Identification and validation of SPP1 as a prognostic biomarker in interstitial lung disease subtypes by single-cell sequencing

Abstract

Background

Interstitial lung diseases (ILDs) are a heterogeneous group of disorders with diverse clinical trajectories, highlighting the need for reliable biomarkers for prognostic assessment and risk stratification.

Methods

Single-cell RNA sequencing (scRNA-seq) was performed on explanted lung tissue from 3 patients with idiopathic pulmonary fibrosis (IPF) and 3 control subjects.

Secreted phosphoprotein 1 (SPP1) was identified and further validated in serum samples from two ILD cohorts comprising 74 patients with IPF and 92 with idiopathic inflammatory myopathy-associated ILD (IIM-ILD), as well as 56 healthy controls (HCs). Patients with IPF were further classified as acute exacerbation (AE) or stable IPF, and patients with IIM-ILD as rapidly progressive (RP) or stable IIM-ILD. The association between serum SPP1 levels and 1-year all-cause mortality was assessed using Cox proportional hazards models.

Results

scRNA-seq demonstrated that SPP1 expression was markedly upregulated in explanted lungs from IPF patients compared with controls. Serum SPP1 levels were significantly higher in patients with ILD than in HCs (P < 0.001), and were further elevated in AE-IPF (P < 0.001), RP-IIM-ILD (P < 0.001), and anti-melanoma differentiation-associated gene 5-positive IIM-ILD (P = 0.015). The 1-year mortality rates were 35.1% in patients with IPF and 31.5% in those with IIM-ILD. In multivariable Cox analyses, serum SPP1 was an independent predictor of 1-year mortality in the overall ILD cohort (hazard ratio [HR] 1.007, 95% confidence interval [CI] 1.002–1.012, P = 0.005), as well as in the IPF (HR 1.015, 95% CI 1.003–1.027, P = 0.012) and IIM-ILD (HR 1.007, 95% CI 1.001–1.012, P = 0.022) subgroups.

Conclusions

Serum SPP1 may serve as a potential biomarker for risk stratification and prognostic assessment in IPF and IIM-ILD.

Supplementary Information

The online version contains supplementary material available at https://doi.org/10.1186/s12931-026-03788-y.

Introduction

Interstitial lung diseases (ILDs) comprise a heterogeneous group of diffuse parenchymal lung disorders characterized by varying degrees of inflammation and fibrosis, which may ultimately result in progressive respiratory impairment [1].

Although the epidemiology of ILD varies across subtypes and geographic regions, recent evidence indicates a substantial and increasing global disease burden [2].

Among the various ILD subtypes, idiopathic pulmonary fibrosis (IPF) and idiopathic inflammatory myopathy-associated ILD (IIM-ILD) are clinically important entities associated with considerable morbidity and mortality [3, 4]. In particular, acute exacerbation of IPF (AE-IPF) and rapidly progressive IIM-ILD (RP-IIM-ILD) are associated with abrupt clinical deterioration and poor short-term outcomes.

Accordingly, there is a pressing need to identify reliable biomarkers that reflect disease activity and predict adverse outcomes, thereby informing follow-up strategies, therapeutic decision-making, and healthcare resource allocation.

Secreted phosphoprotein 1 (SPP1), also known as osteopontin (OPN), is a multifunctional secreted glycoprotein involved in immune regulation, macrophage activation, extracellular matrix remodelling, and tissue repair [5]. Increasing evidence suggests that SPP1 plays an important role in fibrotic lung disease and chronic inflammatory conditions, particularly through its association with activated macrophages and aberrant tissue remodelling [5, 6]. In pulmonary fibrosis, SPP1 has been linked to profibrotic macrophage populations, fibroblast activation, and extracellular matrix deposition, suggesting that it may contribute to the transition from inflammation to irreversible fibrosis [5]. Clinical studies have further supported the relevance of SPP1 in ILD. Serum OPN levels have been reported to correlate inversely with forced vital capacity (FVC), and to be significantly higher in ILD patients who died or underwent lung transplantation than in survivors [7]. In addition, proteomic profiling in anti-MDA5-positive dermatomyositis identified SPP1 as being particularly increased in the rapidly progressive subgroup of IIM-ILD [8].

In the present study, we performed single-cell RNA sequencing (scRNA-seq) on explanted lung tissue from patients with IPF and further validated our findings in two ILD cohorts, including patients with IPF and IIM-ILD. These were further stratified into AE-IPF and stable IPF, and RP-IIM-ILD and stable IIM-ILD, respectively. We then measured serum SPP1 levels in patients with IPF or IIM-ILD and healthy controls (HCs), examined their associations with clinically relevant ILD subgroups, and evaluated their prognostic value for 1-year mortality in the overall ILD cohort as well as in the IPF and IIM-ILD subgroups.

Materials and methods

Subjects

Patients diagnosed with IPF or IIM-ILD who were evaluated at a tertiary hospital between July 2018 and July 2024 were retrospectively enrolled (Figure S1). All diagnoses of interstitial pneumonia and ILD subtype classifications were established through a centralized multidisciplinary discussion (MDD), in accordance with international recommendations for ILD diagnosis [9]. The MDD panel included 2 pulmonologists experienced in ILD, 1 thoracic radiologist, 1 rheumatologist, and 1 pulmonary pathologist when histopathological data were available. Clinical history, medication exposure, occupational and environmental exposure, autoimmune symptoms, serological tests, HRCT findings, pulmonary function tests, bronchoalveolar lavage results when available, and pathological findings when available were reviewed systematically. Final diagnoses were made by consensus. The IPF diagnosis was based on clinical manifestations, physical examination and chest high-resolution computed tomography (HRCT) findings in accordance with the 2018 guidelines [10]. IIM was diagnosed and classified according to the 2017 EULAR/ACR classification criteria [11]. The diagnosis of ILD was confirmed based on respiratory symptoms, physical examination, and HRCT features. The exclusion criteria were as follows: (1) age < 18 years; (2) combined with malignant tumors; (3) diagnosis of overlap syndrome; (4) without serum samples; (5) follow-up less than 1 year. IPF patients were classified as AE and stable groups. AE-IPF was defined as an acute worsening or development of dyspnea typically within one month, accompanied by new ground-glass opacities on HRCT against a background of usual interstitial pneumonia, after excluding other causes of clinical exacerbation such as acute heart failure and pulmonary embolism [12]. IIM-ILD patients were classified as RP and stable groups. RP-ILD was defined as the exacerbation of radiological interstitial alterations, progressive dyspnea, and hypoxemia within one month following the onset of respiratory symptoms, irrespective of the treatment administered [13].

Fifty-six age- and sex-matched HCs were recruited from the Physical Examination Center of Nanjing University Medical School Affiliated Drum Tower Hospital. The study was conducted in accordance with the Declaration of Helsinki (protocol number 2025-0795-02, September 18, 2025). All patients signed the written informed consent for clinical data, follow-up and serum samples collection.

Clinical data

Clinical data were extracted from electronic medical records, including demographic information, laboratory examinations, pulmonary function tests (PFTs), HRCT and treatment. Demographic data comprised age, sex, smoking history, and body mass index (BMI). Laboratory data included C-reactive protein (CRP), ferritin and arterial blood gas parameters, including partial pressure of oxygen (PaO₂) and oxygenation index. The oxygenation index was derived from the ratio of PaO₂ to the fraction of inspired oxygen. Myositis-specific antibodies and myositis-associated antibodies were detected in IIM-ILD patients using immunoblotting (EUROLINE, EUROIMMUN AG, Germany) at the same central laboratory during their initial hospital admission.

PFT parameters included FVC, FVC% predicted, forced expiratory volume in one second (FEV1), FEV1% predicted, diffusing capacity of the lung for carbon monoxide (DLCO), and DLCO% predicted. To assess disease severity, patients were scored using the Gender-Age-Physiology (GAP) index. Points were assigned based on sex, age, FVC%, and DLCO% to obtain a GAP index ranging from 0 to 8. Patients were classified as Stage I (0–3 points), Stage II (4–5 points), or Stage III (6–8 points) based on the GAP index [14]. Chest HRCTs were mainly described as having usual interstitial pneumonia pattern (UIP), nonspecific interstitial pneumonia (NSIP) pattern, organizing pneumonia (OP) pattern, and diffuse alveolar damage (DAD) pattern [15].

Treatment information included the use of glucocorticoids, immunosuppressive agents, and antifibrotic therapy. Antifibrotic therapy was defined as treatment with pirfenidone or nintedanib for more than 3 months. Intravenous methylprednisolone pulse therapy was defined as intravenous methylprednisolone 0.5 g/day for 3 consecutive days during hospitalization or 7 days prior to admission.

Measurement of serum SPP1 Levels by ELISA

Fasting venous blood samples were collected from all participants at diagnosis and were immediately stored at − 80 °C until analysis. Serum SPP1 levels were quantified in both patients and HCs using a commercially available enzyme-linked immunosorbent assay (ELISA) kit (Elabscience, China), according to the manufacturer’s instructions. Briefly, all frozen serum samples were thawed at 4 °C and centrifuged briefly, and the supernatants were used for analysis. Standards and diluted serum samples were added to antibody-precoated wells and incubated. After washing, biotinylated detection antibody and horseradish peroxidase-conjugated avidin were added sequentially. Substrate solution was then applied for colour development, and the reaction was terminated with stop solution. Absorbance was measured immediately at 450 nm using a microplate reader, and serum SPP1 concentrations were calculated from the standard curve and expressed as ng/mL.

Follow-up data

Survival status was determined by reviewing medical records or via telephone follow-up. Follow-up continued until July 2025. The primary endpoint was 1-year all-cause mortality.

Single-Cell RNA sequencing and analysis

Lung tissue samples from 3 HCs and 3 patients with IPF undergoing lung transplantation were collected for scRNA-seq. Single-cell suspensions were prepared by mechanical dissociation and enzymatic digestion, and libraries were generated using the 10x Genomics Chromium Single Cell 3′ v3 platform. Raw reads were processed with fastp and Cell Ranger using the GRCh38 reference genome.

Downstream analyses, including quality control, normalization, dimensionality reduction, clustering, and cell-type annotation, were performed using Seurat R package. Cells with > 200 detected genes and < 20% mitochondrial UMI counts were retained. After normalization and regression of mitochondrial gene content, the top 2000 highly variable genes were selected for principal component analysis. The first 10 principal components were used for t-distributed stochastic neighbour embedding, Uniform Manifold Approximation and Projection (UMAP) visualization, and graph-based clustering. Marker genes for each cluster were identified using the Wilcoxon rank-sum test, and cell clusters were annotated according to canonical cell markers. Furthermore, pySCENIC was employed to construct gene regulatory networks and identify key transcription factors. Functional enrichment analyses were performed using QuSAGE, Gene Ontology, and Kyoto Encyclopedia of Genes and Genomes pathways.

For external validation, the public scRNA-seq dataset GSE122960 was re-analysed. 8 control donor subjects and 4 patients with IPF were included.

Processed matrices and metadata were downloaded from the Gene Expression Omnibus database and analysed using Seurat. Major lung cell populations were annotated according to canonical marker genes, and SPP1-positive macrophages were defined as macrophage-lineage cells with detectable SPP1 expression. Their distribution was visualized on UMAP plots according to disease status. SPP1 expression across annotated cell types was further visualized using violin plots stratified by disease status.

Statistical analysis

Continuous variables are presented as mean ± standard deviation (SD) or median (interquartile range [IQR]), whereas categorical variables are expressed as counts (percentages). Group comparisons were performed using Student’s t test or the Mann-Whitney U test for continuous variables, and the chi-square test or Fisher’s exact test for categorical variables, as appropriate. Univariable and multivariable Cox proportional hazards regression analyses were performed to identify independent prognostic factors. For multivariable analysis, age, sex, oxygenation index, SPP1, and CRP were included as covariates. Age and sex were retained in the multivariable models as fundamental demographic variables and potential confounders. Receiver operating characteristic (ROC) curve analysis was performed to determine optimal cut-off values and calculate the area under the curve (AUC) for prognostic prediction.

In the IIM-ILD cohort, ROC curve analysis was additionally performed to evaluate the discriminatory ability of serum SPP1 for identifying RP-IIM-ILD. A dichotomized SPP1 variable based on the optimal cut-off value was used to estimate the odds ratio for RP-IIM-ILD by logistic regression. Model calibration was assessed using the Hosmer–Lemeshow goodness-of-fit test based on a logistic regression model with serum SPP1 as a continuous variable. Survival was analysed using Kaplan-Meier curves and compared using the log-rank test. A two-tailed P value < 0.05 was considered statistically significant. All analyses were conducted using R version 4.5.0 (R Foundation for Statistical Computing, Vienna, Austria).

Results

Single-cell RNA sequencing identifies macrophages as the major SPP1-expressing cell population in IPF lungs

We performed scRNA-seq on lung tissue samples obtained from three patients with IPF undergoing lung transplantation and three controls. After quality control and doublet removal, 22,971 cells were retained for downstream analysis, comprising 12,020 cells from controls and 10,951 cells from IPF patients. UMAP analysis identified 19 distinct cell clusters (Fig. 1A), which were annotated into 10 major cell types based on canonical marker genes (Fig. 1B), including alveolar type 1 and type 2 cells, ciliated cells, club cells, endothelial cells, fibroblasts and myofibroblasts, macrophages, monocytes and dendritic cells, neutrophils, and T and B lymphocytes. IPF lungs exhibited an increased proportion of macrophages, neutrophils, and monocytes/dendritic cells, consistent with enhanced inflammatory and immune cell infiltration.

Fig. 1.

Single-cell landscape of SPP1 expression in healthy and IPF lung tissue. Single-cell RNA sequencing analysis of lung tissue from healthy controls and patients with idiopathic pulmonary fibrosis. A UMAP plot showing the distribution of 19 transcriptionally distinct cell clusters. B Annotation of major cell populations based on canonical marker genes, including AT1 cells, AT2 cells, ciliated cells, club cells, endothelial cells, fibroblast/myofibroblast cells, macrophages, monocytes/dendritic cells, neutrophils and T/B lymphocytes. C Feature plot showing that SPP1 expression is increased in IPF lung tissue and is predominantly localized to macrophage clusters. D Violin plots showing SPP1 expression across annotated cell populations. E Heatmap showing the expression of representative marker genes and extracellular matrix-related genes in SPP1-enriched cell populations. UMAP, Uniform Manifold Approximation and Projection; IPF, idiopathic pulmonary fibrosis; SPP1, secreted phosphoprotein 1; AT1 cells, alveolar type 1 cells; AT2 cells, alveolar type 2 cells

UMAP visualization further revealed that SPP1 expression was significantly enriched in macrophage clusters in IPF lung tissue, displaying a distinct aggregation pattern not observed in controls (Fig. 1C), suggesting that SPP1 is upregulated in IPF.

This observation was further supported by violin plots showing substantially higher SPP1 expression in macrophage-enriched clusters compared with other cell populations (Fig. 1D). Heatmap analysis indicated that high SPP1 expression was accompanied by increased expression of extracellular matrix-related genes, including COL1A1, CTHRC1, and DCN, as well as transcriptional regulators, such as PPARG and RUNX2 (Fig. 1E), suggesting that SPP1-positive macrophages may be involved in extracellular matrix remodelling and fibrotic phenotypic programming.

To externally validate our single-cell findings, we re-analysed the public GSE122960 dataset, including 8 control donor subjects and 4 patients with IPF (Fig. S2). Consistent with our discovery cohort, SPP1 expression was more evident in IPF lungs than in control donor lungs and was mainly localized to macrophage-lineage and monocyte/dendritic cell clusters (Fig. S2B). Violin plot analysis further demonstrated higher SPP1 expression in IPF lungs than in control donor lungs, particularly within macrophage-lineage and monocyte/dendritic cell populations, supporting the expansion of SPP1-associated myeloid populations in IPF (Fig. S2C).

Baseline characteristics of the study population

A total of 166 patients with ILD and 56 HCs were included (Table 1). No significant differences were observed between the two groups in age (62.98 ± 10.31 vs. 64.23 ± 6.43 years, P = 0.228) and sex (male, 59.4% vs. 58.9%, P = 0.887).

Table 1.

Baseline characteristics of HCs and ILD patients

Characteristics		HC	ILD	P
n		56	166
Age, years		64.23 ± 6.43	62.98 ± 10.31	0.228
Sex, M(%)		34(58.9%)	99(59.4)	0.887
SPP1, ng/mL		18.73(15.62–23.23)	53.78(33.12–93.20)	P < 0.001

Data are presented as mean ± SD, median (IQR), or n (%)

HCs, Healthy controls, ILD Interstitial lung disease, SPP1 Secreted phosphoprotein 1

Among the 166 patients with ILD, 74 had IPF and 92 had IIM-ILD (Table 2). Compared with the IIM-ILD group, patients with IPF were older (68.36 ± 8.14 vs. 58.65 ± 9.85 years, P < 0.001), were more frequently male (81.1% vs. 42.4%, P < 0.001), had a higher prevalence of smoking history (55.4% vs. 15.2%, P < 0.001), had a lower BMI (22.79 ± 3.66 vs. 24.64 ± 3.49 kg/m², P = 0.004), and had lower DLCO [2.13 (1.52–3.32) vs. 4.04 (2.87–5.30) mmol/min/kPa, P < 0.001] and DLCO% predicted [28.60 (18.53–46.55) vs. 50.60 (38.45–65.90), P < 0.001] than patients with IIM-ILD. Meanwhile, GAP score [5.00 (3.25-6.00) vs. 3.00 (1.00–4.00), P < 0.001] and GAP stage (P < 0.001) were significantly higher in the IPF group. No significant differences were observed between the IPF and IIM-ILD groups in oxygenation index, CRP, or ferritin levels. Serum SPP1 levels showed a trend towards being higher in patients with IPF than in those with IIM-ILD, although the difference did not reach statistical significance [68.50 (63.00–74.00) vs. 58.00 (52.00–65.00) ng/mL, P = 0.078].

Table 2.

Baseline clinical characteristics of the overall ILD cohort and the IPF and IIM-ILD subgroups

Characteristics	ILD	IPF	IIM-ILD	P
n	166	74	92
Age, years	62.98 ± 10.31	68.36 ± 8.14	58.65 ± 9.85	P < 0.001
Male sex, n (%)	99(59.4)	60(81.1)	39(42.4)	P < 0.001
Smoking history, n(%)	55(33.1)	41(55.4)	14(15.2)	P < 0.001
BMI, Kg/m2	23.73 ± 3.68(n = 132)	22.79 ± 3.66(n = 65)	24.64 ± 3.49(n = 67)	0.004
PO2, mmHg	80.00(67.00–97.00)	83.50(68.00-103.00)	79.00(65.00-94.50)	0.055
Oxygenation index, mmHg	313.75 ± 112.44	318.95 ± 97.42	309.57 ± 123.57	0.594
HRCT pattern, n (%)
UIP pattern	94(56.6)	74(100.0)	20(21.7)	P < 0.001
NSIP pattern	52(31.3)	0(0.0)	52(56.5)	P < 0.001
OP pattern	8(4.8)	0(0.0)	8(8.7)	0.009
DAD pattern	12(7.2)	0(0.0)	12(13.0)	0.001
Diagnostic procedures, n (%)
Bronchoalveolar lavage performed	23(13.9)	11(14.9)	12(13.0)	0.736
Bronchoscopic histopathological sampling	12(7.2)	3(4.1)	9(9.8)	0.157
Pulmonary function testing
FVC, L	2.18 ± 0.74(n = 86)	2.21 ± 0.76(n = 40)	2.15 ± 0.72(n = 46)	0.723
FVC% predicted	70.66 ± 21.07(n = 86)	68.86 ± 19.38(n = 40)	72.23 ± 22.52(n = 46)	0.462
FEV1, L	1.84 ± 0.56(n = 86)	1.91 ± 0.56(n = 40)	1.78 ± 0.55(n = 46)	0.279
FEV1% predicted	74.05 ± 20.07(n = 86)	76.14 ± 18.98(n = 40)	72.24 ± 21.00(n = 46)	0.372
DLCO, mmol/min/kPa	3.53 ± 1.68(n = 71)	2.13(1.52–3.32)(n = 30)	4.04(2.87–5.30)(n = 41)	P < 0.001
DLCO% predicted	45.33 ± 20.72(n = 71)	28.60(18.53–46.55)(n = 30)	50.60(38.45–65.90)(n = 41)	P < 0.001
GAP score	3.83 ± 2.07(n = 86)	5.00(3.25-6.00)(n = 40)	3.00(1.00–4.00)(n = 46)	P < 0.001
GAP stage				P < 0.001
I	41(24.7)	10(13.5)	31(33.7)
II	25(15.1)	12(16.2)	13(14.1)
III	20(12.0)	18(24.3)	2(2.2)
SPP1, ng/mL	53.78(33.12–93.20)	68.50(63.00–74.00)	58.00(52.00–65.00)	0.078
CRP, mg/L	7.50(3.58–21.35)	10.30(4.27–22.08)	6.80(3.40-20.98)	0.325
Ferritin, ng/mL	539.60(143.90-1303.30)(n = 35)	181.70(147.10-552.40)(n = 8)	928.00(143.60–1650.00)(n = 27)	0.179
Treatment, n (%)
Glucocorticoid therapy	125(75.3)	35(47.3%)	90(97.8)	P < 0.001
Immunosuppressive therapy	77(46.4)	0(0.0)	77(83.7)	P < 0.001
Antifibrotic therapy	52(31.3)	45(60.8)	7(7.6)	P < 0.001
1-year mortality, n (%)	55(33.1)	26(35.1)	29(31.5)	0.623

Data are presented as mean ± SD, median (IQR), or n (%)

ILD Interstitial lung disease, IPF Idiopathic pulmonary fibrosis, IIM Idiopathic inflammatory myopathy, BMI Body mass index, PO2 Partial pressure of oxygen, HRCT High-resolution computed tomography, UIP Usual interstitial pneumonia, NSIP Nonspecific interstitial pneumonia, OP Organizing pneumonia, DAD Diffuse alveolar damage, FVC Forced vital capacity, FEV1 Forced expiratory volume in 1 s, DLCO Diffusing capacity of the lung for carbon monoxide, GAP gender-age-physiology, SPP1 Secreted phosphoprotein 1, CRP C-reactive protein

In terms of radiological patterns, UIP was the predominant HRCT pattern in patients with IPF, whereas NSIP, OP, and DAD patterns were observed mainly in patients with IIM-ILD. Bronchoalveolar lavage was performed in 23 patients, and bronchoscopic histopathological sampling was performed in 12 patients. Antifibrotic therapy (60.8% vs. 7.6%, P < 0.001) was more frequently used in patients with IPF than in those with IIM-ILD, whereas glucocorticoid therapy (47.3% vs. 97.8%, P < 0.001) was more commonly used in the IIM-ILD group. During 1 year of follow-up, 55 deaths occurred, including 26 in the IPF group and 29 in the IIM-ILD group, with no significant difference in 1-year mortality between the two groups (P = 0.623).

Among patients with IPF, 47 were classified as stable IPF and 27 as AE-IPF (Table S1). Compared with patients with stable IPF, those with AE-IPF had a significantly lower oxygenation index (269.62 ± 99.13 vs. 351.66 ± 77.89 mmHg, P < 0.001), a higher GAP score [6.00 (5.00–7.00) vs. 5.00 (3.00–6.00), P = 0.042], and higher CRP levels [13.80 (7.00–22.90) vs. 5.55 (3.38–19.87) mg/L, P = 0.039].

Antifibrotic therapy was less frequent in the AE-IPF group than in the stable IPF group (37.0% vs. 74.5%, P = 0.001), whereas glucocorticoid therapy was more frequently used in AE-IPF. 1-year mortality was significantly higher in the AE-IPF group than in the stable IPF group (59.3% vs. 21.3%, P < 0.001).

Among patients with IIM-ILD, 54 were classified as stable IIM-ILD and 38 as RP-IIM-ILD (Table S2). Compared with patients with stable IIM-ILD, those with RP-IIM-ILD had a significantly lower oxygenation index (279.31 ± 132.45 vs. 365.61 ± 88.70 mmHg, P < 0.001). Regarding HRCT patterns, DAD was observed only in the RP-IIM-ILD group (31.6% vs. 0.0%, P < 0.001), whereas UIP pattern was less frequent in the RP-IIM-ILD group than in the stable IIM-ILD group (10.5% vs. 29.6%, P = 0.029). Patients with RP-IIM-ILD also had higher GAP scores [4.00 (3.25–5.00) vs. 3.00 (1.00–3.25), P = 0.009], higher CRP levels [19.30 (3.20–49.20) vs. 4.65 (3.00–8.48) mg/L, P < 0.001], and higher ferritin levels [1303.10 (958.15–1650.00) vs. 143.60 (61.50–539.60) ng/mL, P < 0.001]. Autoantibody profiles and treatment details were further compared between stable IIM-ILD and RP-IIM-ILD. Anti-MDA5 antibody positivity was more frequent in the RP-IIM-ILD group than in the stable IIM-ILD group (69.6% vs. 33.3%, P = 0.009). Nearly all patients with IIM-ILD received glucocorticoid therapy. Intravenous methylprednisolone pulse therapy was used only in the RP-IIM-ILD group (21.1% vs. 0.0%, P < 0.001). Tacrolimus was used in 47.4% of patients with RP-IIM-ILD and 37.0% of those with stable IIM-ILD (P = 0.322), whereas cyclophosphamide was more frequently used in the RP-IIM-ILD group (36.8% vs. 9.3%, P = 0.001). One-year mortality was significantly higher in patients with RP-IIM-ILD than in those with stable IIM-ILD (76.3% vs. 11.1%, P < 0.001).

Serum SPP1 levels in ILD subgroups

Serum SPP1 levels were significantly higher in patients with ILD than in HCs [53.78 (33.12–93.20) vs. 18.73 (15.62–23.23) ng/mL, P < 0.001](Fig. 2A; Table 1). In the IPF cohort, patients with AE-IPF had significantly higher SPP1 levels than those with stable disease [101.76 (87.96-136.77) vs. 45.44 (34.32–65.91) ng/mL, P < 0.001](Fig. 2B, Table S1). In the IIM-ILD cohort, patients with RP-ILD likewise showed markedly higher SPP1 levels than those with stable disease [103.34 (75.20-139.27) vs. 31.86 (25.30-44.22) ng/mL, P < 0.001](Fig. 2C, Table S2). In addition, anti-melanoma differentiation-associated gene 5 (MDA5)-positive IIM-ILD patients had significantly higher serum SPP1 levels than anti-MDA5-negative patients [85.45 (33.31-126.15) vs. 37.87 (24.53–82.29) ng/mL, P = 0.015]( Fig. 2D).

Fig. 2.

Serum SPP1 concentrations in healthy controls and ILD subgroups. (A) Serum SPP1 levels in HCs and patients with ILD. B Serum SPP1 levels in stable IPF and AE-IPF. C Serum SPP1 levels in stable IIM-ILD and RP-IIM-ILD. D Serum SPP1 levels in anti-MDA5-positive and anti-MDA5-negative IIM-ILD. SPP1, secreted phosphoprotein 1; ILD, interstitial lung disease; AE, acute exacerbation; HCs, healthy controls; IPF, idiopathic pulmonary fibrosis; IIM, idiopathic inflammatory myopathy; RP, rapidly progressive; MDA5, melanoma differentiation-associated gene 5

Diagnostic performance of serum SPP1 for identifying RP-IIM-ILD

In patients with IIM-ILD, serum SPP1 showed excellent discriminatory performance for identifying RP-IIM-ILD, with an AUC of 0.943. The optimal cut-off value was 53.78 ng/mL. Patients with serum SPP1 levels ≥ 53.78 ng/mL had significantly higher odds of RP-IIM-ILD than those with lower SPP1 levels (odds ratio, 78.333; 95% CI, 18.905–324.571; P < 0.001). The Hosmer–Lemeshow goodness-of-fit test showed acceptable calibration for the continuous SPP1 model (P = 0.114).

Correlation between serum SPP1 and pulmonary function

In the overall ILD cohort, serum SPP1 levels were negatively correlated with DLCO% predicted (r = − 0.267, P = 0.025), while a trend towards a positive correlation was observed between serum SPP1 and GAP score (r = 0.206, P = 0.056) (Table S3). In the IIM-ILD subgroup, serum SPP1 also showed a trend towards a negative correlation with DLCO% predicted (r = − 0.276, P = 0.081) (Table S3).

Correlation between serum SPP1 and ferritin

To further investigate the relationship between serum SPP1 and ferritin, correlation analyses were performed in the overall ILD cohort and in the IPF and IIM-ILD subgroups. In the overall ILD cohort, serum SPP1 levels were positively correlated with serum ferritin levels (R = 0.503, P = 0.002; Fig. S3A). In the IPF subgroup, no significant correlation was observed between serum SPP1 and ferritin levels (R = − 0.405, P = 0.320; Fig. S3B). In contrast, a significant positive correlation between serum SPP1 and ferritin was observed in the IIM-ILD subgroup (R = 0.486, P = 0.010; Fig. S3C).

Predictors of 1-year mortality in ILD patients

Cox proportional hazards regression analysis was performed to evaluate predictors of 1-year mortality in the overall ILD cohort and in disease-specific subgroups. In the overall ILD cohort (Table 3), univariable analysis showed that male sex (hazard ratio [HR] 1.987, 95% confidence interval [CI] 1.097–3.598, P = 0.023), lower oxygenation index (HR 0.990, 95% CI 0.987–0.993, P < 0.001), higher SPP1 levels (HR 1.010, 95% CI 1.007–1.014, P < 0.001), and higher CRP levels (HR 1.015, 95% CI 1.007–1.023, P < 0.001) were associated with an increased risk of 1-year mortality. In multivariable analysis, SPP1 (HR 1.007, 95% CI 1.002–1.012, P = 0.005) and oxygenation index (HR 0.992, 95% CI 0.989–0.996, P < 0.001) remained independently associated with 1-year mortality.

Table 3.

Survival analysis for 1-year mortality in the overall ILD cohort

	Univariable		Multivariable
	HR (95% CI)	p-value	HR (95% CI)	p-value
Age	1.019(0.992–1.046)	0.175	1.016(0.986–1.047)	0.292
Male	1.987(1.097–3.598)	0.023	1.932(0.986–3.787)	0.055
Oxygenation index	0.990(0.987–0.993)	P < 0.001	0.992(0.989–0.996)	P < 0.001
SPP1	1.010(1.007–1.014)	P < 0.001	1.007(1.002–1.012)	0.005
CRP	1.015(1.007–1.023)	P < 0.001	1.001(0.992–1.011)	0.759

ILD Interstitial lung disease, CRP C-reactive protein, SPP1 Secreted phosphoprotein 1, HR Hazard ratio, CI Confidence interval

In the IPF subgroup (Table S4), univariable analysis showed that lower oxygenation index (HR 0.991, 95% CI 0.987–0.995, P < 0.001), higher SPP1 (HR 1.018, 95% CI 1.009–1.027, P < 0.001), and higher CRP (HR 1.028, 95% CI 1.012–1.044, P < 0.001) were associated with an increased mortality risk. In multivariable analysis, SPP1 (HR 1.015, 95% CI 1.003–1.027, P = 0.012), CRP (HR 1.023, 95% CI 1.005–1.040, P = 0.011), and oxygenation index (HR 0.995, 95% CI 0.991-1.000, P = 0.048) remained independent predictors of 1-year mortality.

In the IIM-ILD subgroup (Table S5), univariable analysis identified older age (HR 1.048, 95% CI 1.005–1.091, P = 0.027), male sex (HR 2.252, 95% CI 1.075–4.718, P = 0.031), lower oxygenation index (HR 0.989, 95% CI 0.984–0.993, P < 0.001), higher SPP1 (HR 1.008, 95% CI 1.005–1.012, P < 0.001), and higher CRP (HR 1.012, 95% CI 1.003–1.022, P = 0.013) as significant predictors of increased 1-year mortality. In multivariable analysis, higher SPP1 (HR 1.007, 95% CI 1.001–1.012, P = 0.022), older age (HR 1.044, 95% CI 1.005–1.085, P = 0.025), male sex (HR 3.079, 95% CI 1.327–7.145, P = 0.009), and lower oxygenation index (HR 0.991, 95% CI 0.985–0.996, P < 0.001) remained independently associated with higher mortality.

Kaplan-Meier survival analysis stratified by serum SPP1 levels

ROC analysis identified 66.655 ng/mL as the optimal cut-off value of serum SPP1 for predicting 1-year mortality, with an AUC of 0.756. Kaplan-Meier survival analysis showed that patients with serum SPP1 levels ≥ 66.655 ng/mL had significantly poorer survival than those with lower SPP1 levels in the overall ILD cohort (P < 0.001) (Fig. 3A). Similar findings were observed in both the IPF subgroup (P = 0.017) and the IIM-ILD subgroup (P < 0.001) (Fig. 3B-C).

Fig. 3.

Kaplan-Meier survival analysis grouped by SPP1≥66.655 ng/mL in ILD subtypes. A ILD. B IPF. C IIM-ILD. SPP1, secreted phosphoprotein 1; ILD, interstitial lung disease; IPF, idiopathic pulmonary fibrosis; IIM, idiopathic inflammatory myopathy

Discussion

In this study, we combined scRNA-seq with clinical serum validation to investigate the role of SPP1 in IPF and IIM-ILD. At the tissue level, scRNA-seq analysis showed that SPP1 was upregulated in IPF lungs. At the circulating level, serum SPP1 was significantly elevated in patients with IPF or IIM-ILD compared with HCs and was further increased in clinically severe subgroups, including AE-IPF, RP-IIM-ILD and anti-MDA5-positive IIM-ILD. In addition, serum SPP1 showed excellent discriminatory ability for identifying RP-IIM-ILD and was positively correlated with ferritin in the overall ILD cohort and in the IIM-ILD subgroup. More importantly, higher SPP1 levels were independently associated with 1-year mortality in the ILD cohort, as well as in the IPF and IIM-ILD subgroups. Our findings suggest that SPP1 is not only involved in the pathobiology of fibrotic lung injury but may also serve as a clinically relevant prognostic biomarker in ILD.

Recent scRNA-seq studies have consistently identified a distinct SPP1-enriched macrophage subset within the fibrotic lung microenvironment. In IPF, Morse et al. described a proliferative SPP1hi/MERTKhi macrophage population expanded in fibrotic regions and linked to myofibroblast activation, while Zhang et al. similarly identified a macrophage SPP1 subset localized to fibroblastic foci and increased in IPF lungs [6, 16]. Notably, SPP1 upregulation in pulmonary fibrosis is not restricted to macrophages, indicating that, beyond SPP1 + macrophages, increased SPP1 expression may also be observed more broadly within the fibrotic lung microenvironment. Extending these observations beyond IPF, Papazoglou et al. showed that SPP1 macrophages in systemic sclerosis-associated ILD exhibited a profibrotic transcriptional programme, supporting the concept that SPP1 marks a disease-associated macrophage state closely linked to fibrosis rather than a nonspecific inflammatory phenotype [17]. These observations are consistent with our scRNA-seq findings, which localized SPP1 primarily to macrophage clusters and demonstrated its co-expression with extracellular matrix-related genes.

Accumulating experimental evidence indicates that SPP1 directly participates in profibrotic signalling. A previous study has shown that activation of IL-7R on interstitial macrophages promotes lung fibrosis through Spp1, supporting a pathogenic IL-7R-Spp1 macrophage axis in fibrotic lungs [18]. Similarly, IL20Rb aggravates pulmonary fibrosis by enhancing the activation of bone marrow-derived profibrotic macrophages via the JAK2/STAT3 and PI3K/Akt pathways; these macrophages express multiple profibrotic mediators, including SPP1, thereby amplifying fibroblast activation and extracellular matrix deposition [19]. SPP1 has also been shown to promote M2 macrophage polarization through the JAK2/STAT3 pathway, thereby accelerating pulmonary fibrosis progression [20]. In parallel, platelet-derived CXCL4 can instruct the generation of profibrotic SPP1 + macrophages, which enhance crosstalk with fibroblasts and myofibroblasts, leading to myofibroblast activation and extracellular matrix deposition [21]. Macrophage-derived SPP1 can also bind to CD44 on alveolar epithelial cells, activate the FAK/AKT/Snail1 pathway, and induce epithelial-mesenchymal transition and profibrotic protein expression, further promoting fibrogenesis [22]. More recent evidence has further suggested that SPP1 may drive profibrotic macrophage polarization through the PI3K/Akt pathway, and that inhibition of this axis alleviates fibrosis progression [23]. Consistent with these findings, silencing SPP1 attenuated bleomycin-induced pulmonary fibrosis by suppressing epithelial-mesenchymal transition [24].

Clinical evidence also supports the relevance of SPP1 in fibrotic lung disease.

Plasma OPN concentrations were significantly elevated in patients with interstitial pneumonia compared with those with sarcoidosis and HCs [25]. In IPF, bronchoalveolar lavage cell transcriptomic analysis identified SPP1 as one component of an inflammation- and hypoxia-related prognostic gene signature and related with poor prognosis [26]. Integrative analyses further demonstrated increased SPP1 expression in both lung tissue and bronchoalveolar lavage fluid, with higher BAL OPN levels associated with poorer prognosis [22]. Plasma SPP1 levels were elevated in anti-MDA5-positive dermatomyositis with RP-ILD [8]. And higher serum OPN levels have also been linked to subsequent lung function decline in systemic sclerosis-associated ILD [27]. These observations are consistent with our findings that serum SPP1 was elevated in more severe disease phenotypes including AE-IPF, RP-IIM-ILD, and anti-MDA5-positive IIM-ILD, and remained independently associated with 1-year mortality in the overall included cohort and in the IPF and IIM-ILD subgroups.

Our study also has several limitations. First, all patients were recruited from a tertiary referral centre. However, because most patients were referred from different regions of China, the cohort may still reflect a relatively broad clinical spectrum of ILD. Second, our findings were strengthened by integrating single-cell transcriptomic discovery with validation in clinical serum samples and across subgroups with different disease severity. However, this study included two clinically important ILD subtypes, namely IPF and IIM-ILD, which did not cover the full spectrum of ILDs.

Therefore, our findings should not be directly generalized to all ILD subtypes.

External validation in prospective multicentre cohorts including broader ILD populations, such as other connective tissue disease-associated ILDs, hypersensitivity pneumonitis, sarcoidosis-associated ILD, and other progressive fibrosing ILDs, is needed to determine whether SPP1 has wider applicability as a biomarker across the ILD spectrum. Third, the lung tissues used for scRNA-seq were obtained from three patients with IPF undergoing lung transplantation and therefore may not fully represent the cellular and molecular features of earlier stages of disease. In addition, the GSE122960 validation analysis supported the enrichment of SPP1-positive macrophage populations in IPF lungs, but it could not validate SPP1 expression patterns in IIM-ILD or directly assess the prognostic value of circulating SPP1. Future studies incorporating larger single-cell cohorts, samples from earlier disease stages, and lung tissues from other ILD subtypes are needed to further define the temporal and disease-specific role of SPP1 in ILD progression. Fourth, because this was a retrospective study, regular follow-up assessments were not available for all patients.

In particular, longitudinal pulmonary function and HRCT follow-up data were incomplete, which limited our ability to systematically evaluate disease progression according to standardized progression criteria. Future studies with larger sample sizes and broader ILD populations are necessary to validate the generalizability of SPP1 as a biomarker in ILD.

Conclusions

In conclusion, our integrated single-cell and clinical analyses suggest that SPP1 is elevated in IPF and IIM-ILD, particularly in AE-IPF, RP-IIM-ILD, and anti-MDA5-positive IIM-ILD, and may serve as an independent prognostic biomarker for 1-year mortality in IPF and IIM-ILD. Larger prospective multicentre studies are warranted to confirm these findings.

Abbreviations

ILD

Interstitial lung disease

IPF

Idiopathic pulmonary fibrosis

IIM

Idiopathic inflammatory myopathy

AE

Acute exacerbation

RP

Rapidly progressive

SPP1

Secreted phosphoprotein 1

OPN

Osteopontin

FVC

Forced vital capacity

scRNA-seq

Single-cell RNA sequencing

HCs

Healthy controls

MDD

Multidisciplinary discussion

HRCT

High-resolution computed tomography

PFTs

Pulmonary function tests

BMI

Body mass index

CRP

C-reactive protein

PaO2

Partial pressure of oxygen

FEV1

Forced expiratory volume in one second

DLCO

Diffusing capacity of the lung for carbon monoxide

GAP

Gender-Age-Physiology

UIP

Usual interstitial pneumonia

NSIP

Nonspecific interstitial pneumonia

OP

Organizing pneumonia

DAD

Diffuse alveolar damage

ELISA

Enzyme-linked immunosorbent assay

SD

Standard deviation

IQR

Interquartile range

ROC

Receiver operating characteristic

AUC

Area under the curve

UMAP

Uniform Manifold Approximation and Projection, MDA5,melanoma differentiation-associated gene 5,HR, hazard ratio

CI

Confidence interval

Authors' contributions

JD is the guarantor of the article and takes responsibility for the integrity of the work as a whole, from inception to published article. HX and JD conceived the idea for this manuscript and contributed equally to the conception of this work. TJ played an active role in the creation of the manuscript, and drafted the report. TJ and YL revised the manuscript. TJ, YL, SZ, CZ, WX, SW and QL were all directly involved in the acquisition of data for the article. All authors read and approved the final manuscript.

Funding

This study was supported by the National Natural Science Foundation of China (No. 82570096) and funding for Clinical Trials from Nanjing Drum Tower Hospital, Affiliated Hospital of Medical School, Nanjing University (No. 2025-LCYJ-MS-03).

Data availability

The data underlying the results of this study are from medical records of Nanjing University medical school affiliated Drum Tower Hospital. Therefore, the data is only available from the corresponding author on reasonable request.

Declarations

Ethics approval and consent to participate

This study was approved by the Ethics Committee of Nanjing Drum Tower Hospital. (protocol number 2025-0795-02, September 18, 2025). All patients signed the written informed consent for clinical data, follow-up and serum samples collection.

Consent for publication

Not applicable.

Competing interests

The authors declare no competing interests.