Liver cancer (LC) is the sixth most commonly diagnosed malignancy and the third leading cause of cancer-related death worldwide (1). In 2020, it was reported that the number of new cases and deaths of LC in the world were 905,677 and 830,180, respectively (1). Hepatocellular carcinoma (HCC) is the most common type of LC, which accounts for 75%–85% of cases. With the advancements in early diagnosis, cancer treatment, and cancer surveillance, the prognosis of HCC has greatly improved (2). For HCC patients diagnosed at an early stage, some of them could even have a 5-year overall survival (OS) of nearly 70% after operation (3). However, among HCC patients with a long-term survival, the second primary malignancy (SPM) is becoming a threat for their lives, and the studies on this problem are urgent. Recently, a growing number of studies have been performed to explore the risk factors of developing SPMs in various cancers, such as colorectal cancer (4), lung cancer (5), gastric cancer (6), and breast cancer (7).

The occurrence of SPMs in HCC was first reported by Riesz et al. (8) in 1979. Afterwards, several studies were published to describe this problem (9–11); however, these studies were limited by the small number of patients and different study design. Recently, using the public database, two population-based studies were published to describe the development of SPMs in HCC (12, 13); however, the conclusions we could get from the two studies were limited. The study by Shan (13) only focused on the simple description of SPMs in HCC, while in the study by Wu (12), only four variables including age, sex, treatment, and comorbidities were included in the exploration of the risk factors of developing SPMs. Less is known about the influence of clinical features including race, laboratory data, and tumor characteristics (such as size, vascular invasion, and stage) on the development of SPMs in HCC.

In this study, using the data obtained from the Surveillance, Epidemiology, and End Results (SEER) research database, we aimed to comprehensively analyze the clinicopathological characteristics of HCC patients with SPMs and to explore the risk factors associated with the development of SPMs. Afterwards, a competing-risk nomogram was established to predict the probability of developing SPMs for HCC patients, and the Harrell’s concordance index (C-index), calibration curves, and decision curve analysis (DCA) were used to evaluate the performance of the nomogram.

The number of cancer survivors is growing; in the U.S., it was reported that the number of cancer survivors was estimated to reach 26.1 million by 2040, which approximately accounted for 6.9% of the total population (22). Due to the special disease history, many factors are threats for the health of the survivors, such as the recurrence of cancer (23) and treatment-related toxicities (24); meanwhile, SPMs, with a growing incidence, is becoming a threat on the health of cancer survivors (25, 26). It was reported that the incidence of SPMs ranged from 1% to 16% according to various cancer types (27). However, cancer survivors were frequently excluded from cancer clinical trials, and little is known about their characteristics, risk factors, and survival (28–30). Consequently, exploring the characteristics of SPMs, identifying the risk factors associated with the development of SPMs and screening patients at high risk of developing SPMs are of both a clinical and public health importance. In this study, we identified 1,593 patients (total population: 40,314) who developed SPMs after the initial diagnosis of HCC. In the further analysis, we found that compared to those in the OOPM group, patients with SPMs had better tumor-related characteristics and had a higher proportion of receiving tumor-related treatment. Furthermore, using the Fine-Gray proportional subdistribution hazards model, risk factors associated with the development of SPMs were explored and a competing-risk nomogram was established to predict the 3-, 5-, and 10-year probability of developing SPMs for HCC patients after the initial diagnosis. The C-index, calibration curves, and DCA all showed that the nomogram had a good performance. To our knowledge, this study provided the largest population-based cohort study of SPM incidence in HCC patients.

The incidence of SPMs in HCC patients was 3.95% in this study. In contrast, the incidence of SPMs in a previous US population-based study which included patients from 1992 to 2011 in the SEER database was reported to be 2.83% (13). Meanwhile, the incidence of SPMs was various in previous studies of different countries, which was around 3.5%–7.5% in the USA (10, 31), 2.4% in Spain (32), 1.63%–8.0% in China (12, 33), and 0.7%–1.9% in Japan (34, 35). This might be influenced by the difference in the demographic characteristics, etiologies, tumor characteristics, and treatment strategies, which were also analyzed in our study. In the analysis of the risk factors of developing SPMs among different subgroups, we could find that the development of SPMs was greatly influenced by the factors including vascular invasion, stage, treatment, tumor size, sex, and age at initial diagnosis. It is clear that the long-term results of HCC survivors could be different depending on these factors.

The top five most common sites of developing SPMs were the lung and bronchus, prostate, non-Hodgkin lymphoma, colon, and breast. Consequently, for HCC survivors at a high risk of developing SPMs, regular and long-term surveillance for lung and bronchus, prostate (male patients), and breast (female patients) was necessary, and there might be a need for a discussion by the multidisciplinary team. Furthermore, in the survival analysis, we found that the patients in the SPM group had a better survival than those in the OOPM group. However, for HCC patients with SPMs after the second diagnosis, they had a comparable survival with those in the OOPM group. We also found that 82.84% patients died of cancer in the OOPM group while only 46.37% patients in the SPM group died of cancer. We thought that the main reason was that patients in the OOPM group were associated with poorer tumor-related characteristics and had a low proportion of receiving tumor-related treatment, consequently, these patients were more likely to have a poorer prognosis and die of cancer.

Since a large proportion of patients died before the development of SPMs, it is necessary to conduct a proper statistical method to handle this competitive event. Using the Fine and Gray proportional subdistribution hazards model, we explored the risk factors of developing SPMs from the demographic and clinical variables. The multivariate analysis suggested that patients with SPMs tend to have an older age at the initial diagnosis, a local or regional disease, and a smaller tumor size. These variables have been found to be associated with an increased risk of developing SPMs in other cancers (36–38). Furthermore, patients who received any tumor-related treatment had a higher risk of developing SPMs, especially for those receiving transplantation. Several reasons could explain our results: first, the immune system has been reported to be related to cancer development, and it was reported that older age was associated with immunity decreasing (“immunosenescence”), which might contribute to an increased cancer incidence (39). Second, since HCC was a malignancy with a poor prognosis, HCC survivors with better prognostic factors were believed to have a better survival, and thus more likely to have an adequate time to develop SPMs (4, 22). Third, for patients who received treatment for HCC, they might have a higher susceptibility to tumorigenesis, especially for those who received liver transplantation, the administration of immunosuppressive therapy might contribute to the development of SPMs (40–42).

Considering the elevated risk of developing SPMs after the initial diagnosis of HCC, a competing-risk nomogram was established for the purpose of improving guideline surveillance and further management of HCC survivors. To our knowledge, this is the first model for the prediction of developing SPMs for HCC after the initial diagnosis, and our nomogram could be useful to predict the 3-, 5-, and 10-year probability of developing SPMs for HCC survivors. Furthermore, the performance of our nomogram was also demonstrated well using the C-index and calibration curves, and DCA further suggested that the nomogram had a superior net benefit in clinical scenarios. Consequently, our nomogram is potentially useful for doctors to screen HCC survivors who are at a high risk of developing SPMs and could contribute to the further management of SPMs.

There are several limitations in our study. First, due to the nature of the SEER database, several important cancer-related risk factors, such as lifestyle, alcohol and tobacco consumption, and family history were unavailable; meanwhile, for HCC, several well-known prognostic factors, including cirrhosis, portal hypertension, postoperative complication, and microenvironment-related immune and inflammatory markers (43, 44) were also unavailable. These might lead to a bias of the results. However, our study still found several risk factors of developing SPMs for HCC and our predictive model showed the ability to discriminate patients who are at a high risk of developing SPMs. Second, the cumulative incidence of SPMs could be overestimated when extrahepatic metastasis was regarded as SPMs or the initial simultaneous cancers could not be distinguished using a 2-month latency exclusion. Certainly, the diagnosis time of an SPM might not be the accurate time when it occurred, and these could be solved with the advancement of detection technology and regular surveillance. Third, although our nomogram had a good performance, further validation with external populations is still needed.

In conclusion, we demonstrated that age at diagnosis, tumor stage, tumor size, and treatment were associated with an increased risk of developing SPMs for HCC patients after the initial diagnosis. Afterwards, a reliable competing-risk nomogram was established to predict the 3-, 5-, and 10-year probability of developing SPMs for HCC survivors. Our findings could be useful to clinicians for the surveillance and management of HCC patients. Further exploration involving more patients and more risk factors should be performed, and better surveillance should be conducted for HCC patients who are at a high risk of developing SPMs.

The raw data supporting the conclusions of this article will be made available by the authors, without undue reservation.

GY, WS, and JL contributed to the conception and design of the study. JK, WS, JC, and GL collected the details of the patients and extracted the data. JK, YL, and JC analyzed the data. JK, GY, and GL drafted the manuscript. JL, WS, JC, and YL contributed with a critical revision of the manuscript. All authors contributed to the article and approved the submitted version.

This study was supported by the National Natural Science Foundation of China (No. 81373172, 81770646).

The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

All claims expressed in this article are solely those of the authors and do not necessarily represent those of their affiliated organizations, or those of the publisher, the editors and the reviewers. Any product that may be evaluated in this article, or claim that may be made by its manufacturer, is not guaranteed or endorsed by the publisher.