AZT: Different for People of Color? by John S. James

Data released this week from a U. S. Veterans Administration (VA) study suggested that early treatment with AZT (for persons with T-helper counts of 200 to 500) might not be helpful to Blacks and Latinos, and might even be harmful. (The study did not question later treatment, for anyone with T-helper count under 200.) But three other studies found no racial difference in the effect of AZT. And scientists reviewing the VA study found the data "fragile," and suggested that it may well have resulted just by unlucky chance. There is widespread concern that results which could well be due to errors or statistical happenstance may discourage people from seeking medical care.

The study, called VA Cooperative Study 298, was conducted at veterans' hospitals in Houston, Los Angeles, Miami, New York, San Francisco, and Washington, D. C., and at the Walter Reed Army Medical Center. Volunteers entering the trial had to have T-helper counts of 200 to 500, and symptoms of HIV infection but not AIDS, to be eligible. They were randomly assigned to either an early treatment group, which received AZT immediately, or a later treatment group, which received a placebo at first. Later, when T-helper counts dropped below 200 on two successive visits, the placebo was stopped and all participants in the study received AZT. All AZT doses were 1500 mg per day -- about three times what most physicians use today. The goal of the study was to learn whether starting AZT early would increase survival and delay progression to AIDS. The trial was not designed to look for racial differences.

For ethical reasons, study participants were offered pneumocystis prophylaxis when it was officially recommended. Also, when AZT was officially approved for early use, study volunteers were notified, and some switched from blinded treatment (either AZT or placebo) to AZT, at their request.

Overall Results

A total of 338 volunteers were enrolled in this study; 170 were assigned to receive early treatment (AZT immediately), and 168 assigned to receive placebo at first. The average age of the volunteers was about 40; about two thirds of them were white, one third Black or Latino.

The trial was stopped as planned in January 1991. When the data was analyzed for all volunteers together -- not broken down by race -- it was found that early treatment did clearly delay progression to AIDS; 44 patients in the delayed-treatment group, but only 25 in the early- treatment group, developed AIDS. But early treatment showed no benefit in preventing death; 23 in the early- treatment group died vs. 19 assigned to delayed treatment. (This difference is too small to be statistically significant, meaning that it could easily have occurred by chance.)

Two notable results of the study were that of six cases of dementia, all were in the late-treatment group, suggesting that AZT may have helped in preventing that condition. Also, of six cases of lymphoma, five were in the late-treatment group, suggesting that AZT may also have reduced the risk of lymphoma.

Racial Differences

The researchers were surprized at these inconsistent results, so they looked more closely at the data to see what was happening. They checked to see if results were different for IV drug users compared to other patients, but no difference was found.

When they checked for racial differences, they combined the data for Blacks and Latinos, in order to have enough data in each group to run statistical tests. For the minority groups, they found no statistically significant benefit of AZT in delaying progression to AIDS. But the statistics on death were especially disturbing; nine Black or Latino volunteers in the early treatment group died, but only one who received later treatment. Also, early AZT treatment did not show the same benefit in T-helper count for the people of color as it did for whites.

No one knows why this entirely unexpected result occurred. Researchers who reviewed the study have suggested a number of reasons to be skeptical about the results until more is known:

* No other study has found a racial difference in response to AZT. Three major AZT studies were analyzed to look for such a result, but none was found. How could three studies find no racial difference in response to AZT, while a fourth finds a nine to one difference among minorities with early AZT treatment, vs. a survival advantage among whites? One obvious possibility is that this difference in deaths happened by unlucky chance, and did not reflect any real differences in how races respond to AZT.

* Another concern is that the VA study was analyzed by "intent-to-treat" rules, meaning that volunteers were counted in the treatment groups to which they were randomly assigned -- regardless of anything that might happen later. There are advantages to this kind of analysis, but there are also disadvantages; in the VA study, for example, deaths were counted the same whether they were AIDS related, or due to other causes including murder, suicide, traffic accident, or diseases not believed to be related to HIV. We have not seen any analysis of the study with these unrelated deaths excluded.

A particular problem with intent-to-treat rules in this case is that when AZT was approved for persons with 200 to 500 T-helper cells, study participants had to be given a choice to switch to AZT if they wanted; it would not have been ethical to give a placebo to persons in that T-helper range without their consent. Many of the volunteers assigned to the later-treatment arm did choose to switch; but under the intent-to-treat rules, they had to be counted as late treatment, even if their AZT actually began early. This change did not affect persons assigned to early treatment, who were receiving AZT already.

* This study was not designed to look for racial differences. It is easy to get misleading results when a study is analyzed later in ways not originally planned or intended.

* Because the study was finished in January and presented to other researchers in February, there was no time to complete the analysis, or to thoroughly check the results.

* One theory being considered is that some races might absorb AZT less well than others, when the drug is taken orally. This possibility seems unlikely to account for the VA results, however, since the dose used in that study was three times too high. Unless the differences were enormous, poor absorption would have been a benefit (in reducing side effects), not a detriment.

On February 14 a number of physicians reviewed the VA data. Almost all of them said that it would not affect their practice of medicine, except that it might become one more item to be discussed with the patient when the decision was made as to whether or not to start AZT. No one wanted to change the official FDA "labeling" which suggests that AZT be considered for HIV-positive persons with T-helper counts under 500.

Concerns

The executive director of the National Minority AIDS Council, Paul Kawata, urged caution in interpreting these preliminary findings. "We must not send people of color with HIV infection underground. This study has the potential to take away hope for HIV infected minorities. It is much too early to draw any definitive conclusions."

And Reggie Williams, executive director of the National Task Force on AIDS Prevention, said that "We can not afford to give Black people...any more excuses not to get tested, into early intervention modes and yes, into clinical trials. Nor can we afford to give those in government research and policymaking positions a reason to further marginalize us from our fair share of whatever is out there that may prolong life with HIV."

**************************************