Dried-Blood Sampling for Epstein-Barr Virus Immunoglobulin G (IgG) and IgA Serology in Nasopharyngeal Carcinoma Screening

Abstract

Dried-blood (DB) samples on filter paper are considered clinical specimens for diagnostic use because of the ease of collection, storage, and transport. We recently developed a synthetic-peptide-based immunoglobulin A (IgA) (EBNA1 plus viral capsid antigen [VCA]-p18) enzyme-linked immunosorbent assay (ELISA) for nasopharyngeal carcinoma (NPC) screening. Here, we evaluate the use of two filter papers for DB sampling, i.e., Schleicher & Schuell (S&S) no. 903 and Whatman no. 3; the DB samples were either taken directly from a finger prick or spotted from a Vacutainer blood collector. The elution of DB samples on filter paper was optimized and tested for IgG and IgA reactivity by ELISA (EBNA1 plus VCA-p18) and compared to simultaneously collected plasma samples. The results showed that both types of filter paper can be used for sample collection in NPC diagnosis by using either finger prick or blood spot sampling. Both DB sampling methods produced comparable ELISA (EBNA1 plus VCA-p18) results for IgG and IgA reactivity in 1:100-diluted plasma samples. DB samples of whole blood or finger prick blood show correlation coefficients (r²) of 0.825 to 0.954 for IgA on S&S no. 903 filter paper, 0.9133 to 0.946 for IgA on Whatman no. 3 filter paper, 0.807 to 0.886 for IgG on S&S no. 903 filter paper, and 0.819 to 0.934 for IgG on Whatman no. 3 filter paper. Using plasma IgA as a reference, DB sampling showed sensitivities and specificities of 75.0 to 96.0% and 93.5 to 100%, respectively. DB samples could be stored at 37°C for 1 to 4 weeks on S&S no. 903 filter paper and 1 to 6 weeks on Whatman no. 3 filter paper without a significant loss of reactivity, with provision of transport options for tropical conditions. IgA proved to be more stable than IgG. Whatman no. 3 filter paper is a more economical yet diagnostically comparable alternative to S&S no. 903 filter paper. Finger prick DB sampling is proposed for NPC diagnosis, particularly for remote hospitals and field screening studies.

Nasopharyngeal carcinoma (NPC) is a disease with remarkable geographic and racial distribution worldwide. NPC is a rare disease in many parts of the world, including in Europe and North America, with an incidence below 1 per 100,000 persons. High-incidence regions are located mainly in southern China (25 per 100,000 persons per year in the Guangzhou area), Taiwan, and some Southeast Asian countries (48). There are also areas of intermediate incidence (3 to 8 per 100,000 persons per year), e.g., North Africa, Alaska, Greenland, and highly populated Asian countries, such as Vietnam and Indonesia (5, 48). In Indonesia, especially in central Java, undifferentiated NPC (WHO type III) ranks among the most common types of cancer. In the Yogyakarta province, hospital-based data showed NPC to be ranked the number 1 cancer in males and the number 3 cancer in females (41), with regional villages representing hot spots with high NPC incidence (unpublished data).

NPC WHO type III is virtually 100% associated with the Epstein-Barr virus (EBV). EBV infection in NPC tumor cells displays a type II latency pattern by the expression of EBV EBNA1, LMP1, LMP2, and noncoding EBV-encoded RNA and BamHI A rightward transcript RNA (4, 34), with the additional expression of the BARF1 oncogene (4, 42). EBV was first linked with NPC on the basis of the serological observations made by Old and colleagues in 1966 (36) and further elaborated by Henle and Henle (20). NPC is characterized by aberrant immunoglobulin G (IgG) and particularly IgA responses directed against various latent and lytic EBV antigens (13). These aberrant responses have diagnostic relevance in screening for early-stage and posttreatment monitoring (7, 22, 23, 43). The diagnosis and screening of NPC are done mostly by indirect antibody detection using cell spot slide tests, one of the earliest serology methods developed, which to date is still used as a “gold standard” (8, 35). However, these slide-based assays are subjective and cumbersome, making their application in mass screening inconvenient (11, 23).

Enzyme-linked immunosorbent assay (ELISA) techniques provide a promising alternative with potential for automation and mass screening (11, 14). Recently, we developed a well-standardized IgA EBV ELISA for the primary diagnosis of NPC using a combination of multiepitope EBNA1- and viral capsid antigen (VCA)-p18-derived synthetic peptides in a single-well format and combined it with the detection of IgG reactivity to the EBV immunoblot strips for the confirmatory test (herein referred to as the IgG and IgA EBV ELISA). For field studies, a simple sample collection and transport system is desirable.

The use of filter paper for blood collection and analysis was implemented as early as the 1960s by Guthrie et al. using dried-blood (DB) samples for newborn phenylketonuria detection (15). DB sampling on “Guthrie card” (or Schleicher & Schuell [S&S] no. 903) filter paper is widely used in many types of tests, including chemical, serological, and genetic applications (26). S&S no. 903 filter paper is made from high-purity cotton linters manufactured to give accurate and reproducible absorption of blood specimens according to CLSI (formerly NCCLS) specifications (32).

The simple puncturing of the skin for the collection of blood onto paper has become a significant tool for screening individuals for clinical purposes and for carrying out epidemiological studies. DB sampling ensures easy sample handling, transport, and storage, especially for samples collected at remote sites where the laboratory equipment, personnel, or infrastructure necessary for the correct handling of blood samples may not be available.

Studies with DB specimens include the surveillance for human immunodeficiency virus (HIV) infection among childbearing women (16); the serodiagnosis of hepatitis C (9), measles (12, 27), rubella, and toxoplasma (33); the serotyping of herpes simplex virus type 1/2 infection (21); and the screening for EBV IgG status (25). In general, any analyte can be measured from whole blood or plasma as DB on filter paper (26). Recently, a special card (FTA card; Whatman technology) was developed for DNA/RNA analysis. This type of card has the ability to maintain the DNA/RNA integrity of the samples and is used for HIV epidemiology studies (2, 24), and it may be of relevance for NPC diagnosis as well (44).

Previously, McDade et al. (25) used a DB sampling method for defining EBV IgG serostatus in psychoneuroimmunology studies. Here, we consider and evaluate DB sampling as a method for NPC screening by comparing two types of sampling, i.e., finger prick (FP) and blood spot (BS), on either S&S no. 903 or Whatman no. 3 filter paper to replace plasma in the IgG and IgA EBV ELISA. We propose the use of DB sampling for NPC screening in “high-risk” regions.

MATERIALS AND METHODS

Blood, plasma, and DB samples.

Blood from healthy donors (n = 98) was taken from volunteers in the Yogyakarta region of Indonesia. NPC samples (n = 42) were taken from first-visit patients enrolled in the ear, nose, and throat clinic at Sardjito Hospital in Yogyakarta as part of a standard serology screening procedure (14). NPC status was confirmed for all samples by computer tomography scanning and pathological biopsy examination. In addition, the EBV-positive status of the tumors was confirmed by immunohistochemistry staining using OT1X antibody directed to EBNA1 (7). For all healthy blood donors, parallel samples were taken from both a fingertip and a vein in the arm, while for NPC patients, samples were taken from only the arm.

Sample collection.

FP samples were taken by pricking the middle-finger tip with a lancet (Baxter, United Kingdom) after it was cleaned with 70% ethanol. The blood was allowed to drip directly onto S&S no. 903 (Schleicher & Schuell, Germany) and Whatman no. 3 (Whatman, United Kingdom) filter papers until a circle with a diameter of about 10 mm formed. BS samples were prepared by drawing 100 μl whole blood from a heparinized Vacutainer vial and by spotting it onto S&S no. 903 and Whatman no. 3 papers. Plasma samples were prepared from the same Vacutainer by whole-blood centrifugation at 1,800 rpm for 15 min and subsequently by plasma isolation. The FP, BS, and plasma samples were stored at −20°C until use. The BS samples were also stored at elevated temperatures where indicated below.

Plasma elution from DB samples.

Using a paper puncher, 25-mm² BS disks were cut. One disk was immersed in sample buffer (1% bovine serum albumin, 0.1% Triton X-100, and 0.05% Tween 20 in phosphate-buffered saline). The elution of IgA was optimized by variation (i) of the volume of the sample buffer, (ii) in the elution solvent, and (iii) in the incubation temperature and time, independently for Whatman no. 3 and S&S no. 903 papers, to achieve an optical density value at 450 nm (OD₄₅₀) comparable with that of the 1:100-diluted plasma samples in our standard EBV ELISA (14).

EBV serology tests.

The standard serology test consisted of our IgG and IgA EBV ELISA for NPC diagnosis/screening (13, 14).

The EBNA1 and VCA-p18 synthetic peptides were made based on the predicted immunodominant epitope defined by Pepscan analysis (30) and prepared as described elsewhere (28, 30, 47). IgG and IgA EBV ELISAs were performed as described previously, and they used EBV-seropositive and -seronegative sera as controls in each run (14). All samples were tested in duplicate. The cutoff value (CoV) was determined to be 0.3536, according to receiver operating characteristic curve analysis, defined as the threshold value optimally separating “healthy” samples from “disease” samples (31). The OD₄₅₀ value of each sample was corrected with that of a negative plasma background reaction as described in detail before (10, 14).

For the confirmation test, EBV immunoblot strips containing nuclear antigens from HH514.c16 cells chemically induced to produce the late lytic phase of EBV proteins were used to detect IgG reactivity to the spectrum of EBV EBNA1 and lytic antigens. The strips were prepared and analyzed exactly as described previously (13, 29). Characteristic EBV antigens on blot strips were defined by known human reference sera and monoclonal/monospecific polyclonal antibodies (13). A sample was determined to have a “normal pattern” when IgG reactivity was detected against any combination of EBNA1 (BKRF1 [72 kDa]), VCA-p40 (BdRF1 [40 kDa]), ZEBRA (BZLF1 [36 plus 38 kDa]; fine doublet), and VCA-p18 (BFRF3 [18 kDa]). A sample was determined to have an “abnormal pattern” when IgG reactivity to an EBV antigen(s) other than those involved in the “normal pattern” was present.

DB sample stability.

To evaluate the stability of stored BS samples on filter paper, we obtained several DB samples from four healthy individuals. Separately, 100 μl of blood from a heparinized Vacutainer was spotted onto either S&S no. 903 or Whatman no. 3 filter paper, dried overnight at room temperature (RT; 18 to 22°C), placed in a paper envelope, and stored at −20°C, 4°C, RT, and 37°C. In addition, RT and 37°C incubations were measured to have similar relative humidities (∼30%). Stored BS samples were processed with the IgG and IgA EBV ELISA using the optimized elution method for each type of paper. Evaluations were done once a week for 4 weeks and then at 2-week intervals for a period of 24 weeks.

Analysis.

The descriptive statistical analysis values (means, medians, standard deviations) and correlation coefficients (r²) were determined by comparing the individual IgG or IgA EBV ELISA results for the DB or plasma samples. The sensitivity and specificity of the DB samples were determined by using plasma IgA EBV ELISA results as the reference for the positive and negative values. After correction with the results of a negative plasma background reaction, OD₄₅₀ values above 0.3536 were stated as “IgA positive” and values below 0.3536 as “IgA negative.” Statistical analysis was done by GraphPad Prism version 4.03.

RESULTS

Plasma IgA and IgG EBV ELISA results.

The samples from healthy donors were collected specifically for this study, while the NPC samples were obtained from patients with histologically confirmed NPC who were enrolled in the NPC treatment program at Sardjito Hospital, Yogyakarta, Indonesia (14). The NPC subjects (n = 42) were 71.1% male and 29.9% female, presenting disease stages III (41.7%) and IV (58.3%), with ages ranging from 18 to 70 years (<30 years, 20.6%; 31 to 40 years, 17.6%; 41 to 50 years, 26.5%; and >50 years, 35.3%). Figure 1 shows IgG and IgA EBV ELISA results of all plasma samples used in this study. The means and standard deviations of the OD₄₅₀ values were 1.879 and 0.643, respectively, for the IgG of the healthy donors and 3.135 and 0.359, respectively, for the IgG of the NPC patients, while they were 0.141 and 0.141, respectively, for the IgA of the healthy donors and 1.498 and 0.887, respectively, for the IgA of the NPC patients. The IgA EBV ELISA, using a CoV of 0.3536 (14), showed 94.9% (93/98) of healthy blood donors to be negative and 97.7% (41/42) of NPC patients to be positive, leading to 5.1% (5/98) of results potentially being false positive and 2.3% (1/42) of results potentially being false negative (Fig. 1). The IgG immunoblot confirmation test (13) showed that one of five false-positive healthy blood donor samples and one false-negative NPC patient sample had an “abnormal pattern” and four of five false-positive healthy blood donor samples had a “normal pattern” compatible with the absence of NPC (data not shown). The results from the IgG and IgA EBV ELISA were almost completely in agreement with the clinicopathological diagnosis, showing 1 healthy donor out of 98 (1.02%) as being a candidate potentially at risk for NPC (no follow-up available).

FIG. 1.

IgG (A) and IgA (B) EBV ELISA results for healthy donor plasma samples (IgG, n = 68; IgA, n = 98) and NPC plasma samples (IgG, n = 19; IgA, n = 42) used within this study. Medians are indicated by the straight lines.

IgA elution from DB samples.

The optimization of IgA elution from DB samples was done in several steps. First, we defined the optimal elution buffer composition, which is described in Materials and Methods. We then defined the optimal elution volume (200, 300, 400, 500, and 600 μl/25-mm² paper disk) and the incubation time in combination with the incubation temperature (4°C and RT) for both S&S no. 903 and Whatman no. 3 filter papers. The optimal condition was defined as one in which OD₄₅₀ values for IgA eluted from BS samples and for IgA eluted from 1:100-diluted plasma samples are comparable in a standard IgA EBV ELISA. The optimum sample buffer volume was 500 μl for S&S no. 903 filter paper and 400 μl for Whatman no. 3 filter paper. Figure 2B shows the yield of IgA measured after elution from filter paper for different time periods, ranging from 0.5 to 24 h. IgA was optimally eluted after 1 h of incubation at RT for S&S no. 903 filter paper, with similar results obtained with additional samples (n = 9). For Whatman no. 3 paper, IgA was optimally eluted after 4 h of incubation in 4°C (Fig. 2B). The eluted IgA was stable for a period of 24 h at 4°C and at RT (Fig. 2A and B). Similar results were obtained for IgG elution (data not shown).

FIG. 2.

Dynamic of IgA elution during a 24-h incubation of BS samples on S&S no. 903 (A) and Whatman no. 3 (B) filter papers. The optimal IgA elution for S&S no. 903 filter paper (500 μl sample buffer/25-mm² filter disk) was 1 h of incubation at RT, while for Whatman no. 3 filter paper (400 μl sample buffer/25-mm² filter disk), it was 4 h of incubation at 4°C. Overall, IgA was stable for 24 h with RT and 4°C incubation.

Comparison of plasma, BS, and FP IgG and IgA EBV ELISA results.

After the optimization of IgA elution, the panel of DB and plasma samples was tested by the IgA EBV ELISA according to the standard protocol (14). The DB samples used in this study had already been stored sealed for up to 12 months at −20°C prior to use. The OD₄₅₀ values for the DB (BS and FP) samples were compared to those for the 1:100-diluted plasma samples. In parallel, we analyzed eluted-IgG EBV ELISA results from both BS and FP samples on both S&S no. 903 and Whatman no. 3 filter papers. To compare the IgG and IgA EBV ELISA results of both DB and plasma samples, we analyzed the correlation coefficients (r²) of the OD₄₅₀ values for IgA and IgG for the BS versus plasma, FP versus plasma, and BS versus FP samples for both filter papers. In general, DB samples from S&S no. 903 and Whatman no. 3 filter papers produced OD₄₅₀ values in IgA and IgG EBV ELISA highly comparable to those of the 1:100-diluted plasma samples. Individual r² values of the tests are presented in Table 1.

TABLE 1.

Correlation coefficients between OD₄₅₀ values of plasma and DB samples tested by our IgG and IgA EBV ELISA

ELISA (total no. of healthy donors and NPC patient plasma samples)	Filter paper	r2 between OD450s of indicated samples (no. of samples tested)
		Plasma vs BS	Plasma vs FP	BS vs FP
IgA (150)	S&S no. 903	0.954 (82)	0.836 (63)	0.825 (63)
	Whatman no. 3	0.946 (105)	0.913 (78)	0.929 (78)
IgG (87)	S&S no. 903	0.886 (55)	0.807 (35)	0.886 (35)
	Whatman no. 3	0.819 (37)	0.865 (31)	0.934 (31)

DB sampling sensitivity and specificity.

DB sampling sensitivity and specificity for BS and FP samples on both types of filter paper were defined by the IgA EBV ELISA only, the preferred method for NPC screening (14). Plasma IgA levels from both NPC and healthy panels were used as references for determining positive or negative status, as well as to define false-positive and false-negative results for each filter paper analyzed. Sensitivity and specificity values for each type of filter paper are shown in Table 2. For S&S no. 903 filter paper, the sensitivity and specificity of the BS samples were 96.0 and 93.6%, and those of the FP samples were 80.0 and 100%, respectively. For Whatman no. 3 filter paper, these values were 89.2 and 97.3 for the BS samples and 75.0 and 97.1% for the FP samples, respectively.

TABLE 2.

Sensitivity and specificity of DB samples compared to those of plasma samples in an IgA EBV ELISA^a

Filter paper	DB sample typea	No. of samples with the indicated ELISA result				Sensitivity (%)d	Specificity (%)e
		IgA positiveb	IgA negativec	False positive	False negative
S&S no. 903	BS	24	58	4	1	96.0	93.5
	FP	4	59	0	1	80.0	100.0
Whatman no. 3	BS	33	72	2	4	89.2	97.3
	FP	12	66	2	4	75.0	97.1

^aFor S&S no. 903 filter paper, 82 BS and 63 FP samples were tested. For Whatman no. 3 filter paper, 105 BS and 78 FP samples were tested.

^bIgA-positive samples were defined as samples with OD₄₅₀ values above 0.3536 (13).

^cIgA-negative samples were defined as samples with OD₄₅₀ values below 0.3536.

^dSensitivity was defined as the number of IgA-positive samples divided by the sum of the number of IgA-positive and false-negative samples.

^eSpecificity was defined as the number of IgA-negative samples divided by the sum of the number of IgA-negative and false-positive samples.

DB sampling stability.

Parallel BS samples (n = 4) were stored in envelopes at −20°C, 4°C, RT, and 37°C for 24 weeks. At different time points, one sample was eluted and tested by the IgA and IgG EBV ELISA and compared to a 1:100-diluted plasma sample. A normalized value was obtained by defining the ratio of OD₄₅₀s for BS and 1:100-diluted plasma samples in the same test. Figure 3 shows one of the four samples tested by our IgG and IgA EBV ELISA on S&S no. 903 and Whatman no. 3 filter paper. When stored at −20°C and 4°C, both types of paper showed relatively stable IgA and IgG values until the 24th week. IgA and IgG reactivities of the samples stored on either type of filter paper at RT showed less stability than those of samples stored at 37°C, with decreasing stability observed at week 2 for IgA and at week 0 to 1 for IgG. IgA was more stable than IgG, with decreasing reactivity from 4 (S&S no. 903 paper) and 6 (Whatman no. 3 paper) weeks onwards.

FIG. 3.

IgA EBV ELISA results showing the stability of IgA in DB samples on S&S no. 903 (A) and Whatman no. 3 (B) filter papers, and IgG EBV ELISA results showing the stability of IgG on S&S no. 903 (C) and Whatman no. 3 (D) filter papers over a period of 24 weeks in storage at −20°C, 4°C, RT, and 37°C. Results showed that IgG and IgA in both filter papers remained stable until the 24th week when DB samples were stored at −20°C and 4°C for 4 to 6 weeks for IgA and at 37°C for 1 week for IgG and for less time when the samples were stored at RT.

DISCUSSION

EBV serology is commonly used to facilitate the diagnosis of suspected NPC and is proposed for large field screening and epidemiology survey studies (35, 49). NPC risks are associated with the elevated responses of IgG and particularly of IgA antibodies to certain EBV antigens. Most people in Southeast Asia are first infected by EBV in early childhood, reflected by a nearly 100% seropositivity for IgG to EBV VCA and EBNA1. Figure 1 shows IgG EBV reactivity in healthy blood donors overlapping with IgG levels in NPC patients, thus precluding diagnostic use. On the other hand, IgA reactivities differed more significantly between healthy donors and NPC patients, in agreement with prior studies (14, 20, 23). The presence of IgA to EBV VCA suggested the reactivation of EBV in epithelia, paralleling NPC development. A 15-year follow-up study in China recently revealed that elevated IgA responses to EBV VCA become apparent within a 2-year “window” period before clinical manifestation (22). Another study showed elevated IgA responses to EBV VCA at 16 to 41 months prior to the clinical manifestation of NPC (50). In those previous studies, an EBV slide test was used, which is suboptimal for screening purposes. ELISA is considered a more suitable tool for serological screening because of its relatively low cost, standardization of reagents, and suitability for automation, allowing the processing of large numbers of samples under identical conditions (11, 14). For the serodiagnosis of NPC in a high-risk population in Indonesia, we recently developed a one-step IgA EBV ELISA by combining immunodominant epitope peptides from EBNA1 and VCA-p18 (14). In agreement with others, we demonstrated that IgA to EBNA1 and VCA-p18 is a highly reliable marker for NPC screening (8, 40). The use of synthetic peptides greatly improves the standardization of EBV ELISA. Our IgA EBV ELISA showed that in a panel of freshly collected blood samples, 5 of 98 (5.1%) healthy blood donors were above the CoV and 1 of 42 (2.4%) NPC patients was below the CoV. These aberrant samples were subsequently tested by using immunoblotting as the confirmation test assessing the EBV IgG diversity according to Fachiroh et al. (13). Confirmation testing revealed that one of five healthy blood donors with elevated IgA values had an “abnormal pattern” but that the other four samples had a “normal pattern.” The one NPC patient with a low IgA EBV value presented an “abnormal pattern,” suggestive of NPC. This sample was later confirmed by positive EBV-encoded RNA staining of tumor tissue. In brief, our screening methodology combining IgA EBV ELISA and a confirmation test, IgG EBV immunoblotting, confirmed the diagnoses of 97 of 98 (98.98%) healthy donors and all (100%) of the NPC samples tested within this study.

In some parts of Java Island in Indonesia, as well as in other parts of the vast Indonesian archipelago, there are “hot spots” of NPC, most of them in rural areas with geographic barriers, and they are localized at a distance from the central diagnostic laboratory. In current practice, most NPC patients first present to the clinic with late-stage (III/IV) disease involving a large primary tumor mass and lymph node metastasis, as also seen in our NPC panel (see Results). When diagnosed and treated early, NPC can be effectively treated with radiotherapy, leading to highly improved cure rates. Therefore, it is desirable to have a simple sample collection system in place for screening and diagnosis in remote populations at risk of NPC. It is considered relevant to combine a reliable screening assay with a simple sampling method. The use of filter paper sampling is proposed to be combined with FP bleeding to replace blood drawing from the arm. This combines simple sampling with the ease of transportation for subsequent testing in a reference laboratory. S&S no. 903 filter paper (known as “Guthrie paper”) is a standard sampling paper and is widely used for many types of analyses, including those for DNA (9), RNA (1), protein (21, 25, 37), and chemical substances (19, 38). Whatman no. 3 filter paper is a thick membrane with tight pores used as a fine-particle filtration device (Whatman product information), while S&S no. 903 filter paper is a special liquid specimen collector made of cotton fiber which has the capacity to absorb/release liquids efficiently. Whatman no. 3 filter paper required longer time than S&S no. 903 filter paper to absorb/elute blood (Fig. 2). When Whatman no. 3 paper is used, attention is required during the collection of FP samples to ensure that spots are completely saturated with blood. Whatman no. 3 filter paper is a good alternative to S&S no. 903 filter paper, with its low cost, local availability, and comparable sensitivity.

The elution of Ig from the DB samples was highly reproducible by using the same buffer used to dilute plasma in our standard ELISA protocol. The eluted Ig solution from the DB samples could be stored at −20°C for a few days without reducing its reactivity (data not shown). Thus, it is possible to prepare samples several days in advance of testing. All Ig samples eluted from the DB samples—either FP or BS samples—could be detected by ELISA. Table 1 shows an excellent correlation between OD₄₅₀ values from the two steps of the IgG and IgA EBV ELISA for either the BS or FP samples on both paper types. By using plasma IgA as the reference, the sensitivity and specificity for the BS and FP samples from both types of filter paper were between 75.0 and 100%, respectively, as shown in Table 2. False-positive and/or -negative samples were found among those with values close to the cutoff point, indicating the necessity for precise elution volume and time. Table 2 also shows that the FP samples have lower sensitivities than the BS samples, and yet they have comparable specificities. This may reflect the occasionally limited volume of blood collected on the filter paper spot, indicating that FP sampling needs to be performed with care.

The antibody contained in the DB samples may decay in a humid atmosphere (17), but when stored properly, DB sampling will ensure Ig stability for a long period. We studied the stability of IgG and IgA in BS samples stored in paper envelopes, without desiccant, under different temperatures for 24 weeks. Results showed that IgG and IgA were relatively stable when stored at 4°C and −20°C until the 24th week, in agreement with other serological studies for EBV(25), HIV (18), and measles virus (39). This demonstrates that DB samples will retain their biological contents when stored at a low temperature.

When stored at RT and 37°C, Whatman no. 3 filter paper provided somewhat better IgG and IgA stability than S&S no. 903 filter paper. The DB samples stored at 37°C maintained IgA for 4 to 6 weeks and IgG for 1 week (Fig. 3). This allows sufficient time for the transport of a sample in a tropical atmosphere from a regional hospital or rural area to the central laboratory by standard mail. Storage at RT provided lower IgA and IgG stability than storage at 37°C. This is dissimilar with other findings (3, 25, 45), which indicated that DB samples stored at RT had better antibody (IgG and IgE) stability than those stored at higher temperatures. McDade et al. (25) showed that DB samples were stable in IgG VCA-p18 ELISAs for at least 8 weeks at 4°C and RT but deteriorated after 1 week at 37°C. Our data showed similar results for IgG DB samples stored in 37°C but longer IgA stability for DB samples collected on both paper types. The specific mucosal origin of IgA may provide enhanced stability under “high-stress” conditions like bacterial contamination or proteolytic degradation.

To obtain longer antibody stability on DB specimens, it was suggested that the humidity needs to be controlled (3). Mei et al. (26) emphasize the importance in avoiding humidity by storing DB samples in ziplock bags with desiccant since moisture may harm the specimens by inducing bacterial growth or altering the elution time of the specimens. Dried BS specimens stored in ziplock bags with desiccant can be stored at −20°C for many weeks or years (3, 6, 46).

In summary, our data indicate that DB samples obtained from FP sampling or from a Vacutainer tube may be transported at ambient temperatures by regular mail without the loss of sensitivity and specificity. IgA DB samples are stable for months when stored in cold temperatures. Both S&S no. 903 and Whatman no. 3 filter papers can be used to replace fresh plasma sample type in the NPC field screening program, with Whatman no. 3 providing a more economical alternative.

DB samples for IgA EBV serology can be prepared by applying a few drops of fresh blood drawn by venipuncture or by FP sampling using a lancet. FP sampling allows access to individuals for whom drawing blood by venipuncture may be problematic, such as children and elderly people. FP DB sampling may be a more practical sampling method, as it is inexpensive and does not require trained personnel. DB samples can be sent at ambient temperatures to a research laboratory or a central hospital for testing. For long periods of storage, DB samples require less space than plasma samples. With these advantages, FP DB samples may replace fresh blood as samples for NPC serological studies.

Therefore, FP DB sampling is proposed as a tool for EBV/NPC screening in combination with IgA EBV ELISA, providing a standardized and economical method. The DB sampling method enables population-based screening in remote areas, which is important in finding early-onset NPC cases.