Biostatistics Home Page Biostat Connections Contents

Biostat Connections
October 1997

An Interview with Steve Lagakos about the Past, Present, and Future of SDAC

by Andrea Rotnitzky

Steve Lagakos has recently stepped down from as Director of SDAC, for which he has been principal investigator since its creation in 1989. We have interviewed Steve to tell us about the history and future of SDAC and its role in AIDS research.

AR What does SDAC stand for?

SL Statistical and Data Analysis Center.

AR Can you tell us about the history of the origins of SDAC?

SL In 1988 the government formed a single federal effort to develop and evaluate treatments for persons with HIV/AIDS. At first, they formed what was called the Statistical Center at Harvard, a data management center in North Carolina (called RTI), and a group of about 20 university-based clinics around the country. At that time they thought all AIDS treatment studies would be Phase I trials because they didn't think they could find something that would be effective for a long time. Almost at the same time the first results for AZT came out that showed evidence that it prolonged survival, so, the government formed a large scale cooperative group of institutions with a statistical center (=SDAC) that was in a position to engage in serious statistical research and design of large scale trials. That's how SDAC officially became an entity in August of 1989. We are part of a larger group called the AIDS Clinical Trials Group (ACTG), consisting of about 100 university-based clinics. For the initial Statistical Center we had four FTEs, that is four full time equivalent people, spread out among 6-8 faculty members. Since then, we have grown to about 160 FTEs!

AR How much participation has SDAC received from the faculty and student body of the department?

SL Of the primary faculty, there are 12 who are supported for from 20% to 70% of their time and then there are about 7 additional faculty whose primary appointment is elsewhere, including Drs. Richard Gelber, Rebecca Gelman, and Dianne Finkelstein. There are also about five faculty from outside the department from other parts of the Harvard School of Public Health and from the Harvard Medical School. As for students, we don't employ them, but rather we support the research carried out by several of them. Right now there are about seven or eight students supported SDAC who are doing their thesis research in topics related to AIDS.

AR What are in your view the major contributions of SDAC's research to the advancement of the understanding of AIDS?

SL It is hard to separate our own individual achievements from those of the whole group. We really work as a team. In terms of treatment of people who have HIV, I would say that 80% of all the treatment advances that have occurred in the world have been through the ACTG, so we have been very central on all of those advances. We have also been involved in the prevention of HIV transmission from mother to child. The first and most important study was designed and analyzed here by Dr. Richard Gelber and other colleagues. I have just heard that Thailand has instituted a national prevention program for maternal-to-child transmission of HIV. All mothers who are pregnant who have HIV will have treatment to prevent the transmission of the disease to their babies. This is a real accomplishment because without this intervention about 30% to 35% of babies of HIV mothers would become infected and they would live a miserable life and die at an early age. Now, that rate is being cut down to 10%, and for those not infected it is a true cure.

AR Has SDAC also played a role in the dissemination of the results and the public awareness of the prevention of the disease?

SL When we write our reports with the results of our analyses, we also write letters to the patients and the clinics and describe the results in simple terms, again throughout the ACTG effort.

AR Can you describe some of the important methodological work that has been initiated by the researchers in SDAC?

SL I would say that in terms of statistical methods the advancements have been comprised largely by extensions of already existing statistical methods: more sophisticated techniques for interim analysis and more general types of experimental situations that can be handled. Also there has been important work done on a number of approaches that have to do with the interface of longitudinal data and failure time data. Some of these methods existed before but they really come to play a role in AIDS because we now constantly analyze trials with those techniques. The interplay between surrogate markers and survival is a special case of this. Within the area of failure time analysis there has been work on variations of problems related to truncation and censoring, in particular interval censoring and doubly interval censoring. Also, there has been important methodological work on the analysis of studies with non-compliance. With the new AIDS drugs, called protease inhibitors, compliance has taken a new role because the effectiveness of these new combinations depends a lot on compliance. The feeling is that if a patient is not compliant, say he stops taking the drug during some period, then the viral load grows up quickly and therefore the risk of developing viral resistance is enhanced. So there is currently the thinking that a big cause of failure of these drugs is that patients are not compliant. This is a whole new aspect of the analysis of studies with non-compliance, compared to the older days when the concern was of the effect of non-com-pliance on the attenuation of a treatment effect or its role as a kind of informative lost to follow-up.

AR What are the challenges for the future?

SL Accounting for non-compliance is certainly one. Another area is the understanding of how the virus works and therefore of strategies of how to fight it. This area changed 180 degrees about two years ago when a whole new model was developed for what goes on inside the body of an infected person. Using some mathematical models, actually very simple deterministic differential equations models, investigators were able to show that what appears to go on is a huge daily battle where hundreds of millions of virions are being neutralized by the body's immune cells and what happens over a period of years is that the body just can't keep producing enough. Much more work has to be done in this area because what has been done so far is largely deterministic and it does not account for heterogeneity. Stochastic models are what is being called for here. The other big issue now is that you can get treatments that keep the virus suppressed almost at undetectable levels for prolong periods of time, but not forever because resistance does eventually happen. So an important question now is which anti-HIV drugs do you start taking first, when do you switch, and what do you switch to? We can't do a clinical trial to answer these questions because it will take too many years to get results and by then our knowledge of the treatment of the disease will have changed. So now there are stochastic models being developed for the whole disease process to try to answer these questions. I see this as a challenging area where we will see important developments in the next few years.

AR Finally, what is CBAR?

SLs CBAR is the Center for Biostatistics and AIDS Research. SDAC refers to an entity formed by a large government grant. CBAR is a research center that Harvard has created that houses the work being done at SDAC but also has other projects and allows us to expand into other areas, such as AIDS in developing countries. A real problem in HIV is that the new drugs are wonderful but very expensive. To treat a patient today could cost $40,000-$100,000 per year. And these costs cannot be afforded by developing countries. The poorest of those countries spends more like $5 a year per person on total health costs! We need different solutions, and one of these is a vaccine. So, one of the goals of CBAR is to create ties in developing countries to help them develop expertise in the study design and then to work with them to institute vaccine trials in the future. Another goal is dissemination of our knowledge of trial design. And finally, another goal is the dissemination of information about infection from HIV because a lot of it is preventable.

Maintained by the Biostatistics Webmaster
Last Update: 9 October 1997