Evaluation of Two Commercially Available, Inexpens
http://www.100md.com
微生物临床杂志 2005年第2期
Department of Molecular Medicine and Haematology, School of Pathology, University of the Witwatersrand Medical School, Faculty of Health Science and the NHLS, Johannesburg, South Africa
Westat, Rockville, Maryland
ABSTRACT
Although human immunodeficiency virus type 1 (HIV-1) RNA is the acknowledged "gold standard" marker for monitoring disease activity in patients receiving highly active antiretroviral therapy (HAART), it remains unaffordable in resource-constrained settings. The present study investigated two commercially available kits for the detection of HIV-1 viral load markers as more affordable alternatives to HIV-1 RNA quantitation. The greatly improved heat-denatured, signal-boosted HiSens HIV-1 p24 Ag Ultra kit (Perkin-Elmer) and the ExaVir Load Quantitative HIV-RT kit (Cavidi Tech AB) were compared with the Amplicor HIV-1 Monitor (version 1.5) assay (Roche Molecular Systems Inc.). A total of 117 samples containing HIV-1 subtype C were analyzed by all three methodologies. Eighty-nine of these samples represented serial measurements from 20 patients receiving HAART. The remaining samples analyzed were from a group of treatment-nave patients. The association between the p24 antigen assay and the RNA assay was fairly strong (R2 = 0.686). The association between the reverse transcriptase (RT) quantitation assay and the RNA assay was strong (R2 = 0.810). Both alternative assays seemed most useful for the serial monitoring of patients receiving HAART (n = 89 plasma samples from 20 patients), as all assays showed a statistically significant downward trend over time, with the trend being either linear or curvilinear. In addition, all three assays showed negative correlations with the CD4 count (CD4 count versus RNA load, r = –0.336 and P = 0.001; CD4 count versus p24 antigen level, r = –0.541 and P < 0.0001; CD4 count versus RT level, r = –0.358 and P = 0.0006). Still of major concern are both the lack of sensitivity and the wide degrees of variability of both assays. However, both assays provide a less expensive alternative to the Roche viral load assay and demonstrate the same trends during treatment.
INTRODUCTION
The World Health Organization (WHO) and the Joint United Nations Programme on HIV/AIDS estimate that by the end of 2001 more than 40 million people had been infected with human immunodeficiency virus (HIV) type 1 (HIV-1) and that approximately 2.5 million people had died (13). More than 25 million people in sub-Saharan Africa are already infected with HIV, with the prevalence of HIV infection among adults estimated to be 8.8% (23). South Africa has been severely affected by a generalized HIV epidemic, with the prevalence among antenatal clinic attendees having risen from less than 1% in 1990 to 24.8% in 2001 (6). The first national survey for the prevalence of HIV in systematically sampled households in 2002 estimated that the prevalence of HIV infection in the general South African population is 11.4% overall and 15.6% among individuals 15 to 49 years old (11).
According to WHO, less than 5% of individuals requiring antiretroviral therapy can access these medicines in resource-poor settings (23). Obstacles to treatment implementation in these regions include the lack of trained personnel, the cost of antiretroviral therapy, the drug distribution mechanisms available, and laboratory monitoring costs.
The focus of this paper is the evaluation of two alternative assays for the quantification of HIV, which has found wide applications in assessing the progression of the disease and monitoring the efficacy of antiretroviral therapy (16). Monitoring of plasma for the viral load is an integral part of the standard of care for HIV-infected patients in first-world settings, and a great deal of debate has ensued locally and internationally around the value of these assays for treatment initiation and monitoring in resource-poor settings. In fact, these assays are classified as optional under the draft WHO guidelines for the implementation of antiretroviral therapy (23). Various approaches have been used for the quantification of HIV. The high cost of nucleic acid-based testing strategies, together with the significant technical demands required for their delivery, has precluded the use of these assays in many resource-poor environments. Two potential candidate assays, a heat-denatured (HD), signal-boosted p24 antigen assay (the HiSens HIV-1 p24 Ag Ultra assay) and a reverse transcriptase (RT) enzyme activity assay (the ExaVir Load assay), have recently received considerable attention as low-cost alternatives to nucleic acid-based testing strategies.
These assays have undergone significant improvements in recent years. The p24 antigen quantitation assay now includes heat dissociation of the interfering antigen-antibody immune complexes and subsequent signal amplification with biotyl-tyramide and streptavidin-horseradish peroxidase (S-HRP) to improve the sensitivity. The linear range of this assay has improved substantially with the introduction of a quantitative kinetic enzyme-linked immunosorbent assay (ELISA) reader and software (2, 14, 17, 18, 19, 20, 21). The ExaVir Load assay measures the activity of the virus-encoded enzyme RT, which is packaged together with the viral RNA in the HIV particle. This method was previously compromised due to the susceptibility of the enzyme to inhibitory antibodies and other interfering molecules (5). In the newer version of the assay, virions are first separated from plasma by passage through a gel column, which removes any potential interfering substances. The purification of the RT enzyme is followed by the measurement of the DNA and RNA hybrid produced as a result of the RT activity (1, 3, 8).
Since a paucity of data on the performance of these assays with non-subtype B HIV subtypes are available, an evaluation was conducted in South Africa, where subtype C is the predominant subtype. We compared the performances of both assays to that of the present in-country "gold standard" assay, the Amplicor HIV-1 Monitor (version 1.5) assay (Roche Molecular Systems Inc., Branchburg, N.J.).
MATERIALS AND METHODS
Sample population. A total of 117 samples were collected from HIV-positive individuals monitored at the Johannesburg Hospital outpatient HIV clinic. All patients were from Johannesburg, South Africa, and surrounding areas and were all receiving triple-combination antiretroviral therapy. Studies were conducted in full conformance with local ethics committee approval. Clinical specimens were fresh and/or frozen plasma collected in Vacutainer tubes containing EDTA. The HIV RNA load was determined by the Amplicor HIV-1 Monitor (version 1.5) assay (Roche Molecular Systems Inc.) according to instructions of the manufacturer. For patients who presented with viral loads below the lower quantitation limit of the assay (<400 RNA copies/ml), the ultrasensitive version of the assay was performed. That assay has a detection limit of 50 RNA copies/ml. In addition, a residual sample stored at –70°C was evaluated by both the heat-denatured p24 ultrasensitive assay and the ExaVir Load assay.
Eighty-nine of the available plasma samples, which represented samples from 20 patients, were selected due to the availability of serial samples while the patients were being monitored on a triple-combination antiretroviral therapy regimen. The majority of patients returned for at least four sequential visits, with a number of patients continuing on to the sixth visit. Blood samples for viral load determinations were scheduled at the baseline; at weeks 4, 8, 12, 16, and 24; and then every 8 weeks thereafter. The numbers of samples analyzed at each of the six visits were 20, 20, 20, 15, 9, and 5, respectively.
HD, boosted p24 antigen assay. We used the HiSens HIV-1 p24 Ag Ultra kit from Perkin-Elmer Life and Analytical Sciences, Turku, Finland. Briefly, 50 μl of plasma was initially lysed in 25 μl of lysis buffer before it was diluted 1:6 in 0.5% Triton X-100 (21). The lysis buffer used to conduct the assay was a new improved version received from Jorg Schupbach. The diluted sample was heat denatured for 5 min at 100°C. Treated plasma samples (250 μl) were then transferred to the wells of a microtiter plate coated with monoclonal antibody to HIV-1 p24, and the plate was incubated at room temperature for 2 h on a microtiter plate shaker. The wells were washed with 1x wash buffer (supplied with the kit), to which 100 μl of biotinylated detector antibody was then added, and the mixture was incubated at 37°C for 1 h. After the mixture was washed, 100 μl of diluted S-HRP was added to all wells and the plate was incubated at 37°C for 15 min. After the plate was washed, 100 μl of biotinyl-tyramide working solution was added to each well and the plate was incubated for 15 min at room temperature to allow signal amplification. After the plate was washed, diluted S-HRP was added and the plate was incubated for a further 15 min at room temperature. Finally, o-phenylenediamine substrate was added to all wells after the plate was washed and the plate was inserted into a kinetic ELISA plate reader. Quanti-Kin detection system software (DL3; Diagnostica Ligure, Genoa, Italy) was used to perform kinetic readings over a 30-min period. A further 10-min incubation at room temperature in the reader was stopped by the addition of 100 μl of stop solution and an end-point reading was taken.
The run was considered valid when the substrate blank had an optical density (OD) 0.05, the negative control had an OD 0.10, and the 6,103-fg/ml standard had an OD 0.15. The Quanti-Kin system incorporates programs for data reduction, and the results were obtained as printouts of calibration curves and concentrations for unknown samples.
ExaVir Load Quantitative HIV-RT kit. HIV RT activity was quantified with the ExaVir Load Quantitative HIV-RT kit (Cavidi Tech AB, Uppsala, Sweden), according to the instructions of the manufacturer. Initially, 1 ml of plasma was treated to inactivate interfering cellular enzymes. The viral particles were then captured and immobilized on a separation gel column and washed to remove other potential interfering factors. Isolated virions were lysed with the kit lysis buffer to recover the intraviral RT.
A reaction mixture containing oligo(dT) primer, bromo-dUTP, and the lysates were added to a 96-well poly(A)-coated plate. Two different amounts of lysates, 75 and 15 μl, were used in order to increase the detection range of the assay. Recombinant HIV-1 RT supplied with the kit was used as a standard for quantitation. A serially diluted standard of a known concentration was also added to the plate in 75- and 15-μl volumes. In the presence of RT, DNA and RNA hybrids could be synthesized by RT during an overnight incubation at room temperature. After the reaction mixture was washed, the DNA-RNA hybrid product was detected with -bromo-dUTP antibodies conjugated with an alkaline phosphatase (AP). The product was quantified by addition of a colorimetric AP substrate. Colorimetric readings were taken on a Bio-Tek Elx800 microtiter ELISA plate reader. The first reading was done immediately after the addition of the AP substrate to establish the baseline signal, and the reading was then repeated 2 h later. A final reading was taken after an overnight incubation. Viral RT activity was expressed as femtograms per milliliter. These dUTP values were then converted to an equivalent number of copies per milliliter by the software provided (15).
Plasma HIV RNA assay. The Amplicor HIV-1 Monitor (version 1.5) assay (Roche Molecular Systems Inc.) was performed with 200 μl (standard) or 500 μl (ultrasensitive) of plasma and was used as the gold standard assay. The assay was performed according to the instructions of the manufacturer.
Statistical analysis. All statistical analyses were performed with log10 transformed values for all the variables except the CD4 counts in the analyses, i.e., log10 p24 antigen level, log10 RT concentration (log10 femtograms per milliliter), and log10 RNA load (log10 copies per milliliter; Amplicor). Schematic (box) plots of the log10 p24 antigen levels, log10 RT concentrations, and log10 RNA loads over the patient visits were constructed to evaluate the changes in these values over time for all patients receiving treatment. Tests for the trends of these values over visits were conducted by using a mixed-effects (repeated-measures) model (7) with orthogonal polynomial contrasts (22) for the number of visits in this study included in the model.
To assess the association between log10 p24 antigen levels and log10 RNA loads and between the log10 RT concentrations and the log10 RNA loads, we used a mixed-effects (repeated-measures) model for each association with a heterogeneous first-order autoregressive-correlation structure for the values for an individual patient. This allows lower correlations between values that are farther apart than those that are adjacent for an individual patient, as well as adjustment for the heterogeneity of the outcome values for an individual patient. The coefficient of determination (R2) was calculated for each model as the square of the correlation between the observed and the predicted values. The Spearman test was applied to show a negative correlation between the CD4 count and the results of all three assays. All statistical tests were two sided, and all statistical analyses were performed with SAS (version 8.2) software.
RESULTS
Four negative human plasma samples of different lots supplied in the Roche Amplicor HIV-1 Monitor (version 1.5) assay kits for RNA count determination repeatedly tested negative by all assay methodologies used in this evaluation. RT enzyme activity and p24 antigen measurements were determined from all 117 frozen plasma samples once the RNA copy number determinations were available by the Amplicor HIV-1 Monitor assay.
Description of data analyzed by all three assays with 117 plasma samples. According to the RNA assay, the viral load results ranged from 1.7 log10 (50 copies/ml) to 5.9 log10 (750,000 copies/ml), which are within the linear range of the assay. Figure 1a illustrates the distribution of the data from this data set within these log10 intervals and demonstrates that 40.5% of the samples tested in this study had viral load results less than 2.8 log10 (630 copies/ml), showing a skewing of the results toward the groups with lower log10 values.
The box plot in Fig. 1b illustrates the range of CD4 counts for 89 samples from 20 patients over six visits. A line drawn through the median shows a median increase from a CD4 count of 186 cells/μl at visit 1 to 304 cells/μl at visit 6.
Statistical analysis. The scatter plots of the log10 p24 antigen levels versus the log10 RNA load and of the log10 RT level versus the log10 RNA load for the 117 plasma samples are given in Fig. 2a and b. In the scatterplot of the log10 p24 antigen concentration versus the log10 RNA load in Fig. 2a, approximately 39% of the results for the RNA load, which covered a range of 2.5 to 5 log10 (316 to 100,000 RNA copies/ml), are represented by no change in p24 antigen levels between visits. In the scatterplot of the log10 RT level versus the log10 RNA load in Fig. 2b, the lowest RNA copy number detected by the RT assay is 3.5 log10; thus, the change in the RNA load between 50 and 3,000 copies/ml is not reflected by a change in the RT level. This was represented by approximately 52% of the results. In addition to this, the plot illustrates the ability of the RT assay to measure viral load beyond the upper limit of the assay for the RNA load (>750 000 copies/ml). These limitations of both assays compared to the viral RNA load center largely on sensitivity in the low viral load ranges. The plots show heteroscedasticity (increasing variance with increasing values of log10 RNA load), which was adjusted for in the statistical analysis.
Analysis of patient viral loads during antiretroviral therapy by all three monitoring assays. The mixed-effect model, performed with data for 89 samples from 20 patients, for the log10 p24 antigen level (the regressor) and the log10 RNA load (the outcome variable) showed a significant association between the two (with adjustment for heterogeneity, replication, and the effect of time). The R2 value for this model equals 0.686; thus, 68.6% of the variation in the log10 RNA load was explained by the model. Also, the slope parameter for the log10 p24 antigen level equals 1.0386, which is very close to 1, thus indicating a fair amount of agreement between the two assays.
Similarly, the mixed-effect model for the log10 RT level (the regressor) and the RNA load (the outcome variable) showed a significant association between the two (with adjustment for heterogeneity, replication, and the effect of time). The R2 value for this model equals 0.8103; thus, 81% of the variation in the log10 RNA load was explained by the model. Also the slope parameter for the log10 RT load equals 0.8663, which is also close to 1 but which indicates less agreement between the two measures.
The box plots in Fig. 3a and b of the log10 p24 antigen level and the log10 RT level versus the log10 RNA load show the range of assay measurements for 20 patients monitored over six visits. In 19 of the 20 patients, the serial p24 antigen and RT levels completely paralleled the RNA loads. After therapy initiation the concentrations of p24, RT, and RNA exhibited distinct decreases in all except two patients, in whom p24 antigen and RT remained undetectable. None of the patients tested showed a significant viral rebound or a lack of response, and thus, the performances of these assays in these scenarios could not be assessed. The repeated-measures analyses with orthogonal polynomial contrasts showed a significant linear decline in log10 p24 antigen level (P < 0.0001) and a curvilinear decline (quadratic trend) in the log10 RT level (P < 0.0001) and the log10 RNA load (P < 0.0025).
The results of all three assays were inversely correlated with CD4 T-cell counts, as shown by the Spearman test (CD4 count versus RNA load, r = –0.336 and P = 0.001; CD4 count versus p24 antigen level, r = –0.541 and P < 0.0001; CD4 count versus RT level, r = –0.358 and P = 0.0006).
DISCUSSION
The costs of antiretroviral drugs have decreased substantially in recent times in resource-constrained environments (4). However, a factor limiting the implementation of antiretroviral therapy in many developing countries still focuses on the challenges of implementation of cost-effective laboratory monitoring (9). The obstacles to implementation of laboratory monitoring in resource-constrained settings are often geographically dependent and may include one or more of the following: (i) the lack of an adequate laboratory infrastructure, (ii) the absence of technical skill and/or laboratory management skills, and (iii) the costs of reagents and reagent suppliers support in remote regions.
On the basis of the findings presented above, the p24 antigen and RT enzyme assays may provide less expensive, simpler alternatives for monitoring the response to therapy in the context of infection with HIV-1 clade C. The reduced sensitivities of these assays compared to that of nucleic acid testing revealed in this data set need further investigation (16). The specificities of both of these assays were not addressed in this study and will require additional exploration in future evaluations.
Although the scatterplots in Fig. 2 showed heteroscedasticity (increasing variance with increasing log10 RNA loads), the mixed-effects model showed a significant association between the p24 antigen level, the RT concentration, and the RNA load. This variability may be influenced by the fact that all three assays measure completely different parameters of viral replication. The RT assay measures the activity of the virus-encoded enzyme RT, the HD HiSens HIV-1 p24 Ag Ultra assay measures the level of either virion-bound p24 antigen or relatively freely circulating p24 antigen in immune complexes (18), and the Roche RNA assay quantitates the virion-associated RNA in blood plasma. The more valuable application of the alternative assays is their interpretation for monitoring of patients on antiretroviral therapy over time, as both assays under evaluation showed significant declines in viral loads over time (Fig. 3). There are, however, still concerns regarding the lack of sensitivity below a certain viral RNA load and how this will influence accurate monitoring of patients on antiretroviral therapy (Fig. 2). In addition, all three assays showed the expected negative correlation with the CD4 count (10), a conventional measure of a patient's response to therapy (12, 21).
A major consideration when evaluating these alternative assays is to consider not only the cost but also the complexity of conducting the investigations. In this study we paid careful attention to the technical aspects of the proposed viral load determination methodologies. The RNA assay can be automated to a large degree, which allows easier implementation and standardization. The assay still, however, requires sophisticated, costly equipment and appropriate laboratory design to reduce the chance of cross contamination commonly seen with PCR technology. Both alternative assays can be implemented with basic ELISA laboratory equipment, such as a heating block, pipettes, a plate reader, and a plate washer. The disadvantages that remain for the RT assay include (i) the time taken to complete the assay (2 to 3 days); (ii) the large sample volume required for analysis, which may pose problems in the pediatric treatment environment; and (iii) the lack of automation, which may significantly increase person-to-person variations as well as interassay and intra-assay variations. The p24 antigen assay has the advantage of facilitating a higher throughput than either the RT or the RNA quantitation assays. The disadvantages of the p24 antigen assay include (i) the variability of the p24 antigen results for a given viral RNA load (Fig. 2a) and the fact that the lysis buffer used to improve the sensitivity of the assay is not commercially available at present.
The full economic cost of each assay is required before implementation in any laboratory environment and should include the costs of (i) reagents and consumables, (ii) labor, (iii) equipment and maintenance, and (iv) transport of samples and fixed overheads. These costs will vary substantially in each geographical region, and informed decision making thus needs to be facilitated to ensure that the appropriate technology is chosen on the basis of the cost, the specimen volume required, and the skill set available.
On the basis of the data generated in this study, these assays represent viable alternatives for further exploration for use in certain resource-constrained environments.
ACKNOWLEDGMENTS
This research was supported by a grant from the NIH CIPRA Project Program and The National Health Laboratory Services.
Ethics approval was obtained through the University of the Witwatersrand (number M00-01-07).
REFERENCES
Awad, R. J., G. E. Corrigan, D. H. Ekstrand, R. Thorstensson, C. Kallander, and J. S. Gronowitz. 1997. Measurement of levels of human immunodeficiency virus type 1 reverse transcriptase (RT) and RT-activity-blocking antibody in humans by a new standardized colorimetric assay. J. Clin. Microbiol. 35:1080-1089.
Boni, J., M. Opravil, Z. Tomasik, M. Rothen, K. Bisset, P. Grob, R. Lüthy, and J. Schüpbach. 1997. Simple monitoring of antiretroviral therapy with a signal-amplification-boosted HIV p24 antigen assay with heat-denatured plasma. AIDS 11:F47-F52.
Braun, J., J. C. Plantier, M. F. Hellot, E. Tuaillon, M. Gueudin, F. Damond, A. Malmsten, G. E. Corrigan, and F. Simon. 2003. A new quantitative HIV viral load assay based on plasma virion reverse transcriptase activity for the different types, groups and subtypes. AIDS 17:331-336.
Brown, P. 2000. Cheaper AIDS drugs due for Third World. Nature 405:263.
Burgisser, P., P. Vernazza, M. Flepp, J. Boni, Z. Tomasik, U. Hummel, G. Pantaleo, and J. Schupbach. 2000. Performance of five different assays for the quantification of viral load in persons infected with various subtypes of HIV-1. Swiss HIV cohort study. J. Acquir. Immune Defic. Syndr. 23:138-144.
Department of Health. 2000. National HIV sero-prevalence survey of women attending public antenatal clinics in South Africa. Directorate, Health Systems Research and Epidemiology, Department of Health, Pretoria, South Africa.
Diggle, P. J., K.-Y. Liang, and S. L. Zeger. 1994. Analysis of longitudinal data. Clarendon Press, Oxford, United Kingdom.
Garcia, J. G., S. Yamamoto, M. Gomez-Cano, V. Soriano, T. A. Green, M. P. Busch, T. M. Folks, and W. Heneine. 1998. Measurement of human immunodeficiency virus type-1 plasma virus load based on reverse transcriptase (RT) activity: evidence of variability of virion-associated RT. J. Infect. Dis. 177:1221-1229.
Glencross, D. K., B. V. Mendelow, and W. S. Stevens. 2003. Laboratory monitoring of HIV/AIDS in a resource-poor setting. S. Afr. Med. J. 93:262-263.
Holland, C. A., J. H. Ellenberg, C. M. Wilson, S. D. Douglas, D. C. Futterman, L. A. Kingsley, A. B. Moscicki, et al. 2000. Relationship of CD4+ T cell counts and HIV type 1 viral loads in untreated, infected adolescents. AIDS Res. Hum. Retrovir. 16:959-963.
Human Sciences Research Council of South Africa. 2002. The Nelson Mandela/SHRC study of HIV/AIDS, South African national HIV prevalence, behavioral risks and mass media, household survey 2002. Human Sciences Research Council of South Africa, Cape Town, South Africa.
Johnson, D. 2000. Principles of treating HIV. S. Afr. Med. J. 90:227-230.
Joint United Nations Programme on HIV/AIDS. 2002. UNAIDS report on the global HIV/AIDS epidemic. Joint United Nations Programme on HIV/AIDS, Geneva, Switzerland.
Ledergerber, B., M. Flepp, J. Boni, Z. Tomasik, R. W. Cone, R. Lüthy, and J. Schüpbach. 2000. Human immunodeficiency virus type 1 p24 concentration measured by boosted ELISA of heat-denatured plasma correlates with decline in CD4 cells, progression to AIDS, and survival: comparison with viral RNA measurement. J. Infect. Dis. 181:1280-1288.
Malmsten, A., X. W. Shao, K. Aperia, G. E. Corrigan, E. Sandstrom, C. F. Kallander, T. Leitner, and J. S. Gronowitz. 2003. HIV-1 viral load determination based on reverse transcriptase activity from human plasma. J. Med. Virol. 71:347-359.
Mellors, J. W., C. W. Rinaldo, P. Gupta, R. M. White, J. A. Todd, and L. A. Kingsley. 1996. Prognosis of HIV infection predicted by the quantity of virus in plasma. Science 272:1167-1170.
Nadal, D., J. Boni, C. Kind, O. E. Varnier, F. Steiner, Z. Tomasik, and J. Schupbach. 1999. Prospective evaluation of amplification boosted ELISA for heat-denatured p24 antigen for the diagnosis and monitoring of pediatric human immunodeficiency virus type I infection. J. Infect. Dis. 180:1089-1095.
Schupbach, J., J. Boni, Z. Tomasik, J. Jendis, R. Seger, C. Kind, et al. 1994. Sensitive detection and early prognostic significance of p24 antigen in heat-denatured plasma of human immunodeficiency virus type 1 infected infants. J. Infect. Dis. 170:318-324.
Schüpbach, J., and J. Boni. 1993. Quantitative and sensitive detection of immune-complexed and free HIV antigen after boiling of serum. J. Virol. Methods 43:247-256. (Erratum, 45:245.)
Schüpbach, J., M. Flepp, D. Pontelli, Z. Tomasik, R. Lüthy, and J. Boni. 1996. Heat mediated immune complex dissociation and enzyme linked immunosorbent assay signal amplification render p24 antigen detection in plasma as sensitive as HIV-1 RNA detection by polymerase chain reaction. AIDS 10:1085-1090.
Schupbach, J., J. Boni, L. R. Bisset, Z. Tomasik, M. Fischer, H. F. Gunthard, B. Ledergerber, M. Opravil, and the Swiss HIV Cohort Study. 2003. HIV-1 p24 antigen is a significant inverse correlate of CD4 T-cell change in patients with suppressed viremia under long-term antiretroviral therapy. J. Acquir. Immune Defic. Syndr. 33:292-299.
Snedecor, G. W., and W. G. Cochran. 1989. Statistical methods, 8th ed. Iowa State University Press, Ames.
World Health Organization. 2002. Scaling up antiretroviral therapy in resource limited settings. WHO guidelines. WHO document, April 2002. http://www.who.int/entity/hiv/topics/arv/en/scaling_exe_summary.pdf. World Health Organization, Geneva, Switzerland.(G. Stevens, N. Rekhviashv)
Westat, Rockville, Maryland
ABSTRACT
Although human immunodeficiency virus type 1 (HIV-1) RNA is the acknowledged "gold standard" marker for monitoring disease activity in patients receiving highly active antiretroviral therapy (HAART), it remains unaffordable in resource-constrained settings. The present study investigated two commercially available kits for the detection of HIV-1 viral load markers as more affordable alternatives to HIV-1 RNA quantitation. The greatly improved heat-denatured, signal-boosted HiSens HIV-1 p24 Ag Ultra kit (Perkin-Elmer) and the ExaVir Load Quantitative HIV-RT kit (Cavidi Tech AB) were compared with the Amplicor HIV-1 Monitor (version 1.5) assay (Roche Molecular Systems Inc.). A total of 117 samples containing HIV-1 subtype C were analyzed by all three methodologies. Eighty-nine of these samples represented serial measurements from 20 patients receiving HAART. The remaining samples analyzed were from a group of treatment-nave patients. The association between the p24 antigen assay and the RNA assay was fairly strong (R2 = 0.686). The association between the reverse transcriptase (RT) quantitation assay and the RNA assay was strong (R2 = 0.810). Both alternative assays seemed most useful for the serial monitoring of patients receiving HAART (n = 89 plasma samples from 20 patients), as all assays showed a statistically significant downward trend over time, with the trend being either linear or curvilinear. In addition, all three assays showed negative correlations with the CD4 count (CD4 count versus RNA load, r = –0.336 and P = 0.001; CD4 count versus p24 antigen level, r = –0.541 and P < 0.0001; CD4 count versus RT level, r = –0.358 and P = 0.0006). Still of major concern are both the lack of sensitivity and the wide degrees of variability of both assays. However, both assays provide a less expensive alternative to the Roche viral load assay and demonstrate the same trends during treatment.
INTRODUCTION
The World Health Organization (WHO) and the Joint United Nations Programme on HIV/AIDS estimate that by the end of 2001 more than 40 million people had been infected with human immunodeficiency virus (HIV) type 1 (HIV-1) and that approximately 2.5 million people had died (13). More than 25 million people in sub-Saharan Africa are already infected with HIV, with the prevalence of HIV infection among adults estimated to be 8.8% (23). South Africa has been severely affected by a generalized HIV epidemic, with the prevalence among antenatal clinic attendees having risen from less than 1% in 1990 to 24.8% in 2001 (6). The first national survey for the prevalence of HIV in systematically sampled households in 2002 estimated that the prevalence of HIV infection in the general South African population is 11.4% overall and 15.6% among individuals 15 to 49 years old (11).
According to WHO, less than 5% of individuals requiring antiretroviral therapy can access these medicines in resource-poor settings (23). Obstacles to treatment implementation in these regions include the lack of trained personnel, the cost of antiretroviral therapy, the drug distribution mechanisms available, and laboratory monitoring costs.
The focus of this paper is the evaluation of two alternative assays for the quantification of HIV, which has found wide applications in assessing the progression of the disease and monitoring the efficacy of antiretroviral therapy (16). Monitoring of plasma for the viral load is an integral part of the standard of care for HIV-infected patients in first-world settings, and a great deal of debate has ensued locally and internationally around the value of these assays for treatment initiation and monitoring in resource-poor settings. In fact, these assays are classified as optional under the draft WHO guidelines for the implementation of antiretroviral therapy (23). Various approaches have been used for the quantification of HIV. The high cost of nucleic acid-based testing strategies, together with the significant technical demands required for their delivery, has precluded the use of these assays in many resource-poor environments. Two potential candidate assays, a heat-denatured (HD), signal-boosted p24 antigen assay (the HiSens HIV-1 p24 Ag Ultra assay) and a reverse transcriptase (RT) enzyme activity assay (the ExaVir Load assay), have recently received considerable attention as low-cost alternatives to nucleic acid-based testing strategies.
These assays have undergone significant improvements in recent years. The p24 antigen quantitation assay now includes heat dissociation of the interfering antigen-antibody immune complexes and subsequent signal amplification with biotyl-tyramide and streptavidin-horseradish peroxidase (S-HRP) to improve the sensitivity. The linear range of this assay has improved substantially with the introduction of a quantitative kinetic enzyme-linked immunosorbent assay (ELISA) reader and software (2, 14, 17, 18, 19, 20, 21). The ExaVir Load assay measures the activity of the virus-encoded enzyme RT, which is packaged together with the viral RNA in the HIV particle. This method was previously compromised due to the susceptibility of the enzyme to inhibitory antibodies and other interfering molecules (5). In the newer version of the assay, virions are first separated from plasma by passage through a gel column, which removes any potential interfering substances. The purification of the RT enzyme is followed by the measurement of the DNA and RNA hybrid produced as a result of the RT activity (1, 3, 8).
Since a paucity of data on the performance of these assays with non-subtype B HIV subtypes are available, an evaluation was conducted in South Africa, where subtype C is the predominant subtype. We compared the performances of both assays to that of the present in-country "gold standard" assay, the Amplicor HIV-1 Monitor (version 1.5) assay (Roche Molecular Systems Inc., Branchburg, N.J.).
MATERIALS AND METHODS
Sample population. A total of 117 samples were collected from HIV-positive individuals monitored at the Johannesburg Hospital outpatient HIV clinic. All patients were from Johannesburg, South Africa, and surrounding areas and were all receiving triple-combination antiretroviral therapy. Studies were conducted in full conformance with local ethics committee approval. Clinical specimens were fresh and/or frozen plasma collected in Vacutainer tubes containing EDTA. The HIV RNA load was determined by the Amplicor HIV-1 Monitor (version 1.5) assay (Roche Molecular Systems Inc.) according to instructions of the manufacturer. For patients who presented with viral loads below the lower quantitation limit of the assay (<400 RNA copies/ml), the ultrasensitive version of the assay was performed. That assay has a detection limit of 50 RNA copies/ml. In addition, a residual sample stored at –70°C was evaluated by both the heat-denatured p24 ultrasensitive assay and the ExaVir Load assay.
Eighty-nine of the available plasma samples, which represented samples from 20 patients, were selected due to the availability of serial samples while the patients were being monitored on a triple-combination antiretroviral therapy regimen. The majority of patients returned for at least four sequential visits, with a number of patients continuing on to the sixth visit. Blood samples for viral load determinations were scheduled at the baseline; at weeks 4, 8, 12, 16, and 24; and then every 8 weeks thereafter. The numbers of samples analyzed at each of the six visits were 20, 20, 20, 15, 9, and 5, respectively.
HD, boosted p24 antigen assay. We used the HiSens HIV-1 p24 Ag Ultra kit from Perkin-Elmer Life and Analytical Sciences, Turku, Finland. Briefly, 50 μl of plasma was initially lysed in 25 μl of lysis buffer before it was diluted 1:6 in 0.5% Triton X-100 (21). The lysis buffer used to conduct the assay was a new improved version received from Jorg Schupbach. The diluted sample was heat denatured for 5 min at 100°C. Treated plasma samples (250 μl) were then transferred to the wells of a microtiter plate coated with monoclonal antibody to HIV-1 p24, and the plate was incubated at room temperature for 2 h on a microtiter plate shaker. The wells were washed with 1x wash buffer (supplied with the kit), to which 100 μl of biotinylated detector antibody was then added, and the mixture was incubated at 37°C for 1 h. After the mixture was washed, 100 μl of diluted S-HRP was added to all wells and the plate was incubated at 37°C for 15 min. After the plate was washed, 100 μl of biotinyl-tyramide working solution was added to each well and the plate was incubated for 15 min at room temperature to allow signal amplification. After the plate was washed, diluted S-HRP was added and the plate was incubated for a further 15 min at room temperature. Finally, o-phenylenediamine substrate was added to all wells after the plate was washed and the plate was inserted into a kinetic ELISA plate reader. Quanti-Kin detection system software (DL3; Diagnostica Ligure, Genoa, Italy) was used to perform kinetic readings over a 30-min period. A further 10-min incubation at room temperature in the reader was stopped by the addition of 100 μl of stop solution and an end-point reading was taken.
The run was considered valid when the substrate blank had an optical density (OD) 0.05, the negative control had an OD 0.10, and the 6,103-fg/ml standard had an OD 0.15. The Quanti-Kin system incorporates programs for data reduction, and the results were obtained as printouts of calibration curves and concentrations for unknown samples.
ExaVir Load Quantitative HIV-RT kit. HIV RT activity was quantified with the ExaVir Load Quantitative HIV-RT kit (Cavidi Tech AB, Uppsala, Sweden), according to the instructions of the manufacturer. Initially, 1 ml of plasma was treated to inactivate interfering cellular enzymes. The viral particles were then captured and immobilized on a separation gel column and washed to remove other potential interfering factors. Isolated virions were lysed with the kit lysis buffer to recover the intraviral RT.
A reaction mixture containing oligo(dT) primer, bromo-dUTP, and the lysates were added to a 96-well poly(A)-coated plate. Two different amounts of lysates, 75 and 15 μl, were used in order to increase the detection range of the assay. Recombinant HIV-1 RT supplied with the kit was used as a standard for quantitation. A serially diluted standard of a known concentration was also added to the plate in 75- and 15-μl volumes. In the presence of RT, DNA and RNA hybrids could be synthesized by RT during an overnight incubation at room temperature. After the reaction mixture was washed, the DNA-RNA hybrid product was detected with -bromo-dUTP antibodies conjugated with an alkaline phosphatase (AP). The product was quantified by addition of a colorimetric AP substrate. Colorimetric readings were taken on a Bio-Tek Elx800 microtiter ELISA plate reader. The first reading was done immediately after the addition of the AP substrate to establish the baseline signal, and the reading was then repeated 2 h later. A final reading was taken after an overnight incubation. Viral RT activity was expressed as femtograms per milliliter. These dUTP values were then converted to an equivalent number of copies per milliliter by the software provided (15).
Plasma HIV RNA assay. The Amplicor HIV-1 Monitor (version 1.5) assay (Roche Molecular Systems Inc.) was performed with 200 μl (standard) or 500 μl (ultrasensitive) of plasma and was used as the gold standard assay. The assay was performed according to the instructions of the manufacturer.
Statistical analysis. All statistical analyses were performed with log10 transformed values for all the variables except the CD4 counts in the analyses, i.e., log10 p24 antigen level, log10 RT concentration (log10 femtograms per milliliter), and log10 RNA load (log10 copies per milliliter; Amplicor). Schematic (box) plots of the log10 p24 antigen levels, log10 RT concentrations, and log10 RNA loads over the patient visits were constructed to evaluate the changes in these values over time for all patients receiving treatment. Tests for the trends of these values over visits were conducted by using a mixed-effects (repeated-measures) model (7) with orthogonal polynomial contrasts (22) for the number of visits in this study included in the model.
To assess the association between log10 p24 antigen levels and log10 RNA loads and between the log10 RT concentrations and the log10 RNA loads, we used a mixed-effects (repeated-measures) model for each association with a heterogeneous first-order autoregressive-correlation structure for the values for an individual patient. This allows lower correlations between values that are farther apart than those that are adjacent for an individual patient, as well as adjustment for the heterogeneity of the outcome values for an individual patient. The coefficient of determination (R2) was calculated for each model as the square of the correlation between the observed and the predicted values. The Spearman test was applied to show a negative correlation between the CD4 count and the results of all three assays. All statistical tests were two sided, and all statistical analyses were performed with SAS (version 8.2) software.
RESULTS
Four negative human plasma samples of different lots supplied in the Roche Amplicor HIV-1 Monitor (version 1.5) assay kits for RNA count determination repeatedly tested negative by all assay methodologies used in this evaluation. RT enzyme activity and p24 antigen measurements were determined from all 117 frozen plasma samples once the RNA copy number determinations were available by the Amplicor HIV-1 Monitor assay.
Description of data analyzed by all three assays with 117 plasma samples. According to the RNA assay, the viral load results ranged from 1.7 log10 (50 copies/ml) to 5.9 log10 (750,000 copies/ml), which are within the linear range of the assay. Figure 1a illustrates the distribution of the data from this data set within these log10 intervals and demonstrates that 40.5% of the samples tested in this study had viral load results less than 2.8 log10 (630 copies/ml), showing a skewing of the results toward the groups with lower log10 values.
The box plot in Fig. 1b illustrates the range of CD4 counts for 89 samples from 20 patients over six visits. A line drawn through the median shows a median increase from a CD4 count of 186 cells/μl at visit 1 to 304 cells/μl at visit 6.
Statistical analysis. The scatter plots of the log10 p24 antigen levels versus the log10 RNA load and of the log10 RT level versus the log10 RNA load for the 117 plasma samples are given in Fig. 2a and b. In the scatterplot of the log10 p24 antigen concentration versus the log10 RNA load in Fig. 2a, approximately 39% of the results for the RNA load, which covered a range of 2.5 to 5 log10 (316 to 100,000 RNA copies/ml), are represented by no change in p24 antigen levels between visits. In the scatterplot of the log10 RT level versus the log10 RNA load in Fig. 2b, the lowest RNA copy number detected by the RT assay is 3.5 log10; thus, the change in the RNA load between 50 and 3,000 copies/ml is not reflected by a change in the RT level. This was represented by approximately 52% of the results. In addition to this, the plot illustrates the ability of the RT assay to measure viral load beyond the upper limit of the assay for the RNA load (>750 000 copies/ml). These limitations of both assays compared to the viral RNA load center largely on sensitivity in the low viral load ranges. The plots show heteroscedasticity (increasing variance with increasing values of log10 RNA load), which was adjusted for in the statistical analysis.
Analysis of patient viral loads during antiretroviral therapy by all three monitoring assays. The mixed-effect model, performed with data for 89 samples from 20 patients, for the log10 p24 antigen level (the regressor) and the log10 RNA load (the outcome variable) showed a significant association between the two (with adjustment for heterogeneity, replication, and the effect of time). The R2 value for this model equals 0.686; thus, 68.6% of the variation in the log10 RNA load was explained by the model. Also, the slope parameter for the log10 p24 antigen level equals 1.0386, which is very close to 1, thus indicating a fair amount of agreement between the two assays.
Similarly, the mixed-effect model for the log10 RT level (the regressor) and the RNA load (the outcome variable) showed a significant association between the two (with adjustment for heterogeneity, replication, and the effect of time). The R2 value for this model equals 0.8103; thus, 81% of the variation in the log10 RNA load was explained by the model. Also the slope parameter for the log10 RT load equals 0.8663, which is also close to 1 but which indicates less agreement between the two measures.
The box plots in Fig. 3a and b of the log10 p24 antigen level and the log10 RT level versus the log10 RNA load show the range of assay measurements for 20 patients monitored over six visits. In 19 of the 20 patients, the serial p24 antigen and RT levels completely paralleled the RNA loads. After therapy initiation the concentrations of p24, RT, and RNA exhibited distinct decreases in all except two patients, in whom p24 antigen and RT remained undetectable. None of the patients tested showed a significant viral rebound or a lack of response, and thus, the performances of these assays in these scenarios could not be assessed. The repeated-measures analyses with orthogonal polynomial contrasts showed a significant linear decline in log10 p24 antigen level (P < 0.0001) and a curvilinear decline (quadratic trend) in the log10 RT level (P < 0.0001) and the log10 RNA load (P < 0.0025).
The results of all three assays were inversely correlated with CD4 T-cell counts, as shown by the Spearman test (CD4 count versus RNA load, r = –0.336 and P = 0.001; CD4 count versus p24 antigen level, r = –0.541 and P < 0.0001; CD4 count versus RT level, r = –0.358 and P = 0.0006).
DISCUSSION
The costs of antiretroviral drugs have decreased substantially in recent times in resource-constrained environments (4). However, a factor limiting the implementation of antiretroviral therapy in many developing countries still focuses on the challenges of implementation of cost-effective laboratory monitoring (9). The obstacles to implementation of laboratory monitoring in resource-constrained settings are often geographically dependent and may include one or more of the following: (i) the lack of an adequate laboratory infrastructure, (ii) the absence of technical skill and/or laboratory management skills, and (iii) the costs of reagents and reagent suppliers support in remote regions.
On the basis of the findings presented above, the p24 antigen and RT enzyme assays may provide less expensive, simpler alternatives for monitoring the response to therapy in the context of infection with HIV-1 clade C. The reduced sensitivities of these assays compared to that of nucleic acid testing revealed in this data set need further investigation (16). The specificities of both of these assays were not addressed in this study and will require additional exploration in future evaluations.
Although the scatterplots in Fig. 2 showed heteroscedasticity (increasing variance with increasing log10 RNA loads), the mixed-effects model showed a significant association between the p24 antigen level, the RT concentration, and the RNA load. This variability may be influenced by the fact that all three assays measure completely different parameters of viral replication. The RT assay measures the activity of the virus-encoded enzyme RT, the HD HiSens HIV-1 p24 Ag Ultra assay measures the level of either virion-bound p24 antigen or relatively freely circulating p24 antigen in immune complexes (18), and the Roche RNA assay quantitates the virion-associated RNA in blood plasma. The more valuable application of the alternative assays is their interpretation for monitoring of patients on antiretroviral therapy over time, as both assays under evaluation showed significant declines in viral loads over time (Fig. 3). There are, however, still concerns regarding the lack of sensitivity below a certain viral RNA load and how this will influence accurate monitoring of patients on antiretroviral therapy (Fig. 2). In addition, all three assays showed the expected negative correlation with the CD4 count (10), a conventional measure of a patient's response to therapy (12, 21).
A major consideration when evaluating these alternative assays is to consider not only the cost but also the complexity of conducting the investigations. In this study we paid careful attention to the technical aspects of the proposed viral load determination methodologies. The RNA assay can be automated to a large degree, which allows easier implementation and standardization. The assay still, however, requires sophisticated, costly equipment and appropriate laboratory design to reduce the chance of cross contamination commonly seen with PCR technology. Both alternative assays can be implemented with basic ELISA laboratory equipment, such as a heating block, pipettes, a plate reader, and a plate washer. The disadvantages that remain for the RT assay include (i) the time taken to complete the assay (2 to 3 days); (ii) the large sample volume required for analysis, which may pose problems in the pediatric treatment environment; and (iii) the lack of automation, which may significantly increase person-to-person variations as well as interassay and intra-assay variations. The p24 antigen assay has the advantage of facilitating a higher throughput than either the RT or the RNA quantitation assays. The disadvantages of the p24 antigen assay include (i) the variability of the p24 antigen results for a given viral RNA load (Fig. 2a) and the fact that the lysis buffer used to improve the sensitivity of the assay is not commercially available at present.
The full economic cost of each assay is required before implementation in any laboratory environment and should include the costs of (i) reagents and consumables, (ii) labor, (iii) equipment and maintenance, and (iv) transport of samples and fixed overheads. These costs will vary substantially in each geographical region, and informed decision making thus needs to be facilitated to ensure that the appropriate technology is chosen on the basis of the cost, the specimen volume required, and the skill set available.
On the basis of the data generated in this study, these assays represent viable alternatives for further exploration for use in certain resource-constrained environments.
ACKNOWLEDGMENTS
This research was supported by a grant from the NIH CIPRA Project Program and The National Health Laboratory Services.
Ethics approval was obtained through the University of the Witwatersrand (number M00-01-07).
REFERENCES
Awad, R. J., G. E. Corrigan, D. H. Ekstrand, R. Thorstensson, C. Kallander, and J. S. Gronowitz. 1997. Measurement of levels of human immunodeficiency virus type 1 reverse transcriptase (RT) and RT-activity-blocking antibody in humans by a new standardized colorimetric assay. J. Clin. Microbiol. 35:1080-1089.
Boni, J., M. Opravil, Z. Tomasik, M. Rothen, K. Bisset, P. Grob, R. Lüthy, and J. Schüpbach. 1997. Simple monitoring of antiretroviral therapy with a signal-amplification-boosted HIV p24 antigen assay with heat-denatured plasma. AIDS 11:F47-F52.
Braun, J., J. C. Plantier, M. F. Hellot, E. Tuaillon, M. Gueudin, F. Damond, A. Malmsten, G. E. Corrigan, and F. Simon. 2003. A new quantitative HIV viral load assay based on plasma virion reverse transcriptase activity for the different types, groups and subtypes. AIDS 17:331-336.
Brown, P. 2000. Cheaper AIDS drugs due for Third World. Nature 405:263.
Burgisser, P., P. Vernazza, M. Flepp, J. Boni, Z. Tomasik, U. Hummel, G. Pantaleo, and J. Schupbach. 2000. Performance of five different assays for the quantification of viral load in persons infected with various subtypes of HIV-1. Swiss HIV cohort study. J. Acquir. Immune Defic. Syndr. 23:138-144.
Department of Health. 2000. National HIV sero-prevalence survey of women attending public antenatal clinics in South Africa. Directorate, Health Systems Research and Epidemiology, Department of Health, Pretoria, South Africa.
Diggle, P. J., K.-Y. Liang, and S. L. Zeger. 1994. Analysis of longitudinal data. Clarendon Press, Oxford, United Kingdom.
Garcia, J. G., S. Yamamoto, M. Gomez-Cano, V. Soriano, T. A. Green, M. P. Busch, T. M. Folks, and W. Heneine. 1998. Measurement of human immunodeficiency virus type-1 plasma virus load based on reverse transcriptase (RT) activity: evidence of variability of virion-associated RT. J. Infect. Dis. 177:1221-1229.
Glencross, D. K., B. V. Mendelow, and W. S. Stevens. 2003. Laboratory monitoring of HIV/AIDS in a resource-poor setting. S. Afr. Med. J. 93:262-263.
Holland, C. A., J. H. Ellenberg, C. M. Wilson, S. D. Douglas, D. C. Futterman, L. A. Kingsley, A. B. Moscicki, et al. 2000. Relationship of CD4+ T cell counts and HIV type 1 viral loads in untreated, infected adolescents. AIDS Res. Hum. Retrovir. 16:959-963.
Human Sciences Research Council of South Africa. 2002. The Nelson Mandela/SHRC study of HIV/AIDS, South African national HIV prevalence, behavioral risks and mass media, household survey 2002. Human Sciences Research Council of South Africa, Cape Town, South Africa.
Johnson, D. 2000. Principles of treating HIV. S. Afr. Med. J. 90:227-230.
Joint United Nations Programme on HIV/AIDS. 2002. UNAIDS report on the global HIV/AIDS epidemic. Joint United Nations Programme on HIV/AIDS, Geneva, Switzerland.
Ledergerber, B., M. Flepp, J. Boni, Z. Tomasik, R. W. Cone, R. Lüthy, and J. Schüpbach. 2000. Human immunodeficiency virus type 1 p24 concentration measured by boosted ELISA of heat-denatured plasma correlates with decline in CD4 cells, progression to AIDS, and survival: comparison with viral RNA measurement. J. Infect. Dis. 181:1280-1288.
Malmsten, A., X. W. Shao, K. Aperia, G. E. Corrigan, E. Sandstrom, C. F. Kallander, T. Leitner, and J. S. Gronowitz. 2003. HIV-1 viral load determination based on reverse transcriptase activity from human plasma. J. Med. Virol. 71:347-359.
Mellors, J. W., C. W. Rinaldo, P. Gupta, R. M. White, J. A. Todd, and L. A. Kingsley. 1996. Prognosis of HIV infection predicted by the quantity of virus in plasma. Science 272:1167-1170.
Nadal, D., J. Boni, C. Kind, O. E. Varnier, F. Steiner, Z. Tomasik, and J. Schupbach. 1999. Prospective evaluation of amplification boosted ELISA for heat-denatured p24 antigen for the diagnosis and monitoring of pediatric human immunodeficiency virus type I infection. J. Infect. Dis. 180:1089-1095.
Schupbach, J., J. Boni, Z. Tomasik, J. Jendis, R. Seger, C. Kind, et al. 1994. Sensitive detection and early prognostic significance of p24 antigen in heat-denatured plasma of human immunodeficiency virus type 1 infected infants. J. Infect. Dis. 170:318-324.
Schüpbach, J., and J. Boni. 1993. Quantitative and sensitive detection of immune-complexed and free HIV antigen after boiling of serum. J. Virol. Methods 43:247-256. (Erratum, 45:245.)
Schüpbach, J., M. Flepp, D. Pontelli, Z. Tomasik, R. Lüthy, and J. Boni. 1996. Heat mediated immune complex dissociation and enzyme linked immunosorbent assay signal amplification render p24 antigen detection in plasma as sensitive as HIV-1 RNA detection by polymerase chain reaction. AIDS 10:1085-1090.
Schupbach, J., J. Boni, L. R. Bisset, Z. Tomasik, M. Fischer, H. F. Gunthard, B. Ledergerber, M. Opravil, and the Swiss HIV Cohort Study. 2003. HIV-1 p24 antigen is a significant inverse correlate of CD4 T-cell change in patients with suppressed viremia under long-term antiretroviral therapy. J. Acquir. Immune Defic. Syndr. 33:292-299.
Snedecor, G. W., and W. G. Cochran. 1989. Statistical methods, 8th ed. Iowa State University Press, Ames.
World Health Organization. 2002. Scaling up antiretroviral therapy in resource limited settings. WHO guidelines. WHO document, April 2002. http://www.who.int/entity/hiv/topics/arv/en/scaling_exe_summary.pdf. World Health Organization, Geneva, Switzerland.(G. Stevens, N. Rekhviashv)