Clonality and field cancerization in intraductal papillary-mucinous tumors of the pancreas

Authors


Abstract

BACKGROUND

Multiple lesions of intraductal papillary-mucinous tumor of the pancreas (IPMT) in the same pancreas often are encountered. To elucidate field (multicentric) cancerization and clonality of IPMT, clonal analyses of IPMT and its precursor lesion of ductal hyperplasia were performed. K-ras codon 12 mutations and X-chromosome inactivation of human androgen receptor gene (HUMARA) were investigated.

METHODS

Paraffin embedded tissue samples from the pancreata of 37 patients who underwent resection for IPMTs were microdissected manually or by laser capture microdissection. Multiple samples from each surgical specimen were microdissected representing each IPMT and discrete ductal hyperplasias. DNA was extracted, and K-ras codon 12 mutations were examined by two-step polymerase chain reaction-restriction fragment length polymorphism (PCR-RFLP). The mutations were analyzed by direct DNA sequence. The HUMARA locus was digested with or without HpaII and HhaI prior to amplification. The HUMARA assay was conducted by fluorescence-labeled PCR-RFLP and was analyzed with specialized software.

RESULTS

All 37 pancreata had at least two lesions of ductal hyperplasia, and 23 of 37 pancreata (62%) had K-ras codon 12 mutations in these precursor lesions. Of 23 pancreata with mutated K-ras hyperplasia, 15 (65%) had multiple, distinct mutations in different lesions of hyperplasia in the same pancreas, suggesting a field defect. Thirty-two of 37 IPMTs (86%) had K-ras codon 12 mutations. Among these, 16 IPMTs (50%) had multiple, distinct mutations at K-ras codon 12. The HUMARA assay showed that 12 of 15 IPMTs were informative, and 9 were considered polyclonal and/or oligoclonal origin in origin. With the combined results of multiple K-ras mutation detection and the HUMARA assay, 12 of 15 IPMTs from female patients (80%) were considered polyclonal and/or oligoclonal in origin.

CONCLUSIONS

The current results suggest that multiple, distinct K-ras mutations of different ductal hyperplasias in a given pancreas are due to a field (multicentric) cancerization effect in IPMTs. Thus, most of IPMTs are polyclonal and/or oligoclonal in origin, i.e., IPMTs may originate from multiple (molecularly distinct) precursor lesions. Cancer 2001;92:1807–17. © 2001 American Cancer Society.

Intraductal papillary-mucinous tumor of the pancreas (IPMT) is a relatively new tumor entity that has unique clinicopathologic features and has been described under various terms, including mucinous ductal ectasia, intraductal papillary neoplasm, mucin-hypersecreting neoplasm, and mucin-producing tumor of the pancreas.1–8 Its characteristic biologic behavior includes a slow-growing and less aggressive nature with a favorable prognosis after patients undergo surgery compared with common pancreatic ductal adenocarcinomas.5, 7, 8 Furthermore, one of the important clinicopathologic and molecular features of IPMT is that multifocal occurrence of IPMTs (9.8–32%) has been observed in the same pancreas,9–11 and genetic heterogeneity in an individual IPMT has been reported.12 These observations raised the questions of whether IPMTs may arise monoclonally from a single precursor cell with subsequent clonal expansion and metastasize or seed through pancreatic ducts to form secondary tumors in the same pancreas or whether there is a field defect that causes multiple primary neoplastic lesions. This would imply that originally separate neoplastic foci may coalesce to form an IPMT, thus prompting us to investigate field (multicentric) cancerization and clonality in the neoplastic evolution of IPMTs.

The hypothesis of field cancerization proposed by Slaughter et al.13 has been invoked to explain the occurrence of multiple, independent, primary neoplasms. Molecular evidence for field cancerization and clonality has been studied extensively using a molecular marker, such as p53 gene mutations, K-ras gene mutations, loss of heterozygosity, and X-chromosome inactivation (XCI) analysis in patients with head and neck carcinoma,14–16 aerodigestive tract carcinoma,17–19 breast carcinoma,20 esophageal carcinoma,21, 22 and bladder carcinoma.23

Likewise, in patients with pancreatic ductal adenocarcinomas, K-ras codon 12 mutations are found frequently in ductal hyperplasia and IPMTs,24–29 and it has been reported that a stepwise increase in the frequency of K-ras codon 12 mutations is correlated with the stage of neoplastic evolution to carcinoma.30 Therefore, K-ras codon 12 mutation is a very early event in tumorigenesis, and a hyperplasia-adenoma-carcinoma sequence in the evolution of IPMT has been recognized.24, 27, 28, 30

An assessment of field cancerization and clonality requires a good clonal marker to investigate and to qualify as a molecular marker: Such an alteration should 1) occur very early in the development of the lesion, 2) be maintained during progression of the lesion, 3) exhibit sufficient variability, and 4) be applicable in the majority of the lesions.19, 31 Because K-ras codon 12 mutation occurs frequently in approximately 30–100% of IPMTs and also in 13–83% of its precursor lesion, ductal hyperplasia,24–30, 32, 33 and because it is believed that K-ras mutation is a stable tumor marker throughout carcinogenesis,34, 35 K-ras codon 12 mutation is a good marker to investigate field cancerization and clonality of IPMT. In addition, the clonality of various tumors has been inferred using a polymerase chain reaction (PCR)-based assay for nonrandom X-chromosome inactivation at the human androgen receptor gene locus (HUMARA) in female patients, and the HUMARA assay, among other X-linked probes, has yielded superior results.36–40

Therefore, we studied K-ras codon 12 mutations by analyzing multiple tissue samples taken from multiple different lesions of ductal hyperplasia distributed in the same pancreas that was resected surgically for IPMT and from multiple foci within an individual IPMT. Moreover, in female patients with IPMT, the pattern of XCI using a highly polymorphic HUMARA assay also was examined to assess clonality using laser capture microdissection (LCM)41, 42 for a precise microdissection of tumor cells from the tissue section. We found that field (multicentric) cancerization or field defect and polyclonality and/or oligoclonality of IPMTs were frequent, and the results are presented in this article.

MATERIALS AND METHODS

Tissue Specimens and Histologic Analysis

Surgically resected pancreatic specimens were obtained from 37 patients with IPMT (22 men and 15 women; mean age, 68.3 years; age range, 44–82 years) who underwent surgery at Asahikawa Medical College Hospital and affiliated hospitals between October, 1987 and February, 2000. The resected pancreas was fixed in 10% buffered formalin, serially cut along with the main duct at 4–5 mm intervals, and embedded in paraffin. Histologically, there were 15 patients with benign (adenoma) IPMT and 22 patients with borderline (moderate dysplasia) and/or malignant (carcinoma) IPMT according to the World Health Organization criteria.3 Among the 22 patients with borderline/malignant IPMTs, there were 3 patients with borderline IPMTs; 13 patients with noninvasive, malignant IPMTs; and 6 patients with invasive IPMTs. Ductal hyperplasias were classified into nonpapillary hyperplasia (NPH), papillary hyperplasia (PH), and atypical hyperplasia (AH), as described previously.24, 25, 32 Forty-eight normal ductal epithelia from 34 pancreata with IPMTs and 34 foci of 9 ductal adenocarcinomas from 9 female patients (mean, 3.8 foci per carcinoma; range, 2–8 foci per carcinoma) also were used for K-ras mutation analysis and the HUMARA assay.

Tissue Microdissection and DNA Extraction

Three to six unstained, 10-μm-thick serial sections and three to six 5-μm-thick serial sections for manual microdissection and for LCM, respectively, were obtained from paraffin blocks, depending on the number of cells in the target area. To obtain multiple samples of ductal hyperplasias distributed in a given pancreas, 196 different lesions of ductal hyperplasia were microdissected from 37 pancreata (mean, 5.3 lesions per pancreas; range, 2–14 lesions per pancreas); 54 NPH lesions from 18 patients (49%; mean, 3.0 lesions per pancreas; range, 1–7 lesions per pancreas), 37 PH lesions from 17 patients (46%; mean, 2.2 lesions per pancreas; range, 1–4 lesions per pancreas), and 105 AH lesions from 30 patients (81%; mean, 3.5 lesions per pancreas; range, 2–4 lesions per pancreas). Similarly, 197 foci were microdissected from 37 IPMTs; thus, a mean of 5.3 foci (range, 2–16 foci) were microdissected from each individual IPMT. All microdissected foci of IPMT were intraductal (even in those that contained invasive components).

Manual microdissection was performed using a 27-gauge fine needle under stereoscopic observation, excluding mesenchymal cells as much as possible. Microdissected samples from several different areas of IPMTs and ductal adenocarcinomas as well as normal ducts contained an estimated number of at least 500 cells of interest. The dissected tissues were digested in 50 μL of lysis buffer (50 mM Tris-HCl, pH 8.0; 1 mM ethylenediamine tetraacetic acid;, 0.5% Tween 20; and 1 mg/mL proteinase K) for 16 hour at 55 °C and then incubated for 10 minutes at 95 °C to inactivate proteinase K.

LCM

The HUMARA assay was performed in IPMT sections from female patients. This necessitated the use of microsampling by LCM (Arcturus Engineering Inc., Mountain View, CA).41, 42 The microdissected cells from at least three to six, serial, 5-μm-thick sections were pooled together with the lysis buffer, and DNA was extracted by treatment with proteinase K. Each LCM microdissection sampling obtained at least 300 cells of interest.

Detection of K-ras Codon 12 Mutations

Mutations at codon 12 of K-ras exon 1 were detected by using two-step PCR-restriction fragment length polymorphism (PCR-RFLP), as described previously by Levi et al.43 The DNA from the K-ras sequence of exon 1 was amplified by a first PCR using the mismatched primers described by Levi et al. After restriction enzyme digestion using MvaI (Toyobo Company, Tokyo, Japan), PCR was performed in a 20 μL reaction mixture using Ampli-Taq Gold DNA polymerase (Perkin Elmer Applied Biosystems Division, Foster City, CA) according to the manufacturer's manual. In a second PCR and after another round of MvaI digestion, wild type fragments were cleaved to yield 29 and 106 base pair (bp) products, whereas mutant fragments yielded 135 bp. Electrophoresis of the digested sample on 3% agarose gel confirmed the mutation band. MIAPaCa-2, a pancreatic carcinoma cell line, was used as a positive control, and placental DNA was used as a negative control for K-ras codon 12 mutation detection.

Genomic DNA derived from MIAPaCa-2 (mutant control) cells was mixed with the DNA derived from placenta (wild type control) using ratios of from 1:0 to 1:10,240 to determine the sensitivity of two-step PCR-RFLP analysis. The difference in the intensity of the two major bands (wild type, 106 bp; mutant, 135 bp) was clearly distinguishable using the 1:1280 mixture, suggesting that mutation can be detected when mutant cells comprise approximately > 0.1% of the total cell population.

Direct Sequencing Analysis

The mutations identified by two-step PCR-RFLP were confirmed by direct sequencing. Mutated DNA was eluted from the agarose gel band using GenElute spin columns (Sigma Chemical Company, St. Louis, MO). The sequence of K-ras codon 12 was determined by automated fluorescent DNA sequencing using a dideoxy chain termination method. PCR products were used for cycle sequencing and were sequenced using Dye Terminator Cycle Sequencing Ready Reaction according to the manufacturer's protocol. PCR products were analyzed on an ABI Prism 310 Genetic Analyzer (Perkin Elmer).

HUMARA Assay

In females, either the maternally derived X-chromosome or the paternally derived X-chromosome in each cell is inactivated randomly and permanently at an early stage of embryogenesis,44 thus leading to somatic mosaicism of normal females with respect to X-linked alleles, with approximately half of the somatic cells expressing the maternal allele and the other half expressing the paternal allele. The pattern of XCI identified by the differential methylation of a site near the highly polymorphic CAG repeat in the HUMARA gene was used to determine the clonality status of IPMTs from 15 female patients according to a modification of the method described by Allen et al.35, 39, 40 Briefly, DNA was digested with 10 U RsaI, and with or without HpaII and HhaI methylation-sensitive restriction enzymes (Toyobo Company) at 37 °C for 12 hours. RsaI, which digests DNA other than the template DNA, enables accurate PCR on smaller amounts of DNA, as described previously by Nomura et al.39 Amplification of a portion of exon 1 of the HUMARA gene was performed by PCR in a final 20-μL solution containing fluorochrome (6-carboxyfluorescein)-labeled sense primer and the same concentration of nonfluorescein-labeled antisense primer. The same primers described by Wu et al.40 were used. Because the HUMARA locus contains many glycine (G) and cytosine (C) nucleotide sites, the performance of PCR is problematic. To avoid this problem, we used a GC-rich PCR system (Roche Molecular Biochemicals, Alameda, CA) according to the manufacturer's manual. The adjacent acinar cell tissue DNA and/or normal pancreatic duct epithelium DNA from the same pancreas and DNA from nine female patients with pancreatic ductal adenocarcinoma also were examined.

Genescan Analysis

Stock solutions of formamide (99.9%), Genescan Rox 400HD (size standard), and amplified, fluorescence-labeled DNA were mixed thoroughly mixed in a ratio of 12.0:0.5:1.0. Then, the mixture was electrophoresed on an ABI 310 Genetic Analyzer. After electrophoresis, the data were analyzed for fluorescent intensity of the two HUMARA alleles using Genescan software (ABI version 2.1; Perkin Elmer).

Evaluation of Clonality

Methylation-sensitive restriction enzymes, HpaII and HhaI, which cannot digest methylated (inactive) alleles, were used to cut the nonmethylated (active) allele before PCR. Thus, only the methylated (inactive) allele was amplified. When the methylation patterns are random (polyclonal), two PCR products are obtained, whereas, when the methylation pattern is uniform (monoclonal), either one PCR product or a significant reduction in intensity of one of the two PCR product peaks is obtained. In monoclonal methylation patterns, if the longer or shorter allele is amplified after digestion, it is referred to as the l or s pattern of monoclonality, respectively.45 Moreover, the peak area data produced by Genescan analysis were used to calculate a clonality ratio. The ratio of both alleles (allelic ratio) after RsaI, HpaII, and HhaI digestion was obtained by comparing the amount of PCR products after normalization with respect to the sample predigested with RsaI alone.46

Statistical Analysis

Differences in incidence of K-ras mutations were compared between each histologic group and calculated by using the Fisher exact test. P values < 0.05 were regarded as statistically significant.

RESULTS

K-ras Codon 12 Mutations in Ductal Hyperplasia

Multiple samples of anatomically discrete ductal hyperplasia were microdissected for the analysis from each patient with IPMT, and there was no significant difference in the number of microdissected samples per pancreas among NPH, PH, and AH samples. In histologically normal epithelium of the pancreatic duct, K-ras mutations were detected in 2 of 34 pancreata (6%) (Table 1). Ninety-five of 196 lesions of hyperplasia (49%) had K-ras mutations. According to histologic grade, K-ras mutations were detected in 30 of 54 NPH samples (56%), in 14 of 37 PH samples (38%), and in 51 of 105 AH samples (49%).

Table 1. Incidence of Multiple K-ras Mutations Samples of Normal Epithelium and of Ductal Hyperplasia from 37 Patients with Intraductal Papillary-Mucinous Tumors of the Pancreas
SampleNo. of patientsK-ras mutations (%)Single identical mutation (%)Multiple distinct mutations (%)
  • NPH: nonpapillary hyperplasia; PH: papillary hyperplasia; AH: atypical hyperplasia.

  • a

    P < 0.05.

  • b

    P < 0.05.

Normal epithelium342 (6)2 (100)0
Hyperplasia3723 (62)8 (35)b15 (65)a
 NPH188 (44)b1 (12)7 (88)
 PH178 (48)5 (62)3 (38)
 AH3022 (73)b10 (48)12 (52)

Twenty-three of 37 pancreata (62%) had K-ras mutations in ductal hyperplasia: Fifteen of these pancreata (65%) had multiple distinct mutations in different ductal hyperplasias, and 8 (35%) had a single identical mutation (P < 0.05) (Table 1). The 15 pancreata with multiple, distinct mutations had two, three, and four different mutation types in seven pancreata, six pancreata, and two pancreata, respectively. K-ras mutations of NPH, PH, and AH were detected in 8 of 18 pancreata (44%), 8 of 17 pancreata (48%), and 22 of 30 pancreata (73%), respectively. The incidence rate (73%) of K-ras mutations in AH was significantly higher compared with the incidence rate (44%) in NPH (P < 0.05). However, the incidence rate (88%) of multiple, distinct mutations in NPH was significantly higher compared with the incidence rate (38%) in PH (P < 0.05) and tended to be higher than the incidence rate (52%) in AH.

Twenty-three pancreata (62%) with IPMTs had K-ras mutations in hyperplasia, and multiple, distinct K-ras mutations in different hyperplastic lesions were identified in 65%. These results demonstrate the high incidence of molecular abnormalities, indicating the multicentric, independent development of ductal hyperplasia, a precursor lesion of IPMT, or field defect in the pancreas with IPMT.

The most frequent type of mutation of K-ras codon 12 mutations in hyperplasias was GTT (valine) (32%), followed by GAT (aspartic acid) (28%), CGT (arginine) (16%), TGT (cysteine) (13%), and AGT (serine) (9%). According to the histologic type of hyperplasia, mutation types GTT, GAT, CGT, TGT, and AGT were identified in 33%, 40%, 10%, 10%, and 7% of NPH samples; in 50%, 14%, 8%, 14%, and 14% of PH samples; and in 25%, 25%, 22%, 14%, and 10% of AH samples, respectively. There was no correlation between the prevalence of K-ras codon 12 mutation types and individual hyperplasias: NPH, PH, and AH.

Multiple, Distinct K-ras Codon 12 Mutations in IPMT

K-ras codon 12 mutations were detected in 32 of 37 IPMTs (86%), similar to the incidence rate (89%) in 9 ductal adenocarcinomas of the pancreas (Table 2). There was no significant difference between the incidence of K-ras mutations in patients with benign IPMTs (87%) and patients with borderline/malignant IPMTs (86%). All eight ductal adenocarcinomas with mutated K-ras had a single identical mutation type at different sampling sites within each carcinoma. In contrast, 16 of 32 IPMTs (50%) with mutated K-ras had multiple, distinct mutations within an individual IPMT, demonstrating intratumoral heterogeneity of K-ras mutations, and these IPMTs with multiple, different mutations were considered to have a polyclonal and/or oligoclonal origin. There was no significant difference between the incidence of multiple, distinct mutations in patients with benign IPMTs (46%) and patients with borderline/malignant IPMTs (53%). Of 16 IPMTs with multiple, distinct mutations, two, three, and four different mutation types were identified in 12 IPMTs, 3 IPMTs, and 1 IPMT, respectively.

Table 2. Incidence of Multiple K-ras Mutations in 37 Intraductal Papillary-Mucinous Tumors of the Pancreas and 9 Ductal Adenocarcinoma
Tumor typeNo. of patientsK-ras mutations (%)Single identical mutation (%)Multiple distinct mutations (%)
  1. IPMT: intraductal papillary-mucinous tumor of the pancreas.

Ductal adenocarcinoma98 (89)8 (100)0
IPMT3732 (86)16 (50)16 (50)
 Benign1513 (87)7 (54)6 (46)
 Borderline/malignant2219 (86)9 (47)10 (53)

Mutation types of K-ras codon 12—GAT, AGT, GTT, TGT, and CGT—were identified in 30%, 22%, 20%, 19%, and 8% of 94 foci of IPMTs with mutated K-ras, respectively. In contrast, all eight ductal adenocarcinomas with mutated K-ras had a single mutation type—GAT (50%) and GTT (50%).

HUMARA Assay in 15 Female Patients with IPMT

HUMARA assay and K-ras mutation analysis were performed using DNA extracted from multiple foci within an individual IPMTs in 15 female patients (5 benign IPMTs and 10 borderline/malignant IPMTs) (Table 3). Two patients (Patients 2 and 10) were homozygous, and PCR amplification was not successful in one patient (Patient 7). Thus, the remaining 12 patients with IPMT were informative. All tumor samples that showed a monoclonal pattern had a clonal ratio < 0.3, indicating that monoclonal cells constitute > 70% of the total cell population according to our results of the standard curve of fluorescence intensity (data not shown). One IPMT (Patient 13) showed a polyclonal pattern. Of 11 IPMTs that showed a monoclonal pattern in each focus, 8 demonstrated both l and s patterns among multiple foci taken from each tumor (Fig. 1). These eight IPMTs were regarded as polyclonal and/or oligoclonal: These tumors were comprised of multiple, independently derived, monoclonal cell populations. When the results of the HUMARA assay were analyzed in combination with the K-ras mutation analysis, 12 of 15 IPMTs (80%) were polyclonal and/or oligoclonal, and the remaining 3 IPMTs (20%) were monoclonal.

Table 3. Clonality Analysis of 15 Intraductal Papillary-Mucinous Tumors of the Pancreas in 15 Female Patients
DesignationStatusClonalityDesignationStatusClonality
HUMARAK-ras codon 12HUMARAK-ras codon 12
  1. HUMARA: human androgen receptor gene; M: malignant intraductal papillary-mucinous tumor of the pancreas (IPMT); Poly: polyclonal; N: normal pancreatic duct epithelium; WT: wild type (GGT); BL: borderline IPMT; Homo: homozygosity of HUMARA alleles; Mono(s): short pattern of monoclonality; Mono(l): long pattern of monoclonality; AC: pancreatic acinar cells; PCR: polymerase chain reaction; BN: benign IPMT.

Patient 1Patient 9
 M-1, 2, 3, 4, 5Mono(s)GATMonoclonal BL-1, 2, 3, 4, 5Mono(l)WTMonoclonal
 NPolyWT ACPolyWT
Patient 2Patient 10
 BL-1, 2HomoAGTMonoclonal BL-1, 2, 3, 4, 5, 6HomoWTPolyclonal
 NHomoWT BL-7, 8HomoTGT
Patient 3 BL-9HomoGTT
 M-1Mono(s)AGTPolyclonal NHomoWT
 M-2Mono(l)TGTPatient 11   
 M-3Mono(s)GAT BN-1Mono(l)TGTPolyclonal
 M-4Mono(l)GTT BN-2Mono(l)AGT
 M-5, 6PolyAGT and TGT BN-3Mono(l)GAT
 NPolyWTPatient 12   
Patient 4    BN-1, 2, 3Mono(s)GTTPolyclonal
 M-1, 2, 3, 4Mono(s)WTPolyclonal BN-4Mono(l)GTT
 M-5, 6, 7, 8, 9, 10Mono(l)WT ACPolyWT
 M-11Mono(s)GATPatient 13   
 M-12PolyGAT and AGT BN-1, 2, 3, 4PolyCGTPolyclonal
Patient 5 ACPolyWT
 M-1Mono(l)WTPolyclonalPatient 14   
 M-2Mono(s)AGT BN-1, 2, 3Mono(l)GTTPolyclonal
 M-3, 4, 5Failed PCRWT BN-4Mono(s)GTT
 ACPolyWT BN-5, 6, 7PolyGTT
Patient 6 ACPolyWT
 M-1Mono(s)GATPolyclonalPatient 15   
 M-2Mono(s)AGT BN-1Mono(l)WTPolyclonal
 M-3Mono(l)WT BN-2Mono(s)GAT
 M-4Mono(s)WT NPolyWT
 NPolyWT
Patient 7
 M-1Failed PCRTGTPolyclonal
 M-2Failed PCRAGT
 M-3, 4, 5Failed PCRWT
 NFailed PCRWT
Patient 8
 M-1, 2, 3, 4, 5, 6, 7, 8, 9Mono(s)WTPolyclonal
 M-10, 11Mono(l)WT
 M-12, 13, 14, 15PolyWT
 NPolyWT
Figure 1.

Clonality analysis in a representative patient with intraductal papillary-mucinous tumor of the pancreas (IPMT) (see Table 3, Patient 3, M-2 and M-3). (A) A malignant IPMT represented in the histologic section adjacent to the section used in microdissection by laser capture microdissection (LCM) (original magnification, ×4). (B) Two different foci (a and b) within the same IPMT were microdissected by LCM (original magnification, × 4). (C) Sample a exhibits a GGT (wild type; glycine) to TGT (cysteine) mutation in the K-ras codon 12 analysis. (D) Sample b exhibits a GGT to GAT (aspartic acid) mutation. (E,F) Sample a demonstrates a longer (l) pattern of monoclonality, whereas sample b demonstrates a shorter (s) pattern of monoclonality on human androgen receptor gene assay. Two different foci (a and b) within the same IPMT have distinct K-ras mutation types and monoclonal patterns of different parental origin.

Conversely, among nine patients with ductal adenocarcinoma, PCR amplification was not successful in two patients, and two patients were homozygous. Four patients showed either an l pattern or an s pattern of monoclonality. One IPMT was polyclonal and/or oligoclonal, because it showed both an l pattern and an s pattern among multiple samples within the tumor (data not shown). When combined with the results of K-ras mutation analysis, 8 of 9 patients had a single, identical mutation among multiple samples taken from each tumor, 8 of 9 patients (89%) with ductal adenocarcinoma had monoclonal patterns.

DISCUSSION

Recent studies on K-ras mutations of ductal hyperplasia demonstrated a high prevalence of K-ras mutations in patients with pancreatic ductal adenocarcinoma33, 47 and chronic pancreatitis.24 Furthermore, K-ras mutations were identified frequently in patients who had a disease free pancreas.25, 32 Thus, K-ras mutations occur very early, and hyperplasia, in general, is regarded as an early precursor lesion of pancreatic adenocarcinomas.24, 25, 33, 47, 48 Although K-ras mutations are found in ductal hyperplasias and probably play an important role in pancreatic carcinogenesis, it also has been suggested that K-ras mutations have a limited role and that only a small fraction of hyperplastic lesions with mutated K-ras progresses to carcinoma (< 1%).25, 47, 49 Nevertheless, K-ras mutation is an early key event leading to later genetic alterations, including inactivation of the p16 tumor-suppressor gene.47, 48 Multiple distinct K-ras mutations reportedly were identified among different hyperplasias in 47–50% of patients with ductal adenocarcinoma,33, 47 and 6–12% of patients with carcinoma had multiple, distinct mutations,47, 50 adding molecular support for field (multicentric) cancerization. Such a field defect is compatible with multiple, independent, transforming events in each lesion occurring in the same pancreas.

Compared with ductal adenocarcinoma, IPMT is a slow-growing tumor with a favorable prognosis after patients undergo surgery.5, 7, 8 However, multifocal occurrences and recurrences after surgical resection have been reported,9–11, 51 and these facts may represent a field defect23 in the multicentric, independent development of IPMTs. Herein, we report for the first time field (multicentric) cancerization and clonality in IPMTs. K-ras mutation analysis and the HUMARA assay were performed on multiple tissue samples that were microdissected from different hyperplasias distributed in the same pancreas and multiple foci within the same IPMT. Because K-ras mutation is a very early event, hyperplasia, which is a precursor lesion of IPMT, occurs almost exclusively at codon 1228, 30, 33 and stable, K-ras codon 12 mutations34, 35 can be used as a clonal marker to investigate field cancerization and clonality in the development of IPMTs as well as in ductal adenocarcinoma of the pancreas. If multiple, different hyperplasias in the same pancreas have distinct K-ras mutation types, then these hyperplasias are multicentric and independent lesions, which suggests field (multicentric) cancerization or field defect whereby carcinogenic insults result in the independent transformation of different epithelial cells.23

The current results from K-ras mutation analysis in hyperplasia and IPMT are consistent with previous studies,28, 30 which reported that K-ras mutation in IPMT is a very early event of carcinogenesis and that the incidence of K-ras mutations increases with pathologic atypia. Z'graggen et al.30 clearly demonstrated a stepwise increase in the incidence of K-ras mutations in the evolution of IPMT, suggesting a field defect. Multiple samples were microdissected from various lesions of different histologic grades, although there were not enough for investigating multiple, distinct mutations in the same IPMT-bearing pancreas. There is only one study in which extensive multiple sampling from hyperplasia and IPMT was performed: A recent study by Matsubayashi et al.33 demonstrated that K-ras mutations in hyperplasia were recognized in 11 of 12 patients with IPMTs (92%), and multiple K-ras mutations were identified in 5 of 11 patients with hyperplasia (45%). In the current study, K-ras mutations were found in hyperplasia in 62% of patients with IPMTs, 65% of whom had multiple, distinct mutations in different hyperplasias. Our results are in agreement with those of Matsubayashi et al. and support the hypothesis of field (multicentric) cancerization or field defect in the pancreas with IPMT. There has been accumulating molecular evidence to support field cancerization in various organs.14–22, 33 We also demonstrated molecular evidence for field cancerization or field defect in the pancreas bearing IPMT, and patients with IPMT are at risk for synchronous and/or metachronous, multiple tumors before and after surgery, which may affect the current clinical approach.10, 11, 30, 51

Clonality of the tumor is a fundamental issue directly relevant to tumorigenesis, and neoplastic tissues are believed to be comprised of a monoclonal cell population. However, some neoplasms, such as colon adenoma in patients with familial adenomatous polyposis,52 pancreatic endocrine tumors,53 and fibroadenoma of the breast,54 seem to be polyclonal in origin. Clonal analyses of tumors yield conflicting results: Most indicate monoclonality, and some indicate polyclonality. Polyclonality is attributable either to contamination with stromal cells or to lesions comprised of more than one monoclonal, single tumor cell population.55, 56 To minimize contamination for clonality analysis, we employed LCM, a novel, membrane-based microdissection technique recently established as a valuable tool by Emmert-Buck et al.41, 42 Thus, a precise microdissection and a highly sensitive PCR technique enabled us to analyze multiple, different foci within each IPMT as well as epithelial cells from small ductal lesions.

An assessment of clonality requires a good clonal maker based on genetic alterations,19, 31 as described above. Among the clonal markers applicable to IPMT, a combination of K-ras mutation analysis and the HUMARA assay has a greater impact on clonality analysis than either one used alone. Each method has some drawbacks: K-ras mutation analysis may underestimate the prevalence of different clones, because certain preferential mutation types occur at a relatively high frequency, resulting in the same mutation type in lesions that are probably of different clonal origin. The HUMARA assay is applicable only to female patients, and the results from approximately 10% female patients are uninformative. In addition, XCI has the issue of patch size56: If a sample was taken within a patch, then it shows a monoclonal pattern. The patch size of normal epithelium of the pancreatic duct has not been determined, but normal epithelium was polyclonal in our study, and the cell numbers from microdissection were not remarkably different among various ductal lesions and tumor foci. Consequently, we employed a combination of both analyses to obtain a more accurate assessment of clonality.

Clonality of IPMT has not been investigated to date, and this is the first study on the clonality of IPMT with the HUMARA assay and K-ras analysis. In accordance with field cancerization in IPMT, we found that most of IPMTs are polyclonal and/or oligoclonal, resulting from the fusion of separate lesions. Fifteen IPMTs from 15 female patients were subjected to both analyses on multiple, different foci taken from each tumor. K-ras analysis (Table 3) demonstrated that 2 of 15 IPMTs (13%) had wild type alone and 13 of 15 IPMTs (87%) had mutations, 6 of which (46%) had multiple, different mutations. If a given IPMT has multiple different mutations at K-ras codon 12, then the IPMT should be of polyclonal and/or oligoclonal origin—a mixture of daughter cells derived from two or more progenitor cells independently initiated by K-ras activation57: Thus, the presence of intratumoral heterogeneity of K-ras mutations is considered of polyclonal and/or oligoclonal origin rather than genetic heterogeneity occurring during tumor progression. HUMARA analysis of 12 informative tumors showed that only 3 IPMTs had a monoclonal pattern (Patients 1, 9, and 11 in Table 3), and the remaining 9 IPMTs were polyclonal and/or oligoclonal. Therefore, combining these results with the K-ras mutation status demonstrated that 12 of 15 IPMTs (80%) were polyclonal and/or oligoclonal in origin. These results suggest that IPMTs are frequently polyclonal and/or oligoclonal and that most individual IPMTs can be considered a result of the fusion of two or more independent monoclonal cell populations that arose as a consequence of field cancerization.

Although field cancerization also has been suggested33 for ductal adenocarcinoma, in our study, 8 of 9 ductal adenocarcinomas (89%) were monoclonal. It is conceivable that the most likely explanation for the discrepancy of clonality between ductal adenocarcinoma and IPMT may be rapid and aggressive growth compared with a slow-growing and less aggressive nature, respectively. It is known that the evolution of IPMT from hyperplasia through atypical hyperplasia and from adenoma to carcinoma requires a long time, which may allow multiple, originally independent lesions to coalesce, thus resulting in polyclonality. Conversely, in ductal adenocarcinoma, even if field defects yield multiple preneoplastic and neoplastic lesions, these initially may have a polyclonal and/or oligoclonal cellular composition. Nevertheless, the presence of a more aggressive single clone characterized by a selective growth advantage results in monoclonality.

Although there have been no reports focusing on the clonality of IPMT, Matsubayashi et al.33 reported that multiple K-ras mutations were detected in 3 of 11 IPMTs (27%) with mutated K-ras. In our series, 16 of 32 IPMTs (50%) showed multiple, distinct mutations. Thus, analysis of K-ras mutations alone gives rise to lower frequencies of polyclonality, because the same mutation type occurs often.

In conclusion, clonality analysis using a combination of K-ras codon 12 mutation and HUMARA analysis allowed the identification of frequent, multiple, distinct mutations in different ductal hyperplasias from the same pancreas. These findings support the hypothesis of field cancerization or field defect and results show that polyclonal and/or oligoclonal IPMT arises as a consequence of the fusion of two or more monoclonal neoplastic cell populations of independent origin.

Acknowledgements

The authors are very grateful to Drs. Tsuneshi Fujii, Yasuhiro Nakano (Department of Gastroenterology, Asahikawa Kosei Hospital, Japan), Hiroyuki Maguchi and Toshiya Shinohara (Center of Gastroenterology and Hepatology, Teine Keijinkai Hospital, Japan), and Masaru Suga (Doutou Hospital, Japan) for providing tissue blocks.

Ancillary