Performance of the Abbott RealTime HIV-1 Viral Load Assay Is Not Impacted by Integrase Inhibitor Resistance-Associated Mutations

Abstract

The Abbott RealTime HIV-1 viral load assay uses primers and probes targeted to integrase, which is also the target of integrase inhibitors such as raltegravir. Viral loads of 42 raltegravir-susceptible and 40 raltegravir-resistant specimens were determined using RealTime HIV-1 and Roche Monitor (v1.5). The differences in viral load measurements between assays were comparable in the two groups, demonstrating that the RealTime HIV-1 assay can tolerate raltegravir-selected mutations.

The addition of the integrase (IN) strand transfer inhibitor (INSTI) raltegravir (RAL) to the list of approved antiretroviral drugs (ARVs) for treatment of HIV-1 infection has enabled many patients failing existing treatment regimens to achieve viral load (VL) suppression (10). As for most ARVs, VL rebound during RAL treatment is typically associated with the development of one or more mutations in the target enzyme. RAL resistance is associated with changes at positions 92, 143, 148, and 155 and other, secondary locations in IN (6, 8).

The Abbott RealTime HIV-1 assay (17) utilizes a novel partially double-stranded linear probe design and optimized cycling conditions and targets a highly conserved region of IN. The assay was designed to tolerate polymorphisms whether they occur naturally or through drug selection and has been shown to perform well on genetically and geographically diverse strains of HIV-1 (15, 16).

INSTI resistance-associated mutations (RAMs) are predominantly located in the catalytic core domain; the list of positions where changes associated with resistance to RAL occur includes 51, 66, 68, 74, 92, 95, 97, 121, 128, 138, 140, 143, 148, 151, 155, 157, 163, 203, and 206 (6, 7, 11, 13). No RAM sites are located within the RealTime HIV-1 probe or reverse primer regions; only two potential RAMs (143 and 148) are within the forward primer (Fig. 1). A previous study utilizing genetically engineered RNA transcripts with RAMs at position 143 or 148 revealed no impact on assay performance (5).

Fig. 1.

Schematic diagram of HIV-1 integrase. Mutations associated with resistance to INSTI (7, 11, 13) are shown above; primary RAL mutations are in bold type. Approximate locations of the RealTime HIV viral load assay forward (F) and reverse (R) primer and probe (P) binding sites are indicated. Only the forward primer is located in an area potentially affected by integrase inhibitor resistance.

(This work was first presented at the 50th Interscience Conference on Antimicrobial Agents and Chemotherapy, Boston, MA, 12 to 15 September 2010 [18].)

The objective of the present study was to investigate whether there is any discernible impact of RAL RAMs on the performance of the RealTime HIV-1 assay utilizing actual clinical specimens. Eighty-two plasma samples previously submitted for RAL susceptibility assessment using the PhenoSense Integrase assay (1, 2, 9) (Monogram Biosciences, South San Francisco, CA) were tested with the Abbott RealTime HIV-1 (RealTime HIV-1; Abbott Molecular Inc., Des Plaines, IL) and Roche Cobas Amplicor HIV-1 Monitor v1.5 (Monitor v1.5; Roche Molecular Systems, Inc., Branchburg, NJ) VL assays. Many of the specimens were collected and stored frozen in plasma preparation tubes (PPT; Becton, Dickinson & Co). Sequence data covering the RealTime HIV-1 target region were obtained for all samples (GeneSeq Integrase; Monogram Biosciences, South San Francisco, CA). Consensus sequences for each sample were compared to the NL4-3 reference to identify RAL RAMs. Nucleotide mismatches relative to the RealTime HIV-1 primers/probe were assessed. Correlation data were stratified based on the fold change (FC) in RAL 50% effective concentration (EC₅₀) being below or above the biological cutoff of 1.5 (2) to evaluate the impact of both naturally occurring polymorphisms and RAL RAMs on Abbott RealTime HIV-1 assay performance relative to that of Monitor v1.5. To assign subtype, IN sequences were aligned against HIV-1 group M reference strains (http://www.hiv.lanl.gov) and phylogenetic reconstructions were generated using PHYLIP v3.573 (J. Felsenstein, University of Washington, Seattle, WA). All sequences were subtype B.

Viral load results within the dynamic range for both assays were obtained for 82 samples (Table 1). As previously reported (15), RealTime HIV-1 results tended to be lower than those from the Monitor v1.5 assay (mean difference of 0.23 log₁₀ copies/ml, paired t test, P < 0.0001). Correlation between VL measurements was high (R² = 0.86) (Fig. 2A). VL values for 69 samples (84%) were within 0.5 log₁₀ copies/ml; values for 80 samples (98%) were within 1 log₁₀ copies/ml. Based on the PhenoSense Integrase assay, 42 samples were RAL sensitive (FC ≤ 1.5) and 40 samples had FCs equal to or greater than the biological cutoff of 1.5 (Table 1). For the 42 RAL-sensitive samples, the correlation was lower (R² = 0.79) (Fig. 2B); values for 37 samples (88%) were 0.5 log₁₀ copies/ml, values for 40 samples (95%) were within 1.0 log₁₀ copies/ml, and values for 2 samples (5%) were within >1.0 log₁₀ copies/ml. Among the 40 samples with reduced RAL susceptibility, the correlation was high (R² = 0.90) (Fig. 2C); 32 (80%) had VL results within 0.5 log₁₀ copies/ml and all were within 1.0 log₁₀ copies/ml. Ten of the 40 samples with reduced RAL susceptibility had the N155H mutation, which is located outside the RealTime HIV-1 assay primer or probe binding sites, as the only RAM. However, 30 of the 40 samples contained mutations at position 143 or 148, located within the RealTime HIV-1 forward primer site; 3 of these samples contained mixtures of up to 4 possible amino acids at position 143. Among these 30 samples, which are the most likely to contain sequences that have reduced binding affinity for any of the RealTime HIV-1 primers or probes, correlation of VL values between assays remained high (R² = 0.90) (Fig. 2D). Notably, VL values for 26 of the 30 (86.7%) were within 0.5 log₁₀ copies/ml.

Table 1.

Viral load results in groups of clinical specimens grouped by RAL resistance pattern

Group	n	VL differencea
		Mean	Minimum	Maximum
All	82	−0.23	−1.80	0.83
RAL FC ≤ 1.5 (no RAMs)	42	−0.28	−1.80	0.23
RAL FC > 1.5	40	−0.18	−0.92	0.83
N155H	10	−0.26	−0.64	0.17
Any RAM at position 143 or 148	30	−0.15	−0.92	0.83
Q148H	17	−0.13	−0.64	0.23
Q148R	6	−0.42	−0.92	−0.07
Q148H or -R	23	−0.21	−0.92	0.23
Y143R	4	0.00	−0.23	0.19
Y143C, -R, or mix	7	0.04	−0.41	0.83

^aLog₁₀ copies/ml for RealTime HIV-1 minus the value for Monitor v1.5.

Fig. 2.

Scatter plots of viral load measurements from the RealTime HIV assay (x axis) and Monitor v1.5 assay (y axis). (A) All samples (n = 82). (B) Samples with RAL fold changes of ≤1.5 (n = 42). (C) Samples with RAL fold changes of >1.5 (n = 40). (D) Samples with RAL fold changes of >1.5 and with RAMs at position 143 or 148 detected (n = 30). The median RAL fold change values (25th percentile, 75th percentile) for these 4 groups of samples are as follows: 1.2 (0.93, >100) (A), 0.93 (0.78, 1.1) (B), >100 (45, >100) (C), and >100 (>100, >100) (D).

Bland-Altman analysis for all samples showed a mean bias (RealTime HIV-1 minus Monitor v1.5) of −0.23 log₁₀ copies/ml (95% limits of agreement, −0.88 to 0.42) (Fig. 3A); for RAL-sensitive samples, it was −0.28 log₁₀ copies/ml (95% limits of agreement, −0.94 to 0.38) (Fig. 3B), and for samples with reduced RAL susceptibility, the bias was −0.18 log₁₀ copies/ml (95% limits of agreement, −0.81 to 0.46) (Fig. 3C). For the subset of samples with mutations at position 143 or 148, the mean bias was −0.15 log₁₀ copies/ml (95% limits of agreement, −0.79 to 0.49) (Fig. 3D). Mean differences in VLs for subgroups defined by specific RAL RAM patterns did not reveal any statistically significant differences (Mann-Whitney test, P > 0.05, for all pairwise comparisons between groups) (Table 1).

Fig. 3.

Bland-Altman plots of viral load measurements; differences in viral loads as determined with Monitor v1.5 versus RealTime HIV (y axis) versus the averages of the paired results (x axis). (A) All samples (n = 82). (B) Samples with RAL fold changes of ≤1.5 (n = 42). (C) Samples with RAL fold changes of >1.5 (n = 40). (D) Samples with RAL fold changes of >1.5 and with RAMs at position 143 or 148 detected (n = 30). The Fig. 2 legend gives RAL fold change ranges in each group.

The two samples with the most divergent VL measurements (log₁₀ differences in VL of >1.0) both were RAL sensitive and lacked RAL RAMs. Further sequence analysis revealed no molecular basis for the discrepancy in VL results. However, as many of these samples were stored frozen in PPT, the explanation might be related to previous reports of VL overestimation by the Monitor 1.5 assay under these conditions (3, 4, 12, 14).

A limitation of this study is that not all mutations associated with RAL resistance were represented in the samples tested. However, no other known RAL RAM is located in the primer or probe binding sites, and furthermore, these RAMs are less common than those observed here in RAL-treated patients. Formal confirmation that other INSTI RAMs do not affect the accuracy of the RealTime HIV-1 VL assay or that of other VL assays targeting the IN region awaits additional studies. Of note, evaluation of genetically engineered transcripts containing all possible permutations of nucleotide substitutions resulting in INSTI RAMs at positions 143, 146, 147, and 148 demonstrated no impact on assay performance (5).

In conclusion, analysis of RAL RAMs revealed that very few are localized in the RealTime HIV-1 assay target region. In fact, RAMs were identified at only two codons within the forward primer site. The data presented here demonstrate that the RealTime HIV-1 assay has the capacity to tolerate RAL-selected mutations with no apparent impact on assay performance.