Pre-diagnostic anti-EBV antibodies and primary liver cancer risk: a population-based nested case-control study in southern China

The present study is a nested case-control study based on a population-based prospective cohort for NPC screening in Xiaolan town, Zhongshan City, southern China, between 2009 and 2017. Details of cohort recruitment were previously described [14]. Briefly, eligible participants were aged 30–59 years and residing in the study area. In total, 39,409 participants completed the initial screening. After excluding subjects with ages outside 30–59 years (n = 3585), those with missing birth date (n = 61), duplicate subjects (n = 232), and those with missing identity (n = 58) or sex (n = 9), 35,653 participants remained in the cohort. As of 30 June 2021, 127 confirmed incident primary liver cancer cases (presenting as 10th version of International Classification Disease C22) were identified through linkage to the cancer registry of Zhongshan City – two cases were excluded due to diagnosis prior to the baseline study examination, leaving 125 (83 image-based and 42 pathological-based) confirmed primary liver cancer cases. The index date for cases was the date of diagnosis.

For each case, we used risk-set sampling to randomly select 30 controls within the screening cohort who were alive on the case index date, did not migrate out of Zhongshan City, and were not diagnosed with primary liver cancer before the case index date on birth year, date of initial screening, and sex. Five cases were not matched to controls exactly on birth year; therefore, we matched three cases to controls within 1 year, and two cases within 8 years of the birth year (Fig. 1).

Screening cohort participants donated 6 mL of whole peripheral blood at study enrollment and during cohort follow-up. The current analysis was restricted to the initial (earliest) blood sample. EBNA1-IgA (Zhongshan Bio-Tech Company, Zhongshan, China) and VCA-IgA (UROIMMUNAG, Lübeck, Germany) were measured in sera using relative optical density (rOD) by enzyme-linked immunosorbent assay (ELISA). According to the manufacturers’ instructions, antibodies in sera were classified as positive (rOD ≥1) or negative (rOD < 1). In addition, we divided the participants into three groups: double-negative (i.e., negative for both EBNA1-IgA and VCA-IgA), single-positive (i.e., positive for either EBNA1-IgA or VCA-IgA, but not both), and double-positive (i.e., positive for both EBNA1-IgA and VCA-IgA).

HBsAg in serum was detected by ELISA following the manufacturer’s instructions (Autobio Diagnostics Co., China). Samples were collected and tested in 2012 when a mass liver cancer screening program was launched in Zhongshan City [9]. Because chronic HBV infection is nearly universally acquired at birth or during early childhood in southern China [15], HBsAg seropositivity in adulthood generally reflects long-term chronic infection.

The distributions of anti-EBV antibodies in primary liver cancer cases and controls were compared by Wilcoxon rank-sum tests. To evaluate the association between pre-diagnostic EBV seropositivity and primary liver cancer risk, we performed conditional logistic regression to estimate odds ratios (ORs) and 95% confidence intervals (CIs) with or without adjustment for age at enrollment (in years) as a continuous variable and HBsAg status (positive or negative) in matched case-control analyses accounting for risk sets.

Joint associations of HBV and EBV with primary liver cancer risk were tested by cross-classifying HBsAg status and anti-EBV antibody status as positive or negative (four categories: both positive, both negative, or either HBsAg or anti-EBV antibody-positive), and then testing for multiplicative interaction [16] or additive interaction by calculating relative excess risk due to interaction (RERI) [17, 18]. Only the 72 primary liver cancer cases and 953 matched controls with pre-diagnosis information on HBsAg status were included in this analysis.

To rule out a reverse-causal effect of preclinical liver cancer on anti-EBV antibodies, we performed a sensitivity analysis restricted to cases with blood samples for anti-EBV antibodies obtained at least 2 years before diagnosis, and their corresponding matched controls.

Among the 125 confirmed incident primary liver cancer cases (Table 1), 84.8% (106/125) were males; 58.4% (73/125) were aged between 50 and 59 at diagnosis; 88.8% (111/125) were HCC. Seventy-six cases had HBsAg tests, including four who were tested after cancer diagnosis (Fig. S1); 80.2% (61/76) of cases were positive for HBsAg. Characteristics of the 72 cases with pre-diagnosis HBsAg results and the 49 cases without HBsAg results did not show systematic differences (Table S1). For EBV serology, 9.6% (12/125) of cases and 4.9% (102/2077) of controls were EBNA1-IgA-positive, while 12.8% (16/125) of cases and 7.2% (150/2077) of controls were VCA-IgA-positive. Two cases (1.6%) and 20 controls (1.0%) were seropositive for both anti-EBV antibodies. The median time between anti-EBV antibody testing and cancer diagnosis was 4.74 years (range: 0.08, 110.2), and that for HBsAg testing was 3.45 years (range: − 0.74, 8.31) years (Fig. S1).

The distributions of EBNA1-IgA (p = 0.003) and VCA-IgA (p = 0.002) were significantly different between cases and controls (Fig. 2). Participants with positive EBNA1-IgA (adjusted OR: 2.19, 95% CI: 1.06, 4.51), positive VCA-IgA (adjusted OR: 1.93, 95% CI: 1.03, 3.65) or single-positive anti-EBV (adjusted OR: 2.00, 95% CI: 1.17, 3.42) antibodies had around two-fold odds of developing liver cancer, compared with seronegative subjects in adjusted models (Table 2). The age-adjusted ORs controlled for rOD of EBNA1-IgA and VCA-IgA were 1.52 (95% CI: 1.17, 1.98) and 1.63 (95% CI: 1.13, 2.34), respectively (Table 2).

Subgroup analysis (Table 3) showed that rOD of EBNA1-IgA remained associated with increased risk of primary liver cancer among males (adjusted OR: 1.55, 95% CI: 1.13, 2.14) and with increased risk of HCC specifically (adjusted OR: 1.51, 95% CI: 1.10, 2.09). Additional stratified associations between binary anti-EBV antibodies and primary liver cancer risk by age, sex and classification are presented in Tables S2 and S3. There was no significant difference between binary anti-EBV antibodies and primary liver cancer risk both in female and male strata.

When HBsAg and anti-EBV antibody status were cross-classified, the strongest association with primary liver cancer risk was observed for both serologically positive HBsAg and anti-EBV antibodies. Specifically, the adjusted OR for subjects who were both HBsAg-positive and anti-EBV antibody-positive was 50.67 (95% CI: 18.28, 140.46). The RERIs of VCA-IgA (35.07, 95% CI: 2.99, 144.76) and anti-EBV antibodies (30.81, 95% CI: 3.42, 114.93) with HBsAg were both significant, whereas no multiplicative-scale interactions were detected (Table 4).

Restricting to cases with anti-EBV antibodies measured at least 2 years before diagnosis (Tables S4, S5), the adjusted OR for positive EBNA1-IgA was 2.37 (95% CI: 1.03, 5.45) and that for positive VCA-IgA was 1.91 (95% CI: 0.93, 3.95) (Table S5).

To our knowledge, this study is the first to prospectively investigate whether pre-diagnostic anti-EBV antibodies are associated with primary liver cancer risk. The prospective design minimizes the potential for reverse causation. Moreover, the population-based setting broadens the generalizability of our results to the southern Chinese population and enables the interpretation of ORs as rate ratios due to the incidence density sampling approach [24, 25].

We also acknowledge the limitations of this study. First, we only have two-thirds of subjects has HBsAg, including 58% of cases (72 of 125) with pre-diagnosis HBsAg status. Comparing cases with pre-diagnostic HBsAg and without HBsAg information, we found that the characteristic did not vary significantly (Table S1). Second, due to a lack of information on tumor EBV status, we could not determine whether anti-EBV antibodies were associated with EBV-positive liver cancer cells, EBV-positive lymphocytes infiltrating tumor tissues, or the general humoral immune response to EBV infection. Third, we acknowledge that there might exist uncontrolled confounding due to lack of information of some risk factors, for instance, alcohol consumption and red meat intake. Finally, in this population of predominantly HBV-related primary liver cancer, we could not determine whether risk associations with anti-EBV antibodies varied across different liver cancer etiologies, such as HBV-positive, HCV-positive, alcohol-related, and nonalcoholic-fatty-liver-disease-related. We were also unable to address whether our findings are specific to the southern Chinese population, where chronic HBV infection and EBV-associated NPC are endemic, or whether they might be generalizable beyond this region.