Systemic evaluation of the relationship between psoriasis, psoriatic arthritis and osteoporosis: observational and Mendelian randomisation study

Abstract

Objectives and methods: With 432 513 samples from UK Biobank dataset, multivariable linear/logistic regression were used to estimate the relationship between psoriasis/psoriatic arthritis (PsA) and estimated bone mineral density (eBMD)/osteoporosis, controlling for potential confounders. Here, confounders were set in three ways: model0 (including age, height, weight, smoking and drinking), model1 (model0 +regular physical activity) and model2 (model1 +medication treatments). The eBMD was derived from heel ultrasound measurement. And 4904 patients with psoriasis and 847 patients with PsA were included in final analysis. Mendelian randomisation (MR) approach was used to evaluate the causal effect between them.

Results: Lower eBMD were observed in patients with PsA than in controls in both model0 (β-coefficient=-0.014, p=0.0006) and model1 (β-coefficient=-0.013, p=0.002); however, the association disappeared when conditioning on treatment with methotrexate or ciclosporin (model2) (β-coefficient=-0.005, p=0.28), mediation analysis showed that 63% of the intermediary effect on eBMD was mediated by medication treatment (p<2E-16). Patients with psoriasis without arthritis showed no difference of eBMD compared with controls. Similarly, the significance of higher risk of osteopenia in patients with PsA (OR=1.27, p=0.002 in model0) could be eliminated by conditioning on medication treatment (p=0.244 in model2). Psoriasis without arthritis was not related to osteopenia and osteoporosis. The weighted Genetic Risk Score analysis found that genetically determined psoriasis/PsA were not associated with eBMD (p=0.24 and p=0.88). Finally, MR analysis showed that psoriasis/PsA had no causal effect on eBMD, osteoporosis and fracture.

Conclusions: The effect of PsA on osteoporosis was secondary (eg, medication) but not causal. Under this hypothesis, psoriasis without arthritis was not a risk factor for osteoporosis.

Keywords: arthritis, psoriatic; bone density; epidemiology; osteoporosis.

Figure 1.. Study design overview (A) and the quality control of participants (B).

wGRS, weighted genetic risk score, 2SLS, two-stage least-squares regression, MR, Mendelian Randomization, HLA region, humanleukocyte antigen region, eBMD, heel bone mineral density, IVW, inverse variance-weighted method, MR-PRESSO, MR pleiotropy residual sum and outlier test. Note: The green color was analysed in the UK biobank dataset, the light red was using the publicly available data.

Figure 2.. Summary of effect estimates from the different methods for the psoriasis and PsA on eBMD.

CI, confidence interval, wGRS, weighted genetic risk score, MR, Mendelian Randomization.

Figure 3.. Association of eBMD with the psoriasis and PsA using reverse-direction MR analysis.

IVW, inverse variance-weighted method, MR-PRESSO, MR pleiotropy residual sum and outlier test.