Molecular marker identification for relapse prediction in 5-FU-based adjuvant chemotherapy in gastric and colorectal cancers

Abstract

To confirm the clinical significance of NF-κB and JNK protein expression from experimentally identified candidates for predicting prognosis for patients with 5-FU treatment, we evaluated the protein expression of surgically removed specimens. A total of 79 specimens were obtained from 30 gastric and 49 colorectal cancer patients who underwent R0 resection followed by postoperative 5-FU based adjuvant chemotherapy. Immunohistochemical examinations of NF-κB and JNK on tissue microarrays (TMAs) revealed that significantly shorter time-to-relapse (TTR) in both NF-κB(+) and JNK(-) subgroups in both gastric (NF-κB(+), p = 0.0002, HR11.7. 95%CI3 3.2-43.4; JNK(-), p = 0.0302, HR4.4, 95%CI 1.2-16.6) and colon (NF-κB(+), p = 0.0038, HR36.9, 95%CI 3.2-426.0; JNK(-), p = 0.0098, HR3.2, 95%CI 1.3-7.7) cancers. These protein expression patterns also show strong discriminately power in gastric cancer patients for overall survival rate, suggesting a potential utility as prognostic or chemosensitivity markers. Baseline expression of these proteins using gastric cancer cell lines demonstrated the reciprocal patterns between NF-κB and JNK, while 5-FU exposure of these cell lines only induced NF-κB, suggesting that NF-κB plays a dominant role in the response to 5-FU. Subsequent siRNA experiments confirmed that gene knockdown of NF-κB increased 5-FU-specific sensitivity, whereas that of JNK did not affect the chemosensitivity. These results suggest that the expression of these proteins may aid in the decisions involved with adjuvant chemotherapy for gastrointestinal tract cancers.

Figure 1. Hierarchical clustering of three different matrices and results of immunohistochemical examinations of candidate markers.

(A) Based on a chemosensitivity assay of a cancer cell line panel, the A (activity) × C (cells)  =  AC matrix was created. (B) Quantitative protein expression data of each cell line determined by “reverse-phase” lysate microarray generates the C×P (protein)  =  CP matrix. (C) A heatmap with hierarchical clustering representation of the AP matrix, which is generated from AC and CP matrices. (D) Immunohistochemical stainings of candidate markers for 5-FU treatment.

Figure 2. Time-to-relapse (TTR) rates on the basis of NF-κB and JNK protein expression in gastric and colon carcinomas.

Figure 3. Induction of biomarkers by 5-FU treatment.

(A) Baseline protein expression of NF-κB and JNK in five gastric cancer cell lines. Tublin was used as a loading control. (B) Induction of candidate biomarkers in response to 5-FU treatment in different concentrations in MKN45 and KE39. (C) Examination of protein localization by fluorescent immunocytochemistry using MKN45. (D) Increased inhibitory growth effect by anticancer agents in gastric cancer cell lines after transfection of siRNA for NF-κB p65 and JNK transcripts. Control samples are the corresponding cell lines transfected with the indicated siRNAs without anticancer agents. Abbreviations are: CIS, cisplatinum; DTX, docetaxel; and PXL, paclitaxel; and 5FU, 5-fluorouracil. *p<0.05, Student t-test.