Comparison of the Aptima HIV‐1 Quant Dx assay with the COBAS AmpliPrep/COBAS TaqMan HIV‐1 v2.0 Test for HIV‐1 viral load quantification in plasma samples from HIV‐1–infected patients

Abstract Background and aims HIV‐1 RNA viral load (VL) in plasma samples of HIV‐1–positive patients is used to assess the level of viral replication, the risk of disease progression, and the response and efficacy to antiretroviral treatment. Knowing the performance of different tests for HIV‐1 RNA detection is, therefore, important for clinical care. This study compared the performance of the recently introduced Aptima HIV‐1 Quant Dx assay (Hologic, Inc) and the standard COBAS AmpliPrep/COBAS TaqMan HIV‐1 v2.0 Test (CAP/CTM2) (Roche Molecular System, Inc) for HIV‐1 RNA quantitation. Methods Assay performance was assessed using 335 clinical samples, a standard HIV‐1 low VL panel, and 2 diluted samples from well‐characterized patients infected with different HIV‐1 subtypes tested in 5 replicates over 3 days. All samples were tested on both assays to evaluate inter‐assay agreement, both qualitatively and quantitively. Altogether, we evaluated assay sensitivity, linearity, accuracy, precision, repeatability, and reproducibility. Results Assay agreement for qualitative results in 335 clinical samples was fair (80.6%). Correlation of quantitative assay results (n = 164) was excellent (R 2 = 0.97), with 96.3% of the results within the 95% limit of assay agreement (−0.42 to +0.86 log), and 98.8% within 1 log of each other. Aptima‐HIV‐1 yielded results, on average, 0.22 log higher than CAP/CTM2. Both assays accurately quantitated the HIV‐1 standard at low VL (R 2 ≥ 0.94), with all samples within 0.5 log of the target. Conclusion Aptima‐HIV‐1 assay demonstrated sensitivity, accuracy, reproducibility, and precision for the detection and quantitation of HIV‐1 RNA across a wide dynamic range of VLs. Its performance, together with full automation and high throughput, suggests that Aptima‐HIV‐1 could be a suitable assay for reliable monitoring of HIV‐1 VL in patients undergoing treatment.


| INTRODUCTION
The introduction of new antiretroviral agents in the last decade has significantly improved the efficacy and safety of antiretroviral therapy (ART) in HIV-1-infected patients. 1 Besides clinical and immunological monitoring, which are used as complementary evaluations, HIV-1 RNA quantitation in patients' plasma samples is currently considered the main approach to monitor ART compliance and success. [2][3][4] Optimal control of HIV-1 infection is reached when the complete viral suppression achieved persists over time. Even if full viral suppression is achievable in most patients (both treatment-naïve and experienced), some show a transient, low viremia ("blips"). 5,6 Some blips might be considered artifactual variations in viral load (VL) because of assay variability and laboratory processing inconsistencies, 7 not associated with an increased risk of treatment failure or drug resistance. Key treatment decisions made at VLs ranging from 1.7 to 3 log copies/mL need accurate monitoring in clinical samples and, consequently, require a highly sensitive, precise, and reproducible HIV-1 RNA quantitation assay. Other important attributes of a VL assay include the ability to quantitate HIV-1 RNA precisely over a wide range of VLs and an equally good performance on all HIV-1 subtypes. As assay agreement on quantitative values is generally good at high VLs but tends to decrease substantially at low VLs, [8][9][10][11]

| HIV-1 samples
The 335 plasma samples collected in EDTA tubes were tested side-byside in the Aptima-HIV-1 and CAP/CTM2 assays, without further criteria for selection other than available sample volume.
Tubes were centrifuged for 10 minutes at 1000 to 3000 g for plasma preparation. All samples were first tested with the CAP/ CTM2 assay.
If the residual plasma volume was ≥1.2 mL, the same samples were immediately tested in primary tubes on the Hologic Panther instrument. For samples with less than 1.2 mL residual plasma volume, 0.70 mL plasma was transferred to Hologic specimen aliquot tubes.
Among the samples evaluated in the study, 248 specimens were derived from HIV-1 patients infected with B HIV-1 strains and 87 samples from other subtypes (A, C, F, G, and CRFs), characterized by phylogenetic analysis of HIV-1 pol gene (RT and PR). 14,15 In particular, 12 samples belonged to subtypes A, 9 to subtypes C, 23 to subtypes F, 14 to subtypes G, and 29 were circular recombinant forms (CRFs).

| HIV-1 VL assays
Samples in the 2-assay platforms were processed and tested by trained operators, by Aptima HIV-1 Quant Dx assay (cat. no. PRD-03000) and Roche CAP/CTM2 (cat. no. 05212294190) according to the assay manufacturers' package inserts.

| Aptima-HIV-1 assay
All the samples were tested in specimen aliquot tubes. Samples were loaded onto the Panther system (Hologic, Inc). HIV-1 genomic RNA was first released using target capture technology and then bound to magnetic particles. The Aptima HIV-1 Quant assay uses the TMA method to amplify 2 regions of HIV-1 RNA (pol and LTR) from the sample and amplifies and detects the amplified targets, all in an automated manner.
The assay's reported lower limit of quantification (LLOQ) is 1.47 log copies/mL, and its upper limit of quantitation is 7 log copies/mL (Hologic Inc, PI). The reported limit of detection (LoD) of the Aptima-HIV-1 assay is 12 cp/mL. Panther system allows random access testing of various analytes, processing up to 275 samples in an 8-hour shift.
The system provides results from 120 samples in about 2.5 hours.

| CAP/CTM2 assay
All the samples were tested in Roche S-tubes. The sample volume used was 1 mL. Tubes were loaded onto the Cobas Ampliprep instrument, which extracts HIV-1 LTR and gag targets from the sample. Tubes were then transferred to the COBAS Taqman Analyzer (Roche Molecular Systems, Inc, cat. no. 03121453001), which amplifies and detects the target sequence in an automated fashion. The reported assay's LLOQ is 1.39 log copies/mL, and its upper limit of quantitation is 7 log copies/mL (Roche Inc. PI). The reported LoD of this assay is 20 cp/mL. The CAP/CTM platform has an initial capacity for 72 samples with continuous feeding, which allows 168 samples (1 mL/sample) to be processed per 8-hour shift. This system returns results in 4.5 hours.

| Assay assessment using an external quality panel by Aptima-HIV-1 assay
The Acrometrix HIV-1 linearity panel (ThermoFisher Scientific, Benicia, California, cat. no. 950470) was used to evaluate both the assays' lin-

| Data analyses
VL values were expressed as log copies/mL. Agreement of the assays' qualitative results (ie, defining samples as "negative," "detected <LLOQ," and "quantitated") was determined using a tabular format. For a very small number of plasma samples with discordant results between the 2 assays (eg, negative or detected <LLOQ in one assay and quantitated in the other assay), the patients' immunological data CD4 count and CD4/CD8 ratio, determined as previously described, 15

| Assay performance comparison in clinical samples
Assay performance was compared using all 335 clinical samples (248 HIV-1 type B and 87 HIV-1 non-B samples) with VLs identified by CAP/CTM2, ranging from undetectable (105/335) to detectable HIV-1 RNA amounts (up to 7 log copies/mL). The numbers in circles indicate the agreement between the assays' qualitative results.  Deming regression analysis ( Figure 1A).
Only 2 samples ( Figure 1B) Figure 1B). Of the 6 outlier samples (outside the limit of agreement), three differed by <0.5 log in both assays, one differed by >0.5 log but <1 log, and two by >1 log but <1.5 log.
As shown in Table 1

| HIV-1 subtypes
Results obtained by the analysis of 87 samples from HIV-1 patients infected with subtypes A, C, F, G, and CRFs showed that similar findings could be obtained from samples derived from HIV-1-infected subjects with subtypes A and C, irrespective of HIV-1 RNA amounts, with no significant difference between the assays used. On the other hand, Aptima HIV-1 Quant DX assay was able to detect higher level of viral replication in samples containing HIV-1 subtype F, G, and CRFs, revealing important differences (≥0.5 log) in 13 samples (5 belonging to subtype F, 3 to subtype G, 5 to CRFs), as shown in Table 3.

| Assay accuracy with a standard panel
The assays' ability to accurately quantitate HIV-1 RNA at low VLs was evaluated using the Acrometrix standard at target concentrations ranging from 1.2 to 2.8 log copies/mL. Assay results showed a very good precision, with all assay results differing by <0.5 log copies/ mL from the target values, and excellent linearity (R 2 ≥ 0.94) ( Figure 2).

| Repeatability, reproducibility, precision, and linearity with patient's samples of 2 subtypes
Two well-characterized 14,15 clinical samples (subtype B and F) were tested by Aptima-HIV-1 to determine within-run repeatability, between-run reproducibility, precision, and linearity of the assay.
Within-run repeatability was substantially good, as reflected by an SD ≤0.16 for the 5 replicates tested (Table 4). Between-run repeatability was good, with 23 of the 24 results, obtained on 3 different days, being within 0.5 log of each other (results for the subtype F at low VL on day 1 differed by >0.5 but <1 log from results on day 2 or 3). Overall precision of the assay was excellent, with 23/24 %CV <4% and all %CVs being <8%.
Accuracy was also excellent, with 22 of 24 results being within 0.5 log of the target, and 2 results differing by >0.5 but <1 log from the target. Linearity was excellent, as demonstrated by all correlation coefficients >0.98 (data not shown).

| DISCUSSION
As HIV-1 VL monitoring has become the cornerstone for the management of HIV-1-positive patients during their lifelong treatment regimen(s), the assays used to measure HIV-1 VL must be highly sensitive, specific, accurate, and precise. In the present study, the TMA-based Aptima-HIV-1 assay demonstrated sensitivity, reproducibility, and precision for the detection and quantitation of HIV-1-RNA across a wide dynamic range of VLs (including very low VLs).
Assay agreement for qualitative results in 335 clinical samples was fair (80.6%). Quantitative results for Aptima-HIV-1 and CAP/CTM were highly correlated (R 2 = 0.97) and only 2 of 164 samples quantitated by both assays had results that differed by >1 log copies/mL but <1.5 log copies/mL. Although Aptima-HIV-1 results were on average slightly higher than CAP/CTM2 results (by 0.22 log), the difference between assay results was minimal at low VLs (0.05 at VLs 1-1.99 log; 0.04 at VLs 2-2.99 log, and 0.02 at VLs 3-3.99 log).
Among the 65 discordant results, most samples did not show substantial variations. Indeed, in 45 samples the differences were very small (less than 30 or 20 copies/mL by one test and undetectable by the other test) and may not be considered as real conflicting results.
In fact, optimal viral suppression is generally defined as a VL persistently below the level of detection (HIV RNA <20 to 75 copies/mL, depending on the assay used). 2  detectable VL of 50 to 400 copies/mL is preceded or followed by an undetectable result, testing the sample again is recommended to avoid an assay artifact or to establish a true viral rebound. [16][17][18] Moreover, the occurrence of so called viral blips (50-400/1000 copies/mL) during treatment are important events, which could be misinterpreted as treatment failure and hence may lead to a change in medication, since the goal of ART is VL suppression to TND or <50 copies/mL.
While we cannot rule out the probability that proviral HIV-1 genomic sequences in the plasma could be responsible for the blips (eg, originating from latently infected cells in the pellet), the small number of samples prevents any definitive conclusion.
Combined with full automation, high throughput, and superior workflow, 31 Aptima-HIV-1 is suitable for VL monitoring of HIV-1 patients during treatment.

CONFLICT OF INTERESTS
The authors declare that they have no conflict of interests.

ETHICAL APPROVAL
The study was approved by the local ethics committee (737/2016), and the patients provided written informed consent.