Factors Associated With Persistence of Plasma HIV-1 RNA During Long-term Continuously Suppressive Firstline Antiretroviral Therapy

Abstract Background Persistence of plasma HIV-1 RNA during seemingly effective antiretroviral thereapy (ART) is incompletely understood. Using an ultrasensitive assay, this cross-sectional study investigated residual plasma HIV-1 RNA in subjects maintained on firstline ART with continuous viral load suppression <50 copies/mL for ≤15 years without recognized viral load blips or treatment interruptions and explored its relationship with the duration of suppressive ART, efavirenz concentrations in plasma, 2-LTR circular HIV-1 DNA (2-LTRc DNA) in peripheral blood mononuclear cells, and cellular (CD4 plus CD26/CD38/CD69; CD8 plus CD38/HLA-DR/DP/DQ) and soluble (sCD14, sCD27, sCD30, IL-6) markers of immune activation in peripheral blood. Methods Residual plasma HIV-1 RNA, total HIV-1 DNA and 2-LTRc DNA were quantified by real-time and digital droplet PCR. Cellular (CD4 plus CD26/CD38/CD69; CD8 plus CD38/HLA-DR/DP/DQ) and soluble (sCD14, sCD27, sCD30, IL-6) markers of immune activation were measured by flow cytometry and ELISA. Results Residual plasma HIV-1 RNA and 2-LTRc DNA were detected in 52/104 (50%) and 24/104 (23%) subjects, respectively. Among subjects with detectable HIV-1 RNA, 50/52 showed levels ≤11 copies/mL. In adjusted analyses, HIV-1 RNA levels were 0.37 log10 copies/mL higher with each log10 U/mL increase in sCD27 (95% confidence interval, 0.01–0.73; P = .02). No significant association was found between residual plasma HIV-1 RNA and other explored parameters. Conclusions These findings point to an ongoing relationship between plasma HIV-1 RNA and selected markers of immune activation during continuously suppressive ART. The novel direct association with levels of sCD27 warrants further investigation.

Maintaining plasma HIV-1 RNA suppression below the limit of detection of commercial viral load assays defines clinical success of antiretroviral therapy (ART) [1,2]. Ultrasensitive assays can detect traces of HIV-1 RNA in the plasma of many successfully treated patients [3], but the source remains controversial [4,5]. On the one hand, it is proposed that detection of residual plasma HIV-1 RNA reflects short-lived bursts of virus production from latently infected memory CD4 T cells, but with ongoing virus replication and genetic evolution inhibited by ART [6]. An alternative model derives from evidence of ongoing virus replication in compartments where drug penetration or activity is suboptimal [7][8][9][10]. While the 2 models are not mutually exclusive, published studies have often included patients with heterogeneous treatment histories and varying definitions of suppression, which may in part explain conflicting findings [11][12][13].
Identifying the determinants of HIV-1 RNA persistence during seemingly effective ART is central to our understanding of how the virus may be eradicated [14,15]. Such studies must also consider the interplay between virus replication and host immunity, and the role of chronic immune activation and inflammation in disease pathogenesis [16][17][18]. Effective ART reduces but does not always normalize inflammatory markers, raising the possibility that residual immune activation may both sustain and be sustained by residual virus replication [18]. Available data are not conclusive, however, and they rather point to virologic and immunologic events that take place before the initiation of ART as key determinants of the HIV-1 reservoirs and immune activation that can persist despite ART [19].
We previously reported that treated patients with plasma HIV-1 RNA detected at levels ranging between approximately 10 and 49 copies/mL by a commercial viral load assay were more likely to experience viral load rebound above 50 copies/mL and above 400 copies/mL during 12 months of follow-up when compared with treated patients with undetectable HIV-1 RNA [20]. The aim of this study was to take the converse approach and assess a population that experienced continuous viral load suppression <50 copies/mL for up to 15 years of firstline ART and use ultrasensitive testing to detect HIV-1 RNA below the 10-copies/mL cutoff. The cohort was selected according to precise inclusion criteria, and residual plasma HIV-1 RNA was investigated cross-sectionally in relation to the duration of suppressive ART, plasma concentrations of efavirenz, 2-LTRc DNA in peripheral blood mononuclear cells (PBMCs), and cellular and soluble markers of immune activation in peripheral blood.

Study Population
In this cross-sectional study, eligible patients had started firstline ART with 2 nucleoside/nucleotide reverse transcriptase inhibitors (NRTIs) plus 1 non-nucleoside reverse transcriptase inhibitor (NNRTI), had achieved viral load suppression <50 copies/ mL within 6 months of starting ART, and during subsequent follow-up had remained on the initial NNRTI, undergone ≥2 viral load measurements per year, and maintained continuous viral load suppression <50 copies/mL without any recorded viral load elevation >50 copies/mL or treatment interruption. To ensure balance in terms of duration of suppressive ART, recruitment was stratified to span from 1 year of treatment to 10 years or longer. Changes of the initial NNRTI were not allowed. Changes of the initial NRTIs were allowed to reflect evolving practice, provided they did not coincide with a viral load >50 copies/mL or a treatment interruption. The study took place at multiple centers across the United Kingdom. It was approved by the National Research Ethics Service (London-Dulwich) and included in the National Institute for Health Research Clinical Research Network Portfolio. Patients provided written informed consent.

Residual Plasma HIV-1 RNA
Plasma was separated from venous blood in EDTA within 2 hours of collection and stored at −80°. HIV-1 RNA was quantified using a modified version of the RealTime HIV-1 assay (Abbott, Maidenhead, UK). Validation studies previously showed that the assay 50% and 95% detection rates were 1 and 3 HIV-1 RNA copies/mL, respectively, with high reproducibility across diverse HIV-1 subtypes [10,20,21].

Cellular and Soluble Markers of Immune Activation
Flow cytometry was performed at the Immunology Service of the Royal Free Hospital in London using freshly collected blood.

Plasma Drug Concentrations
Efavirenz (EFV) and nevirapine (NVP) concentrations were measured in untimed plasma samples using high-performance liquid chromatography-tandem mass spectrometry, as previously described [10]. The time of sample collection relative to last dosing was recorded.

Analysis
Duration of suppressive ART was defined as the length of time following the first viral load measurement <50 copies/mL, which in all subjects was recorded within 6 months of starting ART. When analyzing residual plasma HIV-1 RNA and 2-LTRc DNA as continuous variables, undetectable values were replaced with the midpoint value between 0 and the lower limit of detection of the assays. EFV and NVP concentrations were categorized as above or below the recommended minimal effective concentration (MEC) for wild-type virus: 1000 ng/mL for EFV [24] and 3400 ng/mL for NVP [25]. Participants' characteristics were compared by analysis of variance (ANOVA), the Kruskal-Wallis test, and the Fisher exact test for continuous and categorical variables, as appropriate. Data distribution was assessed through descriptive scatter plots. The correlation between EFV concentrations and levels of residual plasma HIV-1 RNA and 2-LTRc DNA was measured by Spearman's rank test. Given that all but 2 HIV-1 RNA values were ≤11 copies/mL, in order to reduce heterogeneity, the main statistical models excluded 2 subjects with residual plasma HIV-1 RNA >11 copies/mL (referred to as outliers). This was based on our previous data indicating that patients on stable ART with a viral load above approximately 10 copies/mL are a distinct population at risk of viral load rebound [20]. Three main models were run. The mean difference (with 95% confidence interval [CI]) in residual plasma HIV-1 RNA, total HIV-1 DNA, 2-LTRc DNA, and markers of immune activation according to duration of suppressive ART (normalized over 1 year) was analyzed by univariable linear regression analysis after log-transformation of the variables. Factors associated with levels of residual plasma HIV-1 RNA or 2-LTRc DNA were investigated by univariable and multivariable linear regression analysis. Variables associated (P ≤ .20) with the outcome of interest in the univariable analysis were included in the multivariable models, with the following modifications: (a) when modeling 2-LTRc DNA, total HIV-1 DNA was excluded due to high colinearity (P < .0001); (b) age and duration of ART were excluded in favor of duration of suppressive ART. Separate models explored: (a) the effect of forcing the nadir CD4 cell count and pre-ART viral load into the multivariable analyses and (b) the association between residual plasma HIV-1 RNA or 2-LTRc DNA and plasma concentrations of EFV (as a continuous variable) in EFV recipients, after adjustment for the time of sampling in relation to the last EFV dose. IBM SPSS v. 22.0 (Armonk, NY, USA) and SAS v9.4 (Cary, NC, USA) were used for the analyses.

Study Population
The population comprised 104 patients that at the time of recruitment had experienced continuous viral load suppression <50 copies/mL for a median (IQR) of 5 (3-8) years while receiving firstline ART with 2 NRTIs plus either EFV (n = 86, 83%) or NVP (Table 1). Reflecting evolving guidelines about when to start ART and which antiretrovirals to use, patients with longer duration of suppressive ART had a lower nadir CD4 cell count and a higher pre-ART viral load, were more often receiving NVP, and were more likely to have experienced changes of the NRTI backbone after first starting ART: 48/104 (48%) patients changed ≥1 NRTIs prior to recruitment, with a median (IQR) of 1 (1-2) NRTI change per subject. At recruitment, NRTI backbones comprised tenofovir disoproxil fumarate (78/104, 75%), abacavir (22/104, 21%) or zidovudine (4/104, 4%), plus emtricitabine or lamivudine.

Virologic and Immunologic Parameters in Relation to the Duration of Suppressive ART
Differences in virologic and immunologic parameters according to the duration of suppressive ART, analyzed after exclusion of the 2 outliers, are shown in Table 2. Patients with longer duration of suppressive ART had higher CD4 counts (+19 cells/mm 3 /y) and lower expression of activation markers on CD4 cells (CD38) and CD8 cells (CD38, HLA-DR/ DP/DQ). Levels of sCD14, sCD30, and IL-6 did not differ by duration of suppressive ART, whereas levels of sCD27 were significantly lower with longer duration of suppressive ART. A trend was observed for lower levels of 2-LTRc DNA with longer duration of suppressive ART. There was less evidence of a difference in residual plasma HIV-1 RNA, and no significant difference was seen in total HIV-1 DNA. A linear regression analysis confirmed an inverse association between 2-LTRc DNA and duration of suppressive ART, with an adjusted mean difference of 0.03 log 10 copies/10 6 PBMC per year longer (95% CI, -0.05 to -0.00; P = .02) (Supplementary Table 2). Forcing the nadir CD4 cell count and pre-ART viral load into the multivariable model did not change the findings: 2-LTRc DNA levels were 0.03 log 10 copies/10 6 PBMC lower per year of suppressive ART (95% CI, -0.05 to -0.00; P = .02; not shown).

Factors Associated With Residual Plasma HIV-1 RNA
The univariable linear regression analysis indicated a significant direct association between residual plasma HIV-1 RNA and sCD27 among the 102 subjects with HIV-1 RNA ≤11 copies/mL (Table 3). There was a weaker direct association with sCD30, whereas no association was evident with other markers of immune activation. There was also no association between residual plasma HIV-1 RNA and total HIV-1 DNA and 2-LTRc DNA. The direct association between residual plasma HIV-1 RNA and sCD27 persisted after adjusting for the duration of suppressive ART, frequency of CD8 + CD38 + cells, and levels of sCD30 and IL-6 ( Table 3). There was no apparent association between residual plasma HIV-1 RNA and nadir CD4 cell count and pre-ART viral load in the univariable analysis. Forcing the nadir CD4 cell count and pre-ART viral load into the multivariable model did not change the findings: residual plasma HIV-1 RNA was 0.44 log 10 copies/mL higher per each log 10 increase in sCD27 (95% CI, 0.09-0.79; P = .02; not shown). The association similarly persisted in analyses including the 2 outliers (not shown).

Relationship Between Virologic Parameters and Plasma Drug Concentrations
We first explored the association between plasma EFV concentrations and virologic parameters in 76 EFV recipients (including the 2 outliers) who had taken the last EFV dose a median (IQR) of 12 (10-13) hours prior to sampling, and were therefore in the range of the "mid-dose" interval, where targets have been previously defined [30]. Although EFV concentrations  varied across patients, there was no clear correlation with levels of residual plasma HIV-1 RNA and 2-LTRc DNA (Spearman's rank test) (Figure 2). Two separate linear regression analyses explored the association between EFV concentrations and residual plasma HIV-1 RNA and 2-LTRc DNA among the 74 patients with residual plasma HIV-1 RNA ≤11 copies/mL. In models that included both EFV concentration and timing of sampling relative to last EFV dose, there was no evidence of an association between plasma HIV-1 RNA (P = .30) or 2-LTRc DNA (P = .47) and EFV concentrations (Supplementary Table 3).

DISCUSSION
This study demonstrated that half of subjects who had experienced viral load suppression <50 copies/mL continuously for up to 15 years while maintained on the initial NNRTI had detectable HIV-1 RNA in plasma when tested with an ultrasensitive assay, at levels that rarely exceeded 11 copies/mL. We previously reported that in treated patients, plasma HIV-1 RNA levels above approximately 10 copies/mL were associated with a risk of viral load rebound above 50 copies/mL and above 400 copies/mL over 12 months of follow-up [20]. Taken together, our findings suggest a cutoff level for plasma HIV-1 RNA below which a risk of viral load rebound is not observed. However, an independent, direct association persisted between residual plasma HIV-1 RNA and the immune activation marker sCD27 in this stably treated population. These data do not elucidate the debated source of residual plasma HIV-1 RNA [15,31]. The lack of a correlation with levels of 2-LTRc DNA, drug concentrations, or multiple markers of immune activation may indicate that ongoing virus replication was an unlikely source, as far as it can be extrapolated from measurements made in peripheral blood. In this context, the findings may provide support to the notion that during effective treatment, detection of a few HIV-1 RNA copies in plasma reflects transient bursts of virus production from latent reservoirs that are inhibited from ongoing replication by ART [8,32]. In this model, a direct correlation may be anticipated between the likelihood of HIV-1 RNA persistence and the size of the viral reservoir. We did not detect an association between residual HIV-1 RNA in plasma and total HIV-1 DNA in PBMC, and we similarly did not previously observe an association with integrated HIV-1 DNA in PBMC [21]. Findings from other studies are conflicting, possibly reflecting differences in the composition of the study populations (eg, proportion of patients on protease inhibitors) and the method used for measuring HIV-1 RNA [33]. We also measured 2-LTRc DNA in PBMC as a possible indicator of viral gene expression and virus replication and found no correlation with residual plasma HIV-1 RNA. In a previous study of 30 subjects who had received various ART regimens for 7-12 years, 2-LTRc DNA was detected in 8 subjects (27%) [33], which is highly comparable to our detection rate of 23%. In our study, 2-LTRc DNA levels were lower, with longer duration of suppressive ART, with a reduction of 0.02 log 10 2-LTRc DNA copies/10 6 PBMC per year. This apparent, slow decay in 2-LTRc DNA is consistent with observations from Bushman [34] and Siliciano [35], and potentially in line with the hypothesis that episomal HIV-1 DNA survives for the life of the cell, rather than being a short-lived product of ongoing virus replication [36].
Previous studies did not detect an association between residual plasma HIV-1 RNA and markers of immune activation on CD4 or CD8 cells (including CD38 and HLA-DR) in patients treated for up to 4 years [14,37], and also detected no association with levels of sCD14 and IL-6 [19]. Our data confirm these observations over the longer term. However, we found that levels of sCD27 were independently associated with levels of residual plasma HIV-1 RNA. sCD27 is a 32-kD protein identical to the extracellular domain of CD27, a transmembrane protein and TNF receptor that is expressed on subsets of T, B, NK, and hematopoietic progenitor cells. sCD27 is released primarily from CD4 T cells and provides a costimulatory function that promotes activation and proliferation of T and B lymphocytes [38][39][40][41]. sCD27 can be detected in healthy donors, and increased plasma levels have been observed in conditions characterized by immune activation, especially auto-immune disorders [42][43][44]. There are limited data in the setting of HIV infection. Levels of sCD27 have been shown to be generally higher than in HIV-negative controls and found to be predictive of the risk of HIV-related renal dysfunction and lymphomas [42,45]. Levels have been shown to decline significantly in the first 12 to 24 months of ART and to increase again with viral load rebound or upon treatment discontinuation [29,46,47]. In our study, consistent with the reported effect of treatment, levels of sCD27 were overall low and reduced further with longer duration of suppressive ART. Taken in the context of these previous observations, our findings illustrate a close association between sCD27 and potential indicators of HIV replication during suppressive ART that warrants further investigation. There are scarce data on the effect of ART on sCD30 [26]. Levels in our population were considerably lower than those measured in the first 4 weeks of ART [48], and closer to levels published from HIV-negative volunteers [28]; however, sCD30 showed no independent association with virologic parameters. Levels of sCD14, IL-6, and cellular markers of immune activation in our study population were comparable to those previously measured in HIV-1-positive patients receiving ART for up to 6 years [19,49]. Consistent with previous observations (reviewed in [16]), markers of immune activation on CD4 and CD8 cells (CD38 and HLA-DR/DP/DQ) were lower with longer duration of suppressive ART, and this apparent decline was paralleled by a predictable gain in CD4 cell counts. Overall, as also reported by Gandhi et al. [19], sCD14, IL6, and cellular markers of immune activation showed no apparent association with levels of residual plasma HIV-1 RNA and 2-LTRc DNA.
Our study has limitations. First, findings obtained in a cross-sectional study should be confirmed with prospectively collected data, which is challenging when aiming to study longterm outcomes in populations maintained on the same regimen. To date, prospectively determined virologic indicators of HIV-1 persistence during long-term suppressive ART have been poorly documented. Compared with published studies addressing similar questions, one strength of our study is the relatively large population combined with stringent eligibility criteria and a consistent definition of virologic suppression. It could be argued that our study population was not selected based on the infection stage at the start of ART, which could have introduced a bias. We observed a lower nadir CD4 cell count and a higher pre-ART viral load in subjects with longer duration of suppressive ART; however, the analyses did not identify an effect of nadir CD4 cell count or pre-ART viral load on the measured parameters in our highly selected population. Of note, we excluded 2 outliers from the main analyses with residual plasma HIV-1 RNA levels >11 copies/mL. Whereas the main conclusions were not affected by their inclusion, it would have been of interest to determine any risk of viral load rebound with prospective follow-up in these 2 subjects. Further, we reported total HIV-1 DNA rather than integrated HIV-1 DNA as the indicator of the viral reservoir in this study. However, we previously described integrated HIV-1 DNA levels in a subset of the same population and demonstrated that they were highly correlated with the levels of total HIV-1 DNA, not influenced by the duration of suppressive ART, and not associated with residual HIV-1 RNA [21]. We collected large volumes of blood from patients in order to test for multiple parameters (eg, 8 mL of plasma was required for the ultrasensitive HIV-1 RNA assay alone). Limited sample volumes meant that not all potentially relevant immune parameters were measured. When selecting soluble markers of immune activation, we elected to include 2 markers that have been extensively explored in the published literature (sCD14 and IL-6) to provide a reference and added 2 markers for which there is high plausibility but scarce published evidence (sCD27 and sCD30). In this context, levels of soluble markers of immune activation were found to be similar to those historically reported in HIV-negative volunteers, which is consistent with the established beneficial effects of suppressive ART. However, published data from healthy volunteers were limited, especially for sCD27 and sCD30, and ideally, matched HIV-negative controls should be sampled to confirm reference values. As an additional point, we measured plasma NNRTI concentrations in untimed specimens and for practical reasons did not seek to obtain a measure of C trough . The patients analyzed in the models had taken the last EFV dose at a consistent interval relative to sampling, and time of sampling would be expected to have little impact overall. Finally, we recognize that both drug concentration and antiviral activity measured in peripheral blood do not necessarily reflect the situation in lymphoid tissue [8,9].
In conclusion, this study provides evidence that residual plasma HIV-1 RNA and PBMC-associated 2-LTRc DNA can be detected in a subset of patients receiving long-term, continuously suppressive firstline ART, without evidence of a clear association between the two, and without evidence of an association of either parameter with plasma concentrations of EFV. While residual plasma HIV-1 RNA was not associated with commonly reported markers of immune activation, there was a novel, independent direct association with levels of sCD27. Further studies are warranted to confirm the findings, extend the immune activation profiling, and determine the underlying mechanism, including whether the relationship reflects ongoing virus replication in lymphoid tissue.

Supplementary Data
Supplementary materials are available at Open Forum Infectious Diseases online. Consisting of data provided by the authors to benefit the reader, the posted materials are not copyedited and are the sole responsibility of the authors, so questions or comments should be addressed to the corresponding author.