The performance of the Applied Biosystems ViroSeq HIV-1 Genotyping System was evaluated in laboratories at The Johns Hopkins University School of Medicine, The University of California at Los Angeles School of Medicine, and the University of Medicine and Dentistry of New Jersey (laboratories 1 to 3, respectively in Table ). Each laboratory analyzed baseline samples obtained prior to antiretroviral therapy, as well as failure samples obtained at defined time points while study subjects were still on antiretroviral therapy. A total of 196 plasma samples were analyzed, including 135 baseline samples and 61 failure samples (Table ).
Applied Biosystems recommends use of the ViroSeq system for 0.5-ml plasma samples with viral loads of >2,000 copies/ml. This amount corresponds to an input RNA copy number (i.e., RNA copies/milliliter × milliliters of plasma) of 1,000 copies. In this study, the volume of plasma available for analysis was <0.5 ml for 43% of the samples. The range of plasma volumes, mean plasma volume, and number of samples with plasma volumes of <0.5 ml analyzed in each laboratory are shown in Table . In addition, 47 (24%) of the samples had viral loads of <10,000 copies/ml. The ranges of viral loads and mean viral loads for samples analyzed in each laboratory are shown in Table . Of note, the sample set included eight samples with input RNA copy numbers of <1,000 (median 742 copies; range 480 to 899 copies).
RNA isolation, reverse transcription, and PCR amplification were performed for all 196 samples. The quality and quantity of the PCR products were analyzed by agarose gel electrophoresis. PCR yielded sufficient DNA for sequencing for 98% of the samples tested. This included seven of eight samples for which the input RNA copy number was <1,000 copies. The four samples for which a suitable product was not obtained had the following viral loads and plasma volumes: 1,205 copies/ml and 0.4 ml (482 copies), 17,738 copies/ml and 0.15 ml (2,661 copies), 6,003 copies/ml and 0.5 ml (3,002 copies), and 142,753 copies/ml and 0.33 ml (47,108 copies). The failure to obtain a suitable PCR product for the latter sample was attributed to loss of the viral pellet during viral RNA isolation. PCR products suitable for sequence analysis were obtained from samples collected at other times from all four of these individuals.
DNA sequencing was performed using premixed reagents with seven sequencing primers. Figure shows the position and orientation of the sequencing primers on the PCR product. Sequencing reactions were analyzed on either an ABI Prism 377 DNA Sequencer or an ABI Prism 310 Genetic Analyzer. One of the three laboratories used a core facility for sequence analysis. We analyzed the performance of the seven sequencing primers. For this analysis, we considered a primer to have failed if the resulting sequence did not assemble with the other sequences from the same sample using the Applied Biosystems ViroSeq HIV-1 Genotyping System software package. In this system, the A and D primers are alternate sequencing primers that bind to the genetically heterogeneous gag region. Either of these primers can be used to generate a consensus sequence for analysis. In this study, sequencing with primer A was successful for 176 (92%) of the 192 samples, and sequencing with primer D was successful for 169 (88%) of the 192 samples. Both primers A and D failed in only 3 (1.6%) of 192 samples. When an individual primer fails on the first sequencing attempt, it is sometimes possible to obtain a usable sequence if the analysis with that primer is repeated. In this study, sequencing reactions and/or analysis of sequencing products was repeated if (i) both primers A and D failed (in this case, analysis with both primers was repeated) or (ii) any other primer (B, C, F, G, or H) failed. Using this strategy, each of the three laboratories performed repeat analysis for <4% of the primer reactions. Sequencing analysis was successful for >70% of the repeated reactions in all three laboratories (Table ).
FIG. 1 Orientation and position of PCR and sequencing primers. The position and orientation of primers in the ViroSeq HIV-1 Genotyping System are shown with respect to the protease and RT coding regions. The two PCR primers (PCR-F and PCR-R) and the seven sequencing (more ...)
We considered genotyping to be successful if the analysis yielded a complete sequence for HIV-1 protease amino acids 1 to 99 and HIV-1 RT amino acids 1 to 320. In this report, genotyping was successful for all 192 samples for which adequate PCR products were obtained (Table ). For 180 samples, sequence data from both DNA strands was obtained for the entire region analyzed. For the remaining samples, a small region of the consensus sequence was generated using data from only one DNA strand. This reflected repeated failure of both primers A and D (three samples), the F primer (two samples), or the G primer (two samples). For five additional samples, sequences obtained with the H primer assembled successfully with the other sequences but had a small internal region of sequence ambiguity.
Mutations associated with resistance to the PACTG 377 study drugs (3TC, d4T, NVP, RTV, and NFV), as well as mutations associated with resistance to other drugs (e.g., ddI, ddC, and AZT) were detected in this cohort. Detailed analyses of the mutations detected and their relationship to outcome in protocol PACTG 377 will be reported elsewhere (S. H. Eshleman et al., unpublished results).
For quality control, we used phylogenetic analysis to compare the nucleotide sequences obtained for each plasma sample to one another and to those of laboratory strains of HIV-1. None of the sequences resembled those of strains pNL4-3 or HXB2, which were present in some of the genotyping laboratories. Furthermore, sequences from individual children derived from samples collected at different times clustered more closely with one another than with sequences from other children in the study, and no two sequences were identical. Therefore, it is unlikely that samples were cross-contaminated (e.g., during PCR) or misidentified.