Cerebrospinal fluid HIV escape associated with progressive neurologic dysfunction in patients on antiretroviral therapy with well controlled plasma viral load

Abstract

Objective: 

To characterize HIV-infected patients with neurosymptomatic cerebrospinal fluid (CSF) ‘escape’, defined as detectable CSF HIV RNA in the setting of treatment-suppressed plasma levels or CSF RNA more than 1-log higher than plasma RNA.

Design: 

Retrospective case series.

Setting: 

Four urban medical centers in the United States and Europe.

Participants: 

Virologically controlled HIV-infected patients on antiretroviral therapy (ART) with progressive neurologic abnormalities who were determined to have CSF ‘escape’.

Intervention: 

Optimization of ART based upon drug susceptibility and presumed central nervous system exposure.

Main outcome measures: 

Levels of CSF HIV RNA and inflammatory markers, clinical signs and symptoms, and MRI findings.

Results: 

Ten patients presented with new neurologic abnormalities, which included sensory, motor, and cognitive manifestations. Median CSF HIV RNA was 3900 copies/ml (range 134–9056), whereas median plasma HIV RNA was 62 copies/ml (range <50 to 380). Median CD4⁺ T-cell count was 482 cells/μl (range 290–660). All patients had been controlled to less than 500 copies/ml for median 27.5 months (range 2–96) and five of 10 had been suppressed to less than 50 copies/ml for median 19.5 months (range 2–96). Patients had documentation of a stable ART regimen for median 21 months (range 9–60). All had CSF pleocytosis or elevated CSF protein; seven of eight had abnormalities on MRI; and six of seven harbored CSF resistance mutations. Following optimization of ART, eight of nine patients improved clinically.

Conclusion: 

The development of neurologic symptoms in patients on ART with low or undetectable plasma HIV levels may be an indication of CSF ‘escape’. This study adds to a growing body of literature regarding this rare condition in well controlled HIV infection.

Introduction

HIV-1 infection is present in the central nervous system (CNS) beginning during primary viremia and continuing over the course of untreated infection ^[1–3]. Although most patients with HIV infection do not present with neurologic symptoms, elevated HIV RNA levels in the cerebrospinal fluid (CSF) can be associated with HIV encephalitis (HIVE) and HIV-associated dementia (HAD) in patients with advanced infection ^[4]. Antiretroviral therapy (ART) suppresses plasma and CSF viral levels and improves neurologic outcomes in patients with HIV ^[5]. As a result, the incidence of HAD has substantially decreased over the last two decades ^[6]. Additionally, even in asymptomatic patients, local HIV infection within the CNS may lead to distinct responses to ART in the neurologic and peripheral blood tissue compartments ^[7]. However, a subset of patients may still develop neurologic symptoms in the setting of long-term plasma viral control ^[8,9].

Recently, Canestri et al.^[10] demonstrated the phenomenon of CSF/plasma HIV RNA discordance involving the development of new neurologic symptoms in 11 patients with well controlled HIV. In some cases, genotyping demonstrated significant resistance mutations in the CSF viral subpopulation, suggesting that the current treatment regimen had failed in the CNS. Some patients improved when their ART regimen was optimized based upon the results of genotyping and analysis of presumed CNS drug exposure.

We sought to add to the contributions of Canestri et al. and previous reports ^[11–15] by further investigating the condition of CSF ‘escape’ in patients will well controlled plasma HIV and preserved immune function. One of the criticisms of previous reports has been that some patients have been on monotherapy ^[10] or salvage therapy ^[12,13], and others have had low CD4⁺ T-cell counts ^[10,11]. We also sought to provide more detailed background information regarding HIV disease course in these patients and to emphasize key portions of the clinical picture, including neuroimaging, that have not been described in detail.

Here, we report a group of patients from four institutions in the United States and Europe. Each patient presented with new-onset neurologic symptoms in the context of low or undetectable plasma HIV levels, underwent neurologic studies, including lumbar puncture and CSF analysis, and was noted to have CSF ‘escape’. ART regimens were optimized based upon drug susceptibility and penetration. This study adds to a growing body of evidence regarding the rare condition of CSF ‘escape’ associated with progressive neurologic disease in otherwise well controlled HIV infection.

Methods

Study design and patient characteristics

We retrospectively compiled cases of HIV-infected patients on ART who presented with neurologic signs and/or symptoms in the context of plasma HIV RNA suppression and underwent evaluation including CSF studies. Participants were identified at four urban academic centers in San Francisco, California, USA; Milan, Italy; New Haven, Connecticut, USA; and Gothenburg, Sweden.

All patients were on stable combination ART regimens with either suppressed (<500 copies/ml) or undetectable (<50 copies/ml) plasma HIV RNA. Patients with symptoms attributable to other neurologic or psychiatric causes were excluded. CSF and concurrent plasma samples were obtained either by the primary clinical team for diagnostic purposes or in the context of research studies in separate local protocols that were approved by the institutional review board or local equivalent at each institution. Clinical brain MRIs were obtained prior to lumbar puncture on varied local 1.5 Tesla scanners in the majority of participants. We included patients found to have CSF ‘escape’, defined as detectable CSF HIV RNA in the setting of plasma levels of less than 50 copies/ml or CSF RNA more than 1-log higher than plasma RNA level as previously defined by Canestri et al. ^[10].

Laboratory methods

HIV RNA levels were measured in cell-free CSF and plasma using the ultrasensitive Amplicor HIV Monitor (version 1.5, Roche Molecular Diagnostic Systems, Branchburg, New Jersey, USA), Cobas TaqMan RealTime HIV-1 (version 1 or 2; Hoffmann-La Roche, Basel, Switzerland), or the Abbott RealTime HIV-1 (Abbot Laboratories, Abbot Park, Illinois, USA) assays at local sites. For uniformity, 50 copies/ml was used as the lower limit of quantitative detection in this analysis. Paired blood and CSF measurements used the same assay. CSF total white blood cells (WBCs) and protein, and CD4⁺ and CD8⁺ T-lymphocyte counts by flow cytometry were measured at each local laboratory on fresh samples. Blood and CSF neopterin measurements employed commercially available immunoassays (BRAHMS Aktiengesellschaft, Hennigsdorf, Germany) and were performed in one laboratory. HIV resistance genotyping was performed where available in CSF samples harboring adequate HIV RNA levels for amplification. Genotyping was interpreted according to the International Antiviral Society-USA guidelines ^[16].

As a means to approximate expected effectiveness of ART in the CNS, we used proposed CNS penetration effectiveness (CPE) scores using the 2010 version developed by Letendre and colleagues ^[8,17] to calculate a ‘raw’ CPE score for each regimen at the time when discordance was identified. In an effort to take into account effective resistance in a consistent, quantitative way, we calculated an ‘adjusted’ CPE score based upon the genotyping results of CSF viral isolates. When a mutation to a particular drug in the regimen was identified, the individual CPE score for that drug was arbitrarily designated ‘0’.

Descriptive analyses were undertaken to characterize these patients and are reported as percentages or median value (range) for continuous variables.

Results

Between February 2000 and August 2011, 10 patients with chronic but well controlled HIV infection and preserved immune status presented with new neurologic symptoms and were recognized as meeting the criteria for CSF ‘escape’. The clinical and demographic characteristics of these patients are described in Table 1.

Table 1: Demographic information and HIV history of patients with cerebrospinal fluid/plasma discordance.

The patients consisted of eight men and two women with a median age of 47.5 years (range 26–55). The median time since HIV diagnosis was 16.2 years (range 9.4–21.7). At the time of the neurologic episode, the patients had been on a stable regimen for a median of 21 months (range 9–60). These regimens consisted of at least two nucleoside reverse transcriptase inhibitors (NRTIs) with a protease inhibitor in nine of 10 cases; the protease inhibitor was boosted with ritonavir in eight of nine cases. Individual patients had additional components to their regimen, including integrase or fusion inhibitors. None were on monotherapy or dual therapy.

The median duration of HIV RNA suppression below 500 copies/ml was 27.5 months (range 2–96). The median duration of HIV RNA suppression below 50 copies/ml was 19.5 months (range 2–96). The median CD4⁺ T-cell count at presentation was 482 cells/μl (range 290–660). The median nadir CD4⁺ T-cell count was 35 cells/μl (range 4–222).

Three patients had a previous neurologic abnormality. These included a presumed cerebellar meningioma that had been stable for many years (patient 1000), labyrinthitis and right sensorineural deafness (patient 8000), and CNS lymphoma that had resolved completely with initiation of ART, without radiotherapy, several years before (patient 5168).

Clinical and MRI manifestations

The neurologic abnormalities present in this patient group (Table 1) occurred subacutely (>2 weeks) in nine of 10 patients and were acute (<2 weeks) in one (patient 9000). They comprise a variety of sensory (in three patients), motor (in nine), and cognitive (in eight) manifestations. Imaging in seven of eight patients at the time of presentation showed MRI abnormalities consisting of white matter hyperintensities on T2-weighted and FLAIR sequences (Table 2). Figure 1a–d shows representative imaging examples from patients 2000 and 7000 during the initial studies for neurologic symptoms.

Table 2: Neurologic studies in patients with cerebrospinal fluid/plasma discordance.

Fig. 1: MRI findings at the time of cerebrospinal fluid ‘escape’ and following modification of antiretroviral regimen.Panels (a–j) show selected MRI images for patients 2000 and 7000. Panels (a–d) show imaging at the time of neurologic workup when cerebrospinal fluid (CSF) ‘escape’ was detected for patients 2000 (a, b) and 7000 (c, d), demonstrating diffuse T2 prolongation (a, b) and suggesting focal lesions (d) at the time of CSF ‘escape’. Panels (e–h) show follow-up imaging for patient 2000 at 111 days and patient 7000 at 60 days. Even though neurologic symptoms had resolved in both cases, imaging still shows diffuse leukoencephalopathy (e, f) and hyperintense, diffuse signal alteration of bilateral white matter (g), despite improvement of previous focal lesions (h). Panels (i and j) show imaging for patient 2000 at 567 days follow-up, demonstrating significant interval decrease in T2 prolongation.

Cerebrospinal fluid and brain abnormalities

CSF pleocytosis and biochemical abnormalities were found in all 10 patients (Table 2). Eight of nine patients had elevated CSF protein levels of at least 60 mg/dl. The median protein level was 105 mg/dl (range 46–170 mg/dl). CSF pleocytosis was observed in nine of 10 patients, with median 14.5 cells/μl (range 0–200). All samples were negative for bacteria, fungi, and other viruses by standard microbiological tests at each institution, including JC virus DNA studies for progressive multifocal leukoencephalopathy. Two samples (patients 1000 and 7000) had low-level (<5000 copies/ml) Epstein–Barr virus (EBV) DNA ^[18]. CSF from patient 5168 had previously been positive for EBV in the past when the patient suffered from CNS lymphoma, but CSF EBV titers were negative during and throughout the time of CSF HIV ‘escape’ in this patient.

CSF neopterin was measured in two patients at the time of CSF ‘escape’. Patient 7066 had a CSF neopterin level of 76.3 nmol/l with a plasma level of 12 nmol/l; patient 5168 had a CSF neopterin of 37.6 nmol/l with a plasma level of 8 nmol/l. Reference ranges for HIV-uninfected individuals are less than 5.8 nmol/l in CSF and less than 8.8 nmol/l in plasma ^[19]; for successfully ART-treated HIV-infected individuals, mean neopterin is 10.8 nmol/l in CSF ^[20].

Two patients (1034 and 4065) underwent brain biopsy at the time of CSF ‘escape’, revealing dense, perivascular lymphocytic infiltrates in the white matter with extension into the surrounding parenchyma. Immunoperoxidase staining showed a mixture of mature and immature B and T lymphocytes, with CD8⁺ predominance.

HIV RNA in cerebrospinal fluid and plasma

By definition, all patients had CSF HIV replication at initial evaluation, with a median of 3900 copies/ml (range 134–9056). All had plasma HIV RNA less than 500 copies/ml and five of 10 had a plasma HIV RNA level less than 50 copies/ml at the time CSF ‘escape’ was discovered. The median plasma viral load was 62 copies/ml (range <50 to 380). For the five patients with controlled but detectable plasma HIV RNA (>50 copies/ml but <500 copies/ml), the CSF HIV RNA was at least 1-log higher than the plasma HIV RNA.

Figure 2 shows longitudinal plasma data for these patients, indicating plasma control less than 500 copies/ml in seven of 10 patients over the previous 1000 days. Of these, five of seven patients had HIV RNA below the limit of detection (<50 copies/ml) for the previous 1000 days. One patient had a transient increase in plasma viral load during this period (patient 4065), but had been well controlled previously and following this increase. Two patients had viral loads that had more recently declined to less than 500 copies/ml (patients 1034 and 7000). These patients were included because they presented with new neurologic symptoms in the absence of alternate pathogens or focal lesions as determined through imaging or brain biopsy, with CSF ‘escape’ in the setting of preserved immune status and declining plasma HIV.

Fig. 2: Longitudinal plasma HIV RNA levels for patients with cerebrospinal fluid ‘escape’.HIV RNA is calculated in days prior to time cerebrospinal fluid (CSF)/plasma discordance was detected (time ‘0’). Reference dotted horizontal line indicates 500 log₁₀ copies/ml. Corresponding CSF HIV RNA levels are indicated on the right axis of the graph at the time when CSF escape was identified (’CSF escape’) and at a standardized follow-up time point in which repeat CSF was available (’follow-up’).

Viral resistance and central nervous system penetration

Table 3 indicates the results of CNS genotyping and CNS penetration calculations. Six of seven patients on whom resistance gentoyping was conducted in the CSF had NRTI mutations, five of seven had protease inhibitor mutations, and two of seven had nonnucleoside reverse transcriptase inhibitor mutations. One patient had no mutations detected on CSF genotyping.

Table 3: HIV drug regimens and resistance profiles in patients with cerebrospinal fluid/plasma discordance.

The original antiretroviral regimens for these patients had a median CPE score of 6.5 (range 3–13). When adjusted for resistance, the median adjusted CPE score was 1 (range 0–9). Regimens were revised in nine of 10 patients based on CSF findings. The revised regimens (see Table 3) had a median raw CPE score of 11 (range 7–16) and median adjusted CPE score of 4 (range, 4–10).

Changes after treatment intervention

Eight of nine patients demonstrated clinical improvement following neurologic evaluation and ART regimen optimization. One patient did not improve (patient 5168) and one patient died from septic shock secondary to presumed bowel ischemia before treatment was modified (patient 1034).

Follow-up CSF was available in four of nine patients and demonstrated reduced CSF HIV RNA levels (from median 5775 to 66 copies/ml) at a median of 70 days following change in drug regimen (range 11–189 days). In three of four cases, discordance between CSF and plasma resolved at this follow-up point; in one case, discordance persisted at a lower level (patient 5168 with 340 copies/ml in the CSF); this patient's abnormalities did not improve after 189 days on the new regimen.

Figure 1e–h shows short-term follow-up imaging for patients 2000 and 7000. At 60 days, MRI for patient 7000 showed resolution of most focal lesions, but the development of a diffuse leukoencephalopathy despite resolution of symptoms. Similarly, patient 2000 had persistent diffuse white matter hyperintensities on MRI at 111 days, with subsequent significant decrease in these abnormalities at 346 and 567 days of follow-up.

Discussion

We report 10 cases of elevated CSF HIV RNA in the setting of plasma suppression in patients with well controlled HIV infection, with long-term plasma control and CD4⁺ T-cell counts indicating preserved immune status at the time when neurologic symptoms developed. These cases demonstrate an unusual but clinically important phenomenon of CSF ‘escape’ associated with incident neurologic signs and symptoms in patients with chronic treated HIV infection.

The patients we report comprise a representative sample of those living with antiretroviral-treated HIV. This includes individuals with persistently suppressed plasma HIV RNA over many years (patient 2000), those who have been under control for a number of years (patients 1000, 7066, 7071, 9000, and 5168), those with good control but a recent ‘blip’ (patient 4065), and those with an unclear history who are coming under control (patients 1034 and 7000). The common clinical picture of neurologic abnormalities across this spectrum of patients suggests that the process of CSF ‘escape’ is a relevant consideration in a variety of clinical contexts.

Patients experienced a variety of neurologic symptoms including cognitive, sensory, and motor impairment. Onset was most often subacute, impairment varied in severity, and abnormalities progressed over time. Overall, the neurologic symptoms reflect a level of debilitation that was significant and involved a range of functional domains.

MRI findings were consistent among patients and with those reported in previous cases ^[10–12]. Furthermore, they are similar but not identical to those classically identified in typical HAD in patients off of ART. White matter hyperintensities on T2-weighted and FLAIR imaging suggest a generalized inflammatory process consistent with diffuse encephalitis, and are similar to findings reported in patients failing ART ^[21]. Comparison of MRI results at the time of presentation (Fig. 1a–d) and short-term and long-term follow-up (Fig. 1e–h, and i and j, respectively) suggest that this process is associated with findings on imaging that may persist after the resolution of symptoms, and may take months to years to resolve completely. Still, the nature of these imaging findings remains incompletely understood.

Despite relatively reconstituted immune status at the time of evaluation, all patients had CD4⁺ T-cell nadirs less than 250 cells/μl, with many below 100 cells/μl, which is consistent with a previous report of a median nadir CD4⁺ T-cell count of 55 cells/μl in similar patients ^[10]. A history of advanced immunosuppression may confer increased risk for prior enhanced local CNS infection and compartmentalization ^[22], which, despite peripheral CD4⁺ T-cell improvement, fails to be entirely suppressed by ART. Clinically, the CD4⁺ T-cell nadir might be an important factor to consider in the assessment of a patient with new neurologic abnormalities.

One concern with previously reported cases of CSF ‘escape’ has been that some patients have been on atypical or incomplete regimens ^[10]. No patients in our study were on monotherapy or dual therapy. All were on appropriate multidrug combination ART regimens before they developed symptoms, although some older regimens are outdated by current standards. Preserved immune status and suppression of plasma viremia suggest compliance with ART, although it cannot be ruled out that suboptimal adherence contributed. Theoretically, partially reduced adherence may lead to insufficient drug concentrations in the CSF while maintaining satisfactory concentrations in plasma. CSF drug concentrations may, therefore, be an important consideration in this subset of HIV patients, as has been suggested elsewhere ^[10]. CSF ‘escape’ may arise secondary to differences in susceptibility between HIV subpopulations in the blood and CSF ^[23–26] due to the selection of resistant virus in the context of subtherapeutic drug levels in the CNS compartment ^[27].

Although it has been argued that CNS drug penetration may be an important factor in the pathogenesis of CSF ‘escape’ ^[28], these cases indicate that viral resistance should also be considered. Resistance to at least one drug in the regimen was common. The ‘adjusted’ CPE score represents a first attempt to incorporate resistance in a numerical calculation of drug effectiveness, and assumes that a single mutation will confer complete resistance to a drug, although, in fact, the drugs may remain partially effective despite the mutations. Although it is unclear to what extent clinical improvement resulted from treatment interventions, most patients improved when their regimens were adjusted with regard to both penetration and resistance. This suggests that regimen modifications should be based on more than penetration alone.

Taken together, the range and quality of neurologic dysfunction and the MRI findings in these patients have substantial overlap with typical findings in HAD. However, despite this overlap, these are not identical to those in HAD and we believe that the cause of these findings is slightly different than that of HAD in the absence of treatment. In accordance with previous reports ^[10–12], markedly elevated CSF total protein levels and WBC counts in our patients compared with healthy HIV-uninfected controls and neuro-asymptomatic HIV-infected patients on ‘successful’ ART ^[29] indicate a CNS inflammatory response. The pronounced inflammation/CD8⁺ T-lymphocyte infiltration noted on brain biopsy in two patients suggests that CSF ‘escape’ in the setting of an immune system reconstituted by systemically successful ART is associated with a degree of local inflammation distinct from typical HAD/HIVE. CSF neopterin, which is elevated in HAD and reduced by ART ^[20,30,31], was markedly increased in comparison to plasma neopterin and typical values of CSF neopterin in HIV-infected, ART-suppressed individuals. This provides evidence that in cases of CSF ‘escape’, inflammation may be relatively compartmentalized in the CNS. The role of inflammation in this disorder may determine the distinct neurotropism for these lesions, as reflected in MRI and clinical symptoms.

Given the observation that symptomatic CSF ‘escape’ is accompanied by CNS inflammation, a moderately reconstituted immune system may play an important role in both eliciting a symptomatic inflammatory response and in providing a substrate for ongoing discordant HIV replication within the CNS. As all of the participants had preserved immune function and none had recently initiated ART, typical immune reconstitution inflammatory syndrome (IRIS) was not considered the primary cause of these abnormalities. Nevertheless, the combination of persistent CNS infection and relatively preserved immune response, including an HIV-specific response, may generate immunopathology in cases of CSF ‘escape’. This is analogous to IRIS ^[32], but may differ in that it represents not the effects of immune reconstitution but rather a ‘stable state’ of antigen and immune response within the CNS.

This analysis is limited by its retrospective approach, which utilized chart reviews and was constrained to studies previously performed during clinical evaluation and research protocols. It is unclear what the prevalence of CSF ‘escape’ may be in the general HIV-infected population, as patients with minor neurologic complaints are less likely to undergo detailed CNS evaluations. Our follow-up data are limited in many cases because further studies were not pursued once symptoms resolved.

Conclusion

Physicians should be aware of this unusual but clinically significant manifestation of HIV disease. These cases reflect that new neurologic symptoms in the context of standard ART regimens and well controlled plasma HIV infection warrant an evaluation of the CSF to determine whether viral replication is occurring and, if so, whether the virus in the CSF compartment possesses resistance to the regimen being used to control the virus in the plasma compartment. CSF HIV analysis can be an important diagnostic tool and should be available to clinicians for the purpose of measuring HIV RNA concentration and identifying resistance. These cases add to the growing literature on CSF/plasma discordance that underscores the need for further investigation into the mechanism and consequences of HIV replication and persistence in the CNS.

Acknowledgements

The authors thank the patients who contributed to this study, Dr Teri Liegler of the UCSF/GIVI Virology Laboratory, Dr Marie Landry of the Yale Virology Laboratory, and Dr Simonetta Gerevini of the Head and Neck Department, San Raffaele Scientific Institute.

M.J.P. contributed in data collection; data analysis; and wrote the manuscript.

F.F. contributed in data collection; data analysis; and edited the manuscript.

J.P. contributed in data collection and data analysis.

E.L. performed data collection and data analysis.

D.F. conducted neopterin analyses and edited the manuscript.

A.B. contributed in data collection and edited the manuscript.

M.G. contributed in study conception; data collection; data analysis; and edited the manuscript.

N.A. contributed in study conception; data collection; data analysis; and edited the manuscript.

R.W.P. contributed in study conception; data collection; data analysis; and edited the manuscript.

P.C. contributed in study conception; data collection; data analysis; and edited the manuscript.

S.S. contributed in study conception; data collection; data analysis; and edited the manuscript.

Conflicts of interest

This work was supported by National Institutes of Health (grants R01 MH62701, R01 NS37660, R01 NS43103, R01 MH081772, and NCRR UCSF-CTSI UL1 RR024131), the Sahlgrenska Academy at University of Gothenburg (project ALFGBG-11067), Swedish Research Council (project 2007–7092), the Italian Ministry of Health, AIDS Program 2009–2010, and a grant from the Doris Duke Charitable Foundation to Yale University School of Medicine to fund Clinical Research Fellow M.J.P.

M.J.P., J.P., E.L., D.F., N.A., A.B., and S.S. have no interests to declare. P.C. and F.F. serve on the board for Abbott, Biogen, and Janssen and report consultancy for Biogen and Johnson & Johnson, and support for lectures and educational materials from Abbott, Boehringer BMS, Gilead, Janssen, and Merck. M.G. reports receiving lecture fees from Abbott, BMS, Gilead, GlaxoSmithKline/ViiV, Janssen, and Tibotec. R.W.P. has received speaker honoraria from Abbott.