Efficiency of the Ortho VITROS Assay for Detection of Hepatitis C Virus-Specific Antibodies Increased by Elimination of Supplemental Testing
http://www.100md.com
微生物临床杂志 2005年第5期
Virology Reference and Molecular Diagnostic Laboratories, VA CT Healthcare System, West Haven, Connecticut
Department of Laboratory Medicine, Yale University School of Medicine, New Haven, Connecticut
ABSTRACT
The clinical significance of specimens with low sample-to-cutoff (S/Co) ratios in the Ortho VITROS chemiluminescence assay (CIA) for detection of antibodies to hepatitis C virus (HCV) was evaluated. In one study of 482 CIA-reactive samples, none of the 83 samples with S/Co ratios of <5 was HCV RNA positive. In a subsequent study, 332 samples with S/Co ratios of between 1 and 20 were tested with the recombinant immunoblot assay (RIBA). None of the 163 samples with S/Co ratios of <5 was RIBA positive, 83% were RIBA negative, and 28 samples (18%) were RIBA indeterminate. HCV RNA and/or clinical evidence of hepatitis was not found in the 27 indeterminate cases examined. These results show that over 99% of samples with very low S/Co ratios (5) have no evidence of HCV infection. Therefore, we suggest that the HCV antibody testing algorithm for the VITROS assay might be modified to eliminate supplemental testing of samples with very low S/Co ratios.
TEXT
Assays for the detection of hepatitis C virus (HCV) antibodies have high false-positive rates, particularly in low-prevalence populations. Therefore, other tests such as recombinant immunoblot assay (RIBA) or HCV RNA PCR are used to confirm positive HCV antibody screening tests. To facilitate usage of the supplemental testing, the Centers for Disease Control and Prevention (CDC) published guidelines that incorporate anti-HCV assay signal-to-cutoff (S/Co) ratios into reflex testing algorithms for HCV antibody testing, in order to provide a more systematic approach for the laboratory diagnosis of HCV and minimize the number of specimens that require supplemental testing (2). Data obtained with three anti-HCV screening assays were used for these recommendations: two enzyme immunoassays (EIAs) (Abbott HCV EIA 2.0 and ORTHO HCV version 3.0 enzyme-linked immunosorbent assay) and one enhanced chemiluminescence immunoassay (CIA)(VITROS anti-HCV assay; Ortho-Clinical Diagnostics).
The VITROS anti-HCV CIA has recently been shown to be at least as specific and sensitive as conventional EIAs (7, 9), and use of this CIA has been increasing. However, detailed studies of the CIA S/Co ratio value predictive of a positive supplemental HCV test have been limited (7, 9). The CDC guidelines have suggested reflex supplemental testing for samples with S/C ratios of <8.0 in the VITROS anti-HCV assay based on the evaluation of a total of 1,326 reactive samples, with only 142 of these specimens having S/Co ratios of 7.9 (2).
We started using the VITROS anti-HCV assay shortly after its Food and Drug Administration approval in August 2001 and have prospectively applied reflex supplemental testing to all samples with S/Co ratios of between 1.00 and 20.0 in the setting of a population of U.S. veterans (8). The objectives of the present study were (i) to determine the S/Co ratio predictive of detectable HCV RNA; (ii) to evaluate, in samples with low S/Co ratios (between 1 and 20), the relationship between S/Co ratios and the number and types of bands detected by RIBA; (iii) to assess, for samples with indeterminate RIBA results, the correlation between S/Co ratios and the type of bands detected by RIBA; and (iv) to analyze patient samples and data from patients who had very low CIA ratios (S/Co, <5) and indeterminate RIBA results for the presence of other indicators of HCV infection.
Patient samples used in this study were sent to our laboratory from the nine Department of Veterans Administration (VA) hospitals located in the Veterans Integrated Service Network I (VISN I). The overall HCV seroprevalence for the veteran population evaluated is approximately 8 to 10%. All serum samples were assessed for the presence of antibodies to HCV using the ORTHO VITROS anti-HCV assay (Ortho-Clinical Diagnostics). Only samples that were repeatedly reactive were included in this study. All samples were stored frozen for further evaluation. Supplemental testing was performed using the RIBA (Ortho-Clinical Diagnostics), the Roche Amplicor HCV test, version 2.0 (Roche Molecular Diagnostics Systems), and/or the Roche Monitor HCV test (Roche Molecular Diagnostics Systems).
The VITROS anti-HCV assay is a two-step sandwich CIA for the detection of human antibody to various proteins of HCV. Results are calculated as a normalized S/Co value. During the calibration process, a lot-specific parameter encoded in the lot validation card is used to determine a valid cutoff value. Samples with S/Co ratios of 1.0 were retested in duplicate and considered "repeatedly positive" if S/Co ratios were 1.0 for at least two of three determinations.
Two separate studies were performed: the first in 2002 and the second from May 2003 to March 2004. In the first study, a total of 482 CIA-positive samples were examined prospectively for the presence of HCV RNA. In the second study, 332 CIA low-positive samples were evaluated by RIBA. For samples with CIA S/Co ratios of <5 and indeterminate RIBAs, the patient database was searched for available HCV RNA results either for the sample in question or for samples drawn within six months of the sample in question. Nucleic acid testing was performed on available samples. In addition, available clinical records were reviewed for all patients with both indeterminate RIBAs and S/Co ratios of <5 to further clarify the disease status. The charts were reviewed for clinical signs of past or present hepatitis and laboratory evidence of hepatitis, such as elevated liver enzyme levels. Repeat samples were excluded from the study.
Our initial study focused on the determination of the S/Co ratio predictive of an HCV RNA-positive result. Results from a total of 482 samples evaluated in 2002 are included in Table 1. A majority of samples (n = 349) were tested using the HCV RNA qualitative assay alone, 101 samples were tested by the HCV RNA quantitative assay, and 32 samples were evaluated by both assays. Of the 154 samples with S/Co ratios of between 1 and 20, 130 samples were tested with the HCV RNA qualitative assay, 20 samples were tested with the HCV RNA quantitative assay, and 4 samples were evaluated with both assays. Although the number of samples with detectable HCV RNA increased in relation to the S/Co ratio, the majority of HCV RNA-positive samples had an S/Co ratio of >20 (266 of 328; 81%). In contrast, only 7 of 71 (10%) samples with S/Co ratios of between 5 and 20 were HCV RNA positive. HCV RNA was not detected in any of the samples with CIA S/Co ratios of <5. Based on these results, we have routinely categorized samples with CIA S/Co ratios of >1 and <20 as low positive and used supplemental testing for all of these samples.
In order to examine in detail the significance of low-positive samples (CIA S/Co ratio of between 1 and 20), we evaluated in more detail a total of 332 samples tested between May 2003 and March 2004. Results of RIBA testing for the 332 samples are outlined in Table 2. None of the 163 samples with S/C ratios of <5 was found to be RIBA positive. Instead, the majority (83%) of the samples with S/Co ratios of 1 to 5 were RIBA negative, and 28 samples were indeterminate. The number of positive RIBAs increased as the S/Co ratio increased, with the highest proportion of RIBA-positive samples identified in the group of samples with S/Co ratios of 16 to 20 (51 of 57; 89%). However, it should be noted that for the 129 samples with S/Co ratios of >8 and <20, 12 (9%) samples were found to be RIBA negative.
Details of the number and type of positive RIBA bands in relation to mean and range of S/Co ratios are shown in Table 3. The mean S/Co ratio for the 163 RIBA-negative samples was significantly lower than that of the 77 RIBA-indeterminate samples (3.19 versus 7.36; P < 0.1; t test), although the ranges of values in both groups were similar. For the 77 RIBA indeterminate samples, reactivity at bands C33c and C22p occurred more frequently than reactivity at bands C100 and NS5. For samples interpreted as RIBA positive, the mean S/Co ratios were significantly different between samples with two positive bands as compared to samples with three and four positive bands (t test; P < 0.01). Reactivity at bands C100, NS5, C33c, and C22p was detected in 84, 54, 96, and 84% of the samples, respectively. The presence of C100 or NS5 bands was associated with lower S/Co ratios (means of 2.68 and 2.24, respectively) than the presence of C33c or C22p, and the mean S/Co ratio for the 33 samples that only had reactivity to C22p was highest (11.04).
Among the samples with CIA S/Co ratios of <5, 28 samples were found to be RIBA indeterminate. Twenty-seven of these samples were further evaluated for the presence of HCV RNA and for clinical evidence of HCV infection by reviewing clinical records. Chart reviews indicated that most of the patients were tested as part of routine screening. Twenty-four of the 28 samples (86%) were tested for HCV RNA and were found to be negative. For three of the remaining four samples, examination of the patient's clinical records indicated no clinical evidence of hepatitis. For all of the 27 patients' records examined, there was no laboratory evidence of abnormal liver function tests and there were no other signs of HCV infection. Clinical status could not be resolved in one of 28 patients because no clinical notes were found and no serum was available for further HCV testing.
The increasingly sophisticated methods for diagnosing HCV infection have a direct impact on patient management, and careful use of the available assays is essential for the accurate and efficient diagnosis of HCV infection. Several CDC-sponsored and other anti-HCV seroprevalence studies have indicated that reactive S/Co ratios could be used to predict supplemental positive results (2, 5-7, 11); most of these studies used EIAs (2, 5, 6, 11). In the present report, the first fully automated random access diagnostic test approved by the FDA for the detection of HCV antibody, the VITROS anti-HCV assay, was utilized. The test menu for this analyzer, which initially also comprised hepatitis B surface antigen and antibody, has been recently expanded to include hepatitis B virus core total and immunoglobulin M antibody tests.
Few published studies have documented experience with the VITROS Anti-HCV assay (7-9). Ismail et al. compared the performance of the VITROS Anti HCV assay to the Abbott second-generation EIA; 318 HCV-negative samples and 177 HCV-positive samples in the Abbott assay were evaluated, but confirmatory testing was only performed on the eleven discrepant results (9). Among the seven VITROS anti-HCV-positive samples, RIBA was negative in four samples and indeterminate in three samples; all seven samples had S/Co ratios of >1 and <8, but the true status of the three indeterminate samples was not determined. Dufour et al. reported on the supplemental testing of CIA low-positive samples by RIBA for 224 samples and by HCV PCR for 55 samples (7). With the present report, we add results from a total of 486 samples with CIA S/Co ratios of <20 (154 samples in our first study, and 332 samples in our second study).
We have demonstrated in the first part of our study that the vast majority (81%) of 328 samples with S/Co ratios of >20 were HCV RNA positive. CDC recommendations have suggested that all CIA-positive samples with S/Co ratios of >8 can be reported as positive without further supplemental testing (2). In addition, Dufour et al. have reported that 95% of samples with CIA S/Co ratios of between 8 and 20 were RIBA positive or indeterminate, although HCV RNA was not detected in 25 samples tested (7). Analysis of our results for samples with CIA S/Co ratios of between 8 and 20 indicates that 43 of 129 (33%) were RIBA negative or indeterminate. If the CDC recommendation of an S/Co ratio of 8 is applied to our data set, reflex testing would not be performed for all samples with S/Co ratios of 8 to 20; this would have resulted in the report of 12 false-positive results. In order to avoid this, we have over the past several years performed supplemental testing for all positive samples with S/Co ratios of <20.
Approximately half of the 486 CIA low-positive samples evaluated in the present study were found to have S/Co ratios of <5. The fact that S/Co ratios of <5 can predict a negative result in over 99% of the samples examined in our study is of particular significance. Among samples with a CIA S/Co ratio of <5, none of the 163 samples examined by RIBA was positive and none of the 83 samples tested by PCR had detectable HCV RNA. In addition, only a minority of samples were RIBA indeterminate and none of these samples showed evidence of HCV infection. Dufour et al. found that 13 of 129 (10%) samples with S/Co ratios of <8 were RIBA positive (7). In contrast, we found that only 3% (6 of 203) of samples with an S/Co ratio of <8 were RIBA positive, and all positive samples in that range had an S/Co ratio of >5. The reason for the difference is uncertain but could be due to differences in sample size or population examined.
An indeterminate RIBA pattern, characterized by the presence of only one of the four bands, was found for 77 of our samples, with C33c and C22p band patterns detected most frequently. In a previous study on the significance of indeterminate third-generation RIBAs, a similar observation was made in 59 samples (10). Higher mean S/Co ratios were observed in samples with C33c and C22p bands as compared to samples with C100 or NS5 bands. Furthermore, eight of the nine samples with C100 or NS5 reactivity had S/Co ratios of <5. In addition, among the RIBA-positive samples with two-band reactivity, the C33c band was most frequently detected. Previous studies have made similar observations, with a high proportion of isolated reactivity to NS5 found in HCV RNA-negative subjects (4, 12) and seroconversion studies indicating that anti-C33 and anti-C22 usually are the first detectable HCV antibodies (3).
We have in the present study demonstrated that there was no evidence of HCV infection in any of the samples with S/Co ratios of 5 evaluated; no sample was found to be RIBA positive, a majority of samples were RIBA negative, and there was no laboratory or clinical evidence of HCV infection in the 27 RIBA indeterminate cases examined. Therefore, we suggest that when the VITROS assay is used, samples with S/Co ratios of 5 do not require supplemental testing because over 99% of samples with S/Co ratios of 5 are not associated with laboratory or clinical evidence of HCV infection. Considering that, in our study, 246 of 486 (50.6%) of low-positive samples had an S/Co ratio of <5, elimination of supplemental testing of samples with S/Co ratios between 1.00 and 5.00 will reduce unnecessary testing and preserve resources. This cost savings has the potential to have significant impact on VA facilities nationwide, as HCV has been identified as the most important emerging pathogen in the VA health care system (1). We have recently opted to report all our CIA low-positive samples with S/CO ratios between 1 and 5 as "borderline," with the recommendation that follow-up testing be performed when HCV infection continues to be suspected based on other clinical or laboratory information.
ACKNOWLEDGMENTS
We gratefully acknowledge the technical assistance of Maria Cavaiuolo, Meghan Condon, Kim McGowan, and Denise Miko.
REFERENCES
Bini, E. J. 2003. Hepatitis C in veterans. Curr. Hepatitis Rep. 2:108-115.
Centers for Disease Control and Prevention. 2003. Guidelines for laboratory testing and result reporting of antibody to hepatitis C virus (anti-HCV). Morb. Mortal. Wkly. Rep. 52(RR-3):1-13.
Courouce, A. M., N. Le Marrec, A. Girault, S. Ducamp, and N. Simon. 1994. Anti-hepatitis C virus seroconversion in patients undergoing hemodialysis: comparison of second- and third-generation anti-HCV assays. Transfusion 34:790-795.
Damen, M., H. L. Zaauer, H. T. M. Cuypers, V. Vrielink, C. L. Van Der Poel, H. W. Reesink, and P. N. Lelie. 1995. Reliability of the third generation recombinant immunoblot assay for hepatitis C virus. Transfusion 35:745-749.
Dos Santos, V. A., R. S. Azevedo, M. E. Camargo, and V. A. Alves. 1999. Serodiagnosis of hepatitis C virus. Effect of new evaluation of cutoff values for enzyme-linked immunosorbent assay in Brazilian patients. Am. J. Clin. Pathol. 112:418-424.
Dufour, D. R., M. Talastas, M. D. A. Fernandez, B. Harris, D. B. Strader, and L. B. Seeff. 2003. Low-positive anti-hepatitis C virus enzyme immunoassay results: an important predictor of low likelihood of hepatitis C infection. Clin. Chem. 49:479-486.
Dufour, D. R., M. Talastas, M. D. A. Fernandez, and B. Harris. 2003. Chemiluminescence assay improves specificity of hepatitis C antibody detection. Clin. Chem. 49:940-944.
Griffith, B. P., J. Falcone, D. Miko, C. K. Y. Fong, P. K. Barua, S. Campbell, and G. Stack. 2003. Hepatitis C virus antibody signal to cut-off ratios predictive of HCV RNA detection: comparison of the Ortho Vitros and EIA assays. Nineteenth Annu. Clin. Virol. Symp., abstr. M48.
Ismail, N., G. E. Fish, and M. B. Smith. 2004. Laboratory evaluation of a fully automated chemiluminescence immunoassay for rapid detection of HBsAg, antibodies to HBsAg, and antibodies to hepatitis C virus. J. Clin. Microbiol. 42:610-617.
Pawlotsky, J. M., A. Bastie, C. Pellet, J. Remire, F. Darthuy, L. Wolfe, C. Sayada, J. Duval, and D. Dhumeaux. 1996. Significance of indeterminate third-generation hepatitis C virus recombinant immunoblot assay. J. Clin. Microbiol. 34:80-83.
Tobler, L. H., G. Tegtmeier, S. L. Stramer, S. Quan, J. Dockter, G. Giachetti, and M. P. Busch. 2000. Lookback on donors who are repeatedly reactive on first-generation hepatitis C virus assays: justification and rational implementation. Transfusion 40:15-24.
Vernelen, K., H. Claeys, H. Vernhaert, A. Volclaerts, and C. Vermylen. 1994. Significance of NS3 and NS5 antigens in screening for HCV antibody. Lancet 343:853.(Margret Oethinger, Donald)
Department of Laboratory Medicine, Yale University School of Medicine, New Haven, Connecticut
ABSTRACT
The clinical significance of specimens with low sample-to-cutoff (S/Co) ratios in the Ortho VITROS chemiluminescence assay (CIA) for detection of antibodies to hepatitis C virus (HCV) was evaluated. In one study of 482 CIA-reactive samples, none of the 83 samples with S/Co ratios of <5 was HCV RNA positive. In a subsequent study, 332 samples with S/Co ratios of between 1 and 20 were tested with the recombinant immunoblot assay (RIBA). None of the 163 samples with S/Co ratios of <5 was RIBA positive, 83% were RIBA negative, and 28 samples (18%) were RIBA indeterminate. HCV RNA and/or clinical evidence of hepatitis was not found in the 27 indeterminate cases examined. These results show that over 99% of samples with very low S/Co ratios (5) have no evidence of HCV infection. Therefore, we suggest that the HCV antibody testing algorithm for the VITROS assay might be modified to eliminate supplemental testing of samples with very low S/Co ratios.
TEXT
Assays for the detection of hepatitis C virus (HCV) antibodies have high false-positive rates, particularly in low-prevalence populations. Therefore, other tests such as recombinant immunoblot assay (RIBA) or HCV RNA PCR are used to confirm positive HCV antibody screening tests. To facilitate usage of the supplemental testing, the Centers for Disease Control and Prevention (CDC) published guidelines that incorporate anti-HCV assay signal-to-cutoff (S/Co) ratios into reflex testing algorithms for HCV antibody testing, in order to provide a more systematic approach for the laboratory diagnosis of HCV and minimize the number of specimens that require supplemental testing (2). Data obtained with three anti-HCV screening assays were used for these recommendations: two enzyme immunoassays (EIAs) (Abbott HCV EIA 2.0 and ORTHO HCV version 3.0 enzyme-linked immunosorbent assay) and one enhanced chemiluminescence immunoassay (CIA)(VITROS anti-HCV assay; Ortho-Clinical Diagnostics).
The VITROS anti-HCV CIA has recently been shown to be at least as specific and sensitive as conventional EIAs (7, 9), and use of this CIA has been increasing. However, detailed studies of the CIA S/Co ratio value predictive of a positive supplemental HCV test have been limited (7, 9). The CDC guidelines have suggested reflex supplemental testing for samples with S/C ratios of <8.0 in the VITROS anti-HCV assay based on the evaluation of a total of 1,326 reactive samples, with only 142 of these specimens having S/Co ratios of 7.9 (2).
We started using the VITROS anti-HCV assay shortly after its Food and Drug Administration approval in August 2001 and have prospectively applied reflex supplemental testing to all samples with S/Co ratios of between 1.00 and 20.0 in the setting of a population of U.S. veterans (8). The objectives of the present study were (i) to determine the S/Co ratio predictive of detectable HCV RNA; (ii) to evaluate, in samples with low S/Co ratios (between 1 and 20), the relationship between S/Co ratios and the number and types of bands detected by RIBA; (iii) to assess, for samples with indeterminate RIBA results, the correlation between S/Co ratios and the type of bands detected by RIBA; and (iv) to analyze patient samples and data from patients who had very low CIA ratios (S/Co, <5) and indeterminate RIBA results for the presence of other indicators of HCV infection.
Patient samples used in this study were sent to our laboratory from the nine Department of Veterans Administration (VA) hospitals located in the Veterans Integrated Service Network I (VISN I). The overall HCV seroprevalence for the veteran population evaluated is approximately 8 to 10%. All serum samples were assessed for the presence of antibodies to HCV using the ORTHO VITROS anti-HCV assay (Ortho-Clinical Diagnostics). Only samples that were repeatedly reactive were included in this study. All samples were stored frozen for further evaluation. Supplemental testing was performed using the RIBA (Ortho-Clinical Diagnostics), the Roche Amplicor HCV test, version 2.0 (Roche Molecular Diagnostics Systems), and/or the Roche Monitor HCV test (Roche Molecular Diagnostics Systems).
The VITROS anti-HCV assay is a two-step sandwich CIA for the detection of human antibody to various proteins of HCV. Results are calculated as a normalized S/Co value. During the calibration process, a lot-specific parameter encoded in the lot validation card is used to determine a valid cutoff value. Samples with S/Co ratios of 1.0 were retested in duplicate and considered "repeatedly positive" if S/Co ratios were 1.0 for at least two of three determinations.
Two separate studies were performed: the first in 2002 and the second from May 2003 to March 2004. In the first study, a total of 482 CIA-positive samples were examined prospectively for the presence of HCV RNA. In the second study, 332 CIA low-positive samples were evaluated by RIBA. For samples with CIA S/Co ratios of <5 and indeterminate RIBAs, the patient database was searched for available HCV RNA results either for the sample in question or for samples drawn within six months of the sample in question. Nucleic acid testing was performed on available samples. In addition, available clinical records were reviewed for all patients with both indeterminate RIBAs and S/Co ratios of <5 to further clarify the disease status. The charts were reviewed for clinical signs of past or present hepatitis and laboratory evidence of hepatitis, such as elevated liver enzyme levels. Repeat samples were excluded from the study.
Our initial study focused on the determination of the S/Co ratio predictive of an HCV RNA-positive result. Results from a total of 482 samples evaluated in 2002 are included in Table 1. A majority of samples (n = 349) were tested using the HCV RNA qualitative assay alone, 101 samples were tested by the HCV RNA quantitative assay, and 32 samples were evaluated by both assays. Of the 154 samples with S/Co ratios of between 1 and 20, 130 samples were tested with the HCV RNA qualitative assay, 20 samples were tested with the HCV RNA quantitative assay, and 4 samples were evaluated with both assays. Although the number of samples with detectable HCV RNA increased in relation to the S/Co ratio, the majority of HCV RNA-positive samples had an S/Co ratio of >20 (266 of 328; 81%). In contrast, only 7 of 71 (10%) samples with S/Co ratios of between 5 and 20 were HCV RNA positive. HCV RNA was not detected in any of the samples with CIA S/Co ratios of <5. Based on these results, we have routinely categorized samples with CIA S/Co ratios of >1 and <20 as low positive and used supplemental testing for all of these samples.
In order to examine in detail the significance of low-positive samples (CIA S/Co ratio of between 1 and 20), we evaluated in more detail a total of 332 samples tested between May 2003 and March 2004. Results of RIBA testing for the 332 samples are outlined in Table 2. None of the 163 samples with S/C ratios of <5 was found to be RIBA positive. Instead, the majority (83%) of the samples with S/Co ratios of 1 to 5 were RIBA negative, and 28 samples were indeterminate. The number of positive RIBAs increased as the S/Co ratio increased, with the highest proportion of RIBA-positive samples identified in the group of samples with S/Co ratios of 16 to 20 (51 of 57; 89%). However, it should be noted that for the 129 samples with S/Co ratios of >8 and <20, 12 (9%) samples were found to be RIBA negative.
Details of the number and type of positive RIBA bands in relation to mean and range of S/Co ratios are shown in Table 3. The mean S/Co ratio for the 163 RIBA-negative samples was significantly lower than that of the 77 RIBA-indeterminate samples (3.19 versus 7.36; P < 0.1; t test), although the ranges of values in both groups were similar. For the 77 RIBA indeterminate samples, reactivity at bands C33c and C22p occurred more frequently than reactivity at bands C100 and NS5. For samples interpreted as RIBA positive, the mean S/Co ratios were significantly different between samples with two positive bands as compared to samples with three and four positive bands (t test; P < 0.01). Reactivity at bands C100, NS5, C33c, and C22p was detected in 84, 54, 96, and 84% of the samples, respectively. The presence of C100 or NS5 bands was associated with lower S/Co ratios (means of 2.68 and 2.24, respectively) than the presence of C33c or C22p, and the mean S/Co ratio for the 33 samples that only had reactivity to C22p was highest (11.04).
Among the samples with CIA S/Co ratios of <5, 28 samples were found to be RIBA indeterminate. Twenty-seven of these samples were further evaluated for the presence of HCV RNA and for clinical evidence of HCV infection by reviewing clinical records. Chart reviews indicated that most of the patients were tested as part of routine screening. Twenty-four of the 28 samples (86%) were tested for HCV RNA and were found to be negative. For three of the remaining four samples, examination of the patient's clinical records indicated no clinical evidence of hepatitis. For all of the 27 patients' records examined, there was no laboratory evidence of abnormal liver function tests and there were no other signs of HCV infection. Clinical status could not be resolved in one of 28 patients because no clinical notes were found and no serum was available for further HCV testing.
The increasingly sophisticated methods for diagnosing HCV infection have a direct impact on patient management, and careful use of the available assays is essential for the accurate and efficient diagnosis of HCV infection. Several CDC-sponsored and other anti-HCV seroprevalence studies have indicated that reactive S/Co ratios could be used to predict supplemental positive results (2, 5-7, 11); most of these studies used EIAs (2, 5, 6, 11). In the present report, the first fully automated random access diagnostic test approved by the FDA for the detection of HCV antibody, the VITROS anti-HCV assay, was utilized. The test menu for this analyzer, which initially also comprised hepatitis B surface antigen and antibody, has been recently expanded to include hepatitis B virus core total and immunoglobulin M antibody tests.
Few published studies have documented experience with the VITROS Anti-HCV assay (7-9). Ismail et al. compared the performance of the VITROS Anti HCV assay to the Abbott second-generation EIA; 318 HCV-negative samples and 177 HCV-positive samples in the Abbott assay were evaluated, but confirmatory testing was only performed on the eleven discrepant results (9). Among the seven VITROS anti-HCV-positive samples, RIBA was negative in four samples and indeterminate in three samples; all seven samples had S/Co ratios of >1 and <8, but the true status of the three indeterminate samples was not determined. Dufour et al. reported on the supplemental testing of CIA low-positive samples by RIBA for 224 samples and by HCV PCR for 55 samples (7). With the present report, we add results from a total of 486 samples with CIA S/Co ratios of <20 (154 samples in our first study, and 332 samples in our second study).
We have demonstrated in the first part of our study that the vast majority (81%) of 328 samples with S/Co ratios of >20 were HCV RNA positive. CDC recommendations have suggested that all CIA-positive samples with S/Co ratios of >8 can be reported as positive without further supplemental testing (2). In addition, Dufour et al. have reported that 95% of samples with CIA S/Co ratios of between 8 and 20 were RIBA positive or indeterminate, although HCV RNA was not detected in 25 samples tested (7). Analysis of our results for samples with CIA S/Co ratios of between 8 and 20 indicates that 43 of 129 (33%) were RIBA negative or indeterminate. If the CDC recommendation of an S/Co ratio of 8 is applied to our data set, reflex testing would not be performed for all samples with S/Co ratios of 8 to 20; this would have resulted in the report of 12 false-positive results. In order to avoid this, we have over the past several years performed supplemental testing for all positive samples with S/Co ratios of <20.
Approximately half of the 486 CIA low-positive samples evaluated in the present study were found to have S/Co ratios of <5. The fact that S/Co ratios of <5 can predict a negative result in over 99% of the samples examined in our study is of particular significance. Among samples with a CIA S/Co ratio of <5, none of the 163 samples examined by RIBA was positive and none of the 83 samples tested by PCR had detectable HCV RNA. In addition, only a minority of samples were RIBA indeterminate and none of these samples showed evidence of HCV infection. Dufour et al. found that 13 of 129 (10%) samples with S/Co ratios of <8 were RIBA positive (7). In contrast, we found that only 3% (6 of 203) of samples with an S/Co ratio of <8 were RIBA positive, and all positive samples in that range had an S/Co ratio of >5. The reason for the difference is uncertain but could be due to differences in sample size or population examined.
An indeterminate RIBA pattern, characterized by the presence of only one of the four bands, was found for 77 of our samples, with C33c and C22p band patterns detected most frequently. In a previous study on the significance of indeterminate third-generation RIBAs, a similar observation was made in 59 samples (10). Higher mean S/Co ratios were observed in samples with C33c and C22p bands as compared to samples with C100 or NS5 bands. Furthermore, eight of the nine samples with C100 or NS5 reactivity had S/Co ratios of <5. In addition, among the RIBA-positive samples with two-band reactivity, the C33c band was most frequently detected. Previous studies have made similar observations, with a high proportion of isolated reactivity to NS5 found in HCV RNA-negative subjects (4, 12) and seroconversion studies indicating that anti-C33 and anti-C22 usually are the first detectable HCV antibodies (3).
We have in the present study demonstrated that there was no evidence of HCV infection in any of the samples with S/Co ratios of 5 evaluated; no sample was found to be RIBA positive, a majority of samples were RIBA negative, and there was no laboratory or clinical evidence of HCV infection in the 27 RIBA indeterminate cases examined. Therefore, we suggest that when the VITROS assay is used, samples with S/Co ratios of 5 do not require supplemental testing because over 99% of samples with S/Co ratios of 5 are not associated with laboratory or clinical evidence of HCV infection. Considering that, in our study, 246 of 486 (50.6%) of low-positive samples had an S/Co ratio of <5, elimination of supplemental testing of samples with S/Co ratios between 1.00 and 5.00 will reduce unnecessary testing and preserve resources. This cost savings has the potential to have significant impact on VA facilities nationwide, as HCV has been identified as the most important emerging pathogen in the VA health care system (1). We have recently opted to report all our CIA low-positive samples with S/CO ratios between 1 and 5 as "borderline," with the recommendation that follow-up testing be performed when HCV infection continues to be suspected based on other clinical or laboratory information.
ACKNOWLEDGMENTS
We gratefully acknowledge the technical assistance of Maria Cavaiuolo, Meghan Condon, Kim McGowan, and Denise Miko.
REFERENCES
Bini, E. J. 2003. Hepatitis C in veterans. Curr. Hepatitis Rep. 2:108-115.
Centers for Disease Control and Prevention. 2003. Guidelines for laboratory testing and result reporting of antibody to hepatitis C virus (anti-HCV). Morb. Mortal. Wkly. Rep. 52(RR-3):1-13.
Courouce, A. M., N. Le Marrec, A. Girault, S. Ducamp, and N. Simon. 1994. Anti-hepatitis C virus seroconversion in patients undergoing hemodialysis: comparison of second- and third-generation anti-HCV assays. Transfusion 34:790-795.
Damen, M., H. L. Zaauer, H. T. M. Cuypers, V. Vrielink, C. L. Van Der Poel, H. W. Reesink, and P. N. Lelie. 1995. Reliability of the third generation recombinant immunoblot assay for hepatitis C virus. Transfusion 35:745-749.
Dos Santos, V. A., R. S. Azevedo, M. E. Camargo, and V. A. Alves. 1999. Serodiagnosis of hepatitis C virus. Effect of new evaluation of cutoff values for enzyme-linked immunosorbent assay in Brazilian patients. Am. J. Clin. Pathol. 112:418-424.
Dufour, D. R., M. Talastas, M. D. A. Fernandez, B. Harris, D. B. Strader, and L. B. Seeff. 2003. Low-positive anti-hepatitis C virus enzyme immunoassay results: an important predictor of low likelihood of hepatitis C infection. Clin. Chem. 49:479-486.
Dufour, D. R., M. Talastas, M. D. A. Fernandez, and B. Harris. 2003. Chemiluminescence assay improves specificity of hepatitis C antibody detection. Clin. Chem. 49:940-944.
Griffith, B. P., J. Falcone, D. Miko, C. K. Y. Fong, P. K. Barua, S. Campbell, and G. Stack. 2003. Hepatitis C virus antibody signal to cut-off ratios predictive of HCV RNA detection: comparison of the Ortho Vitros and EIA assays. Nineteenth Annu. Clin. Virol. Symp., abstr. M48.
Ismail, N., G. E. Fish, and M. B. Smith. 2004. Laboratory evaluation of a fully automated chemiluminescence immunoassay for rapid detection of HBsAg, antibodies to HBsAg, and antibodies to hepatitis C virus. J. Clin. Microbiol. 42:610-617.
Pawlotsky, J. M., A. Bastie, C. Pellet, J. Remire, F. Darthuy, L. Wolfe, C. Sayada, J. Duval, and D. Dhumeaux. 1996. Significance of indeterminate third-generation hepatitis C virus recombinant immunoblot assay. J. Clin. Microbiol. 34:80-83.
Tobler, L. H., G. Tegtmeier, S. L. Stramer, S. Quan, J. Dockter, G. Giachetti, and M. P. Busch. 2000. Lookback on donors who are repeatedly reactive on first-generation hepatitis C virus assays: justification and rational implementation. Transfusion 40:15-24.
Vernelen, K., H. Claeys, H. Vernhaert, A. Volclaerts, and C. Vermylen. 1994. Significance of NS3 and NS5 antigens in screening for HCV antibody. Lancet 343:853.(Margret Oethinger, Donald)