Bile duct brushing is the procedure of choice for the assessment of neoplasia of the biliary and pancreatic ducts. Conventional cytopathologic evaluation has been reported to have high specificity but relatively low sensitivity. Although a number of molecular studies regarding biliary tract tissue specimens have been performed, to the authors' knowledge their precise applicability to cytopathology specimens has not been critically analyzed.
Bile duct brushing specimens with the cytopathologic diagnosis of “suspicious” or “positive for malignant cells” along with corresponding surgical pathology specimens demonstrating adenocarcinoma were searched for in the files of UPMC-Presbyterian Hospital for the years 1990–1996. Tumor cells from representative cytopathology and histology slides were microdissected and analyzed for loss of heterozygosity (LOH) in a panel of microsatellite markers. The results obtained from cytopathologic and surgical pathology specimens were compared.
Eight paired surgical and cytopathology cases of adenocarcinoma involving the biliary tract were identified. The fractional allelic loss (FAL) for the surgical specimens (FAL-S) ranged from 12.5–71.4% and the FAL for the cytopathology specimens (FAL-C) ranged from 25–71.4%. However, when evaluating the actual loci of LOH, the concordance rate of the surgical and cytopathology specimens ranged from 71.4–100% (mean, 88.6%). Only 3 of the 8 cases (37.5%) were found to have identical matching of the LOH loci.
Bile duct brushing is recognized as the procedure of choice for the evaluation of biliary and pancreatic neoplasia. Using conventional cytopathologic analysis, this procedure has high specificity but low sensitivity.1, 2 Ancillary studies, particularly molecular analysis, hold promise in improving the diagnostic efficiency of the procedure. However, there are a number of issues that need to be addressed prior to the implementation of such ancillary testing into diagnostic practice. One issue involves the correlation of molecular results from cytopathology and surgical pathology specimens. To our knowledge, comparative studies between cytologic and surgical pathology specimens have not been performed to date. Of the various molecular analyses available, the determination of the frequency of allelic loss (FAL) (loss of heterozygosity [LOH]) at various genomic sites provides a molecular profile for a neoplasm. In carcinogenesis, the first step is usually a point mutation of one copy of the gene, although other mechanisms such as DNA methylation or gene deletion may also result in first-copy loss or inactivation. The second step is the genomic deletion of the residual second functioning copy of the gene as well as genetic material situated in proximity to the gene itself. The microsatellite markers of interest are situated in proximity to known human tumor suppressor genes (cutaneous malignant melanoma [CMM], von Hippel Lindau [VHL], adenomatous polyposis coli [APC], mutated in colorectal cancer [MCC], cyclin dependent kinase 2A [CDKN2A], phosphate and tensin analog [PTEN], max interacting protein [MXI 1], and tumor protein p53 [TP53]). A broad panel of microsatellite markers is used to define the unique profile of an individual tumor, given that each neoplasm may develop its own palette of somatically acquired mutational alterations. The analysis of k-ras-2 oncogene point mutation is also of interest because it is a frequent event in the development of pancreatic ductal carcinoma and progression. The objective of the current study was to compare the frequency and patterns of allelic loss in a panel of microsatellite markers and k-ras-2 mutation in matched cytologic and surgical pathology specimens.
MATERIALS AND METHODS
The archives of the Department of Pathlogy of the University of Pittsburgh Medical Center-Presbyterian Hospital for the years 1990–1996 were searched for bile duct brushing cases with the cytopathologic diagnosis of “positive for malignant cells” or “suspicious for malignant cells.” The cytopathologic criteria for malignancy included nuclear enlargement, pleomorphism (minimum of threefold variation in nuclear size), an elevated nuclear:cytoplasmic (N:C) ratio, nuclear membrane irregularity, and coarse chromatin.3 Cases diagnosed as “suspicious” fulfilled most but not all criteria for malignancy or were noted in association with a confounding factor, such as infection. The study was reviewed and approved by the University of Pittsburgh Medical Center Institutional Review Board. Of the cases identified, eight had corresponding surgical pathology specimens demonstrating bile duct adenocarcinoma.
Representative cytopathology and surgical pathology slides were selected for microdissection and analyzed for loss of heterozygosity (LOH) in a panel of selected microsatellite markers. For cytology specimens, multiple clusters of neoplastic cells (approximately 1000–2000) were identified and marked on representative alcohol-fixed, Papanicolaou-stained slides (Fig. 1). The areas of interest were microdissected manually from the slides and placed in 25 μL of dilute Tris buffer (pH 7.5). For slides generated from formalin-fixed, paraffin-embedded surgical pathology material, normal biliary tissue (negative control) and neoplastic areas were microdissected manually (Fig. 2). The method for formalin-fixed, paraffin-embedded tissue was described previously.4 The normal tissue samples were run to determine microsatellite informativeness. For cytologic and surgical pathology specimens, all analyses were performed from a single extraction.
One-microliter aliquots of microdissected material were used in the polymerase chain reaction (PCR) amplification for a broad panel of 12 microsatellite markers commonly involved in human pancreatic and biliary carcinogenesis.5, 6 The microsatellite markers analyzed were associated with tumor suppressor genes, including CMM, myelocytomatosis viral oncogene homolog1 [MYCL], VHL, APC, MCC, CDKN2A, PTEN, MXI 1, and TP53. PCR amplification was designed to generate an amplicon of < 200 base pairs (bp) long using synthetic oligonucleotide primers flanking each microsatellite. Oligonucleotide primers were created with 5′ fluorescent moieties (FAM, HEX, and NED) suitable for automated fragment analysis. The PCR products were analyzed by capillary electrophoresis with the ABI 3100 according to the manufacturer's instructions (Applied Biosystems, Foster City, CA). Briefly, this system provided a quantitative analysis of mutational damage by using automated fluorescence-based systems of mutation detection. To quantitatively measure allelic loss, fluorescent-labeled PCR products were passed through a capillary system and measured by laser detection in the ABI 3100. The system was highly sensitive in detecting amplicon size to within 1 bp and relative fluorescent allele amount to within a tolerance of < 1% variability. Allelic imbalance was determined automatically by formulation of the ratio of the individual allele fluorescent content. The analysis was performed in replicate fashion whenever possible, providing a measure for internal quality-control assessment.
Allele peak heights and lengths were used to define the presence or absence of allelic imbalance (LOH) for a given sample. Allelic imbalance was reported when the ratio of polymorphic allelic bands for a particular marker was below 0.5 or above 2.7 This ratio provided an acceptable sensitivity of detection if ≥ 50% of the microdissected cells were altered. In addition, the deleted allele was designated as either “B” or “T” depending on whether the shorter (bottom, longer lengthened) or the longer (top, shorter lengthened) allele was relatively diminished. This knowlege of the specific allele was important because the presence of the identical deleted allele in different microdissection targets supported the existence of the same deletion in all affected target sites. Similarly, it was possible to identify two separate mutations of the same genomic region in different topographic tissue samples when deleted alleles were shown to be discordant. A locus was determined to be concordant when the same allele was lost (for example, LOH B in the cytology specimen and in the surgical specimen). For each cytology and surgical pathology case, the FAL was determined to be the ratio of microsatellite loci affected by LOH to the total number of informative loci. The concordance rate for each patient sample was calculated as the number of loci with an exact match between the cytologic and surgical specimen divided by the total number of informative loci. For each microsatellite marker, the frequency of LOH was calculated for cytology and surgical specimens. The frequency of concordance for each microsatellite marker was defined as the number of patients with loci demonstrating the same mutation in the paired cytology and surgical specimens divided by the total number of informative patients.
The patients ranged in age from 51–83 years (mean, 67.6 years); 3 patients were male and 5 patients were female. Of the total of eight cases, five were bile duct cholangiocarcinomas and three were pancreatic ductal carcinomas. The grade of the adenocarcinoma was “moderately differentiated” in seven patients and “poorly differentiated” in one patient. The bile duct brushing samples were diagnosed as “suspicious for malignant cells” in five cases and “positive for malignant cells” in three cases. The five “suspicious” cases were lacking in one of the five criteria (nuclear enlargement, nuclear pleomorphism, an elevated N:C ratio, nuclear membrane irregularity, and coarse chromatin pattern) or were noted in association with a confounding factor. In three of the suspicious cases, all criteria except nuclear pleomorphism were fulfilled. In one case, there was a lack of nuclear pleomorphism and coarse chromatin. The remaining case demonstrated all cytologic features of malignancy; however, there was heavy infestation by Candida species, which precluded a definitive diagnosis. Five of the eight patients had histologic evidence of metastatic disease.
Mutation and LOH Results
The microsatellite marker results for each pair of cytology and surgical pathology specimens were compared (Table 1). The FAL for cytology specimens ranged from 25–71.4%; the FAL for surgical specimens ranged from 12.5–71.4%. The concordance rate was 71.4–100% (mean, 88.6%). The frequency of mutation/LOH for each microsatellite marker by type of specimen is detailed in Table 2. Of the 13 markers studied, 5 markers had the same frequency of mutation/LOH in the paired cytology and surgical specimens. In five markers, the frequency of mutation/LOH in cytology specimens was greater than that in the corresponding surgical specimens, and in three markers, the frequency was greater in the surgical specimens.
Table 1. Comparison of Mutation and LOH in cytology and Surgical Specimens
Case type, tumor type
D1S 407 1p36 CMM
MYC L 1p34
D3S 2303 3p25
D3S 1539 3p26 VHL
D5S 592 5q25 APC
D5S 615 5q25 MCC
D9S 254 9p21 CDK
D9S 251 9p21
D10S52 0 10q23 PTEN
D10S1 173 10q23 MXI1
D17S9 74 17p13
D17S1 289 17p13 TP53
Concordance rate (%)
C: cytopathology specimen; S: surgical pathology specimen; MUT: mutation; LOH: loss of heterozygosity; LOH T: top allele; LOH B: (bottom) allele; NI: not informative.
Table 2. Informativeness Rate and Frequency of Mutations or LOH for Cytological and Surgical Specimens
Informative rate (%) (no.)
Frequency of mutation or LOH in cytological specimens (%) (n)
Frequency of mutation or LOH in surgical Specimens (%) (n)
Frequency of concordance between cytological and surgical specimens (%) (n)
LOH: loss of heterozygosity.
The results of the current study provide a realistic view of the complexity of biliary neoplasia and underscore appreciation for the concept of intratumoral heterogeneity in the application of molecular tests to cytology specimens. Although the mean concordance rate of 88.6% between the cytologic and surgical specimens suggests relatively good agreement, the converse indicates that 11% of the microsatellite markers are discordant.
Why is there discordance, and what is its significance? Currently, there are two leading theories for intratumoral molecular heterogeneity. The first is the development of independent clones by the field cancerization theory; the second is the emergence of divergent clones from a common monoclonal population of neoplastic cells. Both theories appear to be valid, depending on the type of tumor. Certain neoplasms have the propensity to develop multifocality through various pathways. For example, the first theory appears to be applicable to the majority of multifocal prostate carcinomas, whereinthe pattern of allelic loss is consistent with the development of independent clones.8 Nonetheless, in a minority of prostate carcinomas, the multifocal tumors are clonal and consistent with intraglandular dissemination. The second theory is applicable to adenocarcinoma arising in patients with Barrett esophagus and also in those with mammary ductal carcinoma.9, 10 In these examples, based on the pattern of LOH, genetic heterogeneity was attributed to clonal diversity within a monoclonal neoplasm. Similarly, in cases with some degree of divergence between the cytology and surgical samples, the remaining concordant loci harboring LOH exhibited the same allelic loss pattern. For example, in Case 8, there were discordant results at D3S1539 and D9S251. There were four additional loci with concordant results (k-ras, MYCL, D3S2303, and D17S1289). At all of these concordant loci, the precise mutation or allele lost was identical. The same observations were made for Cases 4, 5, 6, and 7 (Table 1). These results support the theory of the emergence of divergent clones from a common monoclonal population of neoplastic cells. In addition, biliary and pancreatic neoplasms often are discovered late; therefore, cumulative genetic changes may result in clonal diversity within a monoclonal tumor.11, 12 Because the biliary brushing procedure preferentially samples the ductal epithelium, the specimens may contain in situ and superficially invasive carcinoma cells but not the cellular population present in the more deeply invasive areas of the neoplasm with divergent and complex clones.
The issue of discordance in molecular profile would have implications not only in samples obtained from the same tumor but also when comparing the primary neoplasm with metastatic deposits. A study by Sasatomi et al. found substantial mutational heterogeneity between the primary neoplasm and the metastatic sites.4 In this study of American Joint Committee on Cancer (AJCC) Stage II non-small-cell carcinomas of the lung, the FAL was found to be significantly less in the metastatic site compared with the primary neoplasm. The authors theorized that this phenomenon was the result of early metastatic spread of the carcinoma with the primary neoplasm acquiring additional genetic changes. This concept should be kept in mind when comparing the molecular profile of the primary neoplasms with that of the metastatic or recurrent sites. However, to our knowledge it is yet to be determined whether similar molecular events are present in biliary carcinoma.
Further investigation would reveal the preferred areas to sample, the methods of sampling, the number of samples necessary to generate a representative molecular profile, and the degree of concordance required to determine whether the samples are derived from a common source. The current study has shown that molecular profiling from biliary cytology specimens yields a detailed pattern of mutational change, but the results are influenced by sampling. The degree of discordance may be attributed to intratumoral heterogeneity arising from divergent clones developing in monoclonal neoplasms.