The goal of antiretroviral therapy is to reduce HIV-1-associated morbidity and mortality, by suppressing viral replication to the maximum degree possible . Unfortunately, failure to achieve complete suppression is common in clinical practice . Although the advent of highly active antiretroviral therapy (HAART) has led to dramatic decreases in HIV-related morbidity and mortality , the development of virus with reduced susceptibility to antiretroviral drugs is frequently observed after virological failure of HAART [4–7]. Because virological failure is often associated with the emergence of drug resistance, there is increased interest in the use of drug susceptibility assays to assist in the selection of optimal combination drug therapy regimens . Retrospective and prospective clinical studies have demonstrated that regimens containing more drugs to which the virus is (or is predicted to be) susceptible are more likely to elicit durable suppression of virus replication than regimens with fewer active drugs [9–15].
We have previously reported 24-week virologic outcomes from a prospective evaluation of twenty subjects treated with a four-drug combination of nelfinavir, saquinavir, abacavir, and either a second nucleoside reverse transcriptase inhibitor (NRTI) or nevirapine after failure of their first indinavir- or ritonavir-based regimens . Based on drug susceptibility at the time of salvage therapy initiation, viral load suppression at 24 weeks was significantly greater in subjects whose virus was susceptible to two or three drugs than subjects with virus susceptible to none or one drug.
This report represents a detailed description of the patterns of drug susceptibility that emerged during the course of this study and long-term follow-up. We sought to explore the potential relationship between reductions in drug susceptibility, detected before or at the time of viral load rebound, and virological outcome. In addition we characterized patterns of reductions in drug susceptibility after failure of a second-line HAART regimen in treatment-experienced patients.
Subjects who experienced viral load rebound while undergoing therapy with indinavir-containing regimens were enrolled in a four-drug salvage therapy trial . Eligibility criteria included: (1) greater than 24 weeks of continuous therapy with a regimen containing either indinavir or ritonavir; (2) screening HIV RNA > 2500 copies/ml; and (3) no prior exposure to nelfinavir, saquinavir, abacavir or non-nucleoside reverse transcriptase inhibitors (NNRTIs). One exception was made for subject no. 21 who had prior NNRTI experience, but did not receive nevirapine in the present study. Ten subjects received twice daily doses of 1250 mg nelfinavir, 1200 mg saquinavir-soft gel capsule and 300 mg abacavir together with a second nucleoside analogue chosen on the basis of prior drug experience. A further ten subjects received the same nelfinavir, saquinavir, abacavir treatment, together with 200 mg nevirapine twice daily. Subjects were seen at least every 4 weeks for 96 weeks.
For the current analysis, we defined virological failure as (a) incomplete response i.e. not reaching a viral load below 500 copies/ml by week 24; (b) viral load rebound (two consecutive measurements above 500 copies/ml) after becoming undetectable; or (c) viral load rebound greater than 0.5 log above the nadir (if > 500 copies/ml) before 24 weeks. Four subjects (three from the NRTI arm, one from the nevirapine arm) discontinued therapy prior to or at week 8 and were not included in the present follow-up analysis. Subject no. 1 switched from nelfinavir, saquinavir, stavudine, and abacavir to nelfinavir, saquinavir, stavudine, and nevirapine at week 4, due to abacavir intolerance. For the purposes of the analysis described here, this subject was considered part of the nevirapine group, and the baseline was defined as 5 days before the time of switch from abacavir to nevirapine.
Stored frozen plasma samples were chosen retrospectively for phenotyping. The selected samples represented time points just before the viral load became undetectable in the responders, and at or near the viral load nadir (if virus remained detectable) or viral load rebound in the failures. Tested samples are indicated by asterisks in Figure 1.
Viral load was measured using standard or ultrasensitive quantitative reverse transcriptase (RT)-polymerase chain reaction (PCR) (Amplicor HIV Monitor Test; Roche Diagnostic Systems Inc, Branchburg New Jersey, USA; lower limit of quantitation of 50 RNA copies/ml) [16,17]. Drug susceptibility was performed in duplicate using the PhenoSenseTM HIV assay  (ViroLogic, Inc., South San Francisco, California, USA). Virus was defined as having reduced susceptibility if the fold change in the 50% inhibitory concentration (IC50) compared with the drug sensitive reference (NL4-3) was greater than 2.5-fold, which is the assay reproducibility cut-off. The assay requires the amplification of a 1.5 kb segment of the HIV pol gene derived from viral RNA in 1 mL of plasma. Greater than 95% of samples with viral loads above 500 copies/ml can be amplified and tested in the phenotypic assay. Samples with viral loads between 100 and 500 copies/ml have an amplification success rate of approximately 70% (unpublished data).
PR and RT genotypes of resistance test vectors were determined using the GeneSeqTM HIV assay, which utilizes ABI dye-terminator chemistry. Reaction products were analyzed on a PE BioSystems 3700 automated sequencer. DNA sequences were analyzed using a customized version of Sequencher 4.0 software (Gene Codes, Ann Arbor, Michigan, USA), and deduced amino acid sequences of PR (amino acids 1 to 99) and RT (amino acids 1 to 305) were compared with the NL4-3 reference virus. The sensitivity of the genotyping methods used for detection of minor species ranges from 10 to 20% (unpublished GeneSeqTM HIV assay validation data). All viruses in this study were identified as clade B. Complete sequences have been deposited in GenBank (accession numbers AF124542- AF124556 and AF294515- AF294564).
Study population and virologic outcome
Ten subjects were initially enrolled in each treatment group (second NRTI or nevirapine). One of the nevirapine-treated subjects and three of the NRTI-treated subjects discontinued study medications , and one additional NRTI-treated subject switched to the nevirapine group (see Methods). Five of the ten subjects treated with nevirapine achieved sustained viral load suppression (< 500 copies/ml), and five experienced viral load rebound (Table 1). After viral rebound, two of the five subjects (no. 12 and 20) had stable viral loads between 500 and 1000 copies/ml. Only one of six subjects treated with a second NRTI had sustained viral load suppression (Table 1). Viral loads were also undetectable using the ultra-sensitive viral load assay (< 50 copies/ml) in all six subjects with sustained suppression.
Phenotypic susceptibility analyses were performed on the 16 subjects who remained on study medication for at least 24 weeks (median baseline CD4 248 × 106 cells/l; baseline HIV RNA 24 600 copies/ml).
Viral evolution in subjects achieving durable viral suppression
Representative examples of viral load profiles from four of the six responders are shown in Figure 1a. All six subjects had at least one plasma sample with a viral load measurement between 50 and 1000 copies/ml before achieving their undetectable HIV RNA. The viral load of the samples that were tested from the responders ranged from 83 to 865 copies/ml (median, 356 copies/ml).
The drug susceptibility profile of virus in plasma collected prior to the first undetectable viral load was similar to that of virus present at baseline (Table 2). Small but reproducible changes in drug susceptibility were noted in several subjects (e.g. no. 13 and 15), although evolution of drug susceptibility from a sensitive (< 2.5-fold versus reference) to a reduced susceptibility (> 2.5-fold versus reference) phenotype was not observed for any treatment drugs. Durable (92–108 weeks) viral load response in these subjects was associated with the maintenance of susceptibility to two or three drugs in the treatment regimen.
Genotypic analysis of viruses from the responders was also performed. In four subjects (no. 5, 16, 17, and 18), no new resistance-associated genotypic changes were detected compared to baseline (Table 3). The appearance of a K101Q mutation in subject no. 5 at week 44 coincided with a 2.2-fold change in nevirapine susceptibility, compared with 0.4- to 0.7-fold in previous samples (Table 2). As this subject was not receiving nevirapine, it is unlikely that this mutation was selected by the drug treatment. In two additional subjects (no. 13 and 15), subtle changes in protease were noted: mixtures of wild-type and mutant amino acids at positions 24, 46, 54, and/or 71 appeared and faded at various time points (Table 3). The appearance of these mixtures coincided with small decreases in protease inhibitor (PI) susceptibility (four- to seven-fold;Table 2).
Viral evolution in subjects experiencing virological failure
Representative examples of viral load profiles from four of the ten failure subjects are shown in Figure 1b. Eight subjects (no. 6, 14, and 21 in Figure 1b, and no. 1, 7, 12, 19, and 20, not shown) experienced viral load rebound and two (no. 10 in Figure 1b, and no. 2, not shown) had incomplete suppression during therapy. Susceptibility profiles of viruses from various time points prior to or during viral load rebound were characterized by additional reductions in susceptibility to one or more treatment drugs. Four of the five nevirapine-treated subjects (no. 1, 12, 14, and 19) had baseline virus that was susceptible to nevirapine only, whereas the fifth (no. 20) had baseline virus susceptible to nevirapine and saquinavir. Each experienced viral load rebound (between 4 and 36 weeks) that was associated with the emergence of virus with nevirapine resistance (Table 4). In subject no. 20, low-level reductions in saquinavir susceptibility were also detected at weeks 24 and 36.
The emergence of nevirapine resistance coincided with viral load nadir in subjects no. 1 and 14 (week 4), when the viral load was 13 300 and 260 copies/ml, respectively. In subjects no. 19 and 20, it was detected in the first sample collected after viral load nadir (viral loads: 14 000 and 477 copies/ml, respectively). In four of the five cases in which nevirapine resistance emerged, reduced susceptibility to delavirdine and efavirenz was also observed.
Virus from five subjects treated with nelfinavir, saquinavir, abacavir and a second NRTI (no. 2, 6, 7, 10, and 21) displayed new reductions in susceptibility to saquinavir and/or nelfinavir prior to, or during, virological failure. In subjects no. 2 and 10, the baseline virus was susceptible to saquinavir, nelfinavir, and stavudine; however, the response to therapy was incomplete (see viral load plot for subject no. 10 in Figure 1b). As early as week 13, and more prominently at later time points, reduced susceptibility to saquinavir and nelfinavir was observed (Table 4). Decreased susceptibility to stavudine was also noticed at week 18 in subject no. 2. Virus from subject no. 6 was susceptible only to saquinavir at baseline; decreased saquinavir susceptibility was first detected at week 5 and viral load rebound occurred at week 9 (Fig. 1b). Virus in subject no. 7 exhibited small reductions in zalcitabine (2.6-fold) and saquinavir (4.8-fold) susceptibility at baseline; saquinavir resistance became dramatic at week 8 (63-fold), the time of initial viral load rebound. The virus in subject no. 21 at baseline was susceptible to saquinavir, abacavir, and lamivudine. At weeks 12 and 16, when the viral load was 305 and 277 copies/ml, respectively, decreased susceptibility to saquinavir and lamivudine was detected.
Changes in the genotypes of viruses from the failures were consistent with the alterations in susceptibilities to the treatment drugs that were observed (Table 5). Viruses displaying reductions in nevirapine susceptibility developed RT mutations at recognized positions such as 106, 103, and 181, as well as less common mutations at 227  in subject no. 1 and 230  in subject no. 14. Reductions in saquinavir susceptibility were associated with the appearance of G48V in PR in subjects no. 6, 7, and 21, and with L90M in subjects no. 2, 10, and 20. A mixture of wild-type (M) and mutant (I) amino acids was present at position 184 in RT at weeks 12 and 16 in virus from subject no. 21, when low level reductions in lamivudine susceptibility were detected. Decreased susceptibility to abacavir became evident at week 24, when the mixture contained predominantly mutant viruses (M184I and M184V). Notably, in three cases (subjects no. 1, 2, and 6), reductions in drug susceptibility (to nevirapine, stavudine, and saquinavir, respectively) were measured prior to the detection of new resistance-associated mutations by population sequencing. Conversely, in no case was a genotypic change detected prior to a reduction in drug susceptibility.
In the group of ten treatment-failure subjects, new reductions in susceptibility to at least one of the treatment drugs was detected prior to virological failure (viral load > 500 copies/ml at week 24, or viral rebound to > 500 copies or > 0.5 log from nadir) in six cases: subjects no. 1, 6, 10, 14, 20, and 21 (Table 6). Using a more stringent definition of failure of rebound to > 50 copies/ml, this number is lowered to four out of ten. In several subjects, resistance to drugs in the treatment regimen progressively increased following the first detection of reduced susceptibility (e.g. subjects no. 2, 6, and 10).
Detection of resistance at low viral load (< 1000 copies/ml)
A total of 31 plasma samples with viral load between 50 and 1000 copies/ml were tested, from all six of the responders (18 samples total), and from six of the ten failures (subjects no. 10, 12, 14, 19, 20, and 21; 13 samples total). Fourteen samples from the responders (78%), and 12 from the failures (92%), yielded sufficient PCR product to perform phenotypic and genotypic testing (median viral load of positive samples: 356 and 513 copies/ml, respectively). In five of the treatment failures, but none of the responders, new reductions in susceptibility to one or more of the treatment drugs were detected in the low viral load sample. In two subjects (no. 10 and 12) a sample from an earlier timepoint with a viral load > 1000 copies/ml also exhibited reduced susceptibility. Samples with HIV RNA between 50 and 1000 copies/ml were not available from subjects no. 1, 2, 6, or 7.
Optimal treatment of HIV-1 infection with combination antiretroviral chemotherapy requires careful selection of a regimen of drugs to which the virus is susceptible. However, in some cases these regimens still fail to durably suppress virus replication and viral rebound occurs. We have shown here that phenotypic susceptibility testing is capable of detecting emerging drug resistance prior to significant viral load rebound, thereby predicting virological failure. Thus, phenotypic testing has the potential to provide clinical benefit by facilitating earlier change in treatment regimens before the virus develops significant cross resistance to related drugs. For example, in subjects no. 2 and 10 studied here, prospective monitoring of drug susceptibility, prompted by slower than expected viral load response, could have led to an alteration in the drug regimen before indinavir and ritonavir cross resistance developed (see Table 4). However in other subjects, reduced drug susceptibility was stable between the time of first detection and the time of virological failure. Currently, the evolution of drug resistance patterns in individual patients is unpredictable.
Genotypic changes predictive of subsequent virological failure were also detected early during failure of therapy. However in these subjects, population genotyping was not as sensitive as phenotyping, and was not predictive of cross-resistance to other drugs such as amprenavir or efavirenz. In addition, genotypic analysis of baseline samples using accepted guidelines for interpretation  was less valuable in predicting the response to salvage therapy than phenotyping.
The detection of resistance before viral load rebound has been reported previously for two subjects who failed treatment with zidovudine, lamivudine, and nelfinavir . In these previously treatment-naive subjects, lamivudine resistance was detected prior to viral load rebound to over 500 copies/ml. In one of the two subjects, reduced susceptibility to nelfinavir was also detected although the viral load was still below 500 copies/mL. The present study confirms the ability of the phenotypic assay to detect the emergence of virus with reduced drug susceptibility at low viral loads in a larger number of subjects. In most cases, the emergence of phenotypic resistance at low viral load was associated with subsequent viral rebound. In contrast, the lack of any change in drug susceptibility was associated with subsequent declines in viral load, and sustained HIV RNA levels < 500 copies/ml. Thus, in this retrospective analysis, phenotypic susceptibility testing appears to be capable of predicting virologic outcome, even when performed at very low copy number.
In all ten subjects who met the definition of virological failure, virus with reduced susceptibility to PIs emerged early during virological failure. This is in contrast to virological failure of initial indinavir-containing regimens [5,6], where emergence of resistance to indinavir was rare. These disparate observations may be explained by an increase in protease gene heterogeneity in PI-experienced compared to PI-naive patients. Therefore, PI-resistant variants may exist before the initiation of a salvage regimen, and emerge rapidly during early virological failure. Among wild-type strains, there may be a high ‘fitness’ barrier that prevents the rapid evolution of protease inhibitor resistance . Such a barrier may not be present once some protease inhibitor resistance is present, enabling more rapid viral evolution in treatment-experienced patients.
The homogenous nature of the patient population at baseline (with respect to prior PI and NNRTI use), and the high degree of adherence observed during the study, allows for careful evaluation of the relationship between baseline phenotype and virologic response. The virologic response of the ten subjects treated with a nevirapine-based regimen was largely predicted by baseline phenotype. All subjects whose baseline virus was susceptible only to nevirapine had rapid viral load rebound. Most subjects experienced durable viral suppression when their baseline virus was susceptible to nevirapine and one or more other agents. Although the number of subjects evaluated was small, these observations underscore the fragile nature of an NNRTI-based regimen in heavily pre-treated patients when no other active drugs are included in the treatment regimen. Patients who have developed high level resistance to the NRTI and protease-inhibitor class may want to preserve the option of using an NNRTI for future regimens, assuming the availability of drugs that are active against viruses which are resistant to currently available drugs.
This study was retrospective and involved a relatively small number of subjects. Samples of interest were identified on the basis of viral load trends that may be less apparent when viral load is monitored prospectively and less frequently. The feasibility of using resistance assays as a monitoring tool requires further evaluation to determine the clinical benefits and economic impact of such testing practices. However, if the viral load is measured frequently after the initiation of a new regimen, and the response is slower than expected, our data suggest that a susceptibility assay could be useful for evaluation of risk for treatment failure. Under this scenario, resistance testing has been recommended by the US Department of Health and Human Services . Prospective studies evaluating the role of resistance testing in patients with low viral load (< 1000 copies/ml) are necessary to determine the clinical relevance of our observations.
The performance of recombinant virus phenotypic assays requires the generation of resistance test vectors from RT-PCR-amplified PR and RT sequences from patient virus samples. In our current analysis, we were able to perform phenotypic assays on plasma samples with low viral loads, although the reproducibility of amplification below 500 copies/ml has not been thoroughly evaluated. When the lower limit of amplification sensitivity is approached, the possibility of disproportionate amplification of one or a few template RNA molecules (`founder effect') may bias resistance test results. Until the viral load at which founder effects begin to bias phenotypic test results is defined, caution must be exercised when interpreting results obtained from samples with viral loads less than 500 copies/ml, especially when reduced susceptibility is not detected.
In summary, our results suggest that drug resistance emerges very early during treatment failure. Thus, phenotypic susceptibility testing may play an important role in predicting treatment failure earlier than previously recognized. Virological failure in previously treated patients is often associated with the development of reduced susceptibility to most or all treatment drugs. The results from ongoing, larger, prospective studies are required to confirm and extend these findings.
We thank members of the ViroLogic Research and Development Laboratory (Jeannette Whitcomb, Kay Limoli, Rainer Ziermann, Gabrielle Heilek-Snyder, and Doug Smith) and Clinical Reference Laboratory (Jan Turczyn, Jonas San Gabriel, Julie Tam, Christina Zrimsek, Sophia Chen) for performing the phenotypic assays. We are grateful for study support from Agouron, Hoffmann LaRoche, Glaxo Wellcome and Boeringher Inglheim. We acknowledge the co-operation of and extend our appreciation to the subjects who participated in the study.
1. Carpenter CC, Cooper DA, Fischl MA. et al
. Antiretroviral therapy in adults: updated recommendations of the International AIDS Society-USA Panel.
JAMA 2000, 283: 381 –390.
2. Deeks SG, Hecht FM, Swanson M. et al
. HIV RNA and CD4 cell count response to protease inhibitor therapy in an urban AIDS clinic: response to both initial and salvage therapy.
AIDS 1999, 13: F35 –43.
3. Palella FJ, Delaney KM, Moorman AC. et al
. Declining morbidity and mortality among patients with advanced human immunodeficiency virus infection.
N. Engl. J. Med 1998, 338: 853 –860.
4. Parkin N, Lie Y, Hellmann N. et al
. Phenotypic changes in drug susceptibility associated with failure of human immunodeficiency virus type 1 (HIV-1) triple combination therapy.
J Infect Dis 1999, 180: 865 –870.
5. Descamps D, Flandre P, Calvez V. et al
. Mechanisms of virologic failure in previously untreated HIV-infected patients from a trial of induction-maintenance therapy.
JAMA 2000, 283: 205 –211.
6. Havlir DV, Hellmann NS, Petropoulos CJ. et al
. Drug susceptibility in HIV infection after viral rebound in patients receiving indinavir-containing regimens.
JAMA 2000, 283: 229 –234.
7. Race E, Dam E, Obry V, Paulous S, Clavel F. Analysis of HIV cross-resistance to protease inhibitors using a rapid single-cycle recombinant virus assay for patients failing on combination therapies.
AIDS 1999, 13: 2061 –2068.
8. Hirsch MS, Brun-Vezinet F, D'Aquila RT. et al
. Antiretroviral drug resistance testing in adult HIV-1 infection: recommendations of an International AIDS Society-USA Panel.
JAMA 2000, 283: 2417 –2426.
9. Harrigan P, Hertogs K, Verbiest W. et al
. Baseline HIV drug resistance profile predicts response to ritonavir-saquinavir protease inhibitor therapy in a community setting.
AIDS 1999, 13: 1863 –1871.
10. Zolopa A, Shafer R, Warford A. et al
. HIV-1 genotypic resistance patterns predict response to saquinavir-ritonavir therapy in patients in whom previous protease inhibitor therapy had failed.
Ann Intern Med 1999, 131: 813 –821.
11. Piketty C, Race E, Castiel P. et al
. Efficacy of a five-drug combination including ritonavir, saquinavir and efavirenz in patients who failed on a conventional triple-drug regimen: phenotypic resistance to protease inhibitors predicts outcome of therapy.
AIDS 1999, 13: F71 –F77.
12. Deeks SG, Hellmann NS, Grant RM. et al
. Novel four-drug salvage treatment regimens after failure of a human immunodeficiency virus type 1 protease inhibitor-containing regimen: antiviral activity and correlation of baseline phenotypic drug susceptibility with virologic outcome.
J Infect Dis 1999, 179: 1375 –1381.
13. Durant J, Clevenbergh P, Halfon P. et al
. Drug-resistance genotyping in HIV-1 therapy: the VIRADAPT randomised controlled trial.
Lancet 1999, 353: 2195 –2199.
14. Cohen C, Hunt S, Sension M, et al
. Phenotypic resistance testing significantly improves resonse to therapy: a randomized tiral (VIRA3001).Seventh Conference on Retroviruses and Opportunistic Infections
. San Francisco, CA, USA, January 2000 [abstract 237].
15. Baxter JD, Mayers DL, Wentworth DN. et al
. A randomized study of antiretroviral management based on plasma genotypic antiretroviral resistance testing in patients failing therapy.
:CPCRA 046 Study Team for the Terry Beirn Community Programs for Clinical Research on AIDS.
AIDS 2000, 14: F83 –93.
16. Mulder J, McKinney N, Christopherson C, Sninsky J, Greenfield L, Kwok S. Rapid and simple PCR assay for quantitation of human immunodeficiency virus type 1 RNA in plasma: application to acute retroviral infection.
J Clin Microbiol 1994, 32: 292 –300.
17. Sun R, Ku J, Jayakar H. et al
. Ultrasensitive reverse transcription-PCR assay for quantitation of human immunodeficiency virus type 1 RNA in plasma.
J Clin Microbiol 1998, 36: 2964 –2969.
18. Petropoulos CJ, Parkin NT, Limoli K. et al
. A novel phenotypic drug susceptibility assay for HIV-1.
Antimicrob Agents Chemother 2000, 44: 920 –928.
19. Balzarini J, Pelemans H, Esnouf R, De Clercq E. A novel mutation (F227L) arises in the reverse transcriptase of human immunodeficiency virus type 1 on dose-escalating treatment of HIV type 1-infected cell cultures with the nonnucleoside reverse transcriptase inhibitor thiocarboxanilide UC-781.
AIDS Res Hum Retroviruses 1998, 14: 255 –260.
20. Huang W, Parkin NT, Lie Y. et al
. A novel HIV-1 RT mutation (M230L) confers NNRTI resistance and dose-dependent stimulation of replication.
Antiviral Ther 2000, 5 (suppl 3): 24. 24.
21. DeGruttola V, Dix L, D'Aquila R. et al
. The relation between baseline HIV drug resistance and response to antiretroviral therapy: re-analysis of retrospective and prospective studies using a standardized data analysis plan.
Antiviral Ther 2000, 5: 41 –48.
22. Department of Health and Human Services and the Henry J. Kaiser Family Foundation. Guidelines for the use of antiretroviral agents in HIV-infected adults and adolescents.
23. Schinazi RF, Larder BA, Mellors JW. Mutations in retroviral genes associated with drug resistance: 2000-2001 update.
Int Antiviral News 2000, 8: 65 –91.