Introduction
Hepatocellular carcinoma (HCC) is one of the most dominant malignant tumours in the world, and there are more than 840,000 new cases and over 780,000 deaths per year [
1,
2]. Risk factors for HCC include hepatitis B virus (HBV) and hepatitis C virus (HCV), alcohol addiction, nonalcoholic fatty liver disease, obesity, diabetes mellitus, moldy food containing aflatoxin and so on [
3]. Patients infected with HBV or HCV accounted for over 80% of HCC cases [
4]. However, changing lifestyles and increasing HBV vaccination rates and more efficacious antiviral treatments have changed the global epidemiology of HCC [
5]. The incidence of non-virus-related HCC (NV-HCC) is increasing due to fatty liver disease, obesity and insulin resistance [
6]. Studies have shown that HBV deoxyribonucleic acid (DNA) may lead to worse liver function and more complications [
7]. Thus, HCC patients with HBV have lower overall survival (OS) and disease-free survival (DFS) than NV-HCC patients [
8,
9]. These results indicate that NV-HCC and hepatitis B-positive HCC have different clinicopathological features, prognostic factors and clinical outcomes. Thus, a distinct prognostic model for NV-HCC is needed. Recently, numerous prognostic survival models have been established for HCC patients [
10,
11]. However, there are few reports on prognostic models for NV-HCC. Therefore, to facilitate clinical counseling and the individualized prediction of survival for NV-HCC, it is necessary to construct a new prognostic model to assess the specific prognosis of NV-HCC.
Numerous studies have reported that clinical characteristics and routine laboratory examinations of blood are prognostic predictors for HCC, including tumor size, HBV DNA [
12], alpha-fetoprotein (AFP) [
13], neutrophil/lymphocyte ratio (NLR) [
14], and alkaline phosphatase (ALP) [
15]. Increasing AFP levels were associated with worse survival and higher recurrence rates in patients with HCC [
16]. Witjes et al. reported that high AST levels were linked to worse OS in patients with HCC [
17]. The NLR is a prognostic factor affecting survival and recurrence in living-donor liver transplantation for HCC [
18]. Based on ALP, tumor size, liver cirrhosis, microvascular invasion, and other factors, a nomogram was established to evaluate the prognosis of HCC [
12]. However, it is a challenge to screen and combine multiple factors into a prognostic system for NV-HCC.
Discussion
HCC is a leading cause of cancer related death worldwide [
20]. Most HCC patients have a poor prognosis, and the 5-year OS is only 12.1% [
21]. HBV and HCV infection are major important risk factors for HCC. Recently, with lifestyle changes and efficient vaccination strategies, the number of HCC virus patients has decreased, and the number of NV-HCC patients is increasing [
5]. In Taiwan, due to the universal newborn vaccination program, HCC incidence is significantly lower in younger persons who were vaccinated than in those who were not vaccinated at birth [
22]. However, there are few studies predicting the occurrence of NV-HCC. TNM stage is commonly used to predict the prognosis for many cancers [
23,
24]. However, studies reported that patients with the same TNM stage had different clinical outcomes [
25]. This phenomenon indicates that the TNM stage utilized for guidance of the different treatments is insufficient. Therefore, we developed a prognostic model for NV-HCC. LASSO-Cox regression is a useful tool for feature selection and regularization to improve the accuracy of statistical models [
26]. In this study, we constructed a prognostic model for NV-HCC to further guide clinical treatment by using LASSO-Cox regression analysis of the pathological results and clinical laboratory test results.
In this study, by using LASSO-Cox regression analysis, six predictive indicators (number of nodules, LMR, PNI, ALP, SLR and CRP) were selected for the prediction of NV-HCC prognosis. Then, we constructed a prognostic model based on the six factors for NV-HCC patients. The risk score was calculated as follows: Risk score = (0.4419 * number of nodules) + (− 0.0156 * LMR) + (− 0.005 * PNI) + (0.001 * ALP) + (0.1301 * SLR) + (0.0001 * CRP). Based on the risk score, the NV-HCC patients were divided into a low-risk group (risk score < 0.59) and a high-risk group (risk score ≥ 0.59). Kaplan–Meier curves revealed that the high-risk group of NV-HCC patients had a poor OS (P < 0.001). The prognostic score model achieved a higher AUC than the TNM stage and treatment for the 1-year OS, 3-year OS, and 5-year OS. Moreover, we constructed a nomogram that can help to predict OS in NV-HCC patients, which integrated the prognostic score, TNM stage and treatment. Notably, according to our study, the nomogram model was a more powerful predictive factor of OS for NV-HCC patients than the prognostic model, and the C-index of the nomogram model (0.78 and 0.85) was higher than the C-index of the prognostic risk score model (0.76 and 0.80) in the development cohort and in the validation cohort. Moreover, in the development cohort and validation cohort, the heatmaps and waterfall plots of the clinical features also indicated that patients who had shorter OS were mainly distributed in the high-risk group, TNM stage III or IV group, and treatment with chemotherapy group.
The prognosis of NV-HCC patients is closely related to the number of nodules in the liver. Mazzotta et al. reported that patients who had more than 5 HCC nodules during the waiting period had a high risk of post liver transplantation recurrence and death [
27]. Markers of the inflammatory response, including LMR, SLR, lymphocytes, NLR, and CRP, play important roles in the progression of many cancers [
28,
29]. Studies have revealed that LMR is associated with survival in patients with breast cancer, and a low LMR indicates poor prognosis in stage I–III breast cancer [
30]. LMR markedly increased the level of tumour-infiltrating Th17 cells and promoted tumour growth in HCC [
31]. Serum ALP, AST, ALT, and CRP are biomarkers of systemic inflammation and immune activation, and can be used to evaluate liver function [
32]. The elevation of ALP has been demonstrated to predict poor prognosis in esophageal squamous cell carcinoma and pancreatic cancer [
33,
34]. Moreover, ALP was incorporated into prognostic models for many cancers, including HCC and gastric cancer [
35,
36]. Our previous study showed that the LSR is an independent prognostic factor for gastric cancer [
37]. We also established a nomogram based on age, stage status, and SLR, which had a more accurate prognostic prediction for patients with gastric cancer [
38]. CRP is an indicator of inflammatory response, which combined with increased cytokines, growth factors, activated stroma, and DNA damage, promotes tumour invasion, migration and metastasis [
39]. Currently, a low PNI has been shown to be a significant predictor of poor postoperative outcomes and increased mortality in various malignancies, including colorectal cancer, breast cancer, and pancreatic cancer [
40‐
42]. In this study, we used LASSO-Cox regression analysis to identify that the number of nodules, LMR, PNI, ALP, SLR and CRP levels can be used to predict the prognosis of NV-HCC.
There are some limitations in this study that should be noted. First, this model employed data from one medical centre. Multicentre data are needed to further verify the performance of the model. Second, this study has small sample size of NV-HCC patients in the development and validation cohorts. Therefore, a larger cohort is urgently needed to further verify the model of our study. In addition, this study only analysed the OS of NV-HCC patients, and it is uncertain whether DFS and progression-free survival (PFS) can be verified.
Publisher's Note
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.