KRAS/NRAS Mutations Associated with Distant Metastasis and BRAF/PIK3CA Mutations Associated with Poor Tumor Differentiation in Colorectal Cancer

Introduction

Colorectal cancer (CRC) is a common malignancy of the gastrointestinal tract. According to the latest estimates of global cancer incidence and mortality, CRC (10.0%) ranks third in incidence after breast cancer and lung cancer (11.7% and 11.4%, respectively), and ranks second in mortality (9.4%) after lung cancer (18%).¹ In China, from 2015 to 2020, the incidence of CRC increased rapidly, while gastrointestinal cancer (stomach, colorectal, liver and esophageal cancer) had the highest mortality rate, and CRC ranked fifth in mortality after lung, liver, stomach and esophageal cancer.² With the clinical application of anti-epidermal growth factor receptor (EGFR) drug cetuximab and anti-angiogenic drug bevacizumab, targeted therapy has become the first-line treatment for CRC.³

The occurrence, progression, and prognosis of CRC are regulated by some molecular signal transduction pathways, such as EGFR-mediated signaling pathways. As a transmembrane tyrosine kinase receptor, EGFR can activate Ras/RAF/MAPK and PI3K/AKT/mTOR pathways after binding to corresponding ligands, thereby inducing tumor cell proliferation, invasion, metastasis and angiogenesis.⁴ Target drug tyrosine kinase inhibitors (TKI) targeting the EGFR have a significant therapeutic effect in cancer patients with EGFR gene mutations.⁵ In addition, mutations in some genes downstream of EGFR signaling pathway, such as KRAS proto-oncogene (KRAS), NRAS proto-oncogene (NRAS), B-Raf proto-oncogene (BRAF) and phosphatidylinositol-4,5-bisphosphate 3-kinase catalytic subunit alpha (PIK3CA), will affect the targeted therapeutic effect of anti-EGFR monoclonal antibody drugs.⁶ Mutations in these genes result in the EGFR signaling pathway always in an active status, making EGFR-targeted drug therapy ineffective.^7,8 In recent years, KRAS/NRAS/BRAF/PIK3CA wild-type CRC patients have significantly benefited from anti-EGFR monoclonal antibodies.

The different populations, lifestyles, and the interaction between some environmental factors and genetic polymorphisms will affect the clinicopathological characteristics of CRC patients.⁹ In addition, the clinical characteristics of CRC are also related to age, and although study has found that older pT4 patients were more likely to develop serious postoperative complications, the prognosis of older patients may vary depending on the stage, tumor site, and pre-existing comorbidities.¹⁰ Study has shown that KRAS, NRAS or BRAF mutations were associated with CRC metastasis.^11,12 However, study has shown that distant metastasis was not related to KRAS, NRAS, BRAF, or PIK3CA mutations.¹³ Study has shown that BRAF mutations were associated with poorer differentiation, but mutant KRAS was associated with greater differentiation,¹⁴ and PIK3CA was not associated with differentiation.^15–18 Therefore, in different sample sizes, different populations and different studies, the results of the correlation between the studied gene status and clinicopathological features are inconsistent.

The association of the studied gene status with clinicopathological features will provide valuable information for clinicians to better assess the disease severity and disease progression, such as colorectal cancer brain metastases. The development of brain metastases is an important factor in overall cancer mortality in patients with advanced cancer, and the prognosis is poor.¹⁹ Clinicians can monitor the progress of such patients with a combination of screening techniques to achieve early detection and treatment. Electroencephalogram (EEG) biomarkers differ significantly in patients with consciousness after severe acquired brain injury, which can be used as diagnosis and prediction for clinical evaluation of patients.²⁰ EEG alpha band indices can be used by clinicians to diagnose, select and evaluate treatment for main neuropsychiatric disorders of the brain.²¹

The Hakka is a Han ethnic group with a unique genetic background formed by the Hakka ancestors from the Han nationality in central China, who migrated southward for many times and fused with the ancient Yue residents in Guangdong, Fujian, and Jiangxi.²² At present, there is no study on the relationship between the mutation status of KRAS, NRAS, BRAF, PIK3CA genes in EGFR signaling pathway and the clinicopathological features of CRC in Hakka population. Therefore, this study retrospectively analyzed mutations of these genes in CRC tissues, and explored their relationship with clinicopathological features, aiming to provide valuable data for precise treatment of CRC individuals.

Materials and Methods

Participants

After approval from the Ethics Committee of the Meizhou People’s Hospital (Clearance No.: 2020-A-57), 408 CRC patients who received KRAS/NRAS/BRAF/PIK3CA molecular testing were retrospectively investigated between January 2019 and October 2020. The inclusion criteria for the study subjects were: (1) colorectal cancer was confirmed by hematoxylin and eosin (HE) staining and histological analysis; (2) at the time of enrolment, patients presenting in stage I to IV (according to the American Joint Committee on Cancer (AJCC) guidelines²³) were enrolled; (3) there was complete medical records; (4) aged more than 18 years. The exclusion criteria for the study subjects as follows: (1) patients with other tumors disease other than colorectal cancer; (2) other circumstances inconsistent with the inclusion criteria mentioned above.

Detection of Mismatch Repair (MMR) Protein by Immunohistochemistry

All specimens were immobilized with 4% neutral methanol solution and embedded with paraffin. Max Vision two-step method was used for immunohistochemistry. Paraffin sections were dewaxed and hydrated, then washed with PBS. Peroxidase blocker was used to block endogenous peroxidase activity. Primary antibody was added and incubated at room temperature, biotin-labeled secondary antibody was added and streptomycin antibiotin peroxidase solution was added. After incubation, Hematoxylin-3,3’-Diaminobenzidine (DAB) was added for coloration, restained with hematoxylin, dehydration, transparent, and tablets were sealed. The expression of mismatch repair proteins (MutS homolog 2 (MSH2), MutS homolog 6 (MSH6), MutL homolog 1 (MLH1), and PMS1 homolog 2 (PMS2)) was observed under the microscope. When samples from patients were positive for all four of these mismatch repair proteins (>30%) were identified as mismatch repair-proficient (pMMR), and mismatch repair-deficient (dMMR) was identified when one or more proteins were missing.

Genotyping for KRAS, NRAS, BRAF and PIK3CA Genes

Formalin-fixed and paraffin-embedded (FFPE) tissue was cut into 5μm thick slices, and 10 slices were loaded into EP tubes for DNA extraction. DNA was extracted using the AmoyDx^® Tissue DNA Kit (Spin Column) (Amoy Diagnostics, Xiamen, China) according to the manufacturers’ instructions. NanoDrop 2000 (Thermo Scientific) was used to measure DNA concentration. The DNA samples should be diluted to 2 ng/μL, and the OD260/OD280 should be 1.8–2.0.

Genetic tests were performed using amplification refractory mutation system polymerase-chain reaction (ARMS-PCR) with corresponding mutation detection kits, respectively (Amoy Diagnostics Co. Ltd, Xiamen, China). The DNA of the sample to be tested, positive quality control DNA, and negative quality control DNA were added to the PCR reaction solution, respectively. PCR amplification was performed according to the following conditions: initial denaturation at 95°C for 5 minutes; 95°C 25 seconds, 64°C 20 seconds, 72°C 20 seconds run 15 cycles; 95°C 25 seconds, 60°C 35 seconds, 72°C 20 seconds run 31 cycles; Amplification was performed in a LightCycler 480 real-time PCR system (Roche Diagnostics, Germany). According to the amplification curve of the experimental results, the mutation cycle threshold (Ct) values of each reaction tube and external control of the sample were determined, and the sample to be tested was determined as mutation positive or negative according to the obtained Ct values. The genetic sites tested mainly included common mutations in exons 2, 3 and 4 of KRAS (codons 12, 13, 61, 117, and 146), exons 2, 3 and 4 of NRAS (codons 12, 61, 117, and 146), exons 9 and 20 of PIK3CA (codons 542, 545 and 1047), and exon 15 of BRAF (codons 600).

Data Collection and Statistical Analysis

Clinical records, including age, sex, tumor site, maximum diameter of tumor, histological type, tumor differentiation, lymphovascular invasion, perineural invasion, and TNM stage were collected. SPSS statistical software version 21.0 (IBM Inc., State of New York, USA) was used for data analysis. The patients’ clinicopathological features were summarized with descriptive statistics. Categorical variables were compared using χ² test and Fisher’s exact test. Gender, age, tumor site, maximum diameter of tumor, tumor differentiation, lymphovascular invasion, perineural invasion, and distant metastasis were selected as covariates in the multivariate logistic regression analysis for KRAS, NRAS, BRAF, PIK3CA mutations, based on estimating the odds ratios (OR) and their 95% confidence intervals (CIs). The significance test was two-sided, and a P value <0.05 was considered statistically significant.

Results

Characteristics of Subjects

Four hundred and eight patients were included in the analysis, including 256 (62.7%) men and 152 (37.3%) women. The age of all patients was ranging from 25 to 86 years old. There were 185 (45.3%) patients with <60 years old, and 223 (54.7%) patients with ≥60 years old. The number of colon, rectum and cecum tumors was 227 (55.6%), 175 (42.9%) and 6 (1.5%), respectively. There were 255 patients (62.5%) with tumor maximum diameter less than 5 cm, and 138 patients (33.8%) with tumor maximum diameter ≥5 cm. Well differentiation, moderate differentiation and poor differentiation tumor detected in 4 (1.0%), 370 (90.7%) and 34 (8.3%) cases, respectively. There were 80 (19.6%) and 54 (13.2%) patients with lymphatic vascular space invasion and perineural invasion, respectively. And 317 (77.7%) cases, and 91 (22.3%) cases were classified as stage I-III, and stage IV, respectively. There were 15 (3.7%) cases with dMMR (Table 1).

Table 1 Baseline Characteristics of CRC Patients

Frequency and Composition Ratio of KRAS/NRAS/BRAF/PIK3CA Mutations

All patients were genotyped for KRAS (codons 12, 13, 61, 117, and 146), NRAS (codons 12, 61, 117, and 146), BRAF (codons 600), and PIK3CA (codons 542, 545 and 1047) mutations. At least one mutation in these genes was detected in 216 (52.9%) patients, while none was detected in 192 (47.1%) patients. Mutation in KRAS, NRAS, BRAF, and PIK3CA was detected in 190 (46.6%), 11 (2.7%), 10 (2.5%), 34 (8.3%) samples, respectively. Among them, mutations in KRAS and NRAS, KRAS and BRAF, KRAS and PIK3CA were detected simultaneously in 1 (0.2%), 1 (0.2%), and 27 (6.6%) samples, respectively.

Ranking the mutations in order of composition, mutations in KRAS exon 2 had the highest proportion (62.5%), followed by mutations in both KRAS exon 2 and PIK3CA exon 9 simultaneously (7.9%), KRAS exon 4 (6.9%), KRAS exon 3 (5.1%), BRAF exon 15 (4.2%), both KRAS exon 2 and PIK3CA exon 20 simultaneously (3.7%), NRAS exon 2 (3.2%), and PIK3CA exon 9 (2.8%) (Table 2).

Table 2 Frequency and Composition Ratio of KRAS/NRAS/BRAF/PIK3CA Mutation in CRC Patients

Relationship of Clinicopathological Characteristics and KRAS/NRAS/BRAF/PIK3CA Gene Mutation Status

Fisher’s exact test was used to compare the clinicopathological differences between the KRAS/NRAS/BRAF/PIK3CA mutant and wild-type patients. KRAS, NRAS, BRAF, and PIK3CA mutations were not related to gender, age, tumor site, tumor size (maximum diameter of tumor), perineural invasion, lymph node metastasis, and MMR status (all P>0.05). The percentage of mutations in KRAS gene in stage IV patients was significantly higher than that in stage I-III patients (56.0% (51/91) vs 43.8% (139/317), P=0.043), and that in patients with distant metastasis was higher than that in patients without distant metastasis (57.1% (52/91) vs 43.5% (138/317), P=0.024). The mutation percentage of BRAF gene was higher in the tumor with poor differentiation than that in the tumor with moderate and well differentiation (13.3% (4/30) vs 1.6% (6/364) and 0 (0/4), P=0.012), and it was also significantly higher in the patients with lymphovascular invasion than without it (6.3% (5/80) vs 1.5% (5/328), P=0.029) (Table 3).

Table 3 Clinicopathological Characteristics According to KRAS/NRAS/BRAF/PIK3CA Gene Mutation Status in CRC Patients

The regression analysis was performed with KRAS mutations, NRAS mutations, BRAF mutations and PIK3CA mutations as the response variables, and gender (X1), age (X2), tumor site (X3), maximum diameter of tumor (X4), tumor differentiation (X5), lymphovascular invasion (X6), perineural invasion (X7), and distant metastasis (X8) as independent variables. And multicollinearity and model goodness-of-fit was checked. The multivariate logistic regression analysis demonstrated that KRAS mutations (adjusted OR 1.675, 95% CI 1.017–2.760, P=0.043) and NRAS mutations (adjusted OR 5.183, 95% CI 1.239–21.687, P=0.024) appeared more frequently in patients with distant metastasis. BRAF mutations (adjusted OR 7.224, 95% CI 1.356–38.488, P=0.021) and PIK3CA mutations (adjusted OR 3.811, 95% CI 1.268–11.455, P=0.017) appeared more frequently in tumor with poor differentiation. Moreover, a relationship still existed between PIK3CA mutations and female patients (male/female adjusted OR 0.443, 95% CI 0.210–0.934, P=0.032) (Table 4).

Table 4 Multivariate Logistic Regression in CRC Patients Between Gene Mutations and Clinicopathological Characteristics

Discussion

CRC is a common malignant tumor in human digestive tracts.²⁴ There is evidence that the occurrence of tumors is related to inflammation, and blood inflammatory markers can be used as markers for risk prediction and prognosis assessment of CRC. In addition, the significance of microRNAs (miRNAs) in tumor diagnosis, prognosis and prediction is constantly being explored.²⁵ In the past decades, a large number of research data have emerged in the studies on the molecular basis,^26,27 drug investigation and usage,^28,29 and gene testing³⁰ of colorectal cancer, leading to the rapid development of gene testing and targeted therapy for CRC. The targeted therapy of CRC patients needs to be based on the genes’ status, and molecular testing has been paid more and more attention.³⁰ Studies have shown that genes downstream of the EGFR signaling pathway can affect the efficacy of targeted therapy with monoclonal antibodies.³¹

KRAS is one of the most important oncogenic genes in human genome, KRAS gene activating mutations are found in >80% of pancreatic cancer, >30% of CRC, cholangiocarcinoma and lung adenocarcinoma.³² The frequency of mutations may vary from region to region and population to population. Approximately 41.5% of Danish CRC patients have KRAS mutations,³³ 45.5% in Slovene patients,³⁴ 49.6% in Saudi Arabian patients,³⁵ 34.7% in sigmoid colon and 58.2% in cecum of patients in the United States,³⁶ 29% in Iranian patients,³⁷ 54.84% in Turks patients,³⁸ 42% in Japanese patients,³⁹ 41.0% in Vietnamese patients,⁴⁰ and 23% in Indian patients.⁴¹ The frequency of KRAS mutations was 35.0%–50.0% in Chinese CRC patients.^42–46 The frequency of NRAS mutations was 4.2% in Danish CRC patients,³³ 7.3% in Tunisian patients,⁴⁷ 2.0% in Saudi Arabian patients,³⁵ 5% in Japanese patients,³⁹ 9.6% in Vietnamese patients,⁴⁰ 2.0% in Indian patients,⁴¹ and 7% in Mexican patients.⁴⁸ The mutation percentage of NRAS was 1.2%–3.85% in Chinese CRC patients.^45,46,49 In this study, at least one mutation in KRAS, NRAS, BRAF and PIK3CA was detected in 219 (52.6%) CRC patients, while none was detected in 197 (47.4%) CRC patients. KRAS and NRAS mutation was detected in 190 (46.6%) and 11 (2.7%) samples, respectively. The mutation percentage of KRAS and NRAS gene of CRC patients in this study is basically the same as that in other populations.

The frequency of BRAF was 18.0% in Danish CRC patients,³³ 0.4% in Saudi Arabian patients,³⁵ 7% in Iranian patients,³⁷ 12.9% in Turks patients,³⁸ 7% in Japanese patients,³⁹ 8.3% in Vietnamese patients,⁴⁰ 17% in Indian patients,⁴¹ and 12.44% in Greek and Romanian patients.⁵⁰ The frequency of BRAF mutations was 2.3%–8% in Chinese CRC patients.^{45,49,51–53} The frequency of PIK3CA mutations was 18.8% in Danish CRC patients,³³ 13.3% in Arab patients,⁵⁴ and 8% in Italian patients.⁵⁵ The frequency of PIK3CA mutations was about 9.4%–18.9% in Chinese CRC patients.^42,51,56,57 In this study, BRAF and PIK3CA mutation status were detected in 10 (2.5%) and 34 (8.3%) samples, respectively. Compared with other populations, the mutation percentage of BRAF and PIK3CA genes of CRC patients in this study was relatively low.

In this study, KRAS or NRAS mutations appeared more frequently in CRC patients with distant metastasis. Studies have shown that KRAS, NRAS or BRAF mutations were associated with lung metastasis in CRC.^11,12 KRAS/NRAS/BRAF mutations may predict late distant metastasis.⁵⁸ KRAS mutations were related to poor tumour differentiation and liver metastasis.⁵⁹ Distant metastatic tumors had a higher mutation percentage of NRAS mutation but not KRAS.¹⁵ BRAF mutations were more common in peritoneal metastasis patients.¹⁶ However, another study has shown that distant metastasis was not related to KRAS, NRAS, BRAF, or PIK3CA mutations.¹³ In addition, BRAF and PIK3CA mutations appeared more frequently in tumor with poor differentiation in this study. Study has shown that BRAF mutations were associated with poorer differentiation, but mutant KRAS was associated with greater differentiation.¹⁴ BRAF mutations were more common in poorly differentiated tumors, but PIK3CA is not.^15–18 PIK3CA mutations showed null associations with tumor differentiation.⁶⁰ In summary, the correlation of KRAS, NRAS, BRAF, PIK3CA mutations with distant metastasis, tumor differentiation was inconsistent. Different sample sizes, different populations, and different detection methods for genetic mutations can partly explain the results. In addition, the prognosis of CRC is also associated with DNA mismatch repair, and CRC patients with microsatellite instability-low (MSI-L) may experience shorter survival; however, there are challenges in the availability of data for MSI testing.⁶¹

In terms of molecular mechanisms, mutation-activated KRAS in tumor cells reprograms macrophages with tumor-associated macrophage (TAM)-like phenotypes, which not only promotes tumor progression but also induces resistance of tumor cells to targeted therapy.⁶² MiR-450b-5p can activate Wnt/β-catenin signaling to promote cell proliferation, tumor growth, and inhibit the apoptosis of CRC cells, while miR-450b-5p can be up-regulated by KRAS/AP-1 signaling.⁶³ The differential expression of some RNAs in exosomes of CRC cells with BRAF V600E mutation is closely related to proliferation, metabolism of tumor cells and tumor microenvironment changes.⁶⁴ B-Raf/MEK/ERK pathway was related to inhibition of tumor cell differentiation.⁶⁵ The possible mechanisms of the relationship between distant metastasis and KRAS and NRAS mutations, as well as the relationship between poor tumor differentiation and BRAF and PIK3CA mutations, are still not completely clear and need further study.

The results of this study suggest that for CRC patients with KRAS and NRAS mutations, more attention should be paid to the possibility of distant metastases, such as liver metastasis, lung metastasis, and brain metastasis. Some techniques can be applied to the diagnosis of liver metastases in CRC.⁶⁶ Tumor brain metastases are the most common tumors of the adult central nervous system,⁶⁷ and the rate of brain metastases in CRC is the highest among gastrointestinal tumors. Brain metastases usually occur in the advanced stages of the disease and the prognosis is poor in most patients. The quality of life and prognosis of patients with colorectal cancer with neurological symptoms can be improved by early detection and treatment of metastatic lesions. CRC patients with KRAS and NRAS mutations with neurological symptoms, through the application of some new technology such as non-invasive brain stimulation techniques (NIBS),^68–71 can be used to identify the risk factors for brain metastases, so that early detection and treatment.

The study has some limitations that are worth noting. First of all, the number of research objects in this study is relatively small, which leads to some deviations in the results. Second, we only studied the common mutation sites of KRAS/NRAS/BRAF/PIK3CA genes; the status of the other mutation sites in these genes is unknown. Third, this study was limited to the correlation between gene mutations and clinicopathological features, and did not analyze the correlation between gene mutations and clinical outcomes and treatment responses in CRC patients receiving chemotherapy or radiotherapy. Therefore, the next step is to conduct a multi-center study with a larger sample size and a comprehensive analysis of KRAS/NRAS/BRAF/PIK3CA gene.

Conclusions

The relationship between clinicopathological features and KRAS, NRAS, BRAF, PIK3CA genes status in CRC patients were studied, and found that there was a significant relationship between distant metastasis and KRAS or NRAS mutations, while poor tumor differentiation and BRAF or PIK3CA mutations. Importantly, more attention should be paid to the possibility of distant metastasis in colorectal cancer patients with KRAS and NRAS mutations. Monitoring these patients for early diagnosis and treatment of distant metastases will have the potential to improve their survival and prognosis.