Restarting treatment after an interruption reduces the risk of serious events but CD4 recovery falls short of baseline levels

Simon Collins, HIV i-Base

Wafaa El-Sadr from the INSIGHT research network, presented an analysis of event rates from the large international CD4-guided treatment interruption study (SMART) that occurred in the 18 month period of follow-up since enrolment was stopped and patients were recommended to restart treatment.

The study was halted following a recommendation by the DSMB in January 2006 after only two of the planed seven years, due to significant benefits in terms of mortality and serious AIDS and non-AIDS morbidity, in favour of continuous treatment.

As well as providing sufficient power to look at the original study question, this trial dataset (5472 patients were randomised) is providing significant insight into other important aspects of HIV management.

Prior to January 2006, patients randomised to the treatment discontinuation arm (DC=drug conservation) spent 34% of follow-up time on treatment compared to 94% patients in the continuous treatment arm (VS=viral suppression). Post-January 2006 this increased to 71% vs 91% respectively, and when the study closed in July 2007, 83% and 95% of patients in each arm were on treatment,

It is important that after the recommendation to restart treatment, the rate of opportunistic infection or deaths all declined in the interruption arm (from 3.4 to 2.1/100 pt yrs) and stayed constant for patients in the continuous therapy group (1.4/100 throughout). This was a significant change in hazard ratio between the pre- and post January 2006 hazard ratios (p=0.03). Rates reduced in inverse proportion to time since restarting treatment.

Although similar trends were reported for other endpoints (death, OI, major CVD renal or hepatic disease) the p-value for the change in hazard ratio pre-and post January 2006 was not statistically significant.

Although the majority of patients in the DC group re-suppressed viral load, mean CD4 count noticeably failed to reach pre-interruption levels. Patients in either arm who had experienced a non-fatal serious event prior to January 2006 (113 in DC and 50 in VS arms) were at 5.8-fold increased risk of death during the follow-up (95%CI 3.2-10.8), p<0.0001).

These differences were not explained by patients in the DC not following the recommendation to restart treatment; an analysis of a subgroup of patients who all restarted treatment confirmed similar results.

Clinics where >85% patients followed the recommendation to restart treatment reported a drop from 3.8 to 1.1 in the DC arm (p=0.02, for difference in HR pre- and post- January 2006). The persistence of increased risk in the DC arm was largely explained by lower mean CD4 count and higher proportion of patients with uncontrolled viraemia.

When looking at the CD4 response to restarting treatment, the researchers found a significant difference between the two groups, even 18 months after restarting treatment: mean 507 vs 648 cells/mm3 in favour of the continuous treatment arm. Baseline CD4 counts in each group was approximately 600 cells/mm3.

This was not explained by patients within the DC group who chose not to restart treatment as an analysis of a sub group of patients who had all followed the recommendation to restart treatment found similar results.

The investigators concluded that these results further strengthened the earlier recommendation not to use CD4-guided treatment interruptions, as this was associated with long-term impact beyond the period of interruption.


This provides additional validation for the decision to stop the study early and for the recommendation to restart treatment.

Although the study provides some evidence that long-term clinical outcome may become normalised over time once treatment is restarted, the significantly lower CD4 count, even 18 months after treatment was resumed, was not expected.


  1. El-Sadr W et al. Re-initiation of ART in the CD4-guided ART Interruption Group in the SMART Study Lowers Risk of Opportunistic Disease or Death. 15th CROI, 3-6 February 2008, Boston. Abstract 36.

This oral presentation is available to view online from the conference website

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