CD4-guided interruption

The idea behind CD4 cell count-guided treatment interruption is to take antiretroviral therapy until the CD4 cell count has risen to a predefined threshold. Treatment is then stopped and is only restarted after the CD4 cell count has fallen to a second, usually lower, cut-off level (although the starting and stopping thresholds may be the same).

Several large studies have shown the risk of disease progression and death rises dramatically below 200 cells/mm³. In the European CASCADE cohort, the six-month probability of developing an AIDS-defining illness with a CD4 count below 200 cells/mm³ ranged from about 5% to 20% (depending on viral load), compared with 0.5% to 5% with 200 to 349 cells/mm^{3_. 1}

However, the ART Cohort Collaboration found that the risk of progression to AIDS or death was similar in people who began treatment with a CD4 count of 200 to 349 cells/mm³ or above 350 cells/mm^{3_. 2} On the other hand, recent studies have suggested that starting treatment with a CD4 cell count above 350 cells/mm³ – or even above 500 cells/mm³ – is associated with a lower risk of illness and death.

The Italian BASTA study provided early evidence supporting CD4-guided treatment interruption. The trial included about 100 participants with a CD4 cell count above 800 cells/mm³ and full viral suppression. Patients were randomised to receive continuous therapy or to take a treatment break as long as their CD4 cell count stayed above 400 cells/mm³. After 18 months, similar proportions saw their CD4 count fall below 400 cells/mm³ (about 30%), but this climbed to 57% amongst people in the treatment interruption arm whose lowest-ever CD4 cell count was less than 350 cells/mm³. No AIDS-defining events occurred in either arm.³

The CD4-guided treatment interruption study that has received the most recent attention is the Strategies for Management of Antiretroviral Therapy trial, or SMART. In this large international study, nearly 5500 treatment-experienced patients with an undetectable viral load and a median CD4 count of about 600 cells/mm³ (but moderately advanced disease in the past) were randomly assigned to a continuous therapy (‘viral suppression’) arm or a treatment interruption (‘drug conservation’) arm that stopped treatment when CD4 cell count was above 350 cells/mm³ and resumed when it fell below 250 cells/mm^{3_. 4}

SMART was halted prematurely in January 2006 after an interim analysis revealed that the risk of opportunistic illness or death was twice as high in the treatment interruption arm (4.4%) compared with the continuous therapy arm (1.7%). This was not entirely surprising, since people who interrupted therapy spent more time off drugs with a lower CD4 cell count.

However, people in the treatment interruption arm also had a small but significantly higher risk of non-AIDS-related cardiovascular, liver and kidney complications (2.4 vs 1.4%). This was unexpected since these conditions are often assumed to be side-effects of antiretroviral therapy, and investigators had hoped that reducing total drug exposure might minimise drug-related toxicities. Based on prior studies, they had expected the rate of cardiovascular disease to be 15% lower in the treatment interruption group.

Amongst people with a viral of 400 copies/ml or less at enrolment, the rate of opportunistic illness or death was four times higher in the interruption arm, but this difference disappeared when viral load was above 400 copies/ml. Stated another way, people who were doing best on antiretrovirals before treatment interruption had comparatively worse outcomes after stopping therapy.

Another unexpected finding was that lowest-ever CD4 cell count did not predict the risk of illness or death, unlike BASTA and other studies. Further, there were few benefits to weigh against the risks of interrupting treatment. Study participants reported that their overall health improved in the continuous therapy arm, but declined in the treatment interruption arm.⁵

The Trivacan trial also demonstrated worse outcomes in the CD4-guided treatment arm, which was stopped early due to an increased risk of disease progression. This Ivory Coast study enrolled more than 800 individuals with viral suppression below 300 cells/mm³ and a CD4 count above 350 cells/mm³ after at least six months on antiretroviral treatment.

Participants were randomly assigned to either continuous therapy, a fixed-length four months on, two months off intermittent treatment schedule or CD4-guided treatment interruption. As in SMART, patients in the latter arm stopped treatment when CD4 count reached 350 cells/mm³ and resumed when it fell below 250 cells/mm³. After 20 months, those in the CD4-guided interruption arm were 2.5 times more likely to develop serious illness (including invasive bacterial infections, severe oral thrush and tuberculosis) than those on continuous therapy, although the risk of death was the same.⁶

Not all studies of CD4-guided treatment interruption studies have yielded poor outcomes. In the Staccato trial, conducted by the HIV Netherlands Australia Thailand Research Collaboration, 430 participants with a median CD4 count of about 500 cells/mm³ and fully suppressed virus were randomly assigned to receive continuous therapy with a boosted protease inhibitor or undergo treatment interruption. In contrast with SMART and Trivacan, drugs were stopped when CD4 count rose above 350 cells/mm³ and restarted when the count fell below this same level. An original week-on, week-off fixed-cycle interruption arm was stopped early due to a high rate of treatment failure.⁷

After a median follow-up period of 22 months, just over 90% in both arms achieved viral suppression below 50 copies/ml, but those in the CD4-guided treatment interruption group reached a significantly lower CD4 cell count (402 vs 619 cells/mm³). Nevertheless, people who interrupted therapy did not have a higher risk of AIDS-defining illness or death (neither of which occurred in either arm).

However, treatment interrupters did have more acute retroviral syndrome, thrombocytopenia and oral or vaginal thrush, an early indicator of impaired immune function. On the other hand, people stopping treatment reduced their total drug exposure by 62% and had less diarrhoea and peripheral neuropathy and lower blood fat levels. Only a small number of people (about 2% in both arms) developed drug resistance mutations. Although this study was much smaller than SMART, the researchers calculated that if the same rates of illness and death in that trial applied, they should have seen 17 such events in Staccato.

Finally, in the Spanish/Italian Tibet study, 201 patients with a CD4 cell count above 500 cells/mm³ and an undetectable viral load were randomised to continuous therapy or CD4-guided treatment interruption stopping when CD4 cell count rose to 500 cells/mm³ and restarting when it fell below 350 cells/mm³. Over two years, most people cycled back onto therapy only once or twice. Similar to Staccato, total drug exposure was reduced by 67%. No one in either arm had an AIDS-defining illness or death.⁸

There were slightly fewer drug side-effects in the treatment interruption arm (2 vs 5% per person year), but this was outweighed by a much higher likelihood of acute retroviral syndrome (56 vs 3%).There was no net reduction in physical symptoms and patients did not report an improved quality of life. More than a third developed drug resistance, but new mutations were only detected amongst people stopping NNRTIs. The researchers concluded that guided treatment interruptions were not as safe as continuing therapy, and did not reduce the overall rate of management-related adverse events.