|Home | About | Journals | Submit | Contact Us | Français|
HIV-1 RNA quantitation in plasma, or virus load testing, is the primary method by which the response to antiretroviral therapy is monitored. Here we describe evaluation of the Aptima HIV-1 Quant Dx assay (Aptima) performed on the automated Panther system. The clinical performance of Aptima was compared to that of the Cobas AmpliPrep/Cobas TaqMan HIV-1 Test v2.0 (CAP/CTM) using 162 EDTA plasma samples collected from patients undergoing HIV-1 monitoring. Overall agreement was 84.0% (136/162), with a kappa statistic of 0.723 (standard error, 0.047; 95% confidence interval [CI], 0.630 to 0.815), indicating substantial agreement. Using the 86 clinical samples quantifiable by both methods, Passing-Bablok regression revealed a regression line of Y = (1.069 × X) − 0.346 (95% CI of the slope [1.003 to 1.139] and intercept [−0.666 to −0.074]), and Bland-Altman analysis demonstrated a mean difference (Aptima-CAP/CTM) of −0.075 log10 copies/ml (95% limits of agreement of −0.624 to 0.475), consistent with negative bias. Comparison of Aptima results for paired dried blood spot (DBS) and plasma specimens archived from participants in the Peninsula AIDS Research Cohort Study (PARC) demonstrated an overall agreement of 94.7% (90/95) when 1,000 copies/ml was used as the threshold. In conclusion, the Aptima HIV-1 Quant Dx assay provides a suitable alternative for HIV-1 monitoring in plasma and DBS.
Infection with human immunodeficiency virus type 1 (HIV-1) remains a significant global health challenge. The World Health Organization (WHO) estimates that there were 36.9 million people worldwide living with HIV in 2014, 2 million new infections that year, and 1.2 million deaths related to AIDS (1). Similar to the guidelines endorsed by the U.S. Department of Health and Human Services (USDHHS) (2), the WHO now recommends antiretroviral therapy (ART) for all HIV-infected patients regardless of their CD4 T-lymphocyte count or HIV-1 RNA plasma virus load (3). The evidence for this recommendation was further strengthened by the results of two randomized controlled trials, START and TEMPRANO, which demonstrated the clinical benefit of initiating ART in patients with pretreatment CD4 cell counts of >500 cells/mm3, instead of a lower CD4 cell count threshold (4, 5).
The efficacy of ART to reduce morbidity and mortality, as well as prevent transmission, has led the Joint United Nations Programme on HIV/AIDS (UNAIDS) to propose new global targets for treatment scale-up. These UNAIDS targets call for 90% of people living with HIV to be diagnosed, 90% of people diagnosed with HIV to receive sustained ART, and 90% of those on ART to have viral suppression by 2020 (6). A critical component of this “90-90-90” plan is the availability of laboratory testing to monitor for both viral suppression and therapy failures.
HIV-1 RNA quantitation or virus load testing is the primary method by which the response to ART is monitored. In the United States, virus load is recommended for all HIV-infected patients at entry into care, at initiation or modification of therapy, and on a regular basis thereafter, typically at 3- to 4-month intervals that are extended for virally suppressed, adherent patients (2). Globally, the WHO recommends testing every 6 to 12 months, or based on immunological or clinical criteria, if access to testing is limited (7). The USDHHS defines viral suppression as a virus load persistently below the level of detection. This value may vary substantially based on the test used for monitoring, ranging from ~20 to 75 copies/ml. However, the U.S. AIDS guidelines and the AIDS Clinical Trials Group (ACTG) define virologic failure as an HIV-1 RNA level of >200 copies/ml of plasma (2). This threshold is thought to reduce unnecessary changes in ART due to virus load “blips” or assay variability. The WHO currently sets their threshold for treatment failure at ≥1,000 HIV-1 RNA copies/ml of plasma (7).
HIV-1 RNA quantitation is routinely performed using nucleic acid amplification techniques. In the United States, the most commonly performed virus load test is the Cobas AmpliPrep/Cobas TaqMan HIV-1 Test v2.0 (CAP/CTM) (Roche Molecular Systems, Pleasanton, CA), a U.S. Food and Drug Administration (FDA) in vitro diagnostic (IVD) that provides high-throughput, sample-to-answer testing with real-time, reverse transcriptase PCR (RT-PCR)-based quantitation. This assay is well studied and provides a benchmark for virus load testing performance in the United States and globally (8,–24). A recently developed test for virus load is the Aptima HIV-1 Quant Dx assay (Aptima) performed on the fully automated Panther system (both from Hologic Inc., San Diego, CA). This new assay relies on real-time transcription-mediated amplification (TMA) for HIV-1 RNA quantitation, and limited data are available on its performance compared to that of CAP/CTM (25, 26).
The global scale-up of ART and the concomitant increased requirement for virus load monitoring represent an opportunity to build laboratory capacity in both resource-poor and resource-rich settings. Therefore, the process of implementing and scaling up HIV-1 testing programs requires careful consideration of testing platforms, including throughput, footprint, automation, turnaround time, relation to care (point-of-care versus laboratory-based testing), and cost per test, in addition to the analytical and clinical test performance characteristics (27). Furthermore, the large number of virus load tests that will be necessary to support the UNAIDS goals will likely require the use of a variety of testing strategies, including monitoring suppression through centralized testing of dried blood spots (DBS). The DBS approach allows increased accessibility of virus load testing in remote, low-resource settings, as DBS are simple and inexpensive to both collect and transport (28). Moreover, modeling data suggest that virus load monitoring using DBS provides a mechanism for the cost-effective delivery of ART in low-resource settings (29). In this report, we describe the evaluation of Aptima and CAP/CTM for the quantitation of HIV-1 in plasma and DBS, with particular emphasis on the thresholds for clinical decision-making.
This study was reviewed and approved by the Institutional Review Board of Stanford University.
The Aptima HIV-1 Quant Dx assay (Aptima) was performed on the automated Panther system (Hologic Inc., San Diego, CA). Though Conformité Européene in vitro diagnostic (CE-IVD) approval was obtained, this test was not approved by the U.S. Food and Drug Administration (FDA) at the time of the study. This assay is composed of three steps: (i) target capture, (ii) amplification of the HIV-1 long terminal repeat (LTR) and pol gene (integrase) by transcription-mediated amplification (TMA), and (iii) real-time amplicon detection using fluorescently labeled probes. Quantitation is achieved by monitoring the time taken to reach a fluorescent threshold value, which is proportional to the starting HIV-1 concentration. Each reaction mixture includes an internal calibrator/internal control (IC) that is used to account for the effects of inhibition and control for variations in specimen processing and TMA. A single positive calibrator is run in triplicate, along with one replicate each of the low positive, high positive, and negative controls, each time a reagent kit is loaded on to the Panther system. The calibrator and controls are then valid for 24 h. Sample aliquot tubes require an input volume of 0.7 ml, and 0.5 ml is processed by the Panther system. For samples for which multiple replicates were to be tested, a larger volume was transferred to a single tube and multiple tests were ordered from the Panther software interface. The protocol for Aptima DBS testing differed from the plasma protocol only in that there was a 30-min room temperature preincubation of a single DBS in 1.0 ml of sample transport medium provided by Hologic. The Panther system reports values for results of ≥30 and ≤10,000,000 copies/ml (1.47 to 7.00 log10 copies/ml), and low-level positives that fall below this range are reported as detection of <30 copies/ml (<1.47 log10 copies/ml).
To estimate the HIV-1 RNA copies per milliliter of EDTA plasma obtained from DBS testing, the HIV-1 RNA copies per milliliter of sample transport medium obtained from the Panther instrument was multiplied by a factor of 34.48. This value was obtained using the following equation: volume of sample transport medium, in microliters/[volume of whole blood collected on DBS, in microliters, × (1 − estimated hematocrit)]. The hematocrit value was estimated to be 42% based on the mean adult value from the National Health and Nutrition Examination Survey (NHANES) (30).
The Cobas AmpliPrep/Cobas TaqMan HIV-1 Test v2.0 (CAP/CTM) (Roche Molecular Systems, Pleasanton, CA) is comprised of three major steps: (i) automated specimen processing and assay setup, (ii) reverse transcription of the HIV-1 LTR and gag gene, and (iii) real-time PCR amplification and detection of target cDNA using fluorescently labeled hydrolysis probes. Each reaction mixture includes an HIV-1 Quantitation Standard Armored RNA that compensates for the effects of inhibition and controls for variations in specimen processing and reverse transcriptase PCR. One replicate each of the low positive, high positive, and negative controls is run each time a sample rack (24 positions) is loaded on to the Cobas AmpliPrep system, thus allowing testing of up to 21 clinical samples per rack. Sample aliquot S-tubes require an input volume of 1.0 ml, and 0.8 ml is processed by the Cobas AmpliPrep system.
An undocked CAP/CTM system was used in this study, so reaction mixtures were transferred manually to the Cobas TaqMan analyzer for amplification and detection. The AmpliLink software (ver 3.3) reports values for results of ≥20 and ≤10,000,000 copies/ml (1.30 to 7.00 log10 copies/ml), and low-level positives that fall below this range are reported as <20 copies/ml (<1.30 log10 copies/ml).
Analytical characterization of Aptima and CAP/CTM, including linear range, precision, lower limit of detection, and performance against external quality assessment panels, is provided in the supplemental material. For all experiments in which Aptima and CAP/CTM were compared, testing was performed on the same day and with clinical specimens that had undergone the same number of freeze-thaw cycles.
For the comparison of Aptima and CAP/CTM, this study utilized 162 deidentified clinical specimens that were submitted to the Stanford Health Care (SHC) Clinical Virology Laboratory between October 2012 and September 2014 for HIV-1 testing. All specimens were collected as EDTA plasma and stored at −80°C prior to testing. Specimens were selected if sufficient volume was left for both the Aptima and CAP/CTM. Retrospective virus load information was collected to ensure that specimens were selected that spanned the assays' linear ranges. Ninety-five paired DBS and EDTA plasma specimens were obtained from participants in the Peninsula AIDS Research Cohort Study (PARC) (31) and were archived at −80°C prior to testing. DBS were composed of 50 μl of EDTA whole blood collected on Whatman 903 sample collection cards.
Passing-Bablok, Deming, and Bland-Altman analyses were performed in R using the mcr package (https://cran.r-project.org/web/packages/mcr/index.html), and plots were generated using Prism (GraphPad Software, Inc., La Jolla, CA).
A total of 162 EDTA plasma samples from HIV-1-infected patients were tested for HIV-1 RNA using Aptima and CAP/CTM. Results of this comparison are shown in Table 1. Overall agreement was 84.0% (136/162), with a kappa statistic of 0.723 (standard error, 0.047; 95% confidence interval [CI], 0.630 to 0.815), indicating substantial agreement. Five samples that were detected at <30 copies/ml by Aptima were not detected by CAP/CTM. Similarly, 6 samples that were detected at <20 copies/ml by CAP/CTM were not detected by Aptima. In addition, there were a total of 15 samples quantitated by CAP/CTM that were either not detected (4 samples) or detected at <30 copies/ml (11 samples) by Aptima. The mean concentration for these samples by CAP/CTM was 1.87 log10 copies/ml (standard deviation [SD] = 0.32). Using a composite reference, in which detection by either method was considered a true positive, Aptima and CAP/CTM had similar sensitivities (92.3% [120/130] and 96.2% [125/130], respectively; P = 0.287).
To investigate the quantitative agreement between Aptima and CAP/CTM, the log10 concentrations of the 86 clinical samples quantifiable by both methods were plotted against one another and Passing-Bablok regression was performed (Fig. 1A). This analysis resulted in a regression line of Y = (1.069 × X) − 0.346. The 95% confidence intervals of the slope (1.003 to 1.139) and intercept (−0.666 to −0.074) do not include 1 and 0, respectively, indicating that Aptima shows positive proportional bias and negative systematic bias compared to CAP/CTM. Deming regression resulted in a similar regression line [Y = (1.045 × X) − 0.249] and similar 95% confidence intervals of the slope (0.991 to 1.010) and intercept (−0.473 to −0.025). Next, the differences in log10 concentrations were plotted against the average values to generate a Bland-Altman plot (Fig. 1B). This analysis revealed a bias of −0.075 log10 copies/ml (Aptima − CAP/CTM) and 95% limits of agreement of −0.624 to 0.475. These samples were comprised of 1 subtype C, 3 subtype CRF01-AE, and 82 subtype B viruses. Two samples harbored virus with integrase drug resistance mutations, one with the major N155H mutation and the other with the minor E157Q mutation. Quantitation of these samples fell within the 95% limits of agreement. All outliers were subtype B viruses without known integrase resistance mutations. These outliers demonstrated log10 differences (Aptima − CAP/CTM) of −0.805, −0.786, −0.746, and 0.535, respectively.
The Aptima was then used to evaluate 95 paired DBS and EDTA plasma samples archived from HIV-infected patients on therapy (31). Applying the WHO-recommended threshold of 1,000 copies/ml threshold for treatment failure, DBS and plasma demonstrated an overall agreement of 94.7% (90/95) (Table 2). There were five discrepancies: 3 pairs showed <1,000 copies/ml in DBS and ≥1,000 copies/ml in plasma, whereas 2 pairs showed ≥1,000 copies/ml in DBS and <1,000 copies/ml in plasma (Table 3). Twelve samples were quantitated at ≥1,000 copies/ml in both specimens, with plasma values ranging from 3.73 to 5.80 log10 copies/ml (mean = 4.90; SD = 0.62). For these samples, the mean difference (plasma − DBS) was 0.17 log10 copies/ml (95% CI, −0.48 to 0.83).
Here we describe the performance characteristics of the Aptima HIV-1 Quant Dx assay performed on the Panther system compared with the Cobas AmpliPrep/Cobas TaqMan HIV-1 Test v2.0 for the quantitation of HIV-1 RNA in EDTA plasma. We evaluated the Aptima DBS protocol using the WHO-recommended threshold of 1,000 copies/ml for treatment failure.
Overall, Aptima and CAP/CTM gave comparable quantitations of HIV-1 RNA. The evaluation of 162 clinical specimens revealed substantial classification agreement and no statistical difference in clinical sensitivity. Both Passing-Bablok and Deming regression analyses of the samples quantifiable by both methods indicated that Aptima shows a slight negative systematic bias compared to CAP/CTM. A slight negative bias was also observed by Bland-Altman analysis, though the mean difference was just −0.075 log10 copies/ml. Given the level of imprecision at clinical decision points and the significant quantitative agreement in clinical EDTA plasma samples, the analytical differences in HIV RNA quantitation should have limited to no clinical impact.
While several studies have compared Aptima to a variety of HIV-1 virus load tests (25, 26, 32,–35), to date, only the studies from Hopkins et al. (25) and Nair et al. (26) have compared Aptima with CAP/CTM. Consistent with the evaluation presented here, in those studies linearity was confirmed using commercially available material, similar levels of precision were observed at low virus loads, and there was good agreement between methods for both qualitative and quantitative results. An advantage of the current study is that preanalytical sources of variability were reduced by carrying out Aptima and CAP/CTM comparison testing on the same day and with samples that had undergone the same number of freeze-thaw cycles.
Importantly, this report also provides an evaluation of Aptima for the semiquantitative detection of HIV-1 RNA in paired DBS and EDTA plasma samples from patients on ART. Using the WHO-recommended threshold of ≥1,000 copies/ml for therapy failure, we observed excellent agreement between DBS and EDTA plasma. These results, coupled with the simple Panther DBS work flow, suggest that DBS testing using Aptima is a promising option to meet the increased demand for virus load monitoring in low-resource settings necessitated by the global scale-up of ART (29, 36). Furthermore, the specificity of TMA for RNA amplification may be of particular advantage for Aptima compared to RT-PCR methods for DBS virus load monitoring, given the confounding presence of proviral HIV-1 DNA in whole blood (37, 38). Future studies will be required to compare Aptima testing to other methods of virus load monitoring from DBS.
Limitations include a clinical sample set comprised primarily of HIV-1 subtype B and an insufficient number of samples with integrase mutations, which precluded detailed analyses of the impact of subtype and integrase sequence variants on Aptima performance, though a negative impact is not expected due to the assay's dual-target design. In addition, DBS were collected from a cohort of primarily suppressed patients in the United States. Studies are ongoing to evaluate Aptima DBS testing performance and cost-effectiveness in low-resource African settings with high levels of therapy failure.
In conclusion, we report the evaluation of the Aptima HIV-1 Quant Dx assay performed on the Panther system. This assay demonstrated performance comparable to that of the Cobas AmpliPrep/Cobas TaqMan HIV-1 Test v2.0 on HIV-1 subtype B samples. Aptima provides a suitable alternative for HIV-1 monitoring, particularly for laboratories already using the Panther system for human papillomavirus (HPV), Chlamydia trachomatis, and Neisseria gonorrhoeae testing. As additional performance data are collected for non-subtype B clinical samples, the Panther system's simple, automated work flow, ease of use, high throughput, and accurate DBS protocol may provide an attractive option for ministries of health as they consider various approaches to meet the virus load testing demands of the UNAIDS 90-90-90 plan (6). Future time and motion studies (work flow) and comparative cost analyses will be needed to determine the optimal platform, or combination of platforms, to effectively achieve this virus load scale-up. The Aptima HIV-1 Quant Dx assay and other assays on the Panther system may ultimately deliver the combination of test performance and overall efficiency required to significantly impact the global HIV epidemic, as well as improve the diagnosis of other sexually transmitted infections.
We thank Jesse Waggoner for providing assistance with study design and statistical analysis.
Financial support for this research was provided by Hologic, Inc.
The funders had no role in data collection and analysis, decision to publish, or preparation of the manuscript.
B.A.P. has received lecture fees from Hologic, Inc.
Supplemental material for this article may be found at http://dx.doi.org/10.1128/JCM.01569-16.