Background:  The SMART study demonstrated that CD4-guided intermittent ART was inferior to continuous ART for major clinical outcomes. On January 11, 2006, based on this finding, ART-experienced patients (95.6% of all drug conservation (DC) patients) in the intermittent ART arm were strongly encouraged to restart ART. Follow-up has continued for all patients to study closure on July 11, 2007 to assess whether the risk associated with intermittent ART could be reversed.

Methods: Between January 2002 and January 2006, we randomized 5472 patients with CD4>350 cells/mm³ to intermittent, CD4-guided ART (stop ART >350 and start <250 cells/mm³) (DC, n = 2720) or continuous ART (viral suppression [VS], n = 2752). Hazard ratios (HR) for major clinical outcomes were estimated using Cox models for 2 time periods: from randomization to January 11, 2006 (pre-January 2006), and from January 11, 2006 to July 11, 2007 (post-January 2006).

Results:  Pre-January 2006, DC patients spent 34% of follow-up time on ART compared to 94% in the VS group. Post-January 2006, DC patients spent 71% of follow-up time on ART compared to 91% for VS patients. At study closure, 83% of DC and 95% of VS patients were on ART. Percentage of follow-up time spent with CD4 <350 cells/mm³ decreased from 31% pre-January 2006 to 23% post-January 2006 for DC patients, and were 8% and 7%, respectively, for VS patients. As noted in table, pre-January 2006, rates for opportunistic disease or death, death, and a composite outcome of serious cardiovascular disease (CVD), renal and hepatic events were significantly greater for DC compared to VS. Post-January 2006, rates for all 3 outcomes declined for DC, while rates for VS patients remained stable. Patients who experienced a non-fatal opportunistic disease, CVD, renal or liver event pre-January 2006 (113 DC and 50 VS) were at a 5.8-fold (95%CI 3.2 to 10.8) increased risk of death post-January 2006 (p <0.0001). HR (DC/VS) for each time period and p values comparing HR are in table.

Conclusions:  Following the recommendation to reinitiate ART for patients in the DC group, risk of opportunistic disease or death was significantly reduced. However, less than full reversal of risk for DC compared to VS patients for opportunistic disease or death was noted and non-significant reductions for other major outcomes. This may be attributed, in part, to some patients not initiating ART, lower CD4 counts for DC patients post-January 2006, and long-term sequellae of morbidity pre-January 2006. These findings reinforce our recommendation not to interrupt ART using the CD4-guided strategy evaluated in SMART and may have implications for other ART interruption strategies.