Lung CCR6-CXCR3- Th2 Cells as an Indicator of Progressive Fibrosing Interstitial Lung Diseases: A Prospective Observational Study

Tsukie Kin Tsukuda Kochi Medical School, Kochi University Minoru Fujimoto School of Medicine, Iwate Medical University Hiroshi Ohnishi (  honi@kochi-u.ac.jp ) Kochi Medical School, Kochi University Yu Nakatani Kochi Medical School, Kochi University Kazufumi Takamatsu Kochi Medical School, Kochi University Tetsuji Naka School of Medicine, Iwate Medical University Akihito Yokoyama Kochi Medical School, Kochi University


Introduction
Interstitial lung diseases (ILDs) consist of more than 200 lung diseases, most of which are associated with pulmonary brosis. Idiopathic pulmonary brosis (IPF) is classi ed as an idiopathic interstitial pneumonia (IIP) with the most frequent progressive brosing phenotype and the worst prognosis. Other ILDs such as idiopathic non-speci c interstitial pneumonia (NSIP), connective tissue disease-associated ILD (CTD-ILD), unclassi able IIPs, sarcoidosis, brotic hypersensitivity pneumonitis (HP), and pneumoconiosis often show similar brotic clinical features: progression of pulmonary brosis, worsening dyspnea on exertion, progressive decline in lung function, resistance to corticosteroids and/or immunosuppressive therapies, and poor prognosis (1)(2)(3)(4). They are categorized as progressive brosing interstitial lung disease (PF-ILD) or ILDs with a progressive brosing phenotype, and anti brotic therapies such as nintedanib and pirfenidone have been shown to improve disease progression of PF-ILD (5-7).
Prediction of future deterioration of lung function and/or progressive pulmonary brosis is di cult in real-world clinical settings, especially in patients with non-usual interstitial pneumonia (UIP) pattern on chest high-resolution computed tomography (HRCT), even though imaging resolution has improved remarkably. In such patients, sequential changes in pulmonary brosis on chest HRCT and/or in forced vital capacity (FVC) are often evaluated to determine whether anti brotic therapy should be started, but this follow-up may delay the treatment of PF-ILD. Identifying PF-ILD at diagnosis is extremely important for improving patients' prognosis.
There is an urgent and unmet need to identify predictors of responsiveness to corticosteroids, immunosuppressants, and/or anti brotic drugs. The underlying pathological mechanism of brosis in PF-ILD remains unclear and can be complicated by the involvement of multiple factors. Bronchoalveolar lavage uid (BALF) containing various immune cells recovered from the lungs is the best research sample to identify such a mechanism. In the present study, cellular pro les of T cell, B cell, macrophage, and monocyte subsets were compared in BALF and the peripheral blood of patients with brotic and nonbrotic ILDs using multicolor ow cytometry, and a characteristic cellular pro le of helper T (Th) cells in BALF to predict PF-ILD was found.  (8,9). Diagnosis of sarcoidosis was based on accepted criteria (10), and only cases with interstitial shadows in the lung elds were enrolled. Diagnosis of CTD-ILD was based on the common diagnostic criteria for each respective CTD and the presence of ILD on chest HRCT (11)(12)(13)(14)(15).

Methods
Participants underwent chest HRCT before bronchoscopy. BALF (n=40) and peripheral blood (n=38) were obtained from these patients on the same day or within 1 day. Serum Krebs von den Lungen-6 (KL-6) levels were measured by the chemiluminescent enzyme immunoassay LUMIPULSE KL-6 using a KL-6 antibody (SEKISUI MEDICAL Co., Ltd, Tokyo, Japan). The cut-off value for serum KL-6 was set at 435 U/ml according to the manufacturer's instructions. Patients with coexisting active respiratory infections, patients being treated with anti brotic drugs, and/or anticancer agents including immune checkpoint inhibitors, or insu cient BALF recovery (recovery rate less than 35%) were excluded. Progressive brosing interstitial lung disease (PF-ILD) was de ned as brotic ILD (traction bronchiectasis, architectural distortion, and/or honeycombing) on chest HRCT and a relative percent predicted value of forced vital capacity (%FVC) decrease of 5% or more over 6 months despite disease management deemed appropriate by the physician (e.g., antigen avoidance and treatment with corticosteroids ± immunosuppressive agents) in this study.
This study was approved by the ethics committee of the Kochi Medical School, Kochi University (Approval number 202-102) and was registered in the UMIN Clinical Trials Registry (Registry ID UMIN000041871). Written, informed consent was obtained from all participants.

Interpretation of chest HRCT images
Chest HRCT ndings were reviewed separately by two independent respiratory specialists who were not aware of the patients' pro les and grouped into brotic and non-brotic ILDs. The presence of traction bronchiectasis, architectural distortion, and/or honeycombing on chest HRCT provided evidence of lung brosis based on the guideline of the Fleischner Society. Cases with traction bronchiectasis and/or architectural distortion without honeycombing were further de ned as brotic NSIP (f-NSIP) patterns, whereas cases with honeycombing were de ned as a UIP pattern. Following the initial independent chest HRCT assessment, discordant evaluations were resolved by consensus reached after consultation between the two observers.  Figure 1. Classi cation of Th subsets (Th1, Th1/17, Th17, and Th2) of conventional T cells by CCR6 and CXCR3 expressions is shown in Table 1(16).

Pulmonary function tests
Pulmonary function tests were performed 6-7 months before bronchoscopy as a baseline and a second time around 1 month before bronchoscopy in 11 cases, and around 1 month after bronchoscopy as a baseline and a second time 6-7 months after bronchoscopy in 10 cases.

Statistical analysis
Data are presented as medians (interquartile range). Differences between two groups were evaluated by the Mann-Whitney U test. The correlation coe cients between blood cells and BALF cells were calculated using Spearman's rank correlation coe cient (Rs). Quantitative differences were tested by the Chisquared test for goodness of t or by Fisher's exact test. Interobserver agreement for chest HRCT ndings was assessed using Kappa statistics. Kappa values greater than 0.61 were considered to indicate good agreement between observers. Receiver operating characteristic (ROC) curve analysis was used to determine the cut-off values that provide the optimal diagnostic accuracy for the proportion of Th2 cells in BALF to predict PF-ILD. All statistical analyses were performed using EZR version 1.54 (Saitama Medical Center, Jichi Medical University, Saitama, Japan) (17). Statistical signi cance was set at p < 0.05.

Patients' characteristics
Chest HRCT ndings were classi ed as brotic ILD in 27 patients and non-brotic ILD in 13 patients ( Table 2). Interobserver variability before reaching consensus agreement, expressed as a Kappa value, was 0.74, indicating good agreement between observers. There was no signi cant difference in age between brotic ILD and non-brotic ILD. There were more males in brotic ILD than in non-brotic ILD, and more never smokers in non-brotic ILD than in brotic ILD. Serum KL-6 and lactate dehydrogenase (LDH) were signi cantly higher in brotic ILD than in non-brotic ILD. In cellular pro les of BALF, the proportion of lymphocytes was signi cantly higher in non-brotic ILD than in brotic-ILD, whereas the proportion of eosinophils and neutrophils was signi cantly higher in brotic ILD than in non-brotic-ILD. IPF, HP, and CTD-ILD were common underlying diseases of brotic ILD, whereas sarcoidosis was the most common underlying disease of non-brotic ILD.
Furthermore, relative decline of % forced vital capacity over 6 months, chest HRCT pattern, cellular pro le of bronchoalveolar lavage uid, and serum KL-6 levels in brotic patients with brotic ILDs are shown in Table 3 (n=21). At the time of bronchoscopy, only one of the patients with brotic ILD was treated with corticosteroids and tacrolimus; the others were untreated. After bronchoscopy, some were started on corticosteroids alone or in combination with corticosteroids and tacrolimus.
Th subsets in BALF and brotic phenotype on chest HRCT Cellular pro les of BALF showed differences in the proportion of Th subsets between brotic ILD and non-brotic ILD on chest HRCT. The proportion of CCR6 -CXCR3 -CD4 + T cells (Th2) was signi cantly increased in BALF of brotic ILD compared with non-brotic ILD (Figure 2A). On the other hand, the proportion of CCR6 + CXCR3 + CD4 + T cells (Th1/Th17) was signi cantly increased in BALF of non-brotic ILD compared with brotic ILD ( Figure 2B). These differences in Th subsets were not observed in cells in peripheral blood when comparing brotic and non-brotic ILD (Figure 2A and 2B). The proportion of Th1/Th17 (CCR6 + CXCR3 + T cells) in BALF, but not of Th2 (CCR6 -CXCR3 -T cells, Rs = 0.181, p = 0.276 for Th2, Figure 2C), was weakly correlated with the proportion in peripheral blood (Rs = 0.485, p = 0.0021 for Th1/Th17, Figure 2D). Cellular pro les in subsets of B cells, macrophages, or monocytes in the peripheral blood and BALF were not different between patients with brotic ILD and non-brotic ILD. These results suggest that the Th2 subset in the lung, but not in peripheral blood, is associated with the brotic phenotype of ILD on chest HRCT.

Proportion of lymphocytes and Th subsets in BALF in all ILD patients
Furthermore, since the proportion of lymphocytes in BALF is often used as a criterion to predict response to corticosteroids in clinical practice, the correlations between the proportion of lymphocytes and Th2 and Th1/17 in BALF were examined. The proportion of Th2 (CCR6 -CXCR3 -CD4 + T cells) was negatively correlated with the proportion of lymphocytes in BALF in all study subjects (n=40, Rs = -0.419, p = 0.0071, Figure 3A), but no correlation was found for Th1/Th17 (CCR6 + CXCR3 + CD4 + T cells) (Rs = 0.211, p = 0.191, Figure 3B). These results suggest that the lower the lymphocytosis, the higher the Th2 in BALF in patients with ILDs.
Th subsets in BALF and PF-ILD phenotype in patients with brotic ILD Based on these results, we hypothesized that Th2 (CCR6 -CXCR3 -CD4 + T cells) in BALF could be an indicator of PF-ILD and then veri ed this hypothesis.
In brotic ILD patients, the proportion of Th2 (CCR6 -CXCR3 -T cells) in BALF was signi cantly higher in patients with PF-ILD (a relative %FVC decline of 5% or more over 6 months) than in patients with non PF-ILD (less reduction) (p = 0.028, Figure 4A). The proportion of lymphocytes in BALF and serum KL-6 levels were not different between patients with PF-ILD and non PF-ILD ( Figure 4B and C). These results suggest that the Th2 subset in the lungs, but not lymphocytes in BALF or the serum KL-6 level, may be associated with the PF-ILD phenotype in brotic ILD.
Diagnostic value of Th2 cells in BALF for determining PF-ILD ROC curve analysis was used to evaluate the diagnostic value of the proportion of Th2 cells in BALF for identifying PF-ILD patients, and the area under the curve (AUC) value calculated from ROC curve analysis was 0.787 [95% con dence interval (CI): 0.565-1.000] ( Figure 4D). The cut-off value of the proportion of Th2 cells in BALF was set at more than 14.80 based on the highest Youden's index; the sensitivity, speci city, and likelihood ratio for predicting PF-ILD were 0.889, 0.750, and 3.56, respectively. Even with the small number of patients included in this study, these data suggest that Th2 in BALF has moderate diagnostic accuracy for PF-ILD.

Discussion
The cellular pro les of BALF and peripheral blood were evaluated in mostly treatment-naïve patients with various ILDs, and it was found that the proportion of Th2 (CCR6 -CXCR3 -CD4 + T) cells in BALF, not of other Th subsets, was signi cantly increased in patients with brotic ILD compared with non-brotic ILD on chest HRCT. and they were increased with a signi cant decrease in %FVC. These results clearly demonstrated that the predominance of Th2 cells in the lungs was associated with progressive lung brosis in patients with ILD. The increased proportion of CCR6 -CXCR3 -Th2 cells in BALF found in this study may help to identify these patients and provide early intervention to improve the prognosis of PF-ILD patients, but further con rmation is needed in the future.
In general, lymphocytosis in BALF is an important predictor for responsiveness to corticosteroids in clinical practice, but the present study demonstrated that the proportion of Th2 in BALF, but not of lymphocytes, may be predictive of progressive brosing phenotype. Although serial measurements of serum KL-6 levels were reported to be useful for predicting progression of pulmonary brosis (18, 19), serum KL-6 levels taken at the time of bronchoscopy in this study did not show an association with the progression of pulmonary brosis.
PF-ILD was found to have both brotic ILD (traction bronchiectasis, architectural distortion, and/or honeycombing) on chest HRCT and a relative %FVC decrease of 5% or more over 6 months despite disease management deemed appropriate by the physician (e.g., antigen avoidance and treatment with corticosteroids ± immunosuppressive agents) in the present study, but there are no standard criteria for PF-ILD. One paper has proposed any of the following criteria within 24 months as evidence of disease progression: a relative decline of 10% or greater in FVC; a relative decline of 15% or greater in diffusing capacity of the lung for carbon monoxide; worsening symptoms; or radiological appearance accompanied by a 5% or greater but less than 10% relative decrease in FVC (4). However, another study of pirfenidone used a criterion of more than a 5% FVC decrease for at least 6 months (20).
In the present study, the proportion of Th2 among conventional T cells in BALF was increased with the progression of brosis. Interestingly, several reports have demonstrated the relationship between pulmonary brosis and Th2. Periostin, induced by the type 2 cytokines interleukin-4 (IL-4) and IL-13, has been reported to be involved in pulmonary brosis with crosstalk of TGF-β (21,22), and serum periostin could be a good predictor of decreased lung function, increased radiation brosis, and prognosis (23)(24)(25)(26). Moreover, lines of evidence indicate that M2-like macrophages, known to be induced by type 2 cytokines, are profoundly involved in the pathogenesis of brosis (27)(28)(29). At present, which cell type(s) drive the pathogenesis of lung brosis remains unknown. In the present study, the absolute number of BALF Th2 was not markedly increased in patients with progressive brosis, given that the proportion of BALF lymphocytes in these patients was lower than of other ILD patients. Therefore, although a relative increase in BALF Th2 is likely an indicator of the Th2-dominant pathology in progressive brosis, Th2type T cells may not be the main player in the progression of the disease. Further studies are needed to de ne the pathological roles of Th2-type T cells in lung brosis.
Currently, anti brotic agents such as pirfenidone and nintedanib have been reported to be effective in reducing lung function decline in PF-ILD. The timing of the introduction of anti brotic agents in clinical practice depends on the individual judgment of the attending physicians, and many believe that it is best to start as early as possible if the disease has progressive brosing features. Since poorer performance status is the most common cause of the discontinuation of treatment, it is important to start anti brotic drugs early before performance status worsens (30).
The limitations of this study include the small number of study subjects in a single center dealing with respiratory diseases. Even though a small number of subjects were included, the present study clearly showed that Th2 cells in BALF had moderate diagnostic accuracy for determining PF-ILD. The second limitation is that the effect of anti brotic drugs on Th2-high ILD patients could not be investigated in the present study, because only one patient was treated with nintedanib. Another limitation is the heterogeneity of underlying ILD. Different backgrounds of the diseases may have their own speci c immunological responses, which may have affected the study results. Therefore, further large prospective studies are required to examine the administration of anti brotic drugs to ILD patients with high Th2 cells in BALF.

Conclusion
In conclusion, the predominance of Th2 cells in BALF is a predictor of a greater decrease of lung function in PF-ILD, and these patients can be good candidates for anti brotic therapy.

Declarations
Ethics approval and consent to participate This study was approved by the ethics committee of the Kochi Medical School, Kochi University. Written informed consent was obtained from all participants.

Not applicable
Availability of data and materials The datasets used and/or analyzed during the current study are available from the corresponding author on reasonable request.

Competing interests
The authors declare that they have no competing interests. study supervision: HO, MF, TN, and AY; data access and responsibility: TKT and HO had full access to all the data in the study and take responsibility for the integrity of the data and the accuracy of the data analysis.   Figure 1 Gating strategy of ow cytometry for identifying helper T cell (Th) subsets For analysis of peripheral blood (A), we gated on live lymphocytes using FSC and SSC, followed by exclusion of doublets, then identi ed CD3 + CD4 + T cells. Among these CD3 + CD4 + T cells, we gated on CD45RA -CXCR5populations to gate out CD45RA + naïve T cells and CXCR5 + follicular helper T cells, then gated on CD127 Low or CD127 High Tconv and to gate out Tregs using CD25 and CD127. CCR6 and CXCR3

Tables
for the classi cation of Th subsets (refer to Table 1). The same gating strategy was applied for analysis of BALF cells (B). In the BALF cells, the activation of each Th subset was identi ed using HLA-DR and CD38. A and B are examples of mixed connective disease-associated ILD. ILD, interstitial lung disease; BALF, bronchoalveolar lavage uid; FSC, forward scatter; SSC, side scatter; Tconv, conventional T cells; Tregs, regulatory T cells.  Correlations between the proportion of lymphocytes and CCR6 -CXCR3 -Th2 (A) or CCR6 + CXCR3 + Th1/17 (B) in BALF of all ILD patients (n=40). BALF, bronchoalveolar lavage uid.