Major Study Shows AZT Monotherapy Inferior

A major government study has found that AZT alone was inferior to AZT plus ddC or AZT plus ddI combination therapy, or to ddI alone, in patients starting with a CD4 (T-helper) count between 200 and 500. This finding, first reported in the press on September 14 and later described in more detail in a preliminary analysis at the ICAAC conference in San Francisco, is expected to change the common practice of putting people on AZT alone as their first treatment for HIV infection.

[Note: As this issue went to press, we learned that the results of another major AIDS trial, the Delta study in Europe and Australia, strongly confirmed the ACTG 175 results, showing that AZT plus ddC or AZT plus ddI had a substantial survival benefit over AZT alone. The Delta study did not have a ddI-only treatment arm. The Delta researchers strongly recommended that when people first start anti-HIV therapy, they use one of these combinations instead of AZT alone.]

The U.S. study, called ACTG 175, enrolled over 2,000 volunteers and lasted for two years. It randomly assigned patients to receive one of four treatments: AZT alone, ddI alone, AZT plus ddI, or AZT plus ddC. It looked primarily for three different outcomes: (1) survival; (2) a clinical endpoint consisting of either progression to AIDS or death; or (3) a 50% or greater drop in CD4 count (or progression to AIDS, or death, in case they occurred first).

If somebody did poorly (either because they developed an AIDS-related condition, or because their CD4 count fell by 50% or more) they were switched to a different study regimen which might be more beneficial -- those on single-drug therapy being switched to the combination, and those on one of the combinations being switched to the other one.

In addition, several substudies were "nested" in ACTG 175, to test some of the volunteers more thoroughly in order to answer specific questions. A virology substudy measured viral load from over 300 of the volunteers, at intervals of 8, 20, 56, 80, and 104 weeks; these data have just begun to be analyzed (see "ACTG 175 Viral Load Substudy," below). Other substudies include pharmacology, women's health, and quality of life; these have not yet been analyzed, however.

About half of the volunteers discontinued their medications before the end of ACTG 175. Only seven percent of these discontinuations were mandated by the protocol, due to severe adverse experiences which were considered to be possibly drug related. Others discontinued for different reasons such as milder adverse events, disease progression including CD4 decline, or the large number of pills which needed to be taken (because the study was blinded, everyone had to take pills for all three of the drugs). Those in the ddI-only arm discontinued less often than those in other arms.

Fortunately, most of those who discontinued taking their protocol-specified medications still remained in the study for followup purposes. 19 percent of the volunteers in ACTG 175 were lost to followup by December 1, 1994.

[Statistical note: In the presentation of the clinical results of ACTG 175, distributed as a 49-page executive summary, the data were analyzed by the "intent to treat" method. This means that the patients were averaged based on the medication which they were randomly assigned -- even if they changed that medication later. Intent to treat analysis has certain advantages for statistical interpretation -- and certain disadvantages as well. In this case, an exception to strict intent-to-treat rules was made, in that patients who NEVER took their assigned medication were excluded from the analysis.]

Overall Results of ACTG 175

The most important results were found by separate analysis of the "antiretroviral-naive" and "antiretroviral-experienced" volunteers. The antiretroviral-naive group had never received AZT, ddI, ddC, or other antiretrovirals -- or had taken such treatment for less than a week. The "antiretroviral-experienced" volunteers had been heavily treated with AZT before the study began; 42% had already taken antiretrovirals (almost always AZT) for two years or more, and another 24% had been treated for more than one but less than two years.

Results for Those Starting Antiretrovirals

For those who had not taken antiretrovirals before, the death rate during the trial was 7% for those who started AZT alone -- compared to 4% for ddI alone, 4% for AZT plus ddI, and 3% for AZT plus ddC. (This difference is not statistically significant, however, meaning that it could well have happened by chance alone -- since not many people died.)

For the same patients, the rate of either progressing to AIDS or dying during the trial was 12% for those starting on AZT alone, vs. 9% for ddI alone, 8% for AZT plus ddI, and 6% for AZT plus ddC. (Only the difference between AZT alone and AZT plus ddC is statistically significant here.)

And for these patients, the rate of progressing to either a 50% or greater CD4 count decline or AIDS or death, was 23% for AZT alone, vs. 17% for ddI alone, 14% for AZT plus ddI, and 10% for AZT plus ddC. (Here, all the comparisons between AZT and the other three treatments are statistically significant, as is the comparison between ddI and AZT plus ddC.)

These data clearly suggest that AZT alone is worse than the other three regimens as initial therapy, in the patient group studied in this trial. AZT plus ddC may have performed best of the four regimens, but it is impossible to be sure of that from the data.

Results for Those Previously Treated with AZT

For the antiretroviral experienced group, the comparable results are as follows:

The death rate during the trial was 10% for those taking AZT alone -- compared to 5% for ddI alone, 6% for AZT plus ddI, and 9% for AZT plus ddC. (The differences between AZT vs. ddI, and AZT vs. AZT plus ddI, are statistically significant; the other comparisons are not.)

The rate of either progressing to AIDS or dying during the trial was 18% for those using AZT alone, vs. 14% for ddI alone, 13% for AZT plus ddI, and 17% for AZT plus ddC. (Only the difference between AZT alone and AZT plus ddI is statistically significant.)

And the rate of progressing to either a 50% or greater CD4 count decline or AIDS or death, was 38% for AZT alone, vs. 26% for ddI alone, 22% for AZT plus ddI, and 27% for AZT plus ddC. (Here, all the comparisons between AZT and the other three treatments are statistically significant.)

So AZT alone was also worse than the other three regimens for those who had already taken AZT. AZT plus ddC did relatively poorly for those who had already taken AZT (especially in the clinical endpoints, AIDS and death), although it did not do as badly as AZT alone.

The ACTG 175 Virology Substudy

Viral-load results of ACTG 175 were not included in the executive summary. Instead, a preliminary analysis was quickly put together for the ICAAC late-breaker session.

Viral load was not measured on everybody in ACTG 175, but on a selection of the volunteers; 196 of them were antiretroviral naive, and 152 were experienced. Viral-load samples were taken twice at baseline, and at 8, 20, 56, 80, and 104 weeks. The amount of virus will be measured by several different methods, but only plasma HIV RNA has been analyzed so far, and is described here.

Overall, viral load showed a much smaller decline from the AZT-only treatment than from the other three treatment regimens. By eight weeks, the viral load had already predicted the main conclusion of the two-year study -- that treatment with AZT alone was inferior to the other regimens.

The viral load averages at 80 and especially at 104 weeks were less reliable than at earlier times because of drop-outs from the trial, and also because changes of study medication. (Viral load samples were included in the analysis presented at the late-breaker session only if patients were still on the treatment to which they were assigned; this is different from the "intent to treat" analysis used for the clinical outcomes of ACTG 175, described above). Therefore, the descriptions below apply to the baseline and the first three time points unless otherwise noted.

For the antiviral naive patients, the two combination treatments were most effective in reducing viral load, with AZT plus ddC perhaps being a little better on the whole than AZT plus ddI, although they are close enough that it is not possible to tell for sure. ddI alone was intermediate between AZT alone and the two combination treatments.

For antiretroviral experienced patients, AZT again was clearly the least effective in lowering viral load; in fact, it did not lower viral load at all. The other three treatment were about equivalent to each other in the experienced patients.

Overall, each of the two combinations showed about a 0.7 log (5-fold) decrease in viral load, sustained for the first year of the study. (After the first year, there may have been some drop-off in viral suppression, but it is hard to be sure because of the small number of samples left after patients dropped out or changed medications, as mentioned above.)

The viral-load results we most want to see from ACTG 175 would not be group averages, but a patient-by-patient analysis of how to use viral load, and short-term changes in viral load after starting therapy, to predict the risk of long-term clinical progression if treatment is not changed. This information is not yet available.