Effect of KRAS codon13 mutations in patients with advanced colorectal cancer (advanced CRC) under oxaliplatin containing chemotherapy. Results from a translational study of the AIO colorectal study group

Background To evaluate the value of KRAS codon 13 mutations in patients with advanced colorectal cancer (advanced CRC) treated with oxaliplatin and fluoropyrimidines. Methods Tumor specimens from 201 patients with advanced CRC from a randomized, phase III trial comparing oxaliplatin/5-FU vs. oxaliplatin/capecitabine were retrospectively analyzed for KRAS mutations. Mutation data were correlated to response data (Overall response rate, ORR), progression-free survival (PFS) and overall survival (OS). Results 201 patients were analysed for KRAS mutation (61.2% males; mean age 64.2 ± 8.6 years). KRAS mutations were identified in 36.3% of tumors (28.8% in codon 12, 7.4% in codon 13). The ORR in codon 13 patients compared to codon 12 and wild type patients was significantly lower (p = 0.008). There was a tendency for a better overall survival in KRAS wild type patients compared to mutants (p = 0.085). PFS in all patients was not different in the three KRAS genetic groups (p = 0.72). However, we found a marked difference in PFS between patients with codon 12 and 13 mutant tumors treated with infusional 5-FU versus capecitabine based regimens. Conclusions Our data suggest that the type of KRAS mutation may be of clinical relevance under oxaliplatin combination chemotherapies without the addition of monoclonal antibodies in particular when overall response rates are important. Trial registration number 2002-04-017


Background
The oncogene KRAS belongs to the protein family of small G-proteins and is mutated in 35-40% of colorectal cancers (CRC) [1]. RAS mutations are considered early events in colon carcinogenesis and are well conserved between primary tumor and corresponding metastases [2]. KRAS mutations tested routinely include six mutations in codon 12 and one mutation in codon 13.
KRAS has been studied extensively as a prognostic marker in CRC, but results are still conflicting. Overall, there seems a tendency towards inferior outcome for patients with KRAS mutant tumors even in large randomized trials [3][4][5][6][7]. Importantly, KRAS has recently been identified as a strong predictive marker for patients with advanced CRC under anti-EGFR-treatment. Various studies demonstrated that while patients with tumors expressing wild type KRAS may benefit from anti-EGFR (epidermal growth factor receptor)-antibody treatment, patients carrying a mutated KRAS gene do not [5,[8][9][10]. In contrast, two large trials using oxaliplatin-based chemotherapy-backbones did not confirm KRAS wild type to be a powerful predictor of treatment efficacy in metastatic CRC [4,11]. However, the studies published until recently did not differentiate between the different KRAS mutations and other histopathological or molecular features of their patients. The clinical observation, that some patients with KRAS mutation may respond to anti-EGFR-antibody therapy, as well as experimental data, that the biological effects of KRAS mutations may differ, was addressed recently by de Roock and Tejpar [12,13]. They reported that patients with codon 13 mutations in KRAS exhibit a worse overall prognosis with short overall survival times under standard chemotherapy, but may benefit from anti-EGFR-antibody therapy similar to wild type patients. These observations prompted us to look for the effect of the KRAS mutational status in correlation with response and survival data in patients with advanced colorectal cancer receiving oxaliplatin containing chemotherapy from a prospective randomized multicenter phase III trial of the German AIO study group.

Patients
All patients participated in a prospective randomized phase III first-line palliative chemotherapy trial of advanced CRC of the AIO colorectal study group (Arbeitsgemeinschaft Internistische Onkologie of the German Cancer Society). The study was performed according to the Helsinki declaration and it was approved by the ethics review board of the Central Hospital Bremen and of the Medical Faculty of the Ruhr-University Bochum. The study design, patient characteristics, treatment plans and results of the clinical trial have been reported previously [14]. Briefly, a total of 474 patients were randomized to be treated with either 5-FU/folinic acid (FA) and oxaliplatin (FUFOX: oxaliplatin, 50 mg/m 2 ; FA, 500 mg/m 2 ; continuous 5-FU, 2,000 mg/m 2 /22 h; on day 1, 8, 15, 22; q day 36) or capecitabine and oxaliplatin (CAPOX: oxaliplatin, 70 mg/m 2 on day 1 and 8; capecitabine, 2 × 1,000 mg/m 2 /day consecutively for 2 weeks, q day 22). No clinical factor was found to be predictive for definition of a subgroup of patients benefiting more or less from each fluoropyrimidine backbone. We here present data on a subcohort of 201 patients (42.4%) with available formalin-fixed paraffin-embedded tissue. Samples were retrieved from pathologists responsible for first diagnosis, pseudononymized and forwarded to the Institute of Pathology of the Ruhr-University Bochum, which was blinded to treatment allocation and prognostic outcomes.

DNA extraction and mutation analysis
DNA was extracted from anonymized formalin-fixed paraffin embedded tissue samples. For each patient, five 10-μm sections were prepared. An additional representative 1-μm section was deparaffinized, stained with H&E, and analyzed for detailed morphology. Regions displaying tumor cellularity of >70% were marked and macrodissected. Tissue was extracted using QIAmp DNA Mini kit (Qiagen, Hilden, Germany). Real-Time PCR amplification for the most frequent seven KRAS mutations was performed using commercially available kits from DxS Ltd. (Manchester, UK) according to the manufacturer's instructions. This kit detects >95% of known KRAS mutations. Laboratory staff was blinded to the patient data and clinical outcome. BRAF mutation analysis to detect the V600E substitution was performed by RT-PCR (see supplemental file for further details).

Statistical analysis
Chi-square test was used to evaluate the associations between KRAS status and other dichotomous variables. The analysis of progression-free survival (PFS) and overall survival (OS) was done using the Kaplan-Meier method and differences between subgroups were calculated by log-rank test. Data have been analyzed using SPSS 18.0 (Munich, Germany). All tests were two sided. For all tests, p values <0.05 were considered significant.

Correlation between mutations and response rate
Tumor response evaluation was available for 201 patients. Grouping all KRAS mutations together, mutated tumors were associated with a significantly lower response rate (RR; defined as partial or complete remission by RECIST) as compared to tumors without KRAS mutations (44.4% vs. 63.0%, p = 0.012). When patients with codon 13 mutated tumors were analysed separately the overall response rate in this cohort was 23% as compared to 49% in codon 12 mutated tumors and 63% in wild type tumors ( Table 3, p = 0.008). Disease control rates (DCR) were 77%, 81% and 88%, respectively, which was not statistically significant ( Table 3, p = 0.29).

Correlation between mutations and progression-free survival
During follow-up, 170 of 201 evaluable patients had progressed. The median PFS in all patients of the KRAS subcohort was not statistically different in relation to the KRAS mutational status (wild type: 7.5 months, mutation codon 12: 8.2 months, mutation codon 13: 10.0 months; p = 0.71) ( Figure 1). However, when analysing the two treatment arms separately, we found a substantial, non-significant, difference in PFS in codon 13 mutated tumors versus codon 12 mutated tumors. While median PFS was as low as 6.1 months for codon 13 patients receiving infusional 5-FU, median PFS was 13.3 months in patients treated with capecitabine (HR: 2.52, p = 0.22). Patients with codon 12 mutations showed a trend towards the opposite effect: median PFS was 7.0 months under CAPOX therapy while median PFS was 9.9 months under FUFOX (HR: 0.62, p = 0.12) ( Table 4).

Correlation between mutations and overall survival
During follow-up 135 of 201 evaluated patients had died. We observed a trend towards a better survival time in patients with wild type tumors compared to those with a mutation of KRAS. The overall survival of wild type KRAS patients was 19.2 months, for patients with codon 12 mutations 15.6 months and for patients with codon 13 mutations 16.5 months. These differences were of marginal significance (p = 0.085) (Figure 2).
Evaluating the different treatment arms separately, we found comparable survival times without significant differences (Table 5). Since the survival curves seemed to separate for wild type KRAS and mutant KRAS patients at 14 months we hypothesised that this difference was most likely caused by the influence of post-study treatment. 71 out of 128 (55.4%) wild type patients received further lines of therapy while 32 out of 128 wild type patients (25%) received cetuximab. Out of 73 patients with KRAS mutations 42 were treated in further lines (57.5%) with 10 patients receiving cetuximab (13.7%).

Discussion and conclusions
We assessed the prognostic value of KRAS codon 12 and codon 13 mutations in tumor tissue from patients with advanced CRC recruited into a phase III clinical trial using CAPOX or FUFOX treatment regimens. This is the first randomized phase III trial retrospectively investigating the role of codon 13 mutations in advanced CRC patients treated with oxaliplatin combination chemotherapy only, without the addition of a monoclonal antibody. While the overall response rate in codon 13 patients was significantly lower, PFS was not different in the three KRAS mutational groups. Interestingly, we found a substantial difference in PFS between patients with codon 12 and 13 mutant tumors when looking at infusional 5-FU versus capecitabine based regimens. Patients with codon 13 mutations seem to benefit more in terms of PFS from the oral capecitabine based protocols. Moreover, there was a strong trend towards better overall survival in patients with wild type KRAS compared to all mutant KRAS patients. Lastly, when analysing OS in patients who received second and further line therapy we found that KRAS wild type and KRAS mutant patients alike showed a significantly higher OS post progression when treated with cetuximab.
A number of studies have looked at the potential prognostic or predictive value of KRAS mutations on response rates and survival in patients with CRC and several studies found a negative impact of KRAS mutations on prognosis [3,7,[15][16][17].
Recently, a number of randomized trials have included translational research programs to evaluate certain target genes and their role as prognostic or predictive markers in patients with metastatic disease. Thereby, the mutational status of KRAS has now been established as a strong predictive marker of resistance to anti-EGFRantibody treatment [5,8,10], although some trials could not fully confirm these results [4,11]. We still do not exactly know whether KRAS mutations influence the response to other treatment regimens such as standard chemotherapy or bevacizumab combinations. While bevacizumab efficacy seems independent from the KRAS status [18], the activity of certain chemotherapeutic agents may be influenced by KRAS mutations. There, patients seem to do worse under oxaliplatin combinations when carrying a mutant KRAS gene within their primary cancer [4,5,7,16,19,20]. For example, in the recently reported COIN study patients under oxaliplatin/ 5-FU combinations showed a median PFS of 8.6 months in the wild type KRAS cohort while median PFS was only 6.9 months in the KRAS mutant cohort [4]. Similarily, ORR was lower in KRAS mutant patients receiving chemotherapy with oxaliplatin only (41% vs. 50%). Furthermore, another recent small study evaluated the KRAS status in 66 patients receiving a second line chemotherapy with oxaliplatin and infusional 5-FU  refractory to 5-FU/irinotecan based chemotherapy [16]. This study found a significantly lower response rate (7% vs. 27%, p = 0.026) and significantly shorter median PFS (3.1 vs. 5.2 months, p = 0.007) for patients with mutant KRAS tumors compared to patients with wild type KRAS tumors under oxaliplatin containing therapy. Very recently, experimental and some clinical reports suggested that not all KRAS mutations behave alike [12,13,21]. In fact, there is evidence that patients carrying mutations in codon 13 of the KRAS oncogene which is found in about 8% of patients with advanced CRC have a substantially worse overall prognosis but may, on the other hand, benefit from anti-EGFR-treatment.
In our study we found a similar rate of codon 13 mutations as described before. The overall response rate in patients with codon 13 mutations in our analysis was as low as 23%, significantly lower than in patients with wild type or codon 12 mutations. The CRYSTAL-and the OPUS-studies alike found low response rates in patients with codon 13 mutations treated with combination chemotherapy only (17% for irinotecan combinations and 33% for oxaliplatin combinations) [13]. In contrast to previous reports, ORR was substantially higher in this codon 13 mutant patient cohort when cetuximab was added.
Our study was conducted using combination chemotherapy with oxaliplatin in first line treatment without the addition of monoclonal antibodies. Interestingly, the PFS and OS in our codon 13 cohort compared to the other mutated groups and other previously published works was rather long with a PFS of 10 months and an OS of 16.5 months [13]. We do not know why PFS and OS of the codon 13 cohort in our study was prolonged when response rates were as low as 23%. Either low patient numbers or a yet unknown functional mechanism of codon 13 mutated KRAS proteins within colon cancers may be responsible for the observation.
The present analysis also suggests that there may be an interaction between the type of KRAS mutation and the mode of application of 5-FU, i.e. whether administered intravenously or orally as capecitabine. In particular, patients with codon 13 mutations showed longer median PFS intervals when receiving capecitabine compared to infusional 5-FU. Although the overall efficacy of infusional 5-FU and capecitabine in advanced CRC has been found to be comparable [22] there may be  some patient subpopulations or treatment regimen where the mode of application of the fluoropyrimidine is more crucial. For example, anti-EGFR antibodies in wild type KRAS patients may only be active when infusional 5-FU regimens are used, but not when capecitabine based protocols are applied [4]. The potential resistance of KRAS mutated tumors to oxaliplatin containing regimens seems interesting and may reveal more general mechanisms of drug resistance in cancers. Oxaliplatin belongs to the platinum containing compounds like cisplatinum and carboplatinum. Metabolites of platinum compounds interact with DNA and form crosslinks. In addition, platinum-DNA-adducts strongly inhibit DNA polymerases and therefore act antineoplastic. Some authors have studied in cell culture systems and preclinical models the influence of oncogenic RAS mutations on the activity of platinum compounds and found that the nucleotide excision repair protein ERCC-1 may be upregulated through activated RAS. ERCC-1 may subsequently activate DNA repair capacity and thus mediate platinum resistance. A recent study evaluated the role of ERCC-1 mRNA levels in 191 patients treated with FOLFOX within the CON-FIRM1 and CONFIRM2 studies. Low ERCC-1 gene expression was correlated with higher response rates to FOLFOX chemotherapy and better overall survival. In contrast patients with high ERCC-1 did not have benefit from FOLFOX chemotherapy [23]. Of note, we have examined the expression of ERCC-1 by immunohistochemistry in our cohort and found no obvious correlation with KRAS status, response rates or survival and ERCC-1 expression (data not shown). The analysis of the ERCC-1 expression levels by RT-PCR as reported by other studies has not been performed so far [24].
Median overall survival after first progression correlated with cetuximab-treatment in patients bearing KRAS wild type and mutant tumors alike. The role of codon 13 mutation in this setting appears minor since there were only two patients with a codon 13 mutated tumor in the cohort with KRAS mutations of whom we know about cetuximab application. However, there are at least two limitations to our analysis. First, numbers are low in particular in the cetuximab group and secondly addition of cetuximab was not randomized for. There seems to be a selection of patients with good performance status and good prognosis who received third line therapy compared to those who did not. Therefore,  we can not draw definite conclusions from our analysis whether anti-EGFR antibodies are effective in patients with KRAS G13D mutations or not. We presently can not draw final conclusions regarding patient management from this study. It remains unclear whether chemotherapy backbones with irinotecan are less prone to interactions with mutated KRAS because data regarding this issue are conflicting. Independent validation of the findings is essential. Therefore, the standardized, thorough and comprehensive collection of tissue and blood samples of all trial patients within independent cancer tissue banks should be a major goal of modern clinical cancer trials.