The present data demonstrates that mCRC patients treated with bevacizumab plus combination chemotherapy have similar outcomes regardless of KRAS mutation status. Currently, standard first-line treatment for patients with mCRC includes combination chemotherapy with either an anti-EGFR agent (cetuximab or panitumumab), or bevacizumab. Adding bevacizumab to chemotherapy prolongs survival in advanced colorectal cancer patients. However, not all mCRC patients respond to this treatment and identification of a subgroup of patients who will benefit most from bevacizumab currently represents a challenge. Data on the prognostic and/or predictive significance of KRAS status in mCRC patients treated with first-line bevacizumab-containing regimens were obtained mostly as secondary endpoints in trials comparing bevacizumab vs anti-EGFR therapy efficacy (see Additional file 1: Table S1). In general, these studies failed to demonstrate KRAS status as a predictive factor for anti-VEGF therapy in the first-line treatment of mCRC, consistent with the present study. The results of the MACRO trial indicated that KRAS status may be a prognostic factor in mCRC patients treated with bevacizumab; however, that study setting did not allow the evaluation of the predictive role of KRAS status [10]. Evaluation of OS in the studies evaluating the prognostic role of KRAS in patients treated with first-line bevacizumab, including the present study (summarized in Additional file 1: Table S1), is difficult because the patients’ survival is influenced by subsequent line of therapy and patients with wtKRAS tumors could benefit often from anti-EGFR therapy after progression on bev/chemotherapy. Accordingly, the insignificant prolongation of overall survival by 2.3 months in wtKRAS patients over the mutated KRAS subgroup may be to some extent attributed to subsequent anti-EGFR therapy (Table 1).

The adverse events from bevacizumab treatment observed in this study are consistent with those published in post-registrational trials, including BEAT, BRiTE and ARIES [11-13]. We confirmed that severe (grade 3–5) events, including arterial thromboembolic events, proteinuria, gastrointestinal perforation and wound-healing complications were rare with incidence and spectrum of adverse events comparable across all age groups [14]. Taken together, the adverse events related to bevacizumab are acceptable and well tolerated.

Prospective data on the optimal sequence of chemotherapy regimen used with targeted therapy of metastatic colorectal cancer is lacking. It has been demonstrated that first-line regimens combining fluoropyrimidines with oxaliplatin or irinotecan are of comparable efficacy [15,16]. Accordingly, the optimal choice for the chemotherapy to be used in combination with bevacizumab in the initial treatment of mCRC should be based on the characteristics of the patient and tumor, as well as the drug efficacy and toxicity profile. FOLFOX or XELOX is considered as a convenient first-line treatment option for mCRC patients as they cause less diarrhea, nausea, and alopecia than FOLFIRI or XELIRI, despite a potential for dose-limiting neurotoxicity. In the present study, patients who received in the first line bev/OX-based chemotherapy tended to have longer overall survival over those treated with irinotecan-based chemotherapy with bevacizumab. As we did not observe a difference in performance status (ECOG of 0 or 1) between patients who started treatment with bev/OX versus bev/IRI-based therapy in the first-line setting (data not shown), the trend toward better OS in bev/OX subgroup is unlikely to be attributed to better performance status in these patients. Interestingly, the inferior efficacy of treatment with the irinotecan backbone was restricted to the subgroup of patients with KRAS mutation, with median OS in the bev/OX-based subgroup being 5 months longer than in the bev/IRI-based treatment. Based on in vitro experiments showing that KRAS mutation is a predictor of oxaliplatin sensitivity in colon cancer cells, it has been suggested that mutant KRAS CRC patients might benefit more from receiving first-line oxaliplatin-based regimens [17]. Although we have not observed a better response in KRAS mutated mCRC patients treated with bev/OX-based therapy over those with wtKRAS, as it was reported when patients with advanced CRC were treated in the first line with FOLFOX-6 [18], the reported sensitivity of mutated KRAS carcinoma cells to oxaliplatin may explain the improved clinical outcome when KRAS mutant patients were treated with bev/OX-based over bev/IRI-based therapy.

We have confirmed that patients presenting with synchronous metastases have an inferior prognosis compared to patients with metachronous metastases. Among the evaluated parameters, metastatic involvement of multiple organs at time of treatment initiation was the strongest prognostic factor reducing both PFS and OS. Within the subgroup of patients with metastatic disease limited to one distant organ we did not observed difference in clinical outcome with regard to hepatic or extrahepatic (including pulmonary) involvement. Interestingly, together with Rossi et al. [19], we observe trend toward longer both PFS and OS in patients with extrahepatic disease and KRAS mutation than in wtKRAS subgroup when treated with bevacizumab plus chemotherapy. Apart from this fact, we and others [20-22] have reported that KRAS mutation in colorectal cancer itself is associated with pulmonary metastasis. Findings from the VICTOR trial showing that KRAS mutant tumors are associated with an increased risk of lung relapse in CRC patients supported the role of chest imaging in surveillance of colorectal cancer patients, particularly of those with resected primary mutated KRAS carcinoma [20,22]. The reason for increased incidence of lung metastases in KRAS mutated colorectal tumors remains unknown at this moment. Considering the decreased proportion of lymph node metastasis in mutated KRAS patients compared to wtKRAS subgroup (Table 1, [23,24]), it seems that carcinoma cells with activating mutation in KRAS may exhibit a more hematogenous metastatic spread rather than along a lymphogenous path. Survival of tumor cells within the bloodstream and adhesion in the vasculature at the metastatic sites depend on tumor cell – platelet interactions [25]. We hypothesize that activating mutation of KRAS inducing expression of molecules responsible for interaction with platelets, such as tissue factor [26], cyclooxygenase and metalloproteinase-9 [27], or cathepsin B [28] might contribute to increased protection of these carcinoma cells against shear stress as well as to enhanced adhesion properties which in turn leads to onset of pulmonary metastasis of mutated KRAS carcinoma cells and higher metastatic activity in general [29].

The present study is a retrospective analysis, and thus an unintentional selection bias for a subset of patients is possible. However, the parameters of our analysis that confer substantial reliability to the presented results are, for example, unselective multicenter input of evaluated data and the proportion of tumors with wtKRAS vs mtKRAS, mirroring the proportion of KRAS mutation previously detected in the Czech Republic during a one year survey [9]. We excluded a potential bias caused by inclusion patients with known KRAS status only, as no difference was observed between PFS of mCRC patients treated in the first-line with bevacizumab and OX- or IRI-based chemotherapy with KRAS-known (11.5 months, 95% CI 11.0-12.1) and KRAS-unknown (11.6 months, 95% CI 11.0-12.2) which is further underlined by KRAS-unselected published data on first-line bevacizumab in mCRC patients based on CORECT registry [14]. An existing limitation of the present study is that we had no data specifying the type of KRAS mutation, NRAS mutation or data on BRAF mutation. Mutation of BRAF and KRAS are, in the vast majority cases, mutually exclusive, and thus, the wtKRAS subgroup in the present study included an unknown number of BRAF mutated cases who might have worse survival rates. However, estimating from our findings of BRAF mutations in 3.8% of mCRC patients (unpublished data from diagnostic testing), contamination of the wtKRAS subgroup with mutant BRAF patients is unlikely to have been substantial. Regarding specific mutations in the KRAS gene, there is emerging data showing that mutation in codon 13 leads to a more aggressive disease course with extensive synchronous metastases compared to colorectal carcinoma with mutation in codon 12 [30]. Mutation in codon 13 is reported to confer worse prognosis and outcome than alteration in codon 12 [31]. On the other hand, when treated with cetuximab and chemotherapy, patients with mutation in codon 13 demonstrated a trend toward longer PFS than the codon 12-mutated subgroup but also a better outcome than patients with tumors without KRAS mutation [30,31]. KRAS codon 12 (Cys12 = G12C) or codon 13 (Asp13 = G13D) mutation resulted in different vascular strategy in vitro; the neovascularization in Asp13 cells was associated with high levels of VEGF-A but less effective vasculature compared to their Cys12 counterpart [32]. Moreover, a trend toward increased vascular invasion in codon 13-mutated tumors has been observed in colorectal cancer patients [33]. Thus, the different biological role of KRAS mutation in codon 13 may have also implications for anti-angiogenic treatment with bevacizumab.