Characterization of ELISA Detection of Broad-spectrum Anti- Epstein-Barr Virus Antibodies Associated with Nasopharyngeal Carcinoma

Abstract

Epstein-Barr virus (EBV) infection is associated with undifferentiated nasopharyngeal carcinomas (NPC). A distinct seroreactivity pattern to EBV is predictive of subsequent risk of sporadic and familial nasopharyngeal carcinomas. There are currently no accepted screening tools for guiding the clinical management of individuals at high-risk for nasopharyngeal carcinomas, particularly unaffected relatives from nasopharyngeal carcinoma multiplex families. Therefore, the reproducibility of a panel of largely synthetic peptide-based anti-EBV antibody ELISAs was evaluated and their ability to distinguish nasopharyngeal carcinoma cases from controls was explored. IgG and IgA antibodies against 6 different EBV antigens (10 assays, total) were tested on sera from 97 individuals representing the full spectrum of anti-EBV seroprevalence (i.e., healthy individuals with no known EBV seroreactivity, healthy individuals with known EBV seroreactivity, and nasopharyngeal carcinoma cases). Each specimen was tested in triplicate to assess within-batch and across-batch variation, and the triplicate testing was repeated on two separate days. Reproducibility was assessed by the coefficients of variation (CV) and intraclass correlation coefficients (ICC). All markers were detectable in 17% or more of samples. For all but one marker, the overall, within-batch, and across-batch CVs were below 15%, and the ICCs were above 70% for all but three markers. Sensitivity of these markers to detect prevalent nasopharyngeal carcinomas ranged from 22–100%, and among unaffected controls, most distinguished those with and without known seropositivity. In conclusion, a large number of EBV markers can be measured reliably in serum samples using peptide-based anti-EBV ELISAs.

Introduction

Epstein-Barr virus (EBV) infection is closely associated with the development of nearly all undifferentiated nasopharyngeal carcinomas, a rare cancer with elevated rates in South-East Asian and Chinese populations. Nasopharyngeal carcinoma cases are known to develop a distinct anti-EBV antibody profile that is characterized by the presence of antibodies, particularly IgA antibodies against structural (viral capdid antigen or VCA) and some non-structural (early antigen or EA, Epstein-Barr nuclear anigen or EBNA, deoxyribonuclease or DNase) proteins [Henle and Henle, 1976; Baskies et al., 1979; Cheng et al., 1980; Chen et al., 1987; Fachiroh et al., 2004]. This distinct seroreactivity pattern was shown to be predictive of subsequent risk of disease in large prospective cohorts conducted in nasopharyngeal carcinoma high-risk areas [Chien et al., 2001; Cao et al., 2011]. The EBV seroreactivity pattern observed in nasopharyngeal carcinoma patients has been mapped to molecularly-defined viral proteins and peptide epitopes on these proteins, allowing for the design of more simplified, standardized assays [Dardari et al., 2001; Fachiroh et al., 2006; Paramita et al., 2007].

Individuals with a family history of nasopharyngeal carcinomas have been shown to have a much higher absolute risk of developing the disease [Friborg et al., 2005; Yu et al., 2009]. In a previous study in Taiwan, healthy individuals from 358 families with 2 or more individuals affected with nasopharyngeal carcinomas were followed for incident nasopharyngeal carcinomas. The rates of nasopharyngeal carcinomas were observed to be >10 times higher among unaffected family members from multiplex families compared to the general population (90 per 100,000 person-years vs. 7 per 100,000 person-years) [Yu et al., 2011]. Currently there are no accepted screening tools that can be applied to this high-risk population for guiding their clinical management. In a recent study, it was shown that among individuals within nasopharyngeal carcinoma multiplex families, the presence of IgA antibodies against EBNA-1 was associated with a 6.6-fold increased risk of nasopharyngeal carcinomas [Yu et al., 2011].

Taken together, previous work suggests that EBV serology screening has promise as a clinical tool to guide clinicians in the management of individuals at high-risk of nasopharyngeal carcinomas, but will require assays for EBV markers that are highly reproducible and both highly sensitive and specific for nasopharyngeal carcinomas. Therefore, the goal of this study is to define the reproducibility of a panel of largely synthetic peptide-based anti-EBV antibody ELISA tests and provide additional initial findings exploring their performance for distinguishing nasopharyngeal carcinoma cases from controls. IgA antibodies to EBV peptides previously associated with nasopharyngeal carcinoma risk were the focus and, for completeness, assays that detect IgG antibodies to these same peptides were also evaluated [Cheng et al., 2002; Chan et al., 2003; Tang et al., 2007; Liu et al., 2012].

Methods

Study Population

97 individuals selected for this study were participants in our previously described case-control [Hildesheim et al., 2001] and family studies of nasopharyngeal carcinomas in Taiwan [Yang et al., 2005]. Individuals represented the spectrum of anti-EBV seroprevalence for which the assays would be expected to perform well. Therefore, 20 histologically confirmed nasopharyngeal carcinoma cases from the case-control study along with three groups of controls from the family study, all of whom had a family history of nasopharyngeal carcinomas in 2 or more first-degree relatives, were selected. Controls were selected from the family study because they had a wider range of seroreactivity than the controls from the case-control study and were therefore more informative for purposes of assessing assay performance [Pickard et al., 2004]. Control group 1 were 32 unaffected individuals with known seroreactivity to 2 or more of the three anti-EBV antibody tests (anti-EBNA1 IgA, anti-VCA IgA, and anti-DNase, based on previously published results); control group 2 were 25 unaffected individuals with known seroreactivity to only one of the three anti-EBV antibody test results; and control group 3 were 20 unaffected individuals expected to be EBV infected but with no seroreactivity to the three anti-EBV antibody test results. Sera from nasopharyngeal carcinoma cases were obtained from blood collected prior to the initiation of treatment. Ethical approval by an institutional review board (IRB) was obtained in both the U.S. and Taiwan. All subjects provided informed consent.

Specimen Selection and Batching

For each individual selected for study, three identical, masked aliquots of serum for testing were prepared. Two aliquots were placed randomly within one testing box and a third aliquot was placed randomly in a separate box. One sample had low quantity of serum and therefore was aliquoted to be tested only in duplicate within one box. A total of 7 testing boxes were shipped to the laboratory for testing. At the laboratory, aliquots from each box were further aliquoted onto duplicate wells of a single 96-well ELISA plate. Each plate consisted of 80 to 84 test wells, while the remaining wells on each plate were utilized for positive and negative controls as described further below. Finally, each plate for each assay was prepared and tested on two separate days (by the same technician), with a median of 16 days apart (range: 2 to 43 days). Thus, each specimen was tested in triplicate to assess within-batch and across-batch variation, and the triplicate testing was repeated on two separate days for a total of 6 test results per assay.

Anti-EBV Antibody Tests

High binding 96-well ELISA plates (Costar 9018, Thermo Fisher Scientific, Pittsburgh, Pennsylvania, U.S.A.) were coated at 4°C overnight with 100 μl per well of 50mM carbonate-bicarbonate buffer, pH 9.6 containing either 1.0 μg/ml EBV-peptides (VCA-p18, EBNA1, EAp138, EA-diffuse or EAd and Zebra) or nuclear EA extract (EA ext) in 150 μl, exactly as previously described (Table 1) [Meij et al., 1999; Fachiroh et al., 2006; Paramita et al., 2007; Asito et al., 2010]. Subsequently, wells were emptied and filled with 200 μl 3% bovine serum albumin (BSA) in phosphate-buffered saline (PBS) and incubated for 1 hour at 37°C, and then washed 4 times with PBS containing 0.05% Tween-20 (PBS-T). Preparation of ELISA plates was performed on each testing day using a single lot of reagents prepared for this study. Individual wells were incubated for 1 hour with 100 μl of human sera diluted 1:100 in sample buffer (PBS-T with 0.1% Triton-X, 1% BSA and 1% normal rabbit serum) at 37°C, washed 4 times with PBS-T and incubated for 1 hour at 37°C with rabbit anti-human IgG- or IgA-HRP conjugate (DAKO, Copenhagen, Denmark), diluted 1:1000 in sample buffer. After 4 washings with PBS-T, 100 μl TMB substrate solution (Sigma-Aldrich, St. Louis, Missouri, U.S.A.) was added and after 30 minutes, the reaction was stopped by adding 100 μl 1M H₂SO₄. Absorbance at 450 nm was measured in a Tecan Spectrafluor ELISA reader (software: Xfluor-4, Tecan Benelux, Männedorf, Switzerland). In each ELISA test, 2 known EBV IgG/IgA positive reference sera were tested at 1:100 in duplicate calibrators and the cut-off value (COV) for each ELISA plate was defined by calculating the mean OD450 reactivity + 3 multiplied by the SD of 4 defined EBV negative sera (1:100) tested in duplicate. Values below the COV were considered negative. Results were reported as the mean of the duplicate absorbance value observed for each test specimen, divided by the COV.

Table 1.

Sequences of Target Peptides Used in Anti-EBV ELISAs

Peptide EBNA (BKRF1): “PPRRPPPGRRPFFHPVGEADYFEYHQEDGEPDVPPGAIEQGPADDPGEGPSTGPRG”a

Peptide VCA-p18 (BFRF3): “STAVAQSATPSVSSSISSLRAATSGATAAAAVDTGSGGGGQPHDTAPRGARKKQ”a

Peptide EAd-p47 (BMRF1): “TVSPSPSPPPPPRTPTWESHSSNTALERPLAVQLARKRTSSEARQ”a

Peptide EAd-p138 (BALF2): “KSVRVPLYDKEVFPEGVPQLRQFYNSDLSLGIDAEGKL” +“KTGTNGPGNYAVEHLVYAASFSPNLLARYAYYLQF”a

Peptide Zebra (BZLF1): “MMDPNSTSEDVKFTPDPYQVPFVQAFDQATRVYQDLGGPSQAPLPCV”

^apatent protected and proprietary to Cyto-Barr B.V. (Zuidhorn, Netherlands)

Statistical Analysis

Results obtained from the laboratory were multiplied by 100 and log transformed for normalization. Analyses were performed on the log-transformed values. Overall positivity rates were calculated for all specimens tested, and among positives the geometric mean titer (GMT) was obtained. Coefficients of variation (CVs; overall, within plate and between plate) for each assay were computed to obtain estimates for assay variability, both overall and separately within and between testing batches. To estimate the proportion of the total assay variability attributed to variability between individuals, intraclass correlation coefficients (ICCs) for each assay was calculated using variance components analyses formulated as a linear mixed model, fitted with PROC MIXED (SAS® version 9.2, Cary, North Carolina, U.S.A.). On each testing day, ELISA plates were prepared freshly, with newly diluted antigen and conjugates, introducing variability, not encountered if industrial freeze-dried plates and reagents are used. Thus, the random effect in the mixed models included testing date, box, and repeated aliquots of the same person. Study group (defined by nasopharyngeal carcinomas and previously available EBV antibody status as specified above) was included as a fixed effect. For 3 of the assays (EA ext IgA, EA ext IgG, and EAd IgA), the models including “testing date” as a random effect did not converge, and, instead, we included “test day” (coded as 1 or 2) as a fixed effect.

Percent agreement per assay was calculated based on dichotomous test results by taking the mean of the mean agreement across all test results per individualAnalyses were stratified by study groups. For this exploratory evaluation, the percent seropositivity (and GMT among seropositives) for each assay by study group were additionally estimated.

Results

For both days, a total of 580 serum samples from 97 individuals were tested with a panel of 10 anti-EBV ELISA assays (Table 2). The percent positivity among the samples for all of the markers ranged from 17.8% to 71.0%. The GMT for the markers ranged from 65.3 to 140.0, and the GMT among seropositives ranged from 154.3 to 287.5. The percent positivity and GMT among positives were comparable between the 2 test days for each marker (data not shown).

Table 2.

Percent Detectability and GMT by Anti-EBV ELISA Serology Test - EBV Serology Reproducibility Study

Assay	# Individuals/Specimens Tested	% Positivea	Overall GMT	GMT among Positivesa
EBNA IgA	97/580	48.6%	127.2	287.5
VCAp18 IgA	97/580	58.6%	135.0	225.9
EA ext IgA	97/580	31.2%	84.0	169.9
EAp138 IgA	97/580	18.1%	67.7	154.6
EAd IgA	97/580	17.8%	69.8	195.1
Zebra IgA	97/580	17.4%	65.3	154.3
EA ext IgG	97/580	71.0%	140.0	176.6
EAp138 IgG	97/580	41.6%	94.6	173.1
EAd IgG	97/580	33.6%	82.2	207.0
Zebra IgG	97/580	67.6%	139.3	191.2

^aStatistics are based on specimen not individual-level results

Table 3 presents the CVs (within batch, across batch and overall) and ICCs by anti-EBV ELISA serology test. Percent agreement based on positive/negative test results was high for all 10 assays (81.9% to 90.6%). The within-batch CVs ranged from 7.0% (EA ext IgG) to 13.0% (EA ext IgA), while the across-batch CVs ranged from 2.0% (EAp138 IgA) to 10.9% (EA ext IgA) and the ICCs ranged from 42.3% (EA ext IgA) to 87.7% (VCAp18 IgA). For all assays, with the exception of the EA ext assays, CVs were comparable between experiments conducted on separate dates (p-values for date of testing > 0.05). For all but 1 marker (EA ext IgA), the overall CVs were lower than 15%, and for all but three assays (EA ext IgA, EA ext IgG, and Zebra IgA) the ICCs were greater than 70%.

Table 3.

Assay Performance - Coefficients and Variantion and Interclass Correlation Coefficients by Anti-EBV ELISA Serology Test - EBV Serology Reproducibility Study^a

Assay	# Individuals Tested	% Agreementc	Within Batch	Across Batch	Overall
			CV	CV	CV	ICC
EBNA IgA	97	89.7%	11.0	6.8	12.9	78.1
VCAp18 IgA	97	88.7%	7.3	5.8	9.4	87.7
EA ext IgAb	97	87.6%	13.0	10.9	17.0	42.3
EAp138 IgA	97	87.6%	10.0	2.0	10.2	73.3
EAd IgAb	97	90.7%	10.6	5.1	11.8	73.9
Zebra IgA	97	89.7%	9.1	8.6	12.5	67.1
EA ext IgGb	97	82.1%	7.0	8.2	10.8	65.3
EAp138 IgG	97	81.1%	10.5	5.3	11.8	74.6
EAd IgG	97	90.0%	12.6	6.6	14.2	71.3
Zebra IgG	97	85.2%	8.5	4.5	9.6	82.0

^aRandom effect in the mixed models included testing date, box, and repeated aliquots of the same person; study group (defined by NPC and EBV antibody status) was included as a fixed effect.

^bModels including “testing date” as a random effect did not converge; “test day” (coded as 1 or 2) was included as a fixed effect instead.

^cAgreement based on dichotomous results of each assay as described in the Methods

When percent detection and GMTs were evaluated by study group (Table 4), detection rates and GMTs were typically observed to be highest among nasopharyngeal carcinoma cases and lower among healthy individuals with decreasing evidence of EBV seroreactivity (as defined in the methods). High sensitivity for the detection of nasopharyngeal carcinomas was observed for the EBNA IgA (99.2%), VCAp18 IgA (85.0%), EA ext IgG (100%) and Zebra IgG (85.5%) assays. High specificity (defined as 1 minus percent positive among control group 3) was observed for the EBNA IgA (88.3%), VCAp18 IgA (80.0%), EA ext IgA (87.5%), EAp138 IgA (97.5%), EAd IgA (95.0%) and Zebra IgA (84.2%) assays.

Table 4.

Ability of Assays to Distinguish NPC Cases from Non-Cases by Anti-EBV ELISA Serology Test - EBV Serology Reproducibility Study

Assay	NPC Cases (N=20)	% Positivity (GMT among Positives)			P-value
		Control Group 1 (2+)a (N=32)	Control Group 2 (1)a (N=25)	Control Group 3 (0)a (N=20)
EBNA IgA	99.2% (420.5)	52.6% (291.5)	32.7% (140.3)	11.7% (126.7)	<.0001b
VCAp18 IgA	85.0% (228.9)	72.1% (260.8)	51.3% (207.6)	20.0% (123.1)	<.0001b
EA ext IgA	77.5% (191.0)	24.7% (161.0)	17.3% (143.4)	12.5% (130.8)	<.0001c
EAp138 IgA	43.3% (178.6)	16.3% (149.8)	12.7% (115.7)	2.5% (110.3)	<.0001b
EAd IgA	50.0% (256.9)	12.6% (138.1)	8.7% (123.8)	5.0% (132.3)	<.0001c
Zebra IgA	22.5% (220.9)	18.4% (128.9)	13.3% (143.4)	15.8% (139.3)	0.0026b
EA ext IgG	100% (264.2)	73.2% (161.8)	64.7% (143.9)	46.7% (131.9)	<.0001c
EAp138 IgG	66.7% (236.9)	36.3% (155.0)	36.0% (142.4)	31.7% (144.5)	<.0001b
EAd IgG	76.7% (268.1)	21.1% (193.0)	25.3% (150.2)	20.8% (145.4)	<.0001b
Zebra IgG	85.5% (216.8)	63.7% (190.2)	66.7% (174.5)	56.7% (182.2)	0.0004b

^aControls grouped based on the number of positive results out of three anti-EBV antibody tests as described in the Methods

^bP-value for study group as fixed effect in mixed models also including as random effects, testing date, box, and repeated aliquots of the same person

^cModels including “testing date” as a random effect did not converge; “test day” (coded as 1 or 2) was included as a fixed effect instead.

Discussion

EBV serology has been shown to be a predictor of NPC risk for both sporadic and familial NPC [Zeng et al., 1985; Chien et al., 2001; Yu et al., 2011]. General population screening is unlikely to be practical, but there is a need for markers of risk to guide clinical management within high risk families. However, before any EBV markers can be used clinically, it is important to a) develop and characterize assays with potential to be developed further as “highly standardized” commercial tests and b) to show that these markers perform well with high sensitivity/specificity and predictive value within populations of interest. In this paper, initial steps have been taken in this direction to characterize carefully an ELISA-based set of anti-EBV antibody assays that have the potential for commercialization and standardization.

Most of the 10 assays evaluated had good reproducibility. However, IgG assays, though reproducible, are unlikely to be very useful for screening purposes since they are a good marker of exposure to the highly prevalent EBV but do not distinguish well between nasopharyngeal carcinoma cases and non-cases. In contrast, EBV IgA assays appear to provide promising serological tools for first- line nasopharyngeal carcinoma risk assessment. In particular, the EBNA1 IgA and VCA IgA assays, which use synthetic peptides, were reproducible and higher in nasopharyngeal carcinoma cases than in controls. Interestingly, the same two antigens based on different assays together had high sensitivity and specificity in a recent study in China [Liu et al., 2012]. Finally, the assays based on purified cell extracts, such as EA ext IgA, which are produced in a less standardized fashion than for synthetic peptides, had higher CVs and lower ICCs.

Limitations of this study include the small sample size and the exploratory nature of the specificity and sensitivity analysis. Strengths of this study are the ability to select samples taken from individuals with 4 distinct anti-EBV profiles; that samples were collected, processed and stored based on a standardized protocol; and that 10 anti-EBV markers were comprehensively evaluated.

In conclusion, most of the anti-EBV markers are broadly reproducible and can be used for large-scale screening studies that may eventually lead to the clinical management of individuals at high-risk of nasopharyngeal carcinomas. Future studies should evaluate the performance of these assays (and combination of assays), particularly IgA assays, for predicting risk of nasopharyngeal carcinomas within high-risk individuals.