Anatomic Pathology / MOLECULAR ALTERATIONS IN PANCREATIC NEEDLE ASPIRATES

Kras , p53 , and DPC4 ( MAD4 ) Alterations in Fine-Needle Aspirates of the Pancreas A Molecular Panel Correlates With and Supplements Cytologic Diagnosis Tjarda van Heek, MD,1 Anne E. Rader, MD,2 G. Johan A. Offerhaus, MD,1 Denis M. McCarthy, MD,3 Michael Goggins, MD,3,4 Ralph H. Hruban, MD,3,5 and Robb E. Wilentz, MD3
Key Words: Pancreas; Fine-needle aspiration; Cytology; Cytopathology; Kras ; p53 ; Dpc4; Genetics
Abstract Between January 1997 and February 2000, 101 fine-needle pancreatic aspirates were obtained. After a cytologic diagnosis was made, possible molecular alterations were determined on the 94 aspirates with adequate tissue using a molecular panel (K-ras, p53, and DPC4 [MAD4] genes). The 94 aspirates were categorized as follows: diagnostic of adenocarcinoma, 48 (51%); atypical (suggestive of, but not diagnostic of, adenocarcinoma), 19 (20%); negative for adenocarcinoma, 25 (27%); diagnostic of a neoplasm other than adenocarcinoma, 2 (2%). Clinical follow-up revealed that 3 patients (12%) with negative cytologic diagnoses and 12 patients (63%) with atypical cytologic diagnoses had adenocarcinoma. Of 63 with a final diagnosis of adenocarcinoma, 42 (67%) had an alteration in at least 1 of the genes analyzed. In contrast, only 2 (6%) of 31 patients without adenocarcinoma had an alteration in 1 gene on the panel. Overall, the molecular analyses supported the diagnosis of adenocarcinoma in 6 (32%) of 19 aspirates originally diagnosed as atypical by cytology alone. A molecular panel that includes the Kras, p53, and DPC4 (MAD4) genes correlates with and can supplement traditional cytologic diagnosis of pancreatic fine-needle aspirates.
Pancreatic cancer is a genetic disease with a characteristic molecular genetic profile.1 The genetic alterations in pancreatic cancer include the activation of an oncogene, Kras, and inactivation of tumor-suppressor genes, including p53 and DPC4 (MAD4).2 K-ras is activated in approximately 95% of pancreatic ductal adenocarcinomas, and the p53 and DPC4 (MAD4) genes are inactivated in 70% and 55% of cases, respectively.2-13 While alterations in K-ras and p53 occur often in many other types of neoplasms, inactivation of the DPC4 (MAD4) tumor-suppressor gene is relatively specific for ductal adenocarcinoma of the pancreas.11 Activation of the K-ras gene is easily assessed genetically, because the vast majority of mutations occur in codon 12.3,14-18 The p53 and DPC4 (MAD4) genes are harder to study genetically. This is because they are inactivated in cancer, either by intragenic mutation in one allele coupled with loss of the second allele (loss of heterozygosity) or by deletion of both alleles (homozygous deletion). Because pancreatic adenocarcinomas evoke intense nonneoplastic desmoplastic responses, most of the DNA contained in a tumor is actually nonneoplastic DNA. This intimate admixture of normal and neoplastic tissues can mask the loss of a p53 or DPC4 (MAD4) allele in a cancer. A simple way to overcome this problem is to study these genes immunohistochemically. Immunohistochemical stains for the p53 and Dpc4 (Mad4) proteins have recently become available, and, thus, alterations in these genes have become much easier to recognize.5,8,10,19-32 Thus, genetic and immunohistochemical techniques have provided much information about the molecular biology of pancreatic ductal adenocarcinoma. However, there have been only a few studies examining the diagnostic usefulness of
Am J Clin Pathol 2002;117:755-765 755© American Society for Clinical Pathology
van Heek et al / MOLECULAR ALTERATIONS IN PANCREATIC NEEDLE ASPIRATES
detecting these molecular changes in fine-needle aspirates of the pancreas.19,31,33-40 In addition, while K-ras and p53 alterations have been identified in fine-needle aspirates of the pancreas,19,31,33-40 no group has yet studied inactivation of the DPC4 (MAD4) gene in these specimens. Therefore, we set out to assess K-ras, p53, and DPC4 (MAD4) alterations in fine-needle aspiration specimens of the pancreas. We did this with 2 goals in mind: (1) to determine whether this molecular panel correlates with cytologic diagnosis and (2) to explore whether the use of this molecular panel could support the diagnosis of cancer in cytologic samples interpreted as atypical (suggestive of, but not diagnostic of, adenocarcinoma). In cases of atypical cytology, additional diagnostics, such as biopsy or laparotomy, are necessary to exclude a carcinoma. The morbidity and mortality attending additional diagnostics could be avoided if the use of a molecular panel could help support a benign or malignant diagnosis in patients who have atypical aspirates.
Materials and Methods
Specimen Selection Between mid-January 1997 and mid-February 2000, 101 fine-needle aspirates of the pancreas, each from a different patient, were obtained prospectively at Oregon Health Sciences University, Portland. A routine clinical diagnosis was made on each of these cases at the time of the procedure by one of a group of pathologists at this institution. These cases were then collected and rereviewed by 1 cytopathologist (A.E.R.). Based on diagnoses made on rereview of Papanicolaou- and rapid Romanowsky–stained smears and H&E-stained cell-block sections, each specimen was grouped into 1 of 4 categories: (1) diagnostic for adenocarcinoma, (2) suggestive of adenocarcinoma, (3) no evidence of malignancy, and (4) diagnostic of neoplasm other than adenocarcinoma ?Image 1?. This categorization was performed before molecular analysis began.
Data Procurement and Statistical Analysis Clinical and pathologic data for each case were obtained from patients’ medical records and the Social Security Death Index database (available at http://www.ancestry.com). Once these data were acquired for each case, each specimen was given a unique identification number that could not be traced to the original cytology specimen number. Thus, molecular analysis was performed in a blinded manner, without connection to patient identifiers. This protocol was reviewed and approved by the Oregon Health Sciences University Institutional Review Board. Cross-tabulations were analyzed with
the chi-square or Fisher exact test, where appropriate. Means were compared with the t test. Each of these tests was 2-tailed.
K-ras Gene Analysis Six unstained 5-µm slides were cut from cell blocks corresponding to each specimen. In 7 cases there were not enough cells present for subsequent analysis on some or all of the slides. These cases were excluded from the study, so 94 fine-needle aspirates remained to be studied. One of the 6 unstained slides was stained with H&E for comparison with the original cell-block slide on which a diagnosis had been made. From 1 to 3 of the unstained slides, depending on the number of cells on each slide, cells were scraped off and used for independent K-ras gene analyses. Two of the remaining slides were used for p53 and Dpc4 (Mad4) immunohistochemical analysis. The microdissected tissue for K-ras gene analysis was placed in 50-µL DNA isolation buffer (10-mmol/L concentration of tris(hydroxymethyl)aminomethane hydrochloride at pH = 8, 0.2% polysorbate 20, and 100 µg/mL of Proteinase K). The mixture was incubated at 56°C for 16 to 18 hours, and the Proteinase was inactivated at 95°C for 10 minutes. The mixture was stored at –20°C until use. The mixtures were screened for point mutations in codon 12 of the K-ras gene, according to a modification of a previously described protocol.3,41 First, DNA from each specimen was amplified by polymerase chain reaction (PCR) for 15 cycles with primers 1 (5′ ACT GAA TAT AAA CTT GTC GTA GTT GGA CCT 3′) and 2 (5′ TCA TGA AAA TGG TCA GAG AAA CC 3′). The PCR product was then split into 2 equal portions. One of the 2 portions was digested with MvaI (Boehringer-Mannheim, Mannheim, Germany). Both portions then were amplified with primers 1 and 3 (5′ TCA AAG AAT GGT CCT GGA CC 3′). Because MvaI cleaves wild-type but not mutant K-ras, an unenriched sample and a sample enriched for mutant K-ras were produced. Finally, each set of PCR products was denatured at 95°C for 10 minutes, spotted onto 7 different nylon membranes (GeneScreen Plus, NEN Research Products, Boston, MA), and hybridized to each of 7 phosphorus 32–labeled, sequencespecific oligodeoxynucleotide probes.42 A final stringency wash at 63°C and autoradiography were carried out. PCR products amplified from plasmid clones containing each of 7 possible sequences at codon 12 were used as positive controls on the hybridization filters. Negative control samples (water only) were included throughout the series.
p53 and Dpc4 (Mad4)Immunohistochemical Analysis Two unstained 5-µm slides were deparaffinized by routine techniques. The slides were treated with sodium citrate buffer (diluted to 1× from 10× HIER buffer, VentanaBio Tek Solutions, Tucson, AZ) and then steamed for 20
756
Am J Clin Pathol
2002;117:755-765 © American Society for Clinical Pathology
Anatomic Pathology / ORIGINAL ARTICLE
Am J Clin Pathol 2002;117:755-765 757© American Society for Clinical Pathology
minutes at 80°C. After cooling for 5 minutes, 1 slide was labeled with monoclonal antibody to p53 (antibody D0-7, 1:500 dilution, DAKO, Carpinteria, CA), and the second slide was labeled with monoclonal antibody to Dpc4 (Mad4) (clone B8, 1:100 dilution, Santa Cruz, Santa Cruz, CA). A Bio Tek-Mate 1000 automated stainer (Ventana-Bio Tek Solutions) was used to perform the labeling. The antibodies were detected by adding biotinylated secondary antibodies, avidin-biotin complex, and 3,3′-diaminobenzidine. Sections were counterstained with hematoxylin. Four of us (T.V.H., G.J.A.O., R.H.H., R.E.W.) evaluated the immunohistochemical labeling in each of the cases. The labeling in each case was scored as “positive” or “negative.” Because mutations in the p53 protein result in longer-lived proteins, positive labeling for p53 suggests a p53 alteration.26 Cases were scored as showing positive labeling for p53 when there was unequivocal and specific nuclear expression
of p53 in 1 or more cells. Because mutations in or deletions of DPC4 (MAD4) usually result in the truncation or absence of the protein, loss of labeling for Dpc4 (Mad4) indicates a DPC4 (MAD4) alteration.24 Cases scored as showing loss of labeling for Dpc4 (Mad4) had complete loss of labeling in both the cytoplasm and nuclei of cells, compared with other positive cells on the slide, which served as internal controls. Thus, because one looks for overexpression of protein in the p53 assay, all of the cells on a given slide were evaluated for p53. In contrast, in the Dpc4 assay, only cells with epithelial atypia, as compared with normal-appearing positive controls, were evaluated for Dpc4 loss. The interpretation of immunohistochemical labeling of the cancers was highly robust, with agreement among the observers in all cases. Sections of normal pancreas, which show negative p53 and positive Dpc4 (Mad4) staining, served as positive controls in each immunohistochemical run.
A B
C ?Image 1? Three representative fine-needle aspirates of the pancreas. The first (A) was diagnosed as “diagnostic of adenocarcinoma,” and the second (B) was “negative for malignancy.” The third aspiration (C) contains atypical cells that are suggestive, but not diagnostic, of malignancy. (Papanicolaou; A, ×600; B and C, ×400)
van Heek et al / MOLECULAR ALTERATIONS IN PANCREATIC NEEDLE ASPIRATES
Results
Cytology and Patient Characteristics
Based on the initial review of the cytologic material, 48 (51%) of the 94 fine-needle aspirates included in this study were diagnostic of adenocarcinoma, 19 (20%) were suggestive of adenocarcinoma, 25 (27%) were negative for malignancy, and 2 (2%) were diagnostic of a neoplasm other than adenocarcinoma ?Table 1?. One tumor in the latter category was a solid-pseudopapillary neoplasm; this patient was alive 11 months after her diagnosis was made. The other tumor was diagnosed on cytology as “consistent with either mucinous cystic neoplasm or intraductal papillary mucinous neoplasm”; the patient with this mucinous tumor died 4 months after his initial diagnosis. The 94 patients included in the study were followed up for median and mean periods of 15 and 16 months, respectively. The follow-up data were obtained by reviewing the patients’ medical records and verifying their survival status with the Social Security Death Index database. Clinical follow-up revealed that all of the 48 patients with malignant cytologic diagnoses had evidence of adenocarcinoma, while 3 (12%) of the 25 patients with negative cytologic diagnoses had adenocarcinoma. In the former group, 16 patients had histologic confirmation of their diagnoses by biopsy or Whipple resection, while 32 were shown to have adenocarcinoma by radiology (computed tomography scan, magnetic resonance imaging, arteriogram, or endoscopic ultrasound). In the latter group, 6 patients had their diagnoses confirmed by biopsy or Whipple resection, and 19 had them supported by radiology. The 19 patients with atypical fine-needle aspirates were followed up carefully. Twelve of these patients were shown to have evidence of malignancy. Three of these 12 patients underwent a Whipple resection (pancreaticoduodenectomy) for their disease. Of the 7 patients who had atypical aspirates and who never developed adenocarcinoma clinically, 4 underwent a resection or biopsy that showed either chronic pancreatitis (n = 3) or a serous cystadenoma (n = 1). Therefore, with clinical follow-up information added to the cytologic
diagnoses, 63 patients (67%) in this study had adenocarcinoma of the pancreas, by cytology and/or clinical follow-up (48 with positive cytologic diagnoses, 3 with false-negative cytologic diagnoses, and 12 with atypical cytologic diagnoses who developed adenocarcinoma), and 31 (33%) did not have adenocarcinoma of the pancreas. There were no significant differences in clinical characteristics among patients in each of the 4 cytologic groups (P = 1.000 for age; P = .193 for sex).
K-ras Mutations Twenty-eight (30%) of the 94 specimens harbored an activating point mutation at codon 12 of the K-ras gene. Only 3 of the 6 previously identified different activating mutations at codon 12 were seen among the 28 positive cases: 12 (43%) were GGT (Gly) to GTT (Val) mutations, 11 (39%) GGT to GAT (Asp) mutations, and 5 (18%) GGT to CGT (Arg) mutations. No activating point mutations to TGT (Cys), AGT (Ser), or GCT (Ala) were identified ?Image 2?. The presence of a K-ras mutation was analyzed with respect to original cytologic diagnosis. Twenty-four (50%) of 48 samples diagnostic of adenocarcinoma harbored K-ras mutations, while 3 (16%) of 19 samples suggestive of adenocarcinoma contained K-ras mutations. In contrast, only 1 negative cytologic sample harbored a K-ras mutation (GTT, Val). This negative sample with a K-ras mutation did not originate in 1 of the 3 patients with false-negative cytologic diagnoses. The 2 samples diagnosed as neoplasms other than adenocarcinoma did not have K-ras mutations (Table 1). The differences in prevalence of K-ras mutations among all of these diagnostic categories were highly statistically significant (P=.001). The 3 patients with atypical cytologic samples that also had K-ras mutations were studied. Clinical follow-up showed that each of these 3 patients had adenocarcinoma of the pancreas. The cytologic samples from 2 of the patients also showed alterations in the p53 and DPC4 (MAD4) genes, and these 2 patients died 3 and 4 months after their initial cytological diagnoses. The patient whose cytology sample showed a K-ras mutation but no other molecular change was a 70-year-old man who was alive 23 months after his initial
758
Am J Clin Pathol
2002;117:755-765 © American Society for Clinical Pathology
?Table 1? Correlation Between Cytologic and Molecular Diagnoses*
Molecular Diagnosis
Cytologic Diagnosis Mutant K-ras p53 Labeling Loss of Dpc4 Labeling
Diagnostic for adenocarcinoma (n = 48) 24 (50) 25 (52) 11 (23) Suggestive of adenocarcinoma (n = 19) 3 (16) 5 (26) 2 (11) Negative for malignancy (n = 25) 1 (4) 1 (4) 0 (0) Diagnostic for neoplasm, not adenocarcinoma (n = 2) 0 (0) 0 (0) 0 (0)
* Data are given as number (percentage).
Anatomic Pathology / ORIGINAL ARTICLE
diagnosis. Of the 16 patients with atypical aspirates that did not show K-ras mutations, 9 had adenocarcinoma of the pancreas on clinical follow-up. Therefore, compared with cytologic evaluation alone, a positive K-ras test identified 3 additional patients who had pancreatic adenocarcinoma. The absence of a mutation (ie, a wild-type K-ras gene) did not, however, exclude the diagnosis of cancer. Thus, 27 (43%) of 63 patients with pancreatic cancer on cytology or follow-up had a positive K-ras test result, while 30 (97%) of 31 patients without pancreatic adenocarcinoma by cytology and follow-up had a negative K-ras test result. These numbers reflect the sensitivity and specificity of K-ras mutations for the diagnosis of pancreatic ductal adenocarcinoma in fine-needle aspirates, respectively ?Table 2?, ?Table 3?, and ?Table 4?. The differences in prevalence of K-ras mutations between patients with and without final diagnoses of adenocarcinoma were highly statistically significant (P = .001). Table 4 summarizes the test characteristics of the K-ras assay.
p53Immunohistochemical Analysis Of the 94 cases, 31 (33%) showed nuclear labeling for p53 ?Image 3?. In 21 (68%) of these 31 cases, there were more than 50 cells showing nuclear labeling for p53. In 10 cases (32%), there were fewer than 50 positive cells within the cytology sample, with a range from 5 cells (2 cases) to 25 cells (1 case). The remaining cases contained 6 (2 cases), 10 (2 cases), 15 (1 case), 20 (1 case), and 24 (1 case) cells that labeled for p53. Twenty-five (52%) of the 48 cases that were cytologically diagnostic of adenocarcinoma labeled, as did 5 (26%) of the 19 atypical samples. In contrast, only 1 (4%) of the 25 negative samples stained for p53. The negative sample containing aberrant p53 expression did not originate in 1 of the 3 patients with false-negative cytologic diagnoses. The 2 neoplasms that were not adenocarcinomas were negative for p53. Table 1 summarizes the results of p53 immunohistochemical analysis on the fine-needle aspiration samples. The differences in prevalence of p53 immunoreactivity among all of these diagnostic categories were highly statistically significant (P=.001). The 5 atypical cytologic samples that also had p53 protein expression were analyzed. The cytologic samples from 2 of the 5 patients also showed alterations in the Kras and DPC4 (MAD4) genes. Clinical follow-up showed that each of the 5 patients with atypical cytology that harbored p53 overexpression had adenocarcinoma of the pancreas. Of the 14 patients with atypical aspirates that did not show p53 overexpression, 7 had adenocarcinoma of the pancreas on clinical follow-up. Therefore, compared with cytologic evaluation alone, a positive p53 test result identified 5 additional patients who had pancreatic
adenocarcinoma. Again, the absence of a detected p53 abnormality could not be used to exclude the diagnosis of carcinoma.
Am J Clin Pathol 2002;117:755-765 759© American Society for Clinical Pathology
WT Cys Ser Arg Val Asp Ala
1
2
3
4 5
6 7 8
9 10
11 12
13
?Image 2?An autoradiogram of Kras
mutational analysis in fineneedle aspirates of the pancreas. Seven nylon membranes, each hybridized with a different radioactively labeled oligonucleotide specific for the sequence of the wild-type codon 12 (left) and the 6 possible mutations. On each membrane in the left lane are the unenriched polymerase chain reaction (PCR) products; the right lane contains the mutant-enriched PCR products. A mutant specimen should create a weak signal in the unenriched and a strong signal in the enriched columns, because enrichment increases the proportion of mutant DNA. Row 1 contains the hybridization controls, DNA complementary to the labeled oligonucleotides. Rows 2 and 6 show samples that were negative for malignancy. Rows 3 and 5 contain samples diagnosed by cytology as suggestive, but not diagnostic, of malignancy. They harbor Gly (GGT) to Asp (GAT) and Val (GTT) mutations at codon 12 of the Kras gene, respectively. Rows 4, 7, 8, and 9 represent samples that were diagnostic of adenocarcinoma. Rows 10 and 11 are positive controls showing 1 cell with mutant codon 12, coding for the amino acid valine, mixed in 100 and 1,000 cells with wild-type codon 12, respectively. Row 12 is a contamination control sample (water only). Row 13 is an amplification control sample (placenta DNA). Ala, alanine; Arg, arginine; Asp, aspartic acid; Cys, cysteine; Ser, serine; Val, valine; WT, wild-type = glycine.
van Heek et al / MOLECULAR ALTERATIONS IN PANCREATIC NEEDLE ASPIRATES
760
Am J Clin Pathol
2002;117:755-765 © American Society for Clinical Pathology
?Table 4? Test Characteristics of Cytology and Molecular Markers in All Included Fine-Needle Aspirates (n = 94)*
Likelihood Ratio
Sensitivity Specificity PPV NPV Positive Negative Prevalence
Cytology 76 (68-85) 81 (73-89) 89 (83-95) 63 (53-72) 3.94 0.30 67 (58-77) Kras 43 (33-53) 97 (93-100) 96 (93-100) 45 (35-56) 13.29 0.59 67 (58-77) p53 48 (38-58) 97 (93-100) 97 (93-100) 48 (38-58) 14.76 0.54 67 (58-77) Dpc4 (Mad4) 21 (12-29) 100 100 38 (28-48) — — 67 (58-77) Entire panel, 1 mutation 67 (57-76) 94 (89-99) 95 (91-100) 58 (48-68) 10.33 0.36 67 (58-77) Cytology and entire panel 86 (79-93) 94 (89-99) 96 (93-100) 76 (68-85) 13.29 0.15 67 (58-77)
NPV, negative predictive value; PPV, positive predictive value. * Data are given as percentages except for the likelihood ratios. The 95% confidence intervals are given in parentheses. Prevalence is the proportion of patients with malignancy.
?Table 3? Correlation Between Final Clinical-Cytologic Diagnosis and Number of Positive Molecular Test Results*
None 1 2 3 Total
Adenocarcinoma 21 (33) 24 (38) 8 (13) 10 (16) 63 Negative for adenocarcinoma 29 (94) 2 (6) 0 (0) 0 (0) 31 Total 50 26 8 10 94
* Data are given as number (percentage).
?Table 2? Correlation Between Final Clinical-Cytologic Diagnosis and Molecular Diagnosis*
Molecular Diagnosis
Final Diagnosis Mutant K-ras p53 Labeling Loss of Dpc4 Labeling Entire Panel, 1 Mutation
Adenocarcinoma (n = 63) 27 (43) 30 (48) 13 (21) 42 (67) Negative for adenocarcinoma (n = 31) 1 (3) 1 (3) 0 (0) 2 (6)
* Data are given as number (percentage).
A B
?Image 3? Examples of p53 immunohistochemical assays. A, This specimen read as diagnostic of adenocarcinoma shows overexpression of the p53 protein. B, This specimen read as negative for malignancy does not show p53 overexpression. (×400)
Anatomic Pathology / ORIGINAL ARTICLE
Am J Clin Pathol 2002;117:755-765 761© American Society for Clinical Pathology
Thus, 30 (48%) of 63 patients with pancreatic cancer on cytology or follow-up had a positive p53 test result, while 30 (97%) of 31 patients without pancreatic adenocarcinoma by cytology and follow-up had a negative p53 test result. These numbers reflect the sensitivity and specificity of p53 immunohistochemical analysis for the diagnosis of pancreatic ductal adenocarcinoma in fine-needle aspirates, respectively (Tables 2-4). The differences in prevalence of p53 immunoreactivity between patients with and without a final diagnosis of adenocarcinoma were highly statistically significant (P = .001). Table 4 summarizes the test characteristics of the p53 assay.
Dpc4 (Mad4) Immunohistochemical Analysis Of the 94 fine-needle aspirates, 13 (14%) showed loss of Dpc4 (Mad4) expression ?Image 4?. Eleven (23%) of the 48 samples that were diagnostic of adenocarcinoma showed loss of Dpc4 (Mad4) expression, as did 2 (11%) of the 19 that were suggestive of adenocarcinoma. None of the samples showing either no evidence for malignancy or a neoplasm other than adenocarcinoma showed loss of Dpc4 (Mad4) (Table 1). The differences in prevalence of loss of Dpc4 labeling among all of these diagnostic categories were statistically significant (P = .020). The 2 atypical cytologic samples that also showed Dpc4 (Mad4) loss were from the 2 patients whose cytology also had K-ras and p53 alterations. As discussed in the “K-ras Mutations” section, these patients both had pancreatic adenocarcinomas; they died 3 and 4 months after their initial cytologic diagnoses. Of the 17 patients with atypical aspirates that did not show Dpc4 (Mad4) loss, 10 had adenocarcinoma of the pancreas on clinical follow-up. Thus, compared with
cytologic evaluation alone, a positive Dpc4 (Mad4) test result (actually a loss of Dpc4 [Mad4] expression) identified 2 additional patients who had pancreatic adenocarcinoma. Therefore, 13 (21%) of 63 patients with pancreatic cancer on cytology or follow-up had a positive Dpc4 (Mad4) test result, while 31 (100%) of 31 patients without pancreatic adenocarcinoma by cytology and follow-up had a negative Dpc4 (Mad4) test result. These numbers reflect the sensitivity and specificity of Dpc4 (Mad4) immunohistochemical analysis for the diagnosis of pancreatic ductal adenocarcinoma in fine-needle aspirates, respectively (Tables 2 and 4). The differences in prevalence of loss of Dpc4 labeling between patients with and without a final diagnosis of adenocarcinoma were statistically significant (P = .004). Table 4 summarizes the test characteristics of the Dpc4 (Mad4) assay.
Summary of Molecular and Cytologic Findings Each of the cytologic samples was evaluated with respect to the entire panel of molecular markers. Of the 48 aspirates cytologically diagnosed as diagnostic for malignancy, 12 showed no molecular alterations, 20 showed 1 molecular alteration, 8 showed 2 molecular alterations, and 8 showed 3 molecular alterations (ie, 36/48 samples [75%] showed at least 1 molecular alteration). Of the 19 aspirates that were suggestive of malignancy, 13 showed no molecular alterations, 4 showed 1 molecular alteration, none showed 2 molecular alterations, and 2 showed 3 molecular alterations. Twenty-three (92%) of the 25 negative aspirates contained no molecular alterations; 2 of these samples, however, harbored 1 molecular alteration each. Neither of these samples with a molecular alteration came from any of the 3 patients with false-negative cytologic diagnoses. Both of the samples recorded as diagnostic of
A B
?Image 4? Examples of Dpc4 immunohistochemical assays. A, This specimen read as negative for malignancy shows intact expression of the Dpc4 protein. B, This aspiration was diagnostic of adenocarcinoma; Dpc4 expression is lost. (×400)
van Heek et al / MOLECULAR ALTERATIONS IN PANCREATIC NEEDLE ASPIRATES
neoplasm other than adenocarcinoma were negative for all of the molecular markers. The 19 patients with atypical cytologic diagnoses were studied in more detail. Of the 13 patients with cytologic samples showing no molecular alterations, 6 had pancreatic adenocarcinoma on follow-up. Five of these 6 patients died, with an average survival time of 8.0 months; the remaining patient was alive 14 months after cytologic diagnosis. Of the 6 patients with at least 1 molecular alteration, all 6 had pancreatic adenocarcinoma. As discussed in the “K-ras Mutations” and “p53 Immunohistochemical Analysis” sections, 3 of 4 patients with 1 molecular alteration had p53 overexpression in their aspirates, while 1 had a K-ras mutation. The other 2 had alterations in each of the 3 molecular markers. Therefore, a panel of K-ras, p53, and DPC4 (MAD4) molecular alterations identified 6 additional patients who had pancreatic adenocarcinoma but whose cytologic specimens were not diagnostic of adenocarcinoma. The test characteristics of the molecular markers in the subgroup of samples with “suspicious” cytologic findings are summarized in ?Table 5?. Thus, considering the entire panel of tests together, 42 (67%) of 63 patients with pancreatic cancer on cytology and/or follow-up had at least 1 positive molecular test result, while 29 (94%) of 31 patients without pancreatic adenocarcinoma by cytology and follow-up did not have any positive molecular test results. This difference in the presence of at least 1 positive molecular test was highly statistically significant between these 2 groups (P = .001). These numbers also reflect the sensitivity and specificity of at least 1 positive molecular test result in diagnosing pancreatic ductal adenocarcinoma in fine-needle aspirates, respectively (Tables 2-4). Table 4summarizes the test characteristics of the 3 molecular assays.
Discussion Fine-needle aspiration is a useful technique in the diagnosis of pancreatic ductal adenocarcinoma. While most aspirates are diagnostic of a malignant or benign condition, a
substantial minority is atypical and cannot adequately rule in or rule out the presence of an adenocarcinoma. If ancillary techniques could help determine which of these atypical biopsy specimens were in actuality benign or malignant, patients would be saved additional procedures, which could result in decreased morbidity and mortality. In the present study, we took advantage of the fact that pancreatic ductal adenocarcinoma is a genetic disease. Because most adenocarcinomas harbor alterations in the Kras, p53, and DPC4 (MAD4) genes, the presence of such alterations could suggest the presence of an adenocarcinoma, even if a cytologic sample were only suggestive of malignancy.2-13 If molecular markers could identify or rule out malignancy in patients who did or did not have adenocarcinomas of the pancreas, additional tests would not be needed. The technique chosen for each of the molecular markers in this study—a dot-blot assay for the K-ras gene and immunohistochemical analysis for the p53 and Dpc4 (Mad4) proteins—was easy to perform and interpret.3,5,8,10,19-32,41-43 Molecular techniques already have been used on various samples, including cytologic ones, in making definitive diagnoses. For example, it can be very difficult to distinguish a benign bile duct proliferation, such as a bile duct adenoma, from a well-differentiated pancreatic adenocarcinoma metastasis in the liver on histologic grounds alone.44 However, virtually every liver metastasis of ductal adenocarcinoma harbors an activating point mutation in K-ras, but only a minority of benign bile duct proliferations do.44 Therefore, the presence of a K-ras mutation can help establish the diagnosis of pancreatic ductal adenocarcinoma metastatic to the liver. In addition, several groups have used K-ras and p53 mutational status to help in the diagnosis on cytologic specimens.19,31,33-40 However, no group has yet used a panel of molecular changes that has included the DPC4 (MAD4) gene to assess malignancy on cytologic specimens. This is an important omission because, unlike alterations in the K-ras and p53 genes, DPC4 (MAD4) mutations and deletions are relatively specific for pancreatic adenocarcinoma.11-13,45-50 In the present study, we showed that a panel of molecular-based tests can correlate with and supplement cytologic
762
Am J Clin Pathol
2002;117:755-765 © American Society for Clinical Pathology
?Table 5? Test Characteristics of Molecular Markers in the Subgroup of Samples With “Suspicious” Cytologic Findings (n = 19)*
Likelihood Ratio
Sensitivity Specificity PPV NPV Positive Negative Prevalence
Kras
25 (6-44) 100 100 44 (21-66) — — 63 (41-85) p53 42 (19-64) 100 100 50 (28-72) — — 63 (41-85) Dpc4 (Mad4) 17 (0-33) 100 100 41 (19-63) — — 63 (41-85) Entire panel, 1 mutation 50 (28-72) 100 100 54 (31-76) — — 63 (41-85)
NPV, negative predictive value; PPV, positive predictive value. * Data are given as percentages. The 95% confidence intervals are given in parentheses. Prevalence is the proportion of patients with malignancy.
Anatomic Pathology / ORIGINAL ARTICLE
diagnosis in fine-needle aspirates of the pancreas. Seventyfive percent of all fine-needle aspirates diagnostic of adenocarcinoma had at least 1 molecular alteration, while 23 (92%) of 25 fine-needle aspirates read as negative for malignancy contained none of the molecular alterations assayed in this study. In addition, in 6 of 12 aspirates diagnosed as atypical that came from patients who had evidence of a malignancy, at least 1 positive molecular marker was identified. None of the patients who had atypical aspirates but did not have evidence of malignancy on follow-up had a positive molecular marker. With respect to the 3 genes in this study, alterations in the p53 and K-ras genes seem to be more sensitive for detecting adenocarcinoma within fine-needle aspirates of the pancreas. Close to half of the cases diagnostic for adenocarcinoma contained a p53 or a K-ras alteration each. In contrast, only one fifth of the cases diagnostic of adenocarcinoma showed loss of the Dpc4 (Mad4) protein. The sensitivities for all 3 of these genes, however, are lower than expected, given the known rates of molecular alterations in pancreatic adenocarcinomas. The lower rates produced in this study are most likely due to sampling error. For example, nonneoplastic cells on the p53 or Dpc4 (Mad4) immunohistochemical slides may have obscured the labeling in rare cancer cells and, thereby, may have resulted in a falsenegative result. This is especially true with Dpc4 (Mad4) labeling, in which only 1 admixed cell that labeled for Dpc4 (Mad4) could obscure the loss of labeling in other cells. With respect to specificity, the Dpc4 (Mad4) assay is better than the K-ras or p53 assays. While the latter tests were each positive on 1 sample that was negative for malignancy, there was never a loss of the Dpc4 (Mad4) protein in a sample clinically diagnosed as negative for adenocarcinoma. These results are in keeping with the findings of previous molecular studies, which have shown that Dpc4 (Mad4) loss is never present in benign pancreatic tissue.24,25 Therefore, the strength of the Dpc4 (Mad4) labeling technique lies in its specificity, which is an important test characteristic for clinical practice. One problem with this study is the low number of patients who had histologic confirmation of their cancers. It is not in dispute that these patients indeed had cancer, because radiologic and clinical follow-up showed progression of disease in these cases. However, it is possible that these patients had cancers other than pancreatic ductal adenocarcinomas, for example, other primary pancreatic neoplasms, periampullary carcinomas, lymphomas, or metastases. These cancers are capable of mimicking pancreatic ductal adenocarcinoma both clinically and radiologically, and these neoplasms may have different molecular profiles from pancreatic adenocarcinoma. This may contribute to the low sensitivity of our molecular panel in
that we assumed that all patients with the clinical and radiologic appearance of pancreatic cancer had pancreatic ductal adenocarcinoma. It is important to note, however, that a screening test for pancreatic cancer would be most valuable if it could identify patients with any type of cancer involving the pancreas, since the Whipple procedure is the treatment for all of these neoplasms. It is therefore valid, and indeed preferable, to include all patients with cancer in the pancreas and periampullary region, whether pancreatic ductal adenocarcinoma or not, in calculating test characteristics for the molecular panel used in this study. Another problem with this study is the use of radiology as a surrogate for histologic confirmation, even in patients without cancer and even with only a 16-month follow-up. We believe that this is not particularly troublesome because the overwhelming majority of patients with pancreatic ductal adenocarcinoma will have had progression of their disease by 16 months of follow-up. Thus, patients who have pancreatic ductal adenocarcinoma but who have a falsenegative cytologic diagnosis most likely would have evidence of disease progression by 16 months of follow-up time. Therefore, radiology is a valid way to support the diagnosis of benignancy (or at least “negativity for adenocarcinoma”) in this particular situation. Indeed, it is “negativity for (ductal) adenocarcinoma” that is important in this study, since we presume that the molecular panel used in this study is targeted toward patients with ductal adenocarcinoma of the pancreas. Therefore, taken together, a panel of molecular markers is an excellent way to identify some patients who have adenocarcinoma of the pancreas but whose cytologic specimens are not diagnostic of malignancy. Each of the molecular tests in this panel is easy to perform and interpret. A molecular panel may clarify a diagnosis in patients who have atypical aspirates, and its use may lead to earlier and less invasive diagnosis. Therefore, a panel of molecular markers may be an excellent supplement to traditional cytology.
From the Departments of 1Pathology, Academic Medical Center, Amsterdam, the Netherlands; 2Pathology, Oregon Health Sciences University, Portland; and 3Pathology, 4Medicine, and 5Oncology, the Johns Hopkins Medical Institutions, Baltimore, MD.
Supported in part by grant CA62924 from the National Institutes of Health (Bethesda, MD) Specialized Program in Research Excellence (SPORE) in gastrointestinal cancer and by generous donations from the Michael Rolfe Fund for Pancreas Cancer Research. Address reprint requests to Dr Wilentz: Ross 632, 720 Rutland Ave, Johns Hopkins Medical Institutions, Baltimore, MD 21205. Acknowledgments: We thank Jennifer Galford for excellent work in helping prepare this manuscript and Mirjam Polak for technical research assistance.
Am J Clin Pathol 2002;117:755-765 763© American Society for Clinical Pathology
van Heek et al / MOLECULAR ALTERATIONS IN PANCREATIC NEEDLE ASPIRATES
References 1. Hruban RH, Petersen GM, Ha PK, et al. Genetics of pancreatic cancer: from genes to families. Surg Oncol Clin N Am. 1998;7:1-23. 2. Rozenblum E, Schutte M, Goggins M, et al. Tumorsuppressive pathways in pancreatic carcinoma. Cancer Res. 1997;57:1731-1734. 3. Hruban RH, van Mansfeld ADM, Offerhaus GJA, et al. K-ras oncogene activation in adenocarcinoma of the human pancreas: a study of 82 carcinomas using a combination of mutant-enriched polymerase chain reaction analysis and allele-specific oligonucleotide hybridization. Am J Pathol. 1993;143:545-554. 4. Redston MS, Caldas C, Seymour AB, et al. p53 mutations in pancreatic carcinoma and evidence of common involvement of homocopolymer tracts in DNA microdeletions. Cancer Res. 1994;54:3025-3033. 5. DiGiuseppe JA, Hruban RH, Goodman SN, et al. Overexpression of p53 protein in adenocarcinoma of the pancreas. Am J Clin Pathol. 1994;101:684-688. 6. Casey G, Yamanaka Y, Friess H, et al. p53 mutations are common in pancreatic cancer and are absent in chronic pancreatitis. Cancer Lett. 1993;69:151-160. 7. Scarpa A, Capelli P, Mukai K, et al. Pancreatic adenocarcinomas frequently show p53 gene mutations. Am J Pathol. 1993;142:1534-1543. 8. Weyrer K, Feichtinger H, Haun M, et al. p53, ki-ras, and DNA ploidy in human pancreatic ductal adenocarcinomas. Lab Invest. 1996;74:279-289. 9. Sugano K, Nakashima Y, Yamaguchi K, et al. Sensitive detection of loss of heterozygosity in the TP53 gene in pancreatic adenocarcinoma by fluorescence-based single strand conformation polymorphism analysis using blunt end DNA fragments. Genes Chromosomes Cancer. 1996;15:157-164. 10. Lundin J, Nordling S, von Boguslawsky K, et al. Prognostic value of immunohistochemical expression of p53 in patients with pancreatic cancer. Oncology. 1996;53:104-111. 11. Schutte M, Hruban RH, Hedrick L, et al. DPC4 gene in various tumor types. Cancer Res. 1996;56:2527-2530. 12. Hahn SA, Schutte M, Hoque ATMS, et al. DPC4, a candidate tumor suppressor gene at human chromosome 18q21.1. Science. 1996;271:350-353. 13. Hahn SA, Hoque ATMS, Moskaluk CA, et al. Homozygous deletion map at 18q21.1 in pancreatic cancer. Cancer Res. 1996;56:490-494. 14. Smit VTHBM, Boot AJM, Smits AMM, et al. K-ras codon 12 mutations occur very frequently in pancreatic adenocarcinomas. Nucleic Acids Res. 1988;16:7773-7782. 15. Mariyama M, Kishi K, Nakamura K, et al. Frequency and types of point mutation at the 12th codon of the c-Ki-ras gene found in pancreatic cancers from Japanese patients. Jpn J Cancer Res. 1989;80:622-626. 16. Tada M, Yokosuka O, Omata M, et al. Analysis of ras gene mutations in biliary and pancreatic tumors by polymerase chain reaction and direct sequencing. Cancer. 1990;66:930-935. 17. Motojima K, Urano T, Nagata Y, et al. Detection of point mutations in the Kirsten-ras oncogene provides evidence for the multicentricity of pancreatic carcinoma. Ann Surg. 1993;217:138-143. 18. Pellegata NS, Sessa F, Renault B, et al. K-ras and p53 gene mutations in pancreatic cancer: ductal and nonductal tumors progress through different genetic lesions. Cancer Res. 1994;54:1556-1560.
19. van Es JM, Polak MM, van den Berg FM, et al. Molecular markers for diagnostic cytology of neoplasms in the head region of the pancreas: mutation of K-ras and overexpression of the p53 protein product. J Clin Pathol. 1995;48:218-222. 20. Connolly MM, Dawson PJ, Michelassi F, et al. Survival in 1001 patients with carcinoma of the pancreas. Ann Surg. 1987;206:366-373. 21. van den Berg FM, Baas IO, Polak MM, et al. Detection of p53 overexpression in routinely paraffin-embedded tissue of human carcinomas using a novel target unmasking fluid. Am J Pathol. 1993;142:381-385. 22. Zhang SY, Ruggeri B, Agarwal P, et al. Immunohistochemical analysis of p53 expression in human pancreatic carcinomas. Arch Pathol Lab Med. 1994;118:150-154. 23. Iacobuzio-Donahue CA, Wilentz RE, Argani P, et al. Dpc4 protein in mucinous cystic neoplasms of the pancreas: frequent loss of expression in invasive carcinomas suggests a role in genetic progression. Am J Pathol. 2000;24:1544-1548. 24. Wilentz RE, Su GH, Dai JL, et al. Immunohistochemical labeling for Dpc4 mirrors genetic status in pancreatic adenocarcinomas: a new marker of DPC4 inactivation. Am J Pathol. 2000;156:37-43. 25. Wilentz RE, Iacobuzio-Donahue CA, Argani P, et al. Loss of expression of Dpc4 in pancreatic intraepithelial neoplasia: evidence that DPC4 inactivation occurs late in neoplastic progression. Cancer Res. 2000;60:2002-2006. 26. Baas IO, Mulder JWR, Offerhaus GJA, et al. An evaluation of six antibodies for immunohistochemistry of mutant p53 gene product in archival colorectal neoplasms. J Pathol. 1994;172:5-12. 27. Boschman C, Reddy JK, Rao MS. Immunohistochemical analysis of p53 protein expression in human pancreatic ductal adenocarcinomas and precursor lesions [abstract]. Lab Invest. 1994;70:129A. 28. Chang K, Ding, I, Kern FG, et al. Immunohistochemical analysis of p53 and HER-2/neu proteins in human tumors. J Histochem Cytochem. 1991;39:1281-1287. 29. Bartek J, Bartkova J, Vojtesek B, et al. Aberrant expression of the p53 oncoprotein is a common feature of a wide spectrum of human malignancies. Oncogene. 1991;6:1699-1703. 30. Boschman CR, Stryker S, Reddy JK, et al. Expression of p53 protein in precursor lesions and adenocarcinoma of human pancreas. Am J Pathol. 1994;145:1291-1295. 31. Dowell SP, Wilson POG, Derias NW, et al. Clinical utility of the immunocytochemical detection of p53 protein in cytological specimens. Cancer Res. 1994;54:2914-2918. 32. Hall PA, Lane DP. p53 in tumour pathology: can we trust immunohistochemistry? revisited [editorial]! J Pathol. 1994;172:1-4. 33. Sturm PD, Rauws EA, Hruban RH, et al. Clinical value of Kras codon 12 analysis and endobiliary brush cytology for the diagnosis of malignant extrahepatic bile duct stenosis. Clin Cancer Res. 1999;5:629-635. 34. Fukushima N, Suzuki M, Fukayama M. Analysis of K-ras oncogene mutation directly applied to atypical cell clusters on cytologic smear slides of bile and pancreatic juice. Pathol Int. 1998;48:33-40. 35. Iwao T, Itoh M, Tsuchida A, et al. Detection of p53 protein in cytological specimens of pancreatic diseases obtained by selective endoscopic brushing [abstract]. Gastroenterology. 1995;108:a361. 36. Dillon DA, Johnson CC, Topazian MD, et al. The utility of Ki-ras mutation analysis in the cytologic diagnosis of pancreatobiliary neoplasms. Cancer J. 2000;6:294-301.
764
Am J Clin Pathol
2002;117:755-765 © American Society for Clinical Pathology
Anatomic Pathology / ORIGINAL ARTICLE
37. Sturm PDJ, Hruban RH, Ramsoekh TB, et al. The potential diagnostic use of k-ras codon 12 and p53 alterations in brush cytology from the pancreatic head region. J Pathol. 1998;186:247-253. 38. Tascilar M, Sturm PDJ, Caspers E, et al. Diagnostic p53 immunostaining of endobiliary brush cytology: preoperative cytology compared with the surgical specimen. Cancer. 1999;87:306-311. 39. Apple SK, Hecht JR, Novak JM, et al. Polymerase chain reaction–based K-ras mutation detection of pancreatic adenocarcinoma in routine cytology smears. Am J Clin Pathol. 1996;105:321-326. 40. Uehara H, Nakaizumi A, Baba M, et al. Diagnosis of pancreatic cancer by K-ras point mutation and cytology of pancreatic juice. Am J Gastroenterol. 1996;91:1616-1621. 41. Chung CH, Wilentz RE, Polak MM, et al. Clinical significance of K-ras oncogene activation in ampullary neoplasms. J Clin Pathol. 1996;49:460-464. 42. Slebos RJC, Boerrigter L, Evers SG, et al. A rapid and simple procedure for the routine detection of ras point mutations in formalin-fixed paraffin-embedded tissues. Diagn Mol Pathol. 1992;1:136-141. 43. Wilentz RE, Chung CH, Sturm PDJ, et al. K-ras mutations in the duodenal fluid of patients with pancreatic carcinoma. Cancer. 1998;82:96-103.
44. Hruban RH, Sturm PDJ, Slebos RJC, et al. Can K-ras codon 12 mutations be used to distinguish benign bile duct proliferations from metastases in the liver? a molecular analysis of 101 liver lesions from 93 patients. Am J Pathol. 1997;151:943-949. 45. Hoque AT, Hahn SA, Schutte M, et al. DPC4 gene mutation in colitis associated neoplasia. Gut. 1997;40:120-122. 46. Hahn SA, Bartsch D, Schroers A, et al. Mutations of the DPC4/Smad4 gene in biliary tract carcinoma. Cancer Res. 1998;58:1124-1126. 47. Bartsch D, Hahn SA, Danichevski KD, et al. Mutations of the DPC4/Smad4 gene in neuroendocrine pancreatic tumors. Oncogene. 1999;18:2367-2371. 48. Thiagalingam S, Lengauer C, Leach FS, et al. Evaluation of candidate tumor suppressor genes on chromosome 18 in colorectal cancers. Nat Genet. 1996;13:343-346. 49. Takagi Y, Kohmura H, Futamura M, et al. Somatic alterations of the DPC4 gene in human colorectal cancers in vivo. Gastroenterology. 1996;111:1369-1372. 50. Schutte M. DPC4/SMAD4 gene alterations in human cancer, and their functional implications. Ann Oncol. 1999;10(suppl 4):56-59.
Am J Clin Pathol 2002;117:755-765 765© American Society for Clinical Pathology

PLACE THIS ORDER OR A SIMILAR ORDER WITH US TODAY AND GET AN AMAZING DISCOUNT