Does "regression" of Barrett's mucosa lessen the risk of developing dysplasia and cancer in Barrett's mucosa?
F.H. Ellis (Boston)
The question of whether the risk of patients with Barrett's esophagus (BE) developing dysplasia or cancer is reduced after "regression" of the columnar-lined esophagus (CLE) is difficult to answer for information on this point is scanty and the data conflicting. However, in order to provide a reasonable response there are a number of preliminary questions that must be addressed.
First of all, what evidence do we have that partial or complete regression of CLE occurs, either spontaneously or after medical or surgical treatment?
Secondly, and as a corollary to the first question; since it is now generally agreed that most adenocarcinomas in BE undergo a preliminary stage of high grade dysplasia, what evidence is there that treatment of any sort reverses or prevents the development of dysplastic changes that ultimately lead to cancer?
Finally, since the question implies that there may be a lower risk of cancer developing in patients with short segments of CLE as compared to those with long segments, is there any evidence to support this view?
Partial and complete regression of CLE
In order to provide definitive evidence of regression of CLE after treatment, the precise length of the abnormal segment of the esophagus before and after treatment must be determined. Unfortunately, "regression" is often determined by the level of the squamocolumnar junction as measured from the incisor teeth and is not always confirmed by biopsy. Because of the possibility of sampling errors, multiple biopsies at different levels should be obtained. Histologic evidence of regenerating squamous epithelium must be present, but, as Skinner  and others  have pointed out, the regenerating squamous epithelium may overly persistent columnar epithelium suggesting regression, whereas in actual fact CLE remains, even though the surface epithelium appears normal endoscopically and by biopsy. Following antireflux surgery, the end of the tubular esophagus may be shifted by fundoplication and crural approximation giving a false impression of regression of CLE. Rarely is manometry or pH testing used to measure CLE regression or the success of the antireflux maneuvers, further complicating a correct interpretation of the findings.
There is however abundant literature on this subject. With few exceptions [3, 4], reports of results of medical treatment with antacids, H2 blockers and proton pump inhibitors fail to provide conclusive evidence to suggest that medical therapy leads to either partial or complete regression of CLE [5-9].
The effect of antireflux procedures on CLE was discussed before this organization three years ago by Williams  who reviewed the literature through 1992. Of 249 patients reported, there were only five with complete regression (2%) and 16 with partial regression (6.4%) of CLE. Published reports subsequent to Williamson's review confirm the view that regression of CLE is extraordinarily uncommon and complete regression is at best an oddity rather than a predictable result after antireflux procedures. However, one recent report is an exception .
Regression of CLE was reported in 24 of 56 patients after antireflux surgery, in 5 of whom complete regression occurred. The addition of vagotomy to the antireflux procedure favored regression. Only one patient developed carcinoma and progression of the abnormal mucosa was present in that patient.
Pertinent to this discussion, however is Williamson's follow-up of the 5 patients previously reported  who demonstrated partial regression of CLE after antireflux surgery. Two of these developed invasive adenocarcinoma 3 and 15 years postoperatively . Sampliner et al.  have also reported the developement of adenocarcinoma after partial regression of CLE.
Follow-up of patients exhibiting partial or complete regression of CLE is unfortunately rare but Williamson's experience  and that of Sampliner et al.  would suggest that no protection against the subsequent development of carcinoma is provided thereby. However, there have been no reports of adenocarcinoma developing after complete regression of CLE.
It would seem that the most effective way to achieve complete regression is by use of new innovative approaches including Nd:YAG  and argon  laser ablation and photo dynamic therapy  in an anacid environment . These approaches represent major advances in protecting the patient with benign CLE from ultimate malignant transformation. They are designed to remove all of the abnormal surface epithelium and, coupled with antacid maneuvers either pharmacological or surgical, permit squamous epithelial replacement of the denuded segment. While no data is yet available on the subsequent risk of dysplasia or malignant transformation occurring after these procedures, they could well reduce the risk of subsequent development of adenocarcinoma.
Information concerning the fate of dysplasia in patients with CLE is poorly understood, not only because sampling errors may lead to the faulty impression of its disappearance, but also because of the well known observer variability (up to 90%) in its histologic interpretation .
Two reports suggest that the response to medical and surgical treatment is variable including progression of low grade to high grade dysplasia and carcinoma, stable persistence of high grade dysplasia  or, rarely, complete regression of low grade dysplasia . In another report antireflux surgery appeared to prevent the development of dysplasia .
The fact that high grade dysplasia either remains stable or progresses to cancer regardless of treatment suggests that the risk of developing cancer remains high in patients with CLE and high grade dysplasia.
Length of CLE
If the risk of developing cancer in CLE were to be less in patients who show evidence of "regression" of the specialized columnar epithelium, then there should be convincing evidence that patients with long segments of CLE are at a greater risk for the development of adenocarcinoma. In fact, there are those who claim that patients with long segments of CLE are more likely to develop adenocarcinoma than those with short segments of CLE [22-24]. This however, has not been our experience, for in a review of 241 patients with Barrett's esophagus, the length of CLE did not prove to be a risk factor for development of carcinoma since the extent of Barrett's epithelium did not differ significantly in patients with benign disease as compared to those with malignant disease . A "limited" length of CLE was present in 35% of patients who developed adenocarcinoma.
Evidence is accumulating that current definitions for BE, requiring a 3-cm length of CLE, may be too rigid, since short segments of CLE were identified in 18% of biopsies from the esophagogastric junctional area in the report of Spechler et al. . Furthermore, the frequency of dysplasia, a known precursor of adenocarcinoma in BE, has reported to be as high as 9.1% in such individuals , and adenocarcinoma has been described as developing in such short segments . This has led Cameron to suggest that most cancers arising at the esophagogastric junction arise in short segments of BE rather than from the cardia of the stomach itself .
Finally, further evidence that short segments of CLE are not immune to malignant transformation is provided by the development of a second cancer in the short residual segment of CLE remaining after an incomplete esophagogastrectomy for adenocarcinoma in Barrett's esophagus. I have encountered three such cases in my practice, and it is now widely recognized that all of the Barrett's mucosa must be excised at the time of esophagogastrectomy for adenocarcinoma in Barrett's  in order to prevent this complication.
These reports and findings, while somewhat contradictory, do not provide convincing evidence that short segments of CLE are less likely to undergo malignant transformation than are long segments of CLE.
While medical treatment rarely leads to either partial or complete regression of CLE, anti -reflux surgery may produce partial regression of the abnormal epithelium in up to 6% of cases. However, partial regression does not protect against the subsequent development of carcinoma, (2 out of 5 cases in our experience).
High grade dysplasia has not been shown to regress but rather to remain stable and ultimately to lead to invasive carcinoma. Only mild degrees of dysplasia occasionally undergo complete regression.
Short segments of CLE are just as likely as long segments to undergo malignant transformation.
In conclusion, it would appear that only complete and permanent disappearance of the abnormal epithelium may protect the patient from the subsequent development of carcinoma.
This goal may be achieved by use of one or the other of the new techniques of esophageal mucosal ablation. Only time will tell.
10. Williamson WA. Does complete suppression of acidity and alkaline reflux contribute to healing? In: Giuli R, Tytgat GNJ, DeMeester TR, Galmiche JP, eds. OESO. The esophageal mucosa. Amsterdam: Elsevier, 1994:992-994.
20. Reid BJ, Blount PL, Rubin LE, et al. Flow-cytometric and histologic progression to malignancy in Barrett's esophagus: prospective endoscopic surveillance of a cohort. Gastroenterology 1992;102:1212-1219.