Mitogen Inducible Gene-6 Is a Prognostic Marker for Patients with Colorectal Liver Metastases

Abstract

Purpose: Prognostic schemes that rely on clinical variables to predict outcome after resection of colorectal metastases remain imperfect. We hypothesized that molecular markers can improve the accuracy of prognostic schemes.

Methods: We screened the transcriptome of matched colorectal liver metastases (CRCLM) and primary tumors from 42 patients with unresected CRCLM to identify differentially expressed genes. Among the differentially expressed genes identified, we looked for associations between expression and time to disease progression or overall survival. To validate such associations, mRNA levels of the candidate genes were assayed by qRT-PCR from CRCLM in 56 additional patients who underwent hepatectomy.

Results: Seven candidate genes were selected for validation based on their differential expression between metastases and primary tumors and a correlation between expression and surgical outcome: lumican; tissue inhibitor metalloproteinase 1; basic helix-loop-helix domain containing class B2; fibronectin; transmembrane 4 superfamily member 1; mitogen inducible gene 6 (MIG-6); and serpine 2. In the hepatectomy group, only MIG-6 expression was predictive of poor survival after hepatectomy. Quantitative PCR of MIG-6 mRNA was performed on 25 additional hepatectomy patients to determine if MIG-6 expression could substratify patients beyond the clinical risk score. Patients within defined clinical risk score categories were effectively substratified into distinct groups by relative MIG-6 expression.

Conclusions: MIG-6 expression is inversely associated with survival after hepatectomy and may be used to improve traditional prognostic schemes that rely on clinicopathologic data such as the Clinical Risk Score.

Figure 1

Expression of genes that show significant up- or downregulation in metastatic tumors of the liver relative to primary colon tumors from the same patient in 20% or more of the patient population with 99% certainty based on a sampling of 32 patients. Upregulation is indicated by shades of red, downregulation is indicated by shades of green, black indicates no change, and gray indicates ratios that failed to pass detection or consistency criteria. Tables 1 and 2 indicate which cluster each of the genes belongs to.

Figure 2

Kaplan-Meier survival estimates of survival in patients after resection stratified by MIG-6 expression. MaxStat software was used to optimize the cutoff value for high and low MIG-6 expression. A statistically significant difference in overall survival was found by this analysis (log-rank P = .0024)

Figure 3

Patient stratification for overall survival by CRS. Kaplan-Meier survival estimates comparing low-CRS (0-2) vs. high-CRS (3–5) groups. The median overall survival was 25 months in the high-CRS group, but it was not reached in the low-CRS group (P = .005). Fourteen patients did not have preoperative CEA levels and were excluded from the analysis. Numbers below the figure are life tables designating how many patients are represented in the survival curves.

Figure 4

MIG-6 can substratify patients beyond the CRS. Patients divided into low-CRS (0-2) and high-CRS (3-5) groups were further stratified by median MIG-6 expression. (A) Within the low-CRS group, patients were separated into two groups: those in the upper 50th percentile of MIG-6 expression and those in the lower 50th percentile. The median overall survival was 33 months in the upper–MIG-6 subgroup and was not reached in the lower–MIG-6 group (P = .03). (B) Similarly, within the high-CRS group, patients were separated into two groups: those in the upper 50th percentile of MIG-6 expression and those in the lower 50th percentile. The median overall survival in the upper– and lower–MIG-6 subgroups was 18 and 33 months, respectively (P = .04). Numbers below the figure are life tables designating how many patients are represented in the survival curves.