An Investigation into the Association and Prediction of Serious Infection in Patients on Biologic Therapies for Psoriasis

UoM administered thesis: Phd


Aim: This thesis aimed to investigate the risk of serious infection associated with biologic therapies in people with psoriasis. Methods: A systematic search was undertaken of randomised controlled trials (RCTs) and cohort studies reporting serious infections in people taking any licensed biologic therapy for psoriasis compared with those receiving placebo, non-biologic therapy, or other biologic therapies. The meta-analysis was performed using Peto's method with a fixed effects model. Further studies in the thesis used data from the British Association of Dermatologists’ Biologic Interventions Register (BADBIR). Propensity-score weighted Cox proportional hazards models were performed to adjust for potential confounders in the comparison of serious infection risk between cohorts starting on etanercept, adalimumab, infliximab and ustekinumab and a cohort on non-biologic systemic therapies. A multivariable risk prediction model for serious infection within one year of systemic treatment initiation was developed using potentially important covariates in a multiple logistic regression model. Both internal and external validation was performed to assess model performance, with the German psoriasis registry Psobest utilised as the validation dataset. Results: In the systematic review, low- to very-low-quality RCT data showed no significant difference between any biologic therapy and placebo at week 12-16 (pooled Peto odds ratio(OR) 0.71, 95% confidence interval [CI] 0.36-1.41). No significant increases in the risk of serious infection were observed for etanercept (adjusted hazard ratio [adjHR] 1.10, [0.75-1.60]); adalimumab (adjHR 0.93, [0.69-1.26]) or ustekinumab (adjHR 0.92, [0.60-1.41]) compared against non-biologic systemic therapies, while infliximab was associated with an increase in the risk of serious infection (adjHR 1.95, [1.01-3.75]) in BADBIR. Nine baseline covariates were selected in the multivariable risk prediction model, and starting infliximab was the strongest predictor for serious infection. Evaluation of model performance showed a bootstrap optimism-corrected C-statistic of 0.64 (95% C.I. 0.60 -0.69) and calibration slope of 0.88 (0.70-1.07), while external validation performance was poor [C-statistic 0.52 (95% C.I. 0.42-0.62), calibration slope 0.36 (95% C.I. -0.24-0.97)]. Conclusion: The risk of serious infection should not be a key discriminator for patients and clinicians when choosing between non-biologic systemic therapies, etanercept, adalimumab and ustekinumab for the treatment of psoriasis, but clinicians should take into account an increased risk of serious infection when considering infliximab. The prognostic model may help psoriasis patients and clinicians identify treatment choice and modifiable lifestyle factors associated with a lower risk of serious infection before initiation of systemic therapy.


Original languageEnglish
Awarding Institution
Award date1 Aug 2019