Item #35: Was ddI Approved without a Clinical Trial?

AZT was approved in record time, but that record didn’t stand for long. In 1991, the FDA approved another DNA chain terminator, ddI, without even the pretense of a controlled study.

Farber states that the FDA approved ddI without even the pretense of a clinical trial in 1991. She creates the impression that ddI is an untested medicine.

First, ddI was, of course, tested in a clinical trial prior to its approval by the FDA. The results of the trial were published after the drug was approved. 59 This is neither unusual nor sinister, and the results of the trial were available to the FDA during the approval process. She also fails to point out that ddI has been tested in a number of clinical trials. A Cochrane meta-analysis found that ddI added to AZT regimens reduced death and morbidity. 60

John Lauritsen describes the 1991 approval of ddI, the second AIDS drug, in his book “The AIDS War” [1].

By far the best of the four media reports [on the FDA meeting at which ddI/didanosine was approved] was Marilyn Chase’s [22 July, 1991] Wall Street Journal article, “DDI Decision Heralds a New FDA Activism”. She at least reported the views of two members who opposed ddI approval:

“This is a rush to judgment”, said Deborah Cotton of Harvard University Medical School. And statistician Paul Meier of the University of Chicago abstained from the vote because he was “greatly troubled” by what he called an abandonment of scientific standards.

From Chase’s article and from information that Mark Harrington supplied at [an] Act Up meeting, it is possible to piece together what happened…The manufacturer of ddI, Squibb, presented data allegedly demonstrating the benefits of ddI therapy. Having no data from a controlled study, they fell back upon comparisons to “historical controls”, a statistically unacceptable procedure. Among the “historical controls” was the placebo arm of the fraudulent Phase II AZT trials which were conducted in 1986. In addition…Squibb presented data based on the discredited p24 antigen test. This was a mistake, for an FDA woman then got up to say that results from the p24 antigen test are meaningless in terms of clinical outcomes or survival.

At this point ddI would not have been approved, but the FDA came to Squibb’s rescue. They requested permission from the National Institute for Allergies and Infectious Diseases (NIAID) to have “a peek” at data from an uncompleted study being conducted by James Kahn of the University of California at San Francisco. The committee looked at a slide which seemed to show that ddI was just as good as AZT. CD4 cells went up and down. The head of the FDA, David Kessler, told the committee members to “be creative”. They took the hint and voted to approve ddI but only conditionally – for adults and children who had failed or were intolerant of AZT.

The Gallo document also cites a 2006 Cochrane review [2]. “Six trials were included in the meta-analysis. During a median follow-up of 29 months, 2904 individuals progressed [to AIDS], of whom 1850 died. The addition of ddI to AZT delayed both progression and death…After 3 years the estimated percentages alive and without a new AIDS event were 53% for AZT+ddI…and 44% for AZT alone; the percentages alive were 68%…and 59% respectively. Five of the six trials involved randomised comparisons of AZT+ddI versus AZT+ddC: in these, the AZT+ddI regimen had greater effects on disease progression and death.” Thus, about 15 years after approval, there is only one trial (at least of adequate quality to be included by this comprehensive review) of ddI vs. AZT, the remaining trials just compare AZT with ddI versus AZT plus a similar drug, ddC. Not one trial was quoted that compared ddI against placebo.

The Gallo document does not mention just what the clinical trial of didanosine (ddI) [3] showed, perhaps because the answer is “not very much”. The abstract, quite deviously, stated “In the two didanosine groups, there were improvements in the number of CD4 cells [the immune cells whose depletion is supposed to be a 'hallmark' of AIDS]…as compared with the zidovudine [AZT] group.” This is dishonest, because buried in the paper it states that “In the subjects receiving didanosine, CD4 cell counts increased during the initial 8 to 12 weeks of therapy and then began to decline. In the subjects treated with zidovudine, CD4 counts tended to decline throughout the study period.” In fact, Table 4 in this paper shows that although CD4 cell counts were 5% elevated in the 750 mg didanosine group at week 2, they had declined to 10% lower by week 24. In the 500 mg didanosine group, counts rose 2% at weeks 2 and 8, but had also dropped 10% below baseline by 24 weeks. In the AZT group, counts dropped continuously, ending up 23% lower by week 24]. Without a placebo the best conclusion that can be reached is that didanosine at a lower dose was a little less toxic than AZT, there is no information that it resulted in any clinical improvements or even clinical stabilization of patients in the short term, let alone when used, as AIDS drugs are supposed to be used, for lifelong therapy.

The safety data was not very encouraging on this very short term trial: “The median times to a first modification of the dose, including a reduction in or discontinuation of the study drug, were 33, 39 and 27 weeks for subjects assigned to receive 750 mg of didanosine, 500 mg of didanosine, and zidovudine, respectively. Severe anemia was uncommon in all groups and was less common in the group receiving 750 mg of didanosine than in the zidovudine group. Recipients of the 750-mg didanosine dose had less severe leukiopenia and granulocytopenia than recipients of zidovudine. Treatment with the 500-mg dose of didanosine was associated with less severe granulocytopenia than treatment with zidovudine. The rate of pancreatitis was higher in the 750-mg didanosine group than in either the zidovudine group or the 500-mg didanosine gropu. The number of casees of pancreatitis [pancreas failure] was higher in the 500-mg didanosine group than in the zidovudine group, but the difference was not statistically significant. Fatal pancreatitis developed in two subjects receiving 750 mg of didanosine. The 750-mg and 500-mg didanosine groups both had greater increases in unfractionated serum amylase concentrations than the zidovudine recipients. The rate at which peripheral neuropathy of grade 2 or worse developed did not differ significantly between the three treatment regimens.”

Gallo is correct that there was a clinical trial. But the important trial, but the important question is whether this trial proved, beyond a reasonable doubt, that didanosine is safe and effective. Indeed, the important question of whether it was better than doing nothing was not answered by this trial. And, to our knowledge, backed up by the list of trials cited in reference [2], there has never been a placebo-controlled trial of this drug.

1. Lauritsen J. The AIDS war. Asklepios. 1993.
2. Darbyshire J et al. Zidovudine (AZT) versus AZT plus didanosine (ddI) versus AZT plus zalcitabine (ddC) in HIV infected adults. The Cochrane Database of Systematic Reviews. 2006.
3. Kahn JO et al. A controlled trial comparing continued zidovudine with didanosine in human immunodeficiency virus infection. The NIAID AIDS Clinical Trials Group. N Engl J Med. 1992 Aug 27; 327(9): 581-7.