Introduction
Hepatocellular carcinoma (HCC) is a major end-stage liver disease with the sixth highest incidence rate and third highest cause of cancer-related mortality [
1,
2]. According to global cancer statistics for 2020, there were approximately 906,000 new cases and 830,000 HCC deaths. Early and accurate diagnosis and treatment are crucial for mitigating these threats.
Alpha-fetoprotein (AFP) is currently the most widely used biomarker for the diagnosis, monitoring, evaluation of prognosis, and treatment of HCC [
2‐
6]. The elevation of tumor markers, mainly AFP, is independently associated with poor prognostic features, such as shortened survival [
7‐
11]. However, in the HCC patient population, AFP-negative hepatocellular carcinoma (ANHC), i.e., AFP < 20 ng/mL at initial diagnosis, is an important type that causes many patients to lose early diagnosis and treatment and accounts for approximately 30-40% of patients [
12]. Therefore, the survival and prognosis of ANHC patients are still a matter of concern. Unfortunately, there are relatively few prognostic studies based on the ANHC cohort [
9,
13,
14].
We analyzed the impact of the patient’s general condition, tumor indications, and treatment methods on survival, identified relevant risk factors, and constructed a nomogram to guide prognostic evaluation and early diagnosis of ANHC.
Materials and methods
Research design and patients
This study collected the clinical data of patients with ANHC from the Surveillance, Epidemiology, and End Results (SEER) database. The inclusion criteria for this study were patients diagnosed between 2004 and 2015 with a primary tumor in the liver, and the third edition of the International Classification of Tumor Diseases was 8170/3: hepatocellular carcinoma, NOS. The confirmed cases included in this study required HCC to be confirmed and diagnosed through pathological examination and were initially diagnosed as AFP-negative (< 20 ng/mL). The exclusion criteria were AFP positivity (> 20 ng/mL) or unknown status at initial diagnosis, unknown tumor size, unknown marital status at diagnosis, unknown surgical treatment, and lack of complete survival time.
Data collection and statistical variable definition
The following clinical information was collected for further analysis: baseline demographic data, including age, ethnicity, gender, year of diagnosis, marital status, survival time, cancer-specific survival rate, and survival status; tumor characteristics, including tumor size (maximum diameter, in cm), pathological tumor grading; and TNM stage according to the American Joint Committee on Cancer 6th TNM stage; and treatment strategies, including surgery (liver resection and liver transplantation), chemotherapy, and radiotherapy at the primary site.
Gender was categorized as male or female. The year of diagnosis was categorized according to before and after 2009. Ethnicity was categorized into three racial groups: Caucasians, African Americans, and others. Marital status at the time of diagnosis was divided into married and unmarried/divorced/widowed groups. Surgery types were divided into no surgery, liver resection, and liver transplantation. For radiotherapy and chemotherapy, patients were classified as having, not having, or unknown.
Statistical analysis
We randomly divided all eligible patients with ANHC into two groups at 7:3: training cohort (n = 1267) and validation cohort (n = 544). Univariate COX analysis was conducted in the training cohort with overall survival (OS) as the endpoint, and statistically significant individuals (p < 0.1) were selected for the multivariate COX analysis. Furthermore, a nomogram was constructed using OS and cancer-specific survival (CSS) as endpoints (P < 0.05). The two nomogram groups were validated using a validation cohort. A nomogram model for independent prognostic factors was established based on multivariate Cox regression analysis (p < 0.05), and the fit was evaluated using a consistency index (C-index).
In the training and validation cohorts, calibration curves for the first, third, and fifth years were established based on the nomogram model by comparing the predicted and actual observed values of the nomogram in the OS and CSS groups. Calibration curves of the validation cohort were established in the same way. The Receiver Operating Characteristic (ROC) table and Decision Curve Analysis (DCA) evaluation model was fitted, under which diagnostic efficacy was compared with that of TNM staging. Clinical information extraction was performed using SEER*Stat software version 8.3.8 (
www.seer.cancer.gov/seerstat). Data analysis was carried out using R software version 4.2.2(R-Project,
https://www.r-project.org/). A
P-value < 0.05 was considered statistically significant.
Discussion
Based on SEER database, our study collected the largest cohort of AFP-negative patients available and constructed a new nomogram through data analysis. Compared with traditional TNM staging, it has improved effectiveness in evaluating patient prognosis. The independent risk factors included in our final nomogram were age, tumor diameter, ethnicity, pathological grade, surgical, chemotherapy, and radiotherapy conditions.
Patients with HCC constitute a substantial population characterized by a poor prognosis and a relatively short survival period. The 2022 Barcelona Clinic Liver Cancer guidelines propose that AFP is an important indicator that affects the treatment, prognosis, and clinical decision-making for HCC [
15]. The concentration of AFP in adults is about 5–10 ng/mL, and the commonly used optimal screening threshold is 16–20 ng/mL [
6,
7,
16,
17]. Nearly half of HCC patients are AFP-negative, most of which are early and small HCC; in advanced patients, 15-30% of serum have AFP values at the normal range (< 20 ng/mL) [
18,
19]. Therefore, the ANHC population should be non-negligible.
Usually, OS, with all-cause mortality as the endpoint, is widely used in clinical trials because of its better operability and excellent effectiveness. CSS considers tumor mortality as the endpoint event and has better specificity in evaluating specific diseases. However, sampling and follow-up collection are more difficult in clinical trials, and the SEER database compensates for this difficulty. Previous studies have analyzed HCC based on SEER databases, which extensively include AFP factors. The conclusion revolves around the correlation between AFP positivity and adverse prognosis, tumor progression and metastasis, premature death, and other outcomes [
8,
20‐
22]. However, few studies have been conducted in AFP-negative patients [
9,
13,
14], and most of which did not involve treatment. Our study collected data from all AFP-negative patients in the SEER database covering various treatment options, including surgery, radiotherapy, and chemotherapy.
The overall conclusions drawn by the two groups were similar but with slight differences. Various sociological factors influence the occurrence and prognosis of HCC [
23‐
25]. Our study showed that the CSS group was less affected by social factors and that the age and ethnicity factors included in the OS group were not statistically significant in the CSS group. It is worth mentioning that Chen et al. pointed out the impact of marital status on the prognosis of patients with HCC of TNM stage Ia in that positive marital status is an independent predictor of longer OS and reviewed previous relevant studies to support it [
26]. The overlap between the target populations of the two studies was relatively low. Therefore, the impact of marital factors may require a clear understanding of the patients’ scope of application. Tumor burden factors play a prominent role in predicting tumor prognosis. Tumor size and grade are independent prognostic indicators of OS [
13,
27,
28], and their impact on survival time is consistent with common sense. The difference between the two groups was relatively small. ANHC is more common in the early stages of the disease and has a better overall prognosis. Currently, the most beneficial treatment methods include surgery and radiofrequency ablation [
8,
11,
23,
29]. Particularly for stage 0 disease or patients without surgical contraindications, resection and transplantation are the preferred treatment methods and have a better long-term prognosis than radiofrequency ablation [
29‐
31]. Our results draw the same conclusion that the benefits of resection, especially liver transplantation, are significant. For example, a 60-year-old patient with stage a ANHC who undergoes resection can benefit from approximately 0.25 to 0.75 years of overall survival compared to refusing treatment; the five-year cancer-specific survival rate will increase from 0.3 to 0.8 in this scenario. Chemotherapy and radiotherapy have relatively little impact on the prognosis and survival time. The median survival period for symptomatic late-stage cases treated with systemic therapy is 1-1.5 years [
23]. Considering that both are mostly used for patients with late-stage disease, the extension of survival time is relatively low. However, it is still an important treatment method for unresectable tumors.
Tumor TNM stage is an independent risk factor for OS in patients with HCC [
28]. and it’s widely used in clinical evaluation in the therapy of various kinds of carcinoma. The DCA results indicated that the net benefit of the nomogram in prognostic evaluation was superior to that of TNM staging. And we analyzed the area under the curve (AUC) of ROC, which also indicates that our nomogram has better diagnostic efficiency than TNM staging. With improvements in surgical technologies and the development of new treatment modalities, the prognosis and overall survival of patients with HCC will gradually improve. Over time, the proportion of patients with HCC with lower AFP and earlier BCLC stage has increased, and overall survival has improved [
32].
Improvements in overall survival increased the number of potential patients and led to significant improvements in systemic therapy management [
32]. The cut-off point for the year of diagnosis in our study was 2009, which was the beginning of the advent of tyrosine kinase inhibitors (TKIs) led by sorafenib [
33]. Targeted medicines play an important role in improving patient survival. The application of neoadjuvant or conversion therapy effectively reduced the tumor burden before surgery [
34,
35]. It can also be administered after surgery or ablation to reduce the risk of tumor recurrence or progression [
36].
Starting in 2015, immunotherapy was introduced at this stage, and patient survival after this time point improved [
32]. For example, the IMbrave-150 study established the effectiveness of the classic treatment regimen of atezolizumab combined with bevacizumab, which can be used as the first-line treatment for most advanced HCC [
37]. Considering that the application of systemic therapy depends on excellent liver function, there is room for its application in patients with ANHC [
37,
38]. An important part of the ANHC patient population has BCLC stage B disease, which can also benefit from systemic therapy [
32,
39]. Unfortunately, the data of this study were obtained from the SEER database, and the immunotherapy or targeted therapy status of relevant patients was not collected.
There are still shortcomings in this study. We did not obtain other widely used treatment methods for HCC, such as ablation and interventional surgery, and the prognostic evaluation of these treatment methods may not be comprehensive. The main population in the SEER database comprises Caucasians, and its applicability to other countries or regions may be limited. Third, although T staging is determined based on tumor size, number, and extent of tumor invasion, the SEER database lacks key data on portal vessel invasion status, residual liver function, and underlying liver disease, etc. Therefore, some clinically critical tumor-related variable information, such as Child-Pugh class and MELD score, may be missing in our model. In addition, due to the retrospective nature of the study, we were unable to control the treatment of all patients, and their treatment may be susceptible to differences in residential location and time. More importantly, our prognostic model incorporated treatments, which may limit the utility of our study. In the future, a prospective multi-center clinical study on patients with AFP-negative HCC is still needed to incorporate more clinical parameters and further clarify the impact of these parameters on prognosis.
Publisher’s Note
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.