Ab0944 comorbidities in the spondyloarthritis gisea cohort: an average treatment effect analysis on patients treated with bdmards

Background The treatment of Spondyloarthritis (SpA) has enormously improved thanks to bDMARDs, which are usually safe but they still require some caution because of their interference with the immune system homeostasis. ATE is a still evolving method designed to study the causal relationship between treatment and outcome. Objectives We endeavoured to investigate the impact of TNF-inhibitors (TNFi), anti-interleukin17 or interleukin12/23 monoclonal antibodies (anti-IL) on comorbidities among SpA patients enrolled in the Italian GISEA Registry, using an average treatment effect (ATE) analysis. Methods SpA patients from the GISEA Registry were divided into groups according to pharmacological exposure: no treatment (G0), TNFi (G1) and anti-IL (G2). In each group, the prevalence and incidence of infectious, cardiopulmonary, endocrinological, non-inflammatory gastrointestinal disease (NIGID), oncologic, renal and neurologic comorbidities were evaluated. Thus, each comorbidity was fitted for ATE and baseline features were evaluated for importance. Results This multi-centre Italian GISEA study comprised 4458 SpA patients (G0=495 patients, G1=3113 patients, G2=815 patients). Cardiovascular disease was the most prevalent and incident comorbidity in all groups, with no significant difference between groups. ATE showed no increased risk of solid cancer in G1 and G2 (G1 vs. G0 = 0,42 95% CI 0,20-0,85; G2 vs. G0 = 0,26 95% CI 0,08-0,71), but significantly higher prevalence and incidence in G0 (14.07/1000 patient-years, p=0,0001). Conversely, a significantly higher risk of NIGID and fibromyalgia was found in G1 and G2 vs. G0 (NIGID: G1 vs. G0 =1.56 95% CI 1.06-2.33, G2 vs. G0 =1.91 95% CI 1.05-3.24; fibromyalgia: G1 vs. G0 1.69 95% CI 1.05-2.68, G2 vs. G0 2.13 95% CI 1.14-3.41). No treatment risk modification was observed concerning haematological malignancies, cardiovascular events and endocrinological comorbidities. Conclusion Overall, this study reveals that bDMARDs have only a slight interference of the occurrence of comorbidities in SpA patients, underlining the appropriateness of the use of bDMARDs in current clinical practice. Some caveats pertain to NIGID and fibromyalgia. Importantly, causality may yield more reliable and relevant clinical information, flattening the unbalance between observational data and clinical trials. Table 1. Prevalence and incidence of selected comorbidities in the GISEA Cohort and results of the Average Treatment Effect Analysis (ATE). Incidence of comorbidities (number of events; events/1000 patient-years) Comorbidity No treatment (G0)N=494 TNFi (G1)N =3113 AntiIL (G2)N=815 P value (ANOVA) Cardiovascular disease 10 events; 16.90 114 events; 28.81 26 events; 26.44 0,0794 Endocrinological disease 7 events; 11.35 60 events; 9.30 15 events; 13.83 0,6678 Fibromyalgia 5 events; 6.72 102 events; 13.42 33 events; 25.19 0,0095 Gastrointestinal (non IBD) disease 8 events; 12.41 110 events; 16.48 33 events; 19.68 0,0408 Haematologic malignancy 0 events; 0 2 events; 0.26 0 events; 0 0,6498 Solid cancer 10 events; 14.07 13 events; 1.71 5 events; 3.79 0,0001 Average treatment effect (ATE) analysis Outcome (comorbidity)RR (CI 95% ) TNFi vs. no treatment (G1 vs. G0 ) antiIL vs. no treatment (G2 vs. G0 ) antiIL vs. TNFi (G2 vs. G1 ) Cardiovascular 1.41 (0.99-1.99) 1.66 (1-2.97) 1.16 (0.91-1.55) Endocrinologic 1.27 (0.76-2.05) 1.41 (0.68-2.72) 1.09 (0.74-1.54) Fibromyalgia 1.69 (1.05-2.68 ) 2.13 (1.14-3.41 ) 1.25 (0.95-1.51) GI (non IBD) 1.56 (1.06-2.33 ) 1.91 (1.05-3.24 ) 1.21 (0.92-1.51) Haematologic malignancy 0.78 (0.14-2.88) 0.81 (0.04-4.67) 0.78 (0.20-1.96) Solid cancer 0.42 (0.20-0.85 ) 0.26 (0.08-0.71 ) 0.61 (0.33-0.96 ) Legend: G0= Group 0, no treatment= no biologic DMARD, G1= Group 1, TNFi= Tumour necrosis factor inhibitors, G2= Group 2, antiIL= anti-interleukin17 and anti-interleukin17/23 monoclonal antibodies, N= numerosity of the group, IBD= inflammatory bowel disease, RR= relative risk, CI 95%= confidence interval 95%. Significant results are in bold characters. Acknowledgements Roberta Ramonda and Giovanni Lapadula are part of the GISEA Working Group, comprising the Centre of Centre of Bari, Milan (Presidio Ospedaliero Gaetano Pini), Brescia, Catania, Foggia, Rome (Policlinico Umberto I, Policlinico Gemelli and Policlinico Tor Vergata), Cagliari, Modena, Verona, Turin (Azienda Ospedaliero-Universitaria Città della Salute and Ospedale Mauriziano Umberto I), Siena, Pavia, Messina, Ferrara and Padova. Disclosure of Interests None Declared.