Introduction

Cetuximab and panitumumab, anti-epidermal growth factor receptor (EGFR) monoclonal antibodies, are widely used for the treatment of colorectal cancer. Previous studies have established that therapy with these treatments is only effective in patients bearing wild-type KRAS [1, 2]. KRAS, a small, guanosine phosphate-binding protein, is a key component of the RAS/RAF/MAPK pathway, and is activated by phosphorylation of EGFR. Point mutations in codons 12 and 13 of the KRAS oncogene are detected in approximately 30–40 % of all patients with colorectal cancer [3].

Previous reports comparing the mutation status of KRAS between primary colorectal cancer and distant metastases reported that the percentage of concordance between primary tumors and metastases ranged between 68 and 100 % [4]. One reason for the discordance in KRAS status between primary tumors and metastases may be due to limitations in measurement due to poor quality DNA. Because formalin-fixed paraffin embedded (FFPE) samples are typically used in KRAS genotyping, poor preservation of DNA may result in fragmentation, preventing accurate genotyping of KRAS. Direct sequencing is the most frequently used method for KRAS genotyping; however, this process is technically complicated, and may easily lead to errors in measurement.

In this study we examined the KRAS status in 105 patients with colorectal cancer in both primary tumors and liver metastases by direct sequencing. Additional analysis was performed in patients displaying discordant KRAS status between primary tumors and metastases. To investigate the causes underlying this observation, the DNA quality of these tumors was assessed using a Bioanalyzer.

Materials and methods

Patients and samples

Paired samples of primary colorectal adenocarcinoma and corresponding liver metastases were obtained from 105 patients following surgical resection between 1995 and 2008 at the Department of Gastroenterology, Tokyo Women’s Medical University, Tokyo, Japan. All patients were Japanese and gave their written informed consent according to institutional regulations. Of these, sequencing data were successfully obtained from 90 patients (57 males and 33 females; median age 62.9 years (range 38–91). Synchronous liver metastases, defined as metastases diagnosed within a year from the primary tumor resection, were observed in 76 patients. Metachronous liver metastases, defined as metastases appearing >1 year after the primary tumor resection, were observed in 14 patients. The characteristics of the 90 patients are shown in Table 1.

Table 1 Demographic and clinical parameters of patients with metastatic colorectal cancer
Full size table

This study was approved by the institutional ethics committee and was performed in accordance with the Declaration of Helsinki.

Microdissection

FFPE tumor specimens were cut into serial sections with a thickness of 10 μm. Manual microdissection was performed using a scalpel if the histology was homogeneous and the tissue contained >90 % cancer cells. For all other samples, laser capture microdissection (P.A.L.M. Microlaser Technologies AG, Munich, Germany) was performed to ensure that only tumor cells were dissected.

KRAS mutation screening by direct sequencing

DNA was extracted from FFPE specimens using the QIAamp DNA FFPE Tissue Kit (Qiagen, Tokyo, Japan) according to the manufacturer’s instructions. DNA concentration and quality was assessed using the ND-1000 Spectrophotometer (NanoDrop Technologies, Wilmington, DE, USA). Polymerase chain reaction (PCR) was performed in a final volume of 50 μl containing 500 ng of DNA, 0.6 μl (6 pmol) each of forward and reverse primers, and 25 μl of Quick Tag HS DyeMix (Toyobo, Osaka, Japan). Primers spanned codons 12 and 13 of the KRAS gene, as previously described [5]. Primer sequences were as follows: forward, 5′-GAATGGTCCTGCACCAGTAA-3′, and reverse, 5′-GTGTGACATGTTCTAATATAGTCA-3′. The length of the amplified product was expected to be 217 bp. PCR was performed using the following cycling conditions: 94 °C for 3 min, 40 cycles of 94 °C for 30 s, 56 °C for 30 s, and 72 °C for 45 s, and 72 °C for 10 min.

PCR products were purified using the MinElute PCR Purification Kit (Qiagen) and used as a template for cycle sequencing using the Dye Terminator Cycle Sequencing (DTCS) Quick Start Kit (Beckman Coulter, Tokyo, Japan) according to the manufacturer’s instructions. Nested PCR primer sequences were: forward, 5′-GTCCTGCACCAGTAATATGC-3′, and reverse, 5′-ATGTTCTAATATAGTCACATTTTC-3′. A total of 4.4 μl stop solution (3 μl 3 M-NaOAc2, 0.4 μl 0.5 M Na2EDTA, 1 μl 20 mg/ml glycogen) and 60 μl 99.5 % (v/v) ethanol was added to sequencing reactions and samples were centrifuged at 14,000 rpm for 15 min at 4 °C. The supernatant was decanted and DNA was washed twice with 200 μl 70 % (v/v) ethanol, followed by centrifugation at 14,000 rpm for 2 min at 4 °C. DNA pellets were vacuum dried for 5 min and DNA was dissolved in 40 μl sample loading solution provided in the DTCS Quick Start Kit. Sequencing reactions were run on a CEQ-8800 Genetic Analyzer (Beckman Coulter). All sequencing reactions were performed in both forward and reverse directions. Direct sequencing was performed in duplicate for each sample.

KRAS mutation validation

To confirm the results obtained by direct sequencing as described above, we also assessed the mutational status of KRAS using a KRAS Mutation (codon 12, 13) Detection kit (Exciton ver.) (Dnaform, Tokyo, Japan). This kit is based on smart amplification process version 2 (SmartAmp2) technology [6–9]. It was used in accordance with the manufacturer’s instructions. This method was performed once for each sample, but when the first analysis showed ‘no amplification’, a second analysis was performed. If both the first and second analysis showed ‘no amplification’, the sample was determined as ‘not detected’. A third analysis was performed to investigate reproducibility if the second analysis showed successful amplification.

DNA quality analysis using Bioanalyzer

DNA quality was assessed using the 2100 Bioanalyzer (Agilent Technologies, Tokyo, Japan) and Agilent High Sensitivity DNA kit, according to manufacturer’s instructions.

Statistical analysis

Comparisons of the median length of DNA fragments and the median percentage of degraded DNA between the mismatched group and the control group were assessed using the Mann–Whitney’s U test. A P value of <0.05 was considered statistically significant. All values were two-sided.

Results

KRAS status was successfully measured in both primary tumors and liver metastases of 90/105 (85.7 %) patients examined in this study. The lack of available data for the remaining 15 patients was due to either low DNA yield (<500 ng), misalignment of the sense and antisense strands or low sequencing reproducibility.

Concordance of KRAS mutation status in primary colorectal adenocarcinoma and corresponding liver metastases

Of the 90 patients analyzed, 76 displayed synchronous metastases and 14 displayed metachronous metastases. In patients with synchronous metastases, KRAS mutation was observed in the primary tumors of 34 patients (44.7 %). Analysis of concordance between KRAS status in primary tumors and metastases was 88.2 %: 42 patients (55.3 %) had wild-type KRAS in both primary and liver, 25 patients (32.9 %) had mutations in both, three patients (3.9 %) had wild-type KRAS in primary and mutant KRAS in liver, and six patients (7.9 %) had mutant KRAS in primary and wild-type KRAS in liver. In patients with metachronous metastases, KRAS mutation was observed in the primary tumors of eight patients (57.1 %). The percentage of concordance was 100 % in these patients: six patients were wild-type in both primary colorectal adenocarcinoma and liver metastases, and eight patients were mutant in both primary tumors and liver (Table 2).

Table 2 KRAS status in primary tumor and liver metastases in synchronous and metachronous metastases
Full size table

Validation of KRAS status in patients with discordant KRAS status in primary tumors and metastases

Nine patients displayed discordance in KRAS status between primary tumors and liver metastases. To confirm the KRAS status of these patients using an independent technique, their DNA was re-evaluated using the KRAS Mutation Detection Kit (Dnaform). In primary tumors, samples from six of nine patients were successfully analyzed and all patients displayed wild-type KRAS. In metastatic samples, data were obtained for five out of nine samples, with three patients displaying wild-type and two patients displaying mutant KRAS status (Table 3). Thus, using this independent method, we show that at least three patients previously shown by sequencing to have discordant KRAS status in primary and metastatic tumors had concordant status using the KRAS mutation detection kit. Interestingly, however, we found that in primary tumors only two patients showed equivalent KRAS status comparing direct sequencing and the SmartAmp2 kit, whereas four patients showed different status. Similarly, analysis of liver metastases showed that KRAS status was concordant in three patients using these two methods, whereas two patients showed different statuses.

Table 3 Discrepancy of KRAS genotype comparing direct sequencing and the SmartAmp2 kit
Full size table

Background data of nine patients with mismatched KRAS status between primary tumors and metastases

To investigate whether specific factors may account for the discordance observed in the KRAS status of these nine patients, we examined background data for these patients compared with all patients. Age, gender, location of tumor, pathological type, and number of metastases were not differentially distributed between these nine mismatched KRAS status patients and all 90 patients (data not shown). The median year for collection of samples from these nine patients was 2004 (range 1998–2008), which was similar to collection times for the entire patient cohort (median year 2003; range 1995–2008). Similarly, the median concentration of DNA for these nine patients (including both primary and metastases; total 18 samples) was 230.96 ng/µl, which again was comparable to the median concentration of the entire 90 patients (233.93 ng/µl). Last, we compared the median UV 260/280 and 260/230 absorbance ratios in these patients, which are a measure of protein and organic contaminants, respectively, that may affect downstream applications. In the mismatch patients these values were 1.98 and 0.86, respectively, which were similar to the entire patient cohort (1.99 and 0.91, respectively).

DNA quality analysis using Bioanalyzer

Since no differences were observed when background data and DNA quality as assessed by NanoDrop were compared between the mismatched samples and all samples, we hypothesized that this discrepancy may be due to DNA degradation. To test this, we analyzed DNA quality using a Bioanalyzer. The Agilent 2100 Bioanalyzer and High Sensitivity DNA Kit are capable of analyzing the size and quantity of fragmented DNA and the quality control of DNA sequencing using only 100 pg/µl of DNA. Electrophoresis results are visualized as migration-time plots or computer-generated virtual gels. Figure 1 displays typical electropherograms of degraded and well-preserved DNA samples extracted from FFPE specimens. Table 4 indicates the size of fragmented DNA in the peak concentration. It also shows the analysis of DNA from seven patients (control group) randomly selected from the matched samples used in this study.

Fig. 1
figure 1

Electropherograms (left panel) and gel-image (right panel) displaying DNA quality assessment by the Agilent Bioanalyzer. The horizontal axis represents the migration time of DNA fragments in seconds, and the vertical axis represents fluorescence. The left peak represents a DNA ladder at 35 bp and the right peak a DNA ladder at 10,380 bp. a Examples of well-preserved DNA: the size of DNA samples in the peak concentration was 396 bp. The majority of the DNA fragments were >200 bp. b Examples of poorly preserved DNA: the majority of DNA fragments were <200 bp

Full size image
Table 4 Bioanalyzer data of mismatched group and control group
Full size table

The median length of DNA samples in the peak concentration in the mismatched group was significantly shorter than those in the control group (153.5 vs 276.5 bp, P = 0.0059). In the mismatched group, five out of 18 samples contained DNA that was significantly degraded (>50 % degraded fragments in the range of 0–200 bp). In contrast, zero out of 14 samples was degraded in the control group. The median value of the percentage of degraded DNA in each sample was 35.5 % in the mismatched group, which was significantly higher than the control group (22 %, P = 0.020). Because the PCR amplicon size generated using primers for the direct sequencing method was 217 bp, it is likely that the low concordance of results observed in the mismatched group was due to the low quality of the starting DNA template, resulting in insufficient PCR amplification and unreliable sequencing data.

Discussion

In this study, we performed a detailed investigation in patients displaying discordant KRAS status between the primary colorectal cancer and the corresponding liver metastases. Our initial prediction was that these results may be caused by low DNA yield or contamination of the DNA sample with other agents that may interfere with the PCR reaction. However, spectrophotometric analysis of these DNA specimens using a NanoDrop revealed that the yield and presence of protein or organic contaminants was similar between the mismatched samples and all patients. Instead, we demonstrated that there was significant DNA degradation in the primary and/or metastases specimens in the mismatched group. This DNA degradation was difficult to detect by NanoDrop, whereas the Bioanalyzer was very useful for this level of quality control.

Several factors may underlie the observed DNA degradation. First, DNA fragmentation may occur between the time of sample collection and time of analysis. All samples are paraffin-embedded, which is not ideal for DNA preservation and DNA fragmentation increases over time. In this study, patient samples in the mismatched group were not only collected at an older age, but were also distributed widely from 1998 to 2008. Second, the type of fixative used for DNA preservation can also affect DNA integrity. In most cases, formalin fixative is used; however, its cross-linking impairs DNA and RNA quality. Higher formalin concentration causes more damage to DNA quality. In addition, previous reports have shown that neutral-buffered formalin is a superior preservative for DNA compared with non-buffered formalin [10, 11]. In our institute, 10 % buffered formalin was used before 2000. Since 2001, however, 15 % non-buffered formalin was commonly used in the fixation of surgical specimens, which is not optimal for DNA preservation. In our study, we observed no difference in the success rate of KRAS screening before and after 2000 (data not shown). As previously mentioned, the mismatched samples were collected between 1998 and 2008, suggesting that the type of fixative used was not a significant factor affecting DNA integrity and subsequent KRAS genotyping in this study. Third, the sample fixation time is crucial. Basically, the shorter the duration of formalin fixation, the better DNA is preserved [10–13]. In our institute, fixation time was usually within 3 days. But there were no data for the length of formalin fixation for each of these nine samples, nor the time from surgical resection to formalin fixation. Thus, this factor might be one of the reasons for DNA degradation. In a previous study, DNA fragments ranging between 268 and 1327 bp were PCR-amplified from FFPE samples, demonstrating these samples performed well only for shorter length amplicons [13]. Thus, to increase the success rate of the KRAS screening test, it may be useful to design primers generating a smaller amplicon compared with our study, which amplified a 217 bp product.

In this study, an independent KRAS genotyping method, SmartAmp2, was used to validate the data obtained from direct sequencing. The basic principle of this method is that DNA amplification equals detection. Araki et al. [8] validated KRAS status with four different methods (direct sequencing, SmartAmp2, enzyme-enriched sequencing, and PNA-enriched sequencing) and reported that SmartAmp2 was the most sensitive and accurate method that can detect as little as 1 % of the mutant DNA with no non-specific amplification, which often reduces the sensitivity of sequencing by elevating the background noise. In our data, 18 samples from nine patients (primary and metastases) were analyzed with both direct sequencing and SmartAmp2 kit; seven out of 18 samples showed no amplification, and of 11 samples that were successfully amplified, six showed discordant KRAS status. From the Bioanalyzer data, these 18 samples had degraded low-quality DNA, which led to unsuccessful PCR amplification and/or poor sequencing with high background noise when analyzed by direct sequencing. We obtained reproducible data from the SmartAmp2 kit, which seemed to be more reliable than direct sequencing for these degraded samples. In a clinical setting, however, sensitive methods are not always ideal, because they are sometimes so sensitive that they may exclude the patients who are potentially sensitive to anti-EGFR antibody agents.

Recently, the KRAS mutation screening test has become essential in clinical practice for the treatment of unresectable colorectal cancer, since it is a strong negative biomarker for anti-EGFR antibody therapies [1, 2, 14]. Several KRAS genotyping methods are commonly employed; however, a standardized method has not yet been fixed. The most frequently used method is direct sequencing, which was also used in this study. A number of commercial kits are currently available for KRAS genotyping [6, 7, 15]. A study by Dequeker et al. [16] which aimed to identify variables that need to be assessed in a KRAS genotyping quality control scheme, involved sending 14 different colorectal cancer FFPE samples to independent laboratories in 13 different European countries for KRAS mutation screening. This study illustrated that only 10 of 13 experienced laboratories correctly identified the KRAS mutational status in all 14 cases, and KRAS genotyping methods and DNA extraction methods varied considerably between the laboratories. These results indicate a lack of consistency, even in laboratories that routinely perform KRAS genotyping, suggesting that quality control and quality analysis are essential, particularly as this result may affect patient prognosis and treatment strategy. Although direct sequencing is the most widely used method for KRAS genotyping, the result is affected by several factors including DNA quality and quantity, PCR conditions, primer design, and the condition of the sequencing instrument. While the use of kits is expensive, they may minimize the risk of variation between laboratories.

Previous literature has reported that the concordance rate of KRAS status in primary tumors and metastases ranges between 68 and 100 % [4, 17]. Knijn et al. [4] reported that the concordance rate of 892 patients from 18 previously published studies was 96.4 %. In our study, the success rate of KRAS diagnosis was 85.7 % (90/105), and the overall concordance rate between primary and metastases was 90 %, which seemed to be lower than previous reports. The reason might due to DNA degradation; the first 15 had highly degraded DNA that could not be amplified by the first PCR, and even among the 90 patients with successful amplification, some samples had degraded DNA that could not present reliable, reproducible data. In a separate analysis, examination of patients with synchronous and metachronous metastases revealed concordance rates of 81 and 100 %, which was not statistically significant. In previous studies, the discrepancy in KRAS status between primary tumors and metastases has been attributed to intratumoral heterogeneity [18–21]; however, it was also acknowledged this may be due to the incorrect genotyping caused by DNA degradation. Therefore, careful interpretation is required in these studies.

In conclusion, since KRAS genotyping has become an essential test for clinical practice, it is crucial to be aware that these results may be influenced by several factors, such as DNA quality and methodology. Thorough quality control and analysis and standardization of methods for DNA preservation, extraction and genotyping are critical for high quality, reliable results, or the restriction of laboratories for KRAS testing may be taken into consideration.