Diagnostic utility of narrow-band imaging endoscopy for pharyngeal superficial carcinoma.

AIM: To investigate the endoscopic features of pharyngeal superficial carcinoma and evaluate the utility of narrow-band imaging (NBI) for this disease.
METHODS: In the present prospective study, 335 patients underwent conventional white light (CWL) endoscopy and non-magnified/magnified NBI endoscopy, followed by an endoscopic biopsy, for 445 superficial lesions in the oropharynx and hypopharynx. The macroscopic appearance of superficial lesions was categorized as either elevated (< 5 mm in height), flat, or depressed (not ulcerous). Superficial carcinoma (SC) was defined as a superficial lesion showing high-grade dysplasia or squamous cell carcinoma on histology. The color, delineation, and macroscopic appearances of the lesions were evaluated by CWL endoscopy. The ratio of the brownish area/intervascular brownish epithelium (IBE), as well as microvascular proliferation, dilation, and irregularities, was determined by non-magnified/magnified NBI endoscopy. An experienced pathologist who was unaware of the endoscopic findings made the histological diagnoses. By comparing endoscopic findings with histology, we determined the endoscopic features of SC and evaluated the diagnostic utility of NBI.
RESULTS: The 445 lesions were divided histologically into two groups: a non-SC group, including non-neoplasia and low-grade dysplasia cases, and an SC group. Of the 445 lesions examined, 333 were classified as non-SC and 112 were classified as SC. There were no significant differences in age, gender, or the location of the lesions between the patients in the two groups. The mean diameter of the SC lesions was significantly greater than that of non-SC lesions (11.0 ± 7.6 mm vs 4.6 ± 3.6 mm, respectively, P < 0.001). Comparisons of CWL endoscopy findings for SC and non-SC lesions by univariate analysis revealed that the incidence of redness (72% vs 41%, respectively, P < 0.001) and a flat or depressed type of lesion (58% vs 44%, respectively, P = 0.013) was significantly higher in the SC group. Using non-magnified NBI endoscopy, the incidence of a brownish area was significantly higher for SC lesions (79% vs 57%, respectively, P < 0.001). On magnified NBI endoscopy, the incidence of IBE (68% vs 33%, P < 0.001) and microvascular proliferation (82% vs 51%, P < 0.001), dilation (90% vs 76%, P = 0.002), and irregularity (82% vs 31%, P < 0.001) was also significantly higher for the SC compared with the non-SC lesions. Multivariate analysis revealed that the incidence of redness (P = 0.022) on CWL endoscopy and IBE (P < 0.001) and microvascular irregularities (P < 0.001) on magnified NBI endoscopy was significantly higher in SC than non-SC lesions. Redness alone exhibited significantly higher sensitivity and significantly lower specificity for the diagnosis of SC compared with redness plus IBE and microvascular irregularities (72% vs 52%, P = 0.002; and 59% vs 92%, P < 0.001, respectively). The accuracy of redness plus IBE and irregularities for the diagnosis of SC was significantly greater than using redness alone (82% vs 62%, respectively, P < 0.001).
CONCLUSION: Redness, IBE, and microvascular irregularities appear to be closely related to SC lesions. Magnified NBI endoscopy may increase the diagnostic accuracy of CWL endoscopy for SC.