Triple nucleoside combinations fail patients again

Simon Collins, HIV i-Base

Two studies at the IAS meetings using triple nucleoside combinations reported very poor results and a third study was closed the week before the meeting on a preliminary analysis of the results.

This article will report on all three studies.

Trizivir only arm discontinued in ACTG 5095

In April 2003, the Trizivir arm of the ACTG 5095 study was closed by the trials DSMB (Data and Safety Monitoring Board) because patients receiving the triple nucleoside combination were achieving significantly poorer results than the efavirenz containing arms. The efavirenz arms remain unblended and are continuing. (see HTB April 2003)

These data very quickly resulted in Trizivir unsupported by other drugs being dropped as a preferred option for first line therapy in both the new UK and US treatment guidelines, both produced this summer.

The IAS meeting provided the first opportunity for the results from this study to be presented publicly at a large conference. [1]

ACTG 5085 was a double-blind, placebo trial which randomised 1,147 treatment-naïve patients in 1:1:1 ration to either AZT/3TC/abacavir (Trizivir) or AZT/3TC/efavarenz or the four-drug regimen of AZT/3TC/abacavir/efavirenz.

The study population was 81% male, 40% white, 36% Black, 21% Latin. Eleven percent were IVDUs. Mean baseline viral load was 4.9 log copies/mL and 43% of patients were >100,000 copies/mL. Median CD4 count was 238 copies/mm3.

One hundred and sixty-seven patients reached virological failure at the week 32 analysis (defined as confirmed VL >200 at or after 16 weeks). This occurred in 21% of the Trizivir arm (n=82) compared to 10% in the pooled efavirenz-containing arms (n=85). Time to virological failure was shorter with the triple nucleoside arm compared to the pooled efavirenz arms (p<0.001) and this was true for both baseline viral load > or <100,000 c/mL (p<0.001 for each). It was also shorted for patients who achieved viral load <200 and then rebounded. The proportion of patients with viral load <200 at week 48 in the Trizivir and pooled-efavirenz arms was 74% versus 89% respectively. All analyses are ITT.

Very few patients discontinued due to side effects. After a median 32 wks of follow-up, 93% of patients continued on study and 91% continued study drugs. Grade 3 and 4 signs/symptoms occurred in 12% and 2%, with comparable proportions across study arms, so the main concern focused on efficacy.

Adherence (undefined) was a remarkable 100% but resistance testing of people failing in the Trizivir arm showed 22% failing with wild type virus. About half failing with M184V with or without other RTI mutations. 27% still had <500 copies/ml so were not available for resistance testing.

As the other arms in the study continue, no comparative information is available.

Tenofovir/abacavir/3TC – clearly not recommended

Results from a pilot triple-nuke study of once-daily tenofovir+abacavir+3TC in 20 treatment naïve patients were presented by Charles Farthing. This study was also stopped after the high early failure rate became clear. [2]

The rationale appears to have been based on adding the cumulative antiviral potency of the individual drugs rather a review of previous triple-nucleoside studies – because many of the individuals selected for the study had baseline viral load >100,000 c/mL. In the Atlantic study and previous Trizivir studies higher baseline viral load has correlated with higher failure rates.

Median baseline viral load was approximately 82,000 (range,7650-213,000) but 9/20 were >100,000 copies/mL. Median CD4 count was 59 cells/mm3 (range 59-598.

Of 17 patients who remained in the study, nine (52%) had viral rebound: one at week four, six at week eight and two at week 16. Patients were >95% adherent by pill count over this short period.

In the same week these results were presented, GlaxoSmithKline (GSK) announced that they had been running a larger study in the US (ESS 30,009) comparing abacavir/3TC/tenofovir to abacavir/3TC/efavirenz, again all once daily, with approximately 180 patients in each arm. An interim analysis of their data had showed similarly disastrous results – this has got to be one of the worst sets of data on treatment naïve patients presented to a conference for a long time – and they were therefore closing this study. [3]

Baseline viral load levels for patients in this study outlined in a hurriedly called community meeting showed a patient group who arguably should have been even less exposed to this experimental approach. Although baseline statistics were not presented, plotted individual results showed many patients started treatment with viral load well above 100,000 copies/ml and many with counts into the millions.

Nevertheless, the analysis of the first approximately 200 patients showed that the difference in efficacy again appears early and by week eight only 19% of the triple nucleoside patients had viral load <50 copies compared to 37% of the efavirenz group. By week 16 the difference widened further to 30% versus 95% (though admittedly with small numbers).

At the analysis, 49% of the triple nucleoside group, compared to only 5% of the efavirenz arms, had failed by one or other of the protocol defined definitions of failure. By week eight for example, 31% versus 3% failed to see a 2 log drop in viral load in the triple—nuke and efavirenz arms respectively. Resistance data on seven patients whose genotype was sequenced showed that all seven failed with 184V and three with mixed K65R/K and four with K65R.

This study has now stopped.

Several reasons for the failure of the abacavir/tenofovir combinations were discussed at each meeting.

  • one suggestion is purely related to triple-nucleoside single target therapy as a less potent strategy, and previous studies have shown this with other combinations as well as Trizivir. Indeed, Trizivir has always had a caution in treatment guidelines that it is less potent with high viral loads. (Do guidelines not apply to patients in studies?)
  • a second possibility is an inherent interaction between abacavir and tenofovir, possible at intracellular level, similar perhaps to that between AZT and d4T.
  • a third factor may be several overlapping resistance profiles, compounded perhaps by other factors (ie 3TC and abacavir overlap with M184V and tenofovir and abacavir overlap with K65R)
  • a fourth factor may be that these studies used once-daily abacavir and the intracellular and other data supporting this has not yet been approved by either European or US regulatory agencies.


This data has been sufficient for the European Medicines Evaluation Agency to issue it’s own statement as we went to press: see Treatment Alert on page 3.

It would be important to know whether ongoing studies with Trizivir and tenofovir, or other combinations including both abacavir and tenofovir have raised similar concerns.


Unless stated otherwise, references are to the programme and abstracts of the 2nd IAS Conference on HIV Pathogenesis and Treatment, 13-16 July 2003, Paris.

  1. Gulick RM, Ribaudo HJ, Shikuma CM et al – ACTG 5095: a comparative study of three PI-sparing antiretroviral regimens for the initial treatment of HIV infection. Abstract 41.
  2. Farthing C, Khanlou H, Yeh V – Early virologic failure in a pilot study evaluating the efficacy of abacavir, lamivudine and tenofovir in the treatment of HIV-infected patients. Abstract 43.
  3. Data from GSK presented to EATG community meeting, Paris.


Previous report of ACTG 5095 Trizivir closure:

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