig. 1). The area under the 31 P resonance in each of the spectra from vaccine samples was compared to the calibration graph shown in Fig. 2. The linearity of this graph indicates that free phosphate-related species in solution can easily be quantified using NMR. The 27 Al NMR spectrum of the same sample of vaccine shown in the supplemental Figure S1, the aluminum signal was not detected at physiological pH. Similarly, the 31 P NMR spectrum of the bound phosphate content is shown in Figure S2. This finding supports the expectation that all aluminum present is complexed with the antigen. In this non-symmetrical environment the aluminum undergoes rapid relaxation leading to extremely broad resonances which cannot be detected. The second analysis phase required adjustment of the sample to a suitable pH followed by measurement of total 27 Al and 31 P. Upon lowering the pH of the AlPO4 solutions, both 27 Al and complexed phosphate-related species were released in solution and formed a set of diverse equilibria involving aluminum, water and phosphate-related species. As shown in Fig. 3, these aluminum and phosphate-related complexes were detectable via the 27 Al and 31 P resonances at low pH, ranging between pH 3.14 and 0.61. As expected, the 27 Al spectrum does not present any resonances between pH 7.15 and 4.11, but at pH 0.8, resonances were detected between -0.56 and -8.22 ppm. The 31 P spectra indicate the existence of a phosphate-related resonance between pH 7.15 and 3.14. At lower pH values the resonances representing more complex Al- and phosphate-related species emerge and disappear as the pH changed. At pH 0.8 the 31 P resonances appeared at 0.56 and
Fig. 2. Calibrations for the determination of 31 P not associated with Quadracel® vaccine at neutral pH. Graph A indicates a concentration range between 0.5–3.0 × 10−3 M (R2 = 0.986). Graph B indicates a larger range from 0.5 to 40 × 10−3 (R2 = 0.998).
from -7.01 to -7.67 ppm). The largest total integrated resonance area for both 27 Al and 31 P spectra were measured at pH 0.80 indicating abundant NMR detectable Al and P-related species (Fig. 3). Similar experiments were performed for vaccine samples and NMR detectable resonance peak area for Al and P-related species were similarly found to be highest at pH 0.8 (data not shown). In both calibration and vaccine samples, the similarity of peak shifts indicated that the same species of aluminum and phosphate were released
R. Khatun et al. / Journal of Pharmaceutical and Biomedical Analysis 159 (2018) 166–172
169
Fig. 3. NMR spectra from 30 × 10−3 M AlPO4 solution at various pH values. Each adjacent spectrum of a series is offset slightly to give a perspective view of the resonances.
from the AlPO4 suspension into solution in the acidified vaccine samples. The possibility exists that, even at pH 0.8, there are still aggregates of aluminum and phosphate that do not satisfy the symmetry requirements to be observed by 27 Al NMR. However, we propose that by creating calibration samples using the same stoichiometry of total aluminum and phosphate, the solution equilibrium between aluminum and phosphate has been adequately reproduced. Therefore the amount of aluminum observed in the vaccine sample adequately mimics the calibration standard. The identity of the different forms of aluminum associated with H2 O, H2 PO4 − and H3 PO4 groups were assigned by comparing with the previous studies by Mortlock et al. [22,23]. As indicated in Fig. 4A, the resonances representing the 27 Al complexes present in vaccine samples at low pH are consistent with the [Al(H2 O)6 ]3+ and [Al(H2 O)5 (H3 PO4 )n ]m+ complexes seen at higher concentrations (Fig. 4B). The area under both of these resonances was measured and the sum of the two areas was used to calculate total 27 Al. Since the 31 P nucleus is spin I = 1/2, all nuclei from small molecules or part of small associations or complexes, regardless of symmetry, should be visible as either sharp or somewhat broadened lines. The total peak area obtained from all of the resonances directly correlated to the amount of 31 P present in solution and the summation of these areas gave the total 31 P content. The use of the quadrupolar aluminum nucleus for quantitative analysis has two inherent issues. For all 27 Al nuclei in solution to be NMR visible and quantitative required that the complexes establish a minimal threshold of symmetry about the 27 Al nucleus. Secondly, the effect of Al:P ratio upon the line shape of 27 Al must be evaluated in order to determine if an unfavorable ratio could create a 27 Al and 31 P complex exhibiting insufficient symmetry to be observed by 27 Al NMR. In other words, for the 27 Al and 31 P nuclei to be quantitative, all resonances must be sufficiently narrow to be measurable. While the initial in-house preparation of AlPO4 adjuvant in most manufacturing processes begins with approximately 0.86:1 of the anion and cation, this stoichiometry is not
necessarily maintained throughout the complete vaccine formulation process due to mixing with protein components containing phosphate buffer. The introduction or removal of either excess aluminum or phosphate-related species during the manufacturing process may cause variability in the final concentrations of 27 Al and 31 P which, in turn, may cause the stoichiometry to deviate from the initial adjuvant stoichiometry. In order to properly evaluate the relationship of the calibration of the total 31 P and 27 Al present in the sample, a series of spectra were acquired with variable stoichiometric ratios of aluminum and phosphate ranging from 15 to 40 × 10−3 M for each component. The calibration curves for both 1:1 ratios of 27 Al:31 P, as well as constant 31 P and variable 27 Al, are shown in Figs. 5A and 5B. Each of the calibrations shows excellent linearity over the range. In Fig. 5B, the 27 Al area remains linear even though the stoichiometry deviates significantly from equimolar solutions. This suggests that all complexes formed, retain sufficient symmetry to permit accurate quantitation of the 27 Al resonances, and that the total area under all of the 27 Al resonances in a spectrum is representative of the amounts of 27 Al present in the solution. A similar analysis was performed using 31 P NMR to study the area under the 31 P resonances. Fig. 6 shows the phosphate calibration for 1:1 27 Al:31 P stoichiometry as well as variable 31 P concentration with a constant 27 Al concentration of 35 × 10−3 M (Figs. 6A and 6B). Both graphs show very good linearity, indicating that the sum of the 31 P peak area in each spectrum represents the total 31 P in solution. The limit of detection (LOD) [24] was determined using solutions of AlPO4 at pH 0.8. The signal was evaluated over a 2 ppm chemical shift range centered at the signal. The noise was evaluated using a similar chemical shift range in a region without NMR signal. Using the same acquisition parameters recorded for calibration samples, the aluminum LOD for the largest resonance was evaluated at 0.01 × 10−3 M with a signal to noise ratio (S/N) of 5.8:1. For the largest resonance in the 31 P spectrum, the LOD was 0.05 × 10−3 M with a S/N of 2.8:1 (Table S2). This range may be extended with
170
R. Khatun et al. / Journal of Pharmaceutical and Biomedical Analysis 159 (2018) 166–172
Fig. 4. NMR spectra of solutions of AlPO4 at pH 0.79-0.81. Spectra A and C show, respectively, 27 Al and 31 P NMR spectra of typical vaccine solutions of 30 × 10−3 M AlPO4 . For reference purposes, spectra B and D show, respectively, 27 Al and 31 P NMR spectra of more concentrated solutions prepared from 50 × 10−3 M Al3+ with 70 × 10−3 M PO4 3- . Assignments are adopted from references [17,18].
Fig. 5. Calibrations for the determination of total 27 Al present in Quadracel® solution at pH 0.80. 27 Al calibrations at pH 0.80 show: Graph A 27 Al:31 P ratio of 1:1; Graph B variable aluminum with constant 31 P concentration of 35 × 10−3 M. In both calibrations R2 = 0.99.
the acquisition of additional scans. By application of limit of quantitation (LOQ) values used for 1 H NMR, the LOQ may be between 3.3–30 x LOD [25]. The precision of the NMR technique was evaluated by acquiring 10 successive 27 Al and 31 P spectra and comparing the integration values for the peak area (Table S3). The standard deviations of the total peak areas were 0.2% and 0.7% for aluminum and phosphorous spectra respectively. After determining linearity of calibrations for phosphate-related species and 27 Al, Quadracel® vaccine and several adsorbed protein antigens (which are components of several combination vaccine products, including Quadracel® ) were analyzed for free and total 31 P as well as total 27 Al content. The peak area values for 31 P at pH 7.0 and 27 Al and 31 P at low pH were evaluated using the appropriate calibrations from Figs. 4 and 5. These values are shown in
Fig. 6. Calibrations for the determination of total 31 P present in Quadracel® solution at pH 0.80. 31 P calibrations at pH 0.80 show: Graph A 27 Al:31 P ratio of 1:1; Graph B variable phosphate with constant 27 Al concentration of 35 × 10−3 M. In both calibrations R2 = 0.99.
Table 1 and indicate that the Quadracel® vaccine uniformly contained low amounts of free phosphate (approximately 2.7 × 10−3 M), while free phosphate was not detectible in two samples of adsorbed Diphtheria toxoid drug substance. The total 27 Al content in the Quadracel vaccine ranged between 22 × 10−3 to 26 × 10−3 M in each lot, whereas the concentrations for the same lots determined by the in-house colorimetric method were 24 × 10−3 M. For further verification, two Quadracel® lots were spiked with known amounts of AlPO4 solution. The additional verification from the spiked samples confirmed that the aluminum concentration as determined by NMR was within 1% of the original value (results not shown). The determination of the total 31 P indicated that the concentration range for Quadracel® was between 25 × 10−3 to 27 × 10−3 M. The corroborating verification from sample spiking confirmed the 31 P concentration (Table 1).
R. Khatun et al. / Journal of Pharmaceutical and Biomedical Analysis 159 (2018) 166–172
171
Table 1 Determination of 31 P concentrations at physiological pH and total 27 Al and 31 P concentration at pH 0.80 for different vaccine samples. sample
[PO4 ] pH 7.0(M x 10−3 )
Total [Al](M x 10−3 )
Total [PO4 ](M x 10−3 )
Quadracel Lot 1 Quadracel Lot 2 Quadracel Lot 3 Quadracel Lot 4 Quadracel Lot 5 Quadracel Lot 6 Quadracel Lot 7a Quadracel Lot 8a Diphtheria Toxoid Lot 1b Diphtheria Toxoid Lot 2b Diphtheria Toxoid Lot 3 Tetanus Toxoid Lot 1 Tetanus Toxoid Lot 2 Filamentous Haemagglutinin Lot 1 Filamentous Haemagglutinin Lot 2 Pertussis Toxoid Lot 1 Pertussis Toxoid Lot 2 Fimbriae Lot 1 Fimbriae Lot 2 Pertactin Lot 1 Pertactin Lot 2
2.7 2.9 2.7 2.9 2.9 2.8 2.7 2.4 nd nd nd 1.8 1.4 6.4 6.0 6.8 7.1 7.1 6.8 6.9 6.0
25.91 24.51 23.84 23.92 24.17 21.840 23.87 (23.5)a 22.74 (22.9)a 51.54 46.64 53.81 53.52 54.67 21.35 23.72 20.43 25.99 27.28 24.74 20.71 25.21
25.9 26.16 25.18 26.29 27.8 23.0 27.2 (26.5)a 26.8 (26.9)a 44.61 40.52 39.27 48.90 40.68 28.68 24.27 26.61 27.21 33.61 26.05 26.52 25.19
nd – not detected. a 5mM of AlPO4 was added to the sample to verify the concentration. The value in brackets indicates the original concentration calculated from the spiked sample. b These samples were diluted by 1/2 prior to analysis.
The Al3+ concentrations in other drug substances were also measured and are presented in Table 1. These ranges were between 20 × 10−3 to 26 × 10−3 M, for Pertussis Toxoid, Pertactin, Filamentous Haemagglutinin, and Fimbriae, respectively. However, Diphtheria Toxoid and Tetanus Toxoid, showed higher Al3+ concentrations (47 × 10−3 to 54 × 10−3 M), as these drug substances contain twice the amount of protein content compared to those discussed above. Free phosphate concentrations were in the range of 2 × 10−3 to 7 × 10−3 M and total phosphate concentrations were in the range of 27 × 10−3 to 49 × 10−3 M (Table 1). It was noted that free phosphate was not detectable in two Diphtheria Toxoid samples as they were manufactured predominantly in NaCl solution without phosphate buffer. In these Diphtheria Toxoid samples, the total phosphate detected is therefore part of the vaccine adjuvant complex at neutral pH. Total Al and P concentration in the drug substances was also verified through the ICP-OES method that reported similar ranges of Al and P (Table S1). Comparison of the total aluminum and phosphorous values obtained from the two methods indicates that the accuracy between NMR and ICP-OES was good between Diphtheria Toxoid Adsorbed-2 and Filamentous Haemagglutinin Adsorbed vaccine for aluminum analysis. Agreement was also good between Diphtheria Toxoid Adsorbed-2, Pertussis Toxoid Adsorbed and Pertactin Adsorbed for phosphorous analysis. Taken together, these findings suggest that solution qNMR is suitable method to accurately quantify both Al and P in complex biological samples, such as drug substance and vaccine drug product using a single sample preparation step and the same workflow. Overall, the agreement in total aluminum concentrations was comparable between the qNMR and the validated in-house colorimetric method. Further optimizations to this method could be explored by further investigating the AlPO4 solution equilibria for each vaccine or by collecting data as driven acquisitions with reduced relaxation delays initiated after a steady state has been established. 4. Conclusion An instrumental qNMR method has been developed that is reproducible and suitable for analysis of adjuvant components 27 Al and 31 P in complex sample matrices, specifically AlPO4 -adjuvanted
vaccine formulations. Sample acidification was employed to allow determination of total Al and P concentration in several different samples and product formulations. Measurement of free phosphate concentration is achieved at neutral pH and is an expression of the amount of P that is not physically associated with either the insoluble adjuvant particles or with the protein antigens. Along with total Al and P, the ratio of free to total phosphate may have an impact on the colloidal properties of the adjuvant and/or its antigen adsorption capacity, and is thus a potential product quality attribute. The NMR method provides a faster and simpler alternative to traditional wet chemistry methods for Al content and additionally, can assess P content in a single workflow. The suitability of the method to different AlPO4 -adjuvanted formulations has the potential to broaden its applicability to development and monitoring of novel adjuvant and vaccine manufacturing processes. Although the analyses were performed using high field NMR spectroscopy, the smaller frequency ranges provided by lower magnetic fields, which can be provided by simpler NMR instruments at lower cost, are expected to achieve acceptable linearity for quantitation. As modern NMR spectroscopy is primarily a Fourier transform technique, sufficient sensitivity can be achieved by increasing the number of scans.
Compliance with ethical standards No animals were used in this study. No human or animal samples were used in this study.
Conflict of interest Bruce Carpick and Marina Kirkitadze are employees of Sanofi Pasteur Ltd. Rahima Khatun, Yi Sheng, and Howard Hunter are employees of York University. The authors have no other relevant affiliations or financial involvement with any organization or entity with a financial interest in or financial conflict with the subject matter or materials discussed in the manuscript. Thus includes employment, consultancies, stock ownership or options, or royalties. No writing assistance was utilized in the production of this manuscript.
172
R. Khatun et al. / Journal of Pharmaceutical and Biomedical Analysis 159 (2018) 166–172
Acknowledgements The authors would like to thank the Governments of Ontario and Canada for providing assisting funds through a MITACS grant IT08656. This research was partially funded by Sanofi Pasteur Ltd. The authors of this paper wish to extend their utmost gratitude to Rozalia Nisman and Tristan Ho, QC Chemistry Sanofi Pasteur Ltd for discussion and to Juthika Menon, Liliana Sampaleanu, and Danielle Johnson, Manufacturing Technology, Sanofi Pasteur Ltd for providing samples for testing, and to Georgiana Moldoveanu, University of Toronto for support with ICP-OES experiments. The authors are also grateful to the suggestions provided by the reviewers of the manuscript. Appendix A. Supplementary data Supplementary material related to this article can be found, in the online version, at doi:https://doi.org/10.1016/j.jpba.2018.06. 025. References [1] R.K. Gupta, Aluminum compounds as vaccine adjuvants, Adv. Drug Del. Rev. 32 (1998) 155–172. [2] Y. Wen, Y. Shi, Alum: an old dog with new tricks, Emerg. Microbes Infect. 5 (2016) e25. [3] P. He, Y. Zou, Z. Hu, Advances in aluminum hydroxide-based adjuvant research and its mechanism, Hum. Vaccin. Immunother. 11 (2015) 477–488. [4] C.N. Maughan, S.G. Preston, G.R. Williams, Particulate inorganic adjuvants: recent developments and future outlook, J. Pharm. Pharmacol. 67 (2015) 426–449. [5] European Pharmacopeia, Aluminum in Adsorbed Vaccines, Chapter 2.5.13., 9th ed., 2018, pp. 165. [6] A. Mishra, S.R. Bhalla, S. Rawat, V. Bansal, R. Sehgal, S. Kumar, Standardization and validation of a new atomic absorption spectroscopy technique for determination and quantitation of aluminum adjuvant in immunobiologicals, Biologicals 35 (2007) 277–284. [7] T. Clapp, P. Siebert, D. Chen, L.-T. Jones Braun, Vaccines with aluminum-containing adjuvants: optimizing vaccine efficacy and thermal stability, J. Pharm. Sci. 100 (2011) 388–401. [8] F. Medera, T. Daberkow, L. Treccani, M. Wilhelm, M. Schowalter, A. Rosenauer, L. Mädlerc, K. Rezwana, Protein adsorption on colloidal alumina particles functionalized with amino, carboxyl, sulfonate and phosphate groups, Acta Biomater. 8 (2012) 121–1229.
[9] European Pharmacopeia, Aluminum Phosphate Gel, 2166, 9th ed., 2018, pp. 165 http://online6.edqm.eu/ep904/. [10] European Pharmacopeia, Aluminum Phosphate Gel, Hydrated, 1598, 9th ed., 2018, pp. 1686. [11] U. Holzgrabe, I. Wawer, B. Diehl, NMR Spectroscopy in Pharmaceutical Analysis, 1st ed., Elsevier, Oxford, 2008. [12] R. Khatun, H.N. Hunter, W. Magcalas, Y. Sheng, B. Carpick, M. Kirkitadze, Nuclear magnetic resonance (NMR) study for the detection and quantitation of cholesterol in HSV529 therapeutic vaccine candidate, Comput. Struct. Biotechnol. J. 15 (14) (2017). [13] C. Jones, NMR assays for carbohydrate-based vaccines, J. Pharm. Biomed. Anal. 38 (2005) 840–850. [14] C. Simmler, J.G. Napolitano, J.B. McAlpine, S.-N. Chen, G.F. Pauli, Universal quantitative NMR analysis of complex natural samples, Curr. Opin. Biotechnol. 25 (2014) 51–59. [15] G.A. Barding Jr, R. Salditos, C.K. Larive, Quantitative NMR for bioanalysis and metabolomics, Anal. Bioanal. Chem. 404 (2012) 1165–1179. [16] U. Holzgrabe, C. Deubner, C. Schollmayer, B. Waibel, Quantitative NMR spectroscopy-applications in drug analysis, J. Pharm. Biomed. Anal. 38 (2005) 806–812. [17] M. Weber, C. Hellriegel, A. Rueck, J. Wuethrich, P. Jenks, Obkircher, Method development in quantitative NMR towards metrologically traceable organic certified reference materials used as 31 P qNMR standards, Anal. Bioanal. Chem. 407 (2015) 3115–3123. [18] R. Martino, V. Gilard, F. Desmoulin, M. Malet-Martino, Fluorine-19 or phosphorus-31 NMR spectroscopy: a suitable analytical technique for quantitative in vitro metabolic studies of fluorinated or phosphorylated drugs, J. Pharm. Biomed. Anal. 38 (2005) 871–891. [19] S.J. Karlik, G.A. Elgavish, R.P. Pillai, G.L. Eichhorn, Aluminum-27 NMR studies of Al(III)-phosphate complexes in aqueous solution, J. Magn. Reson. 49 (1982) 164–167. [20] M.A. Curtin, L.R. Ingalls, A. Campbell, M. James-Pederson, Hydrolysis studies and quantitative determination of aluminum ions using 27 Al NMR, J. Chem. Educ. 85 (2008) 291–293. [21] J.W. Akitt, Nuclear magnetic resonance spectroscopy in liquids containing compounds of aluminum and gallium, Ann. Rep. NMR Spectrosc. 5A (1972) 465–559. [22] H. Maki, G. Sakata, M. Mizuhata, Quantitative NMR of quadrupolar nucleus as a novel analytical method: hydrolysis behaviour analysis of aluminum ion, Analyst 142 (2017) 1790–1799. [23] R.F. Mortlock, A.T. Bell, C.J. Radke, 31 P and 27 Al NMR investigations of highly acidic, aqueous solutions containing aluminum and phosphorous, J. Phys. Chem. 97 (1993) 767–774. [24] R.F. Mortlock, A.T. Bell, C.J. Radke, 31 P and 27 Al NMR investigations of the effect of pH on aqueous solutions containing aluminum and phosphorous, J. Phys. Chem. 97 (1993) 775–782. [25] Validation of Analytical Procedures:Text and Methodology. International Conference on Harmonization of Technical Requirements for Registration of Pharmaceuticals for Human Use, 2005.
Contents lists available at ScienceDirect
Journal of Pharmaceutical and Biomedical Analysis journal homepage: www.elsevier.com/locate/jpba
27
Al and 31 P NMR spectroscopy method development to quantify aluminum phosphate in adjuvanted vaccine formulations Rahima Khatun a,1 , Howard N. Hunter b,1 , Yi Sheng a , Bruce W. Carpick c , Marina D. Kirkitadze c,∗ a
Departments of Biology and Chemistry, York University, 4700 Keele Street, Toronto, Ontario, Canada York University, 4700 Keele Street, Toronto, Ontario, Canada c Analytical Research and Development, Sanofi Pasteur Ltd., 1755 Steeles Avenue West, Toronto, Ontario, Canada b
a r t i c l e
i n f o
Article history: Received 23 February 2018 Received in revised form 12 June 2018 Accepted 16 June 2018 Available online 21 June 2018 Keywords: Quantitative qNMR Aluminum Phosphate Adjuvant Vaccine
a b s t r a c t A novel qNMR method is described for the quantitative determination of total aluminum and phosphate in aluminum phosphate (AlPO4 ) adjuvanted vaccine samples using solution 27 Al and 31 P nuclear magnetic resonance (NMR) spectroscopy. External standard calibrations of AlPO4 solutions established excellent linearity in the range of 15–40 × 10−3 M and additional studies determined the level of detection for both nuclei. A commercialized combination vaccine product (Quadracel® ), along with several individual adsorbed antigen components used in the vaccine were employed as model systems for method development. The developed method is also capable of quantitating the free phosphate (i.e. the fraction not bound to AlPO4 particles) in adjuvanted vaccines. This study is the first demonstration of a solution NMR method that is suitable for measuring total aluminum, and free and total phosphate concentrations in vaccine formulations consisting of antigen(s) adsorbed to aluminum adjuvant, in a single analytical workflow. © 2018 The Authors. Published by Elsevier B.V. This is an open access article under the CC BY license (http://creativecommons.org/licenses/by/4.0/).
1. Introduction Adjuvants are substances added to vaccine antigens in order to enhance the immune response. Aluminum salts such as aluminum hydroxide (AlOOH) or aluminum phosphate (AlPO4 ) are the most prevalent adjuvants in commercial vaccines due to their excellent safety profiles and ability to induce robust immunity against pathogens [1]. Although the precise mechanism of action for aluminum adjuvants is not known [2,3], the physical interaction of the adjuvant with the vaccine antigens, typically termed adsorption, may play a role in immunogenicity. Antigen-adjuvant adsorption can occur through different mechanisms, such as electrostatic interactions or ligand exchange [4]. Regardless of the mechanism, the interaction of the antigen with the adjuvant surface in the vaccine formulation is thought to lead to an enhanced immune response, and is furthermore a potential marker of vaccine quality and consistency. On a molecular level, it has been shown that the aluminum and phosphate content of AlPO4 are important for the surface charge and can therefore have an impact on the antigen adsorption capac-
∗ Corresponding author. E-mail address: Marina.Kirkitadze@sanofi.com (M.D. Kirkitadze). 1 Authors contributed equally.
ity [5]. Furthermore, the colloidal stability of the AlPO4 adjuvant suspension can be affected by the components of vaccine matrix, e.g., by the excess of free phosphate [5]. Therefore, it is important to measure both free and total phosphate in vaccine formulations to ensure that surface charge remains stable and, in turn, does not alter absorption capacity of protein antigens to AlPO4 adjuvant [6]. As part of a batch release of vaccine, typically only the aluminum content is measured and this quantitation is completed using a colorimetric method involving the chelation with EDTA, followed by back titration with cupric sulfate [7]. Unfortunately, the European Pharmacopoeia monographs describe two different methods for assessing both free and bound phosphate [8,9]. Clearly, it would be quite convenient and advantageous to have a single, rapid method to measure these three characteristic components of adjuvant content during development, optimization, and monitoring of the vaccine formulation process [5]. A method which can quantitate all three components in vaccines, with minimal sample preparation, offers the potential to streamline the manufacturing process development and optimization for aluminum-adjuvanted vaccines. Advancements in technology and expanding NMR applications have continued to demonstrate the utility of NMR in the pharmaceutical industry. These applications range from the design of innovative drugs to the characterization of polymers, natural products, and metabolites [10]. Quantitative NMR (qNMR) continues to
https://doi.org/10.1016/j.jpba.2018.06.025 0731-7085/© 2018 The Authors. Published by Elsevier B.V. This is an open access article under the CC BY license (http://creativecommons.org/licenses/by/4.0/).
R. Khatun et al. / Journal of Pharmaceutical and Biomedical Analysis 159 (2018) 166–172
gain popularity as a method for rapid, analytical determination of drug metabolites, capsular polysaccharides and residuals in complex solutions, including vaccines [11–14]. The utility of NMR for such application relates to its unique ability to identify and quantify various nuclei in solution with minimal interference from the surrounding solvent matrix, making it well suited for monitoring complex samples and manufacturing processes. Most often, qNMR analysis has focused on the measurement of protons (1 H), however, other nuclei such as 31 P and 19 F have been utilized for quantitative metabolic studies [15–17]. The favorable 100% natural abundance of the 27 Al and 31 P isotopes are an initial indicator that Al and P are ideal targets for qNMR, since both nuclei have a high receptivity similar to that of a proton. However, 27 Al is a quadrupolar nucleus (spin 5/2) and the non-insignificant quadrupole moment of aluminum can dominate nuclear T1 and T2 relaxation resulting in extremely broad, or more often, unobservable resonances in the 27 Al spectrum [18,19]. The effect of the relaxation contribution from the electric field gradient about the 27 Al nucleus can be minimized by a coordinating environment for aluminum which results in higher symmetry. This coordination leads to longer T1 and T2 relaxation resulting in narrower, and more importantly, quantifiable NMR resonances [20]. Under these conditions, quantitation of aluminum by NMR has been shown by Maki et al to be feasible [21]. In a typical vaccine formulation, the high molecular weight of the active component (such as a protein antigen) far exceeds the size limit for the utility of high resolution NMR in solution. Under these conditions, spin diffusion becomes the predominant method of relaxation for any NMR active nucleus associated with the protein, and any NMR spectra obtained would be devoid of sharp resonances. Therefore, the detection of aluminum and phosphorous can only take place through release of phosphate and aluminum ions from the AlPO4 adjuvant suspension solid phase into the solution. As the pH is lowered, phosphate and aluminum ions are released into the solution and phosphate establishes equilibria between its PO4 3− , HPO4 2- , H2 PO4 - and H3 PO4 forms. In turn, the phosphate ions can combine with aluminum, increasing the complexity of the aqueous matrix. However, upon further acidification, the solution matrix can be shifted to smaller and more symmetrical aluminum-phosphate-water complexes as determined by Mortlock et al [22,23]. Although the 27 Al and 31 P spectra may show multiple resonances attributable to an ensemble of aluminum and phosphate complexes in solution, the total 27 Al or 31 P present can be correlated to the sum of the area under all 27 Al or 31 P resonances. The total area of these resonances can therefore be used to calculate the 27 Al or 31 P content. Here, for the first time, we present a novel qNMR method for the analysis of total aluminum and phosphorous in formulated vaccine products. The method was applied to a commercialized adjuvanted combination vaccine product, Quadracel® , along with several individual adsorbed antigen drug substances used to manufacture the combination vaccine. This method requires minimal sample preparation, provides rapid, quantitative analysis, and is amenable to routine monitoring of vaccine adjuvant and formulation processes.
2. Materials and methods 2.1. Materials Aluminum nitrate (Al(NO3 )3 ·9H2 O) (solid), 36% hydrochloric acid and sodium hydroxide solutions were purchased from Caledon Laboratories Ltd. Georgetown, Canada. Sodium phosphate monobasic dehydrate (NaH2 PO4 ) and deuterium oxide were purchased from Sigma-Aldrich (Oakville, Ontario and Milwaukee, WI). Quadracel® , an AlPO4 -adjuvanted combination vaccine product, consisting of adsorbed Diphtheria and Tetanus toxoids,
167
adsorbed acellular Pertussis antigens (Pertussis toxoid, Filamentous Haemagglutinin, Pertactin, and Fimbriae), and inactivated Poliomyelitis vaccine, along with individual adsorbed Diphtheria toxoid, Tetanus toxoid, and acellular Pertussis adsorbed antigen drug substances, were produced by Sanofi Pasteur Limited (Toronto, Ontario). 2.2. Sample preparation Calibration solutions were prepared by combining appropriate amounts of Al(NO3 )3 ·9H2 O and NaH2 PO4 stock solutions and adjusting the pH to either 7.0 or 0.79 to 0.81 with either HCl or NaOH solutions. Additional water and deuterium oxide were added to create a final volume of 2 mL of solution consisting of the desired concentration of 27 Al and 31 P along with 10% deuterium oxide for field-frequency lock. Exactly 0.6 mL of this solution was placed in an NMR tube for measurement. Quadracel vaccine analysis involved the combination of a 1.6 mL sample of vaccine with 177 L D2 O (10%). A 0.6 mL aliquot from this solution was placed in a NMR tube for 31 P NMR acquisition. Following data collection, the test sample was then returned to the stock solution and the pH was adjusted to 0.79-0.81 using concentrated HCl. A 0.6 mL aliquot of this solution was placed in a NMR tube and both the 27 Al and 31 P NMR spectra were acquired. A similar procedure was used, beginning with smaller initial volumes, for the other vaccine samples tested. All data were corrected for dilution factors. One of the Quadracel samples was spiked with 5 mM AlPO4 following initial analysis. The 27 Al and 31 P spectra were re-acquired and the increase in the resonance area of the spiked sample was compared to the original sample peak area to confirm the concentration of total 27 Al and total 31 P. 2.3. NMR measurements All NMR measurements were performed at 21 ◦ C on a DRX 600 NMR spectrometer operating with XWINNMR software version 3.5 (Bruker BioSpin Ltd., Karlsruhe, Germany) at 600.00 MHz using a 5 mm BBO probe. The T1 relaxation values for both 27 Al and 31 P nuclei were determined using the inversion-recovery method and the subsequent 1D spectra were acquired with 5 times the longest T1 value measured. The 27 Al spectra were acquired with 64 K data points, 128 scans, a 1 s relaxation delay and processed with 2 Hz line broadening function. At neutral pH the 31 P spectra were acquired with 64 K data points, 64 scans, and a 35 s relaxation delay. The 31 P spectra of acidified samples were acquired with 64 K data points, 128 scans, and a 25 s relaxation delay. A constant receiver gain value was used for each nucleus and was initially determined using the most concentrated calibration sample. The 27 Al and 31 P spectra were externally referenced to Al(NO3 )3 and 85% H3 PO4 respectively. All spectra were processed using TopSpin version 3.5.b.91 pl7 software (Bruker BioSpin Ltd., Karlsruhe, Germany) with 2 Hz line broadening function and baseline correction using a polynomial fit. For each of the spectra within a grouping (i.e. neutral pH phosphate, total aluminum or total phosphate), peak areas were standardized by integrating one of the calibration samples and then using the lastscale feature of the integration module of the software to permit comparison. With this feature peak areas from an initial data set are stored with a fixed digitally assigned value which can be recalled and directly applied to any subsequent data set. 3. Results and discussion Previous NMR studies showed that various aluminumphosphate complexes are formed in aqueous solution containing aluminum and phosphorous and that narrow resonance lines and the chemical shifts strictly depend on the pH of the solution
168
R. Khatun et al. / Journal of Pharmaceutical and Biomedical Analysis 159 (2018) 166–172
Fig. 1. 1D 31 P NMR spectrum of Quadracel® vaccine sample with 10% D2 O at neutral pH. The spectrum was acquired at 242.88 MHz with 64 scans, 35 s relaxation delay and externally referenced to 85% H3 PO4 .
[21–23]. Furthermore, 31 P NMR resonances of these aluminumphosphate complexes were successfully assigned to P atom at various pH values, while the detection and assignment of 27 Al resonances were only possible at low pH. It therefore seemed reasonable that for total concentration of aluminum and phosphate, the summation of the area under all of the resonances in the respective spectra could be evaluated. As 31 P NMR resonances for free phosphate in solution can readily be assigned at pH > 6.0 [22,23], a two-step analysis of vaccine samples was followed whereby the phosphate content was measured at neutral pH to assess free phosphate followed by adjusting the pH to acidic values to permit the measurement of total aluminum and total phosphate. The 31 P spectrum from Quadracel® vaccine at pH 7.0 shows a single sharp resonance in solution at 2.4 ppm representing free (non-complexed) phosphate-related species (Fig. 1). The area under the 31 P resonance in each of the spectra from vaccine samples was compared to the calibration graph shown in Fig. 2. The linearity of this graph indicates that free phosphate-related species in solution can easily be quantified using NMR. The 27 Al NMR spectrum of the same sample of vaccine shown in the supplemental Figure S1, the aluminum signal was not detected at physiological pH. Similarly, the 31 P NMR spectrum of the bound phosphate content is shown in Figure S2. This finding supports the expectation that all aluminum present is complexed with the antigen. In this non-symmetrical environment the aluminum undergoes rapid relaxation leading to extremely broad resonances which cannot be detected. The second analysis phase required adjustment of the sample to a suitable pH followed by measurement of total 27 Al and 31 P. Upon lowering the pH of the AlPO4 solutions, both 27 Al and complexed phosphate-related species were released in solution and formed a set of diverse equilibria involving aluminum, water and phosphate-related species. As shown in Fig. 3, these aluminum and phosphate-related complexes were detectable via the 27 Al and 31 P resonances at low pH, ranging between pH 3.14 and 0.61. As expected, the 27 Al spectrum does not present any resonances between pH 7.15 and 4.11, but at pH 0.8, resonances were detected between -0.56 and -8.22 ppm. The 31 P spectra indicate the existence of a phosphate-related resonance between pH 7.15 and 3.14. At lower pH values the resonances representing more complex Al- and phosphate-related species emerge and disappear as the pH changed. At pH 0.8 the 31 P resonances appeared at 0.56 and
Fig. 2. Calibrations for the determination of 31 P not associated with Quadracel® vaccine at neutral pH. Graph A indicates a concentration range between 0.5–3.0 × 10−3 M (R2 = 0.986). Graph B indicates a larger range from 0.5 to 40 × 10−3 (R2 = 0.998).
from -7.01 to -7.67 ppm). The largest total integrated resonance area for both 27 Al and 31 P spectra were measured at pH 0.80 indicating abundant NMR detectable Al and P-related species (Fig. 3). Similar experiments were performed for vaccine samples and NMR detectable resonance peak area for Al and P-related species were similarly found to be highest at pH 0.8 (data not shown). In both calibration and vaccine samples, the similarity of peak shifts indicated that the same species of aluminum and phosphate were released
R. Khatun et al. / Journal of Pharmaceutical and Biomedical Analysis 159 (2018) 166–172
169
Fig. 3. NMR spectra from 30 × 10−3 M AlPO4 solution at various pH values. Each adjacent spectrum of a series is offset slightly to give a perspective view of the resonances.
from the AlPO4 suspension into solution in the acidified vaccine samples. The possibility exists that, even at pH 0.8, there are still aggregates of aluminum and phosphate that do not satisfy the symmetry requirements to be observed by 27 Al NMR. However, we propose that by creating calibration samples using the same stoichiometry of total aluminum and phosphate, the solution equilibrium between aluminum and phosphate has been adequately reproduced. Therefore the amount of aluminum observed in the vaccine sample adequately mimics the calibration standard. The identity of the different forms of aluminum associated with H2 O, H2 PO4 − and H3 PO4 groups were assigned by comparing with the previous studies by Mortlock et al. [22,23]. As indicated in Fig. 4A, the resonances representing the 27 Al complexes present in vaccine samples at low pH are consistent with the [Al(H2 O)6 ]3+ and [Al(H2 O)5 (H3 PO4 )n ]m+ complexes seen at higher concentrations (Fig. 4B). The area under both of these resonances was measured and the sum of the two areas was used to calculate total 27 Al. Since the 31 P nucleus is spin I = 1/2, all nuclei from small molecules or part of small associations or complexes, regardless of symmetry, should be visible as either sharp or somewhat broadened lines. The total peak area obtained from all of the resonances directly correlated to the amount of 31 P present in solution and the summation of these areas gave the total 31 P content. The use of the quadrupolar aluminum nucleus for quantitative analysis has two inherent issues. For all 27 Al nuclei in solution to be NMR visible and quantitative required that the complexes establish a minimal threshold of symmetry about the 27 Al nucleus. Secondly, the effect of Al:P ratio upon the line shape of 27 Al must be evaluated in order to determine if an unfavorable ratio could create a 27 Al and 31 P complex exhibiting insufficient symmetry to be observed by 27 Al NMR. In other words, for the 27 Al and 31 P nuclei to be quantitative, all resonances must be sufficiently narrow to be measurable. While the initial in-house preparation of AlPO4 adjuvant in most manufacturing processes begins with approximately 0.86:1 of the anion and cation, this stoichiometry is not
necessarily maintained throughout the complete vaccine formulation process due to mixing with protein components containing phosphate buffer. The introduction or removal of either excess aluminum or phosphate-related species during the manufacturing process may cause variability in the final concentrations of 27 Al and 31 P which, in turn, may cause the stoichiometry to deviate from the initial adjuvant stoichiometry. In order to properly evaluate the relationship of the calibration of the total 31 P and 27 Al present in the sample, a series of spectra were acquired with variable stoichiometric ratios of aluminum and phosphate ranging from 15 to 40 × 10−3 M for each component. The calibration curves for both 1:1 ratios of 27 Al:31 P, as well as constant 31 P and variable 27 Al, are shown in Figs. 5A and 5B. Each of the calibrations shows excellent linearity over the range. In Fig. 5B, the 27 Al area remains linear even though the stoichiometry deviates significantly from equimolar solutions. This suggests that all complexes formed, retain sufficient symmetry to permit accurate quantitation of the 27 Al resonances, and that the total area under all of the 27 Al resonances in a spectrum is representative of the amounts of 27 Al present in the solution. A similar analysis was performed using 31 P NMR to study the area under the 31 P resonances. Fig. 6 shows the phosphate calibration for 1:1 27 Al:31 P stoichiometry as well as variable 31 P concentration with a constant 27 Al concentration of 35 × 10−3 M (Figs. 6A and 6B). Both graphs show very good linearity, indicating that the sum of the 31 P peak area in each spectrum represents the total 31 P in solution. The limit of detection (LOD) [24] was determined using solutions of AlPO4 at pH 0.8. The signal was evaluated over a 2 ppm chemical shift range centered at the signal. The noise was evaluated using a similar chemical shift range in a region without NMR signal. Using the same acquisition parameters recorded for calibration samples, the aluminum LOD for the largest resonance was evaluated at 0.01 × 10−3 M with a signal to noise ratio (S/N) of 5.8:1. For the largest resonance in the 31 P spectrum, the LOD was 0.05 × 10−3 M with a S/N of 2.8:1 (Table S2). This range may be extended with
170
R. Khatun et al. / Journal of Pharmaceutical and Biomedical Analysis 159 (2018) 166–172
Fig. 4. NMR spectra of solutions of AlPO4 at pH 0.79-0.81. Spectra A and C show, respectively, 27 Al and 31 P NMR spectra of typical vaccine solutions of 30 × 10−3 M AlPO4 . For reference purposes, spectra B and D show, respectively, 27 Al and 31 P NMR spectra of more concentrated solutions prepared from 50 × 10−3 M Al3+ with 70 × 10−3 M PO4 3- . Assignments are adopted from references [17,18].
Fig. 5. Calibrations for the determination of total 27 Al present in Quadracel® solution at pH 0.80. 27 Al calibrations at pH 0.80 show: Graph A 27 Al:31 P ratio of 1:1; Graph B variable aluminum with constant 31 P concentration of 35 × 10−3 M. In both calibrations R2 = 0.99.
the acquisition of additional scans. By application of limit of quantitation (LOQ) values used for 1 H NMR, the LOQ may be between 3.3–30 x LOD [25]. The precision of the NMR technique was evaluated by acquiring 10 successive 27 Al and 31 P spectra and comparing the integration values for the peak area (Table S3). The standard deviations of the total peak areas were 0.2% and 0.7% for aluminum and phosphorous spectra respectively. After determining linearity of calibrations for phosphate-related species and 27 Al, Quadracel® vaccine and several adsorbed protein antigens (which are components of several combination vaccine products, including Quadracel® ) were analyzed for free and total 31 P as well as total 27 Al content. The peak area values for 31 P at pH 7.0 and 27 Al and 31 P at low pH were evaluated using the appropriate calibrations from Figs. 4 and 5. These values are shown in
Fig. 6. Calibrations for the determination of total 31 P present in Quadracel® solution at pH 0.80. 31 P calibrations at pH 0.80 show: Graph A 27 Al:31 P ratio of 1:1; Graph B variable phosphate with constant 27 Al concentration of 35 × 10−3 M. In both calibrations R2 = 0.99.
Table 1 and indicate that the Quadracel® vaccine uniformly contained low amounts of free phosphate (approximately 2.7 × 10−3 M), while free phosphate was not detectible in two samples of adsorbed Diphtheria toxoid drug substance. The total 27 Al content in the Quadracel vaccine ranged between 22 × 10−3 to 26 × 10−3 M in each lot, whereas the concentrations for the same lots determined by the in-house colorimetric method were 24 × 10−3 M. For further verification, two Quadracel® lots were spiked with known amounts of AlPO4 solution. The additional verification from the spiked samples confirmed that the aluminum concentration as determined by NMR was within 1% of the original value (results not shown). The determination of the total 31 P indicated that the concentration range for Quadracel® was between 25 × 10−3 to 27 × 10−3 M. The corroborating verification from sample spiking confirmed the 31 P concentration (Table 1).
R. Khatun et al. / Journal of Pharmaceutical and Biomedical Analysis 159 (2018) 166–172
171
Table 1 Determination of 31 P concentrations at physiological pH and total 27 Al and 31 P concentration at pH 0.80 for different vaccine samples. sample
[PO4 ] pH 7.0(M x 10−3 )
Total [Al](M x 10−3 )
Total [PO4 ](M x 10−3 )
Quadracel Lot 1 Quadracel Lot 2 Quadracel Lot 3 Quadracel Lot 4 Quadracel Lot 5 Quadracel Lot 6 Quadracel Lot 7a Quadracel Lot 8a Diphtheria Toxoid Lot 1b Diphtheria Toxoid Lot 2b Diphtheria Toxoid Lot 3 Tetanus Toxoid Lot 1 Tetanus Toxoid Lot 2 Filamentous Haemagglutinin Lot 1 Filamentous Haemagglutinin Lot 2 Pertussis Toxoid Lot 1 Pertussis Toxoid Lot 2 Fimbriae Lot 1 Fimbriae Lot 2 Pertactin Lot 1 Pertactin Lot 2
2.7 2.9 2.7 2.9 2.9 2.8 2.7 2.4 nd nd nd 1.8 1.4 6.4 6.0 6.8 7.1 7.1 6.8 6.9 6.0
25.91 24.51 23.84 23.92 24.17 21.840 23.87 (23.5)a 22.74 (22.9)a 51.54 46.64 53.81 53.52 54.67 21.35 23.72 20.43 25.99 27.28 24.74 20.71 25.21
25.9 26.16 25.18 26.29 27.8 23.0 27.2 (26.5)a 26.8 (26.9)a 44.61 40.52 39.27 48.90 40.68 28.68 24.27 26.61 27.21 33.61 26.05 26.52 25.19
nd – not detected. a 5mM of AlPO4 was added to the sample to verify the concentration. The value in brackets indicates the original concentration calculated from the spiked sample. b These samples were diluted by 1/2 prior to analysis.
The Al3+ concentrations in other drug substances were also measured and are presented in Table 1. These ranges were between 20 × 10−3 to 26 × 10−3 M, for Pertussis Toxoid, Pertactin, Filamentous Haemagglutinin, and Fimbriae, respectively. However, Diphtheria Toxoid and Tetanus Toxoid, showed higher Al3+ concentrations (47 × 10−3 to 54 × 10−3 M), as these drug substances contain twice the amount of protein content compared to those discussed above. Free phosphate concentrations were in the range of 2 × 10−3 to 7 × 10−3 M and total phosphate concentrations were in the range of 27 × 10−3 to 49 × 10−3 M (Table 1). It was noted that free phosphate was not detectable in two Diphtheria Toxoid samples as they were manufactured predominantly in NaCl solution without phosphate buffer. In these Diphtheria Toxoid samples, the total phosphate detected is therefore part of the vaccine adjuvant complex at neutral pH. Total Al and P concentration in the drug substances was also verified through the ICP-OES method that reported similar ranges of Al and P (Table S1). Comparison of the total aluminum and phosphorous values obtained from the two methods indicates that the accuracy between NMR and ICP-OES was good between Diphtheria Toxoid Adsorbed-2 and Filamentous Haemagglutinin Adsorbed vaccine for aluminum analysis. Agreement was also good between Diphtheria Toxoid Adsorbed-2, Pertussis Toxoid Adsorbed and Pertactin Adsorbed for phosphorous analysis. Taken together, these findings suggest that solution qNMR is suitable method to accurately quantify both Al and P in complex biological samples, such as drug substance and vaccine drug product using a single sample preparation step and the same workflow. Overall, the agreement in total aluminum concentrations was comparable between the qNMR and the validated in-house colorimetric method. Further optimizations to this method could be explored by further investigating the AlPO4 solution equilibria for each vaccine or by collecting data as driven acquisitions with reduced relaxation delays initiated after a steady state has been established. 4. Conclusion An instrumental qNMR method has been developed that is reproducible and suitable for analysis of adjuvant components 27 Al and 31 P in complex sample matrices, specifically AlPO4 -adjuvanted
vaccine formulations. Sample acidification was employed to allow determination of total Al and P concentration in several different samples and product formulations. Measurement of free phosphate concentration is achieved at neutral pH and is an expression of the amount of P that is not physically associated with either the insoluble adjuvant particles or with the protein antigens. Along with total Al and P, the ratio of free to total phosphate may have an impact on the colloidal properties of the adjuvant and/or its antigen adsorption capacity, and is thus a potential product quality attribute. The NMR method provides a faster and simpler alternative to traditional wet chemistry methods for Al content and additionally, can assess P content in a single workflow. The suitability of the method to different AlPO4 -adjuvanted formulations has the potential to broaden its applicability to development and monitoring of novel adjuvant and vaccine manufacturing processes. Although the analyses were performed using high field NMR spectroscopy, the smaller frequency ranges provided by lower magnetic fields, which can be provided by simpler NMR instruments at lower cost, are expected to achieve acceptable linearity for quantitation. As modern NMR spectroscopy is primarily a Fourier transform technique, sufficient sensitivity can be achieved by increasing the number of scans.
Compliance with ethical standards No animals were used in this study. No human or animal samples were used in this study.
Conflict of interest Bruce Carpick and Marina Kirkitadze are employees of Sanofi Pasteur Ltd. Rahima Khatun, Yi Sheng, and Howard Hunter are employees of York University. The authors have no other relevant affiliations or financial involvement with any organization or entity with a financial interest in or financial conflict with the subject matter or materials discussed in the manuscript. Thus includes employment, consultancies, stock ownership or options, or royalties. No writing assistance was utilized in the production of this manuscript.
172
R. Khatun et al. / Journal of Pharmaceutical and Biomedical Analysis 159 (2018) 166–172
Acknowledgements The authors would like to thank the Governments of Ontario and Canada for providing assisting funds through a MITACS grant IT08656. This research was partially funded by Sanofi Pasteur Ltd. The authors of this paper wish to extend their utmost gratitude to Rozalia Nisman and Tristan Ho, QC Chemistry Sanofi Pasteur Ltd for discussion and to Juthika Menon, Liliana Sampaleanu, and Danielle Johnson, Manufacturing Technology, Sanofi Pasteur Ltd for providing samples for testing, and to Georgiana Moldoveanu, University of Toronto for support with ICP-OES experiments. The authors are also grateful to the suggestions provided by the reviewers of the manuscript. Appendix A. Supplementary data Supplementary material related to this article can be found, in the online version, at doi:https://doi.org/10.1016/j.jpba.2018.06. 025. References [1] R.K. Gupta, Aluminum compounds as vaccine adjuvants, Adv. Drug Del. Rev. 32 (1998) 155–172. [2] Y. Wen, Y. Shi, Alum: an old dog with new tricks, Emerg. Microbes Infect. 5 (2016) e25. [3] P. He, Y. Zou, Z. Hu, Advances in aluminum hydroxide-based adjuvant research and its mechanism, Hum. Vaccin. Immunother. 11 (2015) 477–488. [4] C.N. Maughan, S.G. Preston, G.R. Williams, Particulate inorganic adjuvants: recent developments and future outlook, J. Pharm. Pharmacol. 67 (2015) 426–449. [5] European Pharmacopeia, Aluminum in Adsorbed Vaccines, Chapter 2.5.13., 9th ed., 2018, pp. 165. [6] A. Mishra, S.R. Bhalla, S. Rawat, V. Bansal, R. Sehgal, S. Kumar, Standardization and validation of a new atomic absorption spectroscopy technique for determination and quantitation of aluminum adjuvant in immunobiologicals, Biologicals 35 (2007) 277–284. [7] T. Clapp, P. Siebert, D. Chen, L.-T. Jones Braun, Vaccines with aluminum-containing adjuvants: optimizing vaccine efficacy and thermal stability, J. Pharm. Sci. 100 (2011) 388–401. [8] F. Medera, T. Daberkow, L. Treccani, M. Wilhelm, M. Schowalter, A. Rosenauer, L. Mädlerc, K. Rezwana, Protein adsorption on colloidal alumina particles functionalized with amino, carboxyl, sulfonate and phosphate groups, Acta Biomater. 8 (2012) 121–1229.
[9] European Pharmacopeia, Aluminum Phosphate Gel, 2166, 9th ed., 2018, pp. 165 http://online6.edqm.eu/ep904/. [10] European Pharmacopeia, Aluminum Phosphate Gel, Hydrated, 1598, 9th ed., 2018, pp. 1686. [11] U. Holzgrabe, I. Wawer, B. Diehl, NMR Spectroscopy in Pharmaceutical Analysis, 1st ed., Elsevier, Oxford, 2008. [12] R. Khatun, H.N. Hunter, W. Magcalas, Y. Sheng, B. Carpick, M. Kirkitadze, Nuclear magnetic resonance (NMR) study for the detection and quantitation of cholesterol in HSV529 therapeutic vaccine candidate, Comput. Struct. Biotechnol. J. 15 (14) (2017). [13] C. Jones, NMR assays for carbohydrate-based vaccines, J. Pharm. Biomed. Anal. 38 (2005) 840–850. [14] C. Simmler, J.G. Napolitano, J.B. McAlpine, S.-N. Chen, G.F. Pauli, Universal quantitative NMR analysis of complex natural samples, Curr. Opin. Biotechnol. 25 (2014) 51–59. [15] G.A. Barding Jr, R. Salditos, C.K. Larive, Quantitative NMR for bioanalysis and metabolomics, Anal. Bioanal. Chem. 404 (2012) 1165–1179. [16] U. Holzgrabe, C. Deubner, C. Schollmayer, B. Waibel, Quantitative NMR spectroscopy-applications in drug analysis, J. Pharm. Biomed. Anal. 38 (2005) 806–812. [17] M. Weber, C. Hellriegel, A. Rueck, J. Wuethrich, P. Jenks, Obkircher, Method development in quantitative NMR towards metrologically traceable organic certified reference materials used as 31 P qNMR standards, Anal. Bioanal. Chem. 407 (2015) 3115–3123. [18] R. Martino, V. Gilard, F. Desmoulin, M. Malet-Martino, Fluorine-19 or phosphorus-31 NMR spectroscopy: a suitable analytical technique for quantitative in vitro metabolic studies of fluorinated or phosphorylated drugs, J. Pharm. Biomed. Anal. 38 (2005) 871–891. [19] S.J. Karlik, G.A. Elgavish, R.P. Pillai, G.L. Eichhorn, Aluminum-27 NMR studies of Al(III)-phosphate complexes in aqueous solution, J. Magn. Reson. 49 (1982) 164–167. [20] M.A. Curtin, L.R. Ingalls, A. Campbell, M. James-Pederson, Hydrolysis studies and quantitative determination of aluminum ions using 27 Al NMR, J. Chem. Educ. 85 (2008) 291–293. [21] J.W. Akitt, Nuclear magnetic resonance spectroscopy in liquids containing compounds of aluminum and gallium, Ann. Rep. NMR Spectrosc. 5A (1972) 465–559. [22] H. Maki, G. Sakata, M. Mizuhata, Quantitative NMR of quadrupolar nucleus as a novel analytical method: hydrolysis behaviour analysis of aluminum ion, Analyst 142 (2017) 1790–1799. [23] R.F. Mortlock, A.T. Bell, C.J. Radke, 31 P and 27 Al NMR investigations of highly acidic, aqueous solutions containing aluminum and phosphorous, J. Phys. Chem. 97 (1993) 767–774. [24] R.F. Mortlock, A.T. Bell, C.J. Radke, 31 P and 27 Al NMR investigations of the effect of pH on aqueous solutions containing aluminum and phosphorous, J. Phys. Chem. 97 (1993) 775–782. [25] Validation of Analytical Procedures:Text and Methodology. International Conference on Harmonization of Technical Requirements for Registration of Pharmaceuticals for Human Use, 2005.