Ann Thorac Surg 2001;72:1130-1135
© 2001 The Society of Thoracic Surgeons
Original article: general thoracic
Correlation between dysplasia and mutations of six tumor suppressor genes in Barrett’s esophagus
Siva Raja, BSa,
Sydney D. Finkelstein, MDc,
Fabien K. Baksh, MDc,
William E. Gooding, MSb,
Patricia A. Swalsky, BSc,
Tony E. Godfrey, PhDa,
Percival O. Buenaventura, MDa,
James D. Luketich, MDa
a Division of Thoracic Surgery, Department of Surgery, University of Pittsburgh Medical Center Health System, Pittsburgh, Pennsylvania, USA
b Department of Biostatistics, University of Pittsburgh Medical Center Health System, Pittsburgh, Pennsylvania, USA
c Department of Pathology, University of Pittsburgh Medical Center Health System, Pittsburgh, Pennsylvania, USA
Address reprint requests to Dr Luketich, Section of Thoracic Surgery, University of Pittsburgh Medical Center, C800, PUH, 200 Lothrop St, Pittsburgh, PA 15213
e-mail: luketichjd{at}msx.upmc.edu
Presented at the Thirty-seventh Annual Meeting of The Society of Thoracic Surgeons, New Orleans, LA, Jan 29–31, 2001.
 |
Abstract
|
|---|
Background. Barrett’s esophagus (BE) may progress to adenocarcinoma through dysplastic progression. Classification of dysplasia in BE has significant interobserver variability. Our objective was to determine whether genetic alterations in BE correlate with degrees of histologic dysplasia.
Methods. Fixed tissue from 37 patients with BE and adenocarcinoma was studied for six tumor suppressor genes. Tissues were microdissected and analyzed for loss of heterozygosity. Microdissection of individual crypts showing metaplasia and dysplasia were performed and analyzed for 23 of the 37 patients whose tumors were heterozygous for at least four of the six genes studied.
Results. Frequency of alterations for MXI1, hOGG1, p53, MTS1, DCC, and APC were 7 of 32 (22%), 12 of 35 (34%), 12 of 26 (46%), 17 of 30 (57%), 17 of 27 (63%), and 23 of 36 (64%), respectively. Analysis of BE demonstrated that crypts with metaplasia, low-grade dysplasia, and high-grade dysplasia strongly correlated with alterations in tumor suppressor genes (p < 0.0001).
Conclusions. This pilot study demonstrates that genetic analysis can be performed on individual crypts in patients with BE, and that alterations may facilitate objective classification of the severity of dysplasia.
|
Introduction
|
|---|
Adenocarcinoma of the esophagus accounts for approximately 11,500 deaths annually in the United States [1]. Over the past 10 years there has been a dramatic increase in both the incidence and mortality from esophageal adenocarcinoma [2]. Unfortunately, most patients with esophageal cancer are diagnosed with advanced regional lymph node involvement or distant metastasis. It has been demonstrated that surgical resection is most likely to be curative when the disease is limited to only high-grade dysplasia or a minimally invasive tumor [3]. Thus efforts to reduce mortality have centered on early detection of disease.
The risk of developing esophageal cancer is increased (30- to 125-fold) [4] in the presence of gastroesophageal reflux disease (GERD) combined with the metaplastic transformation of the esophageal squamous lining into glandular columnar mucosa referred to as Barrett’s metaplasia (BE) [5]. The diagnosis of BE leads to periodic surveillance endoscopy with biopsy [4]; if progression to high-grade dysplasia occurs, surgical resection is generally recommended. From a histopathologic perspective, a particularly challenging problem is the accurate and reliable diagnosis and classification of Barrett’s metaplasia and dysplasia. There is considerable interobserver variation, even among experts in the field, which is compounded by the significant sampling variation inherent in biopsy of a macroscopic field of at-risk mucosa [6]. Objective diagnosis and classification of Barrett’s dysplasia based on molecular characteristics offers the potential to contribute to a more consistent diagnosis of high-grade dysplasia, and perhaps to determine risk factors for progression from simple BE.
The molecular pathogenesis of cancer has been shown to result from an accumulation of multiple genetic alterations in a multistep fashion over time. In esophageal cancer, loss of heterozygosity of tumor suppressor genes such as p53, APC the adenomatous polyposis coli gene (APC), the gene deleted in colon cancer (DCC), and MTS1 (p16) have been reported in numerous previous studies [7–9]. Additionally, alterations in other genes such as MXI1 (MAX interacting protein) and human OGG1 have been implicated in other epithelial cell malignancies [10, 11]. Since adenocarcinoma of the esophagus evolves from dysplastic cells in BE, it is likely that progression from metaplasia to cancer follows a predictable and quantifiable pattern of acquisition of genetic alterations. The aim of this study was to determine whether loss of heterozygosity in individual crypts of BE in the mucosa adjacent to an existing adenocarcinoma would be present in proportion to the degree of histologic dysplasia.
|
Material and methods
|
|---|
Patients and specimens
All studies were conducted with the approval of the institutional review board of the University of Pittsburgh Medical Center (UPMC). Archived samples from 37 patients who underwent esophageal resection at UPMC for primary adenocarcinoma arising from BE between 1997 and 1999 were studied. Slides were reviewed and blocks selected showing carcinoma at its deepest point of invasion as well as associated Barrett’s metaplasia and dysplasia when present. Intestinalized crypts were selected in each case showing histologic features classified as metaplasia without dysplasia, indeterminate for dysplasia, low-grade dysplasia, and high-grade dysplasia. Eight serial 4-µm sections were prepared to serve as a basis for microdissection genotyping. The analysis also included five examples of esophageal ulceration and associated inflammatory change without evidence of dysplasia and malignancy. The latter group was included to evaluate gene damage in nonneoplastic tissue.
Microdissection and DNA extraction
All microdissections were performed by the same pathologist (S.D.F.) using a stereomicroscope (Zeiss SZ-40; Zeiss, Oberkochen, Germany). Tissue extraction from formalin-fixed, paraffin-embedded sections was performed as described previously [12] (Fig 1A and 1B). Briefly, sequential 4-µm-thick sections were cut and one was stained with hematoxylin and eosin to localize lesions of interest for microdissection. Paraffin-embedded sections were deparaffinized in xylene (2x3 minutes) and washed in 100% ethanol (2x3 minutes). In each case, more than 85% neoplastic (> 500 cells) and paired corresponding normal tissues were taken. Similarly, individual crypts of varying levels of dyplasia were microdissected over six to eight serial sections. Microdissected tissues were incubated at 37°C overnight with 10 mg/mL proteinase K. The products were stored at -20°C until polymerase chain reaction (PCR) amplification.
View larger version (71K):
[in this window]
[in a new window]
|
Fig 1. (A) Hematoxylin and eosin–stained section of esophagus with microdissection of adenocarcinoma (A), crypt of Barrett’s esophagus (C), and normal skeletal muscle tissue (N). (B) Serial hematoxylin and eosin–stained sections showing crypt microdissection (arrow).
|
|
PCR amplification and DNA isolation
PCR was performed on an Omn-E thermal cycler (Hybaid, Franklin, MA) using 1 µL of genomic DNA, 10 pmol of each primer (Table 1), 300 nmol/L of deoxynucleotide triphosphates, 1.5 mmol/L magnesium chloride, 0.25 U of TaqGold DNA polymerase, and 5 µL of 10xPCR buffer (Perkin Elmer Cetus, Norwalk, CT) in a final volume of 50 µL. After initial denaturation at 95°C for 10 minutes, 35 cycles PCR were carried out as follows: denaturation at 94°C for 1 minute, annealing at 58°C for 1 minute, and polymerization at 74°C for 1 minute, with a final 10-minute extension at 74°C. PCR-generated DNA was isolated through electrophoresis in 2% agarose gel. The target bands of each gene were cut from agarose gel under ultraviolet light transilluminator and dissolved in 1 µL of 50xGelase buffer (Epicentre Technologies, Madison, WI) at 95°C for 8 minutes, followed by incubation with 1 U of Gelase enzyme at 45°C for 1 hour.
Direct DNA sequencing
DNA sequencing for amplified PCR products of APC, OGG1, DCC, and p53 was performed using T7 Sequenase, Version 2.0, DNA sequencing kit (Amersham Life Science, Cleveland, OH), according to the manufacturer’s instructions. Each tumor suppressor gene except for MTS1 and MXI1 was originally sequenced with four lanes (A, G, C, and T); afterward only two lanes showing the polymorphic bands were necessary to load on the sequence gel. The upstream primers of APC, DCC, and OGG1 and the downstream primer of p53 (Table 1) served as the sequencing primers. For detecting loss of heterozygosity (LOH) in the microsatellites, aliquots of the crude lysate were added into the reaction mixture (10 µL) containing 2.5 µCi of [
-33P] dNTP (dATP, dGTP, dCTP, dTTP; New Life Science Products, Boston, MA), and 0.25 µL of Ampli Taq DNA polymerase (Perkin Elmer Cetus, Norwalk, CT). These were run through 6% polyacrylamide gel electrophoresis at 2000 V for 1 to 2 hours. The gel was dried at 80°C for 40 minutes on a Savant gel dryer SG210D (Savant Instruments Inc, Holbrook, NY) and exposed overnight to Kodak X-Omat AR-2 film (Eastman Kodak, Rochester, NY). The sequence was read visually with the use of a light box.
Definition of LOH
On an autoradiograph of the sequencing gel, heterozygosity for a locus is visualized as the presence of two bands of equal fragment size on lanes corresponding to two different nucleotides (Fig 2). Films were digitized using the Gel Doc 2000 (Bio-Rad Laboratories, Hercules, CA) and densitometry was performed on the appropriate bands and the resultant data were analyzed using the Quantity-one software (Bio-Rad Laboratories). LOH was determined by the method described by MacGrogan and colleagues [13] with the following difference. The cut off value for LOH was set at 1.4 for the ratio of ratios (based on the histogram of all the values from all the loci in this study). The same gels were also analyzed through subjective means by at least two different observers to guard against experimental artifact.
View larger version (35K):
[in this window]
[in a new window]
|
Fig 2. Typical autoradiograph showing the loss of heterozygosity (LOH) in a pentanucleotide polymorphism from a tumor sample (right) relative to a normal sample (left) from the same patient.
|
|
Methods for crypt LOH analysis
Given the small size of the fixative treated, microdissected metaplastic/dysplastic intestinalized crypt samples for genotyping, preliminary steps were taken to determine the minimum amount of sample sufficient to afford balanced representative PCR thereby avoiding allelic dropout. This was done using a 5' fluorogenic assay for ß-actin and standard curve using human genomic DNA. Each microdissected sample was run against a standard derived from freshly extracted DNA ranging from 0.08 to 4 ng of DNA. This was used to control the DNA input into each reaction.
Genetic analysis
The genes of interest were studied using direct DNA sequencing of intragenic sequence polymorphism or microsatellite analysis. LOH in DCC, p53, hOGG1, and APC genes were studied by both techniques. A loss at either the intragenic loci or at the associated microsatellite loci was sufficient to determine loss of the allele. For the genes MTS1 and MXI1 a loss in at least one of the two flanking microsatellite loci was sufficient to determine loss.
|
Results
|
|---|
Esophageal adenocarcinomas were notable for the presence of frequent allelic imbalance affecting the genes studied (Table 2). The mean number of heterozygous gene loci per patient was 5.05 (median 5.00) and the mean number of genes demonstrating LOH was 2.52 (median 2.00). There were three patients who did not show any loss out of the 3, 4, and 5 heterozygous gene loci respectively. There were five samples of esophageal mucosa that were diagnosed as having an inflammatory appearance. These cases did not show allelic imbalance in the genes from our panel. No single gene in our panel was found to be universally mutated or nonmutated in esophageal cancer. Thus reliance on any single gene to indicate the presence or absence of dysplasia or malignancy could not be established. Instead, a combination of gene targets in the form of a panel could achieve the high degree of association with the process of carcinogenesis.
Determination of amplifiable DNA using a standard curve of freshly extracted DNA indicated that microdissection samples varied significantly in their respective amounts of amplifiable DNA. Although certain histologic features such as necrosis correlated with diminished amounts of amplifiable DNA, in many cases cellular morphologic features could not explain the difference in DNA content. Thus it was necessary to determine the amount of amplifiable DNA in each microdissected sample to ensure that adequate quantities of DNA were present in each PCR to avoid allelic dropout [14, 15]. Values of less than 100 pg yielded variable results that could be interpreted as loss of heterozygosity. The polymorphic markers used in this study varied in their requirement for minimum threshold levels of amplifiable DNA from 40 to 100 pg of DNA. To simplify the processing of specimens, a value of 100 pg of amplifiable DNA was applied to each PCR reaction.
Intestinalized crypts of Barrett’s esophagus were microdissected from esophageal resection specimens bearing invasive adenocarcinoma. These microdissected crypts manifested cellular changes ranging from metaplasia without dysplasia to high-grade dysplasia. Several observations on this individual crypt analyses were notable. First, the mutational profile of individual crypts differed significantly from one another, in keeping with independent clonal expansion of mutated clones of dysplastic cells. Also, the genotype of dysplastic Barrett crypts did not necessarily mirror that of the invasive adenocarcinoma that had developed in the esophagus. The findings were consistent with a field effect producing multiple independent clones of tumor cells with one progressing more rapidly into malignancy and accounting for the malignancy in each case. There was a close and strong correlation (p < 0.001, Jonckherre-Terpstra Test) between the extent of allelic loss and grade of dysplasia in Barrett’s esophagus (Table 3). Intestinalized crypts showing metaplasia without dysplasia in all cases but one were without allelic loss. In contrast, most crypts bearing high-grade dysplastic epithelium manifest two or more different allelic loss events. Low-grade dysplasia displayed intermediate levels of mutational change.
|
Comment
|
|---|
The diagnosis of Barrett’s high-grade dysplasia has serious implications. High-grade dysplasia is currently treated by esophagectomy [16] or, in nonsurgical candidates, with ablative therapies such as mucosal stripping or photodynamic therapy [17]. This aggressive approach is taken because of the high risk for developing subsequent adenocarcinomas or already harboring a focus of invasive cancer in the esophagus of patients with high-grade dysplasia [2]. Therefore, given this serious implication, it is vital to make an accurate diagnosis of high-grade dysplasia for BE. Yet there is considerable interobserver variation even among experts in the field, which is compounded by the significant sampling variation inherent in biopsy of a macroscopic field of at-risk mucosa. In a study by Reid and colleagues [6], the interobserver agreement for the diagnosis of high-grade versus lower grades of dysplasia and no dysplasia was only 86%. Furthermore, the concordance for dyplasia versus no dyplasia was even lower at 72%. Inflammatory changes in the metaplastic mucosa can further confound the distinction between reactive hyperplasia and dysplasia. The more objective diagnosis and classification of Barrett’s dysplasia based on molecular characteristics may offer the potential for better diagnosis and treatment planning.
Our multigene analysis of esophageal adenocarcinoma arising in the context of Barrett’s metaplasia demonstrates a striking multiplicity of tumor suppressor gene loss. Recognizing that the genes we selected for study of allelic loss determination represent only a small proportion of known tumor suppressor genes, it was noteworthy to find that even this limited panel of 6 markers were significantly altered in the tumor. It is clear that the panel of genes used in this study may not optimally include those specific genetic targets most intimately involved in esophageal tumorigenesis. Nevertheless, this introductory panel can be employed to study both esophageal dysplasia and carcinoma with the expectation that it may include useful molecular markers for diagnostic and prognostic purposes. Our data suggest that molecular quantitiation of the frequency of LOH in individual intestinalized crypts can potentially be used to classify the microdissected sample as nondysplastic, low-grade dysplasia, or high-grade dysplasia. Moreover, the indeterminate category may potentially be divided into true dysplasia with allelic loss versus nondysplastic crypts lacking mutational change (Table 2). The observation that increasing numbers of mutations correlate with the degree of dysplasia in BE was highly statistically significant (p < 0.0001).
Our data support the contention that loss of tumor suppressor genes is widespread with respect to the involvement of specific genes, that it may be expected to affect many well-characterized genes, and that it is not due to a single genetic alteration. This study adds strength to the current paradigm that the evolution of this tumor from BE may not follow a linear pathway [18]. It might occur through several pathways, or they could converge onto a single pathway at a gene that has yet to be identified or investigated.
The creation of a panel of genes that could be used to distinguish dysplasia and tumor from benign processes would appear to be supported by our genotyping results. In our panel of six genes, with the exception of 3 cases, every patient had at least one altered gene. We also demonstrated the feasibility for performing this type of analysis on very small amount of tissue from microdissected individual columnar lined crypts. The importance of this finding lies in the potential for this type of analysis to be applied to biopsy specimens. By demonstrating that it is possible to microdissect single crypts, we can specifically target those crypts that pose a problem for conventional histopathologic evaluation.
There are some limitations that need to be recognized. First, the amount of DNA available for analysis is small, limiting the genotyping analysis to approximately 8 to 10 separate loci. As such, the presence of heterozygosity would need to be established before sample testing. This can be achieved by initially evaluating germline DNA from the cellular component of blood to establish heterozygous loci for the gene targets. Alternative gene targets would need to be available to substitute for those loci that prove to be homozygous for an individual patient. The results of this study indicate that such an approach is feasible using the markers employed in this analysis. With the sequencing of the human genome, it can be expected that additional polymorphic loci as well as new gene targets providing greater frequency of heterozygosity with respect to esophageal cancer development and progression, will be part of an optimized panel of genotyping markers for this cancer.
The most significant aspect of this work is the strategy used to integrate tissue histopathology and molecular genetics of cancer for diagnostic purposes. By creating a panel of cancer gene analysis, specific questions in the molecular pathogenesis of cancer can be addressed. The genotyping analysis carried out on microdissected, fixed tissue specimens enabled precise sample selection for specific genetic testing. In this way the molecular changes underlying specific cellular events can be defined in a manner that is both objective and especially meaningful, given the causative relationship between gene damage and carcinogenesis. The task remains to define the specific gene targets that will be used in this regard and the best methodology enabling the genotyping to be performed in a rapid, simple, reliable manner without impeding established pathology practice. The results of this work provide a basis on which to pursue this objective.
|
Discussion
|
|---|
DR STEVE R. DeMEESTER (Los Angeles, CA): That was a very nice presentation.
I found it interesting that every patient had tumor, I assume, and within each patient you were able to find areas of no dysplasia, low-grade, high-grade, and the tumor itself. Often you can’t even find Barrett’s in patients with adenocarcinoma, much less find each of those different tissues. How often could you not identify all of those different gradations of dysplasia in a patient, or were these assembled from many patients?
Additionally we have noted that there seems to be quite a field effect. We have not been able to find such a disparity between the genetic changes in low-grade versus high-grade dysplasia in these patients. It is more of a field effect. It is universal throughout the tumor and the Barrett’s tissue. So I was surprised that you found such a big difference. Is this perhaps because the samples are from different patients and there is a lot of variability between patients?
In addition, did you look at methylation at all? We know that genes can be silenced by methylation independently of loss of the gene. Have you looked at that?
Last, it strikes me as a little bit of a circular argument, one of your conclusions that the genetic changes may help you determine the degree of dysplasia, because you categorized things based on dysplasia. If you’re saying that dysplasia is variable by pathologists, yet you use dysplasia to correlate or categorize your genetic changes, that is a bit of a circular argument. How do you get around that?
Thank you.
DR RAJA: To address your first question, in fact, we rarely found all the levels of dysplasia within the same patient. In fact, if you look at the table, there were 11 patients with negative dysplasia, whereas in fact 23 patients had been analyzed for the study, and the number does not add up. That’s because in a given patient we could only identify a certain number of these alterations, maybe high-grade dysplasia only or high-grade dysplasia and low-grade dysplasia without any negative dysplasia. It depends on what is available in the field at that time. So, in fact, we did not see all levels of dysplasia within the same specimen.
The second question, I do agree that it is a field effect, and in fact our data does support that hypothesis. The mutations that were seen in the dysplastic crypts were not necessarily the same genes that were lost in the cancer. Oftentimes they were, but that did not necessarily have to be the case. So we believe that that aspect of our data supports the field effect, meaning that all the cells in the gastroesophageal junction are exposed to the same insult and, hence, undergo similar processes.
Your third question regarding methylation, those studies are currently underway. And specifically one of the genes that we looked at, MTS1, also known as p16, is traditionally characterized as having methylation inactivation, and we found that that gene appeared to be heavily lost in our population of adenocarcinoma. But regarding the specific methylation studies, they are underway, and hopefully we will have the data for you soon.
Lastly, our goal is to standardize the process of calling something as high-grade dysplasia or low-grade dysplasia. We’re not trying to redefine the characteristics necessary for one to be high-grade or low-grade dysplasia. We believe that if we were to use genetic alterations (in fact, a number of genetic alterations, I guess not necessarily the specific genes that are altered), as demonstrated by the field effect that we have seen, I think it would result in better characterization and, hence, more standardized treatment. In the case of high-grade dysplasia, most studies have shown that greater than 50% of cases already have invasive carcinoma or intramucosal carcinoma at the time of diagnosis. So I think it’s essential that we standardize our analysis. And just to add one more thing to that, the study by Reid and colleagues showed that they had one sample which had heavy inflammation, and they found, the pathologists and the panel of experts, that the calling ranged from no dysplasia with inflammation all the way to intramucosal carcinoma on the same slide, and I think that makes a very strong point to some objective method that the pathologists could use as an adjunct to their visual observations.
DR SCOTT J. SWANSON (Boston, MA): I enjoyed your paper. I think it’s excellent work. I think I see where you’re heading, which is some sort of screening method for Barrett’s. My question would be, do you get enough tissue through a biopsy forceps from an endoscopy to be able to do this kind of analysis?
Thanks.
DR RAJA: Actually, our ultimate goal was to be able to utilize this test in the setting of biopsy samples. Biopsy specimens traditionally have several crypts, and based on our ability to individually microdissect out these samples, they provide enough genetic material for us to perform these studies. In fact, we had to do several optimization and validation studies before we proceeded with our analysis to show that we did not see a phenomenon that has been reported in the literature—something called artifactual loss of heterozygosity, a loss of heterozygosity that is due to experimental methods. I think that the next step for our project would be to move on to biopsy specimens from patients with varying levels of dysplasia and to see if a similar pattern of genetic alteration correlates well with their degree of dysplasia.
|
References
|
|---|
-
Ligresti R., Lightdale C.J. Diagnosis of esophagogastric tumors. Endoscopy 1998;30:73-79.[Medline]
-
Wright T.A. High-grade dysplasia in Barrett’s oesophagus. Br J Surg 1997;84:760-766.[Medline]
-
Lightdale C.J. Ablation therapy for Barrett’s esophagus: is it time to choose our weapons?. Gastrointest Endosc 1999;49:122-125.[Medline]
-
Falk G.W. Endoscopic surveillance of Barrett’s esophagus: risk stratification and cancer risk. Gastrointest Endosc 1999;49:S29-S34.[Medline]
-
Fitzgerald R.C., Triadafilopoulos G. Recent developments in the molecular characterization of Barrett’s esophagus. Dig Dis 1998;16:63-80.[Medline]
-
Reid B.J., Haggitt R.C., Rubin C.E., et al. Observer variation in the diagnosis of dysplasia in Barrett’s esophagus. Hum Pathol 1988;19:166-178.[Medline]
-
Wu T.T., Watanabe T., Heitmiller R., Zahurak M., Forastiere A.A., Hamilton S.R. Genetic alterations in Barrett esophagus and adenocarcinomas of the esophagus and esophagogastric junction region. Am J Pathol 1998;153:287-294.[Abstract/Free Full Text]
-
Barrett M.T., Galipeau P.C., Sanchez C.A., Emond M.J., Reid B.J. Determination of the frequency of loss of heterozygosity in esophageal adenocarcinoma by cell sorting, whole genome amplification and microsatellite polymorphisms. Oncogene 1996;12:1873-1878.[Medline]
-
van Dekken H., Geelen E., Dinjens W.N., et al. Comparative genomic hybridization of cancer of the gastroesophageal junction: deletion of 14Q31–32.1 discriminates between esophageal (Barrett’s) and gastric cardia adenocarcinomas. Cancer Res 1999;59:748-752.[Abstract/Free Full Text]
-
Wang D.Y., Xiang Y.Y., Li X.J., Hashimoto M., Tanaka M., Sugimura H. Mxi1 is a potential cellular target of carcinogens and frequently mutated in experimental rat tumors and tumor cell lines. Pathol Int 2000;50:373-383.[Medline]
-
Chevillard S., Radicella J.P., Levalois C., et al. Mutations in OGG1, a gene involved in the repair of oxidative DNA damage, are found in human lung, and kidney tumours. Oncogene 1998;16:3083-3086.[Medline]
-
Safatle-Ribeiro A.V., Ribeiro J.U., Reynolds J.C., et al. Morphologic, histologic, and molecular similarities between adenocarcinomas arising in the gastric stump, and the intact stomach. Cancer 1996;78:2288-2299.[Medline]
-
MacGrogan D., Levy A., Bostwick D., Wagner M., Wells D., Bookstein R. Loss of chromosome arm 8p loci in prostate cancer: mapping by quantitative allelic imbalance. Genes Chromosomes Cancer 1994;10:151-159.[Medline]
-
Dreesen J.C., Bras M., Coonen E., Dumoulin J.C., Evers J.L., Geraedts J.P. Allelic dropout caused by allele-specific amplification failure in single-cell PCR of the cystic fibrosis delta F508 deletion. J Assist Reprod Genet 1996;13:112-114.[Medline]
-
Finkelstein S.D., Przygodzki R., Swalsky P.A. Microdissectionbased p53 genotyping. Concepts for molecular testing. Mol Diagn 1998;3:179-191.[Medline]
-
Nguyen N.T., Schauer P., Luketich J.D. Minimally invasive esophagectomy for Barrett’s esophagus with high-grade dysplasia. Surgery 2000;127:284-290.[Medline]
-
Ackroyd R., Brown N.J., Davis M.F., et al. Photodynamic therapy for dysplastic Barrett’s oesophagus: a prospective, double blind, randomised, placebo controlled trial. Gut 2000;47:612-617.[Abstract/Free Full Text]
-
Barrett M.T., Sanchez C.A., Prevo L.J., et al. Evolution of neoplastic cell lineages in Barrett oesophagus. Nat Genet 1999;22:106-109.[Medline]
Top
Abstract
Introduction
