Detection and Classification of Systemic Sclerosis-Related Interstitial Lung Disease: A Review

Abstract

Purpose of review:

Systemic sclerosis (SSc) is a heterogeneous disease with a variable disease course. Interstitial lung disease (ILD) is one of the leading causes of morbidity and mortality in patients with SSc. This review highlights recent advances in the classification, diagnosis, and early detection of SSc-associated ILD (SSc-ILD).

Recent findings:

Risk stratification through measurement of disease extent on high-resolution computed tomography (HRCT) of the chest, longitudinal declines in pulmonary function tests (PFTs), and mortality prediction models have formed the basis for classifying clinically significant ILD. HRCT may be preferred over PFTs for screening, as PFTs lack sensitivity and have a high false negative rate. Novel imaging modalities and biomarkers hold promise as adjunct methods for assessing the presence and severity of SSc-ILD, and predicting risk for progressive disease. Further validation is required prior to their use in clinical settings.

Summary:

Classification of SSc-ILD has shifted to a personalized approach that considers an individual patient’s probability of progressive disease through identification of risk factors, measurement of disease extent on HRCT, longitudinal declines in PFTs, and mortality prediction models. There remains an unmet need to develop screening guidelines for SSc-ILD.

Introduction

Systemic sclerosis (SSc) is a chronic autoimmune disease characterized by immune dysregulation, vasculopathy, and inflammation resulting in excessive fibrosis of the skin and internal organs [1,2]. Interstitial lung disease (ILD) is one of the most common manifestations of SSc, affecting approximately 40–60% of patients within this population. SSc-associated ILD (SSc-ILD) is the leading cause of hospitalization, morbidity, and mortality in patients with SSc, accounting for approximately 35% of SSc-related deaths [3–6]. The 2013 American College of Rheumatology/European Legion Against Rheumatism (ACR/EULAR) classification criteria defines SSc-ILD as, “pulmonary fibrosis seen on high resolution computed tomography (HRCT) or chest radiograph, pronounced in the basilar portions of the lungs, or occurrence of ‘Velcro’ crackles on auscultation, not due to another cause such as congestive heart failure” [7]. Major risk factors for the development of SSc-ILD identified from observational studies include diffuse cutaneous SSc, African American race, older age at disease onset, shorter disease duration, presence of anti-Scl-70 antibodies, and the absence of anti-centromere antibodies [8–10].

As of this writing, there are no FDA-approved treatments for SSc-ILD. Current management approaches follow either a strategy of close monitoring of symptoms and pulmonary function tests (PFTs), or a regimen of immunosuppression with close follow-up of symptoms and PFTs. First-line therapy for treatment of SSc-ILD is mycophenolate mofetil (MMF), which was shown to have similar efficacy to and less toxicity than cyclophosphamide in Scleroderma Lung Study (SLS) II [11]. In the recently published SENSCIS trial, patients with SSc-ILD who were taking nintedanib had a lower annual rate of forced vital capacity (FVC) decline than those taking placebo (difference, 41 mL per year, 95% confidence interval [CI], 2.9 to 79.0, p-value = 0.04) [12]. Notably, 48% of patients in the trial were taking MMF concomitantly with nintedanib or placebo. At this time, it remains unknown where nintedanib will fit into the treatment algorithm of SSc-ILD. It may become the first FDA-approved medication for the treatment of SSc-ILD. Research into newer therapies, such as pirfenidone [13], is ongoing, yet morbidity and mortality from SSc-ILD remain high. Therefore, it is critical that we investigate methods to screen, classify, and risk stratify patients with SSc-ILD so we can detect disease early and identify those at high risk of progression. Research into early detection and treatment of SSc-ILD may eventually enable prevention of progressive disease. In this review, we aim to summarize fundamental pre-existing literature and provide insight into recent advances in the classification, diagnosis, and early detection of SSc-ILD.

Classification of Systemic Sclerosis-associated ILD

There are several ways to classify SSc-ILD: by histopathology, radiographic pattern, radiographic extent of disease, and likelihood of progression (Figure 1). Histopathologically, SSc-ILD is characterized by early pulmonary infiltration of inflammatory cells into the lung parenchyma with resultant fibrosis and can be classified into specific patterns of disease including non-specific interstitial pneumonia (NSIP), usual interstitial pneumonia (UIP), organizing pneumonia, and lymphoid interstitial pneumonia [14, 15]. The most common radiographic pattern on HRCT scan of the chest is NSIP, present in approximately 65% of cases and characterized by ground glass opacities in a primarily peripheral distribution with sub-pleural and basilar predominance. This contrasts with the UIP pattern, present in approximately 25% of cases, characterized by disrupted lung architecture, dense areas of patchy fibrosis, and honeycombing in a primarily sub-pleural distribution [14]. Lung biopsy is usually not required to confirm the diagnosis of SSc-ILD unless other diagnoses such as malignancy or infection are suspected, as patterns can be determined through HRCT alone with a high degree of reliability [15].

Fig 1.

Heading: Classification of Systemic Sclerosis-associated ILD

Legend: Our approach to the classification of SSc-ILD based on review of available data. SSc-ILD can be classified by histopathology, radiographic pattern of disease, radiographic extent of disease, and likelihood of progression. Factors that help determine likelihood of progression include FVC trajectory, disease subtype, autoantibody status, and demographics. NSIP: nonspecific interstitial pneumonia; UIP: usual interstitial pneumonia; OP: organizing pneumonia; LIP: lymphoid interstitial pneumonia; FVC: forced vital capacity.

While these radiographic and histopathologic classifications are useful, and there is a trend for shorter survival in patients with a UIP pattern compared to those with an NSIP pattern [14], the prognosis of patients with SSc-ILD is quite variable and is more closely linked to both disease extent at baseline and progressive functional decline [14,16,17]. Given this variability, it is important not only to characterize SSc-ILD by histopathologic or radiographic pattern, but also to quantify disease extent and classify patients according to their individual risk of progression. In an era of increasingly personalized medicine, further advances in composite clinical screening algorithms and better characterization of predictive markers for progressive disease promise improvements in our overall management of SSc-ILD.

In 2008, Goh et al. developed a classification system to stage the extent of ILD in SSc. Using a combination of HRCT and PFTs, they classified patients into limited and extensive disease categories [18]. Extensive disease was defined as (1) >20% lung involvement on HRCT, or (2) 10–30% ILD involvement on HRCT and FVC <70% predicted. Limited disease was defined as (1) ≤10% ILD involvement on HRCT, or (2) 10–30% lung involvement on HRCT and an FVC ≥70% predicted [18]. This staging system has been validated as a predictor of mortality, and SSc-ILD patients with extensive disease have an approximately 3-fold increased risk of clinical decline (defined as need for supplemental oxygen or lung transplantation) and death compared to those with limited disease [18–20].

Patients with SSc-ILD can also be stratified by their likelihood of progression. There is marked variability in the clinical course of disease: some patients have a slowly progressive decline in, or even stability of, FVC, while others experience a rapidly progressive course, leading to lung transplantation or death, despite treatment [21]. Group-based trajectory modeling based on retrospective review of longitudinal FVC values from 254 SSc patients has identified 7 distinct FVC trajectories: very low baseline FVC with slow decline (5.5% of patients), very low baseline FVC with improvement (13.8% of patients), low baseline FVC with fast decline (9.5% of patients), low baseline FVC that remained stable (19.7% of patients), low-normal baseline FVC with improvement (31.1% of patients), normal baseline FVC with improvement (16.1% of patients) and normal baseline FVC that remained stable (4.3% of patients) [22]. Similarly, the findings of Goh et al. [18] have since been extrapolated to show that dynamic changes in imaging studies and PFTs hold prognostic value and can be utilized to help predict risk of progressive disease [21, 23, 24]. One-year declines in FVC and diffusion capacity for carbon monoxide (DLCO), for example, have been shown to predict survival in patients with extensive disease, with a decrease in FVC by more than 10% and/or a decrease in DLCO by more than 15% during one year associating with poorer prognosis [21,24]. These parameters have since been incorporated into the Outcome Measures in Rheumatology (OMERACT) definition of progression of connective tissue disease-associated ILD ([25]. More recently, Volkmann et al [26••] developed a mortality prediction model through post hoc Cox regression analyses of patients from SLS I and II. They showed that significant declines in FVC (≥ 10%) and DLCO (≥15%) over 24 months were the most robust predictors of long-term survival, even when adjusting for treatment arm and baseline disease severity [26]. Additional risk factors for worse prognosis of SSc-ILD included elevated baseline plasma C-reactive protein (CRP) levels, gastroesophageal reflex disease, pulmonary arterial hypertension, older age, African American race, and male sex [26–28].

Akin to approaches reported in the idiopathic pulmonary fibrosis (IPF) literature [29–31] several prediction models have been developed to risk stratify patients with SSc-ILD using clinical variables available at the time of a patient’s initial office visit [32,33] The SpO₂ and Arthritis (SPAR) model is one such tool, designed to predict ILD progression, that was developed using two independent prospective cohorts of patients who met 2013 ACR/EULAR Classification Criteria for SSc and had mild ILD (<20% lung involvement) assessed by HRCT at baseline [32]. ILD progression was defined as a relative decrease in FVC by ≥15% or a decline in FVC by ≥10% combined with a decrease in DLCO by ≥15% at 1-year follow-up. In their multivariate analyses, declines in SpO₂ after 6-minute walk test and arthritis (defined as one or more tender and swollen joints as judged by a treating physician) were identified as independent predictors of ILD progression in both cohorts, with an optimal SpO₂ cut-off value of 94% by ROC analysis [32•]. The Smoking history, Age, and DLCO (SADL) model is another validated risk prediction model for all-cause mortality in SSc-ILD developed using two independent prospective cohorts of patients meeting 2013 ACR/EULAR Criteria for ILD. The SADL model uses a patient’s smoking history, age, and DLCO to classify him or her into a low, moderate, or high mortality risk group at 3 years from ILD diagnosis [33•]. Though further validation is required, such models are easily used and can potentially guide clinicians in their management decisions.

While classification of SSc-ILD by radiographic pattern continues to hold important treatment and survival implications, we believe a personalized medicine approach that accounts for an individual’s radiographic pattern and extent, FVC trajectory, autoantibody status, disease subtype, and demographic variables, is critical to determining the likelihood of progression and informing treatment decisions.

Screening and Diagnosis of Systemic Sclerosis

Diagnosing SSc-ILD at an early stage can be challenging, as it may develop insidiously and patients may have asymptomatic, “subclinical” ILD. The most common early clinical manifestations of SSc-ILD include exertional dyspnea and nonproductive cough, both of which are nonspecific [34]. Given that patients may be asymptomatic or have nonspecific symptoms, and that ILD is both highly prevalent and the leading cause of death in SSc, universal screening of SSc patients for ILD is critical. According to reports summarizing the proceedings from the ACR and Association of Physicians of Great Britain and Ireland Connective Tissue Disease (CTD)-ILD Summit, the development of a screening system with the dual objective of identifying early-stage disease and identifying those at greatest risk for progression and functional decline is a major unmet need within the field [35]. The two most widely applied methodologies for ILD screening are PFTs and HRCT.

PFTs are a valuable non-invasive method to assess severity of ILD and monitor disease course. Although PFTs are widely utilized to screen SSc patients for ILD, they lack sensitivity and have a high false negative rate for the detection of ILD [36]. In a single center prospective cohort study, Suliman et al [36] showed that among 64 patients with ILD on HRCT, approximately 62.5% had a normal FVC, defined as ≥ 80% predicted. The sensitivity of an FVC <80% predicted for detection of SSc-ILD was only 38%; this increased only to 72% when the following parameters were combined: FVC <80% predicted or ΔFVC ≥10% or total lung capacity <80% predicted or DLCO <70% predicted and forced expiratory volume in 1 second/FVC > 0.7. Showalter et al [37] performed a similar study of 265 patients meeting 2013 ACR/EULAR Classification Criteria for SSc to identify the sensitivity, specificity, and negative predictive value of PFTs for the presence of SSc-ILD on HRCT, and to determine optimal FVC and DLCO thresholds for the presence of SSc-ILD on HRCT. An FVC <80% predicted (sensitivity 69%, specificity 73%) and DLCO <62% predicted (sensitivity 60%, specificity 70%) were identified as the optimal thresholds to define ILD, however, all FVC and DLCO threshold combinations had a negative predictive value of <0.7. Collectively, these data suggest a high risk of missing SSc-ILD if relying solely on PFTs. Moreover, ILD is a radiographic (and/or histopathologic) diagnosis, thus imaging studies are required for diagnosis.

HRCT of the chest is the gold standard for detection of ILD and enables assessment of the radiographic pattern and extent of disease [35, 38]. Launay et al [39] showed the utility of baseline HRCT, as 34 of 40 (85%) SSc patients with normal HRCT at baseline still had normal HRCT at a mean follow-up of 5 years. Thus, these baseline HRCTs had both diagnostic and prognostic value. Routine use of screening HRCT in SSc may identify asymptomatic individuals with subclinical ILD at high risk for developing clinically significant ILD, just as the presence of subclinical ILD on baseline cardiac computed tomography (CT) scans predicts development of clinically significant ILD in community-dwelling adults [40]. High attenuation areas (HAA), defined as the percentage of imaged lung with CT attenuation values between −600 and −250 Hounsfield units, are a novel CT-based quantitative biomarker of subclinical ILD that have strong construct validity as a biomarker of subclinical lung inflammation and extracellular matrix remodeling, processes that precede ILD, in community-dwelling adults [40,41]. In community-dwelling adults enrolled in the Multi-Ethnic Study of Atherosclerosis, greater HAA is associated with reduced FVC, reduced exercise capacity, elevated serum levels of matrix metalloproteinase-7 (MMP-7) and interleukin-6 (IL-6), the development of interstitial lung abnormalities (ILA, a qualitative visually-identified subclinical ILD phenotype on CT) on CT at 10-year follow-up, and an increased risk of developing clinically evident ILD and ILD-specific mortality at 12-year follow-up [41]. Thus, research is needed into whether quantification of HAA on screening HRCTs in patients with SSc can identify individuals at high risk of developing clinically evident ILD and ILD-specific mortality. Ultimately, identification of subclinical ILD on screening HRCTs in patients with SSc may permit a “window of opportunity” for intervention.

However, there remains substantial practice variation in rheumatologists’ use of HRCT to screen for ILD in their SSc patients. In a survey of 676 ACR member rheumatologists in New York, New Jersey, Pennsylvania, and Connecticut, and 356 SSc experts worldwide, only 51% of the general rheumatologist respondents and 66% of the SSc expert respondents reported routinely performing screening HRCTs in their SSc patients [42••]. Moreover, there was significant global practice variation among SSc experts in their HRCT ordering practices: screening HRCT was ordered by 100% (7 of 7) of SSc experts in Latin America, 80% (4 of 5) in Asia, 79% (45 of 57) in Europe, 60% (28 of 47) in the United States, 33% (2 of 6) in Canada, and 0% (0 of 5) in Australia [42]. Further, there was little agreement regarding indications for HRCT among rheumatologists who do not routinely order screening HRCTs in their SSc patients. In our practice, we routinely order HRCTs to screen for ILD in all newly diagnosed SSc patients.

Given that PFTs lack sensitivity for detection of SSc-ILD, that HRCT is the gold standard for diagnosis of ILD, and that there is significant variation in both general rheumatologists’ and SSc expert rheumatologists’ use of HRCTs to screen for ILD in SSc, there is an urgent need to develop screening guidelines for the detection of ILD in SSc.

Novel Methods for Screening and Classification

Novel Imaging Techniques

Given the lack of any clear screening guidelines and gaps in our ability to prognosticate patients adequately, research into novel imaging is an intense area of focus. Over the past 15 years, lung ultrasound (LUS) has emerged as an attractive non-invasive, radiation-free technique with high sensitivity and specificity for the diagnosis of ILD [43–45]. Assessment for pleural irregularities and increased number of B-lines, discrete vertical hyperechoic reverberations arising from pleural lines, serves as the basis for LUS assessment for ILD. An increased number of B-lines is associated with thickening of the lung parenchyma and is suggestive of ILD [43]. Previous studies identified a greater number of B-lines in patients with ILD than in those without ILD on HRCT, with a concordance rate of 83% (46,47). A recent meta-analysis of 11 studies analyzing LUS for diagnosis of CTD-ILD yielded a pooled sensitivity and specificity of 85.9% and 83.9%, respectively [44]. Current limitations of LUS for ILD screening in SSc include lack of standardization (e.g., number of lung zones or intercostal spaces to examine, and which probe to use), operator skill dependence, possible confounding due to skin fibrosis, and the total length of time required for the procedure [43].

Several promising exploratory methods of ILD detection are being investigated. Lung ultrasound surface wave elastography (LUSWE) is a new ultrasound application that measures the elasticity of superficial lung tissue. Zhang et al [48] found significant increases in the speed of ultrasound wave propagation through the more fibrotic lung surfaces of patients with SSc-ILD compared to healthy controls, which could be useful for screening SSc patients for ILD, though it remains unclear how this technology fares in detecting changes of ILD other than fibrosis (e.g., ground glass opacifications) [48]. Research into additional novel approaches such as magnetic resonance imaging (MRI) and molecular imaging are ongoing and show the ability to detect SSc-ILD with high accuracy without the use of ionizing radiation [49–51]. In 2018, for example, Gargani et al [50] evaluated the utility of lung MRI in 32 SSc patients who underwent concurrent cardiac MRI with dedicated lung scanning (with T1 and STIR imaging) and chest HRCT. Mean T1/STIR times and HRCT semi-quantitative scoring were calculated for each patient. The authors showed that mean STIR was moderately correlated with HRCT scores (r = 0.52, p <0.01). Similarly, in a proof of concept study, Schniering et al [51] successfully targeted integrin αvβ3 (alpha-v-beta-3) and somatostatin receptor 2 (SSTR2), two proteins upregulated in ILD lungs, to illustrate the role of nuclear imaging to visualize ILD in animal models and patients with known ILD. Their study shows the potential for screening to occur on a molecular level through the precise targeting of proteins associated with fibrotic lung parenchyma.

Circulating Biomarkers

In recent years, much research has gone into investigating the clinical utility of biomarkers for the diagnosis of ILD and assessment of disease severity and progression [52–53]. Though the precise utility of biomarkers remains investigational and no single serologic marker has been fully validated, numerous potential biomarkers identified in the IPF literature have been subsequently investigated in SSc-ILD (e.g., MMP-7, surfactant protein-D [SP-D], Krebs von den Lungen-6 [KL-6], and C-C motif chemokine ligand 18 [CCL18]) [53–56]. In addition, novel potential biomarkers continue to be explored, such as antibodies against chemokine receptors CXCR3 and CXCR4, two G-protein-coupled receptors implicated in the pathogenesis of pulmonary fibrosis via mediation of cell migration [57]. Significantly higher levels of CXCR3 and CXCR4 antibodies were found in patients with SSc-ILD compared to controls, and levels of these antibodies were shown to correlate inversely with FVC and DLCO in patients with SSc. Somewhat paradoxically, however, SSc-ILD patients with more progressive disease tended to have lower CXCR3 and CXCR4 antibody titers compared to those with more stable disease [57].

While many of these biomarkers hold promise, there have been conflicting results regarding their sensitivity, specificity, and predictive power [56, 58]. In a combined cohort study of 427 Norwegian and French patients meeting 2013 ACR/EULAR Classification Criteria for SSc, four promising ILD biomarkers (SP-D, KL-6, CCL18, and soluble OX40 ligand [OX40L]) were analyzed for their ability to diagnose and predict progression of SSc-ILD [58••]. Serum levels of KL-6 were significantly inversely correlated with FVC (r = −0.317, p <0.001) and DLCO (r = −0.335, p <0.001) and positively correlated with the extent of fibrosis on HRCT (r = 0.551, p <0.001). Similarly, KL-6 (OR 2.41, 95% CI 1.43–4.07) and SP-D (OR 3.15, 95% CI 1.81–5.48, p <0.001) were significantly associated with the presence of lung fibrosis in multivariable analysis adjusting for age and SSc disease duration, defined as the time between first non-Raynaud’s symptom and blood sample collection [58]. These findings corroborate previous studies of KL-6 in SSc-ILD, but are inconsistent with previous smaller analyses of SP-D [59–62]. In a longitudinal analysis, CCL18 was an independent predictor of >10% decrement in FVC over a mean follow-up period of 3.2 years (HR 2.90, 95% CI 1.25 to 6.73, p = 0.014) and the development of de novo extensive disease per Goh criteria (HR 3.71, 95% CI 1.02 to 13.52, p = 0.048) [58]. These results support previous smaller studies that showed elevated CCL18 levels were predictive of significant declines in FVC [63–66]. In sum, these findings suggest potential roles for SP-D as a diagnostic biomarker of SSc-ILD, KL-6 as a biomarker of lung fibrosis severity, and CCL18 as a biomarker of progressive SSc-ILD [58]. Despite the potential for their use in the diagnosis, classification, and risk stratification of SSc-ILD, however, further validation and standardization is ultimately required before such biomarkers should be utilized in clinical practice.

Conclusion

ILD is a common manifestation of SSc and the leading cause of mortality in this patient population. While radiographic and histopathologic classifications help to define general disease patterns, stark variations in clinical course limits the utility of assessing SSc-ILD on the basis of pattern alone. Recent work suggests a paradigm shift has occurred, with patients classified primarily according to their probability of severe, progressive disease through identification of risk factors, measurement of disease extent on HRCT, longitudinal declines in FVC, and mortality prediction models. Though no clinical practice guidelines for ILD screening in SSc exist, we recommend screening with HRCT and PFT in all SSc patients. Biomarkers, lung ultrasound, and novel imaging modalities serve as promising adjunctive or alternative means of screening and diagnosis. Further validation is required before they should be used in clinical practice.

Key points:

• Classification of SSc-ILD can occur in various dimensions. Current trends suggest a more personalized assessment of an individual patient’s pattern of disease, extent of disease, risk factors, and longitudinal assessment of lung function in the context of their radiographic and/or histopathologic pattern of disease. Such classification and risk stratification can help to guide treatment.

• There are no clinical practice guidelines for ILD screening in SSc. Significant global practice variation still exists.

• HRCT is currently the gold standard for the detection of ILD. While useful, PFTs lack sensitivity for detection of ILD. All patients with newly diagnosed SSc should receive a baseline HRCT to evaluate for underlying ILD.

• Lung ultrasound is a non-invasive, radiation-free technique with high sensitivity and specificity for the screening and diagnosis of SSc-ILD. Lack of standardization and length of time for a scan, however, are major barriers for adoption.

• SP-D, KL-6, and CCL18 have emerged as promising candidate biomarkers for the diagnosis of SSc-ILD, assessment of disease severity, and predictors for progressive disease, respectively. Numerous additional biomarkers are under investigation.

Abbreviations:

ACR

American College of Rheumatology

CCL18

C-C motif chemokine ligand 18

CI

Confidence Interval

CRP

C-reactive protein

CTD

Connective Tissue Disease

DLCO

Diffusion capacity of the lung for carbon monoxide

EULAR

European League Against Rheumatism

FVC

forced vital capacity

HAA

high attenuation areas

HRCT

high-resolution computed tomography

HU

Houndsfield unit

ILD

interstitial lung disease

IPF

Idiopathic pulmonary fibrosis

LUS

lung ultrasound

KL-6

Krebs von den Lungen-6

LUSWE

Lung ultrasound surface wave elastography

MMF

mycophenolate mofetil

MMP-7

matrix metalloproteinase-7

MRI

magnetic resonance imaging

NSIP

nonspecific interstitial pneumonia

OMERACT

Outcome Measures in Rheumatology

OX40L

soluble OX40 ligand

PFT

pulmonary function test

SLS

Scleroderma Lung Study

SP-D

surfactant protein-D

SSc

systemic sclerosis

SSc-ILD

systemic sclerosis associated interstitial lung disease

UIP

usual interstitial pneumonia