Objectives Anti-Ro52 autoantibodies are associated with more severe interstitial lung disease (ILD) in adult myositis patients with antiaminoacyl transfer (t)RNA synthetase autoantibodies. However, few studies have examined anti-Ro52 autoantibodies in juvenile myositis. The purpose of this study was to define the prevalence and clinical features associated with anti-Ro52 autoantibodies in a large cohort of patients with juvenile myositis.
Methods We screened sera from 302 patients with juvenile dermatomyositis (JDM), 25 patients with juvenile polymyositis (JPM) and 44 patients with juvenile connective tissue disease–myositis overlap (JCTM) for anti-Ro52 autoantibodies by ELISA. Clinical characteristics were compared between myositis patients with and without anti-Ro52 autoantibodies.
Results Anti-Ro52 autoantibodies were found in 14% patients with JDM, 12% with JPM and 18% with JCTM. Anti-Ro52 autoantibodies were more frequent in patients with antiaminoacyl tRNA synthetase (64%, p<0.001) and anti-MDA5 (31%, p<0.05) autoantibodies. After controlling for the presence of myositis-specific autoantibodies, anti-Ro52 autoantibodies were associated with the presence of ILD (36% vs 4%, p<0.001). Disease course was more frequently chronic, remission was less common, and an increased number of medications was received in anti-Ro52 positive patients.
Conclusions Anti-Ro52 autoantibodies are present in 14% of patients with juvenile myositis and are strongly associated with anti-MDA5 and antiaminoacyl tRNA synthetase autoantibodies. In all patients with juvenile myositis, those with anti-Ro52 autoantibodies were more likely to have ILD. Furthermore, patients with anti-Ro52 autoantibodies have more severe disease and a poorer prognosis.
- juvenile idiopathic inflammatory myopathies
- anti-ro52 autoantibodies
- myositis associated autoantibodies
- interstitial lung disease
Statistics from Altmetric.com
- juvenile idiopathic inflammatory myopathies
- anti-ro52 autoantibodies
- myositis associated autoantibodies
- interstitial lung disease
What is already known about this subject?
The clinical features and prognosis of patients with juvenile myositis and anti-Ro52 autoantibodies were poorly defined.
What does this study add?
Approximately 15% of a large North American cohort of patients with juvenile myositis have anti-Ro52 autoantibodies.
Patients with juvenile myositis and anti-Ro52 autoantibodies are more likely to develop interstitial lung disease (ILD).
Anti-Ro52 autoantibodies are more common in patients with juvenile myositis and anti-MDA5 and antisynthetase autoantibodies.
Patients with juvenile myositis and anti-Ro52 autoantibodies more often have a chronic disease course and require more medications.
How might this impact on clinical practice or future developments?
Anti-Ro52 autoantibodies are useful prognostic markers for ILD and severe disease in patients with juvenile myositis.
Idiopathic inflammatory myopathies (IIMs) are a heterogeneous group of systemic autoimmune diseases characterised by weakness, chronic inflammation of skeletal muscles and elevated serum muscle enzyme levels.1 Many patients also have extramuscular manifestations, including involvement of the skin, lungs and/or joints. Most patients with IIM have a myositis-specific autoantibody (MSA), defined as an autoantibody found only in patients with IIM, which are typically mutually exclusive.2 In contrast, myositis-associated autoantibodies (MAAs) are found in IIM but may also be present in patients with other autoimmune diseases and may be seen in association with an MSA or other MAAs.
MSAs are associated with specific phenotypes.2 3 For instance, antimelanoma differentiation-associated gene 5 (MDA5) autoantibodies are associated with cutaneous ulceration and palmar papules, minimal muscle involvement, arthritis, interstitial lung disease (ILD) and high fatality rate.4–7 In contrast, patients with autoantibodies recognising histidyl-transfer (t)RNA synthetase (ie, Jo1) have antisynthetase syndrome, a unique multisystem autoimmune disease characterised by a combination of myositis, ILD, arthritis, Raynaud’s phenomenon, fever and/or mechanic’s hands.8 Of note, while many phenotypic features are similar between juvenile and adult IIM with the same MSAs, there are some important differences. For example, adults with anti-p155/140 (transcription intermediary factor 1; TIF-1) autoantibodies have an increased risk of malignancy, whereas anti-p155/140 (TIF-1) autoantibody-positive children do not.2 9
In adult patients with IIM, the most common MAA is anti-Ro52.10 Interestingly, anti-Ro52 autoantibodies often co-occur with anti-Jo1 autoantibodies11 and adult patients with both autoantibodies have more severe ILD and more frequently develop lung fibrosis than those with anti-Jo1 autoantibodies alone.12 13 In addition, higher anti-Ro52 autoantibody titres are associated with the development of more severe ILD,14 myositis and joint impairment in adult patients positive with anti-Jo1.15 Patients with both anti-Jo1 and anti-Ro52 autoantibodies have a poorer response to various immunosuppressive drugs and a decrease in survival.13 15
A recent analysis of 22 children with myositis revealed that 23% had anti-Ro52 autoantibodies, although specific clinical associations were not examined.16 The purpose of this study was to define the prevalence and clinical features associated with anti-Ro52 autoantibodies in a large cohort of patients with juvenile myositis.
Patients and methods
Patients and serum samples
Of the 543 patients from the Childhood Myositis Heterogeneity Collaborative Study who were enrolled between 1989 and 2016, with probable or definite myositis by Bohan and Peter criteria,17 those with a serum sample available for autoantibody testing at the time of enrolment were included in the study. Among the 371 patients with juvenile myositis included, 302 (81.4%) had juvenile dermatomyositis (JDM), 25 (6.7%) had juvenile polymyositis (JPM) and 44 (11.9%) had juvenile connective tissue disease–myositis (JCTM) overlap. The JCTM subgroup included patients meeting the criteria for myositis and another autoimmune disease, including 13 with juvenile systemic lupus erythematosus, 11 with juvenile systemic sclerosis, 7 with juvenile idiopathic arthritis and 13 with other autoimmune conditions including autoimmune hepatitis, eosinophilic fasciitis, diabetes mellitus, lichen sclerosis, linear morphea, psoriasis, Sjögren’s syndrome and ulcerative colitis. Sera from 90 healthy control children enrolled in the same studies were available.
All subjects were enrolled in institutional review board-approved natural history studies as previously described,18 and all provided informed consent. A standardised physician questionnaire captured demographics, clinical and laboratory features, environmental exposures at illness onset or diagnosis, as well as therapeutic usage and responses.18 Seven organ system symptom scores at diagnosis, defined as the number of symptoms present divided by the number of symptoms assessed, and an overall clinical symptom score as the average of the seven individual organ symptom scores were calculated as previously described.19–21 In 7 of 33 patients, the presence of ILD was diagnosed by high-resolution CT (HRCT) and lung biopsy. In 11 of 33 patients, ILD was diagnosed by HRCT alone and in 5 patients, ILD was diagnosed by biopsy alone. Seven patients were diagnosed with ILD by chest radiographic imaging combined with pulmonary function testing and did not undergo HRCT or lung biopsy. Three patients did not have imaging records available, and the diagnosis of ILD was based on physician documentation in the medical record. Complete clinical response and remission were defined as at least 6 months of inactive disease on or off therapy, respectively.20 A course of treatment was defined as a single episode from beginning of administration of a given medication to the termination of treatment with that medication, or combination of medications, in each patient. Medical record review, conducted in >75% of patients, verified the clinical, demographic, laboratory and therapeutic data contained in the physician questionnaires. Follow-up visits occurred in 55% of patients, with an average time from enrolment date to final evaluation of 4.3 years. Patient characteristics in our cohort are comparable with other registry-based JDM cohorts in terms of demographics and disease manifestations.22–25
Anti-Ro52 autoantibodies were detected using an enhanced performance Ro52 ELISA (SS-A 52 ELISA, Quanta Lite, INOVA Diagnostics, San Diego, CA) according to the manufacturer’s instructions. Other myositis autoantibodies were detected as previously described.18 26
Dichotomous variables were expressed as percentages and absolute frequencies, and continuous features were reported as means and SD. Pairwise comparisons for categorical variables between groups were made using χ2 test or Fisher’s exact test, as appropriate, while continuous variables were compared using Student’s t-test. Logistic and linear regression were used to adjust the comparisons for possible confounding variables, including the year of diagnosis, length of follow-up and MSAs. Creatine kinase, a highly positively skewed variable, was expressed as median, first and third quartile for descriptive purposes and transformed through a base-10 logarithm for analysis. All statistical analyses were performed using Stata/MP V.14.1 (StataCorp). As this was an exploratory study, a two-sided p value of ≤0.05 was considered statistically significant.
Prevalence and demographics of patients with anti-Ro52 autoantibodies
Anti-Ro52 autoantibodies were more prevalent in patients with juvenile IIM (JIIM) than in healthy control children (14% vs 1%). Sera from 14% of patients with JDM, 12% with JPM and 18% with JCTM had anti-Ro52 autoantibodies (figure 1, table 1). There were no significant differences in gender, race, age at diagnosis or delay to diagnosis between patients with juvenile myositis, with and without anti-Ro52 autoantibodies (table 2).
Prevalence of anti-Ro52 autoantibodies among myositis-specific autoantibody subgroups
Of those patients positive for anti-Ro52 autoantibodies, 26% had coexisting anti-p155/140 (TIF-1) autoantibodies, 21% had antinuclear matrix protein 2 (NXP2) autoantibodies, 19% had anti-MDA5 autoantibodies, 18% had antiaminoacyl tRNA synthetase autoantibodies, 4% had anti-Mi2 autoantibodies, 4% had anti-3-hydroxy-3-methylglutaryl-CoA reductase autoantibodies and 9% were MSA negative (table 2). Anti Ro52 autoantibodies were significantly increased in the anti-MDA5 and antiaminoacyl tRNA synthetase autoantibody subgroups than in other MSA subgroups (table 1). For instance, anti-Ro52 autoantibodies coexisted in 31% of juvenile IIM sera with anti-MDA5 autoantibodies and 64% of those with antiaminoacyl tRNA synthetase autoantibodies (table 1). Similarly, anti-MDA5 autoantibodies coexisted in 19% of anti-Ro52 autoantibody positive sera and 7% of anti-Ro52 autoantibody negative sera. Antiaminoacyl tRNA synthetase autoantibodies coexisted in 18% of anti-Ro52 autoantibody-positive sera and 2% of anti-Ro52 autoantibody-negative sera (table 2). Less than 15% of those with anti-p155/140 (TIF-1), NXP2, antisignal recognition particle or anti-Mi2 autoantibodies, and only 5% of those without an MSA were anti-Ro52 positive (table 1).
Pulmonary manifestations among patients with anti-Ro52 autoantibodies
After controlling for the presence of MSAs (including antiaminoacyl tRNA synthetase and anti-MDA5 autoantibodies), a multivariate analysis showed anti-Ro52 autoantibodies were highly associated with pulmonary involvement. Overall, patients with anti-Ro52 autoantibodies more often had ILD (36% vs 4%), dyspnoea on exertion (59% vs 25%) and a higher early pulmonary score (mean 0.18 vs 0.08) than those without these autoantibodies (table 3). Within the anti-MDA5 autoantibody positive subgroup, Ro52 reactivity was even more strongly associated with ILD: 70% of those with coexisting anti-Ro52 autoantibodies had ILD compared with only 9% of those who were anti-Ro52 negative (table 4). Similarly, among the antiaminoacyl tRNA synthetase autoantibody subgroup, 100% of anti-Ro52 autoantibody-positive and 40% of anti-Ro52-negative patients had ILD (table 4). Other pulmonary manifestations were also associated with Ro52 reactivity within the anti-MDA5 and antiaminoacyl tRNA synthetase autoantibody subgroups. Specifically, among those patients with anti-MDA5 autoantibodies, patients who also were positive for anti-Ro52 autoantibodies more often had dyspnoea on exertion (90% vs 27%) and higher early pulmonary scores than those who were anti-Ro52 autoantibody negative. Only 1 of 33 patients with ILD in our JIIM cohort had rapidly progressive ILD, and this patient was positive for both anti-MDA5 and anti-Ro52 autoantibodies. In patients with antiaminoacyl tRNA synthetase autoantibodies, anti-Ro52 autoantibody-positive patients had increased frequency of dyspnoea on exertion (89% vs 40%), although this did not reach statistical significance. Patients with coexisting anti-p155/140 (TIF-1) and anti-Ro52 autoantibodies also had an increased frequency of ILD (15% vs 1%) and dyspnoea on exertion (50% vs 16%) compared with anti-p155/140 (TIF-1) autoantibody-positive patients who were anti-Ro52 autoantibody negative (table 4). Of note, in the MSA negative subgroup, none of the 5 anti-Ro52 autoantibody-positive patients had ILD (table 4). The association of anti-Ro52 autoantibodies with ILD was significant within the JDM clinical subgroup: 33% of anti-Ro52 autoantibodies-positive patients with JDM had ILD compared with 1% of anti-Ro52-negative patients with JDM (table 4).
Other clinical manifestations among patients with anti-Ro52 autoantibodies
Independent of MSA status, anti-Ro52 autoantibodies were also associated with Raynaud’s phenomenon (23% vs 14%) (table 3). Furthermore, within the anti-NXP2 subgroup, Ro52 reactivity was associated with more cutaneous involvement: patients with both anti-NXP2 and anti-Ro52 autoantibodies had a higher prevalence of V or Shawl-sign rashes (55% vs 17%) and linear extensor erythema (64% vs 20%) than anti-NXP2 autoantibody-positive patients without anti-Ro52 autoantibodies. Those with both anti-NXP2 and anti-Ro52 autoantibodies also had more frequent gastro-oesophageal regurgitation (55% vs 17%). Within the anti-MDA5 subgroup, however, anti-Ro52 autoantibodies were associated with less frequent linear extensor erythema (11% vs 50%). Patients with anti-Ro52 autoantibodies also had a higher mean early cardiac score, defined by the presence of cardiac symptoms at diagnosis.19 There were no other significant differences in the prevalence of the muscle, lung, joint, cutaneous, gastrointestinal or constitutional manifestations between patients with and without anti-Ro52 autoantibodies in univariate or multivariate analysis or in examining these features in anti-Ro52 autoantibody-positive patients in the presence of another MSA.
Disease severity among patients with anti-Ro52 autoantibodies
Several other differences in outcomes and medications received between patients positive and negative for anti-Ro52 autoantibodies suggested that anti-Ro52 autoantibodies are associated with more severe disease (table 5). The disease course in patients with anti-Ro52 autoantibodies was more often chronic continuous (78% vs 52%) and less often monocyclic (3% vs 25%). Anti-Ro52-positive patients were more often American College of Rheumatology (ACR) functional class 4 (11% vs 4%) at the last clinical evaluation and had a higher mean ACR functional class score at that assessment. Anti-Ro52 autoantibodies were also associated with an increased total number of medications received (mean 4.8 vs 3.8). Anti-Ro52 autoantibody-positive patients more often received intravenous pulse steroids (79% vs 52%). Anti-Ro52 autoantibody-positive patients less often achieved clinical remission (5% vs 27%). Lastly, on univariate analysis, but not multivariable analysis, patients with anti-Ro52 autoantibodies less often experienced a complete clinical response (17% vs 32%) and had more medication treatment trials per year (mean 3.5 vs 2.2).
Those with both anti-NXP2 and anti-Ro52 autoantibodies also more often had a severe (class IV) ACR functional class (27% vs 3%) and more frequent wheelchair use (60% vs 20%) as compared with patients positive for anti-NXP2 who were anti-Ro52 autoantibody negative. There was no other association of coexisting MSAs and anti-Ro52 autoantibodies on clinical outcomes or medications received.
Anti-Ro52 autoantibody titres
Anti-Ro52 autoantibody titres did not significantly differ between JDM, JPM and JCTM groups. Overall, we found that higher anti-Ro52 titres are associated with shorter follow-up time, more treatment trials per year, higher early total symptom score, more total number of medications used, higher total functional class, higher severity at onset, higher early pulmonary score, higher early constitutional symptoms score and higher total functional class in patients with juvenile IIM (all p<0.05; data not shown). However, as the Spearman correlation coefficients were ≤0.2 for each association, the clinical significance of high autoantibody titres is modest.
Here, we used a large cohort of patients with juvenile myositis to study the prevalence and clinical significance of anti-Ro52 autoantibodies in children with IIM. We found anti-Ro52 autoantibodies to be strongly associated with ILD and other pulmonary manifestations in patients with juvenile myositis. We also found that children with anti-Ro52 autoantibodies have more severe disease, underwent more intense treatment regimens and had lower rates of disease remission than those without anti-Ro52 autoantibodies. In children with myositis, anti-Ro52 autoantibodies were associated with antiaminoacyl tRNA synthetase autoantibodies, as previously described in adults.11 We also found that anti-Ro52 autoantibodies were associated with anti-MDA5 autoantibodies in paediatric patients with myositis, which has not been reported previously.
Importantly, our analyses indicate that the presence of anti-Ro52 autoantibodies is strongly associated with ILD, even after adjusting for the presence of MSAs such as anti-MDA5 and antiaminoacyl tRNA synthetase autoantibodies. Indeed, the association of Ro52 reactivity with ILD is not limited to the anti-MDA5 and antiaminoacyl tRNA synthetase autoantibody subgroups but extends to other MSA subgroups that are not classically associated with ILD, such as children with anti-p155/140 (TIF-1) autoantibodies. However, none of the five patients who are anti-Ro52 autoantibody positive and MSA negative had ILD. Current practice encourages screening patients with juvenile myositis for MSAs such as anti-MDA5 and antiaminoacyl tRNA synthetase autoantibodies, as these autoantibodies confer risk for developing ILD, and their presence is a determinant of clinical management and patient prognosis. In light of the current findings demonstrating that anti-Ro52 autoantibodies are an independent predictor of ILD, screening patients with juvenile myositis for these autoantibodies may also be prudent.
In adult patients with IIM, anti-Ro52 autoantibodies have been associated with poorer response to immunosuppressive drugs and decreased survival.13 15 Similarly, in our juvenile cohort, anti-Ro52 autoantibodies are associated with more severe disease and poorer outcomes. Of note, the presence of anti-Ro52 autoantibodies was associated with a higher early cardiac score which is a measure of patient-reported cardiac symptoms including palpitations, chest pain and syncope. However, among the nine anti-Ro52-positive patients with one or more of these symptoms, only three had EKG changes or echocardiogram abnormalities. As the severity of other clinical manifestations, including muscle, joint, skin, gastrointestinal and systemic features were not associated with Ro52 reactivity, it seems likely that disease severity seen in the anti-Ro52-positive patients is a consequence of pulmonary disease. Additional studies are required to clarify this point. Nonetheless, our findings highlight the potential use of anti-Ro52 autoantibodies as a predictor of disease severity and poor prognosis in juvenile myositis, which underscores the potential use of screening patients with juvenile IIM for anti-Ro52 autoantibodies.
Of particular significance is the novel association of anti-Ro52 autoantibodies and anti-MDA5 autoantibodies in our JIIM cohort. In adult patients with IIM, anti-Ro52 autoantibodies often co-occur with anti-Jo1 autoantibodies, and in adult anti-Jo1-positive patients, Ro52 reactivity is associated with more severe ILD. A small case series reported coexisting anti-Ro52 autoantibodies in 6 of 13 anti-MDA5 autoantibody-positive patients, 5 of whom had rapidly progressive ILD.27 Interestingly, only 1 of 33 patients with ILD in our JIIM cohort had rapidly progressive ILD, and this patient was positive for both anti-MDA5 and anti-Ro52 autoantibodies.
Although we have now established an association between antiaminoacyl tRNA synthetase and anti-Ro52 autoantibodies in adults and children, it remains unclear why these autoantibodies co-occur. It has been proposed that local autoantibody production induced by type I interferon (IFN)28 could be a driving force behind the production of both anti-Jo1 and anti-Ro52 autoantibodies, given the increase in B-cell activating factor receptors in the sera of patients with IIM with these autoantibodies.29 In the current study of juvenile IIM, we now also demonstrate an association between anti-MDA5 and anti-Ro52 autoantibodies. Interestingly, both MDA5 and Ro52 are cytosolic, IFN-induced proteins; perhaps concurrent overexpression of these proteins in patients with juvenile IIM leads to the development of autoimmunity against both. However, we do not have adequate type I IFN measurements to further examine this hypothesis.
This current study has several limitations. First, this cohort of patients with juvenile myositis had some data collected retrospectively, resulting in some missing data and were collected over more than 20 years, with potential chronology bias. However, we adjusted the variables of this study for the year of diagnosis and tested the distribution of missing values across groups and did not find evidence of a significant bias. Second, although imaging studies were available to confirm the diagnosis of ILD in more than 90% of patients who had ILD, pulmonary function testing data were not available for many of the patients, as a number of the children were of young age when such testing is unreliable in children. Thus, we were not able to study whether patients with ILD and anti-Ro52 autoantibodies had more severe pulmonary dysfunction than those without these autoantibodies. Also, we cannot confirm the absence of ILD as many of the children without clinical suspicion of ILD did not have imaging and/or pulmonary function testing. This, however, is a limitation of standard clinical care in paediatric patients who have challenges to undergo such testing.
Overall, this study shows that anti-Ro52 autoantibodies are present in 14% of patients with juvenile myositis and are strongly associated with ILD, more severe illness and poorer outcomes, even when correcting for the coexistence of MSAs. In patients with juvenile myositis, anti-Ro52 autoantibodies are associated with the presence of antisynthetase autoantibodies, as previously reported in patients with adult myositis, and with anti-MDA5 autoantibodies, and the coexistence of these MSAs increases the likelihood of ILD and poor outcome. The current standard of care in patients with juvenile myositis who have reactivity to MSAs associated with pulmonary manifestations (such as anti-MDA5 and antiaminoacyl tRNA synthetase autoantibodies) is to have a high index of suspicion for the development of ILD and modify management accordingly. Our data suggest that testing for anti-Ro52 autoantibodies may also have a role in disease monitoring, management and patient prognosis in patients with juvenile myositis.
SS and IP-F are joint first authors.
LGR and ALM are joint senior authors.
SS and IP-F contributed equally.
LGR and ALM contributed equally.
Handling editor Josef S Smolen
Presented at This work was presented in abstract form at ACR 2018 and GCOM 2019.
Collaborators *Members of the Childhood Myositis HeterogeneityCollaborative Study Group who contributed to this project:
Bita Arabshahi, Lilliana Barillas-Arias, Mara Becker, AprilBingham, Ruy Carrasco, Victoria Cartwright, Rodolfo Curiel, Marietta M.DeGuzman, Barbara Anne Eberhard, Barbara S. Edelheit, Terri Finkel, Stephen W.George, Ellen A. Goldmuntz, William Hannan, Michael Henrickson, Adam M. Huber,Anna Jansen, James Jarvis, Lawrence Jung, Ildy M. Katona, Steven J. Klein, WPatrick Knibbe, Bianca A. Lang, Carol B. Lindsley, Gulnara Mamyrova, LindaMyers, Stephen R. Mitchell, Kabita Nanda, Terrance P. O’Hanlon, Murray H.Passo, Maria D. Perez, Donald A. Person, Linda I. Ray, Rafael F. Rivas-Chacon,Tova Ronis, Deborah Rothman, Adam Schiffenbauer, Bracha Shaham, David Sherry,Abigail Smukler, Matthew L. Stoll, Sangeeta H. Sule, Scott A. Vogelgesang, Rita Volochayev, Jennifer C. Wargula, Pamela Weiss.
Contributors SS, IPF, LGR and ALM conceived the work. LGR and INT acquired, analysed and interpreted the data. SS, IPF and TK analysed and interpreted the data. FWM interpreted data. SS, IPF and ALM drafted the work and revised it critically for important intellectual content. TK, IPF, FWM and LGR revised the work for critically important intellectual content. All authors approved the final version of the manuscript. All members of the Childhood Myositis Heterogeneity Collaborative Study Group contributed by (1) providing substantial contributions to the acquisition of data, (2) revising the work critically for important intellectual content and (3) providing final approval of the version published.
Funding This research was supported by the Intramural Research Programs of the National Institute of Arthritis and Musculoskeletal and Skin Diseases (ZIA AR041203) and the National Institute of Environmental Health Sciences (Z01 ES101074 and Z01 ES101081) of the National Institutes of Health.
Competing interests None declared.
Patient and public involvement statement Patients were involved in the research from the time they provided consent to join this natural history study. The research questions were not explicitly developed nor informed by their priorities, experience and preferences. The patients/public were not involved in the design of this study. Patients were not involved in the recruitment to and conduct of the study. Patients were not asked to assess the burden of the intervention and time required to participate in the research. Patients have not and will not be involved in choosing the methods and agreeing plans for dissemination of the study results to participants and wider relevant communities.
Patient consent for publication Not required.
Ethics approval All subjects were enrolled in natural history study approved by the National Institutes of Health Institutional Institutional Review Board, and all patients provided informed consent.
Provenance and peer review Not commissioned; externally peer reviewed.
Data sharing statement All data relevant to the study are included in the article or uploaded as supplementary information.
If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.