Systemic lupus erythematosus in adults is associated with previous Epstein-Barr virus exposure

Systemic lupus erythematosus in adults is associated with previous Epstein-Barr virus exposure
Judith A. James 1 *, Barbara R. Neas 1, Kathy L. Moser 2, Teresa Hall 3, Gail R. Bruner 3, Andrea L. Sestak 3, John B. Harley 4
1University of Oklahoma Health Sciences Center, Oklahoma City
2Case Western Reserve University, Cleveland, Ohio
3Oklahoma Medical Research Foundation, Oklahoma City
4University of Oklahoma Health Sciences Center and US Department of Veterans Affairs, Oklahoma City, Oklahoma


*Correspondence to Judith A. James, Arthritis and Immunology Program, Oklahoma Medical Research Foundation, 825 NE 13th Street, Oklahoma City, OK 73104

Funded by:
NIH; Grant Number: AR-42474, AR-01981, AI-31584, AR-45084, AI-24717, AR-42460, AR-45231, N01-AR-52221
US Department of Veterans Affairs

Abstract


Objective
The possible molecular mimicry of the Epstein-Barr virus (EBV) peptide PPPGRRP by the peptide PPPGMRPP from Sm B/B of the human spliceosome is consistent with the possibility that EBV infection is related to the origin of systemic lupus erythematosus (SLE) in some patients. Association of EBV exposure with SLE was therefore tested for and subsequently found in children and adolescents (odds ratio [OR] 49.9, 95% confidence interval [95% CI] 9.3-1,025, P < 10-11). These results were confirmed at the level of EBV DNA (OR > 10, 95% CI 2.53-, P < 0.002). Much smaller seroconversion rate differences were found against 4 other herpes viruses. Herein, we extend these studies to adults and test the hypothesis that EBV infection is associated with adult SLE.

Methods
We selected 196 antinuclear antibody-positive adult SLE patients (age 20 years) and 2 age-, race-, and sex-matched controls per patient. SLE patients and matched controls were tested for evidence of previous infection with EBV, cytomegalovirus (CMV), herpes simplex virus types 1 and 2 (HSV-1 and HSV-2), or varicella-zoster virus (VZV) by standardized enzyme-linked immunosorbent assays.

Results
Of the 196 lupus patients tested, all but 1 had been exposed to EBV, while 22 of the 392 controls did not have antibodies consistent with previous EBV exposure (OR 9.35, 95% CI 1.45-, P = 0.014). No differences were observed between SLE patients and controls in the seroconversion rate against CMV, HSV-2, or VZV.

Conclusion
These new data from adults, along with the many suggestive features of EBV infection, are consistent with the contribution of this infection to the etiology of SLE.



--------------------------------------------------------------------------------
Received: 15 February 2000; Accepted: 27 September 2000
Digital Object Identifier (DOI)

10.1002/1529-0131(200105)44:5<1122::AID-ANR193>3.0.CO;2-D About DOI


Article Text



Systemic lupus erythematosus (SLE) is an idiopathic, autoimmune disorder in which autoantibodies are universally present. Unfortunately, the mechanisms leading to the production and perpetuation of these aberrant autoimmune responses remain poorly understood. Autoantibodies directed against the spliceosomal proteins, anti-Sm and anti-nuclear RNP, are found in 30-50% of lupus patients' sera ([1]). The Sm B/B protein is a major target of this autoimmune response. Epitope mapping in our laboratory has identified the specific regions of this protein that are commonly targeted by lupus patient sera ([2]). Among the B cell epitopes defined by overlapping octapeptides ([2]), PPPGMRPP consistently appears earliest in the anti-Sm autoimmune response of human lupus ([3]).

Immunization with PPPGMRPP induces lupus autoimmunity ([4-6]). A structurally similar and antigenically crossreacting peptide from Epstein-Barr virus (EBV) nuclear antigen 1 (EBNA-1) also induces lupus humoral autoimmunity ([7]). These results, as well as conflicting reports in the literature, led us to test the hypothesis that EBV is associated with SLE. Seroconversion against EBV failed to occur in only 1 of 117 pediatric lupus patients compared with 46 of 153 controls (odds ratio [OR] 49.9, 95% confidence interval [95% CI] 9.3-1,025, P < 10-11) ([8]). Similar results have been obtained confirming the association of serologic evidence of EBV infection with SLE in children and adolescents ([9]). In both studies ([8][9]), much smaller seroconversion rate differences were found against 4 other herpes viruses: cytomegalovirus (CMV), herpes simplex virus types 1 and 2 (HSV-1 and HSV-2), and varicella-zoster virus (VZV). EBV infection was identified by polymerase chain reaction for EBV DNA in 32 of 32 lupus patients (100%) and in 23 of 32 matched controls (72%). In all 9 of the discordant sets, EBV DNA was found in the lupus patient, but not in the matched control (OR >10, 95% CI 2.53-, P < 0.002) ([8]). These data, along with the many suggestive features of this viral infection, are consistent with the immune response against EBV infection being a potential triggering mechanism for SLE.

Initially, pediatric and adolescent lupus patients were selected because of the lower frequency of previous EBV exposure in these normal populations ([10]). Although pediatric and adult SLE are very similar and even share the same diagnostic criteria ([11][12]), the etiologies of SLE are not necessarily the same in younger and older SLE patients. To determine whether the association of SLE with EBV infection observed in children and adolescents ([8]) could be extended to adults, we selected a group of 196 adult lupus patients and 392 normal controls and evaluated their exposure to EBV.

PATIENTS AND METHODS


Patients and controls

We studied antinuclear antibody-positive SLE patients who satisfied the classification criteria of the American College of Rheumatology ([11][12]). All patients were 20 years of age at the time of specimen collection. Patients were recruited from pedigrees of genetic studies being conducted in our laboratory. Virtually all of the SLE patients were members of multiplex lupus families. Controls were matched by age (±10 years), race (self declared), and sex. We selected 196 adult lupus patients and 392 adult controls from 169 pedigrees in our lupus genetics study. Of the 169 pedigrees, 108 included an SLE patient. We also organized age into 10-year intervals. Serum samples had been stored at -20°C since collection.

Anti-Epstein Barr virus viral capsid antigen (anti-EBV-VCA) antibodies

Evidence of previous EBV infection was detected by measuring IgG antibodies against purified EBV-VCA (Wampole, Cranbury, NJ), following the manufacturer's instructions. The data analysis used the international standardized ratio (ISR), a semiquantitative measure of the relative level of antibody that is standardized between assays by known, characterized, positive controls. The ISR is designed to optimally detect seroconversion.

Occasionally, serum specimens produced an equivocal result (ISR 0.9-1.10), and these samples were repeat-tested. To avoid falsely identifying any serum sample as showing no evidence of seroconversion, we uniformly reported as positive any serum sample showing 2 equivocal readings as positive.

Antiviral assays

Serum specimens were also tested for reactivity with CMV, HSV-1, HSV-2, and VZV. Again, these specimens were tested and ISRs calculated by a standardized enzyme-linked immunosorbent assay (Wampole), following the manufacturer's instructions.

Statistical analysis

The statistical analysis adjusted for the correlated nature of the family data and accounted for the potential confounder of age. The relationship between seroconversion and patient status was assessed using conditional logistic regression with pedigree as the matching factor. The models were adjusted for age by including the variable representing 10-year age intervals. Due to the small number of individuals who were not seropositive for some of the viruses studied, all analyses were performed using LogXact 2.1 ([13]). The results presented are exact P values and 95% CIs. The ISR comparisons were analyzed using a mixed fixed-effects and random-effects model (PROC MIXED; SAS Institute, Cary, NC) to account for the pedigree effect. We also calculated least-squares (LS) means for all mean comparisons.

RESULTS


We studied 196 SLE patients and 392 unaffected controls from a total of 169 pedigrees. No controls reported having SLE, and we adjusted all analyses for age. The patients and controls were 66% European American, 30% African American, and 4% other; 94% were women. Demographics of the patients and controls are presented in Table 1. All patients and controls were recruited from families that were multiplex for lupus. (A separate study will be needed to test whether an association exists between EBV infection and the nongenetic or sporadic instances of SLE in simplex pedigrees. No study has yet shown, however, that SLE in multiplex pedigrees is fundamentally different from that in simplex lupus pedigrees.)


Table 1. Demographics of adult lupus patients and matched controls

--------------------------------------------------------------------------------

Lupus patients Normal controls

--------------------------------------------------------------------------------

Age, mean ± SD years 44.7 ± 12.4 45.9 ± 12.9
Age range, years 20-76 20-84
Women, % 94 94
No. of families represented 108 151
Race, %
Caucasian 65 66
African American 30 30
Other* 5 4


--------------------------------------------------------------------------------


* Participants identifying themselves as predominantly of Hispanic, American Indian, Asian, or Eastern Indian heritage.



Sera from each patient and control were tested for evidence of previous EBV infection. If there was sufficient binding of IgG directed against EBV-VCA, the subject was said to have seroconverted, although a conversion from negative to positive is not ordinarily observed. Of the 196 adult lupus patients evaluated, 195 (99.5%) had been previously exposed to EBV by this criterion, while a smaller proportion of controls (370 of 392, or 94.4%) showed evidence of seroconversion (OR 9.35, P = 0.014) (Table 2).


Table 2. Seroconversion against Epstein-Barr virus viral capsid antigen (EBV-VCA), cytomegalovirus (CMV), herpes simplex virus types 1 and 2 (HSV-1 and HSV-2), and varicella-zoster virus (VZV) in sera from adult lupus patients and controls*

--------------------------------------------------------------------------------

Lupus patients, no. positive/total tested (%) Normal controls, no. positive/ total tested (%) OR 95% CI P

--------------------------------------------------------------------------------

EBV-VCA 195/196 (99.5) 370/392 (94.4) 9.35 1.45- 0.014
CMV 130/196 (66.3) 270/392 (68.9) 0.97 0.56-1.70 1.0
HSV-1 161/196 (82.1) 311/392 (79.3) 1.32 1.02-1.86 0.03
HSV-2 123/196 (62.8) 217/392 (55.4) 1.11 0.68-1.84 0.74
VZV 193/195 (99.0) 385/392 (98.2) - - 0.56


--------------------------------------------------------------------------------


* Sera from lupus patients or their controls were tested for IgG antibodies against EBV-VCA, CMV, HSV-1, HSV-2, and VZV, and were standardized for seroconversion. Ninety-five percent confidence intervals (95% CI) and P values are exact values obtained from the LogXact 2.1 program ([13]). OR = odds ratio.
There were too few discordant values to provide valid estimates of the OR and 95% CI.



In addition, as shown in Figure 1, SLE patients had a higher titer of anti-EBV-VCA antibodies compared with unaffected controls (LS mean patient ISR 4.18, LS mean control ISR 2.59; t = 10.65, P < 4 × 10-19) when we accounted for the familial clustering. None of the remaining responses had so large a difference in the effect, although the LS means were significantly different for CMV (LS mean patient ISR 2.20, LS mean control ISR 1.72; P < 2 × 10-4) and for VZV (LS mean patient ISR 3.11, LS mean control ISR 2.72; P < 1 × 10-5). Neither HSV-1 nor HSV-2 showed a significant increase in the SLE patients. The ISR value difference for EBV was 3.3-fold that for CMV and 4.1-fold that for VZV.


Figure 1. International standardized ratios (ISRs) for reactivity of lupus patient and normal control sera with Epstein-Barr virus (EBV) viral capsid antigen, cytomegalovirus (CMV), herpes simplex virus types 1 and 2 (HSV-1 and HSV-2), and varicella-zoster virus (VZV). Mean ISRs for each antigen by patient and control group are presented as heavy horizontal lines, and confidence intervals are presented as thin vertical lines above and below these means.
[Normal View 39K | Magnified View 56K]


Of the antibodies to the herpes viruses measured in this study, those against EBV showed the strongest statistical and practical association with SLE when we accounted for familial clustering and controlled for age. Although HSV-1 produced a statistically significant result, the OR was only 1.32.

DISCUSSION


EBV has been known for more than 4 decades. Why has the association of EBV infection with SLE been so difficult to establish? First, the near ubiquity of EBV infection among adults means that if EBV contributes to the etiology of SLE, then SLE is an uncommon consequence of EBV infection and there must be another explanation for the sporadic occurrence of SLE. The likely reasons have to do with EBV itself (e.g., risk of infection varying with EBV strain), the circumstances of EBV exposure (i.e., route, dose, concurrent infection, age of host), the host immune response (e.g., stochastic processes), and/or the genetics of the host. Without pertinent data from testing these possibilities, however, we can only speculate about the reasons that we consistently observe an association of EBV exposure with SLE. In some ways, these considerations are not relevant to the validity of the observed association, although they may, of course, form the basis for exploring a possible role of EBV in the etiology of SLE.

Second, studies performed before the 1997 report of an association ([8]) all relied upon assays that were <90% sensitive for the presence of EBV infection. When the expected true seronegative rate in adults is 5%, even a false-negative rate of only 5% in detecting seroconversion against EBV will wreak havoc with the power of a study to detect an effect that is present. This means that studies that showed a possible association ([14-16]), as well as those that did not ([17-19]), contained too few SLE patients and controls to reliably and reproducibly detect a difference. In addition, only one of the previous studies used the strategy of evaluating children and adolescents ([19]), which would increase the ability to detect associations. However, this study was also underpowered and used an anti-EBV antibody assay and EBV DNA assay technologies that were even older and less reliable.

Four studies have previously evaluated the role of CMV in SLE ([8][9][19][20]). Three of them found no association of CMV seroconversion with SLE ([8][9][19]), although one study was performed in a small set of pediatric patients ([19]). The two other studies evaluated a combined total of 143 children and adolescents with SLE, along with a larger control group. A fourth study showed an association in a sample of 97 adult SLE patients compared with 97 controls ([20]). Controls were not matched by ethnic background, geographic location, or socioeconomic status. These may be important variables to control for and could certainly explain the differences these investigators observed. In addition, the prevalence of anti-CMV antibodies in the control subjects (43%) was lower than (and outside the range of) that in previous seroepidemiologic studies (54% to >90%). Therefore, variation in assays may also account for some of the difference and explain why those investigators found an association of SLE with serologic evidence of CMV infection ([20]), while we and others did not ([8][9][19]).

EBV is an attractive component of the environmental risk for SLE for other reasons. First, EBV infection is found worldwide and spares no known segment of the human population, which is similar to the distribution of SLE. Second, after infection, EBV remains viable and actively infects the host for life. Third, EBV infection is a continuous source of chronic immune stimulation. Indeed, the latent form of EBV actually infects B cells and is perpetuated in them. These same cells are critical for the production and propagation of autoantibodies, which are central to the laboratory diagnosis of SLE as well as to many of its clinical manifestations. Fourth, EBV periodically reactivates, which could hypothetically be associated with increased production of lupus autoantibodies and perhaps disease flares. Finally, EBNA-1 (the protein with the PPPGRRP and GRGRGRGR sequences that have significant homology with lupus autoantigens) is expressed in latently infected B cells. However, the ubiquitous nature of EBV suggests that if infection with this virus is necessary to the development of lupus, then there must be something about the usual human host's immune response that prevents lupus from becoming a much more widespread disease.

Acknowledgements


We would like to thank the Lupus Multiplex Registry and Repository (supported by National Institute of Arthritis and Musculoskeletal and Skin Diseases grant AR-52221) for the use of their well-characterized lupus patient and control materials. We would also like to thank Jama Kendall, Amber Davis, and Xana Kim for their technical assistance.

References

1 Harley JB, James JA. Autoepitopes in lupus. J Lab Clin Med 1995; 126: 509-16. Links
2 James JA, Harley JB. Linear epitope mapping of an Sm B/B polypeptide. J Immunol 1992; 148: 2074-9. Links
3 Arbuckle MA, Reichlin M, Harley JB, James JA. The development of lupus humoral autoimmunity for anti-Sm autoantibodies is consistent with predictable sequential B cell epitope spreading. Scand J Immunol 1999; 50: 447-55. Links
4 James JA, Gross T, Scofield RH, Harley JB. Immunoglobulin epitope spreading and autoimmune disease after peptide immunization: Sm B/B derived PPPGMRPP and PPPGIRGP induce spliceosome autoimmunity. J Exp Med 1995; 181: 453-61. Links
5 James JA, Harley JB. A model of peptide-induced lupus autoimmune B cell is strain specific and is not H-2 restricted in mice. J Immunol 1998; 160: 502-8. Links
6 James JA, Harley JB. B cell epitope spreading in autoimmunity. Immunol Rev 1998; 164: 184-200. Links
7 James JA, Scofield RH, Harley JB. Lupus humoral autoimmunity after short peptide immunization. Ann N Y Acad Sci 1997; 815: 124-7. Links
8 James JA, Kaufman KM, Farris AD, Taylor-Albert E, Lehman TJA, Harley JB. An increased prevalence of Epstein-Barr virus infection in young patients suggests a possible etiology for systemic lupus erythematosus. J Clin Invest 1997; 100: 3019-26. Links
9 Harley JB, James JA. Epstein-Barr virus infection may be an environmental factor for systemic lupus erythematosus in children and teenagers [letter]. Arthritis Rheum 1999; 42: 1782-3. Links
10 Evans AS, Niederman JC. Epstein-Barr virus. In: Evans AS , editor. Viral infections of humans: epidemiology and control. New York: Plenum; 1989. p. 270-6.
11 Tan EM, Cohen AS, Fries JF, Masi AT, McShane DJ, Rothfield NF, et al. The 1982 revised criteria for the classification of systemic lupus erythematosus. Arthritis Rheum 1982; 25: 1271-7. Links
12 Hochberg MC. Updating the American College of Rheumatology revised criteria for the classification of systemic lupus erythematosus [letter]. Arthritis Rheum 1997; 40: 1725. Links
13 Mehta C, Patel N. LogXact for Windows. Cambridge (MA): CYTEL Software Corporation; 1996.
14 Dalldorf G, Carvalho RPS, Jamra M, Frost P, Erlich D, Marigo C. The lymphomas of Brazilian children. JAMA 1969; 208: 1365-8. Links
15 Evans AS, Rothfield NF, Niederman JC. Raised antibody titers to EB virus in systemic lupus erythematosus. Lancet 1971; 1: 167-8. Links
16 Yokochi T, Yanagawa A, Kimura Y, Mizushima Y. High titer of antibody to the Epstein-Barr virus membrane antigen in sera from patients with rheumatoid arthritis and systemic lupus erythematosus. J Rheumatol 1989; 16: 1029-32. Links
17 Klippel JH, Decker JL, Grimley PM, Evans AS, Rothfield NF. Epstein-Barr virus antibody and lymphocyte tubuloreticular structures in systemic lupus erythematosus. Lancet 1973; 7837: 1057-8. Links
18 Sculley DG, Sculley TB, Pope JH. Reactions of sera from patients with rheumatoid arthritis, systemic lupus erythematosus and infectious mononucleosis to Epstein-Barr virus-induced peptides. J Gen Virol 1986; 67: 2253-8. Links
19 Tsai YT, Chiang BL, Kao YF, Hsieh KH. Detection of Epstein-Barr virus and cytomegalovirus genome in white blood cells from patients with juvenile rheumatoid arthritis and childhood systemic lupus erythematosus. Int Arch Allergy Immunol 1995; 106: 235-40. Links
20 Rider JR, Ollier WER, Lock RJ, Brookes ST, Pamphilon DH. Human cytomegalovirus infection and systemic lupus erythematosus. Clin Exp Rheum 1997; 15: 405-9. Links