The epidemiological and economic impact of a cell-based quadrivalent inuenza vaccine in an adult population in the US: a dynamic modeling approach

Mutations of the H3N2 vaccine strain during the egg-based vaccine manufacturing process seem to partly explain the suboptimal effectiveness of traditional seasonal inuenza vaccine. Cell-based inuenza vaccines avoid such egg-adaptation, thereby improving antigenic match and vaccine effectiveness. The objective of this study was to evaluate the public health and economic impact of a cell-based quadrivalent inuenza vaccine (QIVc) in adult population (18–64 years) compared to the standard egg-based quadrivalent inuenza vaccine (QIVe), in the US. Methods The impact of QIVc over QIVe in terms of public health and costs outcomes was estimated using a dynamic SEIR transmission model. The model is age-structured and accounts for 4 circulating inuenza strains (A/H1N1pdm9, A/H3N2, B(Victoria), and B(Yamagata)). It was calibrated on US attack rate and strain circulation for the seasons 2013–2018. US specic absolute vaccine effectiveness for QIVe, specic hospitalization rate, mortality rate, Quality-Adjusted Life Years (QALYs) and costs were extracted from published literature. Relative vaccine effectiveness of QIVc over QIVe for subjects 18–64 years of age was obtained from a US retrospective cohort study. Robustness of the results was assessed in univariate and probabilistic sensitivity analyses.


Background
With an estimated average of 410,000 deaths yearly (1), the burden of in uenza infection still weighs heavily on the world population, despite years of improvement in worldwide immunization policies. In the US only, the US CDC estimates that in uenza infections have caused the sickness of 35.5M individuals, 16.5M outpatient visits, 490,600 hospitalizations and 34,200 deaths during the 2018-2019 winter season (2). Vaccination against in uenza is generally considered the most effective way to prevent in uenza infection and its consequences. However, the complexities of in uenza viruses, evading the human immune response with antigenic drift and shift phenomena, competing with a diversity of other strains for dominant circulation, seriously challenge, every year, the effectiveness of in uenza immunization campaigns. Recent years have shown advances in in uenza vaccination policies, with the adoption by some (mainly high developed) countries of quadrivalent in uenza vaccines, containing antigens for two B lineages, instead of a single one, and relatively new adjuvanted and high-dose formulations. Still, conventional egg-based in uenza vaccines seem to offer a suboptimal protection during in uenza seasons dominated by in uenza A(H3N2) circulation. A recent meta-analysis measured a 35% vaccine effectiveness (VE) against A(H3N2) versus 54-73% for other strains, for working-age adults, vaccinated with egg-based quadrivalent in uenza vaccine (QIVe) (3). This poor protection seems to be partially explained by issues related to the production of egg-based vaccines. Mutations of in uenza virus strains during the isolation and propagation steps of egg-based vaccine production seem to be leading to a mismatch between the in uenza vaccine strain and the circulating one (4,5). Solving this issue may rely on the use of mammalian cells rather than chicken eggs during vaccine virus isolation and propagation.
This cell based production process has demonstrated an improved antigenic similarity between the original candidate vaccine virus and the circulating wild-type in uenza virus strains, which may translate into an increased VE for new cell-based quadrivalent in uenza vaccines (QIVc) (6,7).
The assessment of the public health and economic impact of new in uenza immunization policies is routinely done for most countries. Despite WHO recommendations for the assessment of vaccination strategies (8), most of these studies are performed using a simple static epidemiological model (9), unable to grasp the indirect effects of vaccination (herd effects). While the potential impact of improved in uenza vaccine effectiveness is generally acknowledged for the pediatric population (10), its impact on the adult population remains unclear. The purpose of the present analysis is to evaluate the costeffectiveness of QIVc in adults (18-64 years) compared to conventional QIVe using a dynamic in uenza transmission model able to account for the indirect effect of vaccination.

Epidemiology and vaccine e cacy
In uenza incidence estimates were extracted from CDC reports for 5 seasons from 2013 to 2018 (Table   S1) (11), and combined with WHO FluNet (12) virological data to obtain yearly incidence per strain for the US. The model assumed 66% of people infected with in uenza virus were symptomatic (13), that they had an incubation period of 0.8 days, and remained infectious for 1.8 days (14,15).
We used strain speci c QIVe VE estimates obtained by Rolfes (Table 1). In order to compute the speci c QIVc increased e cacy against A(H3N2) (in 2017-2018, only A(H3N2) antigens are cell-based), we used the total rVE estimated on a US cohort comparing QIVc and QIVe for the same season, and then, we recomputed, the QIVc VE against A(H3N2) using the following equations: We have assumed an 0% in the lower bound 95% IC in vaccine effectiveness when negative percentage was reported by Rolfes et al (16).
where VE e and VE c stand for the vaccine effectiveness against all in uenza viruses of QIVe and QIVc respectively, p i and VE i are the proportion of circulating i strain and the associated vaccine e cacy in the season 2017-2018, and is the speci c vaccine effectiveness of QIVc against A(H3N2). We assumed that the QIVc was well matched against circulating A(H3N2) and not superior to QIVe against A(H1N1) and B strains. Details of QIV c and QIV e VE per age-groups are given in Table 1. The model is structured by age-group (6-23mo, 2-4yo, 5-12yo, 13-17yo, 18-49yo, 50-64yo, and more than 65yo) and uses a contact matrix to account for assortative rate of contacts between age-groups. In our analysis, we used in our base case analysis the matrix from Mossong Table S2. The number of cases per age-group estimated by the epidemiological model are taken as inputs of the economic model. Probabilities of general practitioner visit, hospitalization, and death are applied on the number of cases attributed to a high or low risk group, and then translated into public health outcomes and costs (Table S2). The economic analysis is performed from a societal perspective without taking into account productivity loss due to death. We consider a willingness to pay per QALY threshold of US$50 000 to consider a strategy as cost-effective (23).

Economic data
Disease costs and QALYs were extracted from a recent in uenza health economic analysis performed in the US context (22). Cost of a work day for the pediatric population are assumed to be related to parental work loss. Vaccine cost for QIVe and QIVc were set at $17.22 and $24.22 respectively(24). We do not consider administration costs since they are assumed to be the same across the different vaccination strategy (no difference in vaccination coverage). Details of the costs per age-groups are given in Table   S2.
As a reference strategy, we assume that the US population is vaccinated with conventional QIVe for those aged under 65 years of age and TIV HD for those aged 65 years and above. Then we compare this strategy to a scenario where QIVe is replaced by QIVc for people aged 18-64 years, other age-groups keeping their baseline vaccination. For both scenarios, we use age-based vaccination coverage rate documented by CDC (25), in particular we consider that 34.9% of people aged 18 to 49yo, 47.30% of people aged 50 to 64yo, and 68.10% of people older than 65 years are vaccinated against in uenza (Table S2). In this base case comparison, we consider that a seasonal mismatch between the QIVe A(H3N2) strain and the circulating strain due to egg adaptation occurred in 3 out of 5 years in the analysis scope (26). As a sensitivity analysis we also assessed the impact of QIVc when the mismatch due to egg adaptation occurred over 1, 2, 4 or all 5 years.
We also performed a stochastic probabilistic sensitivity analysis in order to assess the robustness of our results regarding uncertainties in vaccine effectiveness, economic inputs (primary care and hospitalization costs), probability of outcomes, listed in Table S3 and Table S4, with their probability distributions. In this analysis, 1,000 sets of the above-mentioned parameters are randomly drawn from distributions indicated in the Table S2 and Table S3. Clinical and economic results are averaged over the 5 in uenza seasons.

Results
In our base case scenario of 3 seasonal A(H3N2) mismatches due to egg adaptation, our analysis shows that using QIVc instead of QIVe in the 18-64yo population would have prevented 5.7M cases of in uenza, 1.8M GP visits, almost 50K hospitalizations and more than 5,400 deaths. In total, QIVc would have saved US$845M in direct costs and saved 128K QALYs. Hence, the switch from QIVe to QIVc in the adult population would be a cost saving strategy.
Over 5 seasons of A(H3N2) mismatch, using QIVc instead of QIVe in the 18-64yo population would have a substantial effect but irregular effect, depending on the season, depending on A(H3N2) seasonal circulation (Fig. 2). Hence, the choice of the season was also randomly varied in the probabilistic analysis to assess the uncertainty related to the epidemiological context. Varying the number of seasons with an A(H3N2) mismatch of egg-based vaccines between 1 and 5 years still show that QIVc would be costsaving or very cost effective (Table 3). Probabilistic sensitivity analysis con rms that 95% of the 1,000 simulations gave a cost-saving result (Fig. 3).

Discussion
This analysis emphasizes the potentially major public health gain which could be achieved through the use of QIVc. Several studies have highlighted the suboptimal VE of egg-based vaccines against some strains of A(H3N2) (6, 27). While the link between VE reduction and egg-related strain mutations is still poorly understood (28) and remains to be fully investigated, recent studies, in different populations, tend to con rm the clinical bene ts (7) of QIVc over QIVe regarding A(H3N2) vaccine strain mismatch.
Assessing the strain-speci c rVE of QIVc compared to QIVe remains a di cult challenge as eggadaptation phenomena, while most common on A(H3N2), may also occur on B lineages. However, we represented 66% of the in uenza positive samples (12). By construction, these analyses were only able to assess non-strain speci c QIVc rVE compared to QIVe, and we had to estimate QIVc strain speci c VE, assuming that the increased total VE was only linked to an increase of VE against A(H3N2). Also, variations in QIVc rVE are likely to occur due to the changing distribution patterns of circulating in uenza strains. Namely, in uenza seasons with a highly dominant A(H1N1)pdm09 circulation (2015-2016) will see a low bene t to QIVc compared to QIVe, while others, like 2014-2015, 2016-2017, or 2017-2018, may see a signi cant one. Hence, analysis on multiple seasons is necessary in order to fully assess the potential "averaged" impact of cell-based vaccines across various realistic epidemiological contexts.
We chose to consider as a base case scenario that a mismatch occurred between the A(H3N2) circulating strain and the egg-based vaccine strain during three seasons. We considered this choice a median scenario between a systematic yearly mismatch and no mismatch at all. In addition, this assumption has a limited impact on our results since 1) we consider the observed in uenza strain distribution, and 2) we have analyzed situation when the number of mismatched years was varied from 1 to 5 years and reached qualitatively similar results. In addition, our results are consistent with previous health-economic analysis of QIVc in Europe (UK, Spain, Italy, Germany) (29)(30)(31), where QIVc has been shown to be either costsaving or highly cost-effective.
Our analysis uses a 4-strain SEIR compartmental model with an age-structure. This kind of approach, previously used in several similar analysis (9,(32)(33)(34), is key to capture the potential indirect effects of in uenza vaccination, accounting for prevented chains of transmission. However, it relies on assumptions regarding age-related contact rate which may be di cult to measure for speci c countries. Despite its advantages, our approach suffers from limitations inherent to any modeling exercise. In absence of better estimates, our transmission model assumes that 30% of the population bene ts from remaining immunity against in uenza based on a British modeling study (14). In addition, uncertainties regarding surveillance-based in uenza incidence estimates, or in uenza strains circulations in the US will directly impact the epidemiological dynamics reproduced by the model.
Finally, our analysis shows the potential public health bene ts of the use of QIVc in the 18-64yo US population during the 2013-2018 period. Results obtained retrospectively may be different from what may be achieved in the future as in uenza strain circulation is currently challenging to predict. In particular, given the potential expected bene ts, and on-going clinical trials, it is likely that QIVc would be also recommended for the pediatric population, which would certainly reinforce its impact.

Conclusions
Estimated public health bene ts related to the use of QIVc in the US adult population, indicate that the use of QIVc could be clinically superior and cost-saving compared to the current vaccination practice (QIVe). Sensitivity analyses on vaccine effectiveness, costs and number of seasons of egg-based strain mismatch show the robustness of QIVc cost-effectiveness.

Funding
The study was funded by Seqirus. The funder had no role in the study design, data collection, analysis, decision to publish or preparation of the manuscript.

Figure 1
Diagram of the in uenza transmission model. Si, Ei, Ii, and Ri stand for susceptible, exposed, infectious, and recovered individuals respectively regarding the in uenza strain I ; VSi, VEi, and VIi stand for susceptible, exposed and infectious individuals experiencing a non-protective vaccination respectively regarding in uenza strain i, VRi are individuals protected and vaccinated against in uenza strain ; i stands for in uenza strains: A/H1N1, A/H3N2, B Victoria, B Yamagata Probabilistic sensitivity analysis of the QIVc scenario compared to QIVe. Health is measured in QALY and cost in US$.