Six-gene Assay as a new biomarker in the blood of patients with colorectal cancer: establishment and clinical validation

Abstract

Colorectal cancer (CRC) is the second most common cancer in men and the third most common cancer in women. Although long-term survival has improved over the past 30 years, at least 50% of patients with CRC will develop metastases after diagnosis. In this study, we examined whether quantifying the mRNA of six CRC-related genes in the blood could improve disease assessment through detection of circulating tumor cells (CTC), and thereby improve progression prediction in relapsed CRC patients. Cell spiking assay and RT-PCR were performed with blood samples from healthy volunteers spiked with six CRC cell lines to generate an algorithm, herein called the Six-gene Assay, based on six genes (CEA, EpCAM, CK19, MUC1, EGFR and C-Met) for CTC detection. The CTCs of 50 relapsed CRC patients were then respectively measured by CEA Gene Assay (single-gene assay control) and Six-gene Assay. Subsequently, receiver operating characteristic analysis of the CTC panel performance in diagnosing CRC was conducted for both assays. Moreover, the 2-year progression-free survival (PFS) of all patients was collected, and the application of CEA Gene Assay and Six-gene Assay in predicting PFS was carefully evaluated with different CTC cutoff values. Encouragingly, we successfully constructed the first multiple gene-based algorithm, named the Six-gene Assay, for CTC detection in CRC patients. Six-gene Assay was more sensitive than CEA Gene Assay; for instance, in 50 CRC patients, the positive rate of Six-gene Assay in CTC detection was 82%, whereas that of CEA Gene Assay was only 70%. Moreover, Six-gene Assay was more sensitive and accurate than CEA Gene Assay in diagnosing CRC as well as predicting the 2-year PFS of CRC patients. Statistical analysis demonstrated that CTC numbers measured by Six-gene Assay were significantly associated with 2-year PFS. This novel Six-gene Assay improves the definition of disease status and correlates with PFS in relapsed CRC, and thus holds promise for future clinical applications.

Keywords: RT-PCR; Six-gene Assay; biomarker; circulating tumor cells; colorectal cancer.

Figure 1

Evaluation of cell detection efficiency of CEA Gene Assay and Six‐gene Assay. A dilution series of cells (1, 10, 100 and 1000) from CRC cell lines SW1116 and SW48 were respectively spiked in 1.0 mL of peripheral blood from a healthy donor. Blood samples were further processed by Ficoll‐Paque gradient separation, RNA extraction and real‐time quantitative PCR. The plot represents number of cells spiked versus number of cells observed. The recovery of spiked numbers of CRC cells was measured by CEA Gene Assay and Six‐gene Assay based on the mRNA expression of corresponding genes in CRC cell lines. Each error bar represents mean ± SD. Inset tables provide detailed numbers for each dilution.

Figure 2

Correlation between CEA Gene Assay and Six‐gene Assay in CTC detection. CEA Gene Assay and Six‐gene Assay were performed with blood specimens from 50 relapsed CRC patients. The CTC numbers respectively measured by Six‐gene Assay (x‐axis) and CEA Gene Assay (y‐axis) for each patient are shown on the chart. Patients with CTC detected by Six‐gene Assay but undetectable by the CEA Gene Assay are indicated with red dots. Spearman's r = 0.89, P < 0.01.

Figure 3

ROC analysis of the CTC panel performance in diagnosing CRC was conducted for CEA Gene Assay and Six‐gene Assay. CEA Gene Assay and Six‐gene Assay were performed with blood specimens from 50 relapsed CRC patients. The specificity (x‐axis) and sensitivity (y‐axis) of CEA Gene Assay (blue) and Six‐gene Assay (red) were shown on the chart, respectively.

Figure 4

Survival analysis plots for 2‐year PFS of 50 CRC patients. Time‐dependent covariate Cox regression was used to analyze the relation between the amount of CTC and 2‐year PFS. The patients were categorized according to three different cutoff values (CTC = 1, 3 and 5, respectively), the cutoff value of 5 being approximately the median of detectable CTC values in Six‐gene Assay. P values in the figures correspond to the categorical univariate likelihood ratio test. N = number of unique patients.