A New Attempt to Solve One of AIDS's Biggest Riddles

The yardstick for measuring HIV's damage to the immune system is the number of virus-fighting cells, called CD4 lymphocytes, in the bloodstream. The normal number is about 750 per microliter of blood. Years ago, research showed that a person should start ART by the time the CD4 level falls below 200. People who do not start treatment by then die sooner. The big question is, at what point above a count of 200 should therapy begin?

Many guidelines (including the U.S. government's and the World Health Organization's) recommend that treatment start when the count falls to between 350 and 200. Some doctors think it should start even sooner, at a count of 500.

When ART arrived in 1996 and dramatically restored thousands of people with advanced AIDS to good health, many people with early-stage infection also chose to take it. But it was burdensome and often unpleasant. The original "triple therapy" was 14 pills taken at five different times during the day. Some of the drugs raised cholesterol and blood sugar, and seemed to boost the risk of heart attacks and diabetes.

To widespread disappointment, studies in the late 1990s proved that ART could not cure the infection, and that taking the drugs for a long time might raise the risk of developing a drug-resistant virus and running out of treatment options later on.

The pendulum then swung away from early treatment. But now it is swinging back.

The are several reasons. Today it is possible to take a day's worth of triple therapy in a single pill. The newer drugs have fewer side effects, and there are hundreds of possible combinations of the nearly two dozen antiretrovirals on the market.

The pendulum got a big push last January from a study called SMART, which tested the hypothesis that periodically interrupting ART when the CD4 count rebounded might be as good as taking the treatment continuously -- and possibly better if it reduced the side effects.

But to the surprise of many researchers, that theory was wrong. Patients on continuous treatment not only had fewer AIDS-related illnesses, the study found, they also had fewer non-AIDS-related problems, such as heart, liver and kidney disease. It appeared that suppressing the virus in the bloodstream might benefit the entire body, not just the immune system. The study offered one more reason to consider early treatment.

The proposed new when-to-start study would actually be two studies in one. One would enroll patients with CD4 counts over 500 and randomly assign them to either start treatment immediately or to wait until they drift into the 300 to 350 range. The other would enroll people with lower counts and assign them to start immediately or to wait until their counts are 200 to 250.

Africa potentially has the most to gain. Most infected people there are not started on ART until their CD4 counts drop below 200, and even then many cannot get treatment. If it turns out that that threshold is indefensibly low, there would be one more impetus to bring AIDS drugs and care to the world's poor.

An attempt to answer the when-to-start question was made 14 years ago, when a San Francisco researcher, Donald I. Abrams, launched a study called ComPACT-1. Patients were randomized to either start taking one or more AIDS drugs immediately -- there were only three on the market at the time -- or to wait until their doctors thought the time was right.

ComPACT-1 was to be a 200-patient pilot for a larger study, which would have involved 20,000 people. But it failed miserably, shutting down in 1994 after enrolling only 86 people. Almost everyone wanted to wait, figuring that better treatments were just around the corner.

They were right. The arrival of the first protease inhibitor drug in late 1995 gave birth to the three-drug combinations that revolutionized AIDS care.

Even five years ago, a when-to-start study was not possible, because treatment advances were still rolling in.

Today, however, antiretroviral therapy is extremely effective, safe and simple -- and not likely to improve much in the near future. It is once again possible to ask, and perhaps finally to answer, the early-or-late question.