Background:  Interim results from the Children with HIV Early Antiretrovirals (CHER) trial demonstrated the need for early ART in HIV⁺ infants. However, the potential risk of lifelong ART from infancy requires innovative approaches.

Methods:  Infants <12 weeks with CD4% ≥25% were randomized to:  arm 1, deferred ART with lopinavir/ritonavir (LPV/r)+zidovudine (ZDV)+lamivudine (3TC); arm 2, immediate ART with interruption after 40 weeks; or arm 3, immediate ART with interruption after 96 weeks. ART initiation and re-initiation were based on WHO immunologic criteria (CD4% <25%; <20% after infancy) or CDC severe B or C clinical disease. The composite primary endpoint was time to failure of first-line ART (CD4/clinical) or death. All analyses were intent-to-treat using time-to-event methods.

Results:  Between 2005 and 2007, 377 infants (median age 7.4 weeks) were enrolled of whom 341(91%) completed follow-up. Median time to starting ART in arm 1 was 20 weeks and to restarting ART after interruption in arms 2 and 3 was 33 and 70 weeks, respectively. Of 125 patients, 39 (31%); of 126 patients, 31 (25%); and of 126 patients, 25 (20%) reached the primary endpoint. Overall, 23 (18.4%), 11 (8.7%), 11 (8.7%) died. Relative to arm 1, HR (95%CI) for the primary endpoint was 0.73 (0.46 to 1.17, p = 0.39) in arm 2 and 0.58 (0.35 to 0.96, p = 0.03) in arm 3; and for death, 0.40 (0.19 to 0.85, p = 0.02) in arm 2 and 0.45 (0.22 to 0.91, p = 0.03) in arm 3. Only 7 children (3 vs 3 vs 1) switched to second-line ART. At 4.5 years, mean CD4 percentage was 34% vs 32% vs 30%. Considering all CDC severe B and C events and death, there were 66 vs 38 vs 29 events occurring in 52, 33, and 26 children, respectively. The HR to first clinical event relative to arm 1 was 0.53 (0.34 to 0.82, p = 0.004) in arm 2 and 0.42 (0.26 to 0.67, p = 0.0002) in arm 3. Differences in clinical events were mainly failure to thrive (16 vs 18 vs 11) and encephalopathy (9 vs 5 vs 2); 227, 173, and 139 grade 3/4 adverse events occurred in arms 1, 2, and 3, respectively; 34 additional children were randomized to arm 2 and arm 3 from June 2007 to February 2008 with CD4 >25% (totals 143 per arm); 30 (21%) vs 46 (32%) were still off ART at the end of follow-up; 34 of 143 vs 29 of 143 in arm 2 vs arm 3 reached a primary endpoint (HR 0.84, 95%CI 0.51 to 1.38, p = 0.49). There were 11 and 12 deaths in arms 2 and 3, respectively:  9 vs 10 in primary ART, 1 vs 2 during ART interruption, and 1 vs 0 after ART restart.

Conclusions:  In HIV⁺ infants, early limited ART until first or second birthday followed by interruption appears safe, with better outcomes and less ART exposure than deferred ART. Longer time on initial ART permits longer interruption with marginally better outcomes.