Although magnetic resonance enterography (MRE) can accurately reflect ileal inflammation in pediatric Crohn disease (CD), there are no pediatric data on the accuracy of MRE to detect upper gastrointestinal tract (UGI) lesions. We aimed to compare MRE and esophagogastroduodenoscopy (EGD) in detecting the spectrum and severity of UGI disease in children.Methods:
This is an ancillary study of the prospective multi-center ImageKids study focusing on pediatric MRE. EGD was performed within 2 weeks of MRE (at disease onset or thereafter) and explicitly scored by SES-CD modified for the UGI and physician global assessment. Local and central radiologists scored the UGI region of the MRE blinded to the EGD. Accuracy of MRE compared with EGD was examined using correlational coefficients (r) and area under receiver operating characteristic curves (AUC).Results:
One hundred and eighty-eight patients were reviewed (mean age 14 ± 1 years, 103 [55%] boys); 66 of 188 (35%) children had macroscopic ulcerations on EGD (esophagus, 13 [7%]; stomach, 34 [18%]; duodenum, 45 [24%]). Most children had aphthous ulcers, but 10 (5%) had larger ulcers (stomach, 2 [1%]; duodenum, 8 [4%]). There was no agreement between local and central radiologists on the presence or absence of UGI inflammation on MRE (Kappa = −0.02, P = 0.71). EGD findings were not accurately detected by MRE, read locally or centrally (r = −0.03 to 0.11, P = 0.18–0.88; AUC = 0.47–0.55, P = 0.53–1.00).No fistulae or narrowings were identified on either EGD or MRE.Conclusions:
MRE cannot reliably assess the UGI in pediatric CD and cannot replace EGD for this purpose.