Accurate quantitative methods for measuring viral nucleic acid levels in patient serum have recently been developed. Several recent studies have indicated that the level of HCV viremia correlates with the clinical stage of disease; patients with advanced stages of liver disease such as severe chronic active hepatitis, cirrhosis, and endstage liver disease had higher serum levels of HCV RNA than patients with mild HCV infections. Moreover, the HCV viremia titer may predict a subsequent response to antiviral therapy [11-15]. Two different technologies have been used to assess HCV viremia: the branched-DNA (bDNA) assay [13, 16] and quantitative PCR [11, 12]. The bDNA assay uses a novel approach to viral detection called signal amplification technology. Briefly, viral nucleic acids in a clinical specimen (for example, plasma or serum) are first solubilized by the addition of denaturing reagents and are then hybridized to microtiter plates through the use of virus-specific capture probes. The bound viral nucleic acid is then reacted with virus-specific extender probes followed by bDNA polymers. The bDNA polymers contain multiple repetitive binding sites for oligonucleotide-linked enzymes that catalyze the activation of chemiluminescent substrates; thus, "signal amplification" is achieved without amplification of viral nucleic acid.

Quantitative PCR relies on amplification of viral nucleic acids in the presence of a specific competitor molecule. In this assay, the ratio of amplification products derived from the native viral RNA and the competitive template are compared. In the quantitative competitive reverse transcription PCR assay, known amounts of synthetic internal control HCV RNA are titrated directly into clinical specimens and are then coextracted with the wild-type HCV RNA before analysis by competitive reverse transcription PCR [11]. Adding the competitor RNA directly into the clinical specimen allows the assay to be internally controlled for variations in efficiency of RNA extraction, complementary DNA synthesis, and PCR because the competitor is present during all steps. Quantitative PCR assays that use synthetic competitor templates as internal controls have been shown to be highly accurate for measuring various viral nucleic acids in clinical specimens. The current study compared the bDNA assay with a quantitative competitive reverse transcription PCR assay (quantitative PCR) for detecting and measuring HCV RNA levels in serum specimens in various patient populations.

Methods

Top Methods Results Discussion Author & Article Info References

Clinical Specimens

We selected for analysis 400 specimens from four different sources: prospective blood donors at the Puget Sound Blood Center (Seattle, Washington) or Irwin Memorial Blood Centers (San Francisco, California), patients attending hepatology clinics at either the University of Washington Medical Center or Harborview Medical Center (Seattle), patients receiving long-term renal dialysis at the Northwest Kidney Center (Seattle), and HCV-infected liver transplant recipients followed at the University of Washington. All serum samples were collected and processed within 4 hours of venipuncture to optimize detection of viral RNA [17] and were analyzed by laboratory personnel blinded to the clinical data. All specimens were tested for HCV antibodies by second-generation enzyme immunoassay (EIA2; Abbott Laboratories, North Chicago, Illinois) and by second-generation recombinant immunoblot assay (RIBA-II; Ortho Diagnostics, Raritan, New Jersey) when results of the enzyme immunoassay were positive. All 400 specimens were initially screened for HCV RNA by reverse transcription PCR combined with a specific radioactive probe hybridization, which is a highly sensitive and specific qualitative (nonquantitative) screening assay for HCV RNA [18, 19]. The analytical sensitivity of our reverse transcription PCR assay is less than 10 molecules of purified synthetic HCV RNA [19]; a clinical specificity of greater than 99% has been shown in previous studies and confirmed by ongoing proficiency testing approved by the College of American Pathology. We selected 101 of the 400 specimens as negative controls because they were negative for antibody to HCV by enzyme immunoassay and for HCV RNA by reverse transcription PCR; 299 specimens were selected as viremic specimens because they were positive for both antibody to HCV and HCV RNA by the screening reverse transcription PCR assay.

To evaluate the bDNA and quantitative PCR assays for detecting and measuring HCV RNA levels in patients followed prospectively, serum specimens were obtained at monthly intervals from 19 consecutive HCV-infected patients; 18 of these were treated for 6 months with recombinant interferon-, 3 million U three times per week, and 1 was treated with an escalating dose regimen of interferon for 48 months. Patients were classified as having a complete virologic response to interferon if their serum tested negative for HCV RNA by reverse transcription PCR at the end of therapy; we chose this virologic end point because reverse transcription PCR is the most sensitive test available for detection of HCV viremia.

Branched-DNA Assay

The bDNA assay was done as recommended by the manufacturer (Chiron Corp.; Emeryville, California); the design and principles of the assay are discussed above and in a previous report [16]. Quantitation of HCV RNA (expressed as equivalents per mL) was calculated by comparing relative luminescence with that obtained from an HCV RNA standard curve. All samples were assayed in duplicate, and the mean value of the duplicate tests was used for data analysis.

Quantitative Polymerase Chain Reaction

Known quantities of synthetic internal control HCV RNA were directly titrated into clinical specimens before nucleic acid extraction. Viral and internal control RNAs were coextracted from serum through the guanidinium thiocyanate method and were reverse transcribed; complementary DNA was amplified by PCR using primers derived from the highly conserved 5'-noncoding region of the HCV genome. Polymerase chain reaction amplification products were detected by either agarose gel electrophoresis plus ethidium bromide staining or Southern blot analysis in which a radiolabeled internal probe was used. The RNA copy number was deduced by comparing the PCR amplification product band intensity with the intensity of the internal control bands. The linear range of quantitative PCR ranged from 10³ to 10¹⁰ RNA molecules per mL (3 to 10 logs of HCV RNA per mL) by comparison with end-point dilution analysis of HCV RNA [11].

Results

Top Methods Results Discussion Author & Article Info References

Linearity of Hepatitis C Virus RNA Quantitation in Vitro

We assessed the linearity of HCV RNA quantitation in vitro using a serum sample obtained from an immunosuppressed patient with previously determined high-titer HCV viremia: The undiluted serum HCV RNA titer was 9.5 logs per mL by quantitative PCR. Eighteen 0.5-log dilutions of the high-titer serum samples into noninfected human sera were prepared, and HCV RNA was tested in duplicate by bDNA assay, quantitative PCR, and the qualitative screening assay, reverse transcription PCR (Figure 1). Dilutions 1 through 15 all tested positive for HCV RNA by reverse transcription PCR, whereas dilutions 16 through 18 tested negative. In contrast, only dilutions 1 through 6 were positive by bDNA assay; dilution 7 was indeterminate; and dilutions 8 to 18 were negative. The bDNA assay showed a nearly perfect slope within the linear range of the assay, from 6.0 logs to 8.5 logs when expressed as molecules per mL, or 5.5 logs to 7.9 logs when expressed as equivalents per mL as defined by the manufacturer (Figure 1). The bDNA assay went off scale at HCV RNA levels greater than 7.9 log-equivalents per mL.

View larger version (22K): [in this window] [in a new window]   Figure 1. Analysis of hepatitis C virus (HCV) RNA in serial dilutions of high-titer patient serum by two different quantitative assays, branched-DNA (bDNA) and quantitative polymerase chain reaction (Q-PCR), and a highly sensitive qualitative assay for HCV RNA, reverse transcription PCR (RT-PCR). A single specimen from an immunosuppressed patient with high-titer HCV viremia (9.5 log-transformed RNA molecules per mL by quantitative PCR assay) was subjected to 18 serial 0.5-log dilutions in uninfected human serum. The serial dilutions were analyzed in duplicate by bDNA assay, quantitative PCR, and qualitative PCR. Quantitative data are expressed as log-transformed equivalents per mL for the bDNA assay (open circles) and as log-transformed molecules per mL for the quantitative PCR assay (closed circles). Qualitative results of the bDNA assay and reverse transcription PCR assay are expressed below the graph as + (positive) or – (negative) for each individual dilution tube. The lower cutoff of the bDNA assay was 350 000 equivalents per mL, or approximately 5.5 log-transformed equivalents per mL, as stated by the manufacturer. By comparison, the PCR lower cutoff was 10 000-fold lower (1.5 log equivalents per mL).

The quantitative PCR assay gave linear results on the serially diluted serum specimen, from 4-log molecules per mL to 9.5-log molecules per mL. The reverse transcription PCR end point was estimated to be 100 molecules per mL or approximately 35 log-equivalents per mL; thus, reverse transcription PCR was approximately 4 orders of magnitude more sensitive than bDNA assay for detecting HCV RNA in vitro. The linear range of HCV RNA quantitation by quantitative PCR exceeded that of the bDNA assay by at least 3 orders of magnitude, although quantitative PCR varied slightly more than the bDNA assay. Within the linear range of the bDNA assay, the in vitro correlation between the bDNA and quantitative PCR assays was excellent, even though equivalents per mL as measured by the bDNA assay were consistently 0.5 logs less than the number of HCV RNA molecules per mL by quantitative PCR.

Evaluation of Clinical Specimens by Branched-DNA Assay Compared with Polymerase Chain Reaction-Based Assays

To evaluate the importance of these in vitro observations, we did bDNA testing on 299 viremic serum samples (HCV antibody positive and reverse transcription PCR positive) and 101 noninfected control serum samples (HCV antibody negative and reverse transcription PCR negative) obtained from 400 persons representing various patient populations. The bDNA assay gave positive results for 262 of 299 (87.6%) viremic specimens compared with only 3 of 101 (3%) nonviremic specimens (Table 1). We further analyzed the three specimens from noninfected persons, which initially tested positive by bDNA assay. Repeated duplicate testing by bDNA assay gave negative results for one specimen, indeterminate results for one specimen, and positive results for one specimen. Results of additional reverse transcription PCR testing using three alternative HCV primer pairs from the conserved 5'-noncoding region [20], the HCV core region [21], or the NS5 region [22] were negative for all three specimens. Thus, one of three positive results by the bDNA assay in the noninfected serum samples was attributed to technical error, one remained false-positive, and one became indeterminate. Therefore, the clinical specificity of the bDNA kit on repeat duplicate testing was 98%.

We next used quantitative PCR and the bDNA assay to assess HCV RNA levels in 255 of the 299 specimens that were originally classified as viremic by nonquantitative reverse transcription PCR (Table 2); these 255 specimens were selected because they had sufficient volume for further testing. Nineteen of 255 specimens (7%) had HCV RNA titers of 5 logs per mL or lower by quantitative PCR; 18 of 19 specimens in this category (95%) were negative by the bDNA assay. Forty-eight of 255 specimens (19%) had HCV RNA titers of either 5.5 or 6.0 logs per mL by quantitative PCR; 79% of these specimens tested positive by the bDNA assay. Eighty-three specimens (33%) had HCV RNA titers of 6.5 or 7.0 logs by quantitative PCR; 95% of these specimens tested positive by the bDNA assay. Finally, 105 of 255 (41%) specimens had HCV RNA titers of 7.5 logs or higher by quantitative PCR; all 105 of these specimens (100%) tested positive by the bDNA assay. Thus, the clinical sensitivity of the bDNA assay was 5% in specimens with HCV RNA titers of 5 logs (100 000 molecules) per mL or less, 80% for specimens with HCV RNA titers between 5 and 6 logs (100 000 to 1 000 000 molecules), and 98% for specimens with HCV RNA titers greater than 6 logs. These data indicate that 12% to 15% of viremic specimens obtained from nonimmunosuppressed patients with HCV infection are missed (that is, the test result was false-negative) by the bDNA assay because of viremia levels below the assay cutoff. This finding corroborates the data from our original in vitro observations.

Correlation between Branched-DNA Assay and Quantitative Polymerase Chain Reaction by Patient Population

Table 3 shows the data on the bDNA assay and quantitative PCR for assessment of HCV RNA levels in the 223 clinical specimens that tested positive by both assays. With both assays, median HCV RNA titers in serum were highest among immunosuppressed, HCV-infected liver transplant recipients. In three groups of patients (blood donors, patients receiving renal dialysis, and patients with chronic hepatitis), mean HCV RNA titers by quantitative PCR (molecules per mL) were 0.3 to 0.6 logs higher than mean HCV RNA levels by bDNA assay (equivalents per mL), a finding that supports the relation between molecules per mL and equivalents per mL defined in Figure 1. For these three patient groups, the quantitative correlation coefficients comparing the two assays were similar (r = 0.73 to 0.74; Table 3, and the bDNA and quantitative PCR assays correlated well [P < 0.001]. However, in the two patient groups with high-titer viremia (patients with end-stage liver disease and liver transplant recipients), the number of mean molecules per mL by quantitative PCR exceeded the number of mean equivalents per mL by bDNA assay by 10-fold or greater, and the quantitative correlation coefficients were less favorable (r = 0.58 and 0.57, respectively) Table 3 and Figure 2. We therefore did a series of experiments to investigate these differences. We initially selected 19 specimens in which HCV RNA titers measured by the bDNA assay were within the linear range of that assay, yet all 19 specimens had at least 10-fold higher levels of HCV RNA when assessed by quantitative PCR assay; indeed, all 19 specimens exceeded our cutoff for high-titer HCV viremia by quantitative PCR ( 7.5 logs of HCV RNA per mL of serum&238;). Ten of these samples were obtained from liver transplant recipients, and 9 were obtained from patients with end-stage liver disease. To assess whether the lower HCV RNA levels by the bDNA assay were caused by a prozone effect (that is, saturation of RNA binding sites in the assay), we diluted and retested the 19 serum samples with a new bDNA kit (Figure 3). The mean HCV RNA level obtained with the diluted and retested serum samples from liver transplant recipients was 7.86 logs per mL (72.4 million equivalents per mL) compared with a mean value of 7.25 logs per mL [17.8 million equivalents per mL] for the undiluted serum sample, an average increase of 300% over the original bDNA assay measurements Figure 3, left panel). The nine specimens from patients with end-stage liver disease originally had a mean HCV titer of 6.5 logs by the bDNA assay (3.2 million equivalents per mL) compared with a mean of 7.6 logs by the quantitative PCR assay. After dilution and retesting with a new bDNA kit, five specimens showed increases ranging from 50% to 530% of the original bDNA result [in equivalents per mL], whereas four specimens tested negative by the bDNA assay Figure 3, center panel). Dilution and retesting of the 10 control specimens from patients with chronic hepatitis and patients receiving renal dialysis gave no significant difference in bDNA results Figure 3, right panel). Thus, dilution and retesting with a different bDNA kit gave results that were more similar to the quantitative PCR results in 15 of the 19 specimens, suggesting that HCV RNA measurement was somewhat inhibited in the original bDNA assay. This effect was corrected by diluting and retesting the samples.

View larger version (13K): [in this window] [in a new window]   Figure 2. Hepatitis C virus (HCV) RNA titers determined by branched-DNA (bDNA) assay compared with quantitative polymerase chain reaction (Q-PCR) for the 223 total specimens (top) and the 31 specimens from patients with end-stage liver disease (bottom) described in (Table 3). The HCV RNA levels are expressed as log-transformed equivalents per mL (log eq/mL) for the bDNA assay on the x-axis and as log-transformed molecules per mL (log molecules/mL) for the quantitative PCR assay on the y-axis. In the top panel, correlation between the two assays was highly significant (P < 0.0001 for the combined patient groups), but the relation appeared nonlinear at the upper and lower ranges defined by the quantitative PCR assay. The specimens from liver transplant recipients showed greater discrepancy in HCV RNA titers determined by the two assays.

View larger version (17K): [in this window] [in a new window]   Figure 3. Variation in branched-DNA (bDNA) assay results on repetitive testing of specimens obtained from liver transplant recipients and patients with cirrhosis. Hepatitis C virus (HCV) RNA titers were determined by the bDNA assay before dilution (data set A) and after 10-fold dilution and retesting in a new bDNA kit (data set B). The data are presented as log-transformed equivalents per mL (that is, the final results are corrected for dilution). = specimens from liver transplant recipients (n = 10), \#9679; = specimens from patients with end-stage liver disease (n = 9), and filled diamond equals specimens from patients attending a hepatology clinic for evaluation of chronic hepatitis C (n = 6) or from patients receiving renal dialysis (n = 4).

We next assessed the within-run variation and kit-to-kit variation of the bDNA assay using specimens from liver transplant recipients. Duplicate diluted and undiluted aliquots of transplant serum samples assayed in the same bDNA run showed a coefficient of variation of 28%, a variation approximately fourfold greater than that observed for the bDNA kit-positive controls. To assess kit-to-kit variation, six different undiluted transplant specimens were assayed in duplicate on three different bDNA kits. We again observed a high coefficient of variation in HCV RNA titers, which ranged from 180% to 330%. Finally, 11 diluted transplant specimens were assayed in duplicate in two different experiments using three different bDNA kits; in the first experiment, a high variation in bDNA results was recorded (mean variation, 79%), whereas in the second experiment, a low variation in bDNA results was recorded (mean variation, 7%). In summary, we identified a subset of specimens from liver transplant recipients and patients with cirrhosis that had higher titers of HCV viremia by the quantitative PCR assay than the bDNA assay; results of repetitive testing of such specimens by the bDNA assay showed a higher coefficient of variation than was observed when we used viremic specimens obtained from other patient populations. This variability may partially explain the increased discordance we observed between the bDNA assay and quantitative PCR when assessing HCV RNA levels in these patients.

Monitoring Hepatitis C Virus RNA Levels during Interferon Therapy

To evaluate whether the differences in the dynamic range of the bDNA and quantitative PCR assays would affect interpretation of therapeutic monitoring of HCV RNA levels in patients treated with interferon-, we analyzed sequential specimens obtained from 19 consecutive patients receiving therapy; at the beginning of therapy, all patients were HCV RNA positive by both bDNA and quantitative PCR assays. In each case, bDNA and quantitative PCR testing was done on identical specimens obtained monthly before, during, and after interferon therapy. Overall, 5 of 19 patients (26%) were classified as having a complete virologic response because they were negative for HCV RNA by reverse transcription PCR at the end of therapy. Fourteen of 19 patients (74%) were classified as having either an incomplete no virologic response because they were positive for HCV RNA by reverse transcription PCR at the end of therapy [nonresponders]. Of the 14 nonresponders, 4 became bDNA negative during interferon therapy and 1 remained bDNA negative at the end of therapy Figure 4, bottom right). None of the 14 nonresponders became negative by quantitative PCR at any point during therapy.

View larger version (26K): [in this window] [in a new window]   Figure 4. Monitoring HCV RNA levels during interferon- (IFN) therapy by bDNA and quantitative PCR (Q-PCR). Patients were treated with IFN, 3 MU three times per week for 24 weeks; two patients (bottom left and right panels) received additional interferon therapy. The HCV RNA titers were determined by the bDNA and Q-PCR assays on a monthly basis during therapy. Serum alanine aminotransferase (ALT) levels are shown in each panel (reference range, 6 to 43 IU/L). The patient described in the top left panel was classified as having no virologic response at the end of therapy, although a partial response was evident at 2 months. The patient described in the top right panel was classified as having a virologic response: Results of both HCV RNA assays became negative after the first 2 weeks of therapy, and the patient has remained HCV RNA negative with normal ALT levels during the 8-month post-therapy follow-up. The bottom left panel shows discordance between the bDNA and Q-PCR assays: The patient had a rapid antiviral response by bDNA (negative by 1 month) but a slow response by Q-PCR (positive until month 5). After stopping therapy at 24 weeks, the patient showed a rapid virologic relapse by Q-PCR (positive at month 6 [week 26]), whereas the relapse was not detected by bDNA until month 8. After 14 months without therapy, this patient was retreated with an increased IFN dose (5 MU three times per week for 24 weeks) and had a complete virologic and biochemical response (HCV RNA negative and normal ALT values). The bottom right panel describes a patient who appeared to have a complete virologic response to an escalated dose of IFN by bDNA assay and by serum ALT values but had only a partial response by Q-PCR assay. This patient became transiently negative for HCV RNA by bDNA in month 4 and became persistently negative by bDNA between months 8 and 11. During this period, serum ALT values returned to normal. However, the patient was continuously viremic by the more sensitive Q-PCR assay. Interferon therapy was discontinued before month 11 after ALT values had been normal for 5 months. The patient had a clinical relapse (abnormal ALT values) within 1 month stopping of therapy. RT-PCR equals reverse transcription polymerase chain reaction. To convert the ALT values to µkat/L, multiply by 0.01667.

Figure 4 shows the results of monthly monitoring of HCV RNA levels during inteferon therapy by both bDNA assay and quantitative PCR in four representative cases. In the first case Figure 4, top left], the results of the bDNA assay transiently converted to negative after 2 months of therapy; however, according to quantitative PCR, the HCV RNA titer in serum was still 5.5 logs per mL. During the latter 3 months of therapy, HCV RNA titers returned to pretreatment levels by both assays. In the second case Figure 4, top right), bDNA and quantitative PCR gave concordant results. This patient rapidly responded to therapy with 3 million U of interferon three times per week: The results of the bDNA assay and quantitative PCR converted to negative after 1 and 2 weeks of therapy, respectively, and results of both assays remained negative throughout treatment. Alanine aminotransferase values also decreased from a pretreatment level of 2.6 µkat/L (156 IU/L) to less than 0.33 µkat/L (20 IU/L) after 2 weeks of therapy. The patient has remained HCV RNA negative with normal alanine aminotransferase values during 8 months of follow-up.

The patient whose findings are shown in the bottom left panel of Figure 4 became negative by the bDNA assay within 4 weeks of the initiation of interferon therapy (3 million U three times per week) and remained bDNA negative throughout the rest of the 6-month therapy. In contrast, results of quantitative PCR testing indicated the persistence of HCV viremia throughout the first 4 months of therapy, with a slow but progressive decrease in viral RNA titers until month 5, when the patient became quantitative PCR negative. This patient's alanine aminotransferase levels progressively decreased with therapy, from 2.13 µkat/L (128 IU/L) before treatment to 0.83 µkat/L (50 IU/L) at the end of therapy. The level increased to 4.62 µkat/L (277 IU/L) by 1 month after therapy. Within 1 week of discontinuation of therapy, the patient's HCV RNA levels rapidly increased according to quantitative PCR, but the patient remained negative by bDNA assay until retesting was done 4 weeks later. In this case, quantitative PCR was more sensitive for detecting persistent viremia during interferon therapy, showing a slow but progressive antiviral effect with 3 million U three times per week. After 14 months without therapy, this patient has subsequently shown a rapid and complete virologic and biochemical response to an increased dose of interferon (5 million U three times per week for 6 months). The patient tested negative for HCV RNA by both bDNA and quantitative PCR 1 month, 3 months, and 6 months after initiation of therapy and had normal alanine aminotransferase values at the first visit after therapy began.

The patient whose findings are shown in the bottom right panel of Figure 4 became transiently negative by the bDNA assay after 4 months of interferon therapy at 3 million U three times per week. However, results of the quantitative PCR assay showed no apparent reduction in HCV RNA titers, and although alanine aminotransferase values were moderately suppressed, they remained abnormal. After 24 weeks of therapy, the interferon dose was increased to 5 million U three times per week because alanine aminotransferase values had not returned to normal. After dose escalation, the patient became HCV RNA negative according to the bDNA assay, and alanine aminotransferase values decreased to less than 0.33 µkat/L (20 IU/L) between months 7 and 11. However, the patient remained HCV RNA positive according to the quantitative PCR assay. When therapy was discontinued (week 42), this patient appeared to be a responder by bDNA assay (HCV RNA negative) and by alanine aminotransferase values [which had remained normal for 3 months], but the patient had low HCV viremia titers by quantitative PCR assay Figure 4, bottom right). Two weeks after discontinuation of therapy (month 11), the patient had a 2.5-log increase in HCV RNA levels by quantitative PCR. However, the bDNA assay result remained negative, and alanine aminotransferase values remained normal. The patient developed a biochemical relapse at 48 weeks (alanine aminotransferase level, 1.86 µkat/L [112 IU/L]) and was positive for HCV RNA by the bDNA assay at 50 weeks; this finding corroborated the absence of complete virologic response as detected by the quantitative PCR assay 6 weeks earlier.

Discussion

Top Methods Results Discussion Author & Article Info References

Our study shows several important facets of the quantitation of HCV RNA levels in clinical specimens using the new chemiluminescence-based bDNA kit assay compared with reference reverse transcription PCR (qualitative PCR) and quantitative PCR assays. In its current form, the bDNA assay detected HCV viremia in 87% of specimens from untreated patients. The specificity of the bDNA assay was 97% compared with our reference reverse transcription PCR assay and improved to 98% or 99% on repeat duplicate testing, depending on whether the indeterminate result is considered to be positive or negative. Furthermore, we found a strong correlation between the HCV RNA levels determined in a single undiluted serum specimen by the bDNA assay and the absolute HCV RNA titers determined by our research quantitative PCR assay, which requires that previously quantitated synthetic HCV RNA standards be diluted directly into aliquots of clinical specimens. The bDNA assay is thus both less time-consuming and labor-intensive than quantitative PCR. A final strength of the bDNA assay is its reproducibility and low within-run variability, with a coefficient of variation that is usually less than 10% for duplicate aliquots of the same specimens. On the basis of repetitive testing of 19 identical aliquots of the viremic patient's serum, the run-to-run coefficient of variation of the bDNA assay was 23%.

Although the bDNA assay is rapid, easy to perform, and reproducible, it has clear limitations in detecting low-level HCV viremia. The analytical sensitivity of bDNA assay was approximately 4 logs (10 000-fold) less than our PCR-based methods by quantitative in vitro analysis of HCV RNA. The manufacturer's stated cutoff for detecting HCV viremia in clinical specimens (350 000 equivalents per mL) roughly correlated with 350 000 to 1 million HCV RNA molecules per mL, as determined by quantitative PCR testing of the same specimens. The sensitivity of the bDNA assay for detecting HCV RNA in clinical specimens was directly proportional to the HCV titer, ranging from 5% in specimens with low-titer viremia (5 logs or less by quantitative PCR) to 100% in specimens from liver transplant recipients with high-titer viremia (7.5 logs or greater). In our sampling, 12% to 15% of viremic HCV-infected patients were not identified as viremic by the bDNA assay; this finding was independent of patient group or HCV genotype. Because the degree of HCV viremia is associated with the clinical stage of infection [11], the percentage of false-negative bDNA test results may vary among patient populations and centers. However, our data show that the bDNA assay is not sensitive enough to rule out the diagnosis of active HCV infection in patients with HCV RNA titers less than 10⁶ molecules per mL (350 000 equivalents per mL). Therefore, persons who are HCV antibody positive but negative for HCV RNA by the bDNA assay should be routinely tested by reverse transcription PCR assay to detect or exclude low-level viremia because patients with low-titer HCV infection appear to be the best candidates for systemic interferon therapy [12-15].

We observed an unexpectedly high discordance between the bDNA assay and quantitative PCR for measuring HCV RNA levels in serum specimens obtained from liver transplant recipients and patients with end-stage liver disease. We identified specimens in which quantitative PCR and bDNA assay gave discordant results: Quantitative PCR indicated high titers of HCV RNA, whereas the bDNA assay results on the same specimens were at least 10-fold lower. In initial experiments, dilution and retesting of the discordant specimens using a different bDNA kit yielded an average fourfold increase in HCV RNA titers by bDNA, which suggested that a prozone effect might be occurring (that is, saturation of RNA-binding sites caused by the high-titer serum samples, which would result in falsely low test values). However, further testing of specimens obtained from liver transplant recipients by the bDNA assay showed a high coefficient of variation (112%) among different bDNA kits using the same undiluted transplant specimens. Specimens from transplant recipients and possibly from patients with cirrhosis may contain substances that cause interference in the bDNA assay; this might help explain recent discrepant reports on the utility of the bDNA assay for predicting histologic recurrence of hepatitis C in liver transplant recipients [23-25]. Potential explanations for this include saturation of binding sites in the bDNA assay by high viral RNA levels (that is, the prozone effect), the presence of dilutable or nondilutable interfering substances (such as drugs or metabolites), unexpected inhibition of HCV RNA binding at high concentrations, or subtle effects of viral RNA sequence or RNA secondary structure on either the bDNA or quantitative PCR assay. Our preliminary experiments suggested that the performance of the bDNA assay may improve when high-titer specimens from transplant recipients are diluted 10-fold before testing.

One important clinical application of HCV RNA quantitation is the assessment of HCV viremia during antiviral therapy. We therefore did a detailed analysis of HCV viremia by both the bDNA and quantitative PCR assays in 19 consecutive patients receiving systemic therapy with interferon-. Our results indicate that the linear range of the bDNA assay is not sufficient to allow assessment of the entire spectrum of HCV viremia during therapy, because many patients became negative by bDNA during interferon therapy but were still viremic when tested by the more sensitive PCR-based methods (Figure 4). All such patients had relapse after discontinuation of therapy. Of particular interest were cases in which only a modest response was detected by quantitative PCR; such patients responded rapidly to retreatment with higher doses of interferon (for example, the patient whose findings are shown in the bottom left panel of Figure 4, who became HCV RNA negative after 1 month of therapy at the higher dose). Thus, potential pitfalls in assessing HCV viremia by the bDNA assay are an underestimation of high-titer viremia in some patients and false-negative test results in other patients in whom low-level viremia persists. Because patients who remain viremic by reverse transcription PCR may require either a longer duration or higher doses of interferon therapy to achieve a complete virologic response, we recommend that bDNA-negative patients receive additional testing by the more sensitive reverse transcription PCR method.

The bDNA assay represents a novel diagnostic technology that is based on signal amplification rather than target amplification, is readily available in kit format, is well standardized and relatively precise, and uses nonradioactive (chemiluminescent) detection. As such, it has great advantages for use in the routine clinical laboratory. However, our study indicates some limitations to bDNA technology with regard to the assay's ability to detect the presence of low-level HCV RNA and to accurately assess high-level HCV RNA in some patient populations. Two distinct advantages of the quantitative PCR assay are its increased sensitivity for quantitating lower levels of HCV viremia and the use of synthetic internal control HCV RNA as a molecular copy number standard that may reduce the possibility of interference. However, quantitative PCR is labor-intensive, and standardization among laboratories would be extremely difficult; therefore, quantitative PCR may be best suited for research applications and as a "gold standard" for evaluation of newer technologies. Currently, the use of bDNA in combination with a reliable reverse transcription PCR assay probably offers the most optimal assessment of viral load during therapy.

Ms. dela Rosa: Hepatitis Laboratory, 11th Floor, 1200 12th Avenue South, Seattle, WA 98144.

Dr. Carithers: University of Washington Medical Center, Box 356174, Room EE-429, 1959 NE Pacific, Seattle, WA 98195.

Dr. Willson: Harborview Medical Center, 325 9th Avenue, Seattle, WA 98104-2499.