Differences of protein expression profiles, KRAS and BRAF mutation, and prognosis in right-sided colon, left-sided colon and rectal cancer

To compare protein expression levels, gene mutation and survival among Right-Sided Colon Cancer (RSCC), Left-Sided Colon Cancer (LSCC) and rectal cancer patients, 57 cases of RSCC, 87 LSCC and 145 rectal cancer patients were included retrospectively. Our results demonstrated significant differences existed among RSCC, LSCC and rectal cancer regarding tumor diameter, differentiation, invasion depth and TNM stage. No significant difference was identified in expression levels of MLH1, MSH2, MSH6, PMS2, β-Tubulin III, P53, Ki67 and TOPIIα, and gene mutation of KRAS and BRAF among three groups. Progression Free Survival (PFS) of RSCC was significantly lower than that of LRCC and rectal cancer. In univariate analyses, RSCC, preoperative chemoradiotherapy, poor differentiation, advanced TNM stage, elevated serum CEA and CA19-9 level, tumor deposit, perineural and vascular invasion were found to be predictive factors of shorter PFS. In multivariate analyses, only differentiation and TNM stages were found to be independent predictors of PFS. In conclusion, compared with LSCC and rectal cancer, RSCC has larger tumor size, poor differentiation, advanced TNM stage and shorter survival. The shorter survival in RSCC might be attributed to the advanced tumor stage caused by its inherent position feature of proximal colon rather than genetic difference.

had similar patterns of genomic alteration, and gene mutations of APC, TP53, SMAD4, PIK3CA and KRAS 10 . So, whether colon and rectal cancers have different gene expression and prognosis is still in debate.
The distinction between RSCC and LSCC has received increasing attention in recent years. Some suggested that they were two distinct categories of colon cancer 11 . Many publications reported that there were significant differences regarding epidemiology, clinical presentation, pathology, genetic mutations and survival between RSCC and LSCC 11 . RSCC had been reported to be older, more often female and more often poorly differentiated tumors, and have more advanced stages, increased tumor sizes and different molecular features [11][12][13][14] . Data regarding prognosis in RSCC versus LSCC are conflicting, and it remains a matter of great debate whether tumor location itself has a significant prognostic impact 12 . Most studies demonstrated a poorer survival in RSCC compared to LSCC [15][16][17] . In contrast, several scholars found no difference in OS between RSCC and LSCC after adjusting for various variables 11,18,19 . Warschkow et al. 12 carried out a study including 91,416 patients, and found that RSCC patients had worse OS compared to LSCC patients; but the prognosis of RSCC was better than LSCC after matching clinical features. In addition, whether molecular features differ between LSCC and RSCC remains unclear 20 . Kuramochi et al. 21 had detected mRNA expression levels of 14 signal transduction genes in 52 cases of CRC, but only identified significant differences in PTEN mRNA expression level.
Furthermore, some authors suggested that LSCC and rectal cancer shared multiple common characteristics and were different from RSCC, which was supported by several histological, genetic and methylation findings 3,22,23 . And a new term, Left-sided Colorectal Cancer (LCRC) which included LSCC and rectal cancer, was created 24 . There are three main types of (epi)genetic instability in CRC: (1) chromosomal instability (CIN) caused by KRAS mutations; (2) microsatellite instability (MSI) resulted from deficient DNA mismatch repair (MMR); (3) CpG island methylator phenotype (CIMP) epigenetic instability 3 . The mutational profiles (KRAS, MMR, CIMP) of LSCC and rectal cancer were similar, but were different from that of RSCC. This result was attributed to their differing origins: RSCC originated from midgut, while LCRC originated from hindgut 3 . BRAF was preferentially mutated in RSCC, and EGFR (epidermal growth factor receptor) was prevalently amplified in LCRC 3 . Class III beta-tubulin (β-Tubulin III) had been reported to express at the invasive margin of CRC, and its expression level was correlated with tumor differentiation and lymphatic metastasis 25 . The mutation incidence of p53 gene was reported to be as high as 42.4% in CRC 26 . P53 plays an important role in the transformation from colorectal normal mucosa to carcinoma through adenoma 27 . Several studies reported that gene mutation and protein expression of P53 differed significantly between RSCC and LSCRC 26,28,29 , but others showed that no significant association was identified between p53 protein expression and tumor site 30 . Ki67 is a marker of cell proliferation, and it plays a vital role in the development of CRC 31 . High Ki67 labeling index had been reported to be an independent prognostic biomarker in TNM III and IV CRC 32 . In addition, our previous study showed that topoisomerase II alpha (TOPIIα) expression was related with T stage, N stage, recurrence and prognosis 33 .
In the light of the aforementioned considerations, it is urgent for us to explore the possible differences of gene expression level and prognosis among RSCC, LSCC and rectal cancers. Knowledge of the molecular differences would help us to improve the diagnosis and treatment strategy in clinical practice. Thus, we investigated the protein expression levels of MLH1, MSH2, MSH6, PMS2, β-Tubulin III, P53, Ki67 and TOPIIα, and the gene mutation of KRAS and BRAF in 289 specimens of sporadic CRC patients, and their implications on survival were also investigated.

Methods
Patients. From January 2015 to December 2016, 289 cases of sporadic CRC patients were recruited from the Department of Colorectal Surgery of Changhai Hospital, Second Military Medical University, Shanghai, China. All patients received radical resection of the primary tumor. The clinicopathological characteristics were extracted from the electronic medical records. All patients were followed up every 3 months, with a median follow up period of 10 months, ranging from 3 to 23 months. Informed consent had been obtained from all patients and the project had been approved by the Ethics Committee of Changhai Hospital, Second Military Medical University. All methods were performed in accordance with the relevant guidelines and regulations.
Immunohistochemistry. Paraffin-embedded tumor tissues were examined for MLH1, MSH2, MSH6, PMS2, β-tubulin III, P53, Ki67 and TopIIα expression, using the Envision method following the manufacturer's instruction. After deparaffinization and re-hydration, antigen retrieval was done with Citrate buffer (0.01 mol/L, pH = 6.0) by pressure cooker. Primary antibodies (Table 1) were incubated on the slides for 2 hours at room temperature in a hydrated chamber. The sections were stained with DAB and counterstained with Mayer's hematoxylin, washed again, dehydrated in alcohol, cleared in xylene, mounted with Pertex mounting medium, and coverslipped. All sections were scored blindly by two pathologists (Xu Y & Bai CG) under microscope, by randomly selecting 10 high-power (×400) view fields in each sample and scoring the protein expression in tumor cells.
Expression defects of MLH1, MSH2, MSH6 and PMS2 proteins were defined as complete absence of detectable nuclear staining in tumor cells. Intact nuclear staining of the colorectal crypts of the peritumoral normal mucosa, stromal cells and lymphocytes served as internal positive control and was required for adequate evaluation 34 .
SCientiFiC RepoRts | 7: 7882 | DOI:10.1038/s41598-017-08413-z Sample preparation, DNA extraction, amplification of KRAS and BRAF genes. Surgically resected primary CRC tumor tissue specimens were fixed in formalin and preserved in paraffin blocks for histological examination. After evaluating the standard hematoxylin/eosin-stained slides from each specimen, appropriate samples were specifically chosen by a pathologist to include predominantly tumor cells without significant necrosis or inflammation 36 . Eight 10 μm-thick formalin-fixed paraffin-embedded (FFPE) sections were used for this study, and placed in 2-mL sterile Eppendorf tubes. AmoyDx FFPE DNA Kit (AmoyDx, Xiamen, China) was used for DNA extraction. Human genomic DNA was amplified for KRAS in exons 2, 3 and 4 (codons 12, 13, 59, 61, 117 and 146), and BRAF in exon 15 (codon 600) by using AmoyDx gene mutation real-time PCR kits (AmoyDx, Xiamen, China). All the experiments are performed following manufacturer's instructions. 5 μL DNA was used for PCR amplification in each reaction, the PCR cycling conditions were: 5 min incubation at 95 °C, followed by 15 cycles of 95 °C for 25 sec, 64 °C for 20 sec, 72 °C for 20 sec and then 31 cycles of 93 °C for 25 sec, 60 °C for 35 sec, 72 °C for 20 sec. Fluorescent signal was collected from FAM and HEX channels. Each PCR run must contain one negative control and one positive control. KRAS and BRAF mutation status were determined according to the Ct value as indicated in the manufacturer's instructions.

Statistical analysis. Associations between tumor position and categorical clinicopathological variables were
analyzed by χ 2 test, using Fisher Exact test if one or more expects in the cross table is less than 5. Numerical data which were consistent with normal distribution were presented as the mean ± standard deviation, and comparisons were performed with the student's t-test or one-way ANOVA analysis. Numerical data which were inconsistent with normal distribution were presented as median (minimum-maximum), and comparisons were performed with nonparametric test. The Progression Free Survival (PFS) was estimated by the Kaplan-Meier method and survival differences were analyzed with the log-rank test. The Cox proportional hazards model was used for multivariate analysis of prognostic factors. P < 0.05 was considered statistically significant (two-sided). All statistical analyses were conducted using SPSS 17.0 statistical software (SPSS, Inc., Chicago, IL, USA).

Results
A total of 289 cases of CRC patients were included, consisting of 57 cases of RSCC, 87 LSCC and 145 rectal cancers ( Table 2). Of the 289 CRC patients, 186 were males and 103 were females, with a mean age of 59.6 years. Thirty-three (33/289, 11.4%) patients received preoperative chemoradiotherapy, 156 (156/289, 54.0%) received postoperative chemoradiotherapy; 82 were resected laparoscopically, and the remaining 207 were removed by open surgery. All of the primary tumors were resected radically, and combined resections of metastatic lesions were performed in 29 patients. There were 60, 82, 89 and 58 cases of stage I, II, III and IV according to the UICC-AJCC TNM stage classification system (7th edition). Table 2, significant differences were observed among the RSCC, LSCC and rectal cancer patients regarding surgical procedure (Open/Laparoscopic), tumor diameter, differentiation, invasion depth (T) and TNM stage (all P < 0.05). No significant difference was observed in other characteristics, including gender, age, preoperative chemoradiotherapy, gross type, lymph node metastasis(N), distant metastasis(M), serum CEA level, serum CA 19-9 level, tumor deposit, perineural invasion and vascular invasion ( Table 2, all P > 0.05). Of the 58 cases of TNM stage IV patients, 38 had received postoperative chemoradiotherapy, 12 had received postoperative biological therapy, and 8 had received second line chemoradiotherapy after recurrence. No significant difference was identified in postoperative chemoradiotherapy, biological therapy and second line chemoradiotherapy among the three groups (Table 3, all P > 0.05).

Influencing factors of Progression Free Survival (PFS).
Our results showed that tumor location was related with PFS ( Fig. 4, P = 0.002). The PFS of RSCC was significantly shorter than that of LRCC (P = 0.003) and rectal cancer (P = 0.004). No significant difference was identified in PFS between LSCC and rectal cancer patients (P = 0.392).

Discussion
The difference among RSCC, LSCC and rectal cancer patients has been remaining a serious debate for a long time.
Our comparative study showed that these three groups had similar baseline in most clinicopathological characteristics. But there were still significant differences in surgical procedure (Open/Laparoscopic), tumor diameter, differentiation, invasion depth (T) and TNM stage among the three groups. Compared with LCRC, RSCC had less laparoscopic resection, larger tumor size and poor differentiation. As far as T stage and TNM stage were concerned, RSCC was also found to have similar tumor stage compared with LSCC in our study, which might have something to do with the small sample size. Our results were consistent with that in the published literature, which reported that RSCC had more advanced tumor stages, increased tumor sizes and poorly differentiation 3, 11-14 . Due to the larger bowel lumen, RSCC usually becomes symptomatic later than LSCC, which in turn leads to later diagnosis, larger tumor size and advanced tumor stage 37,38 . Secondly, RSCC is located far away from the anal verge, so it is more difficult to be discovered by digit rectal examination and sigmoidoscopy. Hugen et al. 39 reported that the frequency of mucinous and signet-ring cell tumors was higher in RSCC (45%) than in that in LCRC (20%). It was consistent with our results, but the underlying cause for poorer differentiation in RSCC is still unknown. Some oncologists hypothesized that it could be attributed to different underlying genetic and biological features 20,40 . The reported differences in age, gender, perineural invasion and vascular invasion by other authors 37 was not found in our study, which may be attributed to the relatively small sample size. In our study, RSCC was less likely to undergo laparoscopic resection, which may have something to do with the relatively higher difficulty of laparoscopic right hemicolectomy.
It had been hypothesized that there were significant differences in molecular features between RSCC and LSCC, which might serve as the cause of clinicopathological differences 40 . RSCC was reported to have a higher      frequency of KRAS mutation than LSCC (57.3% vs 40.4%; P < 0.0001) 41 , and a higher incidence of BRAF mutation with 18.4-22.4% in RSCC and 1.3-7.8% in LCRC 42 . But RSCC had also been reported to be associated with more mutant KRAS and more wild-type BRAF tumors 19 . But no significant difference was found in KRAS and BRAF mutation in our study. Similarly, except for the TOPIIα immunostaining index, no significant difference was found in protein expression levels of MLH1, MSH2, MSH6, PMS2, β-tubulin III, P53, Ki67 and TopIIα between RSCC and LSCC. Our results were consistent with that of Zhu et al. 20 and Cancer Genome Atlas Network 43 . Zhu et al. compared gene expression profiling of RSCC and LSCC using the Human Genome Array gene chip in 100 cases of patients, but only 11 genes were identified to be differentially expressed between RSCC and LSCC 20 . Cancer Genome Atlas Network conducted a genome-scale analysis of 276 samples, analyzing exome sequence, DNA copy number, promoter methylation, mRNA and microRNA expression, and concluded colon and rectal cancers had similar patterns of genomic alteration 43 . No significant relationship was found between p53 protein expression and tumor position in our study, which was consistent with the results of Ghavam-Nasiri 30 .
The reported prognostic role of P53, Ki67 and TopIIα expression in the literature is conflicting [44][45][46] . It confirmed our results, which showed none of β-tubulin III, P53, Ki67 and TopIIα expression level was not prognostic factor in CRC. Our study demonstrated that RSCC patients had shorter PFS compared with LCRC patient. Our result was consistent with that in most literature. Petrelli et al. 47 conducted a systematic review including 66 studies and 1,437,846 patients, and found that RSCC had shorter OS than LSCC. And Lee's systematic review showed that RSCC had shorter OS than LCRC 24 . He also suggested that CRC should be classified into two types: the RSCC and the LCRC 24 .
Since the follow-up is relatively short and few patients died in this period, so it is unreasonable for us to compare OS. In addition, since combined resections of metastatic lesions were performed in only 29 of the 60 stage IV patients with distant metastasis, so the Disease Free Survival (DFS) can't be applied in the other 31 patients with residual tumor. For these reasons, we chose PFS to measure survival in our study. We proved that RSCC patients had longer PFS than LCRC patients, but no difference between LSCC and rectal cancer. In univariate analysis, tumor position (RSCC vs. LSCC&Rectal), preoperative chemoradiotherapy, poor differentiation, TNM stage, serum CEA and CA19-9 level, tumor deposit, perineural and vascular invasion were all found to be predictor of PFS. Petrelli et al. summarized that TNM stage, differentiation and vascular invasion were well-recognized risk factors for CRC 47 . Mutation of KRAS and BRAF had also been reported to predict poor survival in CRC patients 13, 48, 49 , but no similar finding was identified in our study which may be related with relatively small sample size and short follow up period. NRAS mutations (codons 12, 13 and 61) were reported to occur in 3-5% of colorectal cancer 48,49 . NRAS mutation was a predictive factor of response to anti-EGFR biological therapy and OS 49 . In our study, only 9 of the included 289 colorectal cancer patients had undergone the NRAS tests. It is difficult to reach statistical significance with such a small sample size, so NRAS gene mutation status was not analyzed in this study. A meta-analysis of FIRE-3, SWOG 80405 and PEAK trials indicated that RAS wild LSCC patients benefited more from anti-EGFR treatment (P < 0.001), while RSCC patients benefit more from anti-VEGF treatment (P > 0.05) 50 . Since no significant difference was identified in postoperative biological therapy in our study, it might not be a major contributing factor to survivals of our patients.
In the following multivariate analysis, only differentiation and TNM stage were found to be independent predictor of PFS. It indicated that tumor position might not be an independent influencing factor of PFS. Many authors also found that no difference existed in survival between RSCC and LSCC after adjusting for various variables 11,18,19 . In terms of these phenomena, we suggest the survival difference between RSCC and LCRC would be caused by differences in tumor stage. The RSCC's larger lumen and longer distance from anal verge makes it less symptomatic and difficult to be discovered, which then leads to advanced tumor stage and poor prognosis 38 .
In conclusion, RSCC patients have larger tumor size, poor differentiation, advanced TNM stage and poor survival, compared with LCRC patients. Except for TOPIIα immunostaining index, no significant difference was identified in the expression levels of MLH1, MSH2, MSH6, PMS2, β-tubulin III, P53, Ki67 and TopIIα, and gene mutation of KRAS and BRAF. The shorter survival in RSCC might be attributed to the advanced tumor stage caused by its inherent position feature of proximal colon.