Article Text
Abstract
Background/aims The diagnostic properties of conventional diagnostic tests (ACE and chest radiography) for sarcoidosis-associated uveitis are not ideal. The diagnostic value of lymphopaenia for sarcoidosis-associated uveitis is investigated.
Methods A retrospective study of 191 consecutive patients with a first uveitis episode visiting the ophthalmology department (Erasmus Medical Center, Rotterdam, The Netherlands). Receiver operating characteristics (ROC) analysis was performed and compared with known ROC values from literature of conventional diagnostic tests for sarcoidosis-associated uveitis. An ideal cut-off was determined for lymphopaenia by calculation of the highest Youden index.
Results Out of all patients with first uveitis attack, 32/191 or 17% were subsequently diagnosed with biopsy-proven or radiological diagnosis of sarcoidosis. Lymphopaenia (<1.5×109/L) was significantly more often observed in patients with sarcoidosis-associated uveitis compared with patients with non-sarcoidosis-associated uveitis (p<0.05). The sensitivity and specificity of lymphopaenia was 75 % and 77 %, respectively. The optimal cut-off for lymphopaenia for diagnosing sarcoidosis-associated uveitis was 1.47 ×109/L. Lymphopaenia resulted in a 12.0 (95% CI 4.7 to 30.5 fold risk for having sarcoidosis, corrected for sex, race and age at onset of uveitis in patients with a first uveitis attack.
Conclusion Lymphopaenia is a non-invasive and useful marker for diagnosing sarcoidosis-associated uveitis.
- inflammation
- immunology
- epidemiology
Statistics from Altmetric.com
Introduction
Ocular involvement is frequently observed in sarcoidosis, usually manifesting as uveitis.1 Diagnosis of sarcoidosis-associated uveitis can be challenging, since ocular histology (the golden standard for the diagnosis of sarcoidosis) is difficult and uveitis may precede extraocular manifestations of sarcoidosis.2 Non-invasive diagnostic tests for sarcoidosis are therefore attractive in patients presenting with uveitis. Chest X-ray and serum biomarkers (ACE and lysozyme) are regarded diagnostic and classifying tests according to the International Workshop On Ocular Sarcoidosis (IWOS) criteria.3 However, these tests have limited positive predictive values (PPV).3
In sarcoidosis, T-lymphocytes are activated and skew from the peripheral blood to the affected tissue, resulting in a relative (T cell) lymphopaenia.4–6 Soluble interleukin 2 receptor (sIL-2R) reflects activation of T-lymphocytes . The diagnostic value of sIL-2R and ACE in the diagnosis of sarcoidosis-associated uveitis seem similar in earlier investigations.7 It has been suggested that the diagnostic criteria of sarcoidosis-associated uveitis may be modified by inclusion of lymphopaenia.8
Herein, we study the value of lymphopaenia as a diagnostic biomarker for sarcoidosis-associated uveitis in a therapy-naive population with a first episode of uveitis.
Material and methods
The ‘Strengthening the Reporting of Observational studies in Epidemiology’ guidelines were used for reporting this observational study.9 Retrospective use of laboratory investigations adheres to the tenets of the Declaration of Helsinki and the Erasmus Medical Center medical ethical committee approved the biobanking protocol and associated procedures.
Study population
We performed a study of 191 patients with a new onset of uveitis visiting the ophthalmology department at the Erasmus University Medical Center, Rotterdam, the Netherlands, from January 2011 to July 2017. All patients presenting with a first episode of uveitis and available lymphocyte counts within 1 month after the onset of uveitis were included. Data were reviewed retrospectively between January and September 2017. Patients with a known cause of lymphopaenia were excluded (online supplementary table 1).
Supplemental material
Originally, 244 patients with a first episode of uveitis were identified. Fifty-three patients with a known cause for lymphopaenia were excluded (immunosuppressive medication: N=17; infectious disease: N=21; systemic disease: N=5; immunosuppressive medication and systemic disease: N=7; immunosuppressive medication and infection: N=3).
The remaining 191 patients were included in the present study.
Definition of diagnostic categories
The aetiological cause of uveitis was determined after the initial diagnostic work-up in our centre. The diagnosis of sarcoidosis was based on the IWOS criteria (only definitive and presumed ocular sarcoidosis patients; ie, biopsy or radiological confirmations were categorised as sarcoidosis).3 Controls were the remainder of (non-sarcoidosis) patients with first attack of uveitis. Other diagnoses were established according to current international criteria.10–16 A definite diagnosis of tuberculosis-associated uveitis was based on a positive microbiology test anywhere in the body without other explanation of uveitis.
Assessment of variables
Demographic data, the onset of uveitis, laterality, location of the uveitis and the final diagnosis of uveitis were noted. The uveitis onset was defined as the date on which an ophthalmologist first documented uveitis.
A multiple linear regression was performed to identify a possible relationship between absolute lymphocyte counts and the duration of uveitis (the duration between moment of uveitis onset until the day of blood sampling). Corrected for sex and race, there was no association between the lymphocyte counts and uveitis duration (≤1 month, p=0.126). Therefore, only patients that had lymphocytes recorded within 1 month after or before the diagnosis of uveitis were included. For this study, the general cut-off for lymphopaenia was used (<1.5×109/L).
Statistical analysis
The characteristics of patients were summarised using descriptive statistics, such as means and percentages. Non-parametric tests were used to compare characteristics between the groups.
The sensitivity and specificity of lymphopaenia as well as the C-statistic (the area under the receiver operating characteristic (ROC) curve; a measure of test performance) for the diagnosis of sarcoidosis was calculated. The ROC and Youden’s index (sensitivity+specifity−1) was used to summarise test performance.17 The optimal cut-off for lymphopaenia in the diagnosis of sarcoidosis-associated uveitis was calculated by maximising the Youden’s index. Binary logistic regression was used to measure the significance of association between lymphopaenia and sarcoidosis-associated uveitis, corrected for gender, race, age at onset of uveitis, use of any immunosuppressive treatment, immunosuppressive disease or immunosuppressive infection. The statistical analysis was done using Excel, IBM SPSS statistics V.21.0.0 for Windows and R, using the package pROC.
Results
The characteristics of our study population are depicted in table 1. The mean age of onset of uveitis was 46.8±18.0 years and 120/191; 63% of patients were female and 128/191; 67% were of Caucasian descent. The median interval between onset of uveitis and determination of lymphocyte count was 4.0 days. All patients had their first episode of uveitis without a known aetiology of their uveitis at moment of blood sampling.
Diagnoses of uveitis were performed after all relevant tests were performed (table 1).
Sarcoidosis-associated uveitis was diagnosed in 32 patients (17%), and the mean onset of uveitis in this group was 45.5 ±17.3 years and showed slight preponderance of female gender (21/32; 66%) and Caucasian ancestry (19/32; 59%). These characteristics did not differ compared with patients with non-sarcoidosis-associated uveitis patients (N=159; p>0.05).
Panuveitis and bilateral involvement were more common in sarcoidosis-associated uveitis (24/32;75% and 26/32; 81%, respectively) compared with patients with non-sarcoidosis-associated uveitis (70/159;44% and 76/159;48%), p=0.002 and p=0.001, respectively.
Optimal cut-off lymphocyte counts for the diagnosis of sarcoidosis-associated uveitis
The cut-off that corresponded with the highest Youden index (0.54) was (1.47×109/L). The associated sensitivity and specificity was 75% and 79%, respectively.
Lymphopaenia
Lymphopaenia was present in 61 out of 191 patients (32%) out of which 24/61 (39%) had sarcoidosis (21 with biopsy proven and 3 with presumed sarcoidosis). The remaining patients with lymphopaenia included HLA-B27-associated uveitis, Behçet’s disease, multiple sclerosis, granulomatosis with polyangiitis and Vogt-Koyanagi-Harada syndrome (table 2). Lymphopaenia in infectious uveitis was seen in 5/61; 8% (including endogenous endolphthalmitis, active tuberculosis, toxoplasmosis and herpes simplex-associated uveitis). The proportion of lymphopaenia in patients with established cause of uveitis and uveitis of unknown cause did not differ, p=0.52.
Sarcoidosis-associated uveitis
Lymphopaenia was significantly (p=0.0001) more observed in patients with sarcoidosis-associated uveitis than in patients with non-sarcoidosis-associated uveitis (24/32; 75% vs 37/159; 23%, respectively). Furthermore, the mean lymphocyte count in patients with sarcoidosis-associated uveitis was significantly lower than in patients with non-sarcoidosis-associated uveitis (1.3±0.5×109/L and 2.0±0.8×109/L; p=0.0001, respectively).
Sarcoidosis-associated uveitis and lymphopaenia
Sarcoidosis-associated uveitis patients with lymphopaenia (N=24) were mostly female (14/24;58%) of Caucasian origin (14/24; 58%) with panuveitis (18/24;75%) and bilateral involvement (19/24; 79%). Sarcoidosis-associated uveitis patients with and without lymphopaenia were similar in location of uveitis and prevalence of bilateral involvement (panuveitis in 18/24;75% vs 6/8; 75%, p=1.0 and 19/24;79 vs 7/8;88, p=1.0, respectively).
Corrected for sex, race and age at onset of uveitis, the occurrence of lymphopaenia increased the risk to find sarcoidosis with a factor 12.0 (95% CI 4.7 to 30.5 fold risk for having sarcoidosis, corrected for sex, race and age at onset of uveitis (table 3).
Test characteristics of lymphopaenia in sarcoidosis-associated uveitis
Table 4 depicts the various test characteristics of lymphopaenia (<1.5x109/L). Lymphopaenia in the diagnosis of sarcoidosis-associated uveitis has a sensitivitity of 75% (95% CI; 60.0-90.0) and a specificity of 76% (95% CI: 70.2-83.3).This corresponds to a Youden’s index of 0.517 and C-statistic of 0.792, 0.710–0.874. Table 5 summarises commonly used diagnostic tests for the detection of sarcoidosis-associated uveitis, including previous literature.
Sensitivity analysis
A sensitivity analysis was performed, adding patients with known causes of lymphopaenia in their history (in total N=244). When analysing this group, the PPV became lower (28%), and the sensitivity and specificity were different (72% and 68%, respectively) when compared with our population without obvious causes for lymphopaenia (N=191).
Discussion
This retrospective study demonstrates that lymphopaenia was strongly associated with the diagnosis of sarcoidosis in patients with a first episode of uveitis. The cut-off for lymphopaenia with most ideal test characteristics was 1.47×109 L, close to the general cut-off used in this study (1.5×109 L), which also might be used.
Peripheral T-lymphocytes are decreased in sarcoidosis and may be an appropriate screening tool in patients with uveitis.6 8 18 Therefore, lymphocyte counts have recently been proposed by Jones et al to be added to the diagnostic IWOS criteria for sarcoidosis-associated uveitis (with cut-off <1.0×109/L and corresponding OR of 5.7). The lymphocyte values found in Jones et al’s8 study, however, cannot be implemented in patients with a new onset of uveitis, because patients with a second or further episode of uveitis were also included. Furthermore, not all patients have been diagnosed according to the IWOS criteria (patients with elevated serum markers, but without biopsy or radiological confirmations were also labelled as sarcoidosis-associated uveitis). The optimal cut-off, identified in this study (1.47×109/L), is close to the general cut-off for lymphopaenia used in this study (1.5×109/L) but differs from the proposed cut-off by Jones et al (<1.0×109/L). Since the optimal cut-off is very similar to the general cut-off for lymphopaenia, the general cut-off can be used to diagnose sarcoidosis (with comparable test characteristics).
Predictive values indicate the chance of disease in a patient with a positive test result (PPV) or the chance that the patient does not have the disease when the test is negative (NPV). Since conventional diagnostic tests have low PPV values in diagnosing sarcoidosis-associated uveitis, a search for a more sensitive and specific diagnostic test is warranted.7 8 Lymphopaenia has a higher PPV than for ACE and sIL-2R but lower when compared with chest X-ray (table 5).7 19 The NPV (ruling out sarcoidosis when a test is negative) of lymphopaenia is comparable to that of chest X-ray and higher than the NPV of ACE in previous studies (table 5). Absence of lymphopaenia therefore performs better than normal ACE levels in ruling out sarcoidosis in the uveitis population and its performance is comparable to chest X-ray, but is less invasive and less expensive compared with the latter. The diagnostic value of the combinations of various tests including lymphopaenia, sIL-2R and chest X-ray would be interesting to investigate, since this is scarcely touched on in the current literature, but are beyond the scope of this study.
Other predictive factors for sarcoidosis in our uveitis population were panuveitis and bilateral involvement. Preponderance of female gender in ocular sarcoidosis has already been described, an aspect we did not identify in our population as a risk factor.1 20 21 Our study did not contain many Asian patients, a known predictive factor for ocular involvement in sarcoidosis.22–27 Therefore, in this study, this association could not be established. Panuveitis and bilateral involvement were more common in patients with sarcoidosis-associated uveitis compared with patients with non-sarcoidosis-associated uveitis. Future research should elaborate on the value of combining the epidemiological features together with laboratory tests and imaging in differentiating sarcoidosis-associated uveitis from other causes of uveitis. Since the number of patients was limited, the detailed assessment of sarcoidosis-associated uveitis for patients with and without lymphopaenia was not performed.
The differential diagnosis of uveitis is diverse: from infectious aetiologies to autoinflammatory/immune diseases.28 29 In the present study, most patients with lymphopaenia have either sarcoidosis or idiopathic uveitis (45/61;74%). However, the association of lymphopaenia with other aetiological groups cannot be entirely excluded, due to the limited number of patients in specific diagnostic categories of uveitis and might geographically vary. The predictive value of lymphopaenia depends on the prevalence of sarcoidosis in the uveitis population. The use of lymphocytes as a predictor of sarcoidosis-associated uveitis might therefore be limited in settings, where other diseases are more prevalent, such as tuberculosis endemic countries. Our university centre is a sarcoidosis centre for the region of South Holland, thus probably containing a higher proportion of sarcoidosis patients when compared with other university centres.
There are multiple causes for lymphopaenia, such as receiving immunosuppressive medication or presence of diseases that dysregulate the immune system (such as HIV). Patients with a known cause of lymphopaenia have been excluded from this study (inferior test characteristics were observed in our sensitivity analysis including patient with known cause of lymphopaenia). Therefore, in patients with a known reason for lymphopaenia, a determination of a lymphocyte count is probably not helpful for screening purposes.
Confirmation bias, which is introduced if the outcome (sarcoidosis) precedes the assessment of the variable (lymphocyte count) was minimised in our study since the lymphocyte counts were always measured before the diagnosis was made.
In conclusion, lymphopaenia appears an useful diagnostic biomarker for the diagnosis of sarcoidosis in patients experiencing their first uveitis attack. Further avenues of research should concentrate on the development of other non-invasive tests for the diagnosis of ocular sarcoidosis and selecting the optimal combination of available tests.
Acknowledgments
This study was in part financially supported by AbbVie, the Netherlands. Financial disclosures: FG-H received grant support from AbbVie, Netherlands, directed to the Department of Ophthalmology, Erasmus Medical Center, Erasmus University, Rotterdam, the Netherlands. JvL received
lecture fees from AbbVie. The remaining authors have no financial disclosures.
References
Footnotes
FG-H and LE are joint first authors.
Contributors Design of the study (FG-H and AR), conducting the study, data collection, data management (FG), data analysis, interpretation of the data and preparation, review of the manuscript (FG-H, AR and JvL). FG-H and LE contributed equally to this manuscript and should be considered first authors.
Funding This study was in part financially supported by AbbVie, The Netherlands.
Disclaimer The sponsor had no role in study design, in the collection, analysis and interpretation of data nor in the writing of the report and in the decision to submit the article for publication.
Competing interests None declared.
Patient consent Not required.
Ethics approval The Erasmus Medical Centrel Medical Ethics Committee reviewed and approved the biobanking protocol and associated procedures this study.
Provenance and peer review Not commissioned; externally peer reviewed.
Data sharing statement Data are available on request from the corresponding author: f.groen@erasmusmc.nl.