Serum biomarkers of pulmonary damage and risk for progression of rheumatoid arthritis-associated interstitial lung disease

Abstract

Objectives:

To investigate baseline and change of pulmonary damage biomarkers (serum Kreb von den Lungen 6 [KL-6], human surfactant protein D [hSP-D], and matrix metalloproteinase 7 [MMP7]) with rheumatoid arthritis-associated interstitial lung disease (RA-ILD) progression.

Methods:

In the Korean RA-ILD (KORAIL) cohort, a prospective cohort, we enrolled RA patients with ILD confirmed by chest computed tomography (CT) imaging and followed annually. ILD progression was defined as worsening in physiological and radiological domains of the 2023 ATS/ERS/JRS/ALAT guideline for progressive pulmonary fibrosis (PPF). Associations between biomarkers and RA-ILD progression were analyzed using multivariable Cox regression, adjusting for potential confounders.

Results:

We analyzed 136 RA-ILD patients (mean age 66.5 years, 30% male, 60% usual interstitial pneumonia pattern). During a median 3.0 years of follow-up, 47 patients (35%) had progression. Higher baseline KL-6 and hSP-D levels were associated with higher risk of ILD progression (multivariable HRs 1.37, 95%CI 1.03‒1.82 and 1.51, 95%CI 1.09‒2.08, respectively), whereas only the highest quartile of MMP7 showed an increased risk (multivariable HRs 2.60, 95%CI1.07‒6.33). Increasing levels of serum KL-6 at 1 year showed the strongest association with progression (ΔKL-6: multivariable HR 2.00, 95%CI 1.29‒3.11), additionally adjusting for baseline biomarker levels.

Conclusion:

In this first prospective study to apply PPF criteria to RA-ILD, 35% progressed over 3 years. Higher baseline KL-6 and hSP-D were associated with progression. In follow-up, greater changes in KL-6 and MMP7 were associated with progression. Serial measurement of pulmonary damage biomarkers may predict RA-ILD progression and may be helpful in monitoring patients and treatment decisions.

Introduction

Rheumatoid arthritis-associated interstitial lung disease (RA-ILD) is a severe extra-articular manifestation with high morbidity and mortality. Despite advances in RA treatment, mortality among patients with RA-ILD has not been improved compared to those with RA and no ILD.^1–3 This persistent high mortality is likely due to the progression of ILD.^4,5 However, antifibrotic treatment have been shown to slow the rate of lung function decline in RA-ILD patients with a progressive fibrosing phenotype.^6,7 Therefore, determining which patients are likely to experience disease progression – and thus would benefit from early intervention- remains a significant challenge requiring further research. In this regard, biomarkers may play a significant role in offering predictive insights for such progression.

Several biomarkers, including cytokines and pulmonary damage biomarkers, have been investigated in idiopathic pulmonary fibrosis (IPF), especially with respect to prognosis and monitoring of treatment response.^8–13 Similarly, these biomarkers have been studied in RA-ILD, but primarily with a focus on diagnosis and mortality.^14–19 In a previous cross-sectional study, we investigated the association between ILD severity and serum biomarkers related to inflammation (interleukin-6 and tumor necrosis factor-α) and pulmonary damage (Krebs von den Lungen-6 [KL-6], human surfactant protein-D [hSP-D], and matrix metalloproteinase-7 [MMP7]).²⁰ In line with the previous research,^{11,12,14,16,18,19,21–26} our findings indicated that KL-6, hSP-D, and MMP7 were each associated with physiologic measures (forced vital capacity [FVC] and diffusing capacity of the lungs for carbon monoxide [DLCO]) or the semiquantitative assessment of ILD on chest CT scans. However, it is unclear whether these biomarkers may predict ILD progression among patients with RA.

In this study, we aimed to determine the relationship between pulmonary damage biomarkers and RA-ILD progression, using blood samples at baseline as well as their longitudinal changes at 1 year follow-up.

Methods

Study sample and design

We analyzed data from the Korean Rheumatoid Arthritis Interstitial Lung Disease (KORAIL) cohort, a multicenter prospective longitudinal observational study from six tertiary hospitals in Korea. Ethical approval was obtained from the ethics committees of all six participating centers, adhering to the Declaration of Helsinki and Good Clinical Practice guidelines (Supplement Table S1). All participants provided written informed consent.

Participants aged 18 years and above diagnosed with both RA and ILD were recruited at an outpatient rheumatology clinic, spanning from January 2015 to July 2018. RA was diagnosed according to the 2010 American College of Rheumatology (ACR)/European Alliance of Associations for Rheumatology (EULAR) criteria,²⁷ and ILD was identified through chest computed tomography (CT) scans by radiologists of each medical centers. Participants were followed annually for three years (last follow-up: September 2021), with a ±3-month window allowed around the scheduled follow-ups as this observational study aligned with routine clinical practice. We analyzed the participants who had complete results on each biomarker and covariates. Study cohort profiles and comparisons of clinical characteristics between participants excluded due to missing data and those included in the analysis indicated in Supplementary Figure S1 and Supplementary Table S2, respectively.

Exposure variables: Pulmonary damage biomarkers

Serum samples were collected at annual study visits and preserved at −80°C. KL-6 levels were quantified by the Nanopia KL-6 assay (Sekisui Medical, Tokyo, Japan) employing latex-enhanced immunoturbidimetric assay techniques. hSP-D and MMP7 were measured using the R-Plex assay platform (MSD, Catalog numbers K1519XR-2 and K1510KR-2, respectively).

In this study, biomarker levels were co-primary exposure variables. The primary analysis investigated natural log-transformed continuous and standardized for each the baseline biomarker. Secondary analyses examined quartiles and dichotomized variables for each biomarker. We analyzed the association between baseline biomarker levels and ILD progression, and between 1-year changes in these levels and ILD progression.

Outcome: ILD progression

Chest high-resolution computed tomography (HRCT) scans at end-inspiration of 1 to 2 mm sections were performed annually. Two experienced radiologists evaluated chest HRCT scans independently, blinded to the clinical scenario and timing of study visits. ILD patterns were determined according to the official guideline from the American Thoracic Society, European Respiratory Society, Japanese Respiratory Society and Latin American Thoracic Society (ATS/ERS/JRS/ALAT) practice guideline. ²⁸ For the assessment process, we adapted the visual quantitative scoring system of Scleroderma Lung Study, with modifications. ²⁹ Briefly, the lung was divided into six zones (upper, middle, and lower for right and left) for visual lesion assessment. Lung lesions such as ground-glass opacities (GGOs), reticular opacity (RO), traction bronchiectasis and bronchiectasis, honeycombing (HC), and emphysema were scored on a semi-quantitative scale based on extent of involvement (0%, 1–25%, 26–50%, 51–75%, and 76–100%). ILD extent was categorized as ≤10% or >10%.

ATS/ERS/JRS/ALAT clinical practice guidelines define progressive pulmonary fibrosis (PPF) as a composite of worsened respiratory symptoms, pulmonary physiology evidence of disease progression, and radiological evidence of disease progression. ³⁰ The definitions and criteria for both the physiologic and radiological domains, including changes in FVC (% predicted) or DLCO (% predicted) within one year of observation or as well as radiological features, are presented in Table 1. In the KORAIL cohort, physiological and radiological measurements were obtained on an annual basis, but respiratory symptoms were not collected. Thus, to be defined as an ILD progressor, both the physiologic and radiologic domains had to be met. The date of ILD progression was the first abnormal domain item obtained.

Table 1.

Definition of progression and numbers of patients with RA-ILD fulfilling each item during follow-up (n=136).

ILD Progression Items and Domains	N (%)
1. Physiological evidence of disease progression (either of following)	111 (81.6)
Absolute decline in FVC % pred. ≥5% within 1 year of follow-up	108 (79.4)
Absolute decline in DLco ≥10% within 1 year of follow-up	31 (22.8)
2. Radiological evidence of disease progression	56 (41.2)
Increased extent (see categories in Table 2)	28 (20.6)
New GGO + New traction bronchiectasis/bronchiolectasis	10 (7.4)
New reticular opacity	41 (30.1)
New or increased honeycombing	31 (22.8)
Outcome definition (fulfilling criteria 1 and 2): Progressive fibrosing lung disease	47 (34.6)

DL_CO, diffusing capacity for carbon monoxide; FVC % pred., forced vital capacity percent predicted; GGO, ground glass opacity; RA-ILD, rheumatoid arthritis-associated interstitial lung disease

Covariates

Covariates were obtained at baseline and each study visit, including age (continuous), sex at birth (male/female), and ever smoking. For RA-related variables, we collected baseline RA duration (continuous), ESR, RF, Disease Activity Score of 28 joints (DAS28)-ESR as a continuous variable and categorical (remission/low/moderate/severe), serologic status (rheumatoid factor [RF] and anti-citrullinated protein antibody [ACPA]), and RA medications (disease modifying anti-rheumatic drugs [DMARDs], including conventional synthetic DMARDs, biologic or targeted synthetic DMARDs, and glucocorticoid use and dose). For ILD characteristics, we collected ILD duration since ILD diagnosis based on images and interval from RA to ILD diagnosis, FVC, DLCO, and HRCT characteristics.

Statistical analysis

We detailed frequencies and proportions for ILD progression criteria, participant demographics, RA activity, and lung status. Categorical data were reported using frequencies, continuous data as means (SD) if normal, and medians (IQR) if not. The sample was analyzed overall and by ILD progression during follow-up. Pulmonary biomarker levels at baseline and after 1-year visit were compared using Wilcoxon rank sum tests, both between and within groups.

In the primary analysis, we used multivariable Cox regression models with each log-transformed biomarker per SD to obtain hazard ratios (HRs) for ILD progression, confirming the proportional hazards assumption through Schoenfeld residuals. The multivariable model was adjusted for traditional risk factors for RA-ILD, including age, RA duration, sex, smoking, and baseline DAS28-ESR. We included physiologic or radiologic variables for sensitivity analysis since these are likely on the causal pathway between the pulmonary damage biomarkers and ILD progression. Other variables such as specific DMARDs, ESR, and CRP were not included due to similar baseline distributions. We performed secondary analyses using quartiles of each pulmonary disease biomarker to evaluate the risk of ILD progression. We constructed Kaplan-Meier curves for these quartiles and calculated p for trend using the multivariable Cox model. We calculated the 1-year change (represented as delta biomarkers Δbiomarker; ΔKL6, ΔhSP-D, and ΔMMP7) for each pulmonary damage biomarker and analyzed them per SD unit and by quartiles, as in the baseline analysis. The multivariable Cox regression model was applied to the log-transformed Δbiomarkers per SD units and each quartile to determine the HRs for ILD progression, adjusting for each baseline biomarker levels, in addition to the variables in the baseline multivariable model. We defined the outcome date as the earliest date on which either the PFT or CT scan met the ILD progression criteria. Censoring dates were set as the last study visit of either PFT or CT, whichever occurred first. Given that outcome was RA-ILD progression (Table 1), all outcomes in this analysis occurred at or after the 1-year study visits. For the 1-year biomarker change, sensitivity analysis was performed with the time origin at the 1-year visit date (Supplementary Table S9).

To enhance their association with RA-ILD progression, we integrated the three biomarkers by assigning scores based on their quantiles (e.g., Q1 = 1), and summing them into a composite score (Supplementary Table 8). In the multivariable models, we examined the HRs per SD unit, adjusting for age, sex, smoking status, and baseline DAS28-ESR scores; for composite scores of Δbiomarkers, we additionally adjusted for the baseline composite score. To compare the effectiveness of individual biomarkers, their 1-year changes, and the composite scores, we generated receiver operating characteristic (ROC) curves and calculated the area under the curve (AUC).

Statistical significance was a two-sided p value <0.05. We adjusted for multiple comparisons using the Benjamini-Hochberg false discovery rate (FDR) correction method, considering findings significant at FDR < 0.05. All analyses were performed using R version 4.3.2.

Results

Sample size and ILD progression

We analyzed a total of 136 participants with RA-ILD that had pulmonary damage biomarkers measured at baseline. During median follow-up of 3.0 years (IQR: 2.8‒3.4), 47 (34.6%) had ILD progression, while 89 (65.4%) had no progression. The domains and items of ILD progression are shown in Table 1.

Baseline characteristics

Among all 136 participants, the mean age was 66.5 (8.3) years, 30.1% were male, and 25.9% were ever smokers. The median RA and ILD durations were 6.3 years (IQR 1.0‒ 10.7) and 1.7 years (IQR 0.1‒4.9), respectively. The median interval between RA and ILD diagnoses was 1.2 years (IQR 0.0‒7.2). Nearly all (99.3%) were seropositive; 88.2% were RF-positive and 95.6% were ACPA-positive. Seventy-five percent of participants had ever used csDMARDs as of study enrollment, while 27.2% had ever used b/tsDMARDs. Most participants (87.5%) were receiving glucocorticoids. The baseline demographic features of participants with and without ILD progression as well as the entire study population are shown in Table 2. There were no differences in baseline characteristics between the two groups.

Table 2.

Baseline characteristics and treatment summary of patients with rheumatoid arthritis-associated interstitial lung disease (n=136).

	All RA-ILD patients	ILD progression a	No ILD progression
n (%)	136 (100.0)	47 (34.6)	89 (65.4)
Mean age, years (SD)	66.5 (8.3)	66.6 (8.2)	66.5 (8.3)
Male, n (%)	41 (30.1)	15 (31.9)	26 (29.2)
Ever smoking, n (%)	35 (25.9)	11 (23.4)	24 (27.3)
Current	12 (8.8)	2 (4.3)	10 (11.2)
Ex-smoking	23 (16.9)	9 (19.1)	14 (15.7)
Median RA duration, years (IQR)	6.3 (1.0–10.7)	6.0 (0.9–10.0)	6.4 (1.2–13.3)
Median ILD duration, years (IQR)	1.7 (0.1–4.9)	1.2 (0.1–4.3)	1.9 (0.2–5.0)
Median interval between RA and ILD diagnosis, years (IQR)	1.2 (0.0–7.2)	0.4 (0.0–6.3)	1.6 (0.0–8.3)
Within 6 months	50 (36.8)	20 (42.6)	30 (33.7)
RA before ILD	77 (56.6)	23 (48.9)	54 (60.7)
ILD before RA	9 (6.6)	4 (8.5)	5 (5.6)
Seropositive RA, n (%)	135 (99.3)	48 (100.0)	87 (98.9)
RF-positive	120 (88.2)	43 (89.6)	88 (87.5)
ACPA-positive	130 (95.6)	46 (95.8)	88 (95.5)
RA treatmentb, n (%)
csDMARDs	102 (75.0)	35 (74.4)	67 (75.3)
Tacrolimus	26 (19.1)	10 (21.3)	16 (18.0)
Leflunomide	35 (25.7)	11 (23.4)	24 (27.0)
Sulfasalazine	56 (41.2)	16 (34.0)	40 (44.9)
Hydroxychloroquine	68 (50.0)	26 (55.3)	42 (47.2)
Methotrexate	72 (52.9)	23 (48.9)	49 (55.1)
b/tsDMARDs	37 (27.2)	11(23.4)	26 (29.2)
TNF inhibitor	22 (17.2)	7 (14.9)	15 (16.9)
Glucocorticoids	119 (87.5)	38 (80.9)	81 (91.0)
Mean prednisone daily dose, mg/day (SD)	4.5 (3.2)	4.8 (4.5)	4.3 (2.3)
RA disease activity
Swollen joint counts, mean (SD)	2.7 (13.9)	2.6 (3.2)	2.8 (3.6)
Tender joint counts, mean (SD)	3.4 (4.8)	2.8 (15.8)	3.7 (12.8)
ESR (mm/hr), mean (SD)	39.3 (26.5)	39.8 (25.7)	39.0 (27.1)
CRP (mg/L), mean (SD)	8.8 (13.9)	9.9 (15.8)	8.2 (12.8)
DAS28-ESR
Mean score (SD)	4.0 (1.5)	4.0 (1.3)	4.0 (1.5)
Remission	27 (19.9)	6 (12.8)	21 (23.6)
Low	19 (14.0)	9 (19.1)	10 (11.2)
Moderate	60 (44.1)	24 (51.1)	36 (40.4)
High	30 (22.1)	8 (17.0)	22 (24.7)
ILD status
Pulmonary function tests
FVC mL, mean (SD)	2502 (752)	2449 (736)	2530 (763)
FVC % predicted, mean (SD)	84.6 (16.7)	82.4 (15.8)	85.8 (17.2)
FVC % predicted <80%, n (%)	46 (33.8)	17 (36.2)	29 (32.6)
Chest CT image assessment
Pattern
Definite/probable UIP	82 (60.3)	36 (76.6)	46 (51.7)
Others	54 (39.7)	11 (23.4)	43 (48.3)
Extent
≤ 10%	85 (62.5)	20 (42.6)	65 (73.0)
>10%	51 (37.5)	27 (57.4)	24 (27.0)

^aILD progression is defined in Table 1.

^bSpecific medications with >10% prevalence are listed.

ACPA, anti- citrullinated protein antibodies; csDMARDs, conventional synthetic disease-modifying antirheumatic drugs; b/tsDMARDs, biologic/targeted synthetic disease-modifying antirheumatic drugs; ILD, interstitial lung disease; IQR, interquartile range; RA, rheumatoid arthritis; SD, standard deviation; TNF, tumor necrosis factor; CT chest tomography; CRP, c-reactive protein; DAS, disease activity score; ESR, erythrocyte sedimentation rate; FVC, forced vital capacity; NSIP, nonspecific interstitial pneumonia; OP, organizing pneumonia; RA-ILD, rheumatoid arthritis associated interstitial lung disease; UIP, usual interstitial pneumonia

Baseline RA disease activity

RA disease activity of participants with and without ILD progression at the time of study enrollment are shown in Table 2. RA disease activity measurements, including swollen joint counts, tender joint counts, and the mean of serum ESR and CRP level were similar between participants with and without ILD progression. At the time of enrollment, 68.0% of patients with progression exhibited moderate to high RA disease activity, as did 65.1% of patients without progression. The mean of DAS28-ESR was also similar between participants with and without ILD progression (4.0 [SD 1.3] vs. 4.0 [SD 1.5]).

Baseline ILD status

At the time of enrollment, the mean of FVC% was similar between participants with ILD progression and those without ILD progression (82.4% [SD 15.8] vs 85.8% [SD 17.2]). The proportion of participants with definite or probable UIP patterns on chest CT was significantly higher in participants with ILD progression than in those without ILD progression (76.6% vs 51.7%). The more extensive extent of ILD at baseline defined as 10% or more involvement was also greater in participants with ILD progression than in those without ILD progression (27/47 [57.4%] vs 24/89 [27.0%]).

Baseline biomarker levels and 1-year change

At baseline, the median levels of all three biomarkers were higher in participants with ILD progression than in participants without progression (Figure 1 and Supplementary Table S3).

Figure 1.

Pulmonary damage biomarker results at baseline and 1 year follow-up among patients with RA-ILD. (A) KL-6, (B) hSP-D, and (C) MMP7

hSP-D, human surfactant protein-D; ILD, interstitial lung disease; KL-6, Krebs von den Lungen-6; MMP7, matrix metalloprotein 7; RA-ILD, rheumatoid arthritis-associated interstitial lung disease

At subsequent 1 year follow-up, serum KL-6 levels further increased in participants with ILD progression, hence the change in serum KL-6 levels at 1 year follow-up was increased in participants with ILD progression while decreased in those without ILD progression (ΔKL-6, median 28.1 U/mL [IQR −68.9 to 144.0] vs −42.6 [IQR −121.8 to 31.9]). However, serum hSP-D levels decreased in both participants with and without ILD progression, which resulted in similar change in serum hSP-D levels (ΔhSP-D −1.4 ng/mL [IQR −3.1 to 1.0] vs −1.3 ng/mL [IQR −2.5 to 0.4]) at 1 year follow-up. Serum MMP7 levels decreased in both participants with and without ILD progression and change in serum MMP7 levels at subsequent 1 year follow-up was similar (ΔMMP7: −2.0 ng/mL [IQR −3.1 to 0.1] vs −1.1 ng/mL [IQR −2.9 to −0.2]).

Associations of pulmonary damage biomarkers with ILD progression

The unadjusted analysis results are detailed in Supplementary Table S4, with the multivariable analysis presented in Figures 2 and 3, as well as in Supplementary Table S5.

Figure 2.

Multivariable hazards ratios for RA-ILD progression by each biomarker. (A) Continuous variable (per SD), (B) Baseline biomarker levels in quartiles, and (C) 1-year change of each biomarker level in quartiles.^c

^a Values were log-transformed when analyzing.

^b Adjusted for age, sex, smoking, interval between RA and ILD diagnosis, DAS28-ESR at baseline.

^c Adjusted for age, sex, smoking, interval between RA and ILD diagnosis, baseline DAS28-ESR, and baseline value of each biomarker (per ln-unit) in case of Δ biomarker analysis or baseline composite pulmonary damage biomarker score in case of composite score analysis

^d p-value corrected by false-discovery rate correction

CI, confidence interval; DAS, disease activity score; ESR, erythrocyte sedimentation rate; HR, hazard ratio; hSP-D, human surfactant protein-D; ILD, interstitial lung disease; IQR, interquartile range; KL-6, Krebs von den Lungen-6; MMP7, matrix metalloprotein 7; RA-ILD, rheumatoid arthritis-associated interstitial lung disease; ROC, receiver operating characteristic; SD, standard deviation

Figure 3.

ILD progression-free survival stratified by quartile of baseline pulmonary damage biomarker quartiles among patients with RA-ILD. (A) KL-6, (B) hSP-D, and (C) MMP7 levels are presented, showing (1) baseline measurements and (2) changes observed at the 1-year follow-up.

hSP-D, human surfactant protein-D; KL-6, Krebs von den Lungen-6; MMP7, matrix metalloprotein 7; RA-ILD, rheumatoid arthritis-associated interstitial lung disease

KL-6 had a multivariable HR for ILD progression of 1.37 (95% CI: 1.03‒1.82) for each SD increase in the baseline serum levels, adjusted for age, sex, interval between RA and ILD diagnosis, DAS28-ESR baseline (Figure 2). Stratified by KL-6 quartiles, the HR of the highest quartile compared to the lowest was 2.95 (95%CI 1.26‒6.95, Figure 2 and 3). Among participants with available 1-year follow-up KL-6 data (n=121/136, 89.0%), the multivariable HR of ILD progression per SD 1-year change in KL-6 levels was 2.00 (95%CI 1.29‒3.11), additionally adjusted for baseline value of each biomarker. The multivariable HR for of participants in the highest KL-6 quartile was 4.47 (95%CI 1.78‒11.2) compared to those in the lowest quartile. Dichotomizing by the highest quartile versus other quartiles, the multivariable hazard ratio (HR) for the top quartile’s 1-year KL-6 change was 2.83 (95% CI: 1.48‒5.40) with adjustments for ILD extent, and 2.83 (95% CI: 1.50‒5.35) with adjustments for ILD pattern in sensitivity analyses (Supplementary Table S6).

The multivariable HR of ILD progression was 1.51 (95%CI 1.09‒2.08) with each SD increase in the baseline serum levels of hSP-D (Figure 2). Stratified by quartiles, the multivariable HRs of the third and fourth hSP-D quartiles were 3.26 (95%CI 1.30‒8.17) and 3.03 (95%CI 1.15‒7.96), respectively, compared to the first quartile (Figure 2 and 3). Contrary to the patterns observed with KL-6, however, this association did not extend to changes in serum hSP-D levels over the subsequent one-year follow-up (multivariable HR 1.43 [95%CI 0.94‒2.18]), among participants with available 1-year follow-up hSP-D data (n=97/136, 71.3%).

Although the multivariable HR of ILD progression was 2.60 (95%CI 1.07‒6.33) for participants with the highest MMP7 quartile versus the lowest, the HR per SD change in baseline MMP7 levels was 1.22 (95%CI 0.82‒1.80, Figure 2A). Among participants with available 1-year follow-up MMP7 data (n=97/136, 71.3%), the HR of ILD progression per SD change in serum MMP7 levels was 1.32 (95% CI 0.82‒2.12). Stratified by quartiles, the HR of the second highest and the highest quartile was 1.12 (95% CI: 0.40‒3.15) and 1.37 (95% CI: 0.47, 4.00), respectively, while the HR of the third highest was 0.31 (95% CI: 0.07‒1.36) compared to the lowest (Figure 2 and 3). Comparing the highest quartile to the remaining, the HR was 1.94 (95% CI: 1.04‒3.65, Supplementary Table S7).

The HR of composite score with all three of baseline pulmonary damage biomarker measurement was 1.73 (95%CI 1.27‒2.35, Figure 2A). The HR of composite score with all three 1-year follow-up pulmonary damage biomarker measurement was 1.62 (95%CI 1.16‒2.78). Of note, the AUC of composite score with all three 1-year follow-up pulmonary damage biomarker (AUC=0.77, 95%CI 0.68‒0.86) was the greatest followed by a 1-year change in serum KL-6 (AUC=0.71, 95%CI 0.62‒0.80) and composite score with baseline pulmonary damage biomarkers (AUC=0.68, 95%CI 0.59‒0.78) (Figure 4 and Supplementary Table S8).

Figure 4.

Areas under the receiver operating characteristics curves for RA-ILD progression by baseline demographic, RA factors, ILD factors, and biomarkers.

Model descriptions:

- Model 1: KL-6 ^a,b

- Model 2: hSP-D ^a,b

- Model 3: MMP7 ^a,b

- Model 4: Delta KL-6 ^c

- Model 5: Composite score with baseline biomarker measurements ^b

- Model 6: Composite score with changes in biomarker measurements ^c

^a Values were log-transformed when analyzing.

^b Adjusted for age, sex, smoking, interval between RA and ILD diagnosis, DAS28-ESR at baseline.

^c Adjusted for age, sex, smoking, interval between RA and ILD diagnosis, baseline DAS28-ESR, and baseline composite pulmonary damage biomarker score

CI, confidence interval; DAS, disease activity score; ESR, erythrocyte sedimentation rate; HR, hazard ratio; hSP-D, human surfactant protein-D; ILD, interstitial lung disease; IQR, interquartile range; KL-6, Krebs von den Lungen-6; MMP7, matrix metalloprotein 7; RA-ILD, rheumatoid arthritis-associated interstitial lung disease; ROC, receiver operating characteristic; SD, standard deviation

Discussion

In this study, we prospectively evaluated the predictive value of three pulmonary damage biomarkers for RA-ILD progression using biomarker levels, PFTs, and HRCT. We found that 35% of RA-ILD patients progressed according to serial physiologic and HRCT parameters, applying the modified ATS/ERS/JRS/ALAT criteria for PPF for the first time in RA patients. ²⁸ Higher baseline KL-6 levels and their 1-year changes were consistently associated with an increased risk of ILD progression, independent of age, sex, smoking, ILD duration, and RA disease activity. We also identified associations between higher hSP-D levels, as well as the highest quartile of MMP-7, and ILD progression. These novel findings suggest that pulmonary damage biomarkers may be useful to predict RA-ILD progression and also can shed light on the pathogenesis of RA-ILD progression.

KL-6 is an extracellular subunit of mucin 1 (MUC1) expressed on the cell surface of primarily epithelial cells including regenerating type II alveolar epithelial cells (AEC).^{31 32,33} In IPF, serum levels of KL-6 are associated with an increased acute exacerbation risk and mortality, ^22,33 as well as changes in serum levels of KL-6. ^8,12,25 In RA, serum levels of KL-6 are a diagnostic biomarker for RA-ILD independently ^14,15 or conjunction with B-line of ultrasonography. ^34,35 Serum levels of KL-6 were associated with severity of ILD, especially fibrosis or reticular opacities, ^14,36 and higher levels of KL-6 were associated with ILD progression and mortality, especially among patients with UIP patterns. ^24,26 However, most previous research was based on retrospective studies and lacked data on RA disease status. ^{21,24,26,36,37} In our current study, we verified that the risk for ILD progression increased with higher serum levels of KL-6 at baseline and subsequent 1 year change after adjustment covariates including RA disease activity, and used HRCT parameters, rather than only FVC, to define progression. Of note, increased risk of RA-ILD progression was observed in participants with the largest annual changes in KL-6, even with further adjustment for ILD patterns or extent (Supplementary Table S9).

Serum hSP-D levels in IPF are known to correlate with severity, progression, and treatment response to antifibrotic agents, indicating its potential as a biomarker for lung pathology. ^9,11,38 In RA-ILD, elevated serum SP-D levels alone, as well as in combination with MMP7 and other variables, were associated with the differentiation and increased risk of ILD in RA patients.^19,26 Our current study further elucidates that baseline hSP-D levels are associated with RA-ILD progression, though this predictive value may diminish over time. Our study aligns with previous research indicating that stable⁹ or decreased serial serum hSP-D levels in IPF patients in both the anti-fibrotic and placebo groups during up to 52 weeks of follow-up.¹¹ This may be attributed to increased excretion to alveolar space rather than blood circulation^39,40 or to compromised alveolar epithelial cell function while developing lung fibrosis, which lead to impairment of surfactant synthesis.^41,42

MMP-7 is recognized as a useful biomarker for differentiating fibrosing ILD from other ILD types, as well as distinguishing RA-ILD from non-RA-ILD.^16–19 In IPF, studies show varying associations between MMP-7 levels and progression or mortality, though high MMP-7 is consistently tied to worse prognosis.^{13,23,41,43 44} In the current study, only the highest quartile of baseline MMP-7 was associated with an increased risk of RA-ILD, but baseline MMP-7 levels and their 1-year change did not correlate with RA-ILD progression. Notably, in the current study, MMP-7 levels of non-progression group decreased at1-year follow-up from a median of 5.7 ng/mL to 4.5 ng/mL (FDR corrected p value = 0.0003) (Figure 1 and Supplementary Table 2), while previous research indicates that MMP-7 levels in IPF patients either remain stable⁹ or increase with declining forced vital capacity (FVC).⁴⁵ This suggests differing roles of inflammation in fibrosis between IPF and RA-ILD. Studies in inflammatory diseases like psoriasis⁴⁶ and lupus⁴⁷ have shown a relationship between elevated MMP7 and increased disease activity. In RA patients, other MMPs are elevated, facilitating bone erosion and neutrophil influx.⁴⁸ In our cohort, 34% (46/136) of participants had low disease activity at baseline, increased to 52% (66/127) at the 1-year follow-up. These findings, along with KL-6 and hSP-D results, suggest that RA-ILD has distinct pathologic features compared to IPF and implicate usefulness of serial MMP7 levels as negative predictive marker for RA-ILD progression.

Our study has several strengths. We had serial measures of pulmonary damage biomarkers, chest HRCT, and pulmonary function tests allowing us to investigate whether biomarkers are associated with RA-ILD progression throughout a multi-center prospective study with several years of follow-up. This is one of the first studies to implement the ATS/ERS/JRS/ALAT Clinical Practice Guideline criteria for PPF in an RA population.²⁸ We also had rich data on covariates for adjustment, including validated measures of RA disease activity.

This study has several limitations to consider. Most participants had mild ILD, which reflects typical of RA-ILD in Rheumatology out-patient clinics. Annual follow-up may lead to interval censoring, while strict progression criteria and a 10% yearly dropout could underestimate RA-ILD progression or introduce bias. The impact of RA treatments on biomarkers was not analyzed due to the limited number of patients across various treatments. Respiratory symptoms were not recorded and visual CT scoring, rather than automated methods,⁴⁹ was used, potentially limiting precision. There may also be a risk of immortal time bias in the 1-year change analysis; however, the sensitivity analysis still demonstrated the usefulness of the 1-year change in KL-6 (Supplementary Table 9). Lastly, a small sample size from the Korean population may limit generalizability. Future research with larger, diverse populations and consideration of unmeasured factors, including genetic variants and autoantibodies, is warranted.

In conclusion, we identified serum KL-6, hSP-D, and MMP7 as pulmonary damage biomarkers associated with ILD progression in a prospective RA-ILD cohort. We found that 35% met the ATS/ERS/JRS/ALAT definition for PPF. ²⁸ Elevated serum levels of KL-6 and increases their levels at the subsequent 1-year follow-up were significantly associated with increased RA-ILD progression risk. Elevated serum levels of hSP-D and the highest quartile of MMP7 levels were also associated with a higher RA-ILD progression risk. These biomarkers collectively offer insight into the ILD progression, highlighting their potential as key indicators for monitoring disease trajectory in RA-ILD patients.