Use of an individual-based model of pneumococcal carriage for planning a randomized trial of a vaccine

Abstract

12For encapsulated bacteria such as Streptococcus pneumoniae, asymptomatic carriage is 13 more common and longer in duration than disease, and hence is often a more convenient 14 endpoint for clinical trials of vaccines against these bacteria. However, using a carriage 15 endpoint entails specific challenges. Carriage is almost always measured as prevalence, 16 whereas the vaccine may act by reducing incidence or duration. Thus, to determine sample 17 size requirements, its impact on prevalence must first be estimated. The relationship between 18 incidence and prevalence (or duration and prevalence) is convex, saturating at 100% 19 prevalence. For this reason, the proportional effect of a vaccine on prevalence is typically less 20 than its proportional effect on incidence or duration. This relationship is further complicated in 21 the presence of multiple pathogen strains. In addition, host immunity to carriage accumulates 22 rapidly with frequent exposures in early years of life, creating potentially complex interactions 23 with the vaccine's effect. We conducted a simulation study to predict the impact of an 24 inactivated whole cell pneumococcal vaccine-believed to reduce carriage duration-on 25 carriage prevalence in different age groups and trial settings. We used an individual-based 26 model of pneumococcal carriage that incorporates relevant immunological processes, both 27 vaccine-induced and naturally acquired. Our simulations showed that for a wide range of 28 vaccine efficacies, sampling time and age at vaccination are important determinants of 29 sample size. There is a window of favorable sampling times during which the required sample 30 size is relatively low, and this window is prolonged with a younger age at vaccination, and in a 31 trial setting with lower transmission intensity. These results illustrate the ability of simulation 32 studies to inform the planning of vaccine trials with carriage endpoints, and the methods we 33

Herd Immunity Alters the Conditions for Performing Dose Schedule Comparisons: An Individual-based Model of Pneumococcal Carriage

“…Figure 3 compares VE by each of these two measures over time, during the vaccine trial. Notably, vaccine efficacy varies with time since vaccination, due to the dynamic nature of pneumococcal colonization [13]. Also interestingly, the vaccine efficacy measured post-vaccination is generally lower in a trial in a vaccinenaïve community than one in a community already using PCV13.…”

“…The result that it would be difficult to detect a small reduction in immune protection in a context of rare VT colonization was perhaps predictable by simple intuitive reasoning. However, previous work [13] has shown that pneumococcal vaccine trials with a colonization prevalence endpoint have several particularities that differ from those of more traditional vaccine trials using incidence of an acute infection as an endpoint [20,21,22,23,24]. In particular, these studies have shown that several factors complicate the interpretation of such studies:…”

“…Indeed, one part of our results -the finding that vaccine efficacy as measured by reduction in either risk or odds of carriage is higher in the presence of herd immunity -is counter to the usual pattern in vaccine trials using incidence as an endpoint, wherein reduced incidence in all trial arms may occur due to herd immunity, but the relative reduction in incidence due to vaccination (or more vaccination) is assumed to be constant across settings [28]. The strong timedependence of measured relative efficacy of various dosing schedule was also a striking finding that has not been considered in most previous studies; this finding again is a result of the use of prevalence rather than incidence as an endpoint, and of the complex dynamics of age-varying force of infection and developing immunity in the early years of life via natural colonization [24]. In general, simulations of clinical trials for vaccines or other infectious disease prevention measures can enhance intuition and improve study design [29].…”

“…• the competition between different serotypes to colonize [25,26] • the ability of most assays to detect only one serotype among those colonized with one or more serotypes [27] • the nonlinear relationship between incidence rate reduction (the biological parameter assumed to be relatively consistent for the same vaccine in different settings) and prevalence reduction (which is measured in such trials) [16] • the fact that as children are receiving vaccine doses they are also acquiring immunity in both serotype-specific and nonspecific fashions to colonization. [24] It is thus risky to rely on intuition for how vaccine trials would work in various settings; hence, the impetus to perform the studies described here. Indeed, one part of our results -the finding that vaccine efficacy as measured by reduction in either risk or odds of carriage is higher in the presence of herd immunity -is counter to the usual pattern in vaccine trials using incidence as an endpoint, wherein reduced incidence in all trial arms may occur due to herd immunity, but the relative reduction in incidence due to vaccination (or more vaccination) is assumed to be constant across settings [28].…”

“…The transmission parameters were fit using 2001 data from the SPARC project, which sampled children aged 7 or younger in Massachusetts, USA and reported a 28% carriage prevalence [12]. We fit to serotype-specific prevalences using an algorithm previously described [13].…”