The viral load was determined as described by the manufacturers. For AMPLICOR v1.5 and Nuclisens the standard protocols were used (22
); the limit of quantification of these two assays is 400 copies of HIV-1 RNA per ml of plasma. The limit of quantification of Quantiplex v3.0 is 50 copies of HIV-1 RNA per ml of plasma (9
). The lower quantification level of the Retina HIV-1 assay is approximately 100 copies per reaction (5
HIV-1 genetic subtypes were determined for 18 subjects. HIV-1-specific reverse-transcription PCR was done on RNA recovered from 200 μl of plasma. Nested PCR was performed to obtain a 409-bp fragment from the C2-V3 env
region by using outer primer pairs JA167 and JA170 and inner primers JA168 and JA169 and to obtain a 582-bp fragment from the p17 gag
region by using outer primer pairs JA152 and JA155 and inner primers JA153 and JA154. The thermal cycling conditions for PCR and primer numbers and positions have been described previously (13
). Amplified DNA fragments were sequenced with an automated DNA sequencer.
To determine the genetic subtype, the sequences were aligned with the 2001 HIV-1 reference sequence set obtained from the Los Alamos AIDS Database, the genetic distances between sequences were calculated by using the two-parameter substitution model of Kimura, and phylogenetic analysis was performed by using the neighbor-joining method with bootstrapping (S. Kumar, K. Tamura, I. B. Jakobsen, and M. Nei, MEGA2: Molecular Evolutionary Genetics Analysis software, Arizona State University, Tempe, Ariz., 2001). The distribution of the genetic subtypes was as follows (gag
): 1 subtype A1/A1, 2 subtype B/B, 2 subtype B/−, 3 subtype −/B, 1 subtype F1/F1, and 6 subtype G/G (Fig. ). Discordant subtype classifications were found in the gag
gene sequences of the remaining three patients: patients 00PTHSM5 and 00PTHDE10 were subtype G in gag
and subtype B in env
(G/B), whereas patient 00PTHGO21 was subtype B in gag
and subtype G in env
(B/G). We were unable to genotype HIV-1 RNA from three patients due to insufficient amounts of plasma. These genotyping data are consistent with a complex HIV-1 epidemic in Portugal, with cocirculation of multiple subtypes and recombinant viruses (10
). The predominance of HIV-1 subtype G and BG recombinant forms over other subtypes in Portugal resembles the pattern seen in HIV-1 epidemics in Galicia, Spain (7
). Preliminary full-length genomic sequencing studies indicate that two of our BG recombinant viruses (00PTHSM5 and 00PTHDE10) are representative of CRF14_BG (N. Taveira and F. McCutchan, unpublished data). The G/B recombinant virus composed of a subtype B gag
and subtype G env
identified for patient 00PTHGO21 has not been described previously.
FIG. 1. Genetic subtypes and evolutionary relationships of the virus sequenced in this study based on neighbor-joining phylogenetic trees of partial gag (A) sequences (p17) and partial env (B) sequences (C2-V3). The phylogenetic trees were constructed with reference (more ...)
Overall, AMPLICOR v1.5, Quantiplex v3.0, and Nuclisens detected viral RNA in 64 (100%), 64 (100%), and 33 (52%) of the 64 patient samples, respectively (Table ). Retina HIV-1 detected viral RNA in 15 of 21 (71%) samples from 10 patients. There was a direct association between the genetic subtypes of HIV-1 and the low sensitivity of Nuclisens, as it detected all A1 and F1 samples but only 67% subtype B samples and 15.4% subtype G and BG recombinant samples (Table ). This accounted for overall false-negative results for three patients infected with subtype G/G and one patient infected with G/B recombinant. The poor performance of this test with HIV-1 subtype G had already been demonstrated in other studies with a more restricted number of clinical samples (4
). Importantly, two of three samples from one subtype B-infected individual (00PTHDE9) and all samples from one individual (00PTHGO21) infected with a gag
B and env
G recombinant virus were not detected with Nuclisens but were detected with AMPLICOR v1.5 and Quantiplex v3.0. To our knowledge, this is the first demonstration that Nuclisens may fail to detect subtype B samples. To investigate whether these negative results were due to recombination within the gag
gene, the p17 sequence from these patients was analyzed by bootscanning using SimPlot version 2.5 (21
; S. C. Ray, SimPlot for Windows [version 2.5], 1999, Baltimore, Md.; http://www.welch.jhu.edu/~sray/download
). No evidence for recombination was obtained for patient 00PTHDE9. In contrast, there was a B/G recombination breakpoint located within the gag
gene of patient 00PTHGO21 (data not shown). This may explain the negative results obtained with this patient, since Nuclisens primers hybridize within the gag
gene region (5
Numbers and percentages of patient samples in which HIV-1 RNA was detected with AMPLICOR HIV-1 Monitor Test version 1.5, Nuclisens HIV-1 QT, and Quantiplex HIV RNA 3.0 (bDNA) tests
In the small subset of samples that were tested, the Retina HIV-1 assay performed better than Nuclisens. HIV-1 RNA was not detected in 5 of 11 (45%) subtype G/G samples with the Retina HIV-1 assay (data not shown). There was one overall false-negative result for patient 00PTHDE14, who was infected with a subtype G/G virus. Of 21 plasma samples tested by Nuclisens and Retina HIV-1 assay, 5 samples from 3 subtype G-infected patients had no viral RNA that was detectable by either assay. Nuclisens detected one sample that was not detected with Retina HIV-1. Retina HIV1 detected HIV-1 RNA in 11 samples, 5 G/G, 2 −/B, 2B/B, and 2 G/B, that were not detected with Nuclisens. In contrast to what was observed for the Nuclisens assay, HIV-1 RNA was detected in all subtype B/B samples that were tested with Retina HIV-1.
To determine the abilities of the four tests to quantify HIV-1 RNA levels of various subtypes, we compared the viral loads obtained from the patient's samples with each test (Fig. ). The overall mean (SD) viral loads measured in the clinical samples were 4.33 (1.04; n = 64), 4.16 (1.01; n = 33), 3.94 (0.98; n = 64), and 3.70 (1.30; n = 15) log10 copies per ml for AMPLICOR v1.5, Nuclisens, Quantiplex v3.0, and Retina HIV-1 tests, respectively. Excluding the Retina HIV-1 results, the highest mean (SD) viral loads for subtype A1, 5.090 (0.928) log10 HIV-1 RNA copies/ml; subtype G, 4.828 (0.721); and recombinant BG, 4.282 (1.161), samples were obtained with AMPLICOR v1.5. The highest mean (SD) viral load measurements for samples of unknown genotype and subtype B samples were obtained with Nuclisens: 4.161 (1.187) and 4.476 (1.083) log10 HIV-1 RNA copies/ml, respectively. Quantiplex v3.0 yielded the highest mean viral load with subtype F1, 3.798 (0.210). Nuclisens produced the lowest mean viral load for subtype A1, G, and BG; AMPLICOR v1.5 produced the lowest mean viral load for subtype F1; and Quantiplex v3.0 produced the lowest mean viral loads for subtype B and samples not genotyped. Except for subtype F1 samples and samples of unknown genotype, the differences between the higher and lower mean viral load values obtained with the three assays for the different subtypes or recombinant samples were >0.5 log10 copies per ml (range, 0.616 to 1.712). This restates the importance of using the same test for patient follow-up.
FIG. 2. HIV-1 RNA levels (log10 copies/ml) for subtypes A1, B, G, F1, and BG recombinant forms. The box extends from the 25th percentile to the 75th percentile, with a line at the median (50th percentile). The tabs extend above and below the boxes to show the (more ...)
The frequencies of discordant results (i.e., when the difference in log10
copies per ml was >0.5) were 34% between Nuclisens and Quantiplex v3.0, 47% between AMPLICOR v1.5 and Nuclisens, and 45% between Quantiplex v3.0 and AMPLICOR v1.5 (Table ). The poor agreement obtained between Quantiplex v3.0 and AMPLICOR v1.5 in our population is in contrast to the 92.7% concordant results obtained in a recent study involving mostly subtype B samples (1
). The frequency of discordant results obtained in our study was higher (76%) for the subtype G samples. In addition, a higher mean viral load, reaching clinical significance (0.697 log10
copies per ml), was obtained for subtype G with AMPLICOR v1.5. These results, therefore, highlight a significant difference in performance between AMPLICOR and Quantiplex with subtype G virus that may have an impact in patient management.
Numbers and percentages of discordant results between AMPLICOR HIV-1 Monitor Test version 1.5, Nuclisens HIV-1 QT, and Quantiplex HIV RNA 3.0 (bDNA) tests
The Spearman rank test was used to analyze the correlation between assays. AMPLICOR v1.5 and Quantiplex v3.0 were strongly correlated (r = 0.8805, P < 0.0001, n = 64), and this was independent of subtype (subtype B, r = 0.7960, P = 0.0007; subtype G, r = 0.8505, P < 0.0001; recombinant BG, r = 0.8619, P = 0.0045), demonstrating that these assays are well suited to quantify subtype G and recombinant BG virus RNA in clinical samples. In contrast, a weaker but significant correlation was found between Nuclisens and Quantiplex v3.0 (r = 0.7123, P < 0.0001, n = 33) and between Nuclisens and AMPLICOR v1.5 (r = 0.5612, P = 0.0007, n = 33). The coefficient of correlation between Nuclisens and Quantiplex v3.0 increased to 0.9373 for subtype B samples (P < 0.0001) and 0.8929 (P = 0.0123) for samples of unknown subtype. Higher correlation coefficients were also found between Nuclisens and AMPLICOR v1.5 when only subtype B (r = 0.7960, P < 0.0007) and samples of unknown subtype (r = 0.9286, P = 0.0067) were analyzed.
In summary, in 38 plasma specimens from Portuguese children infected with HIV-1 subtype B, G, or recombinant BG the infecting virus was not detected by the licensed Nuclisens HIV-1 QT. In contrast, HIV-1 RNA was detected and quantified in all samples with the two other assays that were tested, AMPLICOR v1.5 and Quantiplex v 3.0. Nuclisens HIV-1 QT is, therefore, unsuited for viral load quantification in HIV-1-infected patients living in Portugal and, probably, other countries with a high prevalence of subtype G virus.