Supervisors info:
Παπαδαυίδ Ευαγγελία, Αναπληρώτρια Καθηγήτρια Δερματολογίας, Ιατρική Σχολή, ΕΚΠΑ
Μπάμια Χριστίνα, Επίκουρη Καθηγήτρια Επιδημιολογίας και Ιατρικής Στατιστικής, Ιατρική Σχολή, ΕΚΠΑ
Σάμης Τρεβαζάς, Λέκτορας Στατιστικής κι Επιχειρησιακής Έρευνας, Τμήμα Μαθηματικών, ΕΚΠΑ
Summary:
Background: Long-term data restricted to biologic-naive psoriatic patients in daily praxis have rarely been used for the identification of predictors for drug survival.
Objectives: The main objective was to compare long-term drug survival among ustekinumab(UST) (IL-12/23 antagonist) and etanercept(ETN), infliximab(INF) and adalimumab(ADL) (anti-TNF agents). Secondary aim was the assessment of predictors for the drug survival and the subgroup analysis for psoriatic arthritis.
Methods: Data from 12 years were extracted retrospectively from a single-centre prospective cohort of biologic-naive psoriatic patients in daily praxis. Kaplan-Meier survival curves, log-rank tests and multivariate Cox regression analysis with confounder correction were performed. Backward selection was used for the identification of predictors for drug survival. Interaction was used for the subgroup analysis and multiple imputations(MICE) for the verification of the regression results.
Results: 134 patients were included; ETN=58, INF=21, ADL=30, UST=25. Initiation of therapy during/after 2009 made the 70.8% of the cohort, when all biologicals were in market. The drug survival rates are higher for UST after 1 (UST=82.6%,ETN=64%,INF=52.4%,ADL=51.9%) and 3 years (UST=47.8%,ETN=26%, INF=42.9%,ADL=22.2%). INF performs better than ETN and ADL as the time of treatment increases. UST has the longest median drug survival (25thquantile=2.38 yrs). Among the anti-TNF factors, INF has the longest median drug survival (3.13 yrs), followed by ETN (1.92 yrs) and ADL (1year). As compared to UST, patients on ETN (HR=4.57 95%CI:1.55,13.43), ADL (HR=3.85 95%CIs:1.29,11.49) or INF (HR=3.46 95%CIs:1.10,10.86) were in significantly greater risk to discontinue therapy, while adjusting for gender, age, initiation of treatment during/after 2009 and BMI. Presence of psoriatic arthritis did not cause a significant subgroup effect (lr-test:p-value=0.063). Significant positive predictor of drug survival was the concomitant use of methotrexate (HR=0.4 95%CI: 0.16,1.02). Cyclosporine was a negative predictor of drug survival (HR=4.65 95%CI: 1.27,17.04). Patients in the highest BMI category [40,50)kg/m2 were in greater risk to discontinue therapy (HR=10.71 95%CI: 2.46,46.65).
Conclusions: The longest drug survival corrected for confounders was observed for UST, followed by INF, ADL and ETN. Methotrexate is a positive determinant of longer biological drug survival, whereas cyclosporine and higher BMI are negative.