Influenza vaccination responses in human systemic lupus erythematosus: Impact of clinical and demographic features

Authors


Abstract

Objective

Vaccination against common pathogens, such as influenza, is recommended for patients with systemic lupus erythematosus (SLE) to decrease infections and improve health. However, most reports describing the vaccination response are limited to evaluations of SLE patients with quiescent disease. This study focuses on understanding the clinical, serologic, therapeutic, and demographic factors that influence the response to influenza vaccination in SLE patients with a broad range of disease activity.

Methods

Blood specimens and information on disease activity were collected from 72 patients with SLE, at baseline and at 2, 6, and 12 weeks after influenza vaccination. Influenza-specific antibody responses were assessed by determining the total serum antibody concentration (Bmax), relative affinity (Ka), and level of hemagglutination inhibition in the plasma. Using a cumulative score, the patients were evenly divided into groups of high or low vaccine responders. Autoantibody levels were evaluated at each time point using immunofluorescence tests and standard enzyme-linked immunosorbent assays.

Results

Compared to high responders, low responders to the vaccine were more likely to have hematologic criteria (P = 0.009), to have more American College of Rheumatology classification criteria for SLE (P = 0.05), and to be receiving concurrent prednisone treatment (P = 0.04). Interestingly, European American patients were more likely to be low responders than were African American patients (P = 0.03). Following vaccination, low responders were more likely to experience disease flares (P = 0.01) and to have increased titers of antinuclear antibodies (P = 0.04). Serum interferon-α activity at baseline was significantly higher in patients in whom a flare occurred after vaccination compared to a matched group of patients who did not experience a disease flare (P = 0.04).

Conclusion

Ancestral background, prednisone treatment, hematologic criteria, and evidence of increased likelihood of disease flares were associated with low antibody responses to influenza vaccination in SLE patients.

Systemic lupus erythematosus (SLE) is a prototypic systemic autoimmune disease characterized by the presence of autoantibodies and the involvement of multiple organs. Infectious diseases are one of the leading causes of morbidity and mortality in SLE patients, accounting for 11–23% of all hospitalizations and 20–55% of all deaths (1, 2). This increased susceptibility to infection is likely due to immunosuppressive therapy and intrinsic immune defects. Indeed, corticosteroid treatment equivalent to ≥20 mg/daily of prednisone has been shown to increase susceptibility to infection (1). Additionally, SLE patients display immune abnormalities, such as decreased antigen presentation and disrupted T and B cell interactions, which could decrease immune responses to pathogens (3–5). This increased risk of infection in SLE patients has led to an emphasis on vaccination in this at-risk population.

Influenza infection is a major cause of morbidity and mortality in the United States, resulting in more than 225,000 hospitalizations (6) and 36,000 deaths (7) annually. Immunocompromised individuals, such as SLE patients, are at high risk, for all of the reasons discussed above. Therefore, vaccination of SLE patients with the influenza vaccine has become part of the standard of care. However, several studies have shown that SLE patients experience lower responses to vaccinations than have been observed in healthy control subjects (8–10). Four studies, performed in the 1970s, assessed the anti-influenza response in SLE patients vaccinated against the circulating H1N1 virus. Two of the reports documented low seroconversion rates, determined according to the serum antibody titer and the level of hemagglutination inhibition (HAI), in SLE patients (47–48%) as compared to healthy controls (62–94%) (11, 12). However, in other studies, no significant differences in the serum antibody titer or HAI titer were observed between patients and controls (13, 14). This issue has remained controversial in more recent studies, as several groups have shown significantly lower HAI titers in SLE patients compared to controls (15, 16), while others have shown that patients have HAI titers that are equivalent to those in controls (17, 18).

Previous findings are also contradictory regarding the impact of vaccination on autoantibody production and clinical disease (9, 10, 15, 16, 18–20). Several groups have shown that vaccination is associated with increased autoantibody levels in SLE patients (8, 13, 19) and healthy individuals (20). Application of these results to patients in general clinical practice has been limited, due to the small number of unique individuals studied, the limited number of racial or ethnic groups evaluated, and the selection of lupus patients with low disease activity or quiescent disease. Thus, it remains unclear whether patients with more active disease would be capable of mounting an effective immune response to influenza following vaccination.

Our objective was to evaluate the association between demographic, therapeutic, disease activity, and clinical features and the responsiveness to the influenza vaccine in SLE patients with various racial backgrounds and a broad range of disease activity. A secondary objective was to monitor autoantibody production and disease activity following vaccination, to determine whether vaccination resulted in increased humoral autoimmunity or disease flares. We hypothesized that select disease activity criteria would be correlated with reduced responsiveness to the vaccine, and that, in some patients, vaccination would result in increased autoantibody production.

PATIENTS AND METHODS

Study population.

Seventy-two unique patients who met ≥4 of the American College of Rheumatology (ACR) classification criteria for SLE (21) were recruited from local rheumatology clinics. The patients provided their informed consent to participate in the study, and information on the patients' demographics (sex, age, and race) was obtained. Seventy-two healthy control subjects, matched to the patients by sex, age, and race, were also recruited, via referrals from patients' friends and local advertising. Exclusion criteria included severe anemia (hemoglobin level <8.5 gm/dl), egg allergies, and pregnancy. The patients' clinical information was extracted using the Lupus Family Registry and Repository (22) collection tool, to obtain information on ACR classification criteria, age at diagnosis, and medication usage.

After enrollment of all patients and controls, peripheral blood samples were collected before vaccination and at 2, 6, and 12 weeks postvaccination. Institutional Review Board approval for the study was obtained from the Oklahoma Medical Research Foundation and Oklahoma University Health Sciences Center.

Evaluation of disease activity.

Disease activity was measured using the SLE Disease Activity Index (SLEDAI) and physician's global assessment of disease activity (23) before vaccination and at 6 and 12 weeks postvaccination. Data on medications and hospitalizations were also gathered. With these data, we computed a composite score for flare, using the Safety of Estrogens in Lupus Erythematosus: National Assessment (SELENA) version of the SLEDAI (24, 25); this index was used to classify flares as mild/moderate or severe.

Trivalent influenza vaccine.

All subjects received the subunit influenza vaccine approved for use in the United States. The 2005–2006 vaccine consisted of an A/New Caledonia/20/99 (H1N1)–like, an A/California/7/2004 (H3N2)–like, and a B/Shanghia/361/2002–like virus (Aventis Pasteur). The 2006–2007 vaccine consisted of an A/New Caledonia/20/99 (H1N1)–like, an A/Wisconsin/67/2005 (H3N2)–like, and a B/Malaysia/2506/2004–like virus (Chiron Vaccines Limited). The 2007–2008 vaccine consisted of an A/Solomon Islands/3/2006 (H1N1)–like, an A/Wisconsin/67/2005 (H3N2)–like, and a B/Malaysia/2506/2004–like virus (Novartis).

Characterization of humoral influenza responses.

To quantify antibodies to native glycoproteins, a sandwich enzyme-linked immunosorbent assay (ELISA) was used (26), and the data were subjected to a nonlinear regression model to calculate the Bmax, indicating the relative measure of total antibody concentration, and the Ka, indicating the apparent overall association constant (relative affinity) of the serum antibodies (26, 27). Assays to detect inhibition of hemagglutinin were performed using human red blood cells (27, 28). The measurements from each of these 3 parameters of the influenza vaccination response were ranked (using data from all patients and controls in a given year), and were then scaled to fall between 0 and 100 for each study year. The sum of these 3 ranked and rescaled measurements (the sum of ranks) has a range of 0–300, with an average of ∼150; this is referred to as the index of antinative antibodies (INA). Therefore, patients were classified into groups of high or low responders by identifying a noticeable gap in the sum of ranks for each study year, resulting in equal groups of high responders and low responders (n = 36 patients per group).

Detection of antinuclear antibodies (ANAs) and double-stranded DNA (dsDNA).

ANAs were detected using NOVA Lite HEp-2 assay (positive titer ≥1:120; Inova Diagnostics) (29, 30). The NOVA Lite Crithidia luciliae assay was used to measure anti-dsDNA antibodies (positive titer ≥1:30) (29, 31).

Detection of autoantibodies.

Ro, La, Sm, and nuclear RNP (nRNP) antibodies (Immunovision) were tested by standard ELISA (29, 31). A sample was considered positive if the optical density (OD) was greater than the mean OD plus 2 times the standard deviation determined in samples from 36 healthy individuals. Antibodies against the ribosomal P antigen (anti-P), the 22–amino acid carboxyl-terminal peptide (R5HUP0) (32), were detected using a standard ELISA. To quantify IgG antibodies to the cardiolipin antigen, Linbro/Titertek Enzyme Immunoassay Microtitration plates (ICN BioMedicals) were used in a modified ELISA (33, 34).

Measurement of serum interferon-α (IFNα) activity.

Cells (WISH cells; American Type Culture Collection) were cultured with 50% patient sera and lysed (35, 36). Messenger RNA (mRNA) was purified from the cultures, and complementary DNA was made from the total cellular mRNA and quantified using real-time polymerase chain reaction (37). Forward and reverse primers for 3 genes known to be highly and specifically induced by IFNα (MX1, PKR, and IFIT) were used, and gene expression levels were normalized to the values for GAPDH. The results in experimental samples were compared to the mean and SD values in samples from healthy controls (n = 141). IFNα activity is expressed as the number of standard deviations above the mean value in healthy donors.

Statistical analysis.

Exact (permutation) chi-square tests were used to determine the association between vaccine response and race, age, and type of ACR criteria. Comparisons of the number of ACR criteria, baseline disease activity scores, and prevaccination-to-postvaccination changes in antibody responses between SLE patients and controls were evaluated using Wilcoxon's 2-sample tests. Differences in the number of responders taking each medication were evaluated using Fisher's exact test. Differences between responders with regard to changes in autoantibodies were assessed using conditional logistic regression. Student's 1-tailed t-tests for paired data were used to evaluate the group differences in IFNα serum activity. P values less than or equal to 0.05 were considered significant.

RESULTS

Reduced humoral immune responses following influenza vaccination in SLE patients.

African American patients comprised 44% of the SLE cohort, and >90% of participants were female (Table 1). We first assessed the log10-transformed postvaccination-to-prevaccination ratios of the vaccine response measurements. Using the INA approach, we classified patients as either high responders or low responders, based on their overall anti-influenza response. As shown in Figure 1A, although healthy control subjects showed a greater increase in the total amount of native antibody (Bmax) after vaccination compared to all SLE patients (P = 0.035), the change in Bmax from pre- to postvaccination in the control group was not significantly greater than that in each patient vaccine response group (high responder patients P = 0.254 and low responder patients P = 0.081 versus controls). Both the high responder patient group and the low responder patient group had a significantly smaller increase in the apparent affinity (Ka) after vaccination when compared to controls (P = 0.008 in high responder patients and P < 0.001 in low responder patients versus controls) (Figure 1B).

Table 1. Demographic features of the cohort of patients with systemic lupus erythematosus who received influenza vaccination*
 Total (n = 72)Low responders (n = 36)High responders (n = 36)
  • *

    ACR = American College of Rheumatology; SLEDAI = Systemic Lupus Erythematosus Disease Activity Index.

  • Race was self-defined by the patients. The “Hispanic and other” designation includes one individual who self-defined race in more than one category.

  • P = 0.028 versus low responders in the African American racial group, by Fisher's exact chi-square test.

  • §

    Baseline refers to the day of vaccination.

  • P = 0.053 versus low responders, by Wilcoxon's 2-sample test.

Age, mean ± SD years43 ± 1443 ± 1343 ± 15
Race, %   
 African American443158
 European American496136
 Hispanic and other786
Sex, % female929786
Age at diagnosis, median ± SD years31 ± 1429 ± 1533 ± 14
Number of ACR criteria, median ± SD5.8 ± 1.66.2 ± 1.45.5 ± 1.7
Baseline SLEDAI, median ± SD score§6.0 ± 3.86.0 ± 4.15.0 ± 3.4
Treatment, %   
 Steroids585364
 Antimalarials696969
 Combination steroids and antimalarials515053
Figure 1.

Reduced humoral immunity in patients with systemic lupus erythematosus (SLE) in whom poor antibody responses occurred following influenza vaccination. SLE patients (cases) were divided into high and low responders to the vaccine based on cumulative ranking, as described in Patients and Methods, and plasma from healthy subjects was used as a control. Humoral immune responses to the influenza vaccine are shown as the log10-transformed ratios of postvaccination-to-prevaccination measurements for antibody concentration (Bmax) (A), antibody affinity (Ka) (B), and hemagglutination inhibition (HAI) (C), as well as the overall vaccine response, calculated as the postvaccination–prevaccination difference in percentile ranks (D). Results are shown as box plots, where the boxes represent the 25th to 75th percentiles, the lines inside the boxes represent the median, and the whiskers outside the boxes represent the minimum and maximum values. P values were determined using nonparametric Mann-Whitney tests, with Bonferroni correction for multiple comparisons.

Although the difference in the HAI titer between SLE patients and controls was nonsignificant (P = 0.17), it should be noted that few of these individuals (patients or controls) had substantial increases in HAI titers after vaccination. Furthermore, low responder patients showed a significantly smaller increase in HAI after vaccination than did either the high responder patients or healthy controls (each P < 0.001 versus low responders) (Figure 1C).

We also utilized the INA approach to assess overall differences in humoral influenza responses from pre- to postvaccination. Using this approach, we observed that controls showed significantly improved humoral influenza responses after vaccination than did patients in either vaccine response group (each P < 0.001 versus controls), and this effect was driven by the poor overall antibody responses in the low responder patients (Figure 1D).

Thus, with patients classified as either high responders or low responders to influenza vaccination, we observed that the high responder group of patients generated an anti-influenza response that was much closer to that seen in the healthy controls than to that seen in patients considered to be low responders. For each measure of responsiveness (the Bmax, Ka, HAI titer, and overall response), the low responder group had impaired responses compared to both the high responder group and the control group, particularly when evaluating differences in the HAI titer and differences in the overall humoral immune response. In the following experiments, we used this classification of high and low responders to further examine the correlates of poor vaccine response in SLE.

Increased likelihood of strong responses to influenza vaccination in African American patients with SLE.

African American patients were more likely to have strong responses to influenza vaccination than were European American patients (P = 0.03) (Table 1). Specifically, African American patients were 3 times more likely (95% confidence interval 1.07–9.94) to be high responders than were European Americans. This racial association with vaccination response was not attributable to age, since the average age of the patients in each group did not differ significantly (mean ± SD 42 ± 12.4 years in African Americans versus 46 ± 15.8 years in European Americans). This discrepancy in vaccine responsiveness may be due to the impact of HLA haplotypes in the different racial groups, although this requires further investigation.

Patients ranged in age from 20 years to 89 years, with 42% between 29 years and 45 years. The relationship between age and the response to vaccination was examined, since older individuals may demonstrate decreased influenza antibody levels (38). Although the number of elderly individuals was small, we found no relationship between age and the vaccine response. There is a known predilection for SLE development in women; however, the small number of male patients did not allow evaluation of potential differences attributable to sex.

Association of poor influenza vaccination responses with hematologic manifestations and more ACR classification criteria for SLE.

We next analyzed age at diagnosis, number of cumulative ACR criteria, and select ACR classification criteria to determine whether these factors were correlated with the subsequent vaccine response (Table 1). Low responders had more ACR criteria when compared to high responders (median of 6 ACR criteria in low responders versus 5 in high responders; P = 0.05). Indeed, 23 (64%) of the 36 low responders met ≥6 of the ACR criteria for SLE, compared to 15 (42%) of 36 high responders.

We then analyzed the types of ACR criteria found in both groups. Twenty-five (69.4%) of 36 low responders exhibited at least one hematologic criterion, compared to 12 (33.3%) of 36 high responders (P = 0.009) (Figure 2A). All of the patients with hemolytic anemia were low responders (P = 0.01) (Figure 2B). The low responders also had a higher prevalence of thrombocytopenia (22.2% versus 13.9%), lymphopenia (38.9% versus 27.8%), and leukopenia (36.1% versus 22.2%), although these differences were not statistically significant (Figure 2B). Moreover, 9 (25%) of the 36 high responders exhibited discoid rash, compared to 3 (12.5%) of 36 low responders, a difference that was not statistically significant (P = 0.1). No differences were seen in the prevalence of renal manifestations or arthritis.

Figure 2.

Association of poor influenza vaccination responses with more hematologic manifestations and more steroid use in patients with systemic lupus erythematosus (SLE). Patients were divided into high responders and low responders to influenza vaccination. Shown are the number of individuals who, prior to vaccination, met the indicated American College of Rheumatology classification criteria for SLE (A), exhibited specific aspects of hematologic disorders (B), were taking specific medications (C), and exhibited specific autoantibodies (D). P values for comparisons between high and low responders were as follows: in A, ∗∗ = P = 0.009 by exact (permutation) chi-square test; in B, ∗∗ = P = 0.01 by exact (permutation) chi-square test; in C, ∗ = P = 0.037 by Fisher's exact test. CNS = central nervous system; MTX = methotrexate; MMF = mycophenolate mofetil; ANA = antinuclear antibodies; dsDNA = anti–double-stranded DNA; nRNP = anti–nuclear RNP; RiboP = anti–ribosomal P; aCL = anticardiolipin.

Increased frequency of poor influenza humoral immune responses in SLE patients receiving steroid treatment.

While some investigators have reported that medication has no effect on anti-influenza antibodies (11), others have shown decreases in antigen-specific responses in patients receiving azathioprine or corticosteroids at a level equivalent to ≥10 mg prednisone daily (19, 39). Twenty-four (67%) of 36 patients who were low responders had been receiving prednisone at a level equivalent to ≥10 mg/day, compared to 17 (47%) of 36 patients who were high responders (P = 0.04) (Figure 2C); higher dosages of corticosteroids (≥20 mg daily) did not appear to further lessen the magnitude of the influenza immune response, although the sample sizes were small. No other immunomodulatory medication was associated with a low response.

The number of individuals receiving these medications was relatively small. Therefore, we divided patients into a low immunosuppressive therapies group (those receiving no medications, hydroxychloroquine only, or hydroxychloroquine in combination with the equivalent of 7.5 mg of prednisone daily) and a high immunosuppressive therapies group (those receiving methotrexate, mycophenolate mofetil, azathioprine, cyclophosphamide, >7.5 mg of prednisone daily, or a combination of these). Of the 39 individuals in the high immunosuppressive therapies group, 21 (54%) were high responders to the vaccine. Of the 33 individuals in the low immunosuppressive therapies group, 18 (55%) were high responders to the vaccine. Thus, whereas steroid use is associated with a lower antibody response to influenza vaccination, other immunosuppressive medications do not account for the decreased responses in some patients.

No differences in autoantibody specificities at baseline in patients with poor vaccination responses.

To help determine what factors could be predictive of a low response to vaccination in SLE patients, autoantibody levels were measured at the time of enrollment. Nearly all patients were ANA positive (Figure 2D), and high and low responders did not differ significantly with regard to ANA titer (median titer 1:1,080 versus 1:953) (results not shown). Furthermore, no significant differences in the dsDNA antibody titer were observed between the responder groups (median 1:90 for each) (results not shown).

There was a trend toward higher frequencies of cardiolipin antibodies and lower frequencies of nRNP antibodies in the low responders, but these differences were not significant, which may be related to the low number of individuals positive for these antibodies (Figure 2D). No differences were seen between the high and low responders in the proportion of individuals positive for any of the other autoantibody specificities at baseline. There was also no difference in the average number of autoantibodies in each patient vaccine response group (mean 2.6 in high responders versus 2.4 in low responders).

Increases in autoantibody levels and new specificities after vaccination in low responders.

We next examined the impact of vaccination on autoantibody specificities and levels. For each antibody assessed, we determined the number of individuals who had developed the antibody for the first time (new onset) or had a ≥2-fold increase or ≥2-fold decrease in the levels of the specified antibody after vaccination. As shown in Figure 3, the results showed patterns of changes in certain autoantibodies, including ANAs, anti-La, and anticardiolipin among the high responders, and ANAs, anti-Ro, and anti-P among the low responders.

Figure 3.

Postvaccination increases in autoantibody levels and new autoantibody specificities in patients with low responses to influenza vaccination. Autoantibody titers were measured in patients' sera at the initial (prevaccination) and postvaccination time points. Shown is the number of low responders (A) or high responders (B) with a 2-fold increase in levels, a 2-fold decrease in levels, or new onset of the specified autoantibody. ∗ = P = 0.045 versus antinuclear antibodies (ANA) in high responders, by conditional logistic regression. aCL = anticardiolipin; RiboP = anti–ribosomal P; nRNP = anti–nuclear RNP; dsDNA = anti–double-stranded DNA.

Nineteen (26%) of the patients experienced a change in their ANA titer at 2 weeks postvaccination, characterized as new onset in 2 patients, an increased titer in 8 patients, and a decreased titer in 9 patients. Low responders were more likely to have an increased ANA titer than were high responders (14% versus 8%, respectively), with 5 of the 36 low responders having an increase in ANA titer postvaccination. Eight (22%) of the high responders had decreased ANA titers, compared to only 1 (3%) of the low responders. Indeed, among high responders, postvaccination decreases in the ANA titers were twice as common as increased or new ANA expression. Among low responders, decreases in ANA titers were one-sixth as common as increased or new ANA expression (P = 0.05) (Figure 3).

The small number of patients exhibiting changes in antibodies to La, Ro, cardiolipin, and ribosomal P did not allow a definitive statistical evaluation of differences between high and low responders. However, both groups had new onset of autoantibodies directed against La and cardiolipin (Figure 3). Three low responders and 1 high responder had new onset/increased levels of antibodies to the La antigen. The high responder group had 2 patients and the low responder group had 3 patients in whom new onset/increased titers of antibodies to cardiolipin were observed.

Increased likelihood of disease flare in low responders to influenza vaccination.

We next examined disease activity following vaccination. Five individuals (6.9%), of whom 4 were low responders, had an increase in the SLEDAI of ≥3 points at 6 weeks postvaccination. Eight individuals (11%), comprising 4 high responders and 4 low responders, had a 3-point change in the SLEDAI at 12 weeks postvaccination. Overall, neither an increased frequency of flares nor an increased severity of the flares was seen in the time period immediately after vaccination (6 weeks) compared to a time period further from vaccination (12 weeks). However, this method does not capture some indications of disease flare, such as medication changes or hospitalizations.

To further refine our method of determining flares, the SELENA–SLEDAI flare index was used (24, 25). Using this index, we determined that 14 patients (19.4%) had experienced a flare by 6 weeks postvaccination, including 10 patients (13.9%) who were classified as having mild/moderate flares and 4 (5.6%) as having severe flares (Figure 4A). At 12 weeks postvaccination, there were 19 patients with flares (26.4%), including 16 (22.2%) with mild/moderate flares and 3 (4.2%) with severe flares (Figure 4A).

Figure 4.

Increased frequency of disease flares at 6 weeks, but not 12 weeks, after influenza vaccination in patients with systemic lupus erythematosus (SLE) who were classified as low responders to the vaccine. High responders and low responders were evaluated for the presence of flares using the Safety of Estrogens in Lupus Erythematosus: National Assessment version of the SLE Disease Activity Index (24, 25) for all individuals at 6 and 12 weeks postvaccination. Shown is the total number of individuals experiencing no flare, mild/moderate flare, or severe flare at 6 and 12 weeks (A) as well as the vaccine responsiveness (high or low) of individuals with a severe or mild/moderate flare at 6 weeks (B) or 12 weeks (C).

Of the patients who experienced a flare within 6 weeks of vaccination, 10 (71.4%) were low responders and 4 (28.6%) were high responders (P = 0.01) (Figure 4B). By 12 weeks after vaccination, no significant difference was observed in the flare rate between high responders and low responders (53% and 47% of patients, respectively) (Figure 4C). Using data collected from previous studies in which patients did not receive influenza vaccination, we observed that 1 of 8 patients had experienced a mild/moderate flare (12.5% of patients, compared with 19.4% of patients in our study) when followed up over a 6–8-week timeframe. Of 41 patients observed over a 9–18-week timeframe without vaccination, 8 had exhibited mild/moderate flares (19.5% of patients, compared with 26.4% of patients in our study). The differences in flare rates over similar time periods between vaccinated patients with SLE and nonvaccinated patients with SLE were not statistically significant.

Association of serum IFNα activity with disease flare following vaccination.

We next determined whether there were characteristics that could be a predictor of which patients would develop a disease flare following vaccination. To this end, we compared the group of patients in whom a flare occurred by 6 weeks postvaccination (n = 14) to a group of matched patients who had not experienced a flare within this time period. We found that age at diagnosis, number of ACR criteria for SLE, and the level of disease activity at baseline were not correlated with the occurrence of a flare postvaccination. Select ACR criteria for SLE were more prevalent in the individuals experiencing a flare compared to the matched control group, and these included renal disease (43% versus 29%), central nervous system involvement (21% versus 0), and hematologic disorder (50% versus 29%). In contrast, fewer individuals with serositis (29% versus 50%) or oral ulcers (57% versus 71%) were in the flare group compared to the group who had not experienced a flare. Unfortunately, the small numbers of patients in these groups did not allow a definitive statistical evaluation of the impact of these criteria on flare.

Since increased IFNα-induced gene expression is correlated with higher disease activity in SLE patients (37), we examined serum IFNα activity. We found that patients experiencing a flare 6 weeks postvaccination had higher serum IFNα activity at baseline when compared with patients who did not have a flare (mean IFNα activity 19.3 versus 2.7; P = 0.04) (Figure 5A). Although patients who experienced a flare had a decrease in IFNα activity at 2 weeks postvaccination (mean IFNα activity 10.95), this activity was still higher than that seen in individuals who did not experience a flare (Figure 5A).

Figure 5.

Higher interferon-α (IFNα) serum activity at baseline in patients in whom a flare occurred within 6 weeks postvaccination. Patients were divided into those in whom a flare occurred and age- and race-matched patients who did not have a flare during the course of the study. Shown is the serum IFNα activity at baseline and at 2 weeks postvaccination in the patients who had developed a flare by 6 weeks (A) and in those who had developed a flare by 12 weeks (B). Each symbol represents a serum sample from 1 patient. Bars show the mean ± SEM. P values were determined by unpaired t-test.

This significant difference in serum IFNα activity was not observed in the individuals who had developed a flare by 12 weeks postvaccination (mean IFNα activity 2.9 versus 5.3 in the group without a flare; P = 0.2) (Figure 5B). Of note, we found no significant correlation between IFNα activity level and the overall magnitude of the influenza-specific response.

DISCUSSION

The goal of this study was to identify factors that are correlated with the immune response to influenza vaccination in SLE patients and examine the impact of vaccination on autoimmune disease activity. We found no differences in the immune response based on age or sex. However, we did find that the group of high responders to the vaccine was enriched with African American patients. Previous studies on the effects of influenza vaccination in SLE patients were performed in other geographic regions, among Israeli (8, 17, 19, 40, 41), Italian (18), Mexican (1616), or Dutch (15, 39, 42, 43) subjects, or were not able to evaluate race (11–14). Surprisingly, little is known about racial differences in immune responses to vaccinations. Healthy young African Americans had higher levels of neutralizing antibodies against human immunodeficiency virus gp120 after vaccination compared to European Americans (44), yet older African Americans had impaired T cell responses to influenza vaccination (45). Additional work is necessary to fully understand the influence of race on vaccination responses.

SLE patients with a history of hemolytic anemia were more likely to be low responders. Our initial assumption was that this association would be driven by the number of patients with significant lymphopenia and/or leukopenia. However, no significant differences were found in the percentage of lymphocytes at baseline between the low responders and high responders (28.1% versus 34.8%, respectively). In addition, severe lymphopenia (≤500 cells/μl) was found in 14% of low responders compared to 11% of high responders, indicating that lymphopenia was not the main factor in this association. Rather, hemolytic anemia was the dominant hematologic manifestation driving the association with low vaccination response. No other studies examining the response to vaccination in patients with hemolytic anemia or thrombotic thrombocytopenic purpura could be found. Examinations of vaccination responses in larger collections of patients with hemolytic anemia are warranted to uncover mechanisms for this epidemiologic association.

Findings from studies of changes in autoantibody concentration and/or autoantibody specificities in SLE patients after vaccination have been conflicting (17–20). These studies have been limited by small sample sizes (n = 14–18) (16, 18) as well as differences in methodologies and the limited number of autoantibodies selected for analysis (16). Abu-Shakra et al found that influenza vaccination had no effect on anti-dsDNA antibodies, had a transient effect on Sm, nRNP, Ro, and La antibodies, and had a prolonged effect on anticardiolipin antibodies (19). These results are supported, in part, by our findings.

We observed 2 striking findings with regard to the pattern of autoantibodies following vaccination. First, patients in whom a strong response to vaccination was mounted were significantly more likely to have decreased ANA titers. This suggests that some SLE patients have an immune system poised to generate antigen-specific immune responses, accompanied by minimization of the stimulation, expansion, and maturation of autoreactive B cells. Second, more than one-half of the low responders developed new autoantibodies or experienced an increase in the concentration of existing autoantibodies, suggesting that these patients are more likely to have immune responses focused, at least in part, against self antigens. Further investigation into the longevity of these responses and associated long-term clinical consequences are under way. Additional evaluation of the influence of genetic, environmental, and other disease-specific factors on immune responses in SLE patients, particularly on whether the response is focused on antigen-specific or autoantigen-specific pathways, is warranted.

Several groups have reported no differences in disease activity measurements between SLE patients who have received the influenza vaccine and unvaccinated patients (41, 46). However, the sample size in the majority of these studies was small (n = 23–24) (41, 46). Holvast et al (15) found that influenza vaccination did not result in significant changes to disease activity scores in 56 patients with SLE. We showed no increased rates of flare within 6 weeks postvaccination when compared to that in the 6–12-week postvaccination time interval. However, patients in whom a flare occurred within 6 weeks postvaccination were predominantly low responders. Being able to identify these individuals before vaccination and modify approaches for influenza vaccination in ways that do not negatively impact their underlying disease would be clinically useful.

Several differences between other published studies and our present study should be noted in the context of the flare rates and disease activity changes detected. First, in previous studies, a flare was defined as an increase of ≥3 points as measured by the SLEDAI. Our study calculated a composite flare score using the SELENA–SLEDAI index, which also takes into account hospitalization and medication changes (24, 25). Second, the studies cited enrolled patients with quiescent disease (SLEDAI score ≤4), whereas our patient population was composed of individuals with either quiescent disease or active disease. Indeed, 41 of our 72 patients (57%) had a SLEDAI score of ≥6 at baseline. Last, as mentioned earlier, our study was composed of roughly equal numbers of European American and African American patients with SLE, whereas the majority of the previously reported studies were composed primarily of patients of European descent.

When we examined patients in whom a disease flare developed following vaccination, we found that select ACR criteria were either overrepresented or underrepresented. Renal disease, central nervous system involvement, and hematologic disorder were more prevalent at baseline in the patients who had a flare. Additionally, patients with a flare by 6 weeks postvaccination had higher serum IFNα activity at baseline. Other investigators have shown that IFNα activity is correlated with SLE disease criteria, and that an elevated IFNα activity score is associated with severe manifestations of SLE (47, 48). The relationships between IFN activity, vaccination responses, and subsequent disease flares need to be confirmed in a larger cohort, and the biologic significance of these processes should be examined. This information is especially important to investigate when considering that several groups have advocated the use of IFNα as an adjuvant for influenza vaccination (48, 49).

Our findings indicate that some SLE patients, especially those with a history of hematologic disorder or individuals taking prednisone, experience a weak response to the influenza vaccine. Furthermore, low responders are more likely than high responders to have increased levels of autoantibody production and to experience disease flares following vaccination. Studies are currently under way to explore the biomarkers that might better serve as predictors of which patients are likely to experience a disease flare following vaccination. Coupled with the knowledge of select serologic biomarkers in patients who are anticipated to have low responses, this information would help identify the subset of patients who may need to have their underlying disease more aggressively treated before receiving the influenza vaccine or other types of vaccination.

AUTHOR CONTRIBUTIONS

All authors were involved in drafting the article or revising itcritically for important intellectual content, and all authors approved the final version to be published. Dr. James had full access to all of the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis.

Study conception and design. Crowe, Dedeke, L. F. Thompson, James.

Acquisition of data. Crowe, Merrill, Vista, Dedeke, Niewold, Franek, James.

Analysis and interpretation of data. Crowe, Vista, D. M. Thompson, Stewart, Guthridge, Niewold, Air, L. F. Thompson, James.

Acknowledgements

We would like to thank all of the study participants for their time and commitment to this work, as well as the referring physicians (Drs. Craig Carson, John Harley, Ana Ahluwahlia Kumar, and Linda Zacharias and physician assistants Teresa Aberle and Jama Kendall Shoemaker). We also thank our clinical coordinator, Virginia Roberts, as well as Jourdan Anderson, Wendy Klein, Susan Macwana, Wade DeJager, Phillip McGhee, Gabriel Vidal, Jeremy Levin, and John Johnson for technical assistance. Finally, we thank Dr. J. Donald Capra for critically reading the manuscript.

Ancillary