Epidemiology & Social
Repeated measures study of human herpesvirus 8 (HHV-8) DNA and antibodies in men seropositive for both HHV-8 and HIV
Laney, A Scotta,c*; Dollard, Sheila Ca; Jaffe, Harold Wb; Offermann, Margaret Kd; Spira, Thomas Ja; Gunthel, Clifford Je; Pellett, Philip Ea,f; Cannon, Michael Ja,c
From the aNational Center for Infectious Diseases and the bNational Center for HIV, STD, and TB Prevention, Centers for Disease Control and Prevention, cRollins School of Public Health, the dWinship Cancer Institute and the eSchool of Medicine, Emory University, Atlanta, Georgia, and the fLerner Research Institute, Cleveland Clinic Foundation, Cleveland, Ohio, USA. *Present address: University of Pittsburgh Graduate School of Public Health, Pittsburgh, Pennsylvania, USA.
Requests for reprints to: Dr M. Cannon, Centers for Disease Control and Prevention, 1600 Clifton Rd, Mailstop G-18, Atlanta, Georgia 30333, USA. E-mail: firstname.lastname@example.org.
Received: 28 January 2004; revised: 2 June 2004; accepted: 28 June 2004.
Objective: To study the natural history and pathogenesis of human herpesvirus 8 (HHV-8) infection in HHV-8-seropositive, immunosuppressed men.
Design: Longitudinal study of 87 HHV-8- and HIV-seropositive men [42 with Kaposi's sarcoma (KS)] during four visits over a 2 month period.
Methods: Patients provided oral fluid and blood. HHV-8 antibody titers were measured with peptide-based enzyme-linked immunosorbent assays (ELISA) for ORF65 and K8.1; HHV-8 DNA was detected with polymerase chain reaction ELISA.
Results: HHV-8 DNA was present in oral fluid or peripheral blood mononuclear cells (PBMC) at one or more of the four visits in 71% of men with KS and 56% of men without KS. The strongest correlate of HHV-8 DNA in PBMC was the presence of KS [odds ratio (OR), 8.7; 95% confidence interval (CI), 3.4–22]. Detection of HHV-8 DNA in oral fluid or PBMC was often intermittent, but individuals who shed virus at one time point were more likely to shed at other times. Some men had incomplete epitope recognition in their anti-HHV-8 antibody response. High antibody titers were associated with the absence of circulating HHV-8, particularly for the ORF65 seroassay (OR, 0.16; 95% CI, 0.05–0.51).
Conclusions: Among HHV-8 seropositive men, circulating virus is common even in the absence of disease. The link between KS and HHV-8 DNA in PBMC suggests that anti-herpes drugs may impede KS development or progression. Seroassays should target multiple epitopes to achieve maximal sensitivity. HHV-8 replication may be limited by high antibody titers or other immune function for which antibodies are a marker.
Since human herpesvirus 8 (HHV-8) was discovered in 1994 , considerable progress has been made in understanding the virus and its relation to human disease. Basic science studies have uncovered much about virus structure, replication, in vitro transmission, genome structure and gene expression and function . Diagnostic assays have been developed for detecting HHV-8 DNA and antibodies to both lytic and latent antigens [3–5]. Using these assays, epidemiological studies have determined the prevalence of infection in different populations , identified modes of transmission [7–11], and shown that HHV-8 infection precedes and predicts Kaposi's sarcoma (KS) [8,12–14]. Furthermore, HHV-8 lytic replication appears important for the development of KS, as suggested by the frequent detection of viral DNA in the blood of patients with KS, higher HHV-8 antibody levels to lytic antigens in these patients and a higher risk of developing new KS lesions among individuals with HHV-8 DNA in peripheral blood mononuclear cells (PBMC) [15,16]. Taken together, these and other studies provide convincing evidence that HHV-8 is necessary but not sufficient for the development of KS .
However, because of the complicated nature of the relationship between infection and chronic disease, epidemiological studies have only partially described the complex HHV-8–KS relationship . Limitations of these studies have included one or more of the following: small sample size, cross-sectional study design, lack of a control group at risk for KS (i.e., HHV-8-seropositive controls), possible contamination of the polymerase chain reaction (PCR), lack of PCR on multiple specimens (e.g., saliva, blood), performance of PCR or serology but not both, lack of end-point dilution antibody titers or lack of statistical methods that account for confounding.
Therefore, certain aspects of the natural history and pathogenesis of HHV-8 infection are not well understood. For example, questions remain about the true prevalence of HHV-8 DNA in body fluids such as saliva or blood, the role of lytic infection in KS development, the contribution of the anti-HHV-8 immune response in limiting viral replication, and how viral and immunological markers change over time. The longitudinal study described here among HHV-8- and HIV-seropositive men with or without KS addressed these questions and sought to obtain a more complete picture of the relationship between HHV-8 and KS.
Two HIV clinics in Atlanta, Georgia enrolled 87 HIV-seropositive men between April 1999 and April 2001 for this study. Men were eligible for inclusion if they had positive results for two of three HHV-8 serological assays. Of the 87 enrollees, 83 were men who have sex with men and four were injection drug users; 42 had a clinical diagnosis of KS. Over a 2 month period, the men were observed at a baseline visit, and at three additional visits at 3 week intervals. The completion rate was 93% (156/168 study visits) for men with KS and 92% (166/180) for men without KS.
A questionnaire was administered to each subject, and medical information was abstracted from clinic records. Men were examined to assess KS status at each visit by using a standardized data collection form. Highly active antiretroviral therapy (HAART) was classified dichotomously and defined as any of the following combinations: three nucleoside reverse transcriptase inhibitors (NRTI), two NRTI and one or two protease inhibitors, two NRTI and one non-nucleoside reverse transcriptase inhibitor, or one NRTI and one protease inhibitor and one non-nucleoside reverse transcriptase inhibitor. Among men on HAART at any given visit, 91% were on a regimen that included a protease inhibitor. Baseline CD4 T cell counts and HIV virus load measurements were available from clinic records for all men; however, this information was only available for approximately one-half of follow-up visits for all men. Missing values were imputed.
Specimen collection and serological testing
At each visit, blood samples were drawn and the men deposited oral fluid samples into an empty tube. At some visits, semen, urine and rectal brush and swab samples were obtained from some men, but collection was discontinued because HHV-8 DNA was seldom detected in these samples . Samples were stored at 4°C until same-day delivery and processing, using a protocol described elsewhere . For each study visit, HHV-8 antibody testing was carried out with two enzyme-linked immunosorbent assays (ELISA) based on peptides from two viral open reading frames (ORF): ORF65 and K8.1 . Serum samples were diluted 1 : 100, and the assay cut-off for each ELISA was the mean corrected absorbance at 450 nm of the negative control specimens (from employees at the Centers for Disease Control and Prevention who had screened negative on a different HHV-8 assay) plus 5SD. Endpoint titers were obtained through two- or four-fold dilutions. To help to determine eligibility prior to enrollment, a monoclonal antibody-enhanced immunofluorescence assay (IFA) was used for detection of antibodies to lytic proteins, as described elsewhere , except sera were tested at a higher dilution (1 : 40). Men who were reactive in at least two of the three seroassays were eligible for enrollment.
DNA extraction and amplification
DNA was extracted by using Qiagen kits (Valencia, California, USA) and tested for the presence of HHV-8 DNA by one-step PCR ELISA targeting the ORF26 and glycoprotein B (ORF8) regions, as previously described . Each PCR amplification contained 600 ng total cellular DNA, and sensitivity was 10–20 copies/600 ng DNA. Samples were reported as positive if results from both primer pairs were positive. The sensitivity of the two primer pairs was not identical, and a small number of samples (15 of 644) were weakly positive by the more sensitive (ORF26) primer pair only. These samples were considered equivocal and were classified as negative for analysis purposes.
To identify statistically significant associations, Fisher's exact test, Pearson's correlation coefficient and the Wilcoxon rank-sum test were used where appropriate. All r values were Pearson product-moment correlation coefficients. Linear regression was used to impute missing CD4 T cell counts and HIV virus loads. The correlation of PCR results within individuals was assessed by counting for each patient all the positive visits (pi) that immediately followed a positive visit, and all the negative visits (ni) that followed a positive visit. The risk ratio (RR) was then computed as the sum of the pi values over all patients divided by the sum of the ni values over all patients. In analyses of the relationship between HHV-8 DNA and high antibody titer, generalized estimating equations were used to account for correlated data and control for KS status, HAART status, HIV virus load, CD4 T cell count and race.
In all, 87 men contributed 322 visits. Men with KS were similar to men without KS with respect to age (mean age 40 and 41 years, respectively; P = 0.23), white race (53% and 62%, respectively; P = 0.09) and past injection drug use (11% and 15%, respectively; P = 0.27). Factors that differed between men with and without KS included median HIV viral load (10 262 and 503 copies/ml, respectively; P < 0.001), median CD4 T cell count (92 and 310 × 106 cells/l, respectively; P < 0.001), anti-HHV-8 serum antibody titer levels (Table 1) and prevalences of HHV-8 DNA in oral fluid and PBMC (Table 1).
Detection of HHV-8 DNA and correlates
HHV-8 DNA was detected in oral fluid or PBMC of 63% (55/87) of study participants at one or more clinic visits (Table 1). Although HHV-8 DNA in PBMC was associated with HHV-8 DNA in oral fluid [adjusted odds ratio (OR), 2.8; 95% confidence interval (CI), 1.6–5.1; P < 0.001], many men were positive in only one of the two specimen types. HHV-8 DNA prevalence was higher in men with KS than in men without KS in both PBMC (37% versus 11%; P < 0.001) and oral fluid (31% versus 22%; P = 0.07) (Table 1). Significant associations between KS and HHV-8 DNA persisted in multivariate analyses (Table 2). As shown in Fig. 1, individuals who shed HHV-8 DNA in oral fluid or PBMC at one time point were more likely also to shed virus at other time points. Specifically, for individuals, a visit with a positive PCR result was approximately three times more likely to follow a visit with a positive result for both PBMC (RR, 3.4; 95% CI 1.7–6.2) and oral fluid (RR, 3.3; 95% CI 1.7–6.2).
Antibody titers and correlates for herpesvirus 8
The geometric mean antibody titer was 1766 for ORF65 and 1096 for K8.1. Very high antibody titers were not uncommon. For one or both of the ELISA, antibody titers were ≥ 25 600 for one-fifth of the specimens and for one-quarter of the men (at one or more visits) (Table 1). Geometric mean antibody titers were higher in men with KS than in those without KS for both ORF65 (2877 versus 690; P < 0.001) and K8.1 (1452 versus 740; P = 0.03). During the study, 36% (15/42) of men with KS had at least one ORF65 antibody titer ≥ 25 600, compared with 11% (5/45) of men without KS. Similarly, 26% (11/42) of men with KS had at least one K8.1 antibody titer ≥ 25 600 compared with 4% (2/45) of the men without KS. Correlation between K8.1 and ORF65 antibody titers was greater in men with KS (r = 0.43; P < 0.0001) than in men without KS (r = 0.29; P = 0.0001) (Fig. 2). Within individuals, between-visit changes in antibody titers were correlated (i.e., antibody titers for the two ELISA systems moved in agreement – both increasing, decreasing or not changing – twice as frequently as would be expected by chance alone).
Association between detection of herpesvirus 8 DNA and antibody titers
More men with KS than without KS had HHV-8 DNA in oral fluid or PBMC at one or more visits (71.4% versus 55.6%), and more men with KS than without KS had high ORF65 and K8.1 antibody titers. However, in multivariate analyses, there was a statistically significant inverse association between high ORF65 antibody titers (≥ 1 : 25 600) and HHV-8 DNA in PBMC (OR, 0.16; 95% CI, 0.05–0.51) and oral fluid (OR, 0.16; 95% CI, 0.05–0.51) (Table 2) and a moderate, non-significant inverse association between high K8.1 antibody titers (≥ 1 : 25 600) and the presence of HHV-8 DNA in PBMC (OR, 0.47; 95% CI, 0.14–1.6) and oral fluid (OR, 0.30; 95% CI, 0.08–1.1).
Among men seropositive for both HIV and HHV-8, our main findings were that: circulating HHV-8 was common; the strongest correlate of HHV-8 DNA in PBMC was the presence of KS; HHV-8 DNA detection in oral fluid and PBMC could be intermittent, but individuals who shed virus at one time point were more likely to shed at other times; there was incomplete epitope recognition in the anti-HHV-8 antibody response; and very high antibody titers, particularly for the ORF65 assay, were associated with the absence of HHV-8 DNA in oral fluid and PBMC.
More than two-thirds of men with KS and more than half of men without KS had detectable HHV-8 DNA in oral fluid or PBMC at some time during the study. These results are similar to those of Pauk et al. , who found evidence of HHV-8 DNA in at least one mucosal sample from 60% of the HHV-8-seropositive men without KS, suggesting that active viral replication is often ongoing even in the absence of disease. The frequent presence of HHV-8 DNA in oral fluid or PBMC of HHV-8-seropositive individuals, along with our related finding of a low concentration of PBMC containing high copy numbers of viral DNA (E. Martró, M. J. Cannon, S. C. Dollard, T. J. Spira, A. S. Laney, C.-Y. Ou, et al., unpublished data), support the hypothesis of Biggar et al.  that antibody titers increase over time as the result of ongoing, low levels of active viral replication.
Among men seropositive for both HIV and HHV-8, the presence of HHV-8 DNA in PBMC was strongly correlated with having KS. This confirms, with nearly four times as many observations, our previous finding  and is consistent with recent studies showing that the absence of HHV-8 DNA in blood is associated with KS regression  and the presence of HHV-8 DNA in PBMC predicts subsequent development of KS and may be used to identify HHV-8-seropositive individuals at high risk for KS . Such identification may be useful for KS prophylaxis or treatment with anti-herpes drugs .
Although HHV-8 DNA detection in oral fluid and PBMC was sometimes intermittent, men who shed virus at one time point were more likely also to shed at other time points. Intermittent detection might result if our stringent definition of PCR positivity (i.e., positive with two independent primer pairs) prevented the detection of true positives with low HHV-8 DNA copy numbers. A study that carried out multiple PCR reactions on each patient specimen, in order to test a much larger volume of blood than can be tested in a single PCR reaction, showed that HHV-8-infected individuals could have very low levels of HHV-8 in PBMC that would not be consistently detected using standard PCR sampling techniques (E. Martró, M. J. Cannon, S. C. Dollard, T. J. Spira, A. S. Laney, C.-Y. Ou, et al., unpublished data). Therefore, it seems probable that some of our men who had visits with HHV-8 DNA-positive results separated by visits with negative results (Fig. 1) in fact had circulating virus at all visits and that intermittent detection reflects the inability to detect HHV-8 rather than true loss of infection . Consequently, cross-sectional studies almost certainly underestimate the true prevalence of HHV-8 DNA in the oral fluid or PBMC of seropositive individuals.
The inability of the lytic antigen serological assays to identify all HHV-8-infected individuals may result from the absence of antigen-specific antibodies rather than the inability of the assays to detect low levels of antibodies. In several instances, men positive for HHV-8 DNA were consistently positive in one of the ELISA systems but not the other (Fig. 1). Furthermore, in many of these instances, the titers for the positive ELISA were too high to be reasonably attributed to a non-specific reaction (Fig. 2). These observations are similar to a previous report of incomplete recognition of a latent or lytic epitope following primary infection , but they add the observation that a particular lytic antigen may be recognized while another lytic antigen is not. Assays that target multiple epitopes will be needed to achieve maximal sensitivity in detecting infection and to improve the determination of the true prevalence of HHV-8 infection.
Circulating HHV-8 is linked to KS, and KS is linked to inadequate immune control; however, the presence of anti-HHV-8 immune markers does not always seem to correlate with the absence of circulating HHV-8 or KS disease. For example, mean HHV-8 antibody titers are higher in patients with KS than in HHV-8-seropositive patients without KS [25,26]. Furthermore, in some studies, CD4 and CD8 T cells and natural killer cells appeared ineffective in clearing HHV-8-infected cells from the circulation [27,28]. Based on these and other findings, some have hypothesized that the anti-HHV-8 immune response can actually encourage KS development by inducing the release of inflammatory cytokines, which, in turn, stimulate HHV-8 reactivation in circulating cells, resulting in viral spread and dissemination in tissues . In apparent concordance with this hypothesis, we found the prevalence of HHV-8 DNA in body fluids and the geometric mean HHV-8 antibody titers to be higher in the group with KS than in the group without KS. However, circulating HHV-8 DNA was not necessarily linked with high antibody titers within individuals. In fact, there was an antibody titer threshold above which HHV-8 DNA was detected less frequently than for lower antibody titers. In particular, individuals with ORF65 antibody titers ≥ 25 600 were six times less likely than those with lower titers to have HHV-8 DNA present in PBMC or oral fluid. Therefore, rather than encouraging KS development, ORF65 antibody or other immune function for which ORF65 antibody is a marker may limit HHV-8 replication when titers are very high. Although such an immune response may not stop KS from occurring (as indicated by the occurrence of KS in men with high antibody titers), it may impede the development of new KS lesions .
In conclusion, we have described several new findings about the natural history and pathogenesis of HHV-8 infection among HIV-infected men with and without KS. Further understanding will be gained from longitudinal studies that measure HHV-8 DNA quantitatively, examine other aspects of host immunity, including inflammatory cytokines and antibodies against latent antigens, and observe KS clinical progression over greater periods of follow-up.
We thank the staff of Grady Infectious Diseases Clinic for their cooperation and assistance, especially Susan Hogan, Jeffrey Lennox, Erika Patrick, and all of the personnel on the infectious disease research floor. From the Decatur Veterans Administration Hospital, we thank Laura Gallagher, Jodie Guest, and David Rimland. We also thank Padmaja Itikala, Laura Solomon, Stephanie Staras, Cynthia Stover Smith and Shanna Zepeda for data collection; Shenoa Rondeau for data entry; Mitesh Patel and Kay Radford for HHV-8 serology, specimen processing and PCR; Anne Li for specimen processing; the Emory University Rollins School of Public Health; and the Centers for Disease Control and Prevention Opportunistic Infections Working Group. We also thank the men who participated in the study.
Note: The results in this paper were presented in part at the Seventh International Conference on Malignancies in AIDS and other Immunodeficiencies: Basic, Epidemiologic and Clinical Research, Bethesda, April 2003 [abstract 3].
Sponsorship: This study was supported by the US CDC Opportunistic Infections Program.
1. Chang Y, Cesarman E, Pessin MS, Lee F, Culpepper J, Knowles DM, et al. Identification of herpesvirus-like DNA sequences in AIDS-associated Kaposi's sarcoma. Science
2. Moore PS, Chang Y. Kaposi's sarcoma-associated herpesvirus.
In: Fields Virology
, 4th edn, Vol. 2. Edited by Knipe DM, Howley PM, Griffin BE, Lamb RA, Martin MA, Roizman B, et al.
Philadelphia, PA: Lippincott Williams & Wilkins; 2001:2803–2833.
3. Corchero JL, Mar EC, Spira TJ, Pellett PE, Inoue N. Comparison of serologic assays for detection of antibodies against human herpesvirus 8. Clin Diagn Lab Immunol
4. Pellett PE, Wright DJ, Engels EA, Ablashi DV, Dollard SC, Forghani B, et al. Multicenter comparison of serologic assays and estimation of human herpesvirus 8 seroprevalence among US blood donors. Transfusion
5. Schatz O, Monini P, Bugarini R, Neipel F, Schulz TF, Andreoni M, et al. Kaposi's sarcoma-associated herpesvirus serology in Europe and Uganda: multicentre study with multiple and novel assays. J Med Virol
6. Chatlynne LG, Ablashi DV. Seroepidemiology of Kaposi's sarcoma-associated herpesvirus (KSHV). Semin Cancer Biol
7. Cannon MJ, Dollard SC, Smith DK, Klein RS, Schuman P, Rich JD, et al. Blood-borne and sexual transmission of human herpesvirus 8 in women with or at risk for human immunodeficiency virus infection. N Engl J Med
8. Martin JN, Ganem DE, Osmond DH, Page-Shafer KA, Macrae D, Kedes DH. Sexual transmission and the natural history of human herpesvirus 8 infection. N Engl J Med
9. Pauk J, Huang ML, Brodie SJ, Wald A, Koelle DM, Schacker T, et al. Mucosal shedding of human herpesvirus 8 in men. N Engl J Med
10. Plancoulaine S, Abel L, van Beveren M, Tregouet D-A, Joubert M, Tortevoye P, et al. Human herpesvirus 8 transmission from mother to child and between siblings in an endemic population. Lancet
11. Regamey N, Tamm M, Wernli M, Witschi A, Thiel G, Cathomas G, et al. Transmission of human herpesvirus 8 infection from renal-transplant donors to recipients. N Engl J Med
12. Gao SJ, Kingsley L, Hoover DR, Spira TJ, Rinaldo CR, Saah A, et al. Seroconversion to antibodies against Kaposi's sarcoma- associated herpesvirus-related latent nuclear antigens before the development of Kaposi's sarcoma. N Engl J Med
13. Moore PS, Kingsley LA, Holmberg SD, Spira T, Gupta P, Hoover DR, et al. Kaposi's sarcoma-associated herpesvirus infection prior to onset of Kaposi's sarcoma. AIDS
14. Whitby D, Howard MR, Tenant-Flowers M, Brink NS, Copas A, Boshoff C, et al. Detection of Kaposi sarcoma associated herpesvirus in peripheral blood of HIV-infected individuals and progression to Kaposi's sarcoma. Lancet
15. Cannon MJ, Dollard SC, Black JB, Edlin BR, Hannah C, Hogan SE, et al. Risk factors for Kaposi's sarcoma in men seropositive for both human herpesvirus 8 and human immunodeficiency virus. AIDS
16. Engels EA, Biggar RJ, Marshall VA, Walters MA, Gamache CJ, Whitby D, et al. Detection and quantification of Kaposi's sarcoma-associated herpesvirus to predict AIDS-associated Kaposi's sarcoma. AIDS
17. Sarid R, Olsen SJ, Moore PS. Kaposi's sarcoma-associated herpesvirus: epidemiology, virology, and molecular biology. Adv Virus Res
18. Cannon MJ, Laney AS, Pellett PE. Human herpesvirus 8: current issues. Clin Infect Dis
19. Spira TJ, Lam L, Dollard SC, Meng Y-X, Pau C-P, Black JB, et al. Comparison of serologic assays and PCR for the diagnosis of human herpesvirus 8 (HHV-8) infection. J Clin Microbiol
20. Lennette ET, Blackbourn DJ, Levy JA. Antibodies to human herpesvirus type 8 in the general population and in Kaposi's sarcoma patients. Lancet
21. Stamey FR, DeLeon-Carnes M, Patel MM, Pellett PE, Dollard SC. Comparison of a microtiter plate system to Southern blot for detection of human herpesvirus 8 DNA amplified from blood and saliva. J Virol Meth
22. Biggar RJ, Engels EA, Whitby D, Kedes DH, Goedert JJ. Antibody reactivity to latent and lytic antigens to human herpesvirus-8 in longitudinally followed homosexual men. J Infect Dis
23. Gill J, Bourboulia D, Wilkinson J, Hayes P, Cope A, Marcelin AG, et al. Prospective study of the effects of antiretroviral therapy on Kaposi sarcoma-associated herpesvirus infection in patients with and without Kaposi sarcoma. J Acquir Immune Defic Syndr
24. Dupon M, Masquelier B, Cazorla C, Chene G, Dumon B, Ragnaud JM, et al. Acquired immunodeficiency syndrome- associated Kaposi's sarcoma and human herpesvirus 8 DNA detection in serial peripheral blood mononuclear cell samples. Res Virol
25. Sitas F, Carrara H, Beral V, Newton R, Reeves G, Bull D, et al. Antibodies against human herpesvirus 8 in black South African patients with cancer. N Engl J Med
26. Chatlynne LG, Lapps W, Handy M, Huang YQ, Masood R, Hamilton AS, et al. Detection and titration of human herpesvirus-8-specific antibodies in sera from blood donors, acquired immunodeficiency syndrome patients, and Kaposi's sarcoma patients using a whole virus enzyme-linked immunosorbent assay. Blood
27. Ensoli B, Sturzl M. HHV-8 and multistep tumorigenesis. Trends Microbiol
28. Sirianni MC, Vincenzi L, Topino S, Giovannetti A, Mazzetta F, Libi F, et al. NK cell activity controls human herpesvirus 8 latent infection and is restored upon highly active antiretroviral therapy in AIDS patients with regressing Kaposi's sarcoma. Eur J Immunol
29. Ensoli B, Sturzl M, Monini P. Reactivation and role of HHV-8 in Kaposi's sarcoma initiation. Adv Cancer Res
Kaposi's sarcoma; herpesvirus; epidemiology; opportunistic infections; risk factors; homosexual men
© 2004 Lippincott Williams & Wilkins, Inc.
Highlight selected keywords in the article text.