DNA flow cytometric analysis of paraffin-embedded tissue for the diagnosis of malignancy in bile duct biopsies

Summary

Differentiation of reactive versus neoplastic epithelial changes can be challenging in bile duct biopsies. The samples are often scant, distorted, and mixed with significant inflammation, ulceration, and/or debris. Histological confirmation of malignancy is often required before the initiation of surgical therapy, and an erroneous diagnosis of malignancy can lead to unnecessary clinical management. Aneuploidy assessment by DNA flow cytometry was performed on formalin-fixed paraffin-embedded (FFPE) tissue from 63 bile duct biopsies: 10 with a malignant diagnosis (7 with adenocarcinoma and 3 with at least high-grade dysplasia [HGD]); 3 with an atypical diagnosis showing rare atypical glands/cells, concerning but not definite for malignancy; 28 likely reactive biopsies with acute/chronic inflammation, ulceration, and/or mild nuclear atypia; and 22 additional benign biopsies without significant inflammation, ulceration, or nuclear atypia. Aneuploidy was detected in 7 (70%) of the 10 biopsies with definite neoplasia (5 of 7 adenocarcinoma cases and 2 of 3 at least HGD cases), all 3 (100%) atypical biopsies, and none of the 50 benign biopsies. All 3 atypical cases with aneuploidy were subsequently found to have adenocarcinoma (n = 2) or HGD (n = 1). Among the 2 cases of at least HGD with aneuploidy, 1 case developed adenocarcinoma, but no follow-up information was available in the other case. The remaining 1 case of at least HGD, despite having normal DNA content, was found to have adenocarcinoma on follow-up. None of the 50 benign cases (further supported by normal DNA content) developed adenocarcinoma within a mean follow-up time of 37 months (range: 0–282 months). The estimated sensitivity of aneuploidy as a diagnostic marker of malignancy (adenocarcinoma and HGD) was 70%, with the specificity of 100%, positive predictive value of 100%, and negative predictive value of 94%. In conclusion, DNA flow cytometry using FFPE tissue from bile duct biopsies demonstrates a high rate of aneuploidy (70%) in malignant cases and normal DNA content in all benign biopsies. Although the sample size is small, the results indicate that this assay can be potentially useful in challenging atypical cases, where morphological evaluation is limited by scarcity of atypical glands/cells, inflammation, and/or ulceration.

1. Introduction

Distinction between benign and malignant bile duct samples can be challenging on histological grounds. Bile duct specimens are often characterized by scant cellularity with a few or no malignant cells, debris, and crush artifact, as well as significant inflammation, ulceration, and/or severe reactive changes that can mimic a malignant process. Approximately 10% of patients undergoing surgery for suspected pancreaticobiliary malignancy are known to have benign disease [1,2], and thus, histological confirmation of malignancy is crucial before considering aggressive surgical therapy. Routinely performed procedures for the diagnosis of bile duct lesions such as brush cytology and biopsy have high specificity (99%) but are limited by low sensitivity (45% and 48%, respectively) [3,4]. Although brush cytology is easier to perform and is the preferred initial method [4], intraductal biopsy remains an important diagnostic tool as it allows for deeper samples and provides additional information about tissue architecture. This is crucial because well differentiated pancreaticobiliary tumors can mimic benign epithelium and glands, especially in the superficial samples obtained by brushing.

To improve the diagnostic yield of different sampling techniques, several ancillary tools have been developed and include DNA flow cytometry [5–7], digital image analysis [8–12], KRAS mutational testing [13,14], fluorescence in situ hybridization (FISH) [15–21], and more recently, next-generation sequencing [22–24]. Although these tests have been shown to improve the diagnostic sensitivity in detecting malignancy over cytology alone, they are usually performed on brushing samples and often require a separate specimen for analysis. This protocol may explain some discrepant results between cytological evaluation and ancillary testing. Yet surgical pathologists continue to encounter intraductal biopsies, and there is no objective ancillary testing that allows for morphological correlation to aid in the distinction between reactive and neoplastic epithelial changes. Although the diagnostic utility of immunohistochemical staining, such as p53, S100P, and mapsin staining, has been studied [25–27], the interpretation of staining results is highly variable, and non-neoplastic epithelial cells often show positive staining.

In this regard, we previously demonstrated that DNA flow cytometric analysis of formalin-fixed paraffin-embedded (FFPE) tissue from small mucosal biopsies can establish a diagnosis of neoplasia in various settings, including neoplastic lesions in the ampulla [28], Barrett’s esophagus [29], inflammatory bowel disease [30–32], and gastric neoplasia [33]. This diagnostic modality can be very useful for surgical pathologists as it allows direct histology-flow cytometry correlation while avoiding obtaining separate samples for flow cytometry and histology (thus permitting easier specimen handling/workflow and mailing in samples to reference laboratories if needed). The purpose of this study was to use DNA flow cytometry to evaluate the DNA content of pancreaticobiliary tumors using FFPE bile duct biopsies and to determine whether DNA flow cytometry can be helpful in differentiating between benign/reactive and malignant bile duct biopsies.

2. Materials and methods

2.1. Patients and data collection

After approval by the institutional review board for human subject research (IRB no. 16–21034), the pathology database was searched from 1990 to 2019 to identify 77 patients who underwent bile duct biopsies. Fourteen cases were excluded owing to the lack of hematoxylin and eosin slides, tissue blocks, residual tissue on the blocks, or cells of interest on the recut slides. As a result, a total of 63 bile duct biopsies from 63 patients were retrieved from the files of the Department of Pathology at the University of California, San Francisco (UCSF), for further analysis. All cases were re-reviewed by at least one gastrointestinal pathologist (W.-T.C. and S.K.) and placed in one of the following five diagnostic categories: negative/benign (n = 22), likely reactive (n = 28), adenocarcinoma (n = 7), at least high-grade dysplasia (HGD, n = 3), or atypical (n = 3). Negative/benign biopsies showed fragment(s) of benign biliary epithelium without significant inflammation, ulceration, or nuclear atypia (Fig. 1A), whereas likely reactive samples demonstrated acute/chronic inflammation, ulceration, and/or mild nuclear atypia (Fig. 1C). Some of the histological changes seen in the reactive cases may raise consideration for possible dysplasia, but overall, they were thought to represent reactive/regenerative/inflammatory atypia, especially in the setting of inflammation and/or ulceration. At least HGD cases illustrated severe cytological (i.e., enlarged, rounded nuclei, pleomorphism, and loss of nuclear polarity) and/or architectural atypia (such as cribriform growth), but there was no definite evidence of invasive tumor cells (Fig. 2). Atypical cases were characterized by highly atypical but rare glands/cells exhibiting some, if not all, of the features of malignancy, but because of a limited number of these cells, they could not be definitely classified as malignant (Fig. 3). Demographic, clinical, and endoscopic details were recorded for each case, including age, gender, race, and presence of primary sclerosing cholangitis (PSC), as well as the location and endoscopic appearance of each biopsy site. A diagnosis of malignant or benign disease was established based on follow-up biopsies, brush cytology, surgical resection, and/or clinical follow-up.

Fig. 1.

A and B, The biopsy shows a fragment of benign biliary epithelium without significant inflammation, ulceration, or nuclear atypia (H&E, 10×). The corresponding DNA histogram shows a normal diploid population (green). C and D, Another biopsy shows hyperchromatic and focally crowded nuclei, raising consideration for possible dysplasia, but the presence of retained basal polarity and neutrophilic inflammation suggest reactive changes (H&E, 20×). A normal diploid population (green) with some apoptotic debris (pink) is seen on the DNA histogram. H&E, hematoxylin and eosin.

Fig. 2.

A and B, The biopsy shows adenocarcinoma (H&E, 10×). The corresponding DNA histogram demonstrates an aneuploid population (red) distinct from a normal diploid peak (green). C and D, Another biopsy shows dysplastic glands with a focus of high-grade nuclear features characterized by enlarged, rounded nuclei and loss of nuclear polarity (H&E, 20×). This was interpreted as at least HGD. The DNA histogram shows a distinct aneuploid population (red) in addition to a normal diploid peak (green) and apoptotic debris (pink). The follow-up biopsy showed extrahepatic cholangiocarcinoma. H&E, hematoxylin and eosin; HGD, high-grade dysplasia.

Fig. 3.

A and B, The biopsy shows a rare atypical, disrupted gland (arrow) associated with mixed inflammation, ulceration, and hemorrhage. Panel A: H&E, 4×; panel B: H&E, 20×. C, In addition to a normal diploid population (green) and large apoptotic debris (pink), a distinct aneuploid population (red) is identified on this DNA histogram. The follow-up biopsy showed extrahepatic cholangiocarcinoma. D and E, Another atypical biopsy shows a focus of scattered atypical cells (arrow) with enlarged, hyperchromatic, and irregular nuclei in a background of hemorrhage and inflammation. Panel D: H&E, 4×; panel E: H&E, 20×. This was interpreted as atypical, suspicious for carcinoma. F, The corresponding DNA histogram demonstrates a distinct aneuploid population (red). The patient was found to have pancreaticobiliary adenocarcinoma on follow-up. H&E, hematoxylin and eosin.

2.2. DNA flow cytometry

As previously described [28–34], three to six 60-micron-thick sections were cut from each tissue block and submitted entirely (without manually dissecting the area of interest) for DNA content analysis owing to scant nature of these biopsies. Each sample was stained with 4,6-diamidino-2-phenylindole (Accurate Chemical and Scientific Corporation; Westbury, NY) and analyzed using a BD LSRII S854 flow cytometer (BD Biosciences; San Jose, CA) with UV laser excitation. All DNA content histograms were analyzed using the computer program Multicycle (De Novo Software; Glendale, CA) based on the published consensus guidelines [35,36]. DNA aneuploidy was defined as an extra G₀/G₁ population that was visually distinguishable from a normal diploid G₀/G₁ peak [35,36]. Two authors (W.-T.C. and P.S.R.) interpreted all flow cytometric histograms without knowledge of the histological diagnoses and clinical history.

2.3. Statistical analysis

For descriptive purposes, continuous data were summarized by mean, and categorical data were summarized by percentage. Fisher’s exact test was used to compare the demographic, clinical, and endoscopic features of benign (including negative and reactive) with those of malignant cases (including adenocarcinoma, at least HGD, and atypical). P-values < 0.05 were considered statistically significant.

3. Results

3.1. Clinicopathologic features of bile duct biopsies

The patients included 33 men (52%) and 30 women (48%), with a mean age of 63 years (range: 18–89 years) (Table 1). Eleven patients (17%) had PSC. The majority of biopsies were obtained from the common bile duct (84%) and showed an endoscopic appearance of either stricture (49%) or mass/polyp (19%). Thirteen malignant cases (including adenocarcinoma, at least HGD, and atypical) more frequently presented as a mass/polyp (46%) than benign cases (including negative and reactive) (12%) (p = 0.012). Eleven patients (17%) were ultimately proven to have adenocarcinoma, including 8 with extrahepatic cholangiocarcinoma, 2 with pancreaticobiliary adenocarcinoma, and 1 with gallbladder adenocarcinoma. Two additional patients had at least HGD. One patient with HGD subsequently underwent surgical resection, which showed intraductal papillary neoplasm with HGD, but no invasive carcinoma was identified. No follow-up information was available in the other HGD case. No significant difference was noted in the other features (including age, gender, race, presence of PSC, and location of biopsy site) between benign and malignant cases (p > 0.05).

Table 1.

Clinicopathologic features of negative, reactive, malignant, and atypical bile duct biopsies.

Clinicopathologic features	Overall (n = 63)	Negative (n = 22)	Likely reactive (n = 28)	Adenocarcinoma or at least HGD (n = 10)	Atypical (n = 3)	P-value
Mean age, years (range)	63 (18–89)	62 (18–89)	61 (28–85)	68 (48–81)	68 (62–72)	0.108
Male gender (%)	33 (52%)	13 (59%)	10 (36%)	8 (80%)	2 (67%)	0.064
Caucasian ethnicity (%)	43 (68%)	16 (73%)	20 (71%)	5 (50%)	2 (67%)	0.315
Primary sclerosing cholangitis (%)	11 (17%)	4 (18%)	6 (21%)	1 (10%)	0 (0%)	0.433
Common bile duct location (%)	53 (84%)	16 (73%)	26 (93%)	8 (80%)	3 (100%)	1.000
Endoscopic appearance of stricture (%)	31 (49%)	11 (50%)	13 (46%)	5 (50%)	2 (67%)	0.763
or mass/polyp (%)	12 (19%)	2 (9%)	4 (14%)	5 (50%)	1 (33%)	0.012
Aneuploidy (%)	10 (16%)	0 (0%)	0 (0%)	7 (70%)	3 (100%)	<0.001
HGD or adenocarcinoma on follow-up	13 (21%)	0 (0%)	0 (0%)	10 (100%)	3 (100%)	–

Abbreviations: HGD, high-grade dysplasia.

3.2. DNA content analysis of benign biopsies

No abnormal DNA content was identified in the 22 negative/benign (Fig. 1A) and 28 likely reactive biopsies (Fig. 1C) from 50 patients. Ten patients (20%) had a history of PSC. None of the cases in this group developed adenocarcinoma within a mean follow-up time of 37 months (range: 0–282 months).

3.3. DNA content analysis of malignant biopsies

Aneuploidy was detected in 7 (70%) of 10 cases: 5 (71%) of 7 adenocarcinoma cases (Fig. 2A) and 2 (67%) of 3 at least HGD biopsies (Fig. 2C). Among the 2 cases of at least HGD with aneuploidy, 1 case developed extrahepatic cholangiocarcinoma, but no follow-up information was available in the other case. The remaining 1 case of at least HGD, despite having normal DNA content, was found to have extrahepatic cholangiocarcinoma on follow-up. Among the 10 patients, only 1 patient (10%) had a history of PSC and subsequently developed extrahepatic cholangiocarcinoma.

3.4. DNA content analysis of atypical biopsies

Of the 3 atypical cases with rare atypical glands/cells, aneuploidy was detected in all 3 cases (Fig. 3). One patient was subsequently diagnosed with extrahepatic cholangiocarcinoma, and another patient was diagnosed with pancreaticobiliary adenocarcinoma. In the remaining 1 atypical case, the subsequent surgical resection showed intraductal papillary neoplasm with HGD, but no invasive carcinoma was identified. Overall, the estimated sensitivity of aneuploidy as a diagnostic marker of malignancy (adenocarcinoma and HGD) was 70%, with the specificity of 100%, positive predictive value (PPV) of 100%, and negative predictive value (NPV) of 94%.

4. Discussion

Differentiating between benign and malignant bile duct biopsies can be challenging because the specimens are often small, fragmented/crushed, and mixed with significant inflammation, ulceration, and/or debris. This can make it difficult to recognize potential malignant cells and/or distinguish reactive atypia from true neoplastic transformation. Even if a biopsy contains some rare cells fulfilling most, if not all, of the criteria for malignancy, pathologists may be reluctant to make a diagnosis of malignancy based on a limited number of such cells, which may result in an atypical diagnosis. In addition, pancreaticobiliary adenocarcinoma can be very well differentiated that the distinction between reactive and neoplastic glands may be difficult. Aneuploidy determined by DNA flow cytometry using FFPE tissue offers a more objective method of delineating benign/reactive from malignant bile duct biopsies.

This study shows that aneuploidy can potentially serve as a diagnostic marker of malignancy (adenocarcinoma and HGD), with high sensitivity (70%) and specificity (100%). The high prevalence of aneuploidy in malignant biliary tumors is consistent with the previous report that aneuploidy is present in up to 86% of bile duct and gallbladder carcinomas [37]. Our data also suggest that the presence of atypical cells in the setting of normal DNA content likely signifies an inflammatory/reactive response rather than a malignant process, whereas the finding of aneuploidy indicates the presence of malignancy as all benign/reactive cases showed normal DNA content. The utilization of FFPE tissue allows a histological assessment to be made first followed by DNA flow cytometric analysis, only if a diagnosis of malignancy cannot be made with certainty by morphology (possibly with immunohistochemical staining, such as p53, S100P, and/or mapsin staining). Considering that the routine use of next-generation sequencing as a diagnostic marker of malignancy is limited by high cost and long turnaround time, DNA flow cytometry can potentially serve as a useful ancillary tool as it can be performed on limited biopsy samples and is a relatively inexpensive, yet objective test that can be completed within 2 days. A positive flow cytometric result can be very helpful in confirming a morphologic suspicion of malignancy, potentially minimizing the need for repeat tissue sampling. Conversely, the finding of normal DNA content provides additional reassurance to negative pathological evaluation.

We note that the surgical resection of one atypical case with aneuploidy showed intraductal papillary neoplasm with HGD, but no invasive carcinoma was identified. Although we cannot entirely exclude the possibility of sampling error in the histological evaluation of this resected specimen, it is important to note that FISH abnormalities associated with cholangiocarcinoma, such as polysomy (gains of 2 or more chromosomes), have been reported in both low-grade dysplasia (LGD, 11%) and HGD (58%), supporting a metaplasia-LGD-HGD-carcinoma sequence in the development of biliary tumors [38]. However, biliary dysplasia appears to be very rare or even absent in patients without cholangiocarcinoma, suggesting that dysplasia is a marker of current or developing malignancy [39–42]. In fact, Fleming et al. [41] identified dysplasia in 5 (19%) of 26 biopsy specimens from patients with concurrent or subsequent cholangiocarcinoma, but in none of the 60 patients without cholangiocarcinoma. Similarly, Lewis et al. [40] demonstrated that cholangiocarcinoma cases more frequently harbor dysplasia (any grade; 83% versus 36%, respectively, p < 0.001) and HGD (60% versus 11%, respectively, p < 0.001) than noncholangiocarcinoma cases. Furthermore, a case has been reported in which biliary dysplasia preceded the diagnosis of cholangiocarcinoma by at least 18 months [42]. Indeed, the presence of abnormal DNA content in biliary dysplasia may explain a high false positive rate of 23% reported in both DNA flow cytometry [5] and image analysis [12] that used a separately submitted brushing sample for testing, underscoring the importance of morphological correlation in interpreting any ancillary test results. Because a definitive tissue diagnosis of malignancy is preferable before performing resection, our diagnostic approach using FFPE tissue can potentially complement the morphological evaluation of bile duct biopsies.

Our study has some limitations. First, the sample size of malignant and atypical cases is small (n = 13), and thus, the results should be confirmed in a larger series. Second, different types of tumors (including extrahepatic cholangiocarcinoma, pancreaticobiliary adenocarcinoma, and gallbladder adenocarcinoma) were evaluated by DNA flow cytometry, so specific sensitivities in detecting different types of tumors cannot be accurately determined in our study. Finally, it is important to note that aneuploidy may not be diagnostic of invasive carcinoma. Although we were not able to specifically evaluate the rate of aneuploidy in premalignant lesions owing to the rarity of these cases, biliary dysplasia has been shown to demonstrate abnormal DNA content [38]. However, given the high risk of progression to invasive carcinoma in these cases [39–42], it could be argued that aggressive management is warranted. This is especially true for patients with PSC because early liver transplantation is an option if biliary dysplasia is repeatedly documented even in the absence of cholangiocarcinoma, which is considered a contraindication for liver transplantation [43]. DNA flow cytometry could possibly identify patients with premalignant conditions who are at increased risk of subsequent detection of cholangiocarcinoma as all 3 atypical cases with aneuploidy were subsequently found to have adenocarcinoma (n = 2) or HGD (n = 1). In support of this notion, we previously demonstrated that the presence of aneuploidy detected by DNA flow cytometry significantly correlates with subsequent detection of malignancy throughout the gastrointestinal tract, including the ampulla [28], Barrett’s esophagus [29], inflammatory bowel disease [30–32], and the stomach [33].

In conclusion, DNA flow cytometry using FFPE tissue from bile duct biopsies can potentially serve as a valuable adjunct in the objective identification of malignant epithelial cells. The high rate of aneuploidy (70%) was observed in malignant cases (adenocarcinoma and HGD), whereas all benign and reactive biopsies showed normal DNA content. When there is a clinical suspicion of malignancy and biopsy is atypical or suspicious (but limited by scarcity of atypical glands/cells, inflammation, and/or ulceration), the finding of aneuploidy by DNA flow cytometry can provide confirmatory evidence to a morphologic suspicion or impression of malignancy.