Purpose:

The favorable prognosis of stage I and II nasopharyngeal carcinoma (NPC) has motivated a search for biomarkers for the early detection and risk assessment of Epstein-Barr virus (EBV)–associated NPC. Although EBV seropositivity is ubiquitous among adults, a spike in antibodies against select EBV proteins is a harbinger of NPC. A serologic survey would likely reveal which EBV antibodies could discriminate those at risk of developing NPC.

Experimental Design:

Lysates from a new EBV mammalian expression library were used in a denaturing multiplex immunoblot assay to survey antibodies against EBV in sera collected from healthy individuals who later developed NPC (incident cases) in a prospective cohort from Singapore and validated in an independent cohort from Shanghai, P.R. China.

Results:

We show that IgA against EBV nuclear antigen 1 (EBNA1) discriminated incident NPC cases from matched controls with 100% sensitivity and 100% specificity up to 4 years before diagnosis in both Singapore and Shanghai cohorts. Incident NPC cases had a greater IgG repertoire against lytic-classified EBV proteins, and the assortment of IgA against EBV proteins detected by the immunoblot assay increased closer to diagnosis.

Conclusions:

Although NPC tumors consistently harbor latent EBV, the observed heightened systemic and mucosal immunity against lytic-classified antigens years prior to clinical diagnosis is consistent with enhanced lytic transcription. We conclude that an expanding EBV mucosal reservoir (which can be latent and/or lytic) is a risk factor for NPC. This presents an opportunity to identify those at risk of developing NPC using IgA against EBNA1 as a biomarker.

Translational Relevance

Nasopharyngeal carcinoma (NPC) is closely associated with Epstein-Barr virus (EBV) infection. Survival can be as high as 90% if detected at stage I. Toward the goal of early diagnosis and NPC screening by risk assessment, EBV molecular properties associated with NPC are being explored. Circulating cell-free EBV DNA is a biomarker for early diagnosis but its utility in risk assessment has yet to be determined. EBV serology has been shown in case–control studies to be informative for risk assessment, but high specificity is needed for population-based screening and implementation. Here, we profiled EBV serology in two independent cohorts by denaturing immunoblot in an attempt to maximize sensitivity and specificity. This study demonstrates that EBV serology can be exploited for NPC risk assessment several years before clinical diagnosis and this is achievable with high accuracy using a single informative EBNA1 IgA biomarker detected by an immunoblot laboratory assay.

Nasopharyngeal carcinoma (NPC) is a leading head and neck cancer in Southeast Asia, particularly in southern China where NPC is endemic. Historically, NPC incidence in the 1980s was high in Chinese men in Hong Kong (30/100,000) and Singapore (19/100,000), moderate in Shanghai, P.R. China (5/100,000; ref. 1), and low in white men in the United States (0.5/100,000; ref. 2). Although NPC incidence has decreased over time, there were an estimated 133,000 new cases and 80,000 deaths from NPC worldwide in 2020 (3). Early-stage NPC (I and II) is often asymptomatic, and therefore most NPC cases are diagnosed later (stage III and IV). Five-year overall survival decreases from 90% when diagnosed at stage I to 58% at stage IV (4). Although environmental exposures and genetic factors may contribute to the risk of NPC (1, 5), more than 97% of tumors are associated with latent Epstein-Barr virus (EBV; ref. 6). In addition, elevated antibodies to EBV lytic proteins are considered a harbinger of NPC (7). Thus, a survey of EBV serology could yield crucial information to predict NPC risk and possibly provide target markers for vaccine development and efficacy evaluation (8).

Several EBV biomarkers have been proposed for NPC screening in high-risk populations. Plasma cell-free EBV DNA, which may reflect the release of EBV from apoptotic and/or necrotic cells in a tumor, can detect early-stage NPC (9). This improved detection at stage I and II is based on the levels, methylation pattern, and fragment size of cell-free EBV DNA (10, 11). IgA antibodies against EBV viral capsid antigen p18 (VCA p18) and nuclear antigen 1 (EBNA1) have been evaluated for early detection of NPC in several high-risk populations (12–15). A two-step ELISA approach to detect IgA against VCA p18 and EBNA1 followed by IgA against EBV early antigen nuclear protein extracts can achieve accurate early detection (sensitivity 96.7%, specificity 98%) in an Indonesian NPC endemic cohort (12). Many of these previous studies were conducted in high-risk populations with cross-sectional design or short duration of follow-up. Therefore, the results only indicated whether NPC was present. Additional biomarkers may be required to assess NPC risk (16). Coghill and colleagues conducted a comprehensive serologic survey for EBV biomarkers using a peptide array and found that a composite score of 14 EBV antibodies, including IgA against VCA p18 and EBNA1, had an estimated 85% sensitivity and 61% specificity in a general Taiwanese population cohort to determine NPC status on average 4.2 years before clinical diagnosis (17). Thirteen of these biomarkers were validated by surveying sera taken from individuals at the time of NPC diagnosis using an independent multiplex assay based on bacterial-expressed EBV proteins conjugated to beads (18). Given the low incidence of NPC in the general population, the next step is to improve upon the specificity of an NPC risk score to warrant implementation as a screening test.

In this study, we set out to develop a comprehensive library of 86 EBV open reading frames (ORF) derived almost exclusively from the EBV Akata strain to be expressed in mammalian cells. These mammalian lysates were used to conduct a serological survey for IgA, IgG, and IgM antibodies against EBV proteins, scored by a denaturing multiplex immunoblot, using serum from healthy individuals who were later diagnosed with NPC (incident cases) from a cohort of Singaporean Chinese, referred to as the Singapore Chinese Health Study (SCHS). We then validated these findings in a similarly designed cohort study in Shanghai, P.R. China, named the Shanghai Cohort Study (SCS). Conventionally, an ELISA is the gold standard for measuring EBV serology, but ELISA is a low throughput assay that is not suitable for screening large numbers of EBV proteins. Because denaturing immunoblots using full-length EBV proteins are often used as a confirmatory assay (12, 19, 20), we hypothesized that this labor-intensive but comprehensive multiplex approach would provide a high degree of accuracy for screening a library of EBV proteins. Our results showed a superior performance of serum IgA against EBNA1 (EBNA1 IgA) as the most informative biomarker for NPC risk assessment in case–control pairs of NPC prediagnostic sera.

Study population

The SCHS is a residential cohort of 63,257 Chinese men and women from Singapore, ages 45 to 74 years at enrollment (1993–1998; ref. 21). Each subject was initially interviewed in person by trained personnel using a structured questionnaire including origin of province in China, dialect group (Cantonese and Hokkien), history of tobacco use, medical history, current diet, incense use at home, and for women, menstrual and reproductive history, and use of hormone replacement therapy. All blood components (plasma, serum, red blood cells, and white blood cells) were separated within 4 hours postcollection and stored at −80°C. Incident cancer cases among cohort participants were ascertained through record linkage with national databases of the Singapore National Cancer Registry and the Singapore National Birth and Death Registry, respectively. Since inception in 1993, <0.1% (56 subjects) were lost to follow-up by migration out of Singapore. As of December 31, 2015, 50 participants who had no history of cancer were diagnosed with NPC, 42 of them with available baseline serum samples were included in this study. For each case, we randomly selected a control subject among all eligible participants who provided a baseline serum sample, were cancer free and alive during the time from blood draw to cancer diagnosis of the index case and had no family history of NPC. The controls were individually matched to the index case by sex, dialect group (Hokkien, Cantonese), age at enrollment (±3 years), date of baseline interview (±2 years), and date of biospecimen collection (±6 months). The controls were followed for a median of 13 years (range, 11.1–19.8), and none were diagnosed with NPC.

The SCS is a residential cohort of 18,244 men from Shanghai, P.R. China, ages 45 to 64 years at enrollment (1986–1989; ref. 22). Approximately 80% of eligible subjects agreed to participate in the study. Each subject was interviewed in person by trained personnel using a structured questionnaire including history of tobacco and alcohol use, current diet, and medical history. Blood samples were processed within 4 hours after blood collection, and multiple aliquots of serum samples were stored at −70°C or lower until analysis. Incident cases of cancer and death among the participants were identified via annual interviews and augmented by record linkage analysis with the datasets of the Shanghai Cancer Registry and the Shanghai Vital Statistics. The cohort follow-up is virtually complete; as of May 2019, 597 (3.3%) original participants refused to continue, and 754 (4.1%) were lost to annual follow-up due to change of residence or migration out of Shanghai. As of 2019, 37 participants who were cancer free at baseline were diagnosed with NPC and included in this study. Similarly, we randomly chose one control subject for each case. The control was matched to the index case by age (± 2 years), date of blood draw (± 1 month), and the same neighborhood of residence at study enrollment. The controls were followed for a median of 28.6 years (range, 7.6–31.6), and none were diagnosed with NPC.

All study participants from the SCHS and SCS provided written informed consent before any research activities were performed. The studies were conducted in accordance with the ethical guidelines of the Declaration of Helsinki. Both studies have been continuously approved by Institutional Review Boards of the National University of Singapore (Singapore; for SCHS), the Shanghai Cancer Institute (Shanghai, P.R. China; for SCS), and University of Pittsburgh (Pittsburgh, PA; for SCHS and SCS). Selected baseline demographic and lifestyle factors of the NPC cases and their matched controls in both the SCHS and the SCS are shown in Supplementary Table S1. The identity and the case–control status of all test samples were masked to personnel running and scoring the immunoblots.

EBV expression library

Of 87 possible EBV ORFs (23), 86 were cloned to generate a new EBV ORF mammalian expression library (Supplementary Table S2). This library was based primarily on the EBV Akata genome (83/86 ORFs; ref. 24). Eighty-five ORFs were cloned at the University of Pittsburgh, Pittsburgh, PA, and one ORF (BPLF1 derived from EBV B95-8) was a gift from Dr. Christopher Whitehurst (New York Medical College, Valhalla, NY). LF3 was the only ORF that could not be cloned or synthesized because of repetitive sequences. Seventy-four ORFs were PCR amplified with Phusion High-Fidelity DNA Polymerase (Thermo Fisher Scientific) using primers based on the EBV Akata strain (GenBank KC207813.1), and one ORF (LMP2A) spanning the terminal repeats was based on the EBV B95-8 strain due to available cDNA (GenBank V01555.2). Ten ORFs were codon optimized and synthesized (Genewiz), nine of which were based on EBV Akata, and one ORF (BHLF1) was based on EBV B95-8 because it is truncated in the Akata genome. Genomic DNA extracted from Akata B cells with the GENEJET Genomic DNA Extraction Kits (Thermo Fisher Scientific) served as template for amplification. For the large ORF LMP2A, RNA was extracted with the GENEJET RNA kit and cDNA was synthesized using Revert Aid First Strand cDNA Synthesis Kit (Thermo Fisher Scientific). The PCR products were cloned in-frame with the 3X FLAG-tag of a modified mammalian expression vector. The modified vector was generated by inserting a linker sequence (5′-AAT TCG CGA TCG CTT AAT TAA ACC ATG GAC TAC AAA GAC CAT GAC GGT GAT TAT AAA GAT CAT GAC ATC GAT TAC AAG GAT GAC GAT GAC AAG CTT GCG GCC GCG GTA CCG-3′) into EcoRI and SalI sites of the p3XFLAG-CMV-7 mammalian expression vector (Sigma-Aldrich). EBV ORF BPLF1 was cloned into pFLAG-CMV-2 (Sigma-Aldrich). All cloned EBV ORFs were sequenced verified.

Cell lines and preparation of lysates

The EBV-positive B-cell line, Akata clone 21 (gift of Clare Sample, Penn State University, Hershey, PA), was maintained in RPMI1640 medium supplemented with 10% FBS. HEK293 cells (CRL-1573, ATCC) were grown in DMEM supplemented with 10% FBS. Cell lines are routinely screened (at least once every 2 months) for mycoplasma contamination by PCR and validated by DNA fingerprint using short tandem repeat profiling. Thawed cells were maintained for a maximum of 15–20 passages. Transfections of the 293 cells were performed using X-tremeGENE 9 DNA Transfection Reagent (Roche). Whole-cell lysates were prepared in RIPA lysis buffer supplemented with 1 mmol/L phenylmethylsulfonyl, 2 mmol/L activated sodium orthovanadate, a 1:100 dilution of protease inhibitor and phosphatase inhibitor cocktails (Sigma-Aldrich), clarified by centrifugation, aliquoted, and stored at −80°C.

Multiplex immunoblot assay

Whole-cell lysates were prepared by boiling or incubation at 70°C (Supplementary Table S2) in 1X protein sample loading buffer (LI-COR) supplemented with 2.5% β-mercaptoethanol. Purified human whole IgG, IgA, IgM (ChromPure), and FLAG-BAP recombinant protein (Sigma-Aldrich) were used as positive loading controls and for normalization of detected signals in each gel. Lysates were separated by SDS-PAGE in 4%–20% acrylamide gradient gels (Bio-Rad Criterion TGX) and transferred to 0.2 μm nitrocellulose membrane (Bio-Rad Trans Blot Turbo). Membranes were blocked in 5% nonfat dry milk in TBS and incubated overnight at 4°C with a 1:500 dilution of preadsorbed human sera and 1 μg/mL anti-FLAG M2 clone primary antibody (AB_259529, Sigma-Aldrich). Serum preabsorption was carried out prior to primary incubation for 24 hours at 4°C in 293 cells that were previously fixed in 4% paraformaldehyde and permeabilized in 0.1% Triton X-100. After primary incubation, membranes were washed in TBS (3× 5 minutes) followed by 1-hour incubation with fluorescent tagged secondary antibodies at room temperature in the dark, diluted in blocking buffer [3.75 μg/mL Cy3-AffiniPure goat anti-hIgAH (AB_2337721); 0.03 μg/mL AlexaFluor 680-AffiniPure goat anti-hIgGH (AB_2889013); 3.75 μg/mL Cy3-AffiniPure goat anti-hIgMH (AB_2337729, Jackson IR); 0.1 μg/mL IRDye680LT goat anti-hIgGH+L (AB_10795013); 0.1 μg/mL IR800CW goat anti-mIgGH+L (AB_621842, LI-COR)]. Each membrane was incubated with an optimized dilution of a mixture of anti-mIgGH+L, anti-hIgAH, and anti-hIgGH; or anti-mIgGH+L, anti-hIgGH+L, and anti-hIgMH antibodies which were validated in house to ensure no cross-reactivity between secondary antibody binding or channel fluorescence. The subscript H or H+L indicate the target “heavy” or “heavy and light” chain, respectively. Targeting the heavy chain is antibody specific because each class of antibody contains a unique heavy chain, whereas targeting the light chain that is common to all antibodies provides a nonspecific antibody marker and indicates the total antibody immune response.

Statistical analysis

We used χ2 (for categorical variables) or t test (for continuous variables) statistics to examine the difference in the distributions of selected baseline demographic and lifestyle variables between NPC cases and controls. Conditional logistic regression method was used to estimate OR and 95% confidence intervals of NPC development for the EBNA1 IgA detection. Fisher exact test was used to examine the difference in proportion of serum samples detecting each EBV protein between cases and controls. A Mann–Whitney test was used to compare the numbers of EBV proteins (nEBV) detected in NPC cases to those in controls. Sparse logistic regression analysis was used to compare the EBV antibody response rates in NPC cases and controls by fitting a logistic regression with the L1 regularization (25). The L1 regularization forces the model weights of unimportant immunoreactivity values to be zero, resulting in a sparse logistic model. The scalar multiplier for the L1 penalty term was set to 0.5. A log-likelihood ratio test of the fitting result was performed to determine the predictive significance of EBV protein classes between NPC cases and controls.

Statistical analyses were carried out using SAS software version 9.4 (SAS Institute), the statsmodels package in the Python (26), and Prism 9. The P values of less than 0.05 were statistically significant.

Data and material availability

The EBV ORF expression library generated for this study is available upon request. Python and R scripts for the conservation plots are available on Github (https://github.com/ShairLab/EBV_ Immunoglobulin).

The data generated in this study are available upon request from the corresponding author but may be subject to institutional approval associated with U.S. provisional patent application no. 63/336,590.

Of 27 prediagnostic serum samples collected within 8 years to diagnosis from the endemic SCHS cohort, 20 case–control pairs were prioritized on the basis of sufficiently available sample to screen the entire EBV ORF library (discovery group; Fig. 1). Among the 86 EBV proteins tested for IgAH (heavy chain–specific) in these sera, EBNA1 IgA was the most informative biomarker such that 19 of the 20 NPC cases, and none of the 20 matched controls tested positive (P < 0.0001; Fig. 2). EBNA1 is classified as a latent protein but is an essential protein that is produced in all types of EBV infection (27, 28). The single incident NPC (case ID: NPC12) negative for EBNA1 IgA showed no IgA reactivity to any EBV protein. This patient was a heavy cigarette smoker (23 pack-years) and the heaviest drinker (6 drinks/day) among the 20 pairs tested. It is unclear whether this NPC case was EBV associated, and we do not have access to tumor tissue for confirming EBV status by EBV-encoded small RNA (EBER) staining. The IgG response indicates that the patient was EBV seropositive (Supplementary Fig. S1). Excluding EBNA1 IgA, 18 of the 26 EBV proteins detected by IgA were unique to incident NPC cases which were named “NPC differential IgA panel” (Fig. 3), and 14 of these appeared only in NPC cases within 4 years to diagnosis (case IDs: NPC1-NPC9; Figs. 2 and 3). For three EBV proteins, IgA detection was enriched in the 20 cases compared with controls including, BFRF3 (12 cases vs. 4 controls, P = 0.0098), BKRF4 (4 cases vs. 0 controls, P = 0.0350), and BALF2 (4 cases vs. 0 controls, P = 0.0350), but combinations of these or other markers did not improve upon the superior EBNA1 IgA test score.

Figure 1.

Study population and design. The SCHS and SCS represent NPC endemic and non-endemic (moderate-risk) populations, respectively. Prediagnostic sera were identified for use based on time to NPC diagnosis (median, range in years) and sufficient serum availability, and were divided into discovery and validation groups. Each NPC case was matched with a healthy control based on predefined criteria in the Materials and Methods. Case–control pairs were surveyed against the entire 86 EBV ORF expression library for IgA, IgG, and IgM, or tested for EBNA1 IgA only.

Figure 1.

Study population and design. The SCHS and SCS represent NPC endemic and non-endemic (moderate-risk) populations, respectively. Prediagnostic sera were identified for use based on time to NPC diagnosis (median, range in years) and sufficient serum availability, and were divided into discovery and validation groups. Each NPC case was matched with a healthy control based on predefined criteria in the Materials and Methods. Case–control pairs were surveyed against the entire 86 EBV ORF expression library for IgA, IgG, and IgM, or tested for EBNA1 IgA only.

Close modal
Figure 2.

Detection of EBV-specific IgA in sera from incident NPC cases and healthy controls in the SCHS discovery group. Heatmap depicting IgA heavy chain–specific (IgAH) normalized detection values from 20 incident NPC cases and 20 individually matched controls, ordered by years to NPC diagnosis (dx). EBV ORFs (common protein names in parentheses) are grouped by latent and lytic [immediate-early (IE), early, late] classification, or unassigned if the class is not defined as described in the Materials and Methods. Fisher exact test–identified EBV proteins that significantly discriminated NPC cases from controls are bolded and denoted as *, P < 0.05; **, P < 0.01; ***, P < 0.0001. Putative proteins are italicized and EBV proteins identified by previous studies for NPC prediction are highlighted yellow as described in the Materials and Methods. All values were determined by immunoblot densitometric analysis and reported as a ratio of the antibody isotype-specific detection value divided by the FLAG-tagged EBV protein detection value as defined in the Materials and Methods. White indicates no detection. Red specifies values above the depicted scale. Detection values are totaled (SUM Intensity) at the bottom and follow a separate scale.

Figure 2.

Detection of EBV-specific IgA in sera from incident NPC cases and healthy controls in the SCHS discovery group. Heatmap depicting IgA heavy chain–specific (IgAH) normalized detection values from 20 incident NPC cases and 20 individually matched controls, ordered by years to NPC diagnosis (dx). EBV ORFs (common protein names in parentheses) are grouped by latent and lytic [immediate-early (IE), early, late] classification, or unassigned if the class is not defined as described in the Materials and Methods. Fisher exact test–identified EBV proteins that significantly discriminated NPC cases from controls are bolded and denoted as *, P < 0.05; **, P < 0.01; ***, P < 0.0001. Putative proteins are italicized and EBV proteins identified by previous studies for NPC prediction are highlighted yellow as described in the Materials and Methods. All values were determined by immunoblot densitometric analysis and reported as a ratio of the antibody isotype-specific detection value divided by the FLAG-tagged EBV protein detection value as defined in the Materials and Methods. White indicates no detection. Red specifies values above the depicted scale. Detection values are totaled (SUM Intensity) at the bottom and follow a separate scale.

Close modal
Figure 3.

EBV proteins detected by IgA in the SCHS discovery group excluding EBNA1. EBV proteins that were detected by IgA in sera from incident NPC cases (18 proteins), controls (1 protein), or both (7 proteins) are shown. Proteins in blue were unique to incident NPCs within 4 years of NPC diagnosis. Proteins that discriminated cases from controls are bolded and denoted as *, P < 0.05; **, P < 0.01.

Figure 3.

EBV proteins detected by IgA in the SCHS discovery group excluding EBNA1. EBV proteins that were detected by IgA in sera from incident NPC cases (18 proteins), controls (1 protein), or both (7 proteins) are shown. Proteins in blue were unique to incident NPCs within 4 years of NPC diagnosis. Proteins that discriminated cases from controls are bolded and denoted as *, P < 0.05; **, P < 0.01.

Close modal

We evaluated other immunoglobulins (IgGH, IgMH heavy chain–specific) for NPC predictive biomarkers. A third parameter, IgG heavy and light chain (IgGH+L), was scored as total antibody response due to the shared light chains between antibody isotypes. There were notable high-ranking IgG targets such as BNLF2b IgG that discriminated NPC cases from controls (12 cases vs. 1 control, P = 0.0002; Supplementary Fig. S1), which is consistent with results for BNLF2b IgGH+L (15 cases vs. 1 control, P < 0.0001; Supplementary Fig. S2), but none of these outperformed EBNA1 IgA. IgM, the antibody isotype produced in abundance during primary exposure, showed no value for risk assessment because it was detected in only one case (case ID: NPC1; Supplementary Fig. S3). A sparse logistic regression affirmed that EBNA1 IgA and BNLF2b IgG were the top-ranking discriminators of incident NPC by the greatest assigned positive coefficients (weight: EBNA1 IgA 5.4198, BNLF2b IgG 3.9207; Supplementary Tables S3 and S4). In contrast, a negative weight would indicate discriminatory antibodies in the control group, which could reveal protective antibodies against NPC, but none were comparable in magnitude (Supplementary Tables S3 and S4). We also evaluated EBV proteins detected by IgA or IgG for discriminating NPC cases closer to diagnosis (within 4 years) from those collected at longer time intervals to diagnosis but did not identify a candidate (all P > 0.05). When data were analyzed for the aggregated numbers of EBV proteins detected by IgA or IgG other than EBNA1, we found that IgG detected a greater number of lytic proteins in cases than controls regardless of time interval from serum collection to NPC diagnosis (Fig. 4). In addition, IgA detected a greater number of lytic proteins in cases than controls but only in sera that was collected within 3 years to NPC diagnosis (Fig. 4). While this did not identify a specific biomarker that could discriminate cases closer to diagnosis, the data revealed an important characteristic of EBV pathogenesis. The expanded EBV IgA repertoire is consistent with the hypothesis that the spread of EBV-infected cells (epithelial or B cells) at mucosal sites is a distinguishing feature of those at imminent risk of developing NPC.

Figure 4.

Number of IgG and IgA EBV proteins other than EBNA1 in 20 NPC case–control pairs in the SCHS discovery group. The nEBV that were detected by IgGH or IgAH (heavy chain–specific, cutoff = 0) in NPC cases (black triangle) were compared with control samples (gray inverted triangle). EBNA1 detection is excluded. EBV proteins are classified as lytic, latent, or all gene classes (including unassigned genes) and organized by time to NPC diagnosis (dx) in years. Each point represents one sample and n = number of case–control pairs for each time interval. NPC cases diagnosed with NPC within 2 years in the <3 group are highlighted (blue triangle). Median values (red bar) are indicated. Statistical significance is labeled as *, P < 0.05; **, P < 0.01; ***, P < 0.001, calculated by a Mann–Whitney test.

Figure 4.

Number of IgG and IgA EBV proteins other than EBNA1 in 20 NPC case–control pairs in the SCHS discovery group. The nEBV that were detected by IgGH or IgAH (heavy chain–specific, cutoff = 0) in NPC cases (black triangle) were compared with control samples (gray inverted triangle). EBNA1 detection is excluded. EBV proteins are classified as lytic, latent, or all gene classes (including unassigned genes) and organized by time to NPC diagnosis (dx) in years. Each point represents one sample and n = number of case–control pairs for each time interval. NPC cases diagnosed with NPC within 2 years in the <3 group are highlighted (blue triangle). Median values (red bar) are indicated. Statistical significance is labeled as *, P < 0.05; **, P < 0.01; ***, P < 0.001, calculated by a Mann–Whitney test.

Close modal

On the basis of these findings, we focused on EBNA1 IgA as the leading NPC predictive biomarker. The test for EBNA1 IgA was expanded to 22 additional incident NPC cases from the SCHS (median, 8.84; range, 0.43–14.3 years to diagnosis; Fig. 1). Overall, EBNA1 IgA sensitivity was 85.7% (36/42 cases) and specificity was 92.9% (3/42 false-positives; Table 1). When the data were stratified by years to NPC diagnosis, all 13 sera collected within 4 years to NPC diagnosis were positive for EBNA1 IgA whereas the matched control sera were negative (Table 1; Supplementary Table S5). We conclude that EBNA1 IgA has a sensitivity of 100% within 4 years to diagnosis, and an overall specificity of 92.9% for all controls tested in the SCHS cohort.

Table 1.

Detection of EBNA1 IgA in prediagnostic sera of study participants who developed NPC and those who remained cancer free (controls), the SCHS.

EBNA1 IgA status by time to diagnosisNPC cases (%)Matched controls (%)Fisher exact test two-sided P
All 
 Positive 36 (85.7) 3 (7.1) 8.98 × 10−14 
 Negative 6 (14.3) 39 (92.9)  
0 to 3.99 years    
 EBNA1 pos 13 (100.0) 0 (0.0) 1.92 × 10−9 
 EBNA1 neg 0 (0.0) 13 (100.0)  
4 or more years 
 EBNA1 pos 23 (79.3) 3 (10.3) 1.61 × 10−9 
 EBNA1 neg 6 (20.7) 26 (89.7)  
EBNA1 IgA status by time to diagnosisNPC cases (%)Matched controls (%)Fisher exact test two-sided P
All 
 Positive 36 (85.7) 3 (7.1) 8.98 × 10−14 
 Negative 6 (14.3) 39 (92.9)  
0 to 3.99 years    
 EBNA1 pos 13 (100.0) 0 (0.0) 1.92 × 10−9 
 EBNA1 neg 0 (0.0) 13 (100.0)  
4 or more years 
 EBNA1 pos 23 (79.3) 3 (10.3) 1.61 × 10−9 
 EBNA1 neg 6 (20.7) 26 (89.7)  

We then tested EBNA1 IgA in sera from participants of the SCS, an independent prospective cohort study in Shanghai, P.R. China, where NPC incidence rates were lower than in Singapore. We identified 37 incident NPC cases with prediagnostic sera (median, 10.4 years; range, 0.1–23.9; Fig. 1). EBNA1 IgA sensitivity was 59.5% (22/37 cases) and specificity was 67.6% (12/37 false-positives; Table 2). Because the SCS focused on cancer diagnosis and risk factors, information on other EBV-related comorbidities (such as chronic herpesvirus reactivation, immune suppression, autoimmunity, or acute EBV infection) that could have explained the EBNA1 IgA false-positive results, were not collected. Interestingly, all seven sera from participants diagnosed with NPC within 4 years were positive for EBNA1 IgA and all seven matched controls were negative (Table 2; Supplementary Table S5). This confirmed that EBNA1 IgA has a sensitivity of 100% within 4 years to diagnosis in both the SCHS and SCS case–control cohorts.

Table 2.

Detection of EBNA1 IgA in prediagnostic sera of study participants who developed NPC and those who remained cancer free (controls), the SCS.

EBNA1 IgA status by time to diagnosisNPC cases (%)Matched controls (%)Fisher exact test two-sided P
All 
 Positive 22 (59.5) 12 (32.4) 3.5 × 10−2 
 Negative 15 (40.5) 25 (67.6)  
0 to 3.99 years 
 EBNA1 pos 7 (100.0) 0 (0.0) 5.8 × 10−4 
 EBNA1 neg 0 (0.0) 7 (100.0)  
4 or more years 
 EBNA1 pos 15 (50.0) 12 (40.0) 6.0 × 10−1 
 EBNA1 neg 15 (50.0) 18 (60.0)  
EBNA1 IgA status by time to diagnosisNPC cases (%)Matched controls (%)Fisher exact test two-sided P
All 
 Positive 22 (59.5) 12 (32.4) 3.5 × 10−2 
 Negative 15 (40.5) 25 (67.6)  
0 to 3.99 years 
 EBNA1 pos 7 (100.0) 0 (0.0) 5.8 × 10−4 
 EBNA1 neg 0 (0.0) 7 (100.0)  
4 or more years 
 EBNA1 pos 15 (50.0) 12 (40.0) 6.0 × 10−1 
 EBNA1 neg 15 (50.0) 18 (60.0)  

We also examined whether a combination of EBV-specific IgA could reduce the false-positive rate of the EBNA1 IgA test. We focused on EBV IgA proteins that differentiated NPC cases (Fig. 3) that were detected more than once within 4 years to diagnosis. From the initial EBV library screen performed on the 20 case–control pairs from the SCHS discovery group, we expanded our analysis on the remaining 31 pairs from both SCHS and SCS whose sera were collected within 10 years to diagnosis. However, these eight EBV proteins (EBNA3A, EA p138, BFRF1, BNLF2b, BORF2, SM, TK, BLRF1) did not increase the specificity (false-positive rate) of EBNA1 IgA alone, nor did they discriminate cases by time (<4 years vs. 4+ years; Supplementary Fig. S4). Because of the limited serum samples available, we were not able to test the whole EBV library panel against these additional samples, but we were able to extend the IgA analysis against lytic proteins with the additional specimens using an alternative method. Immunoblots performed on EBV-reactivated cell lysates from HH514-16 cells is an established method for scoring IgA and IgG molecular diversity against EBV lytic proteins (29, 30). We found that IgA detection of lytic proteins could indeed distinguish NPC cases by time (P < 0.05, <3 years vs. 3+ years to diagnosis), with no difference in the controls (Supplementary Fig. S5). This confirms that IgA diversity is increased in NPC cases closer to diagnosis and this feature might be exploited in the future to stratify EBNA1 IgA test-positive cases by time.

The use of EBNA1 IgA for NPC early detection and risk assessment has been evaluated by numerous studies (12–18, 20, 31–34). Strikingly, we identified incident NPC cases with 100% sensitivity and 100% specificity 4 years before clinical manifestation when control subjects were matched according to predefined criteria. When extended to all control subjects, the specificity was 92.9% in the endemic cohort (SCHS) and 67.6% in a lower risk cohort (SCS). We acknowledge that the results are based on a relatively small sample size within 4 years (20 pairs of prediagnostic sera) from both cohorts and warrants additional validation in prospective cohorts with larger sample sizes. Prospective cohorts with prediagnostic NPC sera are rare as discussed in a recent review of EBV serology to evaluate NPC risk (35). Prior prospective studies with incident NPC cases ranged between 8 and 252 (median 41) incident cases but it is not clear how many of these cases fall within the time interval of 4 years to diagnosis (35). Thus, many studies rely on cross-sectional samples taken at the time of NPC diagnosis to infer a change in serology that mark early stages of NPC. In addition, it would be valuable to explore additional cohorts that collected medical information related to other EBV comorbidities (such as acute EBV infection, chronic herpesvirus reactivation, immune dysfunction or suppression) which could improve our predictive power with an EBNA1 IgA test. A recent meta-analysis including 47 studies involving 8,382 patients with NPC mainly from high-risk populations in China, reported an averaged sensitivity of 86% and specificity of 87% for EBNA1 IgA for the early diagnosis of NPC (36). EBNA1 IgA is often cited as a top candidate biomarker for NPC early diagnosis and risk assessment, but suboptimal specificity (false-positives) often precludes its use for population screening. For example, ELISA-based detection of EBNA1 IgA in a Taiwanese cohort yielded a specificity of 58% when sensitivity was set at an optimized threshold of 80% (16).

The superior performance of EBNA1 IgA in our study may be attributed to the choice in methodology and the EBV strain. Immunoblotting is typically used as a confirmatory assay for ELISA-based tests for early diagnosis of NPC (12, 19, 20, 29, 37). It is known that immunoblots can deliver a high degree of sensitivity and specificity, as demonstrated by the original human immunodeficiency virus (HIV) molecular diagnostic test used in clinical pathology (38). The use of multiplex immunoblots enables the denaturation and size resolution of EBV proteins which greatly reduces the false-positives and false-negatives that are commonly found in semi-denaturing methods (such as protein array and ELISA-based assays). In terms of EBV strain, the prototypic B95-8 strain is often used in serology assays. Both Akata and B95-8 strains are classified as type I EBV, but the Akata strain shares greater sequence similarity and identity with EBV strains found in NPC tumors (24, 39). Importantly, the Akata EBNA1 amino acid sequence is almost identical to EBNA1 from NPC tumors whereas the B95–8 EBNA1 amino acid sequence is dissimilar from NPC tumors in at least three residues in the immunodominant epitopes (amino acids 382–410, 413–452; Supplementary Fig. S6; refs. 19, 40). We also considered the expression system as prior studies have used bacterial expression systems, peptide microarrays, or in vitro translated cell-free expression systems (17, 34, 41). Compared with these alternative methods, our EBV mammalian expression library produces FLAG-tagged full-length EBV proteins in human cells which best mirror the posttranslational modifications of EBV proteins in vivo. Upon refinement of test parameters, it might be possible to scale this assay to a clinical setting possibly in the format of a rapid test (20).

Increased EBV lytic gene expression and expansion of the IgA repertoire against EBV are thought to immediately proceed NPC development (34, 42). This is thought to originate from circulating EBV-reactivated B cells that infiltrate nasopharyngeal epithelia and/or EBV reactivation in infected epithelial cells in response to cellular differentiation in the stratified epithelium (43–45). These reactivated cells can be abortive and therefore the diversity of induced EBV proteins can vary widely from cell to cell as evidenced by single-cell RNA sequencing (44). Thus, it is not clear whether a specific assortment of lytic EBV proteins can discriminate incident NPC cases from healthy individuals. However, EBNA1, which is expressed in all types of EBV infection, was consistently detected by IgA within 4 years in both cohorts (100% sensitivity), and only showed a precipitous drop in detection in samples greater than 10 years to diagnosis. Although EBNA1 IgA is present in all seropositive individuals, we postulate that our assay detects elevated EBNA1 IgA, most likely reflecting heightened EBV immunity at the nasopharyngeal mucosa before the onset of NPC.

In the later time frame (>4 years to diagnosis), the SCS cohort appears to have reduced discrimination by EBNA1 IgA, but this could be explained because most samples were collected beyond 10 years to diagnosis. At present, it is not clear why more SCS control samples tested positive for EBNA1 IgA in this latter time frame than in the endemic SCHS cohort. These individuals may have an EBV-related comorbidity with uncontrolled EBV levels and immunity. Furthermore, identifying an additional parameter that stratifies EBNA1 IgA test-positives into imminent cases would be important. While an increased IgG repertoire was detected some years before diagnosis (4+ years), the increased IgA repertoire was limited to 2 years before diagnosis. This finding adds to our knowledge of EBV immune surveillance in the pathology of NPC. One scenario consistent with these observations would be that persons at risk of developing NPC are distinguished by chronically high levels of EBV lytic infection in their blood some years before NPC presentation, which would traffic to mucosal sites to result in high levels of lytic infection at the nasopharyngeal mucosa a few years before NPC presentation.

Tackling the issue of cancer prevention requires a holistic approach. Toward this goal, EBV serology may be combined with EBV DNA analysis to discriminate EBNA1 IgA test-positives already at early stage NPC. The short half-life of plasma cell-free EBV DNA tracks remarkably well with radiotherapy and the response of the NPC tumor (46). Because the source of the cell-free EBV DNA is likely tumor derived, its potential as a risk assessment marker may be limited although this remains to be thoroughly examined (47). Upon implementation of a population screen, a clear clinical follow-up by MRI or endoscopy would have to be defined for test-positives. Given that HIV-positive patients treated with ganciclovir reduced the risk of Kaposi sarcoma by 93% (48), it might be possible that antiviral therapy can be used prophylactically. An EBNA1 inhibitor effective at restricting NPC tumor cell growth in a xenograft model is in clinical trials but its utility in NPC prevention remains to be determined (49, 50). At present, an EBNA1 IgA risk assessment test would provide individuals identified as high-risk an opportunity for regular follow-up by clinical examination and endoscopy.

S. Paudel reports grants from NIH T32 CA186873 during the conduct of the study. B.E. Warner reports grants from NIH T32 CA186873 during the conduct of the study. J.-M. Yuan reports grants from NIH during the conduct of the study; in addition, J.-M. Yuan has a provisional patent for US 63/336,590 issued. K.H.Y. Shair reports grants from NIH during the conduct of the study; in addition, K.H.Y. Shair has a patent for Serological Biomarker for the Risk Prediction of Epstein-Barr virus-associated nasopharyngeal carcinoma (U.S. provisional patent 63/336,590) issued to University of Pittsburgh. No disclosures were reported by the other authors.

S. Paudel: Data curation, formal analysis, funding acquisition, validation, investigation, visualization, methodology, writing–original draft, writing–review and editing. B.E. Warner: Data curation, formal analysis, funding acquisition, validation, investigation, visualization, methodology, writing–original draft, writing–review and editing. R. Wang: Data curation, formal analysis, validation, visualization, writing–original draft, writing–review and editing. J. Adams-Haduch: Resources, writing–review and editing. A.S. Reznik: Data curation, formal analysis, validation, visualization, writing–review and editing. J. Dou: Data curation, formal analysis, visualization, writing–review and editing. Y. Huang: Data curation, formal analysis, visualization, writing–review and editing. Y.-T. Gao: Resources, writing–review and editing. W.-P. Koh: Resources, writing–review and editing. A. Bäckerholm: Data curation, software, formal analysis, investigation, visualization, writing–review and editing. J. Yuan: Conceptualization, resources, data curation, formal analysis, supervision, funding acquisition, validation, visualization, methodology, writing–original draft, project administration, writing–review and editing. K.H.Y. Shair: Conceptualization, resources, data curation, formal analysis, supervision, funding acquisition, validation, investigation, visualization, methodology, writing–original draft, project administration, writing–review and editing.

We thank Dr. Christopher Whitehurst (New York Medical College, Valhalla, NY) for the EBV B95-8–derived BPLF1 construct and Dr. Sumita Bhaduri-McIntosh (University of Florida, Gainesville, FL) for providing EBV seropositive and seronegative sera with known serologic spectra.

The Shanghai Cohort Study was supported by the NIH (grant nos. R01CA043092, R01CA144034, and UM1CA182876), and the Singapore Chinese Health Study was supported by the NIH (grant nos. R01CA080205, R01CA144034, and UM1CA182876). The current work was also partially supported by the UPMC Hillman Cancer Center Head and Neck Cancer Specialized Program of Research Excellence NIH grant # P50 CA097190 pilot award. S. Paudel and B.E. Warner were supported by NIH training grant no. T32 CA186873.

The publication costs of this article were defrayed in part by the payment of publication fees. Therefore, and solely to indicate this fact, this article is hereby marked “advertisement” in accordance with 18 USC section 1734.

Note: Supplementary data for this article are available at Clinical Cancer Research Online (http://clincancerres.aacrjournals.org/).

1.
Yu
MC
,
Yuan
J-M
.
Epidemiology of nasopharyngeal carcinoma
.
Semin Cancer Biol
2002
;
12
:
421
9
.
2.
Parkin
WSL
,
Ferlay
J
,
Raymond
L
,
Young
J
.
Cancer incidence in five continents. Volume VII
.
IARC Sci Publ
1997
;
i–xxxiv
:
1
1240
.
3.
Ferlay
J
,
Colombet
M
,
Soerjomataram
I
,
Parkin
DM
,
Piñeros
M
,
Znaor
A
, et al
.
Cancer statistics for the year 2020: an overview
.
Int J Cancer
2021
[Online ahead of print].
4.
Lee
AWM
,
Sze
WM
,
Au
JSK
,
Leung
SF
,
Leung
TW
,
Chua
DTT
, et al
.
Treatment results for nasopharyngeal carcinoma in the modern era: the Hong Kong experience
.
Int J Radiat Oncol Biol Phys
2005
;
61
:
1107
16
.
5.
Tsao
SW
,
Yip
YL
,
Tsang
CM
,
Pang
PS
,
Lau
VMY
,
Zhang
G
, et al
.
Etiological factors of nasopharyngeal carcinoma
.
Oral Oncol
2014
;
50
:
330
8
.
6.
Cohen
JI
,
Fauci
AS
,
Varmus
H
,
Nabel
GJ
.
Epstein-barr virus: an important vaccine target for cancer prevention
.
Sci Transl Med
2011
;
3
:
107fs7
.
7.
Middeldorp
JM
.
Epstein-Barr virus-specific humoral immune responses in health and disease
.
Curr Top Microbiol Immunol
2015
;
391
:
289
323
.
8.
Sun
C
,
Chen
X-c
Kang
Y-f
Zeng
M-s
.
The status and prospects of Epstein–Barr virus prophylactic vaccine development
.
Front Immunol
2021
;
12
:
677027
.
9.
Chan
KCA
,
Woo
JKS
,
King
A
,
Zee
BCY
,
Lam
WKJ
,
Chan
SL
, et al
.
Analysis of plasma epstein–barr virus DNA to screen for nasopharyngeal cancer
.
N Engl J Med
2017
;
377
:
513
22
.
10.
Lam
WKJ
,
Jiang
P
,
Chan
KCA
,
Cheng
SH
,
Zhang
H
,
Peng
W
, et al
.
Sequencing-based counting and size profiling of plasma Epstein-Barr virus DNA enhance population screening of nasopharyngeal carcinoma
.
Proc Nat Acad Sci U S A
2018
;
115
:
E5115
E24
.
11.
Lam
WKJ
,
Jiang
P
,
Chan
KCA
,
Peng
W
,
Shang
H
,
Heung
MMS
, et al
.
Methylation analysis of plasma DNA informs etiologies of Epstein-Barr virus-associated diseases
.
Nat Commun
2019
;
10
:
3256
.
12.
Paramita
DK
,
Fachiroh
J
,
Haryana
SM
,
Middeldorp
JM
.
Two-step Epstein-Barr virus immunoglobulin a enzyme-linked immunosorbent assay system for serological screening and confirmation of nasopharyngeal carcinoma
.
Clin Vaccine Immunol
2009
;
16
:
706
11
.
13.
Liu
Y
,
Huang
Q
,
Liu
W
,
Liu
Q
,
Jia
W
,
Chang
E
, et al
.
Establishment of VCA and EBNA1 IgA-based combination by enzyme-linked immunosorbent assay as preferred screening method for nasopharyngeal carcinoma: a two-stage design with a preliminary performance study and a mass screening in southern China
.
Int J Cancer
2012
;
131
:
406
16
.
14.
Liu
Z
,
Ji
MF
,
Huang
QH
,
Fang
F
,
Liu
Q
,
Jia
WH
, et al
.
Two epstein-barr virus-related serologic antibody tests in nasopharyngeal carcinoma screening: results from the initial phase of a cluster randomized controlled trial in southern China
.
Am J Epidemiol
2013
;
177
:
242
50
.
15.
Chang
C
,
Middeldorp
J
,
Yu
KJ
,
Juwana
H
,
Hsu
W-L
,
Lou
P-J
, et al
.
Characterization of ELISA detection of broad-spectrum anti-Epstein-Barr virus antibodies associated with nasopharyngeal carcinoma
.
J Med Virol
2013
;
85
:
524
9
.
16.
Coghill
AE
,
Hsu
W-L
,
Pfeiffer
RM
,
Juwana
H
,
Yu
KJ
,
Lou
P-J
, et al
.
Epstein–barr virus serology as a potential screening marker for nasopharyngeal carcinoma among high-risk individuals from multiplex families in Taiwan
.
Cancer Epidemiol Biomarkers Prev
2014
;
23
:
1213
9
.
17.
Coghill
AE
,
Pfeiffer
RM
,
Proietti
C
,
Hsu
W-L
,
Chien
Y-C
,
Lekieffre
L
, et al
.
Identification of a Novel, EBV-based antibody risk stratification signature for early detection of nasopharyngeal carcinoma in Taiwan
.
Clin Cancer Res
2018
;
24
:
1305
14
.
18.
Simon
J
,
Liu
Z
,
Brenner
N
,
Yu
KJ
,
Hsu
W-L
,
Wang
C-P
, et al
.
Validation of an Epstein-Barr virus antibody risk stratification signature for nasopharyngeal carcinoma by use of multiplex serology
.
J Clin Microbiol
2020
;
58
:
e00077
20
.
19.
Fachiroh
J
,
Paramita
DK
,
Hariwiyanto
B
,
Harijadi
A
,
Dahlia
HL
,
Indrasari
SR
, et al
.
Single-assay combination of Epstein-Barr virus (EBV) EBNA1- and viral capsid antigen-p18-derived synthetic peptides for measuring anti-EBV immunoglobulin G (IgG) and IgA antibody levels in sera from nasopharyngeal carcinoma patients: options for field screening
.
J Clin Microbiol
2006
;
44
:
1459
67
.
20.
Paramita
DK
,
Fachiroh
J
,
Haryana
SM
,
Middeldorp
JM
.
Evaluation of commercial EBV RecombLine assay for diagnosis of nasopharyngeal carcinoma
.
J Clin Virol
2008
;
42
:
343
52
.
21.
Hankin
JH
,
Stram
DO
,
Arakawa
K
,
Park
S
,
Low
S-H
,
Lee
H-P
, et al
.
Singapore chinese health study: development, validation, and calibration of the quantitative food frequency questionnaire
.
Nutr Cancer
2001
;
39
:
187
95
.
22.
Ross
RK
,
Yu
MC
,
Henderson
BE
,
Yuan
JM
,
Qian
GS
,
Tu
JT
, et al
.
Urinary aflatoxin biomarkers and risk of hepatocellular carcinoma
.
Lancet
1992
;
339
:
943
6
.
23.
Sample
JT
,
Marendy
EM
,
Hughes
DJ
,
Sample
CE
.
The Epstein–Barr virus genome
.
DNA tumor viruses
.
New York, NY
:
Springer US
;
2009
. p.
241
58
.
24.
Lin
Z
,
Wang
X
,
Strong
MJ
,
Concha
M
,
Baddoo
M
,
Xu
G
, et al
.
Whole-genome sequencing of the akata and mutu Epstein-Barr virus strains
.
J Virol
2013
;
87
:
1172
82
.
25.
Tibshirani
R
.
Regression shrinkage and selection via the Lasso
.
J R Stat Soc Series B Methodol
1996
;
58
:
267
88
.
26.
Seabold
S
,
Perktold
J
.
Statsmodels: econometric and statistical modeling with python
. Proceedings of the 9th Python in Science Conference
2010
:
92
6
.
27.
Frappier
L
.
EBNA1
.
Curr Top Microbiol Immunol
2015
;
391
:
3
34
.
28.
Dheekollu
J
,
Malecka
K
,
Wiedmer
A
,
Delecluse
H-J
,
Chiang
AKS
,
Altieri
DC
, et al
.
Carcinoma-risk variant of EBNA1 deregulates Epstein-Barr Virus episomal latency
.
Oncotarget
2017
;
8
:
7248
64
.
29.
Fachiroh
J
,
Schouten
T
,
Hariwiyanto
B
,
Paramita Dewi
K
,
Harijadi
A
,
Haryana Sofia
M
, et al
.
Molecular diversity of Epstein-Barr virus IgG and IgA antibody responses in nasopharyngeal carcinoma: a comparison of Indonesian, Chinese, and European subjects
.
J Infect Dis
2004
;
190
:
53
62
.
30.
Rabson
M
,
Heston
L
,
Miller
G
.
Identification of a rare Epstein-Barr virus variant that enhances early antigen expression in Raji cells
.
Proc Natl Acad Sci U S A
1983
;
80
:
2762
6
.
31.
Zeng
Y
,
Zhang
LG
,
Wu
YC
,
Huang
YS
,
Huang
NQ
,
Li
JY
, et al
.
Prospective studies on nasopharyngeal carcinoma in Epstein-Barr virus IgA/VCA antibody-positne persons in Wuzhou City, China
.
Int J Cancer
1985
;
36
:
545
7
.
32.
Cheng
W-m
Chan
KH
,
Chen
H-l
Luo
R-x
Ng
SP
,
Luk
W
, et al
.
Assessing the risk of nasopharyngeal carcinoma on the basis of EBV antibody spectrum
.
Int J Cancer
2002
;
97
:
489
92
.
33.
Yu
KJ
,
Hsu
W-L
,
Pfeiffer
RM
,
Chiang
C-J
,
Wang
C-P
,
Lou
P-J
, et al
.
Prognostic utility of anti-EBV antibody testing for defining NPC risk among individuals from high-risk NPC families
.
Clin Cancer Res
2011
;
17
:
1906
14
.
34.
Simon
J
,
Brenner
N
,
Reich
S
,
Langseth
H
,
Hansen
BT
,
Ursin
G
, et al
.
Nasopharyngeal carcinoma patients from Norway show elevated Epstein-Barr virus IgA and IgG antibodies prior to diagnosis
.
Cancer Epidemiol
2022
;
77
:
102117
.
35.
Sinha
S
,
Dickey
BL
,
Coghill
AE
.
Utility of Epstein-Barr virus (EBV) antibodies as screening markers for nasopharyngeal carcinoma: a narrative review
.
Ann Nasopharynx Cancer
2022
;
6
:
6
.
36.
Liu
W
,
Chen
G
,
Gong
X
,
Wang
Y
,
Zheng
Y
,
Liao
X
, et al
.
The diagnostic value of EBV-DNA and EBV-related antibodies detection for nasopharyngeal carcinoma: a meta-analysis
.
Cancer Cell Int
2021
;
21
:
164
.
37.
Karray
H
,
Ayadi
W
,
Fki
L
,
Hammami
A
,
Daoud
J
,
Drira
MM
, et al
.
Comparison of three different serological techniques for primary diagnosis and monitoring of nasopharyngeal carcinoma in two age groups from Tunisia
.
J Med Virol
2005
;
75
:
593
602
.
38.
Fearon
M
.
The laboratory diagnosis of HIV infections
.
Can J Infect Dis Med Microbiol
2005
;
16
:
26
30
.
39.
Kwok
H
,
Wu
CW
,
Palser
AL
,
Kellam
P
,
Sham
PC
,
Kwong
DLW
, et al
.
Genomic diversity of Epstein-Barr virus genomes isolated from primary nasopharyngeal carcinoma biopsy samples
.
J Virol
2014
;
88
:
10662
72
.
40.
Chen
MR
,
Tsai
CH
,
Wu
FF
,
Kan
SH
,
Yang
CS
,
Chen
JY
, et al
.
The major immunogenic epitopes of Epstein-Barr virus (EBV) nuclear antigen 1 are encoded by sequence domains which vary among nasopharyngeal carcinoma biopsies and EBV-associated cell lines
.
J Gen Virol
1999
;
80
:
447
55
.
41.
Song
L
,
Song
M
,
Camargo
MC
,
Van Duine
J
,
Williams
S
,
Chung
Y
, et al
.
Identification of anti-Epstein-Barr virus (EBV) antibody signature in EBV-associated gastric carcinoma
.
Gastric Cancer
2021
;
24
:
858
67
.
42.
Tsao
SW
,
Tsang
CM
,
Lo
KW
.
Epstein–Barr virus infection and nasopharyngeal carcinoma
.
Philos Trans R Soc Lond B Biol Sci
2017
;
372
:
20160270
.
43.
Temple
RM
,
Zhu
J
,
Budgeon
L
,
Christensen
ND
,
Meyers
C
,
Sample
CE
, et al
.
Efficient replication of Epstein–Barr virus in stratified epithelium in vitro
.
Proc Natl Acad Sci U S A
2014
;
111
:
16544
9
.
44.
Ziegler
P
,
Tian
Y
,
Bai
Y
,
Abrahamsson
S
,
Bäckerholm
A
,
Reznik
AS
, et al
.
A primary nasopharyngeal three-dimensional air-liquid interface cell culture model of the pseudostratified epithelium reveals differential donor- and cell type-specific susceptibility to Epstein-Barr virus infection
.
PLoS Pathog
2021
;
17
:
e1009041
.
45.
Caves
EA
,
Cook
SA
,
Lee
N
,
Stoltz
D
,
Watkins
S
,
Shair
KHY
, et al
.
Air-liquid interface system to understand Epstein-Barr virus-associated nasopharyngeal carcinoma
.
mSphere
2018
;
3
:
e00350
18
.
46.
Lo
YM
,
Leung
SF
,
Chan
LY
,
Chan
AT
,
Lo
KW
,
Johnson
PJ
, et al
.
Kinetics of plasma Epstein-Barr virus DNA during radiation therapy for nasopharyngeal carcinoma
.
Cancer Res
2000
;
60
:
2351
5
.
47.
Tay
JK
,
Siow
CH
,
Goh
HL
,
Lim
CM
,
Hsu
PP
,
Chan
SH
, et al
.
A comparison of EBV serology and serum cell-free DNA as screening tools for nasopharyngeal cancer: results of the Singapore NPC screening cohort
.
Int J Cancer
2020
;
146
:
2923
31
.
48.
Martin
DF
,
Kuppermann
BD
,
Wolitz
RA
,
Palestine
AG
,
Li
H
,
Robinson
CA
, et al
.
Oral ganciclovir for patients with cytomegalovirus retinitis treated with a ganciclovir implant
.
N Engl J Med
1999
;
340
:
1063
70
.
49.
Messick
TE
,
Smith
GR
,
Soldan
SS
,
McDonnell
ME
,
Deakyne
JS
,
Malecka
KA
, et al
.
Structure-based design of small-molecule inhibitors of EBNA1 DNA binding blocks Epstein-Barr virus latent infection and tumor growth
.
Sci Transl Med
2019
;
11
:
eaau5612
.
50.
Jiang
L
,
Lung
HL
,
Huang
T
,
Lan
R
,
Zha
S
,
Chan
LS
, et al
.
Reactivation of Epstein-Barr virus by a dual-responsive fluorescent EBNA1-targeting agent with Zn(2+)-chelating function
.
Proc Natl Acad Sci U S A
2019
;
116
:
26614
24
.

Supplementary data