Clinicopathological Significance of BRAFV600E Mutation in Colorectal Cancer: An Updated Meta-Analysis.

Background and Aims: Numerous studies have identified BRAFV600E mutation as a predictive factor of anti-EGFR antibodies in colorectal cancer (CRC). However, the association between BRAFV600E mutation and clinicopathological features remains unclear. Therefore, we aimed to conduct an updated and comprehensive meta-analysis to evaluate the above issues.
Methods: We performed a systematic literature search from PubMed, Web of Science, Embase, and PMC database examining the association between BRAFV600E mutation and clinicopathological features in CRC patients. Odds ratio with 95% confidence interval were used to estimate the effects of BRAFV600E mutation on each clinicopathological parameter with fixed-effect model or random-effect model.
Results: Sixty-one studies published, including 32407 CRC patients from multiple countries, were included in the meta-analysis. The overall BRAFV600E mutation rate was 11.38%, and BRAFV600E mutation was positively related to high disease stage (OR=0.81; 95% CI=0.72-0.92; P=0.001), high T stage (OR=0.51; 95% CI=0.40-0.65; P<0.00001), proximal colon (OR=4.76; 95% CI=3.81-5.96; P<0.00001) or right colon (OR=5.15; 95% CI=4.35-6.10, P<0.00001) tumor location, poor tumor differentiation (OR=0.27; 95% CI=0.21-0.34; P<0.00001), mucinous histology (OR=2.97; 95% CI=2.37-3.72; P<0.00001), K-ras-wild type (OR=0.04; 95% CI=0.02-0.07; P<0.00001), TP53-wild type (OR=0.50; 95% CI=0.31-0.78; P=0.003), deficient DNA mismatch repair (OR=2.93; 95% CI=1.78-4.82; P<0.00001), high microsatellite instability (OR=11.15; 95% CI=8.51-14.61; P<0.00001) and high CpG island methylator phenotype (OR=0.04; 95% CI=0.03-0.08; P<0.00001). Conclusions: Our updated meta-analysis demonstrated that BRAFV600E mutation was related to poor prognosis of CRC and associated with the distinct molecular phenotypes.

Introduction

Colorectal cancer (CRC), the third most common cancer, causes the fourth most frequent cancer-related deaths worldwide 1. It has been widely recognized that constitutive activation of the RAS-RAF-MEK- ERK (MAPK) pathway plays a critical roles in CRC development and progression 2. Gain-of-function mutations of the key protein BRAF in this pathway will constitutively activate this pathway, suggesting the crucial role of BRAF mutation in CRC 3. The BRAF mutation, inducing the substitution of valine for glutamate at position 600 of the b-raf protein, accounts for approximately 90% of BRAF mutations and has more important significance compared to other BRAF mutation types in CRC, and about 10% of CRC patients harbor the BRAF mutation 3. Increasing studies have discussed the relationship between BRAF mutation and the effect of anti-EGFR inhibitors in CRC, but the effects of BRAF mutation on the clinicopathological characteristics of CRC remains limited. Therefore, in this article we comprehensively estimate the association between BRAF mutation and clinicopathological characteristics of CRC patients.

Methods

Literature search strategy

We searched PubMed, Web of Science, Embase, and PMC database for relevant publications with the following search terms: (“colorectal cancer” or “rectal cancer” or “colon cancer”) and (“BRAF mutation” or BRAF). Original articles about human studies written in English published before June 18, 2018 were included.

Inclusion criteria

The studies were gone through in accordance to the predetermined selection. The inclusion criteria were: (1) the association between BRAF mutation and clinicopathological characteristics was studied; (2) sufficient published data for calculating an odds ratio (OR) and 95% confidence interval (CI) were reported; (3) the most appropriate article was selected when multiple articles associated with the same patient population were published. The exclusion criteria were: (1) review articles; (2) articles without enough data to analyzed; and (3) single case reports. The quality of each study was assessed using the Newcastle-Ottawa Scale (NOS).

Data extraction

For every appropriate study, the relevant data were extracted, including name of the first author, publication year, country where the study was conducted, follow-up time, number of patients with BRAF mutation, total number of patients, patient demographics (age and gender); clinicopathological characteristics including tumor site, disease stage, T stage, N stage, metastasis status, tumor size, tumor differentiation and mucinous histology; molecular characteristics including KRAS mutation status, CpG island methylator phenotype (CIMP), TP53 mutation status, DNA mismatch repair (MMR) status and microsatellite instability (MSI) status).

Statistical analysis

Meta-analysis was performed using RevMan (Cochrane Collaboration, Oxford, UK). The strength of the association between the BRAF mutation and clinicopathological parameters was assessed by odds ratio (OR) with the corresponding 95% confidence interval (CI). In the course of data pooling, statistical heterogeneity was defined by using chi-square-based Q-test. The I2 value indicates the degree of heterogeneity. A P-value<0.10 and/or I2>50% are considered significant heterogeneity, and then a random-effect model is used. Otherwise, a fixed-effect model is used.

Results

Characteristics of eligible literatures

According to the search terms, a total of 1332 eligible citations were obtained. After screening the abstract, 1228 citations were excluded. Among the remaining 104 citations, 43 citations were excluded because of the reasons shown in Figure 1. Finally, 61 studies published from 2006 to 2018 were included in the meta-analysis (Figure ). A total of 32407 CRC patients from China, Japan, South Korea, India, French, Sweden, Greece, American, Netherlands, Italy, Germany, Australia, and so on were included, and among these patients, 3688 patients were with BRAF mutation (11.38%). The study sample sizes ranged from 69 to 1980 cases. BRAF mutation rate among all studies ranged from 3.14% to 23.14%, which was consistent with the results in the previous study 4. All specimens were derived from CRC tissues by either biopsy or surgical resection, and were detected for BRAF mutation status mainly by direct sequencing, pyrosequencing, allele-specific PCR and immunohistochemistry (IHC) method.

Figure 1

A flow chart of the study selection process.

The basic characters of the 61 eligible studies were summarized in Supplementary Table . Thirty-five studies are with sample size below 500 5-37, 64, whereas twenty-six studies are with sample size over 500 38-63. The earliest study was published in July 2005 51, and the latest study was published in August 2017 49. Most of these studies involved patients with stage I-IV CRC 5, 6, 8, 10, 12, 14, 15, 19, 21, 22, 24, 26, 27, 31, 32, 34, 38, 39, 42, 44-47, 51, 52, 54, 55, 57-60, 64, and six studies only involved patients with stage IV CRC 23, 29, 33, 35, 43, 56. All the studies have a NOS score of ≥5, and 18 studies have a NOS score of ≥6 (Supplementary Table ).

Correlation of BRAF mutation with clinicopathological characteristics of CRC patients

Demographic characteristics (Age and Gender)

A total of 14 studies investigated the association between BRAF mutation and age. Of 2434 patients younger than 60 years, 182 (7.47%) patients were BRAF mutation positive, and 551 (14.26%) of 3864 patients 60 year or older were BRAF mutation positive. The association between BRAF mutation and age did not reach statistical significance (OR=0.66; 95% CI=0.43-1.00; P=0.05) (Figure , Table ). Fifty-six studies analyzed the association between BRAF mutation and gender. Of 14453 male patients, 1214 (8.40%) CRC patients were with BRAF mutation, and 1822 (15.04%) of 12048 female patients were with BRAF mutation. There was a significantly negative association between BRAF mutation and male gender (OR=0.53; 95% CI=0.49-0.57; P<0.00001) (Figure , Table ).

Clinical Factures (Disease stage, T stage, N stage, tumor size, metastasis status, tumor site, tumor differentiation, and mucinous histology)

Twenty-six studies analyzed the association between disease stage and BRAF mutation. Of 5457 patients with stage I or II, 528 (9.68%) patients were BRAF mutation positive, and 733 (11.65%) patients was BRAF mutation positive from 6290 patients diagnosed with stage III or IV disease. Stage I or II were negatively related to BRAF mutation (OR=0.81; 95% CI=0.72-0.92; P=0.001), indicating that CRC patients with BRAF mutation trend to have more advanced disease stage (Figure , Table ). Furthermore, patients with BRAF mutation were also negatively associated with low T stage (OR=0.51; 95% CI=0.40-0.65; P<0.00001) (Figure , Table ). However, the overall analysis showed BRAF mutation did not statistically significant correlated with tumor size (OR=0.83; 95% CI=0.45-1.55; P=0.56) (Figure , Table ), N stage (OR=0.85; 95% CI=0.73-1.00; P=0.05) (Figure , Table ) and metastasis status (OR=1.30; 95% CI=0.90-1.88; P=0.16) (Figure , Table ).

In total, forty-five studies investigated the relationship between BRAF mutation and tumor site. And among these studies, twenty-four studies categorized tumors as proximal colon, distal colon or rectal tumor, and another twenty-one studies classified the tumor as right colon, left colon or rectal tumor. The final results showed that BRAF mutation was significantly associated with proximal colon tumor location (OR=4.76; 95% CI=3.81-5.96; P<0.00001) or right colon tumor location (OR=5.15; 95% CI=4.35-6.10; P<0.00001) (Figure , Table ).

Twenty studies assessed the association between BRAF mutation and tumor differentiation. 592 (7.81%) patients were with BRAF mutation of 7579 patients with well or moderate differentiation, and 310 (26.34%) patients were with BRAF mutation of 1177 patients with poor differentiation. It was obvious that BRAF mutation was negatively associated with well or moderate differentiation, indicating that CRC patients with BRAF mutation trend to have aggressive tumor phenotype (OR=0.27; 95% CI=0.21-0.34; P<0.00001) (Figure , Table ). Besides, BRAF mutation was also strikingly related to mucinous histology (OR=2.97; 95% CI=2.37-3.72; P<0.00001) (Figure , Table ).

Molecular Features (K-ras mutation status, TP53 mutation status, MMR capacity, MSI status, and CIMP)

Twelve studies analyzed the association between BRAF mutation and K-ras mutation status. Notably, K-ras mutation and BRAF mutation were negatively related (OR=0.04; 95% CI=0.02-0.07; P<0.00001) (Figure , Table ). Of 1616 K-ras-mutated patients, only nine (0.56%) ones were BRAF mutated, while 468 (15.38%) patients of 3043 K-ras-wild patients were BRAF mutated. Interestingly, BRAF mutation was also negatively associated with TP53 mutation (OR=0.50; 95% CI=0.31-0.78; P=0.003) (Figure and Table ).

Only two studies investigated the relationship between BRAF mutation and mismatch repair (MMR) capacity. The results showed that 24 (18.75%) patients were BRAF mutation positive from 128 patients with deficient MMR (dMMR) capacity, and 90 (7.51%) patients were BRAF mutation positive from 1198 patients with proficient MMR (pMMR) capacity. BRAF mutation was significantly related to dMMR (OR=2.93; 95% CI=1.78-4.82; P<0.00001) (Figure , Table ). Twenty-seven studies investigated the BRAF mutation and microsatellite instability (MSI). Of 1872 patients with high microsatellite instability (MSI-High), 864 (46.15%) patients were BRAF mutated, and of 11668 patients with low microsatellite instability (MSI-low) or microsatellite stable (MSS), 810 (6.94%) patients were BRAF mutated. There was a significant association between BRAF mutation and MSI-high (OR=11.15; 95% CI=8.51-14.61; P<0.00001) (Figure , Table ). Ten studies were analyzed for CpG island methylator phenotype (CIMP) and BRAF mutation. Of 4112 patients with low or negative CIMP, 179 (4.35%) patients were with BRAF mutation, and of 834 patients with high CIMP, 359 (43.05%) patients were with BRAF mutation. According to the result, BRAF mutation was negatively associated with high CMIP (OR=0.04; 95% CI=0.03-0.08; P<0.00001) (Figure and Table ).

Additional analyses

A funnel plot of effects calculated from individual studies examining the association between BRAF mutation and disease stage was conducted to estimate the presence of publication bias. Because there are small studies with negative results in the literature, no strong indication of publication bias exist among the series of studies included in this meta-analysis.

Discussion

BRAF mutation was an important molecular alternation in CRC patients. In our study, the highest BRAF mutation rate reached to 23.14% and the average BRAF mutation rate was 11.35% among all the involved studies, similar to other reports 4. Clinicopathological parameters have crucial roles in predicting the prognosis of cancer patients. It is necessary to clarify the relationship between BRAF mutation and clinicopathological parameters in CRC patients 65. Our meta-analysis indicated that BRAF mutation was significantly associated with female, advanced disease stage, high T stage, proximal or right tumor location, poor tissue differentiation and mucinous phenotype. As high disease stage, high T stage, poor tissue differentiation and mucinous histology were the multiple risk factors of the prognosis in CRC patients, it may be deemed that BRAF mutation was a poor predictive indicator 20, 51, 65. Our study also showed that BRAF mutation had a crucial association with disease stage, T stage, N stage and tissue differentiation, which demonstrated the important role of BRAF mutation in occurrence and development of CRC.

Intriguingly, tumors located in the proximal colon were 4.76-fold more likely to have BRAF mutation than tumor located in the distal or rectal colon. Moreover, the BRAF mutation was 5.15-fold more frequent in tumors located in the right colon than tumors located in the left or rectal colon. The association between BRAF mutation and tumor sites was very strong and the reason of the association has not been clarified clearly. Previous studies have indicated that colorectal tumors located in different sites have totally different outcomes and specific biomolecular characteristics 66. Our study also demonstrated that the difference in regard to BRAF mutation in different tumor location. Moreover, the different BRAFmutation rates among different tumor sites might be useful for formulating treatment therapy for CRC located in different tumor sites 67, 68.

Nowadays, in addition to clinicopathologic stage and histological morphology, molecular markers play increasing roles in making therapeutical decision for cancer patients. Melanoma with BRAF mutation is more sensitive to immunotherapy 69. Deficient MMR status has been demonstrated to predict the response of PD-1 blockade in metastatic CRC 70, and MSI-High also has been recognized as a predictive factor of programmed death ligand-1 inhibitor pembrolizumab in metastatic/refractory CRC 71. Our results revealed that BRAF mutation was significantly related to K-ras-wild type, TP53-wild type, deficient MMR, high MSI and high CIMP. This association between BRAF mutation and other molecular features cloud be important to understand the molecular distinction between CRC patients with or without BRAF mutation. However, the association between BRAF mutation and the therapeutical response in CRC needs more prospective investigations.

In this article, although we conducted comprehensive and detailed meta-analysis, there are still some limitations. Firstly, with regard to some clinicopathologic characteristics, the number of the involved patients was limited. Small studies are prone to introduce unstable results and related to publication bias. Secondly, most of these studies were retrospective or observational studies (data not shown), which might induce heterogeneity. Thirdly, the mutation detection assays were different among these studies. The most two commonly used methods are direct sequencing and pyrosequencing. Different BRAF mutation assays also affected the accuracy and precision of the pooled estimates.

In conclusion, our updated and comprehensive meta-analysis based on a large number of clinical data demonstrated that BRAF mutation is a biological predictor for poor prognosis in CRC patients, which helps to elucidate the mechanisms of progression and metastasis of CRC and to develop novel therapeutic strategies for CRC.

Supplementary tables.

Click here for additional data file.

Table 1

Overall analysis of the association between BRAF mutation and clinicopathological features in CRC patients.

Clinicopathological features	OR	95% CI	P value
Demographic characteristics
age (<60 years)	0.66	0.43-1.00	0.05
gender (male)	0.53	0.49-0.57	<0.00001
Clinical factures
disease stage (stage I-II)	0.81	0.72-0.92	0.001
tumor size (<5cm)	0.83	0.45-1.55	0.56
T stage (T1-2)	0.51	0.40-0.65	<0.00001
N stage (N0)	0.85	0.73-1.00	0.05
metastasis (yes)	1.30	0.90-1.88	0.16
tumor location (proximal colon)	4.76	3.81-5.96	<0.00001
tumor location (right colon)	5.15	4.35-6.10	<0.00001
tumor differentiation (well/moderate)	0.27	0.21-0.34	<0.00001
mucinous histology (mucinous)	2.97	2.37-3.72	<0.00001
Molecular features
K-ras mutation status (mutation)	0.04	0.02-0.07	<0.00001
TP53 mutation status (mutation)	0.50	0.31-0.78	0.003
MMR status (dMMR)	2.93	1.78-4.82	<0.00001
MSI status (MSI high)	11.15	8.51-14.61	<0.00001
CIMP phenotype (CIMP low/negative)	0.04	0.03-0.08	<0.00001