Comparative genomic analysis of head and body/tail of pancreatic ductal adenocarcinoma at early and late stages

Abstract Pancreatic ductal adenocarcinoma (PDAC), one of the most lethal human cancers, can be divided into head and body/tail cancers anatomically. We previously reported a prognostic relevance of tumour location in resectable PDAC. This study aimed to further explore the mechanism underlying the molecular diversity between the head and body/tail of PDACs. We detected tumour genomes in 154 resectable (surgery) and non‐resectable (biopsy) PDACs using a next‐generation sequencing panel. Wilcoxon's rank test or Fisher's exact test was used for evaluating associations between clinical characteristics, mutation frequency and survival probability between the two cohorts. Compared with pancreatic head cancers, pancreatic body/tail cancers showed significantly more enriched genomic alterations in KRAS (97.1% vs 82.4%, P = 0.004) and SMAD4 (42.0% vs 21.2%, P = 0.008). At early stages (I‐II), the SMAD4 mutation rate was significantly higher in pancreatic body/tail cancers than pancreatic head cancers (56.0% vs 26.5%, P = 0.021). At late stages (III‐IV), pancreatic body/tail cancers presented significantly higher KRAS mutation rate (100.0% vs 75.8%, P = 0.001), higher frequency of MAPK pathway mutation (100% vs 87.8%, P = 0.040) and lower rates of druggable genomic alterations (30.8% vs 57.6%, P = 0.030) than pancreatic head cancers. Our work points out that pancreatic body/tail cancer seems to be more malignant than pancreatic head cancer at late stages.

prognosis) and in vitro genetic and tumour biology. 1 We assumed that pancreatic body/tail cancer might be a less malignant phenotype compared with pancreatic head cancer. The conclusion was further supported by a recent large cohort database analysis which indicated that patients with pancreatic body/tail cancers had a better prognosis compared to those with head cancers among resectable PDACs. 2 Because of the genomic heterogeneity of PDACs, identification of the genetic and epigenetic profile of pancreatic head and body/ tail cancers could be of great help to better understand the prognostic relevance of primary tumour location. We previously demonstrated that pancreatic body/tail cancer had less invasiveness and metastasis potential than pancreatic head cancer possibly via miR-501-3p/E-cadherin signalling by using strictly matched resectable pancreatic head and body/tail cancers in both in vitro and in vivo models. 3 In contrast, Birnbaum et al 4

| Sequencing experiment
A total of 154 specimens and matched normal blood were detected and analysed in a College of American Pathologists (CAP) and Clinical Laboratory Improvement Amendments (CLIA) certified laboratory at OrigiMed for CSYS assay 6 with a mean coverage of 900× for tumour samples (minimum 700×) and 300× for matched normal blood samples. This panel covers all the coding exons of 450 cancer-related genes and 64 selected introns of 39 genes that frequently rearranged in solid tumours (Table S1). We analysed multiple genomic variant types, including single nucleotide variants (SNVs), copy number variations (CNVs), short and long insertions/deletions (indels) and gene rearrangements by bioinformatics workflows described previously. 6 White blood cells isolated from whole blood were used as matched normal control to identify somatic genomic alterations from germline mutations.

| Statistics analysis
Wilcoxon's rank test or Fisher's exact test was used for evaluating associations between clinical characteristics, mutation frequency and survival probability between the two cohorts. P < 0.05 was considered statistically significant. Statistical analysis was performed using R 3.3.1.

| Patient characteristics and overall survival
The patient clinicopathological data are summarized in Table 1. At the initial diagnosis, compared with pancreatic head cancers, pancreatic body/tail cancers showed more advanced stage (III-IV) and distant metastasis, but less local invasion. There is no significant difference in overall patient survival between the two subgroups ( Figure 1A). The median survival time was 23 and 28 months for patients with pancreatic head cancers and pancreatic body/tail cancers respectively with no statistic difference. Also, there was no significant difference in survival for tumours at early stages (I-II) (P = 0.180, Figure 1B) and late stages (III-IV) (P = 0.240, Figure 1C).
The comparison of gene alteration frequencies between pancreatic head and body/tail cancers is shown in Figure 2A. Compared with pancreatic head cancers, pancreatic body/tail cancers showed significant higher clinically relevant mutation frequency in KRAS (97.1% vs 82.4%, P = 0.004) and SMAD4 (42.0% vs 21.3%, P = 0.008) ( Figure 2B). The mutation frequencies of these two genes were further analysed according to tumour stage ( Figure 2C). The statistical significance in KRAS and SMAD4 mutation rates was achieved only at late stages and early stages, respectively. Interestingly, all MAP2K4 mutations (n = 6) occurred in advanced PDACs.
The comparison of hotspot mutation sites is shown in Figure S1.

| Pathway analysis
Seven primarily signalling pathways influencing tumour initiation and progression were involved in the pathway comparison analysis, including MAPK signalling, Wnt signalling, cell cycle signalling, homologous recombination (HR) pathway, PI3K-AKT-mTOR signalling, ERBB family pathway and Notch signalling. Among these pathways, the total mutation frequency of genes involved in Wnt pathway was significantly higher in pancreatic body/tail cancers than pancreatic head cancers (56.5% vs 36.5%, P = 0.020) ( Figure 3A). We further investigated whether this difference in mutation frequency in Wnt pathway genes related to tumour stages. The mutation frequency of the Wnt pathway genes did not show statistical difference but a higher trend both in early stage (65.4% vs 42.9%, P = 0.090) and in late stage (48.7% vs 30.3%, P = 0.150) between pancreatic body/tail and head cancers. MAPK pathway was the most dominant mutated pathway both in pancreatic head and in body/tail cancers (91.8% vs 97.1%, P = 0.188), although there was no significant difference. Of note, the mutation frequency of MAPK pathway showed the apparent difference in advanced pancreatic body/tail and head cancers (100% vs 87.8%, P = 0.040), but not in early-stage cancers (92.3% vs 95.9%, P = 0.600).
Besides, we focused on TOP 2 frequently mutated pathways, Wnt and MAPK signalling at the gene levels ( Figure 3B). Higher mutated frequency in Wnt signalling of pancreatic body/tail cancers mainly resulted from distinctly enriched SMAD4 mutations (P = 0.008), and more abundant LRP1/1B and CTNNB1 mutations.
Interestingly, we found that RNF43 mutations (n = 8) were mutually exclusive to SMAD4 in pancreatic head cancers but 3 of 4 RNF43 mutations co-occurred with SMAD4 in pancreatic body/tail cancers.

| Clinical druggable genes
To compare potential clinical benefits among pancreatic head and body/tail cancers in terms of targeted drugs, we analysed actionable alterations from 16 clinical relevant genes with 43 potential therapies according to the widely accepted rule ( Figure 4A). 7 Overall, 43.5% of pancreatic head cancer and 34.8% of pancreatic body/tail cancer carried at least one genomic alteration that could potentially benefit from the targeted drugs, but no significant difference was achieved. In late stages (III-IV), pancreatic body/tail cancer showed significant less druggable mutation than pancreatic head cancer In late stages (III-IV), the difference in KRAS wild-type druggable mutation frequency between the two subtypes was increased (0% vs 24.2%, P = 0.001). In addition, patients with TMB-H (>10 mutations/ Mb) accounted for 4.7% (n = 4) and 1.4% (n = 1), in pancreatic body/ tail and head cancers, respectively (P = 0.381).

| D ISCUSS I ON
It is well known that patients with pancreatic body/tail cancer usually have a poorer prognosis than those with pancreatic head cancer probably because of more advanced pathologic stages at initial diagnosis. 1    TGF-beta signalling is involved in squamous/EMT transition in PDAC. 11 Therefore, our results supported the previous two studies that pancreatic body/tail cancers might be more 'squamous phenotype' than pancreatic head cancers. We also assume that the molecular diversity between pancreatic head and body/tail cancers starts from the clonal expansion phase 10 but not the tumour initiation ( Figure 5).
When PDAC developed to the late stages (III-IV), there was a remarkable difference in cancer genome between the two subtypes.
We found significantly higher KRAS mutations (eg substitution, indels and gene amplification), more enriched mutation frequencies in MAPK pathway and a lower rate of druggable genomic alterations in pancreatic body/tail cancers than pancreatic head cancers.
Activating mutation of KRAS is almost ubiquitous (~95%) in PDAC and is an essential event in both tumour initiation and progression. 12 Mutant KRAS drives PDAC development and promotes tumour cell proliferation via altered metabolic pathways including stimulation of glucose uptake and utilization, reprogrammed glutamine metabolism and increased autophagy. 12

E THI C S APPROVAL AND CON S ENT TO PARTI CIPATE
The study was approved by the Ethics Committee of the First Affiliated Hospital, Zhejiang University School of Medicine and the Ethics Committee of Shulan Hospital, and followed the declaration of Helsinki. All patients had signed the informed consent. Writing-review & editing (lead).

DATA AVA I L A B I L I T Y S TAT E M E N T
The data sets used and/or analysed during the current study are available from the corresponding author on reasonable request.