Nucleic Acid-Based Cross-Linking Assay for Detection and Quantification of Hepatitis B Virus DNA

Abstract

A nucleic acid photo-cross-linking technology was used to develop a direct assay for the quantification of hepatitis B virus (HBV) DNA levels in serum. Cross-linker-modified DNA probes complementary to the viral genomes of the major HBV subtypes were synthesized and used in an assay that could be completed in less than 6 h. The quantification range of the assay, as determined by testing serial dilutions of Eurohep HBV reference standards and cloned HBV DNA, was 5 × 10⁵ to 3 × 10⁹ molecules of HBV DNA/ml of serum. Within-run and between-run coefficients of variation (CVs) for the assay were 4.3 and 4.0%, respectively. The assay was used to determine HBV DNA levels in 302 serum samples, and the results were compared to those obtained after testing the same samples with the Chiron branched-DNA (bDNA) assay for HBV DNA. Of the samples tested, 218 were positive for HBV DNA by both assays and 72 gave results below the cutoff for both assays. Of the remaining 12 samples, 10 were positive for HBV DNA by the cross-linking assay only; the 2 other samples were positive by the bDNA assay only. Twenty-eight samples had to be retested by the bDNA assay (CV, >20% between the results obtained from the testing of each sample in duplicate), whereas only three samples required retesting by the cross-linking assay. The correlation between the HBV DNA levels, as measured by the two tests, was very high (r = 0.902; P = 0.01). We conclude that the cross-linking assay is a sensitive and reproducible method for the detection and quantification of HBV DNA levels in serum.

Hepatitis B virus (HBV) is one of the causative agents of viral hepatitis. An estimated 350 million persons worldwide are chronic carriers of the virus, with 100 million carriers in China and approximately 1 million in the United States (1, 2). Although the majority of individuals infected with HBV resolve the primary infection and develop lasting immunity, clinical studies have shown that 5 to 10% of individuals are chronically infected with the virus and that about 25 to 40% of these individuals may deteriorate and the infection may progress to cirrhosis or liver cancer (6).

HBV is a partially double-stranded DNA virus of the class Hepadnaviridae (6). The virus is composed of a 42-nm outer shell and a 27-nm inner shell. A major component of the outer shell is the hepatitis B surface antigen (HBsAg). The inner shell is composed of the hepatitis B core antigen (HBcAg). The detection of serological markers such as HBsAg and a breakdown product of HBcAg, the hepatitis B e antigen (HBeAg), is useful for diagnosis (6). The presence of HBsAg in serum indicates HBV infection but does not provide information on the replicative state of the virus. Although HBeAg is thought to be a good marker for active viral replication, mutants that do not produce HBeAg have been found (4). The presence of HBV DNA in the serum of chronic carriers is a better indicator of viral replication, and detection and quantification of the viral DNA have been used to study the natural progression of the disease and to monitor the responses of patients receiving antiviral therapy (7).

Several molecular approaches have been used to quantify serum HBV DNA levels, including commercially available assays such as those from Abbott (Genostics HBV-DNA Assay; Abbott Laboratories, Chicago, Ill.), Digene (Hybrid-Capture HBV-DNA Assay; Digene Diagnostics Inc., Silver Spring, Md.), and Chiron (Quantiplex HBV-DNA Assay; Chiron, Emeryville, Calif.) (3, 8, 9, 12). Current commercial HBV DNA detection assays, however, have various levels of sensitivity, and interassay comparisons have shown poor agreement due to a lack of standardization (3, 8, 9, 12).

NAXCOR (Menlo Park, Calif.) has developed diagnostic tests for the detection of DNA and RNA targets based on the use of photo-cross-linkable DNA probes (11, 13). Recently, this technology platform has been expanded to the detection and quantification of HBV DNA in serum. This assay is a direct DNA probe test that uses nucleic acid hybridization and cross-linking technology. The aim of this study was to evaluate and compare the analytical and clinical performance of this test with that of the Chiron assay.

MATERIALS AND METHODS

Source of clinical samples.

Clinical testing of 302 serum samples for HBV DNA with the cross-linking and branched-DNA (bDNA) assays was conducted at Quest Laboratories, Singapore (200 samples), and at the Queen Mary Hospital, University of Hong Kong, Hong Kong (102 samples). Samples were obtained from patients with chronic HBV infection.

Source of HBV reference standards.

Eurohep HBV reference plasma standards 1 (genotype A, HBsAg subtype adw) and 2 (genotype D, HBsAg subtype ayw) were obtained from K. Heermann, Division of Medical Microbiology, University of Goettingen, Goettingen, Germany. The levels of HBV DNA in these standards, as determined by the fourth round of the Eurohep trials, was 4.2 × 10⁹ HBV genomes/ml for plasma standard 1 (95% confidence interval, 3.3 × 10⁹ to 5.1 × 10⁹ HBV genomes/ml) and 3.8 × 10⁹ HBV genomes/ml for plasma standard 2 (95% confidence interval, 2.8 × 10⁹ to 4.8 × 10⁹ HBV genomes/ml) (5). Aliquots of the Eurohep standards were serially diluted in HBV-negative serum to achieve viral titers ranging from 10⁵ to 10⁸/ml of serum (0.1 to 100 Meq/ml). These diluted samples were used to determine the sensitivity of the cross-linking assay.

Procedure of the cross-linking assay.

Oligonucleotides complementary to regions within the entire HBV genome were synthesized for the cross-linking assay as described previously (11). Two types of oligomer were synthesized: biotin-modified capture probes and fluorescein-modified reporter probes. Both types of probe were modified with a photoactive coumarin cross-linking agent derived from 7-hydroxy coumarin, 1-O-(4,4′-dimethoxytrityl)-3-O-(7-coumarinyl)-2-O-(2-cyanoethyl-N,N-diisopropyl phosphoramidite) glycerol (13). To avoid problems in quantifying DNA from different HBV subtypes, the sequences for the probes were chosen from conserved regions of the HBV genome.

A detailed protocol for cross-linking-based diagnostic assays has been described previously (13), and only the essential details are given here. Samples were prepared by the addition of 30 μl of lysis reagent (proteinase K, sodium dodecyl sulfate) to 300 μl of serum and incubation of the solution for 30 min at 65°C. Nucleic acids in the sample were subsequently denatured by the addition of 6.3 μl of alkaline denaturation reagent and boiling for 15 min. The samples were cooled for 5 min and then centrifuged at 12,000 × g for 5 min. Each processed sample was aliquoted (125 μl) into two wells of a 96-well polypropylene microtiter plate. In addition to the samples, each assay plate also contained the following controls supplied with the kit: two negative controls (HBV-negative serum) and four different HBV quantification standards (negative serum containing different concentrations of cloned HBV DNA, subtype adw, ranging from 3 to 3,000 Meq/ml). Each standard was added to two wells in the microplate. Next, HBV probes and neutralization reagent were added to each of the sample and control well and the HBV DNA and probes were hybridized for 20 min at 45°C. Next, the samples were irradiated with a UV type A light source for 30 min to cross-link the probe-target hybrids. Following irradiation, streptavidin-coated magnetic beads (Dynabeads M-280; Dynal Inc., Lake Success, N.Y.) were added to each well to capture the cross-linked probe-target hybrids via the biotin residue attached to the capture probes. After 30 min the beads in the microwells were washed twice and then incubated in the presence of an anti-fluorescein antibody-alkaline phosphatase (AP) conjugate. At the end of this step each well was washed four times. Upon completion of the final wash cycle, 100 μl of Attophos (JBL Scientific, San Luis Obispo, Calif.) was added to each well and the plate was incubated at 37°C for 60 min. Finally, the fluorescent product produced from the reaction of Attophos with AP was detected by measuring the fluorescent signal with a microplate reader. The concentration of HBV DNA in the samples was calculated by comparing the mean fluorescent signal produced from each sample to the signals on a standard curve that had been constructed from the results obtained from the four positive standards. The HBV DNA quantification range of the assay was 0.5 to 3,000 Meq/ml. Samples were retested if the coefficient of variation (CV) between the results obtained from duplicate tests with each sample was greater than 20%.

Procedure of the bDNA assay.

The Chiron bDNA assay was performed as described in the manufacturer’s product insert. Briefly, this assay involved sample preparation to release HBV DNA from viral particles in serum (each sample is tested in duplicate) and overnight hybridization of HBV DNA to capture probes that were bound to the wall of the microtiter plate test well and to solution-phase target probes that bound to different sequences of the HBV genome. Next, the wells were washed and bDNA amplifier probes that bound to the hybridized target probes were added. In the final steps of the assay, AP-conjugated probes that were complementary to multiple sites on the amplifier probe were added to the sample wells. The captured HBV DNA was detected and quantified by measuring the level of chemiluminescence produced after reaction of the bound AP with a dioxetane substrate and comparison of the signal with the signals on a standard curve obtained from an assay with a set of HBV standards performed in parallel with assays with the samples. The HBV DNA quantification range of the bDNA assay is 0.7 to 5,700 Meq/ml. Following the manufacturer’s recommendations, any sample that yielded a test result with a CV of greater than 20% between the two replicates was retested.

Precision of the cross-linking and bDNA assays.

The within-run and between-run precisions of the cross-linking and bDNA assays were determined by testing each assay with HBV-positive serum obtained from two separate individuals. The within-run precision of each assay was determined after performing the cross-linking or bDNA assay with six identical aliquots of each sample at the same time. The between-run precisions of the two assays were measured after the test samples were divided into six identical aliquots and each aliquot was tested in six independent cross-linking or bDNA assay runs.

Statistical analysis.

The t test was used to compare the differences between the means of the results obtained by the two different assays. Regression analysis was used to define the relationships between the results of different assays with the same samples. P values of less than 0.05 were used to indicate statistical significance.

RESULTS

Sensitivity of the cross-linking assay.

The results of testing of serial dilutions of the Eurohep standards in the cross-linking assay are shown in Fig. 1. The assay detected both HBV subtypes (adw and ayw) with equal sensitivities. In these analytical experiments, samples containing as little as 0.1 Meq of HBV DNA per ml could be distinguished from HBV-negative samples. However, before establishing the clinical sensitivity limit of the assay, 50 individual samples from uninfected patients were tested to determine the statistical variation observed between unrelated, negative samples. The mean signal (in relative fluorescence units [RFU]) ± the standard deviation between the signals for these samples was 60 ± 3.75 RFU. We decided that a clinical sample would be determined to be positive in the assay if the signal generated from the sample was greater than or equal to 4 standard deviations of the mean signal obtained from negative serum. Thus, from the data obtained with the 50 negative samples, a sample would be considered positive if the signal obtained from the sample was greater than or equal to 60 + 15 = 75 RFU or, alternatively, if the signal was ≥25% above the signal from the negative control in the assay (75/60 = 1.25). From the experiments with the Eurohep samples described above, this cutoff corresponded to a sample that contained 0.5 Meq of HBV DNA per ml. Subsequently, the Eurohep samples were used to derive a set of HBV assay standards that were prepared by serially diluting different amounts of cloned HBV DNA (subtype adw) in serum. By comparing the signal obtained from these standards with the signal obtained from Eurohep samples with known HBV levels (from 0.5 to 100 Meq/ml), it was possible to assign each standard an HBV DNA concentration on the basis of the signal generated by actual viral DNA and not cloned plasmid DNA. Next, the upper quantitation limit of the assay was determined by testing increasing amounts of the cloned HBV DNA to ascertain at which point the relationship between the signal and the HBV DNA concentration was no longer linear. We found that samples containing HBV DNA concentrations as high as 3,000 Meq/ml still yielded a linear response (data not shown). Consequently, this level of HBV DNA was chosen as the upper quantitation limit of the assay.

FIG. 1.

Sensitivity of the cross-linking assay with different HBV subtypes. Eurohep reference samples 1 (subtype adw) and 2 (subtype ayw) were diluted in HBV-negative sera to give viral titers ranging from 0.1 to 100 Meq/ml of serum. The fluorescent signal obtained from the testing of each diluted sample was divided by the signal generated from control (HBV-negative) serum to give the signal/negative control ratio, as shown. Each point is the mean ratio obtained from tests with each sample performed in duplicate. The CVs between the duplicate samples at each point were <5%.

Precision of the cross-linking and bDNA assays.

Studies were performed to determine and compare the within-run and between-run reproducibilities of the cross-linking and bDNA assays. The results obtained showed that the analytical precisions of both assays were very good; the within-run and between-run CVs of the cross-linking assay were 4.3 and 4.0%, respectively. The precision of the bDNA assay, which was determined to be 5.5% (within-run CV) and 6.3% (between-run CV), was in accordance with previous findings (3).

Performance of the cross-linking and bDNA assays with clinical samples.

The ability of the cross-linking assay to detect and quantify HBV DNA in patient samples was assessed by testing 302 serum samples and comparing the results generated to those obtained by testing the same samples by the bDNA assay. The results of this study are summarized below.

Of the 302 samples tested, 218 contained HBV DNA by both the cross-linking and bDNA assays; 194 of these contained HBV levels within the quantification ranges of both assays. Seventy-two samples contained undetectable HBV DNA levels by both assays. Of the remaining 12 samples, 10 contained measurable levels of HBV DNA by the cross-linking assay (range, 1.266 to 3.993 Meq/ml) but not by the bDNA assay; the 2 other samples contained measurable levels of HBV DNA by the bDNA assay (1.140 and 1.711 Meq/ml) but not by the cross-linking assay. Twenty-four specimens had cross-linking assay results greater than the 3,000-Meq/ml upper quantification cutoff. Nineteen of these samples gave a bDNA assay result greater than 3,000 Meq/ml (the results for nine of these samples were above the 5,700-Meq/ml upper detection limit); the remaining five samples yielded a bDNA result ranging from 2210 to 2589 Meq/ml.

During testing, 28 samples (9.3%) yielded a bDNA assay result that exceeded the recommended 20% CV cutoff (between the sample replicates) for an acceptable result and required retesting. By using the same criteria, only three samples (1.0%) required retesting by the cross-linking assay.

There was no statistical difference between the mean HBV DNA levels obtained in the 194 samples that contained measurable HBV DNA by both assays; the cross-linking assay yielded a mean value of 685.7 Meq/ml, and the bDNA assay yielded a mean value of 718.6 Meq/ml. To show the correlation between the results obtained from the cross-linking and bDNA assays, the HBV DNA levels in the 194 samples that contained measurable HBV DNA in both assays were plotted (Fig. 2). Analysis of the results showed that the HBV DNA levels in these samples, as measured by both assays, was significantly correlated (r = 0.902; P = 0.01).

FIG. 2.

Comparison of the HBV DNA levels in 194 serum samples determined by the cross-linking and bDNA assays. The line passing through the data with a slope equal to 1 is the hypothetical line that all datum points would fall on if the two assays yielded results in complete agreement with one another.

DISCUSSION

A nucleic acid-based diagnostic technology that uses cross-linking offers significant advantages over conventional hybridization-based assays. By covalently joining the hybridized probe and target, the cross-linker guarantees that maximal target will be captured and retained throughout the assay for maximal signal generation. In addition, cross-linking allows the use of stringent wash conditions for effective removal of nonspecific background signal. The coupling of these two significant properties allows the development of sensitive and reproducible diagnostic assays. In this study, the cross-linking technology has been used in an assay for the detection and quantification of HBV DNA.

The results of currently available commercial HBV DNA quantification assays have shown poor agreement after testing of the same samples by each assay. A wide discrepancy was observed when identical samples were quantified by assays from Chiron, Digene, and Abbott; cross testing of the HBV standards supplied in the three kits showed as much as a 120-fold discrepancy in the quantitation values obtained by the assays (3). In addition, Janssen and coworkers (8) found 5- to 10-fold differences in quantitation values between the Abbott HBV DNA assay and the Orion AffiProbe Assay. Taken together, these findings underline the need for internationally recognized HBV reference samples that should be used to standardize different HBV DNA assays.

The cross-linking assay was developed by using the Eurohep HBV reference samples to standardize the HBV quantification levels determined by the assay. In addition to providing a quantification standard, the two Eurohep standards were also used to show that the assay detected the HBsAg subtypes, adw and ayw, with equal sensitivities.

The clinical performance of the cross-linking assay was assessed by testing 302 HBV-positive and -negative serum samples. The results of the assay were compared to those obtained by the Chiron bDNA HBV quantification assay. The bDNA assay was chosen since it appears to be the most sensitive commercial test presently available (3, 9). Overall, the agreement between the two assays was excellent. Comparison of the HBV DNA levels in the 194 samples that yielded signals within the quantitation range of both assays showed that the two sets of results were highly correlated (r = 0.902; P = 0.01).

Only 12 of the 302 samples tested gave a discrepant result between the two assays; 10 of these samples contained measurable levels of HBV DNA by the cross-linking assay but not by the bDNA assay, and the 2 other samples contained measurable levels of HBV DNA by the bDNA assay but not by the cross-linking assay. All 12 samples were determined to contain low HBV DNA levels (1.140 to 3.993 Meq/ml) that were within the quoted quantitation ranges of both assays. Although the exact causes of these discrepant results are unknown, it is possible that variables such as assay-specific sample inhibition, target sequence variation, and nonspecific hybridization could give rise to false-negative and false-positive assay results.

One major difference between the results obtained by the cross-linking and bDNA assays was in the number of clinical samples that needed to be retested. Of the 302 samples tested, 28 (9.3%) gave an initial result by the bDNA assay that exceeded the manufacturers recommended 20% CV cutoff (between the sample replicates) for an acceptable result and required retesting. By using the same criteria, only three samples (1.0%) required retesting by the cross-linking assay. Comparison of the analytical precision of the two assays indicated that although the cross-linking assay yielded slightly lower within-run and between-run CVs (4.3 and 4.0%, respectively) than the bDNA assay (5.5 and 6.3%, respectively), these differences would appear to be too small to account for the large difference in the number of samples requiring retesting.

A possible explanation for the difference in the reproducibilities of the two assays may stem from differences in their respective sample preparation procedures. In the cross-linking assay, each clinical sample is typically prepared in a microcentrifuge tube. The final step in this procedure is a short centrifugation step prior to the addition of the sample to two wells in the microplate. This centrifugation step serves to clarify the sample by separating out any nonspecific components (e.g., precipitated proteins and lipids) that may lead to erroneous assay results if they are mistakenly added to the sample wells. In the bDNA assay, however, the entire sample preparation procedure is performed in the sample wells of the microplate. Consequently, nonspecific materials that either are introduced into the microwells upon the addition of the samples at the beginning of the assay or precipitate out of solution during the sample preparation procedure could lead to an erroneous result and poorer reproducibility between the two replicate tests for each sample. This problem would likely become evident when a large number of independent clinical samples are tested but would be less likely to be observed when a single sample is used for analytical precision experiments.

An important question that remains to be resolved relates to the sensitivity that is required for quantitative HBV monitoring. Technologies based on enzymatic amplification of target DNA such as PCR have the proven capability to detect many orders of magnitude less HBV DNA than current hybridization-based assays (9, 10). However, problems inherent to amplification technologies such as contamination (leading to false-positive results), enzyme inhibition (leading to false-negative results), and difficulties in the development of quantitative assays make this a challenge (10). In addition, some individuals may have a positive PCR result because they harbor a latent infection or integrated HBV DNA genomes and yet show no signs of liver disease (9).

Clearly, the optimal assay will be one that, among other attributes, best balances sensitivity with the clinical relevance of the assay result. Of the assays currently available, the bDNA assay seems to best fit these requirements (3, 9). The analytical and clinical studies performed in this study showed that the cross-linking assay described here has essentially the same sensitivity as the bDNA assay; however, there appears to be two differences in the performance of the two assays that are important from an economic viewpoint. First, the cross-linking assay required substantially less retesting of clinical samples than the bDNA assay, and second, although the total hands-on times of both assays are similar, there is a large difference in the times required to complete the two assays. Since an overnight hybridization is required in the bDNA assay, the test takes ∼20 h to complete. The use of solution-phase hybridization coupled with microparticle capture enables the cross-linking assay to be completed in less than 6 h, including sample preparation time.