Timothy L. Schofield

A Novel Staphylococcus aureus Vaccine: Iron Surface Determinant B Induces Rapid Antibody Responses in Rhesus Macaques and Specific Increased Survival in a Murine S. aureus Sepsis Model

ABSTRACT Staphylococcus aureus is a major cause of nosocomial infections worldwide, and the rate of resistance to clinically relevant antibiotics, such as methicillin, is increasing; furthermore, there has been an increase in the number of methicillin-resistant S. aureus community-acquired infections. Effective treatment and prevention strategies are urgently needed. We investigated the potential of the S. aureus surface protein iron surface determinant B (IsdB) as a prophylactic vaccine against S. aureus infection. IsdB is an iron-sequestering protein that is conserved in diverse S. aureus clinical isolates, both methicillin resistant and methicillin sensitive, and it is expressed on the surface of all isolates tested. The vaccine was highly immunogenic in mice when it was formulated with amorphous aluminum hydroxyphosphate sulfate adjuvant, and the resulting antibody responses were associated with reproducible and significant protection in animal models of infection. The specificity of the protective immune responses in mice was demonstrated by using an S. aureus strain deficient for IsdB and HarA, a protein with a high level of identity to IsdB. We also demonstrated that IsdB is highly immunogenic in rhesus macaques, inducing a more-than-fivefold increase in antibody titers after a single immunization. Based on the data presented here, IsdB has excellent prospects for use as a vaccine against S. aureus disease in humans.

Immunoprecipitation and Virus Neutralization Assays Demonstrate Qualitative Differences between Protective Antibody Responses to Inactivated Hepatitis A Vaccine and Passive Immunization with Immune Globulin

Antibodies to hepatitis A virus (anti-HAV) were measured in children from two separate vaccine trials (n = 70) 4 weeks after a dose of inactivated hepatitis A vaccine (VAQTA). The geometric mean titers (GMTs) of anti-HAV were 49.3 and 45.2 mIU/mL by immunoassay, while reciprocal GMTs of neutralizing anti-HAV were 6.5 and 15.0 by an 80% radioimmunofocus inhibition test (RIFIT) and 55.6 and 92.0 by antigen reduction assay (HAVARNA). The GMT of antibody detected by radioimmunoprecipitation (RIPA) was > or =401. These data establish serologic correlates of protection against disease and show that RIPA is most sensitive for detection of early vaccine-induced antibody. Sera collected from adults (n = 20) 7 days after administration of immune globulin contained similar antibody levels by immunoassay (45.1 mIU/mL) and slightly higher GMTs of neutralizing antibody (27.5 by RIFIT and 146 by HAVARNA) but negligible precipitating antibody (GMT, 5.6). These results are best explained by differences in the affinity of antibodies for virus following active versus passive immunization.

Vaccine stability study design and analysis to support product licensure

Stability evaluation supporting vaccine licensure includes studies of bulk intermediates as well as final container product. Long-term and accelerated studies are performed to support shelf life and to determine release limits for the vaccine. Vaccine shelf life is best determined utilizing a formal statistical evaluation outlined in the ICH guidelines, while minimum release is calculated to help assure adequate potency through handling and storage of the vaccine. In addition to supporting release potency determination, accelerated stability studies may be used to support a strategy to recalculate product expiry after an unintended temperature excursion such as a cold storage unit failure or mishandling during transport. Appropriate statistical evaluation of vaccine stability data promotes strategic stability study design, in order to reduce the uncertainty associated with the determination of the degradation rate, and the associated risk to the customer.

Basic principles of stability.

An understanding of the principles of degradation, as well as the statistical tools for measuring product stability, is essential to management of product quality. Key to this is management of vaccine potency. Vaccine shelf life is best managed through determination of a minimum potency release requirement, which helps assure adequate potency throughout expiry. Use of statistical tools such a least squares regression analysis should be employed to model potency decay. The use of such tools provides incentive to properly design vaccine stability studies, while holding stability measurements to specification presents a disincentive for collecting valuable data. The laws of kinetics such as Arrhenius behavior help practitioners design effective accelerated stability programs, which can be utilized to manage stability after a process change. Design of stability studies should be carefully considered, with an eye to minimizing the variability of the stability parameter. In the case of measuring the degradation rate, testing at the beginning and the end of the study improves the precision of this estimate. Additional design considerations such as bracketing and matrixing improve the efficiency of stability evaluation of vaccines.

Maintenance of vaccine stability through annual stability and comparability studies

Evaluation of vaccine stability does not end with licensure of the product. An annual stability program helps assure continued quality of product throughout the dating period, while comparability studies are performed after a process or facilities change in order to demonstrate that the change has not impacted the stability characteristics of the product. Careful attention to the design and analysis of post licensure studies helps mitigate the risk of missing a meaningful shift in the degradation rate of a vaccine, as well as the possibility of incorrectly earmarking a stability shift when the product remains acceptable.

Monitoring the stability of human vaccines.

Postmarketing stability studies of vaccines that tend to be close to their clinical specification at the end of the expiration dating period may require enhanced annual monitoring. In addition, an early assessment of product stability prior to completion of each individual study is desired. However, predictive measures of individual lots may produce early indication of failure. In many cases, these prove to be false alarms. For such products, continued product quality after marketing should, therefore, depend less on evaluating individual observations or individual lot projections, and more on assuring that the underlying stability profile of the product as a whole has not changed. We propose a monitoring procedure and an index of the average quality of vaccine lots currently on the market. We explore the statistical properties of the index for several experimental designs for sampling of marketed lots, and we describe an optimality property of the index.

Transfer of Methods Supporting Biologics and Vaccines

The transfer of analytical methods supporting biologics and vaccines is complicated by the complexity and variability of biological systems. Many of these assays may be linked to clinical performance, and thus subject to specifications established from materials that were tested in the development laboratory. Thus, transfer must account for the risk that the method characteristics have changed, and may generate results for commercial lots that either earmarks a satisfactory lot as failing, or an unsatisfactory lot as passing specification. Transfer study strategies have been proposed based on method parameters or on tolerance intervals. This article describes a framework for establishing the equivalence between two laboratories with emphasis on the associated risks, and compares and contrasts the parametric and tolerance approaches.

A vaccine measured with a highly variable assay: rabies.

Manufacturers and regulators are challenged when evaluating stability of vaccines when potency is measured using a highly variable assay. Participants in the IABS Workshop on Stability Evaluation of Vaccines, a Life Cycle Approach, were offered a case study from a series of stability studies of a rabies vaccine, using the NIH potency assay. The case study was introduced with a scenario in which a new manufacturer was to formulate, lyophilize and fill the vaccine from bulk supplied by another manufacturer. The regulatory authority requested that data from the new manufacturer be supplied, to supplement that of the original producer. Participants were asked to answer a series of questions posed by the regulator, and critique the study design and data analysis according to principles described during the workshop.

Analyzing highly variable potency data using a linear mixed-effects measurement error model

A house standard lot is tested along with experimental samples in a variable TCID50 assay in order to monitor and control assay performance. Instead of being simply a control, it is proposed to use this lot as a calibration standard to reduce the systematic variability in the assay caused by acknowledged sources of variability such as the age of the cells used in the assay and interlaboratory differences. Because of this new proposal, the consistency of the relationship between the test sample and the house standard is assessed within the acceptance range of the house standard. A linear mixed-effects measurement error model is proposed for the data. The slope curve is then used to assess the dynamic relationship between the sample and the house standard within the house standard range. It is shown with these analyses that the sample and the house standard have uniformly good agreement within the house standard range.

Introduction to Special Issue on Nonclinical Biopharmaceutical Statistics

We are pleased to have had the opportunity to edit this special issue devoted to nonclinical statistical applications in the pharmaceutical industry. The papers were drawn from presentations made at Harvard University in October 2009, during the first U.S. conference devoted solely to nonclinical statistics. In broad terms, “clinical” statistics refers to the design and evaluation of studies involving human test subjects, where the objective is to characterize the pharmacokinetics, safety, and efficacy of pharmaceutical products, whereas “nonclinical” statistics refers to the support of nonhuman studies arising from three main areas: (1) the discovery phase, where candidate pharmaceuticals are first identified; (2) safety and pharmacology studies, where toxicity and systemic effects are being studied in animal and in vitro models; (3) chemical, biological, and pharmaceutical development of processes, formulations and assays, where properties and manufacturability of the pharmaceutical product are being studied and frequently includes post-marketed quality control and engineering studies, especially as related to the recent Quality by Design initiatives where real-time process and product knowledge underlie product quality control rather than a static set of required manufacturing conditions with fixed tolerances. The nonclinical aspects cover the entire time span of the pharmaceutical development process, beginning with the early discovery phase through the submission of the New Drug Application (NDA), the Biological License Application (BLA) or the Marketing Authorization Application (MAA) and into the full commercialization phase. In presenting a broad view of the practice of statistics in this important area of pharmaceutical research, our goal was to underscore the breadth of nonclinical statistical practice. The emphasis is on innovative applications of existing methodology to the pharmaceutical development process as well as touching on the connections between statistical practice and regulatory expectations to acknowledge the nature of statistical applications in a regulated industry. The desire to share important statistical tools in the nonclinical statistician’s technical collection and currently relevant nonclinical questions served to guide us in the selection of manuscripts. Innovative uses of experimental design, modeling, and analysis strategies and Bayesian approaches are found. Although it is impossible to cover the full gamut of topics relevant to nonclinical statistics in a single issue, we hope the reader will gain a sense of the richness of statistical applications in this important subject area. The articles presented here cover the three main areas of nonclinical pharmaceutical development mentioned previously. Brain et al. discuss a generalization of semilinear canonical correlation analysis to detect signals in electroencephalography power spectra to obtain drug “signatures” indicating drug entry into the brain. Handzel et al. discuss data analysis approaches exploiting quantile scores and composite decision rules across