Activating mutations of GNAS and KRAS in cystic fluid can help detect intraductal papillary mucinous neoplasms of the pancreas

Abstract

Evaluation of: Singhi AD, Nikiforova MN, Fasanella KE, et al. Preoperative GNAS and KRAS testing in the diagnosis of pancreatic mucinous cysts. Clin. Cancer Res. 20(16), 4381-9 (2014).

Intraductal papillary mucinous neoplasms (IPMNs) of the pancreas have a risk of malignant transformation following an adenoma–carcinoma sequence. Surgical resection is often required, especially for main pancreatic duct IPMNs (MD-IPMNs). There is an urgent need for novel biomarkers to reliably differentiate IPMNs from more benign pancreatic cysts and therefore avoid unnecessary surgery. DNA sequencing has demonstrated that guanine nucleotide binding protein alpha stimulating (GNAS) activity polypeptide 1 mutations play a driving role in IPMN development. GNAS mutations have been shown to be highly specific for IPMNs, whereas oncogenic KRAS mutations have been associated with mucinous differentiation. The evaluated article by Singhi et al. helps to define the role of these mutations as biomarkers in preoperative endoscopic ultrasound fine-needle aspiration samples for detecting IPMNs. They found that the presence of a GNAS and/or a KRAS mutation was highly specific and sensitive for IPMNs.

Keywords:

• biomarkers
• diagnostic
• endoscopic ultrasound
• GNAS
• intraductal papillary mucinous neoplasm
• KRAS
• pancreatic cyst

Cystic lesions of the pancreas can either be inflammatory or proliferative [1]. Differentiating between low- and high-risk pre-malignant tumors can be difficult and the consequences of missing the chance for a curative resection in patients who are suitable can be devastating. To prevent this, many centers recommend routine excision of all suspicious pancreatic cystic tumors (PCTs), potentially exposing patients with benign disease to the unjustifiable risks of major surgery. PCTs can be classified into non-mucinous tumors with negligible malignant potential, such as the serous cystadenomas (SCA), and those with significant malignant potential, such as intraductal papillary mucinous neoplasms (IPMNs) and mucinous cystic neoplasms (MCNs). The mucinous cysts have the potential to give rise to in situ or invasive carcinoma, via an adenoma–carcinoma sequence. Thus, a correct pre-operative diagnosis and evaluation of a PCT is crucial for clinical decision-making.

Endoscopic ultrasound fine-needle aspiration (EUS-FNA) has become an essential tool for cytopathologic confirmation of pancreatic lesions, and pre-operative planning. Recent meta-analysis has shown that pooled specificity for cytology is 93% (95% CI: 90–95) with a sensitivity of 54% (95% CI: 49–59) for a mucinous cyst [2]. Carcinoembryonic antigen (CEA) levels are raised in mucinous cyst fluid while low in non-mucinous lesions, and the optimum threshold of 192 ng/ml has high discriminatory accuracy [3]. Pooled specificity for CEA measurement is 88% (95% CI: 83–91), with a sensitivity of only 63% (95% CI: 59–67) for mucinous differentiation [2]. Therefore, it is clear that additional markers are required to improve the detection of pancreatic mucinous neoplasms.

Recent deep sequencing studies have determined the genetic alterations in the different types of pancreatic cyst (i.e., SCA, IPMN, MCN, solid pseudopapillary neoplasm [SPN]) to improve diagnosis and patient management [4–6]. Of note, these experiments have revealed frequent activating mutations of GNAS and KRAS in IPMNs. Indeed, GNAS mutations appear to be a hallmark molecular alteration of IPMNs (prevalence ∼66%) [6]. However, the ability and clinical usefulness of identifying these mutations pre-operatively has still not been determined.

In the evaluated article, Singhi et al. [7] tested for GNAS and KRAS mutations in a large cohort of pancreatic cyst fluid samples from EUS-FNA to assess their diagnostic utility and correlation with clinicopathologic factors.

Summary of methods & results

Methods

Included cases

Pre-operative pancreatic cyst fluid was obtained by EUS-FNA and only samples from cases with diagnostic uncertainty were sent for molecular analysis. In total, GNAS and KRAS testing was performed on 91 pancreatic cysts (41 IPMNs, 9 IPMNs with adenocarcinoma, 16 MCNs, 10 cystic pancreatic neuroendocrine tumors [PNETs], 9 SCAs, 3 retention cysts, 2 pseudocysts and 1 lymphoepithelial cyst). Diagnoses were confirmed by surgical histology.

Mutational analysis

Genomic DNA was isolated from all cysts (200 μl of cyst fluid required) and mutations were detected by sequencing and analysis of sequence electropherograms. Of note, the ability to do mutational analysis was not affected by cyst size or fluid viscosity. In addition, eight cyst fluids from the SCA group were assessed for loss of heterozygosity at the von Hippel-Lindau locus.

Results

Pancreatic cyst cohort

The mean age of the patients was 60.2 years (20–87 years) and 60% were female. The average cyst size was 3.0 cm (0.7–9.9 cm). The amount of cyst fluid was insufficient for CEA analysis in 21 (23%). Furthermore, 37% (n = 34) of EUS-FNAs were either less than optimal or unsatisfactory (i.e., due to poor cellularity) for cytological diagnosis. All cysts were subsequently surgically resected except for 8 of 9 SCAs. Even though the diagnosis of a SCA is fairly safe with current clinical tools, it is not ideal to use samples with a ‘presumed’ diagnosis without histological confirmation for association with molecular markers. Indeed, in 6 SCAs (75%), von Hippel-Lindau loss of heterozygosity was detected as previously described [4].

GNAS & KRAS mutations are frequent in IPMNs

GNAS mutations were seen in 16 IPMNs (39%) and 2 (22%) with associated adenocarcinoma. Whereas, KRAS mutations were detected in 28 IPMNs (68%), 7 IPMNs (78%) with associated adenocarcinoma and 1 MCN (6%). Mutations of both were found in 10 IPMNs (24%) and 1 (13%) with associated adenocarcinoma. Previous studies have found 25–51% of IPMNs harbor concurrent mutations in GNAS and KRAS [5,6,8].

However, a mutation in either GNAS or KRAS was found in 34 IPMNs (83%) and 8 (89%) with associated adenocarcinoma. Importantly, mutations were absent in cystic PNETs, SCA, retention cysts, pseudocysts and the lymphoepithelial cyst.

GNAS and/or KRAS mutations were significantly associated with male gender, increasing age, smaller cyst size, increased fluid viscosity and raised CEA (71% of which were either IPMN with adenocarcinoma or mucinous cyst).

KRAS mutations were more frequent in branch-duct IPMNs compared with MD-IPMNs. Furthermore, GNAS mutations were 100% associated with the intestinal subtype of IPMNs, as seen previously [6,9], whereas no relation between KRAS mutations and subtype was found.

The presence of GNAS and/or KRAS mutation is highly specific & sensitive for IPMNs

The identification of GNAS mutations had 100% specificity (95% CI: 89–100), but only 36% sensitivity (95% CI: 23–51) for IPMNs. Whereas KRAS mutations performed better with 98% specificity (95% CI: 86–100) and 70% sensitivity (95% CI: 55–82). However, having a GNAS and/or KRAS mutation detected an IPMN with 98% specificity (95% CI: 86–100) and 84% sensitivity (95% CI: 70–92). This was in contrast to cyst fluid CEA levels which only had a 69% specificity (95% CI: 52–83) and 74% sensitivity (95% CI: 55–86) for IPMN; and a 90% specificity (95% CI: 68–98) and 69% sensitivity (95% CI: 54–81) for mucinous differentiation. However, CEA measurements were only possible in 21 cases (23%). Similarly, a cytological diagnosis suspicious for a mucinous lesion had poor specificity and sensitivity, 71 and 60% respectively. Although the presence of a GNAS and/or KRAS mutation had a specificity of 100% (95% CI: 83–100) for identifying a mucinous cyst, its sensitivity was only 65% (95% CI: 52–76). When the presence of a gene mutation was combined with elevated CEA, the specificity and sensitivity were markedly improved to 90% (95% CI: 68–98) and 86% (95% CI: 72–94) for mucinous differentiation.

Expert commentary & five-year view

The purpose of searching for biomarkers in pre-operative pancreatic cystic fluid samples is to discover whether molecular techniques can improve clinical decision-making and patient management. Singhi et al. [7] have demonstrated that the presence of either a GNAS or/a KRAS mutation in EUS-FNAs can potentially help to differentiate between non-mucinous and mucinous cysts. It is important to note, however, that recent consensus guidelines do not recommend EUS-FNA for cyst fluid analysis for high-risk mucinous cysts (i.e., MCNs and MD-IPMNs) for fear of peritoneal dissemination caused by leak of possible malignant contents [10]. Instead, EUS-FNA is proposed only for the evaluation of small branch-duct-IPMNs without ‘worrisome’ features [10]. These recommendations make the use of novel molecular techniques even more relevant and important to translate into clinical practice.

Oncogenic-mutated KRAS is expressed in up to 95% of PDAC cases and is involved in the initiation or early phase of pancreatic tumorigenesis [11,12]. Indeed, frequent mutations of KRAS have been previously detected in the pancreatic juice, tissues and EUS-FNAs of patients with PDAC and IPMN (Supplementary Table 1 [supplementary material can be found online at www.informahealthcare.com/suppl/10.1586/14737159.2015.1002771_Suppl). Interestingly, the frequency of mutated KRAS remains consistent as IPMN progresses (i.e., from adenoma to carcinoma) [13]. However, KRAS mutations have been shown to have high specificity, but low sensitivity for mucinous differentiation (100–96% and 54–45%, respectively) [14,15].

The GNAS gene encodes the α-subunit of the stimulatory G-protein (Gαs), which mediates the regulation of adenylate cyclase activity through G-protein-coupled receptors and activation results in an elevated cAMP level. Somatic GNAS-activating mutations cause McCune Albright syndrome, pituitary adenomas, endocrine tumors, and/or fibrous dysplasia of bone, primarily through increased cAMP levels, which in some cells can lead to proliferation [16,17]. Activating GNAS mutations have also been frequently observed in gastric and duodenal pyloric-gland adenomas [18], and colorectal villous adenomas and are often associated with KRAS mutations [19]. Interestingly, GNAS and KRAS mutations have been detected in 29 and 32% of intraductal papillary neoplasms of the bile duct respectively (especially those with intestinal differentiation) [20].

Previous studies have shown that GNAS is mutated in IPMNs, and in some IPMNs with associated adenocarcinoma (Supplementary Table 1). However, the functions of GNAS in IPMN development and pancreatic tumorigenesis have not yet been fully elucidated. Activated GNAS may be associated more with tumor initiation, rather than with progression, as mutations have been seen in low-grade tumors [5,21] and are not associated with survival outcomes [5,9]. Indeed, the current study [7] provides further evidence for this, as GNAS mutations did not correlate with grade of dysplasia or presence of adenocarcinoma, suggesting that it is an early driver gene in IPMN pathogenesis. This is also supported by the fact that GNAS mutations have been found in ‘incipient’ IPMNs (<1 cm diameter) that were present in pancreata being resected for other reasons [22], and that they can also be detected in duodenal juice even before the patient has a radiologically noticeable IPMN [21]. These factors make detecting GNAS mutations slightly disappointing as a ‘complete’ IPMN biomarker, because the most difficult task clinically is to identify those lesions containing high-grade dysplasia requiring definitive treatment rather than ‘watchful-waiting’.

Komatsu et al. [23] found that exogenous mutated GNAS does not promote pancreatic ductal cell proliferation in vitro. They suggest that mutated GNAS is therefore unable to induce tumor growth and other synergistic genetic aberrations are required. This is supported by studies in transgenic mice where the GNAS R201C mutation alone was insufficient to induce colorectal tumorigenesis, but when combined with inactivation of the adenomatous polyposis coli (APC) gene, increased adenoma formation resulted [24]. Therefore, it would be interesting to create pancreas-specific transgenic mice with activated GNAS R201H or R201C mutations alone and combined with KRAS^G12D to examine the phenotypes in more detail.

Komatsu et al. [23] also showed that mutated GNAS is able to drastically alter gene expression profiles in pancreatic ductal cells, including expression of MUC2 and MUC5AC, and this may determine IPMN phenotype. For example, MUC2 is specific to intestinal-type IPMNs [23], which also frequently have GNAS mutations, and these are associated with the less aggressive colloid carcinoma [25]. However, gastric-type IPMNs have been shown to be frequently associated with concomitant PDACs, and these lesions may not have GNAS mutations, but instead mutated KRAS and negative MUC2 expression, and can also progress to the more aggressive tubular carcinoma [26].

Finally, measuring molecular markers in cyst fluid is a sensible approach for pre-operative stratification. Indeed, Takano et al. [27] found that pancreatic juice might be more representative of mutations from the entire tumor than tissue samples in certain conditions. Moreover, Wu et al. showed that gene mutations in cyst fluid are derived from the cyst walls, indicating that it can provide an excellent representation of the cells in an IPMN [6].

In conclusion, GNAS and/or KRAS mutations appear to be highly specific and sensitive for detecting IPMNs, but have low sensitivity for mucinous cysts (65%). It may be that the addition of other markers can be used to create a signature and improve accuracy [4].

Key issues

• GNAS and KRAS mutations were able to be detected in endoscopic ultrasound fine-needle aspiration cyst fluid samples pre-operatively.

• Mutational analysis was successfully performed on all cyst fluids, compared with 23% insufficient for Carcinoembryonic antigen levels and 37% unsuitable for cytological diagnosis.

• GNAS and KRAS mutations were absent in benign pancreatic cysts.

• GNAS and KRAS mutations cannot differentiate between grades of dysplasia. Therefore, using a panel of selected genomic markers and detecting accumulating genetic abnormalities may allow correlation with increasing degrees of neoplasia.

• GNAS and/or KRAS mutations are highly specific and sensitive for intraductal papillary mucinous neoplasms, but their ability to identify mucinous cysts is poor. Further markers are required.

Acknowledgements

The authors gratefully acknowledge the critical reading and helpful comments provided by Christopher L Wolfgang Wolfgang (Johns Hopkins Hospital, Baltimore, USA).

Financial & competing interests disclosure

The authors have no relevant affiliations or financial involvement with any organization or entity with a financial interest in or financial conflict with the subject matter or materials discussed in the manuscript. This includes employment, consultancies, honoraria, stock ownership or options, expert testimony, grants or patents received or pending, or royalties.

No writing assistance was utilized in the production of this manuscript.