Characterization of Incident Interstitial Lung Disease in Late Systemic Sclerosis

Abstract

Objective

Interstitial lung disease (ILD) is a common and potentially lethal complication of systemic sclerosis (SSc). Screening by high‐resolution computed tomography (HRCT) is recommended in all patients with risk factors, including early disease. Little is known on late presentations of ILD. This study aimed to characterize the incidence, risk factors, and outcomes of late‐onset SSc‐ILD.

Methods

Study participants enrolled in the Canadian Scleroderma Research Group cohort from 2004 to 2020 without prevalent ILD were included. Incidence and risk factors for ILD (on HRCT) were compared according to disease duration above (late) and below (earlier) seven years from the first non‐Raynaud manifestation. Risk of ILD progression was compared using Kaplan‐Meier and multivariable Cox models.

Results

Overall, 199 (21%) of 969 patients developed incident ILD over a median of 2.4 (interquartile range 1.2–4.3) years. The incidence rate in late SSc (3.7/100 person‐years) was lower than in earlier SSc (relative risk 0.68, 95% confidence interval [CI] 0.51–0.92). Risk factors for incident ILD included male sex, diffuse subtype, myositis, antitopoisomerase I autoantibodies, and higher C‐reactive protein levels. Patients with late‐onset ILD were also less frequently White and more frequently had arthritis and anti‐RNA‐polymerase III autoantibodies. Lung disease severity was similar between late‐ and earlier‐onset SSc‐ILD (forced vital capacity 88% and 87%, diffusion capacity of the lungs for carbon monoxide 64% and 62%, respectively). Progression rates were also similar between late‐ and earlier‐onset SSc‐ILD (log rank P = 0.8, hazard ratio 1.11, 95% CI 0.58–2.10).

Conclusion

ILD can present in late SSc. Risk factors and progression rates overlapped with earlier‐onset SSc‐ILD. Surveillance for ILD should continue in longstanding SSc. Frequency and modality of monitoring remain to be defined.

INTRODUCTION

Systemic sclerosis (SSc) is a complex autoimmune connective tissue disease characterized by vasculopathy, immune dysregulation, and fibrosis in the skin and internal organs. Interstitial lung disease (ILD) is a frequent complication affecting more than 50% of patients with SSc ¹ , ² and is the leading cause of SSc‐related death. ³ Early detection of SSc‐ILD is essential, and screening with chest high‐resolution computed tomography (HRCT) is recommended in all patients with SSc at baseline, particularly in the presence of risk factors, including early disease. ⁴ ILD often develops within the first three to five years of the first non‐Raynaud disease manifestation ⁵ , ⁶ , ⁷ and may be more rapidly progressive in early disease. ⁸ Based on these observations, most clinical trials in SSc‐ILD have restricted study inclusion to patients within seven years of SSc onset. ⁹ , ¹⁰ , ¹¹

However, ILD may also develop in longstanding SSc. In fact, among patients without prevalent ILD from the Canadian Scleroderma Research Group (CSRG) and the Australian Scleroderma Cohort Study registries, despite a median disease duration of 8.6 years, 153 new patients with ILD were diagnosed over a median follow‐up of 4.1 years (incidence rate of 2.4 per 100 person‐years), ¹² with the highest incidence rates among patients with antitopoisomerase I antibodies. Recent studies have also shown similar ILD progression rates regardless of disease duration. ¹³ , ¹⁴ Moreover, immunosuppression has been reported to be effective in late SSc‐ILD, ¹⁵ suggesting that the detection and treatment of ILD are relevant even in late SSc.

Although late‐onset SSc‐ILD may be clinically relevant, few studies have systematically assessed ILD in longstanding SSc. Therefore, in the present study, we assessed the incidence, characteristics, risk factors, evolution, and treatment outcomes of ILD developing in late SSc, in comparison with ILD developing in earlier SSc.

PATIENTS AND METHODS

Source and study populations

Our source population consisted of patients enrolled in the CSRG between 2004 and 2020 (for a list of CSRG Investigators, please see Appendix A). Briefly, study participants in the CSRG were recruited from 14 sites across Canada and one site in Mexico and had a diagnosis of SSc verified by an experienced rheumatologist, were more than 18 years of age, and were fluent in English, French, or Spanish. More than 98% of the CSRG cohort met the 2013 American College of Rheumatology/EULAR classification criteria for SSc. ¹⁶ All study participants recruited in the registry were assessed yearly by standardized clinical examinations, self‐reported questionnaires, and laboratory investigations. The study population consisted of patients with SSc without a diagnosis of ILD at CSRG cohort entry (ie, no prevalent ILD). Ethics committee approval for this study was obtained at the Centre hospitalier de l'Université de Montréal and at all participating CSRG study sites. All study participants provided informed written consent to participate in the study.

Outcome measures and other covariates

The presence of ILD was determined by chest HRCT and recorded as abnormal by the recruiting physician and/or in the investigations section of the questionnaire. Chest HRCT was not performed systematically in all patients at the time of SSc diagnosis (because this was not standard clinical practice during the study period) and was most often ordered in the presence of risk factors, symptoms, or abnormal chest x‐ray or pulmonary function tests. ILD was considered prevalent if present on HRCT at CSRG cohort entry, and incident if only recorded as present on HRCT over subsequent follow‐up visits. Patients with a normal HRCT at any follow‐up visit were assumed to have no ILD on all preceding visits, and patients without any HRCT data were considered as having missing ILD data. Chest HRCTs were performed locally, and data on lung disease characteristics and extent were extracted from radiology reports. Pulmonary function tests were performed annually at local respiratory physiology laboratories and data on forced vital capacity (FVC) and diffusion capacity of the lungs for carbon monoxide (DLco) were extracted from reports. ILD progression was defined as ≥10% relative decline in percent predicted FVC, or ≥5% to <10% relative decline in percent predicted FVC with ≥15% relative decline in percent predicted DLco, ¹⁷ a definition that has been shown to predict mortality. ¹⁸

Exposure and other variables

Disease duration was determined based on the onset of the first non‐Raynaud disease manifestation as recorded by a study physician and stratified as late SSc (≥7 years) or earlier SSc (<7 years). Exposure to immunosuppression was defined as treatment with mycophenolate mofetil, cyclophosphamide, rituximab, or tocilizumab at the visit of interest or since the last registry visit. Medication history was recorded by a study physician at each visit. Demographic variables including age, sex, race and ethnicity (from a fixed set of categories), and smoking history were collected by patient self‐report. Skin involvement was assessed using the modified Rodnan skin thickness score. Limited cutaneous disease was defined as skin involvement distal to the elbows and knees with or without facial involvement; diffuse cutaneous disease was defined as skin involvement proximal to the elbows and knees and/or of the trunk. Presence of inflammatory arthritis and myositis was recorded by a study physician, whereas symptoms of gastroesophageal reflux were collected by patient self‐report. Autoantibody analyses were performed at baseline at the Mitogen Advanced Diagnostics Laboratory, University of Calgary, and detected by Euroline's SSc profile line immunoassay (Euroimmun GmbH) according to manufacturer's instructions. Autoantibodies were reported as absent (negative, equivocal, and low titers) and present (moderate and high titers). C‐reactive protein levels were measured at local laboratories.

Statistical analysis

Incidence rates (and 95% confidence intervals [CIs]) for ILD diagnosis and progression were calculated based on the Poisson distribution and stratified by disease duration. Generalized estimating equations models with an autoregressive correlation structure were used to estimate the association between demographic, clinical, and serologic variables and the risk of incident ILD over follow‐up, with a one‐annual visit lag period before ILD assessment for time‐varying variables. Missing variables were omitted and patterns of missingness were assessed. To estimate whether potential risk factors for incident ILD differed by timing of ILD onset, models were adjusted and stratified for disease duration, and further tested by an interaction between each risk factor and disease duration. The risk of ILD progression was analyzed using unadjusted Kaplan‐Meier and Cox proportional hazard models adjusted for FVC and DLco, stratified by disease duration. To explore the effect of immunosuppressive drugs on lung disease progression, analyses were further stratified according to immunosuppressive drug exposure, modeled as a time‐dependent current/noncurrent exposure with a one‐visit lag period to minimize information bias due to reverse causality, and interaction terms were used to assess effect modification by disease duration. Study participants were observed from the time of ILD diagnosis until disease progression, or were censored due to death, permanent study drop‐out, or last study visit. Baseline characteristics (at the time of ILD onset) of study participants with and without ILD progression were compared using a two‐sample t‐test, Mann‐Whitney U‐test, and Fisher's exact test. Sensitivity analyses were performed with disease duration defined according to time of first Raynaud or non‐Raynaud disease manifestation. P values were considered significant if < 0.05. Multiple testing correction was not applied because analyses were considered exploratory. Statistical analyses were performed with R version 4.4.1.

RESULTS

Incidence of ILD in late SSc

Of 969 patients without prevalent ILD at baseline, 199 (21%) developed incident ILD over a median duration of 2.4 (interquartile range [IQR] 1.2–4.3) years. Of these, ILD was diagnosed at least seven years after the first non‐Raynaud disease manifestation in 131 patients (66%), corresponding to an incidence rate of 3.7 (95% CI 3.1–4.3) per 100 person‐years in late SSc. This incidence rate was lower than that in earlier SSc, which was at 5.4 (95% CI 4.2–6.9) per 100 person‐years (relative risk 0.68, 95% CI 0.51–0.92, P = 0.01; Supplementary Table 1A, Supplementary Figure).

Characteristics and risk factors for late‐onset ILD

Table 1 presents the characteristics of patients with and without incident ILD in late and earlier SSc. In both late and earlier SSc, patients who developed incident ILD were more frequently male and more often had diffuse cutaneous involvement, myositis, antitopoisomerase I autoantibodies, absence of anticentromere autoantibodies, and higher C‐reactive protein levels, compared to patients who did not develop ILD. In addition, patients with late‐onset ILD were less frequently White and more often had arthritis and anti‐RNA polymerase III autoantibodies. Lung disease severity was similar between late‐ and earlier‐onset SSc‐ILD, with comparable FVCs (88% and 87%, respectively) and dlco (64% and 62%, respectively) values at ILD onset and similar proportions of patients having ground‐glass opacities, fibrotic interstitial changes, and honeycombing on chest HRCT. Table 2 presents the associations between risk factors and incident ILD using two alternative approaches, namely analyses adjusted for disease duration and including an interaction term, and analyses stratified by late and earlier disease. These show consistent associations between risk factors in both late and earlier ILD.

Table 1.

Risk factors and lung disease characteristics of late‐ and earlier‐onset SSc‐ILD^*

	Late SSc		Earlier SSc
	ILD (n = 131 person‐visits)	No ILD (n = 3,428 person‐visits)	ILD (n = 66 person‐visits)	No ILD (n = 1,154 person‐visits)
Demographic characteristics
Age at ILD onset, mean ± SD	59.2 ± 11.8	59.2 ± 11.8	54.8 ± 11.4	54.2 ± 11.9
Female, n (%)	114 (87)	3,163 (92)	50 (76)	994 (86)
White, n (%)	105 (82)	2,964 (90)	55 (86)	960 (88)
Smoking (ever), n (%)	78 (61)	1,969 (59)	41 (63)	629 (57)
SSc characteristics
Diffuse, n (%)	51 (39)	907 (27)	32 (49)	410 (36)
mRSS, median (IQR)	7 (3–12)	4 (2–9)	8 (4–14)	6 (2–14)
Arthritis, n (%)	31 (28)	462 (17)	16 (25)	190 (18)
Myositis, n (%)	12 (11)	123 (5)	8 (13)	59 (6)
Gastroesophageal reflux, n (%)	56 (52)	1,183 (53)	21 (39)	433 (46)
Anti‐centromere, n (%)	38 (30)	1,699 (54)	13 (22)	450 (45)
Anti‐topoisomerase I, n (%)	26 (21)	202 (7)	15 (26)	131 (13)
Anti‐RNA polymerase III, n (%)	24 (19)	354 (11)	13 (22)	215 (21)
Anti‐Th/To, n (%)	2 (2)	18 (0.6)	4 (7)	11 (1)
Anti‐fibrillarin, n (%)	0 (0)	22 (0.7)	2 (3)	4 (0.4)
Anti‐Ro52/TRIM21, n (%)	34 (27)	726 (23)	13 (22)	226 (23)
Anti‐PM/Scl‐75/100, n (%)	4 (4)	158 (5)	3 (5)	36 (4)
C‐reactive protein, median (IQR), mg/dL	4.0 (1.5–10.0)	2.6 (1.0–5.6)	5.4 (3.0–10.0)	2.5 (1.0–5.9)
ILD characteristics
FVC (% predicted), mean ± SD	88.3 ± 17.2	–	87.2 ± 17.9	–
dlco (% predicted), mean ± SD	63.5 ± 18.9	–	61.8 ± 20.0	–
Ground‐glass opacities, n/N (%)	19/38 (50)	–	17/32 (53)	–
Moderate to severe	4/31 (13)	–	5/29 (17)	–
Fibrotic interstitial changes, n/N (%)	31/40 (78)	–	24/34 (71)	–
Moderate to severe	5/32 (16)	–	4/28 (14)	–
Honeycombing, n/N (%)	6/34 (18)	–	6/31 (19)	–
Moderate to severe	0/30 (0)	–	2/30 (7)	–

^*Anti‐PM/Scl, anti–polymyositis/scleroderma; dlco, diffusion capacity of the lungs for carbon monoxide; FVC, forced vital capacity; ILD, interstitial lung disease; IQR, interquartile range; mRSS, modified Rodnan skin score; SSc, systemic sclerosis; TRIM21, tripartite motif–containing protein 21.

Table 2.

Associations between risk factors and incident ILD expressed as ORs and 95% CIs

Risk factor	Unadjusted OR (95% CI)	Adjusted OR (95% CI)	P value for interaction	Stratified ORs
				Late ILD	Earlier ILD
Age, yr	1.002 (0.989–1.012)	1.003 (0.992–1.015)	0.892	1.003 (0.988–1.017)	1.004 (0.984–1.026)
Female	0.51 (0.34–0.76)	0.54 (0.36–0.80)	0.794	0.56 (0.33–0.97)	0.50 (0.27–0.93)
White	0.61 (0.41–0.91)	0.61 (0.41–0.92)	0.285	0.53 (0.33–0.85)	0.85 (0.41–1.79)
Smoking (ever)	1.13 (0.83–1.53)	1.13 (0.84–1.54)	0.584	1.07 (0.74–1.55)	1.28 (0.76–2.17)
Diffuse	1.80 (1.33–2.44)	1.74 (1.29–2.36)	0.906	1.77 (1.22–2.57)	1.70 (1.02–2.83)
mRSS (log)	1.42 (1.23–1.64)	1.40 (1.21–1.61)	0.065	1.56 (1.29–1.88)	1.19 (0.96–1.47)
Arthritis	1.78 (1.26–2.52)	1.78 (1.26–2.52)	0.526	1.93 (1.26–2.97)	1.53 (0.86–2.73)
Myositis	2.60 (1.60–4.21)	2.55 (1.58–4.12)	0.902	2.62 (1.39–4.92)	2.45 (1.12–5.40)
Gastroesophageal reflux	0.95 (0.69–1.30)	0.96 (0.70–1.32)	0.948	0.97 (0.66–1.43)	0.95 (0.56–1.62)
Anti‐centromere	0.35 (0.25–0.50)	0.36 (0.26–0.51)	0.974	0.36 (0.24–0.54)	0.36 (0.19–0.68)
Anti‐topoisomerase I	3.27 (2.20–4.86)	3.15 (2.11–4.69)	0.232	3.77 (2.32–6.11)	2.34 (1.23–4.45)
Anti‐RNA polymerase III	1.58 (1.07–2.34)	1.50 (1.01–2.24)	0.171	1.84 (1.14–2.98)	1.07 (0.56–2.03)
Anti‐Ro52/TRIM21	1.14 (0.80–1.62)	1.15 (0.81–1.62)	0.599	1.22 (0.81–1.83)	0.99 (0.52–1.89)
C‐reactive protein (log)	1.430 (1.264–1.617)	1.427 (1.262–1.613)	0.456	1.377 (1.165–1.628)	1.511 (1.266–1.804)

^*Analyses were first adjusted for late‐ or earlier‐onset SSc‐ILD and included an interaction term with disease duration, and separately stratified by disease duration at time of ILD onset. CI, confidence interval; ILD, interstitial lung disease; mRSS, modified Rodnan skin score; OR, odds ratio; SSc, systemic sclerosis; TRIM21, tripartite motif–containing protein 21.

Lung disease progression in late‐onset SSc‐ILD

Of 106 patients with incident ILD and available follow‐up data, ILD progression was observed in 48 patients (45%) over a median duration of 3.1 (IQR 2.1–4.0) years, for an average incidence rate of 14.1 per 100 person‐visits. The incidence of lung disease progression was not different between late‐ and earlier‐onset ILD (Figure 1, log rank P = 0.8; and Table 3, adjusted hazard ratio [HR] 1.11, 95% CI 0.58–2.10). Progressors tended to be more frequently male (28% vs 8%, P = 0.076) and less frequently White (76% vs 95%, P = 0.066) in late‐onset SSc‐ILD, but otherwise similar regarding other demographic and disease characteristics (Table 4).

Figure 1.

Kaplan‐Meier curves for lung disease progression, stratified by late‐ and earlier‐onset of systemic sclerosis–ILD (log rank P = 0.8). ILD, interstitial lung disease.

Table 3.

Associations between the risk of interstitial lung disease progression and disease duration, expressed as HRs and 95% CIs^*

	Events	Person‐visits	Incidence per 100 person‐visits (95% CI)	Crude HR (95% CI)	Adjusted HR (95% CI) a
Total	48	340	14.1 (10.5–18.5)
Late‐onset	29	198	14.6 (9.9–20.6)	1.06 (0.59–1.89)	1.11 (0.58–2.10)
Treated	2	10	20.0 (3.3–61.7)	1.05 (0.24–4.57)	0.90 (0.19–4.29)
Not treated	27	184	14.7 (9.8–20.9)	1.00	1.00
Earlier‐onset	19	142	13.4 (8.2–20.3)	1.00	1.00
Treated	3	19	15.8 (3.9–40.9)	1.39 (0.39–4.94)	0.40 (0.05–3.27)
Not treated	15	119	12.6 (7.3–20.1)	1.00	1.00

^*Models were also stratified by the administration of immunosuppressive drugs for exploratory analyses. CI, confidence interval; HR, hazard ratio.

^a Adjusted for forced vital capacity and diffusion capacity of the lungs for carbon monoxide. P = 0.572 for interaction.

Table 4.

Characteristics of progressors and nonprogressors in late‐ and earlier‐onset SSc‐ILD^*

	Late‐onset SSc‐ILD			Earlier‐onset SSc‐ILD
	Progressors (n = 29)	Nonprogressors (n = 37)	P value	Progressors (n = 19)	Nonprogressors (n = 21)	P value
Demographic characteristics
Age, mean ± SD	57.0 ± 12.2	61.4 ± 8.6	0.090	56.7 ± 8.8	52.1 ± 13.5	0.219
Female, n (%)	21 (72)	34 (92)	0.076	15 (79)	15 (71)	0.855
White, n (%)	22 (76)	35 (95)	0.066	16 (84)	18 (90)	0.951
Smoking (ever), n (%)	21 (72)	23 (62)	0.539	12 (63)	13 (62)	>0.99
SSc characteristics
Diffuse, n (%)	13 (45)	15 (41)	0.921	10 (53)	12 (57)	>0.99
mRSS, median (IQR)	11 (4–18)	8 (4–11)	0.182	6 (4–15)	9 (4–15)	0.516
Arthritis, n (%)	6 (21)	9 (26)	0.860	2 (11)	38 (32)	0.680
Myositis, n (%)	4 (14)	2 (6)	0.546	2 (11)	20 (17)	0.447
Gastroesophageal reflux, n (%)	10 (35)	14 (40)	0.846	8 (47)	41 (38)	0.480
Anti‐centromere, n (%)	5 (18)	12 (33)	0.269	2 (12)	4 (20)	0.818
Anti‐topoisomerase I, n (%)	9 (32)	7 (19)	0.383	3 (18)	5 (25)	0.888
Anti‐RNA polymerase III, n (%)	4 (14)	9 (25)	0.457	7 (41)	4 (20)	0.297
Anti‐Th/To, n (%)	0 (0)	0 (0)	–	1 (6)	1 (5)	>0.99
Anti‐fibrillarin, n (%)	0 (0)	0 (0)	–	1 (6)	0 (0)	0.934
Anti‐Ro52/TRIM21, n (%)	7 (25)	11 (31)	0.834	5 (29)	3 (15)	0.509
Anti‐PM/Scl‐75/100, n (%)	2 (7)	0 (0)	0.365	2 (12)	1 (5)	0.883
C‐reactive protein, median (IQR), mg/dL	3.5 (1.1–4.8)	4.9 (1.7–8.7)	0.131	6.0 (2.3–8.1)	3.7 (3.1–5.5)	0.334
ILD characteristics
FVC (% predicted), mean ± SD	89.3 ± 15.0	86.7 ± 15.5	0.485	88.7 ± 16.1	91.2 ± 16.5	0.637
DLco (% predicted), mean ± SD	66.8 ± 17.3	62.8 ± 16.8	0.365	62.7 ± 20.1	70.3 ± 17.7	0.250
Ground‐glass opacities, n/N (%)	1/3 (33)	1/6 (17)	>0.99	3/4 (75)	1/7 (14)	0.173
Moderate to severe	0/2 (0)	0/6 (0)	–	0/3 (0)	0/7 (0)	–
Fibrotic interstitial changes, n/N (%)	2/3 (67)	4/7 (57)	>0.99	4/4 (100)	4/7 (57)	0.406
Moderate to severe	0/2 (0)	0/7 (0)	–	0/3 (0)	0/6 (0)	–
Honeycombing, n/N (%)	0/3 (0)	0/5 (0)	–	0/3 (0)	0/8 (0)	–
Moderate to severe	0/3 (0)	0/5 (0)	–	0/3 (0)	0/8 (0)	–

^*Anti‐PM/Scl, anti–polymyositis/scleroderma; dlco, diffusion capacity of the lungs for carbon monoxide; FVC, forced vital capacity; ILD, interstitial lung disease; IQR, interquartile range; mRSS, modified Rodnan skin thickness score; SSc, systemic sclerosis; TRIM21, tripartite motif–containing protein 21.

Effect of immunosuppressive drugs on lung disease progression in late‐onset SSc‐ILD

Over the course of follow‐up, 29 person‐visits (8.7%) were exposed to immunosuppressive drugs. Study participants were exposed for a median cumulative duration of 2 (IQR 1–3) annual visits. Among the exposed person‐visits, 21, 11, 3, and 1 patients were exposed to mycophenolate mofetil, cyclophosphamide, tocilizumab, and rituximab, respectively. Exposed person‐visits less frequently had late incident ILD (35% vs 61%, P = 0.01) and had lower mean FVCs (84.9% vs 93.6% predicted, P = 0.004) and numerically lower mean dlco values (63.6% vs 71.3%, P = 0.06). Using time‐dependent multivariable Cox analyses, the adjusted HR was 0.90 (95% CI 0.19–4.29) for ILD progression in study participants exposed to immunosuppression compared to nonexposed study participants in late‐onset SSc‐ILD, whereas in earlier‐onset SSc‐ILD, the adjusted HR was 0.40 (95% CI 0.05–3.27; P = 0.572 for interaction) (Table 3).

Sensitivity analyses using disease duration defined from time of Raynaud or non‐Raynaud onset

Sensitivity analyses using disease duration defined according to time of Raynaud or non‐Raynaud were done and are presented in Supplementary Tables 1–4. Results were mostly consistent with primary analyses.

DISCUSSION

In this retrospective cohort study, we aimed to study the incidence, characteristics, risk factors, evolution, and treatment outcomes of late‐onset SSc‐ILD. We found that ILD can present in late SSc, although its incidence rate was lower than in earlier SSc. This is consistent with previous studies showing that clinically significant ILD develops mostly within the first three to five years from disease onset. ⁵ , ⁶ , ⁷ Risk factors for developing ILD in late SSc were largely similar to those observed in earlier SSc and highly consistent with known risk factors for incident, prevalent and/or severe ILD. These included male sex, ¹⁹ , ²⁰ , ²¹ non‐White race, ¹⁹ , ²² , ²³ , ²⁴ , ²⁵ diffuse cutaneous involvement, ⁶ , ⁷ , ²³ myositis, ¹⁹ , ²³ presence of antitopoisomerase I antibodies, ⁶ , ⁷ , ⁸ , ²² absence of anticentromere antibodies, ⁷ , ²³ and high markers of inflammation. ¹⁹ , ²⁶ In addition, in late SSc, incident ILD was also more frequent in patients with anti‐RNA polymerase III antibodies ⁷ and arthritis.

Late‐ and earlier‐onset SSc‐ILD presented with similar lung disease severity and characteristics, with comparable FVC and dlco values as well as similar proportions of patients having ground‐glass opacities, fibrotic interstitial changes, and honeycombing. However, only a small proportion of patients had data on specific HRCT characteristics, and information on ILD distribution and pattern were unavailable in the CSRG database. In a cross‐sectional study done at one CSRG center, we found that lung disease pattern in 35 patients with late‐onset SSc‐ILD mostly consisted of nonspecific interstitial pneumonia and usual interstitial pneumonia patterns, ²⁷ consistent with patterns classically described in SSc‐ILD. ²⁸

Lung disease progression occurred in nearly half of patients and mostly occurred within the first four years of ILD diagnosis. The incidence of disease progression did not vary according to SSc duration at the time of ILD diagnosis. Although shorter SSc duration was previously reported to be associated with more rapid ILD progression, ¹⁹ the apparent plateau in FVC progression in patients with longer disease duration may have been attributable to survival bias, ¹³ and recent studies do not support early disease as a risk factor for ILD progression. ¹³ , ¹⁴

On the other hand, other predictors of SSc‐ILD progression have been identified, including HRCT extent ¹⁴ , ²⁹ , ³⁰ , ³¹ of ILD > 20%, lower or declining FVC and DLco, ²⁹ , ³² , ³³ presence of antitopoisomerase I autoantibodies, ²² , ³¹ , ³⁴ diffuse cutaneous involvement, ³⁴ , ³⁵ male sex, ³⁶ , ³⁷ , ³⁸ and high C‐reactive protein levels. ²⁶ , ³⁹ In our study, progressors tended to be more frequently male and less frequently White in late‐onset SSc‐ILD, but otherwise similar regarding other demographic and disease characteristics.

A minority (9%) of person‐visits were exposed to immunosuppressive drugs, mostly to mycophenolate mofetil and mostly in earlier‐onset ILD. Patients requiring treatment had lower FVC and DLco values, which are known risk factors for ILD progression and possible indications for treatment initiation. Interestingly, on exploratory analyses, risk estimates for the effect of immunosuppressive drugs on lung disease progression suggested a numerically larger benefit in earlier‐onset SSc‐ILD compared to late‐onset SSc‐ILD. Although this difference was not statistically significant, this study was likely underpowered to detect such an interaction, with small numbers of treated patients in each subgroup. Few studies are available to inform us on the effect of immunosuppressive drugs in late‐onset SSc‐ILD, ¹⁵ as most clinical trials excluded patients with SSc duration above seven years. ⁹ , ¹⁰ , ¹¹ It is possible that late‐onset SSc‐ILD is a condition that is pathophysiologically distinct from earlier‐onset SSc‐ILD and thus more refractory to immunosuppressive drugs, although our preliminary observations on similar ILD risk factors, severity, distribution, and pattern do not support this hypothesis.

There are several limitations to this study. First, patients were not systematically screened by HRCT at baseline or on follow‐up yearly visits, as this was an observational cohort and, as such, procedures were performed as per standard of care in clinical practice settings. Thus, this study may have included study participants with prevalent subclinical ILD at baseline, and results should be interpreted as characterizing the incidence of clinically apparent ILD. Sensitivity analyses excluding patients without proven absence of ILD on HRCT before ILD assessment showed a similar, albeit nonstatistically significant trend for lower incidence of ILD in late SSc (Supplementary Table 1C).

Second, nearly half of the study population was excluded from progression analyses due to missing pulmonary function test results on follow‐up. Excluded patients had a lower DLco and a numerically higher frequency of late‐onset ILD compared to included patients (Supplementary Table 5). This may have introduced some bias leading to underestimation of the risk of ILD progression in late‐onset SSc‐ILD.

Third, there was missingness in time‐varying variables used in the study of associations between risk factors and incident ILD, most notably for C‐reactive protein levels and especially among person‐visits without incident ILD (Supplementary Table 6). If one hypothesizes that C‐reactive protein levels are not tested systematically in routine care in the absence of active disease and that missing values are likely within normal range, then the high frequency of missing levels in person‐visits without ILD would introduce bias leading to the underestimation of the association between C‐reactive protein and risk of incident ILD.

Fourth, we were unable to repeat the analyses using the American Thoracic Society/European Respiratory Society/Japanese Respiratory Society/Asociación Latinoamericana de Tórax definition of progressive pulmonary fibrosis, ⁴⁰ which also considers symptoms and radiologic progression, due to significant missing data on follow‐up HRCTs. Fifth, medication data were nominal for “current” or “past” exposure, with no details regarding specific start and stop dates, dose, intermittent exposure, or total duration of treatment. This could have led to exposure misclassification. Finally, for the analysis of the effect of immunosuppressive drugs on lung disease progression, a marginal structural Cox model incorporating inverse probability of treatment weights would have been preferable to account for confounding by indication and time‐varying confounders; however, the small number of exposed person‐visits within each disease duration stratum precluded such analyses.

In conclusion, in this multicentric retrospective cohort study, we confirmed that ILD can present in late SSc. Risk factors for developing ILD in late SSc were largely similar to those observed in earlier SSc and included male sex, non‐White race, diffuse cutaneous involvement, arthritis, myositis, antitopoisomerase I antibodies, and high C‐reactive protein levels. Late‐onset SSc‐ILD appears to present with similar lung disease severity and characteristics as in earlier SSc‐ILD and may have comparable disease progression rates. Surveillance for incident ILD should continue even in patients with longstanding SSc, especially in the presence of risk factors, as patients may be equally at risk of having progressive disease regardless of disease duration at time of ILD onset. Frequency and modality of monitoring remain to be defined and should be the topic of future research.

AUTHOR CONTRIBUTIONS

All authors contributed to at least one of the following manuscript preparation roles: conceptualization AND/OR methodology, software, investigation, formal analysis, data curation, visualization, and validation AND drafting or reviewing/editing the final draft. As corresponding author, Dr Hoa confirms that all authors have provided the final approval of the version to be published, and takes responsibility for the affirmations regarding article submission (eg, not under consideration by another journal), the integrity of the data presented, and the statements regarding compliance with institutional review board/Declaration of Helsinki requirements.

ROLE OF THE STUDY SPONSOR

Inova Diagnostics, Inc; Boehringer Ingelheim; Janssen; Mallinckrodt; and Fresenius Kabi, Dr Fooke Laboratorien, GmbH; Euroimmun; Mikrogen, GmbH, had no role in the study design or in the collection, analysis, or interpretation of the data, the writing of the manuscript, or the decision to submit the manuscript for publication. Publication of this article was not contingent upon approval by Inova Diagnostics, Inc; Boehringer Ingelheim; Janssen; Mallinckrodt; or Fresenius Kabi. Dr Fooke Laboratorien, GmbH; Euroimmun; Mikrogen, GmbH.

Supporting information

Disclosure form

Appendix S1: Supplementary Information

APPENDIX A. CANADIAN SCLERODERMA RESEARCH GROUP INVESTIGATORS

Investigators of the Canadian Scleroderma Research Group: M. Baron, Montréal, Québec; M. Hudson, Montréal, Québec; G. Gyger, Montréal, Québec; J. Pope, London, Ontario; M. Larché, Hamilton, Ontario; N. Khalidi, Hamilton, Ontario; A. Masetto, Sherbrooke, Québec; E. Sutton, Halifax, Nova Scotia; T.S. Rodriguez‐Reyna, Mexico City, Mexico; N. Maltez, Ottawa, Ontario; C. Thorne, Newmarket, Ontario; P. R. Fortin, Québec, Québec; A. Ikic, Québec, Québec; D. Robinson, Winnipeg, Manitoba; N. Jones, Edmonton, Alberta; S. LeClercq, Calgary, Alberta; J‐P. Mathieu, Montréal, Québec; P. Docherty, Moncton, New Brunswick; D. Smith, Ottawa, Ontario; M. Fritzler, Mitogen Advanced Diagnostics Laboratory, Calgary, Alberta.