Defining serum tumor necrosis factor α concentration–related endotype of sarcoidosis: a real-life, retrospective, observational Polish study

Introduction

Sarcoidosis is a heterogeneous, systemic disease, mostly occurring in young adults. It predominantly affects the lungs and lymph nodes. In most cases, the disease resolves spontaneously or is self-limited. In 10% to 30% of patients, sarcoidosis progresses to irreversible organ damage, despite therapy.¹ Poor outcomes are difficult to predict; none of the laboratory markers have been validated in this regard. Clinical presentation of sarcoidosis depends on multiple pathophysiologic mechanisms driven by complex interactions between pro- and anti-inflammatory cytokines, including tumor necrosis factor α (TNF-α), which determines granulomatous inflammation.² We hypothesized that the serum TNF-α (sTNF-α) concentration may define an inflammatory endotype of sarcoidosis with specific demographic, clinical, radiologic, and functional characteristics.

Patients and methods

Nontreated patients with pulmonary biopsy–confirmed sarcoidosis were enrolled in the study between April 2005 and December 2008. All of them provided written informed consent for participation. The study was conducted according to the principles outlined in the Declaration of Helsinki, and its protocol was approved by the Ethics Committee at the Institute of Tuberculosis and Lung Diseases (KE-31/2004).

Baseline patient evaluation included sex, age at the disease onset, age at the study enrollment, disease duration, and a history of treatment. A binary questionnaire (yes / no) was used to assess the presence of the following symptoms at the disease onset: joint pain / edema, weight loss, night sweats, pyrexia, chronic cough, and exertional dyspnea. Presence of at least 3 symptoms classified a patient as symptomatic.

Pulmonary sarcoidosis was classified according to the Scadding scale and evidence of lung parenchymal disease, without or with signs of fibrosis (reticular lesions, and / or distortion bronchiectasis, and / or honeycomb pattern) on computed tomography (CT) scans. Chest X-rays (CXRs) performed at the enrollment visit were compared with those performed at the onset of sarcoidosis in retrospective analysis. Progression of radiologic changes was defined as the presence of mediastinal / hilar adenopathy on the CXR/CT scan at the study enrollment and its absence on initial imaging carried out at sarcoidosis onset, and / or as progression / onset of disseminated / reticular lesions, consolidations, bronchiectasis, or honeycombing not present on the initial CXRs/CT scans. Absence of significant changes in radiologic presentation between the onset of sarcoidosis and the enrollment visit defined stabilization. Regression was considered to occur if none of the above criteria were fulfilled. The patients were allocated to 1 of 3 groups according to their radiologic presentation: 1) progression, 2) stabilization, and 3) regression.

Pulmonary function (PF) parameters (percent predicted of forced expiratory volume in 1 second [FEV₁], forced vital capacity [FVC], FEV₁/FVC index, total lung capacity [TLC], and diffusing capacity for carbon monoxide [DL_CO]) at the enrollment visit and at sarcoidosis onset were compared. Respiratory functional impairment was diagnosed when at least 1 of the above parameters was below the reference values. Deterioration of PF parameters at enrollment was defined as a decrease in spirometry values and / or TLC by at least 10% and / or decrease in DL_CO by 20%, as compared with the values obtained at sarcoidosis onset. Improvement was considered to occur when the PF parameters that had been decreased at the disease onset normalized by the time of enrollment. The patients who did not fulfill the deterioration / improvement criteria comprised a group with stable PF parameters.

Maximal and minimal oxygen saturation (SpO₂max and SpO₂min, respectively) and desaturation (ΔSpO₂) values, the 6-minute walking distance (6MWD), and the dyspnea Borg score during the 6-minute walking test (6MWT) were recorded at enrollment. The SpO₂min value greater than or equal to 90% during exercise was considered normal. The cutoff value for clinically significant ΔSpO₂was greater than or equal to 4%. Patients were diagnosed with exercise ability impairment (EAI) when their SpO₂min and / or ΔSpO₂were below the cutoff values.

Organ involvement was assessed according to the criteria proposed in the ACCESS (A Case Control Etiologic Study of Sarcoidosis) study.³ The participants were allocated into 1 of 3 groups: 1) no extrapulmonary involvement, 2) a single extrapulmonary organ involvement, and 3) involvement of 2 or more extrapulmonary organs. The patients were eligible for treatment as described in our previous study.⁴

Laboratory workup included assessment of red blood cell count, white blood cell count, blood eosinophil count, platelet count, serum calcium level, total protein level and electrophoresis (albumin, β-globulin, γ-globulin), liver function parameters (alanine transaminase, aspartate aminotransferase, γ-glutamyl transpeptidase [GGTP], and alkaline phosphatase [ALP]), as well as plasma D-dimer, fibrinogen, C-reactive protein (CRP), and sTNF-α concentrations. The sTNF-α level was measured by a sandwich enzyme-linked immunosorbent assay using a commercially available kit (TNF-α BIOSOURCE Europe SA, Nivelles, Belgium).

Statistical analysis

Computations and statistical analysis were performed in the R programming language. The Kruskal–Wallis test was used to investigate if the distribution of sTNF-α values differed between categories of qualitative variables. Kendall τ correlations between the sTNF-α level and values of quantitative variables were assessed. Significance of the correlations was verified with the Z test (H₀: τ = 0; H₁: τ ≠ 0). The Benjamini–Hochberg correction was applied to control the false discovery ratio. The adjusted P value threshold was considered to be 0.05. Therefore, if a given variable had the adjusted P value <⁠0.05, there is a 95% chance it is a true discovery.

The sTNF-α cutoff threshold was calculated for all binary variables. Based on the Kruskal–Wallis test, the distribution of sTNF-α values was significantly different for each category of the binary variables (Table 1). For each category of each variable, the mean sTNF-α level was computed. Consequently, for each variable, there was a pair of mean sTNF-α values. The cutoff was set as a midpoint between the maximum of all lower means and the minimum of all higher means. This calculation method allowed for a nuanced distinction between the groups, emphasizing the role of the sTNF-α cutoff value as a heuristic indicator rather than a strict predictive measure, and highlighting its practical utility. Based on the sTNF-α concentration of a given patient, we can infer the category or value of other variables; thus, it provides a helpful clue to the understanding of the underlying data patterns and their implications on patient categorization.

**Table 1**. Quantitative parameters used to define the clinical endotype of sarcoidosis
Parameter	Median (IQR)	Kendall τ correlation coefficient	Adjusted P value	Group 1: sTNF-α <⁠28.58 pg/ml; median (IQR)	Group 2: sTNF-α >28.58 pg/ml; median (IQR)	Adjusted P value
At sarcoidosis diagnosis
Age, y	42 (36–51)	0.21	<⁠0.001	41 (34.5–49)	45.5 (38–55.25)	0.12
At study enrollment
Age, y	47 (39–58)	0.2	<⁠0.001	45 (37–56)	52 (42–61.75)	0.12
Sarcoidosis duration, y	3 (1–6)	0.11	0.1	2 (1–5.5)	4 (1.81–6)	0.17
Liver left lobe size on US, cm	9.5 (8.5–10.3)	0.15	0.03	9.2 (8.2–10.2)	9.85 (9–11)	0.14
Liver right lobe on US, cm	14.1 (12.9–15.1)	0.09	0.21	14.05 (13–15.07)	14.6 (12.9–15.65)	0.43
Spleen longitudinal dimension on US, cm	10.8 (9.8–12.1)	0.24	<⁠0.001	10.6 (9.5–11.93)	12 (10.25–12.85)	<⁠0.001
Spleen transverse dimension on US, cm	4.2 (3.5–5)	0.18	0.01	4 (3.5–4.8)	4.8 (3.8–5.6)	0.01
Functional evaluation at study enrollment
FEV₁, l	2.95 (2.36–3.58)	–0.22	<⁠0.001	3.21 (2.53–3.75)	2.45 (2.18–2.94)	<⁠0.001
FEV₁, % predicted	91.2 (78.45–100.73)	–0.3	<⁠0.001	95.4 (87.48–104.25)	81.4 (70.72–88.02)	<⁠0.001
FVC, l	3.78 (3.08–4.59)	–0.19	<⁠0.001	3.95 (3.43–4.77)	3.49 (2.92–4.16)	0.01
FVC, % predicted	96.75 (84.6–106.2)	–0.29	<⁠0.001	99.5 (91.58–107.85)	85.4 (77.55–97.08)	<⁠0.001
FEV₁/FVC, %	75.81 (71.25–81.64)	–0.11	0.1	76.37 (72.36–83.2)	74.73 (66.91–79.71)	0.22
TLC, l	5.64 (4.95–6.47)	–0.15	0.03	5.78 (5.11–6.69)	5.12 (4.51–5.9)	0.02
TLC, % predicted	98.55 (90.68–106.25)	–0.27	<⁠0.001	101.2 (96.1–110.3)	93.3 (80.65–99.58)	<⁠0.001
DL_CO, mmol/kPa/min	7.38 (6.12–8.72)	–0.26	<⁠0.001	7.94 (6.81–8.98)	6.38 (5.27–7.38)	<⁠0.001
DL_CO, % predicted	80.9 (70.55–90.9)	–0.36	<⁠0.001	84.5 (76.8–91.2)	67.8 (60.92–75.18)	<⁠0.001
SpO₂max during 6MWT	97 (96–98)	–0.15	0.04	97 (96–98)	97 (95–98)	0.33
SpO₂min during 6MWT	94 (92–95)	–0.25	<⁠0.001	94 (93–96)	93 (91–95)	0.046
ΔSpO₂ during 6MWT	3 (2–4)	0.19	0.01	3 (2–4)	3 (2–5)	0.16
6MWD, m	564 (500.5–620.25)	–0.15	0.03	569 (504–626)	558 (472–594)	0.36
6MWD, % predicted	90.57 (82.46–96.52)	–0.03	0.74	90.51 (82.34–96.18)	91.1 (82.8–96.54)	0.92
Laboratory evaluation at study enrollment
Calcium, mmol/l	4.8 (4.5–5.1)	0.05	0.57	4.7 (4.5–5)	4.8 (4.6–5.1)	0.28
Bilirubin, mg/dl	0.6 (0.5–0.8)	0.13	0.08	0.6 (0.5–0.85)	0.65 (0.6–0.8)	0.36
AST, U/l	23 (19–29.5)	0.17	0.01	23 (19–28.75)	24 (20–34)	0.37
ALT, U/l	22 (15–29)	0.07	0.32	22 (15–33.5)	22 (17–29)	0.92
GGTP, U/l	28 (16.5–46.2)	0.25	<⁠0.001	23.9 (14.35–40.55)	34.9 (19.2–68.6)	0.03
ALP, U/l	81 (63–100)	0.29	<⁠0.001	75 (55.75–93)	98 (72–117)	<⁠0.001
Plasma protein, g/dl	7.5 (7.1–8)	0.15	0.03	7.4 (7.1–8)	7.8 (7.1–8.4)	0.09
Albumin, g/l	51.5 (48.72–54.13)	–0.27	<⁠0.001	52.35 (49.93–55.33)	49.23 (45.17–52.08)	<⁠0.001
α1, g/l	5.04 (4.56–5.65)	–0.14	0.04	5.08 (4.67–6.02)	4.96 (4.51–5.38)	0.22
α2, g/l	8.81 (8–9.71)	–0.03	0.74	8.94 (7.93–9.72)	8.64 (8.04–9.7)	0.79
β-Globulin, g/l	14.59 (13.22–15.75)	0.02	0.79	14.7 (13.09–15.78)	14.5 (13.52–15.59)	>0.99
γ-Globulin, g/l	19.38 (16.97–22.37)	0.33	<⁠0.001	18.41 (16.49–20.09)	22.9 (19.62–25.89)	<⁠0.001
LDH, U/l	301.5 (271.75–329.5)	0.09	0.24	300 (274–324.5)	309 (269–345)	0.71
CRP, mg/l	3.9 (1.83–8.57)	0.28	<⁠0.001	2.6 (1.1–5.65)	7.9 (4.05–12.65)	<⁠0.001
Uric acid, mg/dl	5.7 (4.7–6.9)	0.31	<⁠0.001	5.3 (4.43–6.4)	6.9 (5.48–7.88)	<⁠0.001
Creatinine, mg/dl	0.9 (0.7–1)	0.25	<⁠0.001	0.8 (0.7–1)	0.9 (0.8–1.1)	0.08
Hemoglobin, g/dl	13.9 (13.1–14.9)	0.04	0.6	13.9 (13.1–15.05)	13.85 (13.1–14.8)	0.71
Hematocrit, %	41.1 (38.9–43.7)	0.02	0.79	41.6 (38.95–43.7)	40.8 (38.85–43.38)	0.56
Erytrocytes, ×10⁶/µl	4.73 (4.44–5.01)	0.06	0.42	4.77 (4.39–5.06)	4.7 (4.55–4.94)	0.71
Leucocytes, ×10⁶/µl	5.55 (4.7–6.89)	0.01	0.86	5.55 (4.7–6.76)	5.59 (4.81–7.2)	0.72
Platelets, ×10⁶/µl	212 (178–248)	–0.1	0.13	221 (185–253.5)	200 (163.5–234.75)	0.23
Neutrophils, %	64.2 (57.3–68.5)	–0.12	0.07	64.5 (57.75–69)	63.75 (55.1–67.57)	0.43
Lymphocytes, %	23.3 (19.2–29.4)	0.04	0.6	23.3 (19.15–29.35)	22.9 (19.35–29.4)	>0.99
Monocytes, %	8.2 (7–9.4)	0.23	<⁠0.001	8.1 (6.65–9.4)	8.5 (7.7–9.25)	0.24
Eosinophils, %	3.5 (2–4.8)	0.17	0.01	3.3 (1.8–4.55)	4.15 (2.9–6.46)	0.03
Basophils, %	0.5 (0.4–0.7)	–0.02	0.79	0.5 (0.4–0.7)	0.5 (0.3–0.6)	0.59
Neutrophils, ×10³/µl	3.46 (2.8–4.52)	–0.01	0.88	3.3 (2.78–4.4)	3.54 (2.88–4.62)	0.79
Lymphocytes, ×10³/µl	1.27 (1.01–1.67)	0.05	0.53	1.27 (1.02–1.62)	1.28 (0.99–1.78)	0.79
Monocytes, ×10³/µl	0.45 (0.37–0.53)	0.23	<⁠0.001	0.44 (0.36–0.53)	0.49 (0.39–0.56)	0.12
Eosinophils, ×10³/µl	0.17 (0.11–0.27)	0.18	0.01	0.16 (0.1–0.24)	0.24 (0.13–0.44)	0.03
Basophils, ×10³/µl	0.03 (0.02–0.04)	–0.03	0.74	0.03 (0.02–0.04)	0.03 (0.02–0.04)	0.7
D-Dimer, µg/l	411 (237–664)	0.35	<⁠0.001	319 (212–455)	685 (364.25–1079.75)	<⁠0.001
Fibrinogen, g/l	3.34 (2.73–3.98)	–0.09	0.23	3.43 (2.78–3.93)	3.09 (2.7–4.02)	0.66
Kendall’s τ correlation with serum TNF-α concentration was computed for every variable. Significance of the correlation was tested with the Z test. The Benjamini–Hochberg correction was applied to control the false discovery ratio; adjusted P values <⁠0.05 were considered significant. For the comparison between sTNF-α concentration–related endotypes, the Kruskal–Wallis test was performed to investigate if distributions differed between the groups. The Benjamini–Hochberg correction was applied; adjusted P values <⁠0.05 were considered significant. SI conversion factors: to convert bilirubin to µmol/l, multiply by 17.104; AST, ALT, GGTP, ALP, and LDH to µkat/l, by 0.0167; protein and hemoglobin to g/l, by 10; CRP to nmol/l, by 9.524; uric acid to µmol/l, by 59.485; creatinine to µmol/l, by 88.4; D-dimer to nmol/l, by 0.0056; fibrinogen to µmol/l, by 0.294. Abbreviations: 6MWD, 6-minute walking distance; 6MWT, 6-minute walking test; ALP, alkaline phosphatase; ALT, alanine transaminase; AST, aspartate aminotransferase; CRP, C-reactive protein; DL_CO, diffusing capacity for carbon monoxide; FEV₁, forced expiratory volume in 1 second; FVC, forced vital capacity; GGTP, γ-glutamyl transpeptidase; LDH, lactate dehydrogenase; SpO₂max, maximal oxygen saturation; SpO₂min, minimal oxygen saturation; ΔSpO₂, desaturation during exercise; sTNF-α, serum tumor necrosis factor α concentration; TLC, total lung capacity; US, ultrasonography

We investigated whether belonging to a given sTNF-α concentration–related endotype affected other variables. The patients were divided into 2 groups depending on the sTNF-α concentration (group 1, sTNF-α <⁠28.58 pg/ml and group 2, sTNF-α >28.58 pg/ml). The χ² independence test (H₀: there is no association between variables; H₁: otherwise) was used to verify whether the variable categories were independent of a given group of qualitative variables. For quantitative data, the Kruskal–Wallis test (H₀: F₁ = F₂; H₁: F₁ ≠ F₂, with F_i standing for the cumulative distribution function) was applied to ascertain if samples from the 2 groups originated from the same distribution. The Benjamini–Hochberg correction was used, with the adjusted P value threshold set to 0.05.

Results

Among the 125 patients with sarcoidosis included in the study (64 men [51%], 61 women [49%]; median age, 42 years [interqartile range, IQR, 36–51]; median duration of the disease, 3 years [IQR, 1–6]), 77 individuals (62%) were older than 40 years at the disease onset. At enrollment, 87 patients (70%) were older than 40 years. A total of 56 patients (45%) had comorbidities, with cholelithiasis being the most common (21 patients [17%]). Symptoms at diagnosis were present in 88 patients (70%), 17 (14%) had erythema nodosum, and 57 (46%) reported EAI. Radiologic signs of fibrosis were detected in 8 patients (6%) at the onset of sarcoidosis, and in 65 (52%) at the study enrollment. Radiologic progression of disseminated and / or fibrotic lesions was observed in 75 patients (60%), stabilization in 25 (20%), and regression in 22 (18%). At enrollment, 16 patients (13%) were diagnosed with restriction, 59 (47%) with decreased DL_CO, and 26 (21%) with airflow limitation. Deterioration in PF parameters between the disease onset and enrollment was found in 37 patients (30%), stabilization in 46 (37%), and improvement in 8 (6%).

The sTNF-α level did not differ between the sexes and was not associated with any of the following: disease duration, presence of symptoms at the disease onset, Scadding CXR classification and radiologic signs of fibrosis on CT, the FEV₁/FVC index, or previous / new steroid or methotrexate treatment. Age at sarcoidosis onset and at enrollment was associated with the sTNF-α level (τ = 0.21; P <⁠0.001 and τ = 0.2; P <⁠0.001, respectively). The patients older than 40 years at sarcoidosis onset and at enrollment had higher levels of sTNF-α (P = 0.03). With respect to the radiologic presentation, the individuals with radiologic progression of sarcoidosis presented with higher sTNF-α levels than those with radiologic stabilization or regression (P <⁠0.001). The group with radiologic regression had lower sTNF-α levels (P = 0.02) than the remaining groups.

A significant negative correlation was found between the sTNF-α concentration and PF parameters (both absolute and percent predicted values), except for the FEV₁/FVC index. Negative correlations between the sTNF-α value and 6MWD (τ = –0.15; P = 0.03) as well as SpO₂max (τ = –0.25; P <⁠0.001) and SpO₂min (τ = –0.15; P = 0.04) during the 6MWT were observed. Higher sTNF-α levels were noted in the patients with desaturation greater than or equal to 4% during the 6MWT (n = 43 [34%]; τ = 0.19; P = 0.01), and in those with EAI (n = 43 [34%]; P = 0.03).

At enrollment, 70 patients (56%) were diagnosed with multiorgan sarcoidosis. Splenic sarcoidosis (P <⁠0.001), abdominal lymphadenopathy (P <⁠0.001), peripheral lymphadenopathy (P <⁠0.001), cutaneous sarcoidosis (P <⁠0.001), and a greater number of affected organs (P <⁠0.001) were associated with a higher sTNF-α level. Hepatic (left lobe, τ = 0.15; P = 0.03) and splenic dimensions on abdominal ultrasound (τ = 0.24; P <⁠0.001 and τ = 0.18; P = 0.01 for longitudinal and transverse dimensions, respectively) were associated with higher sTNF-α concentrations. A positive correlation with the sTNF-α level was observed for cholestatic enzymes, GGTP (τ = 0.25; P <⁠0.001) and ALP (τ = 0.29; P <⁠0.001). Moreover, a positive correlation was observed for plasma protein (τ = 0.15; P = 0.03), γ-globulin (τ = 0.33; P <⁠0.001), CRP (τ = 0.28; P <⁠0.001), D-dimer (τ = 0.35; P <⁠0.001), monocyte concentrations (τ = 0.23; P <⁠0.001), and percentage of eosinophils (τ = 017; P = 0.01), but not for percentage of granulocytes in blood smear (inverse association; P = 0.03). Comorbidities, except for cholelithiasis (P = 0.03), were not associated with the sTNF-α level. Clinical and laboratory characteristics of patients stratified according to the sTNF-α concentration–related endotype are presented in Table 1.

Discussion

The novel approach to interstitial lung diseases (ILDs) differentiates between inflammatory and fibrotic phenotypes. Sarcoidosis is an entity distinct from other ILDs, having the potential for self-limitation or a long-lasting course before reaching its end stage, that is, pulmonary fibrosis. Fibrosis in the course of sarcoidosis results from prolonged granulomatous inflammation.⁵ Phenotyping of sarcoidosis should include assessing the active inflammatory process, as it may help differentiate between patients with progressive vs burnout disease. The assessment can be performed using simple and cost-effective laboratory measurements, and radioactive imaging tools, such as positron emission tomography–CT, can be used for confirmation, if needed.⁶

Targeted inflammation control may modify the outcome of the disease.^7,8 As the anti–TNF-α medications show contradictory effectiveness,^9,10 we presumed that defining the sTNF-α concentration–related endotype of sarcoidosis could help select the patients that may benefit from this therapy.

TNF-α is produced by various immune cells.¹¹ Its levels in bronchoalveolar lavage fluid fluctuate depending on the disease activity and, if increased, indicate a more severe form of the disease.^11,12 In Asian Indian patients, insidious onset and chronic form of sarcoidosis were associated with higher sTNF-α levels, as compared with patients with acute onset and manifestations.² African-American patients with neurologic disease had significantly higher sTNF-α levels than the individuals lacking this manifestation.¹³ In the studies assessing the clinical efficacy of anti–TNF-α medications, a more pronounced effect was seen in patients with more severe disease, elevated levels of CRP and other unspecific markers of systemic inflammation, and the highest sTNF-α concentration.^9,14-17

Our results are consistent with those reported in previous studies. We found a significant association of the sTNF-α concentration with levels of nonspecific inflammatory markers, extrapulmonary sarcoidosis, and more severe, progressive lung involvement. The disease course is highly variable and unpredictable in the case of persistent lung involvement (Scadding stages II and III). PF depends on the pathophysiologic conditions affecting the pulmonary parenchyma: alveolitis and granuloma infiltration (potentially reversible conditions), and fibrotic changes (potentially irreversible).¹⁸ Abnormal PF test results will not pinpoint the cause of pulmonary damage, although residual airflow obstruction is associated with residual pulmonary fibrosis.¹⁹ In this context, a negative correlation between the sTNF-α level and PF parameters other than the FEV₁/FVC index and exercise capacity parameters (6MWD, SpO₂max, ΔSpO₂) would indicate a disabling respiratory disease related to active inflammation rather than to fibrosis. Levels of sTNF-α greater than 28.58 pg/ml combined with PF abnormalities other than airflow limitation and EAI may be characteristic of patients with potentially reversible, active, disabling sarcoidosis. In our study, radiologic progression, abnormal lung and performance function, older age, and extended disease, but not the presence of fibrosis on CT or reduced FEV₁/FVC index, were related to increased sTNF-α levels. These clinical characteristics could differentiate between an active, progressive endotype vs burnout disease. Initiation of anti–TNF-α treatment in patients with this specific disease endotype (sTNF-α >28.58 pg/ml) may contribute to stopping the granulomatous inflammation that leads to pulmonary fibrosis, and modify the disease course.⁷ sTNF-α values greater than 28.58 pg/ml were associated with extrapulmonary disease. Therefore, patients with sTNF-α concentrations above this cutoff may benefit from active diagnostic screening for extrapulmonary involvement.

In previous studies, treatment with infliximab and adalimumab was more effective in patients with extrapulmonary sarcoidosis than in those with isolated sarcoidosis of the respiratory tract.²⁰

Our study has some significant limitations. We did not compare the sTNF-α levels between patients with sarcoidosis and healthy individuals. Our participants were recruited from a white Polish cohort, and the results may not apply to other ethnicities. The retrospective design did not allow us to conclude on the predictive value of the sTNF-α concentration. To confirm our results, further studies should validate the identification of sarcoidosis endotypes related to sTNF-α levels. Feasibility and cost-effectiveness of sTNF-α measurement make it a potentially valuable additional marker in clinical practice.

Conclusions

The sTNF-α level defines an endotype of radiologically progressive sarcoidosis with impaired PF and EAI. The cutoff value of 28.58 pg/ml allows for discriminating between patients with active, progressive sarcoidosis and those with nonprogressive disease. Measuring the sTNF-α level could provide new insights into personalized management of sarcoidosis.