Prognosis nomogram for transhepatic arterial chemoembolisation with 125I seed implantation for first-branch portal vein tumour thrombus hepatocellular carcinoma：a comparative study of two statistical analysis methods

doi:10.21203/rs.3.rs-2601819/v1

Purpose: To investigate the efficacy of transhepatic arterial chemoembolisation (TACE) combined with ¹²⁵I seed implantation in patients with first-branch portal vein tumour thrombus (PVTT) hepatocellular carcinoma (HCC), we tried to establish a nomogram to determine the prognosis of individual patients by comparing different statistical methods.

Materials and Methods: 97 patients who underwent TACE combined with PVTT 125I seed implantation served as the primary cohort. The median overall survival (mOS) was the main indicator of prognosis nomogram. Univariate analysis and LASSO regression were used to screen for variables. Selected variables were included in the multivariate Cox survival regression model, and a prognostic nomogram was developed. The performance of the nomogram was assessed in terms of calibration, discrimination, and clinical usefulness. The internal and external verifications were also evaluated.

Results: The mOS of all patients was 12.5 months (95%CI: 11.78–13.22). The final determination was made using the tumour burden, PVTT types, and platelet (PLT) count to establish a nomogram prediction model [C-index: 0.770 (95% confidence interval: 0.723–0.817)]. This model demonstrated good discrimination and calibration. Application of the nomogram to external verification data still yielded good discrimination [C-index: 0.764] and good calibration. The decision curve analysis demonstrated that the clinical validity of the model was clinically useful.

Conclusion: This nomogram can be used to predict the prognostic survival time of such patients and can help doctors make decisions in clinical practice. At the same time, the appropriate statistical methods should be selected based on clinical needs.

hepatocellular carcinoma

portal vein tumour thrombus

125I seed

transarterial chemoembolization

nomogram

Hepatocellular carcinoma (HCC) is a malignant tumour of the digestive tract. Portal vein tumour thrombus (PVTT) is a common vascular metastasis route for HCC, and 60–90% of patients with advanced HCC have PVTT.^1–3 According to the Barcelona Clinic Liver Cancer (BCLC) system, HCC with PVTT is categorised as BCLC stage C.⁴ It still has a low median survival of 2.7–4.0 months for those who are untreated.⁵ Transarterial chemoembolisation (TACE) is a standard treatment for advanced HCC. However, PVTT is often accompanied by portal hypertension, intrahepatic metastasis, and liver function deterioration,^6,7 making the effect of TACE treatment extremely limited. Other reported treatment methods, including ⁹⁰Y radioembolization, stereotactic body radiotherapy, systemic therapy, targeted therapy, and immunotherapy, have shown some curative effects; however, there is no consensus regarding PVTT.^8–11

Currently, ¹²⁵I is widely used for the treatment of HCC and other malignant tumours.^12,13 The ¹²⁵I seed implantation exhibited strong targeting and high conformability. The radiation dose received by tumours increased significantly, whereas the radiation dose to normal tissues did not increase. This therapy had the advantages of minimal damage to normal tissue, a high quality of life, and a short hospital stay.¹⁴ Simultaneously, ¹²⁵I seed implantation for PVTT significantly prolonged the median survival time of patients.^15–17 A prospective study also showed that ¹²⁵I seed implantation therapy had a greater effect on type II PVTT than TACE combined with sorafenib (TACE-S).¹⁸

Despite clinical reports on the efficacy and safety of ¹²⁵I seed implantation for the treatment of PVTT, there is no nomogram for predicting the survival of patients with HCC and PVTT after ¹²⁵I seed implantation with TACE. Therefore, this study aimed to evaluate the prognostic factors of patients with type II PVTT who underwent ¹²⁵I seed implantation. Univariate Cox regression and LASSO regression with cross-validation screening methods. By comparing the two methods for selecting the best variable, a nomogram was constructed and validated based on the best clinical characteristics to predict the prognosis of individual patients.

2.1 Patient Information

This study retrospectively collected clinical data from HCC patients with type II PVTT who underwent TACE with ¹²⁵I seed implantation at our hospital and two other hospitals from 1 June 2016 to 31 December 2019. All patients provided written informed consent to receive treatment. This retrospective study was approved by the ethics committee of each hospital.

The inclusion criteria were as follows: (1) HCC was diagnosed according to the American Association for the Study of Liver Diseases Practice Guidelines Committee.²⁰ The diagnosis of PVTT is based on pathological diagnostic criteria or a clear diagnosis of liver cancer and imaging signs of PVTT; (2) Child-Pugh grade A or B (score ≤ 7); (3) Eastern Cooperative Oncology Group (ECOG) performance status 0–1; (4) patients who were not suitable for surgery or did not meet the conditions for resection; and (5) no history of chemotherapy, radiotherapy, or systemic therapy.

2.2 Classification of PVTT

This study adopted Cheng's classification and only type II PVTT was included in this study.²¹ The imaging characteristics of PVTT include space-occupying lesions in the portal vein at each stage, enhancement in the arterial phase, and filling defects in the portal vein. At the same time, multi-phase enhanced computed tomography (CT) or dynamic enhanced magnetic resonance imaging (MRI) were used to diagnose PVTT, which manifested as portal vein lumen thickening and intravascular low-density filling defects.

2.3 TACE Procedure

After using a 5F RH catheter (Terumo, Tokyo, Japan) with a guidewire to guide routine angiography, the arterial cannula was super-selected to enter the blood supply branch of the tumour. Next, 30–50 mg/m2 doxorubicin (Haizheng Pharmaceutical Co. Ltd., Taizhou, China) and 5–20 ml of super-liquid lipiodol (Guerbet, Paris, France) were used and slowly passed through the catheter after mixing. Arterial infusion chemotherapy was injected, and gelatin sponge particles were used to supplement embolisation for those who still had residual blood after embolisation with iodised oil emulsion. The artery supplying the tumour was embolised, and the endpoint of embolisation was the stagnation of blood flow in the artery supplying the tumour.

^{2.4 125}Iodine seed implantation procedure

As we reported previously,¹⁷ four days after the first TACE session, a treatment planning system (TPS) was used to evaluate preoperative CT images to plan and determine the number of particles required, total dose, location of implanted particles, and puncture route. Under CT guidance, following the angle and depth of the puncture needle track designed for the PVTT position, a special puncture needle for seed implantation was used for percutaneous PVTT puncture, and a CT examination was performed. The particles were distributed at a spacing of 5 mm to ensure that they were evenly distributed. After seed implantation, the TPS was used for evaluation. A seed replacement is required if the radiation dose distribution is poor.

2.5 Follow-up and objective

Patients were followed-up every 4–6 weeks after TACE. Follow-up included a detailed evaluation of the clinical manifestations, laboratory indicators, and imaging examinations. The imaging examination was abdominal multiphase enhanced CT or multiphase dynamic enhanced MRI.

The main objective of this study was to construct a prognosis nomogram of patient survival, using two statistical analysis methods (including univariate Cox regression and LASSO regression with cross-validation) for comparison in order to obtain the best clinical nomogram. Also, this study intends to explain the differences between the different models from a clinical perspective.

2.6 Variable screening and nomogram construction

The R program was used (version 4.1.2; Foundation for Statistical Computing, Vienna, Austria). Univariate Cox regression and LASSO regression with cross-validation were used to initially select variables, and a multivariate Cox survival regression model was used to establish prediction models. The best model was determined based on area under the curve (AUC). Subsequently, a prognostic nomogram was established. The performance of the nomogram was assessed by calculating the concordance index (C-index) with 1000 bootstrap resamples. The value of the C-index ranged from 0.5 to 1. The higher the C-index, the higher the prediction performance. Decision curve analysis (DCA) was used to evaluate the accuracy of the model and the net clinical benefits.

2.7 Statistical analysis

The Chi-square and Student's t-test were used to compare the differences in baseline characteristics between the training and validation sets, and the survival curve was calculated using the Kaplan–Meier method. Statistical significance was set at P < 0.05.

3.1 Baseline Characteristics

From 1 June 2016 to 31 December 2019, 97 patients were enrolled in our hospital as the training set. Two other hospitals enrolled 56 patients in the validation cohort. The detailed baseline characteristics included sex, age, ECOG score, Child–Pugh score, tumour number, maximum tumour diameter, total tumour diameter, tumour burden, type of PVTT, alpha-fetoprotein (AFP) level, and blood test results. There were no statistically significant differences in the baseline characteristics of patients in the training and validation sets (Table 1).

Table 1. Baseline characteristics of patients in training set and validation set

Variable	Level	Training set	Validation set	P-value
Variable	Level	N=97	N=56
Sex	Male	85 (87.6)	49 (87.5)	1.000
Sex	Female	12 (12.4)	7 (12.5)	1.000
Age		57.21±10.99	58.96±9.76	0.323
ECOG	0	39 (40.2)	27 (48.2)	0.427
ECOG	1	58 (59.8)	29 (51.8)	0.427
Child–Pugh Score		5.87±0.82	5.82±0.83	0.749
Number	1	45 (46.4)	22 (39.3)	0.317
	2–4	40 (41.2)	22 (39.3)
	≥5	12 (12.4)	12 (21.4)
Maximum tumour diameter	＜60 mm	60 (61.9)	36 (64.3)	0.900
Maximum tumour diameter	≥60 mm	37 (38.1)	20 (35.7)	0.900
Total tumour diameter	＜100mm	69 (71.1)	45 (80.4)	0.285
Total tumour diameter	≥100 mm	28 (28.9)	11 (19.6)	0.285
Tumour burden		7.71 (3.09)	8.29 (2.26)	0.226
Position	Right/Left	50 (51.5)	22 (39.3)	0.195
Position	Biloba	47 (48.5)	34 (60.7)	0.195
PVTT	IIa	56 (57.7)	35 (62.5)	0.683
PVTT	IIb	41 (42.3)	21 (37.5)	0.683
Preoperative AFP	< 400 ng/mL	58 (59.8)	35 (62.5)	0.874
Preoperative AFP	≥400 ng/mL	39 (40.2)	21 (37.5)	0.874
PLT (×109/L)		146.97±39.86	145.79±34.97	0.854
PT (s)		13.12±1.56	13.19±1.36	0.779
TBL (umol/L)		23.69±15.96	20.33±13.27	0.186
ALB (g/L)		38.52±5.96	37.99±4.37	0.564
WBC (×109/L)		5.31±2.25	5.37±1.72	0.876
RBC (×109/L)		4.32±0.61	4.45±0.55	0.201
HGB (g/L)		132.34±18.79	130.12±14.38	0.447

Abbreviations: ECOG, Eastern Cooperative Oncology Group; PVTT, portal vein tumour thrombosis; AFP, alpha-fetoprotein; PLT, platelet; PT, prothrombin time; TBL, total bilirubin; ALB, albumin; WBC, white blood cell; RBC, red blood cell; HGB, haemoglobin.

3.2 Child–Pugh score changes and survival analysis

The overall survival of all patients was > 3 months. The Child–Pugh scores before and after treatment were 5.87 ± 0.83 and 5.61 ± 0.82, respectively, and the difference was statistically significant (t = 2.34, P = 0.02). As of 30 June, 2021, the mOS was 12.5 months (95% confidence interval [CI]: 11.78–13.22) in the training set patients (Fig. 1), and the survival rates at 6, 12, and 18 months were 93.8%, 61.9%, and 24.7%, respectively.

3.3 Preliminary screening of variables

As shown in Fig. 2, the LASSO regression (Fig. 2a) with cross-validation (Fig. 2b) and univariate Cox regression (Fig. 2c) were used as the preliminary screening variables. The results showed that the variables for univariate Cox regression screening were the Child–Pugh score, tumour burden, PVTT type, and platelet (PLT) level. LASSO regression with cross-validation showed that the screening variables were tumour burden, PVTT type, and PLT level.

3.4 Comparison of two different models

The multivariate Cox survival regression model and stepwise backward regression were used to determine the final model variables with the minimum AUC. The results showed that the variables from univariate Cox regression screening to be analysed by the multivariate Cox survival regression model were Child–Pugh score, tumour burden, PVTT type, and PLT level (Model 1). The variables selected by LASSO regression with cross-validation did not change after multivariate Cox survival regression model analysis (Model 2), including tumour burden, PVTT type, and PLT level.

As shown in Fig. 3a, the 12-month AUC of Model 2 (0.850, 95%CI: 0.781–0.933) was slightly larger than that of Model 1 (0.850, 95%CI: 0.770–0.931). However, no statistical difference was observed between Models 1 and 2 (delta.AUC = 0.70, P = 0.60). The time-dependent AUC curve showed that there was no significant difference between the two models for any period (Fig. 3b). Model 2 was selected to make the model more concise.

3.5 Development and validation of an individualized prediction model

A model that incorporated independent predictors (including tumour burden, PVTT type, and PLT level) was developed and presented as a nomogram (Fig. 4a). The calibration curve for 12-months probability showed a moderate level of consistency between the predicted and observed results. For internal validation (Fig. 4b), the C-index for the prediction nomogram was 0.770 (95% CI: 0.723–0.817) for the training set, which was confirmed to be 0.703 (95% CI: 0.628–0.778) via training set validation.

K-fold cross-validation (CV) was used for external validation, and in the K-fold CV procedure, the training and validation sets were randomly and evenly split into K parts. Model 1 was built based on K-1 parts of the training set. The prediction accuracy of this candidate model was then evaluated on K-1 parts of the validation set. By respectively using each of the K parts as the training set and repeating the model building and evaluation procedure, we choose the model with the smallest cross-validation score (the mean squared prediction error [MSPE]) as the “optimal” model. Our data were run 1000 times, where K was equal to 5. At last, the ultimate model was obtained with its parameter estimates as the average values across K candidate ‘optimal’ models. At the same time, we obtained an average of 1000 C-indices (0.766) (Fig. 5a).

3.6 Clinical usefulness

The DCA results for the nomogram are shown in Fig. 5b. The decision curve demonstrated that if the threshold probability of a patient or physician was 25%, using the developed nomogram to predict median survival was more beneficial than using the treat-all-patients or treat-none schemes.

4.1 Analysis of the curative effect of TACE combined with ¹²⁵I seed implantation

Previous studies have shown that PVTT is a risk factor that affects the survival of patients with HCC.^23–25 In this study, TACE combined with ¹²⁵I seed implantation was used to treat HCC patients with type II PVTT, and the mOS reached 12.5 months, similar to previous reports.^15–18 Owing to obstruction of the portal vein caused by PVTT, reperforming TACE greatly reduces blood supply to the liver, leading to liver tissue necrosis and liver function failure. Therefore, it is particularly important to control PVTT progression and reduce the incidence of complete portal vein obstruction. When PVTT is well-controlled, the patient's liver function can be restored to a certain extent.^18,19 Therefore, for HCC patients with PVTT, it is particularly important to actively control PVTT.

It is worth noting that there were differences in the variables screened using the two different methods. Traditional univariate and multivariate Cox regression models consider the Child–Pugh score as an independent prognostic factor affecting patient survival. However, the Child–Pugh score was not statistically significant in the LASSO regression. This needs to be explained from a clinical perspective. As mentioned above, ¹²⁵I effectively alleviated the deterioration of liver function based on PVTT control. This allowed some patients with poor liver function to enter remission and may not differ in overall survival from those with better liver function. These may be Child–Pugh scores, which are not independent prognostic factors for survival. Therefore, we chose to use Model 2.

4.2 Analysis and clinical significance of predictive models

Research based on TNM staging only focuses on the change in the size of the patient’s tumour and whether it has metastasised,²⁰ while ignoring the impact of clinical indicators and overall patient survival. It is undoubtedly one-sided to consider the survival of patients solely based on tumour status, and the development of a nomogram provides a more comprehensive basis for judgment. Therefore, the nomogram is considered more advantageous than the traditional staging method.²⁰

The “Six and Twelve” model proposed in a previous study is an excellent model for HCC treatment with TACE.²² This model indicated that patients with a cumulative tumour size > 12 cm have a poor prognosis, and the mOS was only 15.8 months. This study also used the sum of tumour size (cm) and tumour number to evaluate the tumour burden. According to the characteristics of the enrolled patients, the mean tumour burden was 7.71 cm, indicating that the tumour burden was relatively large. However, it is worth noting that the "Six and Twelve" model does not evaluate vascular metastasis, and the appearance of PVTT undoubtedly increased the tumour burden. This showed that even in patients with portal vein metastasis, the use of the sum of maximum tumour diameter and number to evaluate intrahepatic lesions, combined with the type of PVTT, could be used to predict the survival of HCC patients with type II PVTT who undergo TACE combined with ¹²⁵I seed implantation therapy. According to the prediction model developed in this study, it is feasible to predict the prognosis of such patients.

Simultaneously, we evaluated the model from different angles, conducted external verifications, and obtained satisfactory results. In statistical analysis, the general model usually brings the validation set input directly into the training set model to obtain the C-index. This study used the K-fold CV procedure and CV to achieve multiple C-index calculations. The verification results are more robust than those of traditional methods. ²³ Many studies have preliminarily demonstrated that machine learning models have good applications in survival analysis,^24,25 but traditional multivariate Cox regression models also have irreplaceable advantages. Several methods are currently available for univariate screening. Constructing different Cox survival models and comparing their advantages and disadvantages will undoubtedly make our research more scientific. Further analysis of machine learning models and traditional survival models would complement each other, which may be our next research direction.

4.3 Limitations

This study had certain limitations. First, this was a retrospective study with a small sample size, which may have led to selection bias. These results need to be further confirmed in prospective, multicentre, randomised controlled trials. Although tumour burden and PVTT type were included in the nomogram as predictors, they showed good predictive ability. Due to the small number of indicators included in this study, more effective indicators may not be included. Therefore, the prediction efficiency of the model must be improved further. As discussed above, combining machine-learning models with traditional survival analysis may be the key to solving this problem.

4.4 Conclusion

In this study, by comparing two different statistical analysis methods, a nomogram was established to predict TACE combined with PVTT ¹²⁵I seed implantation to achieve individualised prediction in such patients.

Ethics approval and consent to participate:

The study protocol conformed to the ethical guidelines of the World Medical Association Declaration of Helsinki, and was approved by the Ethics Committee of the Affiliated Cancer Hospital of Zhengzhou University review board. All of the patients sign in the informed consent form.

Consent for publication:

Our version has been presented as a "preprint" in Research Square according to the following link: [1] https://www.researchsquare.com/article/rs-1355598/v1. We declare: A preprint has been published [1]. At the same time, we have also updated the latest manuscripts in the "preprint" system.

Availability of data and materials：

The datasets generated during and/or analysed during the current study are available from the corresponding author on reasonable request.

Competing Interests：

The authors have no relevant financial or non-financial interests to disclose.

Funding：

Grant Support: Henan Province Natural Science Foundation (212300410403). Medical Education Research Project of Henan Province (Wjlx2021334). Medical Science and Technology Research Project of Henan Province (No. LHGJ20190633).Henan Province Natural Science Foundation (222300420574).

Author Contributions：

Conception and design:Hong-Tao Hu; Patient selection and treatment: Xiao-Hui Zhao, Lei Zhao, Hai-Liang Li, Guang-Shao Cao; Data collection, analysis and interpretation: Xiao-Hui Zhao, Li-Lin Zhang, Yan Zhao, Shi-Jun Xu, Yuan Guo, Hai-Liang Li; Data interpretation: Hong-Tao Hu, Xiao-Hui Zhao, Li-Lin Zhang, Yan Zhao, Shi-Jun Xu, Hai-Liang Li; Undertook steering committee activities and critical statistical processing: Hong-Tao Hu, Yan Zhao; Manuscript writing:Xiao-Hui Zhao, Lei Zhao, Li-Lin Zhang, Yan Zhao; Manuscript reviewing: Hong-Tao Hu.

Acknowledgement

Not applicable.

Authors' information (optional)

1.Department of Minimal-Invasive Intervention, The Affiliated Cancer Hospital of Zhengzhou University, Zhengzhou, Henan Province, 450003, China; 2.Health Science Center, Yangtze University, Jingzhou, Hubei Province,434023, China; 3.Screening Department, Henan Red Cross Blood Center, Zhengzhou, Henan Province, 450012, China; 4.Department of Intervention, Henan Provincial People's Hospital, Zhengzhou, Henan Province, 450003, China

Massarweh NN, El-Serag HB. Epidemiology of hepatocellular carcinoma and intrahepatic cholangiocarcinoma. Cancer Control 2017;24:1073274817729245.
Yang JD, Hainaut P, Gores GJ, Amadou A, Plymoth A, Roberts LR. A global view of hepatocellular carcinoma: trends, risk, prevention and management. Nat Rev Gastroenterol Hepatol 2019;16:589–604.
Wei X, Jiang Y, Zhang X, Feng S, Zhou B, Ye X, et al. Neoadjuvant three-dimensional conformal radiotherapy for resectable hepatocellular carcinoma with portal vein tumor thrombus: a randomized, open-label, multicenter controlled study. J Clin Oncol 2019;37:2141–51.
Guarino M, Tortora R, de Stefano G, Coppola C, Morisco F, Salomone Megna A, et al. Adherence to Barcelona Clinic Liver Cancer guidelines in field practice: results of Progetto Epatocarcinoma Campania. J Gastroenterol Hepatol 2018;33:1123–30.
Cerrito L, Annicchiarico BE, Iezzi R, Gasbarrini A, Pompili M, Ponziani FR. Treatment of hepatocellular carcinoma in patients with portal vein tumor thrombosis: beyond the known frontiers. World J Gastroenterol 2019;25:4360–82.
Wang JC, Xia AL, Xu Y, Lu XJ. Comprehensive treatments for hepatocellular carcinoma with portal vein tumor thrombosis. J Cell Physiol 2019;234:1062–70.
Cabibbo G, Enea M, Attanasio M, Bruix J, Craxì A, Cammà C. A meta-analysis of survival rates of untreated patients in randomized clinical trials of hepatocellular carcinoma. Hepatology 2010;51:1274–83.
Rim CH, Kim CY, Yang DS, Yoon WS. Comparison of radiation therapy modalities for hepatocellular carcinoma with portal vein thrombosis: a meta-analysis and systematic review. Radiother Oncol. 2018;129:112–22.
Matsuo Y, Yoshida K, Nishimura H, Ejima Y, Miyawaki D, Uezono H, et al. Efficacy of stereotactic body radiotherapy for hepatocellular carcinoma with portal vein tumor thrombosis/inferior vena cava tumor thrombosis: evaluation by comparison with conventional three-dimensional conformal radiotherapy. J Radiat Res 2016;57:512–23.
Qi X, Guo X. Sorafenib for the treatment of hepatocellular carcinoma with portal vein tumour thrombosis: a systematic review of comparative studies. Prz Gastroenterol. 2015;10:142–7.
Hu J, Bao Q, Cao G, Zhu X, Yang R, Ji X et al. Hepatic arterial infusion chemotherapy using oxaliplatin plus 5-fluorouracil versus transarterial chemoembolization/embolization for the treatment of advanced hepatocellular carcinoma with major portal vein tumor thrombosis. Cardiovasc Intervent Radiol 2020;43:996–1005.
Wang Y, Zhu L, Lin X, He C, An Z, Tang J, et al. [Therapeutic Effect of CT-guided ¹²⁵I Seed Implantation on Advanced Lung Cancer and Pulmonary Metastatic Carcinoma]. Zhongguo Fei Ai Za Zhi 2020;23:424–8. Article in Chinese.
Wang Y, Zhu L, Lin X, et al. [Therapeutic Effect of CT-guided ¹²⁵I Seed Implantation on Advanced Lung Cancer and Pulmonary Metastatic Carcinoma]. Zhongguo Fei Ai Za Zhi. 2020 Jun 20;23(6):424-428. Chinese. doi: 10.3779/j.issn.1009-3419.2020.103.04. PMID: 32517444; PMCID: PMC7309549.
Yuan D, Gao Z, Zhao J, Zhang H, Wang J. 125I seed implantation for hepatocellular carcinoma with portal vein tumor thrombus: a systematic review and meta-analysis. Brachytherapy 2019;18:521–9.
Sun H, Zhang M, Liu R, Liu Y, Hou Y, Wu C. Endovascular implantation of 125I seed combined with transcatheter arterial chemoembolization for unresectable hepatocellular carcinoma. Future Oncol 2018;14:1165–76.
Li Y, Li H, Hu H, Yuan H, Zhao Y. Efficacy and safety of transcatheter arterial chemoembolization combined with either 125I seed implantation or apatinib in hepatocellular carcinoma with portal vein tumor thrombosis: a retrospective comparative study. J Cancer Res Ther 2020;16:1691–7.
Hu HT, Luo JP, Li HL, Guo CY, Yao QJ, Geng X, et al. Transarterial chemoembolization combined with computed tomography-guided 125iodine implantation enhances survival in hepatocellular carcinoma patients with portal vein tumor thrombus. Oncotarget 2017;8:29258–68.
Hu HT, Luo JP, Cao GS, Li Z, Jiang M, Guo CY, et al. Hepatocellular carcinoma with portal vein tumor thrombus treated with transarterial chemoembolization and sorafenib vs. 125Iodine implantation. Front Oncol 2021;11:806907.
Zhao XH, Yuan H, Xia WL, Zhang LL, Li Z, Cao GS, et al. Prospective study of TACE combined with sorafenib vs TACE combined with 125I seed implantation in the treatment of hepatocellular carcinoma with portal vein tumor thrombus and arterioportal fistulas. Front Oncol. 2022 Oct 5;12:977462.
Satala CB, Jung I, Kobori L, Kovacs Z, Fodor D, Szodorai R, et al. Benefits of the 8th American Joint Committee on Cancer system for hepatocellular carcinoma staging. J Gastrointest Cancer 2021;52:243–8.
Iasonos A, Schrag D, Raj GV, Panageas KS. How to build and interpret a nomogram for cancer prognosis. J Clin Oncol 2008;26:1364–70.
Jeong SO, Kim EB, Jeong SW, Jang JY, Lee SH, Kim SG, et al. Predictive factors for complete response and recurrence after transarterial chemoembolization in hepatocellular carcinoma. Gut Liver 2017;11:409–16.
Jung Y, Hu J. A K-fold averaging cross-validation procedure. J Nonparametr Stat 2015;27:167–79.
Wang Q, Xia D, Bai W, Wang E, Sun J, Huang M, et al. Development of a prognostic score for recommended TACE candidates with hepatocellular carcinoma: a multicentre observational study. J Hepatol 2019;70:893–903.
Guo L, Wang Z, Du Y, Mao J, Zhang J, Yu Z, et al. Random-forest algorithm based biomarkers in predicting prognosis in the patients with hepatocellular carcinoma. Cancer Cell Int 2020;20:251

No competing interests reported.

Prognosis nomogram for transhepatic arterial chemoembolisation with ¹²⁵I seed implantation for first-branch portal vein tumour thrombus hepatocellular carcinoma：a comparative study of two statistical analysis methods

Status:

Version 1

Abstract

Figures

1 Introduction

2 Materials And Methods

3 Results

4 Discussion

Declarations

References

Additional Declarations

Status:

Version 1