Signal transduction pathway mutations in gastrointestinal (GI) cancers: a systematic review and meta-analysis

The present study was conducted to evaluate the prevalence of the signaling pathways mutation rate in the Gastrointestinal (GI) tract cancers in a systematic review and meta-analysis study. The study was performed based on the PRISMA criteria. Random models by confidence interval (CI: 95%) were used to calculate the pooled estimate of prevalence via Metaprop command. The pooled prevalence indices of signal transduction pathway mutations in gastric cancer, liver cancer, colorectal cancer, and pancreatic cancer were 5% (95% CI: 3–8%), 12% (95% CI: 8–18%), 17% (95% CI: 14–20%), and 20% (95% CI: 5–41%), respectively. Also, the mutation rates for Wnt pathway and MAPK pathway were calculated to be 23% (95% CI, 14–33%) and 20% (95% CI, 17–24%), respectively. Moreover, the most popular genes were APC (in Wnt pathway), KRAS (in MAPK pathway) and PIK3CA (in PI3K pathway) in the colorectal cancer, pancreatic cancer, and gastric cancer while they were beta-catenin and CTNNB1 in liver cancer. The most altered pathway was Wnt pathway followed by the MAPK pathway. In addition, pancreatic cancer was found to be higher under the pressure of mutation compared with others based on pooled prevalence analysis. Finally, APC mutations in colorectal cancer, KRAS in gastric cancer, and pancreatic cancer were mostly associated gene alterations.


Scientific RepoRtS
| (2020) 10 www.nature.com/scientificreports/ The results of meta-analysis revealed that pooled prevalence of signal transduction pathway mutations in CRC was 17% (95% CI: 14%, 20%) and there was a high heterogeneity between these studies in estimating the prevalence (I-squared = 97.63%, P = 0.0001) (Fig. 3). Also, the subgroup analysis for heterogeneity was performed in CRC included studies based on the different pathways (heterogeneity plot in Fig. 4), detection method (heterogeneity plot in Fig. 5), and involved genes (heterogeneity plot in Fig. 6). The CI of test (Egger's test) included zero, thus there was no significant bias in the results (Egger's test = − 0.692, P = 0.109, 95% CI: − 1.54 to 0.156). The pooled prevalence funnel plot in CRC signal transduction pathway mutations is illustrated in Fig. 7 and the Subgroup analyses of pooled prevalence signal transduction pathway mutations in CRC are summarized in Table 3.
Signaling pathway mutations in liver cancer (Lc). The search on liver cancer resulted in a total of 1056 hepatocellular carcinoma (HCC) and 174 hepatoblastoma participants in 21 studies. There were different ranges of mutations in these studies, which are listed in supplementary Table 4. The Wnt signaling was the most evaluated pathway in which the CTNNB1 gene mutation ranges were evaluated to be 12-75% and the betacatenin genes had the mutation ranges of 2.8-41%. In addition, the mutation ranges in p53 were 1.2 to 61% and the JAKs in the JAK signaling pathway were observed to be 1.2 to 16%.
The results of meta-analysis showed that pooled prevalence of signal transduction pathway mutations in LC was 12% (95% CI: 8-18%) and there was a high heterogeneity between these studies in estimating the prevalence (I-squared = 85.34%, P = 0.0001) (Fig. 8). Also, since the CI of the test (Egger's test) included zero, there was no significant bias in the results (Egger's test = − 0.442, P = 0.411, 95% CI: − 0.65 to 1.53). The pooled prevalence funnel plot in LC signal transduction pathway mutations is illustrated in Fig. 8. Furthermore, the Subgroup analyses of pooled prevalence signal transduction pathway mutations in LC are summarized in Table 3.
Signaling pathways mutations in pancreatic cancer 1 . In a total of 9 studies, 392 PC patients were studied with the KRAS and PIK3CA mutations reported 42 to 92% and 2.7 to 12%, respectively. More data are shown in supplementary Table 5.
The results of meta-analysis showed that pooled prevalence of signal transduction pathway mutations in pancreatic cancer was 20% (95% CI: 5-41%) and there was a high heterogeneity between these studies in estimating the prevalence (I-squared = 97.14%, P = 0.0001) (Fig. 9). Also, the CI of the test (Egger's test) included zero, s no significant bias was present in the results (Egger's test = − 1.351, P = 0.568, 95% CI: − 6.37 to 3.66). The pooled prevalence funnel plot in PC signal transduction pathway mutations is illustrated in Fig. 9. Furthermore, the Subgroup analyses of pooled prevalence signal transduction pathway mutations in pancreatic cancer are summarized in Table 3.
Signaling pathways mutations in other Gi cancers. The other GI cancers included gastro-esophageal cancer, rectal cancer, esophageal squamous cell carcinoma, gallbladder carcinoma, and cholangiocarcinoma. Different signaling pathways in these GI cancers are listed in supplementary Table 6. Briefly, KRAS was the popular gene for mutation analysis ranging from 0% mutation in squamous cell anal carcinoma to 57% in small intestinal adenocarcinoma. BRAF was analyzed in 6 studies with its mutation reported to be 0-45%. Moreover, APC mutations were reported between 9.5 and 47% in different malignancies.
Signaling pathway mutation association with clinic-pathological features and patients survival. The extracted data about clinic-pathological features and patients survival were listed in supplement Tables 2 to 6. As glimpse, the clinic-pathological features statistically significant in association with signaling pathway mutations that they were mentioned in 2 individual studies for gastric cancer and 30, 6, 1 and 2 individual studies for CRC, LC, PC and other GI cancers, respectively.
Survival rate assessment in association with signaling pathway mutations were listed in supplement Tables 2 to 6. The survival rate or prognostic feature in association with signaling pathway mutations were mentioned in 1, 6, 1, 1, 0 and 1 included studies for CRC, LC, PC and other GI cancers, respectively.
The MAPK/ERK signaling was analyzed in the study reported by Sameer et al. 156 who found KRAS mutation to be 24% in 86 CRC patients. Meanwhile, Tong et al. 113 reported the highest rate (75%) of the KRAS mutations in CRC patients in codon 12 in 1506 individuals. Tong's study showed different mutation rates between the separate codons of the KRAS gene with the highest in codon 12 and the lowest (2.5%) in codon 61. Also, in the study conducted by Kawazoe et al. 127 on 264 metastatic colorectal cancers (mCRC), the KRAS exon 2 mutation was calculated to be 34%, as the highest mutation rate. In this study, BRAF mutation rate was reported to be 5.4%. The highest prevalence for the BRAF mutation reported in other studies was 78% 88 . This huge difference in the BRAF mutation rate could be due to the differences in the sample size and the method used for analysis.
The Wnt/beta-catenin signaling and PI3K/AKT signal have been assessed in a variety of studies. The Wnt/ beta-catenin was assessed in 18 different studies and the most evaluated genes were APC, beta-catenin, and CTNNB1. Fujimori et al. 26 showed that 37.5% of the 73 CRC patients had mutations in the exon 3 of the betacatenin gene. Also, Shitoh et al. 32 reported the rate of 3% for beta-catenin mutation in exon 3, and 27% in the high-frequency microsatellite instability (MSI-H). Furthermore, the APC gene mutations were assessed in 10 different studies with the lowest reported to be 33% in the study by Chen 88 study and the highest as 73% reported by Lee et al. 107 . The previous studies showed that the MSI could be associated with the in/del substitutions in genome hot spots which can initiate CRC tumorogenesis by increasing the mismatches indiscriminately [157][158][159] . Investigation on Wnt/beta-catenin signaling was firstly introduced by the association between APC gene and beta-catenin 160,161 . Other studies found the interactions of these genes with beta-catenin-Tcf (T-cell factor) complex suggesting the association of these genes with CRC omplication 162 . The role of APC gene in causing cancer was initially introduced in the familial adenomatous polyposis (FAP) 163 . This gene facilitates beta-catenin distorting. APC gene mutations influence beta-catenin and AXIN protein binding sites 164,165 . Moreover, they could maximize the protein stability and life cycle 166 . Thus, the carcinogenesis process is accelerated by altered signal transduction and cell cycle 167 . www.nature.com/scientificreports/ From among the studies which assessed the PI3 signal transduction pathway, the mutation of PIK3CA gene was reported in 20 studies ranging between 0 and 34%. Meanwhile, Thorstensen et al. 49 found p53 gene mutation rate to be about 18% in CRC patients.
There are variable reports in the matter of clinic-pathological association with mutations in the current study. In the conducted study by Sameer et al. 156 the clinic-pathological assessment indicated that, the SMAD4 mutations are more frequent in colon tumors and statistically associated with tumor grade and lymph nodes involvement. Tong and colleagues 113 reports the KRAS mutations are in association with gender and tumor site. Also, Kawazoe et al. 127 points out the BRAF mutations are associated with tumor location, site of metastasis and differentiation pattern. Meanwhile, Yang and colleagues 168 reports the association of the KRAS mutations with tumor location, type of tumor, differentiation pattern and gender of the patients. Furthermore, there were limited data about the association of the mutations in signaling pathways with survival rate in patients. Some studies suggested BRAF mutations 169 and SMAD4 mutations 140 are association with poor prognosis and survival rate. Highly variable and limited data about clinic-pathological features, survival and prognosis in association with signaling pathway mutations were extracted. The clinic-pathological features and patients survival association with signaling pathway mutations is one of the current study limitations and needs further investigation.
HCC is the fifth cause of death worldwide and is mostly inducted by the chronic liver disorders, such as viral hepatitis 170,171 . In LC patients, the Wnt signaling was the top research interest and the CTNNB1 was the most assessed gene. The CTNNB1 mutation was also investigated in HCC patients in different studies 118,129,131 . Purcell et al. 73 reported CTNNB1 mutations in 15% of hepatoblastoma patients while the reported prevalence in Ueda's study was 75% 74 . Our study subgroup analysis for liver cancer 145 studies showed that beta-catenin has higher mutation rate (20% (95% CI, 10-31%)) and the most altered pathway was Wnt (17% (95% CI, 11-23%)). It has been indicated that the CTNNB1 and P53 genes are the most involved genes in the HCC 172,173 . Moreover, the conducted studies showed that the P53 mutations were mostly associated with the Asian and African countries, while the CTNNB1 mutations were mostly associated with HCC in the Western countries 172,173 .
The pancreatic cancer is known as the forth cause of cancer mortality in the US with only 10% of the cases living more than 5 years 174 . Witkiewicz et al. 130 assessed different genes in MAPK/ERK, PI3K/AKT, and Wnt/ beta-catenin signaling pathways in pancreatic ductal adenocarcinoma patients. They showed that the AXIN1, KRAS, and PI3CA mutations rate were 5%, 92%, and 4%, respectively. Moreover, the high rate of KRAS mutations in pancreatic cancer patients was confirmed by the other studies 55,119,175 . Our study showed that in the subgroup analysis for pancreatic cancer, the KRAS was the most mutated gene (58% (95% CI: 31-83%)) and MAPK was the most altered pathway (31% (95% CI: 5-66%)). In GC, mutations were 14% (95% CI: 2-34%) for KRAS, 7% (95% CI: 1-17%) for MAPK, and 6% (95% CI: 2-12%) for PI3 pathways. In the pancreatic and gastric cancers, the most evaluated pathways were PI3 and MAPK. The KRAS gene generates a GTPase protein which is critical in regulating the cell proliferation and metabolism 176 . The mutations in KRAS leads to impaired cells activity enhancement and malignancy progression 177 .
Gastric Cancer (GC), as another invasive GI cancer, has significant mortality rate worldwide 178 . Zhang et al. 104 studied 100 advanced primary GC cases for the purpose of evaluating PI3K/AKT signaling pathway mutations. They suggested that the MAPK/ERK and PI3K/AKT pathways could be potential therapeutic targets for GC treatment 179,180 . The AKT gene produced a protein in the PI3K/Akt pathway which could play a role in tumorogenesis 80 . The mutations in the PIK3CA and AKT in PI3K/AKT pathway could affect downstream signaling pathway genes, like mTOR (mechanistic target of rapamycin kinase) and caspase 9, which are important in GC progression 104,181,182 . Wang et al. 99 investigated hedgehog pathway in GC patients and showed that the PTCH1 (patched 1) and SMO (smoothened) genes were mutated in 51.2% and 25.6% of the cases. Alterations in PTCH1 gene were associated with the basal cell carcinoma and basal cell nevus syndrome 183,184 .
Moreover, most of the studies included used PCR followed by the Sanger sequencing, as the method of choice. However, some studies used SSCP-PCR (single-strand conformational polymorphism PCR) to detect mutation. The method used the least was the NGS (next generation sequencing) as a preferred method in the recent years. The NGS can be used to analyze numerous samples at the same time and thus reduce the cost and the time required 185 . But the Sanger sequencing is an accurate and sensitive method for mutation analysis and it has been suggested for the confirmation of the NGS results 186 . Also, in the subgroup analysis for the GC, the method of detection could be mentioned as a potent source of the heterogeneity in the current study ( Table 3).
The major limitation in the current study was the extent of subject; it is suggested that further investigations use more narrowing strategies. Also, we aimed at minimizing the author bias in data extraction and screening biases using different authors and double check strategies. Also, it should be mentioned that the p53 signaling is not a canonical signaling pathway but due to the p53 non-transcriptional functions, the importance of this pathway in cancer generation, and its interaction with other signaling pathways, in the present study, we assessed p53 as individual pathway 3 .
In conclusion, progression of GI cancers is affected by signaling pathway mediators. Different studies have shown diverse results based on their population, method, and target gene. Our study concluded that the most important genes that are under mutation pressure include KRAS and PI3CA in the CRC, PC, and GC while beta-catenin and CTNNB1 are genes under mutation pressure for liver malignancies. Subgroup analysis and heterogeneity of the studies could illustrate more valid data between different studies for screening strategies. In this regard, signal transduction pathway mutations pooled prevalence was higher in PC and lower in GC (20% vs. 5%). Thus, PC is the most common cancer involved by signal transduction mediator's mutations. Among studied genes, KRAS in GC and pancreatic cancer and APC in CRC had the most association with cancer outcome. Moreover, MAPK had higher mutation rate among the studied pathways. Furthermore, PCR-SS method had the highest popularity among different methods. Future studies should be carried out to focus on cancer progression and patient's survival assessments. inclusion and exclusion criteria. The studies were screened by two independent authors and all the studies meeting the inclusion criteria were included. Any discrepancy between the two reviewer authors were sorted out by a third expert. Inclusion criteria were the English language writing, publication up to the date of the search, the study setting of cross-sectional or cohort, and the data eligibility for the study. Furthermore, the meta-analysis, conference seminars, and review articles were excluded from the search results.
Data extraction. Selected studies were listed in EndNote software (EndNote X7, Thomson Reuters) and were reviewed by two authors of the study independently; disagreements between them were settled by a third expert. All the relevant studies were screened considering the inclusion criteria and the data were extracted. The extracted data included the first author's name, the publication date (based on year), country, design of the study, type of the cancer, sample size, mutation pathway, gene name, mean age, gender, mutation positive population, and method of detection.