Quantitative computed tomography in interstitial lung disease related to ANCA‐associated vasculitis: First insights from a mono‐centric retrospective cohort

Interstitial lung disease (ILD) is a significant manifestation of anti-neutrophil cytoplasmic antibody-associated vasculitis (AAV) that can precede or follow its onset and consistently worsens prognosis.1, 2 Accurate identification and quantification of these pulmonary lesions are vital for targeted and early management.3 While spirometry has prognostic value, it is not universally performed.4 Currently, detecting a usual interstitial pneumonia pattern on qualitative chest computed tomography (CT) is the only radiological finding with proven prognostic significance.3 Semiquantitative evaluation, though ideal, is difficult to integrate into routine practice as it requires specialized radiologists and suffers from significant inter- and intra-observer variability, limiting its reproducibility.5 Quantitative CT (QCT) overcomes these issues by being largely operator-independent and highly reproducible. Parametric QCT has demonstrated feasibility and validity, meeting Outcome Measures in Rheumatology Clinical Trials criteria for rheumatology tools.6 This densitometric method analyzes voxel attenuation values within the lung volume,7 producing parameters that describe the histogram of lung densities. Strong evidence correlates these parameters—especially kurtosis, skewness, and mean lung attenuation (MLA)—with radiological ILD extent and pulmonary function in diseases like idiopathic pulmonary fibrosis and other rheumatic ILDs.8-10 However, no data exists on QCT's utility in AAV. This study therefore aims to: investigate the role of chest QCT in distinguishing AAV-ILD from non-ILD AAV; determine optimal cutoff values for the most discriminative QCT parameters; and explore whether these parameters predict overall mortality in AAV patients. This cross-sectional observational study was approved by the Ethics Committee of University Hospital of Parma (Protocol AVEN 856, 11/01/2022) and conducted according to the Declaration of Helsinki. The cohort consisted of consecutive patients from the Internal Medicine/Rheumatology and Nephrology Units at the University Hospital of Parma (2009–2019). Inclusion required a confirmed AAV diagnosis (granulomatosis with polyangiitis or microscopic polyangiitis), an available chest CT scan (for symptoms or screening), a consistent CT scanner/protocol, and informed consent. We collected demographic data (age, sex), clinical diagnosis, and survival status. Two board-certified radiologists (each with >15 years of ILD experience) independently reviewed CT scans to characterize lesions. When ILD was present, its extent was quantified using established scoring systems.11 Discrepancies were resolved by consensus. Patients were thus categorized as AAV-ILD or non-ILD AAV. A blinded rheumatologist then performed quantitative analysis using a validated, operator-independent method from systemic sclerosis-ILD,12 generating five key parameters: kurtosis (Kurt), skewness (Skew), standard deviation (sDev), MLA, and fibrosis ratio (%Fib). Kurt, Skew, MLA, and sDev were derived from two volumes: the entire lung (total: tKurt, tSkew, tSDev, tMLA) and radiologically normal regions (voxels from −950 to −700 HU; parenchymal: pKurt, pSkew, pSDev, pMLA). Continuous and categorical variables are reported as median (interquartile range) and counts (percentages), respectively. Group comparisons used Kruskal–Wallis or Fisher's exact tests. QCT discrimination for ILD was assessed by receiver operating characteristic curve analysis. The prognostic value of each QCT parameter for overall mortality (adjusted for age, sex, diagnosis) was evaluated with Cox regression. All analyses used Jamovi (v2.6.2); significance was set at p < 0.05 (two-tailed). The final cohort included 55 AAV patients. Radiological criteria identified ILD in 8 patients (15%), while 15 others (27%) showed non-ILD pulmonary manifestations such as bronchiectasis, nodules, and alveolar hemorrhage. Patient characteristics are detailed in Table 1. QCT analysis showed significant differences between ILD and non-ILD groups for all parameters except tSDev and %Fib. For ILD identification, pKurt (cutoff 1.24; AUC 0.795), tSkew (cutoff 2.5; AUC 0.793), and pSDev (cutoff 98.6; AUC 0.790) demonstrated optimal discrimination, while %Fib (cutoff 8; AUC 0.681) and tMLA (cutoff −732 HU; AUC 0.721) performed the poorest. Weak but significant correlations emerged between semi-quantitative scores and QCT parameters (Spearman's ρ: 0.27–0.37; p < 0.05), except for %Fib. pKurt showed the strongest inverse correlation (ρ = −0.37, p < 0.01). In mortality analysis adjusted for age, sex, and diagnosis, three QCT parameters independently predicted overall survival: pMLA (HR 1.01, 95% CI 1.00–1.02; p = 0.016), pSkew (HR 0.38, 95% CI 0.15–0.96; p = 0.040), and tMLA (HR 1.01, 95% CI 1.00–1.02; p = 0.028). No semi-quantitative parameters retained significance in the final model. This study evaluates QCT parameters in AAV-associated ILD. While preliminary and from a limited cohort, our findings align with QCT applications in other rheumatic disease-related ILDs.7-10 QCT could significantly improve patient stratification by enabling ILD identification even by non-specialists. Its correlation with semiquantitative CT scores underscores this potential utility. Detecting significant pulmonary involvement through QCT might help clinicians select candidates for antifibrotic therapies.13 Beyond its inherent speed and reproducibility, QCT appears to offer prognostic value by identifying patients at higher risk of all-cause mortality. These encouraging results require cautious interpretation due to several limitations. Missing data on smoking history, respiratory symptoms, disease duration, and renal involvement prevent full contextualization, though these factors haven't shown direct influence on pulmonary severity.4 Renal involvement in MPA-ILD appears less frequent than in other MPA forms.14 We also lacked treatment data, though conventional therapy effects on AAV-ILD remain poorly defined.3 The potential influences of pulmonary hypertension and edema on QCT measurements remain unexamined. Furthermore, CT scans were obtained for various indications at different timepoints. While this heterogeneity complicates mortality analysis, it strengthens QCT's demonstrated ability to detect ILD across disease stages. The heterogeneous assessment of death causes limited our analysis to overall mortality. Finally, we performed no clinimetric evaluations, despite the BVAS's known limitations in capturing pulmonary involvement.15 Despite these limitations, QCT parameters show promise in AAV-ILD, similar to their established roles in other rheumatic diseases. Our identified QCT cutoffs align closely with those validated for prognosis in systemic sclerosis and Sjögren's syndrome.8, 9 Although future correlations with disease-specific indices and pulmonary function tests are needed, QCT represents a plausible, rapid, and cost-free method for assessing ILD presence and prognostic significance in AAV. While our data don't conclusively establish QCT's mortality predictive value in comprehensive models including other risk factors, we advocate for testing QCT-based ILD assessment in larger, structured survival models to determine its ability to stratify patients and identify those with poorer prognosis earlier.2 In conclusion, parametric QCT appears readily applicable to AAV, with our findings supporting its potential utility for prognostic purposes and paving the way for personalized treatment strategies in these complex patients. Alarico Ariani: Conceptualization; data curation; formal analysis; writing—original draft. Mario Silva: Radiological analysis; methodology; validation. Giovanni M. Rossi and Marco Delsante: Clinical data collection; interpretation; critical revision. Andrea Becciolini: Statistical analysis; visualization. Enrico Fiaccadori and Augusto Vaglio: Supervision; validation; writing—review and editing. Nicola Sverzellati: Senior radiologist; methodology; supervision. This study adheres to the latest ICMJE guidelines, and all authors meet the criteria for authorship. We thank J. Ratzinger (VPME) for his teachings. The authors declare no conflicts of interest. The data that support the findings of this study are available on request from the corresponding author. The data are not publicly available due to privacy or ethical restrictions.