Rapid changes in serum cytokines and chemokines in response to inactivated influenza vaccination

Background The timing of host cytokine responses to influenza vaccination is poorly understood. Objectives We examined serum cytokine kinetics following inactivated trivalent influenza vaccine (TIV) to better understand potential relationships between markers of inflammation and TIV‐related side effects. Patients/Methods Twenty healthy adult subjects received TIV. Cytokines/chemokines were assessed in intervals from 3 hours to 14 days. Antibody titers were measured at baseline and Day 14. Results Serum cytokine responses to TIV were evident as early as 3 hours post‐immunization. Compared to baseline, IFN‐γ and IP‐10 were significantly elevated 7 hours after TIV administration. Both remained elevated and peaked between 16 and 24 hours before returning to baseline by 44 hours post‐vaccination. Although IL‐8 levels were variable between subjects during the first 24 hours after TIV, by 44 hours, IL‐8 was significantly lower compared to baseline. Interestingly, IL‐8 levels remained significantly lower for up to 2 weeks after receiving TIV. Fifteen of 20 subjects reported mild adverse events. The one subject who reported moderate myalgias and injection site pain after vaccination displayed a distinctive, early cytokine response profile which included IL‐6, IL‐2, IL‐8, IP‐10, MCP‐1, TNF‐α, TARC, and MCP‐4. Conclusions Serum cytokines changed rapidly following TIV and generally peaked at 24 hours. Trivalent influenza vaccine‐induced reductions in IL‐8 occurred later (44 hours) and were sustained for 2 weeks. An outlier response coincided with the only moderate side effects to the vaccine. These data suggest that early cytokine/chemokine responses may provide additional insight into the pathogenesis of adverse events and immune reactivity to vaccination.


| INTRODUCTION
Cytokines and chemokines play an integral, yet somewhat paradoxical role in host defense against influenza. For example, type I interferons have strong antiviral activities and can directly inhibit influenza virus replication. 1,2 Meanwhile, excessive cytokine/chemokine responses have been associated with more severe disease during the 2009 H1N1 pandemic, 3 lung damage in macaques infected with the 1918 influenza virus, 4 and fatal H5N1 infection in humans. 5 The role of cytokines in influenza vaccine responses is less clear.
In the case of the smallpox vaccine, cytokines are linked not only with vaccine efficacy but also with adverse events. 6 A frequently cited rationale for avoiding annual influenza vaccines is concern about experiencing side effects. [7][8][9][10][11] Public concerns about vaccine side effects can undermine immunization programs, including national or statewide seasonal influenza vaccine campaigns. 12,13 Although local and systemic adverse events are generally transient and short-lived after influenza vaccines, 14 predictable post-vaccine reactogenicity events like myalgia and malaise in the first 2 days after influenza vaccination have led many to the misperception that the vaccine "gave them the 'flu'". 15,16 Although this phenomenon has been well described, 16 there are little data to explain the biologic basis of these events and the relationship (if any) with cytokine responses.
Recently, systems biology approaches have been used to prospectively explore the molecular determinants of influenza vaccine responses including efficacy and/or adverse events. 17,18 Although different vaccines were used, both of these studies identified early immune gene expression signatures (1-3 days after immunization) which predicted immunogenicity 17 and the onset of clinical adverse events. 18 In line with these data, another report described an association between serum cytokines and subjective side effects in women 1-2 days after receiving TIV. 19 As the role of cytokines in vaccine responses continues to be defined, we sought to more discretely characterize early serum cytokine kinetics in response to TIV in this proof-of-principle study. The study included 2 groups of vaccinees (n = 10/group) that collectively were assessed at baseline and 3 hours, 7 hours, 16 hours, 24 hours, 44 hours, and 14 days after vaccination for serum cytokines/chemokines, hemagglutination inhibition (HI) titers, and subjective side effects.

| Study design
This was an open-label study. The study was approved by the Johns Hopkins Bloomberg School of Public Health Institutional Review Board (IRB) and was conducted in accordance with the principles of the Declaration of Helsinki and the Standards of Good Clinical Practice (as defined by the International Conference on Harmonisation). All participants provided written informed consent.

| Assays
Hemagglutination inhibition titers were determined at baseline and Day 14 as described previously. 20 The sera were tested for antibodies  In addition, we measured C-reactive protein (CRP), serum amyloid A (SAA), and the soluble cell adhesion molecules sVCAM-1 and sICAM-1 using the MSD Vascular Injury Panel II. Meso Scale Discovery plates were analyzed on the SECTOR Imager 2400 as previously described. 21 All samples were run in duplicate.

| Statistics
Statistical analyses were performed using Prism software v4.0c (GraphPad, San Diego, CA, USA) and STATA (VERSION) (StataCorp, College Station, TX, USA). Differences in median cytokine levels were assessed using the Wilcoxon signed rank test. Corrections were not made for multiple comparisons because this is a pilot study and we wanted to explore potential signals. The baseline statistics for each cohort was summarized using a t-test for continuous data and chi-square for contingency data. For the dynamics of the cytokine response, a nonparametric test was performed to look at paired data on selected cytokines. Nonresponder values were assigned the lower limit of quantitation value.
Cohorts were compared in terms of baseline characteristics (socioeconomic status, demographics, antibody titers at baseline and at 14 days, and the number of non-responders).

| Demographics
This study was conducted in January and February 2012. Twenty-six subjects were screened and consented ( Figure 1). Twenty subjects received TIV. Baseline characteristics of the study participants are shown in Table 1. The average age for all participants was 37 years old, with a range from 24 to 48 years. Nine of 20 subjects were female and all except two were Black, reflecting our population in East Baltimore. The only difference between cohorts 1 and 2 was in gender-in Cohort 1, 70% of subjects were female, compared with 20% in Cohort 2 (P = .03 Wilcoxon signed rank test). One subject in Group 1 did not return for

| Antibody responses to trivalent influenza vaccine
Approximately 90% of the subjects had a ≥4-fold increase in HI titers at Day 14 to each of the three antigens included in the vaccine ( Table 2).
Responses in the 2 cohorts were comparable to all 4 antigens tested.
Most subjects had a ≥4-fold increase in HI titer to the B/Wisconsin

| Cytokine/chemokine responses to trivalent influenza vaccine
Compared to baseline values, significant differences in median cytokine/chemokine levels were noted at one or more time points for all of the proinflammatory and chemokine analytes tested with the exception of eotaxin (Table 3) patterns of change over time for many of the analytes due to considerable interindividual variability. IFNγ, IP-10, and IL-8 had distinct temporal profiles with considerable overlap between the groups ( Figure 2, Table 3). Significant increases in IFNγ occurred by 7 hours and peaked between 16 and 24 hours before returning to baseline by 44 hours post-vaccination. Similarly, IP-10 also increased significantly by 7 hours, with a peak at 24 hours, declining at 44 hours, with a return to baseline by Day 14. No significant differences in IL-8 were observed at the earliest time points, but at 44 hours post-vaccination, IL-8 levels were decreased compared to baseline (Figure 2, Table 3).
As we reported previously, IL-8 remained low at Day 14. 22 One subject (subject 8; S8) had a distinctively robust change in IL-

| Adverse events
Ten subjects reported generalized myalgia after vaccination. Eight were mild in nature; subjects 3 and 8 reported moderate myalgia. The mean time to myalgia onset was 5 hours with a mean duration of 20.7 hours.
In subjects who reported myalgia, median serum MCP-1 levels were greater at baseline and through the first 44 hours post-vaccination than among those who did not report myalgia (Table 4). Eight subjects reported injection site pain; 7 of them reported mild pain, and subject 8 reported moderate pain. The mean time to pain onset post-vaccination was 6.25 hours with a mean duration of 21 hours. Pre-and postvaccination serum IP-10, IL-2, IL-6, IL-8, and TNFα levels tended to be higher, while serum MIP-1β levels tended to be lower in subjects who reported injection site pain than in those who did not report pain.
In addition, subject 2 (who had reported injection site pain and myalgia) also reported 2 episodes of mild diaphoresis. The first started  (Table 4). Subject 8, as mentioned above, had both moderate injection site pain and moderate myalgia.  Meanwhile, another report identified a correlation between injection site soreness and serum cytokines 1-2 days after receiving TIV in healthy pregnant and non-pregnant women. 19 In our study, we extend our analysis into the first hours following TIV administration and identified temporal patterns of serum cytokine and chemokine changes which occurred as early as 3 hours postimmunization, generally peaking at approximately 24 hours.

| DISCUSSION
We observed significantly elevated levels of IFNγ and IP-10 beginning at 7 hours and remained elevated at 24 and 44 hours after vaccination, respectively ( Figure 2). In addition, we also found that serum IL-8 levels were reduced after 44 hours and remained so for up to 14 days.
It is important to acknowledge the limitations of this study. The relatively small number of subjects included in this report represents  For example, our observation that serum IFNγ and IP-10 (CXCL10) levels were both elevated 1 day after vaccination was also reported in volunteers given a monovalent 2009 H1N1 vaccine. 18 In that study, IP-10 was the only soluble marker associated with adverse events.
Furthermore, the decreased levels of IL-8 we reported here replicate our previous findings from a larger independent cohort. 22 Interestingly, Christian et al also found that TIV resulted in decreased serum IL-8 levels in non-pregnant women. 28 As mentioned, the same group followed side effects and serum cytokines daily for 3 days after TIV vaccination in pregnant and non-pregnant women. They found that at baseline, women who reported more arm soreness had lower IL-6 and IL-8 levels and higher IL-1β than those that did not, and those women also had higher TNFα and macrophage migration inhibitory factor (MIF) levels in the days following vaccination. 19 Although we did not measure MIF, the other group did not assess IP-10, and our cohorts are rather different, we did find that volunteers with injection site pain had higher IP-10 and IL-6 levels at baseline and after vaccination, as well. In addition, we noted that volunteers who experienced myalgia had elevated MCP-1 levels at baseline as well as 3 hours and 7 hours after vaccination compared to those who did not report myalgia (Table 4). We did not however see a statistically significant change in IL-8 between those with symptoms and those without, although that may be due to our small sample size.
Like others, we observed considerable interindividual variation in the levels of these markers of inflammation which could explain in part the variability in reported symptoms following influenza vaccination. 29 Although our power to detect effects was somewhat restricted by the sample size, we noted far less intra-individual variability in that subjects with higher median levels tended to remain higher while subjects with lower median levels tended to remain lower.
Taken together, our data support a growing body of literature indicating that soluble markers of inflammation may serve as a muchneeded early indicator of vaccine reactogenicity and risk for adverse events. 18 Importantly, our findings indicate that peripheral cytokines begin to change in the hours immediately following vaccination and warrant further exploration in larger studies to determine the biologic basis of clinical symptoms associated with vaccination. Improving our basic understanding of the immune response(s) to vaccination may enhance influenza vaccine development efforts and public health safety.