Abstract
Objective:
HIV has been shown to persist in cerebrospinal fluid (CSF) in persons on antiretroviral therapy (ART), which may be linked with inadequate ART exposure, potentially contributing to neurocognitive dysfunction. Our objective was to compare ART regimen inhibitory activity in CSF with participant outcomes.
Methods:
A5321 is a prospective study of HIV reservoirs among persons with HIV on long-term ART with documented suppressed viremia; 44 participants who underwent lumbar puncture were evaluated. CSF trough concentrations were determined and inhibitory quotients (IQs) calculated for each antiretroviral (ARV) in a participant’s regimen as the ratio of CSF concentration to literature values for in vitro HIV IC50or90. The geometric mean (GM) of CSF IQs of all drugs in each participant’s ART regimen was calculated (CSF ART-IQGM). Statistical analyses evaluated associations among ART-IQGM and CSF HIV DNA, biomarkers and global deficit score (GDS).
Results:
The median (Q1, Q3) CSF ART-IQGM was higher in those with undetectable vs detectable CSF HIV DNA: 0.9 (0.5, 1.6) vs 0.5 (0.3, 0.9), p=0.027. Higher ART-IQGM was associated with lower GDS (i.e., better global cognitive function, Spearman: −0.30, p=0.05). There was no association between CSF inflammatory biomarkers and ART-IQGM.
Conclusions:
The CSF ART-IQGM is a novel approach to assess ART regimen activity. This metric was associated with lack of CSF HIV DNA detection and better GDS, which suggests ART regimen activity affects HIV persistence in CSF. This tool motivates further investigations of relationships between regimen activity and biomarkers of HIV persistence in CSF and in other anatomic HIV reservoirs.
Keywords: Antiretroviral therapy (ART) regimen activity, CSF, pharmacokinetics/pharmacodynamics (PK/PD), inhibitory quotient (IQ), HIV persistence in CSF
Introduction
Optimizing brain health is an important goal in people with HIV (PWH). HIV has been shown to establish reservoir sites in (PWH), including in the central nervous system (CNS), which may be an obstacle to viral eradication [1, 2]. Furthermore, these reservoir sites may not achieve therapeutic antiretroviral (ARV) concentrations because of inadequate drug penetration through the blood-brain barrier and/or blood-cerebrospinal fluid (CSF) barrier. In some cases, low ARV concentrations have been associated with evidence of low-level ongoing virus replication [3, 4].
The CNS penetration effectiveness (CPE) score was created in 2008 as a proposed method of ranking the CNS penetration and activity of ARVs [5]. Each drug was ranked from 1 to 4 based on pharmacokinetic and pharmacodynamic (PK/PD) data, drug characteristics, results of clinical studies, and effectiveness in reducing CSF viral load or improving cognition. A rank of 4 represented the best penetration or effectiveness. The CPE score for a given ARV therapy (ART) regimen was calculated by summing the ranks of each drug in the regimen. The CPE, however, implicitly assumed all participants on the same ART regimen achieved the same ARV concentrations in the CSF, overlooking known interpatient variability in ARV CSF concentrations.
The objective of this study was to develop a new metric that encompassed person-specific CSF PK/PD based on all ARV components of their ART regimen and to investigate relationships among this ART regimen PK/PD metric and measures of viral persistence in the CSF and cognitive function.
Methods
Study participants and clinical procedures
Advancing Clinical Therapeutics Globally (ACTG) study A5321 is a longitudinal prospective cohort study of HIV reservoirs. Eligible PWH had started combination ART as part of an ACTG treatment-naïve trial and were on ART for ≥2 years with well-documented sustained plasma viral suppression. The full details of A5321 have been previously published [6–8]. Lumbar punctures were performed using 22-gauge pencil-point needles in most cases to withdraw ~20 mL of CSF. CSF samples were promptly centrifuged, and supernatants and cell pellets stored at −80°C according to established ACTG CSF processing protocols. CSF and blood samples were collected from each participant at various time points post-ART dose. All participants in this analysis received tenofovir (TFV, as tenofovir disoproxil fumarate, TDF [300mg daily]) and emtricitabine (FTC [200mg daily]), plus an additional ARV: efavirenz (EFV [600mg daily]), dolutegravir (DTG [50mg daily]), elvitegravir/cobicistat (EVG/c [150mg/150mg daily]), raltegravir (RAL, [400mg twice daily]), atazanavir/ritonavir (ATV/r [300mg/100mg daily]), or darunavir/ritonavir (DRV/r, daily [800mg/100mg daily]).
Virologic and immunologic methods
HIV persistence was measured as cell-associated HIV DNA (CA-DNA) and RNA (CA-RNA) and cell-free HIV RNA. CA-DNA and CA-RNA were measured by qPCR assays in PBMCs and CSF cell pellets derived from approximately 13 mL CSF and normalized by amplifiable CCR5 cell equivalents [8]. Cell-free HIV RNA was quantified by integrase single copy assay in CSF supernatant (3–5 mL) and blood plasma (5 mL) [8]. Six inflammatory biomarkers, selected for the parent A5321 study based on biological relevance as markers of monocyte activation, neuroinflammation, and neuronal injury in treated HIV infection, were measured in cell-free CSF by ELISA (interleukin 6 [IL-6], interferon gamma-induced protein 10 [IP-10], monocyte chemoattractant protein-1 [MCP-1], cluster of differentiation 14 [sCD14], cluster of differentiation 163 [sCD163], and neopterin [GenWay Biotech]).
Neurocognitive outcomes
Methods were adapted from previous A5321 neurocognitive analyses [8]. Briefly, the neuropsychological assessment consisted of 15 individual tests administered prior to or on the same day as the lumbar puncture. Z-scores were calculated by standardizing each raw test score by age, sex, race, and years of education. Deficit scores were calculated for each individual test based on the z-score. If the z-score was missing, then the deficit score was also missing. Deficit scores range from 0 to 5, with higher scores indicating more severe impairment as follows: 0 (Normal), 1 (Mild Cognitive Impairment), 2 (Mild to Moderate Cognitive Impairment), 3 (Moderate Cognitive Impairment), 4 (Moderate to Severe Cognitive Impairment), and 5 (Severe Cognitive Impairment). Individual deficit scores were averaged to give the Global Deficit Score (GDS), which has been clinically validated [9].
Determination of ARV concentrations in plasma and CSF and ART PK analysis
Plasma and CSF concentrations of each ARV (total drug) were quantified by LC-MS/MS using individual standard curves for each matrix as previously described [10]. Base PK models were developed for each ARV using a 2-compartment model with oral absorption and linear elimination from the central compartment and a CSF compartment. PK modeling was accomplished using the nonparametric adaptive grid (NPAG) algorithm within the Pmetrics package version 2.1.0 (Los Angeles, CA) for R version 4.3.1 (R Foundation for Statistical Computing, Vienna, Austria). To standardize pharmacokinetic characteristics across all participants, estimates of participant-specific trough concentrations and exposure metrics (i.e., area under the curve [AUC]) in the CSF were obtained. The full details of these PK methods have been published [11].
Calculation of inhibitory quotient (IQ) ratio for the ART regimen
CSF IQs were calculated for each individual ARV in a participant’s ART regimen as a ratio of model-estimated CSF trough concentration to literature values for in vitro HIV inhibitory concentration (IC). For all ARVs except the NRTIs, IC90or95 were used. For the NRTIs, we used IC50 values as IC90or95 values are not available. IQs based on the IC50 may slightly overestimate inhibitory activity compared with the IC90. This bias is minimized, however, because all participants were receiving the same dual NRTI regimen and therefore any bias was uniform across all IQ and geometric mean IQ calculations. The protein-free IC50or90 values used were [12]: FTCIC50 = 1.73 ng/mL [13], TDFIC50 = 143.7 ng/mL [14], EFVIC95 = 1.6 ng/mL [15], DTGIC90 = 0.64 ng/mL [16], RALIC95 = 9.0 ng/mL [17], EVGIC90 = 0.53 ng/mL [18, 19], ATVIC95 = 6.5 ng/mL [20], and DRVIC95 = 2 ng/mL [20]. The geometric mean (GM) of the CSF IQs for each component of a participant’s ART regimen was calculated (CSF ART-IQGM). As a supplemental approach, the arithmetic mean CSF ART-IQ score for each regimen was also calculated.
Statistical Analysis
Associations between the CSF ART-IQGM and CSF HIV DNA, biomarkers, and GDS were assessed using the exact Wilcoxon rank-sum test and Spearman correlations. All p-values and confidence intervals presented are unadjusted for multiple comparisons as per the statistical analysis plan for this exploratory analysis. Males and females were enrolled in this cohort study; however, the influence of sex was not assessed.
Study Approval
The institutional review boards of all participating institutions approved this study. Participants in A5321 provided written informed consent for an optional single cross-sectional lumbar puncture, a concurrent blood sample, and neuropsychological assessment.
Data Availability
All data underlying the findings are fully available without restriction. The study data are available upon request from the Statistical and Data Management Center (SDAC) of the ACTG; the request may be made by emailing: sdac.data@sdac.harvard.edu.
Results
Participant Characteristics
Forty-four participants on TDF and FTC-containing ART who started ART during chronic HIV and had available paired CSF and blood plasma ARV concentrations were included in this analysis. Participant characteristics at the time of lumbar puncture, including stratification by detection of CSF CA-DNA, are summarized in Table 1. Seventeen participants (39%) were taking EFV, eight (18%) RAL, eight (18%) ATV/r, five (11%) EVG/c, four (9%) DRV/r, and two (5%) were taking DTG.
Table 1.
Participant demographic and clinical factors, overall and by CSF cell-associated HIV DNA viral outcome (not detected, detected)
| Category | Variable | Not detected (N=22) | Detected (N=22) | Total (N=44) |
|---|---|---|---|---|
| A5321 group | Treated during chronic HIV | 22 (100%) | 22 (100%) | 44 (100%) |
| Sex, n (%) | Male | 21 (95%) | 22 (100%) | 43 (98%) |
| Female | 1 (5%) | 0 (0%) | 1 (2%) | |
| Gender, n (%) | Male | 17 (94%) | 19 (100%) | 36 (97%) |
| Female | 1 (6%) | 0 (0%) | 1 (3%) | |
| n | 18 | 19 | 37 | |
| Race/ethnicity, n (%) | White Non-Hispanic | 16 (73%) | 17 (77%) | 33 (75%) |
| Black Non-Hispanic | 4 (18%) | 2 (9%) | 6 (14%) | |
| Hispanic | 1 (5%) | 3 (14%) | 4 (9%) | |
| American Indian/Alaskan Native | 1 (5%) | 0 (0%) | 1 (2%) | |
| Age, years | Median (IQR) | 47 (42, 57) | 51 (47, 56) | 49 (42, 57) |
| Years on ART | Median (IQR) | 8.1 (7.0, 9.3) | 8.2 (6.3, 11.7) | 8.1 (6.3, 9.4) |
| Anchor ARV, n | EFV | 9 (41%) | 8 (36%) | 17 (39%) |
| ATV (/r) | 5 (23%) | 3 (14%) | 8 (18%) | |
| EVG (/c) | 2 (9%) | 3 (14%) | 5 (11%) | |
| RAL | 4 (18%) | 4 (18%) | 8 (18%) | |
| DTG | 1 (5%) | 1 (5%) | 2 (5%) | |
| DRV (/r) | 1 (5%) | 3 (14%) | 4 (9%) | |
| Pre-ART CD4+ T-cell count (cells/mm³) | Median (IQR) | 285 (187, 340) | 218 (82, 298) | 267 (139, 316) |
| CD4+ T-cell count (cells/mm³) most recent prior to LP | Median (IQR) | 670 (539, 790) | 629 (462, 967) | 642 (498, 822) |
| Pre-ART CD4:CD8 ratio | Median (IQR) | 0.3 (0.2, 0.4) | 0.2 (0.1, 0.4) | 0.2 (0.1, 0.4) |
| CD4:CD8 ratio most recent prior to LP | Median (IQR) | 1.0 (0.7, 1.3) | 0.9 (0.7, 1.1) | 1.0 (0.7, 1.3) |
| Pre-ART plasma HIV-1 RNA (log₁₀cps/mL) | Median (IQR) | 4.6 (4.4, 4.8) | 4.7 (4.5, 5.2) | 4.7 (4.5, 5.0) |
| Plasma HIV-1 RNA (cps/mL) most recent prior to LP | <40 | 22 (100%) | 21 (95%) | 43 (98%) |
| ≥40 | 0 (0%) | 1 (5%) | 1 (2%) |
Abbreviations: EFV, efavirenz; DTG, dolutegravir; RAL, raltegravir; EVG, elvitegravir; ATV, atazanavir; DRV, darunavir/ritonavir; LP, lumbar puncture; ARV, antiretroviral; ART, antiretroviral therapy; IQR, interquartile range; /r, with ritonavir; /c with cobicistat
CSF IQs
Model-derived CSF trough concentrations were determined for each participant and each of the eight different ARVs used in these participants’ ART regimens [11]. The CSF IQ for each individual ARV and the GM of each participant’s ART regimen were calculated (i.e., the CSF ART-IQGM). Figure 1 summarizes CSF IQ values for each ARV and for the entire ART regimen by the anchor (or 3rd ARV) drug. The geometric mean IQs correlated reasonably well with arithmetic means (Spearman r = 0.45; p = 0.002). Consistent with CSF penetration and CPE scores, the CSF IQs were highest for FTC, followed by EFV, and lowest for TFV. The calculated regimen CSF ART-IQGM values showed considerable spread among participants on the same ARV combination (Figure 1b), arising from differences in each participant’s measured CSF concentrations.
Figure 1.
CSF IQs of each ARV (a) and CSF ART-IQGM values for the full TDF + FTC + 3rd drug regimen stratified by anchor (3rd drug) ARV (b). Inhibitory quotients were calculated as the ratio of the PK model-estimated CSF trough concentration to the literature-defined inhibitory concentrations in ng/mL shown in (a). 50% inhibitory concentrations were used to calculate the IQ50 where 90% or 95% inhibitory concentrations were unavailable. CSF IQs are plotted by the CPE score for their respective ARVs.
Abbreviations: TFV, tenofovir; FTC, emtricitabine; EFV, efavirenz; DTG, dolutegravir; RAL, raltegravir; EVG, elvitegravir; ATV, atazanavir; DRV, darunavir/ritonavir; IQ, inhibitory quotients; CSF, cerebrospinal fluid; IC, inhibitory concentration; CPE, central nervous system (CNS) penetration effectiveness score; ARV, antiretroviral agent
ART Regimen CSF IQs Exposure-Response Relationships
The CSF ART-IQGM was higher in participants with undetectable CSF CA-DNA versus those with detectable CA-DNA: median (Q1, Q3), 0.9 (0.5, 1.6) for not detected vs. 0.5 (0.3, 0.9) for detected, p=0.027 (Figure 2). In those with detectable HIV DNA in CSF, the median CA-DNA was 2.3 log10 copies/mL (range: 0.1, 7.0 log10 copies/mL). A rank-based analysis gave similar results. Only 4 participants had detectable CA-RNA in the CSF, precluding the ability to investigate this relationship. No association was found between CSF inflammatory biomarkers (IL-6, IP-10, MCP-1 sCD14, sCD163, neopterin) and CSF ART-IQGM (Table 2).
Figure 2.
CSF ART-IQGM and CSF HIV DNA detection. CSF ART-IQGM was higher in those with undetectable vs detectable CSF HIV DNA (median [Q1, Q3]: 0.9 [0.5, 1.6] vs 0.5 [0.3, 0.9], p=0.027)
Abbreviations: ART, combination antiretroviral therapy; CSF, cerebrospinal fluid; IQ, inhibitory quotient; GM, geometric mean
Table 2.
Spearman rank-based correlations between CSF inflammatory biomarkers and ARV drug activity in the CSF
| Geometric mean IQ | ||
|---|---|---|
| Spearman* | P-value | |
| CSF IL-6 (pg/mL) | 0.10 | 0.50 |
| CSF IP-10 (pg/mL) | −0.19 | 0.23 |
| CSF MCP-1 (pg/mL) | −0.26 | 0.08 |
| CSF neopterin (nMol/L) | 0.04 | 0.82 |
| CSF sCD14 (ng/mL) | 0.13 | 0.41 |
| CSF sCD163 (ng/mL) | −0.01 | 0.94 |
Correlations based on 44 participants.
Abbreviations: IQ, inhibitory quotient; CSF, cerebrospinal fluid; IL, interleukin; IP, interferon gamma-induced protein; MCP, monocyte chemotactic protein; CD, cluster of differentiation.
Associations among measures of ARV drug activity in CSF and neurocognitive performance were assessed with respect to the seven domain-specific scores and the GDS. The full analysis is presented in Table 3. Higher CSF ART-IQGM for the full regimen was associated with lower GDS, that is, better global cognitive performance (Spearman: −0.30, p=0.05), although no association was evident with the global z-score (Spearman: 0.12, p=0.47; details not shown).
Table 3.
Spearman rank-based correlation between neurocognitive deficit scores and measures of ARV drug activity in the CSF
| IQ50 for TFV | IQ50 for FTC | IQ90 for 3rd drug | Geometric mean IQ | |||||
|---|---|---|---|---|---|---|---|---|
| Spearman* | P-value | Spearman* | P-value | Spearman* | P-value | Spearman* | P-value | |
| GDS | −0.41 | 0.007 | −0.10 | 0.55 | −0.09 | 0.56 | −0.30 | 0.05 |
| Fine motor DS | −0.10 | 0.54 | 0.09 | 0.57 | −0.08 | 0.60 | −0.14 | 0.38 |
| Verbal learning DS | −0.22 | 0.16 | −0.11 | 0.49 | 0.12 | 0.46 | −0.02 | 0.90 |
| Verbal memory DS | −0.24 | 0.12 | −0.33 | 0.037 | 0.26 | 0.09 | −0.07 | 0.68 |
| Executive functioning DS | −0.24 | 0.13 | −0.05 | 0.76 | −0.07 | 0.68 | −0.19 | 0.24 |
| Speed of processing DS | −0.43 | 0.005 | −0.06 | 0.73 | −0.20 | 0.21 | −0.44 | 0.004 |
| Attention DS | −0.22 | 0.16 | −0.08 | 0.60 | −0.01 | 0.97 | −0.17 | 0.28 |
| Language DS | −0.10 | 0.53 | 0.27 | 0.09 | −0.33 | 0.041 | −0.24 | 0.15 |
Correlations based on GDS for 41 English-speaking participants (39 participants for language GDS).
Abbreviations: IQ, inhibitory quotient; GDS, global deficit score; DS, domain deficit score; TFV, tenofovir; FTC, emtricitabine; CSF, cerebrospinal fluid; IC, inhibitory concentration; ARV, antiretroviral agent
Due to the known adverse neuropsychiatric effects of EFV [21], associations between estimated EFV PK parameters in CSF versus the GDS and domain-specific deficit scores and z-scores were explored. The estimated EFV CSF AUC was negatively correlated with the mean z-score (r=−0.52, p =0.048, n=15), suggesting that increased concentrations of EFV in CSF were associated with lower global z-scores, indicating neurocognitive impairment.
Discussion
The goal of ART is to inhibit HIV replication throughout the body and maintain an undetectable viral load. For PWH with neurocognitive dysfunction an additional consideration is optimizing brain health. In this study of 44 PWH who started combination ART in an ACTG treatment-naïve trial and were on ART for a median of 8.1 years with well-documented sustained plasma viral suppression, 22 (50%) had HIV DNA detected in CSF cells demonstrating viral persistence. Using a novel pharmacometric method we developed to quantify antiretroviral activity in CSF of the entire ART regimen that each participant was receiving, higher regimen activity (quantified as the CSF ART-IQGM) was associated with a lack of CSF CA-DNA detection. Furthermore, a higher CSF ART-IQGM score was associated with better global cognitive function. These findings provide evidence to support that CSF viral outcomes are linked with overall ART regimen activity.
Our work builds upon two important concepts and constructs to classify CSF penetration of various ARVs and assess the activity of an ART regimen: the CPE score and the inhibitory quotient (IQ) [5]. A limitation in the construct of the CPE score is the implicit assumption that all persons taking the same ARV have the same CSF concentrations of that ARV. Thus, individual-level exposure (i.e., interpatient PK variability) was not incorporated and every person receiving the same regimen would have the same CPE score. Our previous work has illustrated the variability in CSF levels of ARVs [11]. Studies have shown a higher CPE score was associated with lower CSF HIV RNA and a correlation between CPE score and CSF viral escape [20]. However, other studies that investigated the effects of higher ART CPE score for treatment and/or development of HIV-associated neurocognitive disorder (HAND) in PWH found conflicting or inconclusive results [22–25]. A recent review of 23 studies that used the CPE score and assessed neurocognitive performance concluded that CPE was not strongly associated with neurocognitive outcomes, but that higher CPE ART regimens may be associated with modestly improved neurocognitive outcomes [26]. The use of IQs in our metric of CSF ART regimen activity builds upon, for example, earlier studies that investigated IQs of individual ARVs in CSF and in lymph nodes as a measure of activity [27, 28]. In a study of CSF IQs, paired plasma and CSF samples were obtained from 127 participants on ART. Univariate analysis found that regimens with CSF IQs >1 for the third drug, along with detectable CSF concentrations for all drugs, were associated with a higher likelihood of having CSF HIV RNA <50 copies/mL (odds ratio = 2.78, P<0.01) [27]. The largest and most recent ART intensification study for neurocognitive impairment compared the addition of dolutegravir (DTG) plus maraviroc (n=61), DTG plus placebo (n=67), and dual placebo groups (n=63) in 191 PWH on suppressive ART [29]. This study found no difference in neurocognitive impairment among the groups and concluded that ART intensification is not recommended as a treatment for neurocognitive impairment in PWH on suppressive ART. The authors noted potential reasons for their negative findings, including the specific drugs chosen for the trial and insufficient power. Additionally, concentrations of the ART drugs were not measured in CSF, and thus the degree to which antiviral activity in the CSF was actually intensified is unknown. In contrast, the CSF ART-IQGM metric provides an approach to address gaps in the CPE and studies of neurocognitive impairment by using measured CSF concentrations of all ARVs in a person’s regimen and referencing those concentrations to how much drug is needed virologically (the inhibitory quotient). CSF IQs were higher for drugs with the largest CPE values, as anticipated, but the IQs showed substantial inter-individual variability, a factor not reflected in the CPE approach.
Most of our participants were taking EFV (n=17). EFV exhibits low CSF penetration as a fraction of plasma concentrations (~0.9%) but has a high CSF IQ (median IQ=8.8, Figure 1) when CSF concentrations are compared with in vitro inhibitory concentrations. EFV also exhibits well-described neurotoxicity [30–33]. In our study, the estimated AUC of EFV in CSF was negatively correlated with the global z-score (r = −0.52, p = 0.048), indicating increased concentrations of EFV in CSF were associated with neurocognitive impairment. [34]. In the clinical setting, studies have shown conflicting results regarding EFV concentrations and neurocognitive performance in PWH [21, 35]. Marzolini et al. showed that high plasma levels of EFV (>4000 μg/L) resulted in three times more frequent CNS toxicity (CNS adverse effects, 9% vs. 24%, p=0.093) than lower plasma levels (1000–4000 μg/L) of EFV [21]. However, a study by Nwogu et al. found that higher EFV concentrations were associated with better neurocognitive performance (r=0.23, p=0.043) based on neuropsychological examination [36]. We have shown in older PWH that better neurocognitive function was associated with higher plasma levels of the EFV metabolite, 8-OH [33]. While EFV has excellent potency and durability in treatment-naïve persons, it is no longer considered a preferred ARV, largely replaced by integrase strand transfer inhibitors, due in part to fewer adverse events.
In this study, there was no association between the CSF ART-IQGM score and six different CSF inflammatory biomarkers (IL-6, IP-10, MCP-1, sCD14, sCD163, neopterin). This finding is consistent with those of Hellmuth et al., who reported that following 24 and 96 weeks of ART, levels of CSF markers in PWH (i.e., neopterin, C-X-C motif chemokine ligand 10 [CXCL10], chemokine ligand 2 [CCL2], and IL-6) normalized while plasma markers remained elevated [35].
They hypothesized that early initiation of ART may have a neuroprotective trajectory and that plasma markers may be affected by multifactorial effects (i.e., potential reservoirs, reservoir sizes, host-specific immunological factors) [37, 38]. Conversely, a study by Ulhammer et al. of 20 neuroasymptomatic PWH on suppressive ART for a median of 13.2 years found that even though CSF neopterin decreased (median 17.8 to 6.1 nmol/L) during treatment (P<0.001), 11 of 20 patients still had CSF neopterin levels above the upper normal reference limit of 5.8 nmol/L [37]. Clearly, further examination of these and other CSF inflammatory biomarkers that may be more consistent with long-term ongoing inflammation and ART regimen plasma and CSF exposure is needed.
The pharmacometric method we developed to quantify ART regimen activity in the CSF advances the study of CSF drug and regimen exposure-response. The notable features of this tool are the use of measured CSF concentrations, which incorporate information on interpatient variability in CSF drug concentrations, and the construction of a CSF exposure metric that includes all drugs the patient is receiving and how much drug is needed for virologic activity. We note some caveats and acknowledge some limitations of this work. We found significantly lower levels of CSF CA-DNA in those with higher vs. lower CSF ART-IQGM scores. However, this CSF CA-DNA persistence does not necessarily equate to ongoing viral replication given most integrated proviruses are replication incompetent [39]. Regarding limitations, only one sample for plasma and CSF was analyzed from each participant. Potential procedural variability such as lumbar puncture technique, sample volume, and presence of red blood cells could influence measured CSF drug concentrations. However, all samples were collected by investigators experienced in lumbar puncture and processed under standardized ACTG protocols. Second, adherence was self-reported, and unreported missed doses may have resulted in low plasma and CSF concentrations, and this could not be accounted for. However, all participants had well-documented sustained plasma HIV-1 viral suppression. Plasma concentrations of TFV and FTC were quantified in 43 (98%) of 44 participants, CSF concentrations of TFV and FTC in 42 (95%) and 44 (100%) respectively, and these concentrations as well as the anchor ARVs in plasma and CSF were consistent with literature expectations [11]. PBMCs, which were not collected at the time of the CSF samples, were available in 16 (36%) of 44 participants. We quantified intracellular tenofovir-diphosphate (TFV-DP) in all 16; the median concentration was 48 fmol/106 cells with an interquartile range of 30 to 81 fmol/106 cells, consistent with previously reported values [40]. Collectively, these virologic and pharmacokinetic data provide evidence of consistent drug-taking behavior. Third, the analysis focused on the relationship between the CSF ART-IQGM metric and measures of HIV persistence and inflammation and did not investigate other factors or confounders. Although CSF drug concentrations are often used as a surrogate for CNS exposure, several studies have shown discordance between CSF and brain parenchymal concentrations. Direct measurement of brain tissue levels is rarely feasible in living individuals; thus, CSF serves as a practical though imperfect proxy. This limitation parallels other infectious disease settings (e.g., TB meningitis) where CSF measurements inform CNS penetration. Our findings, therefore, should be interpreted as reflecting ART activity in the CSF compartment, which may not fully capture drug exposure at sites of viral persistence within brain tissue. Fourth, we did not perform either genotypic or phenotypic resistance testing. Participant-specific IC50 values from phenotypic susceptibility testing might enhance the precision of the ART-IQ metric and should be considered in future studies with this method. Fourth, the total number of participants on each anchor ARV varied. This resulted in small sample sizes for some of the ART regimens. Related, the most common anchor ARV was efavirenz, reflecting ART use patterns during A5321 enrollment. Further evaluations of the CSF ART-IQGM method with contemporary ARV regimens are needed. Finally, this approach is based on measured CSF concentrations, which are not routinely available.
In summary, these data demonstrate the ability to use person-specific drug exposure data to obtain a standardized measure of ART regimen activity in the CSF and assess virologic and cognitive outcomes. We found this ART regimen metric was associated with a lack of detection of CSF HIV DNA and better GDS, which suggests ART regimen activity affects HIV persistence in CSF. This tool provides a basis for further investigations of relationships between regimen activity and biomarkers of HIV persistence in the CSF, and for similar investigations in other anatomic reservoirs for HIV.
Acknowledgments:
We thank the Antiviral Pharmacology Laboratory for quantitation of plasma and CSF concentrations, John W. Mellors, M.D. for his support of the work, the ACTG sites that conducted the study, and the participants who contributed blood and CSF samples.
Funding:
This work was supported by awards UM1 AI068634, UM1 AI068636 and UM1 AI106701 from the National Institute of Allergy and Infectious Diseases and grant 1R01 AI-124965 from the National Institute of Allergy and Infectious Diseases to CVF, grant K23MH125734 to SNA from the National Institute of Mental Health, and 5P30MH062261-20 from the National Institute of Mental Health to SNA. RTG receives grant funding from the Harvard University Center for AIDS Research (NIH P30 AI060354) and the AIDS Clinical Trials Group (NIH/UMAI069412).
Footnotes
Conflict of Interest
The authors have no conflicts of interest to disclose.
References
- 1.Fletcher CV, Dyavar SR, Acharya A, Byrareddy SN. The Contributions of Clinical Pharmacology to HIV Cure Research. Clin Pharmacol Ther 2021; 110(2):334–345. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 2.Whitney JB, Hill AL, Sanisetty S, Penaloza-MacMaster P, Liu J, Shetty M, et al. Rapid seeding of the viral reservoir prior to SIV viraemia in rhesus monkeys. Nature 2014; 512(7512):74–77. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 3.Estes JD, Kityo C, Ssali F, Swainson L, Makamdop KN, Del Prete GQ, et al. Defining total-body AIDS-virus burden with implications for curative strategies. Nat Med 2017; 23(11):1271–1276. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 4.Fletcher CV, Staskus K, Wietgrefe SW, Rothenberger M, Reilly C, Chipman JG, et al. Persistent HIV-1 replication is associated with lower antiretroviral drug concentrations in lymphatic tissues. Proc Natl Acad Sci U S A 2014; 111(6):2307–2312. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 5.Letendre S. Central nervous system complications in HIV disease: HIV-associated neurocognitive disorder. Top Antivir Med 2011; 19(4):137–142. [PMC free article] [PubMed] [Google Scholar]
- 6.Gandhi RT, Cyktor JC, Bosch RJ, Mar H, Laird GM, Martin A, et al. Selective Decay of Intact HIV-1 Proviral DNA on Antiretroviral Therapy. J Infect Dis 2021; 223(2):225–233. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 7.Gandhi RT, McMahon DK, Bosch RJ, Lalama CM, Cyktor JC, Macatangay BJ, et al. Levels of HIV-1 persistence on antiretroviral therapy are not associated with markers of inflammation or activation. PLoS Pathog 2017; 13(4):e1006285. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 8.Spudich S, Robertson KR, Bosch RJ, Gandhi RT, Cyktor JC, Mar H, et al. Persistent HIV-infected cells in cerebrospinal fluid are associated with poorer neurocognitive performance. J Clin Invest 2019; 129(8):3339–3346. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 9.Carey CL, Woods SP, Gonzalez R, Conover E, Marcotte TD, Grant I, et al. Predictive validity of global deficit scores in detecting neuropsychological impairment in HIV infection. J Clin Exp Neuropsychol 2004; 26(3):307–319. [DOI] [PubMed] [Google Scholar]
- 10.Mykris TM, Weinhold J, Winchester LC, Scarsi KK, Fletcher CV, Podany AT, et al. Quantification of nine antiretroviral drugs in cerebrospinal fluid: An approach to overcome sample collection tube adsorption. J Chromatogr B Analyt Technol Biomed Life Sci 2023; 1227:123810. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 11.Avedissian SN, Mu Y, McCarthy C, Bosch RJ, Spudich S, Gandhi RT, et al. Pharmacokinetic approaches to standardize antiviral exposure in cerebrospinal fluid. Pharmacotherapy 2025; 45(5):251–263. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 12.Avedissian SN, Dyavar SR, Fox HS, Fletcher CV. Pharmacologic approaches to HIV-associated neurocognitive disorders. Curr Opin Pharmacol 2020; 54:102–108. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 13.Emtriva (emtricitabine) [package insert]. In. Foster City (CA): Gilead Sciences, Inc.; 2018. [Google Scholar]
- 14.Viread (tenofovir disoproxil fumarate) [package insert]. In. Foster City (CA): Gilead Sciences, Inc; 2019. [Google Scholar]
- 15.Best BM, Koopmans PP, Letendre SL, Capparelli EV, Rossi SS, Clifford DB, et al. Efavirenz concentrations in CSF exceed IC50 for wild-type HIV. J Antimicrob Chemother 2011; 66(2):354–357. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 16.Sculier D, Doco-Lecompte T, Yerly S, Metzner KJ, Decosterd LA, Calmy A. Stable HIV-1 reservoirs on dolutegravir maintenance monotherapy: the MONODO study. HIV Med 2018; 19(8):572–577. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 17.Yilmaz A, Gisslen M, Spudich S, Lee E, Jayewardene A, Aweeka F, et al. Raltegravir cerebrospinal fluid concentrations in HIV-1 infection. PLoS One 2009; 4(9):e6877. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 18.Custodio JM, Rhee M, Shen G, Ling KH, Kearney BP, Ramanathan S. Pharmacokinetics and safety of boosted elvitegravir in subjects with hepatic impairment. Antimicrob Agents Chemother 2014; 58(5):2564–2569. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 19.Podany AT, Scarsi KK, Pham MM, Fletcher CV. Comparative Clinical Pharmacokinetics and Pharmacodynamics of HIV-1 Integrase Strand Transfer Inhibitors: An Updated Review. Clin Pharmacokinet 2020; 59(9):1085–1107. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 20.Calcagno A, Di Perri G, Bonora S. Pharmacokinetics and pharmacodynamics of antiretrovirals in the central nervous system. Clin Pharmacokinet 2014; 53(10):891–906. [DOI] [PubMed] [Google Scholar]
- 21.Marzolini C, Telenti A, Decosterd LA, Greub G, Biollaz J, Buclin T. Efavirenz plasma levels can predict treatment failure and central nervous system side effects in HIV-1-infected patients. AIDS 2001; 15(1):71–75. [DOI] [PubMed] [Google Scholar]
- 22.Caniglia EC, Cain LE, Justice A, Tate J, Logan R, Sabin C, et al. Antiretroviral penetration into the CNS and incidence of AIDS-defining neurologic conditions. Neurology 2014; 83(2):134–141. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 23.Carvalhal A, Gill MJ, Letendre SL, Rachlis A, Bekele T, Raboud J, et al. Central nervous system penetration effectiveness of antiretroviral drugs and neuropsychological impairment in the Ontario HIV Treatment Network Cohort Study. J Neurovirol 2016; 22(3):349–357. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 24.Casado JL, Marin A, Moreno A, Iglesias V, Perez-Elias MJ, Moreno S, et al. Central nervous system antiretroviral penetration and cognitive functioning in largely pretreated HIV-infected patients. J Neurovirol 2014; 20(1):54–61. [DOI] [PubMed] [Google Scholar]
- 25.Ellis RJ, Letendre S, Vaida F, Haubrich R, Heaton RK, Sacktor N, et al. Randomized trial of central nervous system-targeted antiretrovirals for HIV-associated neurocognitive disorder. Clin Infect Dis 2014; 58(7):1015–1022. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 26.Arentoft A, Troxell K, Alvarez K, Aghvinian M, Rivera Mindt M, Cherner M, et al. HIV Antiretroviral Medication Neuropenetrance and Neurocognitive Outcomes in HIV+ Adults: A Review of the Literature Examining the Central Nervous System Penetration Effectiveness Score. Viruses 2022; 14(6). [DOI] [PMC free article] [PubMed] [Google Scholar]
- 27.Calcagno A, Simiele M, Alberione MC, Bracchi M, Marinaro L, Ecclesia S, et al. Cerebrospinal fluid inhibitory quotients of antiretroviral drugs in HIV-infected patients are associated with compartmental viral control. Clin Infect Dis 2015; 60(2):311–317. [DOI] [PubMed] [Google Scholar]
- 28.Fletcher CV, Kroon E, Schacker T, Pinyakorn S, Chomont N, Chottanapund S, et al. Persistent HIV transcription and variable antiretroviral drug penetration in lymph nodes during plasma viral suppression. AIDS 2022; 36(7):985–990. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 29.Letendre SL, Chen H, McKhann A, Roa J, Vecchio A, Daar ES, et al. Antiretroviral Therapy Intensification for Neurocognitive Impairment in Human Immunodeficiency Virus. Clin Infect Dis 2023; 77(6):866–874. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 30.Asundi A, Robles Y, Starr T, Landay A, Kinslow J, Ladner J, et al. Immunological and Neurometabolite Changes Associated With Switch From Efavirenz to an Integrase Inhibitor. J Acquir Immune Defic Syndr 2019; 81(5):585–593. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 31.Decloedt EH, Maartens G. Neuronal toxicity of efavirenz: a systematic review. Expert Opin Drug Saf 2013; 12(6):841–846. [DOI] [PubMed] [Google Scholar]
- 32.Ranzani A, Castelli F, Di Biagio A, d’Arminio Monforte A, D’Avolio A, Soria A, et al. Influence of efavirenz and 8-hydroxy-efavirenz plasma levels on cognition and central nervous system side effects. HIV Med 2024; 25(4):491–497. [DOI] [PubMed] [Google Scholar]
- 33.Sandkovsky U, Podany AT, Fletcher CV, Owen A, Felton-Coleman A, Winchester LC, et al. Impact of efavirenz pharmacokinetics and pharmacogenomics on neuropsychological performance in older HIV-infected patients. J Antimicrob Chemother 2017; 72(1):200–204. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 34.Tovar-y-Romo LB, Bumpus NN, Pomerantz D, Avery LB, Sacktor N, McArthur JC, et al. Dendritic spine injury induced by the 8-hydroxy metabolite of efavirenz. J Pharmacol Exp Ther 2012; 343(3):696–703. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 35.Hellmuth J, Slike BM, Sacdalan C, Best J, Kroon E, Phanuphak N, et al. Very Early Initiation of Antiretroviral Therapy During Acute HIV Infection Is Associated With Normalized Levels of Immune Activation Markers in Cerebrospinal Fluid but Not in Plasma. J Infect Dis 2019; 220(12):1885–1891. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 36.Nwogu JN, Gandhi M, Owen A, Khoo SH, Taiwo B, Olagunju A, et al. Associations between efavirenz concentrations, pharmacogenetics and neurocognitive performance in people living with HIV in Nigeria. AIDS 2021; 35(12):1919–1927. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 37.Ulfhammer G, Eden A, Mellgren A, Fuchs D, Zetterberg H, Hagberg L, et al. Persistent central nervous system immune activation following more than 10 years of effective HIV antiretroviral treatment. AIDS 2018; 32(15):2171–2178. [DOI] [PubMed] [Google Scholar]
- 38.Yilmaz A, Yiannoutsos CT, Fuchs D, Price RW, Crozier K, Hagberg L, et al. Cerebrospinal fluid neopterin decay characteristics after initiation of antiretroviral therapy. J Neuroinflammation 2013; 10:62. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 39.Sannier G, Dube M, Dufour C, Richard C, Brassard N, Delgado GG, et al. Combined single-cell transcriptional, translational, and genomic profiling reveals HIV-1 reservoir diversity. Cell Rep 2021; 36(9):109643. [DOI] [PubMed] [Google Scholar]
- 40.Fletcher CV, Podany AT, Thorkelson A, Winchester LC, Mykris T, Anderson J, et al. The Lymphoid Tissue Pharmacokinetics of Tenofovir Disoproxil Fumarate and Tenofovir Alafenamide in HIV-Infected Persons. Clin Pharmacol Ther 2020; 108(5):971–975. [DOI] [PMC free article] [PubMed] [Google Scholar]
Associated Data
This section collects any data citations, data availability statements, or supplementary materials included in this article.
Data Availability Statement
All data underlying the findings are fully available without restriction. The study data are available upon request from the Statistical and Data Management Center (SDAC) of the ACTG; the request may be made by emailing: sdac.data@sdac.harvard.edu.