SARS-CoV-2 and influenza are the main respiratory viruses for which effective vaccines are currently available. Strategies in which COVID-19 and influenza vaccines are administered simultaneously or combined into a single preparation are advantageous and may increase vaccination uptake.
1. Introduction
Vaccination against COVID-19 is a cornerstone public health intervention to tackle the ongoing pandemic
[1]. Analogously, seasonal influenza vaccination (SIV) is considered one of the most effective means of reducing the burden of disease, which in the pre-COVID-19 era caused on average 250,000–500,000 deaths worldwide each year
[2,3][2][3]. Although circulation of the influenza virus has diminished drastically since 2020, the ongoing 2021/22 northern hemisphere winter season is characterized by the circulation of both SARS-CoV-2 and influenza
[4]. Moreover, in the 2021/22 season, there is an overlap between the administration of booster doses of COVID-19 vaccines and the SIV campaign
[5,6][5][6].
Very little is yet known about the interaction between SARS-CoV-2 and influenza viruses. Dadashi et al.
[7] estimated a pooled prevalence of SARS-CoV-2 and influenza co-infection of 0.8% (95% confidence interval (CI): 0.4–1.3%) with marked regional heterogeneity, ranging from 0.4% in the Americas to 4.5% in Asia. A large study conducted in England
[8] reported that, while individuals positive for influenza had 58% (95%: 44–69%) lower odds of also testing positive for SARS-CoV-2, co-infected individuals had worse clinical outcomes. Specifically, co-infected patients were approximately twice as likely to die (odds ratio (OR) 2.27; 95% CI: 1.23–4.19) as subjects positive only for SARS-CoV-2
[8]. Some experimental evidence has provided useful insights into these poorer outcomes in co-infected patients. Indeed, it has been observed
[9] that prior infection with type A influenza virus promotes SARS-CoV-2 entry and infectiousness in both cell and animal models, probably owing to the ability of the former to increase the expression of angiotensin-converting enzyme 2 (ACE2).
2. Non-Specific Effects of Seasonal Influenza Vaccination on COVID-19-Related Outcomes
2.1. Real-World Evidence
Some early ecological studies
[19[10][11],
20], which are useful for hypothesis generation
[21][12], found a negative correlation between SIV coverage rates and COVID-19-related outcomes. For instance, in Italy, which was the first western country where SARS-CoV-2 spread widely, a negative correlation (
r = −0.59,
p = 0.005) between regional SIV coverage rates in the elderly and COVID-19 deaths was demonstrated
[19][10]. However, subsequent cohort studies on this non-specific effect of SIV yielded contrasting results: some
[22,23,24][13][14][15] found a protective effect, while others did not find any significant association
[25,26][16][17]. To summarize the body of available observational studies, Wang et al.
[27][18] conducted a systematic review and meta-analysis of this non-specific association. In their random-effects model, a significant reduction in laboratory-confirmed cases of SARS-CoV-2 was found in subjects immunized with SIV, with a pooled (
n = 9 studies) OR of 0.86 (95% CI: 0.79–0.94). On the other hand, the association between SIV and some other COVID-19-related outcomes, such as hospitalization (OR 0.74; 95% CI: 0.51–1.06), admission to intensive care units (OR 0.63; 95% CI: 0.22–1.81) or mortality (OR 0.89; 95% CI: 0.73–1.09) was not statistically significant
[27][18].
2.2. Underlying Immunological Mechanisms
Different immunological mechanisms beyond the observed non-specific heterologous effects of SIV on COVID-19-related clinical endpoints, have been proposed. These mechanisms may involve either innate (the so-called “trained immunity”) or adaptive (bystander activation and cross-reactivity) compartments of the immune system
[28][19]. Bystander activation refers to a type of heterologous response which is exerted by adjacent, but not relevant, T cells with different specificity. These heterologous T cells are probably activated by cytokines as a result of the activation of cells during the classical response
[29][20]. By contrast, the cross-reactivity theory holds that T cells involved in the classical adaptive immune response may cross-react with an antigen presenting some degree of amino acid similarity
[28][19]. Finally, the trained immunity hypothesis postulates that the innate immune cells may be primed upon encountering exogenous or endogenous insults, causing long-term metabolic and epigenetic reprogramming of these cells and leading to an enhanced response to a second challenge
[30,31,32][21][22][23].
The available experimental data on influenza virus- and/or SIV-induced cross-reactive or even cross-protective antibodies against SARS-CoV-2 are controversial. For instance, with regard to T and B cell reactivity, Reche
[33][24] concluded that influenza viruses do not have epitopes that cross-react with SARS-CoV-2. Murugavelu et al.
[34][25] tested polyclonal sera obtained from SARS-CoV-2-positive subjects with high anti-spike neutralizing antibody titers and found some degree of cross-reactivity with influenza virus hemagglutinin in both enzyme-linked immunosorbent (ELISA) and Western blot assays. However, a subsequent analysis demonstrated that these hemagglutinin cross-reactive binding antibodies were not neutralizing. More recently, Almazán et al.
[35][26] investigated the role of the small NGVEGF peptide—which is identical, or very similar, to a peptide found in most contemporary A(H1N1)pdm09 strains—in inducing cross-reactive antibodies. This peptide is present in the most critical part (N481–F486) of the receptor binding domain (RBD) of the SARS-CoV-2 spike protein, which interacts with the ACE2 receptor, while in influenza A(H1N1)pdm09 strains, the NGVEGF/NGVKGF peptide is located in an immunodominant region of the neuraminidase. Approximately two thirds of blood donors (
n = 328) had detectable levels of antibodies to this peptide. Immunization with a quadrivalent egg-based influenza vaccine (QIVe) enhanced the anti-SARS-CoV-2 response: subjects with no recent influenza infection had low binding inhibitory activity (average of 32.7%), which was enhanced by QIVe administration (average of 55%) and further enhanced by the BNT162b2 (Comirnaty; Pfizer Inc., New York, NY, USA and BioNTech, Mainz, Germany) vaccine (average of 94%). The NGVEGF peptides also activated CD8+ cells in 20% of donors. Finally, the authors identified 11 additional CD8+ cell peptides that potentially cross-reacted with both SARS-CoV-2 and influenza viruses; depending on the type of human leukocyte antigen (HLA), these peptides may protect against SARS-CoV-2 in about 40–71% of individuals
[35][26].
The bystander activation mechanism has been partially proven by Pallikkuth et al.
[36][27]. Specifically, in their cohort of healthcare workers, A(H1N1) antigen-specific CD4+ cells were present in 92% and 76% of SARS-CoV-2-positive and -negative subjects, respectively. The A(H1N1) CD4+ response also showed a strong positive correlation with SARS-CoV-2-specific CD4+ cells
[36][27].
The trained immunity hypothesis has recently drawn particular attention. It was first documented in the case of BCG (Bacillus Calmette–Guérin) vaccine, and then measles, oral polio and, more recently, SIV
[28,29][19][20]. Experimental confirmation of SIV-induced trained immunity against SARS-CoV-2 was recently obtained in a Dutch study
[37][28]. Following the demonstration of a 37–49% relative risk reduction of SARS-CoV-2 infection among healthcare workers vaccinated with QIVe (compared with non-vaccinated subjects), the authors investigated the biological plausibility of this observation in a well-established in vitro model. Specifically, following the stimulation of peripheral blood mononuclear cells with QIVe and BCG, an increase in the production of cytokines was observed. Re-stimulation of these cells with a heat-inactivated SARS-CoV-2 strain induced a higher production of interleukin (IL)-1 receptor antagonist (IL-1RA), while the production of pro-inflammatory IL-1
β and IL-6 was reduced
[37][28]. An Italian study
[38][29] conducted among healthcare workers (
n = 710) who received 2 doses of the BNT162b2 vaccine found that, in subjects previously vaccinated with the quadrivalent cell culture-based influenza vaccine (QIVc; Flucelvax Tetra, Seqirus Netherlands B.V., Amsterdam, The Netherlands) plus pneumococcal vaccines or with QIVc alone, microneutralization titers against SARS-CoV-2 were 58% (
p = 0.01) and 42% (
p = 0.07) higher, respectively, than in subjects who did not receive any vaccine. By contrast, no significant differences were found for the anti-spike and interferon-
γ responses
[38][29].
3. Safety, Immunogenicity and Efficacy of COVID-19 and Influenza Vaccine Co-Administration
As of February 2022, three randomized controlled trials (RCTs)
[40,41,42][30][31][32] on COVID-19 and SIV vaccine co-administration are available. In a phase IV trial conducted in the United Kingdom (UK)
[40][30], adults were randomized (1:1;
n = 679) to receive either a second dose of ChAdOx1 (Vaxzevria, AstraZeneca, Cambridge, UK) or BNT162b2 vaccines, together with an age-appropriate SIV (QIVc, recombinant quadrivalent influenza vaccine (QIVr; Supemtek, Sanofi Pasteur, Lyon, France) for subjects aged 18–64 years and MF59-adjuvanted trivalent influenza vaccine (aTIV; Fluad, Seqirus) for those aged ≥65 years) or ChAdOx1/BNT162b2 together with placebo. Three weeks later, those who had received placebo received SIV, and vice versa. Toback et al.
[41][31] reported the results of a phase III efficacy RCT, in which a subset of adults was randomized (1:1;
n = 431) to receive the first dose of NVX-CoV2373 (Nuvaxovid, Novavax CZ a.s., Jevany, Czech Republic) plus SIV (QIVc and aTIV for subjects aged 18–64 and ≥65 years, respectively) or SIV alone. Finally, the interim results of a phase II RCT have recently been reported
[42][32]; in this trial, elderly individuals (≥65 years) were randomized (1:1:1;
n = 431) to receive a second dose of mRNA-1273 (Spikevax, Moderna, Cambridge, MA, USA) plus a high-dose quadrivalent influenza vaccine (hdQIV; Fluzone High-Dose Quadrivalent, Sanofi Pasteur, Lyon, France), a dose of hdQIV alone or a second dose of mRNA-1273 alone. In all three RCTs
[41[31][32],
42], vaccines were co-administered in opposite arms in each subject.
All three RCTs
[41,42][31][32] reported no major safety concerns regarding COVID-19 + SIV co-administration. Specifically, as reported in
Table 1, the overall rate of solicited local (especially pain in the injection site) and systemic adverse events was similar between subjects who received COVID-19 + SIV and those who received the COVID-19 vaccine alone. The adverse events reported were mostly mild-to-moderate and self-limiting. A similar picture was seen with regard to unsolicited adverse events.
Table 1. Rate (%) of solicited local and systemic adverse events in COVID-19 and influenza vaccine co-administration groups, as compared with groups to whom either vaccine was administered alone.
Adverse Event |
Comparison |
Vaccine Administration Pattern |
Reference |
COVID-19 Vaccine |
SIV |
COVID-19 + SIV |
COVID-19 Alone |
SIV Alone |
Any local, % |
ChAdOx1 | 1 |
QIVc | 1 |
that the humoral IgG response measured by means of the hemagglutination inhibition (HAI) assay approximately 3 weeks after immunization towards any SIV strain is preserved in COVID-19 + SIV co-administration groups. Indeed, apart from the statistical significance, geometric mean ratios of all pairwise comparisons have proved to be >0.67, which is the non-inferiority margin (
Table 2). Lazarus et al.
[40][30] did not observe any significant difference in geometric mean titers (GMTs) between most co-administration groups (ChAdOx1 + QIVc, ChAdOx1 + QIVr, ChAdOx1 + aTIV, BNT162b2 + QIVc or BNT162b2 + aTIV) and groups receiving SIVs alone. The only exception was the immune response to A(H1N1)pdm09, B/Victoria and B/Yamagata; in these cases, GMTs were 20–38% higher in the BNT162b2 + QIVr group than in individuals who received QIVr only. The IgG response to A(H3N2) was similar. Toback et al.
[41][31] did not find any significant difference in HAI titers between COVID-19 + SIV and placebo + SIV groups, regardless of the strain or vaccine (QIVc or aTIV). Analogously, the humoral immune response towards all four SIV strains was similar in individuals who received mRNA-1273 + hdQIV or hdQIV alone
[42][32].
Table 2. Hemagglutination inhibition IgG geometric mean ratios against influenza vaccine strains in COVID-19 + seasonal influenza vaccine co-administration groups, as compared with groups to whom either vaccine was administered alone.
Influenza Vaccine |
Strain |
COVID-19 Vaccine [Reference] |
BNT162b2 [40] | BNT162b2 [30] |
mRNA-1273 [42] | mRNA-1273 [32] |
ChAdOx1 [40] | ChAdOx1 [30] |
NVX-CoV2373 [41] | NVX-CoV2373 [31] |
84 |
81 |
– |
1 Working-age adults (18–64 years); 2 older adults (≥65 years); 3 strain was not present in the vaccine formulation; 4 seasonal influenza vaccine first vs. placebo first; 5 mRNA-1273 + hdQIV vs. hdQIV alone, geometric mean ratios were calculated from the geometric mean titers provided by the authors; 6 NVX-CoV2373 + seasonal influenza vaccine vs. placebo + seasonal influenza vaccine, geometric mean ratios were calculated from the geometric mean titers provided by the authors; aTIV, MF59-adjuvanted trivalent influenza vaccine; hdQIV, high-dose quadrivalent influenza vaccine; ns, non-significant (p > 0.05); QIVc, cell-based quadrivalent influenza vaccine; QIVr, recombinant quadrivalent influenza vaccine.
Similar results were reported with regard to anti-spike neutralizing antibody titers (
Table 3). Lazarus et al.
[40][30] and Izikson et al.
[42][32] did not find any significant differences between the co-administration groups and individuals who received COVID-19 vaccines alone. By contrast, some immunological inference was reported in the RCT by Toback et al.
[41][31] After adjustment for baseline titers, age and the treatment arm, the geometric mean ratio of the anti-spike humoral response in NVX-CoV2373 + QIVc/aTIV versus NVX-CoV2373 alone was 0.57 (95% CI: 0.47–0.70). On the other hand, this decrease did not seem to translate into a corresponding decrease in vaccine efficacy. For instance, absolute vaccine efficacy against laboratory-confirmed symptomatic COVID-19 in adults (18–64 years) was 87.5% (95% CI: −0.2–98.4%) and 89.8% (95% CI: 79.7–95.5%) in the influenza sub-study and main study, respectively
[41][31].
Table 3. Anti-spike IgG geometric mean ratios in COVID-19/influenza vaccine co-administration groups, as compared with groups to whom either vaccine was administered alone.
Influenza Vaccine |
COVID-19 Vaccine [Reference] |
BNT162b2 [40] | BNT162b2 [30] |
mRNA-1273 [42] | mRNA-1273 [32] |
ChAdOx1 [40] | ChAdOx1 [30] |
NVX-CoV2373 [41] | NVX-CoV2373 [31] |
[ | 40 | ] |
QIVc | 1 | [ |
A(H1N1)pdm09 |
1.05 (0.91–1.21) | 4 | 30] |
– |
1.05 (0.91–1.21) | 4 |
1.09 (ns) | 6 |
QIVc | 1 |
0.90 (0.80–1.01) | 3 |
– |
0.92 (0.81–1.04) | 3 |
0.66 (NA) | 5 |
BNT162b2 | 1 |
QIVc | 1 |
96 |
94 |
A(H3N2) |
1.06 (0.95–1.18) – |
4 |
– |
1.08 (0.96–1.21) | 4 |
1.08 (ns) | 6[ |
QIVr | 1 | 40 |
0.86 (0.72–1.03) | 3 |
– | ] | [ |
0.92 (0.81–1.04) | 3 | 30] |
– |
ChAdOx1 | 1 |
QIVr | 1 |
85 |
B/Victoria |
1.03 (0.93–1.14) 86 |
4 |
–– |
1.05 (0.94–1.18) | 4[40 |
1.03 (ns) | 6] | [30] |
aTIV | 2 |
0.97 (0.83–1.13) | 3 |
– |
1.02 (0.91–1.14) | 3 |
0.71 (NA) |
BNT162b2 | 1 |
QIVr | 1 |
96 |
89 |
– |
[40] | [30] |
B/Yamagata |
0.94 (0.85–1.05) | 4 |
– |
0.98 (0.88–1.10) | 4 |
0.99 (ns) | 6 |
5 |
hdQIV | 2 |
– |
0.97 (0.79–1.19) | 4 |
– |
– |
ChAdOx1 | 2 |
aTIV | 2 |
77 |
QIVr | 65 |
1 | – |
[40] | [30] |
A(H1N1)pdm09 |
1.38 (1.11–1.71) | 4 |
– |
0.86 (0.74–0.99) | 4 |
– |
BNT162b2 | 2 |
aTIV | 2 |
76 |
79 |
A(H3N2) | – |
1.03 (0.87–1.23) | 4 |
– |
1.03 (0.91–1.15) | 4[40 |
– | ] | [30] |
mRNA-1273 | 2 |
hdQIV | 2 |
86 |
91 |
62 |
[42] | [32] |
B/Victoria |
1.20 (1.02–1.42) | 4 |
– |
1.07 (0.96–1.19) | 4 |
– |
NVX-CoV2373 | 1 |
QIVc | 1 |
73 |
63 |
B/Yamagata | 39 |
1.24 (1.05–1.47) | 4 | [ | 41] | [31] |
– |
1.06 (0.94–1.18) | 4 |
– |
NVX-CoV2373 | 2 |
aTIV | 2 |
39 |
35 |
aTIV | 46 |
2 |
A(H1N1)pdm09 |
1.05 (0.87–1.27) | 4 |
– | [41] | [31] |
1.15 (1.01–1.32) | 4 |
1.41 (ns) | 6 |
Any systemic, % |
ChAdOx1 | 1 |
QIVc | 1 |
81 |
83 |
– |
A(H3N2) |
1.18 (1.02–1.37) | 4 |
– |
1.00 (0.89–1.11) | 4 | [40] | [30] |
0.87 (ns) | 6 |
BNT162b2 | 1 |
QIVc | 1 |
87 |
B/Victoria |
1.08 (0.94–1.25) | 4 |
–81 |
– |
[40] |
1.01 (0.91–1.12) | 4 | [30] |
0.54 (ns) | 6 |
ChAdOx1 | 1 |
QIVr | 1 |
B/Yamagata | 3 | 74 |
72 |
– |
1.00 (0.86–1.15) | 4 | [ |
– |
0.92 (0.83–1.03) | 4 | 40] |
0.81 (ns) | 6 | [30] |
BNT162b2 | 1 |
QIVr | 1 |
89 |
82 |
hdQIV | – |
2 |
A(H1N1)pdm09 |
– |
0.99 (ns) | 5[40] | [30] |
– |
– |
ChAdOx1 | 2 |
aTIV | 2 |
72 |
62 |
– |
[40] | [30] |
A(H3N2) |
– |
0.91 (ns) | 5 |
– |
– |
BNT162b2 | 2 |
aTIV | 2 |
59 |
71 |
– |
[40] | [30] |
mRNA-1273 |
B/Victoria |
– |
0.97 (ns) | 5 |
– |
– |
2 |
hdQIV | 2 |
80 |
84 |
49 |
[42] | [32] |
B/Yamagata |
– |
0.91 (ns) | 5 |
– |
– |
NVX-CoV2373 | 1 |
QIVc | 1 |
62 |
50 |
47 |
[41] | [31] |
NVX-CoV2373 | 2 |
aTIV | 2 |
39 |
28 |
55 |
[41] | [31] |
1 Working-age adults (18–64 years); 2 older adults (≥65 years); aTIV, MF59-adjuvanted trivalent influenza vaccine; hdQIV, high-dose quadrivalent influenza vaccine; QIVc, cell-based quadrivalent influenza vaccine; QIVr, recombinant quadrivalent influenza vaccine.
It has generally been found
[40,41,42][30][31][32]
1 Working-age adults (18–64 years); 2 older adults (≥65 years); 3 COVID-19 vaccine + placebo vs. COVID-19 vaccine + seasonal influenza vaccine; 4 mRNA-1273 + hdQIV vs. hdQIV alone; 5 NVX-CoV2373 + seasonal influenza vaccine vs. placebo + NVX-CoV2373 alone, geometric mean ratios were calculated from the geometric mean titers provided by the authors; aTIV, MF59-adjuvanted trivalent influenza vaccine; hdQIV, high-dose quadrivalent influenza vaccine; NA, not available; QIVc, cell-based quadrivalent influenza vaccine; QIVr, recombinant quadrivalent influenza vaccine.