Philip Alcabes discusses myths of health, disease and risk.

Early Onset of AIDS Therapy

Late last week, the NY Times reported that the city of San Francisco’s Department of Public Health is going to begin advising people with HIV to begin antiretroviral treatment (ART) immediately, rather than waiting for the CD4 count to decline.

The policy seems to be based primarily on a secondary analysis of longitudinal data from a multi-center study of HIV-infected people in the U.S. and Canada, the NA-ACCORD study.  The results were reported in the New England Journal of Medicine a year ago.  In that analysis, people with HIV whose CD4 counts were between 351 and 500 who began ART immediately were compared to those who deferred ART until CD4 count was 350 or less.  The deferred-ART group was found to have a 69% higher risk of mortality (from any cause) than were those who began ART before CD4 count fell to <350.  Similarly, among HIV-infected people with CD4 counts above 500, those who began ART after CD4 count was <500 had a 94% higher risk of mortality compared to those who began ART immediately.

But is this a good basis for across-the-board policy for a city the size of San Francisco?  Some physicians worry about the development of drug resistance among viral strains.  Others are concerned about toxicity, always a problem worth considering.  Paul E. Sax tracks the history of the idea and includes a few quotes in his blog post yesterday.

Some commentators wonder whether the new policy is meant to be a boon to pharmaceutical companies.  That’s not a crazy concern:  the Bay Area Reporter noted a couple of weeks ago that San Francisco plans to shift to a “status awareness” policy, increasing HIV testing by 70,000 people per year in an effort to halve the rate of new infections by 2015.  If successful, the increase in testing combined with an increase in recommendations for early ART would expand the market for antiviral medications substantially.

A few aspects of the April 2009 report on NA-ACCORD raise worries about whether it should be the basis for broad-based policy.  First, people who deferred therapy were observed very briefly (median 1.3 years, many of them for only 6 months), so any advantage to early therapy appears to refer to the period immediately post onset of therapy.  That’s important because toxicity and/or resistance might not be evident right away.  Second, looking only at people with an initial CD4 count above 500 and holding constant self-reported history of drug injection, there was only weak evidence for a slight effect of early treatment on mortality (the relative mortality hazard was 1.28 (95% confidence interval 0.85 to 1.83)).  Drug users had a higher mortality risk, and this finding—on which the authors of the New England Journal paper do not comment—suggests that early ART did not reduce the hazard of death for drug users.

Also, the authors of the NEJM paper could not possibly account for some of the hard-to-regiment aspects of HIV infection.  These would include variations in cause of death, treatment adherence, and monitoring of treatment effects — all of which would either not be evident in a cohort study or could not be controlled in a secondary analysis.

Finally, the authors are slightly cagey about the effect of drug-injection history in the above-500-CD4-count group, reporting a twofold increase in death hazard for those who delay ART after excluding people with a drug-injection history – but never reporting information on the effect of ART delay among drug injectors alone.

Most important, observations on people who transitioned to the next-lower CD4 compartment (i.e., from above 500 to <500, or from 351-500 to <350) were censored after 6 months if the individual had not yet initiated ART.  Therefore, the real comparison the NEJM authors are making is between immediate-onset ART while CD4 count remains in the same CD4 compartment vs. immediate-onset ART after CD4 count has dropped to the next lower compartment.  It’s not really a study of immediate versus delayed onset ART.

There’s plenty of reason (including the 2009 NEJM paper) to think that suppressing HIV early rather than late should be helpful, and some reason to think that the reduction in viral load produced by ART will lower infectivity in a way that makes transmission to uninfected sexual or drug-sharing partners less likely.  That in turn could be of public-health value.

Of course, nobody is being forced to start ART before he or she wants to, no matter the policy recommendation. Still, it’s worth wondering whether the expansion of testing and extension of early treatment will substantially improve the public’s health in a way that makes the cost, and self-evident advantages to pharmaceutical (and test-kit) manufacturers, worthwhile.