CHICAGO – If scouring data is what makes a gastroenterologist feel good about risk assessment, there may be a lot of unhappy gastroenterologists out there, at least when it comes to the risk of cancer arising from cardiac mucosa in Barrett’s esophagus, according to Nicholas J. Shaheen, MD.
The risk arising from this nonintestinal metaplasia growth is probably quite low in real life, but the extant literature gives doctors a lot of contradictions, he said at the meeting sponsored by the American Gastroenterological Association.
“The risk of cancer with cardiac mucosa is unclear,” said Dr. Shaheen of the University of North Carolina at Chapel Hill. “Some data do suggest that, at least when present in the tubular esophagus in patients with gastroesophageal reflux symptoms, there may be a risk of adenocarcinoma close to what’s seen in patients with intestinal metaplasia. Other data suggest the risk is quite low, perhaps even approximating that of the general population.”
The reasons for what Dr. Shaheen called “remarkable variability” in these data probably arise more from sampling error than real life. The studies are retrospective, and many lack long-term follow-up data, are plagued with insufficient numbers, and – perhaps most importantly – are not grounded in any standard clinical methodology.
“People who do endoscopy for a living understand that the stuff you read about systematic biopsy protocols is hardly ever honored in the breach. None of these studies ever reports the biopsy protocol from which the samples were taken.”
This lack of protocol means that studies on the cancer risk of columnar lined esophagus (CLE), which is negative for intestinal metaplasia are probably flawed from the beginning.
“The truth is that most gastroenterologists do a lousy job of biopsying Barrett’s, so there is probably a lot of sampling error in these studies, and they are contaminated with a high rate of intestinal metaplasia [IM],” said Dr. Shaheen.
And these studies do not report on the length of the CLE segment from which the biopsy was taken. “The likelihood of finding goblet cells [a characteristic of cardiac mucosa] increases with the length of Barrett’s. None of the studies is normalized for Barrett’s length. When we see studies saying the cancer risk is higher in the presence of goblet cells, length could be a partially confounding association.”
A 2009 study with a small sample size of 68 CLE patients found that abnormal DNA was just as likely in IM-negative samples as IM-positive ones. All of the samples were significantly different from the control samples, suggesting that any metaplasia in the CLE may already be on the path to cancer, Dr. Shaheen said (Am J Gastro. 2009;104:816-24)
In fact, a 2007 Scandinavian study supported the idea that IM isn’t even necessary for neoplastic progression of CLE (Scand J Gastroenterol 2007;42:1271-4). The investigators followed 712 patients for 12 years, and found that the adenocarcinoma rate was about 0.4 per patient per year whether the sample was IM positive or not.
“This study was enough to put a little shudder in the endoscopy community. If IM doesn’t matter, you’re talking about increasing the work in the endoscopy lab by 100%, because there are twice as many non-IM patients as those with IM.”
A 2008 study seemingly found something similar – but with a caveat, Dr. Shaheen said. The CLE patients in this study were followed for 3.5 years, and the cancer rate was virtually identical. But as the follow-up progressed, more and more biopsies turned up IM positive. “A first negative biopsy looked like it was associated with disease-free survival, but almost all IM-negative samples eventually became IM positive, so this didn’t really answer our question.”
Other studies have found that non-IM CLE has a very low neoplastic risk, and that IM is almost always a prerequisite for cancer to develop. The largest of these was conducted in the Northern Ireland Barrett’s Esophagus Registry in 2011. It followed more than 8,000 patients for 7 years. Patients with IM were 3.5 times more likely to develop a related adenocarcinoma than were those without IM (J Natl Cancer Inst. 2011;103:1049-57).
The contradictory evidence leads Dr. Shaheen to suggest a specific biopsy protocol for patients with Barrett’s esophagus.
“In my opinion, if you see a long segment of Barrett’s – more than 2 cm – and the biopsy is negative for IM, there is a good chance that you have a sampling error there, and a second endoscopy and biopsy are not unreasonable. If you see a short segment of Barrett’s and the biopsy is negative for IM, the cancer risk is unclear, but in general it’s probably pretty low, whether there are goblet cells there or not. I would say retaining these patients under endoscopic surveillance is of dubious value. [With] the likely low absolute risk of cancer in this patient population, no blanket recommendation for surveillance is advisable.”