Detection of Rheumatoid Arthritis-Interstitial Lung Disease Is Enhanced by Serum Biomarkers

Abstract

Rationale: Interstitial lung disease (ILD), a leading cause of morbidity and mortality in rheumatoid arthritis (RA), is highly prevalent, yet RA-ILD is underrecognized.

Objectives: To identify clinical risk factors, autoantibodies, and biomarkers associated with the presence of RA-ILD.

Methods: Subjects enrolled in Brigham and Women's Hospital Rheumatoid Arthritis Sequential Study (BRASS) and American College of Rheumatology (ACR) cohorts were evaluated for ILD. Regression models were used to assess the association between variables of interest and RA-ILD. Receiver operating characteristic curves were generated in BRASS to determine if a combination of clinical risk factors and autoantibodies can identify RA-ILD and if the addition of investigational biomarkers is informative. This combinatorial signature was subsequently tested in ACR.

Measurements and main results: A total of 113 BRASS subjects with clinically indicated chest computed tomography scans (41% with a spectrum of clinically evident and subclinical RA-ILD) and 76 ACR subjects with research or clinical scans (51% with a spectrum of RA-ILD) were selected. A combination of age, sex, smoking, rheumatoid factor, and anticyclic citrullinated peptide antibodies was strongly associated with RA-ILD (areas under the curve, 0.88 for BRASS and 0.89 for ACR). Importantly, a combinatorial signature including matrix metalloproteinase 7, pulmonary and activation-regulated chemokine, and surfactant protein D significantly increased the areas under the curve to 0.97 (P = 0.002, BRASS) and 1.00 (P = 0.016, ACR). Similar trends were seen for both clinically evident and subclinical RA-ILD.

Conclusions: Clinical risk factors and autoantibodies are strongly associated with the presence of clinically evident and subclinical RA-ILD on computed tomography scan in two independent RA cohorts. A biomarker signature composed of matrix metalloproteinase 7, pulmonary and activation-regulated chemokine, and surfactant protein D significantly strengthens this association. These findings may facilitate identification of RA-ILD at an earlier stage, potentially leading to decreased morbidity and mortality.

Keywords: biomarkers; interstitial lung disease; rheumatoid arthritis; risk prediction; subclinical.

Figure 1.

Study enrollment in BRASS (A) and ACR (B) cohorts. A flow diagram of study enrollment divides participants into groups according to presence and subtype of interstitial lung disease. ACR = American College of Rheumatology; BRASS = Brigham and Women's Hospital Rheumatoid Arthritis Sequential Study; CT = computed tomography; HRCT = high-resolution computed tomography; ILD = interstitial lung disease; RA = rheumatoid arthritis.

Figure 2.

A comparison of receiver operating characteristic curves for the spectrum of RA-ILD in BRASS, ACR, and combined cohorts. A: Age, sex, ever-smoker, RF, anti-CCP. B: MMP7, PARC, SP-D. C: Age, sex, ever-smoker, RF, anti-CCP, MMP7, PARC, SP-D. (A) Spectrum of RA-ILD in BRASS. A: AUC 0.88, B: AUC 0.92, C: AUC 0.97; A versus B, AUC increases 0.04, P = 0.2; A versus C, AUC increases 0.09, P = 0.002. (B) Spectrum of RA-ILD in ACR. A: AUC 0.89, B: AUC 0.95, C: AUC 1.00; A versus B, AUC increases 0.05, P = 0.18; A versus C, AUC increases 0.11, P = 0.016. (C) Spectrum of RA-ILD in combined. A: AUC 0.85, B: AUC 0.92, C: AUC 0.94; A versus B, AUC increases 0.07, P = 0.061; A versus C, AUC increases 0.09, P = 0.002. (D) Clinically evident RA-ILD in BRASS. A: AUC 0.86, B: AUC 0.92, C: AUC 0.98; A versus B, AUC increases 0.06, P = 0.2; A versus C, AUC increases 0.12, P = 0.01. (E) Clinically evident RA-ILD in ACR. A: AUC 0.82, B: AUC 0.97, C: AUC 1.00; A versus B, AUC increases 0.15, P = 0.045; A versus C, AUC increases 0.18, P = 0.017. (F) Clinically evident RA-ILD in combined. A: AUC 0.82, B: AUC 0.92, C: AUC 0.94; A versus B, AUC increases 0.10, P = 0.033; A versus C, AUC increases 0.12, P = 0.002. (G) Subclinical RA-ILD in BRASS. A: AUC 0.89, B: AUC 0.94; C: AUC 0.98; A versus B, AUC increases 0.05, P = 0.2; A versus C, AUC increases 0.09, P = 0.01. (H) Subclinical RA-ILD in ACR. A: AUC 0.98, B: AUC 0.93; C: AUC 1.00; A versus B, AUC decreases 0.05, P = 0.103; A versus C, AUC increases 0.02, P = 0.19. (I) Subclinical RA-ILD in combined. A: AUC 0.88, B: AUC 0.91; C: AUC 0.95; A versus B, AUC increases 0.03, P = 0.24; A versus C, AUC increases 0.07, P = 0.007. ACR = American College of Rheumatology; AUC = area under the curve; BRASS = Brigham and Women's Hospital Rheumatoid Arthritis Sequential Study; CCP = cyclic citrullinated peptides; MMP = matrix metalloproteinase; PARC = pulmonary and activation-regulated chemokine; RA-ILD = rheumatoid arthritis–associated interstitial lung disease; RF = rheumatoid factor; SP-D = surfactant protein D.

Figure 3.

Serum levels of MMP7, PARC, and SP-D across the spectrum of RA-ILD in the combined Brigham and Women's Hospital Rheumatoid Arthritis Sequential Study and American College of Rheumatology cohorts. Box-and-whisker plot depicting levels of MMP7, PARC, and SP-D in serum samples of patients with no RA-ILD, subclinical RA-ILD, and clinically evident RA-ILD. Circles represent outliers. P values identify statistically significant differences between subgroups as determined by unadjusted logistic regression analyses. MMP7: Subclinical RA-ILD versus no RA-ILD, P < 0.0001; clinically evident RA-ILD versus no RA-ILD, P < 0.0001. PARC: Subclinical RA-ILD versus no RA-ILD, P = 0.0001; clinically evident RA-ILD versus no RA-ILD, P = 0.0031. SP-D: Subclinical RA-ILD versus no RA-ILD, P = 0.0001; clinically evident RA-ILD versus no RA-ILD, P < 0.0001. MMP = matrix metalloproteinase; PARC = pulmonary and activation-regulated chemokine; RA-ILD = rheumatoid arthritis–associated interstitial lung disease; SP-D = surfactant protein D.