|Home | About | Journals | Submit | Contact Us | Français|
Monitoring antiretroviral therapy requires that human immunodeficiency virus type 1 (HIV-1) viremia assays are applicable to all distinct variants. This study evaluates the performance of three commercial viral load assays—Versant HIV-1 RNA bDNA v3.0, Cobas AmpliPrep/Cobas TaqMan HIV-1, and NucliSens HIV-1 EasyQ v1.2—in testing 83 plasma specimens from patients carrying HIV-1 non-B subtypes and recombinants previously defined by phylogenetic analysis of the pol gene. All 28 specimens from patients under treatment presented viremia values below the detection limit with the three methods. In the remaining 55 specimens from naive individuals viremia could not be detected in 32.7, 20, and 14.6% using the NucliSens, Versant, or TaqMan tests, respectively, suggesting potential viral load underestimation of some samples by all techniques. Only 32 (58.2%) samples from naive subjects were quantified by the three methods; the NucliSens test provided the highest HIV RNA values (mean, 4.87 log copies/ml), and the Versant test provided the lowest (mean, 4.16 log copies/ml). Viremia differences of greater than 1 log were seen in 8 (14.5%) of 55 specimens, occurring in 10.9, 7.3, and 5.4%, respectively, of the specimens in comparisons of Versant versus NucliSens, Versant versus TaqMan, and TaqMan versus NucliSens. Differences greater than 0.5 log, considered significant for clinicians, occurred in 45.5, 27.3, and 29% when the same assays were compared. Some HIV-1 strains, such as subtype G and CRF02_AG, showed more discrepancies in distinct quantification methods than others. In summary, an adequate design of primers and probes is needed for optimal quantitation of plasma HIV-RNA in non-B subtypes. Our data emphasize the need to use the same method for monitoring patients on therapy and also the convenience of HIV-1 subtyping.
The accurate determination of human immunodeficiency virus type 1 (HIV-1) RNA levels is important for understanding the natural history of HIV infection, predicting disease progression to AIDS, and the efficacy of antiretroviral therapies in a given patient. Early changes in RNA levels in response to treatment appear to predict the long-term clinical benefit (23). Quantitation assays should be able to detect all genetic variants with similar efficiency. However, all currently available tests have been designed mainly for HIV-1 subtype B. Several reports have noticed differences obtained when quantifying HIV-1 non-B subtypes with different methodologies (5, 7, 12, 14, 25, 26, 28). Moreover, failure to detect non-B subtypes by viral load techniques has been previously described (10, 11). The natural polymorphisms associated with genetic non-B subtypes and recombinant strains in the HIV-1 binding sites of primers or probes used in those assays can influence in their efficiency (12, 20, 25, 26).
In many European countries, 20 to 50% of new infections are due to non-B subtypes and recombinant variants (3, 4, 9, 31). In Spain, several studies have estimated HIV-1 non-B subtype and recombinant variant infections in nearly 14% of HIV-infected Spaniards, and this number is increasing (18). Interestingly, nearly two-thirds of foreigners with HIV-1 infection in Madrid carry non-B subtypes, and this value increases to up to 90% in patients infected in sub-Saharan African countries (16, 22), where these variants are prevalent. The increasing frequency of HIV-non-B subtypes in the United States and Europe demonstrates the importance of using appropriate HIV-1 viral RNA quantitation assays. The problem is particularly relevant in patients undergoing antiretroviral therapy.
The objective of the present study was to evaluate the performance of three commercially available viral load tests on 83 well-characterized plasma specimens carrying six different HIV-1 group M non-B subtypes and several intersubtype recombinant strains. All plasma specimens were collected from HIV-infected subjects with clinical follow-up in a reference hospital in Madrid, Spain.
A panel of 83 well-characterized plasma specimens infected with six different HIV-1 group M non-B subtypes (4 A, 4 C, 1 D, 1 F2, 12 G, and 5 H) and 56 intersubtype recombinant strains (49 circulating recombinant forms [CRFs] [43 CRF02_AG, 4 CRF10_CD, and 2 CRF11_cpx] and 7 unique recombinant forms [URFs; 3 A/D and 4 U/C]) defined at pol and obtained from 49 HIV-1-infected patients were selected for the study. A total of 28 (33.7%) of the 83 plasma specimens were collected from 20 subjects receiving highly active antiretroviral therapy (HAART), and the remaining 55 (66.3%) samples were taken from 29 individuals naive for any antiretroviral treatment. All HIV-infected patients lived in Madrid, Spain, and were being followed up at an AIDS/HIV reference hospital. Forty-five of them (94%) were Africans coming from Equatorial Guinea (n = 32), a former Spanish colony, Nigeria (n = 4), Liberia (n = 3), Ghana (n = 1), Bissau Guinea (n = 1), the Democratic Republic of Congo (n = 1), Cameroon (n = 1), Ethiopia (n = 1), and Kenya (n = 1). Three HIV-1-infected subjects were Spanish natives infected in non-B areas of endemicity, and one individual acquired the infection in India. The present study was part of a project approved by a review board and Ethical Committee from our institution. It was designed to protect the rights of all subjects involved under the appropriate local regulations. To maintain subject confidentiality, a unique identification number was assigned to each specimen.
HIV-1 subtypes were previously defined by phylogenetic analysis in the pol gene including protease and reverse transcriptase coding regions (17, 18, 19, 22), using as reference HIV-1 sequences belonging to HIV-1 group M available at the GenBank. The tree topology was obtained by using the neighbor-joining method. Alignment of DNA sequences was performed by using the CLUSTALX program. The pairwise distance matrix was estimated by using the Kimura two-parameter model within the DNADIST program, as implemented in the PHYLIP software package. Bootstrap resampling (1,000 data sets) of the multiple alignment was performed to test the statistical robustness of the tree.
These plasmas were used to compare the performances of three commercial viral load assays done at the institution according to the manufacturer's instructions. The Versant HIV-1 RNA bDNA v3.0 (Bayer Diagnostic, Tarrytown, NY) uses the semiautomated Bayer system 340 (automated incubations, automated washing, and manual addition of reagents) (6). The NucliSens EasyQ HIV-1 v1.2 assay (BioMérieux, Inc., Durham, NC) measures plasma HIV-1 RNA extracted automatically with easyMAG Magnetic silica, followed by a real-time detection method combined with NASBA technology (13, 29). The Cobas AmpliPrep/Cobas TaqMan HIV-1 test (Roche Diagnostics, Branchburg, NJ) uses the Cobas AmpliPrep instrument for automated sample preparation and the Cobas TaqMan analyzer for automated real-time PCR amplification and detection, according to the manufacturer's recommendations. The Versant assay targets highly conserved regions of the pol gene, and both the TaqMan and NucliSens tests target the gag coding region (6, 13, 30). At the time of our study, according to the manufacturers, the assays could quantify HIV-1 RNA over the ranges of 50 to 500,000 (Versant), 50 to 3,000,000 (NucliSens), and 40 to 10,000,000 (TaqMan) HIV-1 RNA copies/ml.
A paired sample t test was used to compare the mean HIV-1 viremia values obtained using the quantification assays. Statistical analysis was done using the SPSS software version 13.0.
All 28 plasmas from HAART patients presented viral load values below the detection limit (<50 or <40 HIV-1-RNA copies/ml for Versant and NucliSens or for TaqMan, respectively) with the three techniques, as was expected when treatment is efficient. Among the 55 specimens from naive subjects, only 32 (58.2%) could be quantified by the three assays. The remaining 23 (52.7%) provided viremia values above the detection limit only with one or two quantification assays. HIV-1 viremia could not be detected in 32.7, 20, and 14.6% of plasmas from naive patients using the NucliSens, Versant, or TaqMan tests, respectively, suggesting potential viral load underestimation of some samples by all techniques (Table (Table1).1). NucliSens, Versant and TaqMan could not quantify 40, 25, and 10% of the pure non-B subtypes and 28.6, 17.2, and 17.2% of the recombinants at pol, respectively (Table (Table11).
The percentage of samples from naive subjects providing viremia values below the detection limit by one quantification assay but detectable for the other two techniques was 10.9% (6 of 55) for NucliSens (variants: one C, one G, two CRF02_AG, one CRF10_CD, and one URF), 5.4% (3 of 55) for Versant (one D and two CRF02_AG), and 0% for TaqMan (Table (Table2).2). Thus, the TaqMan assay provided detectable viremia in some plasma specimens that could not be quantified by the other two assays, presenting higher reliability by quantifying HIV-1 non-B subtypes and recombinants. In contrast, NucliSens did not quantify a higher number of specimens, which were detected by the other two techniques providing viremia values above the detection limit (Table (Table1).1). NucliSens could not quantify the F2 variant, none of the 3 subtypes H, 1 of 2 subtypes C, 3 of 11 subtypes G, and 5 of 7 URFs. The subtype D and F2 specimens could not be detected by Versant. No techniques detected 17.4% of the 23 CRF02_AG variants infecting naive patients (Table (Table11).
Table Table22 shows HIV-1 viremia values and differences testing the 55 HIV-1 non-B subtypes and recombinants from naive subjects. Viremia differences of greater than 1 log were seen in 8 (14.5%) of 55 specimens, occurring in 10.9, 7.3, and 5.4% when Versant versus NucliSens, Versant versus TaqMan, and TaqMan versus NucliSens, respectively, were compared (Table (Table3).3). In more detail, they were observed in the subtype D specimen, in 1 of 2 subtype A, in 2 of 11 (18.2%) subtypes G, and in 4 of 23 (17.4%) CRF02_AG recombinants (Table (Table2).2). Interestingly, these percentages were higher when viremia differences of 0.8 log were considered, increasing to 45.4% (5/11) subtypes G, 34.8% (8/23) CRF02_AG, 50% (2/4) of CRF10_CD, and the CRF11_cpx specimen (Table (Table2).2). Viremia differences among assays that were higher than 2 to 3 logs were observed in subtype D and in 2 of 23 CRF02_AG specimens (specimens 5, 22, and 29 in Table Table2).2). Thus, some HIV-1 strains showed more discrepancies using distinct quantification methods than others.
In clinical practice, viremia differences of >0.5 log are considered significant for clinicians during therapy monitoring of HIV-1-infected patients. We observed that 45.5, 29, and 27.3% of the 55 specimens presented ≥0.5 log of viremia differences when Versant versus NucliSens, TaqMan versus NucliSens, and Versant versus TaqMan were compared (Table (Table3).3). In more detail, one of two clades A, the clade D, the CRF11_cpx sample, 82% (9/11) of subtype G, 74% (17/23) CRF02_AG, and 75% (3/4) CRF10_CD recombinants, and 14.3% (1/7) presented ≥0.5-log viremia differences comparing two assays (Table (Table2).2). Considering the 20 pure non-B subtypes and the 35 intersubtype recombinants separately, a similar number of specimens provided ≥0.5-log differences in viral load values testing Versant versus TaqMan (25 and 28.6%), Versant versus NucliSens (35 and 51.4%), and TaqMan versus NucliSens (35 and 25.7%), respectively (Table (Table3).3). In analyses of only the 32 plasma samples with detectable plasma viremia by all three techniques, differences in viral load values >0.5 log were observed in 31.2% comparing Versant and TaqMan, 50% comparing Versant and NucliSens, and 37.5% comparing TaqMan and NucliSens.
Among the 32 (58.2%) samples quantified by the three techniques, NucliSens provided the highest HIV RNA values (mean, 4.87 log copies/ml), followed by TaqMan (mean, 4.54 log copies/ml) and Versant (mean, 4.16 log copies/ml). Thus, NucliSens provided 0.3- and 0.7-log higher HIV-1 mean viral load values than TaqMan and Versant, respectively.
This study evaluates the performance of three commercial viral load assays (the Versant HIV-1 RNA bDNA v3.0, Cobas AmpliPrep/Cobas TaqMan HIV-1 test, and NucliSens HIV-1 QT) testing a panel of 83 plasma specimens infected with 27 HIV-1 non-B subtypes and 56 recombinants, previously defined by phylogenetic analysis of the pol gene. All subjects were in regular follow-up in a reference hospital in Madrid, Spain. Twenty-eight specimens were collected from HAART patients, which explains the viremia values below the detection limit obtained by the three tests in all of them. Since viremia is not controlled without antiretroviral treatment, all 55 plasma specimens from naive subjects should have had high HIV-1 viremia viral loads, theoretically susceptible to be quantified by the tests. However, only 32 of these 55 specimens (58.2%) presented viremia values above the detection limit in all three assays, and 23 could not be quantified by one or two techniques, indicating an underestimation of HIV-1 viral load in some non-B specimens. In fact, from 10 to 40% of the HIV-1 non-B subtypes and from 17.5 to 28.6% of recombinants infecting naive subjects could not be quantified for some of the three tests (Table (Table11).
Previous studies quantifying 15 different non-B HIV-1 variants revealed that viremia values comparing the Cobas AmpliPrep/Cobas TaqMan HIV-1 test and bDNA v3.0 yielded results within the accuracy limits of ±0.3 log, being lower (±0.1 log) when only subtype B specimens were analyzed (27). Another report quantifying 10 non-B subtypes observed that the mean difference between viremia detected by Cobas AmpliPrep/Cobas TaqMan HIV-1 test and bDNA v3.0 was 0.360-log RNA copies (24), increasing to 0.553 log when Cobas and NucliSens were compared (24). Other authors published that the coefficients of variance in testing NucliSens EasyQ in 35 clade B specimens ranged from 2.3 to 10.4% (21), increasing to differences of 0.32 viremia log at low-end HIV-1 RNA concentrations (13). In our study, 14.5 and 60% of the 55 plasma samples from naive subjects carrying non-B and recombinant variants provided viral load differences above ≥1 log and ≥0.5 log, respectively, when two of those techniques were compared. In fact, the subtype D showed ≥2-log viremia differences since it could not be quantified by the Versant assay (Table (Table1).1). We strongly believe that the observed viremia differences among assays in some of our non-B specimens could have clinical relevance during antiretroviral treatment monitoring of these patients.
Our data revealed that some strains, such as subtype G and CRF02_AG, showed more viremia discrepancies than others. In agreement with our findings, a recent evaluation testing the Roche Cobas TaqMan and Abbott RealTime extraction-quantification systems for HIV-1 subtype quantification also reported large viremia differences in non-B subtypes, particularly in CRF02_AG variants (14). As for other HIV-1 variants, a recent study showed the versatility of both the NucliSens EasyQ v1.1 and the Cobas Amplicor tests in monitoring subtype B and recombinant CRF01_AE viruses prevalent in Southeast Asia (21). It should be noted that neither of these variants was included in our study.
Although TaqMan and NucliSens assays target the gag coding region, significant quantification differences between the two assays were shown in 29% (16/55) of samples from naive subjects, increasing to 37.5% (12/32) in those samples with viremia values above the detection limit by the three methods. Our results revealed that NucliSens EasyQ HIV-1 v1.2 assay was more sensitive than Versant for HIV-1 non-B subtypes quantification in plasma, according to a previous multicenter evaluation using the NucliSens EasyQ HIV-1 v1.1 versus Versant assay (8).
New versions of HIV-1 quantification assays are continuously being developed to improve virus quantification of all viral variants. Several studies have been conducted to test the agreement among different versions of the same assay, but mainly testing clade B specimens. In fact, less-significant viremia differences among tests occurred when only clade B viruses were tested, as was observed in comparisons of the Cobas AmpliPrep/Cobas Amplicor HIV-1 Monitor Ultrasensitive Test and the Cobas Amplicor HIV-1 Monitor test using manual specimen preparation (1). However, disagreement in viremia quantification between two versions of the same assay has been reported testing non-B subtypes (2, 7, 15). According to all of these data, we strongly suggest measuring all HIV-1 viral load determinations in each HIV-1-infected patient during follow-up by the same quantification technique and version.
In summary, an adequate design of primers and probes is needed for optimal quantitation of plasma HIV-RNA in non-B subtypes. Our data reinforce the need for monitoring patients on therapy with the same method and the convenience of HIV-1 subtyping.
This study was supported in part by grants from Red de Investigación en SIDA (RIS, project ISCIII-RETIC RD06/006) and Fondo de Investigaciones Sanitarias (FIS PI060925). A.H. and V.S. are supported by FIS and Agencia Laín Entralgo.
We thank Rocío González for her support in the statistical analysis.
Published ahead of print on 2 July 2008.