|
Antiretroviral penetration into the CNS and incidence
of AIDS-defining neurologic conditions
|
|
|
Download the PDF here
"We estimated that the incidence of HIV dementia increases by more than 70% after initiating an antiretroviral regimen with a high CPE score compared with a low score. However, we found little change in the incidence of toxoplasmosis, cryptococcal meningitis, and progressive multifocal leukoencephalopathy. These results are unexpected, and the interpretation of our effect estimates needs to be tempered by the limitations of our study........These findings should be interpreted cautiously, and additional studies are needed to examine the effect of CPE score on the incidence of HIV dementia more closely. Together with additional data on the safety and effectiveness of different cART regimens, these results may be useful to plan the management of individuals with HIV infection.
......Future studies will be needed to investigate the effect of antiretroviral penetration on the long-term incidence of neuroAIDS, as well as the effect of newer antiretrovirals that are not well represented in current studies, including ours.
antiretroviral penetration into the brain may lead to deposition of β-amyloid plaques, which has been proposed as a possible explanation for a harmful effect of high CPE score on HIV dementia.34 One study observed a higher percentage of extracellular β-amyloid in cART-treated patients than in untreated HIV-positive individuals,35 and HIV-positive individuals with HIV-associated dementia have higher levels of intraneuronal β-amyloid immunoreactivity compared with HIV-positive individuals without HIV-associated dementia.34,-,36 However, the underlying mechanism through which antiretroviral penetration could cause HIV dementia remains unknown. The hypothesis that antiretroviral penetration increases the incidence of HIV dementia via deposition of β-amyloid plaques requires further research to determine whether these associations are in fact causal. Alternative pathways including antiretroviral-related direct neuronal damage and mitochondrial toxicity should also be evaluated.15 Another explanation for the higher dementia risk for regimens with a high CPE score is that these regimens are less effective to treat HIV disease, for example, because of incomplete adherence: 68% of individuals in the study deviated from their initial regimen at some point. However, both the high average proportion of follow-up spent on the initial regimen (58%) and the lack of a strong association between CPE score and opportunistic infections do not support this explanation.
Similar to any other observational study, the validity of our estimates relies on the untestable assumption that the measured covariates were sufficient to adjust for confounding and selection bias. It is possible that consideration of CPE scores is a factor for decisions concerning antiretroviral regimens in patients with neurocognitive symptoms. If individuals with neurocognitive symptoms are more likely to initiate antiretroviral regimens with higher CPE scores, the estimated effect on dementia might be explained by this confounding by indication. If this were the case, we would expect the estimated effect on dementia to disappear or to attenuate after a certain amount of time. However, excluding cases of HIV dementia occurring in the first year of follow-up does not materially change the results. Furthermore, restricting the analysis to individuals who initiated therapy before the introduction of the first CPE scoring system in 2008 does not affect the results.”
Letter to the Editor and author response:
Caniglia et al.1 furthered the debate on the clinical significance of the CNS penetration effectiveness (CPE) score. The authors and the editorialists2 cautioned against selecting high-CPE regimens-often with less-favorable dosing and side-effect profiles-for patients with HIV-associated neurologic disorders. No high-CPE regimens are recommended by the WHO for first-line therapy.3 Although a low-cost combination regimen of nevirapine and zidovudine is widely used in developing countries, the risks of this treatment are unclear.
Selection bias and confounders can undermine observational cohort studies. The authors never mentioned the bias evident when physicians avoid efavirenz in patients with neuropsychiatric issues. These patients would receive regimens with coincidentally higher CPEs but may already have higher risks for developing AIDS dementia. This diagnosis of exclusion takes time. Even the sensitivity analysis excluding cases of AIDS dementia diagnosed within 1 year of enrollment may not address this confounder.
Almost 70% of patients changed regimen in this study, further challenging its validity. Although reflecting the reality of multiple treatment choices in high-income countries, it calls for careful interpretation of the findings. Continued prospective work is needed to elucidate the clinical significance of the CPE score.
Author Response
1. Ellen C. Caniglia and
2. Miguel A. Hernan
3. on behalf of the HIV-CAUSAL Collaboration
Drs. Beardsley and Le offer a possible explanation for the association we found between initiating an antiretroviral regimen with a high CPE score and the incidence of HIV dementia.1 They noted the bias evident when physicians avoid prescribing efavirenz to individuals with neurocognitive symptoms or neuropsychiatric issues. Efavirenz is commonly used as part of the regimens with low and medium CPE scores. However, an analysis that excludes individuals initiating a regimen with efavirenz does not materially affect our results (table). We agree with Beardsley and Le that future studies are needed to examine the effect of CNS penetration on the incidence of HIV dementia.
![EASL1](../images/051616/051616-3/EASL1.gif)
-------------------
Antiretroviral penetration into the CNS and incidence of AIDS-defining neurologic conditions
Neurology 2014
Ellen C. Caniglia Lauren E. Cain, PhD Amy Justice, PhD Janet Tate, PhD Roger Logan, PhD Caroline Sabin, PhD Alan Winston, MD Ard van Sighem, PhD Jose M. Miro, PhD Daniel Podzamczer, PhD Ashley Olson, PhD Jose Ramon Arribas, MD Santiago Moreno, PhD Laurence Meyer, PhD Jorge del Romero, MD François Dabis, PhD Heiner C. Bucher, MD Gilles Wandeler, MD Georgia Vourli, MSc Athanasios Skoutelis, MD Emilie Lanoy, PhD Jacques Gasnault, MD Dominique Costagliola, PhD Miguel A. Hernán, DrPH
On behalf of the HIV CAUSAL Collaboration
Abstract
Objective: The link between CNS penetration of antiretrovirals and AIDS-defining neurologic disorders remains largely unknown.
Methods: HIV-infected, antiretroviral therapy-naive individuals in the HIV-CAUSAL Collaboration who started an antiretroviral regimen were classified according to the CNS Penetration Effectiveness (CPE) score of their initial regimen into low (<8), medium (8-9), or high (>9) CPE score. We estimated “intention-to-treat” hazard ratios of 4 neuroAIDS conditions for baseline regimens with high and medium CPE scores compared with regimens with a low score. We used inverse probability weighting to adjust for potential bias due to infrequent follow-up.
Results: A total of 61,938 individuals were followed for a median (interquartile range) of 37 (18, 70) months. During follow-up, there were 235 cases of HIV dementia, 169 cases of toxoplasmosis, 128 cases of cryptococcal meningitis, and 141 cases of progressive multifocal leukoencephalopathy. The hazard ratio (95% confidence interval) for initiating a combined antiretroviral therapy regimen with a high vs low CPE score was 1.74 (1.15, 2.65) for HIV dementia, 0.90 (0.50, 1.62) for toxoplasmosis, 1.13 (0.61, 2.11) for cryptococcal meningitis, and 1.32 (0.71, 2.47) for progressive multifocal leukoencephalopathy. The respective hazard ratios (95% confidence intervals) for a medium vs low CPE score were 1.01 (0.73, 1.39), 0.80 (0.56, 1.15), 1.08 (0.73, 1.62), and 1.08 (0.73, 1.58).
Conclusions: We estimated that initiation of a combined antiretroviral therapy regimen with a high CPE score increases the risk of HIV dementia, but not of other neuroAIDS conditions.
From jules: you can see just below in Table 3 patients with high CPE score 55% increased risk for dementia.
![EASL2](../images/051616/051616-3/EASL2.gif)
GLOSSARY
cART=combined antiretroviral therapy;
CI=confidence interval;
CPE=CNS Penetration Effectiveness;
ICD-9=International Classification of Diseases, ninth revision;
NNRTI=nonnucleoside reverse transcriptase inhibitor
AIDS-defining neurologic disorders, or neuroAIDS, include HIV dementia and the opportunistic infections toxoplasmosis, cryptococcal meningitis, and progressive multifocal leukoencephalopathy.1,-,3 The incidence of neuroAIDS in developed countries decreased after the introduction of combined antiretroviral therapy (cART) in 1996,2,4,-,9 but antiretroviral drug-related neurotoxicity remains a concern.2,10,-,12
The risk of neuroAIDS may depend on the concentration of antiretrovirals in the CNS, which is a function of their ability to penetrate the blood-brain barrier. Greater exposure of antiretrovirals in the CNS may decrease the HIV RNA in the CSF,13 but may also be neurotoxic.14,15 One proposed method to assess a drug's penetrative ability into the CNS is via the CNS Penetration Effectiveness (CPE) ranking system. In cohort studies to date, lower CPE ranks are associated with higher CSF HIV RNA after adjusting for a number of clinical variables.13,16,-,18
While the association between CPE rank and CSF HIV RNA is reported, the connection between CPE rank and clinical outcomes remains unclear.17,19,-,22 A randomized controlled trial comparing a CNS-targeted therapy to a non-CNS-targeted therapy among 49 individuals with HIV-associated neurocognitive disorders found no difference in improvement of neurocognitive performance after 16 weeks.23 An observational study in the United Kingdom found a similar incidence of neuroAIDS in individuals with high and low CPE scores.24
Herein, we present estimates of the effect of CPE score on the incidence of 4 neuroAIDS conditions among individuals with HIV-1 infection included in a large multinational collaboration of cohort studies from Europe and the United States.
DISCUSSION
We estimated that the incidence of HIV dementia increases by more than 70% after initiating an antiretroviral regimen with a high CPE score compared with a low score. However, we found little change in the incidence of toxoplasmosis, cryptococcal meningitis, and progressive multifocal leukoencephalopathy. These results are unexpected, and the interpretation of our effect estimates needs to be tempered by the limitations of our study.
In the only published study that has considered CPE scores in relation to neuroAIDS among HIV-positive individuals, the incidence of neuroAIDS was similar for a baseline CPE score of 10 or greater and for a baseline CPE score of 4 or less, but the small number of events resulted in a wide 95% CI (hazard ratio: 0.95; 95% CI: 0.53, 1.72). Furthermore, the study was not restricted to antiretroviral therapy-naive individuals and effect estimates for individual neuroAIDS conditions such as HIV dementia were not provided.24 Other studies of the association between CPE scores and risk of cognitive impairment (but not neuroAIDS) have had conflicting results.17,19,-,21,33
Our findings that CPE score does not affect the incidence of opportunistic infections may not be particularly surprising, because the development of these neuroAIDS conditions may be very closely connected to the degree of impaired cell-mediated immunity and not associated with antiretroviral penetration.4,5 In contrast, antiretroviral penetration into the brain may lead to deposition of β-amyloid plaques, which has been proposed as a possible explanation for a harmful effect of high CPE score on HIV dementia.34 One study observed a higher percentage of extracellular β-amyloid in cART-treated patients than in untreated HIV-positive individuals,35 and HIV-positive individuals with HIV-associated dementia have higher levels of intraneuronal β-amyloid immunoreactivity compared with HIV-positive individuals without HIV-associated dementia.34,-,36 However, the underlying mechanism through which antiretroviral penetration could cause HIV dementia remains unknown. The hypothesis that antiretroviral penetration increases the incidence of HIV dementia via deposition of β-amyloid plaques requires further research to determine whether these associations are in fact causal. Alternative pathways including antiretroviral-related direct neuronal damage and mitochondrial toxicity should also be evaluated.15
Another explanation for the higher dementia risk for regimens with a high CPE score is that these regimens are less effective to treat HIV disease, for example, because of incomplete adherence: 68% of individuals in the study deviated from their initial regimen at some point. However, both the high average proportion of follow-up spent on the initial regimen (58%) and the lack of a strong association between CPE score and opportunistic infections do not support this explanation.
Similar to any other observational study, the validity of our estimates relies on the untestable assumption that the measured covariates were sufficient to adjust for confounding and selection bias. It is possible that consideration of CPE scores is a factor for decisions concerning antiretroviral regimens in patients with neurocognitive symptoms. If individuals with neurocognitive symptoms are more likely to initiate antiretroviral regimens with higher CPE scores, the estimated effect on dementia might be explained by this confounding by indication. If this were the case, we would expect the estimated effect on dementia to disappear or to attenuate after a certain amount of time. However, excluding cases of HIV dementia occurring in the first year of follow-up does not materially change the results. Furthermore, restricting the analysis to individuals who initiated therapy before the introduction of the first CPE scoring system in 2008 does not affect the results.
Some limitations of our study should be noted. First, with relatively few events, the 95% CIs around our effect estimates are wide. However, our study is the largest one to date on this topic. Second, the cohorts included in this analysis are from developed countries; our results may not be generalizable to resource-limited settings or to other health care systems. Third, with the exception of VACS, the contributing cohorts used diagnostic procedures that reflect standard clinical practice in Europe. Excluding VACS from the analysis, however, did not significantly alter the results. Fourth, while our effect estimates are adjusted for cohort, we were not able to adjust for the individual centers within each cohort. Thus, some residual confounding due to centers within each cohort is theoretically possible. Finally, the average duration of follow-up in our study was approximately 3 years. Future studies will be needed to investigate the effect of antiretroviral penetration on the long-term incidence of neuroAIDS, as well as the effect of newer antiretrovirals that are not well represented in current studies, including ours.
We estimated that initiation of a cART regimen with a high CPE score increases the risk of HIV dementia, but not of other neuroAIDS conditions. These findings should be interpreted cautiously, and additional studies are needed to examine the effect of CPE score on the incidence of HIV dementia more closely. Together with additional data on the safety and effectiveness of different cART regimens, these results may be useful to plan the management of individuals with HIV infection.
METHODS
Standard protocol approvals, registrations, and patient consents.
Research using the HIV-CAUSAL Collaboration was determined to be nonhuman subjects research by the Institutional Review Board of the Harvard School of Public Health because it involves the study of existing data that are analyzed in such a manner that the subjects cannot be identified, as set forth in US federal regulations. Written informed consent from patients was not required because all data were completely anonymized.
Study population.
The HIV-CAUSAL Collaboration includes prospective cohort studies from 6 European countries and the United States. All cohorts included in the HIV-CAUSAL Collaboration were assembled prospectively and are based on data collected for clinical purposes from national health care systems that offer universal access to care. Each cohort in the collaboration collected data prospectively, including all CD4 cell counts, HIV RNA measurements, treatment initiations, deaths, and AIDS-defining illnesses (including the events of interest: HIV dementia, toxoplasmosis, cryptococcal meningitis, and progressive multifocal leukoencephalopathy).
The individual cohort studies used in these analyses are UK CHIC (United Kingdom), ATHENA (the Netherlands), FHDH-ANRSCO4 (France), SHCS (Switzerland), PISCIS (Spain), CoRIS/CoRIS-MD (Spain), VACS-VC (US veterans), AMACS (Greece), and AQUITAINE (France). Four cohorts of seroconverters with relatively high CD4 counts did not have any neuroAIDS events and were excluded from the analyses.
We restricted our analyses to individuals with HIV-1 infection who met the following criteria at baseline (starting in January 1998): age 18 years or older, no history of AIDS (defined as the onset of any category C AIDS-defining illness),25 antiretroviral therapy naive (as defined elsewhere26), no pregnancy (when information was available), CD4 cell count and HIV RNA measured within the previous 6 months, and initiating a complete antiretroviral regimen (see below) consisting only of drugs with known CPE ranks.
We conducted separate analyses for the following neuroAIDS events: HIV dementia, toxoplasmosis, cryptococcal meningitis, or progressive multifocal leukoencephalopathy. Non-Hodgkin lymphoma was not included as an event because we could not differentiate primary brain lymphoma from other types of non-Hodgkin lymphoma. We also looked at a combined endpoint of any of the 3 opportunistic infections (toxoplasmosis, cryptococcal meningitis, or progressive multifocal leukoencephalopathy). The date of neuroAIDS was identified by the treating physicians. One of the contributing cohorts (VACS) used ICD-9 codes to identify incident neuroAIDS cases. The other contributing cohorts used diagnostic procedures that reflect standard clinical practice in Europe rather than standardized research criteria.
For each patient, follow-up started on the date of initiation of a complete antiretroviral regimen-defined as treatment with at least 2 nucleoside reverse transcriptase inhibitors plus either one or more protease inhibitors, one or more nonnucleoside reverse transcriptase inhibitors (NNRTIs), one entry/fusion inhibitor, or one integrase inhibitor-and ended at death, 12 months after the most recent laboratory measurement, pregnancy (if known), the cohort-specific administrative end of follow-up (ranging between December 2003 and February 2013) or the event of interest, whichever occurred earlier.
Assessment of antiretroviral CNS exposure.
The 2010 CPE ranking system is a proposed method for measuring the penetrative ability of different antiretroviral drugs into the CNS. Each drug is given a rank ranging from 1 to 4 based on pharmacokinetic and pharmacodynamic data, drug characteristics, results of clinical studies, and effectiveness in reducing CSF viral load or improving cognition. A rank of 4 represents the best penetration or effectiveness.14,16 The CPE score for a given regimen is calculated by summing the ranks of each drug in the regimen. We categorized the CPE score for a regimen as low (<8), medium (8-9), or high (>9) based on the distribution of the data (the cut points were approximately at the median and the 75th percentile) (figure 1). Because our estimates may be sensitive to the chosen cut points, we also treated the CPE score as a continuous variable.
Statistical methods.
Using a pooled logistic regression model, we estimated the average “intention-to-treat” neuroAIDS hazard ratio for a high and a medium baseline CPE score compared with a low baseline CPE score. Under the assumption that the monthly probability of an event is small (a condition satisfied in our study), the parameters of our pooled logistic model closely approximate the parameters of a Cox proportional hazards model.27 We computed these estimates separately for each of the 4 neuroAIDS conditions as well as for the combined endpoint of opportunistic infections. The model included month of follow-up (restricted cubic splines with 4 knots at 1, 6, 24, and 60 months) and the following baseline covariates: CD4 cell count (<200, 200-299, ≥300 cells/μL), HIV RNA level (<10,000, 10,000-100,000, >100,000 copies/mL), sex, acquisition group (heterosexual, homosexual/bisexual, injection drug use, other or unknown), calendar year (1998, 1999-2000, 2001-2003, ≥2004), age (<35, 35-50, >50 years), geographic origin (North America or Western Europe, Sub-Saharan Africa, other, or unknown), race (white, black, other, or unknown), years since HIV diagnosis (<1, 1-4, ≥5 years or unknown), whether or not the regimen was an NNRTI-based regimen, and cohort.
The variables we adjusted for in our models are associated with CPE score and widely known to be associated with the outcomes of interest. For example, a baseline CD4 cell count <200 cells/μL, a baseline HIV RNA >100,000 copies/mL, and a baseline age >50 years were associated with an increased odds of both HIV dementia and the combined endpoint of opportunistic infections compared with a baseline CD4 cell count ≥300 cells/μL, a baseline HIV RNA <10,000 copies/mL, and a baseline age <35 years, respectively (data not shown).
To adjust for potential selection bias due to infrequent follow-up, we computed inverse probability weights. Each patient in the above logistic models received a time-varying weight inversely proportional to the estimated probability of not being censored, for each month that patient was followed.28,29 We fit a pooled logistic model using the baseline covariates listed above, the baseline CPE score category, and the most recent measurement of the following time-varying covariates: CD4 cell count (restricted cubic spline with 5 knots at 10, 200, 350, 500, and 1,000 cells/μL), HIV RNA level (<5,000, 5,000-10,000, 10,000-100,000, >100,000 copies/mL), time since last laboratory measure (0, 1-2, 3-4, 5-6, ≥7 months), and AIDS (any category C AIDS-defining illness other than the neuroAIDS condition of interest), and estimated each patient's probability of remaining uncensored in each month of follow-up. The models for the weights were fit before the final models. The weights were stabilized as described elsewhere28 and were then used to fit the final weighted regression model. The estimated weights for each of the 5 outcomes had mean 1.00 (first percentile: 0.96; 99th percentile: 1.20).
We used robust variance estimators that take into account the procedure of weight estimation to compute 95% confidence intervals (CIs) for each of our estimates.30 Under the assumption that the measured covariates are sufficient to adjust for confounding and selection bias, our approach emulates a nonblinded randomized trial in which patients were assigned to regimens with 1 of 3 CPE score categories.28
Several sensitivity analyses were performed for the outcome HIV dementia: we (1) varied the CPE score of regimens that were boosted with ritonavir because the CPE score category for boosted ritonavir is somewhat ambiguous in 2 regimens; (2) restricted the analysis to VACS in order to additionally adjust for the VACS Index31; (3) excluded VACS from the analysis; (4) estimated stabilized inverse probability weights to adjust for potential selection bias due to death, a competing risk32; (5) excluded cases of HIV dementia occurring in the first year; and (6) restricted the analysis to individuals who initiated therapy before the introduction of the first CPE scoring system in 2008. Exclusion of 0.15% of individuals with unusual treatment combinations (e.g., boosted nelfinavir and unboosted darunavir) did not materially affect the estimates. We also considered an alternative categorization of the CPE score by dividing the lowest category into 2 smaller categories (<6 and 6 to <8), and estimated the average log hazard ratios using a model that included a flexible functional form for the continuous CPE score (restricted cubic spline with 4 knots at 5, 7, 9, and 12). We fit a pooled logistic model containing this form of the CPE score as well as the previously listed baseline covariates, and adjusted for potential selection bias due to infrequent follow-up as previously described. All analyses were conducted with SAS 9.3 (SAS Institute, Cary, NC). The LOGISTIC procedure was used to fit the weighted regression models, and a nonparametric bootstrap with 500 samples was used to compute 95% CIs.
RESULTS
A total of 61,938 individuals met the eligibility criteria for our study; 38,786 (62%) initiated a regimen with a low CPE score, 17,687 (29%) with a medium CPE score, and 5,465 (9%) with a high CPE score. The mean score for individuals on an NNRTI-based regimen was 8.2 and the mean for individuals on a non-NNRTI-based regimen was 7.3. Table 1 shows the baseline characteristics of the study population by CPE score category. Individuals with a high CPE score were more likely to be female, heterosexual, initiating therapy before 2004, and of Sub-Saharan African origin. The median (interquartile range) follow-up time was 37 (18, 70) months.
During follow-up, there were 235 cases of HIV dementia, 169 cases of toxoplasmosis, 128 cases of cryptococcal meningitis, and 141 cases of progressive multifocal leukoencephalopathy. Forty individuals developed 2 of the 4 neuroAIDS conditions, and one individual developed 3. The incidence rate (per 10,000 person-years) was 9.0 for HIV dementia, 6.5 for toxoplasmosis, 4.9 for cryptococcal meningitis, and 5.4 for progressive multifocal leukoencephalopathy. The median (interquartile range) time from cART initiation to the event of interest was 14 (2, 39) months for HIV dementia, 4 (1, 16) months for toxoplasmosis, 10 (1, 25) months for cryptococcal meningitis, and 3 (1, 14) months for progressive multifocal leukoencephalopathy.
Table 2 shows the 3 most frequently used cART regimens with high, medium, and low CPE scores. Compared with initiating a cART regimen with a low CPE score, the hazard ratio (95% CI) for initiating a cART regimen with a high CPE score was 1.74 (1.15, 2.65) for HIV dementia, 0.90 (0.50, 1.62) for toxoplasmosis, 1.13 (0.61, 2.11) for cryptococcal meningitis, and 1.32 (0.71, 2.47) for progressive multifocal leukoencephalopathy (table 3). Compared with a low CPE score, the respective hazard ratios (95% CIs) for a medium CPE score were 1.01 (0.73, 1.39), 0.80 (0.56, 1.15), 1.08 (0.73, 1.62), and 1.08 (0.73, 1.58). Figure 2 shows the log hazard ratio against the continuous CPE score for the combined endpoint of opportunistic infections and for HIV dementia (see figure e-1 on the Neurology® Web site at Neurology.org for the individual opportunistic infections).
The hazard ratio of HIV dementia for a high CPE score vs a low CPE score did not vary substantially by CD4 cell count, age, sex, and type of cART regimen (NNRTI-based vs non-NNRTI-based), but the 95% CIs were wide (data not shown). The hazard ratios and 95% CIs did not change with different specifications of the functional form for month of follow-up (data not shown). None of the sensitivity analyses described in the previous section yielded appreciably different results (table e-1). Varying the cut points of the CPE score did not materially change the point estimates but resulted in wider CIs. Excluding cases of HIV dementia occurring in the first year of follow-up (table e-2) did not materially change the estimates of the effect of CPE score on HIV dementia.
|
|
|
|
|
|
|