 |
|
|
| |
Correlates of Time-to-Loss-of-Viral-Response in CSF and Plasma in the CHARTER Cohort: CPE Score predicts CSF suppression
|
| |
| |
Reported by Jules Levin
17th CROI Feb 16-19 2010 SF
Scott Letendre1, Ronald J. Ellis1, Reena Deutsch1, David B. Clifford2, Ann C. Collier3, Benjamin B. Gelman4, Christina M. Marra3, Justin C. McArthur5,
J. Allen McCutchan1, Susan Morgello6, David M. Simpson6, Robert K. Heaton1, Igor Grant1, for the CHARTER Group 1
1University of California, San Diego, CA 2Washington University, St. Louis, MO, 3University of Washington, Seattle, WA, 4University of Texas Medical Branch, Galveston, TX, 5Johns Hopkins University, Baltimore, MD, 6Mt. Sinai School of Medicine, New York, NY.
"No clinical trials have determined the correlates of the TLVR in CSF, in part because few have sampled CSF in sufficiently large numbers to provide adequate power."
AUTHOR SUMMARY & CONCLUSIONS
A substantial minority of this observational cohort (19% CSF, 36% plasma) lost their viral response within a year of their initial visit.
- This may be explained by the advanced HIV disease and treatment history of this cohort (74% had AIDS, 25% had 8 or more prior antiretrovirals) and by the conservative estimate of loss-of-viral-response (50 c/mL).
Predictors of time-to-loss-of-viral-response were similar in CSF and plasma.
- Worse neuropsychological performance was associated with shorter times-to-loss-of viral response in CSF and plasma.
Characteristics of the antiretroviral regimen (use of an NNRTI or a PI, CNS penetration-effectiveness estimates) were associated with time-to-loss-of-viral-response only in CSF, not in plasma.
-These findings reinforce accumulating evidence that antiretrovirals that reach higher concentrations in the nervous system lead to better control of HIV in CSF. From Jules: Figure 3 shows patients with CPE score 8 or greater and >44 yrs old had better CSF suppression).
STUDY OBJECTIVE
Determine the time-to-loss-of-viral-response in a large, longitudinal cohort and compare the results in CSF and plasma, with particular emphasis on the effects of neuropsychological performance and antiretroviral penetration.
BACKGROUND
Analyses of HIV RNA levels (viral loads) in cerebrospinal fluid (CSF) have provided valuable insights into the pathogenesis of HIV-associated neurocognitive disorders (HAND). Many prior analyses were cross-sectional but longitudinal analyses are better suited for inferring causality.
Longitudinal analyses of time-to-loss-of-viral-response (TLVR) in clinical trials have provided important insights into the antiviral effectiveness of antiretroviral therapy (ART) in blood. No clinical trials have determined the correlates of the TLVR in CSF, in part because few have sampled CSF in sufficiently large numbers to provide adequate power.
Certain observational cohorts, such as CHARTER (CNS HIV AntiRetroviral Therapy Effects Research), have sampled CSF in very large numbers over several years, giving them the power to perform TLVR analyses. TLVR analyses based in observational cohorts have important limitations since initiation of ART is typically not observed. As a result, these analyses require assumptions, such as that ART was similarly effective between individuals during the unobserved period between initiation of ART and the initial study assessment. While this limits the strength of the conclusions of TLVR analyses nested in observational cohorts, these analyses may still provide valuable insights into ART effectiveness in the nervous system.
RESULTS
Summary of the CPE 2010 ranking approach
Table 2 summarizes characteristics of the participants. Those in the plasma analysis were similar to those in the CSF analysis, except that they had slightly less advanced immune suppression and, by design, had undetectable plasma viral loads.
CSF analysis: 67 events (19%) occurred over a median of 9.3 months and 279 (81%) maintained suppression over a median of 20.3 months. Baseline predictors of shorter TLVRs (ime to viral rebound) in CSF were detectable plasma viral load, CD4+ cell counts < 200, age ≤ 44 years, black ethnicity, protease inhibitor (PI) use, and absence of antidepressant (AD) use. Two interactions were identified: black people with worse NP performance had shorter TLVRs than others and older people who took better penetrating ART regimens had longer TLVRs than others. The multivariable analysis is summarized in Table 3 (model p < 0.0001). The Figures display the relationships between TLVR in CSF and 3 of the significant covariates (Figure 1: CD4+ cell count
Plasma analysis: 82 events (36%) occurred over a median of 11.0 months and 143 (64%) maintained suppression over a median of 18.0 months. Baseline predictors of shorter TLVRs in plasma were CD4+ counts < 200, age ≤ 44 years, black
Background
Analyses of HIV RNA levels (viral loads) in cerebrospinal fluid (CSF) have provided valuable insights into the pathogenesis of HIV-associated neurocognitive disorders (HAND). Many prior analyses were cross-sectional but longitudinal analyses are better suited for inferring causality.
Longitudinal analyses of time-to-loss-of-viral-response (TLVR) in clinical trials have provided important insights into the antiviral effectiveness of antiretroviral therapy (ART) in blood. No clinical trials have determined the correlates of the TLVR in CSF, in part because few have sampled CSF in sufficiently large numbers to provide adequate power.
Certain observational cohorts, such as CHARTER (CNS HIV AntiRetroviral Therapy Effects Research), have sampled CSF in very large numbers over several years, giving them the power to perform TLVR analyses. TLVR analyses based in observational cohorts have important limitations since initiation of ART is typically not observed. As a result, these analyses require assumptions, such as that ART was similarly effective between individuals during the unobserved period between initiation of ART and the initial study assessment. While this limits the strength of the conclusions of TLVR analyses nested in observational cohorts, these analyses may still provide valuable insights into ART effectiveness in the nervous system.
Demographic and Clinical Characteristics of the CSF (N = 346) and Plasma Groups (N = 225). Durations are in months, 1Mean (Standard deviation), 2 Number (%), 3 Median (Inter-quartile range)
Predictors of Time-to-Loss-of-Viral Response in CSF and Plasma.
*P value was < 0.15 for the interaction that included this variable.
Figures. Kaplan-Meier plots for 3 representative baseline predictors of TLVR in CSF.
Methods
Design: Longitudinal analysis of HIV seropositive participants in the CHARTER study, a North American cohort based at six sites: Baltimore, Galveston, New York, St. Louis, San Diego, and Seattle. The protocol was approved by the Institutional Review Board at each site. All participants gave informed consent for the study procedures.
Participant Inclusion Criteria: Participants were eligible to be included in this analysis if they met three criteria at their initial visit study visit: They used ART at their initial visit, HIV RNA levels were below the lower limit of quantification (LLQ) in CSF (N = 346) or in plasma (N = 225), and they had at least 2 subsequent semiannual visits.
Variable Description: Time-to-loss-of-viral-response was defined as a viral load > 50 c/mL
Neuropsychological (NP) Testing: Participants completed a battery of NP tests that have sensitivity to HAND, including measures of working memory, information processing speed, learning and memory, motor skills, verbal fluency, and executive functions. These data were summarized using the Global Rating methodology, which accounts for both the number and severity of deficits evident in an individual's NP profile (range = 1-9, with higher scores indicating greater levels of impairment) and global impairment ratings.
Clinical Procedures: HIV viral loads were quantified in plasma and CSF by reverse transcriptase-polymerase chain reaction using a commercial assay with LLQ of 50 copies/mL.ART non-adherence was estimated by self-report of taking fewer than 95% of doses in the 4 days prior to assessment. ART penetration was estimated by the revised CPE 2010 method (Table 1).
Statistical Analysis: Kaplan-Meier estimates of TLVR were calculated for dichotomized demographic, clinical, and NP variables. Baseline variables that were significant at the 15% significance level (Wilcoxon) were analyzed by multivariable Cox proportional hazards models and likelihood ratio tests.
|
| |
|
|
|
|
|
