He L, Ji WS, Jin HL, Lu WJ, Zhang YY, Wang HG, Liu YY, Qiu S, Xu M, Lei ZP, Zheng Q, Yang XL, Zhang Q. Development of a nomogram for predicting liver transplantation prognosis in hepatocellular carcinoma. World J Gastroenterol 2024; 30(21): 2763-2776 [PMID: 38899335 DOI: 10.3748/wjg.v30.i21.2763]
Corresponding Author of This Article
Qing Zhang, MD, Chief Physician, Department of Organ Transplantation, The Third Medical Centre of Chinese PLA General Hospital, No. 69 Yongding Road, Haidian District, Beijing 100039, China. zqy6920@sina.com
Research Domain of This Article
Transplantation
Article-Type of This Article
Retrospective Study
Open-Access Policy of This Article
This article is an open-access article which was selected by an in-house editor and fully peer-reviewed by external reviewers. It is distributed in accordance with the Creative Commons Attribution Non Commercial (CC BY-NC 4.0) license, which permits others to distribute, remix, adapt, build upon this work non-commercially, and license their derivative works on different terms, provided the original work is properly cited and the use is non-commercial. See: http://creativecommons.org/licenses/by-nc/4.0/
Li He, Hai-Long Jin, Yu-Yu Liu, Shuang Qiu, Meng Xu, Zi-Peng Lei, Qian Zheng, Qing Zhang, Department of Organ Transplantation, The Third Medical Centre of Chinese PLA General Hospital, Beijing 100039, China
Li He, School of Clinical Medicine, Shandong Second Medical University, Weifang 261053, Shandong Province, China
Wan-Sheng Ji, Clinical Research Center, The Affiliated Hospital of Shandong Second Medical University, Weifang 261053, Shandong Province, China
Wen-Jing Lu, Yuan-Yuan Zhang, Xiao-Li Yang, Department of Laboratory Medicine, The Third Medical Centre of Chinese PLA General Hospital, Beijing 100039, China
Co-corresponding authors: Xiao-Li Yang and Qing Zhang.
Author contributions: He L, Ji WS, and Jin HL designed and performed the research and wrote the paper; Lu WJ, Zhang YY, and Liu YY designed the research and supervised the report; Wang HG, Qiu S, Xu M, Lei ZP, and Zheng Q designed the research and contributed to the analysis; He L, Ji WS, Jin HL, Yang XL, and Zhang Q provided clinical advice; Yang XL and Zhang Q supervised the report; Yang XL and Zhang Q provided funding support and should be considered as co-corresponding authors.
Supported bythe National Natural Science Foundation of China, No. 81372595 and No. 81972696.
Institutional review board statement: This study was reviewed and approved by our hospital’s Ethics Committee (Approval No. 2023-008).
Informed consent statement: Signed informed consent forms were provided by all patients.
Conflict-of-interest statement: The authors declare that they have no conflict of interest to disclose.
Data sharing statement: No additional data are available.
Open-Access: This article is an open-access article that was selected by an in-house editor and fully peer-reviewed by external reviewers. It is distributed in accordance with the Creative Commons Attribution NonCommercial (CC BY-NC 4.0) license, which permits others to distribute, remix, adapt, build upon this work non-commercially, and license their derivative works on different terms, provided the original work is properly cited and the use is non-commercial. See: https://creativecommons.org/Licenses/by-nc/4.0/
Corresponding author: Qing Zhang, MD, Chief Physician, Department of Organ Transplantation, The Third Medical Centre of Chinese PLA General Hospital, No. 69 Yongding Road, Haidian District, Beijing 100039, China. zqy6920@sina.com
Received: February 19, 2024 Revised: April 24, 2024 Accepted: May 13, 2024 Published online: June 7, 2024 Processing time: 104 Days and 20.4 Hours
Abstract
BACKGROUND
At present, liver transplantation (LT) is one of the best treatments for hepatocellular carcinoma (HCC). Accurately predicting the survival status after LT can significantly improve the survival rate after LT, and ensure the best way to make rational use of liver organs.
AIM
To develop a model for predicting prognosis after LT in patients with HCC.
METHODS
Clinical data and follow-up information of 160 patients with HCC who underwent LT were collected and evaluated. The expression levels of alpha-fetoprotein (AFP), des-gamma-carboxy prothrombin, Golgi protein 73, cytokeratin-18 epitopes M30 and M65 were measured using a fully automated chemiluminescence analyzer. The best cutoff value of biomarkers was determined using the Youden index. Cox regression analysis was used to identify the independent risk factors. A forest model was constructed using the random forest method. We evaluated the accuracy of the nomogram using the area under the curve, using the calibration curve to assess consistency. A decision curve analysis (DCA) was used to evaluate the clinical utility of the nomograms.
RESULTS
The total tumor diameter (TTD), vascular invasion (VI), AFP, and cytokeratin-18 epitopes M30 (CK18-M30) were identified as important risk factors for outcome after LT. The nomogram had a higher predictive accuracy than the Milan, University of California, San Francisco, and Hangzhou criteria. The calibration curve analyses indicated a good fit. The survival and recurrence-free survival (RFS) of high-risk groups were significantly lower than those of low- and middle-risk groups (P < 0.001). The DCA shows that the model has better clinical practicability.
CONCLUSION
The study developed a predictive nomogram based on TTD, VI, AFP, and CK18-M30 that could accurately predict overall survival and RFS after LT. It can screen for patients with better postoperative prognosis, and improve long-term survival for LT patients.
Core Tip: This is a retrospective study to research the influencing factors affecting the prognosis of liver cancer transplantation, including pathological factors, tumor morphology and biomarkers, exploring the prediction model of liver transplantation with high accuracy, thereby optimizing the allocation of liver transplant resources by striking a balance between maximizing the number of beneficiaries and reducing the risk of hepatocellular carcinoma recurrence. By establishing predictive models and implementing a stratification system, we hope to improve the overall efficacy and success rate of liver transplantation.
Citation: He L, Ji WS, Jin HL, Lu WJ, Zhang YY, Wang HG, Liu YY, Qiu S, Xu M, Lei ZP, Zheng Q, Yang XL, Zhang Q. Development of a nomogram for predicting liver transplantation prognosis in hepatocellular carcinoma. World J Gastroenterol 2024; 30(21): 2763-2776
Hepatocellular carcinoma (HCC) is the third-leading cause of cancer-related mortality worldwide[1]. When HCC has already progressed to a moderate or advanced stage, there is a poor prognosis[2]. Liver transplantation (LT) is considered one of the most effective treatments for end-stage chronic liver disease or liver cancer. However, there was a high risk of tumor recurrence and mortality following LT before Mazzaferro created the Milan criteria (MC) in 1996[3,4]. The appearance of the MC effectively reduced the recurrence rate and improved the long-term outcomes of transplanted patients[5,6]. However, some have argued that the MC is too strict, excluding HCC patients with better prognoses who may benefit from timely treatment[7,8]. Consequently, several extension criteria have emerged, such as University of California, San Francisco (UCSF), Metroticket 2.0, Hangzhou criteria, etc. Nevertheless, these extension criteria still have limited predictive accuracy. Therefore, accurately predicting the prognosis for patients with HCC and optimizing the use of donors of LT are ongoing challenges for medical professionals and researchers.
Pathological factors, such as tumor diameter, number, capsule, microvascular invasion, portal vein tumor thrombosis, and tumor differentiation are high-risk factors for tumor recurrence and metastasis[9]. However, relying solely on these factors to judge LT prognosis is insufficient, as the importance of tumor biomarkers in predicting prognosis after LT has been increasingly acknowledged[10,11]. Alpha-fetoprotein (AFP) is the most commonly used biomarker for HCC and is closely associated with the biological aggressiveness and vascular invasion (VI) of the tumor[12-14]. High AFP levels have been linked to poor prognosis in patients with HCC[15,16]. Changes in peri-transplant AFP levels have clinical significance as predictors of HCC recurrence[17]. However, approximately 30% to 40% of patients with HCC have normal AFP levels, and reduced AFP levels are not necessarily indicative of cure for HCC, so using AFP alone is not recommended as a screening tool for assessing HCC prognosis[18]. Additional prognostic markers and clinical indicators should be considered in conjunction with AFP. Des-gamma-carboxy prothrombin (DCP) is a complementary biomarker together with AFP[19]. Elevated DCP levels may indicate more aggressive tumor behavior and a worse prognosis in patients with HCC[20]. A recent study showed elevated serum DCP levels in liver transplant patients with recurrent tumors[21].However, DCP’s utility as a separate application is limited. Golgi protein 73 (GP73) is an HCC biomarker with significant diagnostic and prognostic value. Overexpression of GP73 is strongly associated with advanced HCC stage, higher tumor grade, and poor overall survival (OS)[22]. However, few studies report the prognostic value of GP73 in HCC post-LT. One study reported that serum GP73 level at 6 months was an independent risk factor in LT[23]. However, the value of the preoperative GP73 expression level for the prognosis of LT merits further investigation. Cytokeratin 18 (CK18) is a major intermediate filament protein in the liver, and serum CK18 is associated with apoptotic cell death in hepatocytes[24]. CK18 epitopes M30 (CK18-M30) antigen, a fragment of CK18, can be used to detect apoptosis[25]. CK18-M30 has been suggested as a potential marker for early liver tumor invasiveness[26]. However, the cytokeratin-18 epitopes M65 (CK18-M65) antigen, released during cell necrosis, can be detected in intact CK18[27]. Elevated CK18-M65 levels are associated with reduced survival and an increased rate of tumor recurrence[28]. However, these findings are still controversial, and further studies are needed to clarify their significance.
Thus, the present study used clinical data combined with biomarkers to develop a robust and accurate predictive model for patients with HCC considering LT. While related studies have utilized existing standards combined with clinical pathology and AFP to construct a nomogram, the factors included were complex, and the predictive values of these nomograms were unsatisfactory[29,30]. With the advent of new biomarkers, the surveillance of HCC continues to evolve. Therefore, a model based on biomarkers may more accurately classify patients with HCC into different risk groups, thereby optimizing the allocation of liver transplant resources by striking a balance between maximizing the number of beneficiaries and reducing the risk of HCC recurrence. By establishing predictive models and implementing a stratification system, we hope to improve the overall efficacy and success rate of LT.
MATERIALS AND METHODS
Study population selection
A total of 160 patients with HCC underwent LT between 2016 to 2019 at the PLA General Hospital Third Medical Center. There were 147 males and 13 females; the mean age was 52.5 years ± 9.4 years; 138 patients with HCC had viral hepatitis B. Inclusion criteria: (1) Patients with imaging and pathological diagnosis of HCC; (2) patients treated with surgical LT; (3) no significant distant metastasis or other irreversible disease; (4) age ≥ 18 years; and (5) detailed clinical data and serum specimens available for prognostic assessment and analysis. Exclusion criteria: (1) Active systemic malignancy unless related to HCC and controllable; (2) severe extrahepatic systemic diseases, such as advanced renal insufficiency or severe heart failure; (3) human immunodeficiency virus infection; (4) deceased from non-tumor factors in the postoperative perioperative period or within 2 months; (5) necessary clinical and laboratory data unavailable for prognosis evaluation; and (6) other important non-tumor-related limitations such as mental disability, lack of social support, etc.
This study was conducted in accordance with both the Declarations of Helsinki and Istanbul and approved by our hospital’s Ethics Committee. Signed informed consent forms were provided by all patients. All materials for this project were funded by the National Science Foundation.
Data collection
Clinical data of all patients with HCC was collected, including age, sex, number of tumors, tumor diameter, presence of microsatellite lesions, lymph node metastasis, presence of VI, Child-Pugh score, degree of tissue differentiation, MELD score, aspartate transaminase, alanine transaminase, alkaline phosphatase, and gamma-glutamyl transferase. The OS of the patients was defined as the duration between the LT and either the time of death or last follow-up. Similarly, the recurrence-free survival (RFS) was calculated as the period between the transplantation and the occurrence of relapse or last follow-up. Study data collection was concluded on 31 June 2023.
Measurements of tumor biomarkers
Before LT, 5 mL of peripheral blood was collected from each patient after fasting time is not less than 8 h. Samples were centrifuged at 3000 RPM for 5 min, then the serum was immediately frozen at -80 °C until testing. Values of serum AFP, GP73, DCP, CK18-M65, and CK18-M30 were obtained using a magnetic particle chemiluminescence analyzer (C2000). The use of a fully automated biochemical analyzer and the application of specific kits for the detection of these biomarkers allowed for efficient and accurate assessment of these biomarker levels in the serum. (The Analyzer of C2000 and kit are from Beijing Hotgen Biotech Co.,Ltd.).
Follow-up
After the LT surgery, a comprehensive immunosuppressive regimen was implemented to prevent rejection using tacrolimus or cyclosporine, together with corticosteroids. The corticosteroid dose was gradually reduced over three months following the surgery. After three months, consideration was given to switching to sirolimus for primary immunosuppression. During the postoperative follow-up, monthly follow-ups were conducted throughout the first year to closely observe the patient's progress, then extended to every 3 to 6 months. In the event of suspicious tumor lesions, further tests were obtained. In uncertain cases, a liver biopsy was performed to confirm the diagnosis. The diagnostic criteria for tumor recurrence were similar to those used for the initial HCC.
Statistical analysis
We utilized the t-test to compare normally distributed means of the groups, and the χ2 test to examine the association between categorical variables. The “pROC” package was used to draw the receiver operating characteristic (ROC) curve, and the Younden index to determine the best cutoff value. Survival rates were calculated using the Kaplan-Meier method, and multivariate analysis was performed using the Cox regression analysis. The “randomForestSRC” package was used for the random forest analysis, the “rms” package to build the nomogram, the “timeROC” package to draw the time-dependent ROC curve and determine the area under the curve (AUC) of the nomogram, and the “rms” package to draw the calibration curve. The “dcurves” package was used to perform the decision curve analysis (DCA) curve analysis. The xtile software was used for risk classification. The “survminer” package was used to draw the survival curve. All these statistical analyses were conducted using SPSS version 25.0, xtile software version 3.6.1, and R version 4.3.1.
RESULTS
Demographic and clinical characteristics
The clinical characteristics of 160 patients with HCC are shown in Table 1. In the cohort, 105 patients survived (mean survival 63.03 months ± 11.70 months) and 55 patients died (mean survival 22.24 months ± 15.23 months). The proportion meeting the MC was 40.0%, UCSF 51.9%, and Hangzhou 68.1% (Table 1). The optimal cut-off value and efficacy analysis of the tumor biomarkers are presented in Table 2.
Table 1 Characteristics of hepatocellular carcinoma patients, n (%).
The construction of the Cox proportional hazards model
In univariate analysis, we found that the total tumor diameter (TTD) > 8 cm, maximum tumor diameter (MTD) > 5 cm, VI, membrane invasion (MI), microsatellite lesions (ML), AFP > 400 ng/mL, GP73 > 222 ng/mL, DCP > 26 ng/mL, CK18-M65 > 1066 UL, and CK18-M30 > 530 U/L were risk factors influencing the OS of patients with HCC post-LT. Additionally, multiple tumor (MT), TTD > 8 cm, MTD > 5 cm, VI, MI, ML, AFP > 400 ng/mL, GP73 > 222 ng/mL, DCP > 26 ng/mL, CK18-M65 > 1066 UL, and CK18-M30 > 530 U/L were risk factors influencing the RFS of HCC post-LT (Table 3). In multivariate analysis, AFP, TTD, and CK18-M30 were independent risk factors for predicting OS (Table 4), and RFS in HCC (Table 5).
Table 3 Univariate analysis for overall survival and recurrence-free survival.
Variable
n
Overall survival (%)
P value
Recurrence-free survival (%)
P value
1 yr
3 yr
5 yr
1 yr
3 yr
5 yr
Number of tumour
0.054
0.006
Single tumor
79
0.87
0.78
0.75
0.78
0.77
0.75
Multiple tumors
81
0.88
0.64
0.57
0.69
0.55
0.54
Maximum tumor diameter (cm)
< 0.001
< 0.001
> 5
48
0.79
0.50
0.43
0.58
0.45
-
≤ 5
112
0.92
0.80
0.73
0.80
0.75
0.73
Total tumor diameter (cm)
< 0.001
< 0.001
> 8
46
0.74
0.43
0.38
0.43
0.37
0.34
≤ 8
114
0.93
0.82
0.76
0.86
0.78
-
Vascular invasion
< 0.001
< 0.001
Yes
42
0.69
0.50
0.37
0.50
0.33
0.31
No
118
0.90
0.79
0.74
0.82
0.78
-
Lymph node metastasis
0.098
0.129
Yes
20
0.75
0.60
0.49
0.60
0.50
-
No
140
0.92
0.72
0.68
0.75
0.68
0.67
Membrane invasion
0.001
< 0.001
Yes
59
0.78
0.54
0.50
0.76
0.52
0.46
No
101
0.92
0.78
0.72
0.95
0.82
0.76
Microsatellite lesions
0.002
0.001
Yes
79
0.83
0.60
0.52
0.65
0.53
0.51
No
81
0.92
0.81
0.77
0.81
0.79
0.77
AFP
< 0.001
< 0.001
≤ 400
115
0.90
0.82
0.78
0.86
0.79
-
> 400
45
0.82
0.42
0.33
0.42
0.31
0.28
GP73
0.001
< 0.001
≤ 222
126
0.91
0.76
0.73
0.78
0.73
0.72
> 222
34
0.76
0.52
0.34
0.55
0.38
-
DCP
< 0.001
< 0.001
≤ 26
92
0.91
0.82
0.81
0.85
0.82
0.81
> 26
68
0.83
0.55
0.44
0.57
0.42
0.41
CK18-M65 (U/L)
< 0.001
< 0.001
≤ 1066
95
0.93
0.84
0.81
0.85
0.80
0.80
> 1066
65
0.80
0.52
0.41
0.56
0.43
0.41
CK18-M30 (U/L)
< 0.001
<0.001
≤ 530
95
0.94
0.86
0.83
0.86
0.83
0.82
> 530
65
0.78
0.49
0.39
0.54
0.40
0.39
MC
< 0.001
< 0.001
Yes
64
0.97
0.86
0.82
0.88
0.83
0.81
No
96
0.82
0.61
0.55
0.63
0.54
0.53
UCSF
< 0.001
< 0.001
Yes
77
0.97
0.89
0.85
0.90
0.87
0.85
No
83
0.79
0.54
0.46
0.57
0.47
0.45
Hangzhou
< 0.001
< 0.001
Yes
109
0.95
0.83
0.78
0.86
0.81
0.80
No
51
0.80
0.45
0.37
0.47
0.33
0.31
Table 4 Multivariate Cox hazards analysis for overall survival.
Variable
B
SE
HR
95%CI
P value
AFP > 400 ng/mL
1.22
0.28
3.38
1.95-5.88
< 0.001
CK18-M30 > 530 U/L
1.14
0.31
3.15
1.69-5.87
< 0.001
TTD > 8 cm
0.92
0.28
2.52
1.45-4.40
0.001
Table 5 Multivariate Cox hazards analysis for recurrence-free survival.
Variable
B
SE
HR
95%CI
P value
AFP > 400 ng/mL
1.45
0.28
4.28
2.47-7.40
< 0.001
CK18-M30 > 530 U/L
0.97
0.31
2.66
1.44-4.91
0.002
TTD > 8 cm
1.10
0.28
3.03
1.73-5.29
< 0.001
The construction of the random forest model
Using the random forest method to predict OS after LT, the minimum error rate corresponded to 200 trees. Using 200 to refit the model, we identified AFP, VI, CK18-M30, and TTD as important variables (Figure 1A). In predicting RFS after LT, the minimum error rate corresponded to 190 trees. Using 190 to refit the model also identified AFP, VI, CK18-M30, and TTD as important variables (Figure 1B). The Cox proportional hazards model constructed predicted an AUC of 0.73 [95% confidence interval (95%CI): 0.60-0.86] for 1-year OS, 0.83 (95%CI: 0.76-0.91) for 3-year AUC, and 0.85 (95%CI: 0.77-0.92) for 5-year AUC (Figure 1C). It also predicted an AUC of 0.85 (95%CI: 0.78-0.93) for 1-year RFS, 0.86 (95%CI: 0.79-0.92) for 3-year AUC, and 0.87 (95%CI: 0.81-0.94) for 5-year AUC (Figure 1D). But the random forest model constructed from this analysis yields an AUC at 1-year OS of 0.78 (95%CI: 0.68-0.88), the 3-year AUC of 0.85 (95%CI: 0.79-0.92), and the 5-year AUC of 0.87 (95%CI: 0.80-0.94) (Figure 1E). It also predicted an AUC of 1-year RFS as 0.86 (95%CI: 0.80-0.93), the 3-year AUC of 0.88 (95%CI: 0.83-0.94), and the 5-year AUC of 0.91 (95%CI: 0.85-0.96) (Figure 1F).
Figure 1 Selection of the variables of random forest model and the time dependence area under the curve of the models.
A: Number of best trees and variable importance plots obtained from the random forest method for overall survival (OS); B: Number of best trees and variable importance plots obtained from the random forest method for recurrence free survival (RFS); C: The area under the curve (AUC) of 1, 3, and 5 years of the cox model for OS; D: The AUC of 1, 3, and 5 years of the cox model for RFS; E: The AUC of 1, 3, and 5 years of the forest model for OS; F: The AUC of 1, 3, and 5 years of the forest model for RFS. AFP: Alpha-fetoprotein; TTD: Total tumor diameter; VI: Vascular invasion; CK18: Cytokeratin-18; MTD: Maximum tumor diameter; DCP: Des-gamma-carboxy prothrombin; MI: Membrane invasion; GP73: Golgi protein 73; ML: Microsatellite lesions; AUC: Area under the curve.
The construction of the nomograms
We selected the random forest model because of its greater accuracy using VI, TTD, AFP, and CK18-M30 as variables for our nomogram to predict postoperative OS (Figure 2A), and postoperative RFS (Figure 2B). After assigning point values for each of these parameters, the risk prediction formula for OS after 5-year LT is: 1.3e-08 × points3 + -1.2535e-05 × points2 + -0.000563459 × points + 0.900886136 (Supplementary Table 1). The risk prediction formula for RFS after 5-year LT is 2.2e-08 × points3 + -1.7979e-05 × points2 + -0.000665255 × points + 0.901058744 (Supplementary Table 2).
Figure 2 The construction of the nomogram.
A: The nomogram of predicting overall survival after liver transplantation (LT); B: The nomogram of predicting hepatocellular carcinoma recurrence free survival after LT. AFP: Alpha-fetoprotein; CK18: Cytokeratin-18; TTD: Total tumor diameter; VI: Vascular invasion; OS: Overall survival; RFS: Recurrence free survival.
Performance analysis of the nomograms
According to the results of ROC curve analysis, the AUC for predicting OS and RFS was higher than that of the Milan, UCSF, and Hangzhou criteria, and the AUC of 1 year, 3 years, and 5 years predicting postoperative OS was 0.78 (95%CI: 0.70-0.85), 0.85 (95%CI: 0.79-0.92), and 0.87 (95%CI: 0.80-0.94), respectively. The AUC of 1 year, 3 years, and 5 years predicting postoperative RFS was 0.86 (95%CI: 0.77-0.91), 0.88 (95%CI: 0.83-0.95), 0.87 (95%CI: 0.80-0.94), and 0.91 (95%CI: 0.85-0.96), respectively (Table 6). The calibration curve also showed a good fit of the nomogram for post-LT OS and RFS (Figure 3).
Figure 3 Efficacy comparison of nomogram and existing criteria.
A: The 3-year area under the curve (AUC) of nomogram and Milan, Hangzhou and UCSF for overall survival (OS); B: The 5-year AUC of nomogram and Milan, Hangzhou and UCSF for OS; C: The 3-year AUC of nomogram and Milan, Hangzhou, UCSF for recurrence free survival (RFS); D: The 5-year AUC of nomogram and Milan, Hangzhou, UCSF for RFS; E: The calibration curve of nomogram for OS; F: The calibration curve of nomogram for RFS. OS: Overall survival; RFS: Recurrence free survival.
Table 6 The area under the curve comparison of the nomogram with the other standards.
Overall survival
Recurrence-free survival
Nomogram
Milan
UCSF
Hangzhou
Nomogram
Milan
UCSF
Hangzhou
t = 1-yr
0.78
0.66
0.70
0.72
0.86
0.65
0.72
0.73
t = 3-yr
0.85
0.65
0.71
0.71
0.88
0.66
0.73
0.73
t = 5-yr
0.87
0.67
0.72
0.73
0.91
0.68
0.74
0.76
Survival analysis between high-risk, middle-risk, and low-risk patients
Each patient was scored according to the nomogram, and the best cutoff was obtained according to the calculations, dividing patients into low-risk, middle-risk, and high-risk populations. The 5-year survival of the low-risk population was greater than 90%, while the 5-year survival of the high-risk population was less than 30% (Table 7). In the analysis of OS, 29 were in the high-risk group, 40 in the middle-risk group, and 91 in the low-risk group, The OS of the three groups were (26.1 ± 19.2) months, (41.3 ± 23.6) months, and (59.6 ± 17.3) months, respectively (P < 0.001). In the analysis of RFS, 21 patients were at high risk, 36 at middle risk, and 103 at low risk. The RFS was (10.8 ± 15.9), (24.1 ± 23.6), and (57.5 ± 20.5) months, respectively (P < 0.001). The DCA curve showed that the high clinical utility of the nomogram exceeded those of the Milan, UCSF, and Hangzhou standards (Figure 4).
Figure 4 Survival analysis of nomogram and clinical efficacy analysis.
A: The survival curve for overall survival (OS); B: The survival curve for recurrence free survival (RFS); C: The decision curve analysis (DCA) curve for OS; D: The DCA curve for RFS.
Table 7 Division of the risk classification and 5-year survival rate.
Low risk
Middle risk
High risk
OS
Total points
0-98
100-198
215-315
5-yr survival rate
≥ 90%
70%-89%
≤ 30%
RFS
Total points
0-100
108-208
208-280
5-yr survival rate
≥ 90%
45%-89%
≤ 45%
DISCUSSION
LT is a potentially curative treatment for early-stage HCC. However, the number of patients awaiting transplantation far exceeds the available number of organs[31]. highlighting the need to accurately select the most suitable candidates for LT. This study reports the construction of a reliable nomogram incorporating TTD, VI, AFP, and CK8-M30. The nomogram can be used to accurately predict which patients with HCC are likely to have a more favorable prognosis after undergoing LT. The 5-year AUC for predicting OS of the nomogram was 0.87 and the 5-year AUC for predicting RFS was 0.91. By classifying the patients into risk categories, the long-term survival rate of patients with HCC post-LT can be significantly improved.
This study confirmed that tumor morphology, such as TTD and VI, were found to be valid predictors of clinical outcome after LT and were important for the performance of the model, which is consistent with previous studies[32,33]. In addition to these factors, tumor biomarkers also play an important role in HCC prognosis after LT. This study found that AFP and CK18-M30 were influential factors in determining clinical outcome after LT. Importantly, our study is the first to confirmed the prognostic value of CK18-M30 for HCC post-LT. Some scholars had previously reported that M30/M60 has prognostic value for acute-on-chronic liver failure[34]. and other researchers suggested that CK18-M30 expression correlates with the presence of acute cellular rejection post-LT[35]. It may therefore be an effective prognostic indicator for liver outcome[36]. In our study, the expression level of CK18-M30 was important for the clinical outcome after LT, with an optimal cutoff value of 530 U/L. The CK18-M30 prognostic value for LT was significantly higher than that of CK18-M60 and even more valuable than preoperative DCP and GP73. Its importance in the prognostic model should not be underestimated. Serum AFP level has been shown to be strongly associated with post-LT survival[37]. Elevated AFP before LT is a predictor of HCC recurrence and is linked to worse survival[38,39]. Therefore, experts recommend using AFP levels to select patients for LT waiting lists[40], although there may be variations in the optimal cutoff value for AFP. In recent studies, patients with HCC and AFP levels greater than 1000 ng/mL had reduced survival after LT and a higher risk of HCC recurrence[41-43]. In our study, AFP > 400 ng/mL was a reliable predictor affecting OS and RFS after LT and was of great importance. We utilized statistical models such as the Cox risk regression model and random forest model to identify the most accurate predictors of prognosis for HCC after LT, and found that the forest model containing TTD, VI, AFP, and CK18-M30 was superior in predicting HCC outcomes. As serum markers are easily accessible, we recommend adding them to the preoperative examination program for LT patients. Tumor number, tumor size, and venous invasion can be obtained by imaging, avoiding invasive examinations such as tissue biopsy. Miao et al[44] developed a nomogram based on PET/CT, which had a higher AUC than Milan and UCSF, but PET/CT is expensive and not available to all patients. In comparison, our nomogram could achieve the same predictive efficacy using an easily available method. We could classify the risk of recurrence and survival of patients using the nomogram. Patients with low risk, with a five-year survival above 90%, should be considered first for LT, so that available organs can be optimally allocated and patients can get the maximum benefit.
Comparing the nomogram to existing standards, such as the MC and UCSF, which are internationally accepted criteria in selecting patients with HCC who are suitable for LT, we found that our nomogram had better prognostic value and clinical practicality. Although using MC and UCSF can significantly reduce the recurrence rate and mortality rate after LT, they focus excessively on tumor morphology. Our model’s strength is that it utilizes tumor morphology combined with tumor markers. In addition, we found that the proportion of people eligible for LT by MC was 40% and by UCSF 48%, less than half of all patients, leaving many patients without surgical options. The nomogram not only can expand the number of patients suitable for surgery, but also has high prognostic value. The Hangzhou criteria have been standardized for Chinese populations as follows: (1) Absence of portal vein tumor thrombus; and (2) TTD less than or equal to 8 cm or TTD more than 8 cm, with histopathologic grade I or II and preoperative AFP level less than or equal to 400 ng/mL, simultaneously[45]. In our study, 68% of patients met the Hangzhou criteria. Hangzhou criteria require the degree of tumor tissue differentiation, which necessitates a biopsy. Our model avoids this invasive procedure by utilizing CK18-M30. The AUC for predicting the 5-year OS after LT using the nomogram was 0.87, and for predicting the RFS, it was 0.91. In contrast, the AUC for predicting the 5-year OS using the Hangzhou criteria was 0.72, and for predicting the RFS, it was 0.76. This suggests that the nomogram may have a superior predictive AUC compared to the Hangzhou criteria.
The nomogram was designed to provide a visual representation of the predictive model, allowing clinicians to estimate the individualized prognosis for patients with HCC undergoing LT. In addition, the nomogram can be used to monitor changes in the patient's condition in real time. When indicators change, the model can adjust the prediction results to provide more accurate treatment suggestions for the patient. Through the application of the prediction nomogram, doctors can screen out patients with better prognosis after LT and perform LT surgery, making more effective use of donor livers, while striving for a longer survival for patients. In conclusion, the prediction model of LT provides a scientific and accurate decision-making basis for clinicians and is expected to bring better outcomes for patients with HCC without requiring liver biopsy.
This study has certain limitations. Although we have successfully constructed a nomogram for predicting prognosis, it is essential to validate its accuracy and reliability through a larger multicenter prospective study. This additional verification will provide more robust evidence regarding the performance of our prediction nomogram.
CONCLUSION
This study confirmed that the tumor biomarkers AFP and CK18-M30 are important prognosis indicators for patients with HCC after LT. The developed prediction nomogram is a valuable tool for LT prognosis in patients with HCC. The nomogram demonstrates good accuracy and reliability, enabling clinicians to make more informed decisions regarding the transplantation process, patient selection, and post-transplantation follow-up strategies.
Footnotes
Provenance and peer review: Unsolicited article; Externally peer reviewed.
Peer-review model: Single blind
Specialty type: Transplantation
Country of origin: China
Peer-review report’s classification
Scientific Quality: Grade B, Grade B
Novelty: Grade B, Grade B
Creativity or Innovation: Grade B, Grade B
Scientific Significance: Grade B, Grade B
P-Reviewer: Ker CG, Taiwan; Yildirim M, Türkiye S-Editor: Chen YL L-Editor: A P-Editor: Yu HG
Goldaracena N, Mehta N, Scalera I, Sposito C, Atenafu EG, Yao FY, Muiesan P, Mazzaferro V, Sapisochin G. Multicenter validation of a score to predict prognosis after the development of HCC recurrence following liver transplantation.HPB (Oxford). 2019;21:731-738.
[PubMed] [DOI][Cited in This Article: ][Cited by in Crossref: 15][Cited by in F6Publishing: 14][Article Influence: 2.8][Reference Citation Analysis (0)]
Heinrich S, Mittler J, Theurer J, Ridder DA, Marquardt JU, Weinmann A, Scheuermann U, Otto G, Galle PR, Straub BK, Lang H. Microvascular invasion of hepatocellular carcinoma predicts microvascular invasion of its recurrence: potential implications for salvage liver transplantation?Hepatobiliary Surg Nutr. 2023;12:183-191.
[PubMed] [DOI][Cited in This Article: ][Cited by in F6Publishing: 1][Reference Citation Analysis (0)]
Goldberg DS, Taddei TH, Serper M, Mehta R, Dieperink E, Aytaman A, Baytarian M, Fox R, Hunt K, Pedrosa M, Pocha C, Valderrama A, Kaplan DE. Identifying barriers to hepatocellular carcinoma surveillance in a national sample of patients with cirrhosis.Hepatology. 2017;65:864-874.
[PubMed] [DOI][Cited in This Article: ][Cited by in Crossref: 72][Cited by in F6Publishing: 87][Article Influence: 12.4][Reference Citation Analysis (0)]
Koch C, Bette T, Waidmann O, Filmann N, Schrecker C, Trojan J, Weiler N, Vermehren J, Schnitzbauer AA, Bechstein WO, Zeuzem S, Herrmann E, Welker MW. AFP ratio predicts HCC recurrence after liver transplantation.PLoS One. 2020;15:e0235576.
[PubMed] [DOI][Cited in This Article: ][Cited by in Crossref: 5][Cited by in F6Publishing: 5][Article Influence: 1.3][Reference Citation Analysis (0)]
Mazzaferro V, Sposito C, Zhou J, Pinna AD, De Carlis L, Fan J, Cescon M, Di Sandro S, Yi-Feng H, Lauterio A, Bongini M, Cucchetti A. Metroticket 2.0 Model for Analysis of Competing Risks of Death After Liver Transplantation for Hepatocellular Carcinoma.Gastroenterology. 2018;154:128-139.
[PubMed] [DOI][Cited in This Article: ][Cited by in Crossref: 301][Cited by in F6Publishing: 386][Article Influence: 64.3][Reference Citation Analysis (0)]
Trevisani F, D'Intino PE, Morselli-Labate AM, Mazzella G, Accogli E, Caraceni P, Domenicali M, De Notariis S, Roda E, Bernardi M. Serum alpha-fetoprotein for diagnosis of hepatocellular carcinoma in patients with chronic liver disease: influence of HBsAg and anti-HCV status.J Hepatol. 2001;34:570-575.
[PubMed] [DOI][Cited in This Article: ][Cited by in Crossref: 512][Cited by in F6Publishing: 475][Article Influence: 20.7][Reference Citation Analysis (0)]
Villalba-López F, Sáenz-Mateos LF, Sánchez-Lorencio MI, De La Orden-García V, Alconchel-Gago F, Cascales-Campos PA, García-Bernardo C, Noguera-Velasco JA, Baroja-Mazo A, Ramírez-Romero P. Usefulness of PIVKA-II for monitoring after liver transplantation in patients with hepatocellular carcinoma.Sci Rep. 2023;13:5621.
[PubMed] [DOI][Cited in This Article: ][Cited by in F6Publishing: 1][Reference Citation Analysis (0)]
Wang Y, Zhang Z, Zou D, Hao Q, Ma X, Wang C, Zhang J, Zhang X, Wang B, Wang X, Yang X. Serum Golgi protein 73 is a prognostic biomarker of liver transplantation patients.Int J Clin Exp Pathol. 2017;10:8626-8632.
[PubMed] [DOI][Cited in This Article: ]
Leers MP, Kölgen W, Björklund V, Bergman T, Tribbick G, Persson B, Björklund P, Ramaekers FC, Björklund B, Nap M, Jörnvall H, Schutte B. Immunocytochemical detection and mapping of a cytokeratin 18 neo-epitope exposed during early apoptosis.J Pathol. 1999;187:567-572.
[PubMed] [DOI][Cited in This Article: ][Cited by in F6Publishing: 6][Reference Citation Analysis (0)]
Waidmann O, Köberle V, Bettinger D, Trojan J, Zeuzem S, Schultheiß M, Kronenberger B, Piiper A. Diagnostic and prognostic significance of cell death and macrophage activation markers in patients with hepatocellular carcinoma.J Hepatol. 2013;59:769-779.
[PubMed] [DOI][Cited in This Article: ][Cited by in Crossref: 56][Cited by in F6Publishing: 60][Article Influence: 5.5][Reference Citation Analysis (0)]
Jing YY, Liu WT, Guo SW, Ye F, Fan QM, Yu GF, Yu DD, Gao L, Sun K, Han ZP, Li R, Yang Y, Zhao QD, Wu MC, Wang HY, Wei LX. Hepatitis B virus (HBV) receptors: Deficiency in tumor results in scant HBV infection and overexpression in peritumor leads to higher recurrence risk.Oncotarget. 2015;6:42952-42962.
[PubMed] [DOI][Cited in This Article: ][Cited by in Crossref: 16][Cited by in F6Publishing: 16][Article Influence: 2.0][Reference Citation Analysis (0)]
Szymanska S, Markiewicz-Kijewska M, Pyzlak M, Kaliciński P, Teisseyre J, Czubkowski P, Karkucińska-Więckowska A, Kowalski A, Stefanowicz M, Ismail H. Is M30 staining a reliable marker predicting graft rejection and fibrosis in protocolar liver biopsies in post-liver transplant pediatric patients?Minerva Pediatr (Torino). 2023;.
[PubMed] [DOI][Cited in This Article: ][Reference Citation Analysis (0)]
Oweira H, Sadeghi M, Volker D, Mieth M, Zidan A, Khajeh E, Ghamarnejad O, Fonouni H, Weiss KH, Schmidt J, Lahdou I, Mehrabi A. Serum Caspase-Cleaved Cytokeratin (M30) Indicates Severity of Liver Dysfunction and Predicts Liver Outcome.Ann Transplant. 2018;23:393-400.
[PubMed] [DOI][Cited in This Article: ][Cited by in Crossref: 1][Cited by in F6Publishing: 2][Article Influence: 0.3][Reference Citation Analysis (0)]
Sapisochin G, Goldaracena N, Laurence JM, Dib M, Barbas A, Ghanekar A, Cleary SP, Lilly L, Cattral MS, Marquez M, Selzner M, Renner E, Selzner N, McGilvray ID, Greig PD, Grant DR. The extended Toronto criteria for liver transplantation in patients with hepatocellular carcinoma: A prospective validation study.Hepatology. 2016;64:2077-2088.
[PubMed] [DOI][Cited in This Article: ][Cited by in Crossref: 197][Cited by in F6Publishing: 233][Article Influence: 29.1][Reference Citation Analysis (0)]
Duvoux C, Roudot-Thoraval F, Decaens T, Pessione F, Badran H, Piardi T, Francoz C, Compagnon P, Vanlemmens C, Dumortier J, Dharancy S, Gugenheim J, Bernard PH, Adam R, Radenne S, Muscari F, Conti F, Hardwigsen J, Pageaux GP, Chazouillères O, Salame E, Hilleret MN, Lebray P, Abergel A, Debette-Gratien M, Kluger MD, Mallat A, Azoulay D, Cherqui D; Liver Transplantation French Study Group. Liver transplantation for hepatocellular carcinoma: a model including α-fetoprotein improves the performance of Milan criteria.Gastroenterology. 2012;143:986-94.e3; quiz e14.
[PubMed] [DOI][Cited in This Article: ][Cited by in Crossref: 561][Cited by in F6Publishing: 654][Article Influence: 54.5][Reference Citation Analysis (0)]
Lai Q, Avolio AW, Graziadei I, Otto G, Rossi M, Tisone G, Goffette P, Vogel W, Pitton MB, Lerut J; European Hepatocellular Cancer Liver Transplant Study Group. Alpha-fetoprotein and modified response evaluation criteria in solid tumors progression after locoregional therapy as predictors of hepatocellular cancer recurrence and death after transplantation.Liver Transpl. 2013;19:1108-1118.
[PubMed] [DOI][Cited in This Article: ][Cited by in Crossref: 139][Cited by in F6Publishing: 152][Article Influence: 13.8][Reference Citation Analysis (0)]
Xu X, Lu D, Ling Q, Wei X, Wu J, Zhou L, Yan S, Wu L, Geng L, Ke Q, Gao F, Tu Z, Wang W, Zhang M, Shen Y, Xie H, Jiang W, Wang H, Zheng S. Liver transplantation for hepatocellular carcinoma beyond the Milan criteria.Gut. 2016;65:1035-1041.
[PubMed] [DOI][Cited in This Article: ][Cited by in Crossref: 118][Cited by in F6Publishing: 157][Article Influence: 19.6][Reference Citation Analysis (0)]
