High MACC1 expression in combination with mutated KRAS G13 indicates poor survival of colorectal cancer patients

Colorectal cancer (CRC) is the third most common cancer worldwide [1]. According to the state of the art for classification, colorectal carcinoma can be divided into three major pathways based on the underlying molecular alterations [2,3]. The majority of CRC belong to the “Traditional pathway” which reflects the adenoma-carcinoma sequence model [4,5]. Two additional pathways where recently described, the “Serrated” and “Alternate pathway” [2-4], which are characterized by differences in the microsatellite instability (MSI) status and mutations in the KRAS or BRAF oncogenes. In advanced stages metastatic spread shortens the 5-year survival rate from 90% to 10% [6,7]. Risk estimation by using genetic factors for early identification of high risk patients is still limited [8].

One promising candidate is the metastasis-associated in colon cancer 1 (MACC1) gene. MACC1 is an important prognostic marker for metastases formation in CRC [9,10].

Other promising molecular markers are oncogenic KRAS mutations, which are detectable in 30-50% of CRC tumors [11]. The most frequent point mutations in codon 12 (KRAS G12) and 13 (KRAS G13) of exon 2 result in constitutive activation of KRAS and downstream pathways [12]. Although the role of KRAS mutations as predictive biomarkers for anti-EGFR-targeted therapy is well characterized, there is a need to clarify their relevance as prognostic biomarkers for recurrence of metastatic CRC [13].

In this retrospective study, we attempted to improve the prognostic value of MACC1 for CRC metastasis using separate and combined survival analyses of MACC1 expression levels and KRAS G12 or G13 mutations in tumors of CRC patients. In addition, we investigated the prognostic significance of BRAF V600 mutation and MSI status, two of the most promising prognostic and predictive markers in the landscape of molecular biomarkers.

In total, 99 patients with patho-histologically confirmed primary colorectal adenocarcinomas with Union for International Cancer Control (UICC) stages I, II and III were included in this study. This patient study protocol was approved by the Institutional Review Board (IRB) of the Charité Universitätsmedizin Berlin, complied with the Helsinki Declaration. The median metastasis-free survival (MFS) time and overall survival (OS) time of the patients were 56.0 and 79.4 months, respectively. This tumor marker study based on hypothesis-driven approaches was designed in compliance with the REMARK guidelines [14]. MACC1 mRNA expression level of the tumors were quantified using quantitative reverse transcriptase PCR [9]. All patient and tumor characteristics are shown in detail in the Additional file 1: Table S1–S4. For KRAS and BRAF mutation analyses as well as MSI status determination DNA sequencing methods were used [15,16]. All used material and methods are described in detail in Additional file 2. The molecular characteristics in this cohort (Additional file 1: Table S2) demonstrate the general distribution within colorectal tumors and mutual exclusivity of the oncogenes KRAS and BRAF [7,15].

MACC1 expression was significantly higher in primary tumors of patients who developed metachronous metastases during the follow-up period (P = 0.009; Figure 1A), which confirmed previously described coherences [9,10]. We showed for the first time a correlation of high MACC1 expression with tumors harboring KRAS mutations (G12 or G13) (P = 0.049; Figure 1B) [7]. Separate analyses of primary tumors harboring the KRAS G13 mutation exhibited a significantly higher MACC1 expression (P = 0.010; Figure 1D), but not for KRAS G12 mutated tumors (P = 0.156; Figure 1C). High MACC1 expression also correlated with BRAF wild type (wt) (P = 0.027; Figure 1E) and microsatellite stable (MSS)/MSI-Low (MSI-L) (P = 0.009; Figure 1F) tumors.

We performed Kaplan-Meier survival analyses (Figure 2) and univariate Cox regression to estimate hazard ratios (HR) and 95% confidence intervals (CI) for the association of the tumor characteristics with MFS (Additional file 1: Table S5). Interestingly, univariate Cox regression and Log Rank test revealed that in the set of analyzed markers, only high MACC1 expression levels and KRAS G13 mutation emerged as prognostic factors for MFS. Patients with high MACC1 expression levels (Figure 2A, Log Rank P < 0.001; Cox regression: low vs. high, HR: 5.02, 95% CI: 2.19-11.53, P < 0.001) or KRAS G13 mutation (Figure 2D, Log Rank P = 0.022; Cox regression: wt vs. KRAS G13, HR: 4.25, 95% CI: 1.21-14.93, P = 0.024) in their primary tumors exhibited a significantly poorer prognosis. Although the role of KRAS G13 mutation as a predictive biomarker for anti-EGFR-targeted therapy [18] is well characterized, the prognostic value of KRAS G13 mutation in patients with CRC is controversial [19]. Various reports either encourage or dismiss KRAS mutations as prognostic biomarker for disease-specific survival rates or to predict liver or lung metastases [7,19-24]. Data of the Kirsten Ras in Colorectal Cancer Collaborative Group Studies (RASCAL I) suggested that the risk of recurrence and death is increased by the presence of KRAS mutations [11], and the updated RASCAL II study confined the correlation of KRAS mutation and survival prognosis to the G12V mutation [25]. According to our results, comparison of mutated KRAS tumors with KRAS wt tumors had no significant impact on MFS (Figure 2B, Log Rank P = 0.499; Cox regression: wt vs. KRAS G12 or G13, HR: 1.40, 95% CI: 0.63-3.09, P = 0.409). Studies comparing the effects of KRAS G12 with KRAS G13 mutations are rare [19,24]. Therefore separate analyses of KRAS G12 and KRAS G13 mutations in the tumors were included and demonstrated that especially the KRAS G13, but not KRAS G12 (Figure 2C, Log Rank P = 0.654; Cox regression: wt vs. KRAS G12, HR: 1.29, 95% CI: 0.55-3.03, P = 0.556), mutations were prognostic for reduced MFS. Separate analyses of the most prominent G12 mutations (G12V, G12D and G12C) revealed no significant impact of these mutations on MFS prognosis (Log Rank P = 0.826, P = 0.896 and P = 0.766; respectively). Furthermore, BRAF V600 mutation (Figure 2E, Log Rank P = 0.656; Cox regression: wt vs. BRAF V600, HR: 0.75, 95% CI: 0.18-3.17, P = 0.694) or MSS/MSI-L (Figure 2F, Log Rank P = 0.085; Cox regression: MSS/MSI-L vs. MSI-H, HR: 0.04, 95% CI: 0.00-12.05, P = 0.272) were also not associated with worse prognosis.

Next, we analyzed the effects of clinicopathological parameters like age, gender, UICC stage, grading and postoperative treatment of the tumors concerning MFS in combination with MACC1 or KRAS mutations using multivariate Cox regression (Table 1). Only high MACC1 expression (HR: 6.09, 95% CI: 2.50-14.85, P < 0.001) and KRAS G13 mutation (HR: 5.19, 95% CI: 1.06-25.45, P = 0.042) turned out to be independent prognostic markers for shorter MFS in the set of analyzed biomarkers. To our knowledge, this is the first study showing a prognostic relevance of KRAS G13 for MFS in CRC patients and demonstrating KRAS G13 mutation as independent risk factor of age, gender, UICC stage, grading and postoperative treatment for metachronous metastases development.

In addition, we performed multivariate Cox regression combining MACC1 expression level with the tumor characteristics KRAS G12 or G13 mutation, BRAF V600 mutation and MSI status (Additional file 1: Table S5). This analysis revealed that MACC1 was statistically significant independent for MFS prognosis of these molecular markers (HR: 4.20, 95% CI: 1.80-9.81, P = 0.001). Therefore, we conducted combinatorial survival prognosis of the independent prognostic markers MACC1 expression and KRAS G13 mutation. Interestingly, patients with the worst prognosis were characterized by high MACC1 expression and KRAS G13 mutation (HR: 14.48, 95% CI: 3.37-62.18, P < 0.001). These patients had the shortest MFS time (Figure 2G, Log Rank: P = 0.039 mean: 18.99 months, 95% CI: 10.64-27.34) compared to patients with high MACC1 expression and KRAS wt tumors (Additional file 1: Table S6). The identification of CRC patients at high risk for metastasis formation is possible by assessing MACC1 expression in combination with KRAS G13 mutations.

Using integrative 2-step cluster analysis based on the presence of the proven markers in our cohort four clusters of patients with a different risk for metastasis formation could be distinguished and coincide well with the presumed pathways (Figure 3). According to CRC classification by Leggett and Whitehall [2], Cluster 1 represents the “Serrated pathway” with MSI-H, BRAF mutation and low MACC1 expression. These patients showed the lowest risk for metachronous metastases formation. In our study Cluster 2 and 3 represents the “Traditional pathway”, where the majority of CRC cases belong to. This pathway is characterized by MSS and KRAS or BRAF mutation. Cluster 2 with a low risk for metastasis formation was further characterized by low MACC1 expression. In contrast, Cluster 3 with increased risk exhibited high MACC1 expression levels. Thus, MACC1 expression allows a further subdivision of the “Traditional pathway” into a low and high risk group of patients concerning metastasis formation. Patients in Cluster 4 had the highest risk for metachronous metastases development. These tumors belonging to the “Alternate pathway” were defined by MSI-L/MSS, KRAS G13 mutation and can be further characterized by high MACC1 expression.