Deciphering Genetic Alterations of Taiwanese Pancreatic Adenocarcinoma With Targeted Sequencing


 Purpose Pancreatic adenocarcinoma (PAC) is the 8th leading cause of cancer-related death in Taiwan, and its incidence is increasing. The development of PAC involves successive accumulation of multiple genetic alternations. Understanding the molecular pathogenesis and heterogeneity of PAC may facilitate personalized treatment for PAC and identify therapeutic agents.Materials and Methods We performed next-generation sequencing (NGS) with targeted panels to explore the molecular changes underlying PAC in Taiwan. The Ion Torrent Oncomine Comprehensive Panel (OCP) was used for PAC metastatic lesions, and more PAC samples were sequenced with the Ion AmpliSeq Cancer Hot Spot (CHP) v2 panel.ResultsFive fresh-frozen paraffin-embedded (FFPE) specimens were successfully assayed with the OCP, and KRAS was the most prevalent alternation, which might contraindicate the use of anti-EGFR therapy. One PAC patient harbored a FGFR2 p.C382R mutation, who might derive benefit from FGFR tyrosine kinase inhibitor. Additional 38 samples assayed with CHP v2 showed 113 hotspot variants, corresponding to 54 COSMID IDs. The most frequently mutated genes were TP53, KRAS, HRAS, and PDGFRA, (29, 23, 14, 10 hotspot variants), impacting 11, 23, 14, and 10 PAC patients. Highly pathogenic variants including COSM22413 (PDGFRA, FATHMM predicted score: 0.88), COSM520, COSM521, COSM518 (KRAS, FATHMM predicted score: 0.98) were reported.ConclusionsBy using NGS with targeted panel, somatic mutations with therapeutic potential were identified. The combination of clinical and genetic information is useful for decision making and precise selection of targeted medicine.


Introduction
Pancreatic adenocarcinoma (PAC) is the 8th leading cause of cancer-related death in Taiwan, and its incidence is increasing [1]. Most PAC patients are diagnosed when the tumor is relatively large and has extended beyond the pancreas. There are several reasons for such delay: rst, because of its anatomic location, pancreas is not easily accessible with conventional diagnostic imaging tools. Second, initial symptoms of PAC are usually not remarkable, and clinical workup often procrastinates until the onset of more suspicious signs. Third, cystic precursor lesions of PAC are not easily distinguishable from benign cysts and may represent a diagnostic dilemma that eventually delays the correct diagnosis [2]. PAC is often caught at late stages because patients are often asymptomatic, so having more validated genetic biomarkers can augment early diagnosis and proper treatment. Besides, PAC is the most lethal human malignancy, with a dismal ve-year overall survival rate of less than 5%. Even when the tumor seems con ned to the pancreas at the time of surgery, the 5-year survival rate never exceed 15%. It is estimated that 10% of PACs show a familial aggregation consistent with a genetic susceptibility. But in most instances, the genetic basis for the familial aggregation of PAC has not yet been identi ed [3][4]. The development of PAC involves successive accumulation of multiple genetic alternations with signi cant heterogeneity. Understanding the molecular pathogenesis and heterogeneity of PAC may facilitate personalized treatment for PAC and yield potential therapeutic targets [5][6][7].
Previous comprehensive exome sequencing of PAC had revealed that dozens of intragenic mutations accumulated in each cancer while most were infrequently mutated, and were passengers by themselves.
But these studies also identi ed a number of recurrent aberrations, such as driver mutations which played critical roles during carcinogenesis, involving at least 12 cellular pathways implicated in PAC development [8][9][10]. However, all of these studies were conducted in Western countries, and no study ever assessed the molecular alternations of Taiwanese PAC. From past experiences, the patterns of driving mutations might be very diverse across different ethnic groups [11][12]. Therefore, we proposed to use state of the art nextgeneration sequencing (NGS) with targeted panels to explore the molecular alternations underlying PAC in Taiwan. The propose of NGS is to identify genetic alterations that might be targeted with existing drugs or used as biomarkers. PAC is among the most malignant neoplasms while research on PAC relies on clinical, pathologic, and molecular features for biomarker discovery and corresponding treatment. This study aimed to decipher genetic aberrations coupled with clinical characteristics. The combination of clinical and genetic information may help us for precisely selecting targeted therapeutics.

Materials And Methods
In summary, we used NGS to study genetic alternations of Taiwanese PAC. Tumor genomic DNA was extracted from fresh-frozen para n-embedded (FFPE) specimens with nucleic acid checked. The harvested DNA was sequenced by NGS with two targeted panels, and the information of genetic alterations was analyzed and correlated with clinical features.
In the rst part, we took advantage of readily available FFPE specimens from deceased PAC cases to elucidate the feasibility of extracting adequate nucleic acid for NGS experiments. Pathological hematoxylin and eosin (H&E) stained slides were reviewed by one certi ed pathologist (CYL) to ascertain the presentation of adequate PAC cells. Nucleic acid was extracted from 5X5 µm FFPE sections, and the quality was checked by Qubit uorimeter (Invitrogen, part of Thermo Fisher Scienti c, Waltham, MA) and PCR of GADPH fragments to make sure that fragmentation was not too severe and was acceptable for ampli cation and sequencing. The Oncomine Comprehensive Panel (OCP, Thermo Fisher Scienti c) was used as the initial screening tool detecting tumor genetic alterations which could be used as candidates for downstream analyses. There were 143 pre-selected genes on this panel, which were designed to interrogate somatic mutations including single/multi nucleotide variants (SNVs/MNVs), insertions/deletions (INDELs, 73 genes), copy number variation (CNV) including gain (49 genes) and loss (26 genes) recognized as being oncogenes or tumor suppressors recurrently altered in solid tumors with the potential for approved therapeutics or novel ones with near-term clinical relevance [13]. Besides DNA, we also performed RNA extraction; once the RNA was adequate, the test of fusion genes was proceeded with the OCP, which interrogated 22 pre-selected onco-drive fusion genes.
In the 2nd part, more samples were collected to elucidate the oncogenesis of Taiwanese PAC, by using Ion AmpliSeq Cancer Hot Spot (CHP) v2 panel (Thermo Fisher Scienti c), which was designed to amplify 207 amplicons covering approximately 2,800 COSMIC mutations from 50 oncogenes and tumor suppressor genes [14][15]. Somatic mutations with clinical signi cance were identi ed from PAC with distinct precursors. After library generation, NGS experiments were conducted on Ion Torrent benchtop sequencers. The whole study protocol had been reviewed and approved by IRB of Cathay General Hospital; informed consent of part I was waived from IRB of Cathay General Hospital while signed informed consent was obtained from all participants of part II. All methods were performed in accordance with the relevant guidelines and regulations Results There were 6 FFPE PAC specimens retrieved and tested in part I of the study while 5 of which had adequate DNA/RNA for further NGS experiments. The nucleic acid quality was determined by Qubit uorimeter to ensure that the specimen could be successfully ampli ed and sequenced by targeted panels (Table 1 for details). Table 1 Nucleic acid Q/C from part I Taiwanese PAC samples (n = 5). The highly degraded FJU4 specimen was excluded from NGS analysis.  (Table 2 and Supplementary Fig. 1). Although there were hundreds of variants found in these specimens, many of which were probable passengers. Most Taiwanese PACs harbored KRAS mutations, as previous studies had shown [17]. There was one FGFR2 mutant case. The distribution of CNVs found in this study is detailed in Fig. 1. To prove that the identi ed variants were not spurious, we also evaluated targeted sequencing results using the Partek Flow software and Sankey diagrams of the ve assayed samples are detailed in Supplementary Fig. 2. It should be noted that SAM tools were used for variant calling, resulting in discrepancy in the number of variants harvested. Supplementary Fig. 3 shows variant impact heat map of 5 PAC samples assayed with the OCP.  The presence of KRAS mutation may be a predictive biomarker against the use of anti-EGFR antibody. In lung cancer treatment, KRAS is downstream in the EGFR pathway; therefore, tyrosine kinase-based treatment with ge tinib and erlotinib is ineffective when KRAS is constitutively activated [21][22]. However, many clinical trials which combine other tyrosine kinase inhibitors (TKIs) and chemotherapy are ongoing for these KRAS mutant PACs [23]. In addition, knockdown of mutant KRAS with RNA interference may be a potentially therapeutic strategy in near future [24]. Recently, a subset of KRAS wild-type younger PACs has been identi ed for whom the recognition of alternative oncogenic drivers which are also targetable is in eagerly need [25]. There was one FGFR2 mutant case (FJU01) and the NGS result showed the possibility of FGFR TKI treatment in the future as there are several FGFR TKI trials [26][27]. Although RNA sequencing showed no fusion gene in current study, there are more and more evidences suggesting that fusion oncogenes present not only in sarcoma, but also in carcinoma [28]. This kind of alternation could also be a potential biomarker for targeted therapy, such as crizotinib for treating lung cancer with EML4-ALK translocation [29]. Many CNVs were also found in this study while their meaning needs further investigation. Further studies to clarify whether the gain of oncogenes or loss of suppressors represent prognostic or predictive biomarkers for PAC are warranted.
The CHP was adopted for part II of the study with more PAC samples assayed. The choice of the CHP rather than OCP was based on economic consideration. The CHP v2 surveyed hotspot regions of 50 oncogenes and tumor suppressor genes, with wide coverage of KRAS, BRAF and EGFR genes, which were evidenced as being PAC-relevant from part I of the study. It deserves notice that it's also the CHP platform which was adopted in NCI's MATCH trial [30].  Fig. 4). Although Ryan et al. highlighted that more than 90% of PAC were associated with activating KRAS mutation, the frequency was much higher for intraepithelial neoplasm (> 90%) then mucinous neoplasm (40-65%). Other suppressors such as CDKN2A, TP53 and SMAD4, the aberrant rate increased with higher nuclear grade, albeit alterations in tumor suppressor genes rarely led to targeted therpay. The purposed more GNAS oncogenic mutation was not observed in PAC with mucinous precursor in current study, either [23].
Singhi et al. conducted by far the largest study on around 3,600 PACs and found that the most frequently mutated genes, i.e. KRAS, TP53, CDKN21, and SMAD4, can't be targeted with readily available existing drugs [31]. In 2020, FDA approved olaparib for maintenance treatment of germline BRCA-mutated metastatic PAC whose disease has not progressed on at least 16 weeks of a rst-line platinum-based chemotherapy, based on phase III POLO trial [32].The most common pathogenic germline mutations are in BRCA1, BRCA2, and ATM, and more rarely, in PALB2, MLH1, MSH2, MSH6, PMS2, CDKN2A, and TP53, among others, for an aggregate frequency of 3.8-9.7% [33][34]. Although somatic alternations were interrogated in current study, several of aberrant DNA damage repair genes were reported from Taiwanese PAC patients and re ux germline testing should be indicated once alternations in tumor-only sequencing were evidenced.
There were some limitations of the study. First, the modest sample size might hamper externalization of sequencing results. The 50 targeted genes of the CHP had been covered by the OCP, so there was no concern of comparability. In addition to the limited sample size, it should be noted that only 15% of PACs were resectable at time of diagnosis, which means the majority of PAC genetic analyses were conducted on early stage disease [35][36]. In current study, part I samples were derived from metastatic lesions while for part II, all specimens came from primary pancreatic neoplastic tissue in an effort to broaden clinical scenarios of PAC [37].

Conclusions
By using NGS with targeted panel, somatic mutations with therapeutic potential were identi ed. The combination of clinical and genetic information is useful for decision making and precise selection of targeted medicine.

Declarations -Ethical Approval and Consent to participate
The whole study protocol had been reviewed and approved by IRB of Cathay General Hospital; informed consent of part I was waived while signed informed consent was obtained from all participants of part II.
-Consent for publication All authors gave their consent for publication.
-Availability of data and materials Genomic data of the study were secured by the primary investigators as requested by IRB and might be available in an anonymous manner upon reasonable request.

-Competing interests
All authors declared there was no con ict of interest.

Supplementary Files
This is a list of supplementary les associated with this preprint. Click to download.