A total of 133 postbaseline resistance tests were performed over 535 person-years of follow-up (25 tests per 100 person-years). There were a total of 137 second or subsequent changes in antiretroviral therapy over 411 person-years of observation (33 per 100 person-years), defined as the introduction of 2 or more new drugs subsequent to the initial treatment change after randomization. Fifty-two (18%) subjects changed therapy for a second time before 12 months. Of interest, only 30% of second or subsequent treatment changes were preceded by a resistance test. Findings were similar for the 2 randomized groups in all these analyses (data not shown).
The primary end point, reduction in VL between baseline and 12 months, was recorded for 136 (89%) subjects in the G group and 147 (92%) in the G + P group. This included 9 patients who died of an HIV-related cause before 12 months (6 in the G arm, 3 in the G + P arm), who were assigned a notional VL of 100,000 copies/mL (see Methods section). The mean (SE) reduction in VL between baseline and 12 months was 1.37 (0.20) log10 copies/mL in the G arm compared with 1.28 (0.19) in the G + P arm, a difference of 0.08 (95% confidence interval [CI]: −0.46 to 0.62; P = 0.8). A comparison of the proportion of subjects with a VL of <50 copies at 12 months also failed to reveal a significant difference between the groups (G arm: 48 [35%] of 136 patients, G + P arm: 40 [27%] of 147 patients; P = 0.3).
It has been conjectured that resistance testing may be of less benefit in patients with extensive prior exposure to antiretroviral drugs.7,8 We examined this question by repeating the primary end point analysis, stratifying subjects by whether they had received a cumulative total of fewer or more than 8 drugs at trial entry. The difference between the G and G + P groups was similar in the 2 strata (<8 drugs: 0.09 greater reduction in log10 [VL] in the G group, ≥8 drugs: 0.07 greater reduction in log10 [VL] in the G group). Similar analyses based on the number of drug classes previously exposed to and the number of drugs to which the baseline genotypic report indicated resistance also showed no evidence of statistical interaction.
A total of 18 deaths were observed (13 classified as probably or definitely HIV related), 11(7 classified as probably or definitely HIV related) in the G group and 7 (6 classified as probably or definitely HIV related) in the G + P group. During the study, there were 16 and 13 new AIDS-defining events in the G and G + P arms, respectively, all of which occurred in patients who had had an AIDS diagnosis before trial entry.
To date, 3 other trials have performed a “head-to-head” comparison of the 2 modes of resistance testing.7,17,18 The GenPheRex trial randomized 201 patients to therapy guided by a VirtualPhenotype or phenotypic test result at the point of treatment failure.17 At the end of follow-up (48 weeks), virologic response was similar in both groups whether assessed by VL reduction or by the likelihood of achieving a VL load less than 400 copies/mL. Interpretation is complicated by the facts that many patients (53%) switched therapy for a second time before 48 weeks and the use of repeat resistance testing is not described. Using a similar design, the Realvirfen study randomized 300 patients (276 analyzed) to phenotypic testing or a locally performed genotypic test that was submitted for VirtualPhenotype analysis.18 Fifty-six percent of subjects in the genotypic arm and 47% in the phenotypic arm achieved a VL <400 copies/mL at week 24 (P = 0.1); the corresponding mean reductions in VL from baseline were 1.0 and 1.3 log10 copies (P = 0.02). A third trial suggested, albeit inconclusively, that genotypic testing might be superior to phenotypic testing. The NARVAL (ANRS 088) randomized 541 patients to genotyping, phenotyping, or no testing.7 The proportion of patients fulfilling the primary endpoint (VL <200 copies/mL at week 12) was higher in the genotypic arm (44%) than in the phenotypic arm (35%), a difference bordering on statistical significance (P = 0.08). This difference narrowed with longer follow-up, however; surprisingly, there was no effect on VL reduction at any time point. None of these 3 studies found any impact of the method of resistance testing on CD4 cell count changes.
The ERA trial showed no substantive differences between genotypic testing alone and the combined use of genotypic and phenotypic testing in virologic or immunologic response. The G group achieved a slightly greater reduction in VL at 12 months, by an average of 0.08 log10 copies/mL. The 95% CI for this difference is wide, however, with the data being compatible with up to a 0.62 log10 advantage for genotypic testing alone or up to a 0.46 log10 advantage for combined testing. Thus, the lack of statistical significance should not be interpreted as proof of “equivalence” of these 2 strategies of testing. Also, a planned cost-effectiveness analysis was not carried out because of the lack of a clear benefit of the intervention. Based on routine local VL measurements, there was some suggestion that combined testing conferred a delayed virologic advantage, but there is no obvious plausible explanation for this observation. The power of the study was reduced by resistance test failures and patients delaying or not changing therapy after the test. These are natural facets of resistance testing, however, and the intention-to-treat analysis presented is the most valid approach for assessing the impact of different resistance testing policies.19
The negative findings in some trials of resistance testing have been attributed to a high proportion of heavily pretreated subjects, because it is likely that many such patients have archived resistance or cross-resistance to most, it not all, available antiretroviral drugs. As in related studies,7,17,18 most patients in ERA trial had complex treatment histories; on average, therapy had been received for 48 months and a total of different 7.7 drugs had been prescribed. Repeating the main comparisons stratified by prior drug history failed to identify any particular subgroup that gained benefit from the addition of phenotypic testing, however.
The clinical effect of resistance testing is mediated through its influence on the selection of drug regimens and examination of this “process variable” is key in the interpretation of randomized trials. In this study, the additional information provided by phenotypic tests had no appreciable effect on the number of drugs used, the extent to which drugs were recycled, or the specific drugs that were prescribed. Notably, just more than half of the subjects were prescribed lopinavir/ritonavir, which was available only through an open access program during the period of recruitment to the study. Because lopinavir did not appear on the resistance reports used at study entry (see Methods section) and the relative effect of different protease mutations on its susceptibility are still incompletely understood,20 this drug was presumably prescribed on the basis of its high genetic barrier to resistance. This highlights the need for frequent modification of phenotypic tests and refinement of genotypic algorithms to keep abreast of new drug development. It also reveals the inherent limitations of randomized studies of resistance testing in highly drug-experienced patients, because there is always a tendency to use newly available drugs in all arms.
The similarity in the drug regimens prescribed in the 2 groups was perhaps surprising in light of the level of discordance between the genotypic and phenotypic test results. Parkin and colleagues21 have described multiple possible reasons for genotype/phenotype discrepancy, including limitations of current genotypic interpretation algorithms (particularly in the case of complex mutational patterns), difference in test sensitivity as a result of amino acid mixtures, and the choice of phenotypic cutoffs used to designate susceptibility or resistance. In line with this and other studies,17,18 we found that levels of discordance were generally higher among drugs in the NRTI class and that a genotypic test result tended to indicate a higher level of resistance than a phenotypic test result. An important difference between the ERA trial and the NARVAL, GenPheRex, and Realvirfen trials was the provision of a phenotypic test in addition to rather than instead of a genotypic report. The ERA trial did not attempt to measure the relative influence of the 2 test results (in the G + P group) on prescribing decisions. The similarity between the randomized groups suggests that clinicians may have acted conservatively and placed greater emphasis on the genotypic test result, however. Alternatively, prescribing decisions may have been dictated by independent factors such as drug tolerability or the desire to use novel drugs. In a recent experiment, 13 experts were asked to recommend antiretroviral regimens for 5 patients based on a genotypic test alone, a phenotypic test alone, or the 2 tests in combination.22 In pairwise comparisons, differences in regimens were smallest between the second and third groups, suggesting a greater import of the results of the phenotypic tests than of the genotypic tests. Although it may be correct that the 2 types of test provide “complementary” information,4,21 the results of the ERA trial lead us to question whether this necessarily translates to the selection of more optimal antiretroviral regimens.
It is inherently difficult to assess diagnostic tests in a rapidly moving field, and the results of this study largely reflect the effect of tests that have been superseded.23 Nevertheless, in the case of the phenotypic test, the intervention essentially evaluated by part B of the ERA trial, the only change was in the interpretation of the fold resistance values and not the values themselves. Also, the changes in the genotypic test meant it became more similar to the phenotypic test so that, logically, a trial conducted now would be seeking to detect even smaller differences. In conclusion, the ERA trial found no clear evidence of an added benefit of phenotypic resistance testing in conjunction with genotypic resistance testing, but caution should be exercised in generalizing this finding.
1. 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
2. The EuroGuidelines Group for HIV Resistance. Clinical and laboratory guidelines for the use of HIV-1 drug resistance testing as part of treatment management: recommendations for the European setting. AIDS
3. Vandamme AM, Houyez F, Banhegyi D, et al. Laboratory guidelines for the practical use of HIV drug resistance tests in patient follow-up. Antivir Ther
4. Guidelines for the use of antiretroviral agents in HIV-1 infected adults and adolescents. Available at: www.aidsinfo.nih.gov/guidelines
. Accessed December 20, 2003.
5. Durant J, Clevenbergh P, Halfon P, et al. Drug-resistance genotyping in HIV-1 therapy: the VIRADAPT randomised controlled trial. Lancet
6. Baxter JD, Mayers DL, Wentworth DN, et al. A randomized study of antiretroviral management based on plasma genotypic resistance testing in patients failing therapy. AIDS
7. Meynard JL, Vray M, Morand-Joubert L, et al. Phenotypic or genotypic resistance testing for choosing antiretroviral therapy after treatment failure: a randomized trial. AIDS
8. Tural C, Ruiz L, Holtzer C, et al. Clinical utility of HIV genotyping and expert advice: the Havana trial. AIDS
9. Cingolani A, Antinori A, Rizzo MG, et al. Usefulness of monitoring HIV drug resistance and adherence in individuals failing highly active antiretroviral therapy: a randomized study (ARGENTA). AIDS
10. Wegner SA, Wallace M, Aronson A, et al. Evaluation of the clinical efficacy of antiretroviral resistance testing (CERT) [abstract ThOrB1389]. Presented at: XIV International AIDS Conference; 2002; Barcelona.
11. Melnick D, Rosenthal J, Cameron M, et al. Impact of phenotypic antiretroviral drug resistance testing on the response to salvage antiretroviral therapy (ART) in heavily experienced patients [abstract 786]. Presented at: Seventh Conference on Retroviruses and Opportunistic Infections; 2000; San Francisco.
12. Cohen C, Hunt S, Sension M, et al. A randomized trial assessing the impact of phenotypic resistance testing on antiretroviral therapy. AIDS
13. Haubrich R, Keiser P, Kemper C, et al. CCTG 575: a randomized, prospective study of phenotype testing versus standard of care for patients failing antiretroviral therapy [abstract 80]. Presented at: Fifth International Workshop on HIV Drug Resistance and Treatment Strategies; 2001; Scottsdale, AZ.
14. Dunn DT, Gibb DM, Babiker AG, et al. HIV drug resistance testing: is the evidence really there? Antivir Ther
15. Hertogs K, de Bethune MP, Miller V, et al. A rapid method for simultaneous detection of phenotypic resistance to inhibitors of protease and reverse transcriptase in recombinant human immunodeficiency virus type 1 isolates from patients treated with antiretroviral drugs. Antimicrob Agents Chemother
16. Marschner IC, Betensky RA, DeGruttola V, et al. Clinical trials using HIV-1 RNA-based primary end points: statistical analysis and potential biases. J Acquir Immune Defic Syndr
17. Mazzotta F, Caputo SL, Torti C, et al. Real versus virtual phenotype to guide treatment in heavily pretreated patients: 48-week follow-up of the Genotipo-Fenotipo di Resistenza
(GenPheRex) Trial. J Acquir Immune Defic Syndr
18. Perez-Elias MJ, Garcia-Arata I, Muñoz V, et al, for the Realvirfen Study Group. Phenotype or virtual phenotype for choosing antiretroviral therapy after failure: a prospective, randomized study. Antivir Ther
19. Schwartz D, Lellouch J. Explanatory and pragmatic attitudes in therapeutic trials. J Chronic Dis
20. Parkin NT, Chappey C, Petropoulos CJ. Relationship between lopinavir (LPV) susceptibility and HIV-1 protease genotype [abstract 581]. Presented at: Ninth Conference on Retroviruses and Opportunistic Infections; 2002; Seattle.
21. Parkin N, Chappey C, Maroldo L, et al. Phenotypic and genotypic HIV-1 drug resistance assays provide complementary information. J Acquir Immune Defic Syndr
22. Zolopa AR, Bates M, Parkin N. Experts select different antiretroviral regimens when presented with resistance data in the form of genotype, phenotype, or combined genotype plus phenotype. Antivir Ther
23. Dunn DT, McCormack S, Babiker A, et al. Tracker trials. BMJ
A. G. Babiker, A. Breckenridge (Chairman), J. H. Darbyshire, D. T. Dunn, A. Fakoya, M. Fisher, A. Leigh Brown, C. Loveday (Principal Investigator), S. McCormack, D. Pillay, A. Poppa, A. Rinehart, W. Verbiest, and I. Williams
D. T. Dunn, H. Green, C. Loveday, A. Rinehart, D. Pillay, M. Fisher, S. McCormack, A. G. Babiker, and J. H. Darbyshire
Data and Safety Monitoring Committee
A. McLaren, P. Armitage, V. Beral, and H. Lambert
Medical Research Council Clinical Trials Unit
A. G. Babiker, D. T. Dunn, H. Green, P. Kelleher, S. Khan, S. McCormack, and F. Tyrer
W. Verbiest, A. Rinehart, L. Bacheler, P. McKenna, K. V. Bonduelle, P. Schel, C. Jordens, A. Ghys, and M. Van Brandt
Royal Free and International Clinical Virology Centre
C. Loveday, J. Page, J.-L. Faudon, and K. Jones
Participating Clinical Centers
Barts and the Royal London Hospital, London (M. Murphy, G. Roper, P. Davis, J. Norman, and P. Sashi), Central Middlesex Hospital, London (R. Fox, L. McDonald, and B. Patel), Chesterfield and North Derbyshire Royal Hospital, Chesterfield (K. Rogstad and B. Purdon), Doncaster Royal Infirmary, Doncaster (J. Hawkswell and G. Ball), Edinburgh Western General, Edinburgh (C. Leen, S. Morris, C. Wilson, and S. Burns), Essex County Hospital, Colchester (S. Jebakumar), Gartnavel General Hospital, Glasgow (R. Fox, S. Eves, S. Cameron, and J. McCowan), Kings College Hospital, London (P. Easterbrook, C. Taylor, A. Waters, D. Graham, and L. McQueen), Leeds General Infirmary, Leeds (E. Monteiro and G. Booth), Leicester Royal Infirmary, Leicester (M. Wiselka, J. Laurenti, S. Bonnington, and S. Sikotra), Mayday University Hospital, Croydon (M. Rodgers, T. Newell, and P. Riley), Middlesbrough General Hospital, Middlesbrough (B. McCarron, L. Lowery, and J. Bashford), Mortimer Market Center, London (I. Williams, D. Cornforth, D. Aldam, E. MacFarlane, and S. Rice), Newcastle General Hospital, Newcastle (E. Ong, M. Snow, A. Harrison, A. Turner, and A. Rudsdale), Newham General Hospital, London (A. Fakoya and C. Tawana), North Manchester General Hospital, Manchester and St. Thomas's Hospital, Stockport (E. Wilkins, R. Daintith, E. Stockwell, and A. Bailey), North Middlesex Hospital, London (J. Ainsworth and L. DuRibbo), Royal Berkshire Hospital, Reading (R. Manoharan, D. Nelson, and J. Selwood), Royal Free Hospital, London (M. Johnson, M. Tyrer, D. Wilson, T. Drinkwater, Z. Cuthbertson, P. Bryne, F. Turner, and C. Loveday), Royal Bolton Hospital, Manchester (S. Ahmad and E. Morgan), Royal Hallamshire Hospital, Sheffield (C. Bradbury, S. Herman, D. Docrall, C. Care, and G. Ball), Royal Sussex County Hospital, Brighton (M. Fisher, D. Churchill, N. Perry, J. McIntosh Roffey, and A. McCue), St. George's Hospital, London (F. Davidson, P. Hay, A. Adebiyi, M. Ogunfile, M. Wansbrough-Jones, and B. Edwards), St. Thomas's Hospital, London (B. Peters, L. Judges, N. Saint, J. Turpitt, A. Ronald, L. Burghard, and S. O'Shea), Queen Elizabeth Hospital, Woolwich (J. Russell, S. Mitchell, and G. Vosper), and Whittal Street Clinic, Birmingham (M. Shamanesh and G. Gilleran)