KRAS mutation testing in the treatment of metastatic colorectal cancer with anti-EGFR therapies

Survival of patients with metastatic CRC (mCRC) has improved steadily over the past several decades, due largely to the development of new combinations of standard chemotherapy, as well as to the introduction of new targeted therapies. Among the available targeted therapies are two monoclonal antibodies that target the epidermal growth factor receptor (EGFR) – cetuximab and panitumumab – which have demonstrated efficacy in the treatment of mCRC. These therapies are associated with a unique set of toxicities and costs, prompting the need for tools to select patients who are most likely to derive a benefit from them. Mutations in the KRAS oncogene have consistently been shown to predict non-response to cetuximab and panitumumab. The role of KRAS as a marker of efficacy of anti-EGFR therapies is reviewed.


InTRODuCTIOn
Colorectal cancer (CRC) affects more than 21,000 Canadians each year, and is the second leading cause of death from cancer among Canadian men and women 1 . Survival of patients with metastatic CRC (mCRC) has improved steadily over the past several decades, due largely to the development of new combinations of standard chemotherapy such as 5-fluorouracil, irinotecan, and oxaliplatin, as well as to the introduction of new targeted therapies. Among the available targeted therapies are two monoclonal antibodies that target the epidermal growth factor receptor (EGFR) -cetuximab and panitumumabwhich have clearly demonstrated efficacy in the treatment of mCRC. However, the introduction of these therapies has also introduced a unique set of toxicities and increased costs 2 , prompting the need for tools to select the patients who are most likely to benefit from these therapies.
There has recently been heightened interest in the relevance of several biomarkers for the selection of patients who will benefit from EGFR-targeted therapies for the treatment of CRC and other EGFRassociated cancers. In particular, mutations in the KRAS oncogene have consistently been shown to predict nonresponse to cetuximab and panitumumab. This marker is of particular importance, given the prevalence of KRAS mutations among patients with CRC; up to half of patients with CRC are found to have the mutant version of the gene 3-6 .

KRAS and the EGFR pathway
KRAS is a signal transducer downstream of tyrosine kinase receptors including EGFR -a complex signaling cascade involved in the development and progression of cancer. The EGFR pathway is activated by the binding of the cell-surface EGFR/HER family receptors to their ligands, such as transforming growth factor alpha (TGF-α) and EGF. This leads to activation of genes that regulate cell cycle progression, tumor cell survival, metastases and angiogenesis (Fig. 1). Monoclonal antibodies against EGFR, such as cetuximab and panitumumab, block the receptor signaling and its downstream events, including those mediated by KRAS.
Upon stimulation of the EGFR, wild-type KRAS is active for a short period and the signaling activities to the downstream RAF/mitogen-activated protein kinase (MAPK)/extracellular signal-related kinase (ERK) pathway are tightly controlled. Mutated KRAS protein becomes constitutively activated, thereby making the cascade independent of upstream signaling by tyrosine kinase receptors such as the EGFR. Therefore, blocking of EGFR with cetuximab or KRAS mutation testing in the treatment of metastatic colorectal cancer with anti-EGFR therapies D. Soulières MD*, W. Greer PhD, FCCMG † panitumumab may not affect downstream events. Mutations within the KRAS gene resulting in constitutive protein activity are found in approximately 30% to 50% of all CRCs 3-6 .

The role of KRAS and BRAF as markers of
efficacy of the anti-EGFR therapy

KRAS
As reviewed by Fakih and Wong in this supplement, the efficacy of the anti-EGFR antibodies cetuximab and panitumumab in the treatment of mCRC has consistently been shown to rely on the KRAS status of the tumor (Tables 1 and 2). Post hoc analyses of both randomized and single-arm trials of cetuximab or panitumumab have demonstrated that these monoclonal antibodies are only effective against tumors with wild-type KRAS, while patients with KRAS mutations in codon 12 or 13 do not derive any benefit from these treatments 3,4,6-11 . The first study to provide conclusive data showing the relationship between KRAS status and the efficacy of the monoclonal anti-EGFR antibody panitumumab was Amado's analysis of tumors from 427 mCRC patients who were randomly assigned to treatment with panitumumab or best supportive care (BSC) 6 3 . Response rate and PFS were both significantly improved in patients treated with cetuximab; however, these benefits were limited to those with wild-type KRAS tumors, and patients with mu-  tated KRAS receiving cetuximab demonstrated poorer outcomes than those receiving FOLFOX alone. The phase III CO.17 trial, conducted by the National Cancer Institute of Canada Clinical Trials Group (NCIC CTG) in collaboration with the Australasian Gastro-Intestinal Trials Group (AGITG), examined the effect of cetuximab on survival among patients with advanced CRC in whom all chemotherapy had failed and for whom no other standard anticancer therapy was available 12 . Although cetuximab used alone in the third-line setting improved overall survival and PFS and preserved quality of life to a better degree than BSC alone, resistance to cetuximab was high, with more than half of the cetuximab-treated patients showing progression at the first assessment of disease response. With the accumulating evidence demonstrat-SOULIèRES et al.

BRAF
In the absence of KRAS mutations, resistance to anti-EGFR therapies may occur as a result of mutations in other signaling molecules in the RAS-RAF-MAPK pathway. Recent retrospective analyses of mCRC patients treated with cetuximab or panitumumab have shown that BRAF mutations, which are exclusive from KRAS mutations, occur in approximately14% of patients and are also associated with a lack of response to anti-EGFR therapy 13,14 . BRAF mutations have also been associated with significantly shorter PFS 13,14 and overall survival 13 in patients with mCRC.

Guidelines for testing
In Canada, panitumumab is currently restricted to the treatment of EGFR-expressing mCRC with nonmutated (wild-type) KRAS 15 , while KRAS status is not specified in the indication for cetuximab 16 . However, with the knowledge that patients with mCRC who harbour KRAS mutations do not derive any benefit from treatment with EGFR-targeting monoclonal antibodies in the first-, second-, or third-line settings, Cancer Care Ontario (CCO) recommends the two clinically available EGFR inhibitors, cetuximab and panitumumab, be used "for the treatment of patients with advanced CRC after failure of standard chemotherapy and whose tumors have tested negative for KRAS gene mutations" 17 . CCO has also recently approved the use of cetuximab in combination with irinotecan for the third-line treatment of mCRC only in the presence of tumors with the non-mutated KRAS oncogene.
The American Society of Clinical Oncology (ASCO) recently also released a provisional clinical opinion recommending that patients with mCRC who are candidates for treatment with cetuximab or panitumumab undergo tumor testing for KRAS mutations in a CLIA-accredited laboratory 18 . The updated National Comprehensive Cancer Network (NCCN) clinical practice guidelines for colon cancer and rectal cancer also recommend testing for KRAS gene mutations, stipulating that only patients with wild-type KRAS genes should receive treatment with cetuximab or panitumumab 19,20 .
However, BRAF testing is currently not a requirement for treatment with an EGFR inhibitor.

How to test?
By screening patients with mCRC for KRAS tumor status prior to initiating treatment with an anti-EGFR monoclonal antibody, unnecessary toxicity and costs can be avoided for patients who are unlikely to respond. However, there are logistical challenges in testing tumors from mCRC patients, as well as questions surrounding specimen selection, and selection of the appropriate assay.

Specimen selection
The most readily available clinical specimens for mutational analysis are typically formalin-fixed, paraffin-embedded (FFPE) tissue blocks. Until recently, formalin-fixed samples were considered to be of low quality and yield for DNA testing 21 ; however, improvements in techniques have enhanced the ability to use DNA from FFPE tissue 22 . Because the fixation process damages DNA and can potentially introduce artificial mutations in conventional PCR processes, sufficient cellular material is necessary for analysis 23 . DNA from surrounding reactive cells, such as fibroblasts, leukocytes, or endothelial cells, can also potentially compete with mutant DNA in amplification reactions and introduce errors in testing. Tumor cell enrichment by micro-or macro-dissection or selective sampling of the paraffin block by needle core may increase the sensitivity of mutation testing, but care must be taken with these procedures to ensure that sufficient DNA is available for amplification 22 and contamination of normal DNA with mutant DNA does not occur.
Based on current knowledge, the most appropriate specimens for KRAS mutation testing may be obtained from the primary tumor 24 . However, an estimated 20% of patients will present with metastatic disease and, therefore, lack tissues from the primary tumor 24 . In these patients, KRAS testing may be performed using material from the metastatic tumor 24 . Data comparing KRAS status in primary and metastatic tumors are limited and have had inconsistent results [25][26][27][28] . However, in a recent Italian study of 48 patients with mCRC, DNA sequencing revealed an overall concordance of KRAS mutational status between primary tumors and metastases in 92% of patients, suggesting that evaluation of KRAS status can be performed in either the primary tumor or metastatic sites 29 .

Assay selection
Several mutation detection procedures have been described, all of which are based on the polymerase chain reaction (PCR) ( Table 3) 22 . Generally, selection of the best technique for use is based on the sample size, desired level of sensitivity and DNA quality and quantity available.

Direct sequencing of PCR products
Direct sequencing of PCR products detects all mutations in amplified DNA sequences, and this is SOULIèRES et al.  30 . However, this method requires mutant copies to have a minimum concentration of 20% to 50% that of any accompanying wild-type sequences, and will therefore miss mutations that may be present at a lower level 21,31 . Because of the low sensitivity and the expense of direct sequencing, more sensitive and specific assays have been developed to assess KRAS in clinical samples, including methods that employ restriction fragmentation length polymorphism (RFLP), allele-specific oligonucleotide (ASO) hybridization, high resolution melting analysis (HRMA), and amplification refractory mutation systems (ARMS).

RFLP
Whereas gene sequencing compares the sequence of the sample gene with the normal sequence of the gene, nucleotide by nucleotide, RFLP methods detect differences between mutant and wild-type DNA by their susceptibility to digestion by restriction enzymes. Restriction enzymes can be selected to recognize a defined sequence which is present only in the mutated or non-mutated DNA. Knowing what specific size the digested fragment should be in mutated vs. non-mutated DNA allows one to identify if a mutation is present or not. These amplified mutant copies can then be detected by gel or capillary electrophoresis 32 . While highly sensitive, this method is also very complex, requiring tight control of PCR and digestion conditions to avoid replication errors and artificial mutations. Furthermore, when a mutation is detected by this methodology the specific nucleotide change cannot be identified. If the specific mutation identity is required (this is not currently necessary for clinical utility), then direct sequencing can be used after mutation identification by RFLP.

Allele-specific oligonucleotide hybridization
Short segments of synthetically produced DNA (oligonucleotides) can be used to detect mutations in a gene segment. The oligonucleotides are complementary to a corresponding segment of the gene under investigation, hybridizing completely with the wildtype sequence or to one of the possible mutations. A single base mismatch caused by a mutation reduces the melting point temperature of the double-stranded hybrid 33 . The difference in melting points between matched and mismatched sequences can be used to detect single-base mismatches between wild-type and mutant sequences 34 .
Finding rare mutant alleles in a DNA mixture can be challenging however, particularly in samples containing high levels of normal alleles. Therefore, if a sample has a low tumor burden, this may not be the best approach. This technique is also expensive, requiring specialized equipment and software for analysis 35 .

High resolution melting analysis
The presence of a mutation disrupts the affinity of two DNA chains, causing them to bind with less energy and become more easily separated by heat. High resolution melting analysis (HRMA) is performed following PCR, and measures differences in melting point temperatures between matched and mismatched double stranded DNA, caused by polymorphisms or somatic mutations 35 . HRMA has a high sensitivity, and is also inexpensive and fast. However, because any DNA alteration can produce an abnormal melting point curve, abnormal curves need to be confirmed by sequencing, which increases turnaround time and expense, reducing HRMA's advantage over direct sequencing alone. HRMA may therefore be useful for rapid screening; however the need for confirmation by sequencing may limit its utility in the clinical setting.

Amplification refractory mutation system
The amplification refractory mutation system (ARMS) -also known as allele-specific PCR or PCR amplification of specific alleles -utilizes a PCR primer which is designed to discriminate among templates that differ by a single nucleotide residue. The ARMS primer can be designed to amplify a specific allele of a multi-allelic system while remaining refractory to amplification of another allele that may differ by as little as a single base. ARMS is able to detect directly the presence of KRAS mutations in heterogeneous specimens at a low allelic concentration (1%) without the need for confirmation by direct sequencing. A drawback of ARMS is that it is only able to detect known mutations; separate reactions are required for each mutation, thus requiring more DNA material. However, because the amplification step and the diagnostic steps are combined, this may prove to be a time-efficient and practical method for routine diagnosis of KRAS mutations 22 .

Standardization of KRAS testing in colorectal cancer
Because of the potential for variability among the different testing methods, a thorough analytical validation of testing methods, together with a high standard of quality assurance are critical for accurate, reliable testing of KRAS mutations in clinical practice 24 . In Canada, Health Canada has approved the use of the TheraScreen K-RAS testing kit (DxS, Manchester, United Kingdom), which combines ARMS with a real time PCR technology. In the United States, there is currently no FDA-approved test for KRAS testing, and testing can be performed using laboratory-developed tests, provided that the laboratory is accredited by the College of American Pathologists (CAP) and the test has been appropriately validated. 36 To date, one published study has evaluated the concordance between different methods for KRAS mutation testing 37 . Four commercially available assays were used to assess seven common mutations of the KRAS gene in codons 12 and 13 in 40 colorectal tumor samples, with direct sequencing used as a reference. Two of the allelespecific PCR-based methods and one PCR/direct sequencing method demonstrated high to good agreement with direct sequencing, whereas an oligonucleotide hybridization method showed poor agreement.

RESulTS
The authors of the present article conducted a small study involving six Canadian laboratories to compare the accuracy and sensitivity of three methods of KRAS mutation analysis -the TheraScreen K-RAS testing kit only (1 laboratory), the TheraScreen K-RAS testing kit in combination with direct sequencing (2 laboratories), the TheraScreen K-RAS testing kit in combination with direct sequencing and RFLP (1 laboratory), and RFLP plus sequencing (2 laboratories). In the first phase of the study, 10 DNA samples were extracted from seven KRAS mutant (positive) cell lines containing approximately 50-100% mutant cells. In the second phase, dilutions were created from each of the seven positive cell lines (approximately 10-40% mutant cells). To assess the ability of the laboratory to extract DNA from paraffin and the resulting specificity, accuracy and sensitivity of KRAS mutation testing on such samples, 8-10 samples were extracted from paraffin blocks for the third phase of the study. For each phase, KRAS-negative cell lines were used for comparison. All of the labs were able to detect KRAS mutations in samples derived from cell lines containing 50-100% mutant KRAS cells as well as from diluted cells lines containing 10-40% mutant KRAS cells. However, two of the labs experienced some difficulty interpreting two samples from the diluted cell lines when using sequencing methodology; this is probably due to the limits of sensitivity of sequencing.
Concordant results were achieved with five of the eight samples extracted from paraffin blocks. Inconsistent results with RFLP plus sequencing were seen in one lab, which was later discovered to be due to a mix-up of the samples (Lab 5, samples 3 and 4, Table 4). Discordant results were reported for three of the eight samples. In sample four, results were not concordant as the sample had a low level KRAS mutation requiring a very sensitive assay, prompting the question of what the sensitivity cutoff of an assay should be. In sample six, two labs reported inconclusive results using the TheraScreen test, suggesting that labs reporting inconclusive results with this test should reconsider their delta-Ct cutoff criteria, optimize their assay, or use another method to verify the results. In sample eight, most of the labs had some difficulty in interpreting the mutation status due to the limited tumor area on the slides and a non-formalin based fixation method, resulting in low DNA yield and poor DNA quality. The labs that participated are all well-experienced in performing complex genetic analyses on various sample types. These results thus point to some of the challenges of KRAS testing in poor quality samples.

COnCluSIOnS
In all clinical trials, anti-EGFR therapies have been consistently ineffective in mCRC patients with KRAS SOULIèRES et al. Testing techniques need to be standardized and validated externally as well as internally. Cell line materials provided the most accurate results, while paraffin-embedded tissue may be somewhat more problematic, especially if suboptimal.
The role of the pathologist is very important in KRAS testing. The pathologist is responsible for choosing the most appropriate tissue block to be tested, evaluating the tumor content of the tissue block, and ensuring that it is adequate (tissue size, degree of tumor involvement) by assessing the H&Estained section of the tissue area and marking the area with adequate tumor density -preferably >70% carcinoma cells. Testing should be performed by an accredited and licensed testing lab that conforms to quality guidelines for KRAS testing 24 and routinely participates in proficiency testing (e.g. the College of American Pathologists [CAP]).
We propose an algorithmic approach to KRAS testing, where laboratory professionals (pathologists, geneticists) have access to multiple methods wherever possible, especially when assessing suboptimal material. We have found that for small samples with degraded DNA, Sanger sequencing is often still the best method for mutation detection.
Regardless of the testing method used to determine KRAS status in patients with mCRC, the goal is for a sensitive and specific technique that has been standardized and validated externally and internally. As noted above, CAP currently has a proficiency challenge available for labs so they can assess their ability to test for KRAS mutations.
The anti-EGFR monoclonal antibody therapies are currently approved for the treatment of mCRC in the third-line setting. However, it may be several weeks before results of KRAS testing are obtained, which can be a long wait for a patient with advanced disease who requires treatment. Depending on the availability of funding, it would be optimal for KRAS testing to begin immediately following a diagnosis of metastatic disease; currently however, testing can only be undertaken when the Oncologist is considering third-line therapy.

ACKnOWlEDGEMEnTS
The authors acknowledge the support of Bristol-Myers Squibb Inc. in providing funding for third-party medical writing. The authors also acknowledge the medical writing contributions of Maryka Hladki of Science & Medicine Canada.