The adherence questionnaire was introduced into the study from June 1999 and was given to 150 patients at baseline. Of these, 15 refused to fill it in and another eight were unable to complete it without assistance. Therefore, 127 questionnaires (73% of the total patient sample) were reliably completed and were used for the analysis. Baseline characteristics between those who did and those who did not complete the questionnaire were homogeneous (not shown). According to the employed categorization of self-reported adherence, 43% of patients were classified as non-adherent. There was no baseline difference between the study arms, with 45.8% in the SOC arm and 42.7% in the G arm classified as non-adherent.
Treatments after randomization and sensitivity score
At the beginning of the study, the antiretroviral agents registered in Italy were five NRTI (zidovudine, didanosine, zalcitabine, stavudine and lamivudine), one NNRTI (nevirapine) and four PI (saquinavir hard gel, indinavir, ritonavir and nelfinavir). Efavirenz became available at the centre by an unlimited early access programme in March 1999; abacavir was available for a limited number of patients in the expanded access programme from July 1999 and was approved in January 2000.
Panel decisions were applied to the patient in 144 out of 174 cases (83%): 67 out of 85 (79%) in the genotype arm and 77 out of 89 (87%) in the control arm (P = 0.18). The reasons for not prescribing the drugs decided by the consensus panel were patient's concern about potential toxicity and physician-estimated patient non-adherence. The mean number of drugs employed in the study arms after randomization was 3.3 in the SOC arm and 3.4 in the G arm (P = 0.72). In the G arm, the mean number of new antiretroviral drugs employed was not significantly greater than in the control arm (1.1 compared with 0.7, P = 0.16). There was a trend towards a more frequent use of a new antiretroviral class in the genotype arm (29% compared with 18% in the control arm, P = 0.08). The number of active drugs during follow-up did not differ between randomization arms: at 3 months the mean number of active drugs was 2.3 and 2.1 in the G and SOC arm, respectively (P = 0.20), at 6 months the mean numbers were 2.3 and 2.2, respectively (P = 0.32). Also using the Resistance Collaborative Group interpretation algorithm, there was no significant difference in the genotypic sensitivity scores between study arms (mean genotype sensitivity score 1.8 and 1.7 in G and SOC, respectively). Nevertheless, the genotype-guided group showed a more prominent mean increase from the baseline number of active drugs at 3 months (+0.61 compared with +0.22 in the controls, P = 0.004) and at 6 months (+0.69 versus +0.3, respectively, P = 0.017). There was no correlation between the time of enrolment in the trial and the number of new drugs (P = 0.12) or the number of active drugs (P = 0.52) in the salvage regimen employed in both study arms. Only five patients in the control arm received a genotype between month 3 and month 6.
The frequency of primary study endpoints in both treatment arms are shown in Fig. 1 and Table 3. Using an intent-to-treat approach with missing values considered failures, at 3 months 12% in the control arm and 27% in the genotype arm had HIV-RNA levels below 500 copies/ml (P = 0.01). At 6 months, the relative proportions were 17 and 21%, respectively (P = 0.47).
By intent-to-treat analysis, using the last observation carried forward, the mean change from baseline viral load at 3 months was −0.38 (SD ± 0.96) log10 copies/ml for those assigned to standard of care and −0.62 (SD ± 1.16) log10 copies/ml for those assigned to genotype-guided treatment (P = 0.12). Mean changes at 6 months were −0.39 (SD ± 1.04) log10 and −0.57 (SD ± +1.09) log10, respectively (P = 0.28).
In the group of patients with baseline viral loads of less than 4 log copies/ml, the advantage of the genotype information over the standard of care was particularly evident (see Table 4).
Considering the overall patient population (G plus SOC), the proportion reaching HIV-RNA levels of less than 500 copies/ml at 3 months was 23% in those failing their first or second HAART and 10% in those failing three or more HAART regimens [odds ratio (OR) 2.79, 95% confidence interval (CI) 0.86–10.05;P = 0.06]; the proportions were 23 and 5% at 6 months, respectively (OR 6.14, 95% CI 1.34–38.95;P = 0.007). The effect of genotyping was also analysed in the subgroups of patients failing their first or second HAART regimen and in the subgroup failing three or more regimens (see Table 4).
The frequency of virological response at 3 months differed in relation to the patient-reported adherence category, with 21 out of 73 (29%) adherent patients and seven out of 54 (13%) non-adherent patients achieving HIV-RNA levels below 500 copies/ml. The responses according to adherence category in the different randomization arms are reported in Table 4. Adherent patients in the G arm had a significantly greater probability of 3 month virological success than non-adherent patients in the SOC arm (OR 5.80, 95% CI 1.07–41.41;P = 0.02), also in the SOC arm adherent patients had a greater probability of 3 month success than non-adherent patients (OR 4.17, 95% CI 0.70–31.84;P = 0.07).
Using an intent-to-treat last observation carried forward analysis, in the G arm, the mean changes from baseline CD4 cell counts were +9 (95% CI −18–+27) cells/μl and +15 (95% CI −10–+39) cells/μl at months 3 and 6, respectively. In the SOC arm, changes were +19 (95% CI −2–+39) cells/μl and +22 (95% CI −4–+49) cells/μl, respectively. There was no significant difference between the study arms.
When the patient-reported adherence category was used to analyse CD4 cell responses, the mean CD4 cell count changes at 3 months were +50 cells/μl in adherent patients and −12 cells/μl in non-adherent patients (P < 0.01), at 6 months the changes were +62 cells/μl in adherent patients and −13 cells/μl in non-adherent patients (P < 0.01) (see Fig. 2).
Predictors of virological success
Using bivariate logistic regression, we found the following factors showing association with virological response (HIV-RNA levels below 500 copies/ml) at 3 months: belonging to the transmission category of injecting drug users (OR 0.39, 95% CI 0.15–1.00;P = 0.05); a previous history of an HIV-RNA level of less than 500 copies/ml (OR 3.07, 1.36–6.90;P = 0.006); a greater number of experienced HAART regimens (for each more, OR 0.61, 95% CI 0.40–0.94;P = 0.023); a greater baseline viral load (for each log unit increase OR 0.45, 95% CI 0.23–0.90, P = 0.022); patient-reported non-adherence (OR 0.37, 95% CI 0.14–0.95, P = 0.038); the presence of all active drugs in the combination, according to the definition of primary resistance given in Table 1 (OR 4.36, 95% CI 0.96–19.86, P = 0.055); the presence of protease substitution L90M (OR 0.25, 95% CI 0.09–0.68, P = 0.007); and the total (primary plus secondary) number of PI mutations (for each PI mutation more, OR 0.86, 95% CI 0.74–1.01, P = 0.061). The total number of resistance mutations was not associated with virological response (P = 0.196). The time of enrolment in the trial showed an association with virological response (for each month more from trial initiation, OR 1.12, 95% CI 1.00–1.27, P = 0.057): this was observed only in the G arm (OR 1.19, 95% CI 1.01–1.40, P = 0.03), and not in the SOC arm (OR 1.06, 95% CI 0.87–1.30, P = 0.54). A greater number of active drugs was not significantly predictive of the virological success at 3 months (for each drug more, OR 1.37, 95% CI 0.90–2.10, P = 0.140), even when using the interpretation algorithm from the Resistance Collaborative Group  (for each unit sensitivity score increase, OR 1.34, 95% CI 0.89–2.03, P = 0.162).
As a result of the limited number of outcomes, we used two split multivariate models to detect the independent predictors of 3 month virological success (see Table 5). In a first model, we adjusted the randomization arm with variables that were unbalanced at baseline and showed significant odds of virological success at bivariate analysis, together with adherence-related factors. Results showed that being assigned to G and having a previous history of undetectable viral load independently predicted virological success, whereas patient-reported non-adherence independently predicted virological failure. In a second model, the randomization arm was adjusted with the baseline viral load, the number of previous HAART regimens, and the presence of all active drugs in the combination. The number of previous regimens, the baseline viral load and being assigned to G was independently predictive of the 3 month virological outcome (see Table 5). Finally, when analysing all covariates in a single model, being assigned to G retained a significant association with 3 month virological success (OR 2.67, 95% CI 1.09–6.54;P = 0.03).
The variables associated with virological success at 6 months at the bivariate analysis were: a previous history of an HIV-RNA level of less than 500 copies/ml (OR 3.87, 1.71–8.74;P = 0.001); a greater number of experienced HAART regimens (for each regimen more, OR 0.51, 95% CI 0.31–0.83;P = 0.006); a greater baseline viral load (for each log unit more, OR 0.40, 95% CI 0.16–0.71, P = 0.004); the presence of protease substitution L90M (OR 0.15, 95% CI 0.04–0.52, P = 0.003); the total (primary plus secondary) number of PI mutations (for each PI mutation more, OR 0.76, 95% CI 0.64–0.92, P = 0.003); and the total number of resistance mutations (for each mutation more, OR 0.87, 95% CI 0.77–0.97, P = 0.016). The number of active drugs, the randomization arm and patient-reported adherence were not associated with the 6 month virological outcome.
In the multivariate analysis, a previous history of undetectable viral load, the baseline viral load and the number of previously experienced HAART regimens remained the only variables independently predictive of the 6 month virological outcome (see Table 5).
This randomized study confirms the usefulness of genotype-guided treatment decisions in patients failing combination antiretroviral therapy. At 3 months, there was a significantly greater proportion of patients with HIV-RNA levels below 500 copies/ml in the G arm compared with the SOC arm. Multivariate analysis confirmed genotype information as being independently predictive of virological success. Nevertheless, the advantage of genotype information was lost after 6 months. These findings partly confirm results from previous randomized studies, indicating that resistance testing helps choose more effective salvage regimens in patients failing previous antiretroviral regimens. The VIRADAPT study showed that the treatment adaptation with the results from a genotypic assay allowed a better control of HIV replication up to 6 months over the standard of care . The CPCRA 046 study showed a significantly more profound viral inhibition in the G arm at 12 weeks, but the advantage tended to be reduced over time .
Some peculiarities that help interpret the partly discrepant results also distinguish the present study from other randomized trials. First, patients from published studies were less pretreated than patients from the present study. The proportion of patients failing two or more HAART regimens was greater (57%), with 25% failing three to seven regimens. Moreover, both the VIRADAPT study and the CPCRA 046 trial included fewer than 3% or no patients with previous NNRTI experience, whereas 41% of patients enrolled in the present trial had previous exposure to NNRTI, and 38% showed major baseline mutations conferring resistance to this class of drugs. Given the strong cross-resistance among NNRTI, this fact precludes the use of a whole new class of antiretroviral agents in choosing salvage therapy for these patients, which would be a determinant factor for a successful treatment modification . Furthermore, there was a larger proportion of L90M at baseline than in the VIRADAPT study; the presence of this primary mutation in the protease gene confers resistance or partial cross-resistance to all four PI available during the study period [25,26]. Finally, in contrast with CPCRA 046, in which the G arm received expert advice whereas the group assigned to SOC did not, in the present study treatment decisions were taken by the same panel in both randomization groups, therefore the difference between groups relied solely on the adjunctive information of the resistance assay.
This trial was conducted at a single site, and the fact that treatment decisions were taken by the same panel could have led to a contamination of knowledge from the G arm to the SOC arm, so that the benefit conferred by the genotype information could have been reduced as the trial continued and new patients were enrolled. Nevertheless, we found no correlation between the time of randomization and the use of new drugs or the number of active drugs in the salvage regimen in either arm. Furthermore, contrary to the formulated hypothesis, we found that a later time of enrolment was associated with a greater probability of virological success in the G but not in the SOC arm. This finding argues against the possibility of cross-arm contamination, and suggests that genotype-guided decisions could have been improved by increasing expert knowledge about genotypic resistance interpretation over time.
The fact that the advantage of using genotype information did not persist over time was not caused by the fact that the SOC arm could also receive genotype information in the case of a lack of response after the first 3 months, in fact such access was subject to specific request by the treating physician and only five patients in the control group did actually receive genotyping after this timepoint; the results did not change after excluding these patients from the analysis (not shown). Possible further reasons for the lack of a persistent virological advantage of the G arm could be the greater prevalence of baseline resistance mutations than in the control arm, in particular at the key codon 215 from the reverse transcriptase gene, involved in resistance to zidovudine and to some extent stavudine, and at codons 82 and 90 from the protease gene, involved together in viral resistance to all four PI available at the time of the study [9,24,25]. These characteristics further indicated that these patients had few residual treatment options, limiting the clinical usefulness of resistance genotyping. This is confirmed by the finding that a greater number of previously failed HAART regimens was also independently predictive of virological failure. In patients who have failed most if not all available agents, resistance assays will hardly indicate more effective alternative treatment opportunities . Preliminary results from a large, randomized trial in a setting of heavily pretreated individuals (NARVAL, ANRS 088)  also indicated that the virological advantage of genotype-driven therapy over SOC is marginal, in agreement with the findings from this study.
An important finding was that patient-reported adherence was an important factor influencing the virological and immunological responses to salvage therapy in both the G and in the control arm. Patients showing the best control of viral replication were those receiving genotyping and reporting adherence at baseline. On the other hand, those patients in the control group reporting non-adherence had the worst virological responses and those with the mixed characteristics (adherents from the control arm and non-adherents from the G arm) achieved intermediate virological inhibition. Moreover, on multivariable analysis, non-adherence was independently associated with a reduced probability of viral inhibition. The importance of adherence in predicting the virological response was already established in naive patients [5,28,29]. This study underscores the importance of adherence in the context of genotype-driven salvage therapy. More effective salvage strategies must first take into account adherence optimization when necessary, and subsequently resistance-based adaptation.
We also found that a history of virological inhibition before study entry predicted subsequent virological success: this might be interpreted both as a history of previous adherence to HAART, which might be more easily maintained over time and therefore predict better adherence to the new salvage regimen, as well as a lower probability of covert or archived resistance mutations, as shown partly by the lower mean number of total resistance mutations in this group. Patients randomly assigned with lower viral loads had a greater chance of reaching virological suppression, particularly when assigned to the genotype group. This suggests that, when a genotype combined with treatment history indicates that a new option is available with a relevant number of presumably active drugs, an early switch at low HIV-RNA levels should be encouraged.
In agreement with previous studies, despite the virological advantage of genotyping, CD4 cell responses did not differ between randomization groups. Although it can be hypothesized that patients had already reached substantial CD4 cell repopulation before study entry and that CD4 cell responses might take longer, this finding cannot be completely explained at this time. It can be speculated that the reduced viral fitness of resistant strains or reduced pathogenicity might compensate for the difference in viral inhibition observed [29,30]. Interestingly, there was a strong, significant difference in CD4 cell responses according to the category of patient-reported adherence. Patients reporting adherence at baseline showed a significant increase in CD4 cell counts both at 3 and 6 months, whereas patients reporting non-adherence showed a slight reduction in the mean CD4 cell count level over time. Contrary to virological responses, significantly better immunological responses in adherent patients were maintained over time. This phenomenon might have different explanations. In patients harbouring drug-resistant HIV, a sufficient selective pressure by antiretroviral agents might help reduce CD4 cell killing more effectively than reducing HIV replication, either by selecting less pathogenic viruses or by some direct inhibitory action of PI on apoptosis  (W. Malorni, et al., in preparation).
Virological responses in patients on salvage HIV therapy were significantly improved by the guide of a genotypic resistance test in a population of heavily experienced subjects, but this effect was lost over time. Responses were negatively influenced by patient-reported non-adherence, by the greater number of failed HAART regimens and by a greater viral load, whereas a previous history of viral inhibition on HAART was predictive of better virological responses. Although CD4 cell responses were not influenced by the genotype guide, these were extremely sensitive to patient adherence.
In order to optimize further the response to salvage therapy, genotypic resistance assays should be accompanied by a careful assessment and appropriate implementation of patient adherence. More information on the correct interpretation of genotypic mutations should also help improve the usefulness of genotypic resistance testing .
1. 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.
2. Finzi D, Blankson J, Siliciano JD. et al
. Latent infection of CD4+ T cells provides a mechanism for lifelong persistence of HIV-1, even in patients on effective combination therapy. Nat Med 1999, 5: 512–517.
3. Deeks S, 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–F44.
4. Fatkenheuer G, Theisen A, Rockstroh J. et al
. Virological treatment failure of protease inhibitor therapy in an unselected cohort of HIV-infected patients. AIDS 1997, 11: F113–F116.
5. Lucas GM, Chaisson RE, Moore RD. Highly active antiretroviral therapy in a large urban clinic: risk factors for virologic failure and adverse drug reactions. Ann Intern Med 1999, 131: 81–87.
6. Bangsberg DR, Hecht FM, Charlebois ED. et al
. Adherence to protease inhibitors, HIV-1 viral load, and development of drug resistance in an indigent population. AIDS 2000, 14: 357–366.
7. Durant J, Clevenbergh P, Garraffo R. et al
. Importance of protease inhibitor plasma levels in HIV-infected patients treated with genotypic-guided therapy: pharmacological data from the Viradapt Study. AIDS 2000, 14: 1333–1339.
8. Hirsch MS, Conway B, D'Aquila RT. et al
. Antiretroviral drug resistance testing in adult HIV-1 infection. JAMA 1998, 279: 1984–1991.
9. 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.
10. Shafer RW, Jung DR, Betts BJ, Xi Y, Gonzales MJ. Human immunodeficiency virus reverse transcriptase and protease sequence database. Nucl Acids Res 2000, 28: 346–348.
11. Pillay D, Taylor S, Richman DD. Incidence and impact of resistance against approved antiretroviral drugs. Rev Med Virol 2000, 10: 231–253.
12. Miller V, Ait-Khaled M, Stone C. et al
. HIV-1 reverse transcriptase (RT) genotype and susceptibility to RT inhibitors during abacavir monotherapy and combination therapy. AIDS 2000, 14: 163–171.
13. Harrigan PR, Montaner JS, Hogg RS. et al
. Baseline HIV drug resistance profile predicts response to ritonavir/saquinavir protease inhibitors therapy in a community setting. AIDS 1999, 19: 1863–1871.
14. Zolopa AR, Shafer RW, 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.
15. 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. Antivir Ther 2000, 5: 41–48.
16. Lorenzi P, Opravil M, Hirschel B. et al
. Impact of drug resistance mutations on virologic response to salvage therapy. Swiss HIV Cohort Study.
AIDS 1999, 13: F17–F21.
17. Kuritzkes DR, Sevin A, Young B. et al
. Effect of zidovudine resistance mutations on virologic response to treatment with zidovudine–lamivudine–ritonavir: genotypic analysis of human immunodeficiency virus type 1 isolates from AIDS clinical trials group protocol 315. ACTG Protocol 315 Team.
J Infect Dis 2000, 181: 491–497.
18. Tamalet C, Pasquier C, Yahi N. et al
. Prevalence of drug resistant mutants and virological response to combination therapy in patients with primary HIV-1 infection. J Med Virol 2000, 61: 181–186.
19. Durant J, Clevenbergh P, Halfon P. et al
. Drug-resistance genotyping in HIV-1 therapy: the VIRADAPT randomised controlled study. Lancet 1999, 353: 2195–2199.
20. Baxter JD, Mayers BL, Wentworth DN. et al
. A randomized study of antiretroviral management based on plasma genotypic resistance testing in patients failing therapy. AIDS 2000, 14: F83–F93.
21. Meynard JL, Vray M, Morand-Joubert L. et al
. Impact of treatment guided by phenotypic or genotypic resistance tests on the response to antiretroviral therapy: a randomized trial (NARVAL, ANRS 088). Antivir Ther 2000, 5 (Suppl. 3) : 67.67.
22. Tural C, Ruiz L, Holtzer C, et al. Utility of HIV genotyping and clinical expert advice – the Havana Trial.
In:Program and abstracts of the 8th Conference on Retroviruses and Opportunistic Infections.
Chicago, 2001 [Abstract 434].
23. Murri R, Ammassari A, Gallicano K. et al
. Patient-reported nonadherence to HAART is related to protease inhibitors levels. J Acquir Immunodefic Syndr 2000, 24: 123–128.
24. Weidle PJ, Lichtenstein KA, Moorman AC. et al
. Factors associated with the successful modification of antiretroviral therapy. AIDS 2000, 14: 491–497.
25. Lawrence J, Schapiro J, Winters M. et al
. Clinical resistance patterns and responses to two sequential protease inhibitor regimens in saquinavir and reverse transcriptase inhibitor-experienced persons. J Infect Dis 1999, 179: 1356–1364.
26. Churchill DR, Pym AS, Galpin S. et al
. The Rabbit study: ritonavir and saquinavir in combination in saquinavir-experienced and previously untreated patients. AIDS Res Hum Retroviruses 1999, 15: 1181–1189.
27. Harrigan PR, Coté H. Clinical utility of testing human immunodeficiency virus for drug resistance. Clin Infect Dis 2000, 30 (Suppl 2) : S117–S122.
28. Montaner JS, Reiss P, Cooper D. et al
. A randomized, double-blind trial comparing combinations of nevirapine, didanosine, and zidovudine for HIV-infected patients: the INCAS Trial. Italy, The Netherlands, Canada and Australia Study.
JAMA 1998, 279: 930–937.
29. Haubrich RH, Little SJ, Currier JS. et al
. The value of patient-reported adherence to antiretroviral therapy in predicting virologic and immunologic response. California Collaborative Treatment Group.
AIDS 1999, 13: 1099–1107.
30. Erickson JW, Gulnik SV, Markowitz M. Protease inhibitors. Resistance, cross-resistance, fitness and the choice of initial and salvage therapies.
AIDS 1999, 13 (Suppl A) : S189–S204.
31. Stoddart C, Liegler TJ, Mammano F. et al
. Impaired replication of protease inhibitor-resistant HIV-1 in human thymus. Nat Med 2001, 7: 712–718.
32. De Luca A, Cingolani A, Rizzo MG. et al
. Prediction of treatment outcomes by different interpretation systems (IS) for baseline genotypic resistance in a cohort of patients on salvage HIV therapy. Antivir Ther 2001, 6 (Suppl 1) : S33.S33.
Keywords:© 2002 Lippincott Williams & Wilkins, Inc.
Drug resistance; genotypic resistance; adherence; antiretroviral-therapy; salvage therapy