Background
Colorectal cancer (CRC) is the most common gastrointestinal cancer, and one of the leading causes of cancer deaths worldwide [
1]. Recently, the incidences of CRC have been increasing in Asian countries including Korea [
2]. Metastatic or recurrent colorectal cancer (MRCRC) has a poor prognosis. Many recent studies have demonstrated that MRCRC with KRAS mutant-type (MT) are resistant to anti-epidermal growth factor receptor (EGFR) agents such as cetuximab or panitumumab, which showed efficacy as monotherapy or in combination with cytotoxic agents in MRCRC patients with a wild-type (WT) genotype in KRAS codons 12 and 13 [
3‐
6]. KRAS mutational analysis of primary or metastatic tumor tissues is recommended for all MRCRC patients receiving anti-EGFR chemotherapy.
The KRAS mutation is known to occur in about 40% of all CRC cases [
7]. The activating mutation of KRAS induces stimulation of the RAS/MAPK signaling pathway independent of EGFR, subsequently followed by dysregulated cell growth, proliferation, and survival [
8]. However, the actual differences in clinical behaviors between tumors with WT and MT KRAS, except resistance to anti-EGFR agents, remain controversial. There have been some reports showing that MT KRAS tumors have a poorer prognosis than WT KRAS tumors; however, such a relation between KRAS status and prognosis in CRC was not verified in other reports [
9‐
13]. Recently, it has been reported that the recurrence pattern after the curative resection of CRC is determined according to the KRAS mutational status, showing the positive correlation of KRAS mutation with lung relapse [
14,
15]. Furthermore, some reports demonstrated a different prognosis and clinical presentation with respect to the mutation types of the KRAS gene [
16,
17]. However, these findings need to be confirmed by additional studies.
In addition, WT KRAS tumors are not always sensitive to EGFR-targeted therapy. Many studies have focused on the downstream signaling pathways of EGFR with the goal of identifying other predictive factors for patients resistant to anti-EGFR agents. BRAF or PIK3CA mutation and PTEN loss were suggested as other biomarkers predicting a lack of response to anti-EGFR agents [
18‐
21]. Since the BRAF mutation or PTEN loss are rare [
22], other mechanisms of resistance to anti-EGFR agents are expected to play a substantial role. Although the KRAS mutation is believed to be an early event in the carcinogenesis of CRC [
23] and it has been demonstrated that the KRAS mutational status is highly concordant between primary tumors and related metastases [
14,
15,
24‐
28], some recent studies have produced contradictory results in regards to the KRAS mutational status [
11,
29‐
31]. Especially, most previous studies on the concordance of the KRAS mutation status were conducted on easily available hepatic metastatic tissues [
24,
26,
28]. Therefore, if the discordant rate is different according to the respective metastatic organs, it may also contribute to a poor response to anti-EGFR agents.
Therefore, this study was performed to determine whether there are differences in clinical behavior such as metastatic patterns according to the KRAS mutational status in MRCRC patients. Additionally, the concordance rates of KRAS mutation status between primary sites and respective metastatic organs were also evaluated.
Methods
Study population
Among patients with histologically confirmed colorectal adenocarcinoma who had been treated or followed up at medical oncology clinics of Seoul National University Bundang Hospital (SNUBH) between April 2010 and February 2011, 151 MRCRC patients who had undergone surgical resection or biopsy of both primary tumors and related metastatic sites were consecutively enrolled. Eight patients were excluded because of a failure in the process of KRAS mutational analysis on either primary or metastatic site, thus 143 patients were finally included in this study. The clinical data on enrolled patients was retrieved from the CRC database maintained at SNUBH [
32], and additional data collection was retrospectively supplemented after a review of the electronic medical record (EMR). This study was approved by the institutional review board of SNUBH.
Preparation of tumor DNA and sequence analysis of KRAS gene
Formalin-fixed paraffin-embedded primary tumor and metastatic tissue specimens were chosen for each patient and all specimens were microdissected manually under the supervision of experienced gastrointestinal pathologists. After manual microdissection, > 60% of the sample area was shown to contain tumor cells as estimated from the H&E-stained slides. The DNA was extracted using a chelating ion exchange resin (InstaGene Matrix, Bio-Rad). For mutation analyses in codons 12, 13, and 61 of the KRAS gene, extracted tumor DNA samples were subjected to automated sequencing using an ABI 3130xl Genetic Analyzer (Applied Biosystems) and the data was analyzed with GeneMapper Software Version 4.0 (Applied Biosystems). Primer sequences for exon 1 were 5’-AACCTTATGTGTGACATGTTCTA-3’ (forward) and 5’-TGGTCCTGCACCAGTAAT-3’ (reverse); for exon 2, 5’-ACTGTAATAATCCAGACTGTGTT-3’ (forward) and 5’-CCCACCTATAATGGTGAATATCT-3’ (reverse). The polymerase chain reaction (PCR) conditions were as follows: one cycle of 95°C for 5 minutes; 34 cycles of 95°C for 30 seconds, 58°C for 30 seconds, and 72°C for 1 minute; and one cycle of 72°C for 10 minutes.
Determination of microsatellite instability (MSI) status
The MSI status of paired normal and tumor DNA samples was assessed using the Bethesda panel of 5 microsatellite markers. Sequencing was conducted using the same PCR system as the KRAS mutational analysis. The MSI status was classified as high frequency if ≥ 2 of 5 markers exhibited instability, as low frequency if one marker exhibited instability, and as stable if no markers exhibited instability.
Statistical analysis
The concordance rate of KRAS mutational status in primary tumors and related metastases was evaluated; the Kappa index was measured using Cohen’s k-test, which can assess the concordance between categorical variables of the same individuals. The effect of KRAS mutational status on clinicopathologic parameters and initial metastatic sites were assessed using Pearson’s chi square or Fisher’s exact tests. Multivariate logistic regression analyses were performed to evaluate the effect of other clinicopathologic features on liver-only and lung-only metastases besides the KRAS status. The relationship between the discordant rates of the KRAS mutation status and various clinical parameters including respective metastatic sites was also evaluated using univariate (Pearson’s chi square or Fisher’s exact tests) and multivariate logistic regression analyses. In multivariate logistic regression analysis, the forward stepwise regression model including variables with a P-value < 0.10 in univariate analysis was used. Two-sided P-values of < 0.05 were considered significant. All analyses were performed using SPSS for Windows, version 17.0 (SPSS Inc., Chicago, IL, USA).
Discussion
To the best of our knowledge, this is the first large-scale study in which a KRAS mutational analysis was performed between primary tumors and corresponding metastases in Asian MRCRC patients. Overall, a high concordance rate of KRAS status was observed in Korean MRCRC patients, as had been previously reported in Western populations. We observed significant differences in initial metastatic patterns according to the KRAS mutational status. MT KRAS tumors developed lung metastases more frequently as the initial metastatic site; however, liver and distant LN were more frequently involved as the initial metastatic sites in WT KRAS tumors. Additionally, the degree of concordance in KRAS mutational status was significantly different according to the sites of related metastatic organs, where the lung was the most frequent metastatic site showing the discordance of KRAS status.
Our study demonstrated that the clinical presentation of CRC varied according to the KRAS mutational status. KRAS mutational status was shown to affect the presenting pattern of distant metastasis in MRCRC patients. Recurrent cases after curative treatment for localized diseases were more common in MT KRAS patients; however, systemic metastasis was more frequent at the time of initial CRC diagnosis in WT KRAS patients (Table
1). One previous study showed that the risk of recurrence was significantly higher for MT KRAS than WT KRAS tumors in patients with localized CRC [
10]. If localized CRC patients with MT KRAS had more chances of recurrence, then the MT KRAS cases would be selected and thus a higher frequency of MT KRAS in recurrent CRC patients would be expected than in patients with stage IV disease at the time of initial CRC diagnosis. However, as our patient cohort is relatively small, our assumption needs to be further investigated in future large studies.
In the present study, organs initially involved by distant metastasis were shown to be different according to the KRAS mutational status. Patients with MT KRAS had an initial lung metastasis more frequently than patients with WT KRAS. In contrast, the WT KRAS patients had liver or distant LN metastases more frequently as the initial metastatic sites. However, other metastatic sites such as the peritoneum were not affected by KRAS status (Table
5). When the analyses were conducted on patients with initial distant metastasis confined to a single organ, the results were similar except for distant LN metastasis (Table
6). However, since the number of cases with LN-only metastasis was small (N = 6), the difference of percentages (7.7% vs. 3.3%) might not have reached statistical significance. Although the MT KRAS tumors showed a trend for more frequent development of peritoneum-only metastasis (13.5% vs. 27.9%;
P = 0.062), it was not statistically significant (Table
6). In the multivariate analysis, KRAS status, primary tumor site, and clinical situations for the development of systemic metastasis were significant predictors for liver-only and/or lung-only metastases (Table
7). The reason why the clinical situations of developing systemic metastasis influenced the initially involved metastatic organs (liver or lung) is not clear; this may be related to the process of patient enrollment to this study as patients with available tissues from both primary and paired metastatic sites were only included. During this process, recurred CRC patients with tissue-available lung metastasis might be selectively included. However, even after adjusting for these clinical variables, the KRAS mutational status was an independent predictive factor for both liver-only and lung-only metastases in our study. A previous study, which analyzed the KRAS status in primary tumors of CRC patients, showed that there were more MT KRAS tumors in patients with lung metastasis than in patients with liver metastasis [
14]. Based on this finding, they suggest that KRAS-mutated primary CRC tumors can recur with lung metastasis more frequently than with liver metastasis. Although the KRAS mutational status was regarded as positive if KRAS was mutated in any place of primary tumors or related metastases in the present report, our results also support their suggestion. Furthermore, when the analysis was conducted based on the KRAS status of primary tumor, the result of our study was also the same as above. Our work along with previous studies strongly suggests that the sequence of organs involved by systemic metastasis is influenced by KRAS mutational status in CRC patients.
Our results are generally consistent with previous studies that have reported a high concordance rate of KRAS mutation (about 90%) between primary and metastatic tumors [
14,
15,
24‐
28]. Paired metastatic tissues in previous studies were mostly derived from the liver [
24,
26,
28] because these tissues were easily available from hepatic metastasectomy [
33,
34]. However, reports on the degree of KRAS concordance of other metastatic organs other than the liver have been very limited. Our study evaluated the KRAS status of metastatic tissues from various organs besides liver. The degree of KRAS mutational concordance was different according to the related metastatic sites, with a significantly higher rate of discrepancy in lung metastases (32.4%) when compared with other metastatic organs (12.3%) or liver (10.6%) (
P-values < 0.05; Table
8). Although some studies have recently demonstrated a relatively high degree of discordance, with a discordance rate of up to 50% [
11,
29‐
31], there has been no report showing the site-specific KRAS discordance, as was shown in this study. The mechanism behind the discordant KRAS mutational status is still not exactly known [
31,
35,
36]. Sampling errors, heterogeneity within primary tumors, and the development of mutations during the process of metastasis may be the causes of this discordance. The reason why lung is the most frequent site where the KRAS discordance takes place is also unknown. In the present study, we analyzed the pattern of KRAS discordance (i.e. from WT in primary tumors to MT in related metastatic sites or vice versa); as the sample size was small (N = 25), the KRAS discordance pattern did not show any relation to clinical situations for the development of systemic metastasis (stage IV vs. recurred) or metastatic organs from which tumor specimens were obtained. Among 12 cases with KRAS-discordant lung metastasis, 6 cases (50%) had the change of KRAS status from WT in a primary tumor to MT in the lung and 6 cases (50%) had vice versa; there was no statistically significant difference in the discordance pattern between the lung and other paired metastatic organs (Table
4). Therefore, the underlying causes of KRAS discordance need to be further evaluated in future large studies.
Our study has some limitations. First, this study was performed at a single institution and all MRCRC patients diagnosed at our institution were not included. Instead, only MRCRC patients with both primary and paired metastatic tissues were consecutively included. In such a process, unrecognized biases might have influenced our study. Second, a KRAS mutational analysis was not repetitively conducted in cases with discordant KRAS status partly because of insufficient remaining tissue specimens for further examination. This might raise concerns about the sensitivity of the KRAS mutation analysis. We actually used traditional sequencing (Sanger) method with relatively low sensitivity for KRAS mutation analysis [
37]. In addition, biopsied specimens of primary tumor showed a trend for higher discordance rate than resected specimens in this study (40.0% vs. 15.8%;
P = 0.073), although no significance was shown in multivariate analysis. However, we used tumor cell enrichment by microdissection under the supervision of experienced pathologists to increase the sensitivity of the sequencing method. In addition, biopsy of primary tumors in our study was all performed by endoscopy. Actually, a biopsy of distant metastasis may be more problematic than endoscopic biopsy of primary tumors in context of tumor cell percentage [
38]. In the present study, a small number of metastatic specimens (8/143) was obtained from needle biopsy and only 1 case had KRAS discordance [1/8 (12.5%) for biopsied metastatic specimens vs. 24/135 (17.8%) for resected metastatic specimens;
P = 1.000]. Moreover, the concordance rates observed between primary and metastatic tissues including the liver only (89.4%) or all metastatic organs other than lung (87.7%) were similar to the concordance rates reported in previous studies. All these findings suggest that the high KRAS discordant rate of lung metastasis (32.4%) had not simply resulted from types of tumor tissue specimens (biopsied vs. resected) or less sensitive analytic methods performed at our institute. Instead, the results from our study reflect the real situation of clinical fields as the traditional sequencing (Sanger) analysis is the most frequent method used in the real clinical practice setting. More sensitive methods, such as real-time PCR for KRAS mutation analysis, are only used in the investigational setting and not widely spread in the clinical practice.
Despite these limitations, our study provides some clinically meaningful suggestions. The present study demonstrated that the KRAS mutational status was an independently predictive factor for organs initially involved by distant metastasis. This observation implies that surveillance strategies after curative surgery might be tailored to individual CRC patients according to the KRAS mutational status. Postoperative surveillance might be more focused on lung metastasis (i.e., chest computed tomography) in patients with MT KRAS than in patients with WT KRAS, when considering the chance of performing metastasectomy after the early detection of pulmonary metastasis. Our study also raised the hypothesis that the discordant rates of KRAS mutational status might be metastatic site-specific in CRC. Using the sequencing method, we found different discordant rates according to the metastatic sites. A high KRAS discordant rate in patients with lung metastasis, observed in our study, warrants further large validation studies.
Competing interests
The authors declare that they have no competing interests.
Authors’ contributions
MJK carried out patient data collection and statistical analysis and drafted this manuscript. HSL and KUP participated in sequencing analysis, data collection and contributed to the interpretation of the data. JHK, YJK, DWK, SBK and JSK treated many of the enrolled patients and contributed to the interpretation of data. JHK and JOL assisted patient data collection and the statistical analysis of the study. SMB and JSL contributed substantially to the interpretation of data, and helped the statistical analysis. KWL conceived this study, participated in its design and coordination, treated many of enrolled patients, and helped to draft the manuscript. All authors read and approved the final manuscript.