p53 Immunoreactivity in Barrett's metaplasia, dysplasia, and carcinoma

General Thoracic Surgery p53 Immunoreactivity in Barrett's metaplasia, dysplasia, and carcinoma Barrett's esophagus is a metaplastic condition with an unpredictable potential for neoplasia. Mutations of the tumor-suppressor gene p53 have been implicated in the evolution of some carcinomas. These mutations frequently result in intranuclear protein accumulation, which can be detected immunohistochemicauy. This study was undertaken to determine whether p53 immunoreactivity in Barrett's esophagus is a marker of neoplasia and, if so, when it occurs in the metaplasia-dysplasia-carcinoma sequence. Twenty-eight esophageal resection specimens were studied. Barrett's mucosa was present in each specimen, low-grade dysplasia in 27, high-grade dysplasia in 26, intramucosal cancer in 18, and submucosal cancer in 5. Immunohistochemical staining with the monoclonal antibody Pab1801 was used to detect the intranuclear protein product of mutated p53. No p53 immunoreactivity was seen in specimens of Barrett's mucosa or low-grade dysplasia. p53 immunoreactivity was found only in specimens of high-grade dysplasia, intramucosal cancer, and submucosal cancer. Sixty-nine percent (18/26) of these specimens exhibited mutated p53; 18 of 26 specimens of high-grade dysplasia (69 %), 12 of 18 intramucosal cancer specimens (67 %), and two of five submucosal cancer specimens (40 % ) expressed mutated p53. When p53 staining was observed, the spectrum of neoplastic changes (highgrade dysplasia, intramucosal cancer, submucosal cancer) within the specimen was positive. We conclude that (1) p53 immunoreactivity in Barrett's esophagus is a frequent, but not inclusive, marker for high-grade dysplasia, intramucosal cancer, and submucosal cancer and (2) immunoreactivity occurs late in the metaplasia-dysplasia-carcinoma sequence, during the transition to high-grade dysplasia. (J THoRAc CARDIOVASC SURG 1994;108:1132-7) Thomas W. Rice, MD,a John R. Goldblum, MDb (by invitation), Gary W. Falk, MDC (by invitation), Raymond R. Tubbs, DO b (by invitation), Thomas J. Kirby, MD a (by invitation), and Graham Casey, PhDd (by invitation), Cleveland, Ohio Rrrett's esophagus is an acquired metaplastic condition, caused by severe, long-standing gastroesophageal reflux. Barrett's esophagus is clinically important because From the Departments of Thoracic and ; Anatomical and Clinical Pathology,b Gastroenterology,C and Cancer Biology,d The Cleveland Clinic Foundation, Cleveland, Ohio. Read at the Seventy-fourth Annual Meeting of The American Association for Thoracic Surgery, New York, N.Y., April 24-27, 1994. Address for reprints: Thomas W. Rice, MD, Department of Thoracic and, Desk F25, The Cleveland Clinic Foundation, 9500 Euclid Ave., Cleveland, OR 44195. Copyright @ 1994 by Mosby-Year Book, Inc. 0022-5223/94 $3.00 + 0 12/6/58616 of its predisposition to complications, the most important of which is adenocarcinoma of the esophagus. The incidence of adenocarcinoma in studies of patients with Barrett's esophagus ranges from 1 case per 52 to 1 case per 441 patients. I - 4 The estimated risk of the development of adenocarcinoma is 30 to 125 times greater in patients with Barrett's esophagus than in the general population. I, 4-6 However, adenocarcinoma occurs in only a small percentage of these patients and, overall, the survival of patients with Barrett's esophagus is no different from that of the general population. I, 6 It is unclear what predisposes adenocarcinoma to develop in only a small subset of patients with Barrett's 1 132

The Journal of Thoracic and Volume 108, Number 6 Rice et al. 1 I 3 3 Fig. 1. Top, Section of Barrett's esophagus of specialized type, stained with hematoxylin and eosin, characterized by acid-mucin-secreting goblet cells. Bottom, Adjacent section of Barrett's esophagus of specialized type, stained with a monoclonal antibody for mutated p53. No intranuclear immunoreactivity is present in this focus. Fig. 2. Top, Section of Barrett's esophagus with low-grade dysplasia, stained with hematoxylin and eosin. The nuclei show slight nuclear pleomorphism, with hyperchromatism and irregular nuclear contours. Bottom, Adjacent section of Barrett's esophagus with low-grade dysplasia without intranuclear p53 immunopositivity. No foci with low-grade dysplasia stained positively in this study. esophagus. Compelling evidence exists for a metaplasiadysplasia-carcinoma sequence, whereby specialized, columnar epithelial cells progress to low-grade dysplasia, then to high-grade dysplasia and finally to invasive carcinoma. However, this progression from metaplastic epithelium to adenocarcinoma of the esophagus is not preordained. At present, it is not possible to predict in which patients Barrett's esophagus will progress to dysplasia or to adenocarcinoma. Little is known about the molecular genetic events responsible for the development of adenocarcinoma in Barrett's esophagus. Mutations of the p53 gene have been implicated in the development of carcinomas, including carcinoma of the esophagus'?, 8 The p53 gene is a tumorsuppressor gene located on the short arm of chromosome 17 (17p). It encodes a 53 kd, intranuclear, deoxyribonucleic acid-binding protein that regulates cell division, Table I Neoplastic change N Barrett's mucosa 28 (no dysplasia) Low-grade dysplasia 27 High-grade dysplasia 26 Intramucosal carcinoma 18 Submucosal carcinoma 5 p53-posilive specimens N % 0 0 0 0 18 69 12 67 2 40 preventing progression from the G 1 to the S phase. Intranuclear accumulations of this protein can be detected immunohistochemically with monoclonal antibodies for the protein product of the p53 gene. The protein product of the wild-type (nonmutated) p53 gene is biologically

1 1 3 4 Rice et al. The Journal of Thoracic and December 1994 Fig. 3. Top, Section of Barrett's esophagus with high-grade dysplasia, stained with hematoxylin and eosin. Adjacent to a focus of Barrett's esophagus of specialized type, the glands become architecturally complex with severe cytologic atypia, indicating high-grade dysplasia (arrow). Bottom, Adjacent section of Barrett's esophagus with high-grade dysplasia showing strong intranuclear immunoreactivity for mutated p53 (arrow). The adjacent Barrett's esophagus does not stain. active. It has a short half-life, so that intranuclear accumulation is usually not detected by immunohistochemical techniques. Mutations of the p53 gene often cause amino acid substitutions in the regulatory protein, which render it inactive. The half-life of the mutant p53 protein is frequently increased, with resultant intranuclear accumulation. The purpose of this study was to examine the immunohistochemical expression of mutated p53 in Barrett's esophagus, to determine whether it is a marker of neoplasia and, if so, to determine when it occurs in the metaplasia -dysplasia -carcinoma sequence. Materials and methods Resected specimens from patients with Barrett's esophagus who underwent esophagectomy at the Cleveland Clinic Foun- Fig. 4. Top, Section of invasive adenocarcinoma arising in Barrett's esophagus, stained with hematoxylin and eosin. These glands have infiltrated through the muscularis mucosa and into the superficial submucosa. Bottom, Adjacent section of invasive adenocarcinoma showing strong intranuclear immunoreactivity for mutated p53. dation between January 1, 1986, and July 1, 1993, were reviewed. Specimens with stage I1A (T2 NO MO, T3 NO MO) carcinoma or greater were excluded because, in such specimens, the transition from metaplasia to dysplasia and to carcinoma might be masked by large invasive carcinomas. Formalin-fixed, paraffin-embedded sections were reviewed in each specimen and analyzed for the presence and type of Barrett's mucosa. 9 Only those specimens with definite evidence of Barrett's mucosa were included in this study. The presence and grade of dysplasia were analyzed according to the criteria of Riddell and associates. lo Representative paired sections were chosen from areas of Barrett's mucosa, low-grade dysplasia, high-grade dysplasia, intramucosal carcinoma, and invasive submucosal carcinoma in each specimen. For example, in those specimens in which the most advanced neoplastic finding was intramucosal carcinoma, representative sections were chosen from areas of Barrett's mucosa, low-grade dysplasia, high-grade dysplasia, and intramucosal carcinoma within the resection specimen. Immunohistochemistry was performed with the labeled streptavidin biotin system by means of an automated immuno-

The Journal of Thoracic and Rice et al. 1 1 3 5 Volume 108, Number 6 Table II Frequency of p53 immunoreactivity First author Source of tissue N BM LGD HGD lea Rame[l2 Biopsy 43 1/21 2/13 5/11 8/15 Resection 17 (5%) (15%) (45%) (53%) Jankowski 11 Resection 30 1/15 7/15 (7%) (46%) Flejou 13 Resection 7 0/7 0/3 5/5 6/7 (0%) (0%) (100%) (86%) Jones 14 Biopsy 65 7/73 12/20 7/7 7/10 Resection 10 (10%) (60%) (100%) (70%) Rice Resection 28 0/28 0/27 18/26 14/23 (0%) (0%) (69%) (61%) BM, Barrett's mucosa; LGD, low-grade dysplasia; HGD, high-grade dysplasia; ICA, invasive carcinoma. stainer (Ventana Medical Systems, Tucson, Ariz.). Histologic sections that were 4J.(m thick were prepared and affixed to electrostatically charged slides. After deparaffinization and rehydration in xylene and graded alcohols, endogenous peroxidase was blocked with hydrogen peroxide. Mouse monoclonal antip53 (Pab1801, 1:100, Novocastra Laboratories, Burlingame, Calif.), an immunoglobulin M-class murine monoclonal antibody, was incubated with the section at a concentration of 0.02 mg/mt. Negative controls consisted of equivalently diluted nonimmune mouse immunoglobulin M. Sections were sequentially incubated with biotinylated antimouse immunoglobulin MI, streptavidin-horseradish peroxidase, the chromogenic substrate 3, 3-diaminobenzadine/hydrogen peroxide, and hematoxylin counterstaining, with intervening buffer washes. Cells were judged to be positive for the mutated p53 protein when nuclei exhibited dark-brown intranuclear staining. The immunostained slides were then compared with the histologic features observed in paired sections stained with hematoxylin and eosin to correlate p53 immunostaining with metaplasia, dysplasia, or carcinoma. Results Twenty-eight resection specimens were studied. The most advanced neoplastic changes in the resection specimens were intestinal metaplasia with no dysplasia (n = 1), low-grade dysplasia (n = 1), high-grade dysplasia (n -= 8), intramucosal carcinoma (n -= 13), and submucosal carcinoma (n = 5). From these 28 resection specimens 104 paired histologic sections were available for review (Table I). No p53 immunoreactivity was seen in any of the 28 sections of intestinal metaplasia without dysplasia or in any of the 27 sections of low-grade dysplasia (Figs. 1 and 2). p53 Immunoreactivity was seen only in sections of high-grade dysplasia, intramucosal carcinoma, and submucosal carcinoma (Figs. 3 and 4). Of the 26 sections of high-grade dysplasia, 18 (69%) stained for mutated p53. Similarly, 12 of 18 sections (67%) of intramucosal carcinoma and two of five sections (40%) of submucosal carcinomas stained for mutated p53. In those neoplastic resection specimens in which p53 immunostaining was observed (18 of 26 specimens), the entire spectrum of neoplastic changes (high-irade dysplasia, intramucosal carcinoma, and submucosal carcinoma) within the resection specimen stained positively for mutated p53. Discussion Mutations of the p53 gene have been reported in adenocarcinomas associated with Barrett's esophagus, suggesting a possible causal role_ II - 14 In this study, p53 immunoreactivity was not found in any sections of Barrett's mucosa or low-grade dysplasia, regardless of the presence or absence of more advanced neoplasia in other areas of the resection specimen. Immunoreactivity for mutated p53 was seen only in sections of high-grade dysplasia and invasive (intramucosal and submucosal) carcinomas. However, not every resection specimen with advanced neoplasia demonstrated p53 immunoreactivity. Immunostaining identified p53 mutations in 75% of patients with high-grade dysplasia, in 77% of those with intramucosal carcinoma, and in 40% of those with submucosal carcinoma. Within a resection specimen that stained positively for mutated p53, the spectrum of neoplastic changes (high-grade dysplasia, intramucosal carcinoma, submucosal carcinoma) was positive for p53. This study confirms that mutations in the p53 gene are important events in the development and progression of neoplasia in Barrett's esophagus. Our results were obtained by immunohistochemical staining alone. Whether p53 immunoreactivity correlates with mutations of the p53 gene in all cases is unknown. Immunohistochemical techniques may not detect all p53 mutations. For example, we recently identified five lung and ovarian tumor samples that have a premature stop codon mutaton or deletion in exons 9 or 10 that stain

1 1 3 6 Rice et al. The Journal of Thoracic and December 1994 negatively for Pab1801. Furthermore, p53 immunoreactivity does not always correlate with mutations of the p53 gene. In pancreatic carcinomas, p53 immunoreactivity can occur in tumors that have wild-type p53. In some cases, normal stromal contamination may account for this lack of sensitivity in the detection of p53 mutations. 15 However, evidence exists that nuclear immunostaining in some tumor samples is related to inactivation of the p53 protein by mechanisms other than mutations in the p53 gene. 16,17 That intranuclear protein accumulates only in highgrade dysplasia and invasive carcinoma suggests that p53 mutations occur relatively late in the metaplasia-dysplasia-carcinoma sequence in Barrett's esophagus. The existence of p53 mutations in low-grade dysplasia is controversial (Table II). Flejou and colleagues13 identified a similar p53 immunostaining pattern. They found no p53 mutations in three cases of low-grade dysplasia in Barrett's esophagus, whereas all five cases of high-grade dysplasia and six of seven cases of invasive carcinoma exhibited mutant p5 3. Jankowski and colleagues II reported one specimen in 15 specimens of Barrett's mucosa without dysplasia that immunostained positively for p53. As mentioned, some p53 immunoreactivity may not correspond to p53 mutations. Ramel and colleagues l2 found mutant p53 immunoreactivity in one of 21 (5%) foci of Barrett's mucosa and in two of 13 (15%) foci of indefinite or low-grade dysplasia. Flejou and associates 13 pointed out that the specimens used for histologic grading in Ramel's study were not those used for p53 immunostaining. More advanced neoplastic changes may have been present in the specimens immunostained for p53 than were present in the histologic sections interpreted as low-grade dysplasia. This difference may account for the reported immunoreactivity in the occasional specimens of Barrett's esophagus with no, indefinite, or low-grade dysplasia. Using a similar immunohistochemical technique, Jones and colleagues l4 studied 65 biopsy specimens and 10 resection specimens of Barrett's esophagus. They reported that seven of 73 specimens (10%) of Barrett's esophagus without dysplasia and 12 of 20 specimens (60%) of low-grade dysplasia in Barrett's esophagus stained positively for p53. The rate of p53 expression in low-grade dysplasia in this study differs significantly from that in all other reports. The discrepancy may be the result of immunostaining of a population of low-grade dysplasia cells that are undergoing transition to high-grade dysplasia when studied. These transition specimens may stain positively for mutant p53, despite the histologic appearance of low-grade dysplasia. Another possible and more probable explanation for the discrepancy in this study is the inherent observer variability in the histologic diagnosis of low-grade dysplasia, despite well-outlined histologic criteria. ls Gene sequencing studies have provided intriguing data concerning the relationship between Barrett's esophagus and adjacent adenocarcinoma. Casson and colleagues l9 reported that four of seven samples of Barrett's esophagus had p53 sequence mutations, whereas the adjacent adenocarcinoma was wild-type. Furthermore, in a single case of adenocarcinoma in which p53 mutations were detected by sequencing, the adjacent Barrett's mucosa was wild-type. No immunohistochemical staining was performed in these cases, and it is difficult to reconcile these results with those obtained with immunohistochemistry. A comprehensive study combining p53 immunohistochemical staining and sequence analyses of Barrett's esophagus will be required to resolve these issues. We found no immunoreactivity in specimens of Barrett's esophagus with no dysplasia or with low-grade dysplasia. Immunohistochemical evidence of mutations in the p53 gene was limited to specimens with high-grade dysplasia and invasive adenocarcinoma in Barrett's esophagus. However, p53 immunoreactivity is not an allinclusive marker for p53 mutation in these specimens. Mutation of p53 represents a significant transition in the development of Barrett's adenocarcinoma, and the mutation occurs late in the metaplasia-dysplasia-carcinoma sequence during the transition to high-grade dysplasia. REFERENCES 1. Cameron AJ, Ott BJ, Payne WS. The incidence of adenocarcinoma in columnar-lined (Barrett's) esophagus. N Engl J Med 1985;313:857-9. 2. Achkar E, Carey W The cost of surveillance for adenocarcinoma complicating Barrett's esophagus. Am J GastroenteroI1988;83:291-4. 3. Robertson CS, Mayberry JF, Nicholson DA, James PD, Atkinson M. Value of endoscopic surveillance in the detection of neoplastic change in Barrett's oesophagus. Br J Surg 1988;75:760-3. 4. Hameeteman W, Tytgat GN, HouthoffHJ, van den Tweel JG. Barrett's esophagus: development of dysplasia and adenocarcinoma. Gastroenterology 1989;96: 1249-56. 5. Spechler SJ, Robbins AH, Rubins HB, et al Adenocarcinoma and Barrett's esophagus. An overrated risk? Gastroenterology 1984;87:927-33. 6. Van der Veen AH, Dees J, Blankensteijn JD, Van Blankenstein M. Adenocarcinoma in Barrett's oesophagus: an overrated risk. Gut 1989;30:14-8. 7. NigroJM, BakerSJ,Preisinger AC,etal. Mutations in the p53 gene occur in diverse human tumor types. Nature 1989; 342:705-8. 8. Hollstein M, Sidransky D, Vogelstein B, Harris CC. p53 mutations in human cancers. Science 1991;253:49-53.

The Journal of Thoracic and Volume 108, Number 6 Rice et al. 1 1 3 7 9. Paull A, Trier JS, Dalton MD, etal The histologic spectrum of Barrett's esophagus. N Engl J Med 1976;295:476-88. 10. Riddell RH, Goldman H, Ransohoff DF, et al. Dysplasia in inflammatory bowel disease: standardized classification with provisional clinical applications. Hum Patho11983; 14: 931-68. 11. Jankowski J, Coghill G, Hopwood D, Wormsley KG. Oncogenes and onco-suppressor gene in adenocarcinoma of the oesophagus. Gut 1992;33:1033-8. 12. Ramel S, Reid BJ, Sanchez CA, et al. Evaluation of p53 protein expression in Barrett's esophagus by two-parameter flow cytometry. Gastroenterology 1992;102:1220-8. 13. FiejouJF, Potet F, Muzeau F, et al. Overexpression ofp53 protein in Barrett's syndrome with malignant transformation. J Clin Patho11993; 46:330-3. 14. Jones DR, Davidson AG, Summers CL, Murray GF, Quinlan DC. Potential application of p53 as an intermediate biomarker in Barrett's esophagus. Ann Thorac Surg 1994;57:598-603. 15. CaseyG, Yamakana Y, FreissH,etal.p53 Gene mutations are common in pancreatic cancer and are absent in chronic pancreatitis. Cancer Lett 1993;69:151-60. 16. Lu X, Park SH, Thompson TC, Lane DP. Ras-induced hyperplasia occurs with mutation of p53, but activated ras and myc together can induce carcinoma without p53 mutation. Cell 1992;70:153-61. 17. Casey G, Lo-Hsueh M, Lopez ME, Vogelstein B, Stanbridge EJ. Growth suppression of human breast cancer cells by introduction of a wild-type p53 gene. Oncogene 1991; 6:1791-7. 18. Reid BJ, Haggitt RC, Rubin CE, et al Observer variation in the diagnosis of dysplasia in Barrett's esophagus. Hum PathoI1988;19:166-78. 19. Casson AG, Mukhopadhyay T, Cleary KR, Ro JY, Levin B, Roth JA. Oncogene activation in esophageal cancer. J THORAC CARDIOVASC SURG 1991;102:707-9. Discussion Dr. Carolyn E. Reed (Charleston, S.c.). Dr. Rice, you have reported that no p53 immunoreactivity was seen in Barrett's mucosa alone or in low-grade dysplasia. This is in contrast to the recent paper reported in The Annals a/thoracic Surgery by Dr. Jones and his group from West Virginia, who showed p53 reactivity in 10% of specimens of Barrett's mucosa and 60% of specimens of Barrett's with low-grade dysplasia. This group used the anti-p53 monoclonal antibody DO-I. You have used a different monoclonal antibody. A number of commercial monoclonal antibodies are available to us. Do you believe that this contributes to the differences seen by your group? Our group in South Carolina has been actively looking at p53 expression in squamous cell carcinoma of the esophagus. Of note is the fact that 33% of our patients with esophageal cancer show labeling of nuclei for p53 in the germinal layer of their normal mucosa also. Of even greater interest is a 39% overexpression in the germinal layer from normal esophageal samples taken from random autopsies of black men. To us, this may indicate that p53 immunoreactivity is an early marker, rather than a late marker, of a genetic predisposition to the development of cancer. The trick, of course, will be to discover what conditions halt or allow the sequence of malignant transformation to continue. Dr. Rice. p53 Mutations may be indirectly studied by p53 immunoreactivity. The abnormal protein product of mutated p53 will stain and indirectly verify the presence of p53 mutations. However, some p53 mutations will not be detected by p53 immunoreactivity. Furthermore, the mutations in the p53 gene are not the only cause of p53 immunoreactivity. Because it is an indirect marker of what is truly happening in the gene, p53 immunoreactivity may not always reflect gene mutations. This may explain disparate results. A problem exists as well with the interpretation of p53 immunoreactivity in patients with low-grade dysplasia and Barrett's esophagus. Most studies have found no or a low incidence of staining in this group. When staining exists it may be the result of sampling errors, especially in biopsy specimens. Furthermore, some specimens have been selected for further study by flow cytometry, and the actual area studied histologically may not have been the same as that sampled by flow cytometry. Besides sampling errors, there is the possible occurrence oflow-grade dysplasia in transition. In such a case the p53 mutation has already occurred genetically, but histologically there is still the appearance of low-grade dysplasia. Perhaps the main cause of discrepancies is the large interobserver variation in the diagnosis of low-grade dysplasia. Dr. Jeffrey H. Peters (Los Angeles, Calif). It seems clear from this presentation as well as from the data in the literature that p53 mutations are present in a high percentage of patients with esophageal carcinoma, and, in particular, esophageal adenocarcinoma. The question is how this information should be used. Because of sampling bias, it is not likely that markers like p53 can be used to select patients and make decisions with regard to high-grade dysplasia or low-grade dysplasia, because only small portions of the esophageal mucosa are samples. It may be more interesting to use this sort of information to select patients for either curative or palliative surgical therapy. Do you have any information on the long-term outcome of the patients whose samples were p53 positive versus p53 negative, and do you think p53 status is a prognostic factor? Dr. Rice. We have not yet evaluated the prognostic significance of p53 mutations in these patients. However, this mutation is easily studied by immunostaining and we hope to investigate its value as a prognosticator. At present, p53 immunostaining may be used in situations in which differentiating low-grade dysplasia from high-grade dysplasia is problematic. I would not suggest that it be used in every case, but it may be considered when the diagnosis of high-grade dysplasia is in questions.