How should endoscopic surveillance of Barrett's esophagus in children be conducted?
E. Hassall (Vancouver)
Adenocarcinoma does occur, albeit rarely, in the childhood age group, as a complication of Barrett's esophagus (BE) , and BE in children may be complicated by the development of adenocarcinoma in adulthood. Therefore, the approach to surveillance of BE in children has important implications for both children and adults.
An approach to surveillance of BE in children must first be put into the context of certain special features of BE that occur in the pediatric age group.
Special features of Barrett's esophagus in children
Rarity of Barrett's esophagus in children
In most studies of BE in children, some or all cases have been diagnosed solely on the presence of a "prominent" Z-line at endoscopy or on the presence of only fundic or cardiac columnar epithelium in a surgical specimen, or in one, perhaps two biopsies said to be from the tubular esophagus [5-16]. In most, biopsy sites relative to the endoscopic landmarks were not documented. One may conclude from this that the fundic or cardiac mucosa containing biopsies have come from hiatal hernias .
Table I shows data from 13 studies 1984-1996, in which 117 children were reported to have BE; of these only 41 had Barrett's specialized metaplasia [5-17]. Many of these studies accrued patients retrospectively from as early as 1974, therefore these do not represent incidence data.
The true prevalence of BE in children in unknown, however two reports do give some insight into this issue. In unpublished data from British Columbia Children's Hospital in Vancouver, BC, over an 11 year period through 1995, only 6 children were newly diagnosed as having BE (i.e. with specialized metaplasia) for a prevalence of 0.02% of all pediatric upper gastrointestinal endoscopies during that period. Their ages were 8-17 years (mean 14 years). From Children's Hospital in Boston  were reported only 7 "children" with BE (specialized metaplasia) over a 5 year period through 1986, for a prevalence of 0.6% of all pediatric upper gastrointestinal endoscopies. Of note is that, in this report, the patients' ages were 13-27 years with a mean of 20 years.
From the above data, one may conclude that BE is rare in children, and the true prevalence depends upon the definition of "child", as well as the nature of the referral population.
The rarity of BE in children compared with that in adults implies that the overall cost of screening children (below 18 years of age) with BE is much lower than that of screening a population of 55-year-olds (i.e. the mean age of diagnosis in adults ).
High prevalence of co-morbidities
Snyder and Goldman  were the first to draw attention to the presence of major underlying systemic disorders in children with BE. They reported 7 children/young adults (mean age 20 years) with Barrett's specialized metaplasia, in all of whom underlying mental retardation and or cerebral palsy was present. Table II summarizes the co-morbidities reported in several studies of BE in children; of the co-morbidities, by far the most prevalent are those in which significant neurologic impairment is present (i.e. mental retardation/cerebral palsy/syndromes). Neurologically impaired children, including those with various syndromes, often have a marked predisposition to gastroesophageal reflux (GER) [1, 8], as do children with cystic fibrosis . In both groups, reporting of symptoms and signs may be impaired or delayed, thus severe and chronic reflux may go undiagnosed for many years [4,17]. The third major co-morbidity said to be associated with BE is the group of children with repaired esophageal atresia [11, 14, 15]; although they have many factors predisposing to severe GER, Barrett's specialized metaplasia has not been described in this disorder . The fourth group at high risk for severe GERD  consists of otherwise healthy children (i.e. without co-morbidity) with hiatal hernia or transient lower esophageal sphincter relaxations (TLESRs), or perhaps acid hypersecretion in a few.
It is apparent from these reports that a major underlying systemic disorder is present in at least 60-70% of children with Barrett's specialized metaplasia. Of these major co-morbidities, two i.e. neurologic impairment and cystic fibrosis result in a markedly shortened life expectancy in children, and perhaps a higher morbidity of endoscopic surveillance, though the latter has not been shown. The implication of a shortened life expectancy is that perhaps an aggressive endoscopic surveillance approach might not be indicated, since the underlying systemic disease is a much more likely cause of mortality than adenocarcinoma arising in BE. This approach should be on a case-by-case basis, since these disorders (especially "neurologic impairment") are very heterogenous.
In contrast, children with esophageal atresia do not have a shortened life span, neither do otherwise normal children with hiatal hernia or TLESRs.
All evidence in children points towards Barrett's specialized metaplasia being an acquired, not a congenital condition . Age is relevant in that the youngest child reported with Barrett's specialized metaplasia was 5 years old , thus it appears that the lesion takes some time to develop. Another important age-related consideration is the annual incidence of development of cancer - incidence figures in the literature are derived from adults 50-60 years of age, and range from as low as 0.2% per year  to as high as 2.1 % per year . It is possible that the annual incidence of development of cancer is lower in the childhood age group, becoming higher only with time.
Approaches to determine surveillance guidelines
for children with Barrett's esophagus
Three possible approaches are as follows:
- use an existing model designed for adults, and extrapolate it to children,
- modify the existing model using childhood parameters,
- use anecdotal data.
Approach 1: use an existing model designed for adults, extrapolate it to children
Recently, Provenzale et al.  reported the results of a decision analysis using a computer cohort simulation of 10,000 hypothetical 55-year-old men with Barrett's esophagus, this being the mean age of diagnosis of BE in adults. They measured life expectancy, quality-adjusted life expectancy, and incremental cost-effectiveness. The model also considers operative morbidity, endoscopic and surgical morbidity, and false-positive and -negative rates for endoscopic surveillance, and a time period from endoscopically detectable esophageal cancer to symptomatic cancer. An annual incidence of cancer of 1.3% per year was used for the baseline incidence of cancer, as this represents an average of the reported values. The model illustrated 12 possible strategies for BE, ranging from endoscopic surveillance every year, with esophagectomy for high-grade dysplasia (the most aggressive strategy) through to no surveillance and esophagectomy only for cancer (the least aggressive). For the purposes of this article, Provenzale extrapolated her existing model to children ages 5, 10, and 15 years. Table III shows the results when the model is applied to 5-year-olds. For a 5-year-old patient with BE and no dysplasia or cancer, even the most aggressive strategy results in a remaining quality-adjusted life expectancy of only 52 years. Similarly, the most aggressive strategy in a 15-year-old results in a remaining life expectancy of only 48 years. The possible reasons for this relatively small gain in remaining life expectancy may be that: (a) the incidence of cancer in childhood Barrett's is unknown; it may be lower earlier in life, perhaps rising into the 1-2% range only in mid-life and/or (b) the younger the patient, the greater the cumulative (i.e. life-time) incidence of cancer.
Approach 2: modify existing model, using "childhood parameters"
In designing a model for children, it may be more appropriate to use a changing incidence of cancer, i.e. low early, rising with age, as in ulcerative colitis . The model should also include modifiers for those co-morbidities affecting longevity, i.e. neurologic impairment and cystic fibrosis. In addition, more data are needed regarding the biological behavior of BE, i.e. duration in years from no dysplasia through to high-grade or cancer. A modified model, using parameters more appropriate for children has not yet been constructed.
Approach 3: use anecdotal data
Some information is available regarding the biologic behavior of BE. Reid et al.  performed prospective endoscopic surveillance of a cohort of 62 adults using flow-cytometry and histology and monitoring the progression to malignancy in BE. They noted that 37 of 62 patients who were dysplasia-negative showed no progression over a mean of 34 months (range 12-62 months), and the time from dysplasia-negative to cancer was
55 months and 66 months respectively in two patients. They noted that progression from Barrett's specialized metaplasia negative for dysplasia to cancer "appears to require substantial time intervals".
Adenocarcinoma of the esophagus has been documented in 10 children or young adults . Nine of the 10 had Barrett's specialized metaplasia documented at diagnostic endoscopy in 5 and on the resected specimen in 4. The youngest documented case was 11 years, this child having presented with a mass. From the paucity of reports in the literature, and the absence of encounters with adenocarcinoma in the experience of most pediatric gastroenterologists and surgeons, one may reasonably conclude that esophageal adenocarcinoma rarely develops in children. In using anecdotal data, it is also important to note that the youngest child reported with Barrett's specialized metaplasia was 5 years of age , and that the progression from BE dysplasia negative to cancer appears to take some years.
Using these data, the following recommendations for screening children with BE can be made:
1) consider endoscopic surveillance only in those children where Barrett's specialized epithelium is present;
2) after the initial diagnostic endoscopy and biopsies have revealed the presence of Barrett's specialized epithelium, treat with potent acid suppression for 3 months; this will remove any exudate camouflaging landmarks, and decrease the likelihood of inflammation leading to an erroneous diagnosis of some degree of dysplasia. Then perform baseline screening as follows:
- take multiple biopsies from any suspicious lesions, specifically ulcers still unhealed after 3 months acid suppression, or raised lesions,
- have the biopsies reviewed by at least 2 pathologists experienced in this area. If absence of dysplasia is confirmed:
a) below 10 years of age, perform endoscopic surveillance with biopsies every 3-4 years,
b) 10-20 years of age, perform endoscopic surveillance every 2-3 years with the biopsy protocol above;
3) modify surveillance frequency on a case-by-case basis if a major comorbidity is present, specifically life-limiting neurologic impairment or cystic fibrosis.
2. Reid BJ, Haggitt RC, Rubin CE. Barrett's esophagus and esophageal adenocarcinoma. In: Hill L, et al, eds. The esophagus: medical and surgical management. Philadelphia: WB Saunders Company 1988:157-166.
22. Skinner DB. The incidence of cancer in Barrett's esophagus varies according to series. In: Giuli R, McCallum RW, eds. Benign lesions of the esophagus and cancer: answer to 210 questions. New York: Springer Verlag, 1989:764-765.
24. Reid BJ, Blount PL, Rubin CE, et al. Flow-cytometric and histological progression to malignancy in Barrett's esophagus: prospective endoscopic surveillance of a cohort. Gastroenterology 1992;102:1212-1219.