Tobacco Documents Online

htt. J. Cancer (Pred. Onc'ol.): 74, 212-219 (1997) © 1997 Wiley-Liss, Inc. 8088 HX~3~ 212 LAH ~NT J OAHOER 97 (C)H~LEY-LZSS NY PREVALENCE AND PREDICTIVE VALUE OF p53 MUTATION IN PATIENTS WITH OESOPHAGEAL SQUAMOUS CELL CARCINOMAS" A PROSPECTIVE CLINICO-PATHOLOGICAL STUDY AND SURVIVAL ANALYSIS OF 70 PATIENTS King Y. LAM1-*, Sai W. TSAO2, Dekai ZHANG2, Simon LAW3, Dan.HE2, Lily MAI and John WONG3 tDepartment of Patholog3; Universi~." of Hong Kong, Hong Kong 2Department of Anatotny, Universit3" of Hong Kong, Hong Kong 3Department of Surgery, Universi~. of Hong Kong, Hong Kong The tissues from 70 Chinese patients with oesophageal squamous cell carcinoma were prospectively collected to study for the pattern of p53 mutations and its relationship with clinico-pathological features and prognosis using immu- nohistochemistry, polymerase chain reaction-single strand conformational polymorphism (PCR-SSCP) analysis and DNA sequencing, p53 over-expression and p53 mutations were detected in 73% and 44% of the patients. These p53 aberra- tions had no relationship with the patient age, sex, smoking/ drinking habits and tumor site, size or stage. The p53 over-expression was more intense in moderately~poorly- differentiated squamous cell carcinomas. Thirty.three p53 mutations were noted in 31 patients; 18.2% in exon 5, 15.2% in exon 6, 33.3% in exon 7 and 33.3% in exon 8. Mutations were primarily point mutations and common in codons 248, 273 and 285. There were 46% transversions, 36% transitions and 18% frameshift. The survival of the patients depended mainly on the extent of resection. In patients with stage I!1 oesopha- geal cancer, the median survival of those with p53 mutations was 6.8 months whereas those without was 12.5 months. The results were of clinical importance although the value did not reach statistical significance. Thus, there was a definite role of p53 mutations in the pathogenesis of oesophageal squamous cell carcinomas, p53 mutations were not synonymous with p53 over-expression. The distribution of p53 mutations in oesophageal cancers suggested that the etiologic contribu- tion might be complex and probably involve different exog- enous and endogenous exposures, p53 mutations also appear to play a role in predicting the survival of patients with stage III oesophageal squamous cell carcinomas. Int. J. Cancer 74:212-219, 1997. © 1997 Wiley-Liss, Inc. Oesophageal squamous cell carcinoma is one of the most common human cancers world-wide (Parkin et al., 1988). The incidence and mortality rate is particularly high among Chinese populations, i.e., in Hong Kong (Lee, 1991). Despite various therapeutic interventions, including surgery, radiotherapy, and chemotherapy, the 5-year survival rate for this disease remains low (Law et al., 1992). Thus, to improve therapeutic and preventive interventions, a biological understanding of the nature of the disease is essential. The p53 gene coding for a 53-kDa nuclear phosphoprotein is believed to be a tumour suppressor gene. It is located on the short-arm of chromosome 17 and is a critical regulator of cell growth, differentiation and apoptosis through its actions in cell- cycle checkpoint control (Greenblatt et aL, 1994). It is also well known that mutation of the p53 gene is the most common genetic alteration in human cancers. A number of studies have shown that p53 protein over-expression may provide insight into the carcino- genesis in oesophageal squamous cell carcinomas (Lamet al., 1995b). However, there is lack of well planned molecular studies to correlate p53 mutations with the clinico-pathological parameters (including the survival data) in patients with oesophageal squa- mous cell carcinomas. In this study, we have analysed the correlation of p53 gene mutations with the clinico-pathological features. The profile of p53 mutations was also studied by DNA sequencing in order to gain THIS ARTTOLE IS FOR TNOTV]~DUAL USE ONLY AND HAY NOT EE FURTHER REPRODUCED OR STORED ELEOTRON,I:OALLY NITHOUT IARITTEN PERMISSION FROM THE COPYRIGHT HOLDER. UNAUTHORIZED REPRODUCTION HAY RESUL TN FINANCIAL ~JD~ OTHE~ ,~Vd~.TZ,E~. '~.,~' ~ International Union Against Cancer , Union Intemationale Contre le Cancer insight on the carcinogenesis of oesophageal squamous cell carcinomas in Hong Kong Chinese. MATERIAL AND METHODS Patient characteristics The patients who were chosen for our study had resections for oesophageal cancers at some point during the period between July 1990 and June 1994 at Queen Mary Hospital. The selection criteria for this study included primary oesophageal squamous cell carcino- mas from Chinese patients: patients with pre-operative chemother- apy given were excluded from the analysis. The clinical data collected included age, sex, smoking habits, alcohol use and types of resection (palliative vs. curative). The patients were classified as smokers when they smoked I or more pack of cigarette/day for over 2 years: as drinkers if they had alcohol of or more than 2 pints of beer/day for over 2 years. Resection was considered curative when the tumour was confined to the oesophagus and the adjacent lymph nodes were either uninvolved or less than 2 cm in diameter. discrete, and within the vicinity of the primary tumour. Both the tumour and the adjacent lymph nodes were removed completely. On the other hand, palliative resection was performed if the tumour infiltration extended beyond the oesophagus into mediastinal organs, with gross lymph nodes metastases, and when there was an actual or a strong likelihood of tumour being left over after resection. The actuarial survival rate of the patients was measured from the date of resection of the oesophageal cancers to the date of death or April 30, 1996. Collection of tissues The tissue samples were collected prospectively. The resection specimens of the selected patients were dissected when fresh. Macroscopically, the site and size (maximum length) of the oesophageal tumours were recorded. One representative frozen block from the tumour and one from the non-neoplastic tissue at a distance from the tumour (acting as a control tissue) were taken from each of the resected specimens. In the same site next to each frozen block, a block was also taken, fixed in 10% formalin and embedded in paraffin wax. The frozen blocks were snap frozen in isopentane cooled in liquid nitrogen at -70°C until used. DNA from these samples was extracted by the phenol/chloroform method as described previously (Tsao et al., 1991). The paraffin blocks taken next to the frozen blocks were used for immunohisto- chemistry. Other additional blocks from the tumour, lymph nodes and resection margins were also obtained from each specimen. Contract grant sponsor: University of Hong Kong, contract grant number 335/046/0065. *Correspondence to: Dr K Y Lam. Department of Pathology, Queen Mary Hospital. Hong Kong. Fax: 011-852-28179935. E-mail: akylam@hkucc.hku.hk Received 23 October 1996: revised 13 December 1996 II I, 1, ! I

i p53 MUTATION IN OESOPHAGEAL CARCINOMAS 213 They were processed through paraffin alter routine fixation in 10% paper with a vacuum gel dryer and exposed to X-ray film at -70°C formalin for light microscopic analysis, for 6 to 12 hr with an intensifying screen. Pathological review and selection of tissues The histological sections of primary oesophageal squamous cell carcinomas were studied. Only squamous cell carcinomas were includ.ed; tumours containing any mucin-se.creting component were excluded from analysis. The grades of turnout differentiation (well, moderately or poorly-differentiated) were classified using the WHO criteria (Watanabe et aL, 1990). The carcinomas were staged according to the TNM classification (Beahrs et aL. 1992). The sections from the frozen blocks and the corresponding paraffin blocks taken were also examined histologically to determine if they were suitable for further investigations. The selection criteria were based on: (i) whether there were adequate (amour in the selected blocks (Le.. at least 10% of the surface areas should have been occupied by turnout cells) and (ii) whether the histologic features of the turnout in every chosen block were representative of the whole tumour (since a few carcinomas exhibited slight variations in tumour differentiation at different foci of the tumours). hnmunostaining From the selected paraffin blocks, 5 p.-thick sections were cut. They were de-paraffinized with xylene and rehydrated through graded concentration of alcohol. The sections were then treated with microwave irradiation at 95°C for 14 rain in 10 mM citrate buffer at PH 6. They were washed in water, rinsed with Tris- buffered saline (TBS) and treated with 3% hydrogen peroxide in methanol for 10 min at room temperature (RT) to block intrinsic peroxidase activity. The sections were again washed with water and rinsed with TBS. Ten per cent normal rabbit serum was then added at RT for 10 min. Afterwards, primary mouse monoclonal anti-p53 antibody (NCL-p53-DO7, Novocastra, Newcastle upon Tyne. UK) was added at a dilution of I: 100 and incubated in a moist chamber overnight at 4°C. The antibody recognises both wild and mutant p53 proteins. The slides were again washed 3 times in TBS ~br 3 min. Rabbit anti-mouse IgG-biotinylated (diluted in 10% normal rabbit serum; E354, Dako, Glostrup, Denmark) and pre-incubated (30 min at RT) with avidin-biotin-complex (1:100; Amersham. Aylesbury, UK) was added for 30 min at 37°C. They were washed in TBS as before and were then developed in freshly prepared DAB/HzO~ solution for 10 min at RT. The sections were then washed in wa~er, counter-stained with Mayer's haematoxylin for I min at RT, dehydrated, cleared and mounted. Negative controls were sections treated the same as above but with omission of the primary antibody. A block of squamous cell carcinoma known to be strongly positive for p53 protein (obtained in our previous study) was used as positive control (Lam et al., 1995b). Examination of p53 staining was done at high power fields (×400) using a standard light microscope. Brown nuclear stainabil- ity was regarded as positive. The level of p53 accumulation was graded semi-quantitatively into 4 categories (0. +, + +, + + +) as described in our previous studies CLam etal., 1995a, b). PCR atnplification and SSCP analysis The details of the PCR-SSCP analysis was carried out as described previously (Mok et al., 1993). Briefly, the extracted DNA of the turnout and control tissue were amplified using PCR. The left and right oligonucleotide primers of exons 5 to 8 of the p53 gene (Genosys Biotechnologists, The Woodlands, TX) were radiola- beled with T4 polynucleotide kinase tBoehringer Mannheim, Indianapolis, IN). The PCR products were then mixed with 45 ~1 butler containing 95% formamide. 20 mM EDTA. 0.05% bromo- phenol blue. and 0.05% xylene cyanol FF (Sigma, St. Louis. MO) and loaded into an 8% polyacrylamide gel (49: I ratio of arcylamide to bisacrylamide). Afterwards, they were electrophoresed under each of the 2 conditions (with or without 3% glycerol added to the gelh respectively at 30 watts at 4°C. The gel was dried on a filter Direct sequenchtg of the PCR products Tumour samples containing mutations of the p53 (as revealed by a shift in mobility of the mutated exon in the SSCP gel from the (amour as compared with control tissue) and the corresponding normal tissues were re-amplified by PCR. The PCR products were directly sequenced by using a cycle sequencing kit (USB, Cleve- land. OH) by the dideoxy chain termination method. The sequences of the various p53 gene exons were confirmed by sequencing both sense and antisense complementary DNA strands in an 8% polyacrylamide gel containing 7 M urea. The mutated sequence was then aligned with the reported p53 sequence using the BLAST (Basic Local Alignment Search Tool) from NCBI (National Centre for Biotechnotogy Information) available in Internet. Statistical analysis The data were computerised and statistical tests were performed with the program. Statistical Package for Social Sciences (SPSS-X version 3.1, Chicago, IL). The significance of various parameters on survival were analysed by the univariate and multivariate Cox regression. Survival curve was plotted using the Kaplan-Meier method. Other tests used were X-' test, Fisher's exact test, Student's t-test. The tests were considered significant when their overall p values were below 0.05. RESULTS Clhffcopathological features Seventy patients (59 males. 11 females) with resections (25 curative. 45 palliative) of oesophageal tumours were studied. The mean age of these patients was 63 years (ranging from 41 to 89). Eighty one percent (n = 57) and 67% (n = 47) of the patients were smokers and drinkers, respectively. The oesophageal cancers comprised 29% (n = 20) well differentiated. 51% (n = 36) moder- ately differentiated, and 20% (n = 14) poorly differentiated squa- mous cell carcinomas. Seven per cent (n = 5) of the turnouts were in the upper oesophagus, 47% (n = 33) in the middle oesophagus and 46% (n = 32) in the lower oesophagus. The lengths of the carcinomas ranged from 1 cm to 9.5 cm (mean: 5 cm). The majority (79%~ 55 cases) of these patients had Stage III lesions (T3, N l, M0 ofT4. any N. M0). The rest of the patients comprised 19% (n = 13) Stage II (T2/'f3. N0, M) or TI/T2. NI, M0) and I% (n = 1) stage I (TI, NO. M0) and 1% (n = 1) Stage IV (any T, any N, Ml)-Iesions. hnmtmostab~h~g fi~r p53 p~vtein Overall. 73% (51 of 70) of the patients had nuclear p53 expression detected. The clinico-pathological features of p53 positive and negative cases are listed in Table I. It appears that age, sex. tobacco use. drinking habits of the patients and the size, site, stage of the tumours had no significant relationship with either the positivity or the staining intensity for p53, as detected by immuno- histochemistry. The p53 staining was seen in the poorly-differentiated tumour cells in the periphery of the tumour cells nests in all the cases while the expression was weaker in the better differentiated foci. The central keratinized areas and the immediately adjacent tumour cells were always negative lbr p53 (Fig. 1). The well-differentiated squamous cell carcinomas were often weakly stained for p53 but the moderately/poorly differentiated tumours were strongly stained (p = 0.04. X-" test). On the other hand, there was no statistically significant difference between the p53 positive and negative cases with regard to tumonr differentiation (p = 0.9. X-' test). p53 mttutthm detected by PCR-SSCP Using PCR-SSCP techniques to screen for mutations in exons 5-8 of the p53 gene. we detected mutations in 44% (31 of 70) of the

214 LAM ETAL. TABLE i - CLINICO-PATHOLOGICAL DATA ON ()ESOPHAGEAL SQUAMOUS CARCINOMAS: p53 EXPRESSION Clinico-path~logical features p53 overexpressionI Significant tests P(~sitive vs. Positive ( + + + + + +) Negative negative +1++ vs, +++ Number of patients 51 ( 19 17 15) 19 Mean age 63 (65 64 60) 62 Male/female 44/7 ( 18/1 15/2 11/4) 15/4 Chronic smokers 80% (84% 88% 67%) 84% Chronic drinkers 73% (74% 76% 67%) 52% Mean lon.gitudinal length of the 5.1 (5.6 5.1 4.3) 5.0 carcinomas (cm): Site: Upper 4 (8%) (3 0 1) 1 (5%) Middle 22 (43%) (7 7 8) 11 (58%) Lower 25 (49%) (9 I0 6) 7 (37%) Differentiation: Well 14 (27%) (8 5 I) 6 (32%) Moderate/poor 37 (73%) ( 11 12 14) 13 (68%) Stage: I/II 10 (5 3 2) 4 III/IV4t (14 14 13) 15 p = 0.9 (t-test) p = 0.3 (t-test) p = 0.5 (Fisher's exact test) p = 0.2 (Fisher's exact test) . p = 1.0 (Fisher's exact test) p = 0.1 (Fisher's exact test) p = 0.1 (X2 test) p = 1.0 (Fisher's exact test) p = 0.9 (t-test) p = 0.07 (t-test) p = 0.5 (X2 test) p = 0.6 (×2 test) p = 0.9 (Xz test) p = 0.04 (X" test) p = 1.0 (Fisher's exact test) p = 0.7 (Fisher's exac't test) I+: 10% to 30% of the turnout cells positive; + +: 30% to 50% of the tumour cells positive; + + +: more than 50% of the tumour cells positive. FIGURE 1 - p53 staining in well-differentiated squamous cell carcinoma showing nuclear staining of tumour cells in the periphery of the tttmour (Scale bar: 0.02 c,n). patients analysed. The clinico-pathological features of the patients with and without p53 gene mutations are shown in Table 11. There was no association between p53 mutations and age. sex. smoking and drinking habits of the patients. No relationship was detected between p53 mutations and the size, site. stage and grade of squamous cell carcinomas. The sequencing of the mutated exons of p53 gene The exact position and nature of the p53 mutations and their relationship with the clinico-pathological data are shown in Table IIL Thirty-three mutations were detected in 31 out of the 70 patients analysed. One patient (case 2) had 2 mutations in exon 5 and one (case I 1 ) had mutations in both exon 6 and exon 7. Overall. I o ! [ [

p53 MUTATION IN OESOPHAGEAL CARCINOMAS 215 TABLE I1 - CLINICO-PATHOLOGICAL DATA ON THE OESOPHAGEAL SQUAMOUS CARCINOMAS: DETECTION BY PCR-SSCP Clinico-patht~logical p53 Significant tests features Positive Negative Number of patients 31 39 -- Mean age 63 63 p = 0.8 (t-test) Male/female 26/5 33/6 p = 1.0 (Fisher's exact test) Chronic smokers 74% 87% p = 0.2 (X-' test) Chronic drinkers 68% 67% p = 0.9 (×-~ test) Mean longitudinal 4.9 cm 5.1 cm p = 0.9 (t-test) length of the carcinomas: Site: Upper 3(I0%) 2(5%) . p=0.1(X-~test) Middle 18 (58%) 15 (38%) Lower 10 (32%) 22 (57%) Differentiation: Well 8 (26%) 12 (31%) p = 0.9 (X" test) Moderate 19 (61%) 17 (43%) Poor 4 (13%) 10 (26%) Stage: I/II 7 7 p = 0.7 (X~- test) Ill/IV 24 32 completion of our study. The median survival of the whole group was 12 months. The survival was not related to patients' age, sex, smoking and drinking habits of the patients and tumour differentia- tion. p53 immunostaining and.p53 mutation ( p = 0.9, 0.6.0. I, 0.9, 0.1, 0.9 and 0.2, respectively by the Wilcoxon test). Conversely, survival of the patients was related to the longitudinal length of the cancers (p = 0.01), stage (p = 0.04) and type of resections performed (p = 0.001) using univariate analysis. Using multivari- ate analysis, only the type of resection performed significantly affected patient survival (p=0.001); the medial survival of patients with curative resection was 33 months whereas those with palliative resection was 8 months. In order to evaluate the relationship between role of p53 gene and survival rates within a given stage, the tumours in stage III (n = 55) were analysed. Patients whose tumours had positive p53 mutation had a median survival of 12.5 months whereas those negative had a survival of 6.8 months (Fig. 3). The former group had a shorter median survival, although the results did not reach statistical significance (p = 0.05, Witcoxon test). On the other hand. there was no significant difference in survival between those with or without p53 over-expression (p = 0.4, Wilcoxon test). For patients with lesions in other stages, the number were too few to be stratified into groups. 18.2% (n = 6) of mutations occurred in exon 5, 15.2% (n = 5) in exon 6, 33.3% (n = 11) in exon 7 and 33.3% (n = 11) in exon 8. With regard to the types of mutations, 5 of 33 mutations (15%) were deletions, 1 (3%) was insertion and 27 (82%) were point mutations (12 transitions, 15 transversions). All the transition mutations resulted in amino acid changes: 7 of which (58%) had mutations at the CpG' site. The 15 transversions were seen in specimens from 13 patients. Amino acid changes were seen in 11 cases (13 transversions). Amongst these, 2 cases (cases 2 and 11) had 2 transversions, respectively. Other than these, I case (case 26) with transversion ended up in the formation of a stop codon (nonsense mutation) and a silent mutation was detected in 1 specimen (case 2). Transversions were not seen in well-differentiated squamous cell carcinomas while transitions were not found in poorly-differenti- ated squamous cell carcinomas (well vs. moderately/poorly differ- entiated, p = 0.002, Fisher's exact test). The types of mutations had no relationship with the patients' sex (p = 0.2, X-' test), intensity of p53 over-expression in the tumours (p = 0.2, X~- test), sit.e (p = 0.6, ×~- test), stage (p = 0.2, Xz test) and location (exon) in the gene (p = 0.3, X2 test). In exon 7, 4 specimens (36%) had mutations in codon 248. Another mutation hot spot was codon 273 in exon 8:4 cases had mutations at this site. In addition, 3 cases had mutations in codon 285. Other than these, 2 mutations were identified in codon 195 (exon 6) and codon 241 (exon 7), respectively. p53 analysis by hnmunohistochemistry vs. PCR-SSCP-DNA sequenchzg Of the 31 cases with p53 mutations detected by PCR-SSCP, 5 (16%) were negative for p53 over-expression. Of these, 3 were deletions resulting in frameshift, 1 was transition. The other case (case 2) had double mutations in exon 5; one mutation ended up in no change in amino acid produced and one was transversion at codon 176. There was no statistically significant difference be- tween the positivity detected by PCR-SSCP and immunostaining (p = 0.06, X-' test). The intensity of p53 staining had also no relationship with the location (exons) in the p53 gene (p = 0.8, X" test) and types of mutation (p = 0.11, X-" test). Survival analysis The mean follow-up period of the whole group was 16 months: 79% of the patients (n = 55) had already died by the time of DISCUSSION The majority of p53 mutations in human cancers are localised within 4 evolutionary conserved domains of the gene: exons 5 through 8 (Greenblatt et al., 1994: Montesano et al., 1996). Many of these mutations are point mutations. Six mutation hotspots have been found in human tumours in this area of the gene at codons 175, 245, 248, 249, 273 and 282, respectively (Greenblatt et aL. 1994). As a whole, point mutations are seen at high frequency in all the 6 mutation hotspots in oesophageal squamous cell carcinomas (Andr6zet et aL, 1993; Bennett et al., 1991; Casson et aL. 1991: Est~ve etal., 1993; Gap etaL, 1994; Gates etal., 1994; Hollstein et aL, 1991; Imazeki et al., 1992; Liang et al., 1995; Shibagaki et aL. 1995). The frequency of mutations appears to be vary among geo- graphic regions (Montesano et aL, 1996). In Western and Japanese studies, the mutations are relatively evenly distributed in exons 5, 6 and 8 (Andr6zet et al., 1993; Casson et aL, 1991; Est~ve et aL, 1993: Gates et al., 1994; Hollstein et al., 1991 ; lmazeki et al., 1992; Shibagaki et al., 1995). The mutations are highly frequent in codons 175 and 278. In Chinese, most of the point mutations are seen in exon 5, many of which in codon 175 (Bennett et aL. 1991; Gap et al., 1994; Liang et aL, 1995). In our study, 82% of p53 mutations in oesophageal squamous cell carcinomas were point mutations. Unlike the results obtained from other Chinese populations, point mutations were commonly seen in exons 7 and 8 (Bennett etaL, 1991; Gap etal., 1994; Liang et aL, 1995). Mutations were often noted in codons 248, 273 (exon 7) and 285 (exon 8) in oesophageal squamous cell carcinomas from Hong Kong Chinese. Besides, point mutations at codons 180, 232, 246, 250, 286 which had never been reported previously in oesophageal squamous cell carcinomas were found in our investi- gation. A high prevalence of mutated codon 176 of the p53 gene bas been reported in Chinese patients from Mainland China with oesophageal cancers (Lung etal., 1996). In this study, the mutation was also found in the samples taken from Hong Kong Chinese (44%; 6 of 18 patients with p53 mutation). On the other hand, our results (analysing a higher number of Hong Kong patients) do not support such high frequency of mutated codon 176. We postulate that one of the reasons for this discrepancy is related to differences in the sampling of patients; the Hong Kong patients in the study of Lung etal. (1996) were from a hospital in the northern part of Hong Kong, where many patients may well be recent immigrants from

216 LAM ETAl~ TAIILE III - PATTERN OFp53 MUTATIONS IN THE OESOPHAGEAL SQUAMOUS CARCINOMAS Exons GradeI S-' D-" lmmuno- Case Nucteotide Amino acid stain~ number Codon change change Type of mutation 5 P Y Y + + 41 157 GTC to TTC Val to Phe 5 W N Y + + + 60 162 1 bp (C) deletion -- 5 P Y Y - 2 173 GTG to GTT No change (Val to Val) 5 W Y Y + I 175 CGC to CAC Arg to His 5 P Y Y - 2 176 TGC to TTC Cys to Phe 5 P Y Y + + + 26 180 GAG to TAG Glu to stop codon 6 M N N + 31 193 CAT to CGT His to Arg 6 M Y Y + + 49 195 ATC to AAC lie to Ash 6 M Y N + I 1 195 ATC to AGC Ile to Ser 6 P Y N + 15 208 1 bp (T) insertion -- 6 P Y Y - 63 209 2 bp (CT) deletion -- 7 M Y N + 11 232 ATC to AGC lie to Ser 7 M Y N - 4 241 1 bp (C) deletion -- 7 M Y Y + + 9 241 TGC to TTC Cys to Phe 7 W Y Y + 44 246 ATG to GTG Met to Val 7 M N N + + + 20 248 CGG to TGG Arg to Trp 7 W Y ¥ + + 36 248 CGG to CAG Arg to Gln 7 M Y Y + + + 39 248 1 bp (G) deletion -- 7 M Y N + + + 58 248 CGG to CAG Arg to Gin 7 M Y Y + + + 24 249 AGG to AGT Arg to Ser 7 W Y Y + 61" 250 CCC to CTC Pro to Leu 7 M N N - 42 251 1 bp (T) deletion -- 8 M Y Y + + 56 266 GGA to GTA Gly to Val 8 M N N + + + 22 271 GAG to GTG Glu to Val 8 M Y Y + 3 273 CGT to TGT Arg to Cys 8 M Y N + + 6 273 CGT to CTT Arg to Leu 8 M N Y + + 19 273 CGT to CAT Arg to His 8 M Y Y + + + 53 273 CGT to CTT Arg to Leu 8 W Y N + + 8 282 CGG to TGG Arg to Trp 8 M N Y + 17 285 GAG to GTG Glu to Val 8 W Y Y + 33 285 GAG to AAG Glu to Lys 8 W Y Y - 64 285 GAG to AAG Glu to Lys 8 M N Y ~ + 32 286 GAA to CAA Glu to Gln Transversion (G:C to T:A) Deletion Transversion (G:C to T:A) Transition (G:C to A:T at CpG,~ Transversion (G:C to T:A) Transversion (G:C to T:A) Transition (A:T to G:C at non CpG) Transversion (A:T to T:A) Tr'ansversion (A:T to C:G) Insertion Deletion Transversion (A:T to C:G) Deletion Transversion (G:C to T:A) Transition (G:C to A:T at non CpG) Transition (G:C to A:T at CpG) Transition (G:C to A:T at CpG) Deletion Transition (G:C to A:T at CpG) Transversion (G:C to T:A) Transition (G:C to A:T at non CpG) Deletion Transversion (G:C to T:A) Transversion (A:T to T:A) Transition (G:C to A:T at CpG) Transversion (G:C to T:A) Transition (G:C to A:T at CpG) Transversion (G:C to T:A) Transition (G:C to A:T at CpG) Transversion (A:T to T:A) Transition (G:C to A:T at non CpG) Transition (G:C to A:T at non CpG) Transversion (G:C to C:G) ~W: well-differentiated: M: moderately differentiated; P: poorly differentiated.--'S: smoker; D: drinker: Y: positive: N: negative.-3+: 10% to 30% of the tumour cells positive: + +: 30% to 50% of the tumour cells positive; + + +: More than 50% of the tumour cells positive. Mainland China rather than Hong Kong Chinese. The final explanation would require further epidemiological studies. Analysis of published results on p53 point mutations in oesopha- geal cancer revealed conflicting findings on the predominance of transversions and transitions (Andr6zet et al., 1993; Bennett et aL, 1991; Casson et a/..'1991: Est~ve et aL, 1993; Gao et al., 1994; Gates et al., 1994; Hollstein et al., 1991; Imazeki et al., 1992; Liang et al., 1995; Shibagaki et al., 1995). In Hong Kong, base substitution by transversion (46% of p53 mutations) is the most common genetic change detected in oesophageal cancer. Among these, G:C to T:A transversion (60%, 9 out of 15) was predominant. The G ---, T transversions have been associated with some chemical carcinogens, like cigarette smoke, alfatoxin, and are consistent with the lesions caused by these agents in vitro (Greenblatt et aL, 1994). In our study, all patients with this type of transversion were smokers and all except one were drinkers (Table III). However, it is difficult to draw a definite conclusion on the relation between smoking, drinking habits and p53 mutation because the majority of the patients without p53 mutations were also smokers and drinkers. In our investigation, transition accounted for 36% (12 cases) of the p53 mutations. The G:C to T:A transitions at the CpG nucleotides were found in 7 cases. CpG dinucleotides are regarded as the target of about one third of transition mutations found in human tumours (Tornaletti and Pfeifez, 1995). Methylation at these CpG dinucleotides is thought to be the cause of spontaneous mutation. Thus, as opposed to other studies, spontaneous mutations were not uncommon in oesophageal squamous cell carcinomas (Shibagaki et aL, 1995). These discrepancies in the mutation patterns observed in the local population and both Western and Oriental studies suggest the presence of both genetic and environ- mental differences among these populations. The presence of p53 mutations was not related to the clinico- pathological parameters like age and sex of the patients and the site. size, stage of cancer. The grades of the tumour seemed to be an important criteria in assessing p53 mutations. Transversions were not seen in well-differentiated squamous cell carcinomas while transitions were not found in poorly-differentiated squamous cell carcinomas. In addition, similar to our previous findings, the positivity of p53 staining was found to be higher in the poorly- differentiated carcinoma and in the relatively poorly differentiated tumour cells at the periphery of the tumour nests (Lamet al.. 1995a,b). This suggests that oesophageal squamous cell carcino- mas differentiation characterizes distinct tumours. Expression of the p53 gene has usually been studied at the protein level using immunohistochemistry. However, the presence of readily detectable protein is not necessarily synonymous with p53 mutations. In our study, we addressed this issue by testing p53 mutation using both immunohistochemical and molecular methods. We detected an incidence of 73% positivity by immunohistochem- istry and 44% positivity by PCR-SSCP-DNA sequencing. Overesti- mation of p53 mutations by immunohistochemistry could be related to mechanisms including inactivation of an enzymatic pathway responsible tbr p53 degradation t Wynford-Thomas, 1992 I. On the other hand, a small number of p53 mutations may be missed using the molecular methods due to technical limitations (Soussi et al.. 1994). In the current investigation, 16% (5 of 31 ) of the patients with p53 mutations detected by the molecular method .showed no evidence of protein over-expression by immunohistochemistry. It is worth noting that 4 out of these 5 patients had frameshift mutations that could probably end up with no detectable protein. Thus. we believe that p53 mutations as detected by molecular analysis is not

N T AGCT AGCT p53 MUTATION IN OESOPHAGEAL CARCINOMAS N T AGCT AGCT 217 FIGURE 2 - Representative examples of DNA sequence analysis; T = tumour; N = normal control. The sequence of the complementary strand are shown. A. exon 5. codon 175. C ~ T mutation. B. exon 6, codon 209, 2 base pair (TC) deletion. C. exon 7. codon, C ~ T mutation. D. exon 8. codon, G --~ A mutation. synonymous with p53 over-expression as found by immunohisto- chemistry. We have previously reviewed the incidence of p53 mutations in other published series, which ranged from 10% to 84%. Actually. in many of these, the incidence of mutation is around 50%. The incidence of 44% p53 mutation in oesophageal sqt, amous cell carcinomas as detected in the present series thus tell within the range reported b.v others. We believe that the results in our study are representative as (i) the number of patients analysed is large (n = 70): fib we have included normal tissues as a control for every patient; (iii) the blocks for the study were carefully chosen after histologic review to confirm the presence of typical features; (iv) the clinical and pathological information on patients was accurate since data were collected prospectively from people of the same ethnic group and the patients were operated by the same team of surgeons, handled by same team of pathologists and were closely followed-up. Investigations pertaining to p53 aberrations in predicting progno- sis of patients with the oesophageal cancers reveal conflicting results (Chanvitan et aL, 1995; Furihata et al., 1993: Goukon et 1994; Maesawa et al., 1994; Nakamura et aL, 1995: Sarbia et al., 1994; Shimaya et aL, 1993; Vijeyasingam et aL. 1994: Wang et aL, 1994). We believe that these discrepancies can not be explained solely by genetic and etiologic factors which affect different ethnic groups. First. different conclusions were not supported by studies within the same ethnic group (Furihata et aL, 1993: Goukon et al., 1994; Maesawa et aL, 1994; Shimaya et al., 1993). Second. immunohistochemistry was only used in all except one of the previous studies (Maesawa et aL. 1994) to detect p53 protein over-expression, which is not absolutely associated with gene

.218 Survival Function o lO ~o P53 gene mutiffon • No ~ 7O Survival time (months) FIGURE 3-- Relationship of the p53 mutation and survival rate in patients with Stage III oesophageal squamous cell carcinomas. LAM ETAL mutations. The latter study has used loss of heterozygosity (LO'H) to detect mutations in 49 oesophageal cancers and found no relationship between p53 aberrations and patient survival. Our study is the only one that analysed the relationship between prognosis and p53 mutations using both immunohistochemistry and PCR-SSCP/DNA sequencing. We found that among stage IIl oesophageal squamous cell carcinomas patients, those with p53 mutations exhibit a shorter survival than those without the muta- tions, although the results did not appear statistically significant. In addition, we noticed that the survival of patients depended mainly on the types of operations performed. This parameter was not taken into considerations in previous investigations (Chanvitan et al., 1995; Furihata et al., 1993; Goukon et aL. 1994; Maesawa et al., 1994; Nakamura et al., 1995: Sarbia et al., 1994: Shimaya et al., 1993; Vijeyasingam et aL. 1994: Wang et al., 1994). We thus conclude that p53 mutations might play a role in predicting the survival of patients with oesophageal squamous cell carcinomas, although the types of operation (curative vs. palliative) performed have a more important impact on clinical decision making. ACKNOWLEDGEMENT The authOrs thank the staff of the Departments of Pathology and Surgery, Queen Mary Hospital, for their help in handling of the specimens. This study was supported by a grant from the Univer- sity of Hong Kong (CRCG/research grant 335/046/0065). REFERENCES ANDR~Z~r, M.E, ROBASZKIEWICZ, M., MERCIER, B., NOUSBAUM. J.B.. BAIL~ J.E. HARDY. E., VOLANT, A., LOZACH, P., CHARLES, J.E, GOUr~RON, H. and FEREC, C., TP53 gene mutation profile in oesophageal squamous cell carcinomas. Cancer Res., 53, 5745-5749 (1993). BEAHRS, O.H., HENSON. D.E., HtrrrER, R.V.P. and KENNEDY. B.J.. Manual for staging of cancer, pp. 57-59, JB Lippincott, Philadelphia (1992). BENNETt, W.P., HOLt,STEIN, M.C., HE, A., ZHU, S.M., RESAU, J.H.. TRUMI', B.E, METCALF, R.A., WELSH, J.A., MIDGLEY, C., LANE, D.P. and HARRIS. C.C., Archival analysis of p53 genetic and protein alterations in Chinese oesophageal cancer. Oncogene. 6, 1779--I 784 ( 1991 ). CASSON, A.G., MUKHOPADHYAY, T., CLEARY, K.R., Ro, J.Y., LEVIN. B. and ROTH, J.A., Oncogene activation in oesophageal cancer. J. Thoracic Cardiovasc. Surg., 102, 707-709 ( 1991 ). CHANVITAN, m.o NEKARDA, H. and CASSON, A.G., Prognostic value of DNA index, S-phase fraction and p53 protein accumulation alter surgical resection of oesophageal squamous-cell carcinomas in Thailand. htt. J. Cm,cer, 63, 381-386 (1995). EST~VE, A., LEHMAN, T., JIANG, W., WEINSTEIN, B., HARRIS, C.C., RUOL, A., PERACCHIA, A., MONTESANO, R. and HOLLSTEIN, M.. Correlation of p53 mutations with epidermal growth factor receptor overexpression and absence of mdm2 amplification in human oesophageal carcinomas. MoL Carcinogenesis, 8, 306-311 (1993). FURIHATA, M., OHTSUKL Y., OGOSHI, S., TAKAHASHI, m., TAMIYA, W. and OGATA, T., Prognostic significance of human papillomavirus genomes (type- 16, - 18) and aberrant expression of p53 protein in human oesophageal cancer, h~t. J. Cancer, 54, 226-230 (1993). GAO, H.. WANG, L.D., ZHOU, Q., HONG, J.Y.. HUANG. T.Y. ~nd YANG, C.S., p53 turnout suppressor gene mutation in early oesophageal precancerous lesions and carcinoma among high-risk populations in Henan, China. Cancer Res., 54, 4342--4346 (1994). GATES, C.E.. REED. C.E., BROMBERG, J.S.. EVERET'r. E.T. and BARON, P.L.. Prevalence of p53 mutations in patients with squamous cell carcinoma of the oesophagus. J. Thoracic Cardiovasc. Surg.. 108, 148-152 (1994). GOUKON. Y.. SASANO. H.. NIStItttRA. T.. NAGURA. H. and Mo~l, S.. p53 overexpression in human oesophageal carcinoma: a correlation with turnout DNA ploidy and two parameter flow cytometric study. AnticancerRes.. 14, 1305-1312 (1994). GREENBLAq-r, M.S., BENNETT, W.P.. HOt.LSTEIN. M. and HARRIS, C.C., Mutations in the p53 tumour suppressor gene: clues to cancer actiology and molecular pathogenesis. Cancer Res., 54, 4855-4878 (1994). HOLLSTEIN. M.C., PERI. L.. ~IANDARO. A.M., WELSH. J.A.. MONTES.*NO, METCALF. R.A.. BAK. M. and HOLLSTEIN. Genetic analysis of human oesophageal turnouts from two high incidence geographic areas: frequent p53 base substitutions and absence°of ras mutations. Cancer Res., 51, 4102-4106 ( 1991 ). IMAZEKI. F., OMATA. M., NOSE, H., ONTO, M. and ISONO, K.. p53 gene mutations in gastric and oesophageal cancers. Gastroentemlog3; 103, 892-896 (1992). LAM. K.Y.. CHAN. A.C.L.. CHAN. K.W.. LEL'NG. M.L., and SRIVASTAVA. G.o Expression of p53 and its relationship with human papillomavirus in penile carcinomas. Etlrop. J. Stirs. Otlcol.. 21, 613-616 (1995b). LAM. K.Y., LOKE. S.L., CHEN, W.Z.. CHEL'NG, K.N. and MA. L.. Expression of p53 in oesophageal squamous cell carcinoma in Hong Kong Chinese. Europ. J. surg. OncoL. 21,242-247 (1995a). LAW. S.Y.K.. FoK, M., C~EN~, S.W.K. and WONG. J.. A comparison of outcome after resection tbr squamous cell carcinomas and adenocarcino- mas of the oesophagus and cardia. Surg. Gynecol. Obstet.. 175, 107-112 ( 19921. LEE, S.H., Annual Report fronl Departnlent of Health. Hong Kong 1990/91. Table 23. Hong Kong Government. Hong Kong ( 1991 ). LIANG, Y.Y., Esrf~vE. A., MARTEL-PLANCHE, G.. TA~ZAH.',SHI. S.. LU. S.H.. MONTESANO, R. and Ht)LLSTEIN, M.. p53 mutations in oesophageal tumours from high-incidence areas of China. htt. J. Cuncen 61, 611-614 (1995). LUNG, M.L., CtlAN. W.C.. ZONG. Y.S.. TANG, C.M.C., FOK, C.L.. WONG. K.T.. CnA~, L.K. and LAU, K.W., p53 mutational spectrnm of oesophageal carcinomas from five different geographic locales in China, Cancer EpidermioL Biomarker Preu, 5, 277-284 (1996). MAESAWA, C., TAMURA. G.. SUZUKI. Y.o OGASAWARA. S.. ISHIDA. K., SAtTO, K. and SATOI)ATE, R.. Aberrations of tumour-suppressor genes ( p53. ape. race and Rb) in oesophageal squamous-cell carcinoma, bit. J. Cancen 57, 21-25 (1994). MoK. S.C.H., Lo. K.W. and TsAo. S.W., Direct cycle sequencing of mutated alleles detected by PCR single-strand conformation polymorphism analy- sis. Bh~techniques, 14, 790-794 (1993). MON'rESANO. R.o HOLLSTEIN. M. and H \INAL'T. P., Genetic alterations in oesophageal cancer and their relevance to etiology aud pathogenesis: a review, hit. J. Cancer, 69, 225-235 ( 1996~. NAKAMURA. T.. [DE, H., EGUCHI. R., HAYASHI. K., HANYU. F.. NAGASAKO, K., YU~AWA. M., ASAKA. K.. Ft:J]MORL T. and M.~,EDA, S., Expression of p53 protein related to human papillomavirus and DNA ploidy in .,,uperficial oesophageal carcinoma S trg Today. 25° 591-597 ( 1995 ). P.,,~N, D.M.. LAARA. E. and MLIP,. C.S., Estimate.,, of the world-wide frequency of sixteen rnajor cancers in 1980. htt. J, Cancm: 41, 184-197 (1988).

p53 MUTATION IN OESOPHAGEAL CARCINOMAS 219 SARBIA. M.. PORSCHEN. R.. BORCHARD. F.. Ht)RSTMANN. O., WILL[~.RS. R. and GABBERT, H.E. p53 protein expression and prognosis in squamous cell carcinoma of the oesophagus. Cancer. 74, 2218-2223 (1994). SHIBAGAKI. 1., TANAKA, H.. SHIMADA. Y.. WAGATA. T.. IKENAGA. M.. IMA,".IURA. M. and |SHIZAKI. K.o 1~53 mutation, routine double minute 2 amplification, and human papillomavirus infection are frequently involved but not associated with each other in oesophageal squamous cell carcinoma, Clht. CancerRes.. 1, 769-773 (1995). SHIMAVA. K.. SHtOZAKt. H.. INOUE. M.. TAHA~A. H.. MONOEN. T.. S~MANO. T. and MO~L T.. Significance of p53 expression as a prognostic factor in oesophageal squamous cell carcinoma. Vi~vhows. Archi~: (A. PathoL Anat.), 422, 271-276 (1993). Soussh T., LEGROS. Y., LUBIN. R., ORY, K. and SCHLICHTHOLZ. B.. Multifactorial analysis of p53 alteration in human cancer: a review, h~t. Z Cancel; 57, 1-9 (1994). TORNALE'rr~. S. and PFEIFER. G.P., Complete and tissue-independent methyl- ation of CpG sites in the p53 gene: implications for mutations in human cancers. Oncogene. 10, 1493-1499 (1995). TSAO. S.W., MOK. C.H., O~KE. K.. MUTO. M., GOO[~MAN. H.M.. SHEETS. E.E., BERKOWITZ, R.S., KNAPP, R.C. and L,xu, C.C., Involvement of p53 gene in the allelic deletion of chromosome 17p in human ovarian turnouts. Anticancer Res., 11, 1975-1982 ( 1991 ). VIJEYASINGAM, R., DARNTON, S.J., JZNNER. K., ALLEN, A., BILLINGHAM. C. and MAl-rr~EWS. H.R., Expression of p53 protein in oesophageal carcinoma: clinicopathological correlation and prognostic significance• Brit. J. Surg.. 81, 1623-1626 (1994). WANG, D.Y., XIANG. Y.Y., TANAKA, M., LI. X.R., LI. J.L,, SHEN. Q.. SUGIMURA. H. and KtNO, L, High prevalence of p53 protein overexpression in patients with osophageal cancer in Linxian. China and its relationship to progression and prognosis. Cancer, 74, 3089-3096 (1994). WATANA~E, H., JASS, J.R. and SOBIN, L.H. and in collaboration with pathologists in eight countries• World Health Organization btternational histological classification of tumours: histological t3.'ping of oesophageal and gastric ntmours, pp. 1-8. Springer• Berlin (1990). WYNFORD-THoMAs, D., p53 in tumour pathology: can we trust immunocyto- chemistry? J. PathoL. 166~ 329-330 (1992).