Abstract
Objective To perform a scoping review focusing on osteolysis in systemic sclerosis (SSc).
Methods This review was performed in line with the Preferred Reporting Items for Systematic reviews and Meta-Analyses extension for Scoping Reviews (PRISMA-ScR) recommendations.
Results From a total of 351 results, 29 articles were included for the final analysis. The publications included proved to be heterogeneous regarding the population and inclusion criteria. The lack of a standardized method of detection of osteolysis further enhanced these inequalities. Most studies reported location/prevalence of osteolysis and associations with other manifestations, with only a minority focusing on topics like predictors of osteolysis and its prognostic value. None of the authors addressed treatment approach. The most frequently analyzed and prevalent location was acro-osteolysis (AO). Diffuse cutaneous subtype and anti-topoisomerase I antibody correlated positively with AO. Disease duration, calcinosis, and digital ischemia were the features more frequently associated with AO, but only the last 2 predicted AO. Ultrasound showed high sensitivity for detection of AO.
Conclusion Despite the effect that osteolysis has on patients with SSc, there is a significant lack of studies on this area. Notably, there are no studies that we know of focused on treatment. Also, there is a lack of longitudinal studies that would allow a reliable assessment of its prognostic value and predictors.
Systemic sclerosis (SSc) is a connective tissue disease characterized by a triad of immune dysregulation, vasculopathy, and fibrosis.1,2 It can afflict multiple organ systems, including the musculoskeletal system, with arthritis and tenosynovitis.1,3
Osteolysis is also a feature of SSc that consists of active resorption of bone matrix by osteoclasts. When located in the distal phalanx of fingers/toes, it is referred to as acro-osteolysis (AO) and can result in finger clubbing/shortening. Osteolysis can also appear in the mandibulae (OM), leading to facial deformities and masticatory problems. In long bones, it might even be mistaken for a malignant lesion.4
Despite the negative effect that osteolysis has in SSc, it is still overlooked, with few investigations focusing on its characteristics and treatment options. The authors performed a scoping review of the literature, aiming to systematize the knowledge to date concerning osteolysis in SSc, and to identify areas with need of further research.
METHODS
Data source and search strategy. This review was conducted in line with the Preferred Reporting Items for Systematic reviews and Meta-Analyses extension for Scoping Reviews (PRISMA-ScR). The authors performed a search string in 2 databases, PubMed and Web of Science, from database inception up to October 31, 2021, without filters, using the following keywords:
• Pubmed: ([systemic scleroderma (Medical Subject headings [MeSH] Terms)] or [SSc (Title/Abstract)] or [calcinosis, Raynaud phenomenon, esophageal dysmotility, sclerodactyly, telangiectasias (CREST) syndrome; Title/Abstract)] or [systemic scleroderma (Title/Abstract)]) and ([osteolysis (MeSH Terms)] or [osteolysis (Title/Abstract)] OR [AO (Title/Abstract)])
• Web of Science: (systemic scleroderma [Topic] or SSc [Topic] or CREST [Topic]) and (osteolysis [Topic] OR AO [Topic])
Screening process and selection criteria. After the search, MGG and MR screened the records in 3 stages (title, abstract, and full text) to assess for inclusion. A manual search through the references of the retained manuscripts was also performed in order to detect additional reports. Records were considered eligible when both reviewers included them for the next step. When opinions differed, consensus was reached by discussion with MO.
Manuscripts selected were original articles that focused on any relevant aspect of osteolysis (prevalence; skeletal locations; demographic, clinical, and serologic associations; prognostic role; predictors; treatment options; and imaging diagnosis), in patients aged ≥ 18 years with a diagnosis of SSc. The authors accepted articles in 3 languages: English, Spanish, and Portuguese. Reviews, letters, editorials, abstracts of scientific meetings, and case series with less than 5 patients were excluded. Studies that included patients with overlap syndromes or patients aged < 18 years were also rejected.
After screening, manuscript data were structured in a standardized electronic spreadsheet that included the following: author, year of publication, study design, population origin/nationality, SSc classification criteria, SSc subtypes included, sample size, and, last, the areas of focus. Areas of focus were identified after full-text reading, in order to categorize the major subjects studied by authors. A total of 10 areas of focus were identified: physiopathology, osteolysis location/prevalence, association to autoantibodies (Ab), association to SSc skin subtype, association to nailfold capillaroscopy findings, association to other disease features, osteolysis predictors and prognostic value, patient-reported outcomes (PROs), radiographic evaluation, and ultrasound (US).
Quality appraisal. The quality of the selected studies was evaluated by MGG and MR using the National Heart, Lung, and Blood Institute Study Quality assessment tools.
RESULTS
Three hundred fifty-one results were obtained from the search string in both databases. After duplicate exclusion, 252 were screened sequentially by title, abstract, and full text (Figure). Through quality appraisal and cross-reference, 29 were finally included in the review (Supplementary Materials, available from the authors upon request).
Flowchart of systematic review according to PRISMA guidelines. PRISMA: Preferred Reporting Items for Systematic reviews and Meta-Analyses.
Table 1 summarizes all included manuscripts, dated from as early as 1977. Almost one-third (n = 8) did not use any classification criteria,5-12 with the remaining using at least 1 of the following: American College of Rheumatology (ACR) 1980 criteria,13-19 ACR/European Alliance of Associations for Rheumatology (EULAR) 2013 criteria,20-27 Leroy and Medsger 2001 criteria,22,28-31 or Medsger and Masi 1971 criteria.32,33 This reflects the extended time period represented, roughly half a century, and contributes to a heterogenous population, worsened by representations from several countries from North and South America, Europe, and Asia. Concerning study design, most manuscripts relied on case-control8-10,12,19,23,30-33 or cross-sectional7,11,13-18,20-22,24-28 methodologies, with only 3 cohorts.5,6,29 As for quality assessment (Supplementary Materials, available from the authors upon request), only 2 articles were considered “good.”20,26
Main characteristics of studies included in the scoping review.
Physiopathology. Despite the absence of a consensual physiopathologic pathway to explain osteolysis, only 3 studies focused on this topic.
Park et al19 studied the association between AO, osteoclastogenesis, and vascular endothelial growth factor (VEGF) levels. They concluded that patients with SSc with AO had increased levels of VEGF and raised osteoclastogenesis compared to patients with SSc without AO (Mann-Whitney U test, P = 0.005, P < 0.001) and controls (Mann-Whitney U test, P = 0.02, P < 0.001). Moreover, radiographic extension of AO also correlated positively with osteoclastogenesis (Spearman ρ 0.741, P = 0.01).
Braun-Moscovici et al,15 on the other hand, focused on the levels of parathormone (PTH) and vitamin D levels, having reported a correlation between the presence of AO and increased PTH (chi-square test, P = 0.02), and increased PTH with low vitamin D serum levels (Pearson r −0.312, P = 0.01). In patients with hyperparathyroidism, other possible causes like chronic kidney failure were excluded.
Zaghlol et al20 studied the association between AO, osteoporosis, bone mineral density (BMD), and peripheral vascularity in 30 premenopausal women with diffuse cutaneous SSc (dSSc). In a similar way to Park et al,19 the authors also hypothesized that hypoxia generated osteoclastogenesis and, subsequently, osteoporosis and AO. They reported that patients with AO correlated with low distal radius BMD (multivariate logistic regression, odds ratio (OR) 1.9, 95% CI 1.1-3.0, P = 0.002), and had more frequent radial/ulnar arterial impairment (reduction in velocity of at least 50% compared to the brachial artery; OR 2.2, 95% CI 0.7-6.5, P = 0.04 and OR 1.6, 95% CI 0.9-3.0, P = 0.05, respectively). However, contrary to the results presented by Braun-Moscovici et al,15 there was no difference in vitamin D, calcium, and PTH levels between patients with vs without AO.
Osteolysis location and prevalence. Osteolysis was reported in many sites (Table 2), including the terminal tufts of fingers (AO) and toes, distal ulna, distal radius, ribs, collarbone, and OM. The most commonly reported and prevalent location was AO, with variable percentages.5,6,11,13-15,17-31 Considering studies without skin subtype restriction, Braun-Moscovici et al15 presented the highest AO prevalence (70%), followed by Sakchaikul et al,24 with 64.1%. Frech et al reported the lowest, with 5%.6
Osteolysis location and prevalence in included studies.
OM was the second most reported location, mainly in earlier studies that relied upon smaller samples.8-10,12,14,32,33 The most recent study, that of Yalcin et al, which included the largest sample, reported a prevalence of 8%. As for other studies with no skin subtype restriction, percentages varied between 19.1% and 29%.10,14,32,33
Cutaneous subtype. Eleven studies explored the relation between SSc cutaneous involvement and AO (Table 3). Eight found no association,7,11,17,19,24,26,28,30 including 3 of them using multivariate analysis.7,26,28
Association between AO and cutaneous subtype in included studies.
Two studies reported a positive correlation between AO and the diffuse subtype,18,27 one of which used multivariate regression.27 Azarbani et al27 also evaluated the distal ulna, but without significant results.
Autoantibodies. Ten studies evaluated the association between AO and SSc-specific Ab (Table 4). The most commonly reported was with antitopoisomerase I (anti-topo I),16,18,24,27 which was reported in 2 studies through multivariate analysis.24,27 Sakata et al18 and Johnstone et al7 found a negative correlation with anticentromere Ab. In the study by Jacobsen et al,16 patients with negative antinuclear Ab were less likely to present with AO.
Association between AO and auto-antibodies in included studies.
Nailfold videocapillaroscopy. Only Morardet et al26 studied capillaroscopic findings, specifically concerning the reported SSc patterns: early, active, and late. Through multivariate logistic regression, the authors confirmed an association between late pattern and presence of AO (OR 4.57, 95% CI 1.66-12.55, P = 0.003). This finding was further confirmed when evaluating capillaroscopic parameters separately: patients with AO presented lower capillary density (ANOVA, 2.88 ± 1.30 vs 5.60 ± 2.26; P = 0.001), and patients with severe AO had lower capillary density (ANOVA, 2.60 ± 1.20 vs 5.60 ± 2.26; P = 0.001), the presence of neoangiogenesis (chi-square, P = 0.02), and the absence of microhemorrhages (chi-square, P = 0.02). A late pattern was also more common in patients with severe AO (chi-square, P < 0.001).
Other disease features and organ damage. Multiple studies focused on other disease features and organ damage, with very heterogenous results (Table 5). Once again, the focus was on AO. However, 3 studies also approached OM in bivariate analysis,8,9,33 and 1 study investigated distal ulna resorption in multivariate analysis.27
Clinical associations reported in the studies included in the scoping review.
For AO, manifestations of digital ischemia, calcinosis, and prolonged disease duration were the most commonly reported associations. Three studies also correlated osteolysis with disease severity: AO with Medsger Severity Scale,13,30 or OM with number of organ systems afflicted.33 Notably, there is a single reported negative association, with puffy hands, by Sakchaikul et al.24
Two studies,8,9 conducted on the same population, evaluated different outcomes. One explored the relationship between OM and mouth opening,8 and the other OM and masseter abnormalities in magnetic resonance.9 Unfortunately, both failed to find a significant association. However, OM correlated with longer disease duration. In the study by Wood and Lee,33 patients with OM had worse mouth opening measurements.
Osteolysis as prognostic factor/osteolysis predictors. Only Avouac et al5 focused on this subject, more specifically AO. They evaluated 103 patients through approximately 5 years with hand radiographs at baseline and end of follow-up. Radiographic progression was observed in 22 out of 103 patients. In an initial bivariate analysis, several predicting factors of AO progression were found: calcinosis, digital ulcers, flexion contractures, pulmonary hypertension, lung fibrosis, and anti-topo I. However, after multivariate Cox regression analysis, only calcinosis at baseline (OR 6.17, 95% CI 1.16-47.23, P = 0.03) and digital ulcers (OR 12.43, 95% CI 1.97-88.40, P = 0.01) remained significant. Conversely, AO predicted progression in flexion contractures, but significance was lost in multivariate analysis.
Patient-reported outcomes. Frech et al6 posed a peculiar and unique question: how did hand impairment affect PRO measure completion rate in 339 patients with SSc for less than 5 years? The study included multiple questionnaires: patient global assessment, Scleroderma Health Assessment Questionnaire, Scleroderma Clinical Trials Consortium Gastrointestinal Tract questionnaire, Patient-Reported Outcome Measurement Information System–29, patient skin assessment, modified Medical Research Council dyspnea scale, Functional Assessment of Chronic Illness Therapy Dyspnea questionnaire, and Resource Use Questionnaire (RUQ). Through logistic regression adjusted for multiple variables (including questionnaire size and modality [electronic vs paper]), AO correlated with PRO noncompletion (OR 3.57, 95% CI 1.37-9.32, P = 0.001). RUQ was the least completed PRO, especially in the younger population.
In another study, Valenzuela et al21 explored if AO correlated with higher scores in PRO (Health Assessment Questionnaire–Disability Index, Cochin Hand Functional Scale, and visual analog scale pain), but no significant association was found (multivariate linear regression).
Radiographic scales. To date, osteolysis has been generically accepted and defined as the presence of bone resorption in a given imaging method. Being the most commonly available and inexpensive exam, radiography has been the most used technique to document its presence. This is reflected in all the included manuscripts—they relied solely on radiographic exams to identify osteolysis. Additionally, none of the existing radiographic scales to grade osteolysis are fully validated to the present date, resulting in inconsistency between authors when interpreting radiographs. For example, 4 groups created their own scale for AO.7,19,26,30
Johnstone et al7 developed a grading scale for AO in each finger that ranged from 0 (no resorption) to 4 (complete resorption of the terminal tuft, with obvious penciling). Finger scores were then added together to obtain an overall rating: normal/minimal AO (maximum score for an individual finger of 1 and total score up to 8), moderate AO (maximum score for an individual finger of 2 and total score from 9 to 16), and severe AO (maximum score for an individual finger of 3 or more or a total score of at least 17). Two observers (1 being a radiologist) would view the radiograph and rate each finger through consensus. This scale was later used in the studies of Sakchaikul et al24 and Zaghlol et al.20
Park et al19 also graded AO, using a scale from 0 to 3: 0 = none, 1 = minimal/doubtful, 2 = evident in less than 50% of the tuft, and 3 = evident in more than 50% of the tuft. Scores were afterwards added together. Images were reviewed and rated by consensus by 2 team members (both musculoskeletal radiologists).
Morardet et al26 defined AO as any resorption of bone of at least 1 distal phalanx. If it reduced the length of at least 1 distal phalanx by more than 50%, it would be classified as severe AO. The images were reviewed by 2 independent investigators and, if discrepant, a consensus would be achieved by 2 other distinct elements.
Koutaissoff et al30 also rated each finger tuff for AO, through a scale from 0 to 3 (0 = normal, 1 = doubtful, 2 = evident, 3 = severe). For a finger to have AO, the given score should be > 1. Two musculoskeletal radiologists rated the images, and the interobserver 
coefficient was calculated not only for AO but also for calcinosis, periarticular calcifications, surface/marginal erosions, joint space narrowing, periarticular osteopenia, and collapse arthropathy. The authors reported a 
ranging from 0.64 to 1, reflecting good to excellent agreement.
The remaining authors considered osteolysis as a binary outcome (present vs absent). Out of these, Azarbani et al,27 White et al,32 Wood and Lee,33 Yalcin et al,12 and Sakata et al18 had radiographs evaluated by 2 independent investigators. Sakata et al18 additionally reported an interobserver 
coefficient of 0.60 to 0.85. Avouac et al5 decided to have 3 readers, with an additional fourth if needed for consensus.
Musculoskeletal ultrasound. One study focused on the potential role of US in the detection of AO.31 The authors evaluated 44 patients with SSc and 30 healthy controls with hand radiography and US plus Doppler. US was 90% sensitive in the detection of AO in patients with SSc. Further, they reported an increased Doppler signal adjacent to AO lesions, possibly reflecting the presence of granulation tissue as an attempt of osteogenesis.
DISCUSSION
Most of the included manuscripts in this review focused on location, prevalence, and clinical associations of AO in SSc. Only 1 group each explored the prognostic value/predictors,5 association with capillaroscopy,26 and the usefulness of US.31 Surprisingly, no original studies about treatment options were found.
As a frequently overlooked manifestation of SSc, it is not unexpected that only 3 articles studied the physiopathology of osteolysis. A possible explanation to this manifestation, which is not completely understood, is a reduced vascular supply in the extremities that leads to hypoxia, initially by an obliterative microvasculopathy and later with the contribution of vessel compression by skin fibrosis, flexion contractures, calcinosis, and autonomic disfunction.34
The main regulator of the response to hypoxia is the hypoxia-inducible factor-1α. As the name suggests, its production is stimulated by hypoxia, and it stimulates proangiogenic gene transcription, including VEGF.34 On the other hand, osteoclasts are known to express VEGF receptors.35 With this in mind, and with the results of Park et al,19 one can hypothesize that through a reduced vascular supply and subsequent hypoxia, the VEGF production culminates not only in neoangeogenesis but also in osteoclastogenesis.
The findings of Zaghlol et al20 also strengthen the hypoperfusion hypothesis. In this study, AO correlated with lower radius BMD, possibly reflecting a regional increase in osteoclastic activity. Additionally, AO also associated with macrovascular impairment (ulnar and radial artery), raising again the theory of local hypoperfusion/hypoxia as the causative mechanism. Nevertheless, this concept is mostly applicable to extremities and osteolysis can occur in places where hypoperfusion is not a hallmark, such as the collarbone or mandibulae.34
A possible explanation for vitamin D deficiency reported by Braun-Moscovici et al15 is the small bowel impairment that occurs in SSc that might result in impaired vitamin D absorption. This leads to raised PTH, which is able to increase nuclear factor-
B ligand (RANKL) levels and stimulate osteoclastogenesis. However, this would affect bone homeostasis diffusely and not in a localized manner, as seen in osteolysis. It is therefore clear that more studies are necessary, as it is indispensable to have a complete understanding of osteolysis mechanisms to successfully treat this manifestation.34
As per anatomic locations, AO was the most frequently reported, followed by OM. However, very few authors explored other skeletal areas.14,27 Proximal bones (such as the collarbone), where apparently hypoperfusion is not an inciting mechanism, are underrepresented, further impairing the understanding of the physiopathology of osteolysis.
The simultaneous association of AO with dSSc and anti-TopoI is not surprising, as this skin subset and Ab are also interlinked. However, these correlations need further clarification, more specifically dSSc, as only 2 of the included manuscripts corroborate it.18,27 Moreover, SSc is one of the connective tissue diseases with more breakthroughs in detection of new Ab. More than a diagnostic tool, these appear to associate with specific phenotypes. It is therefore necessary to further clarify their relation to osteolysis, as they were scarcely evaluated.16,18,19
Studies assessing the relation of osteolysis with other disease features and organ damage focused mostly on AO. An association with calcinosis, digital ischemia, and longer disease duration was the most frequently reported.7-9,15,18-23,26,27,30 Such results seem to point toward a common pathophysiological origin with osteolysis. If ischemia leads to osteolysis, its contribution to calcinosis is also known.36 Further, the deposition of calcium in tissues may be related to its reabsorption from the bone that occurs in osteolysis, contributing, in a way, to the normalization of serum calcium values.15 Finally, ulcers are a manifestation that occurs mainly due to endothelial damage with proliferation of the intima of the vessels, narrowing of the lumen, and consequent ischemia.1 The presence of possible granulation tissue (translating the attempt of osteogenesis) on US and the occurrence of neoangiogenesis on capillaroscopy are also worth noting as both suggest an attempt to counteract osteolysis. This is reinforced by the importance of VEGF both for osteoclastogenesis and neoangiogenesis.19,26,31
Knowledge of the effect of AO on daily life may be biased by the patient’s inability to answer questionnaires that require the use of fingers.6 Paradoxically, this might lead clinicians and investigators to underestimate osteolysis impact. To reverse this paradox, healthcare professionals must be instructed to question, report, and investigate patients’ musculoskeletal complaints. Then, we can better understand the findings of Frech et al6 and Valenzuela et al,21 and take measures based on more trustworthy data.
Currently, the assessment of the severity of SSc is performed using the Medsger Severity Scale, a dated tool that requires obsolete exams and needs to be updated. AO is a potential new component, since its diagnosis is accessible and inexpensive, and there is a relationship between the severity of SSc and the manifestation of AO throughout the disease.13,30
This scoping review is not devoid of limitations. Above all, the quality of studies is highlighted; most were “reasonable,” with only 2 of “good” quality (Supplementary Materials, available from the authors upon request), with most of the reported results based on bivariate analysis. Also, only a minority used the most recent and reliable ACR/EULAR 2013 classification criteria. This, as stated before, has led to highly heterogeneous populations, thus not allowing for a more accurate depiction of osteolysis in SSc. As for prevalence, the variability of percentages reported can further be explained by the fact that some authors focused only in specific subpopulations of SSc. For example, Frech et al,6 with the lower prevalence of AO of 5%, included only patients with shorter disease duration (less than 5 years). Simeón-Aznar et al,29 who only included patients with limited cutaneous and sine scleroderma SSc, also reported a low value (7.8%). In both, individuals with a higher probability of presenting AO were excluded; that is, those with longer disease duration and with diffuse cutaneous SSc. Zaghlol et al,20 on the contrary, included only patients with the diffuse cutaneous subtype and found a higher prevalence (80%).
The effectiveness of treatments for osteolysis was not explored in any study. Calcium channel inhibitors and bisphosphonates are examples of options currently used, but without any evidence of efficacy.19,26 Considering the role of VEGF-induced osteoclastogenesis, denosumab may be a promising agent.19 Therapies could also focus on angiogenesis stimulation or vasodilation—prostaglandins and endothelin receptor antagonists are potentially promising.26
Our review also identified a lack of cohort studies that grant a correct assessment of the cause-effect relationship. The sole one that allowed conclusions on the evolution/prognosis of osteolysis was conducted by Avouac et al.5
Strikingly, the manuscripts results and conclusions were hampered by the absence of a consensual definition of radiographic osteolysis and validated radiographic scales. Further, the scales created to date concern only AO,7,19,26,30 and solely Koutaissoff et al30 evaluated interobserver variability. Investment in the creation and full validation of scales is essential for a correct assessment of patients with SSc, as well as for advancing research in this area. In the future, other imaging methods known to better depict bone tissue, such as computed tomography, could also be performed to better document the prevalence and severity of osteolysis.
Footnotes
M.G. Guerra and M. Rodrigues contributed equally as co–first authors.
The authors declare no conflicts of interest relevant to this article.
- Accepted for publication December 28, 2022.
- Copyright © 2023 by the Journal of Rheumatology
