Type III and type VI collagen neoepitopes are associated with disease severity in systemic sclerosis

Abstract

Objective:

Dysregulated collagen turnover is implicated in systemic sclerosis (SSc) pathogenesis. We evaluated collagen turnover biomarkers in relation to the severity of fibrotic manifestations, key cytokines, and progression in SSc.

Methods:

Baseline and 6-month serum samples of early SSc patients in the CONQUER cohort were analyzed for type III (PRO-C3 and C3M) and type VI (PRO-C6 and C6M) collagen turnover biomarkers, as well as C-reactive protein (CRP), interleukin-6 (IL-6), and interferon (IFN)-inducible proteins. The modified Rodnan skin score (mRSS) and forced vital capacity percent predicted (FVC%) served as surrogate markers of disease severity.

Results:

222 patients were included. PRO-C3 (p<0.001) and PRO-C6 (p<0.001) concentrations were higher in patients with diffuse disease, while C6M (p=0.041) was higher in those with ILD. Baseline PRO-C3 (p<0.001) and PRO-C6 (p<0.001) positively correlated with mRSS, whereas C3M (p=0.029) and C6M (p=0.011) negatively correlated with FVC%, although the magnitude of the observed correlations was in the weak range (Rs<0.4). Collagen biomarker concentrations positively correlated with CRP, IL-6, and IFN-inducible proteins at baseline. While changes in CRP correlated positively with changes in collagen degradation protein levels (C3M and C6M), they did not correlate with changes in collagen formation protein levels (PRO-C3 and PRO-C6). In contrast, changes in IFN score showed the highest correlation with changes in PRO-C6.

Conclusion:

PRO-C3 and PRO-C6 correlated with skin disease severity, while C3M and C6M correlated with lung disease severity. Collagen turnover biomarkers correlated with CRP, IL-6, and IFN-inducible proteins, providing support for the link between inflammation and fibrosis in SSc.

Introduction

Systemic sclerosis (SSc) is a chronic systemic autoimmune rheumatic disease characterized by immune dysregulation, vasculopathy, and fibrosis ¹. Although fibrosis of the skin is the hallmark of the disease, fibrosis of the internal organs is the leading cause of morbidity and mortality. Specifically, interstitial lung disease (ILD) is present in up to 60% of SSc patients ² and is the most common cause of disease-related death in patients with SSc ^3,4.

Fibrosis results from excessive extracellular matrix (ECM) deposition in the parenchyma, with collagen as the most abundant protein. An imbalance between collagen production and degradation is implicated in the pathogenesis of fibrosis. Different types of collagens, including types I, III, VI, and VII, are overexpressed in the tissues of patients with SSc ⁵. Type I collagen is the most abundant protein in the skin, while type III collagen, though similarly distributed, is less abundant. Type III collagen is also an integral component of the gastrointestinal, lung, and vascular tissues and comprises 15% of skin collagen ^6,7. Type VI collagen is less studied and found in various tissues, including the dermis, lungs, kidneys, blood vessels, and adipose tissue ⁸. It is located near the basement membrane and anchors different membrane proteins. Our previous gene expression studies indicate that both Type III and Type VI collagen gene expression levels are elevated in SSc skin ^9,10.

SSc is a chronic progressive systemic disease in which patients can experience highly variable trajectories. There is an unmet need to identify better biomarkers to predict future disease severity and progression in SSc. The formation and degradation products of different types of collagens have been shown to correlate with disease severity and progression in other chronic fibrotic diseases. Previous studies in various diseases, such as inflammatory bowel disease ¹¹, hepatic fibrosis ¹², chronic kidney disease ¹³, and idiopathic pulmonary fibrosis (IPF) ¹⁴, suggested that these products might help monitor disease activity and predict prognosis. In SSc, earlier studies ^15–17 have shown that the concentrations of collagen biomarkers are elevated compared to healthy controls and correlated with the severity of skin and lung fibrosis. Moreover, a subsequent study ¹⁸ suggested that baseline concentrations of collagen biomarkers could be used to predict skin disease progression but not ILD progression. To our knowledge, no study has investigated whether the change in these biomarkers correlates with disease progression in SSc.

In the current study, we aimed to investigate whether the formation (pro-collagen III [PRO-C3] and pro-collagen VI [PRO-C6]) and degradation (collagen III M [C3M] and collagen VI M [C6M]) products of type III and type VI collagen are associated with the fibrotic complications of SSc and predict disease progression over time. In a separate analysis, we also examined whether these collagen turnover proteins correlate with key inflammatory cytokines in order to characterize the relationship between these collagen neoepitope fibrotic markers and markers of immune dysregulation in SSc To address these aims, we leveraged the longitudinal clinical data and biorepository of the nationally representative, US-based CONQUER (Collaborative National Quality and Efficacy Registry) cohort ¹⁹.

Methods

Study participants.

CONQUER is a multicenter, US-based registry that has been enrolling early SSc patients since 2018. The registry is representative of different geographical areas in the US, and 14 specialized SSc centers across the country were actively enrolling patients into the CONQUER cohort at the time of this study. All enrolled patients were adults (age ≥ 18 years) and fulfilled the 2013 ACR/EULAR Classification Criteria for SSc ²⁰. Patients were required to have a disease duration of less than five years based on the onset of the first non-Raynaud’s phenomenon symptom attributable to SSc. The study visits are scheduled every six months. This study was approved by each participating institution’s Institutional Review Boards (HSC-MS-18–0359) and was planned and conducted according to the Declaration of Helsinki. All participants gave written informed consent.

Clinical data collection.

Baseline characteristics of the patients, including disease manifestations and demographic information, were collected at the first visit using a standardized case report form. The presence of ILD was determined based on local reads of high-resolution computed tomography scans (HRCT) assessing for signs of ILD (reticular changes, ground glass opacity, honeycombing, or traction bronchiectasis). The modified Rodnan skin score (mRSS) was recorded at every visit by SSc expert rheumatologists to evaluate skin thickening, and patients were classified as having diffuse or limited disease according to the LeRoy criteria ²¹. Pulmonary function tests (PFTs) were performed at every visit. Forced vital capacity % predicted (FVC%) was used as a surrogate for ILD severity. FVC% values were calculated using the Global Lung Initiative (GLI) reference values ²².

Biomarker measurements and calculation of IFN-inducible protein score.

Blood samples were obtained from patients at the baseline and 6-month visits at the participating centers. Samples were shipped overnight to a central biorepository at the University of Texas Health Science Center at Houston and processed and stored according to a standardized protocol. Serum samples were stored in 0.5 ml aliquots in −80 °C freezers. Serum protein assays for collagen proteins were performed at the Nordic Bioscience Laboratory (Herlev, Denmark). A panel of four collagen turnover biomarkers quantifying type III and VI collagen formation (PRO-C3 and PRO-C6) and MMP (matrix metalloproteinase)-degraded type III and VI collagen (C3M and C6M) were measured in serum by competitive enzyme-linked immunosorbent assays (ELISAs) or the Immunodiagnostic Systems robotics platform (IDS_i10; Immunodiagnostic Systems (IDS), Bolden, Tyne & Wear, UK). All biomarkers are validated to measure in human serum samples ^23–26. Samples were rerun if the duplicate coefficient of variation was >15.0%. Intra- and inter-assay variation was <15.0%. All runs included the same three quality control samples, which were to be within the 20% range of the set target value.

The levels of PRO-C3, PRO-C6, C3M, and C6M were examined for correlation with serum C-reactive protein (CRP), interleukin-6 (IL-6), and a composite score of six interferon (IFN)-inducible proteins ²⁷. The CRP and IFN-inducible proteins were analyzed at the Rules Based Medicine Laboratories (Austin, TX, USA) using multianalyte profiling (MAP) multiplexed immune assay. IL-6 was measured by Simoa assays at the Rules Based Medicine Laboratories (Austin, TX, USA). A composite score of IFN-inducible proteins (IFN-score) was calculated using six proteins: monokine induced by IFNγ (MIG), IFNγ-inducible 10-kd protein (IP−-10), monocyte chemotactic protein 2 (MCP-2), β2-microglobulin (β2-m), tumor necrosis factor receptor type II (TNFRII), and macrophage inflammatory protein 3β (MIP-3β). The protein concentrations were first normalized to calculate the composite score by dividing each by the 95th percentile specific to that protein. Then, any values in the top 5% were assigned a score of 1.0. Lastly, the normalized values of the six proteins were summed to calculate the IFN-inducible protein score ²⁷.

Statistical analysis.

Summary statistics were used to describe baseline patient characteristics. The Wilcoxon signed-rank test was used to compare baseline and 6-month samples. Spearman’s correlation was used to evaluate the baseline relationship between biomarker concentrations, clinical features, and other proteins. The association between the biomarker concentrations and presence of diffuse disease, ILD, baseline mycophenolate use, and antibody status was analyzed using the Mann–Whitney 𝑈 test. A linear regression model, including mRSS and FVC% as predictors in the same model, was used to evaluate the relationship between the biomarkers and lung and skin disease. Log base 2 (log₂)-transformed PRO-C3, PRO-C6, C3M, and C6M concentrations were used in linear regression analyses to achieve normal distribution.

The change in biomarker concentrations was calculated as the percent change from baseline to 6 months for collagen biomarkers, CRP, IL-6, and the IFN-score. Pearson’s correlation was used to evaluate the relationship between the concentration changes. Similarly, Pearson’s correlation was used to assess the relationship between the percent change in mRSS and the percent change in collagen biomarker concentrations in patients with diffuse disease, as well as between the percent change in FVC% and the percent change in collagen biomarker concentrations in ILD patients over six months. To determine whether the change in the collagen biomarker is primarily driven by a change in mRSS or FVC%, we constructed a linear regression model with the change in biomarker concentration as the outcome and changes in mRSS and FVC% as the independent variables. To assess the predictive significance of individual collagen biomarkers for the progression of skin disease during follow-up visits in patients with diffuse disease, we used a linear mixed-effects model. The longitudinal mRSS obtained at all available follow-up visits served as the outcome variable, while baseline biomarker concentration, baseline mRSS, time, and mycophenolate use at each visit (time-varying) were used as independent variables. This analysis examines the course of mRSS as a continuous outcome variable during follow-up visits after adjustment for baseline mRSS level and mycophenolate treatment status. We accounted for between-patient variability in mRSS with a random intercept and between-patient variability in the change of mRSS over time with a random slope. This analytic approach has the advantage that all follow-up measurements can be included in the outcome variable. Next, the model was rerun, substituting the baseline biomarker level with the change in biomarker level. The same model was also applied to predict the progression of ILD using all available longitudinal FVC%s during follow-up visits as the outcome and biomarker concentration, baseline FVC%, time, and mycophenolate use at each visit as a time-varying variable, as the predictors in patients who had ILD.

Two-sided p-values less than 0.05 were considered statistically significant. All statistical analyses were performed using STATA/BE 18.0 (StataCorp LP, College Station, TX, USA) or R (R version 4.3.2).

Results

Baseline characteristics.

A total of 222 patients with baseline blood samples were included. The cohort included 185 female patients (83%), 29 African American patients (13%), and 27 Hispanic patients (12%). The mean disease duration from the first non-Raynaud’s symptom was 2.7 ± 1.4 years, and the mean (± SD) follow-up time was 23.2 ± 14.0 months. Anti-Scl-70 antibodies were present in 65 patients (33%), anti-RNA polymerase III antibodies in 48 patients (24%), and anti-centromere antibodies (ACA) in 22 patients (11%). Diffuse cutaneous disease was present in 132 patients (60%), and 153 patients (77%) had ILD (Table 1).

Table 1.

Baseline demographic and clinical characteristics of the CONQUER patients

Variable	n = 222
Female sex, n (%)	185/222 (83.3)
Age at the baseline visit, mean ± SD years	51.0 ± 13.9
Race, n (%)
White	175/221 (78.8)
African American	29/221 (13.1)
Asian	10/221 (4.5)
Others	8/221 (3.6)
Hispanic ethnicity, n (%)	27/221 (12.2)
Diffuse cutaneous involvement, n (%)	132/222 (59.5)
ILD diagnosis based on HRCT, n (%)	153/200 (76.5)
Disease duration, mean ± SD years	2.7 ± 1.4
Anticentromere antibody positive, n (%)	22/198 (11.1)
Anti-Scl-70 antibody positive, n (%)	65/198 (32.8)
Anti-RNA polymerase III antibody positive, n (%)	48/198 (24.2)
mRSS, mean ± SD	13.1 ± 10.3
FVC% predicted, mean ± SD	82.0 ± 19.1
DLCO% predicted, mean ± SD	77.9 ± 27.0
On mycophenolate at the baseline visit, n (%)	130/222 (58.6)

DLCO% = diffusing capacity for carbon monoxide percent predicted; FVC% = forced vital capacity percent predicted; HRCT = High-resolution computed tomography; ILD = interstitial lung disease; mRSS = modified Rodnan skin thickness score.

Biomarker concentrations according to clinical features.

Baseline concentrations of biomarkers were compared between patients with diffuse disease and limited disease, patients with ILD compared to those without ILD, and patients on mycophenolate compared to those not on mycophenolate at baseline. PRO-C3 and PRO-C6 concentrations were higher in patients with diffuse disease compared to those with limited disease (median [IQR]: 94 ng/ml [79.5–126.8] vs. 79.1 ng/ml [63.5–97.1], p < 0.001; 12.9 ng/ml [10.0–16.6] vs. 9.9 ng/ml [8.1–12.5], p < 0.001, respectively). C3M and C6M concentrations did not differ significantly between diffuse and limited patients (Figure 1). C6M concentration was significantly higher in patients with ILD compared to those without (20.2 ng/ml [16.2–24.7] vs. 17.6 ng/ml [15.8–21.8], p = 0.04; Supplementary Figure S1). Patients on mycophenolate at baseline had significantly higher PRO-C6 concentrations compared to those not on mycophenolate (12.2 ng/ml [9.3–15.5] vs. 10.6 ng/ml [8.7–14.6], p = 0.04) (Supplementary Table S1). Patients who were ACA-positive had lower PRO-C3 concentrations compared to ACA-negative patients (75.7 ng/ml [60.5–83.6] vs. 91.3 ng/ml [70.7–115.2], p = 0.01), and anti-Scl-70-positive patients had lower PRO-C3 concentrations compared to anti-Scl-70-negative patients (82.6 ng/ml [62.0–94.8] vs. 92.9 ng/ml [73.1–122.7], p = 0.004). In contrast, anti-RNA polymerase III (RNAP3) positive patients had higher concentrations of PRO-C3 (98.9 ng/ml [79.6–130.4] vs. 84.7 ng/ml [69.2–112.9], p < 0.001) and PRO-C6 (12.95 ng/ml [10.05–17.90] vs. 10.70 ng/ml [8.80–15.10], p = 0.001) compared to anti-RNAP3 negative patients. C3M and C6M concentrations did not differ according to the antibody status (Supplementary Table S2).

Figure 1.

PRO-C3 (A), PRO-C6 (B), C3M (C), and C6M (D) concentrations according to the type of skin involvement

^a Five outliers were omitted from the boxplot for visualization clarity; all data points were included in the statistical analyses.

Baseline correlations of biomarkers with disease duration, mRSS, and FVC%.

Baseline collagen turnover protein concentrations showed no significant correlation with disease duration (Table 2). In patients with diffuse disease, mRSS positively correlated with PRO-C3 (Spearman’s correlation Rs = 0.24, p = 0.005) and PRO-C6 (Rs = 0.34, p < 0.001). In patients with ILD, FVC% negatively correlated with PRO-C6 (Rs = −0.19, p = 0.02) and C6M (Rs = −0.19, p = 0.02). In a linear regression model that included mRSS and FVC%, PRO-C3 (r = 0.02, p < 0.001) and PRO-C6 (r = 0.02, p < 0.001) correlated with mRSS, while C3M (r = −0.003, p = 0.03) and C6M (r = −0.004, p = 0.01) correlated with FVC% (Supplementary Table S3).

Table 2.

Baseline correlations of biomarkers with mRSS, FVC%, and disease duration

	mRSSa		FVC%b		Disease duration
Biomarker	Spearman’s correlation (Rs)	p	Spearman’s correlation (Rs)	p	Spearman’s correlation (Rs)	p
PRO-C3	0.24	0.005	−0.13	0.12	−0.04	0.53
PRO-C6	0.34	< 0.001	−0.19	0.02	−0.11	0.11
C3M	0.02	0.79	−0.16	0.05	0.11	0.11
C6M	0.08	0.37	−0.19	0.02	0.13	0.06

FVC% = forced vital capacity percent predicted; mRSS = modified Rodnan skin thickness score.

^aIn patients with diffuse disease.

^bIn patients with interstitial lung disease diagnosis based on HRCT.

Baseline correlation of biomarkers with CRP, IL-6, and IFN-score.

PRO-C3, PRO-C6, C3M, and C6M positively correlated with each other (Supplementary Table S4). The strongest correlation was between C3M and C6M with Rs = 0.71 (p < 0.001), followed by the correlation between PRO-C3 and PRO-C6 Rs = 0.46 (p < 0.001).

Baseline CRP concentrations positively correlated with PRO-C6 (Rs= 0.16, p = 0.02), C3M (Rs=0.31, p < 0.001), and C6M (Rs=0.45, p < 0.001) (Table 3) Additionally, PRO-C3, PRO-C6, C3M, and C6M concentrations showed statistically significant correlations with both IFN-score and IL-6 concentrations. However, none of the observed correlations were in the moderate range except for the correlation between C6M and CRP (Rs = 0.45, p < 0.001).

Table 3.

Baseline correlation of biomarkers with other proteins and IFN-score

	CRP		IL-6		IFN-score
Biomarker	Spearman’s correlation (Rs)	p	Spearman’s correlation (Rs)	p	Spearman’s correlation (Rs)	p
PRO-C3	0.04	0.52	0.16	0.02	0.21	0.002
PRO-C6	0.16	0.02	0.26	< 0.001	0.37	< 0.001
C3M	0.31	< 0.001	0.22	< 0.001	0.26	< 0.001
C6M	0.45	< 0.001	0.17	0.01	0.22	0.001

CRP = C-reactive protein; IFN-score = interferon-score

Changes in biomarker concentrations over time.

Collagen biomarker concentrations at baseline and six months are shown in Supplemental Figure S2. PRO-C6 concentrations at six months were significantly lower than at baseline (p = 0.02). Concentrations of the other three collagen biomarkers did not differ significantly between baseline and six months.

Correlation of changes in collagen biomarker concentrations with clinical features..

The changes in collagen biomarker concentrations did not correlate with the change in mRSS. Only the change in C6M correlated with the change in FVC% (r=0.24, p=0.04) (Supplementary Table S5). In a linear regression model that included both changes in mRSS and FVC%, no significant relationship was observed between the changes in any of the collagen biomarkers and changes in mRSS and FVC% (Supplementary Table S6).

Correlation of changes in collagen biomarker concentrations with CRP, IL-6, and IFN-score.

The change in CRP correlated with changes in C3M (r=0.30, p < 0.001) and C6M (r=0.36, p < 0.001) concentrations, while the change in IL-6 did not correlate with changes in any of the collagen biomarkers. Change in IFN-score correlated with changes in PRO-C3 (r=0.19, p=0.01), PRO-C6 (r=0.36, p < 0.001), and C3M (r=0.21, p=0.007) concentrations (Table 4).

Table 4.

Correlation of change in collagen biomarker concentrations with changes in CRP and IL-6 concentrations, and IFN-score at 6 months

	ΔCRP		ΔIL-6		ΔIFN-score
Biomarker	Pearson’s correlation (r)	p	Pearson’s correlation (r)	p	Pearson’s correlation (r)	p
ΔPRO-C3	−0.08	0.34	−0.03	0.69	0.19	0.01
ΔPRO-C6	−0.05	0.55	−0.12	0.14	0.36	< 0.001
ΔC3M	0.30	< 0.001	−0.02	0.84	0.21	0.007
ΔC6M	0.36	< 0.001	−0.02	0.81	0.10	0.19

CRP = C-reactive protein; IFN-score = interferon-score.

Predictive significance of biomarkers for mRSS and FVC% over time.

In a multivariable linear mixed effects model adjusted for baseline mRSS, mycophenolate use, and time, baseline PRO-C3 concentration predicted longitudinal mRSS during follow-up visits in patients with diffuse disease (point estimate = −0.02, p = 0.03) (Table 5), whereas baseline PRO-C6, C3M, and C6M did not. The changes in PRO-C3, PRO-C6, C3M, and C6M concentrations at six months did not predict longitudinal mRSS in patients with diffuse disease (data not shown). In patients with ILD, neither the baseline concentrations of collagen biomarkers (Supplementary Table S7) nor the changes in concentrations predicted longitudinal FVC% at follow-up.

Table 5.

Predictive significance of baseline concentrations of collagen biomarkers on the skin disease course (mRSS)

Biomarker	Point estimate	95% CI	p
PRO-C3	−0.02	−0.03 to −0.00	0.03
Baseline mRSS	0.79	0.69 – 0.88	<0.001
Timea	−1.72	−2.32 to −1.12	<0.001
Time-varying MMF use	−1.53	−3.38 – 0.32	0.11
PRO-C6	−0.00	−0.07 – 0.07	0.98
Baseline mRSS	0.77	0.67 – 0.87	<0.001
Timea	−1.72	−2.32 to −1.12	<0.001
Time-varying MMF use	−1.74	−3.61 – 0.12	0.07
C3M	0.10	−0.16 – 0.36	0.46
Baseline mRSS	0.77	0.67 – 0.87	<0.001
Timea	−1.73	−2.32 to −1.12	<0.001
Time-varying MMF use	−1.72	−3.58 – 0.14	0.07
C6M	0.10	−0.07 – 0.28	0.23
Baseline mRSS	0.76	0.66 – 0.86	<0.001
Timea	−1.73	−2.32 to −1.12	<0.001
Time-varying MMF use	−1.40	−5.76 – 2.96	0.53

MMF = mycophenolate mofetil; mRSS = modified Rodnan skin thickness score.

^aTime is a continuous variable measured in months.

Discussion

In the present study, we showed that type III and type VI collagen formation and degradation biomarkers are associated with skin and lung disease severity in a nationwide early SSc observational cohort. In addition, we showed that these collagen neoepitopes correlate with immune-related cytokines, supporting a biological link between immune dysregulation and fibrosis in SSc. Discovering new biomarkers to expand the armamentarium for tracking disease activity or predicting disease course is crucial, given the ongoing need for objective measurement tools ^28,29.

We found that increased concentrations of PRO-C3 and PRO-C6 were associated with the diffuse subtype and anti-RNAP3 positivity. These results align with previous studies showing increased concentrations of collagen formation biomarkers in patients with diffuse SSc compared to limited SSc ^16,30. Additionally, formation biomarkers have been previously shown to be elevated in early diffuse SSc patients compared to late diffuse patients ^15,16; however, we could not confirm these findings in our study as CONQUER is an early SSc cohort. Our finding that anti-RNAP3-positive patients had increased collagen formation biomarkers, which was also previously reported ¹⁷, may be attributable to their tendency for aggressive skin disease onset, with an early peak and higher cutaneous fibrosis burden in the early stages of the disease ^31,32. Of note, ACA was associated with lower PRO-C3 levels; this antibody is associated with lower peak mRSS levels compared to anti-RNAP3 antibodies (33), and PRO-C3 correlates with mRSS in our study. This might also explain the somewhat unexpected finding that anti-Scl-70 positive patients had lower PRO-C3 levels. Anti-Scl-70 positive patients had lower mean PRO-C3 levels than those with anti-RNAP3 (82.6 vs. 98.9), which can be due to the fact that the latter group tends to have the highest peak mRSS . In contrast to formation biomarkers, levels of collagen degradation biomarkers were not elevated in patients with diffuse disease; however, C6M, but not C3M, was increased in patients with ILD at the baseline visit. Kubo et al., who evaluated types I, III, IV, V, and VI collagen biomarkers in patients with SSc, also found that C6M was the only biomarker elevated in patients with ILD compared to those without ILD ¹⁷. Interestingly, in our study, patients on mycophenolate at baseline had higher levels of PRO-C6 compared to those not taking mycophenolate. However, we believe this finding was confounded by indication because mycophenolate is more often prescribed to patients with diffuse cutaneous disease and ILD.

Consistent with the above findings, we demonstrated that at baseline, PRO-C3 and PRO-C6 positively correlated with mRSS in patients with diffuse disease, while C6M negatively correlated with FVC% in patients with ILD. To account for the skin and lung disease in the same model, we also analyzed the whole cohort with individual biomarkers as the outcomes and mRSS and FVC% as the independent variables in linear regression models. In these models, we found that PRO-C3 and PRO-C6 positively correlated with mRSS, while C3M and C6M negatively correlated with FVC%. One of the three studies that previously evaluated the relationship between mRSS and collagen biomarkers analyzed formation biomarkers only and found that mRSS correlated with PRO-C3 and PRO-C6 ¹⁶. The other two studies evaluated both formation and degradation biomarkers and found that mRSS correlated with PRO-C3 and PRO-C6 but not with C3M or C6M ^17,30. Overall, our results suggest that serum collagen formation biomarkers correlate with skin disease severity, while degradation biomarkers correlate with lung disease severity. It is important to underscore that although these correlations were statistically significant, the magnitude of correlations was all in a weak range, indicating that these serum protein markers cannot replace the clinical measures of skin and ILD severity (i.e., mRSS and FVC%).

For the first time in an observational SSc cohort, we demonstrated that collagen turnover biomarkers correlated with CRP, IL-6, and IFN-score. As an acute-phase reactant, CRP has been previously linked to more severe skin and lung disease ^33,34 and used as an entry criterion for clinical trials ^35,36. Notably, serum CRP showed a stronger correlation with concurrent serum collagen degradation than formation markers. Consistent with our findings, in a cross-sectional study investigating the correlation of PRO-C3, PRO-C6, and C3M with markers of systemic inflammation in 215 prospectively recruited patients with advanced chronic liver disease, CRP showed a stronger correlation with collagen degradation marker, C3M, than with the two investigated collagen formation markers, PRO-C3 and PRO-C6 (Rs=0.42 vs. Rs=0.34 vs. Rs=0.29) ³⁷. IL-6 is a pro-inflammatory and pro-fibrotic cytokine implicated in SSc pathogenesis that is blocked by tocilizumab ³⁸. The IFN-score has been correlated with disease severity ³⁹ and used to predict response to mycophenolate ²⁷. Our findings further reinforce the connection between fibrosis and inflammation in SSc.

We capitalized on the two sets of blood samples collected from patients six months apart and analyzed longitudinal trends. Except for PRO-C6, which decreased, none of the biomarker levels changed significantly over six months. Furthermore, the changes in biomarker levels did not correlate with the changes in mRSS or FVC%. These findings speak against the utilization of the investigated collagen neoepitopes as dynamic biomarkers that track changes in disease severity. However, changes in C3M and C6M positively correlated with the change in CRP in our study. We speculate that the breakdown of collagen might induce inflammation, as a post-hoc analysis of the nintedanib trial (SENSCIS) showed an initial increase in CRP followed by a decrease, paralleled by the change in C3M ⁴⁰. Consistently, the post-hoc analysis of the tocilizumab trial reported that changes in CRP from baseline to week 48 visit in the placebo arm showed the highest correlation with the changes in C3M (ρ=0.6) among all investigated serum proteins ⁴¹. Interestingly, the change in the serum IFN-score also showed a positive correlation with the change in collagen formation biomarkers (PRO-C3 and PRO-C6), which was not observed with the change in serum CRP or IL-6 levels. This finding provides additional evidence for the profibrotic effect of type I IFN in SSc ⁴². Of note, although the aforementioned correlations were statistically significant, the magnitude of the correlations was weak, indicating that changes in the CRP, IL-6, and IFN score are also driven by other biological factors beyond collagen formation and degradation markers.

In this analysis of patients enrolled in the CONQUER study, collagen biomarkers had no predictive significance for skin or lung disease, except for PRO-C3. PRO-C3 showed predictive significance for skin disease progression; however, the point estimate was in the opposite direction compared to the cross-sectional correlation between PRO-C3 and mRSS (i.e., PRO-C3 and mRSS were positively correlated cross-sectionally, but the point estimate for prediction was negative). We suspect that this may be due to PRO-C3 peaking early in the disease course, serving as a surrogate marker for patients with rapidly progressive skin disease who reach their peak mRSS early and subsequently experience improvement in skin involvement during later visits ^31,32,43,44, as seen in the RNAP3 positive group. A previous study evaluated the predictive value of PRO-C3, PRO-C4, C3M, and C4M for disease progression in two cohorts of patients with longstanding SSc, including individuals with both disease subtypes and with or without ILD. Disease progression was defined as a decline in FVC% > 10% among SSc-ILD patients or worsening mRSS > 5 points and 25% at a 1 year follow-up visit. With respect to overall disease progression, findings were inconsistent between the derivation and validation cohorts: C3M levels were lower among progressors in the derivation cohort, whereas they were higher among progressors in the validation cohort. Similar inconsistencies were observed for C4M. However, the study also included subgroup analyses focused on ILD progression in SSc-ILD patients and on skin severity progression among patients with diffuse SSc. Consistent with our results, none of the collagen neoepitopes were differentially expressed between progressors and non-progressors with regard to SSc-ILD. In contrast, their findings regarding skin fibrosis progression differ from ours. They reported that higher PRO-C3 was associated with an increased risk of skin fibrosis progression, whereas in our study, higher PRO-C3 levels predicted lower subsequent mRSS levels. We believe this discrepancy is likely attributable to differences in disease duration between the cohorts (8.5 years in the cited study vs. 2.7 years in ours). ¹⁸. Generally, we believe the phase-dependent directionality of PRO-C3 predictive significance for mRSS course limits its utility in clinical practice among patients with SSc. Prediction of change in mRSS or FVC% using dichotomized threshold values was not pursued in the present study because our analysis for examining the predictive significance of baseline collagen neoepitopes for the course of mRSS and FVC% as continuous outcome did not yield consistent, statistically significant results. Several studies in IPF found that PRO-C3 ^14,45, PRO-C6 ¹⁴, C3M, and C6M ⁴⁶ were associated with progressive disease. In our study, we couldn’t confirm the predictive significance of collagen formation or degradation epitopes for fibrosis progression in SSc. The fact that SSc is a multicompartment fibrosing disease with collagen turnover products produced at least by two major organs (skin and lung) complicates the development of general fibrosis-based prognostic/predictive biomarkers as the skin tends to improve ²⁸, and the lung tends to worsen ⁴⁷ over time.

This study had some limitations which merit discussion. Because this is an observational study, patients were not treated with uniform treatment regimens. Hence, we cannot investigate the predictive significance of the collagen markers for response to treatment. We tried to minimize the impact of heterogenous treatment regimens in an observational cohort by including a time-varying mycophenolate use variable in our multivariable model. In addition, a six-month interval might have been too short to capture significant changes in biomarker levels, and longer intervals with repeated sample collection could be sought in future studies. Moreover, the present study was limited to four collagen neoepitopes/degradation products and did not include other serum fibrosis biomarkers.

In summary, we showed that type III and type VI collagen neoepitopes are associated with disease severity in SSc. PRO-C3 and PRO-C6 tended to associate more with skin disease, whereas C3M and C6M tended to associate more with lung disease. While PRO-C3 shows a phase-dependent predictive significance for skin severity progression, the other investigated collagen biomarkers do not predict mRSS or FVC% course. Moreover, the serum fibrosis markers correlated with immune-related cytokines, providing additional evidence for a biological link between immune dysregulation and fibrosis in SSc.

Supplementary Material

1

Supplementary Table S1. Biomarker concentrations according to clinical features

Supplementary Table S2. Biomarker concentrations according to antibody status

Supplementary Table S3. Linear regression model evaluating the relationship between collagen biomarker concentrations and FVC% and mRSS at baseline

Supplementary Table S4. Pairwise Spearman’s correlations between the collagen biomarkers

Supplementary Table S5. Correlation of change in collagen biomarker concentrations with changes in mRSS and FVC% at 6 months

Supplementary Table S6. Linear regression model evaluating the relationship between change in collagen biomarker concentrations and change in FVC% and mRSS

Supplementary Table S7. Predictive significance of baseline concentrations of collagen biomarkers on the interstitial lung disease course (FVC%)

Supplementary Figure S1. PRO-C3 (A), PRO-C6 (B), C3M (C), and C6M (D) concentrations according to the presence of interstitial lung disease

Supplementary Figure S2. PRO-C3 (A), PRO-C6 (B), C3M (C), and C6M (D) concentrations at baseline and 6 months

Data availability statement:

The data underlying this article will be shared on reasonable request to the corresponding author.