You are here

Article Text

Article menu

Download PDFPDF

Oesophagus
Original article
Diagnosis and risk stratification of Barrett’s dysplasia by flow cytometric DNA analysis of paraffin-embedded tissue
  1. Won-Tak Choi1,
  2. Jia-Huei Tsai2,
  3. Peter S Rabinovitch3,
  4. Thomas Small3,
  5. Danning Huang4,
  6. Aras N Mattis1,
  7. Sanjay Kakar1
  1. 1 Department of Pathology, University of California at San Francisco, San Francisco, California, USA
  2. 2 Department of Pathology, National Taiwan University Hospital, Taipei, Taiwan
  3. 3 Department of Pathology, University of Washington, Seattle, Washington, USA
  4. 4 Department of Public Health and Preventive Medicine, SUNY Upstate Medical University, Syracuse, New York, USA
  1. Correspondence to Dr Won-Tak Choi, Department of Pathology, University of California at San Francisco, 505 Parnassus Avenue, M552, Box 0102, San Francisco, CA 94143, USA; Won-Tak.Choi{at}ucsf.edu

Abstract

Objective The diagnosis of dysplasia in Barrett’s oesophagus (BO) can be challenging, and reliable ancillary techniques are not available. This study examines if DNA content abnormality detected by flow cytometry can serve as a diagnostic marker of dysplasia and facilitate risk stratification of low-grade dysplasia (LGD) and indefinite for dysplasia (IND) patients using formalin-fixed paraffin-embedded (FFPE) BO samples with varying degrees of dysplasia.

Design DNA flow cytometry was performed on 80 FFPE BO samples with high-grade dysplasia (HGD), 38 LGD, 21 IND and 14 negative for dysplasia (ND). Three to four 60-micron thick sections were cut from each tissue block, and the area of interest was manually dissected.

Results DNA content abnormality was identified in 76 HGD (95%), 8 LGD (21.1%), 2 IND (9.5%) and 0 ND samples. As a diagnostic marker of HGD, the estimated sensitivity and specificity of DNA content abnormality were 95% and 85%, respectively. For patients with DNA content abnormality detected at baseline LGD or IND, the univariate HRs for subsequent detection of HGD or oesophageal adenocarcinoma (OAC) were 7.0 and 20.0, respectively (p =<0.001).

Conclusions This study demonstrates the promise of DNA flow cytometry using FFPE tissue in the diagnosis and risk stratification of dysplasia in BO. The presence of DNA content abnormality correlates with increasing levels of dysplasia, as 95% of HGD samples showed DNA content abnormality. DNA flow cytometry also identifies a subset of patients with LGD and IND who are at higher risk for subsequent detection of HGD or OAC.

  • Aneuploidy
  • Barrett’s esophagus
  • DNA flow cytometry
  • dysplasia
  • esophageal adenocarcinoma

https://doi.org/10.1136/gutjnl-2017-313815

Statistics from Altmetric.com

    Request Permissions

    If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.

    Significance of this study

    What is already known on this subject?

    • There is a relatively high degree of interobserver variability in grading dysplasia in Barrett’s oesophagus (BO).

    • Severe reactive changes, especially in the setting of intense acute/chronic inflammation and/or ulceration, can mimic dysplasia, which can lead to an erroneous diagnosis of dysplasia and unnecessary clinical management.

    • There is no reliable ancillary technique that can aid the diagnosis of dysplasia and/or identify those patients at increased risk for subsequent detection of high-grade dysplasia (HGD) or oesophageal adenocarcinoma (OAC).

    • Flow cytometric analysis of DNA content has been shown to be a good biomarker for predicting OAC, but it typically uses fresh tissue which requires obtaining separate biopsies for flow cytometry and does not allow correlation of morphological findings with flow cytometric results.

    What are the new findings?

    • The presence of DNA content abnormality correlates with increasing levels of dysplasia, as abnormal DNA content was identified in 0% of BO without dysplasia, 9.5% of indefinite for dysplasia (IND), 21.1% of low-grade dysplasia (LGD) and 95% of HGD.

    • The estimated sensitivity of DNA content abnormality as a diagnostic marker of HGD was 95% with the specificity of 85%, 90% positive predictive value (PPV) and 92% negative predictive value (NPV).

    • One-year, 4-year, 5-year and 12-year detection rates of HGD or OAC for LGD patients with DNA content abnormality were 85.4% (p<0.001), 85.4% (p=0.003), 100% (p<0.001) and 100% (p<0.001), respectively, whereas LGD patients in the setting of normal DNA content had 1-year, 4-year, 5-year and 12-year detection rates of 13.5%, 30.8%, 30.8% and 30.8%, respectively.

    • All IND patients with DNA content abnormality were subsequently found to have HGD or OAC within 2 years (p=0.001), whereas 1-year, 2-year and 13-year detection rates of HGD or OAC in the setting of normal DNA content remained stable at 5.9%.

    • The univariate HRs for subsequent detection of HGD or OAC in patients with DNA content abnormality detected at baseline LGD or IND were 7.0 and 20.0, respectively (p =<0.001).

    Significance of this study

    • DNA content abnormality detected by flow cytometry using formalin-fixed paraffin-embedded (FFPE) tissue from BO samples can serve as a diagnostic marker of dysplasia and facilitate risk stratification of patients with LGD and IND.

    • Given that the diagnosis of HGD usually prompts resection or endoscopic ablative therapy, DNA flow cytometry can be selectively performed on borderline HGD cases (where a definite diagnosis of HGD is difficult based on histology) to confirm a morphological impression or suspicion of HGD, and in situations where there is discordant interpretation among expert GI pathologists before initiating aggressive clinical management.

    • DNA flow cytometry can be performed on LGD or IND samples to identify a subset of patients who may be at increased risk for subsequent detection of HGD or OAC and may benefit from more frequent endoscopic surveillance or ablative therapy.

    • Considering that advanced endoscopic imaging techniques are now allowing endoscopists to perform targeted mucosal biopsies throughout the length of BO, the selective approach of analysing DNA content in FFPE dysplastic tissue has the potential to complement targeted mucosal biopsies, thus potentially reducing the number of biopsies and endoscopic procedures as well as the associated costs.

    Introduction

    Barrett’s oesophagus (BO) is a major risk factor for the development of oesophageal adenocarcinoma (OAC).1–3 BO is defined as endoscopically visible columnar epithelium extending upwards from the gastro-oesophageal junction that is histologically confirmed to have goblet cells (‘intestinal metaplasia’).2 4–6 BO is a genetically unstable metaplastic epithelium that accumulates multiple genetic and chromosomal alterations that eventually progress to dysplasia and OAC.7–12 Because dysplasia is the best available marker of cancer risk in patients with BO, the current endoscopic surveillance practice aims to detect dysplasia prior to the development of OAC, with the appropriate surveillance interval determined by the grade of dysplasia.6

    Dysplasia is classified as ‘low-grade dysplasia (LGD)’ and ‘high-grade dysplasia (HGD).'3 5 13 14 LGD is characterised by a distinct lack of surface maturation with atypical nuclei limited to the basal half of the cytoplasm,5 13–15 whereas HGD shows more severe cytological and/or architectural abnormalities.5 13 14 16 The category of ‘indefinite for dysplasia (IND)’ is used most often in the setting of intense acute/chronic inflammation (often with ulceration) and/or technical issues (including the lack of surface epithelium, marked cautery effect or tangential section), where a definite distinction between dysplasia and regeneration cannot be made with certainty by histology.5 13 14 17 In fact, the pathological evaluation of dysplasia is limited by a relatively high degree of interobserver variability in grading dysplasia.13–15 17–19 As a result, there is a need for ancillary techniques to aid the diagnosis of dysplasia and/or to identify those patients at increased risk for subsequent detection of HGD or OAC.

    In this regard, immunohistochemical stains, especially α-methylacyl-coenzyme A racemase 17 20 and p53,19 21–23 have been extensively studied. However, these studies have demonstrated that the interpretation of staining results is highly variable (thus not easily translatable into clinical practice), and that non-neoplastic epithelia frequently show positive staining (often strong and diffuse).24 Similarly, the diagnostic utility of genetic and chromosomal abnormalities detected in BO (including 9p loss of heterozygosity (LOH), 17p LOH, and mutations of p53 and cyclin-dependent kinase N2 (tumour suppressor genes)) is also limited,7–12 as these changes tend to occur early and frequently in BO (even without dysplasia), often before DNA flow cytometric biomarkers of progression (aneuploidy or elevated 4N fraction) have developed.7–10 12 Alternatively, DNA flow cytometry that measures nuclear DNA content abnormality can potentially serve as an adjunct method for the diagnosis of dysplasia and risk assessment among patients with BO, but such analyses typically use fresh tissue (which does not allow direct histology-flow cytometry correlation).25–27 As such, this study explores the utility of assessing DNA content abnormality for the diagnosis and risk stratification of dysplasia in BO using formalin-fixed paraffin-embedded (FFPE) tissue.

    Materials and methods

    Patients and data collection

    Using our pathology information system (CoPath), all samples of BO categorised as having HGD (119 samples), LGD (121 samples), or IND (49 samples) at University of California at San Francisco (UCSF) Medical Center between 1990 and 2015 were identified. A total of 139 samples from 124 patients with adequate tissue and in which the diagnosis was confirmed on re-review (ie, histologically unequivocal cases of HGD and LGD) were included in the study. Since it is extremely rare to find abnormal DNA content in FFPE BO tissue without dysplasia,21 22 28 29 14 BO samples without dysplasia were randomly selected from 14 additional patients with BO as the control group. All samples with the reported diagnosis of HGD, LGD, IND or negative for dysplasia (ND) were reviewed and confirmed by at least two pathologists (W-TC, J-HT, and SK) using published criteria.5 13 14 In cases where multiple biopsies from the same patient showed dysplasia, all samples were reviewed, and one sample with the largest dysplastic area was selected and assessed for DNA content. DNA flow cytometry was performed using FFPE tissue from 80 samples of HGD (26 biopsies, 12 endoscopic mucosal resections (EMRs), 42 surgical resections), 38 LGD (34 biopsies, 4 surgical resections), 21 IND (all biopsies) and 14 ND (all biopsies). table 1 shows demographic characteristics of our cohort. For patients with LGD or IND, hospital electronic medical records were further reviewed to retrieve pertinent data, including endoscopic findings (length of BO segment (short <3 cm and long ≥3 cm of Barrett’s mucosa), nodule/nodularity and hiatal hernia) as well as demographic risk factors (body mass index (BMI) ≥30 kg/m2, age ≥60 years, gender and ethnicity). The UCSF Institutional Review Board for human subjects research approved our study (IRB # 15–17416).

    View this table:
    Table 1

    Characteristics of patients with BO diagnosed with ND, IND, LGD and HGD at UCSF Medical Center between 1990 and 2015

    DNA flow cytometry

    Depending on the size of the dysplastic area, three to four 60-micron thick sections were cut from each tissue block. The area of interest was manually dissected from each section to maximise the chance of finding a potentially abnormal cell population (by increasing its percentage in a background of normal diploid cells) and collected into a tissue biopsy bag. The tissue sample was deparaffinised with 100% xylene and rehydrated through graded ethanol to distilled water. The sample was then removed from the biopsy bag and placed into a 10 mL Falcon tube for 1 hour incubation with 2 mL 1% pepsin in phosphate-buffered saline (PBS, pH 1.5) at 37°C. After incubation, the sample was vortexed vigorously to break off any clumped tissue fragments. The pepsin digestion was stopped by adding 10 mL NST/bovine serum albumin (BSA) buffer (8.5 grams NaCl, 1.2 grams Tris Base, 0.111 grams CaCL2*2H2O, 0.123 grams MgSO4*7H2O and 0.5 grams BSA in 1 L sterile water, pH 7.8). The sample was filtered through an 80-micron nylon mesh and recovered by centrifuging at 1700 rpm for 10 min at 2°C. The pellet was resuspended in 1 mL working DAPI solution (4,6-diamidino-2-phenylindole; Accurate Chemical & Scientific Corporation, Westbury, New York, USA), vortexed gently and incubated overnight at −80°C. To prepare working DAPI solution, 800 mL NST/NP-40/BSA buffer (1 mL NP-40 (IGEPAL CA-630 or octylphenoxy poly(ethyleneoxy)ethanol, branched; Sigma-Aldrich, St. Louis, Missouri, USA) into 1 L NST/BSA buffer) was first mixed with 200 mL 106 mM MgCl2, 10 mg DAPI (final concentration=10 µg/mL) and 10% dimethyl sulfoxide (DMSO, 100 mL). The resulting stock NST/NP-40/DAPI solution was diluted 1:4 with NST/NP-40/BSA buffer plus 10% DMSO. After overnight incubation, potential aggregates were further dissociated by forceful passage 15–20 times through a 26-gauge needle. The specimen was analysed with a BD LSRII S854 flow cytometer (BD Biosciences, San Jose, California, USA) with UV laser excitation. The published consensus guidelines for clinical DNA flow cytometry were followed,30 and DNA aneuploidy was defined as a discrete extra resting phase (G0)/gap phase 1 (G1) peak that was visually distinguishable from the normal DNA diploid G0/G1 peak.30 The finding of gap phase 2 (G2)/tetraploid (4N) fraction greater than 6% (with DNA index of 1.9–2.1) was also classified as abnormal due to its strong association with dysplasia or OAC.7 25 27 30 The average coefficient of variation (CV) of normal diploid cells and background aggregates and debris (BAD) across all samples were 6.9% and 9%, respectively, which were less than the recommended CV (<8%) and BAD (<20%) based on the published consensus guidelines.30 All DNA content histograms were analysed by the computer program Multicycle (De Novo software, Glendale, California, USA) using algorithms to detect and compensate for cut nuclei and aggregates as well as debris.31 Flow cytometric histograms were interpreted by two pathologists (W-TC and PSR) independently of any other information.

    Statistical analysis

    Statistical analysis was performed using methods appropriate for censored data (Kaplan-Meier (KM) curves and Cox proportional hazards (PH) model), as follow-up time varied among patients. In other words, since not all patients reached the endpoint of HGD or OAC before follow-up ended, subsequent detection of HGD or OAC was descriptively summarised by the KM curves. A null hypothesis of equal distribution of detection times was assessed with the log-rank test. Detection rates at specific time points were calculated from the KM curves. The presence of longer BO segment, nodule/nodularity, hiatal hernia, or abnormal DNA content as a potential risk factor for subsequent detection of HGD or OAC was assessed using univariate and/or multivariate Cox PH models. Potential demographic risk factors (including BMI ≥30 kg/m2, age ≥60 years, gender and ethnicity) were also evaluated using univariate and/or multivariate Cox PH models. Both 95% CIs and p values associated with the null hypotheses of no hazard difference between groups were calculated using the asymptotic Wald test. Statistical significance was set at p<0.05. All analyses were performed using SAS V.9.4. As for the calculation of sensitivity, specificity, positive predictive value (PPV) and negative predictive value (NPV) of DNA content abnormality as a diagnostic marker of HGD, only the samples with a confirmed diagnosis of HGD (n=80), LGD (n=38) and ND (n=14) were used (using the histological diagnosis of dysplasia as the gold standard for this particular purpose), since IND samples (n=21) cannot be classified as either dysplasia or regeneration based on histology. PPV is the probability that a sample with a morphological impression of HGD truly represents HGD in the setting of abnormal DNA content, whereas NPV is the probability that the same biopsy may not represent HGD, if there is no DNA content abnormality.

    Results

    DNA content abnormality in ND

    All 14 BO samples without dysplasia demonstrated normal diploid DNA histograms (figure 1; table 2). The patients included 12 men and 2 women ranging in age from 48 years to 83 years (mean 65 years) (table 1). Ten of 14 patients had follow-up biopsies, and none of them developed HGD or OAC within a mean follow-up time of 56 months (range: 1–170 months) (table 2).

    Figure 1
    Figure 1

    (A) Barrett’s oesophagus (BO) without dysplasia is characterised by the presence of intestinal metaplasia. (B) DNA flow cytometric histogram shows a normal diploid population (green). DAPI, 4,6-diamidino-2-phenylindole.

    View this table:
    Table 2

    Frequency of DNA content abnormality with respect to increasing histological grade of dysplasia and subsequent detection of HGD or OAC in samples with DNA content abnormality

    DNA content abnormality in HGD

    Seventy-six (95%) of 80 HGD samples from 65 patients showed DNA content abnormality (aneuploidy or elevated 4N fraction) (figure 2; table 2). Sixty-five (85.5%) of 76 HGD samples demonstrated aneuploidy (figure 2B), whereas the remaining 11 HGD samples (14.5%) showed elevated 4N fraction without aneuploidy (figure 2D). Abnormal DNA content was identified in 61 (93.8%) of 65 patients with HGD. Four samples (5%) showed normal flow cytometric results. The patients with HGD included 57 men and 8 women with a mean age of 64 years (range: 36–87 years) (table 1). The frequency of DNA content abnormality was similar in both biopsies (including EMR specimens, 94.7%) and surgical resections (95.2%). Fifteen patients with HGD had prior oesophageal biopsies that showed HGD before undergoing surgical resections. There was 100% concordance between the specimens, as abnormal DNA content was identified in all 15 biopsies and surgical resections. A total of 42 (52.5%) of 80 HGD samples had concurrent or subsequent OAC detected on definite treatment for HGD (EMR or surgical resection, 40 samples) or evidence of metastatic disease (2 samples). The remaining 38 samples had a confirmed diagnosis of HGD on EMR or surgical resection but without OAC. Among 76 HGD samples with DNA content abnormality, 40 samples (52.6%) had concurrent or subsequent OAC (table 2). The samples with concurrent or subsequent OAC included 33 patients (30 men) with a mean age of 65 years (range: 36–86 years) at the time of HGD diagnosis. The mean time to concurrent or subsequent detection of OAC for patients with baseline HGD was 2.4 months (range: 0–19 months). Those who did not develop OAC included 32 patients (27 men) with a mean age of 63 years (range: 38–87 years). The estimated sensitivity of DNA content abnormality as a diagnostic marker of HGD was 95% with the specificity of 85%, 90% PPV and 92% NPV.

    Figure 2
    Figure 2

    ( A) HGD is characterised by marked cytological atypia and architectural complexity. (B) DNA histogram shows two discrete aneuploid peaks (red and blue) that are visually distinguishable from the normal diploid population (green). (C) Another example of HGD shows atypical glands lined by highly pleomorphic cells with enlarged, rounder nuclei. (D) DNA histogram shows elevated 4N fraction greater than 6%, in addition to the normal diploid cells (green), but there is no distinct aneuploid peak. DAPI, 4,6-diamidino-2-phenylindole; HGD, high-grade dysplasia.

    DNA content abnormality in LGD

    Eight (21.1%) of 38 LGD samples showed DNA content abnormality (figure 3A,B; table 2). The patients included 34 men and 4 women ranging in age from 41 years to 83 years (mean 66 years) (table 1). Mean follow-up time was 33 months (range: 1–144 months). Interestingly, the presence of abnormal DNA content detected at baseline LGD was a significant predictor of subsequent finding of HGD or OAC with the estimated univariate HR of 7.0 from the Cox model (p<0.001, 95% CI (2.184 to 24.073)) (table 3). Seven (87.5%) of 8 LGD patients  with DNA content abnormality were subsequently found to have HGD (3 patients) or OAC (4 patients) within a mean follow-up time of 4.7 months (range: 2–8 months), whereas 5 (16.7%) of the remaining 30 patients with LGD in the setting of normal flow cytometric results developed HGD (3 patients) or OAC (2 patients) within a mean follow-up time of 15.5 months (range: 1–40 months) (table 2). Using HGD or OAC as the outcome, KM curves showed that LGD patients without DNA content abnormality had 1-year, 4-year, 5-year and 12-year detection rates of 13.5%, 30.8%, 30.8% and 30.8%, respectively, whereas LGD patients with abnormal DNA content had 1-year, 4-year, 5-year and 12-year detection rates of 85.4% (p<0.001), 85.4% (p=0.003), 100% (p<0.001) and 100% (p<0.001), respectively (95% CIs (42% to 100%), (18.8% to 90.5%), (28.7% to 100%) and (28.7% to 100%), respectively) (figure 3C).

    Figure 3
    Figure 3

    ( A) LGD shows cytological atypia that extends to the surface, but the degree of atypia is less severe than that of HGD. (B) DNA histogram shows a distinct aneuploid population (red). HGD was identified on follow-up. (C) Detection of HGD or OAC in LGD patients with abnormal DNA content at baseline. One-year, 4-year, 5-year and 12-year detection rates of HGD or OAC for LGD patients  with DNA content abnormality were 85.4% (p<0.001), 85.4% (p=0.003), 100% (p<0.001) and 100% (p<0.001), respectively, whereas LGD patients  in the setting of normal DNA content had 1-year, 4-year, 5-year and 12-year detection rates of 13.5%, 30.8%, 30.8% and 30.8%, respectively. Each tick represents a person being censored. DAPI, 4,6-diamidino-2-phenylindole; HGD, high-grade dysplasia; LGD, low-grade dysplasia; OAC, oesophageal adenocarcinoma.

    View this table:
    Table 3

    Univariate and multivariate Cox proportional hazards models with HGD or OAC as the outcome in BO patients  with LGD

    View this table:

    Since the presence of longer BO segment, nodule/nodularity and hiatal hernia have been previously reported as major risk factors for the development of BO, dysplasia and/or OAC,6 32–34 the available endoscopic reports for patients with LGD were reviewed. Although the sample size for some of these variables may be small for HR analyses, the length of BO segment (HR=1.1, p=0.883, 95% CI (0.240 to 3.876)), nodule/nodularity (HR=0.2, p=0.018, 95% CI (0.053 to 1.020)) and hiatal hernia (HR=0.5, p=0.562, 95% CI (0.030 to 2.875)) were not significantly associated with an increased risk for subsequent detection of HGD or OAC within the study period (table 3). Similarly, BMI ≥30 kg/m2 (HR=0.3, p=0.062, 95% CI (0.056 to 1.136)), age ≥60 years (HR=2.0, p=0.228, 95% CI (0.590 to 6.378)), gender (p=0.231) and ethnicity (p=0.359) were not significant risk factors for subsequent detection of HGD or OAC (table 3). The HRs and CIs for gender and ethnicity were not estimable, because the majority of patients with LGD were white men (table 1). Only abnormal DNA content remained as a significant risk factor for subsequent detection of HGD or OAC in the multivariate Cox model with the estimated HR of 17.9 (p=0.042, 95% CI (1.623 to 581.173)) (table 3).

    DNA content abnormality in IND

    Twenty-one patients, including 16 men and 5 women with a mean age of 62 years (range: 35–77 years), had biopsy specimens that were classified as IND (table 1). Mean follow-up time was 53 months (range: 1–158 months). Nineteen (90.5%) of 21 samples had normal diploid cells (table 2), and they all had negative follow-up biopsies except one patient who was subsequently found to have HGD. In conjunction with the normal DNA content, the lack of subsequent detection of HGD or OAC in the majority of our IND samples supports the validity of initial IND diagnoses, where a definite distinction between dysplasia and regeneration could not be made with certainty by histology due to extensive acute/chronic inflammation and/or technical issues (figure 4A,B). The two IND patients with abnormal DNA content were subsequently found to have HGD (1 patient in 14 months) or OAC (1 patient in 15 months) within 2 years (p=0.001, 95% CI (25.2% to 100%)) (figure 4C,D; table 2). By contrast, 1-year, 2-year and 13-year detection rates of HGD or OAC in the setting of normal DNA content remained stable at 5.9% (figure 4D). The univariate HR for subsequent detection of HGD or OAC in IND patients with baseline DNA content abnormality was estimated to be 20.0 from the Cox model (p=<0.001, 95% CI (1.872 to 436.409)) (table 4). However, it should be noted that due to the small sample size, the Cox model did not converge when all the potential risk factors were included in the multivariate analysis, and thus the multivariate HR for abnormal DNA content could not be calculated in this case.

    Figure 4
    Figure 4

    ( A) In IND, the presence of intense acute/chronic inflammation as well as loss of surface epithelium makes it difficult to discern if the glandular atypia is reactive or dysplastic. (B) Granulation tissue (ulceration) is also present in the same biopsy. (C) DNA histogram demonstrates an aneuploid population (red) distinct from the normal diploid peak (green). (D) Detection of HGD or OAC in IND patients with abnormal DNA content at baseline. Two IND patients with abnormal flow cytometric results were subsequently found to have HGD or OAC within 2 years (p=0.001). By contrast, 1-year, 2-year and 13-year detection rates of HGD or OAC in the setting of normal DNA content remained stable at 5.9%. Each tick represents a person being censored. DAPI, 4,6-diamidino-2-phenylindole; HGD, high-grade dysplasia; IND, indefinite for dysplasia; OAC, oesophageal adenocarcinoma.

    View this table:
    Table 4

    Univariate Cox proportional hazards model with HGD or OAC as the outcome in BO patients  with IND

    Although the sample size for some of the other potential risk factors may be small for HR analyses, the length of the BO segment (HR=0.7, p=0.802, 95% CI (0.034 to 7.693)), nodule/nodularity (p=0.630), hiatal hernia (HR=1.2, p=0.858, 95% CI (0.058 to 13.009)), BMI ≥30 kg/m2 (HR=0.4, p=0.515, 95% CI (0.016 to 10.412)), age ≥60 years (HR=3.8, p=0.240, 95% CI (0.365 to 82.223)), gender (p=0.341) and ethnicity (p=0.530) were not significantly associated with an increased risk for subsequent detection of HGD or OAC within the study period (table 4). The HRs and CIs for gender, ethnicity and nodule/nodularity were not estimable, because the majority of our patients with IND were white men (table 1), and only 1 of 21 IND samples showed nodularity.

    Discussion

    There is sequential progression from Barrett’s metaplasia to dysplasia and to OAC.3 5 33 The recognition of HGD or early OAC usually prompts resection or endoscopic ablative therapy,1 2 35 but consistent and objective detection of dysplasia by histology may not be always possible.13–15 17–19 This is especially true in the setting of intense acute/chronic inflammation (often with ulceration) and/or technical issues (including the lack of surface epithelium, marked cautery effect or tangential section), where a definite distinction between dysplasia and regeneration cannot be made with certainty by histology (thus a diagnosis of IND is rendered).5 13 14 17 It can be similarly difficult to distinguish LGD from HGD in this setting. Given that the accurate diagnosis and grading of dysplasia determines the appropriate surveillance interval and/or potential treatment decision, there is a need for ancillary techniques to aid the diagnosis of dysplasia and/or to identify those patients at increased risk for subsequent detection of HGD or OAC.

    In this regard, abnormal DNA content detected by DNA flow cytometry can serve as a diagnostic marker of dysplasia, as up to 95% of HGD samples showed DNA content abnormality. The estimated sensitivity of DNA content abnormality as a diagnostic marker of HGD was 95% with the specificity of 85%, 90% PPV and 92% NPV. The high level of concurrent or subsequent OAC in our HGD cohort supports the validity of HGD diagnoses, which is consistent with the previous report that HGD is found adjacent to OAC in 50%–100% of cases.35–39 This close concordance of HGD with DNA content abnormality is also higher than previously reported.22 29 40 41 This is likely due to DNA flow cytometry being performed on FFPE tissue in which (1) the flow cytometry and histology are determined from the identical biopsy and (2) the macrodissected region of the biopsy selected for flow cytometry is much more certain to contain the epithelial cells with the designated histology. By contrast, the lack of DNA content abnormality among BO samples without dysplasia is consistent with the previous report that diploid DNA content is found in the vast majority of non-dysplastic FFPE BO samples, with several studies reporting the aneuploid rate of 0% in BO samples without dysplasia.28 29

    Considering that overdiagnosis of HGD can lead to unnecessary medical management and potential complications, DNA flow cytometry has the potential to improve diagnostic accuracy when HGD is a consideration (ie, borderline HGD cases with a diagnosis of ‘suspicious’ or ‘possible’ HGD). Although an excellent interobserver agreement for HGD has been reported among GI pathologists,13 14 one recent study demonstrated that on review of 485 HGD samples from both academic and private centres by experienced GI pathologists, up to 40% were overdiagnosed with HGD and had to be reinterpreted as LGD, IND, ND or no BO.16 In this regard, abnormal flow cytometric results may serve as objective evidence to confirm a morphological impression or suspicion of HGD. However, it should be noted that DNA content abnormality in the setting of HGD does not provide any additional prognostic value for which patients will develop OAC, as no significant difference was observed in the frequency of DNA content abnormality in HGD samples with (95.2%) or without OAC (94.4%). Rather, the potential value of finding DNA content abnormality in HGD samples lies in confirming the diagnosis in challenging cases with confounding factors like intense acute/chronic inflammation, ulceration and/or technical issues, as well as in situations where there is discordant interpretation among expert GI pathologists.

    Mean time to concurrent or subsequent detection of OAC was relatively short (2.4 months, range: 0–19 months) among our patients with HGD, as many of these patients were diagnosed with OAC on definite treatment for HGD (at the time of EMR or surgical resection). It should also be noted that only diploid DNA content was observed in 5% of the HGD samples. Although we cannot entirely exclude the possibility that HGD cells may not have been present in the deeper levels used for DNA flow cytometric analysis, the finding of normal DNA content, especially in borderline HGD cases, should raise the possibility that the samples may represent reactive changes rather than dysplasia. Thus, patients may benefit from repeat endoscopic procedures rather than aggressive clinical management. However, in a histologically unequivocal case, pathologists should not be deterred from making a diagnosis of HGD even if no abnormal DNA content is found.

    It is well known that the diagnosis of LGD is limited by poor interobserver agreement,6 13 14 18 and a recent study further illustrated suboptimal interobserver agreement for LGD (κ=0.11) even among GI pathologists.15 In this regard, our study demonstrates a significant correlation between abnormal DNA content and LGD samples that were subsequently found to have HGD or OAC within a year, with the estimated univariate and multivariate HRs of 7.0 and 17.9, respectively, from the Cox model (table 3). An increased risk for subsequent detection of HGD or OAC among LGD samples in the setting of DNA content abnormality further supports the validity of abnormal DNA content to define a subset of patients with LGD at highest risk of developing HGD or OAC. This is consistent with prior studies, although our predictive value is higher than that previously reported,22 29 40 41 likely due to the advantages of performing flow cytometry from FFPE tissue, as noted above. This is also consistent with a recent report that a combination panel of LGD, abnormal DNA ploidy (using image cytometry) and Aspergillus oryzae lectin can identify a subset of patients who are at higher risk for subsequent detection of HGD or OAC.42

    Although endoscopic ablative therapy is increasingly being recommended for patients with LGD,43 endoscopic surveillance every 12 months continues to be an acceptable alternative.6 44 In this regard, abnormal flow cytometric results at baseline LGD could potentially enable clinicians to recommend endoscopic therapy to this patient subset, due to its high risk for subsequent detection of HGD or OAC, whereas continued surveillance may be an acceptable approach in the setting of normal flow cytometric results. However, it should be noted that 5 (41.7%) of 12 LGD samples with subsequent detection of HGD or OAC did not show evidence of DNA content abnormality at baseline. As such, negative flow cytometric results should not deter pathologists from making a diagnosis of LGD in a histologically unequivocal case, whereas the presence of DNA content abnormality can support a histological impression of LGD that has a high likelihood of finding HGD or OAC within a year.

    Although p53 immunohistochemistry has been recommended by the British Society of Gastroenterology (but not by the American College of Gastroenterology) as an aid to diagnose dysplasia in BO, the interpretation of staining results is highly variable. We also note that the sensitivity and specificity of abnormal DNA content as a diagnostic marker of HGD are superior to the reported sensitivity and specificity of p53. For instance, Kastelein et al found that p53 was aberrantly expressed in 11% of BO without dysplasia, 38% of LGD and 83% of HGD,23 whereas abnormal DNA content was identified in 0% of BO without dysplasia, 21.1% of LGD and 95% of HGD in our cohort. The same study also reported that aberrant p53 staining at baseline LGD was associated with a twofold increase in disease progression from 15% to 33%.23 In our cohort, up to 87.5% of LGD patients  with DNA content abnormality were subsequently found to have HGD or OAC within a year, whereas only 16.7% of patients with LGD in the setting of normal flow cytometric results developed HGD or OAC within a mean follow-up time of 15.5 months.

    The diagnostic category of IND is used most often in the setting of intense acute/chronic inflammation, ulceration and/or technical issues in which there is cytological atypia suggestive of possible dysplasia, but a definite distinction between dysplasia and regeneration cannot be made with certainty by histology.5 13 14 17 DNA flow cytometry can be very helpful in this setting, because mucosal erosion, ulceration, granulation tissue and/or increases in acute or chronic inflammatory cells typically do not cause aneuploidy and/or increased 4N fraction.25 In our cohort, only 1 (5.3%) of 19 IND samples without DNA content abnormality developed HGD or OAC on follow-up, whereas all patients with abnormal DNA flow cytometric results were subsequently found to have HGD or OAC within 2 years (figure 4D, univariate HR=20.0, p<0.001). This is consistent with the recent data that suggest a high risk for subsequent detection of HGD or OAC for a subset of IND patients within the first 3 years of diagnosis.45 46 This result is also consistent with our earlier findings based on flow cytometry of fresh-frozen biopsies, although the estimated HR in that study was somewhat lower (HR=5.7, p=0.003).45 Overall, IND diagnosis in the setting of DNA content abnormality may warrant more frequent follow-up surveillance intervals or may even support a diagnosis of dysplasia. Alternatively, 18 of 21 IND samples with normal DNA content that showed no evidence of dysplasia or OAC on follow-up most likely represent reactive changes rather than dysplasia, further supporting the notion that it is extremely rare to find abnormal DNA content in FFPE BO tissue without dysplasia.21 22 28 29

    Several studies have demonstrated that the frequency of DNA aneuploidy increases with increasing histological grade of dysplasia using FFPE tissue.22 29 40 41 Montgomery et al reported that DNA aneuploidy was present in 77% of HGD,29 whereas only 32% of HGD showed DNA content abnormality in another study.22 In our cohort, abnormal flow cytometric results were identified in 95% of HGD, 21.1% of LGD, 9.5% of IND and no sample without dysplasia. There can be several reasons for these discrepant results, the most likely being the lack of selective analysis of macroscopically dissected dysplastic epithelial cells in earlier studies. An increase in the proportion of non-dysplastic epithelial cells in a specimen can decrease the sensitivity for detection of aneuploidy and/or increased 4N fraction in the dysplastic epithelial cells (‘dilutional effects’).25 The methods followed in our study are based on the established consensus guidelines, and a pathologist (PSR) with long experience in flow cytometry of Barrett’s epithelium validated the flow cytometric results.

    Although a number of studies have used image cytometry to demonstrate a similar correlation between aneuploidy and subsequent detection of dysplasia or OAC,42 47–50 these studies have employed different techniques to process samples and interpret DNA histograms.41 47–49 In fact, the lack of standardised guidelines for image cytometry30 makes direct comparisons between these studies difficult. Also in BO, there is a limited number of studies that compared the results of image cytometry and traditional flow cytometry (the current gold standard). Although one study reported a concordance rate of 93%,47 another study concluded that flow cytometry missed the diagnosis of aneuploidy in 29%, whereas image cytometry detected all cases of aneuploidy.41 However, by virtue of identifying a small number of abnormal nuclei from the area of interest, image cytometry may identify much higher rates of rare aneuploid cells that are not detected by flow cytometry.41 For instance, Yu et al reported that more than two-thirds of BO without dysplasia (69%) and LGD (82%) samples have aneuploid populations.49 Similarly, Fang et al demonstrated that up to 13% of BO without dysplasia and 60% of LGD samples showed DNA content abnormality.50 The clinical significance of these rare abnormal cells, if they truly represent aneuploid populations, is unclear at this time.

    Advanced endoscopic imaging techniques are now allowing endoscopists to detect subtle mucosal abnormalities and perform targeted mucosal biopsies,6 potentially eliminating a need for random, four-quadrant mucosal biopsies throughout the length of BO. Given that it is extremely rare to find abnormal DNA content in FFPE BO tissue without dysplasia21 22 28 29 and a recent meta-analysis reporting lower risk for progression for non-dysplastic BO (1 per 300 patients per year),51 our data seem to suggest that DNA flow cytometry can be selectively performed on borderline HGD cases (where a definite diagnosis of HGD may not be possible based on histology) to confirm a morphological impression or suspicion of HGD and/or to facilitate risk stratification, especially in the cases of LGD or IND, for increased surveillance or endoscopic therapy. In other words, this selective approach of analysing DNA content in FFPE dysplastic tissue has the potential to complement targeted mucosal biopsies, thus potentially reducing the number of biopsies and endoscopic procedures as well as the associated costs.

    There are several advantages of using FFPE tissue for DNA flow cytometric analysis. First, it enables DNA flow cytometry to be selectively applied to test areas that are morphologically abnormal. Since pathologists usually do not have a difficulty in identifying a histologically abnormal region (although diagnosing and/or grading dysplasia can be challenging at times), there should be no major interobserver variability in selecting the area of interest to be tested for DNA content analysis. In fact, the entire abnormal region should be macrodissected and submitted for DNA flow cytometric analysis. Second, it avoids obtaining separate biopsies for flow cytometry and histology (thus permitting an easier and more routine specimen handling/workflow compared with fresh tissue), and it allows direct histology-flow cytometry correlation. Third, DNA flow cytometry is a relatively simple and inexpensive test that can be completed within 2–3 days. The majority (69%) of BO samples tested in this study were from small mucosal biopsies, from which three to four 60-micron thick sections were cut and macrodissected (without exhausting tissue) to generate high-quality histograms, demonstrating the feasibility of this methodology.

    One possible limitation of our study is that all the patients in our cohort were referred to or seen at UCSF Medical Center, which implies that referral bias cannot be ruled out, but the direction of such bias, if it exists, is difficult to predict in this setting. Also because the study was designed to identify factors predictive of subsequent detection of HGD or OAC occurring within 12–13 years of initial LGD or IND biopsy, we cannot completely exclude the possibility that the presence of longer BO segment, nodule/nodularity and hiatal hernia as well as potential demographic risk factors may be associated with an increased risk of subsequent detection of HGD or OAC with a longer follow-up time. Furthermore, although the current analyses using FFPE tissue represent the evaluation of DNA content in a clinical setting, even ‘routine’ specimen handling and storage may vary between institutions and even at the same institution over time, thus the possible impact of such preanalytical variables on the assay cannot be entirely excluded. However, we note that high-quality histograms were observed across all FFPE samples obtained over a 25-year period at UCSF Medical Center. Finally, the power analysis based on the univariate Cox model suggests that 80% power can be achieved with the sample sizes of 40 for LGD and 20 for IND to detect the estimated HRs of 8.4 and 17.3, respectively, at a significance level of 0.05 (p<0.05). If other variables were considered in the multivariate Cox model, the sample size of 40 for LGD would yield 80% power to detect the estimated HR of 9.6. In this regard, our preliminary results are promising in that based on the sample sizes of 38 for LGD and 21 for IND, only abnormal DNA content detected at baseline LGD or IND was found to be a significant risk factor for subsequent detection of HGD or OAC, with the estimated univariate HRs of 7.0 and 20.0, respectively (p=<0.001). For patients with LGD, abnormal DNA content remained as a significant risk factor for subsequent detection of HGD or OAC in the multivariate Cox model with the estimated HR of 17.9 (p=0.042). While our preliminary results are promising in that the study is approaching adequate power based on these sample size calculations (power analysis), we recognise that the assessment of DNA content abnormality as a predictive marker of HGD or OAC in the setting of LGD or IND will require further validation in a larger, prospective study.

    In conclusion, the prevalence of DNA content abnormality increases with increasing histological grade of dysplasia and is indicative of a higher risk for subsequent detection of HGD or OAC in patients with LGD or IND histology. If further validated in additional studies, our findings support the use of DNA flow cytometry from FFPE tissue to confirm a morphological impression or suspicion of HGD and to identify a subset of patients with LGD or IND who may be at increased risk for subsequent detection of HGD or OAC.

    References

    1. ↵
      2. Spechler SJ ,
      3. Sharma P ,
      4. Souza RF , et al
      . American Gastroenterological Association technical review on the management of Barrett’s esophagus. Gastroenterology 2011;140:e18–e52.doi:10.1053/j.gastro.2011.01.031
      OpenUrlPubMedWeb of Science
    2. ↵
      2. Wang KK ,
      3. Sampliner RE
      . Practice Parameters Committee of the American College of Gastroenterology. Updated guidelines 2008 for the diagnosis, surveillance and therapy of Barrett’s esophagus. Am J Gastroenterol 2008;103:788–97.doi:10.1111/j.1572-0241.2008.01835.x
      OpenUrlCrossRefPubMedWeb of Science
    3. ↵
      2. Naini BV ,
      3. Souza RF ,
      4. Odze RD
      . Barrett’s Esophagus: A Comprehensive and Contemporary Review for Pathologists. Am J Surg Pathol 2016;40:e45–66.doi:10.1097/PAS.0000000000000598
      OpenUrl
    4. ↵
      2. Chandrasoma P ,
      3. Wickramasinghe K ,
      4. Ma Y , et al
      . Is intestinal metaplasia a necessary precursor lesion for adenocarcinomas of the distal esophagus, gastroesophageal junction and gastric cardia? Dis Esophagus 2007;20:36–41.doi:10.1111/j.1442-2050.2007.00638.x
      OpenUrlCrossRefPubMedWeb of Science
    5. ↵
      2. Goldblum JR
      . Current issues in Barrett’s esophagus and Barrett’s-related dysplasia. Mod Pathol 2015;28(Suppl 1):S1–S6.doi:10.1038/modpathol.2014.125
      OpenUrl
    6. ↵
      2. Shaheen NJ ,
      3. Falk GW ,
      4. Iyer PG , et al
      . ACG clinical guideline: diagnosis and management of Barrett’s Esophagus. Am J Gastroenterol 2016;111:30–50.doi:10.1038/ajg.2015.322
      OpenUrlCrossRefPubMed
    7. ↵
      2. Galipeau PC ,
      3. Cowan DS ,
      4. Sanchez CA , et al
      . 17p (p53) allelic losses, 4N (G2/tetraploid) populations, and progression to aneuploidy in Barrett’s esophagus. Proc Natl Acad Sci U S A 1996;93:7081–4.doi:10.1073/pnas.93.14.7081
      OpenUrlAbstract/FREE Full Text
    8. ↵
      2. Galipeau PC ,
      3. Prevo LJ ,
      4. Sanchez CA , et al
      . Clonal expansion and loss of heterozygosity at chromosomes 9p and 17p in premalignant esophageal (Barrett’s) tissue. J Natl Cancer Inst 1999;91:2087–95.doi:10.1093/jnci/91.24.2087
      OpenUrlCrossRefPubMedWeb of Science
    9. ↵
      2. Brankley SM ,
      3. Fritcher EG ,
      4. Smyrk TC , et al
      . Fluorescence in situ hybridization mapping of esophagectomy specimens from patients with Barrett’s esophagus with high-grade dysplasia or adenocarcinoma. Hum Pathol 2012;43:172–9.doi:10.1016/j.humpath.2011.04.018
      OpenUrlPubMed
    10. ↵
      2. Ross-Innes CS ,
      3. Becq J ,
      4. Warren A , et al
      . Whole-genome sequencing provides new insights into the clonal architecture of Barrett’s esophagus and esophageal adenocarcinoma. Nat Genet 2015;47:1038–46.doi:10.1038/ng.3357
      OpenUrlCrossRefPubMed
    11. ↵
      2. Varghese S ,
      3. Newton R ,
      4. Ross-Innes CS , et al
      . Analysis of dysplasia in patients with Barrett’s esophagus based on expression pattern of 90 genes. Gastroenterology 2015;149:1511–8.doi:10.1053/j.gastro.2015.07.053
      OpenUrl
    12. ↵
      2. Walch AK ,
      3. Zitzelsberger HF ,
      4. Bruch J , et al
      . Chromosomal imbalances in Barrett’s adenocarcinoma and the metaplasia-dysplasia-carcinoma sequence. Am J Pathol 2000;156:555–66.doi:10.1016/S0002-9440(10)64760-8
      OpenUrlPubMedWeb of Science
    13. ↵
      2. Reid BJ ,
      3. Haggitt RC ,
      4. Rubin CE , et al
      . Observer variation in the diagnosis of dysplasia in Barrett’s esophagus. Hum Pathol 1988;19:166–78.doi:10.1016/S0046-8177(88)80344-7
      OpenUrlCrossRefPubMedWeb of Science
    14. ↵
      2. Montgomery E ,
      3. Bronner MP ,
      4. Goldblum JR , et al
      . Reproducibility of the diagnosis of dysplasia in Barrett esophagus: a reaffirmation. Hum Pathol 2001;32:368–78.doi:10.1053/hupa.2001.23510
      OpenUrlCrossRefPubMedWeb of Science
    15. ↵
      2. Vennalaganti P ,
      3. Kanakadandi V ,
      4. Goldblum JR , et al
      . Discordance among Pathologists in the United States and Europe in diagnosis of Low-Grade dysplasia for patients with Barrett’s Esophagus. Gastroenterology 2017;152:564–70.doi:10.1053/j.gastro.2016.10.041
      OpenUrl
    16. ↵
      2. Sangle NA ,
      3. Taylor SL ,
      4. Emond MJ , et al
      . Overdiagnosis of high-grade dysplasia in Barrett’s esophagus: a multicenter, international study. Mod Pathol 2015;28:758–65.doi:10.1038/modpathol.2015.2
      OpenUrlCrossRefPubMed
    17. ↵
      2. Sonwalkar SA ,
      3. Rotimi O ,
      4. Scott N , 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;56:900–7.doi:10.1111/j.1365-2559.2010.03571.x
      OpenUrlCrossRefPubMedWeb of Science
    18. ↵
      2. Moole H ,
      3. Patel J ,
      4. Ahmed Z , et al
      . Progression from low-grade dysplasia to malignancy in patients with Barrett’s esophagus diagnosed by two or more pathologists. World J Gastroenterol 2016;22:8831–43.doi:10.3748/wjg.v22.i39.8831
      OpenUrl
    19. ↵
      2. Kaye PV ,
      3. Haider SA ,
      4. Ilyas M , et al
      . Barrett’s dysplasia and the Vienna classification: reproducibility, prediction of progression and impact of consensus reporting and p53 immunohistochemistry. Histopathology 2009;54:699–712.doi:10.1111/j.1365-2559.2009.03288.x
      OpenUrlCrossRefPubMedWeb of Science
    20. ↵
      2. Kastelein F ,
      3. Biermann K ,
      4. Steyerberg EW , et al
      . Value of α-methylacyl-CoA racemase immunochemistry for predicting neoplastic progression in Barrett’s oesophagus. Histopathology 2013;63:630–9.doi:10.1111/his.12216
      OpenUrlPubMed
    21. ↵
      2. Sikkema M ,
      3. Kerkhof M ,
      4. Steyerberg EW , 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;104:2673–80.doi:10.1038/ajg.2009.437
      OpenUrlCrossRefPubMedWeb of Science
    22. ↵
      2. Kerkhof M ,
      3. Steyerberg EW ,
      4. Kusters JG , et al
      . Aneuploidy and high expression of p53 and Ki67 is associated with neoplastic progression in Barrett esophagus. Cancer Biomark 2008;4:1–10.doi:10.3233/CBM-2008-4101
      OpenUrlPubMedWeb of Science
    23. ↵
      2. Kastelein F ,
      3. Biermann K ,
      4. Steyerberg EW , et al
      . Aberrant p53 protein expression is associated with an increased risk of neoplastic progression in patients with Barrett’s oesophagus. Gut 2013;62:1676–83.doi:10.1136/gutjnl-2012-303594
      OpenUrlAbstract/FREE Full Text
    24. ↵
      2. Panarelli NC ,
      3. Yantiss RK
      . Do Ancillary Studies Aid detection and classification of Barrett Esophagus? Am J Surg Pathol 2016;40:e83–e93.doi:10.1097/PAS.0000000000000654
      OpenUrl
    25. ↵
      2. Reid BJ ,
      3. Haggitt RC ,
      4. Rubin CE , et al
      . Barrett’s esophagus. Correlation between flow cytometry and histology in detection of patients at risk for adenocarcinoma. Gastroenterology 1987;93:1–11.
      OpenUrlPubMedWeb of Science
    26. ↵
      2. Reid BJ ,
      3. Levine DS ,
      4. Longton G , et al
      . Predictors of progression to Cancer in Barrett’s esophagus: baseline histology and flow cytometry identify low- and high-risk patient subsets. Am J Gastroenterol 2000;95:1669–76.doi:10.1016/S0002-9270(00)00988-6
      OpenUrlPubMedWeb of Science
    27. ↵
      2. Rabinovitch PS ,
      3. Longton G ,
      4. Blount PL , et al
      . Predictors of progression in Barrett’s esophagus III: baseline flow cytometric variables. Am J Gastroenterol 2001;96:3071–83.doi:10.1111/j.1572-0241.2001.05261.x
      OpenUrlCrossRefPubMedWeb of Science
    28. ↵
      2. Giménez A ,
      3. Minguela A ,
      4. Parrilla P , et al
      . Flow cytometric DNA analysis and p53 protein expression show a good correlation with histologic findings in patients with Barrett’s esophagus. Cancer 1998;83:641–51.doi:10.1002/(SICI)1097-0142(19980815)83:4<641::AID-CNCR3>3.0.CO;2-N
      OpenUrlCrossRefPubMed
    29. ↵
      2. Montgomery EA ,
      3. Hartmann DP ,
      4. Carr NJ , et al
      . Barrett esophagus with dysplasia. Flow cytometric DNA analysis of routine, paraffin-embedded mucosal biopsies. Am J Clin Pathol 1996;106:298–304.doi:10.1093/ajcp/106.3.298
      OpenUrlCrossRefPubMed
    30. ↵
      2. Shankey TV ,
      3. Rabinovitch PS ,
      4. Bagwell B , et al
      . Guidelines for implementation of clinical DNA cytometry. International Society for Analytical Cytology. Cytometry 1993;14:472–7.doi:10.1002/cyto.990140503
      OpenUrlCrossRefPubMedWeb of Science
    31. ↵
      2. Rabinovitch PS
      . Practical considerations for DNA content and cell cycle analysis. In: Bauer KD , Duque RE , Shankey TV , eds. Clinical flow cytometry: principles and applications. Baltimore: Williams and Wilkins, 1992:117–42.
    32. ↵
      2. Buttar NS ,
      3. Wang KK ,
      4. Sebo TJ , et al
      . Extent of high-grade dysplasia in Barrett’s esophagus correlates with risk of adenocarcinoma. Gastroenterology 2001;120:1630–9.doi:10.1053/gast.2001.25111
      OpenUrlCrossRefPubMedWeb of Science
    33. ↵
      2. Wild CP ,
      3. Hardie LJ ,
      4. Reflux HLJ
      . Reflux, Barrett’s oesophagus and adenocarcinoma: burning questions. Nat Rev Cancer 2003;3:676–84.doi:10.1038/nrc1166
      OpenUrlCrossRefPubMedWeb of Science
    34. ↵
      2. Whitson MJ ,
      3. Falk GW
      . Predictors of Progression to High-Grade Dysplasia or Adenocarcinoma in Barrett’s Esophagus. Gastroenterol Clin North Am 2015;44:299–315.doi:10.1016/j.gtc.2015.02.005
      OpenUrl
    35. ↵
      2. Zhu W ,
      3. Appelman HD ,
      4. Greenson JK , et al
      . A histologically defined subset of high-grade dysplasia in Barrett mucosa is predictive of associated carcinoma. Am J Clin Pathol 2009;132:94–100.doi:10.1309/AJCP78CKIOJWOVFN
      OpenUrlCrossRefPubMed
    36. ↵
      2. Skinner DB ,
      3. Walther BC ,
      4. Riddell RH , et al
      . Barrett’s esophagus. Comparison of benign and malignant cases. Ann Surg 1983;198:554–65.
      OpenUrlCrossRefPubMedWeb of Science
    37. ↵
      2. Haggitt RC ,
      3. Tryzelaar J ,
      4. Ellis FH , et al
      . Adenocarcinoma complicating columnar epithelium-lined (Barrett’s) esophagus. Am J Clin Pathol 1978;70:1–5.doi:10.1093/ajcp/70.1.1
      OpenUrlCrossRefPubMed
    38. ↵
      2. Schmidt HG ,
      3. Riddell RH ,
      4. Walther B , et al
      . Dysplasia in Barrett’s esophagus. J Cancer Res Clin Oncol 1985;110:145–52.doi:10.1007/BF00402729
      OpenUrlCrossRefPubMedWeb of Science
    39. ↵
      2. Williamson WA ,
      3. Ellis FH ,
      4. Gibb SP , et al
      . Barrett’s esophagus. Prevalence and incidence of adenocarcinoma. Arch Intern Med 1991;151:2212–6.
      OpenUrlCrossRefPubMedWeb of Science
    40. ↵
      2. Menke-Pluymers MB ,
      3. Mulder AH ,
      4. Hop WC , et al
      . Dysplasia and aneuploidy as markers of malignant degeneration in Barrett’s oesophagus. The Rotterdam Oesophageal Tumour Study Group. Gut 1994;35:1348–51.doi:10.1136/gut.35.10.1348
      OpenUrlAbstract/FREE Full Text
    41. ↵
      2. Huang Q ,
      3. Yu C ,
      4. Zhang X , et al
      . Comparison of DNA histograms by standard flow cytometry and image cytometry on sections in Barrett’s adenocarcinoma. BMC Clin Pathol 2008;8:1–9.doi:10.1186/1472-6890-8-5
      OpenUrlCrossRefPubMed
    42. ↵
      2. Bird-Lieberman EL ,
      3. Dunn JM ,
      4. Coleman HG , et al
      . Population-based study reveals new risk-stratification biomarker panel for Barrett’s esophagus. Gastroenterology 2012;143:927–35.doi:10.1053/j.gastro.2012.06.041
      OpenUrlCrossRefPubMedWeb of Science
    43. ↵
      2. Phoa KN ,
      3. van Vilsteren FG ,
      4. Weusten BL , et al
      . Radiofrequency ablation vs endoscopic surveillance for patients with Barrett esophagus and low-grade dysplasia: a randomized clinical trial. JAMA 2014;311:1209–17.doi:10.1001/jama.2014.2511
      OpenUrlCrossRefPubMedWeb of Science
    44. ↵
      2. Wani S ,
      3. Rubenstein JH ,
      4. Vieth M , et al
      . Diagnosis and Management of Low-Grade Dysplasia in Barrett’s Esophagus: Clinical Practice Updates Expert Review From the Clinical Guidelines Committee of the American Gastroenterological Association. Gastroenterology 2016;151:822–35.
      OpenUrl
    45. ↵
      2. Choi WT ,
      3. Emond MJ ,
      4. Rabinovitch PS , et al
      . "Indefinite for Dysplasia" in Barrett’s Esophagus: Inflammation and DNA Content Abnormality are Significant Predictors of Early Detection of Neoplasia. Clin Transl Gastroenterol 2015;6:e81.doi:10.1038/ctg.2015.7
    46. ↵
      2. Horvath B ,
      3. Singh P ,
      4. Xie H , et al
      . Risk for esophageal neoplasia in Barrett’s esophagus patients with mucosal changes indefinite for dysplasia. J Gastroenterol Hepatol 2015;30:262–7.doi:10.1111/jgh.12696
      OpenUrlCrossRefPubMed
    47. ↵
      2. Dunn JM ,
      3. Mackenzie GD ,
      4. Oukrif D , et al
      . Image cytometry accurately detects DNA ploidy abnormalities and predicts late relapse to high-grade dysplasia and adenocarcinoma in Barrett’s oesophagus following photodynamic therapy. Br J Cancer 2010;102:1608–17.doi:10.1038/sj.bjc.6605688
      OpenUrlCrossRefPubMed
    48. ↵
      2. Wang Y ,
      3. McManus DT ,
      4. Arthur K , et al
      . Whole slide image cytometry: a novel method to detect abnormal DNA content in Barrett’s esophagus. Lab Invest 2015;95:1319–30.doi:10.1038/labinvest.2015.98
      OpenUrl
    49. ↵
      2. Yu C ,
      3. Zhang X ,
      4. Huang Q , et al
      . High-fidelity DNA histograms in neoplastic progression in Barrett’s esophagus. Lab Invest 2007;87:466–72.doi:10.1038/labinvest.3700531
      OpenUrlPubMed
    50. ↵
      2. Fang M ,
      3. Lew E ,
      4. Klein M , et al
      . DNA abnormalities as marker of risk for progression of Barrett’s esophagus to adenocarcinoma: image cytometric DNA analysis in formalin-fixed tissues. Am J Gastroenterol 2004;99:1887–94.doi:10.1111/j.1572-0241.2004.30886.x
      OpenUrlCrossRefPubMedWeb of Science
    51. ↵
      2. Desai TK ,
      3. Krishnan K ,
      4. Samala N , et al
      . The incidence of oesophageal adenocarcinoma in non-dysplastic Barrett’s oesophagus: a meta-analysis. Gut 2012;61:970–6.doi:10.1136/gutjnl-2011-300730
      OpenUrlAbstract/FREE Full Text

    Footnotes

    • Contributors W-TC, J-HT, PSR, ANM and SK contributed to the study concept and design, analysis and interpretation of data, and drafting of the manuscript. W-TC, J-HT and TS performed the experiments. DH contributed to analysis and interpretation of data as well as statistical analysis. All authors have read and approved the manuscript for publication.

    • Funding UCSF Department of Pathology.

    • Competing interests None declared.

    • Ethics approval University of California at San Francisco (UCSF).

    • Provenance and peer review Not commissioned; externally peer reviewed.

    • Correction notice This paper has been amended since it was published Online First. Owing to a scripting error, some of the publisher names in the references were replaced with ’BMJ Publishing Group'. This only affected the full text version, not the PDF. We have since corrected these errors and the correct publishers have been inserted into the references.