KRAS, GNAS, and RNF43 mutations in intraductal papillary mucinous neoplasm of the pancreas: a meta-analysis

Background The prevalence and clinical significances of KRAS, GNAS, and RNF43 mutations in patients with pancreatic intraductal papillary mucinous neoplasm (IPMN) remain elusive. To evaluate the incidence of the gene mutations and clinicopathologic differences between KRAS and GNAS mutations in pancreatic cystic lesions, we performed a meta-analysis of published 33 KRAS, 11 GNAS, and 4 RNF43 studies including 1253, 835, and 143 cases, respectively. Methods We pooled the results of relevant studies identified using the PubMed and EMBASE databases. The effect sizes of outcome parameters were computed by the prevalence rate, weighted mean difference, or odds ratio (OR) using a random-effects model. Results The pooled prevalence of KRAS, GNAS, and RNF43 mutations in IPMN was 61, 56, and 23 %, respectively. The KRAS (OR 7.4 and 71.2) and GNAS (OR 30.2 and 15.3) mutations were more frequently found in IPMNs than in mucinous cystic neoplasms and in serous cystadenomas, respectively. Of the microscopic subtypes of IPMN, KRAS and GNAS were frequently mutated in gastric type (OR 2.7, P < 0.001) and intestinal type (OR 3.0, P < 0.001), respectively. KRAS mutation was infrequently found in high-grade dysplasia lesions of IPMN (OR 0.6, P = 0.032). GNAS mutation was associated with male (OR 1.9, P = 0.012). Conclusions This meta-analysis supports that KRAS and GNAS mutations could be diagnostic markers for IPMN. In addition, the frequencies of KRAS and GNAS mutations in IPMNs are highly variable according to the microscopic duct subtypes, reflecting their independent roles in the IPMN-adenocarcinoma sequence. Electronic supplementary material The online version of this article (doi:10.1186/s40064-016-2847-4) contains supplementary material, which is available to authorized users.


Background
Intraductal papillary mucinous neoplasm (IPMN) of the pancreas is a mucin-producing and cystic tumour growing inside the pancreatic duct and forming papillary projections (Klöppel et al. 2014;Klöppel and Kosmahl 2001). IPMN is considered as a precursor of pancreatic adenocarcinoma and comprised of about 16-24 % of cystic pancreatic lesions (Klöppel et al. 2014;Klöppel and Kosmahl 2001). IPMN forms a multilocular cystic lesion and is difficult to distinguish from mucinous cystic neoplasm (MCN) (Klöppel et al. 2014;Klöppel and Kosmahl 2001).
Therefore, in this meta-analysis, we aimed to know the exact prevalence of KRAS, GNAS, and RNF43 mutations in IPMN patients, and the difference between the frequency of these mutant genes in pancreatic cystic lesions. In addition, we investigated whether KRAS and GNAS mutations have clinicopathologic significances in patients with IPMN.

Data collection and selection criteria
We searched PubMed (http://www.ncbi.nlm.nih.gov/ pubmed) and EMBASE (www.embase.com) using the keywords "KRAS", "GNAS", "RNF43", "pancreas" and "intraductal papillary mucinous neoplasm". We also manually searched the reference lists of the identified articles. Duplicate data or overlapping articles were excluded by examining the authors' names and affiliations. Original articles reporting cases of KRAS, GNAS, and RNF43 mutations published before June 2015 were included. When multiple articles were published by the same authors or institutions, the most recent or single informative article was selected. Articles lacking clinicopathologic data for meta-analysis, review articles without original data, conference abstracts, case reports, and articles that dealt with cell line or animal were excluded. In addition, immunohistochemical studies of RAS mutation were also excluded. There were no geographic or language restrictions. The selection process of the articles is shown in Fig. 1.

Data pooling and statistics
Meta-analysis was performed as previously described (Lee et al. 2011). Briefly, effect sizes for each study were calculated by prevalence rate or odds ratio (OR) and the corresponding 95 % confidence interval (CI) using the Mantel-Haenszel method. The prevalence rate, weighted mean difference (WMD), or OR was combined using a random-effects model (DerSimonian-Laird method). Statistical heterogeneity among studies was evaluated using the Cochrane Q test and I 2 statistics. The I 2 statistic refers to the percentage of variation across studies that is due to heterogeneity rather than chance and does not inherently depend on the number of studies considered [I 2 = 100 % × (Q-df )/Q]. We clarified the cutoff of I 2 statistics for assignment of low (<25 %), moderate (25-50 %), and high (>50 %) heterogeneities. If I 2 value was more than 50 %, subgroup analysis was done. Sensitivity analyses were performed to examine the influence of each study on the pooled prevalence rate, WMD, or OR by serially omitting an individual study and pooling the remaining studies. Publication bias was examined by funnel plots and Egger's tests for the degree of asymmetry. Publication bias was assumed to be present if the P value was less than 0.1. The pooled analysis was performed using Comprehensive Meta-analysis Software version 2.0 (Biostat, Englewood, NJ, USA).

Age and sex
The incidence of KRAS and GNAS mutations in patients with IPMN according to the patient's sex was compared in eleven (Fritz et al. 2009;Hosoda et al. 2015;Kobayashi et al. 2008;Kondo et al. 1997;Mulligan et al. 1999;Schönleben et al. 2008;Singhi et al. 2014;Tada et al. 1991;Uemura et al. 2003;Wada et al. 2004;Wu et al. 2011b) and six (Hosoda et al. 2015;Ideno et al. 2015;Kanda et al. 2013;Singhi et al. 2014;Takano et al. 2014;Wu et al. 2011b) studies, respectively. Five (Fritz et al. 2009;Kobayashi et al. 2008;Schönleben et al. 2008;Singhi et al. 2014;Wada et al. 2004) and three (Kanda et al. 2013;Singhi et al. 2014;Wu et al. 2011b) studies presented mean age with standard deviation (SD) or p value according to KRAS and GNAS mutations, respectively. The mean age of IPMN patients with KRAS mutation ranged from 63.67 to 70.15 years, whereas the mean age of those with wild-type KRAS ranged from 64.3 to 68.58 years. The mean age of IPMN patients with GNAS mutation ranged from 62.14 to 69.54 years, 138 papers were retrieved for the key words and their abstracts were reviewed 70 full texts were reviewed 36 studies were included in this meta-analysis 68 articles were excluded, due to: (1) 22 papers were review articles (2) 7 papers were case report (3) 10 papers were animal study (4) 29 articles were obviously irrelevant or not associated with GNAS, KRAS, or RNF43 mutations of intraductal papillary mucinous neoplasm of pancreas 34 articles were excluded due to: (1) 23 papers were duplicated publications (2) 7 papers did not provide sufficient data (3) 4 papers were immunohistochemical studies

Sensitivity analysis and publication bias
The sensitivity analyses showed that none of the studies affected the pooled prevalence rate, OR, or WMD with CIs, except the pooled analysis of GNAS mutation between the genders and of KRAS mutation between high grade and the other grades of IPMNs (Additional file 3: Fig. S1). Through the funnel plots and the Egger's regression tests, the pooled results from KRAS mutation between tumour locations within the pancreas, GNAS mutation between mean tumour size and between intermediate grade and the other grades, and KRAS and GNAS mutations between microscopic duct subtypes of IPMNs showed the possibility of publication bias. However, other pooled analyses showed no evidence of publication bias (Additional file 4: Table S3) (Additional file 5: Fig. S2).

Discussion
This pooled analysis using data from 1253, 835, and 143 pancreatic IPMN patients revealed that overall KRAS, GNAS, and RNF43 mutations were detected in 61, 56, and 23 %, respectively. These gene mutation rates did not differ according to the ethnicity, detection methods, and specimen type. This meta-analysis showed that the frequencies of KRAS and GNAS mutations in IPMN patients were considerably variable among microscopic duct subtypes. The most common preoperative challenge is to distinguish IPMN from other cystic lesions of the pancreas. There are three primary types of pancreatic cystic neoplasm: IPMN, MCN, and SCA (Wu et al. 2011b). Most of the pancreatic adenocarcinomas develop from IPMNs, followed by MCNs. In contrast, SCAs do not give rise to invasive adenocarcinomas (Wu et al. 2011b). Until now, preoperative cystic fluid evaluation for CEA, amylase, DNA methylation, and microRNA expression remains suboptimal, partly because of their lack of disease specificity (Weinstein et al. 2004). Based on the high frequencies and significant ORs of KRAS and GNAS mutations in IPMNs compared to the other cystic lesions, particularly MCNs, the combined detection of KRAS and GNAS mutations from the cystic fluid would be highly valuable in the preoperative diagnosis of IPMNs. This pooled analysis found that different mutational profile between KRAS and GNAS was significantly related to the microscopic subtypes of IPMNs, which are a determinant for the subtypes of invasive adenocarcinomas. Over 30 % of intestinal and pancreaticobiliary type IPMNs develop colloid and tubular type adenocarcinomas, respectively (Klöppel et al. 2014 type IPMNs rarely develops into invasive adenocarcinomas. When gastric type IPMNs progress to adenocarcinomas, it is the tubular type (Klöppel et al. 2014). The IPMNs with colloid adenocarcinoma is known to have a better prognosis than those with tubular adenocarcinoma (Klöppel et al. 2014;Tan et al. 2015). Colloid adenocarcinomas arising from IPMNs were associated with a high frequency of GNAS mutation (Tan et al. 2015). In agreement with the previous study (Tan et al. 2015), our results suggest that KRAS and GNAS mutational pattern may represent different pathways in the IPMN-adenocarcinoma sequence.
The GNAS gene encodes the α-subunit of the stimulatory G-protein (Gαs). Somatic activating GNAS mutation results in an elevated level of cyclic adenosine monophosphate (cAMP) and in uncontrolled growth signalling (Landis et al. 1989;Weinstein et al. 2004). GNAS mutation has been found in various tumours, fibrous dysplasia, and McCune-Albright syndrome (Landis et al. 1989;Weinstein et al. 2004). Interestingly, most of the GNAS mutations at codon 201 in IPMNs result in either an R201H or an R201C substitution, which are the same mutation as in endocrine neoplasms (Landis et al. 1989;Weinstein et al. 2004). The endocrine tumours with activating GNAS mutations have been supposed to be associated with hormonal secretion.
Recently, inactivating nonsense mutations of RNF43 gene that encodes a protein with E3 ubiquitin ligase activity were found in IPMNs (Amato et al. 2014;Macgregor-Das and Iacobuzio-Donahue 2013;Sakamoto et al. 2015;Tan et al. 2015;Wu et al. 2011a). Our metaanalysis found that RNF43 mutation was not associated with clinicopathologic parameters of patients with IPMN (Additional file 6: Table S4). Due to the lack of published articles, further studies need to clarify the roles and characteristics of RNF43 mutation in IPMN patients.
This meta-analysis revealed that KRAS and GNAS mutations are not associated with the malignant potential or prognosis in patients with IPMN. This meta-analysis showed that frequency of KRAS mutation was rather lower in high-grade dysplasia than low-and/or intermediate-grade dysplasia. However, further more studies are needed to confirm the results. The frequency of GNAS mutation in IPMN patients did not differ among the three grades of dysplasia and in the absence or presence of associated adenocarcinoma. The association between GNAS mutation and the prognosis of patients with IPMN has been the subject of considerable controversy. GNAS mutation was significantly associated with high-grade dysplasia (Wu et al. 2011b), whereas wild-type GNAS in IPMNs was significantly related to the development of adenocarcinoma (Ideno et al. 2015). However, as with this meta-analysis, other studies failed to show significant relationships between GNAS mutation, dysplasia grades, and the presence of adenocarcinoma (Amato et al. 2014;Hosoda et al. 2015;Kuboki et al. 2015;Singhi et al. 2014).
It has been well known that pancreatic cancers are more frequent in Ashkenazi Jews and African groups. However, the frequency of IPMN between races has not been known because of its rare incidence. Therefore, we simply classified IPMN patients into the Caucasian and Asian groups in this study, although the genetic changes of diverse and detailed ethnicity would be an interesting issue and broaden the novel biological pathway of IPMN.
The present meta-analysis has some limitations. First, the possibility of publication bias could not be completely excluded. Second, the individual study used in this metaanalysis was done with relatively small sample sizes, due to the rare occurrence of IPMNs. Lastly, the different studies did not only use different methods for mutation detection but also different tumour materials, such as cystic fluid versus tissue specimen. This might confound the mutation detection rates.
In summary, this meta-analysis provides sensitive and specific diagnostic roles of KRAS and GNAS mutations for detecting the IPMN among the pancreatic cystic lesions. Furthermore, KRAS and GNAS mutations hint a possibility that patients with IPMN have which form of microscopic subtype.