Estimating the protective concentration of anti-pneumococcal capsular polysaccharide antibodies

Abstract

Estimates of minimum protective antibody concentrations for vaccine preventable diseases are of critical importance in assessing whether new vaccines will be as effective as those for which clinical efficacy was shown directly.

We describe a method for correlating pneumococcal anticapsular antibody responses of infants immunized with pneumococcal conjugate (PnC) vaccine (Prevenar) with clinical protection from invasive pneumococcal disease (IPD). Data from three double blind controlled trials in Northern Californian, American Indian and South African infants were pooled in a meta-analysis to derive a protective concentration of 0.35 μg/ml for anticapsular antibodies to the 7 serotypes in Prevenar. This concentration has been recommended by a WHO Working Group as applicable on a global basis for assessing the efficacy of future pneumococcal conjugate vaccines.

The WHO Working Groups anticipated that modifications in antibody assays for pneumococcal anticapsular antibodies would occur. The principles for determining whether such assay modifications should change the protective concentration are outlined. These principles were applied to an improvement in the ELISA for anticapsular antibodies, i.e. absorption with 22F pneumococcal polysaccharide, which increases the specificity of the assay for vaccine serotype anticapsular antibodies by removing non-specific antibodies. Using sera from infants in the pivotal efficacy trial in Northern California Kaiser Permanente (NCKP), 22F absorption resulted in minimal declines in pneumococcal antibody in Prevenar immunized infants but significant declines in unimmunized controls. Recalculation of the protective concentration after 22F absorption resulted in only a small decline from 0.35 μg/ml to 0.32 μg/ml. These data support retaining the 0.35 μg/ml minimum protective concentration recommended by WHO for assessing the efficacy of pneumococcal conjugate vaccines in infants.

Introduction

When antibodies are the major mechanism by which vaccines provide protection, it is theoretically possible to establish an antibody concentration which predicts protection in an immunized population [1]. In practice, protective antibody concentrations have been proposed and generally accepted for a number of vaccine preventable diseases, including tetanus, diphtheria, polio, Japanese encephalitis, measles, mumps, rubella, hepatitis B, varicella, influenza, meningococci and Haemophilus influenzae type b (Hib). [2] Recently a protective concentration has also been recommended for pneumococci [3], [4], which is the subject of this report.

The most important application of protective antibody concentrations is for establishing the protective efficacy of new or improved vaccines when placebo controlled efficacy trials are no longer feasible or ethical. The protective concentration is also used as the benchmark for assessing interference between vaccines given concomitantly. In these assessments, the proportion of the test population receiving the new vaccine or new vaccine combination which achieves an antibody concentration equal or greater than the protective concentration is compared to the control population receiving the control vaccine or vaccine combination which is the current standard of practice and for which efficacy has already been established in controlled clinical trials. Achieving the protective concentration thus serves as the primary outcome in determining whether a new vaccine is not inferior to an already licensed vaccine with clinically documented efficacy and thus can be inferred also to be effective.

Certain fundamental principles must be followed in order to develop a valid estimate of the protective concentration. First, it must be established that the immune mechanism that is measured correlates with protective activity. In all cases for which protective antibody correlates have been established to date, protection is mediated by antibody [2]. It has however not been possible to establish an antibody based protective correlate for certain vaccines which are given by mucosal routes, including the intranasal cold adapted influenza vaccine [5], oral rotavirus vaccine and oral typhoid vaccine [2], perhaps because local cellular and humoral immunity may protect in the absence of measurable systemic antibody.

Second, it is important to verify that the target antigen is indeed a virulence factor of the organism or serves as a protective antigen when antibody binds to it. For many vaccines, the target antigen is the major virulence factor, including tetanus, diphtheria and pertussis toxins, capsular polysaccharides of pneumococcus, meningococcus and Hib, and critical surface exposed viral antigens such as influenza hemagglutinin.

Third, the antibody assay chosen ideally should directly measure the functional activity mediating protection. Examples of such functional assays include antitoxin assays for tetanus and diphtheria, bactericidal assays for Hib and meningococcus, opsonophagocytic assays for pneumococcus and direct virus neutralization assays for measles and other viruses. Because of technical difficulties or variability associated with complex bioassays, it is often desirable to use a binding or binding inhibition assay which correlates highly with functional activity as surrogate assays. Examples are influenza hemagglutination inhibition (HAI) as a surrogate for viral neutralization and ELISA assays for IgG class anticapsular polysaccharide antibodies to Hib and pneumococcal capsular polysaccharides as surrogates for bactericidal and opsonic activity respectively.

Recently a WHO working group has proposed a protective concentration for pneumococcal conjugate (PnC) vaccine in infants [3], [4]. A concentration of IgG anticapsular polysaccharide antibodies measured by ELISA ≥ 0.35 μg/ml measured one month after primary immunization was recommended as the protective threshold based on three double-blind controlled efficacy trials for invasive pneumococcal disease (IPD) performed in Northern California Kaiser Permanente (NCKP) [6], American Indians [7] and South Africa [8].

This report describes the serologic data and statistical methods used to derive this estimate. In addition, we present information on the effect on pneumococcal antibody concentrations and on the protective correlate of performing an additional absorption of the test sera with pneumococcal type 22F polysaccharide. This modification has been introduced to remove antibodies to non-capsular pneumococcal antigens and thereby make the ELISA more specific for antibodies to vaccine-type capsular polysaccharides which confer protection [9].