Background: Magnetic Resonance Imaging (MRI) is the preferred test for assessing fistulising perianal Crohn's disease and its response to anti-TNF therapy. Currently, radiological assessment of perianal fistulas and whether it has responded to treatment is subjective and variable according to radiologist interpretation. This study investigates a complementary objective measurements to reduce inter and intra-observer variability.
Methods: A cohort of patients with perianal Crohn's fistulas treated at a tertiary centre with anti-TNF therapy was identified, and for whom pelvic MRI was available at baseline and follow-up.
Two gastrointestinal radiologists measured fistula volumes, mean signal intensity and time taken to measure fistula volumes on baseline and follow-up MRI scans using a validated open-source segmentation software programme. One gastrointestinal radiologist repeated the measurements after 2 months.
Three different gastrointestinal radiologists also assessed fistula response to treatment (improved/worse/unchanged) by comparing the MRI scans for each patient. Agreement between these radiologists was assessed using the kappa statistic.
Intra-class correlation coefficient (ICC) was used to compare agreement between radiologists and to assess intra-observer variability for one radiologist.
Results: Eighteen patients were recruited of which 6 (33%) were female and median age was 29 years old (range 19–52).
Inter-observer variability was very good for volume and mean signal intensity with ICC of 0.95 (95% CI 0.91–0.98) and 0.95 (95% CI 0.90- 0.97) respectively. Intra-observer variability for one radiologist was also very good for volume and mean signal intensity with ICC of 0.99 (95% CI 0.97–0.99) and 0.98 (95% CI 0.95- 0.99) respectively. Average time taken to calculate volume measurements was 202 and 250 seconds for readers 1 and 2.
Subjective assessment of response between the 3 radiologists was good; kappa of 0.69 (95% CI, 0.49–0.90). All 3 radiologists agreed in 72% of patients.
A significant association was found between percentage volume change and subjective consensus agreement of response made by the radiologists (p=0.001). Median volume change for improved, stable or worsening fistula response was -67% (IQR −78, −47), 0% (IQR −16, +17), and +487% (IQR +217, +559) respectively.
Conclusions: Objective measurements of fistula volume and signal intensity on MRI scans can be made rapidly using segmentation software with good agreement between radiologists. Percentage changes in measured volume are significantly correlated with judgements of response made by radiologists and could potentially be used to decrease variability in fistula assessment. These measurements can potentially be used to track fistula response over time.