INTRODUCTION

Top Abstract Introduction Materials and Methods Results Discussion References

Clinical management of human immunodeficiency virus type 1 (HIV-1) infection and implementation of treatment strategies rely on the quantitative measurement of HIV-1 RNA in plasma (26, 35). A variety of quantitative viral load assays have been developed; they differ in sensitivity, dynamic range, target region, sample volume, specimen preparation, methods of nucleic acid amplification, and detection (20a, 21, 22, 28). Most of these tests were calibrated against a well-characterized standard stock of subtype B HIV-1 RNA prepared by the Viral Quality Assurance Laboratory of the AIDS Clinical Trials Group (43). Although the majority of viral load monitoring is currently performed in North America and Europe, where HIV-1 group M subtype B infections predominate, subtype B represents only about 3% of HIV-1 infections worldwide. With the continued evolution of the HIV-1 epidemic, increasing numbers of non-subtype-B infections are being identified in Europe and the United States (1, 3, 7, 17). In France, a study of blood donors from 1985 to 1995 revealed an increase in the prevalence of non-subtype-B infections from 4% to more than 20% over the 10-year period (4). Moreover, as a result of coinfection of individuals with viral strains of more than one subtype, increasing numbers of intersubtype recombinant forms of HIV-1 are being observed (31, 34). Intergroup (group M-group O) recombinant viral strains have also been identified (30, 40). In fact, of the completely sequenced HIV-1 genomes, nearly 20% have a mosaic structure consisting of at least two subtypes (33).

Given the continued global expansion of non-subtype-B and recombinant forms of HIV-1, it is important that commercial nucleic acid-based tests detect and quantify accurately even the most genetically divergent HIV-1 strains. The impact of genetic variation on quantification by viral load assays has been well documented and may include underquantification or even complete lack of detection (2, 8, 10, 18, 19, 24, 29). Recombination between distantly related strains may further contribute to the emergence of HIV-1 variants that are not detected efficiently by current molecular tests.

Comparison of viral load assays has been complicated by the lack of universal standards and/or quality control panels that encompass non-B subtypes. A panel of 30 HIV-1 isolates representing group M subtypes A to G was recently developed and characterized at the Walter Reed Army Institute of Research (WRAIR, Bethesda, Md.) (20, 27). Ultimately, calibrated standards prepared from these or similar isolates will be used to evaluate the performance of viral load assays.

In the present study, we evaluated the performance of three commercial HIV-1 ultrasensitive viral load assays, LCx HIV RNA Quantitative assay (LCx HIV; Abbott Laboratories, Abbott Park, Ill.), AMPLICOR HIV-1 MONITOR version 1.5 ultrasensitive test (MONITOR v1.5; Roche Diagnostics, Branchburg, N.J.), and Quantiplex HIV-1 RNA version 3.0 assay (bDNA v3.0; Bayer Diagnostics, Emeryville, Calif.), with dilution series of 39 virus isolates. The viruses represent HIV-1 group M subtype A to G and group O strains and include the WRAIR clade panel. In addition, gag-encoded p24 (gag p24), pol-encoded integrase (pol IN), and env-encoded gp41 (env gp41) immunodominant region (IDR) sequences were characterized for each isolate to more definitively establish group and subtype designations and to evaluate the degree of genetic diversity at primer and probe binding sites.

	MATERIALS AND METHODS

Top Abstract Introduction Materials and Methods Results Discussion References

Viral isolates. WRAIR, provided 30 of the HIV-1 group M isolates, including 1 A, 6 B, 5 C, 3 D, 8 E, 3 F, 1 G, and 3 circulating recombinant forms (CRF), through generous donations from Merlin Robb and Nelson Michael. Virus concentrations were determined by electron microscopy (EM), p24 antigen concentration, and quantitative viral load analysis (20, 27). An additional nine isolates, including seven HIV-1 group M isolates (two A, one B, one C, one D, one F, and one mosaic) and one group O isolate, were obtained through a Collaborative Research and Development Agreement with the Centers for Disease Control and Prevention (Atlanta, Ga.); the other group O isolate was received from Serologicals, Inc., Atlanta, Ga. Cell-free virus stocks from the 39 HIV-1 isolates were prepared by SRA Technologies, Rockville, Md.

Molecular characterization of the viral isolates. Three regions of the HIV-1 genome were targeted for sequence analysis: gag p24 (399 nucleotides [nt]), pol IN (864 nt), and env gp41 IDR (369 nt). Each virus stock was diluted 100-fold in HIV-1-seronegative human plasma. Total nucleic acid was extracted from 200 µl of each sample using a QIAamp blood kit (Qiagen Inc., Chatsworth, Calif.). Primers and conditions used for reverse transcription (RT)-PCR amplification of gag p24, pol IN, and env gp41 IDR and for automated sequence analysis and phylogenetic analysis have been described previously (5, 6, 14, 39, 41).

Preparation of the clade dilution panel. A 273-member panel was prepared by diluting each of the 39 virus isolates in defibrinated HIV-1-seronegative human plasma to achieve a target concentration of approximately 4.5 to 5.5 log₁₀ HIV-1 RNA copies/ml (31,623 to 316,223 copies/ml), followed by six fivefold serial dilutions of each isolate. Panels were distributed into 1.2-ml aliquots and stored at 70°C until testing.

HIV-1 load determination. (i) LCx HIV assay. The LCx HIV assay was performed according to the manufacturer's specifications. This competitive RT-PCR assay targets the pol IN region of HIV-1 (20a, 39). With the 1.0-ml sample preparation protocol, the upper limit of quantitation (ULQ) and the lower limit of quantitation (LLQ) are 1,000,000 (6.0 log₁₀) copies/ml and 50 (1.7 log₁₀) copies/ml, respectively.

(ii) MONITOR v1.5 test. MONITOR v1.5 test was performed at LabCorp (Research Triangle Park, N.C.) according to the manufacturer's instructions. The procedure requires 0.5 ml of plasma and uses RT-PCR to target the gag p24 region of HIV-1 (38). The ULQ and LLQ are 75,000 (4.88 log₁₀) copies/ml and 50 (1.7 log₁₀) copies/ml, respectively.

(iii) bDNA v3.0 assay. bDNA v3.0 assay was performed at the Instituto de Salud Carlos III (Madrid, Spain) according to the manufacturer's instructions. The procedure requires 1.0 ml of plasma and relies on signal amplification technology. In this assay, HIV-1 RNA is hybridized to a series of oligonucleotide probes complementary to highly conserved regions of the HIV-1 pol gene (9). The ULQ and LLQ are 500,000 (5.7 log₁₀) copies/ml and 50 (1.7 log₁₀) copies/ml, respectively.

	RESULTS

Top Abstract Introduction Materials and Methods Results Discussion References

Genetic analysis of viral isolates. Thirty-nine HIV-1 isolates were used to evaluate the three commercial viral load assays. The country of origin and group and subtype assignments for each isolate are shown in Table 1. Subtypes for the WRAIR isolates (27) were based on limited sequence analysis of regions of the gag and/or env genes obtained from proviral DNA. The complete genome sequence was available for only four of the samples: CRF-AG (IbNG), DJ263; C, ETH2220; CRF-AE, CM240; and G, HH8793. To increase the integrity of group and subtype assignments three distinct regions (gag p24/pol IN/env gp41 IDR) of the genome were sequenced from each isolate. Based on the phylogenetic analysis (data not shown), the panel was composed of the following group M subtypes: three A, seven B, six C, four D, eight CRF-AE, four F, one G, and four intersubtype recombinants (two CRF-AG [IbNG], one A/G/G, and one B/A/B); it also contained two group O viruses. Of note, the additional sequence information obtained for isolate CM237, previously reported as subtype B, revealed that it is an intersubtype A-B recombinant with pol IN derived from subtype A. 

Viral load determination. To assess the linearity of quantification across the dynamic ranges of the viral load assays, seven dilutions each of the 39 viral isolates were prepared with HIV-1-seronegative human plasma. Concentrations of the serial fivefold dilutions were expected to target from the ULQ to below the LLQ (50 RNA copies/ml) of all three assays. All 273 dilution panel members were evaluated in each of three commercial assays: LCx HIV assay, MONITOR v1.5 test, and bDNA v3.0 assay. The linear regression of each isolate dilution series was calculated from panel members quantified within the assay dynamic ranges. Results for representative HIV-1 group M subtypes A through G and group O are shown in Fig. 1a and b. For all 37 group M isolates, a linear decrease in quantification was observed across each dilution series in all three assays. No subtype-specific deficiency was apparent for any of the assays. Correlation coefficients for the LCx HIV assay, MONITOR v1.5 test, and bDNA v3.0 assay ranged from 0.9786 to 0.999, 0.9765 to 0.9985, and 0.9736 to 0.999, respectively. In contrast, group-specific differences between the assays were readily apparent. Relative to the LCx HIV assay, bDNA v3.0 detected but consistently underquantified both group O isolates by 0.7 to 1.21 log₁₀ copies/ml across the assay dynamic range. MONITOR v1.5 failed to detect any of the group O panel members.

View larger version (44K): [in this window] [in a new window]   FIG. 1.   Representative fivefold dilution series of HIV-1 group M and group O isolates. (a) Group M subtypes A, B, C, and D. (b) Group M subtypes CRF-AE, F, and G and group O. Log10 RNA copies per milliliter are plotted versus the dilution series order, whereby numbers 1 through 7 correspond to neat, 1:5, 1:25, 1:125, 1:625, 1:3,125, and 1:15,625 dilutions, respectively. Quantitative values and linear regression trend lines for each of the three commercial viral load tests are shown as follows: solid circle and solid line, LCx HIV assay; open circle and dotted line, MONITOR v1.5 test; and triangle and dotted-dashed line, bDNA v3.0 assay.

View larger version (18K): [in this window] [in a new window]   FIG. 2.   Comparison of HIV-1 group M virus isolate dilution panel members quantified by the LCx HIV assay and the MONITOR v1.5 test. Group (Grp) O dilution panel members were not quantified by MONITOR v1.5 but are shown for comparison. The lower limit of quantification is shown by broken lines at 1.7 log10 RNA copies/ml (50 copies/ml).

View larger version (19K): [in this window] [in a new window]   FIG. 3.   Comparison of HIV-1 group M and O virus isolate dilution panel members quantified by the LCx HIV assay and the bDNA v3.0 assay. The lower limit of quantification is shown by broken lines at 1.7 log10 RNA copies/ml (50 copies/ml).

View larger version (19K): [in this window] [in a new window]   FIG. 4.   Comparison of HIV-1 group M virus isolate dilution panel members quantified by the MONITOR v1.5 test and the bDNA v3.0 assay. Group O dilution panel members were not quantified by MONITOR v1.5 but are shown for comparison. The lower limit of quantification is shown by broken lines at 1.7 log10 RNA copies/ml (50 copies/ml).

Nucleotide mismatches at primer and probe sites. Genetic characterization of the viral isolates included sequence analysis of pol IN and gag p24, target regions of the LCx HIV assay and the MONITOR v1.5 test, respectively. This characterization provided the opportunity to examine nucleotide conservation within primer and probe binding sites for these assays. Fewer mismatches were observed for the LCx HIV assay than for the MONITOR v1.5 test (Table 2). For group M isolates, the LCx HIV assay had a mean number of total nucleotide mismatches of 1.0 (range, 0 to 3 mismatches). In contrast, the MONITOR v1.5 test had three times more total nucleotide mismatches, with a mean of 3.2 changes (range, 0 to 8). At the forward primer, reverse primer, and probe sites, the mean numbers of nucleotide mismatches for group M samples were 0.6, 0.3, and 0.1, respectively, in the LCx HIV assay and 1.2, 0.6, and 1.3 in the MONITOR v1.5 test. For the two group O samples, the LCx HIV assay had a total of 5 mismatches, in contrast to a mean of 22 (range, 21 to 23) changes observed with the MONITOR v1.5 test. In the LCx HIV assay, the largest number of mismatches was observed for the forward primer and group O sequences (mean and range, four). For the 39 pol IN sequences examined, 92.3% of samples had two or fewer total nucleotide mismatches at the primer and probe binding sites in the LCx HIV assay. In contrast, only 38.5% of the gag p24 sequences had two or fewer mismatches at the MONITOR v1.5 test primer and probe binding sites.

	DISCUSSION

Top Abstract Introduction Materials and Methods Results Discussion References

First-generation HIV-1 quantitative tests were developed and standardized by use of sequence information derived primarily from subtype B infections. Because nucleic acid-based assays depend on hybridization, primer and probe mismatches with non-subtype-B isolates can significantly reduce the efficiency of quantification, thus severely limiting the utility of these tests. Within the last few years, viral load assays with improved performance characteristics for non-subtype-B HIV-1 variants have become available. The ultrasensitive MONITOR v1.5 test has incorporated a new primer set, modifications in cycling conditions, and increased sample volume, resulting in significant improvement in subtype detection and assay sensitivity (13, 42). The ultrasensitive LCx HIV assay utilizes RT-PCR and targets a highly conserved region of pol IN to provide subtype-independent quantification (20a, 39). The addition of a new target probe set, optimization of the buffer to increase signal, and incorporation of isoC and isoG into the probe binding regions have resulted in enhanced bDNA v3.0 assay specificity and sensitivity (25). Recently, increasing emphasis has been placed on the development of assays capable of quantifying both group M and group O strains (11, 39; H. Schiltz, C. Wild, N. Kilgore, P. Singh, M. Chorley, and M. Garcia-Meijide, Abstr. 7th Conf. Retroviruses Opportunistic Infections, p. 220, 2000).

The availability of well-characterized viral isolates composed of different subtypes is valuable for comparing the performance of viral load assays. A panel of 30 isolates representing HIV-1 group M subtypes A to G was assembled by WRAIR (20, 27). A current limitation of this panel is that in most cases, the subtype was designated based on limited sequence information. Assignment of a subtype based on analysis of an individual region of the genome can be unreliable, particularly for isolates derived from regions where multiple subtypes cocirculate (34). Thus, when target regions differ between assays, such a subtype designation may not be predictive of the ability of each test to quantify HIV-1 accurately. In fact, based on additional data for the gag p24 and pol IN regions generated in this study, isolate CM237, previously designated as subtype B based on env, was shown to be a B/A/B (gag, pol, env) recombinant virus. Complete genome characterization would increase the value of the panel and would allow examination of nucleotide mismatches regardless of assay target region.

Of interest, although the viral loads determined by the three assays were in close agreement, they were 0.5 to 1.5 log₁₀ copies/ml lower than the theoretical virion counts predicted by either EM or p24 antigen measurements. This finding highlights a critical issue in the construction of quality control panels, namely, the method(s) used to calibrate virus stocks. For the WRAIR panel, determination of p24 antigen concentration and virus particle count by EM were previously shown to be highly correlated (27). Based on our data, use of multiple quantitative assays to establish a nominal value would be the most reliable method of calibrating virus stocks to be used for nucleic acid testing.

In the present study, dilution series of 39 virus isolates were quantified using three commercial ultrasensitive viral load assays: LCx HIV assay, MONITOR v1.5 test, and bDNA v3.0 assay. For group M isolates (subtypes A to G), there was a high degree of correlation between quantified values obtained across the dynamic ranges for all three tests, and no subtype-specific differences were observed. These data are consistent with the results obtained with HIV-1-infected plasma, showing broad subtype specificity of the LCx HIV assay and improved subtype detection by MONITOR v1.5 relative to earlier-generation tests, particularly for subtypes A, E, F, and G (2, 39, 42). As observed previously, there was close agreement between the MONITOR v1.5 test and the bDNA v3.0 assay for subtype B specimens (12, 16). Our data extend the observation to other group M subtypes (A to G) and to circulating intersubtype recombinant forms of the virus. These results contrast with results obtained with earlier versions of these assays (10, 18).

When dilution profiles of each isolate were examined, all quantified isolates had linear correlation coefficients of better than 0.97 for all three assays. For group M isolates, values quantified across the assay dynamic ranges were within 1 log difference among the three commercial tests. Although there was a tendency of MONITOR v1.5 to underquantify high-end panel members relative to the other tests, the molecular basis for this reduction is unclear. At the low end of the titration series (near 50 copies/ml LLQ), 8 to 12% of the group M dilution panel members were quantified by a single assay. Since quantitative assays are less precise around the LLQ and values fluctuate statistically, single-replicate testing of samples with low viral titers may not predict accurately the virus concentrations in these dilution panel members.

Although HIV-1 group O strains are endemic to West Central Africa and represent a small proportion of total HIV infections, they have attained a relatively widespread distribution, having been identified in France, Spain, Germany, Belgium, and the United States (15, 32, 36, 37). Moreover, the recent identification of intergroup (M-O) recombinant strains (30, 40) raises the possibility that genetically divergent subgenomic regions of group O viruses may become even more widespread when placed in the context of group M genomes. This type of recombination has the potential to dramatically increase the effective global distribution of HIV-1 group O. Thus, it has become increasingly important that viral load assays accurately quantify even the most genetically divergent HIV-1 variants.

Of the three assays examined, only the LCx HIV assay demonstrated the ability to detect and reliably quantify group O viruses. The linear quantification of the two group O strains by the LCx HIV assay is consistent with its performance with group O virus-infected plasma samples (39). The performance of the bDNA v3.0 assay with serial dilutions of group O isolates was considerably less robust. Relative to the LCx HIV assay, values ranged from 0.7 to 1.2 log₁₀ copies/ml lower with the bDNA v3.0 assay, and several of the low-end dilutions were not detected. Thus, although the bDNA v3.0 assay is capable of detecting high-titer group O viruses, the accuracy of quantification is suspect, limiting its utility for monitoring group O virus-infected patients (23). The MONITOR v1.5 test failed to detect any of the group O dilutions analyzed. This result is particularly striking when one considers that both the LCx HIV and the bDNA v3.0 assays measured >4 log copies of virus/ml at the lowest dilution of group O isolate 3012, yet the MONITOR v1.5 test failed to detect this dilution panel member.

Sequence characterization of the viral isolates included the target regions of the MONITOR v1.5 test (gag p24) and the LCx HIV assay (pol IN). For the group M isolates examined, a higher degree of nucleotide conservation was observed for the LCx HIV assay primer and probe sites. A maximum of three total mismatches were observed in the LCx HIV pol IN target region, compared to eight total mismatches in the MONITOR v1.5 gag p24 target region. The isolates had two or fewer mismatches in 100, 97, and 100% of cases for the forward primer, reverse primer, and probe in the LCx HIV assay and 62, 100, and 92% of cases in the MONITOR v1.5 test, respectively. The maximum number of mismatches in the MONITOR v1.5 test was observed with subtypes A, F, G, and CRF-AE. A similar analysis of 290 genetically diverse HIV-1-infected plasma samples revealed up to five total nucleotide mismatches in the primer and probe binding sites in the LCx HIV assay, whereas up to 12 were identified in the MONITOR v1.5 test (39). The CRF-AG (IbNG) isolate, DJ263, trended slightly lower in the LCx HIV assay than in either the MONITOR v1.5 test or the bDNA v3.0 assay, particularly at the high end of the dynamic range, but there was no apparent molecular basis for the difference in quantification. The subtype A isolate, UG273, also trended slightly lower in the LCx HIV assay. For this isolate, three nucleotide mismatches were identified within the reverse primer binding site, although none was present at the critical 3' end of the primer.

Analysis of the primer and probe target regions in the genetically divergent group O isolates revealed a significant difference between the assays. Whereas the LCx HIV assay had up to 5 total mismatches at primer and probe binding sites, the MONITOR v1.5 test had 21 to 23 mismatches. The large number of nucleotide mismatches provides a molecular basis for the observed inability of MONITOR v1.5 to detect group O samples.

Based on the quantification of 37 dilution series of genetically diverse group M isolates, the LCx HIV assay, MONITOR v1.5 test, and bDNA v3.0 assay perform equivalently for group M strains. All group M subtypes (A to G) showed linear dilution profiles across the assay dynamic ranges. In contrast to group M isolates, group O isolates were efficiently quantified by only the LCx HIV assay. The bDNA v3.0 assay detected but consistently underquantified group O isolates, and the MONITOR v1.5 test failed to detect them. Based on these data, the LCx HIV assay provides a desirable alternative to existing commercial assays for monitoring viral loads. The high level of nucleotide conservation within the target region allows quantification of even the most genetically diverse HIV-1 isolates, those of group O. In an era of increasing numbers of infections with recombinant forms of HIV-1 and an ever-changing global distribution of subtypes, the subtype- and group-independent performance of the LCx HIV assay makes it a valuable method for monitoring of patients.