How does Barrett's mucosa heal?
Th.P.J. Hennessy (Dublin)
Barrett's esophagus is a metaplastic epithelial change in which squamous epithelium is replaced by columnar cell epithelium. Robbins and Cotran  have defined metaplasia as a reversible change in which one adult cell type is replaced by another. Despite the scanty clinical evidence it would seem, therefore, since the condition is alleged to be reversible, that regression of the metaplastic Barrett's epithelium ought to be possible.
Regression of Barrett's epithelium after treatment with omeprazole has been reported by Deviere et al.  and Gore et al. . The latter observed both island regression and encroachment of squamous epithelium over the glandular columnar epithelium. This latter form of regression which Skinner  also observed in his patients may represent the process by which the new squamous epithelium advances down the esophagus.
Reports of complete regression, however, are rare. In other reports of response to medical treatment, inflammation has subsided and Barrett's ulcers have healed but without evidence of regression.
Antireflux surgery has induced regression in four out of 10 patients reported by Brand , and Henrion  identified complete regression of columnar epithelium 2.5 years after biliary division. Skinner's patients experienced reversion from dysplastic to nondysplastic epithelium and other patients progressed to regression to squamous epithelium, following antireflux surgery. Nevertheless, several authors reported no regression of metaplastic epithelium after antireflux surgery, and both Williamson  and Perniceni  concluded that regression of Barrett's epithelium after antireflux surgery was the exception rather than the rule.
Such disparate results are difficult to understand on clinical grounds and both the question of the development of the metaplastic epithelium and its regression are inextricably linked.
Although the derivation of the metaplastic epithelium remains speculative, various suggestions have been put forward. Allison and Johnstone  suggested an origin from proximal extension of the columnar epithelium lining the gastroesophageal junction and the cardia. Trier  proposed the esophageal mucous glands or embryonic remnants as the source of the ectopic mucosa.
Experimental studies carried out in our department suggest that the metaplastic epithelium is derived from cuboidal cells lining the ducts of the esophageal mucosal glands. In these studies, a fixed sliding hiatus hernia and a Wendel cardioplasty were established in a canine model. A strip of squamous mucosa was then excised from the posterior wall of the lower esophagus proximal to the squamocolumnar junction. A hyperacidity state was then induced by a daily subcutaneous injection of pentagastrin. After 3 months it could be demonstrated that healing of the mucosal defects was by columnar epithelium and that this columnar epithelium arose from the cuboidal cells lining the deeper portion of the esophageal gland ducts.
In the second part of the experiment this columnar epithelium was removed, thus recreating the mucosal defect. The hiatus hernia was then repaired and the cardioplasty reversed. Pentagastrin was discontinued and omeprazole administered. Three months later the mucosal defect had healed largely by columnar epithelium, but on this occasion there were pockets or islands of squamous epithelium present.
The explanation of the different epithelial types involved in the healing of the defect is as follows: the two proximal thirds of these ducts are lined with columnar epithelium and the distal third is lined by squamous epithelium. After stripping off the mucosa the gland ducts are exposed to the surface and luminal contents at varying levels along their length. Where the mucosa has been removed down to the level of the columnar duct lining, attempts at repair will be due to migration of columnar cells from the ducts that will survive in an acid milieu. Where the stripping has been more superficial some squamous duct lining will remain, but attempts at repair by migration of squamous duct cells will fail when reflux is present because the latter will not survive in an acid milieu. However, if the reflux is corrected such migrating squamous cells will give rise to squamous islands. At the junction of the two epithelium types, overlapping can be seen to occur similar to that seen in the clinical situation.
Extrapolating from the experimental to the clinical situation, if the mucosal injury is superficial and squamous duct lining remains, correction of the reflux will allow squamous regression to occur. However, if the mucosal injury has been severe and deep so that no squamous duct lining remains, no regression will be possible even if the reflux is corrected.
The above theory offers an acceptable and logical explanation for island regression, but would not convincingly account for the complete regression reported by some workers. This latter type of regression may be due to distally advancing squamous epithelium growing over the ectopic columnar epithelium in the absence of reflux.
6. Henrion J, Schapira M, Pourbaix A, Heller FR. Regression complete d'un endobrachyoesophage après correction du reflux biliaire chez un malade gastrectomise. Gastroenterol Clin Biol 1989; 13:745-746.
9. Allison PR, Johnstone AS. The oesophagus lined by gastric mucous membrane. Thorax 1953:8:87-101. 10. Trier JS. Morphology of the epithelium of the distal esophagus in patients with mid-esophageal stricture. Gastroenterology 1970:58:441-461.