Transcriptionally Active Human Papillomavirus Is Strongly Associated With Barrett's Dysplasia and Esophageal Adenocarcinoma

OBJECTIVES: The role of human papillomavirus (HPV) in Barrett's esophagus (BE) remains unclear. The few studies that have previously investigated HPV and esophageal adenocarcinoma (EAC) or BE have produced either negative data or positive results of doubtful clinical/etiological significance or have detected only low-risk HPV types. We therefore prospectively determined the prevalence of biologically active HPV in esophageal epithelium of patients representing the Barrett's metaplasia dysplasia-adenocarcinoma sequence. METHODS: HPV DNA was estimated by nested PCR and viral transcriptional activity detected by E6/7 oncogene mRNA expression and p16INK4A immunohistochemistry in fresh frozen and paraffin-embedded esophageal biopsies of patients with BE, Barrett's dysplasia (BD), and EAC, as well as controls. Biopsies were obtained from the transformation zone (squamocolumnar junction (SCJ)) and the lesion, or corresponding site in controls, i.e., 2 cm above the gastroesophageal junction (GEJ). RESULTS: Of the 261 patients, 81 were positive for HPV DNA. In controls and BE, the virus was mostly detected at the transformation zone. Compared with controls (18.0 %), HPV positivity was significantly more common in BD (68.6 %, incidence rate ratio (IRR) 2.94, 95 % confidence interval (CI) 1.78-4.85, P < 0.001) and EAC (66.7 %, IRR 2.87, 95 % CI 1.69-4.86, P < 0.001), but not in BE (22.1 %, IRR 1.06, 95 % CI 0.60-1.85, P = 0.85). Of the patients, 92.6 % were high-risk (HR) HPV, i.e., types 16 and 18. Again, p16INK4A positivity was greatest in BD and EAC and much less in BE patients (44.1 %, IRR 17.0 (95 % CI 4.86-59.6, P < 0.001), 44.4 %, 17.0 (95 % CI 4.87-59.4, P < 0.001), and 10.6 %, 3.93 (95 % CI 1.01-15.3, P = 0.048) respectively). In 66 HPV DNA-positive patients tested for E6/E7 mRNA, none of the control (n = 16) or BE (n = 13) individuals were positive, whereas 9/22 BD and 9/15 EAC patients demonstrated oncogene expression (P < 0.001). When HPV DNA, p16INK4A, and E6/E7 mRNA were all positive, there was a very strong association with disease severity (SCJ: odds ratio (OR) 104, 95 % CI 20.3-529, P < 0.001; lesion: OR 62.2, 95 % CI 12.4-311, P < 0.001) than when all were negative. CONCLUSIONS: Transcriptionally active HR-HPV was strongly associated with BD and EAC, but was largely biologically irrelevant in BE and controls, suggesting a potential role in esophageal carcinogenesis. These data provide robust justification for further detailed longitudinal, interventional, and molecular studies.