Current treatment guidelines in the developed world recommend the use of drug-resistance testing for the management of patients receiving antiretroviral therapy. However, because such testing relies mainly on sequencing or recombinant virus phenotyping, which are costly and complex, it has remained largely unavailable for the vast majority of patient populations receiving ART in resource-limited settings. Thus, simpler and less expensive drug resistance technologies may improve patient management. Previously explored technology relied on detecting and phenotyping HIV-associated RT activity in plasma by ultrasensitive biochemical assays 
. In this study we describe an improved biochemical approach to resistance testing for NNRTI and 3TC/FTC. The principle is based on our previously reported Amp-RT assays 
but has been improved in two major ways. First, we include real-time PCR for the detection of Amp-RT product which eliminates the need of ELISA quantitation of the PCR products, thus reducing time and cost. Secondly, we describe a method of HIV capture from plasma by magnetic beads, eliminating the need for ultracentrifugation and reducing cost and complexity.
Conventionally, phenotypic resistance is reported as fold changes in IC50 values over a reference WT virus. Since this approach requires testing of multiple drug concentrations we sought to evaluate a screening method with only one drug concentration to help simplify testing. We have found previously that this approach was feasible, given the high level of resistance conferred by M184V and many NNRTI mutations. By testing large numbers of mutant and WT clinical specimens we show here that inhibition with the 50 µM screening concentration correlated well with the absence of genotypic markers of resistance to 3TC/FTC. This screening concentration is close to the 3TC-TP concentration that inhibits WT enzymatic activity by 100% (IC100). We demonstrate that enzymatic 3TC resistance was strongly associated with M184I/V. In addition, we also found that the mutation V118I was associated with enzymatic 3TC resistance in the absence of M184V in 3 of 4 cases. In vitro, V118 is reported to confer resistance to 3TC and other nucleoside analogues only in combination with other mutations like D67N and T215Y 
. However, consistent with our results, Girouard and colleagues detected a 21.7-fold enzymatic resistance to 3TC-TP in RTs harboring just V118I 
. The clinical significance of this resistance is unclear and requires more study. NNRTI mutations are common in ART-treated populations. Our data also show that the screening Amp-RT format detecting resistance to NVP correlated with the presence of K103N, Y181C/I, Y188L, and G190A/Q. Thus, all primary mutations conferring resistance to NNRTIs are detected. While K103N or Y188L confer high cross-resistance between NVP and EFV
, Y181C/I/V is associated with only moderate resistance to EVF. Thus, the use of NVP in the assay allowed the detection of resistance mutations that may be selected by either NVP or EFV. However, further work may be needed to determine if EFV is more suitable than NVP for detecting resistance by particular NNRTI mutations that confer higher resistance to EFV than NVP and whether the use of EFV would have detected resistance in few mutant samples that showed WT results in the current assay.
Despite the strong correlation between genotype and enzymatic resistance, few specimens showed discrepant results. The reasons for the discrepancies are not clear but could be related to a number of factors including a wrong genotype because of sequencing error, imperfect cut-offs values, mutations currently not known to cause enzymatic resistance, or mixed genotypes that are inconsistently detected genotypically or enzymatically. While enzymatic IC50 determination clarified the discrepancies in many samples, further phenotypic testing by culture-based methods and additional genotypic analysis by ultradeep sequencing for minority or mixed resistance, and inclusion of mutations in the connection domain in RT that confer NNRI resistance could help resolve the source of the residual discrepant results 
Although this work has focused on assays for 3TC/FTC and NNRTI resistance because of the high prevalence in virologic failure and the high level of resistance, this testing approach can be adapted to other RT inhibitors, including new-generation NNRTIs and other nucleoside analogs. Such assays will require careful development that takes into account the level of enzymatic resistance found in the mutants, and the need for a screening or IC50-based testing format. One important improvement in our testing is the capture of virus by magnetic beads, which eliminated the need for ultracentrifugation and the associated cost, labor, and maintenance. We demonstrate the utility of virus capture with magnetic beads containing antibodies against CD44, a host-cell protein that is commonly incorporated on the surface of HIV-1 particles. This method has the advantages of allowing high throughput and automated processing. Overall, in contrast to conventional sequencing or cell-culture based phenotypic assays, the enzymatic Amp-RT testing described in this work does not require special and expensive instrumentation or equipment but instead a real-time PCR cycler and common equipment available in clinical virology laboratories with biosafety level 2 capacity. Additional studies geared to evaluating the assay in a resource-limited setting would include further simplified storage conditions, sensitivity to repeated freezing and thawing of the reagents and interference with other infectious agents present in the samples. The described assays have a short turnaround time and have the advantage of being subtype-independent. The reaction carried out in the absence of drug can also be used as a marker of virus load as demonstrated in the good correlation with RNA loads in this and previous studies 
. In conclusion, we demonstrate the strong potential for a simplified drug-resistance testing strategy based on sensitive enzymatic assays for two key drug-resistance profiles that are of particular importance for first-line regimens. Our data strongly support expanded clinical evaluation and standardization of this testing approach.