Ki-Ras Codon 12 Point and P53 Mutations:A Molecular Examination of the Main Tumor,Liver, Portal Vein, Peripheral Arterial Blood andPara-Aortic Lymph Node in Pancreatic CancerKoji Yamaguchi, M.D., Ph.D., Kazuo Chijiiwa, M.D., Ph.D., Nobuhiro Torato, R.T.,Moritoshi Kinoshita, Ph.D., and Masao Tanaka, M.D., Ph.D.Department of Surgery and Oncology, Graduate School of Medical Sciences, Kyushu University, Fukuoka,Japan; and Gene-Diagnostic Center, Otsuka Assay Laboratories, Otsuka Pharmaceutical Company, Ltd.,Tokushima, Japan

OBJECTIVE: Frequent P53 mutations and Ki-ras codon 12point mutations have been reported in pancreatic cancer.Pancreatic cancer often recurs in the liver and/or lymphnodes shortly after a surgical resection. The purpose of thisstudy is to elucidate the occurrence of microcirculatingcancer cells and micrometastasis in pancreatic cancer.

METHODS: P53 mutations and Ki-ras codon 12 point muta-tions were examined in the main tumor, liver, portal vein,and peripheral arterial blood, and para-aortic lymph nodesof patients with pancreatic cancer using molecular exami-nations.

RESULTS: P53 mutations in the main tumor were present innine (29%) of 31 patients with pancreatic cancer, whereas aKi-ras codon 12 point mutation was evident in 18 (62%) of29 examined patients. The peripheral arterial and portal veinblood and liver were positive for gene abnormalities in one(5%) of 21, in none (0%) of 19, and in one (1%) of 20,respectively. A P53 mutation in the main tumor was evidentin none (0%) of seven stage I or II carcinomas and in nine(38%) of 24 stage III or IV cases, whereas a Ki-ras codon 12point mutation was present in four (67%) of six stage I or IIcases and in 14 (61%) of 23 stage III or IV cases. Inaddition, 15 (71%) of 21 patients with gene abnormalities(Ki-ras codon 12 point and/or p53 mutation) in the maintumor showed lymph node metastasis at surgery, whereasfive (42%) of 12 without gene abnormalities did not dem-onstrate lymph node metastasis. Two (29%) of six patientswith gene abnormalities in the main tumor and withoutmetastatic disease at surgery developed liver metastasiswithin 6 months after surgery, whereas all five (100%)without the gene abnormalities and metastatic disease atsurgery did not develop the metastasis, with the sensitivitybeing 100%, specificity 44%, the predictive value of thepositive test 36%, and the predictive value of the negativetest 100%. Two patients who had gene abnormalities in thepara-aortic lymph node were free from histopathological

metastasis and these two patients developed para-aorticlymph node metastasis within 6 months after surgery.

CONCLUSIONS: A molecular examination of Ki-ras codon12 and p53 mutations therefore enables us to predict, tosome degree, the occurrence of liver and lymph node me-tastasis in pancreatic carcinoma. (Am J Gastroenterol 2000;95:1939–1945. © 2000 by Am. Coll. of Gastroenterology)

INTRODUCTION

Activation of promoter gene and/or the loss of such sup-pressor genes as Ki-ras point mutation (1) and P53 mutation(2, 3) have been reported in human carcinomas includingpancreatic carcinoma. In addition, a few gene abnormalitieshave also been reported in pancreatic carcinoma includingDCC (4), DPC4 (5), MTS1 (6), and BRCA2 (7). The rasgene family encodes the 21-kDa membrane-bound proteinsinvolved in signal transduction. The mutation in the Ki-rasoncogene that leads to the activation of p21 protein involvesthe 12th codon. These mutations almost exclusively involvea single base pair substitution, which leads to an amino acidsubstitution, from glycine to asparic acid, from glycine tovaline, or from glycine to arginine. The Ki-ras codon 12point mutation is frequently seen in carcinoma of the gall-bladder (8), bile duct (8), intrahepatic bile duct (9) andpancreas (10). The P53 gene is located at 17p 13.3 and is acommon tumor suppressor gene seen in human carcinomas.This gene is frequently mutated in a variety of humancancers, with the incidence being about one-half that ofpancreatic carcinoma (11–13) and 30% of that in gallblad-der carcinoma (14). The presence of mutations is detected inexons 5–8, which contain 86% of all mutations reported forthe p53 gene (12). Thanks to recent advances in molecularbiology, circulating cancer cells and micrometastasis havebeen reported in colorectal cancer (15, 16), breast cancer(17), and pancreatic cancer (18–21) using immunohisto-chemical analyses for cytokeratin or biological analyses for

THE AMERICAN JOURNAL OF GASTROENTEROLOGY Vol. 95, No. 8, 2000© 2000 by Am. Coll. of Gastroenterology ISSN 0002-9270/00/$20.00Published by Elsevier Science Inc. PII S0002-9270(00)00873-X

carcinoembryonic antigen, cytokeratin, Ki-ras point muta-tions, and P53 mutations. Such evidence shows that circu-lating cancer cells or micrometastasis are seen early in theabsence of clinical metastasis. Concerning pancreatic can-cer, the possible existence of micrometastasis in the liver iscompatible with the clinical experience that the majority ofpatients with pancreatic carcinoma develop hepatic metas-tasis after a surgical resection (22). In this study, molecularanalyses of the main tumor, liver, portal vein, and peripheralarterial blood and para-aortic lymph node were performed toexamine the prevalence and frequency of both Ki-ras andP53 mutations in pancreatic cancer, and also to elucidateany possible clinical implications of performing a molecularexamination in pancreatic cancer patients.

MATERIALS AND METHODS

Tissue samples and DNA ExtractionThis series consisted of 35 Japanese patients with pancreaticcarcinoma. All 35 patients underwent a laparotomy forpancreatic carcinoma in the Department of Surgery andOncology, Graduate School of Medical Sciences, KyushuUniversity, Fukuoka Japan, from March 1994 through May1998. They included 20 men and 15 women, ranging in ageranged from 47 to 76 yr with a mean age of 64.66 1.2 yr.They were all diagnosed with pancreatic carcinoma by his-topathology and/or cytology. The tumor was located in thehead of the pancreas in 28 patients, in the body in two andin the tail in the other five. A surgical resection was per-formed in 25 of the 35 patients. Of the 35 laparotomypatients, a sample of the main tumor was available in 33,peripheral arterial blood in 21, portal vein blood in 19, liverin 20, para-aortic lymph node in 13, and the surroundingnoncancerous pancreatic tissue in 19. The main tumorranged from 2 cm to 15 cm in size with a mean of 4.56 0.3cm. Staging was determined according to the cancer stagingmanual of the American Joint Committee on Cancer (23).The 35 tumors were classified as follows: five were in stageI, three stage II, seven stage III, 15 stage VIa, and five stageVIb. No lymph node metastasis was evident based on ahistopathological analysis in 12 of the 35 patients but waspresent in the other 23 (regional lymph node metastasis in17, and distant lymph node metastasis in the other six). Thehistopathological grade of differentiation was well-differen-tiated adenocarcinoma in 17, moderately differentiated ad-enocarcinoma in 11, and poorly differentiated adenocarci-noma in seven. Radiation therapy was performed as a rulewhen the situations permitted it and was done in 29 patients(intraoperative radiation therapy in 14 and postoperativeexternal radiation in 27). Chemotherapy (mainly with flu-orouracil) was also routinely performed when the situationpermitted it, and was done in 24 patients after the operation.The disease-free intervals and clinical outcome were obtain-able in 25 surgically resected patients up to May 31, 1999(disease-free interval: 14–790 days with a mean of 248651 days and a median of 172 days, survival: 64–1463 days

with a mean of 3946 73 days and 269 days). Metastaticdisease was found in 11 of the 25 patients on the clinicalfollow-up CT. Of the 25 surgically resected patients, 16 diedfrom recurrence of pancreatic carcinoma while the remain-ing nine were doing well with metastatic disease in one andwithout metastatic disease in the other eight. The follow-upcomputed tomographic findings of the abdomen were ob-tainable within 6 months after operation in the 25 patientswho underwent a surgical resection. Early liver and lymphnode metastasis was defined as the occurrence of metastasison the follow-up computed tomographic findings within 6months after surgery in patients without metastatic depositsat the time of surgery.

In the operating room, the peripheral arterial blood wassampled from the arterial line cannulated in the radial arteryto monitor the blood pressure just after the patient wasanesthetized. The portal vein blood was obtained from thehepatoduodenal ligament just after the removal of the tumor.Para-aortic lymph node was dissected and sampled afterKocher’s maneuver (mobilization of the duodenum andpancreatic head from the inferior vena cava) at the begin-ning of the operation. Liver sampling (segment IV) wasdone just after the main tumor was removed. A dissectedpara-aortic lymph node and liver sample were cut into twoparts in the operating room. One was subjected to a his-topathological examination, whereas the other underwent amolecular examination. The surrounding noncancerous pan-creatic tissue, areas of the normal pancreas that were as faras possible from the pancreatic cancer, was also sampledfrom the resected specimen in the operating room and thenunderwent a molecular diagnosis. The main tumor samplingwas done by the surgical pathologists themselves. At thetime of sampling, careful attention was paid so not to in-terfere with the histopathological examination of the tumorstaging and resected margins.

Informed consent was obtained from each patient. Be-cause no ethics committee for human research yet exists atour institute, the protocol was submitted to and preapprovedby the Senior Staff Committee of the Department.

The tissue samples were frozen in liquid nitrogen in theoperating room and stored at280°C. The blood sampleswere stored with an RNAase inhibitor at24°C in theoperating room. Genomic DNA was extracted from thetissue and blood specimens by proteinase K digestion andphenol/chloroform extraction using the method of Sam-brook et al. (24) for the Ki-ras codon 12 point mutationassay and P53 mutation examination.

Primers for Ki-Ras Codon 12Point Mutations and P53 MutationsIn the examination of Ki-ras codon 12 point mutations, weused either a mixture of three synthetic oligonucleotides thatcorresponded to the possible variants of the first nucleotide,or three oligonucleotides that corresponded to the threevariants of the second nucleotide of codon 12 of the Ki-rasgene for one primer for PCR. The primers used for PCR

1940 Yamaguchi et al. AJG – Vol. 95, No. 8, 2000

were synthesized in an ABI DNA synthesizer model 392(Applies Biosystem, Foster City, CA) and two sets of prim-ers were used (25).

In analyzing the P53 mutations, oligonucleotides weresynthesized as primers for PCR based on the published P53gene sequence for each region flanking the intron/exon 5–8(26, 27). The number in each designation indicates theregion of the exon of the P53 gene examined by PCR-FSSCP (fluorescence-based single-strand conformationpolymorphism) analysis. All primers were labeled at their 59ends with fluorescein derivatives by the Fluore primemethod (Pharmacia Biotech, Tokyo, Japan).

Polymerase Chain Reaction forKi-Ras Codon 12 Point Mutation andP53 Mutation Ki-ras Codon 12 Point MutationPCR was performed in 40 cycles of 0.5 min at 95°C, 2 minat 53°C, and 2 min at 70°C, as described by Bakeret al.(28). Mutant-allele-specific amplification (MASA) was per-formed in 32 cycles of 0.5 min at 95°C, 2 min at 59°C or60°C, and 2 min at 70°C, by a modification of the methodof Takedaet al. (25) except without dimethyl sulfoxide(DMSO). PCR products (5ml) were electrophoresed in a 3%agarose gel containing 0.5mg/ml ethidium bromide.Ethidium-bromide staining of the PCR products was per-formed for the detection of point mutations in the Ki-rasgene and visualized under ultraviolet light (Fig. 1).

PCR-FSSCP Analysis for P53 MutationsA quantity of 100 ng of genomic DNA was amplified in atotal volume of 50ml in the buffer containing 100mmmol/Lconcentrations of dATP, dCTP, dGTP, and TTP, and 0.125U of Amplitaq in the buffer recommended for the enzyme,using the primers labeled at 59 ends with fluorescein. ThePCR products were diluted 50 times by a stop solution(Pharmacia Biotech, Tokyo, Japan), heated at 95°C for 5min, and then placed on ice for 5 min. Then 4ml of this

solution was applied to each lane of FSSCP gel fitted to anALF II automated DNA sequencer (Pharmacia Biotech).The FSSCP gels used were 7% polyacrylamide gel contain-ing 5% glycerol. Electrophoresis was performed at 30 W for3–4 h, depending on the length of the amplified nucleotide.The temperature of the gel was kept at 25°C with a built-inwater jacket connected to an external thermostat-regulatedwater circulation system (Fig. 2).

Statistical AnalysisAll values were expressed as the mean6 standard error(SE). When the distribution of the patients was compared,thex2 test was used. The survival curve was obtained by theKaplan-Meier method, and the differences in the curveswere measured by the generalized Wilcoxon test;p values of,0.05 were considered to be statistically significant.

RESULTS

Main Pancreatic Tumor and theSurrounding Noncancerous Pancreatic TissueP53 mutations of the main pancreatic tumor were present innine (29%) of 31 with pancreatic cancer and Ki-ras codon12 point mutation in 18 (62%) of 29 (Table 1). P53 muta-tions at exon 5 were seen in five of the nine patients, and aKi-ras codon 12 point mutation at the second in 14 of the 18.P53 positivity was 0% in stages I or II and 38% in stages IIIor IV, whereas Ki-ras codon 12 positivity was 67% in stagesI or II and 61% in stages III or IV. A P53 mutation was seenin four (44%) of nine patients with nonresectable pancreaticcarcinoma, and in five (23%) of 22 with resectable pancre-

Figure 1. Ethidium-bromide staining of PCR products for detectionof Ki-ras codon 12 point mutation (second letter). Lane 1: Sizemarker pUC19/Msp1. Lanes 2–4: Positive control (53 105 copies(2), 53 102 (3), 53 101 (4)). Lanes 5–7: Healthy control (negativecontrol). Lane 8: Para-aortic lymph node of lane 10. Lane 9: Liverof lane 10. Lane 10: Pancreatic carcinoma. Lane 11: Surroundingpancreas of lane 10. Lane 12: Peripheral blood of lane 10. Lane 13:Blood of portal vein of lane 10. Band in lane 10, which is strongerthan lane 4 (positive control, 53 101 copies), is judged as positivefor Ki-ras codon 12 point mutation.

Figure 2. P53 mutation (exon 7). Pancreatic carcinoma. An uppercurve shows peaks (arrows) which are not seen in a lower curve ofhealthy control and indicate p53 mutation (PCR-FSSCP).

1941AJG – August, 2000 Gene Mutations and Pancreatic Cancer

atic carcinoma. A Ki-ras codon 12 point mutation wasevident in six (75%) of the eight with nonresectable cancerand in 12 (57%) of the 21 patients with resectable carci-noma. A P53 mutation was seen in eight (32%) of 25patients with well to moderately differentiated adenocarci-noma, and in one (17%) of six patients with poorly differ-entiated adenocarcinoma. A Ki-ras codon 12 point mutationwas seen in 12 (52%) of 23 patients with well to moderatelydifferentiated adenocarcinoma and in all six (100%) patientswith poorly differentiated adenocarcinoma.

In the surrounding noncancerous pancreatic area, a Ki-rascodon 12 point mutation (second letter) was seen in only oneof 17 patients with pancreatic cancer, whereas no P53 mu-tation was seen in the other 18.

In one of the 33 patients examined, the gene abnormalityin the main tumor was the same as in the lymph nodemetastasis. The patients with pancreatic head carcinomashowed a Ki-ras codon 12 point mutation (second letter),both in the main tumor and in paraaortic lymph node me-tastasis.

When the presence or absence of a gene abnormality in aprimary tumor was compared with that of micrometastatic/circulating cancer cells, three of 11 patients with mutation-positive primary tumors showed a micrometastasis or cir-culating cancer cells, and seven of eight patients withmutation-negative tumors did not show either micrometas-tasis or circulating cancer cells. The sensitivity of a geneabnormality in primary tumors to micrometastasis or circu-lating cancer cells was 75% and the specificity was 47%.

Gene Abnormalities (Ki-ras Codon 12 Point and/orP53 Mutation) and Metastasis at the Time of SurgeryTwo patients had liver metastasis at the time of surgery.These patients were negative for gene abnormalities in themain tumor. One of the two patients examined was negativefor gene abnormalities in the noncancerous liver (Table 2).A gene abnormality in the noncancerous liver was evident in

one (5%) of 19 patients without liver metastasis at surgery.This patient had pancreatic head carcinoma with a Ki-rasmutation (second letter) and the liver was positive for thesame mutation. The liver was free from metastasis on com-puted tomography at the time of this writing (12 monthsafter the operation).

In two (15%) of the 13 patients examined, the paraaorticlymph node was positive for gene abnormality, with a Ki-rasmutation (second letter) in two. In these two patients, his-tology failed to reveal any lymph node metastasis (Table 3).The two patients developed lymph node metastasis aroundthe paraaortic area in the early postoperative days. In onepatient with histopathologically verified lymph node metas-

Table 1. P53 Mutations and Ki-Ras Codon 12 Point Mutations in Pancreatic Cancer (All Patients)

Sites

P53 Exon Ki-Ras Codon 12Gene

Abnormalities

5 6 7 8 5–8 1st 2nd 1st–2nd Yes No

Main pancreatic tumor 5 1 2 1 9/31 4 14 18/29 21 12/33AJCC Staging

I 0 0 0 0 0/4 1 1 2/4 3 2/5II 0 0 0 0 0/3 1 1 2/2 2 1/3III 1 1 0 1 3/5 1 1 2/6 3 3/6IV 4 0 2 0 6/19 1 11 12/17 14 5/19

Grade of differentiationWell dif. 3 0 1 1 5/15 1 8 9/15 9 7/16Mod. dif. 1 1 1 0 3/10 1 2 3/8 6 5/11Por. diff. 1 0 0 0 1/6 2 4 6/6 6 0/6

Noncancerous pancreatic area 0 0 0 0 0/18 0 1 1/17 1 18/19Liver 0 0 0 0 0/19 0 1 1/16 1 19/20Para-aortic lymph node 0 0 0 0 0/12 0 2 2/12 2 11/13Peripheral arterial blood 0 0 1 0 1/19 0 0 0/20 1 20/21Portal vein blood 0 0 0 0 0/18 0 0 0/17 0 19/19

Well dif. 5 well differentiated adenocarcinoma; mod. dif.5 moderately differentiated adenocarcinoma; por. dif.5 poorly differentiated adenocarcinoma.

Table 2. Genetic Abnormalities and Metastasis in the Liver andLymph Node at Surgery in Pancreatic Carcinoma Patients

Gene abnormality

LiverMetastasis

LymphNode

Metastasis

Yes No Yes No

Overall (35)Yes 1 22 16 7No 1 11 7 5

Main pancreatic tumor (33)Yes 0 21 15 6No 2 10 7 5

Liver (20)Yes 0 1 0 1No 1 18 13 6

Para-aortic lymph node (13)Yes 0 2 2 0No 0 11 6 5

Peripheral arterial blood (21)Yes 1 0 1 0No 0 20 13 7

Portal vein blood (19)Yes 0 0 0 0No 0 19 13 6

Total number of patients examined is given in parentheses.

1942 Yamaguchi et al. AJG – Vol. 95, No. 8, 2000

tasis, a molecular examination failed to reveal any geneabnormality. When gene abnormalities of the main tumorwere compared with the presence or absence of lymph nodemetastasis at the time of surgery, 15 (71%) of 21 patientswith gene abnormalities in the main tumor showed lymphnode metastasis, whereas five (42%) of 12 patients withoutgene abnormalities in the main tumor had no lymph nodemetastasis, with the sensitivity of gene abnormalities in themain tumor to the presence or absence of lymph nodemetastasis being 68%, specificity 45%, the predictive valueof a positive test 71%, and the predictive value of a negativetest 42%.

Gene Abnormalities (Ki-Ras Codon 12 Pointand/or P53 Mutation) and Occurrence of MetastasisTwo of 12 patients without liver metastatic disease at sur-gery developed liver metastasis according to the computedtomographic findings within 6 months after surgery (Table4). In these two patients, a gene analysis was negative in theperipheral arterial blood, portal vein blood, and liver at the

time of surgery. When gene abnormalities of the main tumorand occurrence of liver metastasis were compared, two ofsix patients with gene abnormalities in the main tumordeveloped liver metastasis, whereas none of the five withoutgene abnormalities showed liver metastasis, with the sensi-tivity in gene abnormalities of the main tumor to occurrenceof liver metastasis being 100%, the specificity 56%, thepredictive value of a positive test 33%, and the predictivevalue of a negative test 100%.

Three of the 12 patients without lymph node metastasis atsurgery showed lymph node metastasis (mainly at the ret-roperitoneum) within 6 months after surgery. One of the twopatients examined showed gene abnormalities in the maintumor, whereas the other did not. One of seven with geneabnormalities in more than one sample and without meta-static disease at surgery developed paraaortic lymph nodemetastasis on follow-up computed tomography within 6months after surgery, whereas three of five without eithergene abnormalities or metastatic deposits at surgery did notdemonstrate metastasis. The sensitivity of all gene abnor-malities to the occurrence of lymph node metastasis was14%, the specificity 60%, the predictive value of a positivetest 33%, and the predictive value of a negative test 33%.

Disease Free Intervals and SurvivalBased on the Presence of a Gene AbnormalityNo significant relationship was seen between the overallsurvival and the presence or absence of gene abnormalitiesin the main tumor nor between the disease-free intervals andthe gene abnormalities in the main tumor. There was also nosignificant difference in the survival curves and the diseasefree-intervals based on the p53 mutation status of the maintumor.

DISCUSSION

In 1985, Prassolovet al. (29) first reported the activation ofthe c-Ki-ras gene in lymph node metastasis of human pan-creatic cancer. Thereafter, many papers have reported a highincidence of this point mutation in human pancreatic cancer,with the mutation usually being from GGT: Gly convertedinto GAT: Asp, into GTT: Val and into CGT: Arg. TheKi-ras codon 12 mutation in human pancreatic carcinoma isusually seen in the first and second letters of Ki-ras codon12, not in the third letter. Therefore, the presence or absenceof a mutation in the first and second letters was only exam-ined in the present series and the exact transition was notexamined. A Ki-ras codon 12 point mutation has beenreported in the peripheral blood of patients with pancreaticcancer, with the incidence reported to be 100% (10 of 10) byNomotoet al. (19), 84% (16 of 19) by Mulcahyet al. (30),and 33% (two of six) by Tadaet al. (20). Ki-ras mutationwas also reported in the liver and portal vein blood, with theincidence being 76% (13 of 17) by Inoueet al. (22), usinga two-stage polymerase chain reaction/restriction fragmentlength polymorphism (PCR/RFLP) analysis, and 11% (two

Table 3. Comparison of the Gene Analysis andHistopathological Examination Findings in the Para-AorticLymph Node in Pancreatic Carcinoma

Histopathological LymphNode Metastasis in the

Para-Aortic Lymph Node

Yes No

Gene abnormality in thepara-aortic lymph nodeYes 0 2No 1 10

Table 4. Gene Abnormalities and the Occurrence of Early Liverand Lymph Node Metastasis in Pancreatic Carcinoma PatientsWithout Metastatic Disease at Surgery

Gene abnormalities

LiverMetastasis

LymphNode

Metastasis

Yes No Yes No

Overall (12)Yes 2 5 1 6No 0 5 2 3

Main pancreatic tumor (11)Yes 2 4 1 5No 0 5 1 4

Liver (7)Yes 0 1 0 1No 1 5 1 5

Para-aortic lymph node (5)Yes 0 0 0 0No 1 4 0 5

Peripheral arterial blood (7)Yes 0 0 0 0No 1 6 1 6

Portal vein blood (6)Yes 0 0 0 0No 0 6 1 5

Total number of patients examined is given in parentheses.

1943AJG – August, 2000 Gene Mutations and Pancreatic Cancer

of 18) by Tamagawaet al. (31), respectively. In the presentstudy of pancreatic carcinoma, the gene mutation was evi-dent in 5% of the peripheral arterial blood, 0% of the portalvein blood, and 5% of the liver. The peripheral arterial bloodwas obtained from the radial artery just after the patient wasanesthetized. When the main tumor was removed, the portalvein blood was sampled from the hepatoduodenal ligamentand the liver from segment IV. The low frequency of mi-crometastasis in the liver in our series may indicate aninappropriate sampling site (segment IV), different sam-pling methods and PCR protocols, and/or the definition ofpositivity.

The survival of patients with pancreatic cancer even aftera potentially curative resection remains poor, and patientsgenerally succumb to metastatic diseases. Pancreatic carci-noma has a high potential of liver metastasis, and a molec-ular examination is expected to play an important role in thediagnosis of micrometastasis in the liver. In the presentstudy, two of the six patients with gene abnormalities in themain tumor and without metastatic diseases at surgery de-veloped hepatic metastasis shortly after resection, whereasnone of five without gene abnormalities and liver metastasisat surgery had liver metastasis. Our ability to detect grossmetastatic disease at the time of surgery is good, but thedetection of micrometastases and circulating cancer cells isnot part of clinical practice. A molecular examination seemsto be somewhat useful to predict the occurrence of livermetastasis in patients with pancreatic cancer and to deter-mine which patients would benefit from aggressive adjuvanttherapies.

It is supposed that molecular abnormalities in the primarysite and metastatic lesions should be the same. However, inthis study, only one patient showed the same molecularabnormality in both the main tumor and in the para-aorticlymph node. This may be explained by the heterogeneity ofthe tumor from area to area in one tumor (32, 33). When thepresence or absence of gene abnormalities in the primarytumor was compared with micrometastasis or circulatingcancer cells, the sensitivity was 75% and the specificity47%.

The lymph node status has been reported to influencestrongly the postoperative long-term survival of patientswith pancreatic head cancer (34–37). Although hematoge-nous metastasis, such as in the liver, is beyond the surgeon’sability to control, the management of the lymph node is inthe surgeon’s hands and the effects of radical lymph nodedissection have been reported in pancreatic cancer (38, 39).The para-aortic lymph nodes examined in the present seriesare at the final site of the lymphatic stream of the abdomen(40) and, when the para-aortic lymph node is metastasizedby cancer cells, the effects of radical surgical resection arequestionable. Therefore, lymph node status of the para-aortic area is crucial. Andoet al. (18) reported that in eightof 13 patients with pancreatic carcinoma, 42 of 101 lymphnodes showed Ki-ras point mutation in the para-aorticlymph nodes and suggested that the PCR/RFLP analysis is

potentially highly sensitive for the detection of microme-tastasis in the lymph nodes and it may be useful in reachingan accurate assessment of lymphatic dissemination in pan-creatic carcinoma at the molecular level. In our series, twolymph nodes that were negative based on a microscopicexamination were positive based on a molecular analysis. Inthese two patients, even though the number of patientsexamined was limited, the clinical follow-up by imagingtechniques demonstrated lymph node metastasis in the para-aortic area.

It was reported that, with the application of the mutantallele specific amplification method used in the presentstudy, five cancer cells among 104 normal lymphocytescould be detected. (Personal communication, 1999, M. Ki-noshita, Gene-Diagnostic Center, Otsuka Assay Laborato-ries, Otsuka Pharmaceutical, Tokushima, Japan.) Based onthe present PCR-FSSCP analysis for P53 mutation, onecancer cell among 102 normal lymphocytes could be de-tected. (Personal communication, 1999, M. Kinoshita,Gene-Diagnostic Center, Otsuka Assay Laboratories, Ot-suka Pharmaceutical, Tokushima, Japan.) DNA derivedfrom tumor cells can migrate from the original site andpersist in the lymph nodes for as long as 4 days, thuspossibly giving rise to positive reactions by genetic testing.It remains uncertain as to whether or not the mutationsdetected by enriched PCR reflect the presence of viabletumor cells in the nodes. In this study, we compared themolecular examination, histopathological diagnosis, andfollow-up imaging findings. The molecular examination ofthe liver, peripheral arterial blood, portal vein blood, andpara-aortic lymph node were rarely positive for gene abnor-malities. The molecular examination of the main tumor alsoshowed a relationship (although not so strong) to clinicalstaging, tumor differentiation, liver metastasis, and lymphnode metastasis.

In conclusion, a molecular examination may be useful forselecting subgroups of patients with pancreatic cancer whohave a possibly high potential of liver metastasis and para-aortic lymph node metastasis and thus may benefit fromaggressive adjuvant therapy, although many problems re-main to be solved regarding gene analysis.

ACKNOWLEDGMENTS

We are grateful to Ms. Junko Kusaka and Mr. Koji Tokuda,Otsuka Assay Laboratories, Otsuka Pharmaceutical Co.,Tokushima, Japan, for their technical assistance. We alsothank Mr. Brian Quinn for his critical reading of this manu-script.

Reprint requests and correspondence:Koji Yamaguchi, M.D.,Department of Surgery and Oncology, Graduate School of MedicalSciences, Kyushu University, Fukuoka 812-8582, Japan.

Received Feb. 4, 1999; accepted Feb. 23, 2000.

1944 Yamaguchi et al. AJG – Vol. 95, No. 8, 2000

REFERENCES

1. Bos JL, Fearon ER, Hamilton SR, et al. Prevalence of ras genemutations in human colorectal cancers. Nature 1987;327:293–7.

2. Redston MS, Caldas C, Seymour AB, et al. p53 Mutations inpancreatic carcinoma and evidence of common involvement ofhomocopolymer tracts in DNA microdeletions. Cancer Res1994;54:3025–33.

3. Hruban RH, Petersen GM, Ha PK, et al. Genetics of pancreaticcancer. From genes to families. Surg Oncol Clin North Am1998;7:1–23.

4. Hohne MW, Halatsch ME, Kahl GF, et al. Frequent loss ofexpression of the potential tumor suppressor gene DCC inductal pancreatic adenocarcinoma. Cancer Res 1992;52:2616–9.

5. Hahn SA, Schutte M, Hoque AT, et al. DPC4, a candidatetumor suppressor gene at human chromosome 18q21.1. Sci-ence 1996;271:350–3.

6. Caldas C, Hahn SA, da Costa LT, et al. Frequent somaticmutations and homozygous deletions of the p16 (MTS1) genein pancreatic adenocarcinoma. Nature Genet 1994;8:27–32.

7. Goggins M, Schutte M, Lu J, et al. Germline BRCA2 genemutations in patients with apparently sporadic pancreatic car-cinomas. Cancer Res 1996;56:5360–4.

8. Watanabe M, Asaka M, Tanaka J, et al. Point mutation ofK-ras gene codon 12 in biliary tract tumors. Gastroenterology1994;107:1147–53.

9. Levi S, Urbano-Ispizua A, Gill R, et al. Multiple K-ras codon12 mutations in cholangiocarcinoma demonstrated with a sen-sitive polymerase chain reaction technique. Cancer Res 1991;51:3497–3502.

10. Almoguera C, Shibata D, Forrester K. Most human carcinomaof the exocrine pancreas contain mutant c-K-ras genes. Cell1985;53:549–54.

11. Scarpa A, Capelli P, Mukai K, et al. Pancreatic adenocarci-nomas frequently show p53 gene mutations. Am J Pathol1993;142:1534–43.

12. Nigro JM, Baker SJ, Preisinger AC, et al. Mutations in the p53gene occur in diverse human tumours types. Nature 1989;342:705–8.

13. Pellegata NS, Sessa F, Renault B, et al. K-ras and p53 genemutations in pancreatic cancer: Ductal and nonductal tumorsprogress through different genetic lesions. Cancer Res 1994;54:1556–60.

14. Takagi S, Naito E, Yamanouchi H, et al. Mutation of the p53gene in gallbladder cancer. Tohoku J Exp Med 1994;172:283–9.

15. Hayashi N, Arakawa H, Nagase H, et al. Genetic diagnosisidentifies occult lymph node metastases underdetectable by thehistopathological method. Cancer Res 1994;54:3853–6.

16. Hayashi N, Ito I, Yanagisawa A, et al. Genetic diagnosis oflymph-node metastasis in colorectal cancer. Lancet 1995;345:1257–9.

17. Dowlatshahi K, Fang M, Snider HC, et al. Lymph node me-tastases from breast carcinoma. Cancer 1997;80:1188–97.

18. Ando N, Nakao A, Nomoto S, et al. Detection of mutant K-rasin dissected paraaortic lymph nodes of patients with pancreaticadenocarcinoma. Pancreas 1997;15:374–8.

19. Nomoto S, Nakao A, Kasai Y, et al. Detection of ras genemutations in perioperative peripheral blood with pancreaticadenocarcinoma. Jpn J Cancer Res 1996;87:793–7.

20. Tada M, Omata M, Kawai S, et al. Detection of ras genemutations in pancreatic juice and peripheral blood of patientswith pancreatic adenocarcinoma. Cancer Res 1993;53:2472–4.

21. Tamagawa E, Ueda M, Takahashi S, et al. Pancreatic lymphnodal and plexus miscrometastases detected by enriched poly-merase chain reaction and nonradioisotopic single strand con-formation polymorphism analysis: A new predictive factor forrecurrent pancreatic carcinoma. Clin Cancer Res 1997;3:2143–9.

22. Inoue S, Nakao A, Kasai Y, et al. Detection of hepatic micro-metastasis in pancreatic adenocarcinoma patients by two-stagepolymerase chain reaction/restriction fragment length poly-morphism analysis. Jpn J Cancer Res 1995;86:626–30.

23. American Joint Committee on Cancer. AJCC cancer stagingmanual, 5th ed. New York: Lippincott-Raven, 1997.

24. Sambrook J, Maniatis T, Fritsch EF. Molecular cloning. Alaboratory manual, 2nd ed. New York: Cold Spring HarborLaboratory Press, 1989.

25. Takeda S, Ichii S, Nakamura Y. Detection of K-ras mutationsin sputum by mutant-allele-specific amplification (MASA).Hum Mutation 1993;2:112–7.

26. Buchman VL, Chumakov PM, Ninkina NN, et al. A variationin the structure of the protein-coding of the human p53 gene.Gene 1988;70:245–52.

27. Katsuragi K, Chiba W, Matsubara Y, et al. A sensitive andhigh-resolution method for the detection of mutations in thep53 gene using hluorescence-based PCR-SSCP. Biomed Res1995;16:273–9.

28. Baker SJ, Preisinger AC, Jessup JM, et al. p53 Mutationsoccur in combination with 17-allelic deletions as late events incolorectal tumorigenesis. Cancer Res 1990;50:7717–22.

29. Prassolov VS, Sakamoto H, Sugimura S. Activation of c-Ki-ras gene in human pancreatic cancer. Jpn J Cancer Res 1985;76:792–5.

30. Mulcahy HA, Anker P, Chen X, et al. K-ras mutation in DNAextracted from plasma of pancreatic cancer patients. Gastro-enterology 1997;112:A620.

31. Tamagawa E, Uematsu S, Kobayashi K, et al. Molecularbiological approach to pancreatic cancer. J Clin Surg 1998;53:863–7 (in Japanese).

32. Mulder MP, Keijzer W, Verkerk A, et al. Activated ras genesin human seminoma: Evidence for tumor heterogeneity. On-cogene 1989;4:1345–51.

33. Giaretti W, Monaco R, Pujic N, et al. Intratumoral heteroge-neity of K-ras 2 mutations in colorectal adenocarcinomas.Association with degree of DNA aneuploidy. Am J Pathol1996;149:237–45.

34. Cameron JL, Crist DW, Sitzmann JV, et al. Factors influenc-ing survival after pancreaticoduodenectomy for pancreaticcancer. Am J Surg 1991;161:120–5.

35. Cameron JL. Long-term survival following pancreaticoduode-nectomy for adenocarcinoma of the head of the pancreas. SurgClin North Am 1995;1995:939–51.

36. Yeo CJ, Cameron JL. Pancreatic nodal metastases: Biologicsignificance and therapeutic considerations. Surg Oncol ClinNorth Am 1996;5:145–57.

37. Yeo CJ, Cameron JL. Prognostic factors in ductal pancreaticcancer. Langenbecks Arch Surg 1998;383:129–33.

38. Nagakawa T, Nagamori M, Futakami F, et al. Results ofextensive surgery for pancreatic carcinoma. Cancer 1996;77:640–5.

39. Ishikawa O, Ohhigashi H, Sasaki Y, et al. Practical usefulnessof lymphatic and connective tissue clearance for the carcinomaof the pancreas head. Ann Surg 1988;208:215–20.

40. Nagakawa T, Kobayashi H, Ueno K, et al. Clinical study oflymphatic flow to the paraaortic lymph nodes in carcinoma ofthe head of the pancreas. Cancer 1994;73:1155–62.

1945AJG – August, 2000 Gene Mutations and Pancreatic Cancer