The impact of KRAS mutations on prognosis in surgically resected colorectal cancer patients with liver and lung metastases: a retrospective analysis

KRAS mutations are common in colorectal cancer (CRC). The role of KRAS mutation status as a prognostic factor remains controversial, and most large population-based cohorts usually consist of patients with non-metastatic CRC. We evaluated the impact of KRAS mutations on the time to recurrence (TTR) and overall survival (OS) in patients with metastatic CRC who underwent curative surgery with perioperative chemotherapy. Patients who underwent curative resection for primary and synchronous metastases were retrospectively collected in a single institution during a 6 year period between January 2008 and June 2014. Patients with positive surgical margins, those with known BRAF mutation, or those with an unknown KRAS mutation status were excluded, and a total of 82 cases were identified. The pathological and clinical features were evaluated. Patients’ outcome with KRAS mutation status for TTR and OS were investigated by univariate and multivariate analysis. KRAS mutations were identified in 37.8 % of the patients and not associated with TTR or OS between KRAS wild type and KRAS mutation cohorts (log-rank p = 0.425 for TTR; log-rank p = 0.137 for OS). When patients were further subdivided into three groups according to mutation subtype (wild-type vs. KRAS codon 12 mutation vs. KRAS codon 13 mutation) or amino acid missense mutation type (G > A vs. G > T vs. G > C), there were no significant differences in TTR or OS. Mutational frequencies were significantly higher in patients with lung metastases compared with those with liver and ovary/bladder metastases (p = 0.039), however, KRAS mutation status was not associated with an increased risk of relapsed in the lung. KRAS mutation was not associated with TTR or OS in patients with metastatic CRC who underwent curative surgery with perioperative chemotherapy.


Background
Colorectal cancer (CRC) is the fourth leading cause of cancer-related death worldwide [1]. Although the development of molecular-targeted therapy has improved the survival of patients with metastatic CRC [2,3], the majority of patients with stage IV CRC who undergo complete resection die from metastatic disease. Nevertheless, a good proportion of patients demonstrate good recurrence-free survival. CRC tumorigenesis is characterized by the accumulation of genetic alterations, and V-Ki-ras2 Kirsten rat sarcoma viral oncogene homolog (KRAS) mutations are an early event in tumorigenesis [4]. KRAS mutations occur in approximately 30 to 40 % of patients with CRC, and 90 % of KRAS mutations occur in codon 12 or 13 [2,5,6]. KRAS mutations lead to constitutive activation of downstream pathways, including the Ras/Raf/MAP/MEK/ERK and/or PTEN/ PI3K/Akt pathways [7][8][9][10]. KRAS mutations are established biomarkers for predicting the poor efficacy of anti-epidermal growth factor receptor (EGFR) monoclonal antibodies in patients with stage IV CRC [2,5,11], but the prognostic relevance of KRAS mutations remains controversial [12][13][14][15][16]. Recent studies, in patients with resected stage II and/or III CRC, have highlighted the prognostic value of KRAS codon12 and 13 mutations, showing correlations between mutation subtype, cancer recurrence, and poor overall survival [13][14][15].
Large population-based cohorts usually consist of patients with non-metastatic CRC [12,14,16,17]. The prognostic impact of KRAS mutation in patients with synchronous metastatic CRC who undergo curative resection with perioperative chemotherapy is unknown. The current study investigated the impact of KRAS mutations on the time to recurrence (TTR) and overall survival (OS) in patients with stage IV CRC who underwent curative surgery with perioperative chemotherapy. In addition, the recurrence pattern according to KRAS mutation status after complete resection was evaluated.

Patients
In this retrospective study, patients who underwent curative resection for primary and synchronous metastases at our institution between January 2008 and June 2014 were identified from the hospital records. Patients who underwent separate colorectal resection and metastasectomy were excluded if the duration between the two procedures exceeded 2 months. Patients with positive surgical margins, those with known v-Raf murine sarcoma viral oncogene homolog B (BRAF) mutations, or those with an unknown KRAS mutation status were also excluded. All patients included in the study were administered 5-FU with/without oxaliplatin or irinotecanbased chemotherapy. Clinical and pathological data including sex, patient age, tumor location, resection site, staging at surgery (performed in accordance with the classification of the 6th Edition of the American Joint Committee on Cancer guidelines), BRAF mutation status, perioperative chemotherapy regimens, use of molecular targeting agents including cetuximab and bevacizumab, were collected. The study protocol was reviewed and approved by the SMC institutional review board.

Perioperative chemotherapy regimens
Oxaliplatin based chemotherapy was FOLFOX (oxaliplatin 85 mg/m 2 on day 1, infused during 2 h; LV 200 mg/ m 2 , infused during 2 h, followed by 5-FU as a 400 mg/ m 2 intravenous bolus then a 1200 mg/m 2 infusion during 22 h on days 1 and 2) in 2 week treatment cycles or XELOX(oxaliplatin 130 mg/m 2 on day 1 followed by oral capecitabine 1000 mg/m 2 twice daily (day 1 to 14) in 3 week treatment cycles. Irinotecan based chemotherapy was FORFIRI (irinotecan 180 mg/m 2 on day 1, infused during 2 h; LV 200 mg/m 2 , infused during 2 h, followed by 5-FU as a 400 mg/m 2 intravenous bolus then a 1200 mg/m 2 infusion during 22 h on days 1 and 2) in 2 week treatment cycles or XELIRI (irinotecan 250 mg/ m 2 on day 1 followed by oral capecitabine 1000 mg/m 2 twice daily (day 1 to 14) in 3 week treatment cycles. If bevacizumab or cetuximab was used, patients received cetuximab (initial dose 400 mg/m 2 infused during 2 h, and 250 mg/m 2 weekly) or bevacizumab (5 mg/kg) followed by FOLFOX or FOLFIRI.

DNA extraction and mutation analysis
DNA was isolated from 10-μm formalin-fixed, paraffinembedded tumor specimens using FFPE-DNA isolation kit (Qiagen, Hilden, Germany). A Qiagen the rascreen KRAS mutation kit was used to detect the seven most common KRAS codon 12 and 13 mutations. Specifically, the mutation was detected by real-time polymerase chain reaction based on amplification-refractory muta-

Statistical analyses
Patients were subdivided into wild-type KRAS and mutant KRAS cohorts. The primary objective was to investigate the effect of KRAS mutation on the TTR. TTR was defined as the time from the date of operation to the date of local or metastatic recurrence. As of November 2014, overall survival data are not yet available for the mutant KRAS group. Data from recurrence-free patients were censored at the date of the last follow-up.   To compare baseline characteristics, categorical outcomes were analyzed using the chi-square test or Fisher's exact test. Continuous variables are presented as medians and ranges. TTR and OS were calculated using the Kaplan-Meier method, and data was compared using the log-rank test. The Cox proportional hazard model was used to assess hazard ratios (HRs) of prognostic factor. All factors of statistical significance (p < 0.10) in univariate analysis were included in the multivariate analysis. Two-sided p values of <0.05 were considered as statistically significant. All statistical analyses were performed using the SPSS statistical software version 21 (IBM, Armonk, NY. USA).

Patient characteristics
Between January 2008 and June 2014, 82 patients who were diagnosed with synchronous metastatic CRC and underwent curative resection of primary and metastatic lesions with perioperative chemotherapy were included in the analyses. Table 1 summarizes the patient characteristics according to KRAS mutation status. There was no significant difference in clinicopathologic features between the two groups. Baseline characteristics including age, sex, tumor location, tumor grade, T stage, N stage, synchronous metastasectomy site, and recurrence site were similar between the KRAS wild type and KRAS mutation cohorts. Regarding BRAF mutation status, all of the tested cases (76.8 %) were BRAF wild type.  frequently observed mutation, followed by the G > T and G > C mutations.

The impact of KRAS mutations on TTR and OS
The median follow-up durations were 25 months (range, 4-74) and 34 months (range, 9-63) for patients with KRAS wild type and KRAS mutation status, respectively. During follow-up in surviving participants, there were 57 events for TTR analysis and 25 events for OS analysis. There were no significant differences in survival time distributions according to KRAS wild type and KRAS mutation status (logrank p = 0.425 for TTR; log-rank p = 0.137 for OS, Fig. 1).
In univariate and multivariate analyses, there were no significant differences in TTR or OS between KRAS wild type and KRAS mutation cohorts (Tables 2, 3 and 4). When patients were further subdivided into three groups according to mutation subtype (wild-type vs. KRAS codon 12 mutation vs. KRAS codon 13 mutation) or amino acid missense mutation type (G > A vs. G > T vs. G > C), there were no significant differences in TTR or OS.

Discussion
The majority of studies evaluating the prognostic impact of KRAS mutational status in CRC have been conducted in patients with stage II/III disease. The QUASAR trial, which mainly evaluated patients with stage II CRC, revealed that KRAS mutations had a detrimental effect on recurrence and OS, despite adjuvant chemotherapy [17].
In contrast, the CALGB 89803 and PETACC-3 trials demonstrated that KRAS mutation status had no significant effect on recurrence or OS in patients with stage II/ III colon cancer or CRC treated with adjuvant chemotherapy [12,16]. However, conflicting findings were reported simultaneously in two large studies conducted by The Kirsten ras in-colorectal-cancer collaborative group, the RASCAL and RASCAL II trials, which were comprised of 2721 and 4268 patients, respectively [18,19].
Although the first RASCAL study reported an association of KRAS mutations with an increased risk of recurrence and death for patients with all stages of CRC, recurrence in patients with Dukes' D tumors was less than might be expected. The RASCAL II study concluded that there was a significant prognostic value in failure-free survival alone in patients with Dukes' C cancer harboring a KRAS G12V mutation. Few studies have evaluated the relationship between patients with stage IV disease at the time of diagnosis and KRAS mutations [20][21][22][23]. Patients with metastatic CRC with limited metastases undergo curative primary resection with or without metastasectomy, anti-EGFR antibody therapy, and heterogeneous chemotherapy regimens, making it difficult to evaluate the precise prognostic value of KRAS status in this setting. To overcome this limitation, in this study, we included only patients who underwent curative resection of the primary and metastatic sites who received perioperative chemotherapy. To our knowledge, this study is the first to report TTR in such patients. In this homogenous cohort of Korean patients with metastatic CRC, we observed that KRAS mutation was not associated with TTR or OS, which is congruent with previous studies [20][21][22]. Phipps et al., reported that KRAS mutations did not differ by stage at diagnosis, and that the prognostic value of KRAS mutations only became evident in patients with stage I-III disease [22]. Furthermore, Nash et al., reported that the prevalence of KRAS mutations did not vary with stage, but that KRAS mutations were strong independent predictors of survival for patients with stage I-III CRC [21].
We also investigated the association KRAS mutations with recurrence pattern in our cohort. KRAS mutations were significantly more common in lung metastases compared with liver and bladder/ovary metastases. These finding were concordant with those of Tie et al., who observed a significantly higher prevalence of KRAS mutations in patients with lung metastases compared with those with liver metastases [24]. In addition, in their study, KRAS mutations were associated with an increased risk of lung relapse in patients with stage II/III CRC who were enrolled on the VICTOR clinical trial [21]. However, in the present study, we did not observe recurrence-specific associations with KRAS mutation status. The differential impact of KRAS mutations on recurrence-specific sites according to disease stage requires evaluation in further studies.
Limitations of the present study included the relatively short follow-up, where the median OS was not reached in the KRAS mutation group. Nevertheless, sufficient TTR events occured enabling analysis of recurrence. In addition, the BRAF mutation status was not determined for 19 (33 %) patients, but BRAF mutations were only detected in a small proportion of patient and were not significantly different between KRAS wild type and KRAS mutated patients. In addition, the small sample size did not allow us to evaluate the impact of different KRAS mutation subtypes.
In conclusion, KRAS mutation was not associated with TTR or OS in curatively resected, metastatic CRC. Further validation of these finding is needed in metastatic CRC patients treated with curative resection in prospective controlled trials.

Conclusions
The present study, to our knowledge, is the first report on the effect of KRAS mutations on prognosis in surgically treated CRC patients with synchronous metastases. The most of previous studies evaluating the prognostic impact of KRAS in CRC have been conducted in patients with non-metastatic CRC, and the influence of KRAS mutations on outcome is conflicting. In our study, KRAS mutation was not associated with TTR or OS in metastatic CRC patients who undergo curative surgery and perioperative chemotherapy. KRAS mutation status was also not linked to recurrence pattern. Prospective studies will be necessary to evaluate the prognostic effect of KRAS mutation in metastatic CRC patients.

Consent
This research is strictly retrospective and involving the collection of existing data and records. The study protocol was reviewed and approved consent exemptions by the SMC institutional review board.

Competing interests
The authors declare that they have no competing interests.
Authors' contributions HSK, JL, JH conceived and designated the study; JYL, ML and SHL helped to conceive the study and revised manuscript critically for important intellectually content; SK reviewed the pathologic specimens; WYL, YAP, YBC and SY critically revised the manuscript; STK, JOP and HYL helped acquisition and interpretation of data; YSC and WIK participated in statistical analysis and interpretation of