Background/Purpose: There has been no data
reported evaluating MRI parameters of inflammation and structural damage. We
aimed to identify factors influencing survival on anti-TNF therapy in real
world practice specifically focusing on the role of MRI parameters of inflammation
and structural damage.

Methods: In the FOllow-up
Research Cohort in AS (FORCAST), AS patients from Northern Alberta attending community
and academic practices are assessed for clinical and laboratory outcomes every
6 months, radiography at baseline and 2 years, MRI at baseline, at 3-6 months for patients starting anti-tumor necrosis factor
alpha (anti-TNFα), and annually. MRI
inflammation was assessed using Spondyloarthritis
Research Consortium of Canada (SPARCC) Sacroiliac Joint (SIJ) and Spine scores
while structural change was assessed independently using the SPARCC SIJ
Structural Score (SSS) for fat metaplasia, erosion, backfill, ankylosis and the Fat AS Spine Score (FASSS) for fat
metaplasia in the spine. MRI scans were scored independently by 2 readers and
adjudicated by a third reader according to pre-specified rules. We used Kaplan–Meier plots, log rank tests and univariate and multivariate Cox regression analyses to assess the
effects of patient demographics, smoking, B27, NSAID utilization, and baseline
CRP, ASDAS, mSASSS, SPARCC scores, SSS and FASSS
scores on drug survival. We also assessed early attainment post-treatment of
CRP<6mg/L, ASDAS<1.3, and SPARCC remission (SIJ <2, Spine <3) as
predictors of anti-TNF survival.

Results: We recruited
480 patients on
anti-TNF, mean (SD) age 41.0 (12.7) years, 74.4% males, mean (SD) symptom
duration 18.4 (11.6) years, mean (SD) survival on first anti-TNF 1228.1 (1036.9) days. The number discontinuing first-time anti-TNF prescription was 126
(26.3%) after mean (SD) follow up of 814.8 days from first prescription date,
of which 28% was for lack of efficacy (LOE), and 17% for adverse events. There
were 45 primary and 82 secondary failures. 125 patients had MRI at baseline and
100 had at least one follow up MRI. Univariate
analysis showed that male sex (HR 0.56, p=0.002), baseline CRP (HR 0.99, p=0.03)
and early post-treatment attainment of ASDAS<1.3 (HR 0.57, p=0.02), and
CRP<6mg/L (HR 0.56, p=0.02) were significant predictors of drug survival. Early
attainment of SPARCC SIJ remission was the best MRI predictor of drug survival
(HR 0.58, p=0.14). In multivariate analysis of clinical predictors, SPARCC SIJ
< 2  (adjusted OR=2.2[1.02-4.74];
p=0.043) was a significant predictor.

Conclusion: From an extensive array of
patient demographic and disease severity variables, attainment of normalized
CRP or low disease activity state within first year of starting an anti-TNF was
most strongly associated with survival on treatment. Early remission of MRI
inflammation may also be a factor but this requires further study with larger
sample size.

    Table. Multivariable Cox Regression

*Within
first year of treatment