Page 97 - OR-1-1

Table 3. Univariate and multivariate Cox regression of RFS Table 3. (Continued)
Variables Univariate Multivariate Variables Univariate Multivariate
HR (95% CI) p‑value HR p‑value HR (95% CI) p‑value HR p‑value
(95% CI) (95% CI)
PDOs-based drug sensitivity test Local invasion
Matched Reference Reference No Reference
Unmatched 2.52 (1.25 – 5.05) 0.009 2.27 0.023 Yes 0.63 (0.28 – 1.44) 0.276
(1.12 – 4.59) Note: p-values in boldface indicates statistical significance.
Gender Abbreviations: AFP: Alpha-fetoprotein; CA19-9: Carbohydrate antigen
Female Reference 19-9; CEA: Carcinoembryonic antigen; LN: Lymph node;
PDOs: Patient-derived tumor organoids; RFS: Recurrence-free
Male 0.65 (0.34 – 1.26) 0.201 survival, CI: Confidence interval.
AFP, μg/L
<20 Reference
≥20 1.06 (0.38 – 3.02) 0.907
CA19-9, U/L
<19
≥19 1.26 (0.63 – 2.52) 0.514
CEA, U/L
<5 Reference
≥5 0.45 (0.18 – 1.16) 0.514
Diameter, cm
<5 Reference Reference
≥5 2.87 (1.01 – 8.16) 0.048 1.97 (0.68 – 0.212 Figure 6. Nomogram for the prediction of RFS generated on the basis of
5.74) the PDOs-based drug sensitivity testing, vascular invasion, and tumor
Number number.
Single Reference Reference Abbreviations: PDO: Patient-derived tumor organoid; RFS: Recurrence-
free survival.
Multiple 3.08 (1.58 – 6.01) <0.001 2.06 0.044
(1.02 – 4.17)
Cirrhosis ratio offers a suitable tool for discriminating drug response
phenotypes. Importantly, this streamlined approach may
No Reference reduce the time and labor involved in large-scale PDO
Yes 0.65 (0.20 – 2.16) 0.486 screening – a critical factor in facilitating more clinically
Surgical margin, cm relevant applications.
<1 Reference Moreover, the observational research design –
≥1 0.59 (0.29 – 1.23) 0.159 wherein clinicians remained blinded to the PDOs-
Hepatectomy based drug sensitivity findings at the time of making
Minor Reference adjuvant chemotherapy decisions – mitigates the ethical
Major 1.31 (0.63 – 2.72) 0.471 and decision-making concerns that often accompany
Differentiation experimental or interventional trials. This “matched versus
unmatched” paradigm ensures that outcome differences
Well Reference can be attributed primarily to the inherent predictive value
Poor 1.04 (0.49 – 2.21) 0.406 of the PDOs-based drug sensitivity results, thus enhancing
Vascular invasion the internal validity of our comparisons. Although
No Reference Reference prospective clinical trials informed by PDOs-based drug
Yes 2.79 (1.44 – 5.43) 0.002 2.15 0.028 sensitivity testing would offer more definitive evidence
(1.08 – 4.26) for its utility, 17-19 our present findings provide encouraging
LN metastasis preliminary support for further validation in this direction.
No Reference Finally, the integration of XGBoost machine learning
Yes 1.32 (0.68 – 2.56) 0.406 model and the construction of a nomogram model
strengthen the predictive power of our analysis. These
(Cont’d...) methods not only reaffirm the prognostic influence of

Volume 1 Issue 1 (2025) 10 doi: 10.36922/or.8571