SAN FRANCISCO-Long-term estimates of survival utilizing data on 4,500 patients suggest that some people with HIV taking HAART (highly active antiviral therapy) may have a near-normal life expectancy. In a session on HIV at the 39th Interscience Conference on Antimicrobial Agents and Chemotherapy (ICAAC), A.C. Justice, MD, of the Veterans Administration Pittsburgh Healthcare System, presented 2 years of recent data estimating long-term or immediate survival in the post-HAART era.
SAN FRANCISCOLong-term estimates of survival utilizing data on 4,500 patients suggest that some people with HIV taking HAART (highly active antiviral therapy) may have a near-normal life expectancy. In a session on HIV at the 39th Interscience Conference on Antimicrobial Agents and Chemotherapy (ICAAC), A.C. Justice, MD, of the Veterans Administration Pittsburgh Healthcare System, presented 2 years of recent data estimating long-term or immediate survival in the post-HAART era.
HAART therapy has been associated with prolonged short-term survival but has not been available long enough to directly measure long-term survival. Yet long-term survival estimates are needed to help with patient counseling, medical decision making, and clinical research design, Dr. Justice said.
The study, Collaboration in HIV Outcomes Research US (CHORUS), was sponsored by Glaxo Wellcome and supported by an independent advisory board. It included data on 4,524 patients at four private clinic sites: New York, San Francisco, Beverly Hills, Calif, and Nashville.
The median age for patients at baseline was 39 years. Patients were 91% male and 75% white; 87% had reported homosexual or bisexual exposure, 81% had prior exposure to antiviral agents, 62% had been given a protease inhibitor, and 93% were on multidrug regimens by the end of the first year of observation.
Data collection began in August 1997 using real-time electronic medical records. Observations from August 1997 to August 1998 were used to develop predictions for September 1998 to August 1999. Independent data analysis was conducted at the University of Pittsburgh.
Two models were used to analyze data: an exponential model that assumed a single, constant underlying hazard of mortality and a rival model that assumed a changing hazard rate.
To generate predicted death, models were adjusted for CD4 cell counts, viral load, AIDS, and number of AIDS-defining conditions at baseline. Unadjusted models were used to generate the extrapolated overall median survivals. Predictions were then compared to several benchmarks in the published literature.
Results for median survival showed that patients with CD4 cell counts under 200 cells/mm³ have the shortest survival time: a median of 10 years. Patients with AIDS-defining events have a median survival of 15 years, and those with detectable viral loads, 26 years. Overall, the median expected remaining survival was 32 years, with a 95% confidence interval of 27 to 38 years.
In comparison, US life tables report that median survival for all 39-year-old white males in the United States is 36 years. Although the median long-term survival approaches that of all 39-year-old white males, Dr. Justice said, short-term survival and survival among those with lower CD4 cell counts continue to fall substantially short of normal lifeexpectancy.
The population in this sample was largely restricted to white males under aggressive treatment, which represents survival under optimized treatment conditions. However, Dr. Justice said, the CHORUS study results are consistent with other published HAART data for mortality, both for short- and long-term survival estimates.
Dr. Justice noted that many of these patients had switched therapies multiple times at baseline and then continued to switch during treatment. So its not as though these patients were all suppressed to begin with and stayed suppressed throughout the observation, she said.
Survival expectations have dramatically improved with HAART therapy, Dr. Justice said. If hazard for mortality with HIV remains constant, people under treatment may begin to approach normal median life expectancy, she said. However, it is impossible to know whether it will hold for 30 to 40 years. It is likely that over long intervals of time, disease and aging will play an increasingly important role. And viral resistance may play a role.
Analysis of CHORUS data will be updated on an annual basis to develop more precise estimates and to see how well the models are continuing to predict long-term survival.