 |
|
|
| |
Non-HIV Markers Add to HIV Signals in Predicting VA Cohort Deaths
|
| |
| |
16th Conference on Retroviruses and Opportunistic Infections, February 8-11, 2009, Montreal
Mark Mascolini
combined HIV and non-HIV factors proved a stronger mortality predictor in the top four quintiles .....The VA team sees this study as "an essential first step" in developing a combined HIV and non-HIV biomarker index......disease. The non-HIV markers were FIB 4 liver injury, renal injury by the modification of diet in renal disease (MDRD) method, two anemia thresholds (hemoglobin 10-12 g/dL and under 10 g/dL), and hepatitis B or C seropositivity
Four markers not specific to people with HIV infection-liver injury, renal injury, anemia, and hepatitis B or C serology--improved the ability of CD4 count, viral load, and AIDS diagnoses to predict death in the largely male US Veterans Administration (VA) cohort [1].
The non-HIV markers "improve differentiation of mortality risk achieved by HIV markers, are strongly associated with HIV biomarkers, and therefore likely reflect HIV pathology," Amy Justice and VA colleagues concluded. The findings reflect the dominance of non-AIDS morbidity and mortality in a SMART analysis, which found that the association between non-AIDS death rates, low CD4 count, and high viral load held true across several groups of non-AIDS conditions, including liver disease, renal disease, and non-AIDS malignancy [2].
To get a better handle on how non-HIV markers influence mortality in an HIV cohort, Justice focused on 9789 VA beneficiaries who began antiretroviral therapy from January 1, 1997 through August 1, 2002 and had complete data on HIV and non-HIV markers and survival. They split the group into a development set including people who started antiretrovirals after December 31, 1998 and a validation set of people who began treatment before that date.
HIV markers included CD4 strata (under 50, 50-99, 100-199, 200-349, and 350 or more), viral load above 100,000 copies, and presence of an AIDS-defining disease. The non-HIV markers were FIB 4 liver injury, renal injury by the modification of diet in renal disease (MDRD) method, two anemia thresholds (hemoglobin 10-12 g/dL and under 10 g/dL), and hepatitis B or C seropositivity. The model also adjusted for age (under 50, 50-64, and 65 or older) and substance abuse and dependence.
Of the 9789 cohort members, about half were in the development set and half in the validation set. For the entire group, 70% were under 50 years old, 26% were 50 to 64, and 4% were 65 or older. Almost all, 97.8%, were men, 51% were black, 32% white, and 17% Hispanic or another race or ethnicity. While 12.5% had a CD4 count under 50, 7.5% had 50-99 CD4s, 16.4% had 100-199, 24.0% had 200-349, and 39.5% had 350 or more. Nearly one in five cohort members (18.3%) had a viral load above 100,000 copies, and almost one third (31.2%) abused either drugs or alcohol.
During follow-up, 2566 cohort member (26%) died. HIV biomarkers were strongly associated with non-HIV markers (P < 0.0001). The analysis predicted death by C statistic, in which the strength of a value is indicated by its proximity to 1.0. In the development set, the C statistic was 0.69 with HIV biomarkers alone, 0.72 with non-HIV markers alone, and 0.74 with combined markers. In the validation set, the C statistic was 0.69 with HIV markers alone, 0.71 with non-HIV markers alone, and 0.74 with the combined markers.
Dividing the cohort into quintiles according to risk of death, Justice found that combined HIV and non-HIV factors proved a stronger mortality predictor in the top four quintiles:
· Quintile 2: HIV markers alone 4 deaths/100 person-years (py), non-HIV markers alone 4 deaths/100 py, combined markers 4.3 deaths/100 py
· Quintile 3: HIV markers alone 5.8 deaths/100 py, non-HIV markers alone 5.5 deaths/100 py, combined markers 6.2 deaths/100 py
· Quintile 4: HIV markers alone 7.3 deaths/100 py, non-HIV markers alone 9.6 deaths/100 py, combined markers 10.8 deaths/100 py
· Quintile 5: HIV markers alone 13.9 deaths/100 py, non-HIV markers alone 9.6 deaths/100 py, combined markers 17.1 deaths/100 py
Justice listed four advantages to this analysis: (1) sufficient sample size and follow-up to analyze mortality with uniform data sources and data collection methods, (2) near-complete mortality ascertainment, (3) an older patient population, and (4) validation of results by independent data. The researchers suggested three limitations: (1) only 75% of veterans are antiretroviral naive when entering the cohort, (2) men dominate the cohort, and (3) veterans without HIV have higher rates of substance abuse, morbidity, and mortality than the general population, though these differences are less pronounced among people with HIV.
The VA team sees this study as "an essential first step" in developing a combined HIV and non-HIV biomarker index. They believe the next steps should include validating the approach in a nonveteran population including more women and calculating response to antiretroviral therapy over time. Justice and coworkers suggested that, "once more completely tested, our index may offer a superior prognostic index and integrated surrogate endpoint for clinical research." In the meantime, the model confirms the growing contribution of non-AIDS diseases to death risk among people with HIV infection.
References
1. Amy Justice, Veterans Aging Cohort Study. Non-HIV biomarkers independently predict mortality and are associated with HIV markers. 16th Conference on Retroviruses and Opportunistic Infections. February 8-11, 2009. Montreal. Abstract 735.
(Poster online at. http://www.retroconference.org/2009/PDFs/735.pdf.)
2. Phillips AN, Neaton J, Lundgren JD. The role of HIV in serious diseases other than AIDS. AIDS. 2008;22:2409-2418.
|
| |
|
|
|
|
|
