To evaluate the role of HLA typing as a diagnostic test in patients with uveitis.
Perspective derived from a literature review and analysis of reported results.
Published data on the HLA associations of several uveitis entities and their prevalence among patients and the general population were used to calculate the positive predictive value of HLA testing as a diagnostic test for these disorders.
For nearly all diagnostic entities evaluated (including multiple sclerosis–associated intermediate uveitis, birdshot chorioretinitis, sympathetic ophthalmic, Behçet disease, and Vogt-Koyanagi-Harada disease), the positive predictive value was low (< 0.50), indicating the limited usefulness of routinely applied HLA typing as a diagnostic test. HLA-B27 testing may be of value in identifying a previously undiagnosed or misdiagnosed spondyloarthropathy among patients with recurrent acute anterior uveitis.
In general, HLA typing has limited usefulness as a diagnostic test in patients with uveitis.
When evaluating a patient with uveitis, the clinician will characterize the disease along several axes, including: (1) anatomic location of the inflammation; (2) disease course (e.g., acute vs recurrent acute vs chronic); (3) presence of a causative infection (e.g., syphilis or Lyme disease); (4) presence of an associated systemic disorder (e.g., sarcoidosis or a spondyloarthropathy); and (5) presence of a morphologic syndrome (e.g., birdshot chorioretinitis [BSCR]). All of these characteristics help to guide therapy and may provide information on prognosis. For example, a chronic noninfectious posterior uveitis typically requires chronic systemic therapy, whereas a noninfectious, recurrent, acute anterior uveitis typically requires topical corticosteroid therapy of the acute attacks. Proper identification of a morphologic syndrome also guides therapy. For example, some multifocal choroidopathies, such as acute posterior multifocal placoid pigment epitheliopathy, are self-limited and spontaneously remitting with a good prognosis, whereas others, such as BSCR, are chronic, progressive diseases, requiring immunosuppressive drug therapy to maintain vision. In this process, proper, selective use of laboratory testing is critical. Testing for syphilis is carried out routinely because: (1) syphilis can produce any type of uveitis, such that morphologic features are inadequate to diagnose it, and (2) the treatment is radically different (2 weeks of intravenous antibiotics vs corticosteroids and immunosuppression). Identification of a systemic disease may be important for the patient’s systemic health (e.g., interstitial lung disease in some patients with sarcoid uveitis, which will require systemic corticosteroid therapy) or may dictate therapy (e.g., a systemic necrotizing vasculitis, such as Wegener granulomatosis, in a patient with scleritis).
HLA typing has been important in understanding the pathogenesis of several rheumatic disorders, particularly the spondyloarthropathies. In selected situations, HLA typing may be useful clinically, such as testing for HLA-B27 in the identification of a patient with what was previously termed incomplete Reiter syndrome and is now known as reactive arthritis. HLA-B27 typing also may have usefulness in the evaluation of a patient with uveitis. Patients with HLA-B27–associated uveitis typically have a recurrent, acute, unilateral or unilateral alternating, anterior uveitis. Indeed, a patient with recurrent, acute, unilateral, alternating anterior uveitis is nearly 80% likely to be HLA-B27 positive. In a patient with acute anterior uveitis, HLA-B27 testing has modest prognostic value, because patients who are HLA-B27 positive are more likely to have recurrent disease, although the disease may be morphologically indistinguishable. However, the real value of HLA-B27 testing comes when a positive value prompts the clinician to search for a previously undiagnosed systemic disease that may have permanent, debilitating effects. Among patients with HLA-B27–associated uveitis, two thirds to three quarters will have an associated spondyloarthropathy, of whom approximately one half will not have been diagnosed or will have been misdiagnosed (hence approximately one third of all patients with HLA-B27–associated anterior uveitis). Therefore, HLA-B27 testing’s primary benefit in this situation is in improving management of the patient’s systemic health.
However, as a diagnostic test, the value of HLA typing is limited. An important concept in understanding the value of laboratory testing is the positive predictive value of a test ( Table 1 ). Positive predictive value is the likelihood that an individual with a positive test result will have the disease in question. It is influenced by the frequency of the disease in the group of patients being tested and the characteristics (sensitivity and specificity) of the test. The problem with using HLA typing diagnostically is illustrated in the case of BSCR. BSCR is associated with HLA-A29, and 85% to 95% of patients with BSCR possess HLA-A29, with an overall estimate of 90%. HLA-A29 is present in approximately 7.7% of the general population, and BSCR has among the highest relative odds of any HLA-associated disease. BSCR accounts for approximately 7% of posterior uveitis, and some clinicians advocate screening patients with posterior uveitis with HLA-A29 to diagnose BSCR. The problem with this approach is illustrated by Table 2 , which shows the results of this strategy applied to 1000 patients with posterior uveitis. Despite the tight association of BSCR with HLA-A29, the positive predictive value of HLA-A29 is less than 50%, which means that diagnosing BSCR based on HLA-A29 results is wrong more often than it is right. If all uveitis patients were screened (as opposed to just posterior uveitis), the positive predictive value would be even lower, because the prevalence of BSCR is even lower among all patients with uveitis than it is among those with posterior uveitis. One could improve the positive predictive value of HLA-A29 testing by using it only in those situations where the patient had a multifocal choroiditis with yellow-orange ovoid spots, 1+ vitreous cells, no anterior chamber cells, and no evidence for syphilis, Lyme disease, or sarcoid. However, regardless of the HLA-A29 result, that patient would be diagnosed with BSCR (5% to 15% of patients with BSCR are HLA-A29 negative), so the test adds little to the diagnosis in this situation, either.
|Sensitivity||True positives/(true positives + false negatives)||Proportion of those with disease with positive test results|
|Specificity||True negatives/(false positives + true negatives)||Proportion of those without disease who will have negative test results|
|Positive predictive value||True positives/(true positives + false positives)||Likelihood of a patient with positive results of having the disease|
|Negative predictive value||True negatives/(true negatives + false negatives)||Likelihood of a patient with negative results not having the disease|
|Disease Status||HLA Typing Results||Totals|
|Positive predictive value||0.47|
|Negative predictive value||0.99|
The situation typically is more problematic for other uveitides, where the association of the HLA type with the disease is less strong, either because a lower percentage of patients have the HLA type or because the population frequency of the HLA type is higher. Table 3 lists the positive predictive value for several uveitic syndromes where HLA associations have been reported. Values for the frequency of the specific antigens among patients and among the general population were taken from the literature on HLA typing in these diseases and averaged. The median values for disease prevalence among patients with a given anatomic type of uveitis in this table were taken from 4 large, retrospective studies of patients with uveitis. Calculations of positive predictive values and odds ratios were based on screening all patients with uveitis affecting the anatomic segment in question for the antigen in question.