
Junjie YIN1,2*, Zhu LI3*, Jinyan CHEN1, Qingyang QUE1, Huigang LI1, Ruijie ZHAO1, Jianyong ZHUO6, Chiyu HE1, Wei SHEN1, Peiru ZHANG5, Yifan GU1, Chenghao CAO1, Jiaqi ZHENG1, Yuhang LI1, Rongsen WANG3, Zuyuan LIN1, Shengjun XU2, Xuyong WEI2, Shusen ZHENG4, Di LU3, Xiao XU1,3,4,6. Adverse impacts of body composition abnormalities on outcome after liver transplantation for hepatocellular carcinoma: Development of a multi-task classifier chain surrogate model[J]. Journal of Zhejiang University Science B, 1998, -1(-1): .
@article{title="Adverse impacts of body composition abnormalities on outcome after liver transplantation for hepatocellular carcinoma: Development of a multi-task classifier chain surrogate model",
author="Junjie YIN1,2*, Zhu LI3*, Jinyan CHEN1, Qingyang QUE1, Huigang LI1, Ruijie ZHAO1, Jianyong ZHUO6, Chiyu HE1, Wei SHEN1, Peiru ZHANG5, Yifan GU1, Chenghao CAO1, Jiaqi ZHENG1, Yuhang LI1, Rongsen WANG3, Zuyuan LIN1, Shengjun XU2, Xuyong WEI2, Shusen ZHENG4, Di LU3, Xiao XU1,3,4,6",
journal="Journal of Zhejiang University Science B",
volume="-1",
number="-1",
pages="",
year="1998",
publisher="Zhejiang University Press & Springer",
doi="10.1631/jzus.B2600062"
}
%0 Journal Article
%T Adverse impacts of body composition abnormalities on outcome after liver transplantation for hepatocellular carcinoma: Development of a multi-task classifier chain surrogate model
%A Junjie YIN1
%A 2*
%A Zhu LI3*
%A Jinyan CHEN1
%A Qingyang QUE1
%A Huigang LI1
%A Ruijie ZHAO1
%A Jianyong ZHUO6
%A Chiyu HE1
%A Wei SHEN1
%A Peiru ZHANG5
%A Yifan GU1
%A Chenghao CAO1
%A Jiaqi ZHENG1
%A Yuhang LI1
%A Rongsen WANG3
%A Zuyuan LIN1
%A Shengjun XU2
%A Xuyong WEI2
%A Shusen ZHENG4
%A Di LU3
%A Xiao XU1
%A 3
%A 4
%A 6
%J Journal of Zhejiang University SCIENCE B
%V -1
%N -1
%P
%@ 1673-1581
%D 1998
%I Zhejiang University Press & Springer
%DOI 10.1631/jzus.B2600062
TY - JOUR
T1 - Adverse impacts of body composition abnormalities on outcome after liver transplantation for hepatocellular carcinoma: Development of a multi-task classifier chain surrogate model
A1 - Junjie YIN1
A1 - 2*
A1 - Zhu LI3*
A1 - Jinyan CHEN1
A1 - Qingyang QUE1
A1 - Huigang LI1
A1 - Ruijie ZHAO1
A1 - Jianyong ZHUO6
A1 - Chiyu HE1
A1 - Wei SHEN1
A1 - Peiru ZHANG5
A1 - Yifan GU1
A1 - Chenghao CAO1
A1 - Jiaqi ZHENG1
A1 - Yuhang LI1
A1 - Rongsen WANG3
A1 - Zuyuan LIN1
A1 - Shengjun XU2
A1 - Xuyong WEI2
A1 - Shusen ZHENG4
A1 - Di LU3
A1 - Xiao XU1
A1 - 3
A1 - 4
A1 - 6
J0 - Journal of Zhejiang University Science B
VL - -1
IS - -1
SP -
EP - 0
%@ 1673-1581
Y1 - 1998
PB - Zhejiang University Press & Springer
ER -
DOI - 10.1631/jzus.B2600062
Abstract: Background: Sarcopenia and adipose abnormalities are co-occurring risks in liver transplantation for hepatocellular carcinoma (HCC), yet reliance on computed tomography (CT) segmentation limits routine assessment. Thus, we aimed to quantify the cumulative prognostic burden of these correlated phenotypes and develop a multi-task classifier chain surrogate to predict them using standard clinical data. Methods: In a cohort of 409 HCC transplant recipients, we analyzed the impact of four CT-derived phenotypes on recurrence-free survival (RFS) and overall survival (OS). To predict these correlated comorbidities without CT, we designed a Multi-task Classifier Chain framework using 37 routine peri-transplant variables. This architecture was explicitly trained to exploit inter-phenotype dependencies, comparing performance against standard multi-task baselines. Results: All four phenotypes could independently predict poor outcomes. A profound "dose-response" relationship was observed: 5-year OS dropped from 77.3% (
≤1 abnormality) to 19.6% (≥3 abnormalities; P<0.0001). In the test set, the Classifier Chain framework significantly outperformed baselines by capturing inter-phenotype dependencies. It achieved high discrimination for sarcopenia (area under the receiver operating characteristic curve (AUC), 0.88), myosteatosis (AUC 0.88), visceral obesity (AUC 0.87), and high subcutaneous adipose tissue density (SATD) (AUC 0.79), yielding the greatest net clinical benefit. Conclusions: body composition abnormalities exert a severe, additive negative impact on post-transplant outcomes. By modeling inter-phenotype dependencies, our multi-task classifier chain provides an accurate, non-invasive surrogate, enabling robust risk stratification using routine clinical data alone.
CLC number:
On-line Access: 2026-07-13
Received: 2026-02-01
Revision Accepted: 2026-06-01
Crosschecked: 0000-00-00
Cited: 0
Clicked: 9
Open peer comments: Debate/Discuss/Question/Opinion
<1>