Nonlinear relationship between age and the risk of hepatocellular carcinoma in patients with hepatitis B virus-related cirrhosis.

Introduction
Persistent hepatitis B virus (HBV) infection remains a predominant factor contributing to the incidence of hepatocellular carcinoma (HCC) in Asia and Africa [1]. The use of nucleoside analogues (NAs) has been documented to reduce the risk of HCC by a range of 45% to 63% [1–4]. However, due to advancements in the treatment of decompensated cirrhosis, the effects of cohort dynamics, and population aging, a significant proportion of individuals with HBV-related cirrhosis continue to progress to HCC [5]. Therefore, it is crucial to further enhance therapeutic interventions and implement tailored monitoring measures for such patients.
In recent years, the prolonged application of the next generation of NAs has significantly improved patient prognosis, and interferon is also acknowledged for its potential to reduce the occurrence of HCC [6–8]. Besides these two categories of drugs, despite various new-generation therapeutic drugs undergoing clinical trials, widespread clinical application is still at a considerable distance [8]. Therefore, the key to further improving measures aimed at reducing HCC currently lies in optimizing the implementation of existing protocols.
As a crucial demographic characteristic, the disease features and clinical prognosis of chronic HBV infection vary across different age groups. For individuals with chronic hepatitis B in distinct age brackets, clinical practice should employ varied treatment and follow-up strategies to achieve the goal of refined disease management. Previous studies have posited that age is a pivotal factor influencing the occurrence of HCC [9, 10]. Current guidelines also recommend HCC surveillance tailored to different age groups [11, 12]. However, there are variations or ambiguities among different guidelines regarding the specific age groups that should be the focal point of attention. Simultaneously, further scrutiny is required to determine whether the relationship between age and the risk of HCC is entirely linear.
Herein, the present study aimed to elucidate the relationship between age and the risk of HCC in individuals with HBV-related cirrhosis, a crucial aspect for the refined disease management tailored to distinct age groups.

Patients and methods

Study population
This retrospective cohort study comprised individuals with HBV-related cirrhosis recruited between January 2016 and December 2023 from two participating medical institutions: the Second Hospital of Shandong University and Weifang People’s Hospital. All enrolled patients underwent treatment with first-line potent NAs within 6 months before enrollment or promptly after enrollment. The inclusion criteria were as follows: (1) individuals aged ≥18 years; (2) confirmed diagnosis of cirrhosis based on clinical, radiological, endoscopic, or histological evidence. Exclusion criteria encompassed: (1) co-infection with hepatitis C virus or human immunodeficiency virus; (2) presence of alcoholic liver disease, severe non-alcoholic fatty liver disease, drug-induced liver disease, autoimmune liver disease, or other liver disorders; (3) prior history of HCC; (4) development of HCC or mortality within 6 months after enrollment; (5) follow-up duration of less than 12 months. Ethical approval for the study was obtained from the Second Hospital of Shandong University, conducted in adherence to the principles of the Helsinki Declaration.

Data collection
All data were extracted from electronic medical records. Baseline data was defined as information collected at the commencement of antiviral therapy. Demographic characteristics encompassed age and sex. Laboratory data included platelet counts (PLT), alanine aminotransferase (ALT), gamma-glutamyl transpeptidase (GGT), total bilirubin (TBIL), albumin (ALB), creatinine, serum sodium (Na), alpha-fetoprotein (AFP), international normalized ratio (INR), HBeAg status, and serum HBV DNA (with a lower limit of detection of 20 IU/mL). Imaging data included ultrasonography, computed tomography (CT), or magnetic resonance imaging (MRI). Clinical data covered complications associated with cirrhosis (ascites, hepatic encephalopathy, and esophageal variceal bleeding) and comorbidities such as diabetes. The Model for End-Stage Liver Disease (MELD) and Child-Pugh scores were computed using relevant data.

Definitions
Liver cirrhosis was diagnosed based on the presence of any of the following criteria: clinical manifestations of portal hypertension, ultrasonography/CT/MRI findings indicative of cirrhosis (surface nodularity, fissure widening, enlargement of lateral segments of the left lobe and caudate lobe, splenomegaly, collateral venous circulation, and an enlarged portal vein), and endoscopic or histopathological evidence. To assess the severity of cirrhosis related to HCC, cirrhosis was classified into three types: compensation, recompensation, and decompensation. The compensation group comprised patients who maintained a compensated cirrhosis state throughout the follow-up period. The definition of recompensation is primarily based on the Baveno VII criteria which requires meeting all the following criteria: (1) sustained viral suppression for hepatitis B: undetectable serum HBV DNA ( < 20 IU/ml) or HBsAg seroclearance; (2) resolution of ascites (discontinuation of diuretics) and HE (discontinuation of lactulose/rifaximin), and the absence of recurrent variceal bleeding for at least 12 months; and (3) stable improvement in liver function tests (MELD score < 10 and/or Child-Pugh Class A) [13, 14]. Decompensation included patients who encountered at least one decompensating event during the study period and did not achieve recompensation [15].

Study outcomes
The primary outcome was the occurrence of HCC. Regular HCC surveillance was conducted for all patients at intervals of no more than 6 months, involving abdominal imaging and AFP measurements. Diagnosis of HCC relied on either histological confirmation or met specific imaging criteria determined through dynamic CT and/or MRI, including the presence of a nodule > 1 cm with arterial enhancement followed by portal/delayed-phase wash-out. Data were censored at the date of the last follow-up, the occurrence of HCC, or death, whichever occurred earliest.

Statistical analyses
Continuous variables were presented as mean±standard deviation (SD) or median (interquartile range [IQR]), and comparisons between groups were conducted using either one-way ANOVA or Kruskal-Wallis test, depending on the appropriateness of the data distribution. Categorical variables were described in terms of frequencies and percentages and analyzed using Chi-squared test or Fisher’s exact test. The cumulative incidence of HCC was assessed using both unadjusted and adjusted Kaplan-Meier methods. To investigate the association between age and HCC risk, univariate and multivariate Cox proportional-hazards regression models were applied. Hazard ratios (HR) with 95% confidence intervals (CI) were reported. Restricted cubic spline regression was utilized to explore potential nonlinear relationships between age and HCC risk [16]. Nonlinearity was assessed using the likelihood ratio test. Additionally, a two-piecewise Cox proportional-hazards regression model was employed to elucidate any nonlinear relationships between age and HCC risk. To identify subpopulations susceptible to demographic-related disparities, stratified analyses were conducted based on sex and different types of cirrhosis. Sensitivity analysis was performed by excluding data from a single center. All statistical analyses were carried out using R Statistical Software (Version 4.3.2, http://www.R-project.org, The R Foundation) and the Free Statistics Analysis Platform (Version 1.8, Beijing, China). Two-tailed p-values less than 0.05 were considered statistically significant.

Results

Clinical characteristics of the included population
A total of 498 individuals with HBV-related cirrhosis undergoing first-line potent NAs therapy were initially screened, with 434 eventually included in the final cohort (Fig. 1). The mean age of the entire cohort was 51.2 years, and 66.8% were male. The average Child-Pugh and MELD scores were 6.9 and 11.3, respectively. Among these patients, 239 (55.1%) exhibited compensated cirrhosis, 108 (24.9%) demonstrated recompensated cirrhosis, and 87 (20.0%) manifested decompensated cirrhosis during the study period.
Age at baseline was stratified into three categories based on the median values and the threshold derived from restricted cubic spline regression as follows (one inflection point was identified at age 59 based on the peak of the RCS curve, and another at age 51 based on the median value): the youth group (≤50 years old), the middle-aged group (51–59 years old), and the elderly group (≥60 years old). Patients below the age of 50 exhibited a higher likelihood of being male, elevated ALT, GGT, and ALB levels, as well as lower Child-Pugh scores. Detailed baseline characteristics are outlined in Table 1.
At a median (IQR) follow-up of 43.0 (25.0–57.0) months, 45 patients developed HCC, with 12 (5.8%) in the youth group, 18 (14.5%) in the middle-aged group, and 18 (14.6%) in the elderly group (Table 1). Additionally, 22 patients succumbed to complications of cirrhosis.

Non-linear relationship between age and HCC risk
The adjusted multivariate Cox regression models with penalized splines revealed a non-linear association between age (as a continuous variable) and the risk of HCC (p for nonlinearity = 0.017, Fig. 2). According to the two-piecewise Cox proportional hazards regression model, age was associated with HCC risk in a non-linear manner, with the threshold value identified as 59 (Fig. 2, Tabel 2). Below the threshold, age was linked to an increased risk of HCC (aHR 1.104, 95% CI 1.032–1.181; p = 0.004), while this heightened risk persisted above the threshold (aHR 1.010, 95% CI 0.921–1.108; p = 0.833; Table 2).We examined the association between baseline variables and the risk of HCC (Table S1). Both univariate and multivariate Cox regression analyses indicated that age and cirrhosis stage were significant risk factors for HCC. Given that male sex is a well-established risk factor for HCC, stratified analyses were performed based on sex and different types of cirrhosis, revealing that all p-values for interaction were above 0.05, indicating a stable non-linear pattern and a consistent threshold value (Figs. 3A and 3B).

Association of different age groups with HCC risk
The cumulative incidence of HCC development at 2, 4, and 6 years was 1.6%, 4.3%, and 19.1% in the youth group, 5.1%, 13.7%, and 34.1% in the middle-aged group, and 6.1%, 10.7%, and 34.3% in the elderly group, respectively. In unadjusted Kaplan-Meier analyses, the youth group demonstrated lowest cumulative incidence of HCC, while the risk was not significantly different between the middle-aged and the elderly groups (Figure S2A, Table 3). A similar result was obtained after adjusting for male sex, types of cirrhosis, PLT, GGT, Child-Pugh grade, HBeAg status, and diabetes in adjusted Kaplan-Meier analyses (Figure S2B).
A crude and fully adjusted association between age and HCC risk is presented in Table 3. In the unadjusted Cox regression model, compared to the middle-aged group, the youth group exhibited a lower likelihood of HCC development (HR, 0.3; 95% CI 0.14–0.63; p = 0.001), while the elderly group had a similar HCC risk (aHR, 0.96; 95% CI 0.48–1.91; p = 0.899). These associations remained consistent in both the minor adjusted model and the fully adjusted model (Table 3).

Sensitivity analyses
To ensure the robustness of the findings, a sensitivity analysis was conducted, excluding data from 52 patients at Weifang People’s Hospital. The adjusted multivariate Cox regression models with penalized splines continued to demonstrate a nonlinear association between age (as a continuous variable) and the risk of HCC (p for nonlinearity = 0.013, Figure S1A). The threshold value was approximately 59 based on the two-piecewise Cox proportional hazards regression model (Figure S1A, Table S2). The youth group exhibited the lowest cumulative incidence of HCC, while the risk did not differ between the middle-aged and elderly groups in both unadjusted and adjusted Kaplan-Meier analyses (p = 0.002, Figure S1B, C). Similar results were obtained in the unadjusted and adjusted Cox models (Table S3). These findings suggest that age is a significant risk factor for HCC development, and it maintains a non-linear relationship with HCC risk even after excluding patients from a single center.

Discussion
This study reveals a non-linear, parabolic-like relationship between baseline age and HCC risk during long-term NAs treatment in individuals with HBV-related cirrhosis. For patients aged 59 years and below, increasing age is associated with a rising risk of HCC, while above the threshold, this association lacks statistical significance, and heightened risk persists. Patients aged 50–59 years exhibited the highest constituent ratio. These results suggest that the early initiation of antiviral therapy is preferable to mitigate the progressively increasing risk of HCC in patients with hepatitis B-related cirrhosis. Simultaneously, heightened surveillance for HCC is recommended among individuals above 50 years of age, particularly in the age group of 50–59 years.
To our knowledge, this is the first cohort study to examine the potential nonlinear relationships between age and HCC risk using restricted cubic spline regression. It is well-known that old age is a significant demographic risk factor for HCC. Multiple multi-parameter HCC prediction models, including THRI (Toronto HCC Risk Index), HBV-related REACH-B, PAGE-B, and SAGE-B, as well as the aMAP model, applicable for predicting HCC risk associated with various etiologies, all incorporate age as a contributing factor [9, 17–20]. These models indicate that with increasing age, there is a corresponding escalation in the risk of liver cancer. In statistical methodology, previous studies have commonly employed Cox regression, treating age as a univariate, binary, or multiclass variable for analysis. However, a more nuanced examination is warranted to ascertain whether the association between age and HCC risk adheres strictly to a linear pattern. In this study, restricted cubic spline regression was utilized. We found that the risk of HCC may increase irreversibly during the early stages, progressing rapidly with a steeper slope. Those with a baseline age of 50–59 years demonstrated the highest on-treatment HCC risk. When the age exceeds 59, the risk of HCC remained at a relatively high level. Although there appeared to be a declining trend in the risk of HCC with increasing age, this trend lacks statistical significance. These findings suggest that, in patients with HBV-related cirrhosis, the risk of HCC increases significantly with age before 59 years, whereas it remains stable thereafter. This indicates that initiating antiviral therapy before the age of 59 may mitigate the age-related increase in HCC risk, potentially providing greater clinical benefit. In contrast, the benefit of antiviral therapy may be relatively limited when initiated after the age of 59. Our results are consistent with previous epidemiological investigations of HCC. Cancer registry data from the International Agency for Cancer Research’s Cancer Incidence in Five Continents (CI5) database from 1978 through 2012 showed incidence rates of HCC and age are directly correlated until approximately 75 years of age in most populations [21]. Nevertheless, in Chinese and black African populations, the mean age of patients with the tumor is appreciably younger [22, 23]. The HCC-BRIDGE study analyzed 18,031 HCC patients from 14 countries. The mean ages of these patients were 69, 65, and 62 years in Japan, Europe, and North America, respectively, while in China and Korea, the mean ages were 52 and 59 years [24]. This may be attributed to variations in the etiology of HCC in different regions, as well as differences in the age of individuals at the time of hepatitis B or C infection.
Antiviral therapy for individuals with HBV has been a subject of intense scholarly debate. Although there are slight variations in the antiviral indications for individuals with hepatitis B-related cirrhosis in the current guidelines, all guidelines recommend initiating antiviral treatment at an early stage considering that HBV-associated hepatocarcinogenic events can develop during the early stage of chronic hepatitis B, even without apparent signs of significant hepatic inflammation or fibrosis [11, 25, 26]. Our study comprehensively assessed the risk of HCC in individuals with hepatitis B-related cirrhosis across various age groups and found a non-linear relationship between baseline age and HCC risk during long-term NAs treatment. This finding further emphasizes the importance of early initiation of antiviral therapy in individuals with liver cirrhosis, adding novel evidence to the screening and treatment strategies for individuals with hepatitis B-related cirrhosis.
In this study, we also conducted a subgroup analysis to explore discrepancies in the HCC incidence between different cirrhosis and sex groups. Many studies have demonstrated that NAs can improve the long-term prognosis of patients with decompensated cirrhosis, and some patients can achieve recompensation according to the Baveno VII criteria [13, 15, 27]. Our previous study showed that the decompensated group had a significantly increased risk for the development of HCC compared with the recompensated group, while the compensated and recompensated groups had a similar HCC risk [28]. Additionally, previous studies reported that HCC exhibits a strong male preponderance, with a male-to-female ratio estimated to be 2–2.5:1 [29]. Our stratified analyses showed that the effect of age on HCC risk was consistent across different cirrhosis and sex groups, implying the robustness of the non-linear relationship between age and the risk of HCC across different subgroups.
This study has several limitations. First, being a retrospective study, our findings are inevitably subject to bias and confounding. To address these limitations, we implemented strict inclusion criteria and employed multiple rigorous statistical methods to minimize the impact of confounding factors on the study conclusion. Second, we only included patients with cirrhosis due to hepatitis B; therefore, whether the non-linear relationship between baseline age and HCC risk can be extrapolated to patients with other etiologies of cirrhosis warrants further validation. Third, patients were treated with first-line antiviral agents such as entecavir, tenofovir disoproxil fumarate, or tenofovir alafenamide at baseline. Some patients switched or modified their antiviral regimens due to cost, tolerability, or treatment response. Therefore, we did not include drug classification as covariates in the statistical analysis. Fourth, we did not assess the genotypes of HBV at baseline. Previous studies have shown that genotype C accounts for 85% of HBV genotypes in our region (northern China) [30]; therefore, we speculated that the differences in HCC risk resulting from genotype variations were not very noticeable.

Conclusion
The relationship between baseline age and on-treatment HCC risk demonstrated a non-linear, parabolic-like trajectory. Early initiation of antiviral therapy is recommended to mitigate the escalating risk of HCC in individuals with HBV-related cirrhosis. Additionally, enhanced monitoring for HCC is advised for individuals initiating antiviral therapy above the age of 50 years.

Electronic supplementary material
Below is the link to the electronic supplementary material.