Abstract
Background
The recently developed Delphi consensus provides a simplified high-resolution computed tomography (HRCT) phenotype classification system for pulmonary sarcoidosis. This study aimed to investigate the clinical value of HRCT phenotype classification from the Delphi consensus study on pulmonary sarcoidosis in treatment‑naïve patients and to identify potential refinements for classifying early‑stage sarcoidosis.
Methods
This retrospective single-center study included 278 treatment‑naïve patients with pulmonary sarcoidosis. HRCT signs were scored, and clinical data were recorded for all patients. Patients were followed up to observe the treatment effect, recurrence, and onset of pulmonary fibrosis. Associations between HRCT findings, pulmonary function tests (PFTs), early corticosteroid response, recurrence, and fibrosis progression were analyzed. The imaging data were then classified into seven Delphi-defined HRCT phenotypes to evaluate their clinical and prognostic relevance.
Results
HRCT signs, including nodules, consolidation, and ground-glass opacity (GGO), were significantly associated with decreased PFT parameters. The presence of central consolidation and nodules was predictive of a favorable early treatment response, while the presence of coarse granular GGOs was associated with a significantly higher risk of recurrence. In terms of Delphi HRCT phenotype, there were 196 patients with the nonfibrotic phenotype (70.5%) and 28 patients with likely-to-be fibrotic phenotypes (10.1%) in our cohort. However, there were unclassifiable cases, including those with isolated lymphadenopathy (12.2%), GGO (5.4%), and pleural effusion‑dominant (1.8%) patterns. Likely-to-be fibrotic phenotypes were associated with the greatest impairment of pulmonary function and the highest risk of fibrosis. The consolidation-dominant type was also associated with an increased risk of fibrosis. According to Cox proportional hazards regression, the isolated lymphadenopathy type was associated with a lower risk of recurrence.
Conclusions
The Delphi HRCT phenotype classification is clinically useful but may not be sufficient to fully characterize early-stage pulmonary sarcoidosis. The isolated lymphadenopathy and GGO-dominant types should be considered in the future framework due to their distinct functional and prognostic relevance.
Keywords: Sarcoidosis, high-resolution computed tomography (HRCT), imaging phenotypes, pulmonary fibrosis
Introduction
Sarcoidosis is a multisystem granulomatous disorder, with pulmonary involvement occurring in approximately 90% of patients who have this condition (1,2). Pulmonary involvement may include the airways, lung parenchyma, pleura, and mediastinal lymph nodes, with certain imaging patterns correlating with clinical outcome (3,4). Lymphadenopathy and pulmonary parenchymal involvement are characteristic radiologic features of sarcoidosis. However, the heterogeneity of morphologic manifestations on imaging poses significant challenges to the accurate diagnosis of atypical cases. More than 50% of patients with pulmonary sarcoidosis require therapeutic intervention, and a delayed diagnosis may incur unnecessary costs. In addition, approximately 10–30% of these patients progress to chronic, nonremitting, and severe forms of the disease, including progressive pulmonary fibrosis and recurrent episodes, leading to respiratory failure and mortality (5,6). The underlying mechanisms and the factors responsible for the unfavorable outcomes among these patients remain unclear.
The chest X-ray staging system proposed by Scadding in 1958 provides some prognostic value but does not delineate phenotypic subgroups or facilitate early diagnosis (7). Crucially, it applies only to confirmed cases and lacks diagnostic precision. High-resolution computed tomography (HRCT) has demonstrably superior diagnostic accuracy compared to chest radiography and provides valuable insights into disease activity and prognosis (8). Although several studies have introduced HRCT-based scoring systems, the complexity and limited clinical utility of semiquantitative methods have restricted their adoption in practice (9,10). In 2024, an expert consensus for the simplified classification of pulmonary sarcoidosis phenotypes was published in The Lancet (11). This system distinguishes nonfibrotic from likely-to-be-fibrotic disease and classifies imaging manifestations into seven distinct phenotypes. Evidence from an independent study demonstrated a relatively high reproducibility of this classification system, with an intraclass correlation coefficient (ICC) of 0.69 (12). However, no study has evaluated the utility of this classification in relation to pulmonary function impairment or its association with long-term outcomes.
In our study, we assessed the clinical applicability of the Delphi classification for pulmonary sarcoidosis, especially in treatment‑naïve patients. We further examined the associations between Delphi-defined phenotypes, disease course, and therapeutic outcomes. By integrating HRCT patterns with clinical and functional data, we sought to determine the classification’s value in prognostic stratification and imaging interpretation and thus to develop individualized treatment strategies for patients with pulmonary sarcoidosis. We present this article in accordance with the STROBE reporting checklist (available at https://qims.amegroups.com/article/view/10.21037/qims-2026-0866/rc).
Methods
Study population
This retrospective single-center study included 278 consecutive treatment-naïve patients with pulmonary sarcoidosis who were admitted to China-Japan Friendship Hospital, National Center for Respiratory Medicine, a tertiary referral center, between June 2016 and June 2025 for diagnostic evaluation and baseline clinical assessment. The patients met the current clinical and histopathological diagnostic criteria for pulmonary sarcoidosis. The inclusion criteria included the following: (I) endobronchial ultrasound-guided transbronchial needle aspiration or transbronchial lung biopsy indicating nonnecrotizing granulomatous inflammation but no evidence of tuberculosis or other granulomatous diseases; (II) laboratory tests excluding active tuberculosis and nontuberculous mycobacteria; (III) availability of HRCT imaging at the time of hospitalization; and (IV) definitive diagnosis of sarcoidosis from a multidisciplinary team. Meanwhile, the exclusion criteria were as follows: (I) previous corticosteroid therapy or other treatments for sarcoidosis before initial diagnosis; (II) coexisting infectious diseases, severe systemic diseases, or cancer; and (III) poor-quality imaging with significant motion artifacts. All patients who satisfied the inclusion and exclusion criteria during the study period were included. No formal sample size calculation was performed because of the retrospective observational design. Demographic and clinical data were collected from medical records. Patients were followed up to assess treatment response, recurrence, and fibrosis. This study was conducted in accordance with the Declaration of Helsinki and its subsequent amendments and was approved by the Institutional Review Board of China-Japan Friendship Hospital (No. 2021-125-K83; date of approval: August 31, 2021). The requirement for informed consent was waived due to the retrospective nature of the analysis. A flowchart of the study is presented in Figure 1.
Figure 1.
Flowchart of the study. HRCT, high-resolution computed tomography; MDT, multidisciplinary team; NTM, nontuberculous mycobacteria; TB, tuberculosis; TBLB, transbronchial lung biopsy; TBNA, transbronchial needle aspiration.
Analysis and classification of imaging features
Baseline HRCT scans were acquired at the time of initial diagnosis and before initiation of any therapy. To reduce observer bias, all images were independently reviewed by two board-certified radiologists, one with more than 5 years and the other with more than 15 years of experience in thoracic imaging. Both reviewers were blinded to all clinical information. Disagreements were resolved via discussion or adjudication by a senior thoracic radiologist. Imaging features were assessed according to a predefined scoring system (Table 1 and Figure 2), adapted from the framework proposed by van den Heuvel et al. (13). Central consolidation was defined as a hilar-centered area of consolidation, typically exhibiting an air bronchogram. Peripheral consolidation was defined as consolidation abutting the pleura, with a broad base directly continuous with the pleural surface. The extent of parenchymal involvement was graded separately for each pulmonary lobe as follows: 0, absent; 1, 1–25%; 2, 26–50%; 3, 51–75%; and 4, 76–100%. Micronodular (<3 mm) burden in terms of lobar involvement was categorized according to the following scheme: 0, minimal (0–2 micronodules); 1, mild (3–10 scattered micronodules); 2, moderate (11–50 micronodules); and 3, severe (>50 micronodules with a tendency to coalesce). The lingula was regarded as a separate lobe; therefore, the total lobar score for both lungs ranged from 0 to 24. Nodules larger than 3 mm were quantified via artificial intelligence (AI)-based pulmonary nodule detection software (uAI; United Imaging Intelligence, Shanghai, China). All AI-detected nodules were reviewed by the radiologists, and false-positive detections were excluded.
Table 1. Scoring system of HRCT signs in sarcoidosis.
| HRCT sign | Value/unit | Meaning |
|---|---|---|
| Classification variable | ||
| Lymphadenopathy | 0 | Absent |
| 1 | Nonconfluent lymph node enlargement | |
| 2 | Confluent lymph node enlargement | |
| Septal thickening pattern | 0 | Absent |
| 1 | Present | |
| Cavity | 0 | Absent |
| 1 | Present | |
| Bronchovascular bundle thickening | 0 | Absent |
| 1 | Proximal bronchovascular bundle thickening (inner 1/3 lung zone) involvement | |
| 2 | Distal bronchovascular bundle thickening (outer 2/3 lung zone) involvement | |
| Pleural effusion | 0 | Absent |
| 1 | Unilateral | |
| 2 | Bilateral/pericardial | |
| Mosaic sign | 0 | Absent |
| 1 | Present | |
| GGO pattern | 0 | Absent |
| 1 | Coarse granular GGO | |
| 2 | Diffuse fine GGO | |
| Architecture distortion | 0 | Absent |
| 1 | Present | |
| Continuous variable | ||
| Short-axis diameter of the largest lymph node | mm | Short-axis diameter of the largest mediastinal lymph node |
| Micronodule | Total score | Score per lobe: 0, minimal (0–2 micronodules); 1, mild (3–10 micronodules); 2, moderate (11–50 micronodules); and 3, severe (>50 micronodules with a tendency to coalesce) |
| Nodule | Count | Diameter >3 mm |
| Central consolidation | Total score | Score per lobe: 0, absent; 1, 1–25% involvement; 2, 26–50% involvement; 3, 51–75% involvement; and 4, 76–100% involvement |
| Peripheral consolidation | Total score | Same as above |
| Interlobular septal thickening | Total score | Same as above |
| GGO | Total score | Same as above |
The total score represents the sum of the individual lobe scores. GGO, ground-glass opacity; HRCT, high-resolution computed tomography.
Figure 2.
Different HRCT signs of sarcoidosis. (A) Nonconfluent lymph node enlargement (red arrow). (B) Confluent lymph node enlargement (red arrow). (C) Nodules (red arrows). (D) Central consolidation (red arrow) and micronodules (yellow arrow). (E) Peripheral consolidation (red arrow). (F) Septal thickening (red arrow). (G) Coarse granular GGOs (red arrow). (H) Diffuse fine GGOs (red arrow). (I) Proximal bronchovascular bundle thickening (red arrow) and distal bronchovascular bundle thickening (yellow arrow). (J) Cavity (red arrow). (K) Pleural effusion (red arrow). (L) Air trapping (red arrow). GGO, ground-glass opacity; HRCT, high-resolution computed tomography.
The CT acquisition parameters were as follows: a tube voltage of 120 kVp, automatic tube current modulation, detector collimation with an acquired slice thickness of 0.5 mm or 0.625 mm, images reconstructed with a slice thickness/increment of 1.0 mm/1.0 mm or 1.25 mm/1.25 mm, and a matrix size of 512×512. Iterative reconstruction was applied with a high-spatial-frequency lung kernel (FC86 or “Lung”). The field of view was automatically adjusted according to each patient’s body size.
Clinical assessment and outcomes of patients with pulmonary sarcoidosis
Comprehensive laboratory evaluations included measurement of serum angiotensin-converting enzyme (sACE) levels. Pulmonary function tests (PFTs) assessed forced vital capacity (FVC), forced expiratory volume in one second (FEV1), the FEV1/FVC ratio, and diffusing capacity for carbon monoxide (DLCO), all of which are expressed as percentages of predicted values. Bronchoalveolar lavage fluid (BALF) samples were collected, and lymphocyte counts were conventionally analyzed.
For patients with clinical indications for corticosteroid therapy, treatment was generally initiated shortly after diagnostic confirmation and completion of baseline clinical assessment during the initial management period. Early treatment response was assessed using HRCT performed 1 month after corticosteroid initiation. According to the HRCT signs, treatment effects were classified into improved (>30% radiographic resolution) and persistent disease (radiographic resolution <30%, lack of regression of existing lesions, or emergence of new lesions). Patients were followed up at regular intervals according to standard clinical protocols. Patients were monitored for a minimum of 6 months to assess recurrence, which was defined as the development of new lesions or enlargement of pre-existing lesions on follow-up HRCT. To determine which baseline imaging features are more likely to progress to fibrosis, we evaluated new-onset fibrosis on serial CT scans in patients with a minimum follow-up of 1 year. New-onset fibrosis was defined as the appearance of traction bronchiectasis or architectural distortion, either focal or extensive.
Validation of Delphi consensus HRCT phenotypes for pulmonary sarcoidosis
According to the Delphi consensus, we allocated the patients’ HRCT imaging into seven distinct phenotypes (Figure 3). The seven phenotypes included nonfibrotic and likely-to-be-fibrotic phenotypes. The nonfibrotic subtypes included: (I) multiple peribronchovascular, perifissural, or subpleural micronodules (Type 1); (II) multiple larger peribronchovascular nodules (Type 2); (III) scattered larger nodules (Type 3); and (IV) consolidation as the predominant or sole abnormality (Type 4). The likely-to-be-fibrotic subtypes included: (I) bronchocentric reticulation with or without dense parenchymal opacification and without cavitation (Type 5); (II) bronchocentric reticulation and dense parenchymal opacification with cavitation (Type 6); and (III) large bronchocentric masses (Type 7).
Figure 3.
Representative examples of the Delphi consensus HRCT phenotypes of pulmonary sarcoidosis. (A) Multiple peribronchovascular, perifissural, or subpleural micronodules (Type 1). (B) Multiple larger peribronchovascular nodules (Type 2). (C) Scattered larger nodules (Type 3). (D) Consolidation as the predominant or sole abnormality (Type 4). (E) Bronchocentric reticulation with or without dense parenchymal opacification, without cavitation (Type 5). (F) Bronchocentric reticulation and dense parenchymal opacification, with cavitation (Type 6). (G) Large bronchocentric masses (Type 7). (H) Isolated lymphadenopathy type. (I) GGO-dominant type. GGO, ground-glass opacity; HRCT, high-resolution computed tomography.
Statistical analysis
HRCT signs were systematically analyzed, and interobserver agreement was assessed via the ICC. Multivariate linear regression was applied to examine correlations between HRCT findings and PFT results. Binary logistic regression was conducted to identify imaging predictors of new-onset fibrosis, recurrence, and treatment response.
After patients were allocated into subtype groups, intergroup differences in laboratory indices and pulmonary function parameters were assessed. Categorical variables were compared via the chi‑squared or Fisher exact test, while continuous variables were compared via one‑way analysis of variance or the Kruskal-Wallis test, as appropriate. Treatment response was assessed via logistic regression, and Cox proportional hazards models were used to analyze recurrence and new‑onset fibrosis, with adjustments made for clinical covariates. Time-to-event outcomes were calculated from the date of diagnosis to the occurrence of recurrence or new-onset fibrosis, with the last available follow-up defined as the censoring point. A two-sided P value <0.05 was considered statistically significant. Analyses were performed with available data, and denominators are reported for variables with missing data. No imputation was performed.
Results
A total of 278 patients were included in the study (85 males and 193 females; mean age 50.0 years; age range 24–79 years). Among these patients, 57 (20.5%) had a smoking history, and 42 (15.1%) reported environmental exposure. PFT data were available for 242 patients, with 114 (47.1%) and 16 (6.6%) of these patients exhibiting impaired DLCO (<80% predicted) and obstructive defects (FEV1/FVC <70% postbronchodilator), respectively. Elevated sACE levels were found in 148 of 249 (59.4%) patients. Among patients with data from BALF analysis (n=205), 111 (54.1%) had a CD4+/CD8+ ratio >3.5.
Corticosteroid therapy was applied in 200 of 278 patients (71.9%). Early treatment response, assessed using HRCT performed 1 month after corticosteroid initiation, was evaluated in 167 patients, of whom 130 (77.8%) achieved improvement. During follow-up, recurrence occurred in 45 of 163 patients (27.6%), with a median time to event of 23.1 months (range, 4.7–73.3 months). Signs of new-onset fibrosis were observed in 75 of 159 patients (47.2%), with a median time to event of 22.2 months (range, 1.26–90.9 months). The detailed characteristics of the patients are provided in Table 2.
Table 2. Patient characteristics.
| Characteristic | Values (N=278) |
|---|---|
| Age (years) | 50.04 [24–79] |
| Sex | |
| Male | 85 (30.6) |
| Female | 193 (69.4) |
| Smoking history | 57 (20.5) |
| Environmental exposure | 42 (15.1) |
| PFT available | 242 (87.1) |
| Impaired DLCO | 114 (47.1) |
| Obstructive defects | 16 (6.6) |
| sACE available | 249 (89.6) |
| sACE elevated | 148 (59.4) |
| BALF available | 205 (73.7) |
| CD4+/CD8+>3.5 | 111 (54.1) |
| Treatment | |
| Received treatment | 200 (71.9) |
| Treatment response assessed | 167 (60.1) |
| Improved | 130 (77.8) |
| Nonresponse | 37 (22.2) |
| Recurrence (n=163) | 45 (27.6) |
| New-onset fibrosis (n=159) | 75 (47.2) |
Data are presented as number (percentage) or mean [range]. BALF, bronchoalveolar lavage fluid; CD4+/CD8+, ratio of CD4-positive to CD8-positive T lymphocytes; DLCO, diffusing capacity for carbon monoxide; PFT, pulmonary function test; sACE, serum angiotensin-converting enzyme.
HRCT findings
Mediastinal lymphadenopathy was the most common abnormality, appearing in 259 of 278 (93.2%) patients. Micronodules were observed in 209 patients (75.2%) and nodules in 143 (51.4%). Peripheral and central consolidations were present in 134 (48.2%) and 95 patients (34.2%), respectively. Ground-glass opacity (GGO) was uncommon (n=36, 12.9%) and included coarse granular and diffuse fine patterns. Additional findings included bronchovascular bundle thickening (20.9%), pleural effusion (10.4%), mosaic sign (6.8%), and cavitation (3.6%). ICCs ranged from 0.48 to 0.95, indicating moderate-to-excellent inter-observer agreement. The detailed quantitative results are provided in Table 3.
Table 3. Baseline descriptive analysis of HRCT findings.
| Imaging feature | Prevalence | Total score | ICC |
|---|---|---|---|
| Lymphadenopathy | 259 (93.2) | 0.57 | |
| Nonconfluent lymph node enlargement | 195 (70.1) | ||
| Confluent lymph node enlargement | 64 (23.0) | ||
| Maximum diameter (mm) | – | 17.36±7.76 (0–40) | 0.95 |
| Nodule count | |||
| Micronodules (diameter <3 mm) | 209 (75.2) | 5.17±5.58 (0–18) | 0.87 |
| Nodules (diameter >3 mm) | 143 (51.4) | 3.38±5.74 (0–33) | |
| Consolidation | |||
| Central consolidation | 95 (34.2) | 0.95±1.80 (0–11) | 0.75 |
| Peripheral consolidation | 134 (48.2) | 1.33±1.86 (0–8) | 0.64 |
| Septal thickening | 128 (46.0) | 0.75 | |
| Septal thickening count | – | 1.85±2.80 (0–15) | 0.76 |
| GGO pattern | 36 (12.9) | 0.66 | |
| Coarse granular GGO | 20 (7.2) | ||
| Diffuse fine GGO | 16 (5.8) | ||
| GGO count | – | 0.97±3.30 (0–18) | 0.48 |
| Bundle thickening | 58 (20.9 ) | – | 0.69 |
| Proximal bronchial thickening | 41 (14.7) | ||
| Distal bronchial thickening | 17 (6.1) | ||
| Other features | |||
| Pleural effusion | 29 (10.4) | – | 0.93 |
| Unilateral | 14 (5.0) | ||
| Bilateral/pericardial | 15 (5.4) | ||
| Mosaic sign | 19 (6.8) | – | 0.64 |
| Cavity | 10 (3.6) | 0.79 |
Data are presented as number (percentage) or mean ± standard deviation (range). ICC values >0.60 indicate good agreement. GGO, ground-glass opacity; HRCT, high-resolution computed tomography; ICC, intraclass correlation coefficient.
Associations of HRCT findings with pulmonary function and clinical outcomes
Scores of micronodules, central consolidation, peripheral consolidation, and pleural effusion independently predicted a reduced DLCO (P<0.05). In addition, the presence of cavities and GGOs was a strong negative predictor of DLCO (P<0.001). For FEV1, significant negative predictors included central consolidation (P<0.01), peripheral consolidation (P<0.01), and cavities (P<0.05). A decline in FVC was associated with nodules (P<0.05), central consolidation (P<0.05), peripheral consolidation (P<0.001), and cavities (P<0.01). The correlation of HRCT findings with pulmonary function is summarized in Table 4.
Table 4. Significant predictors of pulmonary function parameters according to multiple linear regression.
| Predictor | DLCO (𝛽) | FEV1/FVC (𝛽) | FEV1 (𝛽) | FVC (𝛽) |
|---|---|---|---|---|
| Micronodules | −0.468* | – | – | – |
| Nodules | – | – | – | −0.426* |
| Central consolidation | −1.618* | — | −2.727** | −2.120* |
| Peripheral consolidation | −1.558* | −1.017* | −2.191** | −2.450*** |
| Cavity | −23.074** | – | −15.562* | −20.723** |
| Pleural effusion | −7.367* | – | – | – |
| GGO count | −1.561*** | – | – | – |
*, P<0.05; **, P<0.01; ***, P<0.001. β, regression coefficient; DLCO, diffusing capacity for carbon monoxide; FEV1, forced expiratory volume in 1 second; FVC, forced vital capacity; GGO, ground-glass opacity.
Multivariable logistic regression identified the independent predictors of favorable treatment response to be central consolidation [odds ratio (OR) 1.34; 95% confidence interval (CI): 1.01–1.77; P=0.043] and nodules (OR 1.095; 95% CI: 1.03–1.17; P=0.005).
Coarse granular GGOs were significantly associated with an elevated recurrence risk (OR 12.26; 95% CI: 1.66–90.56; P=0.014). Independent risk factors for fibrosis included higher scores of micronodules, central consolidation, and peripheral consolidation (P<0.05). The results of the regression are shown in Table 5.
Table 5. Associations between HRCT imaging signs and clinical outcomes in patients with sarcoidosis.
| Outcome | Predictor | B (SE) | P value | OR (95% CI) |
|---|---|---|---|---|
| Treatment response | Nodules | 0.09 (0.03) | 0.005 | 1.10 (1.03-1.17) |
| Central consolidation | 0.29 (0.14) | 0.043 | 1.34 (1.01–1.77) | |
| Recurrence | Coarse granular ground-glass opacities | 2.51 (1.02) | 0.014 | 12.26 (1.66–90.56) |
| New-onset fibrosis | Central consolidation | 0.85 (0.26) | 0.001 | 2.34 (1.41–3.90) |
| Peripheral consolidation | 0.49 (0.14) | <0.001 | 1.63 (1.24–2.13) | |
| Micronodules | 0.14 (0.05) | 0.005 | 1.14 (1.04–1.26) |
B, regression coefficient; CI, confidence interval; HRCT, high-resolution computed tomography; OR, odds ratio; SE, standard error.
HRCT phenotypes of sarcoidosis according to the Delphi consensus
The cohort of 278 patients with pulmonary sarcoidosis was classified by two independent reviewers into seven imaging subtypes according to the Delphi consensus. In cases of disagreement, the final classification was adjudicated by a senior thoracic radiologist with more than 20 years of experience. Among untreated patients, the most common phenotype was Type 1 (n=94, 33.8%), followed by Type 4 (n=46, 16.5%), and Type 3 (n=45, 16.2%). The least frequent phenotype was Type 6 (n=2, 0.7%). Among patients in the “unable to classify” category (n=54, 19.4%), 34 (12.2%) exhibited isolated lymph node enlargement without pulmonary parenchymal involvement. Moreover, 5 (1.8%) patients had lymphadenopathy with concurrent pleural effusion, and 15 (5.4%) presented with lymphadenopathy accompanied by GGOs. The detailed demographic and clinical characteristics of each subgroup are presented in Table 6.
Table 6. Baseline characteristics of patients according to the Delphi classification.
| Delphi category | Total | Age (years) | Female |
|---|---|---|---|
| Nonfibrotic subtypes | 196 (70.5) | ||
| Type 1 | 94 (33.8) | 51.1±11.24 | 65 (69.1) |
| Type 2 | 11 (4.0) | 48.26±10.30 | 8 (72.7) |
| Type 3 | 45 (16.2) | 48.22±10.49 | 30 (66.7) |
| Type 4 | 46 (16.5) | 49.57±10.65 | 32 (69.6) |
| Likely-to-be-fibrotic subtypes | 28 (10.1) | ||
| Type 5 | 10 (3.6) | 55.2±12.35 | 6 (60.0) |
| Type 6 | 2 (0.7) | 60.00±5.66 | 2 (100.0) |
| Type 7 | 16 (5.8) | 58.25±10.52 | 11 (68.8) |
| Unable to classify | 54 (19.4) | 46.53±12.07 | 39 (72.2) |
| Total | 278 (100.0) | – | 193 (69.4) |
Data are presented as number (percentage) or mean ± standard deviation.
Utility of the HRCT subtypes in clinical practice
Due to the small number of cases in each likely-to-be-fibrotic phenotype, Types 5, 6, and 7 were merged into a single likely-to-be-fibrotic group for statistical analysis. In addition, cases that could not be classified according to the Delphi system were also included in the clinical analyses.
Pulmonary function indices demonstrated significant variation across the imaging subtypes of pulmonary sarcoidosis. The likely-to-be-fibrotic subtypes exhibited the greatest impairment in FVC, FEV1, and DLCO compared with the other phenotypes (P<0.05). The GGO-dominant type also demonstrated reduced DLCO (P<0.05). In contrast, the isolated lymphadenopathy type and Type 1 exhibited preserved lung function. The FEV1/FVC ratio did not differ significantly among the HRCT phenotypes (P=0.204). The likely-to-be-fibrotic type had the highest level of sACE, and the sACE level of Type 1 was higher than that of Type 3. A comparative summary of pulmonary function across HRCT phenotypes is provided in Figure 4.
Figure 4.
Comparison of laboratory test among different HRCT phenotypes of patients with sarcoidosis. Bar plots show mean values of (A) FVC, (B) FEV1, (C) FEV1/FVC, (D) DLCO, and (E) sACE across the seven radiological phenotypes. *, P<0.05; **, P<0.01; ***, P<0.001 (compared between groups as indicated by the post hoc test). DLCO, diffusing capacity for carbon monoxide; FEV1, forced expiratory volume in 1 second; FVC, forced vital capacity; GGO-Dt, ground-glass opacity-dominant type; HRCT, high-resolution computed tomography; L-Dt, isolated lymphadenopathy type; LTBF, likely to be fibrotic; sACE, serum angiotensin-converting enzyme.
Binary logistic regression identified imaging subtype as a significant predictor of treatment outcome (P<0.05). In the multivariable model, Type 3 was independently associated with a favorable response (OR =5.94; 95% CI: 1.44–24.55; P=0.014).
Cox regression showed that isolated lymphadenopathy type was associated with a lower recurrence risk [hazard ratio (HR) =0.36, 95% CI: 0.12–0.93; P=0.048]. Several HRCT phenotypes were independently associated with fibrosis risk. Compared with Type 1, Type 3 (HR =0.143, 95% CI: 0.023–0.497; P=0.009) and isolated lymphadenopathy type (HR =0.076, 95% CI: 0.004–0.368; P=0.012) were associated with reduced risk, whereas Type 4 (HR =2.004, 95% CI: 1.115–3.673; P=0.021) and the likely-to-be fibrotic type (HR =3.739, 95% CI: 1.844–7.535; P<0.001) were associated with increased risk. The HRCT phenotype model could predict fibrosis (likelihood ratio, χ2=81.70; P<0.0001), and Kaplan-Meier curves (Figure 5) demonstrated distinct survival differences between the phenotypes.
Figure 5.
Survival curve for recurrence (A) and fibrosis (B) for patients with different HRCT phenotypes of sarcoidosis. GGO, ground-glass opacity; HRCT, high-resolution computed tomography; LRT, likelihood ratio test.
Discussion
The purpose of this study was to evaluate the clinical utility and prognostic value of the recently established Delphi consensus classification for pulmonary sarcoidosis in our cohort. Interstitial nodule burden and parenchymal consolidation burden were both found to correlate with restrictive ventilation and impaired diffusive pulmonary function and are also risk factors for the development of pulmonary fibrosis. Patients that were defined as having the likely-to-be-fibrotic subtypes accounted for only a small proportion of the cohort, but showed the highest risk of new-onset fibrosis. Meanwhile, Type 4 was also associated with a significantly increased risk of fibrosis. In addition, the GGO-dominant pattern, which is not defined as a distinctive type in the Delphi consensus classification, was significantly associated with a reduced diffusive capacity and a higher risk of recurrence.
Regression analysis indicated that micronodule burden was associated with impaired DLCO. This finding can be explained by the distribution of granulomas along bronchovascular bundles and within the interstitium, where diffuse involvement may restrict gas exchange and reduce DLCO (6,7,14). Coarse GGOs were also associated with a decline in DLCO, possibly reflecting fine granulomatous accumulation and subtle fibrosis (15). Meanwhile, central and peripheral parenchymal consolidations were associated with a lower FVC and FEV1, likely reflecting airway involvement and consequent ventilatory impairment (4,7,16). Peripheral consolidations were also associated with reduced DLCO and obstructive ventilatory defects, possibly due to the compression of the peripheral small airways by granulomatous infiltration (17). Cavitary lesions were strongly associated with pulmonary function impairment, likely resulting from granulomatous airway narrowing and a ball-valve mechanism that led to air trapping and parenchymal destruction (18). It was further found that pleural effusion was also associated with impaired DLCO. This association may be attributable to concomitant parenchymal involvement and the reduced effective lung volume caused by pleural effusion.
In our study, early treatment response was assessed at 1 month. This time point was chosen because systemic prednisolone is typically initiated at moderate doses for approximately 4 weeks, after which tapering is performed according to disease response (19). Since there are no standardized radiological criteria for evaluating treatment response in patients with pulmonary sarcoidosis, we adapted the principles of the Response Evaluation Criteria in Solid Tumors (RECIST) to objectively define radiologic improvement in our cohort (20). With respect to early treatment response, central consolidations and nodules were associated with more favorable outcomes following oral glucocorticoid therapy. Murdoch et al. reported that nodular lesions are the most likely type to regress after treatment (21), and other studies have confirmed that consolidations may also completely resolve following therapy (7). Nonetheless, recurrence occurred in a proportion of patients, although the reasons for this remain unclear. Extensive parenchymal involvement, particularly coarse granular GGO, was linked to an increased risk of recurrence. Moreover, progression to fibrosis was independently associated with higher scoring for micronodules, central consolidations, and peripheral consolidations. In a previous study of 74 patients with sarcoidosis, patients with nodules and consolidations on HRCT had elevated serum cancer antigen 15-3 (CA15-3) levels, which may be a biomarker for progressive fibrosis (22,23).
In our cohort of treatment‑naïve patients, the likely-to-be-fibrotic subtype accounted for only 10.1% of cases. As expected, patients with this subtype exhibited ventilation impairment and reduced diffusive capacity. These patients also had a high risk of fibrosis progression. Our results support the clinical and prognostic relevance of the Delphi classification. However, patients with the likely-to-be-fibrotic subtype accounted for a small portion of our cohort, and this may reflect a late stage of sarcoidosis. Nonfibrotic patterns predominated at the first visit but had variable clinical relevance. Interestingly, while patients with Type 1 had relatively preserved FVC and FEV1, Type 4 within the nonfibrotic group was associated with pulmonary function decline. These findings suggest that pulmonary function, particularly diffusive capacity, deteriorates as radiologic severity increases. Furthermore, sACE levels in patients with Type 1 disease were significantly elevated compared to those in patients with Type 3, which is likely due to the greater severity of active granulomatous inflammation and a higher total granulomatous burden during the earlier stages of the disease (24,25). In terms of treatment response, patients with a predominant scattered nodular pattern exhibited relatively good outcomes with corticosteroid therapy, consistent with earlier reports (21). However, no significant differences were observed for treatment response between the other phenotypes, suggesting that the Delphi classification may have limited predictive value for short-term clinical outcomes. Overall, the Delphi classification demonstrated good model fit for the progression of fibrosis. This highlights the potential value of the Delphi system in identifying patients at increased risk for fibrosis and supports its prognostic relevance for longitudinal monitoring. Notably, although Type 4 is categorized as a nonfibrotic phenotype in the Delphi system, our findings suggest it may be associated with a high risk of fibrosis, warranting closer monitoring. A substantial proportion of patients (19.4%) could not be classified within the existing Delphi framework, yet many of these cases exhibited imaging features with clear prognostic relevance. These included isolated lymphadenopathy without parenchymal involvement, corresponding to Scadding Stage I. In our cohort, this phenotype was associated with a lower risk of recurrence. Another subset presented with diffuse or patchy GGO as the predominant abnormality. Although the 70% threshold for consensus classification in the Delphi study was not reached for GGOs, agreement was 69% after two survey rounds, suggesting substantial reproducibility. In our cohort, the GGO-dominant type correlated with significantly reduced DLCO, supporting its consideration as a clinically relevant imaging phenotype. Additionally, a small number of patients presented with mediastinal lymphadenopathy accompanied by pleural effusion, a pattern not included in the current consensus categories. However, previous studies suggest that pleural effusion may indicate pleural involvement and adverse outcomes (3,17). Given their clinical impact, these unclassified imaging patterns warrant consideration for inclusion in future refinements of phenotypic schemes.
Limitations
This study involved several limitations that should be noted. First, as we employed a single-center design with a relatively small sample size, the risk of errors in subgroup comparisons was high. Second, the retrospective nature of the study resulted in variations in follow-up intervals, potentially hampering the accurate assessment of clinical endpoints. Third, because all patients were recruited from an inpatient cohort at a tertiary referral center, there was potential selection bias, and patients with milder disease managed exclusively in outpatient settings may be underrepresented. Finally, bronchoscopic mucosal findings were not systematically incorporated into the phenotyping analysis, although such findings may reflect airway involvement not fully captured by HRCT. The exclusion of these factors may undermine the reliability of the conclusions. Despite these limitations, our cohort was derived from the National Center for Respiratory Medicine in China and, as a highly enriched sample, may represent a broad regional population. Future prospective multicenter studies with standardized imaging acquisition and longitudinal follow-up are needed to confirm these findings and refine the Delphi classification. Such an expanded framework would transform the Delphi consensus from a valuable classification system into a more robust clinical tool capable of guiding individualized risk stratification and management for all patients with pulmonary sarcoidosis.
Conclusions
This study validates the clinical utility of the Delphi HRCT phenotypic classification for pulmonary sarcoidosis and proposes a broader framework for future precision subtyping.
The recognition of these imaging phenotypes is critical for enhancing diagnostic precision, enabling individualized prognostication, and guiding targeted follow-up strategies. Future multicenter prospective studies should be conducted to validate and refine the classification system, ensuring broader applicability across a greater diversity patient populations, especially for patients with early-stage pulmonary sarcoidosis.
Supplementary
The article’s supplementary files as
Acknowledgments
None.
Ethical Statement: The authors are accountable for all aspects of the work in ensuring that questions related to the accuracy or integrity of any part of the work are appropriately investigated and resolved. The study was conducted in accordance with the Declaration of Helsinki and its subsequent amendments. The study was approved by the Institutional Review Board of China-Japan Friendship Hospital (No. 2021-125-K83; date of approval: August 31, 2021), and the requirement for informed consent was waived due to the retrospective nature of the analysis.
Footnotes
Reporting Checklist: The authors have completed the STROBE reporting checklist. Available at https://qims.amegroups.com/article/view/10.21037/qims-2026-0866/rc
Funding: None.
Conflicts of Interest: All authors have completed the ICMJE uniform disclosure form (available at https://qims.amegroups.com/article/view/10.21037/qims-2026-0866/coif). The authors have no conflicts of interest to declare.
Data Sharing Statement
Available at https://qims.amegroups.com/article/view/10.21037/qims-2026-0866/dss
References
- 1. Rossides M, Darlington P, Kullberg S, Arkema EV. Sarcoidosis: Epidemiology and clinical insights. J Intern Med 2023;293:668-80.. 10.1111/joim.13629 [DOI] [PubMed] [Google Scholar]
- 2. Zhao M, Zhou Y. Diagnosis of Pulmonary Sarcoidosis. Clin Chest Med 2024;45:15-24.. 10.1016/j.ccm.2023.06.005 [DOI] [PubMed] [Google Scholar]
- 3. Chopra A, Foulke L, Judson MA. Sarcoidosis associated pleural effusion: Clinical aspects. Respir Med 2022;191:106723.. 10.1016/j.rmed.2021.106723 [DOI] [PubMed] [Google Scholar]
- 4. Belperio JA, Fishbein MC, Abtin F, Channick J, Balasubramanian SA, Lynch Iii JP. Pulmonary sarcoidosis: A comprehensive review: Past to present. J Autoimmun 2024;149:103107.. 10.1016/j.jaut.2023.103107 [DOI] [PubMed] [Google Scholar]
- 5. Obi ON, Saketkoo LA, Maier LA, Baughman RP. Developmental drugs for sarcoidosis. J Autoimmun 2024;149:103179.. 10.1016/j.jaut.2024.103179 [DOI] [PubMed] [Google Scholar]
- 6. Drent M, Crouser ED, Grunewald J. Challenges of Sarcoidosis and Its Management. N Engl J Med 2021;385:1018-32.. 10.1056/NEJMra2101555 [DOI] [PubMed] [Google Scholar]
- 7. Bailey GL, Wells AU, Desai SR. Imaging of Pulmonary Sarcoidosis-A Review. J Clin Med 2024;13:822.. 10.3390/jcm13030822 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 8. Zhang Y, Du SS, Zhao MM, Li QH, Zhou Y, Song JC, Chen T, Shi JY, Jie B, Li W, Shen L, Zhang F, Su YL, Hu Y, Lower EE, Baughman RP, Li H. Chest high-resolution computed tomography can make higher accurate stages for thoracic sarcoidosis than X-ray. BMC Pulm Med 2022;22:146.. 10.1186/s12890-022-01942-y [DOI] [PMC free article] [PubMed] [Google Scholar]
- 9. Benamore R, Kendrick YR, Repapi E, Helm E, Cole SL, Taylor S, Ho LP. CTAS: a CT score to quantify disease activity in pulmonary sarcoidosis. Thorax 2016;71:1161-3.. 10.1136/thoraxjnl-2016-208833 [DOI] [PubMed] [Google Scholar]
- 10. Walsh SL, Wells AU, Sverzellati N, Keir GJ, Calandriello L, Antoniou KM, Copley SJ, Devaraj A, Maher TM, Renzoni E, Nicholson AG, Hansell DM. An integrated clinicoradiological staging system for pulmonary sarcoidosis: a case-cohort study. Lancet Respir Med 2014;2:123-30.. 10.1016/S2213-2600(13)70276-5 [DOI] [PubMed] [Google Scholar]
- 11. Desai SR, Sivarasan N, Johannson KA, George PM, Culver DA, Devaraj A, Lynch DA, Milne D, Renzoni E, Nunes H, Sverzellati N, Spagnolo P, Baughman RP, Yadav R, Piciucchi S, Walsh SLF, Kouranos V, Wells AU, Sarcoid Delphi Group . High-resolution CT phenotypes in pulmonary sarcoidosis: a multinational Delphi consensus study. Lancet Respir Med 2024;12:409-18.. 10.1016/S2213-2600(23)00267-9 [DOI] [PubMed] [Google Scholar]
- 12. Van Woensel J, Krdzalic J, de Jaegere T, Janssen MTHF, Ramiro S, Magro Checa C, Landewé RBM, Mostard RLM. Radiological phenotypes in pulmonary sarcoidosis: a reliability study of newly defined high-resolution computer tomography phenotypes. BJR Open 2025;7:tzaf017.. 10.1093/bjro/tzaf017 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 13. Van den Heuvel DA, de Jong PA, Zanen P, van Es HW, van Heesewijk JP, Spee M, Grutters JC. Chest Computed Tomography-Based Scoring of Thoracic Sarcoidosis: Inter-rater Reliability of CT Abnormalities. Eur Radiol 2015;25:2558-66.. 10.1007/s00330-015-3685-4 [DOI] [PubMed] [Google Scholar]
- 14. Spagnolo P, Rossi G, Trisolini R, Sverzellati N, Baughman RP, Wells AU. Pulmonary sarcoidosis. Lancet Respir Med 2018;6:389-402.. 10.1016/S2213-2600(18)30064-X [DOI] [PubMed] [Google Scholar]
- 15. Nunes H, Uzunhan Y, Gille T, Lamberto C, Valeyre D, Brillet PY. Imaging of sarcoidosis of the airways and lung parenchyma and correlation with lung function. Eur Respir J 2012;40:750-65.. 10.1183/09031936.00025212 [DOI] [PubMed] [Google Scholar]
- 16. Aleksonienė R, Zeleckienė I, Matačiūnas M, Puronaitė R, Jurgauskienė L, Malickaitė R, Strumilienė E, Gruslys V, Zablockis R, Danila E. Relationship between radiologic patterns, pulmonary function values and bronchoalveolar lavage fluid cells in newly diagnosed sarcoidosis. J Thorac Dis 2017;9:88-95.. 10.21037/jtd.2017.01.17 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 17. Malaisamy S, Dalal B, Bimenyuy C, Soubani AO. The clinical and radiologic features of nodular pulmonary sarcoidosis. Lung 2009;187:9-15.. 10.1007/s00408-008-9118-2 [DOI] [PubMed] [Google Scholar]
- 18. Franco G, Debray MP, Anzani N, Marruchella A, Cazes A, Le Guen P, Taillé C, Faverio P, Borie R, Luppi F, Crestani B. Cystic lung in sarcoidosis: Clinico-radiologic characteristic and evolution. Respirology 2024;29:1067-76.. 10.1111/resp.14823 [DOI] [PubMed] [Google Scholar]
- 19. Grunewald J, Grutters JC, Arkema EV, Saketkoo LA, Moller DR, Müller-Quernheim J. Sarcoidosis. Nat Rev Dis Primers 2019;5:45.. 10.1038/s41572-019-0096-x [DOI] [PubMed] [Google Scholar]
- 20. Litière S, Collette S, de Vries EG, Seymour L, Bogaerts J. RECIST - learning from the past to build the future. Nat Rev Clin Oncol 2017;14:187-92.. 10.1038/nrclinonc.2016.195 [DOI] [PubMed] [Google Scholar]
- 21. Murdoch J, Müller NL. Pulmonary sarcoidosis: changes on follow-up CT examination. AJR Am J Roentgenol 1992;159:473-7.. 10.2214/ajr.159.3.1503008 [DOI] [PubMed] [Google Scholar]
- 22. Leštan Ramovš Z, Sodin-Šemrl S, Lakota K, Čučnik S, Manevski D, Zbačnik R, Zupančič M, Verbič M, Terčelj M. Correlation of the High-Resolution Computed Tomography Patterns of Intrathoracic Sarcoidosis with Serum Levels of SAA, CA 15.3, SP-D, and Other Biomarkers of Interstitial Lung Disease. Int J Mol Sci 2023;24:10794.. 10.3390/ijms241310794 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 23. Bandyopadhyay D, Mirsaeidi MS. Sarcoidosis-associated pulmonary fibrosis: joining the dots. Eur Respir Rev 2023;32:230085.. 10.1183/16000617.0085-2023 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 24. Tana C, Drent M, Nunes H, Kouranos V, Cinetto F, Jessurun NT, Spagnolo P. Comorbidities of sarcoidosis. Ann Med 2022;54:1014-35.. 10.1080/07853890.2022.2063375 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 25. Li Y, Xu G. Diagnostic Value of Imaging and Serological Biomarkers in Pulmonary Sarcoidosis. Adv Respir Med 2024;92:190-201.. 10.3390/arm92030020 [DOI] [PMC free article] [PubMed] [Google Scholar]
Associated Data
This section collects any data citations, data availability statements, or supplementary materials included in this article.
Supplementary Materials
The article’s supplementary files as
Data Availability Statement
Available at https://qims.amegroups.com/article/view/10.21037/qims-2026-0866/dss