Five biomarker mRNAs in combination improve detection of tumor cells and characterize their aggressiveness in lymph nodes from CRC patients

Lymph node (LN) metastasis is the single most important prognostic risk factor for tumor recurrence after curative surgery for colorectal cancer (CRC). However, a large number of patients with no detected LN metastasis recur. The aim was to improve tumor cell detection in LNs and determine tumor aggressiveness by analysis of selected biomarker mRNAs. Biomarker mRNAs expression was analyzed in a panel of LNs, primary tumors, and immune cells by quantitative reverse transcriptase-polymerase chain reaction (qRT-PCR). Biomarker mRNAs indicative of aggressiveness were selected from genome-wide hybridization bead array analysis of LNs with metastasis individually compared to a panel of primary tumors and control tissues. Specific qRT-PCR assays were constructed for selected genes. Levels were calculated by normalization to 18S rRNA. A statistical method to define a cut-off mRNA level above which patients have increased risk for recurrence was developed. Using this cut-off, each biomarker was analyzed for prognostic value in a clinical material of 174 CRC patients. Differences between patient groups in disease-free survival after surgery were calculated using Kaplan-Meier survival model and risks for recurrent disease by Cox´s regression analysis. A formula was constructed for evaluation of the prognostic value of the 5-biomarker combination. A side-by-side comparison between LN-metastasis detection by biomarker mRNA analysis and histopathology was performed on 185 LNs from 57 patients. The impact of the LN tissue volume subjected to analysis was investigated in a subgroup of 107 LNs from 30 patients. CEACAM5 mRNA was chosen as proxy for tumor cells. CEACAM5 levels were high in tumor cells, not detected in immune cells, proportional to the number of tumor cells, and high levels were an indicator of poor prognosis. Detection of metastasis/micrometastasis-positive LNs was 1.33-fold higher by CEACAM5 level compared to routine histopathology in side-by-side analysis. Differential screening identified KLK6, SLC35D3 and POSTN as indicators of poor prognosis while MUC2 was an indicator of good prognosis. Increasing LN tissue volume from an 80 μm section to half a LN increased CEACAM5-positive LNs from 34/107 to 80/107 (P < 0.0001). When combined, the 5 biomarkers could identify patients at risk for recurrence with higher sensitivity than histopathology (P < 0.0001), and the formula KLK6:CEACAM5 + SLC35D3:CEACAM5 - MUC2:CEACAM5 + POSTN:18S rRNA allowed allocation of CRC patients to different risk categories with respect to recurrence. Risk category was an independent prognostic factor to TNM-stage and tumor grade. Determination of CEACAM5 mRNA levels improves detection of tumor cells in LNs from patients surgically treated for CRC, and, together with KLK6, SLC35D3, POSTN and MUC2, further allows characterization of the LN tumor both with respect to aggressiveness of the tumor cells themselves and the tumor promoting effect of the microenvironment. Their combined expression levels allow allocating patients to different risk categories. These results became even clearer when a larger volume of LN tissue was examined. We suggest that analysis of the five biomarker mRNAs in combination may replace the routine histopathology examination of LNs in CRC patients. A national multicenter validation study is ongoing.