Objectives Patients with connective tissue diseases can develop interstitial lung disease (ILD), leading to a progressive fibrosing ILD (PF-ILD) phenotype in some cases. We aimed to investigate the occurrence of PF-ILD in idiopathic inflammatory myopathies (IIMs), and factors potentially predicting this phenotype. Secondary aims were to assess the radiological pattern and factors associated with IIMs-ILD.
Methods Patients with IIMs from our multicentric prospective cohort were retrospectively evaluated. Data were recorded at IIMs and ILD diagnosis, and during follow-up. Patients with ILD were classified according to the predominant high-resolution CT (HRCT) pattern: non-specific interstitial pneumonia (NSIP), usual interstitial pneumonia (UIP) and organising pneumonia (OP). PF-ILD was defined according to the 2022 American Thoracic Society (ATS), European Respiratory Society (ERS), Japanese Respiratory Society (JRS) and Latin American Thoracic Society (ALAT) guidelines. Univariate and multivariate analyses were performed to identify factors associated to ILD and to PF-ILD.
Results Of 253 patients with IIMs, 125 (49%) had ILD: 99 (78%) at IIMs diagnosis and 26 (22%) during follow-up (21/26 within 5 years). Multivariate analysis identified anti-Jo-1, anti-MDA5, anti-Ro52, high score on manual muscle test, mechanic’s hands and Raynaud’s phenomenon as independently associated with ILD. The predominant HRCT pattern was NSIP (50% of patients), followed by UIP (28%) and OP (22%). At 1-year follow-up, PF-ILD occurred in 18% of IIMs-ILD. PF-ILD was predicted by anti-MDA5, heliotropic rash, xerostomia and xerophthalmia at univariate but not at multivariate analysis.
Conclusion Patients with IIM should be carefully screened for ILD at IIMs diagnosis and yearly during follow-up. All patients with IIMs-ILD should be carefully monitored to capture ILD progression since a consistent proportion of them are expected to develop PF-ILD.
- pulmonary fibrosis
- autoimmune diseases
Data availability statement
All data relevant to the study are included in the article or uploaded as supplementary information.
This is an open access article distributed in accordance with the Creative Commons Attribution Non Commercial (CC BY-NC 4.0) license, which permits others to distribute, remix, adapt, build upon this work non-commercially, and license their derivative works on different terms, provided the original work is properly cited, appropriate credit is given, any changes made indicated, and the use is non-commercial. See: http://creativecommons.org/licenses/by-nc/4.0/.
Statistics from Altmetric.com
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.
WHAT IS ALREADY KNOWN ON THIS TOPIC
Patients with connective tissue diseases (CTDs) may develop progressive fibrosing interstitial lung disease (PF-ILD). Among patients with idiopathic inflammatory myopathies (IIMs), ILD is detectable in up to 50% of cases. However, unlike in other CTDs (eg, systemic sclerosis, rheumatoid arthritis), the occurrence of PF-ILD has been poorly investigated in patients with IIMs.
WHAT THIS STUDY ADDS
This study provides real-life data on the occurrence of PF-ILD in a large multicentre cohort of patients with IIMs. The reported association between different radiological patterns and clinical/serological features may allow a better stratification of patients with IIMs in different clinical phenotypes.
HOW THIS STUDY MIGHT AFFECT RESEARCH, PRACTICE OR POLICY
Our findings suggest that clinicians ought to carefully monitor patients with IIMs-ILD for ILD progression. The considerable proportion of patients with IIMs who develop PF-ILD highlights the importance of future research to better define the treatment of these patients, especially the potential role of antifibrotics.
Interstitial lung disease (ILD) can be observed in up to 50% of patients with idiopathic inflammatory myopathies (IIMs).1 Although this rate is similar to that observed in systemic sclerosis (SSc)—wherein experts agree to perform high-resolution CT (HRCT) of the lung at SSc diagnosis and pulmonary function tests (PFTs) at least yearly—to date there is no consensual screening strategy for ILD in IIMs.2 3 Nevertheless, early and timely detection of IIMs-ILD is of the utmost importance, as ILD may require a close follow-up and tailored treatment approach.
The ILD radiological pattern may differ among patients with IIMs. The most common HRCT feature is the non-specific interstitial pneumonia (NSIP), followed by organising pneumonia (OP), and usual interstitial pneumonia (UIP).4 Notably, recent studies have highlighted that all patients with ILD, not only those with idiopathic pulmonary fibrosis (IPF), can potentially develop a progressive fibrosing ILD (PF-ILD).5–7 PF-ILD is a self-sustaining process which is identified based on radiological, clinical and functional decline features and may occur at any time during follow-up (ie, in prevalent ILD cases).8 Data from nine centres across the UK showed that patients with PF-ILD have a significantly higher mortality compared with those with non-PF-ILD, and similar to patients with IPF who are characterised by particularly poor prognosis.9 Early identification of predictors of PF-ILD may help improve patient management (eg, adopting a more aggressive therapeutic strategy in earlier stages of IIMs-ILD) and prognosis. Moreover, there has been growing interest for the study of PF-ILD in connective tissue diseases (CTDs) after the INBUILD trial results showed the efficacy of nintedanib—a tyrosine-kinase inhibitor—in reducing the rate of forced vital capacity (FVC) decline versus placebo in patients with PF-ILD related to rheumatic systemic diseases.10 Although only two patients with IIMs were included in this randomized clinical trial (RCT), recent evidence11 12 suggests that patients with IIMs may develop PF-ILD not unlike patients with other CTDs (eg, SSc or rheumatoid arthritis). However, its occurrence has been poorly investigated so far in IIMs-ILD.
The primary aim of our study was to assess the prevalence and predictors of PF-ILD in a large, multicentre cohort of patients with IIM-ILD. Secondary aims were to explore clinical and serological features associated with ILD occurrence in IIMs, and the association between different chest HRCT patterns and clinical-functional data of patients with IIMs-ILD.
We conducted a retrospective cohort study comprising consecutive patients affected by IIMs—according to Bohan and Peter’s criteria, the European Neuromuscolar Center (ENMC) criteria or 2017 EULAR/ACR classification criteria13–15 followed-up in four third-level referral centres (Padova, Firenze, Udine, and Paris-Cochin Rhumatologie)15 between 2002 and 2020. Enrolled patients had to have at least one visit with PFTs, after the diagnosis of IIMs. Exclusion criteria were: juvenile idiopathic inflammatory myopathies (ie, IIM diagnosis at <18 years), incomplete and/or unavailable clinical and/or serological baseline data, myositis in other CTDs, unavailable follow-up visits.
The following demographic, clinical and laboratory variables were collected at diagnosis and every year during the follow-up: constitutional symptoms (fever, asthenia and weight loss); articular involvement (arthritis and/or inflammatory arthralgia); muscular involvement (myalgia, muscular weakness, dyspnoea and dysphagia); skin manifestations (eg, Gottron’s papules and signs, mechanic’s hands) and other manifestations, including Raynaud’s phenomenon (RP), xerostomia and xerophthalmia. All patients underwent manual muscle test 8 (MMT-8), PFTs including total lung capacity, FVC and diffusing capacity for carbon monoxide (DLCO), serum levels of creatine kinase (CK), lactate dehydrogenase and myoglobin. PFTs were recorded at diagnosis and yearly during follow-up. All patients were screened for ILD by medical evaluation (eg, bibasilar crackles), PFTs and/or X-rays. When ILD was suspected due to abnormal signs during clinical examination or imaging findings, HRCT was performed.
Further diagnostic investigations—that is, muscle MRI and/or electromyography and/or muscle biopsy—were performed according to the physician’s judgement.
Autoantibodies were tested in all patients at diagnosis: serum anti-nuclear antibodies were analysed by immunofluorescence assay on HEp-2 cells, anti-extractable nuclear antigen antibodies by ELISA and immunoblot, myositis specific autoantibodies (MSA) and myosistis associated autoantibodies (MAA) by commercial line blots (Euroline Myositis Profile, Euroimmun, Lubeck, Germany) including recombinant human proteins for Mi-2 alpha, transcription intermediary factor 1-gamma, small ubiquitin-like modifier-1 activating enzyme, Ku, PM-Scl75/100, MDA-5, signal recognition particle, Jo-1, PL-7, PL-12, EJ, and OJ.16
Comparison between patients with and without ILD and radiological evaluation in patients with IIMs-ILD
The diagnosis of ILD was based on HRCT. Clinical and serological data at IIM diagnosis were compared between IIM-non-ILD and IIM-ILD population. As some patients developed lung involvement during follow-up and not at the time of IIM diagnosis, we also compared clinical and serological data between IIM non-ILD population, patients with ILD at IIM diagnosis and those who developed ILD during follow-up.
Patients with fully available imaging for review were evaluated by four experienced thoracic radiologists (CG, LC, FG, AF) and classified into three subgroups according to the predominant radiological pattern: NSIP as group 1; UIP (definite and probable according to the latest guidelines17) as group 2; and OP, as group 3. The same IIM-ILD population was also grouped based on the three main IIM clinical phenotypes: polymyositis (PM), dermatomyositis (DM) and antisynthetase syndrome (ASyS). The three radiological subgroups (NISP, UIP and OP) were compared considering clinical, functional and serological data collected at the time of IIMs-ILD diagnosis, and differences in predominant radiological features among the main IIMs clinical phenotype were analysed.
Identification of patients with progressive fibrosing IIM-ILD
Patients with IIM-ILD were further subgrouped in progressors (PF-ILD) and not-progressors (non-PF-ILD) according to the latest definition of PF-ILD provided by the 2022 ATS/ERS/JRS/ALAT guidelines.17 More specifically, patients displaying at least two of the following three characteristics in the first year after IIMs-ILD diagnosis were classified as PF-ILD: (1) worsening of respiratory symptoms; (2) functional deterioration (defined by an absolute decline in FVC of more than 5% and in DLCO (adjusted for haemoglobin) of more than 10% within 1 year of follow-up); and (3) radiological worsening (identified as the increased extent of traction bronchiectasis, reticular abnormality, honeycombing, a new ground-glass opacity and increased lobar volume loss). Among patients with IIM-ILD, only those with functional and/or radiological data at ILD diagnosis and after 1-year follow-up were included in the evaluation of progression.
Clinical, functional and serological features at the time of IIM-ILD diagnosis were also compared between patients with PF-ILD and non-PF-ILD. Finally, we evaluated the immunosuppressive pharmacological treatment during the first year after IIMs-ILD diagnosis to explore potential differences between the two groups.
This was a retrospective study on anonymised patient data collected from medical records. The study protocol was in accordance with current national regulation on retrospective observational studies.
Continuous variables were expressed as mean±SD or median (25th–75th IQR) and categorical variables as frequencies and percentages. Comparisons between groups were performed using the Mann-Whitney U test for continuous variables and the χ2 test or Fisher’s exact probability test for categorical data, where appropriate. Variables which were found to be different (p<0.1) between patients with and without ILD, and between PF-ILD and non-PF-ILD at univariable analysis were included in a multivariable logistic regression model, adjusted for age and sex. Two-sided p<0.05 was considered statistically significant. Kaplan-Meier survival analysis was performed to assess ILD-free survival in the subgroup of patients who developed ILD during follow-up. The statistical analysis was performed using the SPSS statistical package, V.22.0.
Demographic and clinical characteristics of the study population
Two hundred and fifty-three patients affected with IIMs were included in the study: 183 (72%) women, median age at diagnosis 55 (IQR 46–66) years. Among the overall population, a total of 125/253 (49.4%) patients exhibit ILD (figure 1): 99/125 (79%) had ILD at IIMs diagnosis and 26/125 (21%) developed ILD during follow-up. Baseline demographic characteristics of the study population are reported in table 1. The median follow-up (range) was 6 (3–10) years and was comparable in patients with and without ILD (p=0.44). Considering IIMs clinical phenotype (table 1), most patients were diagnosed as DM (96/253, 38%), followed by ASyS (77/253, 30%) and PM (67/253, 27%). The remaining IIM clinical phenotypes necrotizing autoimmune myopaty (NAM) and inclusion body myositis (IBM) were found exclusively in patients without ILD. Nine patients (3.5%), all with positive PM-Scl antibodies, had an overlap syndrome with SSc.
Evaluation of factors associated with ILD
Serological and clinical features, and treatment in patients with and without ILD
Among patients who exhibited ILD during the follow-up, 21 out of 26 (81%) developed ILD within the first 5 years from IIMs diagnosis (online supplemental figure 1).
Compared with patients with ILD at IIM diagnosis, those who developed ILD during follow-up showed similar baseline demographic, serological and clinical variables (online supplemental table 1), with a similar follow-up period in the two groups (p=0.19).
In patients developing ILD (either at IIM diagnosis or during follow-up) anti-Jo1 (p<0.0001), anti-SSA/Ro (p<0.0001), anti-Ro52 (p=0.0002) and anti-MDA5 (p=0.01) were more frequently found in patients with IIMs with ILD than in those without (table 1). By contrast, anti-Mi 2 was more frequently detected in patients without ILD (p=0.005). In patients without ILD, CK values were higher (p=0.001), MMT-8 score was lower (p<0.0001) and histological and/or MRI features of myositis were more frequent (p<0.0001) than in those with ILD. Patients with ILD more frequently had fever (p=0.004), arthritis (p=0.0003), RP (p=0.02) and mechanic’s hands (p<0.0001). (table 1). Glucocorticoids were taken by the same proportion of patients in the two groups, although at lower dosage in patients with ILD compared with those without. Patients without ILD were more frequently treated with methotrexate and high-dose intravenous immunoglobulins, whereas those with IIMs-ILD with mycophenolate mofetil, azathioprine, cyclophosphamide, or rituximab.
At multivariable analysis, high scores of MMT-8, anti-Jo1, anti-MDA5 and anti-Ro52, RP and mechanic’s hands were independently associated with ILD. However, heliotrope rash was negatively associated with the occurrence of ILD (table 2).
Subgroup analysis evaluating radiological patterns and their association with clinical and functional variables
HRCT scan at IIMs-ILD diagnosis was available for fully radiological assessment in 78 out of 125 patients (62%). Patients with IIMs-ILD without complete HRCT images for review and those with available HRCT showed no substantial differences in terms of clinical and serological variables, except for a longer follow-up in the former group (data not shown). We performed a subgroup analysis according to the predominant radiological pattern: NSIP (n=39, 50%), UIP (n=22, 28%) and OP (n=17, 22%). Demographic, clinical, serological and functional data in the three subgroups at IIM-ILD diagnosis are summarised in online supplemental table 2. Notably, lung involvement was always detected before the diagnosis of IIM in the OP group versus NSIP group (p=0.01), and UIP group (p=0.04). Moreover, the OP group displayed significantly higher levels of baseline CK compared with UIP (p=0.01), but not NSIP (p=ns), and a lower MMT-8 score compared with both NSIP (p=0.008) and UIP (p=0.003).
When IIM-ILD population was subgrouped based on the three more frequent IIM clinical phenotypes (PM, DM and ASyS) no differences were found in clinical and radiological presentation among groups (online supplemental table 3).
Assessment of ILD progression
A functional and/or radiological 1-year follow-up was available in 78/125 patients (62%).
According to the 2022 guidelines definition,18 14/78 (18%) patients were subgrouped into PF-ILD and 64/78 (82%) patients in not PF-ILD. An example of radiological progression on the CT scan of one patient with IIM-ILD enrolled in our study is reported in online supplemental figure 2.
Demographic, clinical, serological and functional data are summarised in table 3. The two subgroups did not differ with regard to age, sex and IIM clinical phenotype and three patients (4%) died during the follow-up. The PF-ILDs presented a higher prevalence of positive anti-MDA5 antibodies compared with non-PF-ILDs (29% vs 6%, p=0.03), with no other significant differences in the autoantibody profile. The PF-ILDs presented a higher prevalence of xerophthalmia (29% vs 5%, p=0.02) and xerostomia (36% vs 6%, p=0.007) at ILD diagnosis compared with non-PF-ILDs, as well as a higher prevalence of heliotrope rash (29% vs 5%, p=0.02). Regarding pharmacological treatment, the two groups did not differ during the first year after ILD diagnosis (online supplemental table 4). Glucocorticoids were taken by a high percentage of patients in both groups (93 vs 86%, p=0.49) with similar initial dosage, that is, 25 mg/day of prednisone (or equivalent). Mycophenolate mofetil was the immunosuppressant more frequently used in both groups (36% vs. 32%, p=0.81).
Univariate and multivariate evaluation of ILD progression risk
At univariate analysis, anti-MDA5 (OR: 6.10; 95% CI 1.30 to 28.4, p=0.02), heliotrope rash (8.00; 95% CI 1.55 to 41.23, p=0.01), xerostomia, xerophthalmia at ILD diagnosis (8.19; 95% CI 1.84 to 36.36, p=0.006; and 8.00; 95% CI 1.55 to 41.2, p=0.01, respectively), were predictive of progression (table 4). However, none of the significant features at the univariate analysis were confirmed as independent predictors of PF-ILD at multivariate analysis.
In our large cohort of patients with IIMs-ILD, we found that about 20% developed PF-ILD within 1 year, according to the 2022 guidelines definition. Moreover, we detected some key clinical and serological features strongly associated with ILD diagnosis and HRCT pattern in IIMs, highlighting that fibrotic changes may occur early in IIMs-ILD course.
Among patients with IIMs, we found that those with amyopathic or hypomyopathic features carry a particularly high risk of developing ILD, highlighting that muscular inflammation and pulmonary involvement often have an independent course in IIMs, and supporting a systematic ILD screening in patients with non-severe muscular impairment. The presence of anti-Ro52 antibodies was independently associated with ILD in our multivariate analysis, thus confirming its strong association with lung fibrosis, independently of concomitant anti-synthetase antibodies positivity.19 Although MSAs and MMAs autoantibodies are useful in identifying IIMs subsets,20 21 their detection is limited to referral centres. Our patients with mechanic’s hands had an estimated 9-fold increased risk of developing ILD versus those without, an association that was recently reported in Asian patients.22 23 Although mechanic’s hands and RP are hallmarks of the anti-synthetase syndrome,18 these are manifestations that do not occur in all patients with ASyS and may be observed in other IIMs phenotypes. Mechanic’s hands are easily detectable and may prompt a thorough ILD screening, hence the importance of clinical features in ILD risk stratification of patients with IIMs.24 Over 80% of our patients with IIMs-ILD showed signs of ILD at IIMs diagnosis, or developed ILD during the first 5 years of follow-up. Overall, our findings suggest that patients with IIM at risk for ILD should be carefully screened via PFTs and HRCT at baseline, and yearly at least for the first 5 years from IIMs diagnosis.
Half of our IIMs-ILD cohort had an NSIP pattern on HRCT. Patients with OP require particular attention due to their association with higher CK and lower MMT-8 score versus other subgroups, often a precursor of IIM diagnosis. The association between consolidation in the lung and active muscular disease suggest a more inflammatory phenotype of IIM, which may require higher doses of glucocorticoids. In addition, the detection of lung consolidation on HRCT should prompt myositis antibodies testing, even in the absence of concomitant signs/symptoms indicative of IIMs. It also bears noting that over 20% of our patients had a UIP pattern at diagnosis, supporting the emerging concept that fibrosis may be an early event in CTD-ILD.
Almost 20% of our IIM-ILD population developed PF-ILD. In the literature, very little data is available on the risk of developing PF-ILD in patients with CTDs other than SSc. Takei et al11 reported that 27% of patients with CTD in their cohort developed a progressive form of ILD within 24 months, without accounting for each CTD subtype. The Canadian ILD Registry25 recently published specific data on patients with IIMs-ILD (without describing any features of the IIMs population) showing that over 40% developed PF-ILD, according to the INBUILD criteria which evaluate the occurrence of PF-ILD within 24 months. On the other hand, the 2022 ATS criteria consider a tighter 1-year time frame and are more accurate in the definition of radiographic progression on HRCT. Even in our study, PF-ILD occurred at 1 year in a considerable percentage of patients, suggesting that patients with IIMs-ILD should be more carefully evaluated for ILD progression, not unlike patients with SSc and rheumatoid arthritis. Given the EMA approval of the antifibrotic nintedanib for the treatment of PF-ILD (July 2020), there is a need to refine PF-ILD estimates in specific real-world CTD subgroups.9 Although IIMs-ILD features are generally thought to be less fibrotic than SSc-ILD, our findings highlighted a considerable proportion of patients with early fibrotic changes; in this regard, two small retrospective studies evaluating the use of nintedanib and pirfenidone in IIMs-ILD provided some positive signs of effectiveness, in addition to the results from the INBUILD trial.26 27
We found no differences in clinical and serological characteristics of progressors versus non-progressors. Unlike previously reported data on SSc-ILD,28 29 male patients with IIMs and those with lower FVC and DLCO values at diagnosis did not appear to have an increased risk of progression. Similarly, there was no difference between progressors versus non-progressors in HRCT pattern, indicating that the PF-ILD phenotype may occur not only in patients with fibrotic UIP-like pattern but also in those with ground glass and consolidation. Although anti-MDA5 antibodies, heliotrope rash, xerostomia and xerophthalmia were associated with PF-ILD at univariate analysis, none were confirmed as independent predictors. Given the conflicting reports on the prognostic value of positive anti-Ro52 in CTD-ILD,30 31 it should be noted that this autoantibody specificity was not associated with PF-ILD in our cohort, as found by Vojinovic et al.30 Taken together our data suggest monitoring all patients with IIMs-ILD with PFTs, and HRCT reassessment at 1 year, independently of radiographic phenotype or clinical/serological IIMs characteristics.
The main strength of our study lies in the large number of patients included, given that IIMs are recognised as rare diseases. To the best of our knowledge this is one of the first large multicentre observational studies specifically evaluating the PF-ILD occurrence in patients affected with IIMs, in a well-characterised IIMs cohort with a long-standing follow-up. As limitations, although centres involved in the project are third-level referral centres for CTDs, we cannot rule out slight differences in patient management and follow-up. Second, due to the retrospective design of the study, a limited percentage of incomplete data was tolerated, and we cannot rule out that the lack of a systematic evaluation of respiratory muscle involvement may have partially affected the evaluation of PF-ILD occurrence. Finally, the small sample size of patients with PF-ILD may have limited the identification of independent predictors; in this regard it should be noted that about 50% of our patients with positive MDA5 developed PF-ILD, suggesting that these patients warrant particular attention, nonetheless.
We found that almost 20% of patients with IIMs-ILD in our cohort developed a PF-ILD phenotype, and we were unable to identify independent predictors. Therefore, it seems reasonable that all patients with IIMs-ILD should be screened via PFTs and repeat HRCT to monitor ILD progression. In addition, we confirmed the close association between specific autoantibodies and some clinical manifestations, and ILD in IIMs. Whereas glucocorticoids and immunosuppressants are the cornerstone of management of IIMs-ILD, the optimal treatment for patients with PF-ILD IIMs is yet to be determined. Further studies are needed to ascertain whether these patients would substantially benefit from antifibrotic treatment as it pertains to morbidity, mortality, and quality of life.
Data availability statement
All data relevant to the study are included in the article or uploaded as supplementary information.
Patient consent for publication
The study protocol was in compliance with the ethical guidelines of the 1975 Declaration of Helsinki, and it was approved by the Ethics committee—University Hospital of Padova—number 5505/AO/22. Informed consent was waived due to the retrospective nature of the study.
EZ and EC are joint first authors.
EB and LIa are joint senior authors.
EZ and EC contributed equally.
EB and LIa contributed equally.
Contributors EZ and EC contributed equally to this paper as first authors. EB and LIa contributed equally to this paper as senior authors. EZ, EC, EB and LIa—study conception and design. EZ, EC—drafted manuscript. EZ, EC, GC, CG, AF, EDZ, MG, LIe, ET, DM, LC, GE, FG, SB, AM, BM, AG, JA, LQ, YA, AD, PS, EB and LIa—collected clinical data, contributed to analysis and interpretation of the results and critically reviewed the manuscript. All authors approved the submitted manuscript. EZ, EC, EB and LIa are guarantors for this manuscript and attest that all listed authors meet authorship criteria.
Funding The authors have not declared a specific grant for this research from any funding agency in the public, commercial or not-for-profit sectors.
Competing interests EZ and EB report personal and consulting fees from Boehringer Ingelheim. JA and YA report consulting fees from Boehringer Ingelheim. PS reports institutional grants and personal fees from PPM Services, Boehringer Ingelheim and Roche; consulting fees from Boehringer Ingelheim; non-financial support from PPM Services and Boehringer Ingelheim, and personal fees from Galapagos, Chiesi, Novartis, Lupin, Pieris and Santhera Pharmaceuticals.
Provenance and peer review Not commissioned; externally peer reviewed.
Supplemental material This content has been supplied by the author(s). It has not been vetted by BMJ Publishing Group Limited (BMJ) and may not have been peer-reviewed. Any opinions or recommendations discussed are solely those of the author(s) and are not endorsed by BMJ. BMJ disclaims all liability and responsibility arising from any reliance placed on the content. Where the content includes any translated material, BMJ does not warrant the accuracy and reliability of the translations (including but not limited to local regulations, clinical guidelines, terminology, drug names and drug dosages), and is not responsible for any error and/or omissions arising from translation and adaptation or otherwise.