Infectious Disease Postdoc/Clinician
Department of Pediatrics, University Hospital, Syracuse NY
Send response to journal:
Mr Tyler has a very important message, for both sides of the discussion here: that focussing in on one endpoint is a Bad Idea!
He his quite right that the followup of infants who were infected with HIV despite AZT prophylaxis appears to be worse than that of infected infants who didn't recieve prophylaxis. I am aware of the publications he mentions and am not aware of serious flaws in their observations.
However, the point is that it only looks as INFECTED infants. AZT prophylaxis, of course, prevents HIV transmission. One very obvious explanation therefore is that perhaps only those strains of HIV which were more pathogenic were transmitted. Another, equally valid, is that maybe in those infants who were just unlucky enough to get infected despite AZT, the drug did indeed have an adverse effect, either directly or through faster HIV progression. At this time I do not know of a study capable of disentangling the story.
We must look at the whole picture, and consider the fact that a considerable proportion of the entire cohort were prevented from acquiring HIV at the time of birth. In PACTG076 HIV transmission was lowered by around 66% (8.3% versus 25.5%). Therefore, AZT would have to increase mortality in the ENTIRE group by 300% in order to be considered worse than placebo, assuming that HIV is going to kill the infected infant. The fact that the disease appeared in the infected infants argues it's an effect of HIV (either a more active strain or a toxic effect of AZT preventing proper immune-mediated virus control in the infant). I'm sure that Mr Tyler would agree with me wholeheatedly, albeit for perhaps different reasons, that preventing HIV infection and therefore future antiviral administration is a worthwhile goal. The fact that those infants who do get infected have a worse prognosis highlights the inadequecy of current "Gold Standards" of care, but the evidence so far suggests that is the best we have. I hate to say it, but better interventions are obviously needed.
The French data is, last I heard, open to dispute since the followup of the actual children from PACTG076 doesn't yet suggest any kind of neurological or oncological problems from AZT. Maybe it's a difference in administration in France?
I am extremely grateful, and someone discouraged, for Mr Tyler referring to me the very recent "real-life" follow up of the HIVNET 012 protocol. To tell the truth, having heard the liver toxicity story I had not been following it recently. The discrepancy between the initial trial data (which showed a 50% increase in efficacy above and beyond AZT) and the more recent results (which showed much lower effects) is worth noting. In the HIVNET012 trial, (with breastfeeding in over 95% of mothers, a difference between HIVNET012 and PACTG076) AZT had a 14-16 week transmission rate of 25.1% versus 13.1 for NVP. I agree that 21.7% versus 18.1% isn't so impressive a difference, and even with the pricing and administration advantage (oral versus IV), one wonders if the overall benefit is worthwhile if liver toxicity is included. This story also highlights, from the opposite angle, how different centers can have quite different outcomes from the same intervention.
On a sobering note, one hopes that the rosy results from the initial trials were not biased in any way...
As regards teratogenicity, AZT has not been shown to induce fetal malformations or future cancers in the same way as some animal models have suggested (rodents at a near-lethal dose). One 6 year study failed to show tumourogenesis in humans exposed in utero . There is no evidence of human teratogenicity by AZT. One study reported in 2000 showed a rate of 2.0 defects per 100 for first-trimester antiretroviral exposure, a 2.7/100 rate for 2nd or 3rd trimester exposure, and a 2.2/100 rate in the general population (the study size was 916 infected mothers). 
Vitamin A on the other hand, if taken in excess (obviously!) can double the risk of birth defects. Certain vitamin A derivatives, such as Roaccutane and other anti-acne medications come with strict warnings, and advise two pregnancy tests and two forms of contraception during and for up to a month after administration. A psoriasis medication called Soriatane advises the same for three years after finishing the drug (note that topical retinoid preparations are much less likely to affect fetuses compared to systemic drugs such as these). This information is freely available from most nutritional websites.
I heartily agree that the current situation is far from ideal, but it is nowhere near as bad as Mr Tyler (and others) make out. And yes, judging from the data AZT is less teratogenic than vitamin A. I stand by my statements. I will admit though, that there is perhaps an argument for low-dose vitamin A during pregnancy in true instances of vitamin A deficiency. Looking through the data, I can't see a strong argument against this suggestion and it might well be a good trial to run.
As regards the specific questions asked I reply:
AZT is being used because it's mitochondrial toxicity (indisputable) and oxidative ability (extremely disputable) are outweighed by the benefits of preventing transmission of HIV, and future death due to infection. The GSH issue is not even considered, perhaps wrongly, but until someone performs an interventional trial we'll never know.
I hope that clarifies my position.
1. Hart CE, Lennox JL, Pratt-Palmore M, et al. Correlation of HIV type 1 RNA levels in blood and the female genital tract. J Infect Dis 1999; 179:871-82).
2. Garcia. Seventh Conference on Retroviruses and Opportunistic Infections.
Competing interests: None declared