Pharmacological treatments for SSc-ILD: Systematic review and critical appraisal of the evidence

Many therapies have been investigated for systemic sclerosis-associated interstitial lung disease (SSc-ILD), including immunosuppressive therapies, antifibrotic agents, immunomodulators and monoclonal antibodies. There is a high unmet medical need to better understand the current evidence for treatment efficacy and safety. This systematic review aims to present the existing literature on different drug treatments investigated for SSc-ILD and to critically assess the level of evidence for these drugs. A systematic review was performed in accordance with Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) guidelines. A structured literature search was performed for clinical trials and observational studies on the treatment of SSc-ILD with pharmaceutical interventions from 1 January 1990 to 15 December 2020. The quality of each reference was assessed using the Grades of Recommendation, Assessment, Development, and Evaluation (GRADE) criteria. A total of 77 references were reviewed and 13 different treatments were identified. We found high-quality evidence for the use of cyclophosphamide, nintedanib, mycophenolate and tocilizumab. Therefore, we would posit that the clinical community has four valid options for treatment of SSc-ILD. Further research is mandatory to provide more evidence for the optimal treatment strategy in SSc-ILD, including the optimal time to initiate treatment, selection of patients for treatment and upfront combination therapy

within the first 5 years following the onset of SSc symptoms [8].Of the patients who do develop ILD, 25-30% develop progressive disease, with worsening fibrosis and poorer outcomes [8].
Although most patients with SSc-associated ILD (SSc-ILD) experience a slow decline in lung function, some patients progress rapidly, with progression defined as declined lung function and signals of increased fibrosis on high-resolution computed tomography (HRCT) scans [9].In the EUSTAR registry, 23-27% of patients with SSc-ILD experienced ILD progression during any 12-month period, and 67% experienced progression at any time over the mean 5-year follow-up period [10].Due to this variable clinical course, treatment decisions need to be made on a case-by-case basis.However, although treatment recommendations are available [11], there is no established treatment algorithm for SSc-ILD.
Many therapies have been investigated for SSc-ILD, including immunosuppressive therapies, antifibrotic agents, immunomodulators, monoclonal antibodies, haemopoietic stem cell transplant (HSCT) and lung transplant.Given that there were no approved drug treatments available [11] until the approval of the tyrosine kinase inhibitor nintedanib in 2019 [12], patients with SSc-ILD had a high unmet medical need.The lack of approved therapies for SSc-ILD and known immune system involvement mainly lead to the use of immunosuppressive therapies such as cyclophosphamide, methotrexate and mycophenolate mofetil (MMF), as they are used for SSc.Methotrexate has been shown to improve skin score in early diffuse cutaneous SSc (dcSSc), but beneficial effects in other systems, including the lungs, have not yet been established [11].The two most common drugs used in treatment of SSc are cyclophosphamide and MMF, supported by a positive randomised controlled trial (RCT) [13] and a negative RCT [14], respectively, but with similar efficacy results.The better safety and tolerability profile of MMF in the Scleroderma Lung Study (SLS) II and the toxicity of cyclophosphamide in the long term has made MMF the more commonly used drug in clinical practice for continued treatment.The latest available guidelines for SSc that included ILD recommended cyclophosphamide and HSCT as these were the only treatments with completed RCTs at that time.Additional drugs, such as nintedanib and tocilizumab, have been approved for slowing the rate of decline in pulmonary function in SSc-ILD (US label) [12,15] since the last guidelines were published.
Previous reviews in this area have focussed on a single treatment [16,17], or on ILD in other types of CTD or rheumatic diseases [18].There is a need to better understand the current evidence for treatment efficacy and safety specifically in SSc-ILD.This systematic review aims to present the existing literature on different drug treatments investigated for SSc-ILD and to critically assess the level of evidence for these drugs.

Literature search
This systematic literature review was performed in accordance with the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) statement and the PRISMA 2009 checklist.A structured literature search was performed for studies on the treatment of SSc-ILD with pharmaceutical interventions.The PubMed database was searched from 1 January 1990 to 23 March 2020 (the date of the literature search); the search was updated on 15 December 2020.The search terms were '((systemic sclerosis OR SSc OR scleroderma) AND (interstitial lung disease OR ILD)) OR (SSc-ILD)'.Results were filtered to select clinical trials and observational studies only.

Study selection
The results were screened to select full peer-reviewed manuscripts of studies of SSc-ILD in humans, describing outcomes following pharmaceutical-based interventions.Studies of heterogeneous ILD populations (for example, CTD-associated ILD [CTD-ILD]) were selected if they included patients with SSc-ILD.Study designs, case reports, review articles, letters to the editor, conference abstracts, editorials and guidelines were excluded from the search results for consistency.Other exclusion criteria included preclinical studies, non-pharmaceutical interventions (for example, lung transplants) and studies not reporting outcomes.The reference lists of selected publications were manually searched for additional relevant publications that met the inclusion criteria above but did not appear in the original PubMed search.

Data extraction
Study information and outcomes data were independently extracted from each publication by one of three reviewers.Data collection included study design, sample size, treatment details, patient baseline characteristics, changes in pulmonary function, HRCT outcomes, patient function or quality of life measures, survival and safety outcomes.

Critical assessment of evidence
Each publication was assessed using the Grades of Recommendation, Assessment, Development, and Evaluation (GRADE) criteria [19] by one of three reviewers, independently checked and then agreed by all authors.GRADE assessments were conducted to assign the quality of evidence from each reference as high, moderate, low, or very low according to factors that include the study methodology, consistency and precision of the results, and directness of the evidence.

Search results
In total, 124 publications were identified in the PubMed searches (Fig. 1).Screening of titles and abstracts resulted in 68 being excluded, leaving 56 publications.A manual search of the references of these selected articles found an additional 21 publications that met the inclusion criteria but were not identified in the PubMed search.A total of 77 publications were reviewed in full.There were 45 separate studies in SSc-ILD only (multiple publications for the same study not counted).Thirteen studies included SSc-ILD data as part of the larger patient population, with SSc-ILD results reported separately in eight of these.Thirteen different pharmaceutical interventions were identified and grouped together by type.See Table 1 for summary data and GRADE outcomes from these studies.Phase III data was available for nintedanib (N = 576) [20] and cyclophosphamide (N = 158) [13], and Phase II data is available for MMF (N = 142) [21]; tocilizumab had randomised Phase II data in SSc (N = 87) [22] and Phase III data in early dcSSc with raised inflammatory markers (N = 210) [23].

Nintedanib
We identified three publications evaluating nintedanib published in 2019 and 2020 [24][25][26].Data from these studies are shown in Supplementary Table 1.Nintedanib was investigated in two Phase III placebocontrolled studies: one included only patients with SSc-ILD (N = 576) [24,26], whereas the other included patients with SSc-ILD as part of a wider progressive fibrosing ILD population (n = 39/663) and results for SSc-ILD were not reported separately [25].In the nintedanib study in SSc-ILD, the median time from onset of first non-Raynaud symptom was 3.4 years, with a relatively equal split between dcSSc and limited cutaneous SSc [24], and almost half of the patients were receiving MMF at baseline.The certainty of the evidence for nintedanib was high [24,26] or moderate [25], based on the Phase III trials, suggesting high confidence in the finding that nintedanib significantly reduces the annual decline in forced vital capacity (FVC) in SSc-ILD or progressive fibrosing ILD including SSc-ILD.

Lung function
In patients with SSc-ILD, nintedanib (n = 288) significantly reduced the annual rate of decline of FVC compared with placebo (n = 288) (primary endpoint), with a between-group difference of 41.0 mL/year (95% confidence interval [CI] 2.9-79.0;P = 0.04; FVC% predicted difference 1.2%; 95% CI 0.1-2.2) [24].Loss of FVC in mL/year was lowest in those patients that had been receiving MMF at baseline in both treatment arms, and nintedanib reduced the annual rate of decline of FVC both in patients receiving and those not receiving MMF [24].In the study of nintedanib in progressive fibrosing ILD [25], patients with SSc-ILD were not analysed in isolation, although the results across the autoimmune subset have been presented [27].

Tocilizumab
We identified three references for tocilizumab, an interleukin-6 (IL-6) monoclonal antibody, which was investigated in two randomised, placebo-controlled studies with the first publication in 2016.Data from these studies are shown in Supplementary Table 2.The Phase II study investigated tocilizumab in patients with SSc with a mean baseline FVC % predicted of 80-82% (N = 87) [22,28], and the Phase III study investigated 210 patients with early dcSSc with raised inflammatory markers, including 136 with ILD by HRCT [23].The quality of evidence was mixed, with high-quality evidence based on the Phase III trial [23], and low-or moderate-quality evidence based on the Phase II trial [22,28].While the modified Rodnan skin score (mRSS) primary endpoint was not met in either trial, there is evidence that tocilizumab may have important, clinically relevant results on lung function.

Lung function
In the Phase III tocilizumab study in SSc, in which mRSS was the primary endpoint, there was a smaller decline in the secondary endpoint of FVC% predicted in the SSc-ILD subpopulation with tocilizumab (n = 104) compared with placebo (n = 106) at 48 weeks (between-group difference 3.4%; 95% CI 0.4-5.6;P = 0.002) [23]; these results were reflected in the overall population.Fewer participants treated with tocilizumab had a decline of ≥10% in FVC% predicted compared with placebo (9% vs 25%, respectively, of those with ILD and 5% vs 17%, respectively, of all patients).In the Phase II tocilizumab study, there was a smaller decline in FVC (an exploratory endpoint) for tocilizumab (n = 43) compared with placebo (n = 44) at 24 weeks (least square mean difference 136 mL; 95% CI 9-264; P = 0.0368); however, at 48 weeks the difference was not significant (least square mean difference 120 mL; 95% CI -23 to 262; P = 0.099) [22].In the open-label extension of the Phase II study, 51 patients received tocilizumab for up to 96 weeks; no patients experienced a >10% decline in FVC% predicted at Week 96 [28].Because neither tocilizumab study met its primary mRSS endpoint, all P-values were considered nominal.

HRCT pattern
In patients with ILD treated with tocilizumab in the Phase III study, there was a difference compared with placebo in change from baseline at 48 weeks in median quantitative lung fibrosis (QLF) for whole lung (− 0.6; 95% CI 1.2 to − 0.3; P = 0.0008) and most affected lobe (− 1.6; 95% CI -3.3 to − 0.4; P = 0.002), and quantitative ILD (QILD) for whole lung (− 3.3; 95% CI -4.3 to − 0.7; P = 0.008) scores on HRCT (P-values were nominal due to failure to meet primary endpoint) [23].These results were reflected in the overall study population.Oral CYC (1-2 mg/kg/day) and low-dose prednisone (<10 mg/day) The majority of pts.treated with oral CYC and low-dose prednisone showed significant improvements in FVC at 12 months, which were maintained in those completing 24 months of treatment; there was no significant improvement in DLco but it remained stable in the majority of cases Very low N = 14 Akesson, et al.Arthritis

Records
Rheum 1994 [95] Initial daily dose of oral CYC (2.0-2.5 mg/kg) and prednisolone (30  IV CYC (0.75 mg-1 g/m 2 body surface area; maximum dose 1 g) monthly for 6 months and bimonthly for the remaining 6 months and low-dose prednisone (1 mg/kg/day for 4 weeks, tapered to 5 mg q2w up to 10 mg) OR high-dose oral prednisone (10 mg/day) A combination of CYC with low-dose steroids is effective in ILD, especially in active disease; results did not show differences between the high-and low-dose groups but differences in disease severity between the groups at baseline may have affected the findings  3. The efficacy of MMF was investigated in the Phase II SLS II study, an RCT evaluating MMF for 2 years against oral cyclophosphamide for 1 year followed by 1 year of placebo in SSc-ILD (N = 142) [14].Seven analyses of the SLS II trial [14,[29][30][31][32][33][34] presented high-or moderate-quality evidence, suggesting that while there was no difference in clinical efficacy compared with cyclophosphamide after 2 years, MMF was better tolerated and significantly reduced the extent of SSc-ILD on HRCT but did not result in improvement in lung fibrosis [14,31].However, mycophenolate was mostly investigated in smaller cohorts or observational studies [35][36][37][38][39][40][41][42][43][44][45][46] with very low-or low-quality evidence.

Lung function
For most studies, including SLS II [14], the baseline mean FVC% predicted was >60%, with Nihtyanova et al. and Iudici et al. reporting baseline FVC% predicted >80% [39,44].The effect of MMF on lung function was non-significant in most studies, including in SLS II where FVC% predicted improved from baseline by +2.17% in the MMF arm (n = 69) and +2.86% in the cyclophosphamide arm (n = 73), with no significant difference found (N = 126; P = 0.24) [14].However, significant improvement in mean FVC% predicted from baseline was demonstrated in a cohort of patients with CTD-ILD (N = 125) treated with MMF for 156 weeks (+7.3%;P = 0.004) [37].Gerbino et al. also demonstrated that MMF treatment significantly improved FVC% predicted by a mean of 4% per year (P = 0.002) in a small cohort of patients with SSc-ILD (N = 13) [40].
Liossis et al. showed that MMF significantly improved mean diffusing capacity for carbon monoxide (DLco) % predicted compared with pretreatment after 4-6 months in patients with dcSSc-associated alveolitis of recent onset (N = 6; 75.4% vs 64.2%, respectively, P = 0.033) [45].Vanthuyne et al. also observed significant improvements in DLco% predicted when patients with early SSc-ILD or extensive skin disease were treated with MMF in combination with methylprednisolone and glucocorticoids for 12 months [46].

HRCT pattern
In SLS II, treatment with MMF or cyclophosphamide was not significantly correlated with changes in QLF scores.However, QILD score in the whole lung was improved in both the MMF (− 2.51; 95% CI -4.9 to − 0.15; n = 51) and cyclophosphamide arms (− 2.78; 95% CI -5.17 to − 0.40; n = 47) [14].Goldin et al. demonstrated that both MMF and cyclophosphamide treatments were associated with a significant improvement in QILD score in the whole lung (pooled group: − 2.51%; 95% CI -4.00 to − 1.03; P = 0.001), with no significant differences observed between the two treatments [31], and FVC% predicted was significantly correlated with change in the extent of fibrosis on HRCT assessed by computer-aided diagnosis scores [31].
In studies that recruited patients with early SSc-ILD (N = 5-16), there were non-significant improvements in ground-glass opacities by chest HRCT following MMF treatment [45,46].In a small case series (N = 12), 54.5% of patients with SSc-ILD who experienced an inadequate response to cyclophosphamide, MMF treatment led to stabilisation of the Warrick score (a semi-quantitative scoring system composed of a severity and extent score calculated across four pulmonary zones) [43].However, a study of patients with progressive SSc-ILD found a similar deterioration in Warrick score at 2 years of 2.7 and 2.0 for MMF and cyclophosphamide, respectively [35].
For cyclophosphamide, the certainty of evidence was generally very low or low (23/33 references).Studies with high-quality evidence did not find a significant effect of cyclophosphamide on FVC% predicted at 12 months; however, there was significant improvement in fibrosis on HRCT scans [47][48][49][50][51]53,54].Moderate-quality evidence found that cyclophosphamide significantly improved Health Assessment Questionnaire-Disability Index (HAQ-DI) scores at 12 months, with a higher proportion of patients achieving a minimally clinically important difference [48,52].This is in alignment with the assessment of HAQ-DI evidence in a previous systematic review focussing on cyclophosphamide [16].

Lung function
Out of the 33 studies, 28 reported lung function data, although results were variable across studies.In SLS I, there was a smaller decline in FVC% predicted at 12 months in the cyclophosphamide arm (n = 79) compared with placebo (n = 79) (− 1.0 ± 0.92 vs − 2.6 ± 0.9) [13].The mean absolute difference in FVC% predicted at 12 months for cyclophosphamide compared with placebo was 2.53% (95% CI 0.28-4.79;P < 0.03); this difference was also seen at 24 months [13].A comparison of the cyclophosphamide arms from SLS I and SLS II using an inferential joint model approach found significant improvement from baseline in lung function at 3-12 months, but not after that time point [55].
The quality of evidence was low or very low for azathioprine [71][72][73], pomalidomide [80] and four out of six rituximab studies [74][75][76].There were two moderately rated studies on rituximab: there was moderate evidence that rituximab stabilised or improved lung function after 6 months of treatment [77,79]; however, other studies with similar findings were very low quality [74,75].In a recent systematic review of rituximab in CTD-ILD, including SSc-ILD, the authors concluded that while it was a promising therapeutic tool, more data were needed from multicentre prospective trials [17].This matches our assessment of the quality of the evidence for rituximab in SSc-ILD.

Lung function
Rituximab treatment was associated with significant improvements in FVC after both 1 year (n = 8) [74] and 2 years (n = 8-33) [75,76] of treatment in patients with SSc-ILD.In an observational study using the EUSTAR cohort, rituximab treatment was associated with preventing further FVC decline from baseline in patients with SSc-ILD (N = 9) compared with matched controls (P = 0.02) [77].In an open-label randomised study, rituximab (n = 30), but not cyclophosphamide (n = 30), was associated with a significant increase in FVC% predicted [79].In patients treated with azathioprine following cyclophosphamide therapy, a 2-year experimental study (N = 27) and a 1-year observational study (N = 9) showed beneficial effects on lung function [71,73]; however, there was no improvement in a 1-year experimental study with 45 patients [72].

HRCT pattern
Only three studies reported HRCT results following immunosuppressive treatment [72,74,75].There were no significant improvements in serial HRCT scans in patients with SSc-ILD who were treated with oral prednisolone, cyclophosphamide and azathioprine at 1 year (n = 22) [72].In a small study of 18 patients with SSc-ILD, there were no significant improvements in HRCT scores at 24 weeks following rituximab treatment [74].In another rituximab study, a 5-10% decrease in ground-glass lesions was reported in 5 out of 8 patients [75].

Other tyrosine kinase inhibitors
We identified five publications investigating two additional tyrosine kinase inhibitors during the literature search: imatinib mesylate and dasatinib.The earliest publication for each treatment was in 2008 and 2017, respectively.Data from these studies are shown in Supplementary Table 6.The evidence for imatinib and dasatinib was rated low or very low quality [81][82][83][84][85]. Imatinib mesylate was investigated in SSc-ILD in four studies, including one Phase I/IIa study (N = 20) [82], two Phase II studies (both N = 30) [81,85] and a small case series (N = 5) [84], with doses ranging from 200 to 600 mg once daily, either as monotherapy or in combination with cyclophosphamide.There was one Phase I/II singlearm safety/biomarker study investigating dasatinib in SSc-ILD (N = 31) [83].Most patients in these studies had a baseline FVC% predicted >70%, except the Khanna et al. imatinib study, where the baseline mean value was <70% [82].

Lung function
Two imatinib studies reported on changes in lung function.In Spiera et al. [85], a mean increase of 2.1% in FVC% predicted from baseline at 12 months was reported for the 16/30 patients with ILD.In Fraticelli et al. [81], 4/26 patients had a good response at 6 months of treatment, defined as an increase in FVC >15%, and/or increase in DLco >15% and partial pressure of oxygen >90%, and HRCT pattern unchanged or improved from baseline; 19/26 patients had stabilised ILD.In the singlearm dasatinib study, there were significant differences in FVC and DLco at 6 months between patients classed as improvers (n = 3) and nonimprovers (n = 9).Improvers had stable FVC and DLco at 6 months, whereas non-improvers had a decline in both these measures (P = 0.1289 and P = 0.0195, respectively) [83].

HRCT pattern
Only two studies reported HRCT results.After 6 months of imatinib treatment, there was a significant reduction in the number of lung segments with ground-glass opacities compared with baseline in stabilised patients (n = 15; P = 0.0002), but not in the number of lung segments with honeycombing [81].In patients treated with dasatinib, 23/31 patients had matched baseline and 6-month follow-up HRCT scans [83].Of these, improved or stable HRCT scores were observed in the most severe lobe in 9/23 patients and in the whole lung in 10/23 patients.In the most severe lobe, 39% of patients showed no progression in ILD by quantitative HRCT following dasatinib treatment.

Other treatments
We identified four studies in the literature search investigating other treatments for SSc-ILD: bosentan, riociguat, pirfenidone and D-penicillamine [86][87][88][89].Data from these studies are shown in Supplementary Table 7. Bosentan, a nonselective endothelin receptor antagonist, was investigated in a randomised, placebo-controlled Phase II/III study in SSc-ILD (N = 163), where the primary endpoint was change in 6-min walking distance, and secondary endpoints included changes in pulmonary function tests, published in 2010 [86].The soluble guanylate cyclase inhibitor riociguat was investigated in a Phase IIb placebocontrolled trial of patients with dcSSc at high risk of skin fibrosis progression, including a subset of patients with ILD (n = 25/121) [89].The quality of evidence for bosentan and riociguat was assessed to be high and moderate, respectively, although neither study met its primary endpoint [86,89].An open-label Phase II trial investigated the safety and tolerability of the antifibrotic pirfenidone, using two different dosetitration schedules (2 weeks vs 4 weeks) in patients with SSc-ILD (N = 63) published in 2016 [88].The quality of evidence for pirfenidone was low, and the study found no significant differences between treatment groups [88].D-penicillamine was investigated in a retrospective observational study (N = 122) comparing cyclophosphamide, D-penicillamine, prednisone and other immunosuppressive treatments used to treat SSc-ILD as well as a no-treatment group published in 1994 [87].The evidence from Steen et al.'s study was assessed as very low quality, suggesting low confidence in the findings that only cyclophosphamide had a significant effect on lung function over 2 years [87].

Discussion
ILD secondary to SSc has emerged as the leading cause of morbidity and mortality in SSc [5][6][7].Accordingly, there have been a large number of interventional studies evaluating candidate therapies.These studies have included retrospective case reviews and open-label observational studies, but surprisingly few RCTs.This diversity in the reporting of potential therapies lends itself to standardised assessment of the quality of evidence.
In the present critical assessment of published evidence from the last 30 years, we found high-quality evidence for the use of cyclophosphamide, nintedanib, mycophenolate and tocilizumab.We would posit that the clinical community has four options for treatment of SSc-ILD.Nintedanib has been shown to reduce decline in pulmonary function, both alone and in combination with MMF.Based on data from tocilizumab in SSc [22,23,28], including patients with ILD [23], there is evidence for an important, clinically relevant effect on lung function in patients with early diffuse cutaneous disease and raised inflammatory markers.Both nintedanib and tocilizumab have been approved for slowing the rate of decline in pulmonary function in patients with SSc-ILD (the latter in the US only).Treatment with cyclophosphamide and MMF has been shown to preserve pulmonary function.However, treatment with cyclophosphamide is limited in duration due to its toxicity.MMF, in our opinion, should be considered particularly in combination with targeted agents.Some of the other treatments discussed may also have potential.
In SSc and SSc-ILD we are mainly interested in reducing or stopping the ongoing processes that lead to irreversible damage and loss of function, whether inflammatory, fibrotic or both, thereby slowing down clinical deterioration.This was achieved as the primary endpoint in nintedanib clinical trials, and as secondary or exploratory endpoints in the tocilizumab trials.Long-term follow-up of lung function and markers of inflammation and/or fibrosis are essential parameters for selecting the most reliable approach in patients with ILD.
As outlined in our review, there are many different endpoints that can be assessed.Different patient subgroups may also require different treatments.Future approaches may stratify patients as more data become available.

Future and other treatments
Several clinical trials are investigating possible future treatments for SSc-ILD.The SLS III trial is a Phase II study investigating MMF in combination with pirfenidone (NCT03221257).Another Phase II placebocontrolled trial of MMF combination therapy (NCT02370693) is investigating MMF with or without the proteasome inhibitor bortezomib.Other treatments that have been studied but lack data include a Phase III trial of the phosphodiesterase 5 inhibitor tadalafil (NCT01553981) that completed in 2015 (although no results have been published) and a dose-comparison trial of abituzumab (an immunoglobulin G2 monoclonal antibody targeted at the integrin CD51) in patients receiving a steady dose of MMF that was terminated due to difficulty in recruitment (NCT02745145).
Autologous HSCT is another potential treatment option, although as a non-drug-based treatment it was excluded from the current literature review.One single-centre study and two larger RCTs have shown improvement of skin involvement and stabilisation of lung function in patients with SSc, including SSc-ILD [90][91][92].Event-free and overall survival were improved in patients treated with HSCT compared with cyclophosphamide in both trials, with survival benefits maintained at 5-10 years [91,92].However, HSCT was associated with increased treatment-related mortality in the first few years after treatment.In view of the high risk of treatment-related side effects, the European League Against Rheumatism guidelines recommend careful selection of patients with rapidly progressive SSc at risk of organ failure for HSCT treatment, and state that the experience of the medical team is of key importance [11].
In the current treatment approaches discussed in this review, there has been no stratification on markers for disease.Instead, there has effectively been a blanket approach with immunosuppressive therapy.Disease marker-led treatment stratification could be useful for better targeting patients, with which drug or treatment combination determined for each patient based on markers for inflammation or fibrosis.More data from long-term follow-up of markers would also give insight into disease progression and treatment efficacy over time.These questions may inform future clinical research approaches.

Conclusions
The literature to date leaves a number of critical questions unanswered.Is there an hierarchy of effectiveness that should influence selection of therapy?Are there effective combinations of therapy that provide optimum results?If combination therapy is employed, is there a preferred order of drug initiation?In the nintedanib study [93], nintedanib reduced the loss of pulmonary function both in those who were and were not using MMF at baseline, with no heterogeneity in treatment effect between the subgroups, but there are few other data addressing this question.What clinical, laboratory and HRCT findings might lead to identification of subsets most likely to benefit from one therapy versus another?
Further research is required to provide more evidence for the optimal treatment strategy in SSc-ILD, including the optimal time to initiate treatment, selection of patients for treatment and upfront combination therapy.
gifts or other services from DS Medica.del Galdo reports grants or contracts from Mitsubishi-Tanabe, Capella Biosciences, Chemomab and Kymab; consulting fees from Mitsubishi-Tanabe, Capella Biosciences, Chemomab, Boehringer Ingelheim, Actelion and AstraZeneca; and meeting or travel support from AbbVie and Janssen.James Seibold reports consulting fees from Boehringer Ingelheim and Prometheus Bioscience; payment or honoraria from Boehringer Ingelheim; and participation on a data safety monitoring board for Boehringer Ingelheim.
SLS I: CYC oral ≤2 mg/kg/day (n = 73) vs placebo (n = 72) for 1 year In SLS II, MMF and CYC treatment resulted in improvements in mRSS in pts.with dcSSc over 24 months High SLS II: MMF ≤3 g/day for 2 years (n = 69) vs CYC oral ≤2 mg/kg/day for 1 year, followed by placebo BID for 1 year (n = 73) N = 287 (SLS I n = 145; SLS II n = 142) Volkmann, et al.Ann Rheum Dis 2019 [30] SLS I: 1-year oral CYC (n = 79) vs placebo (n = 79) In addition to traditional mortality risk factors in SSc (skin score and age), SLS I and II found that a decline in FVC and DLco over 2 years is a better predictor of mortality than baseline FVC and DLco Moderate SLS II: 1-year oral CYC (n = 73), 1-year placebo vs 2 years of MMF (n = 69) N = 300 Volkmann et al.ACR Open Rheumatol 2020 [33] MMF 1500 mg BID for 24 months (n = 69) In SLS II, MMF and CYC treatment improved overall HRQoL in pts.with SSc-ILD Moderate Oral CYC 2.0 mg/kg/day for 12 months followed by placebo for 12 months (n = 73) N = 142 Cyclophosphamide Silver, et al.J Rheumatol1993 [94] M.C.Vonk et al.effect of cyclophosphamide on lung function = 75)[64].
M.C.Vonk et al.

Table 1
Evidence profiles for treatments in SSc-ILD.
Significantly lower frequency of clinically significant pulmonary fibrosis in the MMF-treated cohort.Significantly better 5-year survival from disease onset and from start of treatment.No significant difference in median change in FVC%pred MMF >1 g/day (majority received 2 g/day in divided doses) MMF was associated with a significant improvement in VC.DLco did not change significantly during MMF treatment MS Week 1: 360 mg BID, then 720 mg BID for 12 months After 12 months of MS therapy, median values for FVC, FEV 1 and DLco did not change significantly and fulfilled the definition of stable disease by the American Thoracic Society Mycophenolate (MMF n = 3; MS n = 7) 1500 mg/day, 22-72 months FVC, TLC and DLco did not change significantly in either mycophenolate or CYC group after 1 or 2 years.A deterioration of lung HRCT findings at 2 years was noticed after mycophenolate but not after CYC.The study does not support replacement of CYC with mycophenolate for pts.CYC 750 mg/m 2 monthly IV injections/oral equivalent for 6-12 months Oral AZA 2 mg/kg/ day as maintenance at 6 months if there was a good response.If there was an inadequate response, MMF 500 mg BID for 1 month and then, if tolerated, 1500-2000 mg/day PFTs and imaging scores were stabilised by MMF in SSc-ILD pts.who were inadequate responders to CYC Very low N = 12 (continued on next page) M.C.Vonk et al.

Table 1 (
[31]inued ) CYC weekly pulses of 500 mg up to 20 pulses.Oral corticosteroids, proton pump inhibitors, calcium channel blockers, antiplatelet agents and vitamin D. Mesna 100 mg and oral trimethoprim-sulfamethoxazole 160-180 mg The incidence of improvement or stabilisation of lung function parameters was significantly higher in AZA-treated than in MMF-treated pts.MMF 1500 mg BID for 24 months (n = 69)In SLS II, frequent cough correlated with both the presence and severity of GERD and ILD at baseline and improved in parallel with improvements in both ILD and GERD over 2 years of MMF or CYC therapy Moderate Oral CYC 2.0 mg/kg/day for 12 months followed by placebo for 12 months (n = 73) N = 142 Goldin, et al.Ann Am Thorac Soc 2018[31] [32]<3.0 g BID, for 2 years (n = 50); n = 48 completed 24 months of treatmentIn SLS II, CYC treatment for 1 year followed by placebo, or MMF treatment for 2 years was associated with a significant improvement in the extent of HRCT SSc-ILD assessed by computeraided diagnosis scores High CYC 1.8-2.3mgQD,for 1 year followed by 1-year placebo (n = 47); n = 32 completed 24 months of treatment N = 97 Namas, et al.Arthritis CareRes 2018[32]

Table 1 (
Six pulses of IV CYC (15 mg/kg) and IV MP (10 mg/kg) given at 3-4 weekly intervals Treatment with IV CYC and MP may stabilise disease activity during the course of treatment and for 6 months afterwards; however, in the long term, deterioration still occurred in most pts., sometimes at a significant rate continued )Oral CYC (1 mg/kg/day, increased every month up to a maximum of 2 mg/kg/day) vs placebo In SLS I, FIBmax score, mRSS and BDI were independent correlates of the change in FVC%pred over time; severity of reticular infiltrates on baseline HRCT and baseline mRSS may be predictive of response to CYC therapy High N = 158 (12 months: n = 136; 18 months: n = 112) 24/day; range10-40;tapered to between 5 mg every other day and 10 mg/day over 10 weeks) VC and Cst improved after treatment, with reduced skin involvement; the improvements mainly occurred in pts.with biochemical evidence of an acute-phase reaction IV CYC (1000 mg/m 2 of body surface monthly for 6 months) and oral prednisone (25 mg/day for first month and 5 mg/day of maintenance dosage for remaining 5 months) CYC pulse stabilised alveolitis in the majority of cases; the association with prednisone may help to control disease evolution in the lung.CYC plus prednisone did not change FVC but, at least in part, improved DLco Very low N = 23 (continued on next page) M.C.Vonk et al.Oral CYC (1 mg/kg/day, increased every month up to a maximum of 2 mg/kg/day) for 1 year vs placeboIn SLS I, CYC improved lung function, skin scores, dyspnoea and health status/disability for at least 12 months; however, except for a sustained impact on dyspnoea, these effects waned and were no longer apparent at24months Moderate N = 145/158 completed ≥6 months of treatment Yiannopoulos, et al.Rheumatol Int 2007 [67] Monthly IV pulses of CYC (750-1000 mg/m 2 ) and 1 g of MP Combination of IV pulses of CYC and MP is well tolerated and effective, mainly in stabilising respiratory function; this goal is more realistic when treatment is given before At least 6 months of CYC treatment (record review; typically, daily oral CYC 1 mg/kg of body weight, titrated to 2 mg/kg depending on tolerability) While the majority of pts.treated with CYC for active ILD experienced long-term lung function stability and survived, more than a third of pts.experienced lung function decline, death, or required a lung transplant Very low N = 38 (continued on next page) M.C.Vonk et al.