KRAS Mutations in Duodenal Lavage Fluid After Secretin Stimulation for Detection of Pancreatic Cancer

Abstract

Objective:

Although pancreatic ductal adenocarcinoma (PDAC) is still a devastating disease, the survival rate for surgically removed PDACs has significantly improved in recent years. Early detection is essential in managing PDAC.

Background:

The presence of KRAS mutations in PDAC leads to the initial genetic abnormality and offers a significant timeframe for identifying resectable PDACs. A minimally invasive and highly specific PDAC screening test is necessary to prevent the need for invasive follow-up tests.

Methods:

Between July 2021 and March 2023, 169 cases were enrolled in 7 institutions. By administering secretin before esophagogastroduodenoscopy (EGD), the excretion of pancreatic juice into the papillary fluid can be stimulated, creating a resource for testing. Washing fluid was collected using a specialized catheter from control individuals (n=75) and patients with resectable PDAC (n=89) at the initial diagnosis. A highly sensitive technique was employed to study KRAS gene mutations.

Results:

This study obtained an AUC of 0.934 (95% CI: 0.904, 0.964) when using KRAS mutations in duodenal lavage fluid to differentiate between patients with resectable PDAC and healthy controls. The estimated sensitivities were calculated with specificity set at 100%, resulting in a sensitivity of 83.1% (95% CI: 71.7%, 91.2%). The McNemer test showed a significantly higher sensitivity for KRAS mutations than serum CEA and CA19-9 (P<0.0001).

Conclusions:

We created a method to identify resectable PDACs by analyzing KRAS mutation levels in duodenal fluid collected during EGD with secretin stimulation of pancreatic juice secretion.

Pancreatic ductal adenocarcinoma (PDAC) remains one of the most devastating cancers due to its silent nature and the lack of screening tests for the detection of the disease in its early stages.¹ Due to limited sensitivity and specificity, tumor markers (eg, serum CA19-9) are not effective for detecting resectable stage PDACs, so the search for better indicators continues.² Regarding CA19-9, about 5% to 10% of the Japanese population has the Lewis-antigen negative phenotype, meaning they lack the enzyme needed for antigen production.³ Detailed tests, such as endoscopic retrograde cholangiopancreatography (ERCP), can be highly invasive if a screening test is positive, and the United States Preventive Services Taskforce (USPSTF) stated in 2020 that current PDAC screening tests are more harmful than beneficial due to false positives.⁴ We need a minimally invasive PDAC screening test that is highly sensitive and has near-perfect specificity. The 5-year survival rate for resected PDACs has dramatically improved in recent years due to postoperative adjuvant chemotherapy.⁵ Nevertheless, even the latest drugs can only prolong the prognosis by half a year in most cases.⁶ Detecting just a fraction of latent PDACs at an operable stage would make a significant difference.

We have reported that there are broad opportunities for early detection of PDACs to prevent deaths from metastatic disease.^7–9 The high occurrence of KRAS mutations found early in ∼93% of cases,¹⁰ and the more limited mutation patterns than with other cancers point to a most promising biomarker candidate. However, only about 20% of patients with resectable PDACs exhibit KRAS mutations that can be detected in cell-free DNA derived from peripheral blood.^11,12 PDAC is a cancer that begins in cells of the peripheral pancreatic ducts¹³ and involves the main pancreatic duct at an early stage.^14,15 Concentrations of cancer cells and DNA are, therefore, higher in the pancreatic duct and near the papilla of Vater than in the blood.¹⁶

Gastric cancer is one of the world’s most common epithelial malignancies¹⁷ ranking second in Japan, where the Ministry of Health, Labor and Welfare (MHLW) guidelines advise biennial gastric cancer screening through esophagogastroduodenoscopy (EGD).¹⁸ This recommendation is also in accordance with the Korean Practice Guidelines for Gastric Cancer.¹⁹ For this group of screenees, we have developed a revolutionary practical method for detecting PDAC (Video 1) involving the administration of synthetic human secretin to stimulate pancreatic juice secretion^20,21 before EGD. The ampulla of Vater is then washed with a saline solution, which is collected (Fig. 1A–C) for identification of KRAS mutations achieved with high-sensitivity mutation analysis technology.²² This study analyzed whether this test can accurately detect PDAC during EGD without any false positives.

FIGURE 1.

Technique and new catheter for this examination. A, Before the esophagogastroduodenoscopy (EGD) procedure, a sedative and synthetic human secretin are administered intravenously. The esophagus, stomach, and duodenal bulb are observed through EGD, and the ampulla of Vater is identified. The specialized catheter is then inserted through the endoscopic channel. B, The ampulla of Vater is flushed with a 20 mL saline solution via the side hole of the specialized catheter, and the fluid is collected through the holes at the tip from the duodenum and stimulated ampulla of Vater (Video 1). C, The contrast catheter of a standard EGD has only one hole at the tip. Aspiration applies pressure to the hole at the tip, which hinders the collection of sufficient duodenal lavage solution. However, the new specialized catheter features a dual-layer design. The washing hole is located horizontally and specifically designed to facilitate the cleansing of the ampulla of Vater. The 6 suction holes at the tip serve to decrease pressure and facilitate the collection of duodenal fluid and ampulla of Vater wash solution without absorption of the duodenal mucosal wall. The new catheter is now being introduced to the market by SB-KAWASUMI Laboratories Inc. (Tokyo, Japan). The blue arrow symbolizes the movement of the saline solution. Orange arrows are used to represent the direction of fluid flow during duodenal lavage.

VIDEO 1.

Download video file ^{(83.8MB, mp4)}

Procedure for cleansing the duodenal papilla and collecting the cleansing fluid. 1. Preparation and Positioning: Advance the tip of the esophagogastroduodenoscope to the duodenal papilla. 2. Catheter Insertion: Insert a specialized catheter through the channel of the endoscope and extend it beyond the tip of the esophagogastroduodenoscope. 3. Cleansing Process: Identify the catheter's cleansing hole (side hole), marked in black, and cleanse the duodenal papilla using 20 mL of saline solution introduced through the side hole. 4. Fluid Collection: Collect the cleansing fluid through the six holes located at the tip of the catheter. 5. Observation: Synthetic human secretin facilitates the flow of bile-containing pancreatic juice from the duodenal papilla, which can be observed during the procedure. The entire process is completed within approximately 1 to 2 minutes.

METHODS

The study was conducted in line with the “Clinical Trials Act” (jRCTs05120116, jRCTs05120117) as it involved the use of synthetic human secretin (ChiRhoStim, ChiRhoClin Inc., Burtonsville, MD), which is not approved in Japan under the Pharmaceutical and Medical Devices Law. The study obtained samples and clinical information with written informed consent and approval from the Certified Review Board at Osaka University Hospital (Osaka University Clinical Research Review Committee, CRB5180007).

Study Design and Patients

Seven institutions participated in this study to evaluate diagnostic performance (Supplemental Table S1, Supplemental Digital Content 1, http://links.lww.com/SLA/F396).

The inclusion criteria were defined as follows:

1. Provision of informed consent and over 20 years of age.

2. Diagnosis or suspicion of PDAC.

3. Control: individuals who had provided written consent to participate in this research.

The criteria for exclusion were as follows:

• Hypersensitivity to secretin or any other agents, specifically cysteine hydrochloride and mannitol.

• A history of remission from acute pancreatitis or acute exacerbation of chronic pancreatitis within 2 weeks of starting the study (synthetic human secretin [ChiRhoStim]).

• Use of anticholinergic medications within at least 5 half-lives before starting the study drug.

• Pregnancy or a lactating state.

• Receipt of alternative research drugs within 3 months before the study drug.

• Unsuitability as determined by the attending physician.

The primary outcome was the sensitivity achieved at a specificity of 100% in resectable PDAC. Secondary outcomes were amounts of duodenal juice collected and any adverse events within a day of taking the study drug.

Procedures

A synthetic secretin injection of 0.2 µg/kg was administered over 1 minute before an EGD to stimulate pancreatic juice secretion. The esophagus, stomach, and duodenal bulb were examined. Then, the ampulla of Vater, the exit of the main pancreatic duct, was flushed with a saline solution collected using a specialized catheter called an S&Y tube (see below, Figs. 1B, C). Inspection time was ∼1 to 2 minutes from washing to collection (Video 1). Detection of KRAS gene mutations was accomplished by BNA clamp Taqman PCR.

Specialized Catheter

Traditional contrast catheters only have one hole at the tip, which can become blocked with duodenal mucosa, hindering suction recovery. Hence, the development of a specialized catheter (S&Y tube) was undertaken (Fig. 1C). At a distance of 3 cm from the catheter’s tip, there is a black marker with a side hole to introduce the washing fluid into the duodenal papillary area. To ensure consistent suction pressure and optimize the collection of duodenal lavage fluid, 6 suction holes are positioned near the tip. The fluid obtained is automatically collected into a 50 mL suction trap conical tube positioned between the EGD and the suction tube. The catheter received authorization for manufacture and marketing in Japan as a general medical device (05B1X00002000034) in August 2020.

Highly Sensitive Mutation Analysis

BNA clamp PCR is a technique that prevents the amplification of normal DNA, allowing the identification of mutants involving 4 mutations (G12D, G12V, G12A, and G12C) in KRAS codon 12 in duodenal lavage fluid. The procedure was executed using a Roche Light Cycler 480 II (Thermo Fisher Scientific Inc., Waltham, MA) with a reaction mixture volume of 40 µL, containing 0.5 µM forward primer, 0.5 µM reverse primer, 0.125 µM BNA clamp, 0.25 µM Taqman probes (Supplemental Table S2, Supplemental Digital Content 1, http://links.lww.com/SLA/F396), 2.5 mM MgCl₂, 50 mM NaCl, 10 mM Tris (pH 8.5) (NIPPON GENE Co. Ltd., Tokyo, Japan), 0.2 mM dNTP (Thermo Fisher Scientific Inc.), 0.2× ROX reference dye (Takara Bio Inc., Kusatsu, Japan), 2 units Uracil-DNA Glycosylase (New England Biolabs Inc., Ipswich, MA), 0.25 units OneTaq Hot Start DNA Polymerase (New England Biolabs Inc.) and 105 copies genomic DNA. The cycling conditions were as follows: 65 °C for 1 minute, followed by 94 °C for 30 seconds, and a total of 60 cycles at 94 °C for 20 seconds and 57 °C for 60 seconds. The threshold cycle (Ct) calculation involved using a threshold line set at 1.5. Correcting batch effects was achieved using identical positive and negative controls.

Duplex DNA Sequencing for Detection With High Sensitivity

The Covaris M200 system (Covaris LLC., Woburn, MA) was used to fragment cell-free DNA purified from duodenal lavage in microTUBE-50 (Covaris) with the settings: Peak incident power=75 W; duty factor=10; cycles per burst=200; treatment time=100 seconds. DNA libraries were prepared using the xGen cfDNA & FFPE DNA Library Prep v2 MC Kit [IDT (Integrated DNA Technologies Inc.), Coralville, IA] per the manufacturer’s instructions. The library preparation used 105 GE (genome equivalents) of cell-free DNA as input materials. NextSeq. 500 Mid Output (Illumina Inc., San Diego, CA) was used to sequence a pooled library of 9 samples at 1.8 pM concentration, generating 151 bp × 2 paired-end reads. Unique molecular identifiers were used for data processing, following the single-read families method as per IDT’s analysis guidelines.

Quantification of KRAS Mutations by Digital PCR (dPCR) Combined With Melting Curve Analysis

KRAS mutations were quantified using dPCR and melting curve analysis, as described in a previous report.²³ The sequences of primers and probes are provided in Supplemental Table S3, Supplemental Digital Content 1, http://links.lww.com/SLA/F396. The composition of the reaction solution used for measuring PCR and Tm in the 8-plex assay in the wells was as follows: 1 × QuantStudio 3D Digital PCR Master Mix v2 (Thermo Fisher Scientific Inc.), 0.25 μM KRAS forward primer, 2.0 μM KRAS reverse primer, 0.5 μM KRAS wild-type/G13D detection probe, 0.5 μM KRAS G12A detection probe, 0.5 μM KRAS G12R detection probe, 0.5 μM KRAS G12D detection probe, 0.5 μM KRAS G12V detection probe, 0.5 μM KRAS G12S detection probe, 0.5 μM KRAS G12C detection probe, 0.5 μM KRAS pseudogene blocker, and 60 ng of cfDNA in a final reaction volume of 15 μL. After adding 14.5 μL of PCR solution to a QuantStudio 3D Digital PCR 20K Chip (Thermo Fisher Scientific Inc.), PCR was conducted using a thermal cycler. The amplification procedure was conducted in the following manner: 10 minutes at 96 °C, followed by 60 cycles consisting of 30 seconds at 98 °C and 2 minutes at 60 °C. The chip was subjected to melting temperature measurement using a device following the PCR procedure. The wild-type and 7 mutants (G12D, G12R, G12V, G13D, G12A, G12C, and G12S) on the chip were quantified digitally, according to a previously published method.²³

Statistical Analysis

The baseline characteristics of PDAC patients were tabulated and compared with those of the controls. P values were obtained using the χ² test for categorical variables and the Wilcoxon rank-sum test for continuous variables. The distributions of KRAS mutation, serum CEA, and serum CA19-9 between PDAC patients and controls were compared using the Wilcoxon rank-sum test. The receiver operating characteristic (ROC) curve was plotted for three biomarkers (ie, KRAS mutations, serum CEA, and serum CA19-9), and the cut-off value was established at 100% specificity.

A comparison was made between the current method and the previous one, which did not involve using a specialized catheter and synthetic human secretin for KRAS mutations. The comparison focused on sensitivity, using a cut-off value of 100% specificity and referring to their 2-sided 95% Pearson Clopper CIs. In addition, a Bayesian comparison was conducted to determine the posterior probability of the current method’s superiority over its predecessor. This was done using the beta-binomial distribution with a non-informative prior distribution Beta (1,1). The diagnostic capacity of the current method was compared with that of the previous methods using the AUC, with an estimated AUC of 0.72 for the previous method. The rationale for determining the sample size was as follows: our objective was to achieve a power of 0.9, ensuring statistically significant superiority in sensitivity for the current method while maintaining a specificity of 100%. The KRAS mutations, CEA, and CA19-9 diagnostic performance are evaluated by examining sensitivity, utilizing a cut-off of 100% specificity, and the area under the ROC curve (AUC). Sensitivity is assessed using the McNemar test, and the Delong test is employed to compare AUC.

Dr. Yachida had full access to all the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis.

RESULTS

Enrollees in this Prospective Study

A total of 169 cases were enrolled in 7 institutions between July 15, 2021, and March 15, 2023 (Fig. 2 and Supplemental Table S1, Supplemental Digital Content 1, http://links.lww.com/SLA/F396). The data for this study was gathered using a case-control sampling strategy: we included the maximum number of patients eligible for surgery for PDAC, while we explicitly selected controls from individuals who did not have PDAC or cancer in other organs at each institution. The present study design was determined by a medical statistician (S.H.), who determined the required number of cases. Only one case of general fatigue post-examination was identified as an adverse event with a potential link to EGD. Among the controls, 6 had chronic pancreatitis (Group 1B), 7 autoimmune pancreatitis (Group 1C), and 16 had intraductal papillary mucinous neoplasm (IPMN, Group 2). In addition, 46 healthy controls (HC) exhibiting a normal pancreas were included (Group 1A).

FIGURE 2.

Subject details. A total of 169 cases were enrolled; 75 were controls at enrollment (without PDAC or carcinoma in other organs), 89 were PDAC patients diagnosed as operable at the initial visit, and the remaining 5 individuals withdrew their consent after registration, rendering them ineligible for testing (Supplemental Table S1, Supplemental Digital Content 1, http://links.lww.com/SLA/F396). Controls with IPMN were classified as Group 2. Within those without neoplastic lesions in the pancreas, we observed chronic pancreatitis in 6 (Group 1B) and autoimmune pancreatitis (AIP) in 7 individuals (Group 1C), leaving a total of 46 participants with a normal pancreas (Group 1A). Among the initial visit cohort of 89 patients diagnosed with PDAC eligible for surgery, a total of 70 patients (78.7%) underwent preoperative chemotherapy. Subsequently, curative surgery was performed on 65 out of the total 89 patients, constituting 73.0% of the PDAC cases. The remaining 24 patients underwent disease progression during preoperative chemotherapy or comprehensive examination, thus making them unsuitable candidates for surgery.

Out of the patients initially diagnosed with resectable PDAC, 24 individuals were unable to undergo surgery due to disease progression during preoperative chemotherapy or comprehensive examination (Group 4). A total of 65 patients underwent radical surgery (Group 3).

Table 1 and Supplemental Tables S1, Supplemental Digital Content 1, http://links.lww.com/SLA/F396 present clinical information for each Group, including duodenal lavage fluid volume, KRAS mutations, and serum CEA and CA19-9 concentrations. Considering that the study’s enrollment criteria necessitated control participants to be men and women aged 20 years or older, that is, not age-matched to PDAC patients, it is important to note that Group 1 displayed a statistically significant younger age compared with Groups 3+4. The number of KRAS mutations in duodenal lavage fluid, as well as serum CEA and CA19-9 levels, showed a significant increase in PDAC patients (Groups 3+4).

TABLE 1.

Characteristics of Participants

Classification	Group 1A (n=46)	Group 1B (n=6)	Group 1C (n=7)	Group 2 (n=16)	Group 3 (n=65)	Group 4 (n=24)	Group 1 vs. 3+4 P
Sex
Male	25 (54%)	4 (33%)	4 (57%)	9 (56%)	33 (51%)	13 (54%)	0.6119
Female	21 (46%)	2 (67%)	3 (43%)	7 (44%)	32 (49%)	11 (46%)
Age
Median, yr	66	59	70	70	71	69	0.0022
Range, yr	35–83	44–82	49–80	44–85	50–84	54–81
Volume of duodenal juice
Median, mL	15.68	17.44	17.82	18.20	19.24	18.99	0.5468
Range, mL	5.93–43.68	5.55–23.91	6.26–27.26	7.42–36.52	3.69–37.92	10.5–34.84
No. mutations in KRAS 100,000 copies
Median	21	20	25	320.5	376	232.5	<0.0001
Range	0–92	0–66	0–75	2–6756	0–63535	0–4366
CEA (ng/mL)
Median	1.95	1.4	1.6	2.9	2.8	3.6	0.0111
Range	0.5–8.1	1.1–6.3	0.7–8.2	1.2–5.8	0–15.9	1.2–39.6
CA19-9 (U-/mL)
Median	9.5	12.65	11.0	14.0	79.2	181.0	0.0034
Range	0–378.0	0–27.3	0–51.2	0–438.0	0–4061.0	0–5241.0

Group 1A: controls at initial diagnosis (normal pancreas).

Group 1B: controls at initial diagnosis (chronic pancreatitis).

Group 1C: controls at initial diagnosis (AIP).

Group 2: controls at initial diagnosis (IPMN).

Group 3: patients with resectable PDAC at initial diagnosis (operation).

Group 4: patients with resectable PDAC at initial diagnosis (deterioration of disease stage).

AIP indicates autoimmune pancreatitis; IPMN, intraductal papillary mucinous neoplasm.

Figure 3 illustrates data for KRAS mutations in duodenal lavage, serum CEA, and serum CA19-9 based on Group: note that the Y axes for KRAS and serum CA19-9 utilize the logarithm base 10. Notably, none of the patients with chronic pancreatitis (Group 1B) or AIP (Group 1C) demonstrated KRAS mutations above the cut-off (92 copies). Among the 7 AIP patients, KRAS mutations were lower in all, while 2 had increased CEA levels, and 1 had increased CA19-9 above the baseline. This finding agrees with previous reports²⁴ and implies that high tumor marker levels may make diagnosing AIP more complicated.

FIGURE 3.

KRAS mutations and serum CEA and CA19-9 levels. A, Numbers of mutations per 100,000 copies of KRAS gene in duodenal lavage collection fluid, shown as distribution plots and box-and-whisker plots; note that the Y axis is the logarithm with base 10. The boxes represent 25th to 75th percentiles, and horizontal lines indicate the medians and whiskers extend to the maximum and minimum values within 1.5× the interquartile range. B, Concentrations of serum CEA levels are shown as distribution plots and box-and-whisker plots. C, Concentrations of serum CA19-9 levels, shown as distribution plots and box-and-whisker plots; note that the Y axis is the logarithm with base 10.

Collection of Duodenal Lavage Fluid

Duodenal fluid is generally acidic. In a preliminary assessment, the duodenal lavage fluid exhibited a pH value of 5.0 (with a range of 2.5–6.0, n=73) before the utilization of synthetic human secretin and a specialized catheter (Supplemental Fig. S1A, Supplemental Digital Content 1, http://links.lww.com/SLA/F396). The median volume of the fluid collected from the duodenal lavage was 8.5 mL, ranging from 2.0 to 21.5 mL (Supplemental Fig. S1B, Supplemental Digital Content 1, http://links.lww.com/SLA/F396). Secretin greatly enhances the secretion of alkaline pancreatic juice.²⁵ Consequently, the utilization of secretin led to a pH of 6.0 in most cases (94.5%), thereby leading to an improved recovery of DNA. In addition, there was a significant improvement in the median volume of duodenal lavage fluid, which measured 17.9 mL (range, 3.7–43.9 mL).

The Number of Mutated Copies With BNA Clamp Taqman PCR

The BNA (bridged nucleic acid) clamp PCR method,²⁶ developed by Sysmex Corporation (Kobe, Japan), has the capacity to discern one mutant allele from a pool of 10,000 copies (detection sensitivity: 0.01%). In the present study, 4 distinct patterns of KRAS mutations (ie, G12R, G12D, G12V, and G12C) were investigated in duodenal lavage fluid. To confirm the performance of the BNA clamp Taqman PCR method, 7 mutations (ie, G12R, G12D, G12V, G12A, G13D, G12S, and G12C) were also examined using high-precision digital PCR by Hitachi Ltd.²³ The mutant copy number in the digital PCR results was calculated by summation of the copy numbers of 4 specific mutant types. The results showed a strong correlation [Spearman rank correlation coefficient, 0.660 (95% CI: 0.530, 0.757); P<0.0001] between the 2 analysis methods (Supplemental Fig. S2, Supplemental Digital Content 1, http://links.lww.com/SLA/F396).

Initially, a comparison was made between 65 patients who underwent radical surgery (Group 3) and healthy controls with normal pancreas (Group 1A) (Table 2). The quantity of duodenal lavage fluid collected did not show any variation. Group 3 exhibited statistically significant rises in KRAS mutations within the duodenal lavage fluid (P<0.0001), serum CEA (P=0.0453), and serum CA19-9 (P=0.0222). Construction of the receiver operating characteristic (ROC) curves and determination of AUC were performed [AUC: KRAS gene mutation, 0.934 (95% CI: 0.904, 0.964); CEA, 0.666 (95% CI: 0.609, 0.723); CA19-9, 0.767 (95% CI: 0.717, 0.818)] (Fig. 4).

TABLE 2.

Comparison of Clinical Information and Tumor Markers Conducted Between Group 1A (Healthy Controls With Normal Pancreas) and Group 3 (Patients Undergoing a Curative Operation for PDAC)

Classification	Group 1A (n=46)	Group 3 (n=65)	Group 1A vs 3 P
Sex
Male	25 (54%)	33 (51%)	0.7100
Female	21 (46%)	32 (49%)
Age
Median, yr	66	71	0.0014
Range, yr	35–83	50–84
Volume of duodenal juice
Median, mL	15.68	19.24	0.8546
Range, mL	5.93–43.68	3.69–37.92
No. mutations in KRAS 100,000 copies
Median	21	376	<0.0001
Range	0–92	0–63,535
CEA (ng/mL)
Median	1.95	2.8	0.0453
Range	0.5–8.1	0–15.9
CA19-9 (U/mL)
Median	9.5	79.2	0.0222
Range	0–378.0	0–4061.0

Category 1A: controls with a normal pancreas at initial diagnosis.

Category 3: patients with resectable PDAC at initial diagnosis (operation).

FIGURE 4.

Diagnostic significance of KRAS mutations in duodenal lavage fluid and serum CEA and CA19-9 levels in patients undergoing surgery. A, Calculation of an AUC [area under the ROC (receiver operating characteristic) curve] for KRAS mutations in duodenal lavage fluid to distinguish between patients with resectable PDAC (Group 3) and healthy controls (Group 1A) resulted in a value of 0.934 (95% CI: 0.904, 0.964). The determination provided an optimal cut-off value of 92 mutations per 100,000 KRAS gene copies. The specificity and sensitivity of positive KRAS mutations in the duodenal lavage fluid were found to be 100% and 83.1%, respectively. B, The AUC value for serum CEA was determined to be 0.666 (95% CI: 0.609, 0.723). With the optimal value set at 2.3 ng/mL as the cut-off, serum CEA exhibited a specificity of 67.6% and a sensitivity of 63.5%. The optimal cut-off value was determined as the highest point of the Youden index, as indicated by the red asterisk. C, Serum CA19-9 exhibited an AUC of 0.767 (95% CI: 0.716, 0.818]. With the optimal threshold established at 33 U/mL, the specificity and sensitivity values were 94.4% and 66.7%, respectively.

Achieving a specificity rate of 100% in the PDAC screening test is essential to avoid invasive and extensive testing. The estimation of sensitivities at the cut-off values and their 2-sided Pearson-Clopper CIs was conducted with specificity set at 100%. Accordingly, the determined cut-off value for KRAS was set at 92 per 100,000 copies, displaying a sensitivity of 83.1% (95% CI: 71.7%, 91.2%). The prescribed cut-off points at 100% specificity for CEA and CA19-9 were 8.2 ng/mL and 378 U/mL, respectively. The sensitivity rates were 10.8% (95% CI: 4.4%, 20.9%) for CEA and 29.2% (95% CI: 18.6%, 41.8%) for CA19-9. McNemar test was performed, assigning a range value of 100% to the specificity of each test. The data presented a statistically significant advantage in sensitivity for KRAS mutations in duodenal lavage fluid (P<0.0001) (Supplemental Fig. S3, Supplemental Digital Content 1, http://links.lww.com/SLA/F396). The AUC for PDAC patients with KRAS mutation was determined to be 0.934 (95% CI: 0.904, 0.964), signifying a notable improvement relative to the AUC of the previous method, which did not incorporate specialized catheters and synthetic human secretin, and yielded 0.72 (unpublished data). The preceding approach exhibited a sensitivity of 46% (95% CI: 27%, 79%). The current method exhibited a significantly higher sensitivity than the previous method, with a small overlap in the CIs. The Bayesian posterior probability of the current method’s superiority over the previous method has been computed as 0.997. The P values for the pairwise comparison between AUC and KRAS mutations were statistically significant: P<0.0001 for CEA and P=0.0030 for CA19-9.

Given that the ultimate goal of this assay is to detect early-stage PDAC that is potentially curable, we performed a comparison between the pathologic stage (pStage) 0+I group (n=41) and the control group (Group 1A, n=46). The pStage 0+I group exhibited higher levels of KRAS mutations (P<0.0001, Wilcoxon test), serum CEA (P=0.0411), and CA19-9 (P=0.0051) (Figs. 5A–C). ROC analysis and Youden’s index were used to evaluate the performance and cut-off values, yielding an AUC of 0.912 (95% CI: 0.859, 0.961) for KRAS mutations, 0.629 (95% CI: 0.543, 0.715) for CEA, and 0.701 (95% CI: 0.619, 0.783) for CA19-9 (Figs. 5D–F). The cut-off values were determined to be 92 per 100,000 copies for KRAS mutations, 2.2 ng/mL for CEA, and 27.4 U/mL for CA19-9. At these cut-off values, specificity and sensitivity were 100% (95% CI: 92.3%, 100%) and 80.9% (95% CI: 65.1%, 91.2%) for KRAS mutations, 67.6% (95% CI: 50.8%, 80.9%) and 61.0% (95% CI: 45.7%, 74.3%) for CEA, and 94.4% (95% CI: 81.6%, 98.5%) and 56.1% (95% CI: 41.0%, 70.1%) for CA19-9, respectively. These findings are consistent with the results of the earlier comparison between Group 3 and the control group (Group 1A), suggesting that this assay is a promising tool for the detection of early-stage PDAC. The correlation between clinical stage (cStage) 0+I, as determined by pre-treatment imaging, and KRAS mutations in duodenal lavage fluid, as well as serum CEA and CA19-9 levels, was also evaluated (Supplemental Fig. S4, Supplemental Digital Content 1, http://links.lww.com/SLA/F396). The results are similar to those observed at pStage.

FIGURE 5.

Comparative analysis of KRAS mutations in duodenal lavage fluid, serum CEA, and CA19-9 levels between patients diagnosed with pathologic stage (pStage) 0+I PDAC (n=41) and healthy controls (HC, Group 1A) to assess diagnostic value. A–C, A comparison between the 2 groups revealed significant differences in KRAS mutations (P<0.0001), serum CEA levels (P=0.0441), and CA19-9 levels (P=0.0051). D–F, ROC curve analysis and determination of cut-off values were conducted. For KRAS mutations, a threshold of 92 mutations per 100,000 copies of the KRAS gene was identified. For serum CEA, a cut-off value of 2.3 ng/mL was adopted, and for CA19-9, a threshold of 30 U/mL was used, based on the highest point of the Youden index (indicated by the red asterisk). At these cut-off values: ● For KRAS mutations, specificity was 100% (95% CI: 92.3%, 100%) and sensitivity was 80.9% (95% CI: 65.1%, 91.2%). ● For CEA, specificity was 67.5% (95% CI: 50.8%, 80.9%) and sensitivity was 65.3% (95% CI: 54.1%, 75.1%). ● For CA19-9, specificity was 94.4% (95% CI: 81.6%, 98.5%) and sensitivity was 56.1% (95%CI: 41.0%, 70.1%).

No significant differences were observed in KRAS mutations in duodenal lavage fluid (P=0.6483) or serum CEA levels (P=0.9530) between the pStage groups. Conversely, serum CA19-9 levels were significantly higher (P=0.0406) in pStage IIA+IIB cases compared with pStage IA+IB cases (Supplemental Fig. S5, Supplemental Digital Content 1, http://links.lww.com/SLA/F396 and Table 3). We expected a low number of KRAS mutations in the pancreatic tail due to its distance from the ampulla of Vater. The absence of a relationship between tumor location and these 3 markers is depicted in Supplemental Figure S6, Supplemental Digital Content 1, http://links.lww.com/SLA/F396. The number of KRAS mutations in the duodenal wash collection fluid showed a statistically significant increase with the increasing size of PDAC (P=0.0083, Kruskal-Wallis test) (Supplemental Fig. S7, Supplemental Digital Content 1, http://links.lww.com/SLA/F396). KRAS mutations were identified above the specified threshold in 6 of 10 patients with small PDAC measuring <10 mm.

TABLE 3.

Correlations Between Pathologic Findings and KRAS Mutations, Serum CEA, and CA19-9 in Patients With Resected PDACs

		No. mutations in KRAS 100,000 copies		CEA (ng/mL)		CA19-9 (U/mL)
	No. patients (%)	Median	Range	Median	Range	Median	Range
cStage (n=89)
0	1 (1.1%)	61	61	2.8	2.8	3.3	3.3
IA + IB	75 (83.3%)	320	0–63,535	2.9	0–39.6	81	0–4061
IIA +IIB	13 (14.6%)	5845	30–4366	4.7	1.2–10.3	188	0–5241
pStage (n=65)
0	1 (1.5%)	61	61	2.8	2.8	3.3	3.3
IA + IB	40 (61.5%)	354.5	0–63,535	2.5	0–13.1	59.45	0–2672
IIA +IIB	23 (35.4%)	529	30–4679	2.8	1.2–13.5	115	0–4061
III	1 (1.5 %)	1280	1280	15.9	15.9	750	750
Location (n=65)
Head	34 (52.3%)	375.5	4–63,535	2.5	0.2–11.6	98.1	0–2672
Body	20 (30.8%)	375.5	0–4679	2.7	0–15.9	64.45	0–3538
Tail	11 (16.9%)	422	61–2572	3.5	1.2–13.1	14	0-4061
Tumor size (mm) (n=65)
≤10	10 (15.4%)	141	23–873	2.5	0.8–6.1	47	0–1163
11-20	21 (32.3%)	226	21–4242	1.9	0–5.5	44	0–4061
21-30	19 (29.2%)	520	0–63,535	3.7	1.3–13.1	76.9	0–2672
≥31	15 (23.1%)	772	138–4679	3.0	1.2–15.9	115	0–3538
Histology (n=65)
High-grade PanIN	2 (3.1%)	257	61–453	3.0	1.5–4.5	3	0–6
Ductaladenocarcinoma	61 (93.8%)	350	0–4679	2.7	0–15.9	81	0–4061
Adenosquamouscarcinoma	2 (3.1%)	32,832	2128–63,535	3.75	3.4–4.1	219.5	17–422

PanIN indicates pancreatic intraepithelial neoplasia.

Subsequently, we examined 2 different sets of patients: Group 1, those without pancreatic tumors (ie, PDAC and IPMN), and Group 3 + Group 4, with PDACs. The AUC for PDAC patients with KRAS mutation was 0.909 (95% CI: 0.873, 0.945), with a specificity of 100% and a sensitivity of 79.8% (Supplemental Fig. S8A, Supplemental Digital Content 1, http://links.lww.com/SLA/F396). The AUC for CEA and CA19-9 were 0.681 (95% CI: 0.623, 0.739) and 0.788 (95% CI: 0.737, 0.839), respectively (Supplemental Figs. S8B and S8C, Supplemental Digital Content 1, http://links.lww.com/SLA/F396). The AUC derived from the KRAS mutations detected in the recovery fluid of the duodenal wash demonstrated a statistically significant superiority over the AUC determined using CEA (P<0.0001) and CA19-9 (P=0.0030). As stated before, the McNemar test was performed, assigning a range value of 100% to the specificity of each test. The data presented a statistically significant advantage in sensitivity for KRAS mutations in duodenal lavage fluid (P<0.0001) (Supplemental Fig. S9, Supplemental Digital Content 1, http://links.lww.com/SLA/F396).

IPMN Patients

Of the IPMN patients (Group 2), 9 were diagnosed as low-grade and followed up (Supplemental Fig. S10A, Supplemental Digital Content 1, http://links.lww.com/SLA/F396). Seven were suspected to be malignant and underwent radical surgery. The excised specimens indicated 5 patients had IPMN with associated invasive carcinoma (invasive IPMN), and 2 had high-grade IPMN. All 7 patients requiring surgical intervention were classified as mixed duct-type IPMN (Supplemental Fig. S10B, Supplemental Digital Content 1, http://links.lww.com/SLA/F396). Subclassification based on histopathologic diagnoses of the surgical specimens revealed the following: 5 cases were pancreatobiliary type (all invasive IPMN) and 2 cases were gastric-type (both high-grade IPMN). The number of KRAS mutations in duodenal lavage collection fluid (P=0.0453) was significantly higher in invasive IPMN compared with low-grade IPMN on follow-up (Supplemental Fig. S10C, Supplemental Digital Content 1, http://links.lww.com/SLA/F396). Conversely, no statistically significant differences were observed in serum CEA (P=0.3785) or CA19-9 (P=0.1871) levels between low-grade and invasive IPMN (Supplemental Figs. S10D and S10E, Supplemental Digital Content 1, http://links.lww.com/SLA/F396).

Duplex Sequencing

KRAS mutations were unexpectedly found in the duodenal lavage fluid of controls, albeit at a low frequency. We used Duplex sequencing,²⁷ a sensitive next-generation sequencing technology, to study the factors involved. The chosen cases included PDAC cases and controls, both near and below the cut-off level for KRAS mutations. Out of the nine cases studied, four cases had 58 to 96 mutations per 100,000 KRAS copies using BNA clamp PCR (Fig. 6). Three cases, including 2 with PDAC, had a mutation rate of 0.05% or higher in codon 12 with Duplex sequencing. In cases where the BNA clamp PCR detected <41 KRAS copies per 100,000, a 0.02% to 0.05% mutation rate was observed at codon 12 through Duplex sequencing. In addition, there is a possibility of rare mutations happening inside and outside of codons 12 and 13, which might be susceptible to selection pressure. The results indicated a clonal expansion of KRAS mutations in minimal quantities within the duodenal lavage fluid, observed not only in patients with PDAC but also in the control group of healthy individuals.

FIGURE 6.

Validation of KRAS mutations in duodenal wash recovery solution by Duplex sequencing. A Duplex sequencing analysis was performed on cases with KRAS mutations in duodenal lavage using BNA clamp PCR. These cases were selected based on their threshold values, which were set at or below 92. Cases A and B belong to the resectable PDAC category (Group 3) and Case C to the control Group 1; the count of KRAS gene mutations in the duodenal lavage fluid, as determined by BNA clamp PCR, was comparatively high, albeit remaining below the predetermined cut-off value of 92. Despite this, Duplex sequencing identified a mutated allele at codon 12 that exceeded 0.05% in Case C. Cases D, E, F, G, H, and I belong to Group 1. Mutant alleles were unexpectedly observed in all cases, particularly at and around codons 12 and 13, but with a low prevalence (<0.02%).

DISCUSSION

The presently described test is safe and minimally invasive, requiring only a short time (∼1–2 min) in patients undergoing EGD (Video 1). Despite its bleak prognosis, PDAC has a distinctive biomarker, the KRAS-constrained mutation pattern. Here, we found that all patients with chronic pancreatitis (Group 1B) and AIP (Group 1C) lacked elevation in KRAS mutations, allowing distinction from PDAC cases and differential diagnosis otherwise being complicated (Graphic Abstract). KRAS mutations in duodenal wash fluid were frequently detected in pStage 0+I PDAC, irrespective of tumor location, suggesting that this approach has the potential to detect early-stage PDAC regardless of its site.

A study conducted in 1996 by Iguchi et al¹⁶ involved the administration of secretin and placement of a duodenal tube (Dreiling tube) to collect pancreatic juice for an hour. The PCR-SSCP method detected mutations of codon 12 of the KRAS gene in 12 out of 19 PDAC patients. However, using a nasal tube for 60 minutes to examine the duodenum is invasive and not suitable for physical examination or cancer screening in healthy people. Our new method effectively identified KRAS mutations in small PDACs (<10 mm) that are challenging to detect using serum CEA or CA19-9.

The addition of secretin in this study neutralized the pH of duodenal fluid, resulting in improved recovery of DNA, and a specialized catheter was employed to optimize the collection of washing fluid. In the United States, efforts have been made to collect pancreatic juice in high-risk individuals, such as unaffected members of familial PDAC families, using a special hood over the ampulla of Vater.²⁸ This trial is interesting, but the approach presents difficulties for endoscopists as a routine screening test for PDAC.

In Japan and Korea, EGD is replacing upper GI x-rays for gastric screening but might be considered too invasive a procedure for Europe and the United States with their low rates of gastric cancer. The test’s limitation lies in its reliance on geographic variations in gastric cancer incidence for widespread implementation. Nevertheless, considering the persistently high prevalence of gastric cancer in East Asia, Eastern Europe, and South America, we believe the test has significant applicability to the target population.

The duodenal lavage fluid exhibited a superior recovery rate with the new catheter employed in the present study (Supplemental Fig, S1B and S1C, Supplemental Digital Content 1, http://links.lww.com/SLA/F396). This study used a new specialized catheter under manufacturing and marketing approval as a medical device by MHLW (Fig. 1C). The recovery rate of the duodenal lavage fluid was better with the new catheter, but sometimes it was <50% (Supplemental Fig. S1B, Supplemental Digital Content 1, http://links.lww.com/SLA/F396). Therefore, a more recent catheter has been developed to address this issue (Supplemental Fig. S11, Supplemental Digital Content 1, http://links.lww.com/SLA/F396), and we have applied to MHLW for its approval as a medical device in Japan. By using the newer catheter incorporating a balloon in future large-scale validation studies, we expect to further improve sensitivity.

Limitations on use extend beyond countries where gastric cancer is uncommon. The fact that PDAC is much less common than the very prevalent breast and colorectal cancers means that a population-based screening modality makes no economic sense. Therefore, it should be restricted to high-risk individuals for PDAC (eg, new-onset diabetes, hereditary PDAC families). In the scenario where the incidence of early PDAC in the high-risk population is 1%,^29,30 the significant difference in sensitivity between the current method and the previous method corresponds to 270 cases, as denoted by the Number Needed to Screen.³¹ That is, the new approach can uncover one early PDAC case that was missed by the previous method after screening 270 cases in the high-risk population. Given the number needed to screen for high-risk individuals, this test might be applicable even in countries with a low prevalence of gastric cancer. We are conducting a large prospective study on high-risk individuals without a PDAC diagnosis. It should be noted that not all PDAC cases exhibit KRAS mutations, thus resulting in a sensitivity of <100%. It is inevitable that false negatives will occur. The established cut-off values for the KRAS mutations can aid in determining the necessity of invasive and comprehensive testing. Nevertheless, making decisions without careful consideration is not recommended since estimation errors can occur in the cut-offs, and each situation presents its own set of circumstances. Although not achievable given the sampling design limitations of this study, it is essential to develop a probability prediction system for PDAC using the KRAS mutations, which should be pursued in subsequent cohort studies.

The potential impact of identifying latent PDACs drives our enduring vision of uniting PDAC with gastric cancer detection. Using improved catheters, screening for pancreatic and gastric cancer at the same time appears warranted in countries that regularly perform EGD.

Supplementary Material

sla-284-e62-s002.docx ^{(4.5MB, docx)}