Abstract
Objective Systemic sclerosis (SSc) is a multisystem disease with heterogeneity in presentation and prognosis.
An international collaboration to develop new SSc subset criteria is underway. Our objectives were to identify systems of SSc subset classification and synthesize novel concepts to inform development of new criteria.
Methods Medline, Cochrane MEDLINE, the Cumulative Index to Nursing and Allied Health Literature, EMBASE, and Web of Science were searched from their inceptions to December 2019 for studies related to SSc subclassification, limited to humans and without language or sample size restrictions.
Results Of 5686 citations, 102 studies reported original data on SSc subsets. Subset classification systems relied on extent of skin involvement and/or SSc-specific autoantibodies (n = 61), nailfold capillary patterns (n = 29), and molecular, genomic, and cellular patterns (n = 12). While some systems of subset classification confer prognostic value for clinical phenotype, severity, and mortality, only subsetting by gene expression signatures in tissue samples has been associated with response to therapy.
Conclusion Subsetting on extent of skin involvement remains important. Novel disease attributes including SSc-specific autoantibodies, nailfold capillary patterns, and tissue gene expression signatures have been proposed as innovative means of SSc subsetting.
Systemic sclerosis (SSc) is a multisystem autoimmune rheumatic disease characterized by microvascular injury and accumulation of collagen in skin and other organs, such as the musculoskeletal system, lungs, kidneys, and gastrointestinal (GI) tract.1,2,3,4,5,6 SSc is associated with poorer patient outcomes and lower quality of life when compared to other rheumatic diseases.7 The 2013 American College of Rheumatology/European League Against Rheumatism (ACR/EULAR) classification criteria for SSc include skin thickening, fingertip lesions, abnormal nailfold capillaries, and the presence of SSc-related autoantibodies, but do not differentiate subsets of patients with SSc.8 Subclassification of SSc into a number of pathogenetically homogenous subsets with similar clinical manifestations and outcomes would help segregate clearly between prognostically distinct disease subgroups. Despite the complex multiorgan nature of SSc, the subsets are frequently defined as being limited cutaneous (lcSSc) or diffuse cutaneous (dcSSc), based on the location of skin involvement.9 This classification system gives insight into disease progression; however, within lcSSc and dcSSc, the course of disease is highly variable between patients.10,11 With a more modern perspective, our understanding of SSc subsets is changing. A combination of multisystem involvement, antibody profiling, genetic markers, and differences in proteomics may play a role in prognosis and treatment options.12,13,14,15,16 Further defining subsets of patients with SSc may help to prognosticate, especially in early disease.17
An international collaboration to develop new criteria to subset SSc is underway.18 Current perceptions around SSc subset criteria were identified by leading international experts. In a survey of 30 SSc experts from 13 countries, 90% of experts use > 2 subsets for classifying and treating their patients.19 Concepts such as progression rates and likely organ involvement are considered for subsetting patients with SSc informally in clinical practice.
There is a need for criteria to identify subsets of patients with SSc for recruitment into clinical trials of novel therapeutic agents, to inform management, and for prognosis in clinical care. Previous attempts to outline SSc subset classification criteria have relied mainly on clinical manifestations.20 However, in recent years, novel disease attributes including autoantibody profiles, nailfold capillary patterns, and gene expression signatures have been proposed as means of subsetting. The objectives of this study were to identify existing systems of subset classification in SSc and to synthesize novel concepts in subsetting through a systematic review of the literature.
METHODS
Data sources and search strategy. A search of publications related to SSc and subsets was performed using Medline, Cochrane MEDLINE, the Cumulative Index to Nursing and Allied Health Literature, EMBASE, and Web of Science from their inceptions to December 2019 (for search strategy and key terms, see Supplementary Table 1, available with the online version of this article). The research question was, “What are the advantages and disadvantages of existing systems of subset classification in patients with systemic sclerosis?”
Searches were supplemented by hand searching the bibliographies of relevant articles (including citation searching). Studies were limited to humans, without language or sample size restrictions. Non–English-language articles were translated by native-language speakers or machine software. EndNoteX9 software (Clarivate) was used to check for duplications.
Studies were screened and excluded if they (1) reported localized scleroderma or scleroderma-like syndromes; (2) were abstracts, case reports, or review articles; or (3) were studies for which updated manuscripts were available. All articles were divided between 4 research groups (DK/CD, JF/FV, MM/JP/JS/TN, MB/SJ/TN) and independently reviewed by investigators from each group using a standardized data abstraction form. Abstracted data included classification schema, number of SSc subsets, number of subjects, country of origin, stated and perceived advantages and disadvantages of the classification system, and external validation. The systematic review conforms to the Preferred Reporting Items for Systematic Reviews and Meta-Analyses statement. The Strengthening the Reporting of Observational Studies in Epidemiology (STROBE) checklist was used to assess the reporting quality of the included studies.
RESULTS
Search results. Our literature review identified 5686 citations, of which 5584 were excluded because they were not relevant (conditions other than SSc, no classification system proposed), they had insufficient data, the data were not original, and/or they did not involve humans. The remaining 102 studies reported schema to subset patients with SSc (Figure 1).
SSc subset criteria. Subset classification systems have historically relied on clinical manifestations, most commonly extent of skin involvement (n = 20; Table 1),9,10,11,21–37 molecular, genomic, and cellular patterns (n = 12; Table 2),38–49 SSc-specific autoantibodies (n = 46, including 5 studies exploring both clinical and serological subsets10,21,27,29,37; Table 3),10,21,27,29,37,50–90 and abnormal nailfold capillary patterns (n = 10; Table 4).91–100 Twenty-one studies reporting associations between capillary abnormalities and clinical features or serology were included (Table 5).94,99,101–119 Using the STROBE checklist, the majority provided a clear presentation of what was planned, done, and found (Supplementary Table 2, available with the online version of this article).120
SSc subsets based on the extent of skin involvement. The diffuse vs limited SSc criteria of LeRoy, et al9 is the most commonly used system of SSc classification. The differences in development of visceral (renal and myocardial) disease and survival were shown for the subsets.9,11,25,26 The system has a good discriminative value to identify the groups of patients with different dominant features (vascular vs fibrotic), internal organ damage, and outcome. It enables identification of patients with early SSc with poor prognosis who will need close monitoring and facilitates the comparison of more homogenous groups of patients in epidemiological studies and clinical trials. The LeRoy 1988 classification system9 has the advantage of comprising only 2 groups and requires criteria other than cutaneous involvement. To classify as diffuse SSc (dSSc), the prerequisites are the onset of Raynaud phenomenon (RP) within 1 year of the onset of skin involvement, early and significant visceral involvement, and the absence of anticentromere antibodies (ACA). When using these strict LeRoy criteria, dSSc represents only a small portion (8.5%) of the total group with definite SSc.23 Two SSc-specific autoantibodies were included in the original LeRoy criteria: antitopoisomerase I antibodies (ATA) and ACA.
Acknowledging the important role of autoantibodies and capillary abnormalities, LeRoy updated the classification in 2001, proposing 4 subsets: limited SSc (lSSc), lcSSc, dcSSc, and diffuse fasciitis with eosinophilia. The classification includes lSSc as RP only in association with serological and/or capillary abnormalities.32 Considering that SSc is a multistage multiorgan disorder, lSSc is likely an early stage of disease and corresponds to very early SSc in the classification of Avouac, et al.28
Others have proposed 3 subset systems based on the extent of cutaneous involvement within the first year of presentation: type I digital (finger or toe skin involvement), type II intermediate (skin involvement proximal to metacarpophalangeal [MCP] joints, but excluding trunk), and type III diffuse (truncal sclerosis).10,24,29,33 The latter type was characterized by male predominance, shorter RP before skin changes, and worse prognosis.11 The clinical distinctiveness of the types was confirmed by difference in autoantibody profile: ACA was found more frequently in type I, while ATA was more frequent in intermediate SSc (iSSc) and dSSc. In the study, the authors included only SSc patients with disease duration ≤ 2 years after the onset of skin lesions, and none of the patients had received any treatment that could potentially affect skin sclerosis prior to the enrollment. That ruled out the possibility that the iSSc group consisted of patients with SSc that would evolve into dSSc later or who originally had dSSc with skin regression under the treatment. Compared to the 2-subset LeRoy system, this classification better reflects the clinical heterogeneity of disease and identifies the subgroups with milder or more severe clinical prognostic evolution.
The simplicity of this 3-subset classification, which is based on clinical examination of skin only and does not require special equipment or tests, makes it highly reproducible and suitable for clinical care and research studies. Notably, this classification system includes a time determinant reflective of the pace of disease, and thus has a prognostic value. Barnett, et al10 emphasized the importance of assessing the extent of skin involvement within the first year of presentation to place a patient into a specific type. Indeed, type I and II patients had a better prognosis in terms of life expectancy compared to type III. However, only slight difference in survival was found between patients with iSSc and those with lSSc.
Patients with iSSc were found to have variable clinical features and represented a serologically heterogeneous group. It raises the question of iSSc as a distinct variant. Some authors suggested that further subdivision of iSSc might be necessary to identify the subsets with particular patterns of internal organ damage and outcome. Scussel-Lonzetti, et al25 divided iSSc into “above elbow” and “below elbow” groups but found them similar with respect to internal organ involvement, mortality, and autoantibody profile. Although the authors supported the concept of an iSSc subset, differentiation was shown only between the LeRoy subsets (“normal + limited” vs “intermediate + diffuse”) in terms of heart involvement, disease activity (elevated erythrocyte sedimentation rate [ESR], anemia), and pulmonary fibrosis. The most significant difference in survival rates was found between lSSc and dSSc, whereas the difference between other subsets was absent (lSSc vs iSSc, P = 0.2) or very low (iSSc vs dSSc, P = 0.03). ATA positivity was similar between iSSc and dSSc while ACA frequencies gradually decreased from lSSc through iSSc to dSSc (50%, 34%, and 3.4%, respectively). Supporting the LeRoy system, the skin involvement proximal to MCP joints was one of the strong predictors of mortality. In line with those findings, Vayssairat, et al23 showed the advantages of LeRoy subset system and disutility of adding iSSc as a subset. When patients with proximal skin thickening were divided into intermediate and truncal subsets, no difference in severity score was found between them.
The patients with calcinosis, RP, esophageal involvement, sclerodactyly, telangiectasia (CREST) syndrome, suspected secondary RP, and/or visceral SSc without skin involvement were not acknowledged in the aforementioned 2 classification systems.9,10 The recently developed immunoblotting technique to detect SSc-related autoantibodies and nailfold capillary microscopy allows the detection of these probable connective tissue diseases. Expanding the subsets, Maricq, et al22 added undifferentiated connective tissue disorder with SSc features, SSc sine scleroderma, and CREST. This classification allows the inclusion of patients who are in earlier stages of their disease.
Boonstra, et al27 identified 4 clinical subgroups by hierarchical clustering using skin, musculoskeletal, cardiac, pulmonary, and GI manifestations; demographics; and risk assessment using follow-up data. Subgrouping patients allowed the prediction of severity and mortality with 2 subgroups showing higher-than-average 5-year mortality rates: subgroup 1 (male predominance, dcSSc, higher modified Rodnan skin score [mRSS], scleroderma renal crisis (SRC), ATA, less frequent interstitial lung disease [ILD]); and subgroup 2 (female and non-White predominance, more frequent pulmonary arterial hypertension [PAH], gastric antral vascular ectasia [GAVE], ILD, and lower diffusing lung capacity for carbon monoxide [DLCO] and forced vital capacity [FVC]). Low-risk clusters (subgroups 3 and 4) included patients with lcSSc who were predominantly female, had more frequent GI manifestations (dysphagia, diarrhea, constipation) for both subgroups, as well as peripheral vascular involvement (digital ulcers), ACA, and White predominance for subgroup 3, and less frequent ILD, FVC, and DLCO for subgroup 4. Three subgroups (1, 3, and 4) were similar to the clusters (6, 3, and 1, respectively) in another subclassification system developed by Sobanski, et al as a European Scleroderma Trials and Research Group clustering initiative.37 However, 2 main clusters, A and B, in the latter study strongly support the LeRoy 200132 subclassification into dcSSc and lcSSc.
SSc subsets based on molecular gene expression profiling. Another approach to classifying patients with SSc into subsets is molecular phenotyping identified through gene expression profiling in tissue samples. Four subsets characterized by distinct molecular pathway signatures have been described and validated in multiple studies: fibroproliferative, inflammatory, normal-like, and limited.38–45,49,121 The intrinsic molecular subsets are consistent for each patient, as well as across the different skin biopsy sites, regardless of clinically affected or unaffected status.38,122 The subsets are also consistent across the organ systems38,39,42,122; however, highly lung-specific innate immune and cell proliferation processes were shown within the immune-fibrotic axis, suggesting that there are gene pairs that are more likely to interact in one tissue than the other (Table 2).123
SSc subsets according to SSc-related autoantibodies. The classification system according to serum antibodies is based on the findings of mutually exclusive, SSc-specific autoantibodies that did not change during the course of disease. The autoantibody subsets are distinguished by patterns of cutaneous involvement, specific clinical features, and prognosis (Table 3). SSc-specific autoantibodies were found to be stronger predictors of disease outcome and organ involvement than the extent of skin involvement.27 The subset of patients with SSc positive for ACA represents a clinically homogenous group with distinct clinical features and seems to have a better prognosis: less severity; less frequent ILD, SRC, inflammatory arthritis, and inflammatory myositis; and patients had lower rates of GI tract involvement, finger ulcers, digital tuft resorption, or finger contractions. The patients are also older at disease onset, predominantly female, and more likely to have limited disease, lower skin scores, telangiectasia and pulmonary hypertension.10,21,29,51–57,59,61–63,65,69–71,73,74,84,86,89,124 ACA status was found to be predictive of the extent of skin involvement over time.59 Patients with limited disease who were ACA-negative at baseline were more likely to progress to diffuse disease. ACA-negative patients also had a greater extent of cutaneous involvement, worse survival, and more severe internal organ involvement.29,65
Another study supported subdivision of lcSSc into 2 serological subtypes, Th/To-positive and ACA-positive, with different internal organ involvement and outcome.50 Compared to the ACA-positive patients, Th/To-positive patients were younger at disease onset and predominantly male, with less PAH development, but more ILD (38% vs 4.5%). The highest mortality was found in ATA+ and ATA+/ACA– subgroups, while ACA+/ATA– and Pm/Scl+/RNA polymerase antibody (RNAP)-negative patients were classified as low risk.26 Some patients were not within described serological subsets; for example, ACA was commonly found in association with mild skin involvement, but 9% of dcSSc patients with truncal involvement were positive for ACA.10
Caetano, et al described those patients who had a more insidious onset of skin and major organ involvement, a lower incidence of ILD and SRC, and better survival than expected for dcSSc as a distinct clinical subtype (dcSSc ACA+).70 Thus, further subgrouping within each autoantibody profile may be promising from a clinical point of view. Indeed, 2 subgroups of anti-CENPA can explain variable clinical manifestations in an ACA-positive subset.87 Subgrouping among patients with SSc positive for anti-RPC155 antibodies (RNAP III large subunit, 155 kDa) revealed that anti-RPA194 was associated with a lower cancer risk and less severe GI disease, while anti-RNAP I/II/III was associated with SRC.75 Therefore, different autoantibody combinations have utility as tools for organ involvement and cancer risk stratification in SSc.
Patterson, et al86 reported subgrouping RNAP III–positive patients into 2 clusters; a strongly positive cluster was associated with an increased risk of GAVE, lower risk of esophageal dysmotility, and shorter disease duration. A strong positivity for anti-RNAP III (a higher ELISA index) was associated with the development of SRC.75 Although 3 main autoantibodies (ACA, ATA, and anti-RNAP III) have strong mutually exclusive relationships, coexpression of other antibodies are relatively common.86,90,125,126 A combination of 2 SSc-related autoantibodies was revealed in one-third of patients in the study by Patterson, et al.86 Anti-Ro52 most frequently occurred in combination with other autoantibodies, but coexpressions of ATA with anti-RNAP III (0.6%) and ACA (3%) were also found in a small proportion of patients with SSc.86 In cases with coexistence of ≥ 2 autoantibodies, the autoantibody of highest titer determined the clinical phenotype.
SSc subsets according to nailfold capillary abnormalities. Capillary abnormalities seen on nailfold video capillaroscopy (NVC) can be used to subgroup SSc patients with different clinical manifestations and prognoses. There are 2 classification systems based on the NVC changes (Table 4). First, Maricq, et al127 described 2 capillary patterns: “slow” and “active.” Slow pattern was characterized by capillary telangiectasias and high number of extremely large (giant) capillary loops with a relatively well-preserved capillary distribution. The main feature of active pattern was moderate-to-extensive capillary loss associated with considerable distortion of the nailfold capillary bed and new blood vessel formation (bushy capillaries). Associations between capillaroscopic findings and disease activity, degree of progression, and prognosis were found. SSc patients with slow pattern predominantly had slowly progressive disease (new symptoms/signs during follow-up were found only in 1/11 patients), longer RP prior to entry, and were ACA-positive, while all patients with active pattern were ACA-negative and half showed disease progression. Capillary loss (active pattern) reflected disease progression that was confirmed in other publications.98,114 Ostojic, et al103 found that enlarged capillaries without a significant capillary loss (slow pattern) were more frequently seen in lcSSc, whereas giant capillaries (GCs) with advanced capillary loss (active pattern) occurred in dcSSc.
The Maricq NVC classification system has been further subdivided within the active pattern into “active” and “late,” whereas slow pattern was renamed as “early” by Cutolo, et al.95,128 The principal change was the interpretation of patterns as consecutive phases of progressive obliterative microangiopathy.128 Early pattern is characterized by a relatively well-preserved capillary distribution and density with a few enlarged capillaries/GCs, few capillary microhemorrhages, and no evident loss of capillaries. The following moderate loss of capillaries is a sign of the next active phase, with a mildly disturbed architecture of capillaries, frequent GCs and microhemorrhages, capillary derangement, and absent or few ramified capillaries (neoangiogenesis). The capillary changes typical for this phase (hemorrhages and GCs) are closely associated with disease activity. Sambataro, et al showed that NEMO score (cumulative number of microhemorrhages and microthrombosis) ≥ 6 was the best predictor of disease activity, followed by the GC score (number of GCs) ≥ 3.118 The active pattern had more severe disease manifested as extensive skin involvement and greater visceral involvement (muscle, kidney), and patients were ACA-negative in comparison with the early pattern.91 In the most advanced phase of SSc microangiopathy, represented by the late NVC pattern, the disorganization of the normal capillary array is generally seen, with severe loss of capillaries and large avascular areas, irregular enlargement of the capillaries, few or absent GCs, microhemorrhages, and ramified/bushy capillaries. Normal NVC pattern is rarely seen in SSc (4–12%), nearly exclusively in the lcSSc subset.103,129 Numerous studies confirmed that patients with more advanced NVC patterns had more severe disease.91,92,93,98,103,127,129 Significant capillary loss was more common in patients with lcSSc who met ACR criteria compared to those who did not.115
Classifying patients with SSc according to NVC patterns may predict development of a new organ involvement within 1 year.98,100 In 2 studies,98,100 the odds ratio to develop severe organ involvement (defined as a category 2 or higher in any of the 9 organ systems assessed according to the Medsger Disease Severity Scale, or new PAH or ILD at 18–24 months’ follow-up) was stronger according to more severe NVC patterns, adjusting for disease duration, subset, and vasoactive medications. These findings were externally validated in an Italian cohort.100 Associations between certain manifestations and NVC patterns are controversial, such as reduced capillary density and PAH.107,108 Sample size was sometimes too small to detect possible associations.104
All 3 NVC patterns can be observed in both clinical disease subsets (lcSSc and dcSSc)128; however, early pattern is more common in lcSSc, especially early lcSSc,93 whereas the late pattern is more prevalent in dcSSc.92,93 Classifying patients into NVC subsets is important early in the disease course because capillary loss is a reliable indicator of rapidly progressive early disease.25,94 Shenavandeh, et al showed that late pattern in patients with early SSc was associated with severity of finger contractures and significantly reduced pulmonary function, compared to active and early patterns.94 Table 4 demonstrates that the reduced number of capillaries typical for active and late patterns was more commonly seen in patients with longer disease duration, higher mRSS, more severe lung (including PAH), GI, and peripheral vascular involvement, a higher number of organs affected, and elevated ESR and C-reactive protein.94,101–103,105,107,109–114,117,118,119 The ACR criteria sensitivity may be improved by adding the NVC patterns.115,116 More severe NVC patterns (active and late) occurred in patients seropositive for ATA and anti-RNAP III, and negative for ACA.93,95,117,119 ANA-negative99 and ACA-positive94 patients had the most favorable early pattern. However, SSc-related autoantibodies are not directly linked with the development of a distinct SSc NVC pattern (Table 4 and Table 5).129
The limitations included small proportions of patients with each NVC pattern (especially early pattern), resulting in limited power to detect statistically significant differences. Some outcomes were omitted from the analysis (i.e., GI involvement and SRC), while others might have been interrelated (i.e., abnormalities in the cardiac measures might be secondary to pulmonary involvement, rather than present as primary cardiac involvement). Further, follow-up duration in the prospective studies varied and was relatively short. Definitions of organ involvement also varied between the studies, which made the comparison of the results difficult.
DISCUSSION
SSc subset classification is a rapidly evolving field. Our systematic review highlights both the continued importance of skin involvement and the novel role of SSc-specific antibodies, abnormal nailfold capillary patterns, and molecular profiling in assessing patients to determine a subset.
The dcSSc subset comprises patients with rapidly progressive disease who require more aggressive treatment. However, disease progression assessed as severity-duration ratio (early significant visceral and skin involvement) suggests disease activity only in early dcSSc.23,130,131 In later stages of disease, patients classified as rapid progressors in the beginning may still have a high disease severity due to the accumulated significant damage, but low disease activity as a result of treatment or spontaneous remission. Some patients with SSc first develop severe skin involvement and/or visceral disease late in the disease course. Thus, the limited/diffuse system loses its predictive value in more advanced disease and should be supplemented with a necessary determination of disease activity and severity when it comes to choosing treatment. With the recent advances in antibody detection, some novel SSc-specific autoantibodies could be added to SSc subset classification autoantibody profiling to the skin involvement while determining a subset.
Based on gene expression profiling, patients with lcSSc can be assigned to the limited, inflammatory, or normal-like subsets, whereas fibroproliferative subset can be seen in patients with dcSSc. The molecular subsets seem to be a universal feature of SSc end-target organ pathology, not affected significantly by heterogeneity of skin involvement within a patient and/or fibroblast heterogeneity in tissues.38,39,122 The molecular intrinsic subset assignment could represent a valuable approach for matching patients with SSc to appropriate therapies. Molecular phenotyping may aid personalized medicine by identifying therapies with higher potential for success in each individual patient, as well as to select patients with SSc who will improve naturally as part of their disease course.47
Some limitations of subgrouping by molecular phenotyping include the relatively small sample sizes of clinical trials due to the rarity of disease itself, specific inclusion criteria that misrepresents the full spectrum of SSc, lack of controls, and differences in methods of transcript quantification and in the exact list of genes between studies. Moreover, not all therapy- or disease-relevant genes are regulated at the mRNA level. The use of molecular subsetting in clinical practice for individual patients is limited, as paired skin samples from each individual are often not available, analyses are not standardized, and large numbers of samples in a dataset are needed to identify the molecular subset with accuracy. Recently, supervised machine learning algorithms have been developed and may be successfully used to assign single samples to intrinsic gene expression subsets according to predefined criteria.47 The method utilizes a multinomial elastic net classifier and an optimized set of genes. Classifier accuracy in that study was proved using concordance of samples (83.3%) reporting Cohen κ coefficient (0.7391), and was externally validated. Further efforts are needed to explore molecular heterogeneity and intrinsic subsets in other tissues and particularly in peripheral blood, given its accessibility.
Attempts to identify SSc subsets considering SSc-specific autoantibodies have faced a variety of challenges. Boonstra, et al reported that adding autoantibody status to the cluster process resulted in correct classification of patients with ILD, PAH, and SRC.27 All high-risk patients were correctly identified by taking autoantibodies into account, but the number of patients incorrectly identified as possibly high risk increased significantly (by 66%), suggesting limited additional value of autoantibody status for clustering.27 The limitations of studies on SSc-specific autoantibodies included underestimation of the number of antigens due to the limitations of the techniques not allowing the identification of membrane proteins, or to a loss of proteins at each step, small sample size, a lack of validation groups, and/or limited generalizability (e.g., SRC is rare in Japanese patients; clinical features in each SSc-related ANA-based subgroup appear to vary among populations of different backgrounds). Feasibility is another consideration, as some autoantibodies are identified by immunoprecipitation, which is not widely used in clinical laboratories, and/or some detection kits are not commercially available. Limitations of classification systems developed by cluster analysis are the exclusion of a significant number of patients due to missing data and/or loss to follow-up that affects the extrapolation of the results. Finally, there have been inconsistent definitions of variables between the studies, a lack of analysis of the potential effect of treatment regimens on survival, and the influence of disease duration on the clustering process.
In conclusion, modern methods to subset SSc include skin involvement, immunologic profile, molecular signatures, visceral involvement, and age. Classifying on the basis of skin involvement, NVC, and autoantibody profile may allow early prediction of internal organ involvement. Molecular subsetting may identify those who are likely to respond to therapy. Longitudinal prospective studies to track subsets are needed to provide insight into disease trajectory, assess their predictive value, and confirm a possible transition between subsets and evolution under treatment.
ACKNOWLEDGMENT
We are thankful to Melanie Anderson, an information specialist at the University Health Network Library Services, and Keshini Devakandan, a clinical research analyst in the Toronto Scleroderma Program, for their assistance with the literature search.
Footnotes
This work was supported by a grant from the Scleroderma Foundation and the World Scleroderma Foundation. SRJ is supported by a Canadian Institutes of Health Research New Investigator Award and the Gurmej Kaur Dhanda Scleroderma Research Award. DK was funded by the National Institutes of Health/National Institute of Arthritis and Musculoskeletal and Skin Diseases grant 5K24AR063120-07.
The authors declare no conflicts of interest relevant to this article.
- Accepted for publication April 26, 2021.
- Copyright © 2021 by the Journal of Rheumatology
This is an Open Access article, which permits use, distribution, and reproduction, without modification, provided the original article is correctly cited and is not used for commercial purposes.
REFERENCES
- 1.↵
- 2.↵
- 3.↵
- 4.↵
- 5.↵
- 6.↵
- 7.↵
- 8.↵
- 9.↵
- 10.↵
- 11.↵
- 12.↵
- 13.↵
- 14.↵
- 15.↵
- 16.↵
- 17.↵
- 18.↵
- 19.↵
- 20.↵
- 21.↵
- 22.↵
- 23.↵
- 24.↵
- 25.↵
- 26.↵
- 27.↵
- 28.↵
- 29.↵
- 30.
- 31.
- 32.↵
- 33.↵
- 34.
- 35.
- 36.
- 37.↵
- 38.↵
- 39.↵
- 40.
- 41.
- 42.↵
- 43.
- 44.
- 45.↵
- 46.
- 47.↵
- 48.
- 49.↵
- 50.↵
- 51.↵
- 52.
- 53.
- 54.
- 55.
- 56.
- 57.↵
- 58.
- 59.↵
- 60.
- 61.↵
- 62.
- 63.↵
- 64.
- 65.↵
- 66.
- 67.
- 68.
- 69.↵
- 70.↵
- 71.↵
- 72.
- 73.↵
- 74.↵
- 75.↵
- 76.
- 77.
- 78.
- 79.
- 80.
- 81.
- 82.
- 83.
- 84.↵
- 85.
- 86.↵
- 87.↵
- 88.
- 89.↵
- 90.↵
- 91.↵
- 92.↵
- 93.↵
- 94.↵
- 95.↵
- 96.
- 97.
- 98.↵
- 99.↵
- 100.↵
- 101.↵
- 102.
- 103.↵
- 104.↵
- 105.↵
- 106.
- 107.↵
- 108.↵
- 109.↵
- 110.
- 111.
- 112.
- 113.
- 114.↵
- 115.↵
- 116.↵
- 117.↵
- 118.↵
- 119.↵
- 120.↵
- 121.↵
- 122.↵
- 123.↵
- 124.↵
- 125.↵
- 126.↵
- 127.↵
- 128.↵
- 129.↵
- 130.↵
- 131.↵