The Body Covers: The 13th Conference on Retroviruses and Opportunistic Infections
Antiretroviral Exposure Not Associated With Liver-Related Deaths in D:A:D Cohort
February 7, 2006
Reference Abstract: Exposure to ART and the risk of liver-related death: is there an association? Results from the D:A:D study (Poster 770)
The initial purpose of the Data Collection on Adverse Events of Anti-HIV Drugs (known as D:A:D) study was to assess the incidence of myocardial infarction among HIV-infected patients receiving antiretroviral therapy. Over time, however, the study has expanded to other areas. It's now one of the largest, prospective, multi-cohort studies of HIV-infected persons under active follow-up.
Eleven cohorts worldwide are participating in D:A:D, with a total current enrollment of more than 35,000 patients from 188 clinics in 21 countries in Europe, the United States and Australia. As of Feb. 1, 2004, more than 75,000 person-years of follow-up were contributed by the patients.
The original study population of 23,441 was enrolled from December 1999 to April 2001, and is referred to as D:A:D Cohort I. An additional 12,900 people were enrolled in D:A:D Cohort II throughout the spring of 2004. The study is projected to continue at least until 2006 and provide more than 130,000 person-years of data.
The data collection for D:A:D takes place at least every eight months. Each cohort gathers and computerizes its data, which subsequently are merged in a database in Copenhagen.
D:A:D collects information about any deaths that occur in the cohort over the previous year. Investigators are required to complete a specific form that collects information about the immediate cause of death, and the presence of hepatitis B (HBV) and hepatitis C (HCV) infection.
To learn more about this cohort, the investigators have put together a Web site (maintained by the Copenhagen HIV program). Here you can view all the forms used in the study as well as the manual of operations, which allows anybody to evaluate the quality of the data collected. This openness is refreshing.
In this study presentation, Rainer Weber et al evaluated if longer exposure to highly active antiretroviral therapy (HAART) was associated with an increased risk of hepatic death. They looked at the deaths in this large cohort -- 1,248 (5.3%) persons died (1.6/100 person-years) with 183 of these deaths (15%) considered to be liver-related. The patients died of liver failure, bleeding and infections, which are what patients with terminal liver disease die of. Among the people who died, 16.9% had active HBV and 66.1% HCV, with 7.1% having both.
More than 97% of the patients who died had been exposed to HAART for a median of 3.3 years. Not surprisingly, the researchers found that the factors that were associated with liver-related death were a low CD4+ cell count, age, coinfection with HBV/HCV and intravenous (IV) drug abuse. Antiretroviral exposure was not clearly associated with liver-related deaths even after a lot of modeling and tweaking of the data.
These results are reassuring and suggest that treatment has a beneficial effect on this population too. However, the problem with these large databases is that the results of any study are extremely dependent on the quality of the data that are entered and the ongoing quality assurance monitoring. The D:A:D Web site explains how this process works: "Each participating cohort appoints a monitor, who is not in any way associated with the particular site in the cohort. The monitoring must be performed at regular visits to all sites at least once annually. ... The cohort is responsible for selection of patients for review, however 100% of the records of patients who have died or experienced an event/developed signs of cardiovascular [problems] must be reviewed" (D:A:D Manual of Operations, February 2005 edition). I'm not sure if this level of monitoring is enough or not.
My concern about the process is that it is not really random (the site selects the monitor and the charts to be reviewed), and I suspect it is prone to missed events and underestimates the incidence of some problems. Although the data are prospectively collected, they are obtained from the chart record. So it is really a retrospective chart abstraction study with the limitations of that type of study and it is very dependent on the quality of the data abstraction. However, even with all these concerns, it is difficult to envision a system where ongoing toxicities in a large number of patients can be monitored at a reasonable cost.
Authored by: R Weber, C Sabin, N Friis-Møller, A D'Arminio Monforte, P Reiss, W El-Sadr, F Dabis, M Law, C Pradier, S De Wit, on behalf of the D:A:D Study Group
Affiliations: SHCS, Univ Hosp Zürich, Switzerland; Royal Free Ctr for HIV Med, Royal Free and Univ Coll, London, UK; Copenhagen HIV Prgm, Hvidovre, Denmark; ICONA, L Sacco Hosp, Univ of Milan, Italy; HIV Monitoring Fndn, Academic Med Ctr, Univ of Amsterdam, The Netherlands; CPCRA, Columbia Univ, Harlem Hosp, New York, NY, US; Bordeaux Univ Hosp, INSERM U593, France; Australian HIV Observational Database, Natl Ctr in HIV Epidemiology and Clin Res, Sydney; Nice Cohort, CHU Nice Hosp de l'Archet, France; Ctr Hosp Univ St Pierre, Brussels, Belgium
View poster: Download PDF
This article was provided by The Body PRO. Copyright © Body Health Resources Corporation. All rights reserved.