The Prognostic Utilities of DNA Mismatch Repair Status and KRAS and BRAF Mutation in Korean Colorectal Cancer Patients: The KASID Multicenter Study

Abstract

Introduction

KRAS, BRAF, and DNA mismatch repair (MMR) mutations aid clinical decision-making for colorectal cancer (CRC) patients. To ensure accurate predictions, the prognostic utilities of these biomarkers and their combinations must be individualized for patients with various TNM stages.

Methods

Here, we retrospectively analyzed the clinicopathological features of 904 Korean CRC patients who underwent CRC surgery in three teaching hospitals from 2011 to 2013; we also assessed the prognostic utilities of KRAS, BRAF, and MMR mutations in these patients.

Results

The overall frequencies of KRAS and BRAF mutations were 35.8% and 3.2%, respectively. Sixty-nine patients (7.6%) lacking expression of ≥1 MMR protein were considered MMR protein deficient (MMR-D); the remaining patients were considered MMR protein intact. KRAS mutations constituted an independent risk factor for shorter overall survival (OS) in TNM stage I–IV and stage III patients. BRAF mutations were associated with shorter OS in TNM stage I–IV patients. MMR-D status was strongly positive prognostic in TNM stage I–II patients.

Discussion/Conclusion

To our knowledge, this is the first multicenter study to explore the prognostic utilities of KRAS, BRAF, and MMR statuses in Korean CRC patients. Various combinations of KRAS, BRAF, and DNA MMR mutations serve as genetic signatures that affect tumor behavior; they are prognostic in CRC patients.

Introduction

Colorectal cancer (CRC) is the third most commonly diagnosed cancer and the second leading cause of cancer death worldwide [1]. The American Institute for Cancer Research has reported that the CRC incidence in South Korea is the second highest globally [2]. Various molecular mechanisms are involved in the CRC pathogenesis, of which chromosomal instability (CIN) (75%) and microsatellite instability (MSI) (15%) play major role [3]. CIN affects critical genes such as APC, KRAS, PI3K, and TP53 [4]. MSI is caused by the inactivation of genes involved in the correction of nucleotide − base mismatches in DNA [5]. Loss of expression of mismatch repair (MMR) genes including MLH1, MSH2, MSH6, and PMS2 can be caused by spontaneous events or germinal mutations such as Lynch syndrome [6, 7].

In order to correlate cancer cell phenotype with clinical implication and guide specific targeted therapies, the CRC Subtyping Consortium made a single consensus, based on gene expression data, into 4 distinct groups, known as the consensus molecular subtypes (CMSs) [8]. CMS1, known as MSI-immune subtype, is characterized by high BRAF mutation, hypermethylation of CpG islands, an association with an impaired MMR system, and the infiltration of immunogenic lymphocytes in the tumor microenvironment (14%) [9]. The majority of CRC previously described as CIN was divided into 3 subtypes based on transcriptomic profiling. CMS2, known as canonical subtype, displayed epithelial signatures with prominent WNT and MYC signaling activation (37%) [10]. CMS3, known as metabolic subtype, has genomic features consistent with CIN but has fewer DNA somatic copy number alterations and more hypermutated/MSI than CMS2 and CMS4 (13%) [8]. CMS4, known as mesenchymal subtype, exhibits high transforming growth factor β (TGFβ), extremely low levels of hypermutation, MSS status, and very high somatic copy number alteration counts (23%) [11].

KRAS and BRAF mutations are associated with upregulation of the RAS/RAF/MAPK signaling pathway, which has a critical role in CRC development [12]. The mechanisms that underlie CIN include changes in chromosomal segregation and the DNA damage response (e.g., KRAS mutations) [13]. Both BRAF mutation and MMR statuses distinguish sporadic tumors from Lynch syndrome-related tumors [14]. Among MMR protein deficient (MMR-D) CRCs, 34–70% have BRAF mutations [15]. MMR-D and the CpG island methylator phenotype are closely associated [16]. Most MMR-D tumors develop through CpG island methylator phenotype-associated methylation of MLH1 [17]. Moreover, the presence of a BRAF mutation in an MLH1-deficient CRC patient suggests a sporadic tumor rather than a Lynch syndrome tumor [14].

However, the prognostic utilities of KRAS and BRAF mutations, as well as MMR status, in CRC patients with specific TNM stages remain unclear [18]. We performed a large retrospective study in such Korean patients and evaluated their clinicopathological features and prognoses according to TNM stage. In addition, we assessed the associations of KRAS mutations, BRAF mutations, and MMR status with each TNM stage.

Materials and Methods

Patients and Treatment

Clinical and pathological information was collected regarding 904 patients who underwent CRC surgery in three teaching hospitals (Seoul St. Mary's Hospital, Gangdong; Kyung Hee University Hospital, and Yeungnam University Hospital) from 2011 to 2013.

Inclusion criteria were complete data concerning tumor markers, survival, and stage. Cases without any one of tumor marker records about KRAS mutation, BRAF mutation, or MMR status were excluded. A total of 210 patients with incomplete tumor marker records were excluded. Fifteen patients with histologically confirmed Tis were also excluded from the study. Medical records and pathology reports were analyzed. Adjuvant chemotherapy was recommended to stage III or high-risk (with cancer-related obstructions or perforations; with poorly differentiated cancers; with lymphovascular/perineural invasion; or with MMR protein intact [MMR-I] status) stage II patients, based on the National Comprehensive Cancer Network guidelines. FOLFOX (fluorouracil, leucovorin, and oxaliplatin), FOLFIRI (fluorouracil, leucovorin, and irinotecan), LF (leucovorin and fluorouracil), XELODA (capecitabine), and XELOX (capecitabine and oxaliplatin) systemic chemotherapies were prescribed according to each patient's clinical situation. Depending on KRAS status, target therapy was offered as an adjunct to systemic chemotherapy.

This study was approved by the Institutional Review Board of Seoul St. Mary's Hospital (approval number: KC17RCDI0781). This study has been designed to review medical records retrospectively. Medical records that can perceive identities are anonymized. No one other than the researcher could browse these medical records. Consents from patients were waived.

Analysis of KRAS, BRAF, and MMR Status

Tissue samples from tumors and normal colonic mucosae were obtained from all patients after resection. Formalin-fixed paraffin-embedded tissue samples were sectioned at a thickness of 10 μm. After deparaffinization, tissue samples were digested with proteinase K (Qiagen Inc., Valencia, CA, USA); KRAS exon 2 DNA was then isolated and amplified via polymerase chain reaction (PCR) using the forward primer (5′-AAGGCCTGCTGAAAATGAC-3′) and reverse primer (5′-TGGTCCTGCACCAGTAATATG-3′). The PCR products were purified using QI-Aquick PCR Purification Kits (Qiagen) and sequenced on an ABI 3730 automated platform (Applied Biosystems Inc., Foster City, CA, USA). Similarly, exon 15 of BRAF was amplified via PCR using the forward primer (5′-AATGCTTGCTCTGATAGGAAAAT-3′) and reverse primer (5′-TAATCAGTGGAAAAATAGCCTC-3′).

MMR status was determined by the immunohistochemical (IHC) absence of proteins encoded by MLH1, MSH2, MSH6, or PMS2. IHC for MLH1, MSH2, MSH6, and PMS2 was performed on FFPE tumor tissue block on a Ventana BenchMark Ultra device (Ventana Medical Systems, Arizona, USA). As a multicenter study was conducted, each result was confirmed by different pathologists. IHC for the detection of MMR proteins in CRC samples is a simple tool to determine MMR deficiency [19, 20, 21]. To minimize interobserver variation between different pathologists, reporting format from the Gastrointestinal Pathology Study Group of the Korean Society of Pathologists has been used [22]. Tumors that had lost at least one MMR protein microsatellite marker were considered MMR-D; the remaining tumors were considered MMR-I.

Statistical Analysis

The Mann-Whitney U test was employed for univariate analysis of continuous variables; the results are expressed as means ± standard deviations. The χ² or Fisher's test was used for univariate analysis of categorical variables. Variables significant in univariate analyses were included in multivariate logistic regression analysis. Overall survival (OS) was determined using the Kaplan-Meier method. All statistical analyses were conducted using R, version 4.0.4 (www.r-project.org), based on a significance level of 0.05.

Results

Clinicopathological Features

The clinicopathological features of the 904 patients are summarized in Table 1. Three patients with both KRAS and BRAF mutations were excluded. Mutations in KRAS codons 12 and 13 were detected in 242 (26.8%) and 76 (8.4%) patients, respectively. Four patients (0.4%) had mutations in both KRAS codons 12 and 13.

Table 1.

Clinicopathological features of patients

Clinicopathological features	N (%)
Patients, n	904
Age, median	62.64±11.78
Sex
Male	541 (59.85)
Female	363 (40.15)
HTN
No	518 (57.30)
Yes	386 (42.70)
DM
No	727 (80.42)
Yes	177 (19.58)
Smoking
No	697 (77.10)
Yes	207 (22.90)
Cancer family history
No	759 (83.96)
Yes	145 (16.04)
Tumor location
Cecum	32 (3.54)
Ascending colon	150 (16.59)
Transverse colon	54 (5.97)
Lt. colon	385 (42.59)
Rectum	262 (28.98)
Multiple	21 (2.32)
TNM stage
I	184 (20.35)
II	271 (29.98)
III	335 (37.06)
IV	114 (12.61)
Mucinous type
No	874 (96.68)
Yes	30 (3.32)
KRAS mutant	324 (35.84)
BRAF mutant	29 (3.21)
MMR-D	69 (7.63)

Univariate and Multivariate Analysis of KRAS, BRAF, and MMR Status

The clinicopathological features of patients, stratified according to KRAS, BRAF, and MMR statuses, are summarized in Tables 2, 3, 4. KRAS mutations were significantly associated with smoking, tumor location, and mucinous-type tumors. BRAF mutations and MMR-D status were significantly associated with tumor location and tumor differentiation extent. In our study, smoking has been divided into ever smoker versus never smoker. In univariate analysis, KRAS mutations were significantly associated with never smoking.

Table 2.

Univariate analysis of KRAS mutation

KRAS mutation	Mutant (N = 324)	Wild (N = 580)	p value
Age	63.4±12.9	62.2±11.1	0.054
Sex
Male	175	366	0.128
Female	149	214
HTN
No	184	334	0.871
Yes	140	246
DM
No	269	458	0.165
Yes	55	122
Smoking
No	265	432	0.0153
Yes	59	148
Cancer family history
No	272	487	0.996
Yes	52	93
Tumor location, n (%)
Cecum	21 (6.5)	11 (1.9)
Ascending colon	73 (22.5)	77 (13.3)
Transverse colon	19 (5.9)	35 (6.0)	0.0001
Lt. colon	117 (36.1)	268 (46.2)
Rectum	91 (28.1)	171 (29.5)
Multiple	3 (0.9)	18 (3.1)
TNM stage
I	69	115
II	95	176	0.894
III	117	218
IV	43	71
Mucinous type
No	307	567	0.026
Yes	17	13
Differentiation
Well	28	44
Moderate	290	508	0.087
Poor	6	28

Table 3.

Univariate analysis of BRAF mutation

BRAF mutation	Mutant (N = 29)	Wild (N = 875)	p value
Age	63.2±13.8	62.63±11.7	0.834
Sex
Male	19	522	0.659
Female	10	353
HTN
No	15	503	0.670
Yes	14	372
DM
No	23	704	0.815
Yes	6	171
Smoking
No	22	675	0.825
Yes	7	200
Cancer family history
No	25	734	0.938
Yes	4	141
Tumor location, n (%)
Cecum	4 (13.8)	28 (3.2)
Ascending colon	8 (27.6)	142 (16.2)
Transverse colon	4 (13.8)	50 (5.7)	0.001
Lt. colon	7 (24.1)	378 (43.2)
Rectum	5 (17.2)	257 (29.4)
Multiple	1 (3.4)	20 (2.3)
TNM stage
I	4	180
II	6	265	0.119
III	14	321
IV	5	109
Mucinous type
No	27	847	0.250
Yes	2	28
Differentiation
Well	0	72
Moderate	23	775	0.001
Poor	6	28

Table 4.

Univariate analysis of MMR status

MMR	MMR-D (N = 69)	MMR-I (N = 835)	p value
Age	59.6±14.9	62.9±11.4	0.073
Sex
Male	36	505
Female	33	330	0.221
HTN
No	45	473	0.209
Yes	24	362
DM
No Yes	62 7	665 170	0.058
No	62	665
Smoking
No	53	644	0.953
Yes	16	191
Cancer family history
No	54	705
Yes	15	130	0.241
Tumor location, n (%)
Cecum	8 (11.6)	24 (2.9)
Ascending colon	22 (31.9)	128 (15.3)
Transverse colon	9 (13.0)	45 (5.4)	0.0001
Lt. colon	15 (21.7)	370 (44.3)
Rectum	13 (18.8)	249 (29.8)
Multiple	2 (2.9)	19 (2.3)
TNM stage
I	18	166	0.116
II	23	248
III	21	314
IV	7	107
Mucinous type
No	64	810	0.071
Yes	5	25
Differentiation
Well	3	69
Moderate	51	747	0.001
Poor	15	19

Multivariate analysis was performed by collecting variables that were likely to be highly related in univariate analysis. Table 5 summarizes the results of multivariate logistic regression analysis. Even though highly related, differentiation and KRAS mutation had no statistical significance in univariate analysis. When analyzed in multivariate analysis, differentiation was statistically meaningful in KRAS mutation. Similarly, in MMR-D, age and TNM stage were found to be statistically significant only in multivariate analysis.

Table 5.

Multivariate analysis of KRAS mutation, BRAF mutation, and MMR status

	KRAS mutant			BRAF mutant			MMR-D
	OR	95% CI	p value	OR	95% CI	p value	OR	95% CI	p value
Age	1.01	0.99–1.02	0.313				1.03	1.01–1.06	0.014
TNM stage				1.34	0.88–2.05	0.173	1.42	1.02–1.97	0.038
Location	0.74	0.62–0.89	0.001	0.47	0.27–0.81	0.006	2.19	1.45–3.33	0.001
Differentiation	0.67	0.45–1.00	0.047	5.31	2.04–13.8	0.001	0.15	0.06–0.35	0.001
Mucinous type	2.59	1.21–5.51	0.013
Smoking	0.64	0.45–0.90	0.011

Survival Analysis

For OS analysis, all patients were categorized into three groups: wild type, KRAS mutant, and BRAF mutant. OSs were compared according to TNM stage (shown in Fig. 1) (i.e., stages I–IV, I–II, III, and IV). The results of Cox multivariate analyses are summarized in Table 6.

Fig. 1.

OS at stage I–IV (p value = 0.001, KRAS HR = 2.105, BRAF HR = 5.285) (a), stage I–II (p value = 0.09) (b), stage III (p value = 0.02, KRAS HR = 2.281, BRAF HR = 4.527) (c), and stage IV (p value = 0.2) (d) in wild type, KRAS mutant, and BRAF mutant.

Table 6.

Cox analysis of OS in stage I to stage IV patients

	Stage I–IV		Stage I–II		Stage III		Stage IV
	HR	95% CI	HR	95% CI	HR	95% CI	HR	95% CI
Sex	4.64	1.67–12.85	1.54	0.25–9.43	0.69	0.3–1.58	1.06	0.38–2.97
Age	1.01	0.99–1.04	0.99	0.93–1.07	1.03	0.99–1.07	1	0.96–1.04
Location	1.02	0.71–1.45	0.9	0.29–2.79	0.89	0.55–1.45	1.05	0.57–1.91
Differentiation	1.58	0.66–3.76	0.3	0.03–2.84	2.55	0.66–9.87	0.67	0.13–3.44
Mucinous type	1.54	0.47–5.04			3.74	1.01–13.76
KRAS	2.12	1.18–3.81	6.89	0.75–62.99	2.73	1.17–6.39	1.32	0.48–3.63
BRAF	4.64	1.67–12.85			3.63	0.92–14.38	4.27	0.7–26
MMR	0.93	0.35–2.47	0.13	0.02–0.97	1.37	0.27–6.86	1.51	0.2–11.43

For stage I–IV patients, significant OS differences were apparent among the three groups in both univariate (p value = 0.001, KRAS HR = 2.105, BRAF HR = 5.285) and multivariate analyses. However, such differences were not apparent for stage I–II patients or stage IV patients. For stage III patients, the OSs of KRAS- and BRAF-mutant patients were shorter in univariate analysis (p value = 0.02, KRAS HR = 2.281, BRAF HR = 4.527). However, in multivariate analysis, only KRAS mutations were significantly prognostic.

OSs were also compared according to TNM stage and MMR status (shown in Fig. 2). In stage I–II, the OSs significantly differed between MMR-D and MMR-I patients in both univariate (p value = 0.01, MMR-I HR = 6.913) and multivariate analyses.

Fig. 2.

OS at stage I–IV (p value = 0.7) (a), stage I–II (p value = 0.01, MMR-I HR = 6.913) (b), stage III (p value = 0.9) (c), and stage IV (p value = 0.6) (d) in MMR-D and MMR-I.

Conclusion

We evaluated KRAS, BRAF, and MMR statuses in 904 Korean CRC patients. Recent studies have explored the prognostic utilities of these parameters, as well as their combinations, in CRC patients [23, 24, 25, 26, 27, 28, 29]. The 5-year survival rates of patients with early stage CRC (TNM stage I–II) approach 90%. However, the Surveillance, Epidemiology, and End Results Program database indicates that the survival rate of patients diagnosed with late-stage CRC is only 15.1% [30]. Thus, early diagnosis of CRC is important. There is a need for biomarkers that allow CRC screening and facilitate predictions of prognosis and treatment response.

We found a KRAS mutation rate of 35.8%, consistent with a previous report [31]. However, our BRAF mutation rate was lower than the rates previously reported (8–12%) [32]. Asian countries exhibit lower BRAF mutation rates than Western countries [33]; ethnic differences may influence these findings. Approximately 3–13% of Asian CRC patients have exhibited MMR-D status in previous studies; our results are similar [34]. In contrast, MMR-D CRC constituted approximately 15–20% of all CRCs in Western studies. MMR-D status can be detected by IHC or MSI analyses. CRC patients exhibit good concordance between MLH1/PMS2/MSH2/MSH6 IHC data and the results of MSI analysis via fluorescent PCR combined with fragment length measurement [35].

In univariate analysis of clinicopathological parameters, we found that the tumor locations significantly differed among patients. Compared with wild-type CRCs, KRAS- and BRAF-mutated CRCs were more common in the right colon. Furthermore, MMR-D CRCs were more frequent in the right colon compared with MMR-I CRCs. Salem et al. [36] reported KRAS mutations in 61–71% of right colon CRCs. Yamauchi et al. [37] found high MMR-D and BRAF mutation rates in CRCs of the ascending and transverse colons, as well as CRCs of the hepatic flexure. We subdivided right colon CRCs into cecal cancers, ascending CRCs, and transverse CRCs. MMR-D and BRAF mutations were common in all such tumors; however, KRAS mutations were only frequent in cecal and ascending colon tumors.

In our study, smoking was closely associated with CRCs characterized by KRAS mutation-negative. Consistent with our findings, Samadder et al. [38] reported that smoking-related risk estimates were higher for KRAS mutation-negative than for KRAS mutation-positive tumors. Chen et al. [39] performed a meta-analysis that smoking showed positive correlation with BRAF mutation and MSI positivity. However, in our study, smoking had no statistically significant relationship between BRAF mutation and MMR status. In the study by Chen et al. [39], smoking status was classified as nonsmokers who had less than 100 cigarettes in their lifetime and smokers (former or current). Smoking pack-years or cessation period has not been recorded in our study. Additionally, there were insufficient number of individuals with MMR-D and BRAF mutations who smoked (n = 16 and 7, respectively) to provide accurate assessment. Tailored smoking status and abundant subgroup patient number should be needed to validate precise relationship.

Patients with stag I–IV CRC who had KRAS mutations exhibited poor OS. However, the prognostic utility of KRAS mutations requires further investigation in patients with stages I–II and IV CRC; the results are controversial [18, 40, 41]. Hutchins et al. [40] reported that the risk of recurrence was higher in stage II and III patients with KRAS mutations than in such patients with wild-type tumors [42]. However, Roth et al. [18] found that KRAS mutations were not prognostic in stage II-III CRC patients. It may be difficult to predict prognosis using KRAS status alone; MMR status greatly influences chemotherapy decision-making for stage II patients. It may be necessary to examine multiple factors when determining the effects of KRAS mutations according to disease stage.

Most KRAS mutations are in codons 12 and 13 of exon 2 [31]. For the various RAS genes (HRAS, NRAS, and KRAS), specific exon/codon mutations are associated with typical clinical, pathological, and molecular features [42]. The RAS genes are highly homologous; they are nearly identical in the regions that encode the first 90% of all proteins [43]. Of all RAS mutations in human tumors, mutations in KRAS constitute approximately 85%, mutations in NRAS constitute approximately 15%, and mutations in HRAS constitute <1% [44]. In this study, we were unable to analyze differences according to codon mutation site; the sites were not defined in all patients. Because tumor activity can be affected by codon mutation site, any prognostic difference between KRAS codon 12 and 13 mutations requires further investigation.

We found that BRAF status was significantly associated with prognosis only in stage I–IV patients. A few small cohort studies have reported that BRAF mutations are poorly prognostic in stage II and III patients [45]. However, the prognostic utility of BRAF mutations varies when MMR status is considered. In other studies, BRAF mutations constituted an independent negative prognostic factor in MMR-I patients with stages II and III CRC [46, 47]. However, in MMR-D patients, BRAF mutations were not independently prognostic [48]. The differences may be attributed to the low BRAF mutation rate in Asian countries, where BRAF mutations are present in only 8–12% of all CRCs [49]. Our BRAF mutation rate was only 3.2%.

We found that MMR-D status was a useful prognostic factor in stage I–II patients. Klingbiel et al. [50] analyzed 1,254 patients; they found that MMR-D status was associated with better OS in stage II patients and longer relapse-free survival in stage III patients. In contrast, Phipps et al. [51] pooled data from seven studies with 5,010 patients; they found that non-MMR-D patients exhibited significantly shorter disease-specific survival (compared with MMR-D patients), regardless of other tumor markers or disease stage. The differences may be attributed to differences in sample size. As mentioned, various classifications by molecular subtype including CMS were introduced recently [5]. However, we could not compare prognosis by these classifications with our results because biomarkers from our study were limited. Further studies are required to validate prognosis with these classifications in Korean CRC patients.

Our study had some limitations. First, the number of patients with BRAF mutations was insufficient to yield reliable results in Cox analysis of OS among patients with stag I–II CRC. Second, we could not evaluate NRAS mutations because of insufficient data. Although KRAS- and NRAS-mutant CRCs exhibit similar clinical characteristics, previous studies suggested that NRAS mutations were predictive of resistance to anti-EGFR monoclonal antibodies [52]. Thus, further analysis is required. Third, we did not subdivide the BRAF mutations according to V600E/non-V600E status. Most BRAF mutations are V600E (>90%). BRAF non-V600E mutations are often accompanied by RAS mutations; such patients exhibit longer OS [53].

In conclusion, to our knowledge, this is the first multicenter study to explore the prognostic utilities of KRAS, BRAF, and MMR statuses in Korean CRC patients. In Korean CRC patients, the prognostic utilities of KRAS and BRAF mutations, as well as MMR status, suggest that the various TNM stages are associated with different risk factors. KRAS mutation was an independent risk factor for shorter OS in stage I–IV and stage III. BRAF mutation was an independent risk factor for shorter OS in stage I–IV tumors. MMR-I status was an independent risk factor for shorter OS in stage I–II.

Statement of Ethics

This study was conducted according to the guidelines of the Declaration of Helsinki and approved by the Institutional Review Board of Seoul St. Mary's Hospital (approval number: KC17RCDI0781). The Institutional Review Board of Seoul St. Mary's Hospital decided that written informed consent was not required because anonymizing the identifiable information in the patients' medical records would not have a negative impact on their rights or welfare. Also, without the waiver of consent, this study could not practicably be conducted.

Conflict of Interest Statement

The authors have no conflicts of interest to declare.

Funding Sources

This article was not funded.

Author Contributions

Tae-Woo Kim has contributed in writing and editing the paper as the first author; Soon Woo Hwang has contributed in drafting of the manuscript; Kyeong Ok Kim, Jae Myung Cha, and Young-Eun Joo have contributed in data acquisition and study design; and Young-Seok Cho has revised the manuscript critically.

Data Availability Statement

All data generated or analyzed during this study are included in this article. Further inquiries can be directed to the corresponding author.

Funding Statement

This article was not funded.