Liquid Biopsy for ctDNA in Peritoneal Lavage and Blood in Pancreatic Cancer

1. Introduction Approximately 1.000 new cases of pancreatic cancer (PC) are diagnosed each year in Denmark. Ductal adenocarcinoma is by far the most frequent histological subtype of cancer in the pancreas and accounts for more than 90% of all cases \[1\]. Thus, in this protocol the term PC refers to pancreatic ductal adenocarcinoma. PC is a devastating disease with an overall 5-year survival of 3-7% \[2\]. Less than one in four patients are candidates for intended curative resection, but even after microscopically radical resection (R0 resection) the majority of patients will have recurrent disease within 2 years \[3\]. The discrepancy between pathological assessment and clinical outcome indicates that most patients will harbour non-detected (i.e. occult) cancer cells at the time of resection. It is of utmost importance to improve our ability for prognostication and prediction of PC, aiming at a tailored and personalized approach for treatment decisions regarding this aggressive type of cancer. This project aims to take an important step in this direction, by a 2-tiered approach: 1) Examination of the peripheral blood (detection of circulating tumor DNA), and 2) the peritoneal lavage fluid (detection of tumor DNA in the peritoneal space). 2. Background PC is a deadly disease and surgical resection of the tumor is the only hope of cure. Approximately 20-25% of the PC patients are candidates for intended curative resection at the time of diagnosis, but despite microscopically radical resection the majority of patients will have recurrent disease within 2 years \[3\]. This indicates that most patients will harbour non-detected (i.e. occult) cancer cells at the time of resection. Studies suggest that free tumor cells in the peritoneum and in the blood are part of this occult disease burden, and that patients with such findings should not be operated but treated as having metastatic disease. However, the exact incidence of these tumor cells in an unselected cohort of patients undergoing pancreatic resection is unknown, and the potential impact on postoperative survival is also uncertain. In recent years, molecular biomarkers are increasingly being regarded as both predictive and prognostic tools for cancer patients. This study will use the most optimal available methods to investigate the incidence of tumor cells in the peritoneum and blood in PC patients, and to relate these findings to the final outcome of the resected patients. This project has become highly relevant since new treatment methods (i.e. Pressurized IntraPeritoneal Aerosol Chemotherapy (PIPAC)) may be used to eradicate free tumor cells. 2.1 Positive peritoneal lavage indicates a poor prognosis in PC patients Occult malignant cells may be detected in fluid obtained by peritoneal lavage (PLF) during staging laparoscopy, open surgery or sampled from postoperative drain fluid in patients with PC. The malignant cells are diagnosed by conventional cytology combined with immunocytology, but real-time polymerase chain reaction (RT-PCR) of CEA can increase the detection rate of occult malignant cells in peritoneal lavage fluid \[4\]. The majority of data report on the use of CEA \[4\], but also evaluation of EpCAM have been subject of interest. A recent study reported the combination of CEA/EpCAM mRNA as an optimized set-up for detection of free intraperitoneal tumor cells of various origins with a high sensitivity and an excellent specificity \[5\]. Another marker, CA 19-9, has gained interest, however mostly in studies including gastric cancer patients and presenting the results in protein levels rather than mRNA. These studies suggest that elevated levels of CA 19-9 in PL is associated with more advance stages of disease, is a more reliable predictive factor for staging than serum CA 19-9 levels and is a more sensitive marker than CEA \[6, 7\]. While mRNA- and protein-based analyses have shown promising results, only casuistic data exists on the detection of KRAS mutations in PLF \[8-10\]. However, in a Norwegian study including patients resected for rectal cancer, KRAS mutations were found in PLF collected immediately after surgery in 19 out of 237 patients. These KRAS positive patients had significantly poorer survival \[11\], and similar results might be found in patients resected for PC as well. A recent systematic review and meta-analysis demonstrated that PC patients with positive peritoneal cytology (Cy+) had a significant poorer survival than patients with negative peritoneal cytology (Cy-)(HR 3.18), and the authors concluded that Cy+ patients should not undergo surgery \[12\]. This conclusion was supported by a significant lower overall survival and a higher peritoneal recurrence rate after resection of Cy+ patients when compared to Cy- patients \[13\]. Agreement that Cy+ in resectable PDAC is a negative predictor of prognosis came from another recent meta-analysis and systematic review. However, this study also indicated that the median OS was worse in patients without than in those with resection among patients with Cy+, thereby emphasizing need of further careful assessment of indications for radical resection in Cy+ patients \[14\]. Additionally, it should be noted that manipulation of tumor during surgery significantly increases the rate of patients with positive PLF (based on EpCAM mRNA measurements) from 10% to 54%, although this study was unable to detect a worse prognosis in EpCAM positive patients \[3\]. Despite such evidence, the majority of international guidelines do not include peritoneal lavage and cytology, protein analysis or RT-PCR as part of the preoperative evaluation of PC patients. This observation may be explained by the retrospective origin of most of these (mainly Japanese) data, lack of uniformity regarding definition of methods used for analysis of PLF, lack of studies evaluating preoperative treatment of PLF+ patients, and a general reluctance to accept that the majority of resectable PC patients have disseminated disease at the time of diagnosis. 2.2 Circulating tumor DNA (ctDNA) in the blood is a marker of poor prognosis in PC patients Major improvements have been made in understanding the molecular carcinogenesis and the genetic landscape of PC. One significant milestone is the revelation of four common driver genes in PC carcinogenesis. KRAS is mutated in 90-95%, and is therefore the most frequently mutated gene in PC patients, followed by less frequent mutations in TP53 (50%), SMAD4 (19%) and CDKN2A/P16 (6%) \[15, 16\]. Interestingly, KRAS mutations have also been detected in circulating tumor DNA (ctDNA) in the blood (liquid biopsies) from patients with metastatic PC, and ctDNA is considered a marker of poor prognosis \[17\]. Similar, KRAS mutations were found in the plasma of one-third of patients with a resectable tumor, and ctDNA positive (ctDNA+) patients had a significantly poorer overall survival (13.6 months vs 27.6 months, p\<0.0001) \[18\]. Similar conclusions were drawn in recent systematic reviews and meta-analyses \[19,20\], while one study failed to confirm these results. The detection of KRAS mutations in cell-free DNA has also been identified as a prognostic biomarker in PC patients \[21\]. Moreover, the combination of KRAS mutation analysis and CA 19-9 assay is useful for detection and prognostic evaluation of pancreatic carcinoma. If looking at studies including all stages of PC patients, the prevalence of KRAS mutations in liquid biopsies was 40.8%, and these mutations had a negative impact on overall survival with a HR of 3.16 \[19\]. Different ctDNA detection methods have been used, however the recent introduction of digital droplet PCR (ddPCR), a new robust PCR method for quantifying low-abundance point mutations in cell-free circulating DNA, shows promising results and offers increased sensitivity and reproducibility relative to quantitative PCR (qPCR) \[22\]. This high-sensitivity method for ctDNA detection has been clinically and analytically tested in a recent study also confirming the promise of ctDNA as a clinically useful prognostic biomarker in pre-surgery samples, in immediate post-operative period and in post-surgery follow up \[23\]. Hopefully, applying this sensitive detection method together with the possibility of multiple assessments over time, the use of ctDNA may be able to predict treatment response and patterns of resistance earlier leading to truly personalized medicine, with molecular-guided treatment decisions based on a single blood draw \[20\]. As with the detection of free intraperitoneal malignant cells, data on the consequences of circulating tumor DNA has so far not had significant implications on the recommendations and guidelines regarding the treatment of PC patients. 2.3 Why study the value of PLF analysis in PC patients now During the last decades, the utility of different molecular biomarkers for detection of tumor cells in PLF and blood have been subject of interest. Many studies, including recent reviews and meta-analyses, have proven detection of KRAS mutation in plasma as a predictive and prognostic biomarker as well as a monitor of treatment response \[17, 19-21, 24-27\]. Moreover, introduction of the high-precision ddPCR method has further increased the expectations to the utility of KRAS mutation analysis in plasma as a prognostic biomarker \[18, 23, 28\]. In recent years, also studies investigating the applicability of biomarkers in PLF have yielded promising results. The sensitivity of using PCR for detection of CEA is superior to conventional cytology \[4\], especially when combined with analysis of EpCAM \[5\]. Surprisingly, only a few, almost historical studies of KRAS mutational analysis in PLF have been performed \[18-20\]. However, the promising results and the obvious advantages of ddPCR analysis of KRAS in plasma in PC leads to the inevitable question whether KRAS mutational analysis is feasible and suitable as a prognostic biomarker in PLF as well. The treatment of resectable, locally advanced and metastatic PC has changed significantly over the past few years. New chemotherapy regimens have improved survival in metastatic PC \[29\], and these regimens (+/- radiation therapy) are presently being tested in both resectable and locally advanced PC with promising preliminary results \[30\]. In theory, these new regimens may be potentially effective against ctDNA in PC patients, whereas the effect on peritoneal lavage positive (PLF+) PC patients is more speculative due to the low intraperitoneal concentrations of systemic chemotherapy. However, the latter problem may be solved by using Pressurized IntraPeritoneal Aerosol Chemotherapy (PIPAC) which allows better intraperitoneal distribution, concentration and accumulation of chemotherapy, without the systemic side effects. Initial PIPAC experience has shown up to 70% histological response rates in manifest peritoneal metastases \[31\], PIPAC may convert patients from PLF+ to a PLF- status \[32\], and the applicant's research group was the first in the world to show an effect of PIPAC directed treatment in PC patients with manifest peritoneal metastases \[2\]. Given these potential new treatment options against occult cancer cells in PC patients, the time has come for assessing the extent of the problem \[4\] with the ultimate goal of improving treatment strategies and results. Thus, if the present study identifies a significant prognostic factor that may be diagnosed through PLF, then the next step would be to use PIPAC directed adjuvant treatment in these patients. In fact, a prospective study in high risk colon cancer patients is currently ongoing, where these patients are treated with PIPAC directed adjuvant therapy in order to reduce the risk of developing peritoneal metastases \[5, 33\]. PCR for mRNA was superior to conventional cytology with immunocytochemistry \[5\]. However, it may be speculated that the highly sensitive ddPCR of KRAS may be a better tool for PLF+ detection when focusing on PC patients, as up to 95% of these harbour mutations in this gene. 3\. Aims 1. Investigators aim to investigate the incidence of PLF+ and KRAS ctDNA in the blood from an unselected cohort of PC patients scheduled for attempted curative surgery. 2. Secondly, investigators will study the prognostic impact of PLF+ and KRAS ctDNA positivity in PC patients. 4\. Materials and methods 4.1 Study design and study group Prospective and descriptive multi-centre study investigating the incidence of malignant cells and circulating tumor DNA in the peritoneum and of circulating tumor DNA in the blood before and after intended curative resection of patients with pancreatic cancer. Approximately 200 patients with PC are resected each year in Denmark, and 50 (25%) of them are resected at Odense University Hospital (OUH) (www.dpcg.dk). Based on a national approach and the inclusion of 1-2 international high-volume centers (e.g. Karolinska University Hospital (Sweden) and Lübeck University Hospital (Germany)) and 2-3 large international centers (e.g. Halle (Saale (Germany)), at least 200 PC patients may be recruited for participation within one year. Patients scheduled for intended curative resection of PC are eligible for inclusion. However, patients with a postoperative histological diagnosis other than pancreatic adenocarcinoma and patients having explorative laparotomy/laparoscopy only will be excluded. The ultimate goal is inclusion of 200 consecutive patients. However, the number of included patients may be adapted if more data on the prevalence of PLF+ in surgically treated PC appear. Investigators will collect blood samples at 4 different time points and peritoneal lavage fluid (PLF) during preoperative (PL-1) and postoperative (PL-2) diagnostic laparoscopy. Conventional cytology with immunocytochemistry and ctDNA (KRAS) analyses will also be performed in a negative control group (benign inflammatory ascites from patients with decompensated liver cirrhosis, n=20-30) and in a positive control group (malignant ascites or PLF specimens from patients with peritoneal metastasis from PC, hence patients who are candidates for or already receive PIPAC treatment, n=20-30). Also, for the purpose of validation, ddPCR for KRAS in the blood will also be performed on a positive control group consisting of patients with metastatic PC (n=20) and on a negative control group consisting of healthy donors (n=20). 4.2 Patient enrolment Patients with PC are evaluated at the local multidisciplincary (MDT) conference. After the MDT conference, candidates are screened for potential inclusion. Demographic data, previous treatment and co-morbidity will be examined in the patient file to see if the patients meet the inclusion criteria. After screening, potential patients and - by patient request - relatives are individually informed by a surgeon performing the procedure in calm surroundings without disturbances. Both oral and written information will be given, and after one week - upon permission from the patient - the candidates are contacted by telephone. If participants accept, both oral and written consent are mandatory and will be obtained during consultation. Candidates are encouraged to bring relatives to this meeting. Oral information may be given by phone, but only if the potential participant has agreed to this contact form. In order to ensure the optimal selection and treatment of the patients included in the study, the investigators must have access to medical records. The following demographic and procedure related data will be stored in a REDCap electronic CRF, delivered and secured by OPEN (Open Patient data Explorative Network): age, gender, inclusion criteria, informed consent, performance status, co-morbidity, day and time of operation, possible side effects and complications, data regarding oncological treatment, data from analysis of PLF, blood and tissue and results of laparoscopic examination and CT. Thus, patients are screened for potential inclusion during the MDT, but the above information will only be gathered and recorded after inclusion of the patients. 4.3 Peritoneal lavage 1. The included participants will undergo diagnostic laparoscopy with LUS (LAP/LUS) in general anaesthesia immediately prior to resection. This is a standard procedure in all PC patients prior to resection. During laparoscopy peritoneal lavage (PL) with 500 ml isotonic saline is performed, and at least 200 ml is collected for analyses. 2. After resection, the specimens containing the primary tumor undergo pathological examination and all patients with N+ disease (cancer in one or several regional lymph nodes = high risk patients) are offered follow up by CT after 6-9 months. In case of recurrence, patients will go for a new MDT conference to determine treatment according to national guidelines, while diagnostic LAP/LUS with PL will be repeated in patients without signs of recurrence (PL-2). If the diagnostic laparoscopy and/or PLF analysis shows evidence of peritoneal metastases, the patient will be referred to a multidisciplinary conference according to national guidelines. If there is no sign of metastases the patient will switch to standard follow-up. 3. PLF will be processed as follows: If a spontaneous coagulum is present, it will be fixed in formalin and embedded in paraffin (FFPE). Four centrifuge tubes will be filled with 50 ml of the fluid and centrifuged. From the sediment of the first tube, two smears will be produced, dried and stained with Papanicolaou and May-Giemsa Grünwald. From the sediment of the second tube, a cell block will be prepared after addition of three drops of plasma and two drops of thrombin. The cell block will be FFPE. From each of the two FFPE blocks, a 4-5 µm thick section will be cut with a microtome and stained with H\&E for microscopic analysis. If appropriate, as judged by the pathologist and based on the findings at conventional cytology, sections from the paraffin embedded material will be used for immunocytochemical analyses for PC tumor markers, such as carcinoembryonic antigen (CEA), EpCAM, IMP3, MUC5AC, S100P and/or maspin as well as markers for mesothelial cells, such as calretinin and vimentin, on 4 μm thick sections from the paraffin block(s), as described previously \[5\]. The sediments of the third and fourth tubes will be stored at -80°C in MagNA Pure LC Lysis Buffer (Roche), for subsequent analysis by ddPCR. From the frozen sediment of the third and fourth tube, DNA will be isolated and examined by ddPCR for the KRAS mutation. Primers are designed to target mutations in codon 12 and 13 alone (G12D, G12V, G12R and G13D), since these patterns account for 90% of all KRAS mutations \[28\]. 4.4 Liquid biopsies Blood samples are taken after diagnosis, but prior to neoadjuvant therapy (when provided) (BS-1), after neoadjuvant therapy prior to surgery (BS-2), approximately one month postoperatively and prior to adjuvant therapy (BS-3) and after adjuvant therapy (when provided) (BS-4). At each time-point 20 ml (2 x 10 ml) peripheral blood is collected in ethylenediaminetetraacetic acid (EDTA) vacutainers. BS 1-4 are analysed as follows: DNA will be purified from blood samples (BS-1 - BS-4). Detection of the KRAS mutation will be performed as described above. 4.5 Primary tumor One formalin-fixed and paraffin-embedded (FFPE) or frozen tumor block from the included patients and from the positive BS and PLF controls (if available - otherwise a FFPE block with metastastic tissue) will be selected by a pathologist for DNA purification in order to perform dd-PCR for KRAS and/or next-generation sequencing (NGS) for KRAS and other relatively frequently mutated genes (e.g. BRAF, p53, SMAD4, CTNNB1). The amount of analysed tissue is approximately 5 mg (corresponding to 3 x 10 μm paraffin sections) and from this tissue around 50 ng of DNA will be purified for NGS analyses. Besides, histological tumor type and grade as well as TNM status will be recorded. Thus, in this project the approach is targeted sequencing for a selected number of mutated genes, and therefore, methods are used that are not extensive mapping of the genome. In KRASwt patients, PLF and BS will be analyzed using dd-PCR for a driver mutation other than KRAS, identified by NGS. 4.6 Patient safety Diagnostic laparoscopy with ultrasound (LAP/LUS) is a safe procedure. The goal of LAP/LUS is to detect signs of metastatic disease (e.g. liver mets, peritoneal mets) before open exploration. Collection of fluid during this laparoscopy is a simple and safe procedure, which can easily be performed by any institution. However, both the LAP/LUS and the postoperative diagnostic laparoscopy (see above) are only performed by experienced laparoscopic surgeons and the risk of complications is low (\<2%, bleeding, abscess, perforation). The laparoscopic examination is performed during a short-term general anaesthesia, which for some can cause nausea, vomiting, headache, fatigue, dizziness, fever or throat soreness. However, serious side effects are very rare and will most often be caused by the procedure and the patients' comorbid medical condition rather than the anaesthetic itself. Examples of serious side effects are myocardial infarction (up to 5%), pulmonary embolism (\<2%), aspiration (0.03%), malignant hyperthermia (0.03%), anaphylaxis (0.01%) and death (0.001%) \[34\]. Also, a small proportion of the patients will be offered a control CT-scan, which is associated with a radiation exposure corresponding to approximately 3 years background radiation (cancer.dk). However, most of the patients, who undergo surgery for pancreatic cancer will have this scan done at some point during the follow-up period. Thorough preoperative assessment will be performed in order to identify risk factors and stratify patients so that optimization and planning can occur preoperatively. Also, blood sampling is a simple procedure with a very small risk of non-serious complications such as infection and superficial hematoma at the insertion site. The risk of these side effects is minimized by pre-insertion disinfection with alcohol swaps and brief post-insertion compression over insertion site, respectively. It is standard clinical practice to perform LAP/LUS in patients diagnosed with PC immediately prior to resection. However, the PL and the subsequent processing of PLF, is not standard clinical procedure in these patients. Also, the second LAP/LUS with PL-2, that is offered to only a small fraction of the patients, and the 4 blood samples for KRAS mutation analysis deviate from standard clinical practice. The execution of NGS on FFPE from primary tumor is not standard procedure, but will not impact the patients' diagnosis or treatment. All other aspects of this study is in accordance with standard clinical practice (National clinical guidelines) and therefore does not differ between patients included in this study and all other patients with PC. 4.7 Outcomes Primary outcomes • The rate of pre- and postoperative Cy +/- and/or ctDNA (KRAS mutation) +/- in PLF in patients undergoing intended curative resection for PC • The rate of ctDNA (KRAS) +/- in blood samples obtained in patients undergoing intended curative resection for PC Secondary outcomes • Overall survival, median survival and recurrence free survival (RFS) in Cy+ versus Cy- and/or ctDNA+ versus ctDNA- in PLFs • Overall survival, median survival and recurrence free survival (RFS) in ctDNA+ versus ctDNA- patients 5\. Statistics Descriptive statistics will be used to describe the included patients and the rate of Cy+ and ctDNA+. Survival is estimated with the Kaplan-Meier method and the log-rank test is used to compare survival curves. For all tests p\<0.05 is considered statistically significant. Sample size calculation The reported frequency of PLF+ vary, probably mainly due to different methods of obtaining and analysing PL and different patient characteristics, but the pooled rates of occult peritoneal tumor cells are 8% (2-24%) before and 33% (15-58%) after manipulation (i.e. surgery) \[4\]. The literature suggests that at least one-third of patients with resectable PC will harbor ctDNA (KRAS mutations) at the time of the diagnosis \[18\]. Since the main endpoints are descriptive no sample size can be made for these. PLF+ resected patients have a reported hazard ratio of 3.2 for overall survival but with a very wide confidence interval (1.9-5.4) \[12\]. For disease free survival a hazard ratio of 2.9 is reported (2.4-3.5). Given the uncertainty regarding both the hazard ratio and frequency of PLF+ a sample size was calculated using the lowest value of the confidence interval for the OS hazard ratio (1.9) and a frequency of PLF+ of 8%. With a power of 0.8 and alfa at 0.05 a total of 161 patients (149 PLF- and 12 PLF+) is needed to detect a hazard ratio of 1.9 between the two groups. Given the uncertainty of both estimates the power was estimated for three different proportions of PLF+ and two different HR for the three different sample sizes. With an expected dropout rate of 10% a total of 180 patients must be included in the study, but due to the multi-centre setup the ultimate goal is the inclusion of 200 consecutive patients. 6\. Economy Partial funding of this project has been obtained from Danish Cancer Society ("Kræftens Bekæmpelse") (Allocation grant 2019-20: 342.900 DKK, Sagsnr. R218-A13057-18-S66), the Research Council ("Forskningsrådet"), Odense University Hospital (Allocation grant 2019: 296.000 DKK, Sagsnr. A3064) and Axel Muusfeldts Fond (300.000 DKK, sagsnr. 2020-0172). Both grants are allocation grants scheduled to cover analyses of liquid biopsies and PL fluids. Investigators are currently in the process of applying for additional funds. No member of the research team has financial interests in the present study. There are no relations to any donating funds. 7\. Ethics The collection of PLF and PB material is a simple and safe procedure that can easily be performed at the institutions that perform PC surgery; hence this study will not expose the patients to any risks or burden besides the already known low risk of this procedure. Some of the patients with N+ disease will undergo an additional laparoscopy with peritoneal lavage (PL-2), but the risk of this second procedure is not significantly different from the first procedure prior to resection. The second laparoscopy may also be a benefit for these high-risk patients, since it may allow early detection of local (peritoneal) and asymptomatic recurrence and thus enable them a quicker access to relevant treatment. As all analyses of the PB and PLF, with the exception of the analysis of PLF-2, will be performed at the end of the study period, a research biobank will be established for this project. At the end of this project, leftovers of the stored material will be destroyed. Unfortunately, it is to be expected that many (most) patients will prove to have died from their disease. Hence, we apply the Ethics Committee for an exception from the rule that there must be informed consent from the patients. We will ensure that patients have not advocated against the use of tissue samples in the 'Vævsanvendelsesregister'. A proportion of positive PLF controls have only signed the original statement of consent regarding permission to perform molecular analyses on PLF, without specific consent to perform targeted NGS for around 20 cancer-related genes on archived histologic material (from primary tumor or metastases). This is, however, necessary to be able to identify KRAS wt patients, whose PLF and PB will be analyzed with dd-PCR for a driver mutation other than KRAS. A few of these patients are deceased or have been terminated from the surgical department due to disease progression. For these patients, we apply for dispensation from obtaining a supplementary consent to perform targeted NGS for around 20 cancer-related genes also on material from primary tumor or a metastasis.