P296 MAGNIFI-CD index is appropriate for treatment monitoring in perianal Crohn’s Disease
Beek, K.(1);Mulders, L.(2)*;Tielbeek, J.(1);Horsthuis, K.(1);Buskens, C.(3);D'Haens, G.(2);Gecse, K.(2);Stoker, J.(1);
(1)Amsterdam UMC, Radiology & Nuclear Medicine, Amsterdam, The Netherlands;(2)Amsterdam UMC, Gastroenterology & Hepatology, Amsterdam, The Netherlands;(3)Amsterdam UMC, Surgery, Amsterdam, The Netherlands;
Background
Perianal Crohn’s Disease (pCD) is a frequent and debilitating complication of Crohn’s Disease (CD). Magnetic Resonance Imaging (MRI) is the imaging modality for diagnosis, classification, and treatment monitoring in pCD. To evaluate treatment response, a validated radiological index is mandatory. Recently, the MAGNIFI-CD index has been developed. The aim of our study was to validate the MAGNIFI-CD index; evaluating reliability, responsiveness and accuracy.
Methods
At two tertiary IBD referral centers, patients (>16yrs) with complex pCD that had pelvic MRI before and 3-24 months after medical and/or surgical treatment were included. Two independent, blinded abdominal radiologists randomly scored total MAGNIFI-CD and its separate items. Two clinicians scored clinical outcomes (remission, response, non-response) using the “fistula drainage assessment” (FDA). Interobserver agreement (ICCs or weighted K), responsiveness (according to FDA; Wilcoxon signed rank test), and test accuracy (ROC analyses for the follow-up (FU) MAGNIFI-CD) were determined. Youden indices (YI) were used to choose optimal cut-off values for the FDA categories.
Results
Sixty-seven patients (median age 30 [IQR 23-47], 51% female) were eligible with mean CD and pCD duration of 7.0 yrs (SD 8.7) and 3.7 yrs (SD 4.3), respectively. 44% was biological naïve, 9% had defunctioning ostomy and 20% had proctitis. Baseline MAGNIFI-CD was 18 [IQR 9–20]. MAGNIFI-CD had an almost perfect ICC of 0.88 (95%CI 0.78-0.92); reliability of the separate items was moderate to substantial. Between baseline and FU MAGNIFI-CD a significant decrease was observed in responders and remitters (p<0.05), while a non-significant decrease (p=0.14) was seen in clinical non-responders. AUROCFU MAGNIFI-CD was 0.80 (SD 0.05) discriminating remission and non-response/response, YImax of 0.61 resulted in a cut-off value ≤10 with a sensitivity and specificity of 70% and 80%. The AUROCFU MAGNIFI-CD was 0.75 (SD 0.07) discriminating response/remission and non-response, YImax of 0.50 resulted in a cut-off value ≤14 with a sensitivity and specificity of 73% and 75%. 
Conclusion
The MAGNIFI-CD index is a reliable radiological index supported by an almost perfect agreement. Given the moderate agreement, the item “dominant feature” would need additional training or better definitions. The index shows a robust responsiveness to clinical change and does not improve significantly in non-responders. The MAGNIFI-CD seems to be an acceptable to excellent test to accurately determine response or remission. We suggest a cut-off value of ≤14 for response and ≤10 for remission at follow-up MAGNIFI-CD. These results support that the MAGNIFI-CD is suitable for treatment monitoring in clinical trials.