You are here

Article Text

Article menu

Download PDFPDF

Clinical and epidemiological research
Extended report
Infections and the risk of incident giant cell arteritis: a population-based, case-control study
  1. Rennie L Rhee1,
  2. Peter C Grayson2,
  3. Peter A Merkel1,3,
  4. Gunnar Tomasson4
  1. 1Division of Rheumatology, University of Pennsylvania, Philadelphia, Pennsylvania, USA
  2. 2Systemic Autoimmunity Branch, National Institute of Arthritis and Musculoskeletal and Skin Diseases, National Institutes of Health, Bethesda, Maryland, USA
  3. 3Department of Biostatistics and Epidemiology, University of Pennsylvania, Philadelphia, Pennsylvania, USA
  4. 4Department of Public Health Sciences, University of Iceland, Reykjavik, Iceland
  1. Correspondence to Dr Rennie L Rhee, Department of Rheumatology, Perelman School of Medicine at the University of Pennsylvania, 3400 Spruce Street, White Building 5th Floor, Philadelphia, PA 19104, USA; Rennie.rhee{at}uphs.upenn.edu

Abstract

Objectives Alterations in the immune system and infections are suspected to increase susceptibility to giant cell arteritis (GCA). Recently herpes zoster has been directly implicated in the pathogenesis of GCA. We examined the association between prior infections, in particular herpes zoster, and incident GCA in a population-based cohort.

Methods A nested case-control study was performed using an electronic database from the UK. Cases with newly diagnosed GCA were identified using a validated algorithm and compared with age-matched, sex-matched and practice-matched controls. Conditional logistic regression was used to examine the relationship between any infection or herpes zoster infection on the development of GCA after adjusting for potential confounders; results were expressed as incidence rate ratios (IRRs).

Results There were 4559 cases of GCA and 22 795 controls. Any prior infection and herpes zoster were associated with incident GCA (IRR 1.26 (95% CI 1.16 to 1.36), p<0.01; and 1.17 (95% CI 1.04 to 1.32), p<0.01, respectively). A greater number of infections was associated with a higher risk of developing GCA (IRR for 1, 2–4 and ≥5 infections was 1.28, 1.60 and 2.18, respectively).

Conclusions Antecedent infections and, to a lesser extent, herpes zoster infections are modestly associated with incident GCA. These data provide population-level support for the hypothesis that long-standing alterations of the immune system are associated with susceptibility to GCA and suggest that herpes zoster is unlikely to play a major causal role in the pathogenesis of GCA.

  • Giant Cell Arteritis
  • Infections
  • Epidemiology

https://doi.org/10.1136/annrheumdis-2016-210152

Statistics from Altmetric.com

    Request Permissions

    If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.

    Introduction

    Giant cell arteritis (GCA) is a systemic vasculitis affecting medium and large arteries in adults over the age of 50 years. Potential causative mechanisms are still being elucidated but evidence suggests that GCA may be an antigen-driven disease although the exact trigger is still unknown.1 The involvement of vascular dendritic cell activation, T lymphocytes, interferon-γ and macrophages in the pathogenesis of GCA may indicate that an infectious agent is the cause.2

    Prior studies have investigated the potential association between microbes and GCA using temporal artery specimens but these studies had conflicting findings.3–6 Recently, varicella zoster virus (VZV) has been directly implicated in the pathogenesis of GCA after VZV antigen was found in a large majority of temporal arteries of patients with GCA and use of antiviral medication has been suggested as adjunctive therapy in GCA;7 ,8 however, independent confirmation of these findings are still needed. These studies raise new questions regarding the role of microbes in GCA and whether their involvement in the pathogenesis begins months and years prior to the onset of GCA. Alternatively, alterations in the immune system seen in patients with GCA, including ageing of the immune system (or immunosenescence), may also increase susceptibility to infections.9 ,10

    The objective of this study is to determine the degree to which infections, including clinically evident herpes zoster, are associated with incident GCA using data from a large population-based cohort. Additionally, because prior infections may be temporally associated with the onset of GCA without playing a causal role in disease pathogenesis, the relationship between common infections and incident GCA was explored.

    Methods

    Study design and data source

    We conducted a nested matched case-control study with incidence density sampling using The Health Improvement Network (THIN), a population-based database from the UK.11 This design is computationally more efficient than a cohort study and produces ORs that are unbiased estimates of incidence rate ratios (IRRs). The THIN database contains electronic medical records of over 11 million persons in the UK and is representative of the general UK population in terms of demographics and common illnesses.12 THIN includes information on demographics, medical diagnoses and drug prescriptions. Medical diagnoses are recorded using Read codes, the standard classification system in the UK. The study was approved by the THIN Scientific Review Committee and considered exempt by the Institutional Review Board of the University of Pennsylvania.

    Case definition

    All patients receiving medical care from 1994 to 2015 from a THIN practitioner were eligible for inclusion. Using a previously validated algorithm, cases were defined as patients over the age of 50 years with at least one Read code for GCA, temporal arteritis or Horton disease, and a prescription for glucocorticoids.13 Glucocorticoid use was defined as at least two prescriptions for oral glucocorticoids: one within 6 months of the diagnosis date and the second within 6 months of the first prescription. The index date was the date of the first diagnosis code. To avoid the inclusion of prevalent cases, cases were included only if the initial Read code occurred at least 6 months after registration into the THIN database and after the practice implemented Vision software, a program which automatically codes and enters clinical data into the database. Patients with a prior diagnosis of polymyalgia rheumatica, based on Read codes, were excluded.

    Control selection

    The control group was selected based on incidence density sampling. For each case, five controls that were alive and free of GCA at the index date were selected and matched on age (within 5 years), sex and general practice.

    Exposure and covariate assessment

    The main exposures of interest were clinically evident herpes zoster infection or any infection prior to the index date based on a single Read code. Herpes zoster codes coupled with antiviral prescriptions written within 1 month were also examined to improve specificity for herpes zoster infection. Infections were identified using Read codes that were analogous to a list of international classification of disease V.9 codes for infections used previously.14 ,15 Aside from herpes zoster infections, a prescription for antimicrobials was not required in order to include viral infections. We also examined the occurrence of any infection, number of infections and five common infections in the elderly (infections affecting the respiratory tract, urinary tract, gastrointestinal tract, conjunctiva, and skin and soft tissue).16 The timing of infections prior to the index date were evaluated to determine if infections closer to the index date were more highly associated with GCA compared with more remote infections. Due to the possibility of misdiagnosis of symptoms of GCA as infections, infections that occurred within 6 months prior to the date of the diagnosis of GCA were not included in the primary analysis but were analysed in a sensitivity analysis (see online supplementary text).

    Covariates included receipt of zoster vaccination, prior use of immunosuppressive therapies (glucocorticoids, azathioprine, ciclosporin, leflunomide, methotrexate, mycophenolate or sulfasalazine) earlier than 6 months from index date, any prior alcohol use, smoking history (ever—yes/no) and comorbidities at index date categorised using the Charlson Comorbidity Index.17 ,18 The majority of prescriptions for oral immunosuppressive drugs are accurately captured in THIN.19 ,20

    Statistical analysis

    For comparison of the baseline characteristics, we used Student's t-test (or, if not normally distributed, Wilcoxon rank sum) and χ2 test for continuous and categorical data, respectively. We used conditional logistic regression and, because the resulting ORs from the models accurately estimated the IRRs, the relationship between infection and the development of GCA was expressed with IRR and 95% CIs. Linear test for trend was performed to examine whether increasing categories of infection count was associated with a linear increase in GCA risk. All analyses were adjusted for prior glucocorticoid use, prior use of non-glucocorticoid immunosuppressive therapies, alcohol use, smoking history and the Charlson Comorbidity Index. To determine if receipt of the zoster vaccination was an effect modifier on the relationship between herpes zoster infection and GCA, a herpes zoster-by-vaccination interaction was included in the model along with herpes zoster infection and receipt of zoster vaccine as separate variables. Patients were also analysed according to the time period prior to index date in which the infection occurred to determine if recent infections were more highly associated with GCA compared with earlier infections; infections diagnosed within the 6 months prior to diagnosis were included in this analysis. To account for potential confounding due to differential use of glucocorticoids or other immunosuppressive medications among cases and controls, a sensitivity analysis excluding patients who previously received at least one prescription for an oral glucocorticoid or immunosuppressive medication was performed. Additional sensitivity analyses that were performed are listed in the online supplementary text. A significance level of 0.05 was used for all tests of hypothesis. All analyses were performed using Stata V.12.1 (Stata, College Station, Texas, USA).

    Results

    Study population

    There were 4559 cases of GCA and 22 795 matched controls. Patients with GCA were more likely to have received a prescription for oral glucocorticoids and other immunosuppressive therapy ≥6 months prior to the index date. There was no difference in the proportion who received the zoster vaccination between cases and controls (table 1). Evaluation of the individual diagnostic categories within the Charlson Comorbidity Index revealed that, prior to the index date, patients with GCA were more likely to have cerebrovascular disease (9% vs 7%, p<0.01), chronic pulmonary disease (22% vs 15%, p<0.01), mild liver disease (0.6% vs 0.4%, p=0.02), peptic ulcer disease (6% vs 4%, p<0.01), peripheral vascular disease (5% vs 3%, p<0.01) and renal disease (13% vs 11%, p<0.01), and less likely to have a pre-existing rheumatic disease (3% vs 5%, p<0.01) or dementia (1% vs 2%, p<0.01) compared with controls.

    View this table:
    Table 1

    Characteristics of study subjects

    Herpes zoster and GCA

    GCA cases were more likely to have had a prior herpes zoster infection (9% vs 7%, p<0.01; table 2). However, when the definition of a herpes zoster infection was restricted only to Read codes accompanied by prescription of an antiviral, the difference was attenuated and no longer significant.

    View this table:
    Table 2

    Unadjusted occurrence of infections among cases of GCA and controls

    After adjustment for potential confounders, a prior history of herpes zoster infection greater than 6 months before the index date was associated with an increased risk of developing GCA (adjusted IRR 1.17 (95% CI 1.04 to 1.32), p<0.01) (table 3). Similar results were found even after excluding patients who previously received oral glucocorticoids or immunosuppressive therapies (data not shown). The results of additional sensitivity analyses (see online supplementary text) led to similar results (see online supplementary table S1).

    View this table:
    Table 3

    The association of infections with incident giant cell arteritis

    After stratifying by time period prior to the index date, there was no association between herpes zoster infection and GCA among the different time strata except for herpes zoster infections that occurred 5–10 years before the index date. Any infection was significantly associated with incident GCA in all time periods with the greatest association seen with infections which occurred in the first year before the index date (table 4).

    View this table:
    Table 4

    The association of herpes zoster and all infections with incident giant cell arteritis stratified by time period prior to index date

    There were 115 (0.4%) patients in the cohort who received the zoster vaccination. The herpes zoster-by-vaccination interaction term was statistically significant (p for interaction=0.04). That is, patients who received the zoster vaccine had a stronger association between herpes zoster infection and GCA (adjusted IRR 3.53 (95% CI 1.17 to 10.62)) compared with those who did not receive the vaccine (adjusted IRR 1.16 (95% CI 1.03 to 1.30)), even after adjusting for potential confounders.

    Other infections and GCA

    There was a higher occurrence of any infection in patients with GCA compared with controls (68% vs 61%, p<0.01) (table 2). Patients who had at least one prior infection were more likely to develop GCA (adjusted IRR 1.26 (95% CI 1.16 to 1.36), p<0.01) (table 3). A higher number of prior infections was also associated with a greater risk of GCA: 0 (reference), 1, 2–4 or ≥5 infections yielded IRRs of 1, 1.28, 1.60 and 2.18, respectively (test for trend: p<0.01).

    Analysis of the five common types of infections showed a higher risk of GCA with each type of infection: respiratory tract (IRR 1.25 (95% CI 1.16 to 1.35), p<0.01), urinary tract (IRR 1.26 (95% CI 1.15 to 1.38), p<0.01), gastrointestinal tract (IRR 1.33 (95% CI 1.15 to 1.53), p<0.01), conjunctiva (IRR 1.25 (95% CI 1.14 to 1.37), p<0.01), and skin and soft tissue (IRR 1.10 (95% CI 1.00 to 1.21), p=0.04). A sensitivity analysis excluding patients who previously received glucocorticoids or other immunosuppressive therapies led to similar results (data not shown).

    Infection was associated with incident GCA within all strata of time periods of infection prior to index date with the greatest association being within 6 months prior to the index date (table 4).

    Discussion

    This study used a large population-based cohort to examine the association between infections and newly diagnosed GCA. We found that prior infections were associated with the development of GCA in a ‘dose-dependent’ fashion, but only a modest association was found between herpes zoster infection and GCA.

    There is limited information on the relationship between common infections and incident GCA. We found that the risk of incident GCA was higher among patients who had prior infections and was similar among several common types of infections. Except for the 1st year prior to the index date, the association between infection and GCA was similar regardless of the time period in which the infection occurred, although these subgroup analyses were limited by smaller sample sizes. However the greatest association between infection and GCA was seen in the 1st year prior to the diagnosis. This finding may be due to misclassification bias if symptoms of GCA (such as fever) were misdiagnosed as infection. Alternatively, it is possible infections are directly involved in the pathogenesis of disease or that greater immune dysregulation occurs prior to the onset of disease. One prior study found an increased short-term effect of infections on the occurrence of GCA such that infections were three times more likely to occur in patients who developed GCA compared with matched controls.21 Our study adds to this finding and shows that even years prior to onset of GCA, patients with GCA are more likely to have infections compared with controls. It is also notable that the occurrence of infection in GCA only modestly increases after diagnosis and initiation of glucocorticoids, also suggesting that medications alone cannot explain the increased association of infections in patients with GCA.22

    Recently, the VZV antigen was found in the great majority of temporal artery biopsy specimens of patients with GCA7 whereas prior studies also using histological specimens found no evidence of VZV in temporal artery biopsies.5 ,6 ,23 Our study found a minimal-to-no association of clinically overt herpes zoster with GCA, thus not providing population-level support to recent observations on the association between VZV and GCA. The lack of a strong temporal association between herpes zoster infections and GCA further contradicts the possibility that herpes zoster has a causal role in the onset of GCA. However, it is possible that the histological findings can be explained by latent or subclinical VZV (eg, without classic skin manifestations) which our study was not able to assess. Similarly, reactivation of VZV at time of GCA onset may also explain the histological findings; however, when we included herpes zoster infections within 6 months of GCA diagnosis, results were similar suggesting that there was not a higher occurrence of clinically overt herpes zoster infections at the time of diagnosis.

    There are several possible explanations for the association between infection and development of GCA. Infections have long been theorised to induce and perpetuate autoimmunity through alterations in the immune system.24 ,25 Additionally, patients with GCA may simply be more predisposed to having infections possibly as a result of immune dysregulation. The association with GCA and older age suggests that immunosenescence, or the ageing of the immune system, is involved in susceptibility for GCA.9 ,10 Whether infections are involved in the causal pathway or are simply a marker of immune dysfunction is still unclear. Lastly, the possibility of a vascular microbiome is emerging as studies now refute the idea that blood vessels are sterile.26 Perturbances in the vascular microbiome from infections and/or antimicrobial therapy may increase susceptibility to GCA, although no studies have yet demonstrated this directly.3 ,27 ,28

    Interestingly, the zoster vaccine was a significant effect modifier on the relationship between herpes zoster infection and GCA, such that herpes zoster infection conferred a much higher risk of GCA among patients who did versus patients who did not receive the zoster vaccine. However, this finding should be interpreted with caution since receipt of the zoster vaccine in this study was exceedingly rare (n=115 (0.4%)) given that the zoster vaccine became routinely available in the UK in September 2015 and only among individuals aged 70–78 years.29 If a true interaction between herpes zoster infection and zoster vaccine exists, the possibility of an adjuvant effect to the immune system by re-inoculation with the antigen may explain this association. Alternatively, other risk factors may be present in patients with GCA prior to diagnosis which increase their likelihood of receiving vaccination. Additional studies with a larger number of patients who received the vaccine are needed to confirm this finding.

    Our study has several strengths. Performing a nested case-control study within a population-based cohort enabled us to obtain a large sample of patients with a relatively rare disease. This study is one of the largest cohorts of patients with GCA. Use of the THIN database has several advantages over other administrative databases including availability of electronic medical record data (not just claims-based data) which provide greater depth of information as well as long periods of follow-up, and at times the lifelong health record of an individual patient, enabling us to examine long latency periods between exposure and outcome. The THIN data have been used extensively in epidemiological studies including studies of GCA.13 ,22 ,30

    There are also several limitations of our study to consider. Misclassification may have occurred as Read codes for infections were not validated and may not have accurately identified the exposures. Misclassification of the diagnosis of GCA may have occurred but would likely bias to the null, further strengthening our results. Glucocorticoids and other immunosuppressive therapies were more often prescribed to patients who later developed GCA suggesting that other potential comorbidities and/or immunosuppressive therapies not accounted for in the analysis may have confounded results. While we did not perform an in-depth exploration for the reasons for antecedent use of glucocorticoids, examination of the diagnostic categories comprising the Charlson Comorbidity Index revealed that a significantly greater proportion of patients with GCA had a chronic pulmonary disease and patients with GCA were less likely to have a pre-existing rheumatic disease (which included juvenile idiopathic arthritis, lupus, myositis, polymyalgia rheumatica, rheumatoid arthritis and scleroderma); the higher prevalence of pulmonary disease in patients with GCA may possibly explain the higher use of glucocorticoids prior to diagnosis. Furthermore, significant associations between infection and the risk of incident GCA were observed in multivariable models that adjusted for the use of immunosuppressive therapies and in analyses that excluded patients who had previously received glucocorticoids or other immunosuppressive medications. Lastly, patients with pre-existing polymyalgia rheumatica, which can often precede the diagnosis of GCA, were excluded which limited the ability to examine this important subgroup.31

    In summary, this study found that antecedent infections are associated with incident GCA, although infections are probably a minor determinant of overall risk of GCA. The modest association seen between herpes zoster and GCA suggests that herpes zoster infections are unlikely to play a major causal role in the pathogenesis of GCA. These data provide population-level support for the hypothesis that long-standing alterations of the immune system are associated with susceptibility to GCA.

    Acknowledgments

    The authors thank Junko Takeshita, MD, PhD for allowing them to use her compiled list of Read codes for identifying infections in the THIN database.

    References

    1. ↵
      1. Weyand CM,
      2. Goronzy JJ
      . Giant cell arteritis as an antigen-driven disease. Rheum Dis Clin North Am 1995;21:1027–39.
      OpenUrlPubMedWeb of Science
    2. ↵
      1. Weyand CM,
      2. Goronzy JJ
      . Immune mechanisms in medium and large-vessel vasculitis. Nat Rev Rheumatol 2013;9:731–40. doi:10.1038/nrrheum.2013.161
      OpenUrlCrossRefPubMed
    3. ↵
      1. Bhatt AS,
      2. Manzo VE,
      3. Pedamallu CS, et al
      . In search of a candidate pathogen for giant cell arteritis: sequencing-based characterization of the giant cell arteritis microbiome. Nat Rev Rheumatol 2014;66:1939–44. doi:10.1002/art.38631
      OpenUrl
    4. ↵
      1. Gabriel SE,
      2. Espy M,
      3. Erdman DD, et al
      . The role of parvovirus B19 in the pathogenesis of giant cell arteritis: a preliminary evaluation. Arthritis Rheum 1999;42:1255–8. doi:10.1002/1529-0131(199906)42:6<1255::AID-ANR23>3.0.CO;2-P
      OpenUrlCrossRefPubMedWeb of Science
    5. ↵
      1. Rodriguez-Pla A,
      2. Bosch-Gil JA,
      3. Echevarria-Mayo JE, et al
      . No detection of parvovirus B19 or herpesvirus DNA in giant cell arteritis. J Clin Virol 2004;31:11–15. doi:10.1016/j.jcv.2004.05.003
      OpenUrlCrossRefPubMed
    6. ↵
      1. Alvarez-Lafuente R,
      2. Fernández-Gutiérrez B,
      3. Jover JA, et al
      . Human parvovirus B19, varicella zoster virus, and human herpes virus 6 in temporal artery biopsy specimens of patients with giant cell arteritis: analysis with quantitative real time polymerase chain reaction. Ann Rheum Dis 2005;64:780–2. doi:10.1136/ard.2004.025320
      OpenUrlAbstract/FREE Full Text
    7. ↵
      1. Gilden D,
      2. White T,
      3. Khmeleva N, et al
      . Prevalence and distribution of VZV in temporal arteries of patients with giant cell arteritis. Neurology 2015;84:1948–55. doi:10.1212/WNL.0000000000001409
      OpenUrlCrossRefPubMed
    8. ↵
      1. Gilden D,
      2. Nagel MA
      . Varicella zoster virus triggers the immunopathology of giant cell arteritis. Curr Opin Rheumatol 2016;28:376–82. doi:10.1097/BOR.0000000000000292
      OpenUrl
    9. ↵
      1. Ginaldi L,
      2. Loreto MF,
      3. Corsi MP, et al
      . Immunosenescence and infectious diseases. Microbes Infect 2001;3:851–7. doi:10.1016/S1286-4579(01)01443-5
      OpenUrlCrossRefPubMedWeb of Science
    10. ↵
      1. Mohan SV,
      2. Liao YJ,
      3. Kim JW, et al
      . Giant cell arteritis: immune and vascular aging as disease risk factors. Arthritis Res Ther 2011;13:231. doi:10.1186/ar3358
      OpenUrlCrossRefPubMed
    11. ↵
      1. Zhang Y,
      2. Lu N,
      3. Peloquin C, et al
      . Improved survival in rheumatoid arthritis: a general population-based cohort study. Ann Rheum Dis Published Online First: 23 Jun 2016. doi: 10.1136/annrheumdis-2015-209058 http://dx.doi.org/10.1136/annrheumdis-2015-209058
    12. ↵
      1. Blak BT,
      2. Thompson M,
      3. Dattani H, et al
      . Generalisability of The Health Improvement Network (THIN) database: demographics, chronic disease prevalence and mortality rates. Inform Prim Care 2011;19:251–5.
      OpenUrlPubMed
    13. ↵
      1. Smeeth L,
      2. Cook C,
      3. Hall AJ, et al
      . Incidence of diagnosed polymyalgia rheumatica and temporal arteritis in the United Kingdom, 1990–2001. Ann Rheum Dis 2006;65:1093–8. doi:10.1136/ard.2005.046912
      OpenUrlAbstract/FREE Full Text
    14. ↵
      1. Patkar NM,
      2. Curtis JR,
      3. Teng GG, et al
      . Administrative codes combined with medical records based criteria accurately identified bacterial infections among rheumatoid arthritis patients. J Clin Epidemiol 2009;62:321–7, 327 e1–7. doi:10.1016/j.jclinepi.2008.06.006
      OpenUrlCrossRefPubMedWeb of Science
    15. ↵
      1. Grijalva CG,
      2. Chen L,
      3. Delzell E, et al
      . Initiation of tumor necrosis factor-alpha antagonists and the risk of hospitalization for infection in patients with autoimmune diseases. JAMA 2011;306:2331–9. doi:10.1001/jama.2011.1692
      OpenUrlCrossRefPubMedWeb of Science
    16. ↵
      1. Htwe TH,
      2. Mushtaq A,
      3. Robinson SB, et al
      . Infection in the elderly. Infect Dis Clin North Am 2007;21:711–43, ix. doi:10.1016/j.idc.2007.07.006
      OpenUrlCrossRefPubMed
    17. ↵
      1. Charlson ME,
      2. Pompei P,
      3. Ales KL, et al
      . A new method of classifying prognostic comorbidity in longitudinal studies: development and validation. J Chronic Dis 1987;40:373–83. doi:10.1016/0021-9681(87)90171-8
      OpenUrlCrossRefPubMedWeb of Science
    18. ↵
      1. Khan NF,
      2. Perera R,
      3. Harper S, et al
      . Adaptation and validation of the Charlson Index for Read/OXMIS coded databases. BMC Fam Pract 2010;11:1. doi:10.1186/1471-2296-11-1
      OpenUrlCrossRefPubMed
    19. ↵
      1. Ogdie A,
      2. Langan SM,
      3. Parkinson J, et al
      . Medical Record Databases. In: Strom BL, Kimmel SE, Hennessy S, eds. Pharmacoepidemiology West Sussex, UK: Wiley-Blackwell Publications, 2012:224–7.
    20. ↵
      1. Ogdie A,
      2. Alehashemi S,
      3. Love TJ, et al
      . Validity of psoriatic arthritis and capture of disease modifying antirheumatic drugs in the health improvement network. Pharmacoepidemiol Drug Saf 2014;23:918–22. doi:10.1002/pds.3677
      OpenUrlCrossRefPubMed
    21. ↵
      1. Russo MG,
      2. Waxman J,
      3. Abdoh AA, et al
      . Correlation between infection and the onset of the giant cell (temporal) arteritis syndrome. A trigger mechanism? Arthritis Rheum 1995;38:374–80. doi:10.1002/art.1780380312
      OpenUrlPubMedWeb of Science
    22. ↵
      1. Durand M,
      2. Thomas SL
      . Incidence of infections in patients with giant cell arteritis: a cohort study. Arthritis Care Res (Hoboken) 2012;64:581–8. doi:10.1002/acr.21569
      OpenUrl
    23. ↵
      1. Nordborg C,
      2. Nordborg E,
      3. Petursdottir V, et al
      . Search for varicella zoster virus in giant cell arteritis. Ann Neurol 1998;44:413–14. doi:10.1002/ana.410440323
      OpenUrlCrossRefPubMedWeb of Science
    24. ↵
      1. Root-Bernstein R,
      2. Fairweather D
      . Complexities in the relationship between infection and autoimmunity. Curr Allergy Asthma Rep 2014;14:407. doi:10.1007/s11882-013-0407-3
      OpenUrlCrossRefPubMed
    25. ↵
      1. Sfriso P,
      2. Ghirardello A,
      3. Botsios C, et al
      . Infections and autoimmunity: the multifaceted relationship. J Leukoc Biol 2010;87:385–95. doi:10.1189/jlb.0709517
      OpenUrlAbstract/FREE Full Text
    26. ↵
      1. Reszka E,
      2. Jegier B,
      3. Wasowicz W, et al
      . Detection of infectious agents by polymerase chain reaction in human aortic wall. Cardiovasc Pathol 2008;17:297–302. doi:10.1016/j.carpath.2007.11.002
      OpenUrlCrossRefPubMedWeb of Science
    27. ↵
      1. Koening CL,
      2. Peterson SG,
      3. Podnecky NL, et al
      . Absence of bacteria in the temporal arteries of patients with giant cell arteritis. J Clin Rheumatol 2016;22:43–4. doi:10.1097/RHU.0000000000000344
      OpenUrl
    28. ↵
      1. Clifford A,
      2. Hoffman GS
      . Evidence for a vascular microbiome and its role in vessel health and disease. Curr Opin Rheumatol 2015;27:397–405. doi:10.1097/BOR.0000000000000184
      OpenUrl
    29. ↵
      1. Costello EK,
      2. Lauber CL,
      3. Hamady M, et al
      . Bacterial community variation in human body habitats across space and time. Science 2009;326:1694–7. doi:10.1126/science.1177486
      OpenUrlAbstract/FREE Full Text
    30. ↵
      1. Tomasson G,
      2. Peloquin C,
      3. Mohammad A, et al
      . Risk for cardiovascular disease early and late after a diagnosis of giant-cell arteritis: a cohort study. Ann Intern Med 2014;160:73–80. doi:10.7326/M12-3046
      OpenUrlPubMedWeb of Science
    31. ↵
      1. Salvarani C,
      2. Cantini F,
      3. Hunder GG, et al
      . Polymyalgia rheumatica and giant-cell arteritis. Lancet 2008;372:234–45. doi:10.1016/S0140-6736(08)61077-6
      OpenUrlCrossRefPubMedWeb of Science
    View Abstract

    Footnotes

    • Handling editor Tore K Kvien

    • Contributors All authors were involved in the conception and design of study, analysis and interpretation of data, and drafting of manuscript.

    • Funding RLR receives support from the Rheumatology Research Foundation (Scientist Development Award) and the Vasculitis Foundation (Fellowship Award).

    • Competing interests Intercontinental Marketing Services (IMS) Health Real World Evidence Solutions is an UK-based expert in anonymous patient data for the healthcare industry. IMS Health is a commercial organisation that supplies data and trains and supports researchers in the use of primary care patient data. Data are available for us in medical research in the academic setting as well as in industry for a fee which varies depending on the type of data requested. Aside from undergoing ethical review by The Health Improvement Network (THIN) Scientific Review Committee, independent academic groups who voluntarily act as an ethical review body, this protocol was not in any way discussed with IMS Health nor were any changes made by the company. We did not receive financial support or other forms of computational or analytical support from IMS/THIN. The data were collected by IMS and the general practitioners without knowledge of the study objectives and hypotheses.

    • Provenance and peer review Not commissioned; externally peer reviewed.

    • Data sharing statement Unpublished data are available to anyone after request. Please contact rennie.rhee@uphs.upenn.edu.