CROI: CD4-guided treatment interruptions unsafe, SMART study concludes

The SMART trial

The SMART trial (Strategies for Management of Antiretroviral Therapy) is a large international study designed to examine the effects of intermittent treatment in HIV-positive patients. Although it was supposed to run for over nine years, the study was stopped last month after two years, following the discovery of the risks of the interruption strategy.

The study’s investigators set out to recruit 6000 patients, all with CD4 cell counts of 350 cells/mm³ or greater. They were randomised to receive continuous anti-HIV treatment, or to receive intermittent treatment on the basis of their CD4 cell count.

The patients randomised to the treatment interruption group stopped taking antiretroviral therapy at the start of the study. When their CD4 cell counts had fallen to below 250 cells/mm³, treatment was started again. However, once the CD4 cell count rose to above 350 cells/mm³, treatment was stopped, and so on. The investigators’ hypothesis was that this strategy would allow the patients to maintain CD4 cell counts at a safe level while minimising the drugs' side-effects.

The study’s data safety and monitoring board (DSMB) decided to stop the trial on 10 January, after they discovered the increased risk of HIV disease progression, and recruitment to the trial was stopped on the following day.

Treatment interruptions guided by CD4 cell counts put HIV-positive patients at an increased risk of disease progression, AIDS and death, according to the results of a large international study presented today at the Thirteenth Conference on Retroviruses and Opportunistic Infections in Denver.

Surprisingly, the study found that patients on intermittent treatment were more likely to experience serious complications usually associated with HIV treatment. It also failed to detect a link between the risk of disease progression and the patients’ lowest-ever CD4 cell count.

Although preliminary, initial analysis of the study’s results has suggested that the increased progression risk in the patients undergoing treatment interruption may be explained in part by their lower CD4 cell counts over the course of the study.

In search of an explanation

Although the data are only weeks old, researchers involved in the SMART trial have been eager to examine the potential causes of the dramatic increase in disease progression in the treatment interruption arm, particularly the raised incidence of ‘treatment-associated’ complications.

Wafaa El-Sadr, presenting the data in Denver, showed an analysis of the time spent by the two groups at low CD4 cell counts. Unsurprisingly, the patients in the treatment interruption arm spent longer at CD4 cell counts below 350 cells/mm³ (32 vs. 7%). They also spent longer below 250 cells/mm³ (8 vs. 2%).

Since these proportions are low, it is unlikely that the study investigators were failing to monitor their patients closely enough and exposing them to long periods at suppressed CD4 cell counts. However, they do reveal that the patients undergoing treatment interruption were living at lower CD4 cell counts across the study. Dr El-Sadr stated that the increased disease progression could be linked to the ‘proximal’ CD4 cell count, the CD4 cell count measured at the time of the disease progression event.

She also suggested that increased levels of inflammation, possibly due to the increased amount of time spent at lower CD4 cell counts or higher HIV levels in the treatment interruption group, could be responsible for the increased incidence of severe complications such as cardiovascular disease in the treatment interruption group.

Results

By the end of recruitment in January, SMART had enrolled 5472 patients from 318 research sites in 33 countries. The patients had been followed up for a median of 14 months, during which there had been 164 recorded instances of disease progression, defined as death, or the development of a serious AIDS-related condition or a serious complication.

Overall, the patients in the interruption group had spent 33% of the time on treatment, compared with 93% in the continuous treatment arm. The patients in the intermittent therapy arm had had a median of three treatment interruptions, with a median length of 18 months.

By the 10 January meeting, there was a significantly elevated incidence of disease progression in the patients undergoing intermittent HIV treatment, with 117 cases (3.7%), compared to 47 (1.5%) in the continuous treatment arm. Although these absolute rates are low, the comparison of the two arms reveals a 2.5-fold elevated risk in the patients undergoing treatment interruptions (p

The DSMB concluded that this elevated risk was so great that collecting further data would be very unlikely to alter the conclusion that treatment interruption was inferior to continuous therapy. Coupled with the risk posed to patients by continuing the study, this led them to recommend stopping enrolment and switching all patients to continuous treatment.

When the investigators broke down the disease progression, they found that the patients undergoing treatment interruptions were more likely to have died (1.7 vs. 0.9%, p = 0.01), or to have experienced serious progression of disease (0.6 vs. 0.1%, p = 0.004).

Only a small proportion of deaths were due to AIDS-defining conditions, with most deaths being due to cardiovascular complications, violence and accidents. The majority of disease progression events were due to thrush in the gullet (oesophageal candidiasis).

Surprisingly, however, the investigators found that the patients undergoing intermittent treatment were at a significantly elevated risk of serious complications (2.1 vs. 1.4%, p = 0.04). These included heart attack, stroke, coronary artery disease requiring surgery and kidney or liver disease. The investigators had expected these complications to be more common in the patients taking continuous anti-HIV treatment.

The overall risk of disease progression was unaffected by gender, race or CD4 cell count at the start of the study. There was also no influence of a prior AIDS diagnosis, which was present in 24% of the study sample.

It was also unaffected by the lowest-ever or ‘nadir’ CD4 cell count. This was unexpected, since patients with lower nadir CD4 cell counts have been shown to be at risk of disease progression in previous studies of treatment interruptions.

One factor that was linked to an increased risk of disease progression was viral load at the start of the study. Among the patients with a viral load below 400 copies/ml at enrolment, those taking CD4-guided treatment interruptions had a risk of progression 3.8 times greater than those on continuous therapy. In contrast, there was no difference in risk among the patients entering the trial with viral loads above 400 copies/ml.

Implications for the future?

Far from being a ‘failure’, the SMART study has provided an answer to its primary goal, demonstrating that CD4-guided treatment interruptions were inferior to continuous treatment within the study.

It is, however, unlikely that these results will spell the end of treatment interruption studies entirely. SMART was a large and important study. However, 95% of its participants were treatment-experienced, with a median of six years’ antiretroviral treatment in both groups. Although the SMART results provide strong evidence that CD4-guided treatment interruptions are dangerous in treatment-experienced patients, it would be imprudent to apply this conclusion to other groups of patients, such as those starting HIV treatment for the first time.

The SMART study included a number of planned sub-studies, designed to investigate other aspects of HIV treatment and treatment interruptions. One of these is an analysis of the quality of life as perceived by patients enrolled in the study. The results of this analysis are eagerly awaited as they will shed light on the benefits of CD4-guided treatment interruptions as perceived by the patients taking part in the study. This sub-study will also have longer follow-up than the main study, as it included fewer patients, and recruitment had already been completed.

The study investigators also plan to analyse further data and samples taken from patients in the SMART study, such as stored blood samples to analyse changes in blood fats and sugars, and measurements of body composition. In addition, although recruitment has stopped and patients in the treatment interruption group are being urged to switch to continuous therapy, the investigators plan to follow the SMART study participants, to monitor the long-term effects of treatment interruption.

However, the final verdict on the results of this study will emerge once the full dataset from the study has been analysed, reviewed and published. Once this has happened, and more time has passed, the results of this trial can be compared with findings from similar studies before changes to treatment recommendations and guidelines can be considered.