Raised Serum Markers of T Cell Activation and Exhaustion in Granulomatous-Lymphocytic Interstitial Lung Disease in Common Variable Immunodeficiency

Common variable immunodeficiency (CVID) is the most common symptomatic primary immunodeficiency in adults with a prevalence of 1:25,000–1:50,000 [1]. The patients are characterized by decreased levels of IgG and IgA with or without low IgM levels, with poor antibody response to vaccines (and/or low switched memory cells), and should have either increased susceptibility to infection, autoimmune manifestations, granulomatous disease, and/or unexplained polyclonal lymphoproliferation [2]. Up to 70% of CVID patients develop non-infectious inflammatory/immune-mediated complications, such as autoimmune cytopenia, interstitial lung disease (ILD), enteropathy, and liver disease [3]. CVID is a heterogeneous disorder where monogenic defects are found in ~ 10%, implying that several mechanisms of immune dysregulation can lead to these non-infectious complications [4]. It is unclear to which degree the individual clinical inflammatory complications have a shared pathogenesis, and various cells such as B cells, T cells, monocytes/macrophages, and type 3 innate lymphoid cells seem to be involved [5‐9].

Approximately 20–30% of CVID patients develop ILD, which is associated with increased morbidity and mortality [10]. Pulmonary histological abnormalities in these patients reflect diversity and overlap, and include patterns of pulmonary lymphoid hyperplasia such as follicular bronchiolitis, lymphocytic interstitial pneumonitis, and nodular lymphoid hyperplasia as well as granulomas, organizing pneumonia, and fibrosis [10‐13]. The term “granulomatous-lymphocytic interstitial lung disease” (GLILD) was proposed by Bates et al. to describe ILD in CVID with lymphocytic infiltrates and/or granulomas [10]. However, it is not clear if different entities of CVID-related ILD, also within the term GLILD, represent different pathological processes or variations within a disease spectrum. GLILD can often be distinguished radiologically from other ILDs by CT findings, and is characterized by reticulation, bronchial wall thickening, pulmonary nodules, and ground glass opacities [14‐16].

The pathogenesis of GLILD is poorly understood. Small case series of immunohistochemistry performed on lung biopsies in GLILD have shown lymphocytic infiltrates with the presence of T cells and variable findings of B cell follicles within the infiltrates [11, 17, 18]. Maglione et al. found increased serum levels of B-cell activating factor (BAFF) in CVID patients with a progressive course of ILD compared to stable GLILD, suggesting that a BAFF-mediated resistance to apoptosis drives pulmonary B cell hyperplasia within the lungs [6, 18]. Friedmann et al. found that bronchoalveolar lavage (BAL) fluid of patients with CVID-related ILD had lower numbers of regulatory T cells, higher numbers of T follicular helper (T_FH)-like memory cells skewed toward Th1 cells, and a larger fraction of B cells, mostly the inflammatory CD21_low B cell subtype, as compared to patients with sarcoidosis [19].

In idiopathic pulmonary fibrosis and connective tissue disease related ILD, biomarkers such as surfactant protein D (SP-D), Club Cell protein 16 (CC16), and matrix metalloproteinase 7 (MMP-7) have been shown to work as diagnostic and prognostic biomarkers [20‐23]. SP-D and CC16 appear to reflect airway epithelium injury, and MMP-7 is a marker of extracellular matrix (ECM) remodeling and fibrosis [24‐26]. To the best of our knowledge, these markers, and other markers of ECM remodeling and growth factors with relevance for fibrogenesis, have not previously been examined in GLILD.

CVID patients with an inflammatory phenotype often develop several non-infectious complications. In this study, we aimed to identify biomarkers that could distinguish GLILD, or reflect the pathogenesis of GLILD distinctively, in a population of CVID patients where all have non-infectious complications. Accordingly, we designed this study by dividing the cohort of CVID patients with non-infectious complications in those with and those without GLILD, and kept CVID with infections only in a separate group. We analyzed selected soluble serum markers of inflammation, immune homeostasis, pulmonary epithelium injury, fibrogenesis, and ECM remodeling (see Table 1 for overview) based on previous studies of ILD, CVID, and various inflammatory disorders, and aimed to distinguish the serum profile of GLILD from other non-infectious complications in CVID. This could highlight novel aspects of GLILD pathogenesis and identify new therapeutic approaches and clinically useful biomarkers.

In the present study, we compared CVID patients with GLILD to CVID patients with other non-infectious complications to explore whether GLILD patients are characterized by a distinct profile of serum markers reflecting inflammation, pulmonary epithelial cell injury, ECM remodeling, and/or fibrogenesis. Our main findings were (i) GLILD patients had higher levels of sCD25, sTIM-3, TNF, and IFN-γ than CVID patients with OC, pointing to T cell activation and exhaustion as potentially central in GLILD pathogenesis. (ii) GLILD patients had higher levels of SP-D, CC16, and MMP-7; biomarkers of pulmonary epithelium injury; and ECM remodeling, compared to OC. (iii) Levels of BAFF were not significantly higher in our GLILD cohort, but its receptor, sBCMA, had increased levels in GLILD patients compared to the OC group. (iv) The significant markers reflecting T cell activation, airway epithelium injury, and ECM remodeling were shown to be consistent over time.

An increasing number of studies have highlighted the importance of T cell activation in the pathogenesis of the non-infectious complications of CVID [7, 8, 27, 28]. We have previously shown that sCD25 is increased in CVID compared to healthy controls, and also that sCD25 is significantly higher in CVID patients with non-infectious complications compared to infection only [29]. However, only a few studies have explored the role of T cell activation in GLILD. Berbers et al. found increased expression of Ki67 and IFN-γ in effector/memory CD4⁺ T cells as a sign of activation in patients with non-infectious complications compared to those with infections only, and pointed to more extreme findings in the GLILD patient subgroup [7]. Van Stigt et al. found elevated levels of sCD25 in a cohort of 12 patients with CVID and granulomatous disease (of which ten had GLILD/granulomas in lungs) compared to CVID patients with infection only [30]. In the present study, we extend these findings in several ways. By examining a larger CVID population than in previous reports, we found elevated levels of sCD25 in GLILD patients not only as compared to IO patients, but also compared to CVID patients with other non-infectious complications. sCD25 is released from the cell membrane of T cells as a result of activation, and has shown to be a useful biomarker in diseases such as sarcoidosis and hemophagocytic lymphohistiocytosis [31]. Our findings suggest that sCD25 could also be a relevant biomarker in GLILD.

Persistent T cell activation is often accompanied by T cell exhaustion, a dysfunctional state of activation, rendering the cell refractory to further stimuli. This can be reflected by upregulation of various check-point inhibitors like TIM-3. The soluble form of TIM-3 is a result of ADAM10-mediated shedding of its membrane-bound form, mainly on IFN-γ-producing CD4⁺ and CD8⁺ T cells [32]. Although the functional consequences of this shedding is not clear, levels of sTIM-3 have been shown to be a reliable marker of T cell activation/exhaustion in various disorders like progressive HIV, hepatitis virus C infection, and COVID-19 [32‐34]. A few authors have suggested that T cell activation in CVID may be accompanied by exhaustion [8, 35], but the present study is the first to relate elevated sTIM-3 to CVID and GLILD. Although our finding of elevated sTIM-3 in GLILD is indicative of T cell exhaustion, parallel cellular studies should be performed in order to draw more firm conclusions. In relation to T cell homeostasis, we also found that fractions of T_reg that we previously have reported lower in CVID patients with an inflammatory phenotype were lower in the GLILD group compared to the OC group [36]. This points to a distinctive immune dysregulation in GLILD.

In line with the role of IFN-γ-producing CD4⁺ and CD8⁺ T cells as cellular sources of sTIM-3, we found elevated levels of the Th1-derived cytokine IFN-γ in GLILD patients compared to the other CVID groups. Although TNF have several cellular sources, Th1 cells are important contributors, and the raised TNF levels in GLILD patients further underscore the role of Th1 cells in GLILD. To summarize, higher levels of sCD25 and sTIM-3 combined with the Th1 response signature cytokines TNF and IFN- γ in GLILD are consistent findings suggesting that activated T cells play an important role in GLILD pathogenesis.

In the present study, we have showed that GLILD patients have higher levels of CC16 (an anti-inflammatory protein secreted by non-ciliated respiratory epithelium), SP-D (a neutralizing, anti-viral lipoprotein complex secreted by type II alveolar cells and bronchiolar epithelium [25, 37]), and MMP-7 (a protease expressed by epithelial cells, fibroblasts, and macrophages that activates microbicidal α-defensins [24]) than OC patients. In ILDs, elevated levels of CC16 and SP-D may reflect airway epithelial injury [21, 26]. Levels of CC16 are demonstrated to correlate with disease activity of ILD in systemic sclerosis, and SP-D has shown to be useful to diagnose and predict risk of exacerbation and mortality in idiopathic pulmonary fibrosis [38, 39]. SP-D in combination with MMP-7 can facilitate the identification of rheumatoid arthritis associated ILD at an early stage [40]. In GLILD, more than half of the patients have a progressive phenotype, with declining DLCO and FVC over time [16, 18]. This study suggests that CC16, SP-D, and MMP-7 could serve as biomarkers in GLILD even if their prognostic value is currently uncertain. In fact, SP-D came out as one of the strongest predictors of GLILD by stepwise regression. In line with previous reports, we found that GLILD patients had higher percentage of CD21^low and lower percentage of class-switched memory B cells [41, 42]. However, unlike Maglione et al., we were not able to demonstrate significantly higher levels of BAFF in the GLILD group. Instead, we did find higher levels of sBCMA in the GLILD group compared to the OC group. BCMA is shedded as sBCMA from plasma cells, and regulates plasma cell proliferation in bone marrow [43]. CVID patients overall are known to have substantially lower levels of sBCMA than healthy subjects, as confirmed in our cohort, reflecting the maturation defect of the B-cell lineage [44]. Although BCMA is mainly expressed on plasma cells, we cannot say if these or other cells are the source of sBCMA in GLILD. This issue may be subject for further studies, preferentially of bone marrow and lung biopsies of GLILD patients.

Currently, there is no consensus regarding optimal treatment of GLILD; however, rituximab combined with azathioprine or mycophenolate is often used and has shown effect in retrospective studies [6, 45]. Our results suggest that the use of treatment targeting T cells should be further explored. Although rituximab directly targets CD20⁺ cells, it also limits T cell activation by impairing B cell antigen presentation. Data on more T cell-specific therapy for GLILD are scarce. However, in a pilot study of treatment with the CTLA-4 Ig fusion protein abatacept, Warnatz et al. reported favorable results in five of eight CVID patients with ILD [46], and a case report of two GLILD patients describes successful treatment with the mTOR inhibitor sirolimus [47]. Inhibitors of the Janus kinase (JAK)/signal transducer and activator of transcription (STAT) pathway suppress intracellular signaling mediated by multiple cytokines, also afflicting T cells directly [48]. JAK inhibitors have been used to treat inflammatory conditions of some inborn errors of immunity: STAT3 and STAT1 gain-of-function mutations and type I interferonopathies, but there are as far as we know no reports of their use in CVID or GLILD.

The main strength of this study is a well-characterized CVID cohort with a substantial number of patients in all three subgroups, GLILD, OC, and IO. Another strength is that we used a simple and predefined statistical design with handpicked relevant serum markers for CVID and ILD/GLILD rather than using fixed multiplex panels. One limitation is that even though we compared GLILD patients to a CVID group of other non-infectious complications, we cannot rule out that other differences between the groups could have affected the results.

Studies in larger GLILD cohorts could clarify if the biomarkers we identified correlate with pulmonary function, treatment effect, and prognosis. The retrospective design is an important limitation of the study, as is the lack of data from bronchial lavage fluid and pulmonary biopsies from GLILD patients.

Our findings suggest that T cell activation and exhaustion, pulmonary epithelium injury, and ECM remodeling are central and distinct features of GLILD pathogenesis potentially reflecting novel targets for therapy and promising biomarkers for clinical use. However, there is a need for larger prospective studies that also include pulmonary biopsy material.