Mutation of K-RAS, CDKN2A, SMAD4 and TP53 in Pancreatic Cancer: Role of Liquid Biopsy in Preoperative Diagnosis

BACKGROUND Pancreatic cancer is the 4th cancer-related cause of death in Western countries and it is expected to become the 2nd leading cause of death by 2030 [1]. Approximately 80% of patients are unresectable at diagnosis [2] including a 41.4% of patients found unresectable at exploratory laparoscopy/tomy [3]. Moreover up to 38% of eventually resected patients will need a vascular resection in order to achieve a radical pancreatectomy [4], just over half of which because of a histological invasion [2]. Last but not least, successfully resected patients will recur within 12 months in more than 40% of cases with dismal prognosis [5]. Those early recurrent patients will show local spread or metastatic disease within 12 months with a post-recurrence survival and overall survival more than halved compared to patients recurring beyond one year from resection [5]. Laparotomy may be therefore useless in such patients. In this setting neoadjuvant treatment plays a key-role. It has the potential advantages to deliver systemic therapy to all patients with an increased efficacy of radio- and chemotherapies distributed in a virgin field, identify patients with aggressive tumor biology that could recur shortly after surgery and thus would not benefit form an operation. Moreover neoadjuvant therapy could downstage the tumor avoiding major vessels resections, decrease positive margins resections and decrease post-operative pancreatic fistulas [6]. Unfortunately imaging is no longer reliable in predicting resectability after neoadjuvant treatment [7]. The total mean direct cost of a patient with a resectable disease is $134,700, while for metastatic or unresectable patients this cost is $49,000-65,300 [8]. A properly staged patient provides a better resource allocation.

Liquid biopsy provides an non-invasive signature of the tumor, it is based circulating genetic material coming from cellular turnover and thus especially from the tumor [9]. Analyzing mutations on cell-free nucleic acids gives translational information on tumor biology and therefore on its clinico-pathological features and likely on its progression. Pancreatic cancer progression has long been studied. What is clear is that activating K-RAS mutations are an early event in most lesions, followed by inactivating mutations in CDKN2A, TP53 and SMAD4 [10]. Those 4 mountain genes are predictive of progression pattern in an autoptic study: patients with 2 or less genes mutations were more likely to develop oligometastatic failure and to harbor earlier disease stage at diagnosis compared to those with 3 or more mutated genes [11]. K-RAS mutations have been associated to a worse overall survival in particular when found in peripheral blood by liquid biopsy [12]. CDKN2A mutations have been associated to lymphatic invasion and widespread metastatic recurrence [13]. TP53 mutations have been associated to poor differentiation and locoregional recurrence [13]. Finally SMAD4 mutations have been associated to portal vein invasion, perineural invasion and lymph vessel invasion [14]. Besides evidences on those mountain genes there is a growing body of evidence of several stage and prognosis predictors: MET protein over expression significantly correlated with increased TNM stage and worsened survival [15]; EDIL3 expression was significantly up-regulated in PDAC in both cell lines and clinical specimens and correlated with patients' TNM stage, T classification and overall survival times [16]; BRCA1/BRCA2 mutation predict significantly shorter disease-free and overall survival [17]; High RAB27A expression was significantly associated with vascular invasion and tumor stage [18].

Those data have been validated on samples of the tumor and thus usually on specimens since only 53,3% of ultrasound guided biopsies provides enough material for histopathology and/or immunohistochemistry [19]. Recent evidences show that pancreatic cancer is one of the 4 cancers, along with colorectal, gastroesophageal and breast cancer, in which detectable ctDNA levels are present either in localized either in metastatic stages [20]. As seen in previous studies, the concordance between plasma and primary tumor mutations is as high as 100% in pancreatic cancer [9, 21].

The study of those mutations on liquid biopsy would provide non-invasive informations on preoperative staging and pattern progression and therefore a more accurate therapeutic planning.

AIM OF THE PROJECT This study is a pilot study. Genomic data will be described along with the clinical features of the disease for each patient.

The primary objective of this study is to evaluate the relationship of mutations of the main genes responsible for pancreatic cancer progression (K-RAS, CDKN2A, SMAD4 and TP53) along with others related to stage and prognosis (MET, BRCA1/BRCA2, EDIL3 and RAB27A) detected on peripheral blood and time to progression and vascular invasion in patients undergoing pancreatectomy. Other outcomes of interest will be clinicopathological and operative features such as operative time, type of vascular resection, clinical stage, pathological T-, N- and M-stage, margin status and overall-survival.

MATERIALS AND METHOD Patients with non-metastatic PaC undergoing pancreatic resection will be enrolled from surgery departments of a multi center international web. Blood samples in EDTA will be centrifugated at 2300 rpm for 10min twice. Liquid biopsies will be preoperatively collected and stored at -80°. An protected anonymous multi parametric database will be filled with all relevant patient's and disease's informations (sex, birth date, comorbidities, BMI, lab tests, imaging clinical stage, operation time, operation type, ICU stay, Dindo-Clavien post-operative morbidity, pathological stage, TNM, Grading, ajcc 8th stage, perineural, vascular and lymphatic invasion, resection margins, disease free survival, overall survival) and K-RAS, CDKN2A, SMAD4, TP53, MET, BRCA1/BRCA2, RAB27A and EDIL3 somatic genetic mutations on liquid biopsy with Next Generation Sequencing (NGS).

Recruitment will be multicentric: AOU sant'andrea di Roma, Policlinico Universitario Gemelli di Roma, St Vincent University Hospital Dublino, Policlinico Universitario di Modena.

STATISTICAL ANALYSIS Assessment of relationship among gene mutations and vascular invasion will be performed by means of logistic regression models. An overall score will be prepared based on rounding of estimated log-odds ratios and its performance evaluated by means of ROC curves and C-statistic.

An evaluation of the conditional association will also be performed by means of multivariable logistic regression models, which will be selected by minimizing the Akaike Information Criterion.

To assess secondary outcomes we will also use Cox regression models in a similar fashion.

To the best of our knowledge, there is information in the literature regarding the frequency of vascular invasion in pancreatectomies with vascular resection (56.7%), and about mutations of the genes of interest in pancreatic ductal adenocarcinoma (the lowest mutated allele frequency being 30% for CDKN2A). Since there is no previous hypothesis on the relationship among genes and vascular invasion, and since this study simply involves the analysis of a small amount of blood posing no additional risk for the patient, we decided to perform a pilot study based on 50 patients. A fully planned and adequately powered study will be then conducted on the basis of the results of this pilot study. The sample size of 50 is deemed small enough (considering, as stated, the no additional risk and no use of invasive techniques) but large enough to guarantee that the final contingency table is not sparse. All analyses will be conducted with the R software.

CONCLUSIONS Non-metastatic PaC represent a challenge in staging and prognosis definition. A patient tailored treatment requires an accurate preoperative staging. Up to now more than 40% of patients thought to be resectable exit the operating room unresected and another 40% of resected patients recur within one year. Among patients who are offered a surgery, some would benefit of a neoadjuvant treatment and some others of palliative treatments. This is why a reliable preoperative staging is essential in therapeutic decisional making. Invasive diagnostic tests of the pancreas are expensive and risky due to its central retroperitoneal location. Liquid biopsy offers a non invasive tumor characterization. This study will be the first analyzing the relationship of the four major genes, along with 5 other prognostic genes, involved in pancreatic cancer genesis and progression and T stage with special attention to early recurrence and vascular invasion, by liquid biopsy. This will result in a better resource allocation, treatment planning and patients outcomes.

REFERENCES

1. Cancer Res. 2014; 74: 2913-2921
2. Ann Surg Oncol. 2016 Jun;23(6):2028-37
3. Cochrane Database Syst Rev. 2016 Jul 6;7:CD009323
4. J Gastrointest Surg 2004 Dec;8(8):935-49
5. Ann Surg 2018 Mar 23 Epub ahead of print.
6. J Gastrointest Oncol. 2015 Aug;6(4):418-29.
7. Ann Surg. 2017 Dec 7. doi: 10.1097/SLA.0000000000002600. [Epub ahead of print]
8. Cancer. 2012 Oct 15;118(20):5132-9
9. Science 2018 Feb 23:359(6378):926-930.
10. Oncogene. 2013 Nov 7;32(45):5253-60.
11. Clin Cancer Res 2012 Nov 15;18(22):6229-47.
12. Cancer 2015 Jul 1;121(13):2271-80.
13. Ann Surg 2013 Aug;258(2):336-46.
14. Pancreas 2015 May;44(4):660-4.
15. World J Gastroenterol 2014 Jul 14;20(26):8458-70.
16. Oncotarget 2016 Jan 26;7(4):4226-40.
17. J Am Coll Surg 2018 Apr;226(4):630-637.
18. Med Oncol (2015) 32:372.
19. Gastrointest Endosc. 2016 Oct;84(4):670-8.
20. Sci Transl Med 2014 Feb 19;6(224):224ra24.
21. Proc Natl Acad Sci U S A 2017 Sep 19;114(38):10202-10207.