MB staining of dysplastic and nondysplastic Barrett's esophagus: An in vivo and ex vivo study

Methylene blue (MB) selectively stains specialized columnar epithelium (SCE) in Barrett's esophagus (BE) with high accuracy, including dysplastic cells. AIMS: 1) To prospectively evaluate the accuracy of MB staining for SCE and the MB staining properties of SCE. 2) To determine the association of MB staining properties with the risk for dysplasia and cancer (CA) in BE. METHODS: In the ex vivo study, we mapped, photographed, and obtained samples from 4 fresh, unfixed esophagectomy specimens from patients with HGD and/or early CA before and after MB staining. We characterized the MB stain intensity and heterogeneity (% of nonstaining, flat columnar mucosal according to a standardized ordinal scale. In the in vivo study, a single experienced endoscopist performed MB staining and obtained stained and unstained biopsies from 31 consecutive patients with biopsy-proven BE undergoing endoscopic surveillance. We photographed and videotaped all procedures. 2 blinded experienced pathologists estimated the proportions of SCE and non-SCE epithelium in each specimen and graded dysplasia according to a standardized scale published elsewhere. We determined the association of the intensity and heterogeneity of MB staining with the risk for low grade dysplasia LCD), high grade dypslasia (HGD) or CA using univariate analysis and multiple logistic regression. RESULTS: We obtained 265 biopsies (246 stained and 19 unstained) at 31 EGDs and 40 sections (29 stained and 11 unstained) from 4 surgical specimens. The accuracy of ex vivo and in vivo MB staining for SCE was identical (92%). 11%, 2%, and 0.8% of endoscopic biopsies showed LCD, HGD, and CA, respectively. 52%, 14%, and 20% of sections from excised esophagi had LGO, HGD, and focal CA (7 submucosal and 1 intramucosal), respectively. In surgical specimens, all CA were associated with macroscopic ulceration or a depressed nodule. There was a significant association between decreasing stain intensity and increasing severity of dysplasia (p=.009). Light MB staining was associated with LGD after controlling for ulceration (p=.01). Moderate to marked stain hetergeneity (p=.01) and light or no MB stain (p=.01) were associated with HGD/CA in a multivariate model including ulceration. When these features were applied to the endoscopic group, marked stain heterogeneity was the single best predictor of HGD and/or CA. Focal non- or light blue staining and moderate to marked heterogeneity predicted the presence of HGD and/or CA in all 3 patients diagnosed endoscopically. CONCLUSION: MB stain heterogeneity and decreased to absent staining when found within diffuse dark MB staining SCE are significant independent predictors of CA in BE. These MB staining characteristics may help direct biopsies in patients with endoscopically normal or monotonously abnormal BE. Future studies which validate these endoscopic features may demonstrate an improved diagnostic yield of endoscopic surveillance for early CA.