October 27, 2008
There's nothing like hearing the results of studies directly from those who actually conducted the research. In this interview, you'll meet one of these impressive HIV researchers and read his explanation of the study he presented at ICAAC/IDSA 2008.
My name is Steve Berry. My poster is called, "A High Risk of Hospitalization Immediately Follows HAART Initiation."1 We were interested in this because we know that in the time period right after you start HAART [highly active antiretroviral therapy], there are some unique illnesses that can happen. You can have immune reconstitution inflammatory syndrome [IRIS]. You can have adverse drug reactions. Nobody had previously examined whether, during those first six, 10 or 12 weeks after starting HAART, you have a very high risk of hospitalization because the risk of those adverse effects outpaces the improvement in your immune system.
Stephen Berry, M.D.
What were the patient demographics?
There were about 1,000 patients: 965 were virologic responders and 362 were non-responders. Those two groups were evenly matched in some categories, but not in others. African Americans were slightly more likely to be non-responders. Injection drug use was also slightly more common in the non-responders. We know that in later eras -- 1999 through 2002, or 2003 through 2005 -- you were more likely to have a virologic response to HAART than in the earliest calendar era of HAART use.
By non-responders, do you mean people who failed or people who didn't take their meds?
For the most part, it's probably people who didn't take their meds. That's defined by having a 1-log decrease in viral load by six months after your HAART initiation date. Some of it potentially is underlying resistance, but I doubt that was the bulk of it. It was probably non-adherence.
When we look over this year after HAART initiation and compare hospitalization rates, the primary finding is: For virologic responders and non-responders, for the first 45 days after starting HAART, your risk of serious illness as measured by hospitalization risk stays similar to what your risk was before you started HAART. It doesn't decrease at all yet. Then, between 46 and 90 days after initiating HAART, we see a big change for virologic responders, whereas non-responders stay just where they were in terms of risk of serious illness.
Finally, after 90 days, the risk of serious illness for virologic responders seems to hit a plateau, which stays constant throughout the rest of the year. The rate of hospitalization after 90 days is the same as the overall rate of hospitalization in our clinic population regardless of being on HAART or not. This group seems to have the majority of its clinical benefit from HAART, at least as measured by hospitalization risk, occur by 90 days, but really between 46 and 90 days.
Did you break out which patients were more likely to be hospitalized?
We know that the CD4 count at the time of HAART initiation is the biggest predictor. That's no surprise there. But accounting for that risk in our model, this pattern of decrease at 46 to 90 days holds true. In other words, if you look at persons who have very low CD4 counts when they start HAART, less than 50 cells, they have an even rate of hospitalization for the first 45 days. Then, their risk of hospitalization falls between 46 and 90 days. If you look at the opposite end of the spectrum, for people who are starting HAART at 350 cells, their rate stays constant for the first 45 days and falls between 46 and 90 days.
In addition to lower CD4 count when you start HAART, there were other predictors in the multivariate model for hospitalization risk. These predictors included being female and injection drug use. There's also a borderline significance with African-American race in our analysis. These are all things that have been shown before in other analyses that looked at overall hospitalization risks, but didn't look at risk just after starting HAART.
I see you broke down, a little bit, what type of antiretrovirals patients were on.
Exactly: looking at whether it was an NNRTI [non-nucleoside reverse transcriptase inhibitor]-based regimen or a PI [protease inhibitor]-based regimen. A few persons were on both PI and NNRTI regimens. All patients in the study were HAART-naive at the onset. Interestingly, the type of HAART didn't at all affect the pattern of falling between 46 and 90 days. Also, calendar era of HAART did not affect that pattern: This pattern of hospitalization risk falling between 46 and 90 days was the same for persons who initiated in 1997 through 1998 -- who for the most part had a large pill burden with lots of the old-fashioned PIs -- versus persons who initiated in 2003 to 2005, who were using simplified pill regimens, often with efavirenz [EFV, Sustiva, Stocrin] or atazanavir [ATV, Reyataz].
Do you have details on the reasons these patients were hospitalized?
We do. In the first 45 days after starting HAART, the bulk of the reasons for hospitalization were infectious causes. Breaking that down, it's evenly split between AIDS-defining opportunistic infections and non-opportunistic infections. Among the opportunistic infections, the most common cause was PCP [Pneumocystis pneumonia], followed by Cryptococcus, followed by recurrent bacterial pneumonia.
These are both groups?
These are both groups.
What's the average CD4 count for the HAART responders?
The average CD4 count for responders at initiation was around 150 cells, which is not different than the average CD4 count for non-responders at HAART initiation.
But even here, it probably didn't go up that much. It was still under 200 for patients in the 1- through 45-day range. How much did it go up between days 46 and 90?
We didn't measure that.
Still, we can assume that the hospitalizations among the responders were probably due to low CD4 counts.
Definitely. I would expect that the persons who were more likely to be hospitalized at 46 to 90 days -- we know that CD4 count predicts that risk.
Did you tease out why women were more at risk than men?
That's a good question. It's not due to pregnancy-related complications. Other than that, I don't have a good theory. We know we've seen it in a lot of studies. Some studies have suggested that women potentially are starting HAART at lower CD4 counts. This study adjusted for CD4 count and we still see that women have a higher risk of hospitalization. It may be that the cutoff of 200 or 350 may be different for women and men in terms of hospitalization risks. Or there are other factors that I don't know of.
It could be that only non-infected women access care more. It could be the men just die and the women access care. [Laughs.]
I wouldn't be surprised if that's also a reason.
What were injection drug users [IDUs] hospitalized for? Were the hospitalizations more related to their drug use, or to opportunistic infections?
Someone else suggested I look closely at that, and we could with this dataset. I haven't looked at it. I can speculate that the IDUs have a lot of skin and soft-tissue infections and endocarditis, but I would bet that PCP is probably still a No. 1 cause of hospitalizations among IDUs.
Are you continuing this study?
What we are continuing, and will hopefully be able to publish, is looking at all those specific reasons for hospitalizations. Certainly, as this study went through 2006 and as time goes on, we can extend the same analysis and see how things may change as we move forward in time.
One more question: How does this compare to a long time ago, in terms of 1996, 1997 and 1998?
What's interesting there is this pattern. When do you get better [after starting HAART]? When do your chances of [having a] serious illness get better [i.e., are reduced]? The pattern we show is that this tends not to happen until after 45 days of being on HAART. It happens between 46 and 90 days of being on HARRT. That pattern was the same in 1997 and 1998 as it is now, or in 2003 through 2005. The time period did not seem to matter for when you get better while being on HAART.
Thank you very much.
This transcript has been edited for clarity.