Globally, the incidence of liver cancer is relatively high. Although improvements in treatment techniques and strategies in recent years have markedly improved the cancer-specific survival (CSS) and overall survival (OS) of liver cancer, the long-term survival rate of liver cancer is still extremely low[21]. The 1-year, 3-year, 5-year overall survival rates were 60.5%, 27.6%, and 8.4%, respectively[22]. However, early death, which was defined as death within 3 months after the initial diagnosis, was more likely to appear in advanced HCC. In this study, the data in the SEER database showed that among patients with stage IVA and IVB, the incidence of early death was 44.6% and 62.8%, respectively. At present, most studies were based on the exploration of early hepatocarcinoma and the long-term survival rate of patients, while there were few studies on the early death of patients with advanced HCC. Therefore, developing a series of prediction tools to identify the risk factors and predict the probability of early death to guide clinical treatment is quite valuable.
Although the AJCC staging system has been commonly used in the prognosis assessment of primary liver cancer, it had certain limitations as some crucial risk factors such as age, gender, race, histological grade, and treatment solutions were not included. The innovation of our research lies in the development of nomograms for early death of advanced HCC for the first time that includes the above risk factors. Nomogram is a straightforward and accessible visual tool, which has been widely used for malignant tumor risk and prognosis assessment in recent years[23]. Its simplicity and intuitiveness are conducive to busy clinicians to identify high-risk patients faster and more accurately, and then choose precise and individualized treatment plans. The SEER database has the characteristics of a large sample size and complete patient follow-up information. Therefore, the nomograms based on the SEER database were more exact and stable[24, 25].
In this study, regardless of stage IVA or stage IVB patients, age was significantly associated with early death, which was consistent with the results of previous studies[26, 27]. From the SEER database data and the existing research, it is obvious that HCC is more likely to occur in male[28], however, gender was not a risk factor for early death of advanced HCC patients. In stage IVA patients, the probability of early death in male and female patients was 45% and 42%, and that of patients in male and female with stage IVB was 63% and 64%, respectively. Also, the results showed that race and marital status were not the risk factors for early death of patients. Except for the above demographic characteristics, the research also showed that in patients with stage IVA, larger tumor size, higher histological grade, AFP positive, higher fibrosis scores, T3 ~ 4 stage, and patients who had not received surgery, radiotherapy, or chemotherapy were in the higher risk of early death. In patients with stage IVB, higher histological grade, AFP positive, T3 ~ 4 stage, N1 stage, bone metastasis, lung metastasis, and those without surgery, radiotherapy, or chemotherapy tended to suffer an early death.
It has been reported that tumor size was an independent risk factor for HCC recurrence and death after HCC resection. Taking 5 cm as the boundary, the 5-year recurrence-free survival (RFS) and overall survival (OS) rates in the ≤ 5 cm group were 38.3% and 61.5%, while those in the > 5 cm group were 25.1% and 59.9%[29]. However, our study showed that tumor size was not a risk factor for early death in patients with stage IVB. This may be because stage IVB HCC was mostly due to the recurrence of early liver cancer after surgery. At this time, the disease progressed rapidly, and the tumor size may have less impact on the early death of patients. Our research also illustrated that, in stage IVB patients, fibrosis score was also not a risk factor for early death. Liver fibrosis was a chronic inflammatory process, and it has been reported that liver fibrosis had no effect on OS and RFS until developed into liver cirrhosis[30]. However, stage IVB HCC had the characteristics of rapid progression and extensive invasiveness, and cirrhosis had no chance to develop, which might be why fibrosis score was not a risk factor for early death in patients with stage IVB. [31]. The brain was one of the most likely sites for metastasis in patients with advanced HCC[32–34], but the results in our research did not show that brain metastasis was significantly associated with early death of patients. This might be relevant to the small number of brain metastasis in liver cancer and the insufficient sample size.
Nowadays, liver cancer has entered a multimodal diagnosis and treatment era. In addition to clinically commonly used ultrasound, magnetic resonance imaging, and computed tomography, tumor markers were also playing a momentous role in the diagnosis of liver cancer[35]. Tumor markers that contribute to the diagnosis of HCC include alpha-fetoprotein (AFP) heterogeneity, Glypican-3, osteopontin, Des-γ-carboxyprothrombin, Golgi protein-73, abnormal pro-thrombin, and heat shock protein[36]. Among them, AFP is the most widely accepted serum biomarker for the diagnosis of HCC, however, the specificity and sensitivity of this tumor marker were 72–90% and 39–65%, respectively. Besides, the early diagnosis efficiency of AFP was only 9–32%, and cholangiocarcinoma did not express AFP, which limited its clinical use[37, 38]. Therefore, if these newly discovered tumor markers can be combined with epidemiology and clinical pathology, it will help establish a more accurate prediction model to guide the individualized treatment of HCC.
Sorafenib is currently considered to be the standard frontline therapy for advanced HCC. However, a great number of adverse events mainly including gastrointestinal or skin diseases have been found in patients taking sorafenib. In some severe cases, sorafenib can cause high blood pressure and abdominal pain, leading to interruption of treatment[39]. Even so, only approximately 30% of patients with advanced HCC can benefit from sorafenib, and these people usually develop drug resistance within 6 months[40]. Therefore, for the treatment of advanced HCC, it is urgent to explore new drugs and further understand the process of tumor resistance.
It is still controversial whether HCC patients with lymph node invasion should accept surgery treatment. However, in this study, the outcome indicated that surgery had an important effect on the improvement of early death in advanced HCC. Moreover, previous studies have stressed that surgery had beneficial value for advanced HCC patients, especially for that with regional lymph node invasion[41]. Therefore, although the guidelines recommend targeted therapy as the frontline treatment for advanced HCC, it might be more targeted at patients with stage IVB. For patients with stage IVA who only had regional lymph node metastasis, surgery might also be a better treatment option. However, considering a small number of patients undergoing surgery in our study, it might be more prudent to set strict indications for surgery in advanced HCC based on the clinical conditions of the patients. Furthermore, large-scale prospective studies are required to prove the surgical value of advanced HCC.
Inevitably, there are several limitations in this study. First, although the SEER database provides a large enough sample size, it still lacks some potential risk factors, which may be related to early death, such as the specific location of regional lymph node metastasis, the patients' past medical history, adverse habits (drinking and smoking history), postoperative tumor remnants, the specific methods of radiotherapy and chemotherapy, and other tumor markers other than AFP. Second, the SEER database has limitations in HCC tumor staging; the AJCC staging system lacks some important prognostic information, including Child-Pugh classification and the patients' performance status, which makes it not widely endorsed for HCC. It is possible that the patients in our research may include Child-Pugh class C, who were in Barcelona Clinic Liver Cancer (BCLC) stage D disease, and surgery, radiotherapy, and chemotherapy were not recommended. Third, as a retrospective study, selection bias caused by censoring data is unavoidable. Last, although internal verification suggested that the nomograms had good predictive capabilities, it still requires multiple centers and large sample size data for external verification to avoid overfitting.