THE USE OF SPECIAL STAINS IN THE DIAGNOSIS OF BARRETT ESOPHAGUS AND BARRETT

DYSPLASIA: RECOMMENDATIONS FROM THE RODGER C. HAGGITT GASTROINTESTINAL

PATHOLOGY SOCIETY

Amitabh Srivastava1, Henry Appelman2, Jeffrey D. Goldsmith, MD3, Jon M. Davison4, John

Hart, MD5 and Alyssa M. Krasinskas, MD6

1Department of Pathology, Brigham and Women's Hospital, Boston, MA; 2Department of

Pathology, University of Michigan, Ann Arbor, MI; 3Children’s Hospital Boston, Beth Israel

Deaconess Medical Center, and Harvard Medical School, Boston, MA; 4Depatment of Pathology,

University of Pittsburgh School of Medicine, Pittsburgh, PA; 5Department of Pathology,

University of Chicago Medical Center, Chicago, IL; and 6Department of Pathology and

Laboratory Medicine, Emory University, Atlanta, GA.

The incidence of esophageal adenocarcinoma continues to increase in the United States,

albeit at a slower rate than that recorded in the mid-1990’s.[1] The estimated number of new

cases in the United states for 2015 is 16,980, with an estimated 15,590 deaths, making

esophageal adenocarcinoma the 7th leading cause of death in men in 2015.[Cancer Facts &

Figures 2015, http://www.cancer.org] Since Barrett esophagus (BE) is a known risk factor for

the development of esophageal adenocarcinoma, various screening and surveillance programs

for BE are currently utilized in an attempt to identify individuals at risk for progression to

cancer. Pathologists play a critical role in these screening and surveillance protocols, as they

confirm the diagnosis of BE and assess for the presence and grade of dysplasia. Currently,

morphology is the gold standard for the diagnosis of BE and BE-associated dysplasia. However,

these diagnoses are not always straightforward. Numerous studies have attempted to identify

ancillary stains that may help pathologists confirm the presence of BE and/or dysplasia, and to

predict the progression to adenocarcinoma. In this review, representatives of the Rodger C.

Haggitt Gastrointestinal Pathology Society evaluated the published literature and provide

recommendations regarding the use of ancillary stains in the diagnosis of BE and BE-associated

dysplasia.

Background

Definition of BE

In 1950, Norman Barrett, a British surgeon, published a learned discourse on peptic

ulcer of the esophagus, in which he described the presence of a columnar-lined lower

esophagus; he ascribed this phenomenon to the stomach being pulled up into the chest as the

result of a congenital short esophagus.[2] Intestinal metaplasia (IM) of the distal esophagus

was first described in 1953 by Allison and Johnstone who published seven cases of clinical reflux

esophagitis with a columnar-lined lower esophagus, two of which had goblet cells.[3] In this

study, the authors were the first to speculate that this columnar lining was an acquired

metaplasia secondary to reflux of acidic gastric contents.

Beginning in the mid-1970s, it gradually became apparent that BE was a precursor, and

thus a risk factor, for adenocarcinoma.[4-6] It was noted that the prevalence of BE and

adenocarcinoma of the gastroesophageal junction (GEJ) rapidly increased in some parts of the

world, including the United States, from the 1960s through the 2000s. At the same time,

squamous cell carcinoma, the most common type of esophageal carcinoma worldwide,

gradually decreased in prevalence in Western societies, including the United States, and was

eventually replaced by adenocarcinoma as the most prevalent esophageal carcinoma.

Several guidelines for the diagnosis and management of BE have been published over

time and in different countries (Table 1). In 1998, the American College of Gastroenterology

(ACG) created the first practice guidelines for the diagnosis, surveillance, and therapy of BE.[7]

These guidelines required IM to be present for a diagnosis of BE. In 2006, the British Society of

Gastroenterology (BSG) issued guidelines for the diagnosis and management of BE.[8] BE was

defined as an endoscopically apparent area above the GE junction suggestive of BE that was

supported by the finding of columnar lined esophagus on histology. Under these guidelines, IM

was not a requirement for diagnosis due to the presumption that inadequate sampling may

result in missed IM.

In 2011, the American Gastroenterological Association (AGA) Institute Medical Position

Panel issued guidelines on the definition of BE.[9] This group defined BE as “the condition in

which any extent of metaplastic columnar epithelium that predisposes to cancer development

replaces the stratified squamous epithelium.” This was the first definition to specify the

association of BE with cancer. As with the ACG definition, the AGA definition requires the

presence of IM due to the assertion that the presence of goblet cells is the risk factor that

predisposes to malignancy. This definition did leave an open question as to the potential risk

of columnar mucosa devoid of IM. However, this definition mentioned that the absence of IM

did not necessitate ongoing surveillance.

In 2014, the BSG published modified guidelines that required an endoscopically

abnormal segment of esophagus that is greater than or equal to 1 cm in length. [10] As in the

2006 guideline, this BSG definition did not require the presence of IM, but did note that the lack

of IM in a segment of columnar-lined esophagus did show a lower risk of malignant

transformation compared to those that contained IM. As in the 2011 AGA Guideline, the BSG

noted that patients without IM may not require ongoing cancer surveillance.

Finally, in January 2016, the ACG issued a modified definition of BE: “BE should be

diagnosed when there is extension of salmon-colored mucosa into the tubular esophagus

extending ≥1 cm proximal to the GEJ with biopsy confirmation of intestinal metaplasia”. The

main difference between this definition and the previous ACG definition from 2008[11] is that

there is now a minimum required length of columnar mucosa identical to that in the BSG

guidelines.[12]

In summary, all authorities agree that BE is a metaplastic phenomenon that is a risk

factor for the development of adenocarcinoma of the GEJ or distal esophagus. All major

professional organizations require an endoscopically abnormal segment of distal esophagus

that currently must exceed 1 cm in length and contains columnar epithelium on biopsy. The

necessity of IM for the diagnosis of BE varies; IM is required in the United States and parts of

Europe, whereas IM is not necessary in The United Kingdom and Japan.

Definition of BE: Endoscopically apparent abnormal mucosa extending ≥1 cm proximal

to the GEJ, and, in the United States, with biopsy confirmation of intestinal metaplasia.

Challenges in the diagnosis of BE

As discussed above, the diagnosis of BE is determined by two factors: (1) endoscopic

identification of columnar metaplasia in the esophagus; (2) histologic confirmation of columnar

epithelium and/or IM.[13] In the absence of inflammation, the salmon pink and velvety

appearance of columnar mucosa is readily differentiated from whitish, pearlescent squamous

mucosa, but mucosal injury in the esophagus due to reflux may obscure this distinction. In

order to identify columnar metaplasia in the distal esophagus, the endoscopist must be able to

identify the GEJ. The proximal extent of the gastric folds[14] and the distal extent of the

longitudinal palisade vessels of the distal esophagus[15] are two landmarks that have been

proposed. The former landmark is preferred by most gastroenterologists in North America and

Europe, while Japanese gastroenterologists favor the latter landmark.[13] The AGA and BSG

guidelines for the diagnosis of BE recommend the proximal extent of the gastric folds due to

moderately superior reproducibility[16] and historical use of this landmark in most Western

studies.[9, 10]

To measure the BE segment length, the endoscopist must localize these landmarks with

respect to the squamocolumnar junction, but localization may be compromised by esophagitis,

peristalsis, patient respiration, degree of insufflation and anatomic variation of the vasculature

of the distal esophagus.[9, 10] In patients with a hiatal hernia, the endoscopist must also

identify the diaphragmatic hiatus in relation to the GEJ in order estimate the extent of BE.[17]

The Prague C&M criteria were proposed to standardize the endoscopic diagnosis and grading of

columnar metaplasia in the distal esophagus in an effort to improve diagnostic

reproducibility.[18] This study specified the anatomic landmarks and criteria for measuring the

circumferential (C) and maximal (M) extent of columnar metaplasia in the esophagus. Reported

kappa scores in the external validation study were excellent for BE≥1 cm in length (0.72), but

substantially worse for BE<1cm in length (0.21).[18] This may be the reason that the recent BSG

and ACG definitions both require a minimum length of one centimeter for a diagnosis of BE.

Challenge in the diagnosis of BE: The GEJ can be difficult to identify in some patients.

Esophageal columnar metaplasia is an amalgam of several types of glandular mucosa

(including cardiac-type, fundic-type and hybrid cardio-fundic glands).[19] In addition, cells

showing multiple lines of differentiation have been identified in the metaplastic glandular

epithelium using electron microscopy and histochemical stains, including small intestinal and

colonic-type goblet cells, intestinal surface epithelial cells, gastric-type foveolar cells,

intermediate cell types, and Paneth cells.[20, 21] Pancreatic acinar metaplasia/heterotopia is

frequently seen in biopsies from the GEJ, but has no correlation with the presence of BE.[22] In

spite of this cellular and glandular heterogeneity, goblet cells are identifiable on routine

hematoxylin and eosin stained sections in the vast majority of cases (Figure 1A). Goblet cells

can be differentiated from distended, goblet-shaped surface mucous cells (‘pseudogoblet cells’)

based on their distinctive (grey-blue) staining characteristics of the mucin vacuole and their

distribution in the epithelium: pseudogoblet cells are often clustered while true goblet cells are

usually single and interspersed among foveolar cells (Figure 1B). The utility of ancillary stains to

detect true goblet cells is discussed later.

Challenge in the diagnosis of BE: Distinguishing true goblet cells from pseudogoblet

cells.

Prior to the current ACG and BSG definitions of BE as requiring ≥1 cm of columnar-lined

esophagus proximal to the GEJ, one major interpretive challenge for clinicians and pathologists

was the distinction of ultrashort segment BE (BE <1cm in length) from IM of the proximal

stomach immediately distal to the anatomic GEJ in biopsies taken near the GEJ. IM is a common

finding in biopsies from the GEJ, reported in up to 18% of patients without endoscopically

evident BE.[23] However, because BE is now defined by both the presence and extent of

columnar metaplasia/IM (≥1 cm proximal to the GEJ), pathologists should be descriptive by

including only the term “IM” in their diagnoses of biopsies obtained from the GEJ or unknown

site (ie, “distal esophagus”) if more proximal biopsies were not taken or do not show IM, rather

than assign a definitive diagnosis of BE. If pathologists want to comment on the presence or

absence of BE in the biopsy, a comment such as the following can be added: “If the biopsy was

taken from the tubular esophagus and there is endoscopic evidence of esophageal columnar

metaplasia extending at least 1 cm proximal to the gastroesophageal junction, then the

presence of intestinal metaplasia confirms the diagnosis of Barrett Esophagus.”

In a subset of cases, there are histologic clues to the esophageal origin of biopsies taken

near the GEJ, including multilayered epithelium, “buried” IM (squamous epithelium overlying

glandular epithelium with IM), and esophageal glands/ducts.[21, 24, 25] In the study by

Srivastava et al., these findings were almost exclusively seen in biopsies from patients with

endoscopic BE compared to controls.[24] Multilayered epithelium (MLE) is a distinctive

epithelial type comprised of squamoid basal (reserve) cells with surface columnar cells usually

at the mucosal surface that can mimic true goblet cell metaplasia.[26] Prospective studies

involving targeted biopsies taken from the GEJ[25, 27] and squamocolumnar junction[28] found

MLE in 10-36% of cases in a background of cardia-type or cardiofundic-type mucosa.[25] It was

strongly associated with the presence of BE [27, 28] and endoscopic evidence of

gastroesophageal reflux.[27] Various authors have suggested that this finding is indicative of

esophageal origin of the epithelium based on its phenotypic similarity and association with

esophageal gland ducts [25-27] and suggest that it may represent an early transitional form of

columnar metaplasia in the esophagus.[27, 28]

Challenge in the diagnosis of BE: Intestinal metaplasia at the GEJ is no longer

considered ultrashort BE. Hence, pathologists should not make a diagnosis of BE when

intestinal metaplasia is present in biopsies taken from the GEJ, as the current definition

requires the Barrett segment to be ≥1 cm in length.

There are no convincing data that cardia IM is associated with an increased risk of

cancer. Cardia and GEJ IM is more frequently associated with Helicobacter pylori gastritis[29],

while BE is more often associated with gastroesophageal reflux.[29, 30] In studies that compare

these two groups of patients, there are low rates of progression to high grade dysplasia (HGD)

or adenocarcinoma in patients diagnosed with BE, but progression was not seen in patients

with IM of the cardia who undergo surveillance endoscopy with no HGD at baseline.[30, 31]

Consequently, the BSG and ACG recommend against endoscopic surveillance of cardia IM and

also recommend against taking biopsies of a normal appearing GEJ.[10] This recommendation

was affirmed by an international consensus of BE experts in a report sponsored by the AGA.[17]

Dysplasia in BE

BE-associated dysplasia has been classified into four categories: (1) Negative for

dysplasia (ND), (2) indefinite for dysplasia (IND), (3) low grade dysplasia (LGD), and (4) HGD.

While an international consensus definition of each of these categories has yet to be created,

the following are histologic criteria that are generally agreed upon by GI pathologists:[32]

- ND (Figure 1B): This epithelium shows a simple columnar epithelium with small nuclei

and abundant apical cytoplasm. The nuclei are round and usually occupy <25% of the

cell surface area; nuclear membranes are smooth. Nuclei are uniformly present at the

basal aspect of the cell and show extreme monotony from cell to cell. Nuclear to

cytoplasmic ratio is generally higher at the base of the pits and shows a progressive

decrease towards the luminal surface. The architecture of non-dysplastic epithelium

shows a well-organized, uniform distribution of pits.

- LGD (Figure 2A): Cytologically, nuclear changes include increased nuclear to cytoplasmic

ratio, hyperchromasia, nuclear membrane irregularities and variation in normal

chromatin distribution with intact nuclear polarity; significant nuclear anisocytosis is

absent. These cytologic changes should extend to the luminal surface. Architecturally,

only minimal gland crowding should be present.

- HGD (Figure 2B): Nuclear changes are more profoundly atypical than LGD with loss of

nuclear polarity, marked hyperchromasia, atypical mitotic figures, and significant

nuclear anisocytosis that extend to the luminal surface. Architectural features of HGD

include significant gland crowding, cribriform architecture, and dilated glands containing

necrotic debris. Single cell invasion of the lamina propria, solid sheets and/or cords of

neoplastic cells (i.e. the ‘never-ending gland’ pattern) should not be seen in HGD and

suggest intramucosal adenocarcinoma.

- IND: This category should be utilized when there are histologic features that are

suspicious, but not diagnostic of dysplasia. This is most often a result of some amount

of neutrophilic inflammation within or adjacent to the abnormal epithelium. Cytologic

atypia that is confined to the crypts/glands without extension to the surface epithelium

is also generally regarded as IND. Some studies have shown that patients with cytologic

features of dysplasia limited to the crypts (crypt dysplasia) may behave equivalently to

patients with conventional dysplasia;[33] however, this has not been validated.

Moreover, the majority of patients with crypt dysplasia were reported to have

conventional dysplasia elsewhere at the same endoscopy,[33] limiting the practical

utility of this morphological finding.

The diagnosis of BE-associated dysplasia is a marker of increased risk of progression to

adenocarcinoma. Progression rates of LGD range from 2-28% annually, while progression of

HGD ranges from 6-60% per year.[34-40] The aggregate rate of annual progression to either

HGD or adenocarcinoma in groups of patients with IND is 0.86% to 1.4%.[41-44] The large range

of rates of progression are very like due to the fact that the diagnosis of both LGD and IND are

plagued by significant interobserver variability with Cohen’s Kappa statistics of approximately

0.4 (moderate agreement); interobserver agreement tends to better at the ends of the

interpretive spectrum (i.e. negative for dysplasia and HGD/intramucosal adenocarcinoma) and

rather poor for LGD/IND.[32, 45] Nevertheless, dysplasia remains the gold standard for

predicting risk of neoplastic progression in BE. The current standard of care prescribes that

patients with LGD receive increased surveillance; patients diagnosed with HGD require

endoscopic interventions, such as radiofrequency ablation, endoscopic mucosal resection of

visible lesions, and, in selected cases, esophagogastrectomy. Because of the important

management implications for a diagnosis of dysplasia coupled with the interobserver variability

in the interpretation of dysplasia, it is recommended that any new diagnosis of dysplasia be

confirmed by two pathologists, including at least one with expertise in gastrointestinal

pathology.[10, 12]

Dysplasia in BE: BE-associated dysplasia is a marker of increased risk of progression to

adenocarcinoma. Any new diagnosis of dysplasia should be confirmed by two

pathologists, including at least one with expertise in gastrointestinal pathology.

Methods for Developing the Recommendations

In 2014, the Rodger C. Haggitt Gastrointestinal Pathology Society (GIPS) asked the

membership via a survey what practice topics members would like to see addressed by the

GIPS. The GIPS executive team chose this topic from among the top three responses that

would best lend itself to the format of a GIPS recommendation. A senior author was identified,

who then created a team of pathologists with interest and expertise in BE and who were willing

to co-author the manuscript. The authors reviewed the literature and referenced papers that

addressed the two chosen questions: (1) Are there scientific data to support the statement that

special stains are necessary to diagnose BE, and (2) Are there scientific data to support the

statement that special stains are necessary to diagnose dysplasia in BE? After analyzing the

literature, the authors summarized the data to formulate the recommendations. The

recommendations were vetted by members of the Executive Committee of GIPS and then

reviewed by a separate group of pathologists with interest and expertise in the field, none of

whom were authors of the manuscript or members of the Executive Committee. After final

acceptance by the GIPS Executive Committee, the recommendations were made available to

the GIPS membership through the society website (http://usgips.com) for a one-month

comment period. Submitted comments were analyzed for validity and incorporated into the

recommendations prior to submission for publication.

Are ancillary stains needed to diagnose BE?

As noted above, in the United States and parts of Europe, the identification of goblet

cells by the pathologist is necessary for a diagnosis of BE. In the vast majority of cases, true

goblet cells can be easily and accurately recognized by their characteristic morphology in

routine H&E sections.[46] However, a common mimic of goblet cells, named “pseudogoblet”

cells, occurs frequently in inflamed cardia-type epithelium. These consist of scattered

mucinous cells with distended cytoplasmic mucin droplets that can be confused with goblet

cells.[47] With experience, these so-called “pseudogoblet” cells can be separated from true

goblet cells in H&E sections. If there is any doubt, an Alcian blue (at pH 2.5)/PAS stain can be

utilized to confirm the H&E impression: The columnar mucinous cells of the foveolar epithelium

contain neutral mucins that stain red or magenta, while the acidic mucin in goblet cells stain

bright blue (Figure 3A).[48, 49]

A pitfall of using an Alcian blue stain alone is the detection of so-called “columnar blue

cells.” Not infrequently, columnar cells without the goblet shaped mucin droplet of true goblet

cells will stain blue with the Alcian blue pH 2.5 stain (Figure 3B). Some investigators have

suggested that these columnar blue cells represent an earlier phase of IM and should be

regarded as BE, or at least a precursor of BE.[48, 50-52] However, subsequent studies have

demonstrated that Alcian blue positive columnar cells in cardiac-type mucosa are sometimes

present in patients without BE.[47] One small prospective study that addressed the utility of the

Alcian blue stain in BE determined that patients with Alcian blue positive pseudogoblet cells

(but no true goblet cells) experienced no increased risk of future dysplasia.[53] There is one

study that demonstrates that columnar blue cells containing sulfomucin (brown staining in the

high iron diamine/Alcian blue pH 2.5 stain) were more commonly present in the metaplastic

esophageal mucosa that contained goblet cells than those that contained only sialomucins

(blue staining with the Alcian blue pH 2.5 stain).[50] The authors suggest that this type of

columnar blue cell may in fact represent an incomplete form of IM, but there are no

prospective studies to address the clinical significance of this finding. Hence, the utility of the

Alcian blue pH 2.5 stain alone is limited by these columnar blue cells that must not be confused

with true goblet cell metaplasia.

Since the detection of goblet cells plays a central role in the diagnosis of BE, there has

been some interest in goblet cell distribution and number. One well-designed study indicated

that at least eight biopsies from the endoscopically defined Barrett segment were necessary to

be confident that goblet cells were in fact present by H&E examination.[46] In this study, Alcian

blue/PAS stains were also performed to identify goblet cells missed in the initial H&E sections.

Obtaining the Alcian blue/PAS stain led to only an additional 5.6% of patients being diagnosed

as having BE. However, this study did not address whether simply performing additional H&E

sections would have led to a similar increase in the detection of goblet cells. Another study that

did look at the effect of obtaining deeper H&E levels found that goblet cells were detected in

4.7% of cases that had no goblet cells in the original 3 levels of serial sections.[54] Obviously,

identification of columnar mucosa devoid of goblet cells can be accomplished by H&E stain

alone, and reflexive use of AB-PAS staining of squamous epithelium in order to "exclude

Barrett's epithelium" is unjustifiable.

Recommendations for the use of Alcian blue (at pH 2.5) and/or PAS stains to diagnose

BE:

Should not be used reflexively on all esophageal biopsies because goblet cells

are almost always identifiable on routine H&E stained sections.

Alcian blue/PAS stains can be utilized to distinguish pseudogoblet cells from

true goblet cells if there is a doubt on the H&E section.

A number of studies have been performed analyzing mucin glycoprotein subtypes by

immunohistochemistry in Barrett epithelium. MUC2 is regarded as a reliable marker of an

intestinal epithelial phenotype, while MUC5AC is expressed in normal gastric foveolar epithelial

cells. Goblet cells in BE are consistently reactive with the MUC2 antibody and non-goblet

columnar cells are reactive with the MUC5AC antibody.[55] Interestingly, MUC2 is also

expressed in a proportion of the non-goblet cell columnar epithelium in BE, particularly in the

epithelium adjacent to goblet cell containing Barrett mucosa.[56] It has been suggested that

MUC2+ non-goblet columnar cells indicates an earlier stage in intestinal differentiation and may

be used as a surrogate marker of goblet cells in Barrett mucosa when goblet cells are not

identified in endoscopic biopsies as a result of sampling issues.[56] However, no study has been

performed to determine whether MUC2 immunostains could be used to predict the future

development of goblet cells. In fact, none of the published studies have advocated the use of

any of the mucin glycoprotein immunostains as diagnostic or prognostic markers.[57-62]

Recommendation for the use of mucin glycoprotein immunostains to diagnose BE: Not

indicated.

Markers of intestinal phenotype have also been studied in immunohistologic studies,

including CDX2, Das-1, villin, and HepPar-1. CDX2 is an important nuclear transcription factor

that promotes epithelial intestinal differentiation during normal development.[63] The Das-1

antibody recognizes an antigen specific to a colonic epithelial cell protein, and is regarded as a

sensitive and specific marker of IM.[64] Likewise, villin and HepPar-1 proteins are highly

expressed in intestinal epithelia and are regarded as markers of intestinal differentiation in

metaplastic epithelia.[65, 66] Many studies have demonstrated reactivity with these markers in

non-goblet cell columnar epithelium in patients with BE.[67-72] In most studies, these markers

are more often present in non-goblet cell columnar epithelial component in patients who also

have goblet cell containing epithelium, suggesting that perhaps they are markers of an earlier

phase of intestinal differentiation. However, none of these studies explored their use as

predictors of the future development of goblet cell containing metaplastic epithelium in the

esophagus. There is one study that reported that immunoreactivity with either CDX2 or SOX9 (a

transcription factor downstream of the sonic hedgehog signaling pathway) in patients with only

non-goblet cell metaplastic epithelium in the esophagus was predictive of the future detection

of goblet cell containing BE with a sensitivity of 85% and specificity of 85%.[73] Because the

vast majority of studies that have evaluated the risk of neoplastic progression in BE are based

primarily on histologic identification of IM,[74] the clinically accepted definition of IM is based

on the identification of goblet cells on routine histology rather than immunophenotypic

evidence of IM.

Recommendation for the use of markers of intestinal phenotype (CDX2, Das-1, villin,

and HepPar-1) to diagnose BE: These stains may be markers of an earlier phase of

intestinal differentiation, but none have shown to be predictors of the future

development of goblet cell containing BE. Thus, the use of these stains is not currently

indicated.

Recommendation for the use of CDX2 and SOX9 immunostains to diagnose BE: These

stains show promise, but too few studies have been performed to date. Thus, a

recommendation cannot be made at this time.

Because there are no entirely reliable endoscopic landmarks to precisely identify the

GEJ, even with the new definition of BE, there may continue to be some degree of uncertainty

regarding the distinction between BE and cardia IM. As discussed above, mucin histochemical

stains reveal overlapping patterns of mucin glycoprotein expression in gastric and esophageal

IM. However, there are some differences in the expression of cytokeratin subtypes between

the two conditions. Barrett epithelium most often exhibits strong cytoplasmic CK7 reactivity of

both the surface epithelium and glandular compartment, with weak reactivity for CK20

confined to the surface epithelium. In contrast, gastric IM less frequently demonstrates this

pattern of reactivity.[75-77] However, this characteristic pattern of expression in BE is not

specific enough to utilize as a diagnostic marker, and is difficult to identify in suboptimally

oriented endoscopic biopsy specimens.[68, 78, 79]

Recommendation for the use of CK7 and CK20 immunostains to distinguish IM of the

gastric cardia from BE: These stains are not specific enough to be utilized as

diagnostic markers. Also, in suboptimally oriented biopsies, these stains can be

difficult to interpret. Thus, their use is not indicated.

Are ancillary stains needed for dysplasia diagnosis or risk stratification in patients with BE?

It is well established that the vast majority of patients with known BE do not develop

esophageal adenocarcinoma on follow up, and that most patients diagnosed with esophageal

adenocarcinoma do not have a prior diagnosis of BE.[80] However, the current standard of care

for patients diagnosed with BE is periodic surveillance for early detection of neoplasia. As noted

above, a morphological diagnosis of dysplasia remains the gold standard for identification of

patients at increased risk for the development of cancer. The surveillance biopsy diagnosis

guides the recommended surveillance intervals in these patients. The recent ACG guidelines

recommend a 3-5 year surveillance interval for patients with non-dysplastic BE, a 12 month

surveillance for an IND diagnosis, and either definitive therapy or a 12 month surveillance for

patients with LGD. Surveillance alone is not recommended for HGD unless significant

comorbidities preclude definitive endoscopic therapy. [12]

In order for an ancillary stain to be helpful in the clinical care of patients with BE, the

stain should ideally either: (1) Aid in the diagnosis of dysplasia, and/or (2) Stratify risk in

patients with BE. The morphological diagnosis of dysplasia lacks interobserver reproducibility,

particularly for the IND and LGD categories. Thus, any special stains that improve interobserver

reproducibility and thereby standardize dysplasia diagnosis would be immensely beneficial to

patient care. Regarding risk stratification, there is wide variation in the reported rates of

progression to adenocarcinoma for BE patients with LGD and HGD. Part of this variation is

obviously related to the problem of interobserver reproducibility mentioned above. Another

reason for this discrepancy may be the presence of heterogeneous biological subsets within

each morphological dysplasia diagnosis category. An ancillary test that can help identify high

risk subsets of patients within each group could be useful in guiding surveillance

recommendations. Finally, given the lack of any progression in the vast majority of BE patients,

an ancillary test that can identify the truly low risk subset of patients whose surveillance can be

extended beyond the usual 3-5 years for non-dysplastic BE would have beneficial consequences

to cost of care for these patients.

Studies analyzing the utility of ancillary immunostains in BE and BE-associated dysplasia

fall into three broad categories. 1) The first and largest group includes those that have used

morphology as the gold standard to classify biopsies and then determined the prevalence of the

marker in question in each morphologically defined category. Progressively increasing

prevalence of positivity in the HGD and carcinoma categories has been then taken as evidence

for the utility of the stain being a marker of high risk in BE. The obvious limitation of such

studies is that they cannot improve upon morphology because they use it as the gold standard

for determining biomarker positivity in the non-dysplastic and dysplastic groups. Moreover,

most of these studies do not directly assess the association of biomarker positivity with

subsequent disease progression. 2) A case control design overcomes some of these limitations

as it can be used to assess the odds of biomarker positivity in BE patients who progress to HGD

or cancer on follow up (“cases”) in comparison to those who remain free of neoplasia during

follow up (“controls”). HGD is considered a relevant end point in some studies because patients

usually exit surveillance and enter a phase of definitive treatment when they are diagnosed

with HGD. However, the use of HGD as an end point is problematic because the distinction

between low and HGD is not always straightforward. What is persistent LGD for some

pathologists may be progression to HGD for others. Another issue with case control studies is

the number of index and surveillance biopsies that need to be stained since both progressors

and non-progressors are likely to have had multiple surveillance biopsies during follow up.

Staining all available biopsies may be feasible in a research study but is not practical in a routine

clinical setting unless very strong performance characteristics can be shown for a given

biomarker that clearly separates the truly high risk patients from the low risk ones. 3) The last

category of study design includes those that have evaluated biomarker status at index

examination and then determined the likelihood of disease progression on follow up. The

studies on ancillary stains described below include all three study designs.

Ancillary stain: p53

TP53 is a tumor suppressor gene that prevents genomic mutations. In normal cells, the

level of p53 protein is low, but when DNA damage or other stress occurs, p53 is upregulated,

which results in growth arrest, DNA repair and apoptosis. TP53 gene mutations are highly

prevalent in esophageal adenocarcinoma as well as HGD, but are uncommon in non-dysplastic

BE. Mutations often lead to increased nuclear accumulation of a faulty protein, but in the case

of truncating mutations, p53 can be completely lost. Thus, faint scattered p53 positivity within

nuclei correlates best with a wild-type gene status whereas either intense nuclear positivity or

complete absence ("null pattern") of staining correlate best with TP53 mutations (Figure 4). The

latter null pattern of staining has been recognized only in recent studies and was largely

ignored in earlier immunohistochemical studies on p53 expression in BE.[81, 82] Abnormal

immunohistochemical expression of p53 can be detected across the entire morphological

spectrum of BE but the prevalence is lower in non-dysplastic BE. [83]

Use in the Diagnosis of Dysplasia. It has been proposed that p53 immunohistochemistry

may aid in improving diagnostic reproducibility in the diagnosis of dysplasia and in risk

stratification in BE. A summary of studies on p53 and the diagnosis of dysplasia in BE is included

in Table 2. One of the recommendations put forth by the 2014 BSG guidelines on the diagnosis

and management of BE states: “the addition of a p53 immunostain to the histopathological

assessment may improve the diagnostic reproducibility of a diagnosis of dysplasia in BE and

should be considered as an adjunct to routine clinical diagnosis” (grade B

recommendation).[10] This recommendation is largely based on studies that focused on the

progression to HGD or adenocarcinoma (further discussed below) rather than pure histologic

evaluation. One study that supported this recommendation examined the utility of p53

immunostaining in improving observer reproducibility in the diagnosis of BE-associated

dysplasia; in this study, biopsies from 143 patients with BE dysplasia and 35 negative for

dysplasia were stained with p53.[84] The conclusions from this study were that use of p53

improves observer reproducibility and disease progression was better predicted by

incorporating p53 immunohistochemical findings into the morphological diagnosis. Pathologists

integrated the p53 findings into their morphologic diagnosis but no details were provided in the

study regarding how exactly this evaluation was performed. Scoring p53 immunohistochemistry

may itself be subject to interobserver variability. In fact, in the study cited above, strongly

positive abnormal glands were considered p53 positive while p53 negative cases were either

“weakly positive or completely negative.”[84] As mentioned earlier, this latter pattern of

staining correlates with truncating mutations and should be considered an aberrant pattern of

staining. A recent study of 72 patients by the same group evaluated reproducibility of p53

interpretation by 10 pathologists from four institutions.[82] P53 immunostain was interpreted

as “significant” in the presence of strong or absent staining and as “not significant” when

neither of the two patterns were present or findings were equivocal. The authors reported

better interobserver reproducibility for p53 interpretation (average kappa value=0.6) when

compared to morphological diagnosis of dysplasia using the four categories in the Vienna

classification (average kappa value=0.3). Since the p53 staining was evaluated in two categories

and the morphological diagnosis split across four, the higher kappa value for p53 is not

surprising. In fact, when the morphological diagnosis was grouped into two categories of

definite dysplasia versus no dysplasia, the average kappa value was 0.55, which is not much

different from the p53 interpretation. Based on the results of this study, the authors suggested

that including p53 can be a useful aid in dysplasia diagnosis but cannot be used to separate LGD

from HGD. Abnormal p53 was not present in any of the 14 non-dysplastic BE cases included in

the study though others have reported a prevalence of about 10% in this category (see more

data below). Moreover, 21/22 cases of LGD showed abnormal p53 staining, even though

molecular studies indicate TP53 mutations generally to be a late event and to be present in only

about 30% of BE-associated LGD.[85] Despite the current widespread use of p53 IHC, we

believe that additional studies are needed to develop and validate precise criteria before p53

staining can be fully endorsed and incorporated into the morphological dysplasia diagnosis

algorithm.

Use in Predicting Disease Progression. Immunoreactivity for p53 is the most widely

studied and most promising marker among the many ancillary immunostains that have been

evaluated as markers of high risk for disease progression in BE. A summary of studies on p53

and disease progression in BE is included in Table 2. Some of the earlier studies reported p53

positivity in biopsies IND or with LGD to be a highly sensitive and specific marker of progression

to HGD or adenocarcinoma. In a study of 61 patients, 25 (41%) were diagnosed either as IND or

as LGD and five (20%) of these developed HGD or adenocarcinoma on follow up. Five of nine

patients with p53 positive indefinite or LGD progressed compared to none of the 16 with p53

negative IND or LGD.[86] Others have reported a 88% sensitivity and 75% specificity for p53

positivity in LGD as a marker for progression to HGD or carcinoma; however, this was based on

a very small series of 16 cases in which agreement for a diagnosis of LGD among three

pathologists was present in only 4/16 cases included in the study and 50% patients had no

evidence of dysplasia on follow up. [39] Weston et al studied 48 patients with LGD (26

prevalent, 22 incident) who were followed up for a mean duration of 41.2 months and five of

these progressed to either multifocal HGD (4 patients) or cancer (1 patient).[87] In this study,

p53 positivity was present at index biopsy in the low grade dysplastic focus in 4/31 patients that

regressed to no dysplasia, 3/12 that showed persistent LGD and 3/5 that progressed to HGD or

carcinoma.[87] The limitations of this study include the use of quantitative image analysis,

which may have scored more than just the strong staining nuclei, and the inherent issue in

many studies, the interobserver variability in grading dysplasia (some of the illustrations of p53

positive LGD show round nuclei with loss of polarity suggestive of HGD). A nested case control

study from the Northern Ireland Barrett’s Esophagus Register (NIBR) included 29 cases with

incident adenocarcinoma and 6 with incident HGD that were matched to at least three control

patients with no progression.[88] With p53 positivity scored semi-quantitatively as 0, 1 (<10%;

focal), 2 (10-50%; diffuse) and 3 (≥50%; intense), 32.4% of cases and 11.7% of controls showed

p53 positivity. No increase in risk was observed with focal (<10%) staining, but patients with

diffuse or intense p53 staining were nearly ten times more likely to progress to carcinoma or

HGD compared to those with focal or no staining (OR=9.28; 95% CI, 1.78-48.3), underscoring

the importance of establishing an appropriate definition of “abnormal” p53 expression. In this

study, a third of all patients with incident adenocarcinoma had initial biopsies with diffuse or

intense p53 positivity, suggesting that p53 is not sufficiently sensitive to predict progression to

cancer in all patients at index biopsy as a single marker; however p53 did outperform the initial

histologic assessment by a substantial margin (of note, on the initial histologic assessment,

none of the cases had LGD and one had HGD).[88] A similar study of 27 patients with incident

HGD or adenocarcinoma matched to an equal number of controls with no progression found

moderate (15-40% positive nuclei) but not strong (>40%) p53 overexpression to be an

independent predictor of progression to HGD or adenocarcinoma, when adjusted for age,

gender and a diagnosis of LGD [89]. It is difficult to come up with a plausible biological

explanation as to why moderate but not strong p53 positivity should be a better predictor of

the risk of neoplastic progression in BE. Although it is less common than the “overexpression”

pattern, the null pattern of p53 staining that correlates with truncating TP53 mutations was not

evaluated in the studies mentioned above. Also, the use of semi-quantitative percentage

thresholds for determining p53 positivity is likely to be problematic in routine clinical practice

because strong expression in a small focus of morphologically atypical glands may not be scored

as positive if it represents only a small percentage of the cells on the slide (Figure 5).

A more recent study from the NIBR included 89 BE patients who progressed to

adenocarcinoma or HGD of the esophagus or gastric cardia six months after the initial BE

diagnosis; these patients were matched to 291 controls with no progression to HGD or

adenocarcinoma.[90] P53 positivity, defined as “dark staining adjacent to background with no

staining”, was present in 41% of cases and 28% of controls. P53 positivity was associated with a

significantly increased risk of progression to adenocarcinoma (OR, 1.95; 95% CI, 1.04-3.67) but

was not a significant predictor of combined HGD and adenocarcinoma outcomes (OR, 1.60; 95%

CI, 0.91-2.82). Moreover, in this study, the combination of a morphological diagnosis of LGD,

abnormal DNA ploidy and positivity for Aspergillus oryzae lectin most accurately discriminated

between cases that progressed to HGD or adenocarcinoma and controls that did not.[90] In the

largest study to date, aberrant p53 staining (defined as either overexpression or complete loss

of staining in at least one crypt or gland), was evaluated in 635 patients, 8% of whom

developed HGD (n=35; 6%) or carcinoma (n=14; 2%) on follow up.[91] P53 overexpression was

present in 18% and complete loss was seen in 2% of all biopsies. Aberrant p53 staining was

seen in 11% of biopsies with non-dysplastic BE, 38% of LGD, 83% of HGD and 100% of

adenocarcinomas. Overall, aberrant p53 staining was seen in 49% of cases that progressed to

HGD or carcinoma compared to 14% of controls with no progression on follow up. Importantly,

abnormal p53 expression was shown to be an independent risk factor (when controlled for

dysplasia diagnosis). The positive predictive value for neoplastic progression increased from

15% for a diagnosis of LGD to 33% for a diagnosis of LGD with aberrant p53 staining.

Immunohistochemistry for p53 was performed on tissue from all surveillance endoscopies in

patients with any form of dysplasia while in those with no dysplasia staining was performed in

biopsies from only one randomly chosen surveillance endoscopy. This makes it more likely to

detect aberrant p53 staining in the former compared to the latter group. A surprisingly high

number (223/635; 35%) of patients were diagnosed with LGD at index or follow up endoscopy

in this study and 85% of them showed no progression raising some concern about

morphological thresholds used for a diagnosis of LGD.

To summarize, most studies suggest that abnormal p53 immunohistochemical staining is

a risk factor for progression to HGD or cancer. However, there is a significant body of literature

on longitudinal outcomes for each dysplasia category based on morphology alone. The studies

discussed above do not precisely define the prospective risk of adverse outcomes based on

abnormal p53 immunohistochemistry, which limits the ability of abnormal p53 findings to alter

the recommended surveillance intervals based on morphology alone. Therefore,

implementation of p53 immunostaining as a routine diagnostic adjunct is hindered by several

factors: (1) there is no widely accepted definition of abnormal p53 staining, particularly

regarding minimum extent of staining; (2) there are no guidelines on how many biopsies need

to be tested and whether it should be performed at each surveillance endoscopy; (3) there are

no recommendations for integrating the histologic diagnosis and p53 immunohistochemical

result and its impact on recommended surveillance interval; and (4) it is unknown if patients

will benefit from the addition of p53 testing. There are no data to indicate that patients with

non-dysplastic p53 positive biopsies would benefit from altering the surveillance interval from

the usual 3-5 year interval to 12 months. False positive results in this setting may result in a

substantial burden of unnecessary surveillance. With the exception of the study by Kaye et al,

all of the studies mentioned above treated p53 immunohistochemistry independent of the

morphologic diagnosis. Hence, there is no evidence at this time that a p53 result should modify

the morphologic diagnosis. Additional studies are needed to determine the best definition of

“abnormal” p53 staining and to show how integrating p53 testing into routine practice could

improve patient care.

Other ancillary stains

A number of other biomarkers assessed by immunohistochemistry have been reported

in small numbers of studies. Immunohistochemistry for cyclin D1 was evaluated in a

prospective study of 307 patients, 12 of whom developed adenocarcinoma during follow up.

Index biopsies positive for cyclin D1, defined as any positive nuclei irrespective of extent or

intensity of staining, were significantly associated with progression to adenocarcinoma

(OR=6.85; 95% CI=1.57-29.91). [92] In the same study, p53 positivity was not predictive of

progression to adenocarcinoma [93] (OR=2.99; 95% CI=0.57-15.76). Others have reported that

cyclin D1 is almost universally expressed across the entire BE morphological spectrum but high

expression (>50% cells positive) is seen more often in HGD and adenocarcinoma. Cyclin D1 was

not found to be predictive of progression to HGD or adenocarcinoma in a retrospective nested

case control study by Murray et al (OR=0.81; 95%CI=0.14-4.5).[88] Surface expression of cyclin

A has also been reported to correlate with degree of dysplasia in BE and proposed as a high risk

marker for progression to adenocarcinoma (OR=7.5; 95% CI=1.8-30.7) but this finding remains

to be validated. [94]

AMACR (-Methylacyl-CoA racemase) immunohistochemistry has been reported in

initial studies to be completely negative in non-dysplastic BE and in BE with reactive atypia but

positive in 11-38% of LGD and 64-81% of HGD cases, raising hopes that this could be a useful

marker of dysplasia.[95] It was also suggested in some studies that AMACR positivity in cases

diagnosed as IND may indicate a need for closer surveillance.[96] However, more recent

longitudinal data from a larger series of patients showed weak AMACR positivity in 46% and

strong expression in 3% of biopsies without dysplasia in BE. Strong but not weak AMACR

expression was predictive of progression to adenocarcinoma in this study. [97] It appears,

therefore, that variable AMACR positivity is not entirely specific for dysplasia and may also be

present in non-dysplastic BE.

IMP3 is an oncofetal protein that has been reported to be expressed in >90% of BE-

associated HGD and esophageal adenocarcinomas. [98] However, it is also expressed in 14% of

LGD and 7% of non-dysplastic BE, limiting its utility in the diagnosis of BE-related HGD. Loss of

SOX2 expression was recently reported to be associated with an increased risk of progression to

adenocarcinoma. Complete absence of SOX2 expression was found in 2% of biopsies with no

dysplasia, 28% of biopsies with LGD and 63% of biopsies with HGD.[99] Finally, MUC stains,

MUC5AC and MUC2, CD10 and CDX2 have been used to diagnose foveolar-type dysplasia in

BE[100], a form of dysplasia that has also been reported as gastric or non-adenomatous-type

dysplasia in the literature. There is no consensus at present on the morphological and

immunophenotypic definition of foveolar-type dysplasia and while these stains may be useful in

categorizing dysplasia as adenomatous or foveolar-type, they do not help in the distinction of

BE with and without dysplasia.

Recommendation for the use of special stains to diagnose dysplasia and for risk

stratification in BE:

A diagnosis of dysplasia remains a morphological diagnosis; ancillary stains are

not recommended for diagnosing dysplasia in BE at this time.

Whether p53 immunostaining can be used as a marker for identifying high risk

BE patients requires additional studies before it can be recommended for

routine use.

Summary

After a thorough and critical review of the literature, the Rodger C. Haggitt

Gastrointestinal Pathology Society recommends that morphology should remain the gold

standard for diagnosing BE and BE-associated dysplasia at this time. The Alcian blue stain at pH

2.5 combined with a periodic-acid Schiff (PAS) stain has limited utility in morphologically

challenging cases of BE, particularly in the distinction of pseudogoblet cells from true goblet

cells, and is not indicated as a reflexive test. Other ancillary stains need further study before

they can routinely be applied to the diagnosis of BE and BE-associated dysplasia. We believe

that these recommendations will provide helpful information to pathologists,

gastroenterologists and others involved in the diagnosis and management of patients with BE.

ACKNOWLEDGMENTS

The authors thank Dr. Kenneth P. Batts, Dr. John R. Goldblum, Dr. Gregory Y. Lauwers, and Dr.

Elizabeth Montgomery for their critical review of the manuscript and constructive comments,

and Dr. Marie Robert for her oversight of the project.

Table 1. Comparison of recent written criteria for the diagnosis of Barrett’s mucosa.

ACG: American College of Gastroenterology; BSG: British Society of Gastroenterology; AGA: American Gastroenterological Association

Group Year of publication

Endoscopic requirement

Minimal length required

Goblet cells required

Specifically mentions cancer risk

ACG 2002, 2008 stated no yes no

BSG 2006 stated no no no

AGA 2011 implied no yes yes

BSG 2014 stated Yes, ±1cm no no

ACG 2016 stated Yes, ±1cm yes no

Table 2: Summary of selected studies on use of p53 IHC for BE dysplasia diagnosis and prediction of disease progression Study

(n) Design Objective P53 IHC Scoring Method Major Findings Limitation

Younes M, 1997 (n=61)

RL P53 IHC as a marker for risk of progression in BE

Positive when intensity similar to staining in HT29 colon cancer cell line

At least one biopsy p53+ in 9/61 (15%) 5/9 (56%) p53+ patients progressed to

HGD or EAC on follow up

Median follow up only 26 months All progressors had at least one prior biopsy diagnosed as

LGD or indefinite for dysplasia

Bian Y-S, 2001 (n=30)

CS

Concordance betweenTP53 mutation and p53 IHC expression in BE

Positive when >10% nuclei positive in clusters or scattered throughout the tissue sample

TP53 mutations in 77% HGD, 20% LGD IHC positive in 85% HGD and 71% LGD

Small sample size Used macrodissected samples (n=77) from esophagectomies

with cancer instead of longitudinal surveillance biopsies

Weston AP 2001 (n=48)

RL p53 IHC as predictor of progression risk in BE with LGD

Percentage of positive nuclei in glandular area of interest scored by digitized image analysis

5/48 LGD patients showed progression 3/5 progressors positive for p53 (in 8.3-

58.1% nuclei)

Only 1 definite carcinoma outcome Study restricted to LGD cases only; some LGD illustrations

provided raise suspicion for HGD

Skacel M, 2002 (n=16)

RL Impact of p53 IHC on IOV and progression risk in LGD

"only when staining occurred in the atypical area"

9/16 (56%) LGD p53 positive 7/8 progressors p53 positive compared

to 2/8 non-progressors

Very small sample size Mean follow up only 23 months 4/6 "progressed from LGD to HGD in ≤ 7 months 3 GI pathologists agreed on the LGD diagnosis in only 4/16

cases

Murray L, 2006 (n=210)

Cases (35; 6HGD/29EAC) Controls (175)

RCC

P53 IHC as marker of disease progression in BE

Focal (<10%) Diffuse (10-50%) Intense (>50%)

Increase risk of progression only in cases with diffuse or intense staining

Diffuse or intense staining present in only 32.4% of cases that progressed

Mean follow up only 3.7 years P53 IHC scored as percentage nuclei staining regardless of

intensity

Kaye PV, 2009 (n=143)

RL Impact of p53 IHC on IOV in dysplasia diagnosis and patient outcome

Compared staining intensity in dysplastic foci with background non-dysplastic epithelium

Moderate overall agreement for dysplasia diagnosis

Agreement improved with p53 IHC

Unclear how IHC findings were incorporated into dysplasia diagnosis

Kappa for indefinite for dysplasia (0.06) and LGD (0.37) remained poor even after p53 IHC

Sikkema M, 2009 (n=54)

Cases (27; HGD/EAC) Controls (27)

RCC

Predictive value of p53, Ki67 and aneuploidy as markers of progression risk in BE

Moderate-intense staining considered positive and graded as normal (<15% positive nuclei), moderate (15-40%) or strong (>40%) overexpression

"Moderate" p53 overexpression associated with increased risk for progression to HGD/EAC

Small number of cases; combined HGD and EAC used as outcome of interest

32% biopsies from cases with progression had a prior diagnosis of LGD compared to 5% of control biopsies

Moderate but not strong staining associated with progression; lacks biological plausibility

Bird-Lieberman EL, 2012

(n=380) Cases (n=89; HGD/ EAC)

Controls (n=291)

RNCC

Test biomarkers for risk of progression to HGD/EAC in BE

Positive when strong dark staining seen adjacent to background with no staining

LGD diagnosis by 2 expert GI pathologists significant predictor of progression to HGD/EAC

P53 positivity associated with increased EAC risk but not combined HGD/EAC outcome

Unclear how many cancer outcomes included Difficult to explain why p53 expression would be associated

with carcinoma but not combined HGD/EAC outcomes 20.2% cases had indefinite or LGD diagnosis at baseline

compared to 2.4% of controls

Kastelein F, 2013 (n=635)

Cases (49; 35 HGD/14 EAC)

Controls (586)

RCC p53 IHC as predictor of neoplastic progression in BE

At least one gland with overexpression or complete loss of expression

Aberrant p53 IHC in 11% biopsies with no dysplasia, 38% with LGD, 83% with HGD and 100% with EAC

Aberrant p53 more common in cases (49%) than controls (14%; RR=6.2)

Only 14 cases with cancer outcome All surveillance biopsies from patients with any dysplasia

stained with p53 but only one set of random surveillance biopsies from patients without dysplasia

Some LGD illustrations with aberrant p53 may be foveolar type HGD

Kaye PV, 2016 (n=72)

IOV Assess reliability of dysplasia grading and p53 interpretation

"Strong" or "absent" staining considered aberrant

No aberrant staining in non-dysplasia cases compared to 21/22 LGD and all 15 HGD cases

Kappa improved from 0.47 to 0.55 after including p53 IHC findings

Small number of cases in each group Unclear how p53 IHC findings were incorporated into

dysplasia diagnosis and to separate LGD from HGD

Figure Legends [note to reviewers: jpeg photos have been inserted here for easy viewing; letters will be added to the final hi-res photos; please make recommendations with this in mind]

Figure 1. (A) Goblet cells are readily identified on this H&E stained biopsy that was taken from

the distal esophagus. True goblet cells tend to be interspersed among foveolar cells and hence

tend to be separated from one another. (B) Pseudogoblet cells are distended, goblet-shaped

surface mucous cells that often occur in clusters and tend to contain grey-blue mucin on the

H&E stain.

Figure 2. (A) An example of low grade dysplasia (H&E stain). (B) An example of high grade

dysplasia (H&E stain).

Figure 3. (A) On an Alcian blue-PAS stain, the neutral mucin within foveolar and pseudogoblet

cells stains bright pink or magenta; if goblet cells were present, they would stain bright blue.

[plan to replace this photo with one containing goblet cells as well] (B) On an Alcian blue stain

alone, in addition to true goblet cells (arrow), foveolar cells may also stain blue (“columnar

blues”).

Figure 4. (A&B) Wild-type pattern of p53 expression in an example of BE without dysplasia.

Scattered, faintly positive nuclei are present (A, H&E stain; B, p53 immunostain). (C&D) p53

overexpression in a case of high grade dysplasia. Strong and diffuse p53 staining correlates with

TP53 mutations (C, H&E stain; D, p53 immunostain). (E&F) This example of high grade dysplasia

shows complete absence of p53 staining (“null pattern”), which also correlates with TP53

mutations (E, H&E stain; F, p53 immunostain).

Figure 5. Challenges in interpreting the p53 stain. (A&B) Two wild-type (non-dysplastic)

examples of BE showing scattered intensely staining nuclei with the p53 stain. (C) Diffuse but

not uniform intense staining with p53. Does it represent wild-type or mutated TP53?

References

1. Hur, C., et al., Trends in esophageal adenocarcinoma incidence and mortality. Cancer, 2013. 119(6): p. 1149-58.

2. Barrett, N.R., Chronic peptic ulcer of the oesophagus and 'oesophagitis'. Br J Surg, 1950. 38(150): p. 175-82.

3. Allison, P.R. and A.S. Johnstone, The oesophagus lined with gastric mucous membrane. Thorax, 1953. 8(2): p. 87-101.

4. Haggitt, R.C., et al., Adenocarcinoma complicating columnar epithelium-lined (Barrett's) esophagus. Am J Clin Pathol, 1978. 70(1): p. 1-5.

5. McDonald, G.B., D.L. Brand, and D.R. Thorning, Multiple adenomatous neoplasms arising in columnarlined (Barrett's) esophagus. Gastroenterology, 1977. 72(6): p. 1317-21.

6. Naef, A.P., M. Savary, and L. Ozzello, Columnar-lined lower esophagus: an acquired lesion with malignant predisposition. Report on 140 cases of Barrett's esophagus with 12 adenocarcinomas. J Thorac Cardiovasc Surg, 1975. 70(5): p. 826-35.

7. Sampliner, R.E., Practice guidelines on the diagnosis, surveillance, and therapy of Barrett's esophagus. The Practice Parameters Committee of the American College of Gastroenterology. Am J Gastroenterol, 1998. 93(7): p. 1028-32.

8. Playford, R.J., New British Society of Gastroenterology (BSG) guidelines for the diagnosis and management of Barrett's oesophagus. Gut, 2006. 55(4): p. 442.

9. Spechler, S.J., et al., American Gastroenterological Association medical position statement on the management of Barrett's esophagus. Gastroenterology, 2011. 140(3): p. 1084-91.

10. Fitzgerald, R.C., et al., British Society of Gastroenterology guidelines on the diagnosis and management of Barrett's oesophagus. Gut, 2014. 63(1): p. 7-42.

11. Wang, K.K., R.E. Sampliner, and G. Practice Parameters Committee of the American College of, Updated guidelines 2008 for the diagnosis, surveillance and therapy of Barrett's esophagus. Am J Gastroenterol, 2008. 103(3): p. 788-97.

12. Shaheen, N.J., et al., ACG Clinical Guideline: Diagnosis and Management of Barrett's Esophagus. Am J Gastroenterol, 2016. 111(1): p. 30-50.

13. Takubo, K., et al., Differences in the definitions used for esophageal and gastric diseases in different countries: endoscopic definition of the esophagogastric junction, the precursor of Barrett's adenocarcinoma, the definition of Barrett's esophagus, and histologic criteria for mucosal adenocarcinoma or high-grade dysplasia. Digestion, 2009. 80(4): p. 248-57.

14. McClave, S.A., H.W. Boyce, Jr., and M.R. Gottfried, Early diagnosis of columnar-lined esophagus: a new endoscopic diagnostic criterion. Gastrointest Endosc, 1987. 33(6): p. 413-6.

15. Vianna, A., et al., Normal venous circulation of the gastroesophageal junction. A route to understanding varices. Gastroenterology, 1987. 93(4): p. 876-89.

16. Amano, Y., et al., Which landmark results in a more consistent diagnosis of Barrett's esophagus, the gastric folds or the palisade vessels? Gastrointest Endosc, 2006. 64(2): p. 206-11.

17. Sharma, P., et al., Quality Indicators for the Management of Barrett's Esophagus, Dysplasia, and Esophageal Adenocarcinoma: International Consensus Recommendations from the American Gastroenterological Association Symposium. Gastroenterology, 2015.

18. Sharma, P., et al., The development and validation of an endoscopic grading system for Barrett's esophagus: the Prague C & M criteria. Gastroenterology, 2006. 131(5): p. 1392-9.

19. Paull, A., et al., The histologic spectrum of Barrett's esophagus. N Engl J Med, 1976. 295(9): p. 476-80.

20. Zwas, F., et al., Scanning electron microscopy of Barrett's epithelium and its correlation with light microscopy and mucin stains. Gastroenterology, 1986. 90(6): p. 1932-41.

21. Takubo, K., J.M. Nixon, and J.R. Jass, Ducts of esophageal glands proper and paneth cells in Barrett's esophagus: frequency in biopsy specimens. Pathology, 1995. 27(4): p. 315-7.

22. Wang, H.H., et al., Prevalence and significance of pancreatic acinar metaplasia at the gastroesophageal junction. Am J Surg Pathol, 1996. 20(12): p. 1507-10.

23. Spechler, S.J., et al., Prevalence of metaplasia at the gastro-oesophageal junction. Lancet, 1994. 344(8936): p. 1533-6.

24. Srivastava, A., et al., Morphologic features are useful in distinguishing Barrett esophagus from carditis with intestinal metaplasia. Am J Surg Pathol, 2007. 31(11): p. 1733-41.

25. Shi, L., et al., Gland ducts and multilayered epithelium in mucosal biopsies from gastroesophageal-junction region are useful in characterizing esophageal location. Dis Esophagus, 2005. 18(2): p. 87-92.

26. Glickman, J.N., et al., Phenotypic characteristics of a distinctive multilayered epithelium suggests that it is a precursor in the development of Barrett's esophagus. Am J Surg Pathol, 2001. 25(5): p. 569-78.

27. Langner, C., et al., Multilayered epithelium at the gastroesophageal junction is a marker of gastroesophageal reflux disease: data from a prospective Central European multicenter study (histoGERD trial). Virchows Arch, 2014. 464(4): p. 409-17.

28. Shields, H.M., et al., Prospective evaluation of multilayered epithelium in Barrett's esophagus. Am J Gastroenterol, 2001. 96(12): p. 3268-73.

29. Hirota, W.K., et al., Specialized intestinal metaplasia, dysplasia, and cancer of the esophagus and esophagogastric junction: prevalence and clinical data. Gastroenterology, 1999. 116(2): p. 277-85.

30. Sharma, P., et al., Relative risk of dysplasia for patients with intestinal metaplasia in the distal oesophagus and in the gastric cardia. Gut, 2000. 46(1): p. 9-13.

31. Jung, K.W., et al., Epidemiology and natural history of intestinal metaplasia of the gastroesophageal junction and Barrett's esophagus: a population-based study. Am J Gastroenterol, 2011. 106(8): p. 1447-55; quiz 1456.

32. Montgomery, E., et al., Reproducibility of the diagnosis of dysplasia in Barrett esophagus: a reaffirmation. Hum Pathol, 2001. 32(4): p. 368-78.

33. Lomo, L.C., et al., Crypt dysplasia with surface maturation: a clinical, pathologic, and molecular study of a Barrett's esophagus cohort. Am J Surg Pathol, 2006. 30(4): p. 423-35.

34. Duits, L.C., et al., Barrett's oesophagus patients with low-grade dysplasia can be accurately risk-stratified after histological review by an expert pathology panel. Gut, 2015. 64(5): p. 700-6.

35. Montgomery, E., et al., Dysplasia as a predictive marker for invasive carcinoma in Barrett esophagus: a follow-up study based on 138 cases from a diagnostic variability study. Hum Pathol, 2001. 32(4): p. 379-88.

36. Phoa, K.N., et al., Radiofrequency ablation vs endoscopic surveillance for patients with Barrett esophagus and low-grade dysplasia: a randomized clinical trial. JAMA, 2014. 311(12): p. 1209-17.

37. Rastogi, A., et al., Incidence of esophageal adenocarcinoma in patients with Barrett's esophagus and high-grade dysplasia: a meta-analysis. Gastrointest Endosc, 2008. 67(3): p. 394-8.

38. Shaheen, N.J., et al., Radiofrequency ablation in Barrett's esophagus with dysplasia. N Engl J Med, 2009. 360(22): p. 2277-88.

39. Skacel, M., et al., p53 expression in low grade dysplasia in Barrett's esophagus: correlation with interobserver agreement and disease progression. Am J Gastroenterol, 2002. 97(10): p. 2508-13.

40. Wani, S., et al., Risk factors for progression of low-grade dysplasia in patients with Barrett's esophagus. Gastroenterology, 2011. 141(4): p. 1179-86, 1186 e1.

41. Horvath, B., et al., Risk for esophageal neoplasia in Barrett's esophagus patients with mucosal changes indefinite for dysplasia. J Gastroenterol Hepatol, 2015. 30(2): p. 262-7.

42. Kestens, C., et al., Risk of neoplastic progression in Barrett's esophagus diagnosed as indefinite for dysplasia: a nationwide cohort study. Endoscopy, 2015. 47(5): p. 409-14.

43. Sinh, P., et al., Clinical outcomes in patients with a diagnosis of "indefinite for dysplasia" in Barrett's esophagus: a multicenter cohort study. Endoscopy, 2015. 47(8): p. 669-74.

44. Thota, P.N., et al., Risk stratification of patients with barrett's esophagus and low-grade dysplasia or indefinite for dysplasia. Clin Gastroenterol Hepatol, 2015. 13(3): p. 459-465 e1.

45. Coco, D.P., et al., Interobserver variability in the diagnosis of crypt dysplasia in Barrett esophagus. Am J Surg Pathol, 2011. 35(1): p. 45-54.

46. Harrison, R., et al., Detection of intestinal metaplasia in Barrett's esophagus: an observational comparator study suggests the need for a minimum of eight biopsies. Am J Gastroenterol, 2007. 102(6): p. 1154-61.

47. Genta, R.M., R.M. Huberman, and D.Y. Graham, The gastric cardia in Helicobacter pylori infection. Hum Pathol, 1994. 25(9): p. 915-9.

48. Peuchmaur, M., F. Potet, and D. Goldfain, Mucin histochemistry of the columnar epithelium of the oesophagus (Barrett's oesophagus): a prospective biopsy study. J Clin Pathol, 1984. 37(6): p. 607-10.

49. Rothery, G.A., et al., Histological and histochemical changes in the columnar lined (Barrett's) oesophagus. Gut, 1986. 27(9): p. 1062-8.

50. Chen, Y.Y., et al., Significance of acid-mucin-positive nongoblet columnar cells in the distal esophagus and gastroesophageal junction. Hum Pathol, 1999. 30(12): p. 1488-95.

51. Gottfried, M.R., S.A. McClave, and H.W. Boyce, Incomplete intestinal metaplasia in the diagnosis of columnar lined esophagus (Barrett's esophagus). Am J Clin Pathol, 1989. 92(6): p. 741-6.

52. Offner, F.A., K.J. Lewin, and W.M. Weinstein, Metaplastic columnar cells in Barrett's esophagus: a common and neglected cell type. Hum Pathol, 1996. 27(9): p. 885-9.

53. Younes, M., et al., Goblet cell mimickers in esophageal biopsies are not associated with an increased risk for dysplasia. Arch Pathol Lab Med, 2007. 131(4): p. 571-5.

54. Chitkara, Y.K. and C.L. Eyre, Evaluation of initial and deeper sections of esophageal biopsy specimens for detection of intestinal metaplasia. Am J Clin Pathol, 2005. 123(6): p. 886-8.

55. Hahn, H.P., et al., Intestinal differentiation in metaplastic, nongoblet columnar epithelium in the esophagus. Am J Surg Pathol, 2009. 33(7): p. 1006-15.

56. McIntire, M.G., et al., MUC2 is a highly specific marker of goblet cell metaplasia in the distal esophagus and gastroesophageal junction. Am J Surg Pathol, 2011. 35(7): p. 1007-13.

57. Arul, G.S., et al., Mucin gene expression in Barrett's oesophagus: an in situ hybridisation and immunohistochemical study. Gut, 2000. 47(6): p. 753-61.

58. Chinyama, C.N., et al., Expression of MUC1 and MUC2 mucin gene products in Barrett's metaplasia, dysplasia and adenocarcinoma: an immunopathological study with clinical correlation. Histopathology, 1999. 35(6): p. 517-24.

59. Guillem, P., et al., Mucin gene expression and cell differentiation in human normal, premalignant and malignant esophagus. Int J Cancer, 2000. 88(6): p. 856-61.

60. Gulmann, C., et al., Cytokeratin 7/20 and MUC1, 2, 5AC, and 6 expression patterns in Barrett's esophagus and intestinal metaplasia of the stomach: intestinal metaplasia of the cardia is related to Barrett's esophagus. Appl Immunohistochem Mol Morphol, 2004. 12(2): p. 142-7.

61. Piazuelo, M.B., et al., Phenotypic differences between esophageal and gastric intestinal metaplasia. Mod Pathol, 2004. 17(1): p. 62-74.

62. Van De Bovenkamp, J.H., et al., Gastric-type mucin and TFF-peptide expression in Barrett's oesophagus is disturbed during increased expression of MUC2. Histopathology, 2003. 42(6): p. 555-65.

63. Moskaluk, C.A., et al., Cdx2 protein expression in normal and malignant human tissues: an immunohistochemical survey using tissue microarrays. Mod Pathol, 2003. 16(9): p. 913-9.

64. Das, K.M., et al., Detection of a shared colon epithelial epitope on Barrett epithelium by a novel monoclonal antibody. Ann Intern Med, 1994. 120(9): p. 753-6.

65. Chu, P.G., Z. Jiang, and L.M. Weiss, Hepatocyte antigen as a marker of intestinal metaplasia. Am J Surg Pathol, 2003. 27(7): p. 952-9.

66. Shi, X.Y., B. Bhagwandeen, and A.S. Leong, CDX2 and villin are useful markers of intestinal metaplasia in the diagnosis of Barrett esophagus. Am J Clin Pathol, 2008. 129(4): p. 571-7.

67. DeMeester, S.R., et al., Cytokeratin and DAS-1 immunostaining reveal similarities among cardiac mucosa, CIM, and Barrett's esophagus. Am J Gastroenterol, 2002. 97(10): p. 2514-23.

68. Glickman, J.N., et al., Phenotype of Barrett's esophagus and intestinal metaplasia of the distal esophagus and gastroesophageal junction: an immunohistochemical study of cytokeratins 7 and 20, Das-1 and 45 MI. Am J Surg Pathol, 2001. 25(1): p. 87-94.

69. Groisman, G.M., M. Amar, and A. Meir, Expression of the intestinal marker Cdx2 in the columnar-lined esophagus with and without intestinal (Barrett's) metaplasia. Mod Pathol, 2004. 17(10): p. 1282-8.

70. Kerkhof, M., et al., Does CDX2 expression predict Barrett's metaplasia in oesophageal columnar epithelium without goblet cells? Aliment Pharmacol Ther, 2006. 24(11-12): p. 1613-21.

71. Phillips, R.W., H.F. Frierson, Jr., and C.A. Moskaluk, Cdx2 as a marker of epithelial intestinal differentiation in the esophagus. Am J Surg Pathol, 2003. 27(11): p. 1442-7.

72. Steininger, H., et al., Expression of CDX2 and MUC2 in Barrett's mucosa. Pathol Res Pract, 2005. 201(8-9): p. 573-7.

73. Zhang, X., M. Westerhoff, and J. Hart, Expression of SOX9 and CDX2 in nongoblet columnar-lined esophagus predicts the detection of Barrett's esophagus during follow-up. Mod Pathol, 2015. 28(5): p. 654-61.

74. Desai, T.K., et al., The incidence of oesophageal adenocarcinoma in non-dysplastic Barrett's oesophagus: a meta-analysis. Gut, 2012. 61(7): p. 970-6.

75. Couvelard, A., et al., Cytokeratin immunoreactivity of intestinal metaplasia at normal oesophagogastric junction indicates its aetiology. Gut, 2001. 49(6): p. 761-6.

76. Ormsby, A.H., et al., Cytokeratin subsets can reliably distinguish Barrett's esophagus from intestinal metaplasia of the stomach. Hum Pathol, 1999. 30(3): p. 288-94.

77. Ormsby, A.H., et al., Cytokeratin immunoreactivity patterns in the diagnosis of short-segment Barrett's esophagus. Gastroenterology, 2000. 119(3): p. 683-90.

78. El-Zimaity, H.M. and D.Y. Graham, Cytokeratin subsets for distinguishing Barrett's esophagus from intestinal metaplasia in the cardia using endoscopic biopsy specimens. Am J Gastroenterol, 2001. 96(5): p. 1378-82.

79. Mohammed, I.A., C.J. Streutker, and R.H. Riddell, Utilization of cytokeratins 7 and 20 does not differentiate between Barrett's esophagus and gastric cardiac intestinal metaplasia. Mod Pathol, 2002. 15(6): p. 611-6.

80. Dulai, G.S., et al., Preoperative prevalence of Barrett's esophagus in esophageal adenocarcinoma: a systematic review. Gastroenterology, 2002. 122(1): p. 26-33.

81. Kaye, P.V., et al., Novel staining pattern of p53 in Barrett's dysplasia--the absent pattern. Histopathology, 2010. 57(6): p. 933-5.

82. Kaye, P.V., et al., Dysplasia in Barrett's Oesophagus: p53 immunostaining is more reproducible then H&E diagnosis and improves overall reliability while grading is poorly reproducible. Histopathology, 2016.

83. Jones, D.R., et al., Potential application of p53 as an intermediate biomarker in Barrett's esophagus. Ann Thorac Surg, 1994. 57(3): p. 598-603.

84. Kaye, P.V., et al., Barrett's dysplasia and the Vienna classification: reproducibility, prediction of progression and impact of consensus reporting and p53 immunohistochemistry. Histopathology, 2009. 54(6): p. 699-712.

85. Bian, Y.S., et al., p53 gene mutation and protein accumulation during neoplastic progression in Barrett's esophagus. Mod Pathol, 2001. 14(5): p. 397-403.

86. Younes, M., et al., p53 Protein accumulation is a specific marker of malignant potential in Barrett's metaplasia. Dig Dis Sci, 1997. 42(4): p. 697-701.

87. Weston, A.P., et al., p53 protein overexpression in low grade dysplasia (LGD) in Barrett's esophagus: immunohistochemical marker predictive of progression. Am J Gastroenterol, 2001. 96(5): p. 1355-62.

88. Murray, L., et al., TP53 and progression from Barrett's metaplasia to oesophageal adenocarcinoma in a UK population cohort. Gut, 2006. 55(10): p. 1390-7.

89. Sikkema, M., et al., Aneuploidy and overexpression of Ki67 and p53 as markers for neoplastic progression in Barrett's esophagus: a case-control study. Am J Gastroenterol, 2009. 104(11): p. 2673-80.

90. Bird-Lieberman, E.L., et al., Population-based study reveals new risk-stratification biomarker panel for Barrett's esophagus. Gastroenterology, 2012. 143(4): p. 927-35 e3.

91. Kastelein, F., et al., Aberrant p53 protein expression is associated with an increased risk of neoplastic progression in patients with Barrett's oesophagus. Gut, 2013. 62(12): p. 1676-83.

92. Bani-Hani, K., et al., Prospective study of cyclin D1 overexpression in Barrett's esophagus: association with increased risk of adenocarcinoma. J Natl Cancer Inst, 2000. 92(16): p. 1316-21.

93. Shi, X.Y., B. Bhagwandeen, and A.S. Leong, p16, cyclin D1, Ki-67, and AMACR as markers for dysplasia in Barrett esophagus. Appl Immunohistochem Mol Morphol, 2008. 16(5): p. 447-52.

94. Lao-Sirieix, P., et al., Non-endoscopic screening biomarkers for Barrett's oesophagus: from microarray analysis to the clinic. Gut, 2009. 58(11): p. 1451-9.

95. Lisovsky, M., O. Falkowski, and T. Bhuiya, Expression of alpha-methylacyl-coenzyme A racemase in dysplastic Barrett's epithelium. Hum Pathol, 2006. 37(12): p. 1601-6.

96. Sonwalkar, S.A., et al., A study of indefinite for dysplasia in Barrett's oesophagus: reproducibility of diagnosis, clinical outcomes and predicting progression with AMACR (alpha-methylacyl-CoA-racemase). Histopathology, 2010. 56(7): p. 900-7.

97. Kastelein, F., et al., Value of alpha-methylacyl-CoA racemase immunochemistry for predicting neoplastic progression in Barrett's oesophagus. Histopathology, 2013. 63(5): p. 630-9.

98. Lu, D., et al., An oncofetal protein IMP3: a new molecular marker for the detection of esophageal adenocarcinoma and high-grade dysplasia. Am J Surg Pathol, 2009. 33(4): p. 521-5.

99. van Olphen, S., et al., SOX2 as a Novel Marker to Predict Neoplastic Progression in Barrett's Esophagus. Am J Gastroenterol, 2015. 110(10): p. 1420-8.

100. Brown, I.S., D.C. Whiteman, and G.Y. Lauwers, Foveolar type dysplasia in Barrett esophagus. Mod Pathol, 2010. 23(6): p. 834-43.