Impact of molecular surgical margin analysis on the prediction of pancreatic cancer recurrences after pancreaticoduodenectomy

Background Pancreatic cancer is one of the lethal cancers among solid malignancies. Pathological diagnosis of surgical margins is sometimes unreliable due to tissue shrinkage, invisible field cancerization and skipped lesions like tumor budding. As a result, tumor recurrences sometimes occur even from the pathologically negative surgical margins. Methods We applied molecular surgical margin (MSM) analysis by tissue imprinting procedure to improve the detection sensitivity of tiny cancerous cells on the surgical specimen surface after pancreatoduodenectomy. Surgical specimens were collected from 45 pancreatic cancer cases who received subtotal stomach preserving pancreatoduodenectomy at Nagoya University Hospital during 2017–2019. Quantitative methylation-specific PCR (QMSP) of the original methylation marker panel (CD1D, KCNK12, PAX5) were performed and analyzed with postoperative survival outcomes. Results Among 45 tumors, 26 cases (58%) were QMSP-positive for CD1D, 25 (56%) for KCNK12 and 27 (60%) for PAX5. Among the 38 tumors in which at least one of the three markers was positive, CD1D-positive cancer cells, KCNK12-positive cancer cells, and PAX5-positive cancer cells were detected at the surgical margin in 8 cases, 7 cases and 10 cases, respectively. Consequently, a total of 17 patients had at least one marker detected at the surgical margin by QMSP, and these patients were defined as MSM-positive. They were associated with significantly poor recurrence-free survival (p = 0.002) and overall survival (p = 0.005) than MSM-negative patients. Multivariable analysis showed that MSM-positive was the only significant independent factor for worse recurrence-free survival (hazard ratio: 3.522, 95% confidence interval: 1.352–9.179, p = 0.010). On the other hand, a significant proportion of MSM-negative cases were found to have received neoadjuvant chemotherapy (p = 0.019). Conclusion Pancreatic cancer-specific methylation marker panel was established to perform MSM analysis. MSM-positive status might represent microscopically undetectable cancer cells on the surgical margin and might influence the postoperative long-term outcomes. Supplementary Information The online version contains supplementary material available at 10.1186/s13148-021-01165-8.

of PC increases, PC is estimated to be the second leading cause of cancer-associated deaths in the United States by 2030 [2]. Although surgical resection is the only potentially curative treatment for PC, only 20% of newly diagnosed patients have an indication for surgical resection [3]. Moreover, the 5-year survival rate after surgery does not exceed 20-25% [4], although surgical techniques and perioperative chemotherapy have been improving [5][6][7].
We previously reported that dissected peripancreatic tissue margin (DPM) negative PC resections could reduce the recurrence rate and improve the prognosis [8]. However, we sometimes experience local recurrences derived from the histologically DPM negative facets. It may partially because the histological diagnosis of surgical margins is usually tricky due to tissue shrinkage, skipped lesions like tumor budding [9,10]. Also, the concept of 'field cancerization' implies that cancer-specific gene alterations possibly occur and spread through the pancreatic duct surrounding the primary tumor-originated pancreatic duct [11].
Molecular surgical margin analysis (MSM) by tissue imprinting procedure [12] has been reported as a super sensitive and quick method for evaluating the surgical margins of operative specimens [13,14]. After the pancreatoduodenectomy for a pancreatic head cancer case, the resected specimen usually has a sizeable surgical margin area, including the forward serosa, backward connective tissue, and portal vein notch. In these cases, tissue imprinting procedure by nitrocellulose membranes [12] is suitable to collect tiny cells on the complicated surgical margin surfaces, compared with surgical margin tissue collection [13].
We used a PC-specific methylation marker panel instead of KRAS mutation to perform a molecularbased quick and super-sensitive diagnosis of the tiny cancerous cell existence because of their high sensitivity and rapidity. Kisiel JB et al. reported several methylation markers for PC by bisulfite DNA sequencing [15]. The top two sensitive markers that can distinguish PC from normal pancreas rather than KRAS mutation and one promising candidate marker from our previous study [16] were chosen for this study. We optimized quantitative methylation-specific PCR (QMSP) assay for them.
In this study, we tried to apply the tissue imprinting procedure to resected PC specimens after pancreatoduodenectomy. We then assessed whether PC-specific methylation detection on the surgical margin could predict cancer recurrences and patients' survival outcomes after surgery more precisely rather than several factors, including preoperative serum tumor markers, histological findings and tumor stages.

QMSP values between tumor and adjacent normal pancreas
QMSP values of all 45 samples were shown in Fig. 1a. In each gene of the panel, Tumor QMSP was significantly higher than normal QMSP (CD1D: p < 0.001, KCNK12: p < 0.001, PAX5: p < 0.001). It means that each candidate gene was significantly methylated in PCs rather than the adjacent normal pancreas. ROC curves of each gene to discriminate tumor from normal were shown in Fig. 1b. Interestingly, we realized that a certain number of surgical margin samples have relatively high QMSP values. Representative assays of molecular surgical margin analysis were shown in Additional file 1: Figure S1. All of the margin samples had an amplification curve on the ACTB assay as a reference gene. While, some of them had the curve on some methylation signals, defined as MSM positive (a). Others had non of methylation signals, which was defined as MSM negative (b).

Diagnosis of molecular surgical margin analysis
For each gene of the methylation panel, MSM analysis positive (MSM-positive) was defined as the case with both tumor and surgical margin QMSP-positive using the optimal cut-off value. The rest of all cases were MSMnegative. As for CD1D, QMSP-positive tumors and adjacent normal tissues were 26 cases (58%) and 11 cases (24%) of 45 cases. Among them, QMSP-positive surgical margins were 8 cases and QMSP-negative in 18 cases. For KCNK12, QMSP-positive tumors and adjacent normal tissues were 25 cases (56%) and 4 cases (9%) of 45 cases, and QMSP-positive surgical margins were 7 cases and QMSP-negative in 18 cases. For PAX5, QMSP-positive tumors and adjacent normal tissues were 27 cases (60%) and 4 cases (9%) of 45 cases, and QMSP-positive surgical margins were 10 cases and QMSP-negative in 17 cases.
We summarized the QMSP results in Fig. 1c . It means that 84% of PC samples can be diagnosed by our methylation marker panel, while the rest of the 7 cases has to be excluded in this study and needs other tumor-specific gene markers. Then, among 38 cases, QMSP-positive surgical margin samples were found in 17 cases diagnosed as MSM-positive. Clinicohistological characteristics of these 17 cases were compared with the rest of 21 MSM analysis negative (MSM-negative) cases in Table 1. Interestingly, cases that underwent neoadjuvant chemotherapy had a significantly low risk of MSM-positive rather than no neoadjuvant chemotherapy cases (p = 0.019).

Correlation between MSM analysis and DPM
We just compared MSM positivity with pathological positivity for all 38 samples. The number of pathological surgical margin (PSM) positive cases was 6 (29%) in MSM-negative cases and 5 (29%) in MSM-positive cases, with no significant difference (p = 0.999). Moreover, the median PSM distance was 1081 µm (range 0-4000) in MSM-negative cases and 372 µm (range 0-2648) in MSM-positive cases, with no significant difference (p = 0.521) ( Table 1). There was no correlation between each QMSP value on surgical margin and PSM distance (p = 0.697 in CD1D, p = 0.557 in KCNK12, p = 0.576 in PAX5, Fig. 2).

Postoperative prognostic outcomes of MSM positive cases
MSM-positive 17 cases were compared with MSMnegative 21 cases in postoperative prognostic outcomes   Figure  S2a, S2b). Considering by recurrence type, MSM-positive cases had a tendency of cumulative local recurrence (median local RFS: not reached vs. 14.3 months, p = 0.072, Additional file 1: Figure S2c) and peritoneal recurrence (median peritoneal RFS not reached vs. not reached, p = 0.071, Additional file 1: Figure S2d). Whereas there was no significant difference in the distant metastatic recurrence (median distant metastatic RFS: 27.9 vs. 14.3 months, p = 0.117, Additional file 1: Figure S2e).

Univariate and multivariate analyses of prognosis
In total 38samples, univariate analysis revealed that the MSM-positive and venous invasion factors were significant predictors of worse RFS. When the multivariable analysis was performed on these predictors, MSM-positive was the only significant independent factor for worse  Table 2).

Discussion
MSM-positive cases were significantly associated with poor survival outcomes. Surprisingly, even in the PSM negative cases, MSM-positive cases exist and are also associated with poor survival outcomes. These results may imply that the judgment of PSM is sometimes tricky because of postoperative deformation of the modification by tissue shrinkage, neoadjuvant chemotherapy, tumor budding [9,10], misunderstanding of actual surgical margin and limited slice of the tumor margin area. On the other hand, MSM analysis can collect cells from a whole area of the dissected surface, and the results were judged digitally. Thus it has fewer false negatives than pathological diagnosis if the ideal molecular marker panel is available [12]. However, it is necessary to consider that MSM analysis may count cancer cell contamination floating on the specimen's surface, which causes false positives due to its high sensitivity.
Interestingly, MSM-negative cases had significantly many neoadjuvant therapy (NAT) treated cases. Initial treatment before resection may effectively cause a tumor shrinkage and achieve not only PSM-negative but also MSM-negative. Regimens of 24 NAT includes 15 gemcitabine-base chemotherapy, six oxaliplatin-based chemotherapy and three S-1-based chemoradiotherapy. Although we cannot distinguish differences depending on each regimen in this small number study, it seems evident that NAT is effective for pancreatic cancer cases for keeping the molecularly safe tumor margins.
PCs often recurs postoperatively due to locally progressive intensity and anatomically resectable limits. Interestingly, postoperative recurrence is present in patients with pathologically positive margins and pathologically negative margins [8,17,18]. There are several MSM-positive cases among PSM negative cases in this study, and occult local cancer remnants may be involved in postoperative recurrences. Besides, MSM-positive cases also correlate distant organ metastases and peritoneal recurrences. Conversely, performing the local dissection with molecularly safe surgical margins is fundamental to avoid or postpone any type of postoperative recurrences.
The whole MSM analysis procedure takes less than 3 h from sampling to acquiring the QMSP result if methylation marker panel was available [13,14]. Further improvement of the DNA extraction kit or real-time PCR procedure might minimize the time for utilizing it as an intraoperative application. If the result of MSM analysis is positive during surgery, additional resection or additional postoperative local therapy can be considered. To improve accuracy, imprinting the residual organ's surface after resection might be suitable rather than imprinting the resected specimen's surface to detect the occult remnant cells with epigenetic abnormalities.
There are a few limitations in this study. First, this study is a single-center analysis with a small number of samples. The positive rate of any methylation marker in tumors was 38 of 45 (84%). Thus 7 cases were excluded from the study. Although it may partially due to the preoperative chemotherapy, it is better to examine another cancer-specific methylation markers derived from some promising papers [19,20] in the future study. Secondly, since MSM analysis is imprinted in manually, there is a possibility of variation in imprinting. However, in this study, sample collection was performed by a single experienced person [13,14]. For clinical application, it is necessary to establish a protocol of imprinting. Finally, this cohort mainly consists of resectable PC without neoadjuvant chemotherapy and borderline resectable or unresectable PC with intensive neoadjuvant chemotherapy just because the routine neoadjuvant therapy still has not been established for resectable PC at that time. This situation may affect univariate analyses of RFS results.

Conclusion
A pancreatic cancer-specific methylation marker panel was established and enabled us to perform MSM analysis. MSM-positive margin is one of the predictors of recurrence and survival in patients who underwent pancreaticoduodenectomy for PC. Additionally, performing NAT has the advantage of securing molecular negative surgical margins.

Patient cohort
Forty-eight PC cases who underwent curative-intent subtotal stomach-preserving pancreaticoduodenectomy at the Department of Gastroenterological Surgery (Surgery II), Nagoya University Graduate School of Medicine, Nagoya, Japan, between January 2017 and December 2019 were included. Of those, excluding 3 cases that were histologically proven to be non-invasive PCs, 45 cases were eligible for the study. Tumors, adjacent normal tissues and surgical margin imprinting samples were collected just after the surgical specimen was picked out from the surgical site. This study was approved by the hospital's ethics committee, and written informed consent was obtained from all patients about the use of surgical samples and their clinical data. The background and clinicopathological features of the patients are summarized in Table 3.

Surgical margin sample collection procedures
After removing the surgical specimen by curative surgeries, float-on contamination of cells on the specimen's surface was removed by running water. Then, margin imprinting samples were collected by pressing 3 × 3 cm Hybond-C Extra nitrocellulose membranes (GE Healthcare, Little Chalfont, UK) directly on the specimen for 10 s following the previous publication [12,13]. The membranes were placed into a coded 50 ml tube with 3 ml 1% SDS-PK solution. Three repeat membrane samples were taken from each facet of the specimen. Thereafter, matched tumor tissue and normal pancreatic ductal tissue were collected and placed in a coded 50 ml tube with 3 ml 1% SDS-PK solution.

DNA extraction and bisulfite treatment
Collected tissues and margin imprinting samples were prepared by four rounds of proteinases K exposure during two overnight periods. These chemically digested samples were applied to dry bead tubes of UltraClean Tissue and Cells DNA isolation Kit (MO BIO Laboratories, Carlsbad, CA) for DNA extraction. One micro gram of DNA samples was subjected to bisulfite treatment by BisulFlash DNA Modification Kit (Epigentek, Farmingdale, NY).

Quantitative methylation-specific PCR
CD1D, KCNK12 and PAX5 were selected as candidate excellent methylation markers for pancreatic adenocarcinomas extracted from the published previous papers [15,16]. ACTB was measured for normalization. Primers and probes of target genes were set at CpG rich region, while those of ACTB were set at no CpG region (Additional file 2: Table S1). The bisulfite-modified DNA was used as a template for fluorescence-based QMSP. It was performed in triplicate using StepOnePlus (Thermo Fisher Scientific, Waltham, MA). Thermal cycling was initiated with denaturation at 95 °C for 10 min, followed by 40 cycles of 95 °C for 15 s and 60 °C for 1 min. Each plate  ) is defined as the ratio of the fluorescence emission intensity values for the target gene-specific PCR products to those of the ACTB (reference gene) and then multiplied by 100 for easy tabulation. QMSP positive was defined as ≥ 1.0 following the previous work [13].

Statistical analysis
Continuous variables were analyzed by the Mann-Whitney U test as a non-parametric test and Student's t test (2-tailed) as a parametric test. Categorical variables were analyzed by Fisher's exact test. RFS was defined as the time from surgery to first documentation of disease recurrence. OS was defined as the time from surgery to the date of death from any cause. RFS and OS were analyzed by the Kaplan-Meier method and compared using the log-rank test. Gray's test was used to evaluate the cause-specific cumulative incidence. Associations of gene methylation and other histopathological factors with RFS were evaluated by the Cox proportional hazards model with hazard ratios (HRs) and 95% confidence intervals (95%CIs). Associations with p < 0.05 in univariate analyses were further evaluated in multivariate regression analyses. All tests were 2-sided and considered statistically significant and clinically promising for values of p < 0.05. Statistical analyses were carried out with R version 3.3.2.