Background
Systemic lupus erythematosus (SLE) is a systemic autoimmune disease that results from a disruption in immune tolerance to self-antigens, leading to inflammation of multiple organs [
1]. T cells play a major role in SLE pathogenesis, amplifying inflammation by the secretion of pro-inflammatory cytokines, helping B cells to generate autoantibodies, and maintaining the disease through the accumulation of autoreactive memory T cells [
2]. Interferon gamma (IFN-γ) is predominantly produced by T cells and natural killer cells [
3], and it plays a critical role in lupus [
4]. Previous studies have shown that IFN-γ mRNA expression is increased in peripheral blood mononuclear cells (PBMC) [
5], and that serum levels of IFN-γ are elevated in patients with SLE [
6,
7]. In addition, in vivo experiments in murine models of SLE have shown that elevations in IFN-γ mRNA levels are correlated with disease progression [
8,
9].
Although there is no clinical laboratory test available to examine serum IFN-γ, the IFN-γ-releasing assay (IGRA) is a diagnostic test used to detect
Mycobacterium tuberculosis (TB) infection by measuring the IFN-γ production by T cells stimulated with TB antigens [
10]. Since biologics such as tumour necrosis factor alpha (TNF-α) inhibitors make patients susceptible to reactivation of TB, screening of latent tuberculosis is recommended in the field of rheumatology [
11,
12]. IGRA is composed of three tubes (nil, mitogen, and TB antigen) for measuring IFN-γ. The nil tube is a negative control used to measure baseline IFN-γ production, whereas the mitogen tube is a positive control used to measure IFN-γ production with phytohemagglutinin (PHA). Therefore, IGRA results of negative and positive control tubes yield IFN-γ produced at baseline and by the activated T cells, respectively.
IFN-γ production by T cells has not been used to assess disease activity in SLE before; therefore, we evaluated ex vivo IFN-γ production following stimulation with PHA in patients with lupus by analysing the results of IGRA.
Methods
Patient selection and study design
We retrospectively analysed the IGRA results of SLE patients who had undergone IGRA at Severance Hospital between November 2009 and December 2016. Patients with a diagnosis of SLE according to the 1997 revised American College of Rheumatology classification criteria [
13] were included in the study. The exclusion criteria were as follows: (i) patients with concomitant autoimmune disease; (ii) patients with active infection on the date of the IGRA; (iii) patients with end-stage renal disease; and (iv) patients with absent complement 3 (C3), C4 or anti-double-stranded DNA (dsDNA) results on the date of the IGRA. Ultimately, 118 patients were included in this study. The flowchart for patient selection is shown in Additional file
1: Figure S1. Among the patients with active lupus, 13 patients had follow-up IGRA results available after the lupus became inactive. For comparison, IGRA results from patients with rheumatoid arthritis (RA) were retrospectively obtained for patients who had undergone IGRA before the administration of biologics such as TNF-α inhibitor. Data from healthy controls (
n = 173) were retrospectively obtained from individuals who had undergone IGRA for a regular health check-up at Severance Hospital. This study was approved by the Institutional Review Board of Severance Hospital (IRB approval number: 4-2016-1115) and conducted in accordance with the principles set forth in the Declaration of Helsinki. The requirement to obtain informed consent was waived because of the retrospective nature of the study.
Assessment of clinical and laboratory data
The clinical data collected included age, sex, disease duration, clinical manifestations, concurrent immunosuppressive agents, classification of new onset SLE, and the SLE Disease Activity Index-2000 (SLEDAI-2 K) [
14]. The disease duration was defined as the period from SLE diagnosis to the date of the initial IGRA, and patients were defined as having new-onset SLE when IGRA was performed within 1 month of the initial diagnosis of SLE. Glucocorticoid dosage was estimated by calculating the total glucocorticoid dosage that was administered 1 week prior to the IGRA, and was expressed in prednisolone equivalent dosage. Clinical manifestations of SLE included skin rash, photosensitivity, oral ulcers, arthritis, serositis, nephritis, and neurological, haematological and immunological disorders, as previously defined [
13]. Laboratory data included white blood cell counts; platelets; lymphocyte counts; erythrocyte sedimentation rate (ESR); levels of haemoglobin, C-reactive protein (CRP), blood urea nitrogen (BUN), creatinine (Cr), aspartate aminotransferase (AST), alanine aminotransferase (ALT), total bilirubin, albumin, C3, C4, and anti-dsDNA; and urine protein/creatinine ratio (urine P/Cr).
Definition of active SLE according to SLEDAI-2 K scores
During the testing period, the SLEDAI-2 K score of each patient was evaluated. Laboratory and clinical abnormalities that were not attributable to SLE were excluded when evaluating SLEDAI-2 K scores. Active SLE was defined as previously described by Franklyn et al. [
15]. Patients with SLEDAI-2 K scores ≥ 5 were defined as having active SLE, while patients with SLEDAI-2 K scores < 5 were defined as having inactive SLE. Poor hospitalisation outcome was defined as in-hospital mortality and/or intensive care unit admission.
Estimation of IFN-γ level assessed by IGRA
For each patient, IGRA was performed in whole blood samples using the QuantiFERON-TB Gold-In Tube test (QFT-GIT; Cellestis, QIAGEN, Germany) according to the manufacturer’s instructions. Briefly, 1 mL of blood was drawn directly into each of the QuantiFERON®-TB Gold blood collection tubes. The kit consists of three blood collection tubes: (i) nil tube (negative control: whole blood without antigens or mitogen); (ii) mitogen tube (positive control: whole blood with phytohemagglutinin); and (iii) TB antigen tube (whole blood with peptides of ESAT-6, CFP-10, and TB7.7 proteins simulating TB-specific antigens). The tubes were incubated overnight at 37 °C, and the concentrations of IFN-γ (IU/mL) were measured using an enzyme-linked immunosorbent assay (ELISA). An automated microplate processor (Evolis Twin Plus system; Bio-Rad Laboratories, Hercules, CA, USA) was used to analyse and calculate the results. Ex vivo IFN-γ production was estimated by calculating the difference in IFN-γ production between the mitogen tube and the nil tube (mitogen minus nil) in order to measure the ability to produce additional IFN-γ after PHA stimulation.
Statistical analysis
Data analysis was conducted using either GraphPad Prism version 5.0 (GraphPad Software, San Diego, CA, USA) or MedCalc statistical software version 16.2.0 (MedCalc Software bvba, Ostend, Belgium). Data were expressed as medians with inter-quartile ranges (IQR) or for categorical variables, as frequencies and percentages. Continuous variables were compared using Student’s t test, and categorical data were compared using the chi-square test or Fisher’s exact test as appropriate. Comparison of nil and ex vivo IFN-γ production in patients with paired IGRA results was performed using the Wilcoxon signed rank test. To compare poor hospitalisation outcome according to ex vivo IFN-γ production, we used Kaplan-Meier analysis and the log-rank test. Correlations between age or disease duration and IFN-γ production in the nil tube, IFN-γ production in the mitogen tube, or ex vivo IFN-γ production, and the correlation between ex vivo IFN-γ production and the SLEDAI-2 K scores were calculated using Pearson’s correlation analysis. Univariate and multivariate logistic regression analyses were performed with forward stepwise logistic regression analysis to compare laboratory variables in differentiating between active and inactive SLE. In multivariate analysis, only variables that were statistically significant in univariate analysis were included. The cut-off value of ex vivo IFN-γ production in discriminating active and inactive SLE and in predicting poor hospitalisation outcome was evaluated using receiver operator characteristic (ROC) curve analysis. In all statistical analyses, a two-tailed p value <0.05 was considered statistically significant.
Discussion
Although IFN-γ mainly mediates host defence against microbial invasion, it is known to have a pivotal role in SLE. The inhibition of IFN-γ has shown benefits in reducing disease activity in murine models of SLE [
16,
17] and a therapeutic monoclonal antibody against IFN-γ is being developed for the treatment of SLE [
18,
19]. In this study, we evaluated IFN-γ production in patients with SLE, using IGRA results obtained during the screening of latent TB prior to immunosuppressive treatment. Our data demonstrated that ex vivo production of IFN-γ is decreased in patients with active disease compared to patients with inactive disease. The results of our study are different from those of previous publications that have reported increased IFN-γ production or IFN-γ-related gene expression in SLE, because our study evaluated IFN-γ production after PHA stimulation in whole blood [
6,
7,
20].
However, our data are not contradictory to previous findings, in that baseline IFN-γ production (nil results) in patients with active SLE is increased compared to that in patients with inactive SLE, in patients with RA, and in healthy controls. While baseline IFN-γ production increased, ex vivo IFN-γ production after stimulation with PHA decreased in patients with active SLE. In a previous study by Hagiwara et al., ex vivo experiments using enzyme-linked immunospot (ELISPOT) assay demonstrated that IFN-γ-producing T cells were decreased in patients with active lupus [
21], which is similar to our finding. Our study used ELISA, which can measure the total amount of ex vivo IFN-γ production in whole blood, excluding baseline production. Since ELISA is known for its high sensitivity and wide dynamic range, measuring ex vivo IFN-γ production by ELISA can be a suitable biomarker for disease activity in lupus. Decreased ex vivo IFN-γ production could be associated with T cell exhaustion, which is a non-functional state that occurs under conditions of antigen persistence, such as those that arise during various infections and cancers [
22,
23]. Similar to infection and cancer, T cell exhaustion has also been described in SLE [
24]. Excessive auto-antigen exposure in active SLE may lead to T cell exhaustion. Alternatively, T cells may become unresponsive due to a negative feedback mechanism during periods of overwhelming inflammation. PD-1, an inhibitory T cell marker, is increased in SLE [
25]. Likewise, critically ill patients who are admitted to the intensive care unit have been shown to demonstrate a high proportion of indeterminate results on the IGRA due to unresponsiveness to mitogen stimulation [
26].
Our observations have raised an important issue regarding the limitations of the IGRA in SLE. Nearly half of all patients with active SLE had indeterminate results from the IGRA, while none of the patients with inactive SLE had an indeterminate IGRA result. The IGRA results are dependent on IFN-γ production by TB antigen-specific T cells. However, when IFN-γ production is hampered by defective T cells, the sensitivity of the IGRA may be reduced. Unreliable IGRA results in patients with T cell defects have been reported in HIV infection [
27]. Our study suggests that the interpretation of the IGRA requires caution in patients with active SLE.
More importantly, decreased ex vivo IFN-γ production correlates well with SLEDAI-2 K scores, and the poor hospitalisation outcome was more frequent in patients with ex vivo IFN-γ production ≤ 0.40 IU/mL. The findings of our study imply that monitoring ex vivo IFN-γ production could aid in assessing disease activity and predicting clinical outcome in SLE. So far, no test is available to evaluate T cell reactivity for SLE disease activity and prognosis, and our data provide the possibility of developing a diagnostic test to measure ex vivo IFN-γ production. Furthermore, we demonstrated that decreased production of IFN-γ in patients with active SLE recovers when SLE becomes inactive following immunosuppressive treatment. Similar to our findings for SLE, decreased IFN-γ production in patients with RA has been shown to recover after TNF-α inhibitor treatment [
28]. Decreased ex vivo IFN-γ production was likewise noted in our group of patients with active RA as compared with healthy controls. However, decreased ex vivo IFN-γ production was more pronounced in active SLE than in active RA in our study.
The strength of our study is that we included a large number of subjects who had undergone IGRA, an ex vivo method to estimate IFN-γ production in patients with SLE. However, the present study has several limitations. First, the clinical and laboratory data and IGRA results of patients were collected by reviewing the medical records. Since the IGRA is not routinely performed in patients with SLE, there could have been patient selection bias, resulting in the selection of patients with more severe SLE in our study. Therefore, our study findings should be validated in future prospective studies. Second, the effect of immunosuppressive treatment was not thoroughly controlled. It is possible that immunosuppressive treatment affects the decrease in ex vivo IFN-γ production. However, the proportion of patients with active SLE treated with glucocorticoids or immunosuppressive agents was lower than that of patients with inactive SLE, as most patients with SLE had undergone IGRA before initiating potent immunosuppressive treatment. Therefore, the effect of immunosuppressive treatment on active lupus may not be significant. Third, we did not evaluate CD4 T cell numbers. Although we demonstrated that ex vivo IFN-γ production ≤ 7.19 IU/mL was an independent predictor for discriminating active and inactive lupus regardless of leukopenia and lymphopenia, there is a possibility that the level of ex vivo IFN-γ production is associated with CD4 T cell numbers.
Conclusions
In conclusion, we have demonstrated that ex vivo IFN-γ production decreases in active SLE, which correlates with SLEDAI-2 K scores. In addition, the prognosis of patients with SLE with low ex vivo IFN-γ production was unfavourable. These findings suggest that ex vivo IFN-γ production might be a useful biomarker for monitoring disease activity in patients with SLE. Furthermore, special caution in the interpretation of results is required when IGRA is performed in patients with active lupus because of a large proportion of indeterminate results.