JAIDS Journal of Acquired Immune Deficiency Syndromes:
Persistent Intrathecal Immune Activation in HIV-1-Infected Individuals on Antiretroviral Therapy
Yilmaz, Aylin MD, PhD*; Price, Richard W MD†; Spudich, Serena MD†; Fuchs, Dietmar PhD‡; Hagberg, Lars MD, PhD*; Gisslén, Magnus MD, PhD*
From the *Department of Infectious Diseases, The Sahlgrenska Academy at Göteborg University, Göteborg, Sweden; †Department of Neurology, University of California San Francisco, San Francisco General Hospital, San Francisco, CA; and the ‡Division of Biological Chemistry, Biocentre, Innsbruck Medical University, Innsbruck, Austria.
Received for publication June 27, 2007; accepted September 10, 2007.
Supported by grants from the National Institutes of Health (NINDS R01 NS43103-01), Medical Faculty of Göteborg University (ALFGBG-2874), Göteborg Medical Society, Swedish Physicians Against AIDS Research Foundation, and Government of the State of the Austrian Tyrol.
Correspondence to: Aylin Yilmaz, MD, PhD, Department of Infectious Diseases, Sahlgrenska University Hospital/Östra, Smörslottsgatan 1, 416 85 Göteborg, Sweden (e-mail: email@example.com).
Background: Neopterin is a well-established marker of macrophage activation. The cerebrospinal fluid (CSF) neopterin levels are elevated in most HIV-1-infected individuals and decrease significantly after initiation of antiretroviral therapy (ART). Unexpectedly, CSF concentrations often remain mildly abnormal even in patients treated for a long time with suppressive ART. The aims of this study were to analyze if persistently elevated CSF neopterin levels were associated with the type of antiretroviral regimen or with low-level CSF HIV-1 concentrations and to evaluate if plasma HIV-1 RNA levels correlated to lingering CSF neopterin concentrations in patients with effective ART.
Methods: One hundred fifty-seven chronically HIV-1-infected patients with stable ART for ≥6 months and no neurologic symptoms were included, and 193 HIV-1-infected patients without ART served as controls. Neopterin was analyzed with a radioimmunoassay or an enzyme-linked immunosorbent assay. HIV-1 RNA quantification was performed with the Roche Amplicor assay (version 1.5; Hoffman-La Roche, Basel, Switzerland). Two quantitative HIV-1 RNA assays with sensitivities ≤2.5 copies/mL were used in 40 samples.
Results: As anticipated, HIV-1 RNA and CSF neopterin levels were markedly lower in patients on ART compared with untreated controls. No significant difference in CSF neopterin concentrations was found between those treated with protease inhibitor- and nonnucleoside reverse transcriptase inhibitor-based regimens in combination with 2 nucleoside analogues. Subjects with CSF HIV-1 RNA loads <2.5 copies/mL had the lowest CSF neopterin levels. Plasma viral load had no impact on intrathecal immune activation in cases with CSF viral loads <50 copies/mL.
Conclusion: The persistent intrathecal cell-mediated immune response was associated with CSF viral load but not with treatment regimen in individuals on ART.
Neopterin is a low-molecular-weight pteridine mainly produced by macrophages and dendritic cells stimulated by interferon-γ (IFNγ). Its concentrations in body fluids rise when the cellular immune response is activated, for example, during infections, autoimmune disorders, tumors, and allograft rejections.1 In cerebrospinal fluid (CSF), increased concentrations of neopterin have been found in several infections of the central nervous system (CNS), including aseptic meningitis, Lyme neuroborreliosis, herpes simplex virus type 1 encephalitis, and HIV-1 infection, but also during conditions like multiple sclerosis and several neurodegenerative disorders.2-4
HIV-1 enters the CNS shortly after transmission, and HIV-1 RNA can be detected in the CSF of nearly all individuals infected by HIV-1, irrespective of disease stage and neurologic symptoms. Elevated levels of CSF neopterin have also been found during all stages of infection, with the highest levels in patients with AIDS dementia complex (ADC) and opportunistic infections.5 After initiation of highly active antiretroviral therapy (HAART), concentrations of CSF neopterin decrease rapidly, although nearly half of the patients have slightly elevated levels after 2 years of successful viral suppression in CSF.6 The reason for this persistent intrathecal immune activation and its clinical significance are unknown. Low-level virus replication, autoimmune phenomena, and a slow downregulation of the immune activity are some pathogenic theories.6
For many years, it was suggested that HIV-1 infection in the CNS might be more difficult to treat than systemic HIV-1, but several studies have indicated that this is not the case. One example is the dramatic decline in the incidence of ADC since the introduction of antiretroviral therapy (ART).7 In addition, the CSF viral load decreases markedly in HIV-1-infected patients on effective ART.8 Even patients with incomplete systemic viral suppression have a pronounced decrease of CSF viral load and of intrathecal immune activation.9
Most nucleoside reverse transcriptase inhibitors (NRTIs), nonnucleoside reverse transcriptase inhibitors (NNRTIs), and some ritonavir-boosted protease inhibitors (PIs) are considered CNS penetrating10-14 and lead to suppression of CSF viral load. It is not known what impact various combinations of antiretroviral drugs have on the intrathecal immune response or if suppressing the CSF viral load to levels <2.5 copies/mL leads to decreased immune activation in the CNS.
The aims of this study were to analyze if the persistent CSF neopterin levels in well-treated HIV-1-infected individuals were associated with antiretroviral regimen or with a CSF viral load <50 copies/mL but detected by more sensitive assays. We also wanted to evaluate whether the plasma HIV-1 RNA levels correlated with elevated CSF neopterin concentrations in patients on HAART having a CSF HIV-1 RNA level <50 copies/mL, because it has been suggested that the increased intrathecal immune activation may reflect viral replication in the periphery and not in the CNS.9
MATERIALS AND METHODS
Subjects were selected from cohorts in Göteborg, Sweden and San Francisco, California. All were neurologically asymptomatic and chronically HIV-1 infected with maintenance on stable ART regimens for ≥6 months. In total, 157 HIV-1-infected patients with ART and 193 HIV-1-infected controls without treatment for ≥6 months (74 female and 276 male) were retrospectively included in the study (232 from Göteborg and 118 from San Francisco) between November 1985 and August 2006. The subjects were divided into groups according to the number of antiretroviral drugs with which they were treated: (1) 1 NRTI; (2) 2 NRTIs; and (3) HAART, defined as ≥3 drugs. If more than 1 comparable lumbar puncture was available, a single interval was randomly selected. One hundred thirty-four of the participants were treated with HAART, 11 with 2 NRTIs, and 12 with 1 NRTI. Patient characteristics are shown in Table 1. Among the controls not receiving ART, 151 (78%) were naive to ART and the 42 previously treated subjects had been off therapy for a median of 1.9 (interquartile range [IQR]: 0.7 to 4.1) years. Study participants were further analyzed regarding the effect of CSF viral load on intrathecal immune activation, and subjects with CSF viral loads <50 copies/mL were additionally evaluated regarding treatment regimens and the impact of plasma viral load on CSF neopterin levels. This study was approved by the Ethics Committees of Göteborg University and the University of California San Francisco.
HIV-1 RNA in CSF and plasma was measured using the Roche Amplicor assay (version 1.5; Hoffman-La Roche, Basel, Switzerland). The assay has a dynamic range down to 50 copies/mL (1.70 log10 copies/mL) and a lower detection limit of 20 copies/mL (1.30 log10 copies/mL) for plasma and CSF. All HIV-1 RNA values <20 copies/mL were set at 19 copies/mL. Forty patients (11 from Sweden and 29 from the United States) had their plasma and CSF viral loads determined by modified quantification assays with detection limits of 2.0 and 2.5 copies/mL, respectively (methods described elsewhere).15,16 Neopterin was measured by a commercially available radioimmunoassay (Henningtest Neopterin; BRAHMS, Berlin, Germany)17 or an enzyme-linked immunosorbent assay (BRAHMS). The same antibodies are used in these 2 assays, and the results are interchangeable.18 Normal reference values were ≤8.8 nmol/L in serum and ≤5.8 nmol/L in CSF.19 The CD4 T-cell count was analyzed by flow cytometry.
Log10 transformation was applied to all HIV-1 RNA and neopterin data in figures and tables. Nonparametric methods were used for group descriptives (median and IQR). When 2 groups were being compared, comparisons were done using the Mann-Whitney U test. Differences among more than 2 groups were detected using 1-way analysis of variance, with multiple-group comparisons done using the Tukey post hoc test.
A larger proportion of patients on HAART (88%; P < 0.001) and 2 NRTIs (64%; P < 0.001) had a CSF viral load <50 copies/mL than patients treated with 1 NRTI (0%). Plasma HIV-1 RNA was <50 copies/mL in 70% of individuals on HAART, 9% of individuals on 2 NRTIs, 0% among subjects treated with 1 NRTI, and 2% for those without treatment. CSF neopterin levels were significantly lower for individuals on HAART (P < 0.001) or 2 NRTIs (P < 0.01) compared with untreated controls, whereas patients on 1 NRTI had levels in the same range as the controls (Fig. 1). The patients on HAART had significantly higher CD4 T-cell counts than individuals on 1 NRTI and subjects without ART. Subjects not receiving ART were more recently diagnosed with HIV-1 than treated subjects. It should be kept in mind that the number of subjects receiving 1 NRTI or 2 NRTIs was quite small.
Individuals on PI- or NNRTI-based HAART (defined as ≥1 PI or 1 NNRTI in combination with NRTIs) and with CSF viral loads <50 copies/mL had lower levels of CSF neopterin compared with untreated controls with CSF HIV-RNA levels <50 copies/mL (P < 0.001 and P < 0.01, respectively; Fig. 2). The CSF HIV-1 RNA and neopterin for subjects on PI- or NNRTI-based regimens was in the same range. The median (IQR) plasma HIV-1 RNA levels did not differ between subjects on a PI-based regimen (1.3 [1.3 to 1.8] log10 copies/mL) and an NNRTI-based regimen (1.3 [1.3 to 1.3] log10 copies/mL) but was significantly higher in untreated controls (3.2 [2.1 to 3.8] log10 copies/mL) (P < 0.001). The patients had been on their PI- or NNRTI-based treatment for a median (IQR) of 1.9 (1.0 to 3.2) and 2.9 (1.1 to 4.0) years, respectively. Of the 70 patients on PI-based treatment, 21 were on single PI treatment with indinavir (n = 11), nelfinavir (n = 8), or saquinavir (n = 2); 8 had a regimen including saquinavir and nelfinavir; and 41 had a ritonavir-boosted PI with lopinavir (n = 22), atazanavir (n = 9), saquinavir (n = 5), or indinavir (n = 5). No differences in CSF neopterin concentrations were found between those groups (data not shown). For the 43 subjects on NNRTI-based regimens, CSF neopterin did not differ between those on efavirenz (n = 24) and those on nevirapine (n = 19).
The individuals who had CSF HIV-1 RNA levels <2.5 copies/mL had significantly lower plasma viral loads than the other groups (P < 0.001). The median (IQR) plasma HIV-1 RNA levels were 0.4 (0.4 to 0.9) log10 copies/mL for subjects with CSF viral loads <2.5 copies/mL, 2.6 (1.6 to 3.5) log10 copies/mL for subjects with CSF HIV-1 RNA levels from 2.5 to 49 copies/mL, 4.1 (2.7 to 4.7) log10 copies/mL for subjects with CSF HIV-1 RNA levels ≥50 copies/mL, and 3.3 (2.3 to 4.4) log10 copies/mL for untreated controls. The subjects with CSF viral loads <2.5 copies/mL also had the lowest CSF neopterin concentrations (P < 0.01; Fig. 3). Fifteen of 35 subjects with CSF viral loads <2.5 copies/mL had CSF neopterin concentrations within the normal reference value compared with 2 of the 29 participants with CSF HIV-1 RNA loads ≥2.5 copies/mL (13 with a CSF HIV-1 RNA level between 2.5 and 49 copies/mL and 16 with a CSF HIV-1 RNA level ≥50 copies/mL).
For subjects on HAART and with CSF viral loads <50 copies/mL, there were no statistically significant differences in CSF and serum neopterin levels for the 92 individuals with plasma viral loads <50 copies/mL and the 26 with plasma viral loads ≥50 copies/mL (Fig. 4). The median (IQR) plasma HIV-1 RNA level for individuals with detectable viremia was 3.1 (2.4 to 3.7) log10 copies/mL.
As expected, HAART effectively reduces plasma and CSF HIV-1 RNA and the degree of immune activation, measured as decreasing concentrations of serum and CSF neopterin. Despite virologically successful ART, many patients demonstrate ongoing intrathecal immune responses, which is in concordance with previous studies.6 The clinical significance of this finding is unknown. Although there have been reports about neurologic deterioration in patients with virologically effective treatment,20 it is an uncommon clinical development. By analyzing various treatment regimens and by using an extremely sensitive HIV-1 RNA quantification assay, we have extended the characterization of the persisting signs of intrathecal immune activation, but the question as to whether low-grade intrathecal immune activation is harmful or not remains.
One weakness with the study is its retrospective approach. CSF analyses are difficult to include in prospective studies because of the reluctance of many to include lumbar punctures in the study protocols comparing treatments. Therefore, retrospective studies in well-categorized patient materials are, to date, the main means of gathering more information. Another weakness with the study might be the fact that the controls were poorly matched to the cases.
Current treatment guidelines recommend a combination of 2 NRTIs plus an NNRTI or a PI when initiating ART.21,22 In this study, we compared patients treated with NRTIs plus a PI and with NRTIs plus an NNRTI. The CSF neopterin levels were similar in these 2 groups, which indicates that there are no large differences between these HAART regimens on persistent low-level intrathecal immune activation. It should be noticed, however, that NNRTI-treated patients had been somewhat longer on their treatment than PI-treated patients. Another finding demonstrating that many treatment regimens are effective in the CSF is that the current nonaccepted treatment with only 2 NRTIs leads to a much more pronounced reduction of viral load in CSF than in plasma,23 which was confirmed in the present study.
There were previously concerns that the CNS could be a sanctuary site that is difficult to reach with antiretroviral drugs. Theoretically, the lower drug concentrations in CSF, with the large mass of infected cells in the CNS being long-lived macrophages and microglia, might lead to a slower and less impressive decline in CSF than in plasma. This type of viral kinetics has been observed in patients with ADC,24 where virus in CSF mostly emanates from local production within the CNS as opposed to a larger ratio of virus trafficking from the periphery in neurologically asymptomatic individuals, as in this study.
Beneficial virologic treatment results in CSF in individuals without neurocognitive symptoms could, at least in part, be attributable to the fact that the plasma viral load in most cases exceeds the CSF viral load25 or that some ART regimens more effectively reduce viral replication in CSF than in blood. In vitro experiments have demonstrated a superior antiviral effect for NRTIs in macrophages compared with NRTIs in activated lymphocytes, because cells like macrophages have much lower levels of 2′-deoxy-nucleoside triphosphates (dNTPs), thereby enhancing the chain-terminating activity of NRTIs.26 Antiretroviral drugs are required to penetrate into the CNS to achieve a substantial decline of the CSF viral load, but the number of CSF-penetrating drugs needed to do this is not established. Monotherapy with zidovudine reduces the viral burden and the cell-mediated immune response in CSF more than in plasma, and even led to a decrease in the incidence of ADC in the late 1980s.27-29 Drugs that are undetectable or barely detectable in CSF, such as saquinavir/ritonavir, didanosine, and nelfinavir, do not lead to viral suppression in CSF when used on their own.27,30,31 Lopinavir/ritonavir, conversely, reaches CSF concentrations exceeding the 50% inhibitory concentration (IC50)13 and suppresses the CSF viral load successfully for a median of 38 weeks when used as monotherapy.32 Thus, 1 potent antiretroviral drug could suffice as long as it is has the capacity to cross the blood-brain barrier (BBB). How long the effect of a single agent lasts in the CSF and whether this can lead to increased development of resistance in the long run need further evaluation.
Among treated subjects, the group with CSF HIV-1 RNA levels <2.5 copies/mL had significantly lower CSF neopterin levels than subjects with higher CSF viral loads. It has previously been observed that higher CSF viral loads are associated with higher CSF neopterin concentrations, but this has not been extended to CSF HIV-1 RNA levels <50 copies/mL. We also found that extremely low viral activity stimulates immune activation, albeit at much lower levels than in untreated patients. If intrathecal immune activation is thought to be harmful, a conclusion drawn from these results would be to include more CSF monitoring analysis and to lower the threshold for viral detection and the goal for treatment.
Interestingly, in another study, subjects who exhibited systemic virologic failure (defined as HIV-1 RNA load >500 copies/mL) still had beneficiary results of their ART, with decreased systemic immune activation and a larger reduction of CSF than plasma viral load.9 The failing patients in that study had lower CSF neopterin than untreated subjects but higher levels than the successful patients. It was hypothesized that reduced activity of the peripheral immune system might lead to fewer T cells trafficking through the BBB, and reduced CSF viral load as a consequence, because short-lived cells from outside the CNS have been shown to sustain a large proportion of the CSF viral load in neurologically asymptomatic subjects.33 Bearing this in mind, we expected the individuals in our study with systemic virologic failure (plasma HIV-1 RNA load ≥50 copies/mL) to have higher CSF neopterin levels than the treatment successes (plasma HIV-1 RNA load <50 copies/mL). This was only true for subjects who also had CSF viral loads ≥50 copies/mL, however. Individuals with plasma viral loads ≥50 copies/mL (failures) but with undetectable CSF viral loads did not have higher CSF neopterin levels than the group with undetectable plasma HIV-1 RNA levels as well. This again points toward the viral burden in CSF as an important factor behind increased CSF neopterin concentrations. Furthermore, no difference in CSF neopterin was found between patients on HAART and a CSF HIV-1 RNA load ≥50 copies/mL and matched untreated controls.
Although ART reduces the CSF viral load dramatically, decreases the incidence of neurologic complications, and even improves symptoms in patients with ADC, it does not have the same impressive effect on intrathecal cell-mediated immune activation, as measured by neopterin. Even though the individuals with CSF viral loads <2.5 copies/mL had the lowest CSF neopterin levels, these levels were still slightly elevated in many (57%) of these virologically well-suppressed patients. We do not yet know the clinical consequences of having an active immune system in the CNS for a prolonged time and if this could lead to neurologic injuries in the future.
1. Wirleitner B, Schroecksnadel K, Winkler C, et al. Neopterin in HIV-1 infection. Mol Immunol
2. Furukawa Y, Nishi K, Kondo T, et al. Significance of CSF total neopterin and biopterin in inflammatory neurological diseases. J Neurol Sci
3. Fredrikson S, Link H, Eneroth P. CSF neopterin as marker of disease activity in multiple sclerosis. Acta Neurol Scand
4. Hagberg L, Dotevall L, Norkrans G, et al. Cerebrospinal fluid neopterin concentrations in central nervous system infection. J Infect Dis
5. Griffin DE, McArthur JC, Cornblath DR. Neopterin and interferon-gamma in serum and cerebrospinal fluid of patients with HIV-associated neurologic disease. Neurology
6. Abdulle S, Hagberg L, Svennerholm B, et al. Continuing intrathecal immunoactivation despite two years of effective antiretroviral therapy against HIV-1 infection. AIDS
7. d'Arminio Monforte A, Cinque P, Mocroft A, et al. Changing incidence of central nervous system diseases in the EuroSIDA cohort. Ann Neurol
8. Mellgren Å, Antinori A, Cinque P, et al. Cerebrospinal fluid HIV-1 infection usually responds well to antiretroviral treatment. Antivir Ther
9. Spudich S, Lollo N, Liegler T, et al. Treatment benefit on cerebrospinal fluid HIV-1 levels in the setting of systemic virological suppression and failure. J Infect Dis
10. Burger DM, Kraaijeveld CL, Meenhorst PL, et al. Penetration of zidovudine into the cerebrospinal fluid of patients infected with HIV. AIDS
11. Haas D, Stone J, Clough LA, et al. Steady-state pharmacokinetics of indinavir in cerebrospinal fluid and plasma among adults with human immunodeficiency virus type 1 infection. Clin Pharmacol Ther
12. Haworth SJ, Christofalo B, Anderson RD, et al. A single-dose study to assess the penetration of stavudine into human cerebrospinal fluid in adults. J Acquir Immune Defic Syndr
13. Yilmaz A, Ståhle L, Hagberg L, et al. Cerebrospinal fluid and plasma HIV-1 RNA levels and lopinavir concentrations following lopinavir/ritonavir regimen. Scand J Infect Dis
14. Antinori A, Perno CF, Giancola ML, et al. Efficacy of cerebrospinal fluid (CSF)-penetrating antiretroviral drugs against HIV in the neurological compartment: different patterns of phenotypic resistance in CSF and plasma. Clin Infect Dis
15. Yilmaz A, Svennerholm B, Hagberg L, et al. Cerebrospinal fluid viral loads reach less than 2 copies/ml in HIV-1-infected patients with effective antiretroviral therapy. Antivir Ther
16. Havlir DV, Bassett R, Levitan D, et al. Prevalence and predictive value of intermittent viremia with combination HIV therapy. JAMA
17. Werner E, Bichler A, Daxenbichler G, et al. Determination of neopterin in serum and urine. Clin Chem
18. Mayersbach P, Augustin R, Schennach H, et al. Commercial enzyme-linked immunosorbent assay for neopterin detection in blood donations compared with RIA and HPLC. Clin Chem
19. Hagberg L, Andersson L, Abdulle S, et al. Clinical application of cerebrospinal fluid neopterin concentrations in HIV infection. Pteridines
20. Cysique LA, Maruff P, Brew BJ. Prevalence and pattern of neuropsychological impairment in human immunodeficiency virus-infected/acquired immunodeficiency syndrome (HIV/AIDS) patients across pre- and post-highly active antiretroviral therapy eras: a combined study of two cohorts. J Neurovirol
21. Gisslén M, Ahlqvist-Rastad J, Albert J, et al. Antiretroviral treatment of HIV infection: Swedish recommendations 2005. Scand J Infect Dis
22. Hammer SM, Saag MS, Schechter M, et al. Treatment for adult HIV infection: 2006 recommendations of the International AIDS Society-USA Panel. JAMA
23. Gisslén M, Svennerholm B, Norkrans G, et al. Cerebrospinal fluid and plasma viral load in HIV-1-infected patients with various anti-retroviral treatment regimens. Scand J Infect Dis
24. Ellis RJ, Gamst AC, Capparelli E, et al. Cerebrospinal fluid HIV RNA originates from both local CNS and systemic sources. Neurology
25. Gisslén M, Fuchs D, Svennerholm B, et al. Cerebrospinal fluid viral load, intrathecal immunoactivation, and cerebrospinal fluid monocytic cell count in HIV-1 infection. J Acquir Immune Defic Syndr
26. Aquaro S, Perno CF, Balestra E, et al. Inhibition of replication of HIV in primary monocyte/macrophages by different antiviral drugs and comparative efficacy in lymphocytes. J Leukoc Biol
27. Gisslén M, Norkrans G, Svennerholm B, et al. The effect on human immunodeficiency virus type 1 RNA levels in cerebrospinal fluid after initiation of zidovudine or didanosine. J Infect Dis
28. Portegies P, de Gans J, Lange JM, et al. Declining incidence of AIDS dementia complex after introduction of zidovudine treatment. BMJ
29. Hagberg L, Norkrans G, Gisslén M, et al. Intrathecal immunoactivation in patients with HIV-1 infection is reduced by zidovudine but not by didanosine. Scand J Infect Dis
30. Gisolf EH, Enting RH, Jurriaans S, et al. Cerebrospinal fluid HIV-1 RNA during treatment with ritonavir/saquinavir or ritonavir/saquinavir/stavudine. AIDS
31. Lafeuillade A, Solas C, Halfon P, et al. Differences in the detection of three HIV-1 protease inhibitors in non-blood compartments: clinical correlations. HIV Clin Trials
32. Yeh R, Letendre S, Novak I, et al. Single agent therapy with lopinavir/ritonavir controls HIV-1 replication in the central nervous system [abstract 381]. Presented at: 14th Conference on Retroviruses and Opportunistic Infections; 2007; Los Angeles.
33. Harrington PR, Haas DW, Ritola K, et al. Compartmentalized human immunodeficiency virus type 1 present in cerebrospinal fluid is produced by short-lived cells. J Virol
HIV-1 RNA; cerebrospinal fluid; neopterin; antiretroviral therapy
© 2008 Lippincott Williams & Wilkins, Inc.
What does "Remember me" mean?
By checking this box, you'll stay logged in until you logout. You'll get easier access to your articles, collections,
media, and all your other content, even if you close your browser or shut down your
To protect your most sensitive data and activities (like changing your password),
we'll ask you to re-enter your password when you access these services.
What if I'm on a computer that I share with others?
If you're using a public computer or you share this computer with others, we recommend
that you uncheck the "Remember me" box.
Highlight selected keywords in the article text.
Data is temporarily unavailable. Please try again soon.