Systemic sclerosis
All patients were recruited consecutively by screening the databases of participating centres in 2015. Serum samples had been collected between 2004 and 2014. SOMAscan assay The relative expression levels of 1129 serum proteins were assessed in the discovery cohort using the SOMAscan platform (SomaLogic, Boulder, Colorado, USA), a highly multiplexed aptamer microarray, as previously described. 17 18 Briefly, samples were deposited in a 96-well plate and incubated with a mixture of the 1129 SOMAmer reagents. Two sequential bead-based immobilisation and washing steps eliminated unbound proteins, non-specifically bound proteins and unbound SOMAmer reagents. The remaining specifically bound SOMAmer reagents were isolated, and each reagent was quantified simultaneously on a custom Agilent hybridisation array. The amount of each SOMAmer measured was quantitatively proportional to the protein concentration in the original sample. Results were expressed in relative fluorescence units. ELISA assays To confirm result reproducibility and cross-platform consistency, serum concentrations of candidate proteins were measured in duplicate in the validation cohort using commercial ELISA assays (Quantikine ELISA Human Chemerin Immunoassay, RD systems; SET Nuclear Oncogene (SET) ELISA Kit (Human), cat. #ABIN1152447, antibodies-online.com) at appropriate dilu- tions (1:100 for chemerin; 1:1 for SET), according to the manu- facturer’s protocol. Statistical analyses For the description of study populations, quantitative vari- ables were expressed as means±SD or medians (IQR) for non- normal distributions, and categorical variables were expressed as numbers (percentage). Normality of distributions was assessed using histograms and tested using the Shapiro-Wilk test. In a first step, data from the discovery cohort were log2 transformed and a differential analysis was performed between cases and controls with the Bioconductor R (V.3.6.1) package limma (V.3.38.3). 19 Limma uses an empirical Bayesian approach to estimate variances in moderated t-tests, which has proven to improve results on standard t-tests, especially when the number of replicates is small. Raw p values were adjusted with the Benja- mini Hochberg method 20 and proteins with adjusted p values <0.05 were considered differentially expressed. For cases, the Spearman correlations between PVR and expressions of these differentially expressed proteins were tested and were consid- ered significant if raw p values were <0.05. In a second step, differential expression of candidate biomarkers identified in the discovery cohort was assessed between cases and controls from the validation cohort using Mann-Whitney test; and correlations with PVR in cases were analysed using Spearman test. In a third step, serum levels of the candidate biomarkers were compared between the five patient groups using the Kruskal- Wallis test. Pairwise comparison was conducted by post-hoc Dunn test followed by Bonferroni correction. As protein production could always be quantified in all data- sets, no imputation for missing data was performed. A heatmap was drawn for differentially expressed proteins with the R package pheatmap (V.1.0.12) after standardisation of the expres- sion data. Other figures were created using GraphPad Prism V.9.1.0 (GraphPad Software, California, USA).
biomarkers for the accurate and non-invasive prediction of PAH in SSc patients. Several works have previously tried to address this issue. 6–12 Most of them focused on identifying surrogates for haemody- namic diagnostic parameters, with biomarkers that could accu- rately discriminate SSc patients with and without PAH. 7–10 Few studies, however, have tried to find surrogate markers for severity parameters, such as PVR. 6 11 This is a matter of importance, since PVR reflects the ongoing vascular remodelling underlying the disease progression and can be used to guide treatment initiation and assess therapeutic efficacy. 11 13 This work was an exploratory study that aimed to identify proteins that correlate with haemodynamic severity in SSc-PAH patients and to serve as a base on which future hypotheses could be tested. To achieve this, we used a wide-scale approach by investigating alterations in the patients’ serum proteome with a high-throughput assay. In a second step, we assessed whether the identified candidate biomarkers were involved in disease pathogenesis. In a first step, to identify candidate biomarkers, a discovery cohort was recruited from Boston University Arthritis Centre. Patients were included if they met the following criteria: (1) diagnosis of SSc according to the 2013 American College of Rheumatology (ACR)/European Alliance of Associations for Rheumatology (EULAR) criteria; 14 (2) limited cutaneous (lc) subset of SSc according to LeRoy’s criteria; 15 (3) absence of extensive interstitial lung disease (ILD) according to Goh’s criteria; and 16 (4) absence of PAH-specific therapy. In that first step, patients with diffuse cutaneous (dc)SSc and/or extensive ILD were excluded to avoid our biomarker screening to be inter- fered with by any active organ involvement other than PAH, and our results to be mediated by any pathogenic process other than pulmonary microangiopathy. In addition, extensive ILD was also considered an exclusion criterion to avoid including patients with secondary (group 3) pulmonary hypertension that probably did not have significant pulmonary microangiopathy. Patients fulfilling these inclusion criteria were classified as cases if they had an RHC-proven diagnosis of PAH according to the 2015 European Society of Cardiology (ESC)/European Respira- tory Society (ERS) guidelines 4 ; and as controls if they had no evidence of PAH. PAH screening modalities, assessment of PAH probability and referral for RHC followed ESC/ERS guidelines in effect at the time of sample collection. For cases, all serum samples were collected on the same day as RHC. In a second step, in order to confirm the validity of the iden- tified candidate biomarkers, an independent validation cohort was recruited from Boston University, Grenoble University Hospital and Lille national reference centre for SSc. Patients were included if they met the same criteria as the discovery cohort. Cases and controls were also similarly defined. For cases, serum samples were collected on the same day as RHC (except for four samples collected within 6 weeks before). METHODS Serum proteome signature of SSc-PAH Study population In a third step, in order to determine if our candidate biomarkers were influenced by other SSc manifestations apart from PAH, three other patient groups were recruited from Boston University Arthritis Centre. The first group included patients with dcSSc, no extensive ILD and no evidence of PAH. A second group included patients with lcSSc, extensive ILD and no evidence of PAH. A third group included healthy controls.
Sanges S, et al . Ann Rheum Dis 2023; 82 :365–373. doi:10.1136/ard-2022-223237
366
Powered by FlippingBook