Spondyloarthritis (SpA) is a heterogeneous group of diseases which include ankylosing spondylitis, reactive arthritis, psoriatic arthritis, arthritis associated with inflammatory bowel disease and undifferentiated SpA. Uveitis is characterised by intraocular inflammation of the iris, ciliary body and/or choroids. Acute anterior uveitis is classically described as the most common extra-articular manifestation of SpA. Symptoms of acute anterior uveitis include generally an acute onset of eye redness and pain, intense photophobia, blepharospasm and myosis. There may be a reduction of visual acuity when there are medium opacities or cystoid macular oedema. Pain is caused by ciliary spasm secondary to anterior chamber inflammation and can irradiate all around the distribution area of the first branch of the trigeminus nerve, including the periorbital area. The initial episode classically tends to subside in 2–4 months without treatment.1 The relationship between uveitis and SpA is well recognised, and uveitis is usually described as the most common extra-articular manifestation in SpA; however, there are no definitive data about the prevalence of uveitis in SpA. It appears that uveitis occurs in 10–50% of patients with SpA.2 3 Thus this prevalence varies greatly. Furthermore, there are conflicting data about potential complications of uveitis in SpA—that is, loss of visual acuity.

The objective of this analysis was to determine the prevalence of uveitis in SpA according to the disease duration and the type of SpA, and to describe the clinical characteristics of uveitis, by a systematic literature review.

MATERIAL AND METHODS

To identify relevant studies, allowing calculation of the prevalence of uveitis or to determine its characteristics in SpA, or both a systematic analysis of literature was performed in October 2006.

Literature search strategy

Publications were identified in the PubMed Medline database up to October 2006 and in the abstracts of EULAR (European League Against Rheumatism) and the ACR (American College of Rheumatology) scientific meetings of the years 2004, 2005 and 2006. Two searches were carried out in the Medline database which used the following key words: for the first search “(spondyloarthropathy OR reactive arthritis) and uveitis” (limits were: English, French, humans); for the second search “spondylarthritis”(MeSH) (limits were: randomized controlled trial, English, French, humans).Original articles were selected if they reported the prevalence or the characteristics of uveitis in SpA. The analysis concerned adults over 16 years old; studies concerning juvenile arthritis were excluded. Lastly, a hand-search of references was also performed. All types of SpA were analysed: ankylosing spondylitis, psoriatic arthritis, reactive arthritis, arthritis associated with inflammatory bowel disease and undifferentiated SpA.

Data collection

The articles were analysed by one author (NZ) according to a predetermined abstraction sheet. The following items were collected: number and type of SpA (and criteria for the diagnosis), number of patients reported as having uveitis (and type of definition: ophthalmologist examination, history of uveitis documented…), sex, age at the moment of the study, disease duration and age at the first uveitis flare, HLA-B27 positivity, characteristics of uveitis (acute or chronic, anterior or posterior, unilateral or bilateral, recurrent and number of recurrences for each patient) and complications of uveitis (decreased visual acuity, posterior synechias, macular cystoid oedema).

Statistical analysis

Statistics were descriptive. The prevalence of uveitis in SpA was calculated as the number of patients with SpA presenting one or more episodes of uveitis over the total number of patients with SpA. Standard deviations and means were calculated based on available data. In some cases, odds ratios were calculated.

RESULTS

Article selection

The Medline database and the congress abstracts search yielded 935 articles. With the hand search of references, a total of 134 articles and 10 abstracts were entered in the analysis (fig 1). Main reasons for exclusion were estimation of prevalence impossible and no characteristics of uveitis (41.9%), articles not related to SpA or uveitis (44.8%).

• Download figure
• Open in new tab
• Download powerpoint

Figure 1 Flow diagram of systematic literature search for articles reporting prevalence or characteristics of uveitis, or both, in spondyloarthropathies.

Prevalence

To calculate prevalence, a total of 126 articles (116 original articles and 10 abstracts) was assessed. The total number of patients with SpA was 29 877. The diagnosis of SpA was based on New York criteria or modified New York criteria for 9076 patients (30.4%), expert opinion for 8024 (26.9%), European Spondyloarthropathy Study Group (ESSG) criteria for 5873 (19.7%), ESSG and Amor’s criteria for 1091 (3.7%), other criteria (Amor’s criteria alone, Rome’s criteria, ESSG and New York criteria, ESSG and New York and Amor’s criteria, ESSG criteria and expert opinion) for 2322 (7.8%). Diagnosis criteria were not identified for 3491 patients with SpA (11.5%).4^–7 Specific diagnosis criteria used for diagnosis of psoriatic arthritis and reactive arthritis were not collected.

Of 26 168 patients, 18 037 (68.9%) were male. Mean (SD) age was 43.9 (2.7) years. Mean disease duration was 17.7 (1.0) years for the whole sample but varied with the type of SpA: 17.0 (1.0) years for ankylosing spondylitis, 17.4 (1.0) years for psoriatic arthritis, 22 (10.0) years for arthritis associated with inflammatory bowel disease, 5.8 (0.6) years for reactive arthritis and 5.8 (0.7) years for undifferentiated SpA. HLA-B27 was present in 9521/11 758 patients (81.0%).

A total of 9757 patients (32.7 (0.5)%) were reported as having presented one or more flares of uveitis. Among these patients, uveitis was defined by a self-reported history of uveitis for 3905 patients (40.0%) and an ophthalmological examination for 3078 (31.5%). Uveitis was not defined for 2774 patients (28.5%).

The prevalence of uveitis in SpA was 32.7 (0.5)% for a mean disease duration of 17.7 (1.0) years; this prevalence increased with the mean disease duration: for a mean disease duration of <5 years the prevalence was 12.3 (3.0)% whereas for a mean disease duration of >30 years it was 43.0 (3.6)% (fig 2).

• Download figure
• Open in new tab
• Download powerpoint

Figure 2 Prevalence of uveitis in spondyloarthritis (%) according to the mean disease duration (years).

Prevalence varied slightly according to the definition used for uveitis: in patients with uveitis defined by a history of uveitis the prevalence was 30.5 (0.8)%; in patients with uveitis defined by an ophthalmological examination it was 32.8 (1.0)% (p<0.001). Prevalence varied also with the type of definition used for SpA: 34.9 (1.2)% for ESSG criteria versus 30.9 (1.0)% for New York and modified New York criteria (p<0.001).

Furthermore, uveitis prevalence varied with the type of SpA (table 1): it was 33.2 (0.8)% in ankylosing spondylitis, 25.1 (2.3)% in psoriatic arthritis, 36.9 (4.4)% in arthritis associated with inflammatory bowel disease, 25.6 (3.8)% in reactive arthritis and 13.2 (2.9)% in undifferentiated SpA.

The prevalence of uveitis was higher in patients with SpA presenting HLA-B27 positivity: 39.8% for B27 positivity versus 13.6% B27 negativity—that is, an odds ratio of 4.2 (95% confidence interval 3.3 to 5.3) (p<0.001).

This prevalence was also different according to sex: women were more affected by uveitis than men (33.3% vs 28.5%)—that is, an odds ratio of 1.3 (95% confidence interval 1.1 to 1.4) (p<0.001).

Characteristics and complications of uveitis during SpA

Characteristics and complications of uveitis could be determined in 36 articles—that is, a total of 1989 patients with uveitis and SpA. For these patients, the diagnosis of SpA was based on: expert opinion for 1108 (55.7%), ESSG or Amor’s criteria for 537 (27.0%), New York criteria or modified New York criteria for 258 (13.0%) and Rome’s criteria for 26 (1.3%). Diagnosis criteria of SpA were not identified for 60 patients (3.1%).

Uveitis was defined by an ophthalmological examination for 1182 patients (59.4%), by anamnesis for 733 (36.9%) and not identified for 74 (3.7%).

Mean (SD) age at the first flare of uveitis was 37.0 (4.8) years. Uveitis was acute in 88.7 (2.5)% of patients (549/619), anterior in 90.5 (2.0)% (724/800) and unilateral (affecting one eye at a time even if alternating) in 87.3 (2.8)% (460/527). It was bilateral in 9.1 (2.5% (48/527) and posterior in 3.5 (1.3)% (26/741). No data about the duration of acute flares were available. More than one episode of uveitis occurred in 50.6 (2.6)% of patients (737/1456 patients). Uveitis was complicated with a reduction of visual acuity in 8.3 (3.8)%? (17/206 available data). There was no precision on the degree of visual loss. The other complications observed were posterior synechias in 39.5 (4.6)% of patients (171/433), macular oedema in 19.7 (4.1)% (70/356).

DISCUSSION

This systematic literature review reports the prevalence and characteristics of uveitis in SpA. In this systematic analysis the prevalence of uveitis in SpA was high: 32.7 (0.5)% for a mean disease duration of 17.7 (1.0) years. Prevalence of uveitis varied with the type of SpA: it was higher in ankylosing spondylitis and lower in undifferentiated SpA. Prevalence increased with disease duration. The main characteristic of uveitis in SpA was the possibility of recurrence with 50.6 (2.6)% of patients having had more than one flare.

The lifetime cumulative incidence of definite acute anterior uveitis is about 0.2% in the general population and 1% in the HLA-B27 positive population.8 This study confirms that uveitis is a common extra-articular feature of SpA whatever the underlying disease, with a prevalence ranging from 13.2 (2.9)% in undifferentiated SpA to 36.9 (4.4)% in patients with arthritis associated with inflammatory bowel disease. However, data about the prevalence of uveitis in arthritis associated with inflammatory bowel disease are scarce and the prevalence was based on only three articles (453 patients).9^–11 It is necessary to confirm these data in larger cohorts. This estimation is higher than the prevalence in inflammatory bowel disease without arthritis previously published—that is, 4–7%.12 13 The prevalence of uveitis in undifferentiated SpA was lowest (13.2 (2.9)%) but the mean disease duration was short (5.8 (0.7) years). There are few studies with a long follow-up of patients with this type of SpA, a concept recently introduced in the classification of SpA.6 7 In ankylosing spondylitis, the prevalence was obtained in a larger population of patients (12 768). These results confirm a high prevalence of uveitis in ankylosing spondylitis.

The prevalence of uveitis in SpA varies not only according to the type of SpA but also according to the sex of the patient: the prevalence was higher in women than men (odds ratio = 1.3 (95% confidence interval 1.1 to 1.4)). These data are discussed in the literature with contradictory results and need to be the object of large cohort studies; the question of the differences between male and female symptoms in SpA has not been elucidated.14^–17 One of the more convincing findings of this analysis is the higher risk of uveitis in B27-positive patients (odds ratio = 4.2 (3.3 to 5.3)). Previous studies reported a prevalence of uveitis ranging from 40% to 48% in patients with SpA presenting HLA-B27 positivity versus 15–25% in negative HLA-B27 patients.18^–21

The main limit of this systematic literature review is a possible selection bias. Some patients may have been included in more than one of the retrieved articles, with a partial “overlap” of data as a consequence. And because it is also likely—at least for the articles reporting results from clinical trials—that patients may have been selected because of a higher activity and/or severity of their disease, the estimated prevalence might be overestimated. Another limit of this work is that the prevalence of uveitis was obtained mainly through patient-reported history. However, it should be noted that the same figures for prevalence were obtained through assessment by ophthalmologists.

The interpretation of prevalence must take into account disease duration. Indeed, in this analysis the prevalence of uveitis in SpA increased with the mean disease duration; this prevalence reached 43.0 (3.6)% for a mean disease duration >30 years. This notion confirmed previous cohort studies which found a prevalence of uveitis >40%.for a mean disease duration of >20 years.22 23 In addition, from this literature review, it can be seen that the prevalence of uveitis in SpA increases linearly with disease duration up to 20 years, but after this time the prevalence seems to reach a plateau, which could be interpreted as follows: the incidence of first uveitis is probably constant during the first 20 years of the disease (with a roughly linear increase of the prevalence as a consequence), but becomes much lower after the 20th year of disease duration. According to these results, a patient who has not had uveitis after 20 years of disease may have a very low (virtually no) risk of occurrence of the disease (fig 2). However, it should be noted that these analyses concerned mean disease duration, and not true categories of disease duration, which is one of the limitations of this work, which is unavoidable in systematic literature reviews.

This work also examined the question of the characteristics and complications of uveitis in SpA. There are conflicting data about the potential complications of uveitis in SpA—that is, recurrences and loss of visual acuity. Furthermore, the literature was scarce and not precise. In this systematic literature analysis, the decrease of visual acuity was estimated at 8.3 (3.8)% of patients: according to reports, it varied between 6% and 12% for a significant decrease of visual acuity (more than two decimal places).24 25 Recurrences occurred in 50.6 (2.6)%. Some publications have studied complications of uveitis HLA-B27 positive versus HLA-B27 negative. According to some authors anterior uveitis associated with the HLA-B27 haplotype has a less favourable prognosis and recurrences are more frequent.24 26

The similarities and differences in the pathogenesis of uveitis and arthritis remain a subject of discussion and research. The HLA-B27 haplotype is decidedly linked to the association between arthritis and uveitis, as described for the first time in 1973 by Brewerton et al.27 28 Polymorphism in the LMP2 gene and in HLA-DR8 may also influence the relative risk for acute anterior uveitis in patients with ankylosing spondylitis.29 30 Finally, there is probably a common genetic predisposition; uveitis and articular manifestations in SpA may have similar pathophysiology but the precise mechanisms underlying the involvement of these tissues remain unclear.31

Some recent articles have reported the efficacy of anti-tumour necrosis factor α in the treatment of uveitis in SpA; in particular, it appears that anti-tumour necrosis factor antibodies (infliximab and adalimumab) decrease the rate of recurrences.32^–34 In this context it was particularly important to establish the basal prevalence of uveitis in SpA, and the rate of recurrences. These data may then be used as baseline comparators for further studies.