Abstract
BACKGROUND: Systemic sclerosis (SSc) exhibits heterogeneous pulmonary involvement, most commonly interstitial lung disease (ILD) and pulmonary arterial hypertension (PAH). How systemic inflammatory, immune, and nutritional states differ across these lung phenotypes remains incompletely understood. OBJECTIVE: To characterize integrated inflammatory-immune-nutritional signatures across SSc lung phenotypes and validate their robustness using multivariable modeling and unsupervised clustering. METHODS: We conducted a retrospective cross-sectional study of 314 consecutive SSc inpatients fulfilling the 2013 ACR/EULAR criteria (2018-2023). Patients were stratified into four lung phenotypes: ILD-/PAH- (n = 45), ILD + /PAH- (n = 183), ILD-/PAH + (n = 12), and ILD + /PAH + (n = 74). Routine laboratory data were used to derive inflammatory markers (hs-CRP, ESR, NLR, PLR, MLR, SII, SIRI, SCI), nutritional parameters (BMI, albumin, lipid profiles), and immune markers (autoantibodies, immunoglobulins, complement). We performed (1) group comparisons; (2) multinomial logistic regression focusing on the three predominant phenotypes (ILD-/PAH- as reference; ILD + /PAH-; ILD + /PAH +), excluding ILD-/PAH + due to small sample size (n = 12) and sparse-data instability; and (3) k-means clustering of standardized biomarker profiles. Internal validation and clustering stability analyses were performed. RESULTS: ILD + /PAH + patients showed a higher systemic inflammatory burden, including higher ESR (38 [21-61.5] vs. 23 [10-45] mm/h, p = 0.008), hs-CRP (10.15 [3.34-28.18] vs. 5.9 [1.5-16.1] mg/L, p = 0.017), CRP/albumin ratio (0.33 [0.11-0.95] vs. 0.17 [0.04-0.49], p = 0.019), and SII (1124.76 [700.8-2034.14] vs. 820.34 [461.33-1197.08], p = 0.013) compared with ILD-/PAH-. Anti-Scl-70 positivity was enriched in ILD-containing phenotypes (p < 0.001), and immunoglobulins varied (IgG p = 0.032; IgA p = 0.028). In multinomial models, female sex and longer disease duration were associated with both ILD + /PAH- (OR 3.48, 95%CI 1.51-8.04; OR 1.02 per month, 95%CI 1.01-1.04) and ILD + /PAH + (OR 3.49, 95%CI 1.31-9.33; OR 1.03 per month, 95%CI 1.01-1.05). Triglycerides (Z-score) were associated with ILD + /PAH- (OR 1.70, 95%CI 1.05-2.77), while log(ESR) (Z-score) was associated with ILD + /PAH + (OR 1.82, 95%CI 1.01-3.27). Clustering supported a three-cluster structure driven by inflammation (ESR), mucosal immunity (IgA), and metabolic dysregulation (triglycerides), with strong stability (Adjusted Rand Index = 1.00). Internal validation showed stable within-cohort performance (accuracy ∼0.62-0.65). CONCLUSION: SSc lung phenotypes defined by ILD/PAH co-occurrence display distinct integrated inflammatory-immune-nutritional signatures. The ILD + /PAH + phenotype reflects a high-burden systemic inflammatory state, and a small set of coherent markers (ESR, triglycerides, IgA, and autoantibodies) provides a reproducible framework for phenotype-oriented stratification.