Predicting HIV disease progression risk in people starting HAART: Risk calculator available online

In 1997, John Mellors and colleagues from the University of Pittsburgh reported data on the risk of HIV disease progression in untreated HIV-positive gay men who were participants in the Multicenter AIDS Cohort Study (MACS). These provided an estimate of the risk of developing AIDS, or dying, within the next three years according to a range of CD4 and viral load levels, so long as the individual continued not to take anti-HIV treatment. These predictions (a tabulated summary of which appears in NAM's booklet Viral load & CD4 here at aidsmap.com) have since been commonly used in discussions between doctors and people with HIV regarding the risks and benefits of starting or delaying HAART.

Now a new cohort study, published in the July 12th issue of the Lancet, has provided important additional information which will further inform this process. The ART Cohort Collaboration is a unique venture (funded by the UK's Medical Research Council and GlaxoSmithKline) which has brought together data from thirteen cohorts of HIV-positive people using a statistical method called 'meta-analysis'. Those included were Europeans and Americans who began a HAART regimen of at least three drugs, having previously taken no antiretroviral therapy. Data on subsequent response to therapy were analysed using an intent to continue treatment approach. This means that people remained within the study regardless of any switches or interruptions in treatment, which probably most accurately reflects how treatment is taken in 'real-life' settings.

The analysis included 12,574 people, 79% of whom were men. Twenty-one per cent had HIV disease classified as CDC stage 3 (AIDS). Eighty per cent began therapy with a protease inhibitor and two nucleoside analogues. At baseline, the median CD4 count was 250 cells, and the median viral load was 4.9 log. After six months of treatment, the median CD4 count had increased to 343 cells, and 73% of the cohort had viral load below 400 copies.

During 24,310 person/years of follow-up there were 870 AIDS events and 344 deaths (a total of 1,094 events of either AIDS or death). The researchers calculated the risk of disease progression at one, two or three years after beginning treatment according to five key baseline variables: CD4 count, viral load, age, transmission category and CDC stage.

Overall, the following baseline factors predicted a poorer prognosis:

• CD4 count less than 200

• viral load above 5 log (100,000 copies); lower levels were not predictive of response

• age above 50

• being an injecting drug user

• CDC stage 3.

Specific risk estimates can be viewed either in the Lancet or online at the ART Cohort Collaboration website. This excellent facility includes a calculator where a risk estimate is provided based on input patient data.