Minimal Detection of Cerebrospinal Fluid Escape After Initiation of Antiretroviral Therapy in Acute HIV-1 Infection

Abstract

Objective:

Despite suppression of HIV-1 replication in the periphery by antiretroviral therapy (ART), up to 10% of treated individuals have quantifiable HIV-1 in the CSF, termed CSF escape. CSF escape may be asymptomatic but has also been linked to progressive neurological disease, and may indicate persistence of HIV in the central nervous system (CNS). CSF escape has not yet been assessed after initiation of ART during acute HIV-1 infection (AHI).

Design:

Prospective cohort study.

Setting:

Major voluntary counseling and testing site in Bangkok, Thailand.

Subjects:

Participants identified and initiated on ART during AHI who received an optional study lumbar puncture at pre-ART baseline or after 24 or 96 weeks of ART.

Main outcome measures:

Paired levels of CSF and plasma HIV-1 RNA, with CSF > plasma HIV-1 RNA defined as CSF escape.

Results:

204 participants had paired blood and CSF sampling in at least one visit at baseline, week 24, or week 96. 29 participants had CSF sampling at all three visits. CSF escape was detected in 1/90 at week 24 (CSF HIV-1 RNA 2.50 log₁₀ copies/mL, plasma HIV-1 RNA < 50 copies/mL), and 0/55 at week 96.

Conclusions:

While levels of CSF HIV-1 RNA in untreated AHI are high, initiating treatment during AHI results in a very low rate of CSF escape in the first two years of treatment. Early treatment may improve control of HIV-1 within the CNS compared to treatment during chronic infection, which may have implications for long-term neurological outcomes and CNS HIV-1 persistence.

BACKGROUND

The central nervous system (CNS) is an important site during early HIV-1 infection, chronic replication, and viral persistence, especially as a unique compartment sequestered by the blood-brain barrier that influences inflammatory response, immune cell trafficking, and antiretroviral drug penetration. Even while on suppressive antiretroviral therapy (ART), the CNS may continue to be affected uniquely in the phenomenon of cerebrospinal fluid (CSF) escape, in which viral presence is undetectable in the periphery but is detectable in CSF. During chronic HIV-1 infection, CSF escape may occur in up to 10% of individuals ^[1]. Recent international consensus guidelines have provided definitions ^[2].

CSF HIV escape may give some insight into interrogating long-term persistence of the CNS reservoir, which remains an important target for achieving ART-free remission ^[3]. Persistent CSF escape likely points to HIV-1 replication from CNS resident cells, i.e. a reservoir. However, some cases of CSF escape appear to be episodic ^[4], though this is difficult to know given the paucity of longitudinal CSF data ^[3]. Episodic CSF escape, also termed CSF HIV viral blips, could stem from CNS reservoirs but may also be explained by transient immune cell trafficking into the CNS that supports local viral replication ^[3].

In as early as eight days of AHI, CNS involvement can be detected by the presence of HIV-1 RNA in CSF, which is consistently lower than concurrent plasma HIV-1 RNA ^[5]. Given this and the proposed mechanism of immune cell trafficking that may introduce HIV-1 into the CNS, it is likely that CSF HIV-1 RNA has a delayed appearance behind that in plasma ^[6]. Additionally, markers of immune activation and inflammation, but not neuronal injury, appear early in the CNS in AHI ^{[5, 7]}. Prompt initiation of ART in AHI suppresses HIV-1 replication in the CNS and reverses neuroinflammation ^{[6, 8]}.

Early ART in AHI appears to mitigate neuroinflammation and neuronal injury but whether it reduces events of CSF escape, a potential window into CNS HIV persistence, remains unknown ^[6]. We investigated the frequency of CSF escape and identified clinical and laboratory factors associated with CSF escape in treated AHI.

METHODS

Study participants

Individuals with AHI identified at the Thai Red Cross AIDS Research Centre in Bangkok were enrolled in the ongoing RV254/SEARCH010 study (clinicaltrials.gov NCT00796146) ^[9], followed by immediate initiation of ART via a standardized protocol (clinicaltrials.gov NCT00796263). Standard first-line ART through 2016 included efavirenz plus two nucleoside reverse transcriptase inhibitors. Efavirenz could be replaced by ritonavir-boosted lopinavir or raltegravir for intolerance or resistance. A subset received a five-drug regimen that added raltegravir and maraviroc ^[10]. The majority were switched to dolutegravir starting in 2017 as the first-line regimen. Participants underwent serial interviews, examinations, and phlebotomy, with optional lumbar puncture (LP). All participants provided written informed consent prior to enrollment in the cohort. The research protocol was approved by institutional review boards at Chulalongkorn University Hospital, Yale School of Medicine, UCSF, and the Walter Reed Army Institute of Research.

This analysis included all participants who initiated ART between April 2009 and April 2019 with paired blood and CSF sampling in at least one visit at study enrollment (baseline), week 24, or week 96. CSF escape was defined as paired CSF HIV-1 RNA greater than plasma HIV-1 RNA at week 24 or week 96, as per recent international consensus definitions ^[2].

Sampling and laboratory testing

Clinical and laboratory parameters were assessed at baseline, week 24, and week 96. CD4+ T cell count was measured by single- and dual-platform flow cytometry (Becton-Dickinson). HIV-1 RNA in plasma was performed using the COBAS AMPLICOR HIV-1 Monitor Test v1.5 or COBAS TaqMan HIV-1 Test v2.0 (Roche Molecular Systems), with lower limits of quantification (LLQ) of 50 and 20 copies/mL, respectively. CSF samples were run in batch and diluted fourfold for volume requirements for detection of HIV-1 RNA, with an LLQ of 80 copies/mL using the TaqMan platform. HIV-1 RNA measurements below the LLQ that tested positive were designated detectable but not quantifiable. Neuropsychological testing was performed at standardized intervals at weeks 0, 12, 24, and 96 ^[11].

RESULTS

Study participant characteristics

Of 627 participants enrolled to date, 204 participants with AHI followed by prompt ART initiation had paired blood and CSF sampling during at least one visit at baseline, week 24, or week 96 (Table 1). Among these, 29 participants had CSF sampling at all three visits. 98% were Thai men, the majority men who have sex with men. Median age was 26 years (range 18-60). At enrollment, participants were mostly in Fiebig stage III, with median CD4+ T cell count 386 cells/mm³ (range 91-1302) and median plasma HIV-1 RNA 5.87 log₁₀ copies/mL (range 2.43-7.89). Study enrollment was at median 19 days post-estimated infection (range 3-49). The only differences in clinical and laboratory characteristics at enrollment between those who consented to LP versus those who declined were earlier Fiebig stage and higher CD4+ T cell count (Supplemental Table 1).

Table 1.

Characteristics of participants treated in acute HIV infection with available study lumbar puncture.

	All participants (n=204)	Baseline (n=165)	Week 24 (n=90)	Week 96 (n=55)
Age at enrollment (range)	26 (18-60)	26 (18-60)	27 (18-60)	28 (18-60)
Male, n (%)	199 (98)	160 (97)	86 (96)	53 (96)
Risk behavior, n (%)
WSM	5 (2)	5 (3)	4 (4)	2 (4)
MSW	7 (3)	6 (4)	1 (1)	2 (4)
MSM	192 (94)	154 (93)	85 (94)	51 (93)
Fiebig stage at enrollmenta, n (%)
Stage I	33 (16)	27 (16)	13 (14)	8 (15)
Stage II	47 (23)	37 (22)	20 (22)	11 (20)
Stage III	96 (47)	76 (46)	42 (47)	29 (53)
Stage IV	19 (9)	17 (10)	10 (11)	3 (5)
Stage V	8 (4)	7 (4)	4 (4)	3 (5)
Stage VI	1 (0)	1 (1)	1 (1)	1 (2)
Infection durationb (days), median (range)	19 (3-49)	18 (3-49)	19 (7-49)	19 (9-42)
CD4+ T-cells, cells/mm3 (range)	386 (91-1302) †	389 (101-1302)	613 (291-1464)	639 (320-1357)
CD8+ T-cells, cells/mm3 (range)	517 (81-4556) †	515 (102-4556)	575 (178-1352)	628 (260-1575)
Plasma HIV-1 RNA, log10 copies/mL, (range)	5.87 (2.43-7.89)†	5.83 (2.43-7.89)	<1.30 (<1.30-5.44)	<1.30 (<1.30-4.42)
Plasma HIV-1 RNA < 50 copies/mL, n (%)	0 (0) †	0 (0)	86 (96)	53 (96)
CSF HIV-1 RNA, log10 copies/mL, (range)	-	3.13 (<1.90-6.61)	<1.90 (<1.90-3.84)	<1.90 (<1.90-3.14)
CSF HIV-1 RNA < 80 copies/mL, n (%)	-	39 (24)	88 (98)	54 (98)
CSF viral escape, n (%)	-	-	1 (1)	0 (0)

Abbreviations: HIV-1, human immunodeficiency virus-1; WSM, women who have sex with men; MSW, men who have sex with women; MSM, men who have sex with men; CSF, cerebrospinal fluid.

^aStage I: RNA+, p24 antigen−; Stage II: p24 antigen+, IgM−; Stage III: IgM+, Western blot−; Stage IV: Western blot indeterminate; Stage V: Western blot+ without p31 protein band; Stage VI: Western blot+.

^bInterval between estimated time of exposure and enrollment. If a range of dates was provided, the mean was used.

^†At time of enrollment.

At baseline, 126 of 165 participants (76%) had quantifiable CSF HIV-1 RNA with median 3.13 log₁₀ copies/mL (range <1.90-6.61). There were no cases where CSF HIV-1 RNA exceeded the paired plasma HIV-1 RNA level. At week 24, of 90 participants with paired blood and CSF samples, four (4%) had detectable CSF HIV-1 RNA, of which two (2%) were quantifiable. Three of these four cases with detectable CSF HIV-1 RNA were associated with plasma virological failure. The remaining one met criteria for CSF escape, with plasma HIV-1 RNA < 50 copies/mL and CSF HIV-1 RNA at 2.50 log₁₀ copies/mL. At week 96, of 55 participants with paired blood and CSF samples, one (2%) had detectable and quantifiable CSF HIV-1 RNA, which was associated with plasma virological failure (Figure 1). This participant did not have quantifiable plasma or CSF HIV-1 RNA at week 24.

Figure 1. Paired blood and CSF HIV-1 RNA at weeks 0, 24, and 96.

Medians and interquartile ranges are indicated by black bars. Dashed or solid lines represent paired blood and CSF HIV-1 RNA from participants who had either blood or CSF HIV-1 RNA greater than the respective limit of detection at week 24 (four participants) or 96 (one participant), with the solid line representing CSF escape.

Clinical course of CSF escape after treatment in acute HIV-1 infection

From 145 on-ART CSF samples, one participant at week 24 post-treatment initiation met criteria for CSF escape. This 23-year-old male identified as bisexual and presented with acute retroviral syndrome. He was diagnosed with AHI in Fiebig stage IV at estimated 21 days post infection and was immediately started on efavirenz, tenofovir, and emtricitabine. Plasma HIV-1 RNA and CD4+ T cell count responded rapidly to ART (Supplemental Figure 1).

This participant had no study lumbar punctures or neuroimaging performed other than that at week 24. CSF white blood cell count was 4 cells/mm³, protein 30 mg/DL, and glucose 62 mg/dL. The participant did not endorse any neurological symptoms at week 24. An MRI performed at this time showed a small nonspecific hyperintense focus in the right high frontal white matter. Performance on neuropsychological testing was lowest at week 12 with a composite z-score of −1.8. At subsequent visits, he returned to a z-score within ±1.0 (Supplemental Figure 1). After 76 weeks on ART, this participant was transitioned to dolutegravir, abacavir, and lamivudine per the study protocol, which was unrelated to the CSF findings from week 24. At baseline, the participant had nonreactive venereal disease research laboratory test (VDRL) and Treponema pallidum hemagglutination assay (TPHA). At week 24, TPHA was nonreactive, yet VDRL was not performed. At week 48, the participant had a reactive VDRL with a titer of 1:32 and reactive TPHA.

DISCUSSION

After immediate ART initiation in Thai participants with AHI, we detected a very low frequency of CSF escape of 1% after 24 weeks and 0% after 96 weeks. Moreover, the single case of CSF escape was a low CSF viral load of 316 copies/mL and was not associated with any neurological findings or dysfunction. This represents asymptomatic escape that was detected incidentally in a research setting. Prevalence of HIV CSF escape has been estimated in 10-20% of individuals who started ART during chronic HIV ^{[1, 12]}. Thus, initiation of ART during AHI may produce a beneficial control of HIV-1 within the CNS compared with treatment during chronic infection.

CSF escape has been described in the context of symptomatic neurological and neurocognitive impact in some patients, along with neuroinflammation and neuroimaging abnormalities ^{[1, 13]}. Additionally, persistent CSF escape may indicate the presence of a CNS reservoir due to HIV-1 replication from CNS resident cells. Episodic CSF escape, otherwise called CSF HIV viral blips, can stem from CNS reservoirs, but may also be explained by transient immune cell trafficking into the CNS that supports local HIV-1 replication or release ^[3]. Many cases of detectable CSF HIV-1 RNA are likely better described as CSF viral blips since they are not elevated on longitudinal measurements ^[4]. Of note, even with CSF viral blips, there may be an association between low-level CSF HIV-1 RNA and increased neuroinflammation as measured by elevated CSF neopterin levels ^[4].

Prior characterizations of this AHI cohort have described pre-ART characteristics of acute CNS infection. At baseline, most participants have detectable CSF HIV-1 RNA, which is consistently lower than that in plasma ^[5]. A higher ratio of CSF to plasma HIV-1 RNA associates with a greater degree of systemic immunodysregulation and neuroinflammation ^[14]. We now demonstrate that despite high pre-ART HIV-1 RNA levels (up to 10⁶ copies/ml), CSF HIV-1 RNA is almost always suppressed to undetectable levels when initiating ART in AHI. This rarity of CSF escape may suggest that very early treatment reduces neuroinflammation, CNS persistence, and possibly viral reservoirs since the CNS harbors unique viral entry dynamics and compartmentalization ^[6]. Further investigations are needed to interrogate levels of HIV-1 DNA in CSF cells and other markers of CNS HIV-1 persistence.

The case of CSF escape was identified in a participant who has undergone only one study lumbar puncture to date. Thus, it is impossible to distinguish between persistent CSF escape and a CSF viral blip. Secondary CSF escape is unlikely given the nonreactive TPHA at time of lumbar puncture ^[15]. The level of 316 copies/mL is too low to perform sequencing or resistance testing using available assays. Applying novel assays with low viral copy number input could be useful in determining whether drug resistance mutations drive this case of CSF escape, degree of compartmentalization between blood and the CNS, and cellular tropism (macrophage versus CD4+ T cell). Such information would provide additional clues about whether the population is produced by resident CNS cells versus transient immune cell trafficking ^[3], though does not exclude the possibility of macrophage migration to and from the CNS through meningeal lymphatics ^[16].

With only one case of CSF escape, we cannot draw any statistically significant associations with markers of neuroinflammation or systemic inflammation. In the cohort at large, markers of immune activation in CSF but not plasma normalize after 96 weeks of ART ^[8]. Due to sample dilution for volume requirements for CSF, the LLQ of HIV-1 RNA differs between CSF and plasma, and thus our study cannot capture low-level CSF HIV-1 RNA < 80 copies/mL. Recent work has demonstrated even low-level CSF HIV-1 RNA < 20 copies/mL associates with decreased blood-brain barrier integrity and executive function ^[17]. Participants who consented for LP had small differences in CD4+ T cell count and Fiebig stage at enrollment compared with those who declined LP, and thus our study population may not be representative of all individuals presenting with AHI.

In conclusion, CSF escape is rare (1%) following initiation of ART during AHI in this cohort. Future work should corroborate the low incidence of CSF escape over longer follow-up and investigate the virological features of any CSF escape viruses detected after treatment in AHI.