Tomoaki Yoh1, Satoru Seo2, Satoshi Ogiso1, Koshiro Morino1, Ken Fukumitsu1, Takamichi Ishii1, Yuji Nakamoto3, Kojiro Taura1. 1. Department of Surgery, Graduate School of Medicine, Kyoto University, Kyoto, Japan. 2. Department of Surgery, Graduate School of Medicine, Kyoto University, Kyoto, Japan. Electronic address: rutosa@kuhp.kyoto-u.ac.jp. 3. Department of Diagnostic Imaging and Nuclear Medicine, Kyoto University, Kyoto, Japan.
Abstract
BACKGROUND: The aim of this study was to establish a quantitative equation to predict microvascular invasion (MVI) for patients with resectable hepatocellular carcinoma (HCC). METHODS: This retrospective study included 219 patients with resected HCC from 2004 to 2015. All had available three pre-operative serological markers (alfa-feto protein (AFP), fucosylated AFP (AFP-L3), and des-gamma-carboxy prothrombin (DCP)), and one imaging marker (tumor to liver ratio of SUVmax (TLR) by 18F-FDG-PET). A multiple linear regression model for predicting MVI was developed (2004-2009, n = 111) and then validated (2010-2015, n = 108). Further, impact on the obtained model on survival outcomes was assessed. RESULTS: Using the derivation cohort, following equation was developed; MVI probability (%) = 14.2 × log10DCP + 9.9 × TLR - 22.0. This model resulted in an area under receiver operating characteristic curve (ROC) of 0.806 and 0.751, in the derivation and validation cohort, respectively. Furthermore, MVI probability ≥40% determined by ROC analysis was associated with worse overall survival and recurrence-free survival in the derivation and the validation cohort (all p < 0.05). CONCLUSION: A quantitative model, using DCP and TLR, was able to preoperatively predict with good performance MVI and long-term outcomes in patients with HCC after liver resection.
BACKGROUND: The aim of this study was to establish a quantitative equation to predict microvascular invasion (MVI) for patients with resectable hepatocellular carcinoma (HCC). METHODS: This retrospective study included 219 patients with resected HCC from 2004 to 2015. All had available three pre-operative serological markers (alfa-feto protein (AFP), fucosylated AFP (AFP-L3), and des-gamma-carboxy prothrombin (DCP)), and one imaging marker (tumor to liver ratio of SUVmax (TLR) by 18F-FDG-PET). A multiple linear regression model for predicting MVI was developed (2004-2009, n = 111) and then validated (2010-2015, n = 108). Further, impact on the obtained model on survival outcomes was assessed. RESULTS: Using the derivation cohort, following equation was developed; MVI probability (%) = 14.2 × log10DCP + 9.9 × TLR - 22.0. This model resulted in an area under receiver operating characteristic curve (ROC) of 0.806 and 0.751, in the derivation and validation cohort, respectively. Furthermore, MVI probability ≥40% determined by ROC analysis was associated with worse overall survival and recurrence-free survival in the derivation and the validation cohort (all p < 0.05). CONCLUSION: A quantitative model, using DCP and TLR, was able to preoperatively predict with good performance MVI and long-term outcomes in patients with HCC after liver resection.