We hypothesized that individuals who develop fever after smallpox vaccination have genetically determined differences in their immune responses to vaccinia virus. We looked for an association between the development of fever and single-nucleotide polymorphisms (SNPs) in 19 candidate genes in 346 individuals previously assessed for clinical responses to smallpox vaccination. Fever after smallpox vaccination is associated with specific haplotypes in the interleukin (IL)–1 gene complex and in the IL18 gene. A haplotype in the IL4 gene was highly significant for reduced susceptibility to the development of fever after vaccination among vaccinia-naive individuals. Our results indicate that certain haplotypes in the IL-1 gene complex and in IL18 and IL4 predict an altered likelihood of the development of fever after smallpox vaccination. Our findings also raise the possibility that these same haplotypes may identify individuals at risk for the development of fever after receipt of other live virus vaccines, providing information that could be useful in anticipating and preventing more-serious adverse events.
Smallpox vaccination with live vaccinia virus causes adverse reactions that range from the very mild (e.g., fatigue) to the life-threatening (e.g., progressive vaccinia) . In a previous report, when 680 vaccinia-naive adults received primary vaccination with Dryvax (New York Board of Health strain), a significant proportion developed a constellation of symptoms including fatigue, headache, muscle aches, and fever that we termed “acute vaccinia syndrome” (AVS) . Fever, defined as a temperature ⩾37.7°C, is the most objective measure of AVS and occurred in ∼13% of the newly vaccinated individuals . In children, the rate and magnitude of fever after vaccination appeared to be higher, with 15%–20% of children showing a temperature >102°F . The incidence of fever appears to be much lower after revaccination in both adults and children, suggesting that host immunity may mitigate the response [3, 4].
Why do only 15% of newly vaccinated individuals develop fever after inoculation with vaccinia virus? What differentiates this population from those individuals who do not develop fever or other adverse events? We hypothesized that the adults who develop fever after smallpox vaccination may have genetically determined differences in their immune responses to vaccinia virus that lead to either higher levels of viral replication or more-vigorous inflammatory responses. To evaluate this hypothesis, we tested for an association between AVS and single-nucleotide polymorphisms (SNPs) in 19 candidate genes linked to the control of poxvirus infections, innate immunity, and inflammation, in a population of individuals previously assessed for clinical responses to smallpox vaccination.
Study design and subjects. This was a callback study for individuals who had participated in 1 of 3 previous studies of smallpox vaccination (National Institute of Allergy and Infectious Diseases Division of Microbiology and Infectious Diseases study numbers 00-005, 01-632, 01-651, and 02-007) conducted at the Saint Louis University Vaccine Treatment and Evaluation Unit (VTEU) [2, 5, 7]. The present study was approved by the Saint Louis University and Washington University institutional review boards. Subjects were contacted by letter and invited to return to the Saint Louis University VTEU for evaluation of a possible genetic basis for adverse events after smallpox vaccination. Informed consent was received from all subjects. Participation involved a single visit, to obtain informed consent and a blood sample; 346 subjects were enrolled during the period of 9 February 2004–12 June 2004.
Genotyping. DNA was obtained from whole blood by standard methods. Genotyping was performed using the high-throughput FP-TDI (template-directed dye terminator incorporation with detection by fluorescence polarization) assay, as described elsewhere . Nineteen genes linked to either inflammatory processes or host defense against viral infections, including vaccinia, were selected for analysis: IL1A (interleukin-1, alpha) , IL1B (interleukin-1, beta) , IL1R1 (interleukin-1 receptor, type I) , IL4 (interleukin-4) , IL8 (interleukin-8) , IL10 (interleukin-10) , IL12A (interleukin-12, alpha) , IL12B (interleukin-12, beta) , IL18 (interleukin-18 [interferon-gamma–inducing factor]) , IL21 (interleukin-21) , IFNG (interferon, gamma) , IRF1 (interferon regulatory factor 1) , FCGR2A (Fc fragment of IgG, low-affinity IIa, receptor [CD32]) , FCGR3A (Fc fragment of IgG, low-affinity IIIa, receptor [CD16a]) , TNFA (tumor necrosis factor, alpha) , TLR1 (Toll-like receptor 1) , TLR4 (Toll-like receptor 4) , MYD88 (myeloid differentiation primary response gene ) , and VDR (vitamin D receptor) . All SNPs were identified from public data bases, and SNP information was downloaded from dbSNP (http://www.ncbi.nlm.nih.gov/SNP/) and/or from the International HapMap project (http://www.hapmap.org/). Assays were designed for all SNPs available from each of the genes studied and included SNPs from both coding and promoter regions. Primers were designed using Primer3 (release 0.9; code available at: http://primer3.sourceforge.net/), with some previously described modifications [18, 19]. A list of all the SNPs for each gene and the primers used for FP-TDI sequencing can be found at http://snp.wustl.edu/AVS/.
Resequencing of the IL1A gene. Chromosomal coordinates for the untranslated regions (UTRs) and coding regions of IL1A were obtained from the University of California Santa Cruz Genome Browser Database for the latest build of the human genome (hg17) . Overlapping regions of interest were combined, resulting in 15 resequencing targets that covered the 7 exons, 2 UTRs, 23 conserved regions, and 1000 upstream bases for IL1A. Resequencing was performed using polymerase chain reaction primers, as described elsewhere . Resequencing was performed on a subset of 24 samples (12 from case patients and 12 from control subjects) from white, non-Hispanic individuals. A χ2 test of the genotype data from the IL1A region was used to identify SNP alleles with the highest correlation to the fever/AVS phenotype. The 12 fever-positive individuals carrying the highest number of IL1A risk alleles were selected for resequencing. An equal number of matched control subjects were randomly selected from the fever-negative individuals. Sequencing traces were compared with the reference sequence on which all known SNPs had been placed. Both the SNPs and reference sequences were downloaded from public databases. Sequence traces were analyzed using Sequencher (version 4.1; Genecodes). Results were tabulated for all unknown and known SNPs in each sample by use of a simple Excel table.
Statistical analyses. SNPs were coded as 0, 1, or 2, depending on whether the subject was homozygous for the major allele or was heterozygous or homozygous for the minor allele. Of the 357 SNPs, 254 were found to be diallelic across the set of subjects studied. SNPs were first analyzed by χ2 analysis and the Cochran-Armitage test for trend; subsequently, linear model analysis was used to identify SNPs that differed between the patients with fever and those without any fever after vaccination. Vaccine status (naive vs. nonnaive) was found to be a significant predictor of development of fever. Therefore, we adjusted for vaccine status in the tests of association between individual SNPs and the outcome measures. To address the multiple-comparisons issue, a permutation test was used to determine the P value cutoff for statistical significance. The multiple-comparisons permutation analysis yielded a P value cutoff of .01. Within each locus, haplotypes were constructed using an expectation-maximization (EM) algorithm–based approach similar to that implemented in the software SNPHAP (D. Clayton, Department of Medical Genetics, Cambridge Institute for Medical Research). Haplotypes with estimated frequencies <1% were not used. A regression score test was then performed on the inferred haplotypes to test for outcome association, using the EM-derived maximum likelihood estimates of the haplotype frequencies and the posterior probabilities of the pairs of haplotypes for each subject. In this way, all possible haplotype configurations were used for the association test, rather than just the most probable haplotype . Haplotype associations were tested simultaneously in the regression model to identify haplotypes that were independently predictive of outcome.
Characteristics of the subjects. A total of 428 subjects who had participated in 1 of 3 previous smallpox vaccine studies [2, 5, 6] received letters of invitation to participate in the present study. Of these, 352 responded by phone, and 346 volunteers were enrolled. Of the 346 subjects enrolled, 176 had been vaccinia naive (never vaccinated) before being vaccinated in 1 of the vaccinia clinical trials, and 170 were vaccinia nonnaive, having been previously vaccinated before receiving Dryvax in 1 of the 3 previous studies [2, 5, 6]. Ninety-four of the enrolled subjects (61 naive and 33 nonnaive) had a documented temperature ⩾37.7°C within 3–15 days after their smallpox vaccination (case patients), with most febrile events recorded on days 7–12 after immunization . The remaining 252 individuals (114 naive and 138 nonnaive) without documented fever after smallpox vaccination served as the control group. The characteristics of the 2 populations are shown in table 1. Almost all of the patients identified themselves as being white (95%), with only 2.2% of individuals identifying themselves as being black/African American, 1% as being Asian, and 1.6% as being multiracial or other. A significantly higher proportion (65%) of the individuals who developed fever after vaccination had received smallpox vaccine for the first time (naive). In contrast, of individuals who did not experience fever after smallpox vaccine, 55% had been previously vaccinated against smallpox (P<.001). Because the individuals previously vaccinated against smallpox were generally older than the naive individuals, the prior vaccine status accounts for what is a statistically significant difference in mean age between those individuals who developed fever after vaccination and those who did not.
Identification of haplotypes associated with fever after smallpox vaccination. A total of 357 SNPs were assessed for each of the 346 subjects. Of the 123,522 possible observations, 9.7% could not be definitively genotyped due to technical issues and are treated as missing values in the analysis. The results for each of the individual SNPs examined are available at http://snp.wustl.edu/AVS/. Initial χ2 and trend tests for differences in genotype frequency between case patients and control subjects indicated that the most significant single SNP effects were located in IL1A, with 3 SNPs in the gene (rs3783550, rs1878319, and rs2071375) having nominal P values <.01 from both tests. Several of the significant SNPs in IL1R1 and IL1A showed additive effects—that is, the effect of the SNP was more pronounced in the homozygous than in the heterozygous state. Single SNP analyses were then refined to incorporate prior vaccination status in a linear model (data are available at http://snp.wustl.edu/AVS/) and used to generate the haplotype analysis.
As shown in table 2, we were able to identify a total of 8 haplotypes in 4 different genes—IL1A, IL1B, IL1R1, and IL18—that were associated with a differential risk for development of fever after smallpox vaccination. Three haplotypes were identified in IL1A, with haplotype 1 associated with a reduced risk for a fever after vaccination, haplotype 2 showing no significant effect, and haplotype 3 associated with an increased susceptibility for fever after smallpox immunization. Eight haplotypes were identified in IL1B in our population, 2 of which (haplotypes 7 and 8) were associated with a marked increase in susceptibility to fever after vaccination. However, both these haplotypes were present at low frequency in the population. A third gene on chromosome 2, IL1R1, had a single haplotype (10) that was associated with a moderately increased susceptibility to fever after smallpox immunization. A total of 10 haplotypes were identified in IL18 in our population, with 1 (haplotype 1) associated with a statistically significant reduced risk for fever after vaccination and 2 (haplotypes 9 and 10) associated with an increased susceptibility to fever.
The 8 haplotypes noted above were predictive for altered susceptibility to fever after smallpox vaccination in our total population, consisting of individuals receiving their first smallpox vaccination (naive) and those receiving a booster immunization (nonnaive). Next, we used a subgroup analysis to determine whether certain haplotypes were more predictive in naive than nonnaive individuals and to identify additional haplotypes that were predictive in naive or nonnaive individuals only and, thus, had been missed by analyzing the entire group. One haplotype in IRF1 (haplotype 3) was associated with increased susceptibility to fever after smallpox immunization in the vaccinia-nonnaive population only (table 3). We also identified a haplotype in IL4 (haplotype 2) that was highly significant for reduced susceptibility to fever after vaccination in the vaccinia-naive individuals only. By the subgroup analysis, 5 IL1R1 SNPs were significantly predictive in the vaccinia-nonnaive population and no IL1R1 SNPs were predictive in the naive population, indicating that the predictive haplotype for IL1R1 is important for nonnaive-individuals only.
Resequencing of the IL1A gene.IL1A contained both the largest number of SNPs  that showed a statistically significant association with altered susceptibility to fever after smallpox vaccination and the most significant single SNP associations from the linear model analysis. To determine whether there were additional SNPs in IL1A that might better predict the development of fever after vaccination, we performed resequencing of the IL1A gene in 24 individuals (12 case patients and 12 controls). We were able to identify 4 additional SNPs in IL1A in this population (data are available at http://snp.wustl.edu/AVS/), but none of the 4 were predictive for altered susceptibility to fever after vaccination.
We have found that the development of fever after smallpox vaccination is associated with specific haplotypes in the IL-1 gene complex on chromosome 2 and with haplotypes within the IL18 gene on chromosome 11. IL-1 is a proinflammatory cytokine that causes fever in humans in subnanomolar concentrations . Polymorphisms in the individual genes located in the IL-1 complex (IL1A, IL1B, IL1RN, and IL1R) have been linked to increased susceptibility to a variety of conditions, including atopy , the development of muscle inflammation after exercise , osteoarthritis , Behcet disease , ankylosing spondylitis , severe malaria , and stroke . Given the direct role played by IL-1 in causing fever, its association with febrile responses to vaccination is not unexpected, but this is the first direct linkage of polymorphisms in the IL-1 gene complex to individual differences in febrile responses.
IL-18, a member of the IL-1 cytokine superfamily, is also a proinflammatory cytokine whose receptor shares some transduction pathways with IL-1R [31, 32]. However, IL-18 does not directly cause fever. IL-18 induces IFN-γ production by T lymphocytes, NK cells, and macrophages and has antiviral effects against vaccinia in a murine model of infection [32–35]. Polymorphisms in IL18 have been linked to a number of conditions, including inflammatory bowel disease  and cardiovascular disease .
The possible involvement of IL-1 and IL-18 in adverse responses to smallpox vaccination is noteworthy in view of the important role played by each cytokine in experimental models of viral myocarditis. Although no patients in our cohort developed clinically apparent myocarditis after smallpox vaccination, among >492,000 US military personnel vaccinated between December 2002 and September 2003, there were 59 confirmed or probable cases of acute myopericarditis [37–39]. The underlying pathophysiology of this complication remains unknown, but an autoimmune component is considered likely . In murine models of coxsackie virus myocarditis, in which automimmunity plays an important role in disease, increased levels of both IL-1 and IL-18 were linked to myocardial inflammation, and blockade of IL-1 activity reduced both inflammation and mortality [40–43]. It would be of significant interest to determine whether the individuals who developed myocarditis after smallpox vaccination possess the IL1 or IL18 haplotypes linked to fever in our study.
A recent study analyzing cytokine levels after smallpox vaccination found an association between increases in systemic levels of IFN-γ and adverse events, including fever, in healthy adults . Direct administration of IFN-γ causes fever and flulike symptoms in recipients, consistent with a potential role of IFN-γ in AVS . Although we did not identify any SNPs or haplotypes in IFNG that were linked to fever after vaccination, our subgroup analysis identified a haplotype in IL4 that predicts a decreased risk for fever after smallpox vaccination in naive individuals but not in previously vaccinated (immune) subjects. This haplotype includes the SNP at −589 in the promoter region of IL4 (rs 2243250), where a C→T substitution is associated with increased production of IL-4 . IL-4 is an important regulatory cytokine that polarizes adaptive immunity toward Th2 responses and suppresses the appearance of IFN-γ–producing cells and Th1 responses . It is interesting to speculate that the association between the IL4 haplotype and a reduced risk of fever could be secondary to IL-4 inducing a relative reduction in Th1 and IFN-γ responses to vaccinia challenge in that subgroup. It should also be noted that increased IL-4 levels are known to have detrimental effects on the ability of hosts to control poxvirus infection in mouse models of disease, but there were no adverse events among our vaccinees with the IL4 haplotype to suggest that this was an issue with vaccinia immunization in this population .
The development of fever after vaccination is not confined to smallpox immunization and is not infrequent after the administration of other live virus vaccines. An important unanswered question is whether the same haplotypes that predict fever after smallpox vaccination may also identify individuals at risk for fever after the administration of other live virus vaccines and whether this information could be useful in anticipating and preventing more-serious adverse events. Fever is the most common adverse event after measles-mumps-rubella (MMR) vaccine administration, and temperatures in excess of 39.5°C were seen in 12% of children receiving MMR vaccine in one study . The rate of febrile seizures is known to increase after MMR immunization, presumably because of vaccine-induced fever, and there has been significant interest in identifying susceptible children . A recent study found that children with a sibling with a history of febrile seizures were at increased risk for febrile seizures after MMR vaccination, consistent with a possible genetic basis for increased susceptibility to this condition . The present study provides a basis for testing the hypothesis that children with febrile reactions after MMR administration are more susceptible due to certain haplotypes in IL1, IL4, or IL18.
We thank Zhi Zhang and Weimin Duan, for technical assistance and helpful discussions.