Elevated antibodies against Epstein–Barr virus among individuals predicted to carry nasopharyngeal carcinoma susceptibility variants

Abstract

Epstein–Barr virus (EBV) is an obligatory factor in the development of nasopharyngeal carcinoma (NPC), and anti-EBV IgA antibodies are elevated many years prior to the development of NPC. Nearly all adults are infected with EBV, but only a few develop cancer, suggesting that additional co-factors, including genetic susceptibility, must be required for the disease to manifest. Individuals were selected from the Taiwan Family Study, a cohort of 3389 individuals from NPC multiplex families. Primary analyses were conducted among 671 individuals from 69 pedigrees with the strongest family history of disease (>3 NPC-affected family members). The likelihood that a given family member carried a NPC susceptibility variant was estimated using Mendelian segregation rules, assuming a dominant mode of inheritance. We compared anti-EBV IgA antibody seropositivity between family members predicted to be carriers of NPC-linked genetic variants and those with a lower likelihood of carrying such variants. Obligate carriers of NPC susceptibility variants (100 % predicted probability of harbouring the genetic mutation) were nine-fold more likely to be anti-EBV IgA positive compared to family members predicted not to carry disease-causing variants (OR=9.2; P-trend<0.001). This elevated risk was confirmed in analyses restricted to both unaffected individuals and pedigrees with EBV-related pathway variants identified through exome sequencing. Our data indicate that family members who are more likely to carry NPC susceptibility variants are also more likely to be anti-EBNA1 IgA seropositive. Genetic susceptibility associated with control over this common herpes virus is likely a co-factor in determining which EBV-infected adults develop NPC.

Introduction

Nasopharyngeal carcinoma (NPC) is an epithelial tumour linked to Epstein–Barr virus (EBV), a ubiquitous herpes virus that establishes lifelong latency in human B cells and periodically reactivates at mucosal sites in the head and neck [1–4]. While nearly all adults are EBV infected, only a small fraction develops NPC, suggesting that viral or non-viral co-factors are important determinants of NPC risk. Early case-control studies and more recent large-scale prospective cohorts from NPC-endemic regions have consistently demonstrated a strong and consistent association between the presence of anti-EBV IgA antibodies and the risk of developing NPC [5–7]. Since IgA is a transient antibody marking mucosal (e.g. oral cavity) pathogen exposure [8], and EBV periodically reactivates in the oral epithelium [4, 9], it is hypothesized that anti-EBV IgA antibodies mark ongoing/recent EBV reactivation at these mucosal surfaces.

Genetic susceptibility is known to be associated with NPC development. Close to 10 % of NPC in endemic regions clusters within families, suggesting that inherited genetic susceptibility is an important risk factor for NPC [10]. The involvement of underlying genetic susceptibility in NPC development is further supported by genome-wide association (GWA) and targeted gene association studies reporting associations with human leukocyte antigen (HLA) genes that are pivotal in the presentation of viruses to the immune system [11–14]. Furthermore, in NPC multiplex families from Taiwan and China (i.e. >two first- or second-degree relatives affected by NPC), we and others have identified rare genetic variants that segregate with NPC status (Yu, submitted; [15–17]).

Taken together, these findings lend support to the hypothesis that only certain EBV-infected individuals develop NPC due to the co-existence of genetic variants that increase susceptibility to disease. However, studies evaluating the direct link between inherited genetic variation and EBV control (i.e. IgA antibody), a strong risk factor for NPC, are lacking. To address this knowledge gap, we correlated the predicted likelihood of carrying purported NPC susceptibility gene variants with anti-EBV IgA antibody seropositivity among pedigrees highly affected by NPC (>3 NPC-affected family members).

Results

Among 671 individuals belonging to 69 pedigrees with a strong family history of NPC (>3 NPC-affected family members), the likelihood of carrying predicted NPC-linked genetic variants was related to anti-EBV IgA levels. (Fig. 1; Table 1) The majority (58 %) of family members who were obligate carriers of the predicted variant tested above the positivity threshold for anti-EBV IgA. In contrast, only 13 % of spouses who married into the pedigrees, and who were predicted not to carry the variant since they were not genetically related to high-risk families, were anti-EBV IgA positive. After accounting for both participant age and sex, obligate carriers were nine-fold more likely to test anti-EBV IgA positive (OR=9.2; 95 % CI 4.1–20.8; P-trend<0.001). The increased IgA antibody positivity rate in those with a higher likelihood of carrying (NPC cases) or inheriting (children and siblings of NPC cases) a predicted NPC-linked genetic variant was present for both EBNA1 IgA (OR for obligate carriers versus non-carriers: 5.8; 2.5–13.6; P-trend<0.001) and VCA IgA (OR=27.9; 3.6–213; P-trend<0.001).

Fig. 1.

The level of anti-EBV IgA antibody was higher in those with an increased likelihood of carrying (NPC cases) or inheriting (children or siblings of NPC cases) genetic variants predicted to be linked to NPC.

Table 1. Anti-EBV IgA antibody responses and probability of carrying a predicted NPC-linked genetic variant among multiplex pedigrees from Taiwan.

Probability (%) of predicted NPC variant	69 pedigrees with strong family history of NPC (>3 NPC-affected family members)				49 pedigrees with rare, EBV-related variants identified in exome sequencing			Analysis restricted to unaffected (non-NPC) family members
	Mean IgA level	N (%) IgA positive	OR*	95 % CI	N (%) IgA positive	OR	95 % CI	N (%) IgA positive	OR	95 % CI
Anti-EBV IgA†
Obligate carriers‡ (100 %)	0.67	66 (57.8 %)	9.19	4.07–20.8	67 (58.8 %)	25.8	5.67–118	6 (37.5 %)	3.29	0.94–11.5
Children/siblings of carriers (50 %)	0.32	90 (18.4 %)	1.75	0.80–3.82	52 (16.7 %)	4.45	0.99–20.1	90 (18.4 %)	1.89	0.85–4.17
Unrelated spouses (0 %)	0.32	9 (13.0 %)	1.0	Ref.	2 (6.9 %)	1.0	Ref.	9 (13.0 %)	1.0	Ref.
				P-trend<0.001			P-trend<0.001			P-trend=0.04
Anti-EBNA1 IgA
Obligate carriers‡ (100 %)	0.27	52 (45.6 %)	5.82	2.50–13.6	54 (47.4 %)	13.4	2.98–60.4	5 (31.3 %)	3.03	0.82–11.2
Children/siblings of carriers (50 %)	0.13	74 (15.2 %)	1.38	0.60–3.15	47 (15.1 %)	3.05	0.68–13.7	74 (15.2 %)	1.43	0.62–3.33
Unrelated spouses (0 %)	0.13	8 (11.6 %)	1.0	Ref.	2 (6.9 %)	1.0	Ref.	8 (11.6 %)	1.0	Ref.
				P-trend<0.001			P-trend<0.001			P-trend=0.12
Anti-VCA IgA
Obligate carriers‡ (100 %)	0.40	32 (28.1 %)	27.9	3.64–213	32 (28.1 %)	–§	–	1 (6.3 %)	3.19	0.18–55.4
Children/siblings of carriers (50 %)	0.19	24 (4.9 %)	4.54	0.58–35.3	8 (2.6 %)	–	–	24 (4.9 %)	6.53	0.82–52.1
Unrelated spouses (0 %)	0.19	1 (1.5 %)	1.0	Ref.	0	1.0	Ref.	1 (1.5 %)	1.0	Ref.
				P-trend<0.001			P-trend<0.001			P-trend=0.17

*Odds ratio (OR) for anti-EBV IgA positivity calculated using logistic regression models adjusted for age and sex.

†Represents a combination of EBNA1 IgA and VCA IgA antibody response, as outlined in Methods.

‡Obligate carriers include both NPC cases and unaffected family members from the pedigree that were deemed to have 100 % probability of carrying the predicted variant based on their relationship to NPC cases across multiple generations (e.g. both siblings and children affected by NPC).

§OR could not be estimated because no unrelated spouses with 0 % predicated carriage of purported NPC-linked variant (ref. group) tested positive for anti-VCA IgA antibody.

Additional evidence supporting a positive link between inherited genetic susceptibility to NPC and the immune response to EBV was observed among 454 individuals from 49 pedigrees that were previously exome sequenced and found to harbour rare, NPC-linked variants in genes from potentially EBV-related pathways. Among this subset of families, obligate carriers of predicted disease-causing variants and siblings/children of those obligate carriers were ~25-fold (OR=25.8; 95 % CI 5.7–118) and ~five-fold (OR=4.5; 1.0–20.1) more likely to be anti-EBV IgA positive compared to genetically unrelated spouses, respectively (P-trend<0.001; Table 1).

We performed a sensitivity analysis that removed NPC cases to avoid potential bias resulting from a disease effect. Despite a limited number of unaffected obligate carriers in these pedigrees (family members who carried the predicted variant but were not NPC cases) for comparison, we still observed a trend towards higher anti-EBV IgA positivity in family members predicted to carry NPC-linked genetic variants (OR=3.3; 95 % CI 0.9–11.5) and their first-degree genetic relations (OR=1.9; 0.9–4.2) compared to predicted non-carriers (P-trend=0.04; Table 1). Reported associations also remained unchanged after further adjustment for inheritance of specific HLA alleles (Table S1, available in the online version of this article).

We observed no consistent association between anti-EBV IgA positivity and the carriage of predicted NPC-linked variants among the 229 pedigrees with a weaker family history of disease (only two NPC-affected family members). Although obligate carriers were more likely to be anti-EBV IgA positive compared to non-carriers when including all 1605 family members (OR=4.2; 95 % CI 3.3–5.4; P-trend<0.001), this association was no longer evident in an analysis that avoided a disease effect by restriction to unaffected family members (OR=1.0; 0.7–1.4; P-trend=0.95).

Discussion

Epstein–Barr virus (EBV) is a necessary factor for the development of anaplastic NPC, but only a small proportion of adults infected with EBV develop cancer. We utilized a study of Taiwanese individuals with a strong family history of the disease to demonstrate that family members carrying putative genetic variants linked to NPC were more likely to exhibit anti-EBV IgA seropositivity. This suggests that genetic variation is related to ongoing exposure to EBV and could represent a potentially important co-factor in the development of EBV-related NPC.

This unique analysis was made possible by the availability of information regarding not only NPC case status and predicted carriage of genetic variants, but also data on serum IgA antibody levels in these multiplex family members. The results from a subset of families with EBV-related genetic variants identified through exome sequencing are notable because the rare variants occurred in genes that are potentially involved in EBV entry and control or are thought to interact with EBV proteins. Differences in anti-EBV antibodies according to whether individuals were predicted to inherit such NPC susceptibility variants provided additional evidence that a proportion of genetic susceptibility to NPC could be EBV-mediated. Furthermore, larger studies should build upon these findings to directly measure anti-EBV IgA antibodies in patients with deep sequencing, proving the inheritance of specific mutations. Additional work is also needed to statistically explore to what degree specific genetic susceptibility to NPC is mediated through an inability to control EBV.

Because NPC is endemic in Taiwan, it is plausible that some families enrolled in this study may have included two individuals affected by NPC simply by chance. Rather than reflecting NPC that arises through unique genetic susceptibility, this instead may more likely represent sporadic disease. Accordingly, we observed inconsistencies in the association between anti-EBV antibody levels and the carriage of predicted disease-causing variants in the 229 families with no more than two family members affected with NPC, the families most likely to represent sporadic disease.

Even among the 69 families with >3 affected family members and the highest likelihood for an underlying genetic susceptibility to NPC, the assumption in our genetic model that all cases were classified as obligate carriers could be a limitation. If any NPC cases in these 69 families were phenocopies (i.e. cases that did not carry a susceptibility variant), this exposure misclassification could have impeded our analysis, likely biasing effect estimates towards the null by classifying ‘unexposed’ family members in the 100 % carriage category. We must also acknowledge that there was not a perfect correlation between genetic variation and anti-EBV antibodies, suggesting that a portion of genetic susceptibility to NPC does not act through EBV-related mechanisms.

In summary, we observed evidence that individuals with a higher probability of inheriting genetic variants linked to NPC were also more likely to be anti-EBNA1 IgA seropositive. We observed additional evidence supporting an association between predicted NPC-related genetic susceptibility and the immune response to EBV in a subset of 49 exome-sequenced families with rare variants in genes thought to be involved in EBV-related pathways. EBV is a necessary factor for anaplastic NPC, and our data suggest that genetic susceptibility that influences control of this ubiquitous herpes virus is likely a co-factor in determining which EBV-infected adults develop NPC.

Methods

Individuals were selected from the Taiwan Family Study (TFS), an ongoing NPC multiplex family study. Details of this study population have been published previously [18, 19]. Briefly, 358 families with >2 relatives affected by NPC were ascertained via the Taiwan National Cancer Registry (1980–2003). A total of 3389 individuals, including 832 individuals diagnosed with NPC (394 alive and sampled at baseline and 438 deceased cases) and 2557 unaffected family members, were identified. A family history questionnaire was administered to each living NPC case to ascertain vital status and pedigree information (i.e. to determine the degree of relatedness between NPC cases and other family members). Proxy interviews were conducted with a close family member. Incident cases of NPC diagnosed after initial recruitment (N=25) were identified through June 2015 via active follow-up and linkage to the Taiwan National Cancer Registry, bringing the total NPC count to 857 (394 prevalent cases sampled at baseline; 438 deceased; 25 incident cases that developed during follow-up).

Within TFS, we [1] assigned a probability to each individual that described the likelihood that a family member carried a predicted NPC-linked genetic variant and [2] measured anti-EBV IgA antibody positivity in each family member. Among those with available anti-EBV IgA antibody data, adequate information on relatedness within the pedigree was available to assign probabilities to 361 NPC cases and 1915 unaffected family members from 298 pedigrees. We focused our analyses on the 69 pedigrees with the strongest family history (>3 NPC-affected family members), increasing the likelihood that the family harboured genetic susceptibility to NPC. These pedigrees included 671 individuals (99 NPC cases and 572 unaffected family members).

Assigning the probability of carrying predicted NPC-linked genetic variants

We assigned the probability of each family member being a carrier of a predicted NPC susceptibility variant based on the observed family structure, including how closely related each individual was to their family members affected by NPC. To assign these probabilities, we assumed the following about how the predicted NPC-linked variants were inherited: 1) autosomal dominant mode of inheritance – only one copy of the variant is sufficient for disease according to Mendelian inheritance rules; 2) variant present in both males and females – mutation was not sex-linked; 3) rarity of the variant in the general Taiwanese population (i.e. spouses not genetically related to the pedigree would not introduce new variants into the pedigrees); and 4) heterozygosity of affected NPC cases.

Using these rules, each family member in the selected pedigrees was assigned a carriage probability of 0, 50 or 100 %. Spouses who married into the pedigree were assigned 0 % (no probability) of inheriting the predicted NPC-linked genetic variant based on the lack of genetic relatedness to other pedigree members. Obligate carriers (100 % probability) of the predicted NPC-linked genetic variants included NPC cases, given the assumption that carrying the predicted variant resulted in the NPC phenotype, and unaffected (non-NPC) family members who ‘linked’ generations of NPC cases (Fig. S1). The remaining unaffected family members with a first-degree relation to an NPC case or obligate carrier, including siblings or children of the predicted mutation carriers, were assigned an intermediate (50 %) likelihood of inheriting a NPC-linked genetic variant.

Because each family member in the pedigrees examined was within one degree of relation (e.g. children, siblings, spouses) to either an NPC case or a family member who was an obligate carrier, these probabilities could be estimated without ambiguity. Perl programming language was used to implement the algorithm described above.

Anti-EBV IgA antibody testing

TFS participants were asked to consent to the collection of blood for anti-EBV antibody testing. As previously described [18], the serum collected at baseline was tested for IgA antibodies against two EBV proteins, viral capsid antigen (VCA) and EBV nuclear antigen 1 (EBNA-1). Participants were considered anti-EBV IgA positive (seropositive) if titres of either VCA or EBNA1 were higher than pre-defined thresholds (VCA: indirect immunofluorescence assay detection >1 : 10; EBNA1 : ELISA OD₄₀₅>0.20). Continuous values for each assay were expressed as fractions, representing the ratio of output standardized to an EBV-negative background distribution. Average anti-EBV IgA antibody levels are reported as the sum of these semi-quantitative ratios from each assay.

Statistical analysis

We report the average level and seropositivity rate for anti-EBV IgA antibody according to groups defined by their likelihood of carrying a predicted NPC-linked genetic variant, as defined above: 0 (unrelated spouse), 50 (children/siblings of obligate carriers) and 100 % (obligate carriers, including NPC-affected family members). We used logistical regression to model the association between the probability of carrying a predicted NPC-linked genetic variant (3-level variable for the 0, 50 and 100 % carrier categories) and anti-EBV IgA antibody seropositivity, adjusted for participant age and sex.

Sensitivity analysis

The fact that anti-EBV IgA antibody levels are elevated in NPC cases raised the possibility of a disease effect. To avoid inducing an overall association between anti-EBV IgA antibodies and carriage of predicted susceptibility variants by including blood samples from NPC cases, we repeated the antibody-genetic variant comparison focusing only on unaffected family members (i.e. obligate carrier group restricted to unaffected family members predicted to carry the purported variant).

We previously reported that variation in the inheritance of specific human leukocyte antigen (HLA) alleles is related to NPC status in the general Taiwanese population [20]. To ensure that the associations we observed between genetic variation and anti-EBV IgA level were independent of a potential effect of HLA on EBV antibody titres, we included a categorical variable for each of the following alleles in logistic regression models: HLA*A1101, HLA*A0207, HLA*A3303, HLA*B4601, HLA*B5801 and HLA*B3802.

Exome sequencing data

We recently reported the identification of specific rare variants that co-segregated with NPC status in 49 of these multiplex pedigrees (N=454 individuals; 100 cases and 354 unaffected family members) using a whole-exome sequencing approach [Yu, submitted]. Relevant for the present effort, the rare variants identified in these families occurred in genes that are potentially involved in EBV entry and control or are thought to interact with EBV proteins (e.g. NOTCH1, DLL3, MAML1, MFNG, PSEN2, ITGB6, NIPAL1, BCL2L12, NEDD4L, LAMC2, LFNG). We leveraged these available exome sequencing data to further evaluate the association between anti-EBV IgA antibody seropositivity and predicted carriage of susceptibility variants in families where it was known that a rare, likely to be functional, mutation that co-segregated with NPC was present.