Introduction

Pancreatic ductal adenocarcinoma (PDAC) is one of the most aggressive cancers. In 2021, PDAC was the fourth leading cause of cancer-related mortality in the United States, with a 5-year survival rate of 10%1, and it is predicted to become the second leading cause of cancer-related deaths by 20302.

Peritoneal dissemination is one of the most common recurrence patterns in patients with resected PDAC, and survival after peritoneal recurrence is dismal3. Peritoneal washing fluid obtained by irrigation of the peritoneal cavity with normal saline solution during surgery (intraoperative peritoneal washing fluid [IPWF]) is used for peritoneal washing cytology. Papanicolaou and Giemsa staining examination by a pathologist is the standard protocol for detecting floating tumor cells in the peritoneal cavity, and positive peritoneal washing cytology is a well-known indictor of poor prognosis and a risk of peritoneal dissemination in patients with resected PDAC3,4,5. However, because IPWF has a high content of non-cancerous cells, such as mesothelial cells and white blood cells, and a low amount of tumor cells, peritoneal washing cytology has a relatively low sensitivity of 28–64%, but a high specificity in the range of 85–100%6,7,8. Therefore, peritoneal washing cytology might be underestimated in most patients who undergo curative resection. Thus, a more sensitive and reliable diagnostic tool for detecting floating tumor cells in IPWF is desired.

KRAS mutations in patients with PDAC are widely accepted as a prognostic factor, even in liquid biopsy9. The rate of KRAS mutations is 80–95% in patients with PDAC, and these mutations are commonly found in codon 12/1310,11,12. Therefore, the detection of KRAS mutations in IPWF may reflect the presence of cancer cells, complement the low sensitivity of peritoneal washing cytology, and help predict peritoneal dissemination recurrence in patients with PDAC.

Various methods are available for detecting tumor-derived mutations, including next-generation sequencing (NGS)-based and PCR-based methods13. Peptide nucleic acid (PNA)-directed PCR clamping is one of the PCR-based methods. A PNA is a synthetic DNA analog that blocks primer annealing and chain elongation of wild-type alleles in PCR; thus, only mutant-type alleles are amplified9,14,15,16,17. PNA-directed PCR clamping has a high sensitivity and low cost, using routine real-time PCR machines, and allows rapid detection compared with other analytical applications18. Detecting low variant allele frequency variants in peritoneal floating-cell DNA is challenging19,20. PNA-directed PCR clamping enhances the sensitivity of amplification of mutant alleles and may be favorable for detecting mutations with a low variant allele frequency.

Previously, we reported that KRAS mutation detection in cell-free DNA from the serum of pancreatic cancer patients can be used to predict the prognosis of pancreatic cancer patients by analyzing threshold cycle (Ct) values using PNA-directed PCR clamping9. The aim of this study was to evaluate the performance of the PNA-directed PCR clamping as a molecular-based peritoneal washing cytology for sensitive detection of KRAS mutations in PDAC by comparing the sensitivity, specificity, and accuracy of this method with those of peritoneal washing cytology.

Results

Patient characteristics and IPWF cytology

IPWF was obtained from 54 patients whose characteristics are presented in Table 1. Resectability was categorized as resectable in 36 patients (66.7%), borderline resectable pancreatic cancer (BR) with portal vein involvement in three patients (5.6%), BR with artery involvement in 10 patients (18.5%), unresectable locally advanced in two patients (3.7%), and unresectable with metastasis in three patients (5.6%).

Table 1 Clinicopathological features of patients enrolled in this study.
Full size table

Seven of the 54 (13.0%) patients were diagnosed with positive peritoneal washing cytology and five (9.3%) with peritoneal dissemination at surgery. For abdominal findings, 45 (83.3%) patients had negative peritoneal washing cytology and absence of peritoneal dissemination, four (7.4%) had positive peritoneal washing cytology and absence of peritoneal dissemination, two (3.7%) had negative peritoneal washing cytology and presence of peritoneal dissemination, and three (5.6%) had positive peritoneal washing cytology and presence of peritoneal dissemination (Table 1).

Among the 45 patients with negative peritoneal washing cytology and absence of peritoneal dissemination, four patients underwent non-curative resection. The reasons for non-curative resection were positive invasion of the proper hepatic artery in two cases, positive extensive invasion of the portal vein in one case, and positive metastasis in lymph node #16 in one case. Among the four patients with positive peritoneal washing cytology and absence of peritoneal dissemination, two patients underwent distal pancreatectomy. Among the total 54 patients, curative resection was successful in 43 patients (79.6%).

Correlation of PNA-directed PCR clamping results with clinical features

The average of total yield of DNA in peritoneal floating-cell DNA was 3.990 µg (0.093–35.568 µg). The associations between PNA-directed PCR clamping results and clinical findings are shown in Fig. 1. The median Ct was significantly lower in patients with PDAC than in non-PDAC patients (36.680 vs. 38.930, P = 0.016). The Ct value was significantly lower in patients with positive peritoneal washing cytology than in those with negative peritoneal washing cytology (33.795 vs. 37.109, P < 0.001) and in patients with peritoneal dissemination than in those without dissemination (34.113 vs. 36.942, P = 0.005).

Fig. 1
figure 1

Boxplots showing the difference between Ct and PDAC, peritoneal washing cytology, and peritoneal dissemination status. Comparison of the median Ct value according to disease type (a). peritoneal washing cytology (b) and peritoneal dissemination (c) statuses were compared using the t-test. The Ct value was significantly lower in patients with PDAC, positive peritoneal washing cytology, and presence of peritoneal dissemination. PDAC: pancreatic ductal adenocarcinoma. *: P < 0.05, **: P < 0.01, ***: P < 0.005.

Full size image

Optimal cut-off of PNA-directed PCR clamping for KRAS mutations in IPWF

The optimal Ct cut-off value for KRAS mutation detection in peritoneal floating-cell DNA determined using receiver operating characteristic (ROC) curve analysis was 36.42, and the area under the curve, sensitivity, and specificity were 0.952, 100%, and 80.0%, respectively (Fig. 2). All patients with non-pancreatic cancer were determined to be peritoneal floating-cell DNA-negative when a cut-off value of 36.42 was used. Molecular cytopathology was positive in 22 of the 54 (40.7%) patients.

Fig. 2
figure 2

Receiver operating characteristics curve for the optimal cut-off value of KRAS mutations in pfDNA. ROC analysis was conducted to determine the optimal Ct value for predicting the presence of KRAS mutations in pfDNA, which was detected using Sanger sequencing or NGS. The optimal cut-off value was 36.42, and the AUC was 0.952. ROC: receiver operating characteristics; pfDNA: peritoneal floating cell DNA; NGS: next-generation sequencing; AUC: area under the curve.

Full size image

Correlations among peritoneal washing cytology status, peritoneal disseminationstatus, and molecular status

The correlations among peritoneal washing cytology, peritoneal dissemination, and molecular diagnosis are shown using a Venn diagram in Fig. 3a. Twenty-two (40.7%) patients and 16 (35.6%) patients with negative peritoneal washing cytology and absence of peritoneal dissemination had a positive molecular diagnosis. One patient with positive peritoneal washing cytology and absence of peritoneal dissemination and two patients with negative peritoneal washing cytology and presence of peritoneal dissemination had a negative molecular diagnosis, and all three patients with positive peritoneal washing cytology and presence of peritoneal dissemination had a positive molecular diagnosis.

Fig. 3
figure 3

(a) Venn diagram of the statuses of molecular cytopathology and peritoneal washing cytology, and the presence of peritoneal dissemination at surgery. The Venn diagram shows the correlation between molecular diagnosis and peritoneal washing cytology/peritoneal dissemination status. Among the 54 patients, 22 (40.7%) were found to have a positive molecular diagnosis, whereas 16 (35.6%) out of the 45 patients with CY0P0 had a positive molecular diagnosis. pfDNA: peritoneal floating cell DNA. (b) Relationship of KRAS mutations between peritoneal-floating cell DNA and FFPE specimen DNA.

Full size image

Correlation between positive molecular diagnosis in cytopathology and first recurrence site

The median observation period was 27.7 months. The first recurrence sites in patients with PDAC who underwent curative resection are shown in Table 2a,b. The associations between molecular diagnosis and first recurrence sites were investigated. Among 43 patients, 18 patients had a positive molecular diagnosis, and 25 patients had a negative molecular diagnosis. Recurrence occurred in 23 patients, including 12 (66.7%) patients with a positive molecular diagnosis and 11 (44.0%) patients with a negative molecular diagnosis. For the recurrence site, peritoneal dissemination recurrence was significantly more frequent in patients with a positive molecular diagnosis than in those with a negative molecular diagnosis (38.9% vs. 8.0%, P = 0.013). Among 41 curatively resected patients with negative peritoneal washing cytology and absence of peritoneal dissemination, peritoneal dissemination was significantly more frequent in patients with a positive molecular diagnosis than in those with a negative molecular diagnosis (37.5% vs. 8.0%, P = 0.021).

Table 2 Correlation of positive molecular cytology and first recurrence site in patients who underwent curative resection with CY0-1P0 or CY0P0.
Full size table

Sequencing of tumor DNA and peritoneal floating-cell DNA

The sequencing results are shown in Table 3. KRAS mutations were evaluated in 45 patients by analyzing DNA extracted from formalin-fixed paraffin-embedded (FFPE) specimens of primary tumors or metastases. Specifically, DNA was extracted from FFPE specimens of 42 pancreatic tumors, two peritoneum nodules, and one lymph node. The KRAS mutation rate in the FFPE specimens was 84.4% (38/45) according to Sanger sequencing. The GGT > GAT (G12D) mutation was the most frequent (n = 22/45, 48.9%), followed by GGT > GTT (G12V) (n = 12/45, 26.7%), GGT > CGT (G12R) (n = 3/45, 6.7%), and GGT > TGT (G12C) (n = 1/45, 2.2%).

Table 3 Detailed description of sequences.
Full size table

Sequencing of peritoneal floating-cell DNA was performed in 54 patients. KRAS mutations were detected by Sanger sequencing of PNA-directed PCR clamping products and NGS of peritoneal floating-cell DNA in 14 patients (n = 14/54, 25.9%). G12D was the most frequent (n = 6/54, 11.1%), followed by G12V (n = 5/54, 9.3%), G12R (n = 2/54, 3.7%), and G12C (n = 1/54, 1.9%).

The pattern of KRAS mutations was assessed in 45 patients for whom DNA extracted from both FFPE specimens and peritoneal floating-cell DNA was sequenced. Ten patients (n = 10/45, 22.2%) had KRAS mutations in DNA extracted from both FFPE specimens and peritoneal floating-cell DNA; five patients (n = 5/45, 11.1%) did not harbor KRAS mutations in DNA extracted from FFPE specimens or peritoneal floating-cell DNA; two patients (n = 2/45, 4.4%) had KRAS mutations only in peritoneal floating-cell DNA; and 28 patients (n = 28/45, 62.2%) had KRAS mutations only in DNA extracted from FFPE specimens (Fig. 3b).

Limit of detection of PNA-directed PCR clamping

To identify detection limit of PNA-directed PCR clamping, serial dilutions of KRAS mutation pancreatic cancer cell lines, ranging from 0 to 100% were used. KRAS mutation was significantly detected at a mutant-to-wild type DNA ratio of 0.1% (Fig. 4a). The linear equation was Ct = –2.645 × log variant allele frequency + 35.00, with a correlation coefficient of 0.954 (Fig. 4b). With a Ct cut-off value of 36.42 for IPWF, the estimated variant allele frequency calculated using the calibration curve was 0.29%.

Fig. 4
figure 4

(a) Bar graph for the correlation between variant allele frequency and Ct. To identify its limit of detection, PNA-directed PCR clamping was assessed using serial dilutions of KRAS mutation-positive pancreatic cancer cell lines. The difference between median Ct values was analyzed using the t-test. There was no statistically significant difference in the median Ct value between the negative control group (variant allele frequency 0%) and the mixture when variant allele frequency was below 0.1%, and the detection limit was variant allele frequency of 0.1%. Ct: cycle threshold; PNA: peptide nucleic acid. (b) Linear plot for relative quantification using diluted mutation allelic fraction in the serial dilutions. Similar line plots for quantification purposes were obtained for the serial cell line mixture. The linear equation was Ct = –2.645 × log variant allele frequency + 35.00, with a correlation coefficient of 0.954. PNA: peptide nucleic acid. ns: not significant, *: P < 0.05, **: P < 0.01, ***: P < 0.005, ****: P < 0.001.

Full size image

Discussion

We identified KRAS mutations in peritoneal floating-cell DNA and evaluated the prognostic value of KRAS mutation detection in predicting peritoneal dissemination. Using PNA-directed PCR clamping, we detected KRAS mutations in 22 out of 54 (40.7%) patients, suggesting that numerous patients harbor a certain amount of peritoneal floating tumor cells in the peritoneal cavity. Among the 45 patients who tested negative for both peritoneal washing cytology and peritoneal dissemination during surgery, 16 (35.6%) had a positive molecular diagnosis, indicating that relying solely on cytology may result in underestimating the risk of peritoneal dissemination in such cases. Furthermore, patients with a positive molecular diagnosis had peritoneal dissemination recurrence in resected PDAC significantly more often. Hence, molecular analysis of peritoneal lavage may have a higher sensitivity than peritoneal washing cytology and might help predict peritoneal dissemination in patients with PDAC.

Peritoneal washing cytology is an indicator of poor prognosis and a predictor for peritoneal dissemination in patients with PDAC4. Recent research by Ariake et al. has highlighted the effectiveness of systemic chemotherapy as a first-line therapy in patients with positive peritoneal washing cytology, and conversion of the peritoneal washing cytology status from positive to negative was an indication of conversion surgery21. Despite its significance, the low sensitivity of peritoneal washing cytology continues to be a clinical issue. Suenaga et al. showed that peritoneal lavage tumor DNA, assessed using digital droplet PCR, was a predictive factor for peritoneal dissemination after surgery7. Peritoneal fluid might be a more suitable liquid sample for predicting peritoneal dissemination than peripheral blood, which is commonly used for liquid biopsies in PDAC8,22,23. In this study, the rate of recurrence of peritoneal dissemination was significantly higher in curatively resected patients who had a positive molecular diagnosis in both negative peritoneal washing cytology and absence of peritoneal dissemination and positive peritoneal washing cytology and absence of peritoneal dissemination cases. Molecular diagnosis showed potential to more accurately stratify the risk for peritoneal recurrence than peritoneal washing cytology, complementing the limitations of peritoneal washing cytology. Neo- and adjuvant therapy are well-known to be important for improving the prognosis of patients with PDAC24,25. As knowledge accumulates, molecular diagnosis may play a crucial role in guiding treatment decisions in the future. For example, patients with a positive molecular diagnosis may more likely benefit from tailored treatment plans, including more intense neo- and adjuvant regimens. Similarly, a longer duration of neo- and adjuvant therapy may be recommended more frequently for those with a positive molecular diagnosis than for those with negative diagnosis.

Highly sensitive analytical methods are required to detect low-variant allele frequency mutations in peritoneal floating-cell DNA19,20,26. PNA-directed PCR clamping, digital droplet PCR, and NGS with ultra-high coverage are available for the detection of mutations with low variant allele frequency (< 1%)14,26,27. NGS allows comprehensive analysis but is costly and time-consuming27,28. Digital droplet PCR has similar sensitivity to PNA-directed PCR, but PNA-directed PCR is likely to have the advantage of not requiring specialized equipment29. PNA-directed PCR clamping has increased sensitivity owing to the amplification of mutant alleles9,14,15,16,17 and may be favorable for detecting mutations with a low variant allele frequency because it is more simple, rapid, cost-effective, and has higher sensitivity than digital droplet PCR and NGS.

In patients who had KRAS mutations in peritoneal floating-cell DNA, the present study revealed 80% (n = 10/12) concordance rate between KRAS mutation detection in peritoneal-floating cell DNA and primary or metastatic tumors, consistent with a previous study that reported concordance between tissue and liquid biopsy results30,31. In the present study, a paired comparison of KRAS mutational patterns between peritoneal floating-cell DNA and tumor DNA showed several discordant patterns. In two patients, KRAS mutations were detected in peritoneal floating-cell DNA, despite the primary tumor being wild-type (patients #5 and #6). KRAS mutation patterns are heterogeneous11,32, and clonal evolution has been implicated in tumor progression33. Patient #6, who had postoperative peritoneal dissemination, had a positive molecular diagnosis. A positive molecular diagnosis is regarded as a precursor state for the recurrence of peritoneal dissemination attributed to clonal evolution32. Indeed, liquid biopsy studies have detected somatic mutations that were missed in paired tumor tissues34; therefore, liquid biopsy can potentially elucidate intra-tumor or special heterogeneity and capture genomic information28. The most prevalent discordant pattern observed was the presence of KRAS mutation in DNA extracted from FFPE specimens, whereas peritoneal-floating DNA showed wild-type KRAS. One possible explanation for this discrepancy is that the low abundance of tumor cells in IPWF resulted in KRAS mutations falling below the limit of detection.

In this study, we set an optimal Ct value threshold of 36.42 through ROC analysis of all PCR results. We employed KRAS mutant and wile-type controls from cell lines for PCR quality control. PCR analysis without PNA was performed for validation purposes. The broader applicability of this optimal cutoff necessitates the establishment of a standard reference. Moreover, in real-time PCR, combining positive and negative control DNA at varying concentrations can facilitate absolute quantitation34, although this requires further validation during future in vitro diagnostic development.

This study had some limitations. First, this was retrospective single-center study with a relatively small cohort, which limited the generalizability of the findings and restricted some statistical analyses due to the small sample size. Nevertheless, we calculated the statistical power retrospectively based on the comparison of Ct values between positive and negative KRAS mutation groups using Sanger sequencing and/or NGS for IPWF via the Student’s t-test. The statical power was 0.993, which indicated that this study was sufficiently powered to detect the intended effects. Second, PNA-directed PCR clamping analysis of IPWF is not effective in patients with wild-type KRAS. However, KRAS mutations were detected in 92–93% of cases in large NGS analyses11,35, with KRAS being most commonly mutated and the RAS isoform being exclusively mutated in PDAC. Wild-type KRAS is a possible reason why two patients (patients #20 and #23) were classified as having positive peritoneal washing cytology but had a negative molecular diagnosis in the present study. PNA-directed PCR clamping still has the advantage of simplicity and cost, making it a good method for screening, but false negatives should be noted. Third, as the presence of non-cancerous cells in IPWF may affect PNA-directed PCR clamping to determine variant allele frequency, assessment of the variability and amount of component cells is crucial. At our hospital, a qualitative evaluation of cellular components in IPWF has been conducted, including neutrophils, eosinophils, lymphocytes, histiocytes, erythrocytes, and mesothelial cells. The amount of tumor cells in the IPWF was not quantified, but the amount of total DNA from IPWF can be referenced as we have shown in this study. A previous study has shown that the detection of tumor mutations may be influenced by the tumor volume present in IPWF, with an optical cutoff of 2% on the NGS platform36. However, the estimated tumor volume in IPWF from patients who underwent curative-intent surgery was found to be almost less than 1%26. In our study, PNA-directed PCR clamping could detect low variant allele frequency mutations at 0.29%. Fourth, the observation period was insufficient to allow for the evaluation of the association between molecular diagnosis and prognosis; this will have to be addressed in future studies with longer observation periods.

In conclusion, molecular diagnosis using PNA-directed PCR clamping for detecting KRAS mutations in peritoneal floating-cell DNA is more sensitive than peritoneal washing cytology. Molecular diagnosis may be useful for proper risk stratification for peritoneal dissemination recurrence.

Methods

Patients and data collection

From January 2020 to December 2021, consecutive patients were enrolled in this study. All patients were histologically diagnosed with PDAC. All eligible patients provided written informed consent prior to participation. Patients diagnosed with gallbladder cholesterol polyps (n = 6) or cholangiectasis (n = 1) were included as negative controls. The study was approved by the Human Experimentation Committee of Keio University Hospital (No. 20120443 and 20,170,086) and conducted in accordance with the Declaration of Helsinki.

Patient follow-up

The patients were followed up on outpatient basis every 3 months after surgery after discharge from the hospital. Clinical examinations and laboratory investigations were performed at each visit, and MDCT scans were performed every three to six months at the direction of the attending outpatient physician. If recurrence was suspected during postoperative follow-up, a multidisciplinary medical team consisting of hepatobiliary-pancreatic surgeons, internal medicine specialists, and radiologists confirmed the recurrence. Locoregional recurrence was defined as recurrence in the remnant pancreas or in the surgical bed, such as soft tissue along the celiac or superior mesenteric artery or around the pancreatojejunostomy site. Distant recurrence was stratified into three different categories: liver, lung, and peritoneal recurrence. Liver, lung, and peritoneal metastasis was defined as a mass detected on computed tomography during the postoperative follow-up. When considering patterns of recurrence, only the first site was documented in this study as indicated in Table 2a,b. Some patients had multiple metastasis. If recurrence was suspected on radiological examination, the histological confirmation was not mandatory.

IPWF collection

IPWF was collected immediately after laparotomy. One hundred milliliters of isotonic saline was introduced into the pelvic cavity. After gentle agitation, as much fluid as possible was collected from the pouch of Douglas using a suction tube.

IPWF cytology

Fifty milliliters of IPWF was used for pathological examination. Peritoneal washing cytology was classified as follows: normal or benign, indeterminate, suspicious for malignancy, or malignant. Positive peritoneal washing cytology was defined as a diagnosis of malignant peritoneal washing cytology. Patients with a peritoneal nodule with histologically proven adenocarcinoma were diagnosed as having peritoneal dissemination.

DNA extraction from IPWF

Fifty milliliters of IPWF was transferred into a sterile conical tube and centrifuged at 800 × g at 4°C for 10 min, immediately after collection. The pellet was resuspended in PBS and centrifuged once more at 800 × g at 4 °C for 10 min. The pellet, containing the cellular component of the IPWF, was stored at – 80 °C until use. Immediately before DNA extraction, the pellet was centrifuged at 10,000 × g at 4°C for 5 min. Genomic DNA was extracted from IPWF using DNA-binding columns from the QIAamp DNA FFPE Tissue Kit (Qiagen, Hilden, Germany), omitting the deparaffinization step from the manufacturer’s instructions9,12.

DNA extraction from FFPE specimens

Genomic DNA was extracted from FFPE specimens and purified using DNA-binding columns from the QIAamp DNA FFPE Tissue Kit (Qiagen, Hilden, Germany) according to the manufacturer’s instructions9,12. For FFPE specimens, DNA was in principle extracted from the primary tumor; however, in cases where the primary tumor was not resected, it was extracted from the resected metastases as indicated in Table 3. The extracted genomic DNA was stored at –80°C.

Measurement of DNA concentration

The genomic DNA was quantitated using the TaqMan RNase P Detection Reagents Kit (Applied Biosystems, Carlsbad, CA, USA)9 and a ViiA 7 Real-Time PCR System (Applied Biosystems). Each sample was assayed in duplicate, and the concentration was calculated using a calibration standard.

Cell lines

Two human pancreatic adenocarcinoma cell lines, AsPC-1 and BxPC-3, were obtained from the American Tissue Culture Collection (Rockville, MD, USA). The cells were maintained in RPMI 1640 medium supplemented with 10% heat-inactivated FBS (Thermo Fisher Scientific K.K., Tokyo, Japan) in a humidified 5% (v/v) CO2 incubator at 37°C. Genomic DNA was extracted from cells using the conventional phenol–chloroform method.

PNA-directed PCR clamping

PCR amplification was performed in 25 μL reaction mixtures containing 2.0 ng genomic DNA, as previously described9. qPCRs were run using the following KRAS primer pair: 5′-GGCCTGCTGAAAATGA-3′ (forward) and 5′-AAGGCACTCTTGCCTA-3′ (reverse). The sequences of the fluorescent resonance energy transfer probe and PNA were 5′-FAM-AGCTCCAACTACCACAAGTTTATATTC-BHQ-1–3′ and 5′-TACGCCACCAGCTCC-3′, respectively. PNA-directed PCR clamping was performed by measuring the Ct value. To maintain constant PCR accuracy, 2 ng of mutant KRAS, G12D (AsPC-1), and wild-type KRAS (BxPC-3) control were used every time PCR was performed, and the Ct value of the mutant KRAS control was maintained at 29.2 ± 0.1 and that of the wild-type KRAS at 37.5 ± 0.2. In addition, each sample was also assayed without PNA for input control, and the Ct value was maintained at 29.3 ± 0.1.

Amplicon sequencing for the sanger method

To detect low variant allele frequency mutations in IPWF sample using the Sanger method, PNA-directed PCR clamping amplification was performed using TaKaRa Ex Primer DNA Polymerase Dye plus (TaKaRa Bio, Osaka, Japan) in 50 μL reaction mixtures containing 1.75 μM of PNA, 1 μM of forward and reverse primers, and 10.0 ng of genomic DNA on a T100 Thermal Cycler (Bio-Rad Laboratories, Hercules, CA, USA). The thermal cycles were as follows: 30 cycles of 98°C for 10 s and 55°C for 15 s, and a final extension at 68°C for 10 s. PCRs were repeated twice using 0.25 μL of the previous PCR product. The amplification products were confirmed using electrophoresis and purified using the QIAquick PCR Purification Kit (Qiagen) according to the manufacturer’s instructions. Sanger sequencing was outsourced to FASMAC (Kanagawa, Japan).

NGS

NGS of peritoneal floating-cell DNA targeting 160 cancer-related genes, including KRAS mutations in PDAC37, was performed at the Genomics Unit using the PleSSision testing platform (Keio University Hospital, Tokyo, Japan).

Determination of the detection limit of PNA-directed PCR clamping

A DNA mixture derived from AsPC-1 (mutant KRAS, G12D) and BxPC-3 (wild-type KRAS) cells was diluted with distilled water (molecular biology grade) at different KRAS mutation ratios: 100%, 50%, 10%, 5%, 1%, 0.1%, 0.01%, and 0%. PNA-directed PCR clamping was carried out in triplicate using 2 ng of DNA, and the experiment was repeated four times. We plotted Ct vs. the logarithm of the mutation allelic fraction in the dilution series and calculated the parameters A and B in the linear equation Ct = A × logMAF + B.

Statistical analyses

Categorical variables are reported as frequency and percentage. Continuous data are reported as median and range. Continuous variables were compared using the t-test and categorical variables were compared using the chi-square test. The Ct cut-off value for positive biological peritoneal lavage was defined based on an ROC curve. Statistical significance was set at P < 0.05. All statistical analyses were performed using JMP 16 (SAS Institute Inc., Cary, NC, USA) and GraphPad Prism version 9.3.0 (GraphPad Software, San Diego, CA, USA).