Risk of NNRTI-resistance during STI: implications for SMART and other STI studies

Simon Collins, HIV i-Base

Several studies addressed the risk of resistance to NNRTIs developing during structured treatment interruptions (STIs).

Interest in two strategies using STIs: stimulating stronger HIV specific immune responses, and reselecting for wild-type virus prior to salvage therapy, is now largely reduced. However, various trials are underway that cycle treatment in order to reduce both toxicity and costs associated with long-term chronic treatment.

The largest of these is the international SMART study in which HAART is initiated and discontinued by CD4 count and response. This study is now recruiting patients in the UK. Several studies at the rsistance meeting provided important information that should inform patient management in this study.

Mireia Arnedo and colleagues from the University of Barcelona analysed baseline and post-interruption drug resistance in both plasma and proviral DNA, from 112 STI cycles in 35 patients from four clinical studies. All drugs were stopped at the same time.

Overall, 9/35 (26%) patients selected resistance mutations during 20/112 (18%) cycles just of half of which were not detectable at baseline in plasma or proviral DNA. Half the patients using 3TC and a quarter using NNRTIs showed resistance, approximately half of which were new mutations. Detailed results are shown in the Table 1 below.

Table 1: New or archived mutations in four STI studies

Total New Archived
RTI (exc-184V) 6% (2/35) 3% 3%
RTI M184V 50% (9/18) 22% 28%
NNRTI 23% (3/13) 23% 0
PI 0% 0 0
Total 26% (9/35) 14% 12%

Results were also presented from a subgroup of eight patients who had genotypic and phenotypic samples tested at maximum levels of viral rebound, and these showed a good correlation between measures.

Lucia Palmisano and colleagues from Istituto Superiore di Sanita, Rome, presented further analysis from a large Italian study that randomised patients on successful HAART to either continue treatment of follow a course of four interruptions (of 1, 1, 2 and 2 month duration) at three month intervals.

Virological resistance occurred in 18% of the STI arm. Samples from a subgroup of 49 patients, with PBMC samples at baseline, half of whom developed resistance during the study were stratified by mutations detected during the STI. HIV RNA was measured with a test with a cut off of 3.5 copies/mL. 33% patients with mutations failed subsequent treatment compared to 12% of those who showed no resistance (p= 0.004). Resistance in resistance in baseline plasma and PBMC correlated.

This link to subsequent failure as a strategy to reduce side effects has ended by limiting long-term treatment options. This is a particular concern given that the study protocol for stopping treatment in this study had already tried to minimise risk of resistance to NNRTIs with long half-lives by stopping nevirapine three days earlier, and efavarenz one week earlier, than the nucleosides in each regimen.


The potential for NNRTI-resistance due to monotherapy when these drugs are discontinued is now widely recognised. Although many doctors already advise discontinuing the NNRTI one week earlier than their nucleosides, this may still not be sufficient as recent studies showed some patients can maintain therapeutic levels for several weeks after stopping treatment. [3]

Guidelines for stopping treatment within intermittent treatment studies such need to ensure resistance is minimised in patients in the intermittent treatment arms. The international SMART study randomises patients on stable therapy to either maintain continuous treatment, or start and stop treatment by predetermined levels of CD4 count (250 and 350 cells/mm3 respectively).

This study has excellent protocols for stopping NNRTI-based treatment, recommending switching the NNRTI to a (preferably) boosted PI for the last three weeks of treatment. They work intensively with site investigators to ensure that doctors with patents in the intermittent arm stop patients in a way that minimises the risk of resistance. The study is also collecting data to record and evaluate these stopping strategies.

Resistance developed in any STI study confounds the real question on the potential benefit of CD4-driven treatment management and reducing long-term cumulative exposure to antiretrovirals. It also jeopardises future options for these patients.

This is also an important issue for patients to understand. If the safest way to avoid resistance is to switch the NNRTI to a PI for the last few weeks, this may involve additional short-term side effects such as nausea, diarrhoea, etc. The potential risk of resistance and implications for their next period on NNRTI is actually far more serious than the short-term side effects, but the temptation will be to stop the PI treatment too early.

Nucelosides with long half-lives and low genetic barriers to resistance such as 3TC clearly also need to be considered. Although tenofovir has a long half-life, it has a higher barrier to resistance than 3TC. In the discussion, Michael Miller form Gilead reported that they have not seen new resistance in the limited numbers of patients stopping tenofovir in registrational studies, but it would be useful to see this reported as a separate study.



  1. Arnedo M, Garcia F, Gil C et al. Risk of developing selected de novo resistance mutations during structured therapy interruption (STI) in chronic HIV-1 infection. XIII International HIV Resistance Workshop 8-12 June 2004, Tenerife. Abstract 142. Antiviral Therapy 2004; 9:S158.
  2. Palmisano L, Giuliano M, Pirillo MF et al. Baseline predictors and virological outcome in subjects developing mutations during intermittent HAART. XIII International HIV Resistance Workshop 8-12 June 2004, Tenerife. Abstract 143. Antiviral Therapy 2004; 9:S159.
  3. Taylor S, Allen S, Fidler S et al. Stop study: after discontinuation of efavirenz, plasma concentrations may persist for two weeks or longer. 11th CROI 2004, Abstract 131. For report see:

Links to other websites are current at date of posting but not maintained.