HIV/AIDS Express Report
The XII International HIV Drug Resistance Workshop
Cabo del Sol, Los Cabos, Mexico

The Emerging Evidence of the Relationship between Adherence and Drug Resistance

Expert Commentary

Daniel Kuritzkes, MD, Director of AIDS Research, Brigham and Women's Hospital, Section of Retroviral Therapeutics, Partners AIDS Research Center, Associate Professor of Medicine, Harvard Medical School, Boston, Massachusetts

The availability of potent combination antiretroviral therapy has resulted in substantial reductions in AIDS-related morbidity and mortality in the developed world. Emergence of drug-resistant virus may limit the benefits of antiretroviral therapy in some patients. Several abstracts presented at The XII International HIV Drug Resistance Workshop reported results of studies related to the development and persistence of drug resistance.

Data from Susan Little and colleagues show that drug resistant variants may persist for up to three years following transmission into a new host. These findings have important implications for the use of drug resistance testing prior to starting antiretroviral therapy. Indeed, guidelines for resistance testing are changing to take these observations into account. Even when not detected, resistant variants can influence treatment outcome. The study by John Mellors and colleagues demonstrates the importance of efavirenz-resistant variants present as a minority species in non-nucleoside reverse transcriptase (NNRTI) inhibitor-experienced patients with apparently sensitive virus by standard resistance testing. Presence of the resistant minority variants significantly increased the risk of failure on efavirenz-containing salvage therapy. Pre-existing mutants also determine the risk of developing high-level lopinavir resistance during salvage therapy. King et al showed that patients at greatest risk of developing increased resistance to lopinavir were those whose virus carried at least 4 protease inhibitor (PI) mutations at baseline and had 40- to 60-fold reduced susceptibility to lopinavir. Lastly, the study by Kempf and colleagues shows the complex relationship between adherence and development of drug resistance. The risk of resistance was greatest in patients with intermediate levels of adherence to a nelfinavir-containing regimen. Once that regimen failed, continued adherence only increased the likelihood that drug resistance would develop. By contrast, a regimen that included lopinavir/ritonavir was unlikely to select for resistant virus at any level of adherence. These data might explain the high success rate and low resistance cost of ritonavir-boosted regimens, specifically lopinavir/ritonavir.

Relationship between Adherence and Resistance

The relationship of adherence to virologic response has been well documented for PI-based therapy, with lower adherence consistently resulting in higher viral loads or higher probability of virologic failure.1-3 However, the relationship between adherence and the development of viral resistance is less well studied. Bangsberg et al have demonstrated a bell-shaped relationship between adherence and viral resistance.4 Results of a local linear regression analysis by King and colleagues of the M98-863 Study have confirmed this relationship.5

M98-863 was a double-blind, phase 3 clinical trial in which 653 HIV-infected adults were randomized to either lopinavir/ritonavir (LPV/r; Kaletra) (400 mg/100 mg) twice daily with nelfinavir (Viracept) placebo or nelfinavir (750 mg) three times daily with LPV/r placebo.6 All patients received stavudine (40 mg twice daily) and lamivudine (150 mg twice daily). Patients with HIV RNA levels above 400 copies/mL at Week 24 of the study or later (up to 2 years follow-up), had samples sent for genotypic resistance testing. In the present analysis, the relationship of adherence to the probability of resistance development (HIV RNA >400 copies/mL any time at or after Week 24) was assessed by logistic regression.

Primary PI resistance was defined as the emergence of a D30N or L90M mutation or M46I/L plus confirmed phenotypic resistance for nelfinavir-treated patients or any primary or active site mutation (8, 30, 32, 46, 47, 48, 50, 54, 82, 84, and 90) for lopinavir. Secondary PI mutations were defined as those appearing at positions 10, 20, 33, 53, 71, 73, 77, and 88, plus positions 46 and 54 for nelfinavir. Lamivudine resistance was defined as the presence of the M184V/I/T mutation in reverse transcriptase.

Adherence was calculated as the number of pills consumed relative to the number of pills expected to be consumed. Pill counts of returned study drug were conducted at each study visit. For PI adherence, only the active PI was used in the assessment of adherence.

Based on logistic regression analysis, adherence was significantly associated with the probability of HIV RNA >400 copies/mL any time at or after Week 24 (P<.001). After adjusting for adherence, the probability of detectable HIV RNA remained lower for LPV/r-treated patients versus nelfinavir-treated patients (P<.001).

Probability of PI Resistance

For nelfinavir-treated patients, a bell-shaped relationship emerged between adherence and the probability of the emergence of primary PI mutations (Figure 1). The maximal probability of resistance was approximately 20%, occurring at an adherence level of 85-90%. No LPV/r-treated patient demonstrated the emergence of primary PI mutations resulting in an estimated probability of zero at all adherence levels.

For nelfinavir-treated patients, the maximal risk of secondary PI mutations was approximately 20% at 85-90% adherence (Figure 2). The shape of the secondary resistance/adherence curve could not be precisely defined for the LPV/r-treated patients since relatively few demonstrated new secondary PI mutations. The maximal risk was approximately 5% at an adherence level of 80-85%.

Probability of Resistance among Viremic Patients

Among viremic nelfinavir-treated patients with genotype data, the probability of the emergence of primary PI resistance mutations increased with increasing adherence, reaching a maximum of approximately 65% at 100% adherence (Figure 3).

Likewise for nelfinavir-treated patients, the probability of secondary PI mutations increased with increasing adherence, reaching a maximum of approximately 60% at 100% adherence (Figure 4). Thus, even at 100% adherence, resistance developed in the less potent regimen.

For viremic LPV/r-treated patients with genotype data, the probability of secondary mutations remained relatively stable across all adherence levels, generally between 10 to 15% at adherence levels between 75 to 95% but lower outside that adherence range (Figure 4).

The bell-shaped relationship between adherence and the development of viral resistance leads to several observations: the probability of resistance is low at low adherence levels because while viral load is more likely to be detectable, selective pressure is absent; the probability of resistance is low at high adherence levels because while selective pressure is high, viral replication is less likely; and at intermediate adherence, patients are most likely to have ongoing viral replication in the presence of selective pressure, leading to a maximal probability of resistance development. Moreover, the risk of resistance development appears to be influenced by lower drug potency allowing viral replication in the presence of drug and a lower genetic barrier to resistance in patients without complete viral suppression.

Prevalence and Persistence of Transmitted Drug-resistant Virus

According to results presented by Susan J. Little, MD, of the University of California—San Diego Antiviral Research Center, San Diego, California, the persistence of transmitted drug-resistant variants has significant implications for the treatment of antiretroviral-naпve subjects and subsequent secondary transmission of drug-resistant variants.7

A total of 10 subjects with primary HIV infection who chose to defer antiretroviral underwent baseline nucleotide sequence analysis of pol to identify primary drug resistance mutations. All 10 subjects (mean time from estimated date of infection—65 days) had at least one major drug resistance mutation identified at baseline with corresponding phenotypic resistance. Seven subjects were identified with NNRTI resistance (103N ± 181C), one with nucleoside reverse transcriptase inhibitor (NRTI) (70R, 74I, 215Y) and PI resistance (46I, 84V, 90M), one with NNRTI (188L) and PI resistance (30N), and one with 3-class drug resistance (103N, 215Y, 84V, 90M). The average time to reversion of 103N variants to mixed 103N/K populations was 196 days following the estimated date of infection (95% Confidence Interval (CI), 153-238 days). In the 3 patients with PI mutations, no reversion was detectable at 64, 191, and 342 days after infection. Transmitted NRTI, NNRTI and PI drug resistance persisted during the entire period of follow-up for 9 of 10 subjects with complete reversion of a transmitted K103N variant documented in only one subject, nearly 3 years after infection.

The durable persistence of transmitted drug-resistant virus is consistent with the establishment of widespread infection with a pure population of resistant clone(s), in contrast to the rapid reversion observed in chronically infected subjects who discontinue therapy after virologic failure. These results corroborate previously published data by Dr. Little that showed initial antiretroviral therapy is more likely to fail in patients who are infected with drug-resistant virus.8 Those results demonstrated that time to viral suppression was significantly shorter among subjects with fully susceptible virus at baseline than among those with a major drug-resistance mutation at baseline (P = .03) and time to virologic failure was significantly shorter among those with a 50 percent inhibitory concentration (IC50) ratio of more than 10 than among those with completely drug-susceptible virus at baseline (P = .05).

The results of both studies suggest that resistance testing should be recommended routinely for patients with new HIV infection. In subjects with primary HIV infection with drug-resistant virus, plasma HIV RNA was not suppressed as readily by potent antiretroviral drugs which may in turn, permit additional viral replication facilitating the selection of variants with greater drug resistance. These results take on clinical relevance in light of data demonstrating that a major source of new drug-resistant infections is the result of high-risk HIV sexual transmission behaviors.9

Analysis of Incremental Lopinavir Resistance in PI-experienced Patients

The results of an investigative analysis indicate that in PI-experienced patients receiving LPV/r, the likelihood of emergence of additional resistance during virologic failure appears to be dependent upon both baseline genotype and phenotype.10 In previous studies of antiretroviral-naпve patients treated with a LPV/r-based regimen, the development of resistance to lopinavir has not been observed.6,11,12 In contrast, the development of resistance to LPV/r has been observed in PI-experienced patients.13 The present analysis explored the selection of incremental lopinavir resistance in those patients during failure of LPV/r therapy as an alternative method for estimating an upper breakpoint for lopinavir.

For analysis of genotype and phenotype, samples were selected from among patients in the above-referenced studies demonstrating virologic rebound or incomplete virologic response. Selection of incremental resistance was defined as having satisfied any of the following: (1) emergence of a new primary PI mutation (D30N, V32I, G48V, I50V, V82A/F/T/S, I84V, L90M); (2) emergence of a new secondary mutation that is not normally observed as a polymorphism (L24I, L33F, M46I/L, I47A/V, I54A/V/L, N88D); and (3) emergence of any other secondary mutation (L10F/I/R/V, K20M/R, M36I, A71V/T, G73S/A, V77I) accompanied by a ≥2-fold change in lopinavir IC50 between baseline and rebound.

Baseline and rebound genotypic results were available from 53 patients (40 single PI-experienced and 13 multiple-PI experienced). Phenotypic results were available from all 53 patients at rebound and from 44 patients at baseline. Selection of additional lopinavir resistance was observed in 19 patients (36%) with viral rebound resistance data available, including 14 (35%) single PI-experienced patients and 5 (38%) multiple-PI experienced patients.

A second-order logistic regression model indicated that maximal selective pressure (highest probability of additional lopinavir resistance) at 4-6 baseline PI mutations with little selective pressure below 2 or above 7 PI mutations. Thus, no lopinavir resistance emerged in the rebound isolates from the 13 patients with 0-1 baseline PI mutations and suggests a high pharmacologic barrier to resistance of lopinavir. This is consistent with results in antiretroviral-naпve patients, where resistance to LPV/r has not been observed to date. Thus, the optimal response to LPV/r therapy would be observed in antiretroviral-naпve patients and patients with 0-1 baseline mutations. In contrast, the selection of additional resistance was evident from patients with 2-3, 4-5, 6-7 and ≥8 baseline PI mutations and suggestive of a compromised pharmacologic barrier to resistance.

A second-order logistic regression model suggested a substantial drop in selective pressure beginning at 40- to 60-fold reduced baseline susceptibility to lopinavir. The probabilities of incremental selection of lopinavir resistance in patients with 40-, 60-, and 80-fold baseline lopinavir IC50 were 46% (95% CI, 25-72%), 31% (95% CI, 11-63%), and 20% (95% CI, 5-56%), respectively. Among patients with ≥4 baseline PI mutations, additional resistance was selected in patients <40-fold, 40- to 60-fold, and >60-fold baseline reduced susceptibility to lopinavir. The magnitude of additional phenotypic lopinavir resistance was highest among patients with at least 4 baseline PI mutations but <60-fold baseline reduced susceptibility to lopinavir. Therefore, the upper clinical breakpoint for LPV/r is derived primarily from patients with 4 or more baseline mutations, where the pharmacologic barrier to resistance is expected to be significantly eroded. Therefore, the selection of resistance by LPV/r is most likely in patients with baseline lopinavir susceptibility of ≤40- to 60-fold and in patients with 4 to 7 baseline PI mutations. Mean fold change in lopinavir resistance is illustrated in Figure 5.

The presence of 2-3 mutations at baseline compromises susceptibility to LPV/r, whereas 4 or more mutations has a significant impact on the magnitude of additional phenotypic LPV resistance. These results are consistent with the Inhibitory Quotient (IQ) pharmacological model for LPV/r activity.

Low Frequency NNRTI-resistant Variants Contribute to Failure of Efavirenz Regimens

According to the results of new study, prior NNRTI experience selects minor NNRTI-resistant variants that are often missed by standard genotyping leading to failure of efavirenz-based regimens.14 A total of 212 NNRTI-experienced and 269 NNRTI-naпve patients were randomized to efavirenz, abacavir, adefovir and amprenavir with a second PI or placebo. The study examined the relationships between NNRTI experience, baseline NNRTI resistance, virological response and the emergence of efavirenz resistance.

Genotypes of baseline plasma were obtained in 452 of 481 patients. Minor NNRTI-resistant variants were sought through two methods: single genome RT-PCR and sequencing and chimeric Ty1/HIV-1 RT retrotransposon system that measured the frequency of efavirenz resistance.

Virologic failure (defined as confirmed HIV RNA >200 copies/mL) was associated with NNRTI experience, baseline NNRTI mutations, and the development of efavirenz resistance (all P<.001). Standard genotyping did not detect NNRTI mutations in baseline samples from 50 of 216 (23%) NNRTI-experienced patients. Virologic outcome for these patients was no better than for patients with identified baseline NNRTI mutations (n = 166). In contrast, among all patients with negative baseline genotypes for NNRTI mutation, virologic outcome was significantly better at 24 weeks (P = 0.15) and 48 weeks (P = .02) in NRRTI-naпve patients (n = 237) compared with NNRTI-experienced patients (n = 50). Investigators suggested that standard genotyping may not have adequately detected NNRTI-resistant variants.

Baseline plasma from 10 NNRTI-experienced and 8 NNRTI-naпve patients who experienced virological failure despite a negative baseline genotype for NNRTI mutations, was tested for minor NNRTI-resistance mutations. Variants encoding NNRTI-resistance mutations were identified by sequencing in 6 of 10 NNRTI-experienced patients with the following frequencies per positive patient: 181C and 190A (5 of 15 sequences); 181C (3 of 19); 181C (3 of 22); 108I (2 of 35); 103N (1 of 33); and 103N (1 of 34). By comparison, NNRTI-resistant variants were found in only 1 of 8 NNRTI-naпve patients (100I, 1 of 33 sequences). The Ty1/HIV-1 RT assay detected efavirenz-resistant colonies in 8 of 10 NNRTI-experienced patients with the following frequencies: 10.9, 6.7, 6.4, 3.3, 2.0, 1.6, 1.3, and 0.8%. In NNRTI-naпve patients, resistant colonies were found in 2 of 8 patients with frequencies of 0.6 and 0.3%.

The investigators concluded that prior NNRTI experience selects minor NNRTI-resistant variants that are often missed by standard genotyping. The undetected variants may lead to failure of efavirenz-based regimens.


1. 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.
2. Bangsberg DR, Hecht FM, Charlebois ED, et al. Adherence to protease inhibitor, HIV-1 viral load, and development of drug resistance in an indigent population. AIDS. 2000;14:357-366.
3. Paterson DL, Swindells S, Mohr J, et al. Adherence to protease inhibitor therapy and outcomes in patients with HIV infection. Ann Intern Med. 2000;133:21-30. Erratum in: Ann Intern Med. 2002;136:253.
4. Bangsberg DR, Kagay CR, Porco T, et al. Modelling the relationship between adherence and accumulation of protease inhibitor drug-resistance mutations based on objectively measured adherence and empirically derived relationships. Antiviral Therapy. 2002;7:S133. Abstract 160.
5. King M, Brun S, Tschamap J, Moseley, Kempf D. Exploring the effects of adherence on resistance: use of local linear regression to reveal relationships between adherence and resistance in antiretroviral-naпve patients treated with lopinavir/ritonavir or nelfinavir. Presented at The XII International Drug Resistance Workshop, June 10-14, 2003, Cabo del Sol, Los Cabos, Mexico.
6. Walmsley S, Bernstein B, King M, et al for the M98-863 Study Team. Lopinavir-ritonavir versus nelfinavir for the initial treatment of HIV infection. N Engl J Med. 2002;346:2039-2046.
7. Little SJ, Dawson K, Hellmann NS, Richman DD, Frost SDW. Persistence of transmitted drug-resistant virus among subjects with primary HIV infection deferring antiretroviral therapy. Presented at The XII International Drug Resistance Workshop, June 10-14, 2003, Cabo del Sol, Los Cabos, Mexico. Abstract 115.
8. Little SJ, Holte S, Routy JP, et al. Antiretroviral-drug resistance among patients recently infected with HIV. N Engl J Med. 2002;347:385-394.
9. Kozal M, Amico KR, Chiarella J, et al. Patients with antiretroviral-resistant HIV infections engaging in high-risk transmission behavior. Presented at The XII International Drug Resistance Workshop, June 10-14, 2003, Cabo del Sol, Los Cabos, Mexico. Abstract 116.
10. King M, Mo H, Molla A, Stevens R, Brun S, Kempf D. Exploration of the methodology for estimating upper clinical breakpoints for lopinavir/ritonavir by analysis of the emergence of resistance during virologic failure in experienced patients. Presented at The XII International Drug Resistance Workshop, June 10-14, 2003, Cabo del Sol, Los Cabos, Mexico. Abstract 106.
11. Kempf D, King M, Bauer E, et al. Comparative incidence and temporal accumulation of PI and an RTI resistance in HIV-infected subjects receiving lopinavir/ritonavir or nelfinavir as initial therapy. 10th Conference on Retroviruses and Opportunistic Infections, Boston, Massachusetts, February 10-14, 2003. Abstract 600.
12. Stevens RC, Cernohous P, King M, et al. SOKRATES: prospective clinical trials to investigate the evolution of protease resistance during lopinavir/ritonavir treatment. First European HIV Drug Resistance Workshop, Luxembourg, France, March 6-8, 2003. Abstract 214.
13. Kempf D, Isaacson JD, King MS, et al. Analysis of the virologic response with respect to baseline viral phenotype and genotype and protease-inhibitor-experienced HIV-1-infected patients receiving lopinavir/ritonavir therapy. Antiviral Therapy. 2002;7:165-174.
14. Mellors J, Palmer S, Nissley D, et al for the ACTG 398 Study Team. Low frequency non-nucleoside reverse transcriptase inhibitor (NNRTI)-resistant variants contribute to failure of efavirenz-containing regimens in NNRTI-experienced patients with negative standard genotypes for NNRTI mutations. Presented at The XII International Drug Resistance Workshop, June 10-14, 2003, Cabo del Sol, Los Cabos, Mexico. Abstract 134.