By Matt Irwin

23 Feb. 2002

In November 2001 JAMA (the Journal of the American Medical Association) published two major studies on "antiretroviral" drugs, one of which showed increased mortality for people taking protease inhibitors. The other study did not look at mortality, but rather focused on "viral load".

Protease inhibitors were introduced on the market in December 1995 with much fanfare, but only two randomized studies were ever done that claimed reduced infections or death from use of protease inhibitors. These studies (Cameron et al and Hammer et al.) were short-term and showed significant toxic side effects. These two studies did not use a true placebo, but instead placed one group on two drugs plus a protease inhibitor and the other group just on the two other drugs. Since the two other drugs had not been shown to reduce morbidity or mortality, this by itself eliminates them from meaningful discourse. One of them (Cameron et al) showed no difference in mortality, which they conceal by lumping mortality stats in with a variety of others. The other (Hammer et al) relied on statistical manipulations which have been heavily criticized. (see for a discussion of the Hammer study as well as some other problems with protease inhibitors)

The two new studies were not randomized, but were much longer term, with follow up for 3 to 4 years, and had much larger sample sizes (3226 and 1219 people, respectively) than the earlier randomized studies. Unfortunately (for the makers of the drugs) the new studies actually found that people on protease inhibitors did much worse than people only started on other medications. They did not compare protease inhibitors to placebo, but rather to regimens of drugs that do not contain protease inhibitors. As usual, efforts were made to minimize or eliminate the impact of their negative findings, and the negative results were explained away with unsupported arguments.

The increased mortality in people using protease inhibitors was found in the second study to be reviewed, but the first study also found many inconsistencies with conventional beliefs about HIV and AIDS. For example, the first study to be reviewed found that 40% of people on any drug regimen had to change drugs, mostly due to toxic side effects. Following is a brief review of both studies, but I of course recommend reading the entire article.

Study #1:

Phillips et al (November 28, 2001). HIV viral load response to antiretroviral therapy according to baseline CD4 cell count and viral load. JAMA 286(20); 2560-2567.

This study followed 3226 people after initiating "antiretroviral" drugs for the first time. They were followed for up to 192 weeks (over 3 1/2 years) with an average follow up time of 119 weeks. The major objective of the study was to see if people with high "viral loads" had more difficulty in attaining a viral load level below 500 "copies" per mL of blood than people with low "viral loads". They also wanted to make a similar comparison between people who started the study with low CD4 counts and people who had high CD4 counts. They found no difference in viral load level after 32 weeks, no matter what the person's initial viral load or CD4 count was. Their only finding was that people with high viral loads took a few weeks longer to reach 500 "copies" per mL, on average, which would be expected. These findings cast doubt on the idea that people should have "antiretroviral" drugs started based on CD4 and "viral load" measurements. People with CD4 counts less than 200 had slightly lower rate of viral load suppression initially, but this disappeared after adjusting for other other variables. They also found no difference in "viral load rebound" based on initial viral load or initial CD4 count. Here is a quote from the study:

"Considering the baseline viral load, we found no evidence that those with viral loads between 10,000 and 100,000 had a worse response than people with lower levels. Those with baseline viral loads greater than 100,000 at baseline tended to experience a slower rate of initially achieving a viral load less than 500 copies/mL, ... (but) there was a similar OR of achieving less than 500 copies/mL at 32 weeks and a similar rate of viral load rebound above 500 copies/mL between groups with greater than 100,000 and groups with less than 10,000 copies/mL." ...

"In conclusion, in our study there was no strong evidence that lower CD4 cell counts and higher viral loads at baseline are associated with poorer virological (i.e. "viral load") outcome of antiretroviral therapy." (page 2565)

They also mention the high rate of medication discontinuation and toxicity in the first 40 weeks of drug treatment:

"Around 40% of the patients in our analysis experienced some change in their antiretroviral therapy during the first 40 weeks... It previously has been shown that most early changes are due to toxicity." (page 2565)

Study #2

Hogg et al. (November 28, 2001). Rates of disease progression by baseline CD4 cell count and viral load after initiating triple-drug therapy. JAMA 286(20); 2568-2577.

This study followed 1219 people who were started on "antiretroviral" drugs for the first time. They were followed for 3 years, and mortality was compared among people who started with high and low CD4 counts and high and low "viral loads". They found that the person's initial viral load had no effect on mortality after adjusting for other variables. After "adjustment" the only variable that remained significant was that people with CD4 counts less than 200 had increased mortality rates.

The mortality rate for all the people in the study was quite low, even for people with very low CD4 counts. For example, 74% of people who started the study with CD4 counts less than 25 were still alive 3 years later. I have previously written a paper showing that low CD4 counts are associated with a wide variety of physical and psychosocial stressors (see, so it is reasonable that very low CD4 counts are a loose and non-specific marker for health status, and that people with very low CD4 counts would be at increased risk for dying or being ill, whether or not they have been diagnosed "HIV-positive".

Protease Inhibitor Use Associated With Increased Mortality

The remarkable finding from this study was that people initiated on "antiretroviral" therapy with a protease inhibitor had significantly increased risk of death prior to statistical adjustments (RR 2.02). This risk was not as high after adjusting for other variables, but still showed a trend towards increased risk of death (RR 1.44). Similarly, a person who carried a diagnosis of "AIDS" prior to drug initiation had an increased risk of death before adjustment (RR 2.57), but after adjustment this risk was completely eliminated (RR 1.04). To find these numbers in the article one must look at Table 3 on page 2576, as they are not mentioned in the text or in the abstract.

What do we make of these confusing findings? If one believes the hype about protease inhibitors, they should confer a survival advantage, or at least a trend towards increased survival. The authors explain this result away by saying simply "Subjects who initiated therapy with a protease inhibitor were more likely to die in the univariate analysis, but use of protease inhibitor was based on worst prognosis." (page 2575) They do not offer any evidence supporting this statement except to say that "This is confirmed by the fact that protease inhibitor use was not significant after adjustment." (page 2575) They fail to explain, however, why their "adjustments" still left a trend with 44% increased risk of death (RR 1.44) even after adjusting for variables such as CD4 count, viral load, and diagnosis of AIDS. Although this 44% increase was not statistically significant, it still represents a very unexpected finding if one believes the hype about protease inhibitors.

The study opens with the usual claims that "Triple drug combination therapy has been shown to dramatically decrease morbidity and mortality" (page 2568), and cites the two randomized studies mentioned at the opening of this message as evidence. They also admit that these triple drug regimens were initiated "starting in 1996", but fail to note that mortality started falling in 1995, before the new drugs were introduced, and that new AIDS cases had been falling sharply since 1993, which is a much more logical explanation for the reduced mortality rates.

The early randomized studies by Cameron et al and Hammer et al noted significant toxicities such as severe nausea, vomiting, or diarrhea in well over half of the study participants who took protease inhibitors. These toxicities could explain why people who started on these drugs showed increased mortality when compared with people who were not started on protease inhibitors, but this possibility is not even entertained by the authors of the study.