We conducted a retrospective review of medical records of patients seen at the ophthalmology clinic at Harborview Medical Center at the University of Washington between 2010 and 2017. Individuals were included if ophthalmology assessment raised a concern for TBU. Specifically, criteria included individuals with clinical signs of uveitis; evidence of tuberculous infection by either tuberculin skin test (TST), QFT-G, or ocular sampling for MTB-PCR; and exclusion of non-tuberculous infectious and non-infectious etiologies by screening tests. Non-tuberculous etiologies included syphilis (syphilis IgG or T. pallidum particle agglutination assay (TPPA)), rheumatoid arthritis (rheumatoid factor (RF)), lupus (antinuclear antibodies (ANA)), sarcoidosis (angiotensin-converting enzyme (ACE) and chest radiography), and toxoplasmosis (toxoplasma IgM and IgG). All patients underwent independent clinical evaluation by an infectious disease specialist following referral from ophthalmology prior to initiation of ATT.
Data were abstracted from the De-identified Clinical Data Repository (DCDR) which contains information collected from the University of Washington clinical systems between 2010 and 2017. A Microsoft Amalga query was run on the DCDR database using the keyword search term “TB of the eye.” No additional cases of interest were captured by further searches which included tuberculous uveitis, tuberculous chorioretinitis, tuberculous conjunctivitis, tuberculous episcleritis, tuberculous interstitial keratitis, tuberculous iridocyclitis, and tuberculous keratitis. Several additional individuals receiving ongoing clinical care were also included in the study.
Abstraction of demographics (age, gender, country of origin), TB risk factors, prior active or latent TB infection (LTBI), medical co-morbidities, time from symptom onset to diagnosis, and site of involvement was performed using a standardized data tool. Site of inflammation was categorized by a fellowship-trained uveitis specialist as anterior uveitis, intermediate uveitis, posterior uveitis, or panuveitis according to Standardization of Uveitis Nomenclature (SUN) guidelines . Unilateral or bilateral disease was also ascertained. Information collected on diagnostic evaluation included TST, QFT-G, ocular sampling for MTB-PCR, sputum sampling for acid-fast bacilli (AFB) smear and culture, chest radiograph or computed tomography (CT), and human immunodeficiency virus (HIV) test.
Information on recommended treatment was also documented and included anti-tuberculous drugs, systemic steroids, topical steroids, and any drug-related toxicities. Completed therapy was categorized by infectious disease-trained clinicians according to the following scheme: (1) Adequate active ATT (adequately treated for active TB): defined according to standard treatment for drug-susceptible pulmonary TB  and included individuals in whom a fluoroquinolone was substituted for ethambutol during the intensive phase. (2) Adequate alternative active ATT: this included patients who were either not started on a four-drug regimen or completed less than 2 months of intensive phase therapy but received at least 6 months of daily isoniazid- or rifampin-containing combined therapy. These treatment courses are sufficient for different types of paucibacillary disease such as culture negative pulmonary TB which is often treated with a shortened 4-month regimen per American Thoracic Society guidelines . (3) Inadequate or no active ATT (inadequately treated for active TB). Inadequate active ATT did not meet any of the definitions above and included treatment for LTBI.
Finally, inflammation was reassessed after recommendation of ATT and improvement was determined using SUN criteria . Improvement was defined for patients with anterior uveitis as decrease of anterior chamber (AC) cell by 2 grades or a final score of < 0.5+ with topical corticosteroid eye drops administered no more than two times daily (low-dose topical therapy). For patients with anterior and intermediate uveitis, a decrease in vitreous haze by two grades or a final score of < 0.5+ with low dose topical corticosteroid therapy or low-dose oral corticosteroid (< 10 mg daily) was considered improvement. For patients with posterior uveitis maintenance of quiescence of any active choroidal, retinal, or vascular lesions noted at presentation with low-dose oral corticosteroid (< 10 mg daily) therapy after an initial pulse and taper was considered improvement. Patients that required topical corticosteroid therapy > 2 drops daily, or corticosteroid sparing therapy with or without oral corticosteroids were designated not improved. Improvement was ascertained within a 1-month window of the date 6 months after cessation of ATT when possible. In the absence of long-term follow-up, the clinical exam at the last recorded follow-up was used. If patients declined ATT therapy, the study endpoint was determined at the office visit falling between 6 and 12 months after initiation of anti-inflammatory therapy. If improvement in inflammation was observed before initiation of ATT, treatment was reclassified as no ATT group for final analysis.
We used Fisher’s exact test to compare clinical improvement according to primary predictors of interest, adequate ATT and site of inflammation. To examine possible confounding effects, we used Fisher’s exact test to compare associations between clinical improvement and age (stratified as < 60 and ≥ 60 years), sex, birth in a high-burden country, duration of symptoms (stratified as ≤ 1 and > 1 year), and underlying immunosuppression. The study was reviewed and approved by the University of Washington Institutional Review Board.