Serum biomarker profile orchestrating the seroconversion status of patients with autoimmune diseases upon planned primary 17DD Yellow fever vaccination

The present study aimed to investigate whether the serum biomarkers of immune response orchestrate the seroconversion status in patients with autoimmune diseases (AID) upon planned primary 17DD-YF vaccination. For this purpose a total of 161 individuals were enrolled in a prospective study, including patients with Rheumatoid Arthritis (RA = 38), Spondyloarthritis (SpA = 51), Systemic Lupus Erythematosus (SLE = 21) and Sjögren’s Syndrome (SS = 30) along with a group of healthy controls (HC = 21). Analysis of plaque reduction neutralization test (PRNT) titers and seropositivity rates along with the 17DD-YF viremia and serum biomarkers were carried out at distinct time points (D0/D3–4/D5–6/D7/D14–28). The results demonstrated an overall lower PRNT titer and seropositivity rate (170 vs. 448; 77 vs. 95%) in AID as compared to HC, especially in SpA and SLE subgroups. No significant differences were observed in the viremia levels amongst groups. In general, a more prominent serum biomarker response was observed in AID as compared to HC, throughout the timeline kinetics. Remarkably, AID/PRNT(−) exhibited higher levels of several biomarkers at baseline as compared to AID/PRNT+. Moreover, while AID/PRNT(+) exhibited earlier increase in serum biomarkers at D3–4/D5–6, the AID/PRNT(−) displayed higher response at later time points (D7/D14–D28). Of note, a synchronic increase of IFN-γ at the peak of viremia (D5–6) was observed in HC and AID/PRNT(+) groups, whereas a later asynchronous IFN-γ response was reported for AID/PRNT(−) at D7. The biomarker profile tends to deflate at post-vaccination timeline, highlighting a putative immunomodulatory effect of live attenuated 17DD-YF vaccine in AID/PRNT(+), but not in AID/PRNT(−). Altogether these data suggested that inflammatory status prior vaccination, low IFN-γ at viremia peak and the occurrence of asynchronous biomarker storm after 17DD-YF vaccination may orchestrate the lack of neutralizing antibody response γ.


Study groups
Seropositivity rates % (n) GeoMean (95%CI)  Table 2. Viremia Levels in Patients with Autoimmune Disease upon Planed Primary 17DD-YF Vaccination. * Data are reported as mean copies of 17DD-YF RNAnemia at peak ± standard error. No significant difference by no-parametric Kruskal-Wallis analysis (p = 0.5056) followed by Dunn's post-test for sequential pairwise comparisons and a threshold p value of < 0.05 was considered for statistical significance.
Baseline profile of serum biomarkers in patients with autoimmune disease according to the neutralizing antibody status after planned primary 17DD-YF vaccination. Aiming at investigating whether the Day0 baseline profile of serum biomarker was associated with the seroconversion status achieved after planned primary 17DD-YF vaccination, the AID patients were categorized according to the PRNT status after primary 17DD-YF vaccination as AID/PRNT(−) and AID/PRNT(+) and the biomarker levels prior planned primary 17DD-YF vaccination compared between AID subgroups. The results are presented in the Fig. 3. Data analysis demonstrated that AID/PRNT(−) exhibited higher levels of several biomarkers at Day0 baseline (CXCL8, CCL11, CXCL10, TNF-α, IL-15, IL-17, IL-10, G-CSF, GM-CSF) as compared to AID/ PRNT(+), suggesting that an inflammatory status prior vaccination may orchestrated the lack of antibody response (Fig. 3). CXCL8  CCL11  CCL3  CCL4  C CL2  C CL5  CXCL10   IL-1β  IL-6  TNF-α  IL-12  I FN-γ  IL-15  I L-17 IL-1Ra I L-4 IL-5 I L-9 I L-10 IL-13   (CXCL8, CCL11, CCL3,  CCL4, CCL2, CCL5, CXCL10), pro-inflammatory cytokines (IL-1β, IL-6, TNF-α, IL-12, IFN-γ, IL-15, IL-17),  regulatory cytokines (IL-1Ra, IL-4, IL-5, IL-9, IL-10, IL-13), and growth factors (FGF-basic, PDGF, VEGF, G-CSF, GM-CSF, IL-2 and IL-7) in serum samples from Autoimmune Disease patients (AID, black filled square, n = 140) and Healthy Controls (HC, white filled circle, n = 21). Measurements were carried out using the Luminex platform, according to manufacturer's instructions as provided in Material and Methods. The results are expressed as median baseline fold changes ± variation at each time point along the kinetic follow-up (D3-4, D5-6, D7 and D14-28) according to the paired sample collected at D0. Data analysis was performed considering the baseline fold value = 1.0 as the reference for decrease (< 1.0) or increase (> 1.0) in biomarker levels along the kinetic timeline. www.nature.com/scientificreports/ Serum biomarker signature in patients with autoimmune disease after planned primary 17DD-YF vaccination. The serum biomarker signature of patients with autoimmune diseases were assessed at distinct time points after planned primary 17DD-YF vaccination and the results presented in Fig. 4. Data analysis demonstrated that, in general, AID patients exhibited a more prominent increased in serum biomarkers levels as compared to health controls. In fact, a massive up-regulation of several soluble mediators was observed in AID patients at Day3-4 and Day5-6 with subsequent decrease at Day7 towards Day14-28 (Fig. 4A). Noteworthy, was that while HC presented a synchronic increase of IFN-γ at the peak of viremia (Day5-6) with late increase of IL-10 at Day7, AID patients displayed a simultaneous and persistent increase of IFN-γ and IL-10 at the peak of viremia (Day5-6) up to Day7 after vaccination (Fig. 4A). Heatmap analysis further illustrated these synchronic/asynchronous phenomena observed for the IFN-γ and IL-10 rhythm nearby the viremia peak (Fig. 4B).
Serum biomarker signatures after planned primary 17DD-YF vaccination according to the type of autoimmune disease. Serum biomarker signatures were further assembled for AID patients according to the type of autoimmune disease (Fig. 5). The results demonstrated that SpA and SLE patients presented early increase in serum biomarkers upon planned primary 17DD-YF vaccination, more prominent at Day3-4  Conversely, RA and SS subgroups exhibited a robust and persistent increase in serum biomarkers upon planned primary 17DD-YF vaccination. While the RA subgroup still presented a decrease in several biomarkers later at Day14-28, the SS subgroup maintained most serum biomarkers elevated throughout the kinetic timeline after planned primary 17DD-YF vaccination (Fig. 5). Of note, was the finding that IFN-γ levels increase at the peak of viremia (Day5-6) in most AID subgroups, except for the SLE subgroup which presented IFN-γ up-regulation early at Day 3-4 ( Fig. 5 and Supplementary Figure1).
Serum biomarker signatures in patients with autoimmune disease according to the neutralizing antibody status after planned primary 17DD-YF vaccination. In order to further characterize the changes in serum biomarkers observed after planned primary 17DD-YF vaccination and its association with the post vaccination PRNT status, the AID patients were categorized as AID/PRNT(−) and AID/PRNT(+) and the fold changes in biomarker levels compared between AID subgroups (Fig. 6). Data analysis showed that AID/ PRNT(+) exhibited an earlier serum biomarker response with higher fold change values at Day3-4 and Day5-6 with following decrease in a range of soluble mediators at Day7 towards Day14-28 (Fig. 6A). On the other hand, AID/PRNT(−) presented higher fold change in serum biomarkers at late time points with a progressive increase www.nature.com/scientificreports/ towards Day7, maintaining a sustained up-regulation of serum biomarkers up to Day14-28 (Fig. 6A). These findings suggest the occurrence of putative immunomodulatory effect of live attenuated 17DD-YF vaccine associated with the seroconversion status observed in AID (Fig. 6A).

CCL3
IL-17 Color  . The results are expressed as the proportion of subjects with increased biomarker levels (baseline fold change values > 1). Data analysis was carried out considering the 50th percentile as the reference to identify the set of biomarkers with high proportion of subjects with levels above the global median cut-off along the kinetic timeline. Those biomarkers with proportion of subjects with levels above the global median cut-off were highlighted with gray scale rectangles. (B) Heatmaps were constructed considering the baseline fold change values at each time point along the kinetic follow-up (D3-4, D5-6, D7 and D14-28). This approach was employed to draw the overall change in the serum biomarkers profile after primary 17DD-YF vaccination of Autoimmune Disease patients (AID, n = 140) and Healthy Controls (HC, n = 21). Data interpretation was carried out based on the color keys employed to underscore the baseline fold value = 1.0 as the reference for unaltered levels (black filled square), the baseline fold value < 1.0 for decreased levels (green filled square) and the baseline fold value > 1.0 for increased levels (red filled square), according to the paired sample collected at D0.

Discussion
The 17DD-YF vaccine induces a safe and effective protective immunity in healthy vaccinees leading to high levels of protection in healthy adults (95-98%) resultant of robust humoral and cellular immunity 7,[15][16][17][18][19] . Recent evidences have shown that planned primary 17DD-YF vaccination is safe and immunogenic but lower seropositivity rate is overall observed in AID patients upon planned 17DD-YF primary vaccination 14 .
The present study aimed to investigate whether the serum biomarkers of immune response orchestrate the distinct seroconversion rates in AID patients, leading to positive or negative results in plaque reduction neutralization test, depending on the type of disease. For this purposes, four subgroups of AID patients (RA, SpA, SLE and SS) were enrolled in an observational phase IV controlled prospective analysis to quantify the serum biomarker timeline profiles (D0, D3-4, D5-6, D7, D14-28) along with the 17DD-YF RNAnemia, taking into account the seropositivity rates of neutralizing antibodies (PRNT ≥ 1:50) after planned primary 17DD-YF vaccination.
Overall, our results demonstrated lower PRNT seropositivity rate in AID patients with decreased GeoMean titers as compared to HC. Campi-Azevedo et al. 20 have previously shown that distinct patterns of viremia kinetics may affect the 17DD-YF vaccine immunogenicity in healthy adults. These authors postulated that early viremia peak after primary vaccination was associated with higher seroconversion rates, while lower and late viremia peak was observed in primary vaccinees exhibiting lower seroconversion rates. According to these authors a sub-optimal antigen exposure would lead to the generation of impaired protective humoral response, regardless the ability of live-attenuated virus to replicate and amplifies the antigenic exposure 20 . Our data demonstrated that the viremia levels at peak did not differ between AID/PRNT(−) and AID/PRNT(+) as compared to HC. Therefore, distinct viremia profiles seem to not substantiate the differences observed in the seroconversion rates in subgroups of AID.  . The results are expressed as the proportion of subjects with increased biomarker levels (baseline fold change values > 1). Data analysis was carried out considering the 50th percentile as the reference to identify the set of biomarkers with high proportion of subjects with levels above the global median cut-off along the kinetic timeline. Those biomarkers with proportion of subjects with levels above the global median cut-off were highlighted with gray scale rectangles. www.nature.com/scientificreports/ It has been proposed, besides the viremia profile, the analysis of systemic immunological mediators should be also evaluated together with PRNT titers to better understand the immune response triggered by the 17DD-YF vaccine and its relationship with the development of protective antibodies 20 . A detailed analysis of several serum biomarkers, including chemokines and cytokines together with the viral load profile and the levels of neutralizing antibodies was carried out in healthy adults receiving distinct doses of 17DD-YF vaccine in a time and dose dependent fashion. Distinct kinetic profiles of chemokines, pro-inflammatory and modulatory cytokines were observed according to the seroconversion rates 20 .
Our data demonstrated that, in general, the 17DD-YF vaccine triggered a more prominent serum biomarker response in AID as compared to HC, throughout the timeline kinetics. Remarkably, AID/PRNT(−) exhibited higher levels of several biomarkers prior vaccination as compared to AID/PRNT(+). These differences in   www.nature.com/scientificreports/ baseline profiles of serum biomarkers suggested that an inflammatory status prior vaccination may orchestrate the lack of antibody response. The analysis of changes in serum biomarker levels throughout the time line kinetics showed that while the AID/PRNT(+) exhibited an earlier (Day3-4/Day5-6) higher fold changes in serum biomarker, the AID/PRNT(−) displayed higher fold changes at late time points (Day7/Day14-D28). Of note, a synchronic pattern of high IFN-γ levels at the day of peak viremia (D5-6) was observed in HC and AID/PRNT(+) groups, whereas an asynchronous IFN-γ response was reported for AID/PRNT-later at Day7. Interestingly, AID/ PRNT(+), biomarkers tend to deflate the biomarker response at post-vaccination timeline highlighting a putative immunomodulatory effect of live attenuated 17DD-YF vaccine. This effect was not observed in AID/PRNT(−). The synchronic/asynchronous rhythm of pro-inflammatory/modulatory serum biomarkers detected along the timeline kinetic, particularly nearby the peak of viremia (Day5-6), may be associated with distinct seroconversion profile observed in AID subgroups. Campi-Azevedo et al. 20 have postulated that asynchronous production of IL-10 opposite to the viremia peaks and pro-inflammatory cytokines (TNF-α and IFN-γ). The hypothesis was that the decreased levels of IL-10 at the peak of viremia support a pro-inflammatory microenvironment to produce neutralizing antibodies 20 . Corroborating this hypothesis, it has been previously demonstrated that the lack of seroconversion after YF-17DD primary vaccination was associated with lower levels of pro-inflammatory cells (TNF-α + neutrophils and monocytes) and correlated with enhanced levels of IL-10 produced by CD8 + T-cells 21 .
A remark of the present investigation was the opportunity to perform a parallel analysis of YF-specific neutralizing antibody and the quantification of serum biomarkers along the kinetics timeline after planned 17DD-YF primary vaccination, focusing on patients with distinct types of AID. Our finding demonstrated that amongst AID subgroups, SpA and SLE patients exhibited the lowest seropositivity rate. The analysis of viremia levels at peak did not show differences amongst AID subgroups (RA, SpA, SLE and SS) as compared to HC. Again, the viremia profiles did not support the differences observed in the seroconversion rates in the subgroups of AID. The results comprising the serum biomarker signatures after planned primary 17DD-YF vaccination according to the type of autoimmune disease demonstrated that SpA and SLE patients presented early increase in serum biomarkers (Day3-4) with subsequent decrease in most soluble mediators (Day14-28). RA and SS subgroups exhibited a robust and persistent increase in serum biomarkers upon planned primary 17DD-YF vaccination. Noteworthy was that most AID subgroups exhibited a synchronic IFN-γ levels increase at the peak of viremia (Day5-6), except the SLE subgroup which displayed early up-regulation of IFN-γ at Day 3-4. However, unlikely observed for HC, all AID subgroups displayed an up-regulation of IL-10 at the peak of viremia.
In SLE patients, it is possible that the asynchronous IFN-γ production and the up-regulation of IL-10 at peak of viremia may explain the lower seroconversion rates observed. Kyogoku et al. 22 have performed a detailed analysis describing the differences in the IFN signature in SLE patients as compared to healthy donors. Transcriptional responses of peripheral blood monocyte subsets and CD4 + T-cells demonstrated that IFN-associated gene signature was distinct in monocytes and CD4 + T-cells and from SLE as compared to healthy controls. In fact, it was observed that healthy donors, even days after YF vaccine immunization displayed a virus-induced monocyte and CD4 + T-cell signatures comprising only approximately 36-47% and 10% of the probe-sets identified in SLE, respectively 22 . The analysis of IFN signatures showed qualitative and quantitative differences in SLE, characterized by a complex compiled gene patterns with increased expression levels. In this sense the assigned "autoimmune-specific" IFN signatures observed in SLE patients may explain the early IFN-γ production observed in our results upon 17DD-YF primary vaccination.
As far as the SpA patients, despite the synchronic IFN-γ levels increase at the peak of viremia (Day5-6), the concomitant up-regulation of IL-10 at Day5-6 may contribute to the lower seroconvertion rate observed. It is possible that a hyper-inflammatory state prior 17DD-YF vaccination plays a role tuning the immune response in these patients. Increasing evidences from mouse studies have been accruing to support that cytokine dysregulation, especially the TNF-α/IL-17/IL-23 axes is an emerging picture of aberrant immune hyper-activation in human SpA. As the role of inflammatory cytokines (TNF-α, IL-17 and IL-23) may differ across the group of spondyloarthritis diseases 23,24 , other factors may impact the seroconversion rates in SpA patients, such as: the disease activity at baseline as well as the type of immunomodulatory therapy schemes prior vaccination. The influence of such conditions remains to be elucidated in further investigations.
The present work has some limitations. The sample size, especially the number of AID/PRNT(−) subjects is relatively small requiring further validation in larger study investigations. Moreover, we did not analyze the cellular immunity profile in AID subgroups upon planned 17DD-YF primary vaccination. In this study, we were unable to categorize the AID patients according to the use of non-biological or biological immunomodulatory therapy, due to restrictions in sample size. Furthermore, we did not investigate the impact of disease activity on the effectiveness of planned 17DD-YF primary vaccination neither follow-up the long-term changes in disease activity over time after vaccination. These issues are currently under investigation in the ongoing multicenter study referred as OCAMO Project -Brazilian cohort of immunogenicity and safety of live-attenuated vaccines, carried out by the Collaborative Group for the study of MMR (Measles, Mumps and Rubella) and Yellow Fever Vaccines.
Altogether these data suggested that inflammatory status prior vaccination, low IFN-γ at viremia peak and the occurrence of asynchronous biomarker storm after 17DD-YF vaccination orchestrate the seroconversion status of patients with autoimmune diseases upon planned primary 17DD Yellow Fever vaccination. It is possible that other factors such as disease activity as well as the use of non-biological or biological immunomodulatory therapy schemes prior vaccination may also impact the seroconversion profiles of AID subgroups. Future studies including these topics will further clarify the distinct seroconversion rates observed amongst types of AID as compared to healthy controls. The inclusion criteria comprised: individuals older than 18 years, with no previous records of YF vaccination. The following criteria were considered for the volunteers comprising the AID group: patients who fulfilled the international classification criteria for autoimmune diseases diagnosis, based on the American College of Rheumatology and/or European League Against Rheumatism guidelines [25][26][27][28][29][30] ; patients in remission or with low disease activity advised by a rheumatologist to receive planned 17DD-YF primary vaccination upon withdrawal of immunomodulatory therapy with interruption interval as specified by Brazilian recommendations for YF vaccination of AID patients 13 . Details about immunomodulatory therapy prior withdrawal are provided in the Supplementary Table 1. The exclusion criteria comprised: previous history of YF vaccination; medical advice to not receive the YF vaccine; subjects who refuse to participate in the study; primary immunodeficiency or immunosuppression by other causes: HIV carriers with CD4 count lower than 200 cells/mm 3 or lymphocyte counts lower than 500 cells/mm 3 , low IgM or IgG levels (according to patient's information for previous medical records); history of organ transplantation; neoplasia; subjects who received another vaccine simultaneously or within a 30 days interval.
The compendium of study design and methods is provided in the Fig. 7.
Biological samples. Blood

YF-plaque reduction neutralizing test (PRNT).
The quantification of YF-specific neutralizing antibody was carried out at the Laboratório de Tecnologia Virológica, Bio-Manguinhos/FIOCRUZ according to the standard protocol of plaque-reduction neutralization test (PRNT) described previously by Simões and colleagues, 2012 31 . The results were expressed as the reciprocal of the last serum dilution which reduced the plaque numbers in 50% (PRNT 50 ) relative to the virus control included in each batch. Seropositivity was considered for PRNT titers higher than 1:50.

Data availability
The datasets generated and/or analyzed during the current study are available from the corresponding author upon request.