Erythema induratum of Bazin and TB-AOI represented uncommon referrals to our provincial TB program during the period of study. Most cases were among immigrants from TB high-burden countries, and the treatments provided were variable. In addition to novel epidemiological and long-term outcome data on these conditions in a TB low-incidence setting, our study highlights key programmatic challenges in the diagnosis and management of these conditions. While 76% of EIB cases had improvement or resolution of symptoms at last contact, similar responses were reported by less than half of TB-AOI cases. These findings underscore a need for improved diagnostics and clinical decision tools to enhance treatment candidate selection, particularly for suspected TB-AOI cases. Furthermore, the variable treatments used for these conditions in our program significantly limits effectiveness evaluation and illustrates a need for treatment standardization if practice improvement is to be achieved moving forward. Small sample size and treatment heterogeneity in this case series means that no conclusions about treatment effectiveness can be made.
Compared with previously published case series, the demographic characteristics of EIB and TB-AOI cases in our series are similar; females, 30–50 years of age, born or living in TB endemic settings (> 20 cases per 100,000 population per year) predominate across case series [
6,
12,
14,
15,
17,
20‐
22]. In prior series, ATT associated clinical improvement of EIB skin lesions has ranged from 69 to 86% [
20,
21], while visual acuity improvement and/or ocular inflammation reduction in TB-AOI has ranged from 24 to 85% [
6,
12,
15,
17,
22]. Treatment outcomes may be influenced by delays in diagnosis, particularly in TB-AOI where prompt initiation of ATT may improve visual prognosis [
22]. Although our treatment outcomes were not dissimilar from previous case series, the large range, particularly for TB-AOI, and substantial residual non-response rates, illustrate the need for improvements in the management of these conditions.
Improved treatment candidate selection
The non-specific nature of EIB and TB-AOI clinical syndromes and a frequent lack of confirmatory diagnostics make treatment candidate selection for these conditions particularly challenging. Whereas tissue sampling for histopathology and microbiology evaluation may aid diagnostic differentiation in cases of suspected EIB [
16], the low diagnostic yield and added risk of morbidity with ocular tissue sampling limits its utility for TB-AOI [
11]. The integration of molecular diagnostics, criteria based histopathologic/ophthalmologic assessments, and novel imaging techniques have the potential to increase diagnostic specificity for these conditions.
Tuberculosis molecular diagnostics (e.g. nested polymerase chain reaction (PCR)) appear to have an emerging role in EIB and TB-AOI diagnosis. The reported sensitivity of ocular fluid PCR for TB has varied considerably across cases series in TB high-incidence settings, ranging from 31 to 47% in a pair of studies in India and Singapore [
23,
24] to as high as 77% in a more recent Indian series using a novel multi-targeted PCR [
25]. False positive rates of 5 to 9% have been reported in ocular fluid sample PCR from disease control groups [
23,
26]. Tissue-based PCR results in EIB have demonstrated equally variable sensitivity with TB isolation ranging from 14 to 77% across cases series [
7]. A single case-control series in China found that 32% of panniculitis and 5% of control biopsies demonstrated TB on PCR [
27]. Over the period of our study, PCR for TB was not routinely done on tissue biopsies specimens that did not show evidence of acid-fast bacilli on pathology, as is common in EIB and was the case in all of our study samples. Continued limited availability, lack of standardization, and low sensitivity of these tests remain as obstacles to their wide spread use, particularly in low-incidence settings were validation studies are lacking [
7,
28].
In the absence of confirmatory molecular diagnostics, histopathology in EIB may significantly aid treatment candidate selection. Lobular panniculitis with vasculitis - some authors consider veins and venules to be predominantly involved [
8,
9] – and progression to tissue necrosis and granuloma formation with disease evolution are characteristic findings in EIB [
7]. That relatively few EIB biopsies in our series demonstrated vasculitis, 48% compared with 90% in a large series by Segura et al. [
9], may in part reflect biopsy technique and timing. In our series, punch biopsies, which often contain less subcutaneous fat [
29], were most commonly obtained. Excisional biopsies with adequate subcutaneous fat - the site most commonly demonstrating vasculitis changes [
9] - may optimize diagnostic yield in EIB [
7]. Further, biopsies in our series may have been predominantly from later stage lesions – participants reported lesions being present a median of 2 years prior to diagnosis - which may contain less extensive vasculitis changes compared with earlier stage lesions [
7]. To maximize diagnostic yield, excisional biopsies of early stage skin lesions should be preferentially utilized in cases of suspected EIB.
In the case of TB-AOI, low culture yield of ocular aspirates and potential risks associated with ocular tissue biopsy place diagnostic emphasis on ophthalmological examinations. Findings of choroidal granulomas, occlusive retinal vasculitis, and multifocal serpiginoid choroiditis have been proposed as specific to TB-AOI in high TB incidence settings [
13]; however, the transferability of these findings to low incidence setting has been questioned [
28]. Serial ophthalmologic examinations also may serve a diagnostic function following initiation of ATT therapy. The authors of two larger case series reported that most TB-AOI cases demonstrated a reduction of ocular inflammation within 2–6 weeks of starting ATT and quiescence by six months [
15,
30]. More recently, circumventing the low yield and risks of ocular tissue biopsy, several researchers have proposed a role for positron emission tomography/CT (
18F-FDG-PET/CT) to identify extra-ocular sites of infection for tissue sampling [
31]. Evaluating a series of 20 TB-AOI cases, Doycheva et al. identified FDG-uptake in mediastinal and hilar lymph nodes in 45% of cases and 40% of those who underwent biopsy cultured TB [
32]. The Collaborative Ocular Tuberculosis Study (COTS)-1 group is working to address the lack of evidence for the diagnostic approaches to TB-AOI and highlight the importance of multipronged approach in a recent study evaluating treatment outcomes [
33].
Therapy standardization
While variations in case definitions (e.g. inclusion of cases with concurrent active TB disease, differing or absent pathologic criteria used for EIB cases) and outcome measurements (e.g. timing of post treatment follow-up, use of serial ophthalmologic exams for TB-AOI) limit cross-series ATT comparisons. Acknowledging these limitations, a trend of positive clinical response and relapse reduction is apparent with extended duration triple-drug ATT for both EIB and TB-AOI.
Cases-series evaluating EIB treatments have shown higher non-response and relapse rates among cases receiving mono- or dual- drug ATT compared with those receiving triple-drug ATT for at least 6 months [
20,
21]. Similar increased effectiveness with triple-drug ATT has been shown for TB-AOI with several studies supporting extension of treatment to 9 or more months [
17,
22]. A case-control study by Ang et al. in Singapore demonstrated improved one-year post treatment visual acuity and relapse reduction among TB-AOI cases that received greater than 9 months of ATT [
15]. Similarly, a retrospective review of TB-AOI by Agrawal et al. in the United Kingdom found TB-AOI patients who received more than 9 months of triple-ATT had a lower likelihood of treatment failure [
12]. While a significant difference in overall disease outcomes between ATT regimens was not found in our case series, small sample size and a high rate of loss to follow-up, particularly among TB-AOI cases, restricted our evaluation. In keeping with findings of prior EIB case series [
16,
20], the two cases with post treatment relapse in our series had not received triple-drug ATT regimens (Additional file
2).
Whether EIB and TB-AOI represent immune-mediated response to LTBI or are manifestations of subclinical TB disease continues to be an area of scientific debate with implications for the selection of appropriate treatment [
7,
10]. Despite this uncertainty, informed by the observational study results outlined above, a number of authors favor treating these conditions similar to active TB disease using standard drug-susceptible TB ATT regimens recommended by the Centers for Disease Control and Prevention [
34]: isoniazid, rifampicin, ethambutol, and pyrazinamide for 2 months followed by isoniazid, rifampicin for at least 4 additional months for EIB [
5,
7,
16] with potential extension to 7 plus months for TB-AOI [
12]. Supporting such active TB disease therapy for EIB, a Japanese review of 66 cases over a 20 year period found that that a quarter of EIB cases also had active TB lymph node disease [
35]. In the case of TB-AOI, the high morbidity associated with delayed or incomplete treatment of potential ocular TB disease as well as concern about limited drug penetration to the eye, favor prolonged active TB disease ATT therapy [
15,
36]. Despite this, treatment remains highly variable, particularly in low incidence settings, with only 58% of suspected TB-AOI cases receiving standard drug-susceptible TB ATT in a recent US series [
6].
Limitations
The medical records used for the retrospective component of this study did not contain detailed ophthalmologic exam findings from baseline and follow-up. While the majority of TB-AOI cases were designated as ‘uveitis’ (90%), further information on location, extent and nature of ocular inflammation could not be reliably captured. Previous studies have shown variable ATT response between predominantly anterior, posterior or panuveitis cases of TB-AOI [
6,
12,
15]; analysis of treatment response by such subgrouping was not possible in our study. Furthermore, lack of serial ophthalmological exam findings to objectively quantify ATT response limits our TB-AOI clinical outcome measures. Our prospective outcome measures for both EIB and TB-AOI were based on unstandardized subjective patient report limiting generalization and potentially introducing recall bias. Use of standardized classifications systems, such as that proposed by the International Uveitis Study Group in the case of TB-AOI [
37], is needed to improve cross study comparison in this field. Finally, unreliable capture of corticosteroid prescription data in are TB program records made us unable to comment on what, if any, effect their adjunctive use may have had. It should be noted that adjunctive use of corticosteroids has not been supported in EIB and findings are conflicting for TB-AOI; case series level evidence suggesting benefit in early TB-AOI disease [
30,
38] needing to be balanced against findings of increase likelihood of relapse [
12].