Abstract
Background/Aim: Detection of pancreatic cancer using small samples of the pancreas obtained by endoscopic ultrasound-guided fine-needle aspiration (EUS-FNA) remains a challenge. The purpose of this study was to investigate whether the detection of KRAS mutations in cell-free DNA (cfDNA) extracted from supernatants of liquid-based fixed cytology (LBC) specimens obtained using EUS-FNA in solid pancreatic cancer can be an auxiliary test for differential diagnosis. The purpose of this study was to investigate whether the detection of KRAS mutations in cell-free DNA (cfDNA) extracted from supernatants of liquid-based fixed cytology (LBC) specimens obtained using EUS-FNA in solid pancreatic cancer can be an auxiliary test for differential diagnosis. Patients and Methods: This was a single-institution cohort study that included 50 patients with pancreatic lesions. cfDNA was isolated from the supernatant of fixed LBC samples, and KRAS mutation status was compared between cfDNA samples and FFPE small fragment tissues. Results: Of the 50 cfDNA samples, 84% (42/50) were valid. KRAS mutations were detected in 57.1% (24/42) of the 42 valid samples. The sensitivity, specificity, and accuracy of KRAS mutation detection using cfDNA samples in the pancreatic lesions were 63.2% (24/38), 100.0% (4/4), and 66.7% (28/42), respectively. KRAS mutation status between FFPE small tissues and cfDNA samples were comparable. Conclusion: Gene mutation analysis using cfDNA from the supernatant of fixed LBC samples is an effective ancillary diagnostic tool for pancreatic cancer.
- Pancreatic cancer
- endoscopic ultrasound-guided fine-needle aspiration cytology
- liquid-based cytology
- cytology cell-free DNA
- KRAS mutation
Pancreatic cancer is one of the most common causes of cancer-related deaths among people worldwide (1, 2). The incidence has steadily increased over the past 30 years in developed countries (2). The high mortality rate of pancreatic cancer stems from late diagnosis, and 53% of patients are diagnosed with a metastatic disease (3).
Endoscopic ultrasound-guided fine-needle aspiration (EUS-FNA) is widely used in the histological diagnosis of abdominal tumors. It is an accurate and safe diagnostic modality, which has become the first-line sampling procedure for the histological/cytological diagnosis of solid pancreatic cancer (4). EUS-FNA cytology is considered a useful tool for the evaluation of solid pancreatic lesions; however, its diagnostic yield may be influenced by multiple factors, such as the experience of the endosonographer and the cytopathologist/cytotechnologist, tumor characteristics, sampling technique, and processing of smears (5). Therefore, the detection of pancreatic cancer using small samples remains a challenge in pancreatic EUS-FNA cytology.
The Kirsten rat sarcoma viral oncogene homolog (KRAS) gene encodes the KRAS protein. Point mutations in the KRAS gene are present in over 90% of pancreatic ductal adenocarcinomas (PDAC), and the mutations typically affect the hotspot codon 12 (6). Detection of KRAS mutations using EUS-FNA samples is useful for initial diagnosis, and KRAS mutations are used to confirm a malignant EUS-FNA diagnosis (7). However, pathological/cytological diagnosis and genetic analysis are often hampered by the small sample volume of tumor cells.
Liquid-based cytology (LBC) was first developed in 1991. Optimal sampling techniques may produce samples of satisfactory quality with high cellularity and especially low blood contamination (8). In addition, the reduction of bloody background allows better visibility in the LBC method. LBC tests (i.e., ThinPrep and Surepath) are more accurate than conventional smears for the cytological diagnosis of many cancers (9-14). We previously performed cytological research using cell-free DNA (cfDNA), and showed that EGFR mutations can be detected in cfDNA obtained from the supernatant fluids of effusion and cerebrospinal fluid (CSF) samples (15, 16). We have also established a technique to extract genomic DNA, such as cfDNA, from the supernatant fluid of CytoRich Red-fixed LBC samples (17). In the biliopancreatic region, cfDNA collected from the supernatant fluid is an effective adjunctive diagnostic tool for pancreatic cancer. It is a viable alternative to formalin-fixed and paraffin-embedded (FFPE) cell block with the potential for reduced additional tissue sampling and rapid turnaround time (18). Furthermore, cfDNA collected from the supernatant fluid of pancreatic bile duct brush cytology specimens is also an adjunct diagnostic tool for pancreatic and cholangiocarcinoma (19).
The aim of this study was to investigate whether the detection of KRAS mutations in cfDNA from the supernatant fluid of LBC samples fixed in CytoRich Red can be used as an ancillary test of EUS-FNA for differential diagnosis of solid pancreatic cancer.
Patients and Methods
Cell culture and clinical patient selection. The human pancreas cancer cell lines MIA PaCa-2 (KRAS_G12C) and PANC-1 (KRAS_G12D), and the human lung cancer cell line PC9 (EGFR_exon19 del) were cultured in Roswell Park Memorial Institute 1640 (RPMI-1640, Nissui Pharmaceutical Co., Ltd., Tokyo, Japan) medium supplemented with 10% fetal bovine serum in an atmosphere of 5% CO2 as described previously (17). Each cell line was fixed in CytoRich Red, and after overnight fixation at room temperature (about 23-25 degrees), all cell lines were divided into sediment and supernatant fluid.
This retrospective study evaluated data from consecutive patients who provided written informed consent before undergoing initial EUS-FNA cytology for suspected solid pancreatic lesions at Kurume University Hospital, Kurume-city, Japan between June 2018 and December 2018. Fifty patients who underwent EUS-FNA cytology sampling with supernatant fluid were enrolled. Final diagnoses were made via biopsy, surgery, or clinical progress, and follow-up was performed for a minimum of 24 months. Patients without final pathological diagnoses were clinically diagnosed based on imaging findings indicating lymph node or liver metastasis. This retrospective study was approved by the Ethics Committee of Kurume University Hospital, Kurume-city, Japan (391). The study was carried out in accordance with the criteria set by the Declaration of Helsinki.
EUS-FNA procedures. Histological and cytological samples were obtained from patients with pancreatic lesions who were referred to Kurume University Hospital and determined to be eligible for EUS-FNAB. The samples were collected using a 22G/25G puncture needle (Olympus, Ltd., Tokyo, Japan) with the following technique: 15-20 strokes per session, 2-3 times (20). After the needle was retracted into the catheter, the entire catheter was withdrawn. At our institution, a skillful cytotechnologist determines specimen adequacy using rapid on-site evaluation (ROSE) (21, 22). Materials aspirated with saline were spread onto a petri dish and fixed by skillful cytotechnologists (Figure 1). Two alcohol-fixed slides for ROSE and cytological final diagnosis, a formalin-fixed vial for histological diagnosis, and a CytoRich Red (Becton Dickinson, Franklin Lakes, NJ, USA) for the LBC method were prepared in most cases (17). On one of the two alcohol-fixed slides, modified ultrafast Papanicolaou staining using Cytocolor® (CYTOCOLOR® Cytological standard stain acc. to Szczepanik for microscopy: Merck KGaA, Darmstadt, Germany) was performed for ROSE. Visible small tissue on the petri dish was selected, fixed in 10% neutral buffered formalin, and embedded in paraffin for histological diagnosis. After extraction of visible small tissue, distilled water was directly placed into a disposable tube. In our laboratory, a disposable tube containing all residual cell material in distilled water was centrifuged at 800 × g for 10 min, and all sediments were fixed in CytoRich Red. After overnight fixation at room temperature (about 23-25 degrees), all samples were divided into cell sediment and supernatant fluid. The supernatant fluids (2.0-6.0 ml) were then immediately stored at −80°C until DNA extraction, and cell sediments were prepared for cytological diagnosis. Each LBC smear was processed using the SurePathTM system according to the manufacturer’s instructions, as described in our previous report (17).
Processing methods for rapid on-site evaluation (ROSE) by endoscopic ultrasound-guided fine-needle aspiration (EUS-FNA) cytology. The blue arrows indicate on-site work, and the green arrows indicate pathological laboratory work.
Histological and cytological definitions. Formalin-fixed and paraffin-embedded (FFPE) tissues were prepared using small tissue samples fixed with 10% neutral buffered formalin, and all FFPE samples were stained with hematoxylin and eosin (HE). All cytological smears were stained using the Papanicolaou (Pap) method. Cytological results of EUS-FNA were classified as inadequate or adequate; adequate specimens were classified as normal or benign, indeterminate, suspicious for malignancy, or malignant (23). Neuroendocrine tumor (NET) was evaluated as indeterminate.
FFPE small tissue samples, cell-free DNA extraction and KRAS mutation analysis. Paraffin-embedded tissues were sliced into thin sections of 8 μm thickness, and 4-6 sections were used for DNA extraction. Genomic cfDNA was purified from 200 μl of supernatant fluids using a Master Pure DNA extractor kit (Wako Pure Chemical, Osaka, Japan) according to the manufacturer’s protocol (15, 16). The DNA quantity was assessed using a NanoDrop One (Thermo Fisher Scientific, Waltham, MA, USA). In the NanoDrop One analysis, low-quality cfDNA (A260/A280 ratio higher than 3.0) was considered invalid.
Briefly, genomic DNA (G12A, G12C, G12D, G12R, G12S, G12V, and G13D) was amplified using a thermal cycler (Bio-Rad Laboratories, Inc. Hercules, CA, USA). Subsequently, point mutations in the KRAS gene codons 12 and 13 (G12A, G12C, G12D, G12R, G12S, G12V and G13D) were examined using the fluorescence resonance energy transfer-based preferential homoduplex formation assay (F-PHFA; Riken Genesis Co., Ltd., Tokyo, Japan) according to the manufacturer’s instructions. F-PHFA is a novel nonenzymatic mutation detection method that uses competitive hybridization between labeled DNA and polymerase chain reaction (PCR) products. The assay was performed using the CFX Connect real-time PCR detection system (Bio-Rad Laboratories, Inc.) and an F-PHFA customized kit (Riken Genesis Co, Ltd., Tokyo, Japan) (12, 17). In this PCR analysis, samples in which wild-type KRAS was not detected were defined as invalid, showing DNA fragmentation. By applying the F-PHFA assay to a set of clinical FFPE samples with known mutational status, five activating KRAS point mutations (G12V, G12D, G12R, G12A, and G13D) were validated in this study. In addition, we confirmed that the threshold of KRAS point mutations is an index of 50.
Statistical analysis. Data are presented as mean (ranges) or as numbers (percentages). The agreement in KRAS mutation detection between FFPE tissues and cfDNA samples was examined using Cohen’s kappa. Cohen’s kappa was interpreted as follows: a kappa coefficient <0.2 was considered “slight;” between 0.2 and 0.4 was “fair;” between 0.41 and 0.6 was “moderate;” between 0.61 and 0.8 was “substantial;” and a kappa coefficient >0.8 was considered ‘almost perfect.’ All statistical analyses were performed using SAS version 9.4 (SAS Institute Inc., Cary, NC, USA).
Results
Activating KRAS point mutations using cell lines. First, we confirmed the detection of KRAS mutations using cfDNA from supernatant fluids of LBC samples fixed with CytoRich Red using MIA PaCa-2 PANC-1 and PC9 cells. Genomic cfDNA was successfully purified and there were no low-quality cfDNAs. In the PCR analysis, KRAS mutations were detected using cfDNA (Figure 2).
KRAS point mutations using cell lines. KRAS mutations were detected using cell free DNA obtained from supernatant fluids with MIA PaCa-2 cells and PANC-1 cells, but not PC9 cells.
Baseline characteristics of patients and cytological diagnoses using EUS-FNA cytology. Fifty patients were diagnosed with 42 pancreatic cancers and eight benign lesions, including three NETs and five non-neoplastic lesions, via biopsy, surgery, or clinical progress, and follow-up (Table I). In pancreatic cancers, there was no significant proportional variance regarding the sex of the patients; the male-female ratio was 23:19. The average age at the time of sampling was 66.0 years (range=33-81 years; median 69). Tumor lesions were located most frequently in the pancreatic head (n=25; 59.5%).
Baseline characteristics of endoscopic ultrasound-guided fine-needle aspiration cytology groups.
The classification of specimens based on EUS-FNA cytological diagnosis was as follows: 31 malignant specimens, including those suspicious for malignancy, eight indeterminate, 10 normal or benign, and one inadequate. To determine the cytological sensitivity, specificity, accuracy, positive predictive value (PPV), and negative predictive value (NPV), both malignant and suspicious for malignancy specimens were classified as malignant, while normal or benign, indeterminate, and inadequate specimens were categorized as benign. The results showed that EUS-FNA cytology had 73.8% (31/42) sensitivity, 100.0% (8/8) specificity, 78.0% (39/50) accuracy, 100.0% (31/31) PPV, and 42.1% (8/19) NPV (Table II).
Diagnostic efficacy of endoscopic ultrasound-guided fine-needle aspiration cytology for pancreatic lesions.
Detection of KRAS mutations in, and quality of, cfDNA extracted from supernatant fluids collected via EUS-FNA cytology. Next, we examined whether KRAS mutations could be detected using cfDNA extracted from supernatant fluids of fixed clinical LBC samples. Genomic cfDNA was successfully purified from all LBC samples (average, 16.3 ng/μl; range=7.7-36.0 ng/μl). The cfDNA mean of pancreatic cancer, neuroendocrine tumor and benign were 15.8 ng/μl (7.7-36.0), 20.5 ng/μl (8.8-30.8), and 17.9 ng/μl (14-28.1). The mean retention period was 116 days (43-179 days). In addition, there were no low-quality cfDNAs (average A260/A280 ratio, 2.20; range=1.60-2.95). Of the 50 cfDNA samples, 84% (42/50) were valid and 16% (8/50) were invalid. Of the 42 samples with valid evaluation, KRAS mutations were detected in 57.1% (24/42), whereas no KRAS mutations were detected in the remaining 42.9% (18/42). Among the 24 samples with KRAS mutations, there were seven G12D (29.2%), 14 G12V (58.3%), and three G12R (12.5%) subtypes. No KRAS mutations were detected in the cfDNA samples from benign lesions, such as NETs (n=3), or non-neoplastic lesions (n=5). The sensitivity, specificity, and accuracy of KRAS mutation detection using cfDNA samples from the pancreatic lesions were 63.2% (24/38), 100.0% (4/4), and 66.7% (28/42), respectively (Table III).
Results of KRAS mutation detection in cfDNA from supernatant fluids.
KRAS mutation status concordance between FFPE small tissues and cfDNA samples. Finally, we investigated the KRAS mutation status concordance between FFPE small tissue samples and cfDNA from supernatant fluids. Of the 50 FFPE small tissue samples, 90% (45/50) were valid and 10% (5/50) were invalid. Of the 45 samples with valid evaluation, KRAS mutations were detected in 51.1% (23/45), whereas no KRAS mutations were detected in the remaining 48.9% (22/45). Among the 23 samples with KRAS mutations, there were six G12D (26.1%), 14 G12V (60.9%), and three G12R (13.0%) subtypes. No KRAS mutations were detected in the FFPE small tissue samples from benign lesions, such as NETs (n=3), or non-neoplastic lesions (n=5). KRAS mutation data from small tissues and cfDNA samples are summarized in Table IV. There was no significant difference in KRAS mutation status between FFPE small tissues and cfDNA samples; the overall concordance rate was 66% (33/50), with a kappa coefficient of 0.438 (95% confidence interval=0.251, 0.625), indicating moderate agreement.
KRAS mutation status concordance between formalin-fixed paraffin-embedded (FFPE) small tissues and cfDNA samples (n=50).
An ancillary diagnostic tool to determine malignancy using cfDNA from supernatant fluids. The relationship between cytological diagnosis and KRAS mutation detection in pancreatic cancers is shown in Table V. The rates of mutant type detection in both indeterminate and suspicious for malignancy specimens were higher than those in normal or benign specimens, suggesting that cfDNA from supernatant fluids can be used for ancillary genetic testing.
Relationship between cytological diagnosis and KRAS mutation detection in pancreatic cancers.
Discussion
In the current study, we investigated whether the detection of KRAS mutations in cfDNA from the supernatant fluid of LBC samples fixed in CytoRich Red can be used as an ancillary test of EUS-FNA for differential diagnosis of solid pancreatic cancer. We observed that 84% (42/50) of the 50 cfDNA samples were classified as valid. In addition, KRAS mutation status between FFPE small tissues and cfDNA samples were comparable. Furthermore, the detection rates of mutant types in the diagnostic category of indeterminate and suspicious for malignancy were higher than those in the normal or benign category. This suggests that, when diagnosis is difficult, a higher diagnostic accuracy may be achieved with the combination of molecular techniques than with a microscopic observation technique alone.
Liquid biopsies, particularly those involving cfDNA from plasma, are rapidly emerging as an important and minimally invasive adjunct to standard tumor biopsies (24). Liquid biopsy has typically referred to the analysis of cfDNA from peripheral blood samples; however, different body fluids (non-peripheral blood samples) can be used as a source of tumor-derived molecular information, such as urine, CSF, and pleural effusions (15, 16, 25, 26). We have previously reported that activating EGFR mutations can be detected in cfDNA extracted from the supernatant fluid of fixed LBC samples via a PCR assay (17). In pancreatic cancer, the usefulness of KRAS mutation analysis using tumor tissues or cytological samples has been reported (12). Moreover, Fonseca et al. reported that cfDNA-based liquid biopsy is an important clinical tool for the monitoring of patients with pancreatic cancer (27).
Specimens obtained using EUS-FNA have traditionally been diagnosed using a 95% ethanol-fixed conventional Pap smear, whereas LBC has been widely used for preparing gynecological and non-gynecological cytological specimens in recent years. Therefore, we used a conventional Pap smear combined with LBC to assess the diagnostic efficacy of EUS-FNA for pancreatic lesions. Hashimoto et al. reported that LBC (Surepath) has a relatively higher diagnostic performance than conventional Pap smears for EUS-FNA samples of pancreatic lesions (12). In contrast, Yeon et al. reported that LBC (CellPrepPlus) shows lower diagnostic accuracy in pancreatic EUS-FNA than conventional Pap smears (8). However, there was no on-site cytopathologist who could evaluate if the small samples at the time of collection were adequate (8). Therefore, sample collection and cell processing methods are important for accurate diagnosis using EUS-FNA.
Histo/cytological examination using microscopic observation of tumor cells obtained by EUS-FNA is the gold standard for the diagnosis of pancreatic cancer. Sekita-Hatakeyama Y et al. reported that KRAS mutation analysis using residual LBC samples was successful in all cases, and the sensitivity, specificity, and accuracy of the combination of cell block examination and KRAS mutation analysis were 90.3%, 92.3%, and 90.7%, respectively (14). Furthermore, the authors concluded that KRAS mutation analyses of EUS-FNA specimens may improve the accuracy of pathological diagnosis. The application of EUS-FNA in cell sampling is limited for a variety of reasons, such as the presence of necrosis, inflammation or a hard stromal component, needle size, and FNA technique (18, 28). To perform accurate gene analysis, genetic analysis in supernatant fluids of fixed LBC samples is also important, because LBC samples obtained by EUS-FNA can be used as a source for investigating KRAS mutations as prognostic and predictive biomarkers. In the current study, we found that genomic cfDNA was successfully purified from the supernatant fluids of all LBC samples, and KRAS mutations were detected in 57.1% (24/42) of the 42 samples with valid evaluation, demonstrating that KRAS mutations can be detected in cfDNA from supernatant fluids of fixed LBC samples obtained using EUS-FNA.
PDAC is by far the most common pancreatic neoplasm (13, 29). Point mutations in the KRAS gene are thought to be an early event in the development of PDAC, already occurring in low-grade pancreatic intraepithelial neoplasia (PanIN) lesions of the pancreas (30). In addition, PanIN lesions are also present in autoimmune pancreatitis with cancer (31).
In the current study, no KRAS mutations were detected in cfDNA samples from benign lesions, such as NETs and non-neoplastic lesions. However, Sekita-Hatakeyama Y et al. reported that, among the 13 specimens that were classified as benign, one had the G12R KRAS mutation subtype, and that the corresponding patient may have had PanIN in association with chronic pancreatitis (14). Although the detection of KRAS mutations in cfDNA is reminiscent of malignancy, it may not be as easily diagnosed as PDAC. Therefore, we believe that KRAS mutation analysis should be understood as an ancillary genetic tool for the histo/cytological diagnosis in EUS-FNA specimens.
Quantification and assessment of the quality of cfDNA in cytology is also very important for DNA analysis using clinical samples. Roy-Chowdhuri S et al. reported that, in effusion cytology, effusion supernatant samples had equivalent cfDNA yield and sequencing metrics to those of matched FFPE samples, and showed higher DNA concentrations than plasma samples of similar volumes (32). In our previous report on CSF cytology, 53.9% (16/26) of the cfDNA samples were invalid because the amount of extracted genomic DNA from cfDNA was very small. In contrast, 16% (8/50) of the cfDNA samples obtained using EUS-FNA in this study were invalid, and the amount of cfDNA obtained from EUS-FNA samples was higher than that from CSF samples. Therefore, in genetic analysis, not only the quality, but also the quantity of the DNA is important.
Study limitations. First, there were a few non-neoplastic lesions, which may have affected the PPV and NPV. Second, there were four cases of discrepancy in the KRAS mutation analysis between FFPE small tissues and cfDNA samples. For example, one case diagnosed as malignant was found to contain KRAS wild-type using KRAS mutation analysis of cfDNA. In another case, the cytological diagnosis was normal or benign, whereas KRAS mutation analysis detected a KRAS mutant type. Thus, the genetic analysis of supernatant cfDNA is not complete, and further studies are needed to increase the number of cases.
In conclusion, we demonstrated that gene mutation analysis using cfDNA from supernatant fluids of fixed LBC samples is an effective ancillary diagnostic tool for pancreatic cancer. We believe that, in the near future, cfDNA analysis from non-blood biological liquids including LBC samples will become routine clinical practice for cancer patient diagnosis and management.
Acknowledgements
The Authors would like to thank Editage (www.editage.com) for English language editing.
Footnotes
Authors’ Contributions
A.K., Y.N., and J.A. designed the experiments; Y.T., H.A. performed the experiments; E.S. analyzed the data; A.K., Y.N., Y.O., and H.A. helped with the discussion; A.K., and Y.N. wrote the manuscript; and J.A. supervised the project. All Authors have reviewed the manuscript.
Conflicts of Interest
The Authors declare that there are no potential conflicts of interest relevant to this article.
- Received March 11, 2023.
- Revision received March 26, 2023.
- Accepted March 27, 2023.
- Copyright © 2023 International Institute of Anticancer Research (Dr. George J. Delinasios), All rights reserved.
This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY-NC-ND) 4.0 international license (https://creativecommons.org/licenses/by-nc-nd/4.0).
