BACKGROUND & AIMS: To develop an in situ molecular signature to predict postsurgical recurrence in hepatocellular carcinoma (HCC) patients. METHODS: Immunohistochemistry was performed using tissue microarrays containing both tumoral and peri-tumoral regions of the advancing tumor edge from 336 HCC patients (289 were positive for hepatitis B virus) who underwent curative resection. Forty-nine variables were analyzed in the training set (n=151) using support vector machine and stepwise algorithms to develop a classifier to predict recurrence within 1 year, which was mainly caused by invasion or metastasis from the primary tumors. The classifier was further validated in an independent cohort of 185 patients (71 internal and 114 external). RESULTS: The final signature was composed of eight IHC features: CD80(T), B7-DC(T), HLA-DR(P), FasL(P), Bcl-2(T), Ki-67(T), cyclin D1(T), and CK19(T). In the independent test set, this classifier reliably predicted recurrence within 1 year (sensitivity, 69.1%; specificity, 65.0%) with an odds ratio of 4.149 (95% CI, 2.189-7.864). Based on a multivariate logistic model, the in situ molecular signature provided significant predictive power independent of tumor number, tumor size, vascular invasion and BCLC classification (p=0.001). The highest potential clinical impact of the classifier was observed in early-stage (BCLC classification 0-A) patients (p<0.0001), and the classifier was also predictive of the time-to-recurrence and overall survival (both p<0.0001). CONCLUSIONS: This in situ molecular classifier could provide a novel approach to identify patients who are at greatest risk for postsurgical recurrence of HCC and may benefit from intensive clinical follow-up or chemopreventive strategies.
BACKGROUND & AIMS: To develop an in situ molecular signature to predict postsurgical recurrence in hepatocellular carcinoma (HCC) patients. METHODS: Immunohistochemistry was performed using tissue microarrays containing both tumoral and peri-tumoral regions of the advancing tumor edge from 336 HCC patients (289 were positive for hepatitis B virus) who underwent curative resection. Forty-nine variables were analyzed in the training set (n=151) using support vector machine and stepwise algorithms to develop a classifier to predict recurrence within 1 year, which was mainly caused by invasion or metastasis from the primary tumors. The classifier was further validated in an independent cohort of 185 patients (71 internal and 114 external). RESULTS: The final signature was composed of eight IHC features: CD80(T), B7-DC(T), HLA-DR(P), FasL(P), Bcl-2(T), Ki-67(T), cyclin D1(T), and CK19(T). In the independent test set, this classifier reliably predicted recurrence within 1 year (sensitivity, 69.1%; specificity, 65.0%) with an odds ratio of 4.149 (95% CI, 2.189-7.864). Based on a multivariate logistic model, the in situ molecular signature provided significant predictive power independent of tumor number, tumor size, vascular invasion and BCLC classification (p=0.001). The highest potential clinical impact of the classifier was observed in early-stage (BCLC classification 0-A) patients (p<0.0001), and the classifier was also predictive of the time-to-recurrence and overall survival (both p<0.0001). CONCLUSIONS: This in situ molecular classifier could provide a novel approach to identify patients who are at greatest risk for postsurgical recurrence of HCC and may benefit from intensive clinical follow-up or chemopreventive strategies.
Authors: Yuan Liao; Bo Wang; Zhi-Liang Huang; Ming Shi; Xing-Juan Yu; Limin Zheng; Shengping Li; Lian Li Journal: PLoS One Date: 2013-04-02 Impact factor: 3.240