Between July 1999 and September 2003, 121 subjects were enrolled in ACTG Protocol 371 at 15 sites in the United States. The leading sites of accrual were the Bellevue Hospital Center (24 subjects, 20%) and the University of California San Diego Medical Center (23 subjects, 19%). The total sample included 50 entered in the AI stratum and 71 in the RI stratum. All subjects began treatment within three days of enrollment. A total of 73 subjects (28 AI and 45 RI) entered the first treatment interruption (step 3). Of these, two did not register for this step but met all criteria including treatment discontinuation and were included in data analysis.
Baseline characteristics of the 121 subjects enrolled are described in . Overall, 115 (95%) were male. The median age was 34 years. Most (84, 69%) were white and most (114, 94%) had no history of injection drug use. The median overall CD4 cell count was 535 cells/mm3 with a range of 87 to 1835 cells/mm3 and not significantly different between the AI and RI groups. The overall median plasma HIV RNA concentration was five-fold higher for the AI stratum (210,000 copies/mL) compared to the RI stratum (43,000 copies/mL). displays the baseline characteristics for the 73 subjects who entered the first treatment interruption and who comprise the analysis group for the primary efficacy endpoint. The median CD4+ T cells prior to treatment interruption were 893 for AI and 829 for RI. The 48 subjects not in the primary analysis had lower baseline CD4+ T cells than the 73 subjects included in the primary analysis (p=0.044) and higher baseline plasma HIV RNA concentration (p=0.016), and these trends were similar when assessed separately for the AI and RI strata.
Baseline Characteristics for the 73 subjects in the primary efficacy analysis
Treatment toxicity was as expected for the drug regimen employed and did not differ between study strata. Abacavir hypersensitivity was seen in nine subjects, 6/73 (8%, 2 AI, 4 RI) of those in the primary analysis and 3/48 (6%, 0 AI, 3 RI) of the remaining subjects. No myocardial infarctions were reported.
During the first 52 weeks after initiating treatment, clinical symptoms associated with drug toxicity of grade two or three were reported in 71 of the overall 121 enrolled subjects, 26/50 (52%) in the AI stratum and 45/71 (63%) in the RI stratum. There were no grade 4 signs or symptoms reported during the first 52 weeks and only one grade 4 chest pain in the AI stratum in the subsequent weeks. Laboratory toxicities of grade 2 and above were reported in 44 of the 121 enrolled subjects in the first 52 weeks (18/50 AI, 26/71 RI). Lipoatrophy was not systematically defined in this trial but no cases were reported as reason for treatment discontinuation. The safety of treatment interruption was examined after the SMART trial results were presented and there was no evidence detected of accelerated disease progression or clinical events.
The disposition of subjects registered into ACTG 371 are shown in . Study treatment was discontinued before week 52 in 24 subjects (11 AI, 13 RI). In 7 (4 AI, 3 RI), toxicity was the stated cause, 8 subjects (4 AI, 4 RI) discontinued treatment for other reasons including intolerance and pill burden, in 3 (1 AI, 2 RI) treatment was stopped for virologic failure and 6 (2 AI, 4 RI) subjects were lost to follow up.
Twenty-four subjects who were on study for at least 52 weeks did not undergo treatment interruption. Of these, 8 subjects (2 AI, 6 RI) completed an early study version, 3 (1 AI, 2 RI) had toxicities, 6 (2 AI, 4 RI) did not interrupt due to patient preference, 4 (3 AI, 1 RI) had virologic failure, and 3 (3 AI) were lost to follow up.
The primary efficacy endpoint, plasma HIV RNA concentration <5000 copies/mL after 24 weeks of treatment interruption, was achieved in 40% of the 73 subjects who underwent treatment interruption and 24% of the 121 subjects enrolled overall. Virologic success was achieved in 12/28 (43%: 95% CI: 24% – 63%) in the AI stratum and in 17/45 (38%, 95% CI: 24% – 53%) in the RI stratum who underwent treatment interruption. Most (25) primary endpoint successes were achieved in the first treatment interruption. There was no statistically significant difference between the two groups (p=0.81; 95% CI on the difference between groups: −31% to 19%). Considering the first treatment interruption, 34 (47%) subjects re-initiated treatment prior to week 24. We also found no differences between the AI and RI strata in having average plasma HIV RNA concentration <5,000 or <10,000 copies/mL between weeks 18–30 of treatment interruption (AI: 13/28, 46%, RI:18/45, 40% p=0.63; AI: 16/28, 57%, RI: 21/45, 47% p=0.47 respectively), or plasma HIV RNA concentration <10,000 copies/mL at week 24 of treatment interruption (AI: 14/28, 50%, RI: 22/45, 49% p=1.0).
Secondary efficacy analysis for the first treatment interruption also revealed no statistically significant difference between the two strata in terms of:
- Time to three consecutive plasma HIV RNA concentrations >=5,000 copies/ml or two >= 50,000 copies/ml (median, weeks AI: 10, RI: 14, p=0.76);
- Peak (maximum) plasma HIV RNA concentration (median, 28,224 copies/ml AI, 36,241 copies/ml RI; p=0.67);
- Time to peak plasma HIV RNA concentration (median weeks AI: 8, RI: 8, p=0.35).
- Rate of initial increase in plasma HIV RNA concentration (median, log10/week AI: 0.29, RI: 0.50 p=0.46).
- Changes in CD4+ T cells between those immediately preceding treatment interruption (median, 893 cells/mm3 AI, 829 cells/mm3 RI) and the average of all measurements between weeks 18–30 of continuous treatment interruption were not significantly different (median changes, −165 cells/mm3 AI, −90 cells/mm3 RI; p=0.25).
Immune activation markers were elevated at baseline, especially in the AI stratum. These markers declined with treatment in both strata in the study population overall () and similarly in the 73 subjects analyzed for primary efficacy endpoint (data not shown).
Percentage of activated (CD38+/HLA-DR+) CD4+ and CD8+ T cells during the induction phase.
The most important predictor of success in meeting the primary efficacy endpoint was baseline plasma HIV RNA concentration (p=0.003 for log 10 RNA) as summarized in . Successful outcomes were more common in the group with baseline plasma HIV RNA concentration < 100,000 copies/mL (22/46, 48%) than in those with baseline plasma HIV RNA concentrations above 100,000 copies/mL (7/27, 26%)). A higher percentage of activated CD4+ T cells at baseline (HLA-DR+/CD38+, overall median: 11%, Q1–Q3: 7%–16%) was also associated with primary endpoint success in a model that controlled for baseline plasma HIV RNA concentration and AI/RI strata (odds ratio = 3.3 comparing CD4+ activation above versus below the median, 95% CI: 1.03–10.7, p=0.036); without controlling for baseline plasma HIV RNA concentration, baseline activated CD4+ T cell percentage showed no association (p>0.2). Baseline percentages of activated CD8+ T cells, naïve and memory CD4+ and CD8+ T cells, and absolute CD4+ and CD8+ T cell counts were not associated with primary endpoint success in models adjusted for baseline plasma HIV RNA concentration.
Primary Endpoint Success: Effect of Baseline Plasma HIV RNA Concentration