We measured levels of six cytokines and human immunodeficiency virus (HIV) RNA in saliva from HIV-seropositive individuals and compared salivary cytokine levels in HIV-seropositives and seronegatives. All of the six tested cytokines were detected in saliva although interleukin-1β, interferon-γ and interleukin-10 were detected more frequently (90%, 68% and 61% of samples, respectively) than interleukin-6, tumor necrosis factor-α and tumor necrosis factor-α receptor II (2–17%). There was no significant association between cytokine levels in saliva and plasma suggesting that cytokines were produced locally. Interferon-γ levels were significantly higher in saliva from HIV-seropositives when compared to seronegatives while interleukin-10 levels were lower in seropositive saliva. Interleukin-10 levels were higher in individuals with low CD4 counts in the seropositive group. HIV RNA was detected in 29% of saliva samples from seropositives and there was a significant correlation between saliva and plasma HIV RNA levels. However, HIV RNA levels in saliva were not significantly associated with any of the saliva or plasma cytokine levels or with CD4 cell numbers. This study shows no association between inflammatory cytokine levels and HIV levels in saliva and suggests that saliva HIV levels are more influenced by blood HIV RNA levels than oral inflammation.
Human immunodeficiency virus (HIV) RNA can be detected in the saliva of HIV-infected individuals however infectious virus is rarely cultured due to relatively low levels of the virus and the presence of substances in saliva that inactivate HIV [1–5]. Epidemiologic studies suggest that transmission of HIV through oral contact occurs much less frequently than through genital contact although some cases of oral transmission have been documented [6,7]. HIV can be detected in paired samples of saliva and blood from a significant proportion of donors  suggesting that HIV levels in saliva may, to some extent, be a reflection of blood levels. Similar to what is observed with HIV RNA in plasma, higher levels of HIV RNA are detected in saliva during primary infection and treatment with anti-retroviral drugs reduces saliva HIV RNA levels [9,10]. However, some individuals express higher levels of HIV RNA in saliva than blood suggesting that HIV can be produced locally in the oral cavity and may be influenced by oral inflammation . Studies of HIV RNA levels in the genital tract of women and men show that this mucosal site generally reflects the amount of virus in plasma although menstruation, infections and inflammation can increase genital tract virus levels without affecting plasma virus levels [12–19]. Those studies suggest that it is possible that saliva RNA levels are influenced by plasma virus levels and/or oral inflammation although those relationships have not been assessed.
Cytokines such as interleukin-1 (IL-1) and tumor necrosis factor-α (TNF-α) are known to increase HIV expression by infected cells  and therefore it is possible that local secretion of proinflammatory cytokines could affect expression of HIV in the oral cavity. In fact, inflammation of the gingiva is associated with higher salivary HIV levels . However, the relationship between oral HIV levels and oral cytokines has not been reported.
Mucosal surfaces such as the oral cavity are important sites for immune responses and saliva has been studied for the presence of cytokines since the presence of these soluble mediators in saliva could influence or be indicative of inflammatory reactions or specific immune responses to infection. Oral candidiasis is associated with increased levels of interferon-γ (IFN-γ), IL-2 and IL-5 in saliva . The T helper (TH) cytokine profiles in saliva have also been investigated and both TH1-type (IL-12, IFN-γ and IL-2) and TH2-type (IL-4 and IL-10) cytokines can be detected in saliva although HIV-seropositive subjects have a profile skewed more toward TH2 than seronegative subjects . Factors such as smoking by donors can affect cytokine levels in saliva . The effect of menstrual cycle on levels of salivary cytokines was also measured . While IL-1β, IL-8, IFN-γ and TNF-α were detectable in the saliva of women; there was no influence of the phase of the menstrual cycle on these cytokines. In contrast, plasma IL-8 was higher during the follicular phase than the luteal phase.
As outlined above, HIV RNA detection in saliva correlates with plasma virus RNA detection suggesting that oral HIV RNA levels may be reflective of virus in blood although other studies suggest that inflammation may affect HIV levels in saliva. Since cytokines, including inflammatory cytokines, have been detected in saliva, the main goal of this study was to determine if HIV levels in saliva were associated with levels of either cytokines or plasma HIV RNA. We also compared levels of cytokines in saliva from HIV-seronegative with seropositive individuals to determine if HIV disease or oral disease affects oral cytokines and further assessed whether cytokine detection and levels were similar between saliva and plasma to determine if cytokines had an oral origin.
2 Materials and methods
2.1 Patients and samples
Saliva and plasma samples were collected from volunteers enrolled in the Oral Substudy of the Women's Interagency HIV Study (WIHS) at the Oral Substudy site at the University of Illinois at Chicago College of Dentistry Ryan White Dental Clinic in accordance with the human investigation committee. WIHS is a prospective epidemiological and natural history study of HIV-infected and high risk uninfected women. Informed consent was obtained from all participants.
In preparation for the single collection of unstimulated saliva by the draining method, the subjects were asked to remove both lipstick with a piece of gauze and any removable dental prosthesis. The mouth was rinsed thoroughly with deionized water and the donor rested for 5 min before sample collection (no talking or reading). The head was tilted forward, the mouth opened and the lower lip placed over the rim of a plastic funnel that emptied into a container in ice. The saliva during the first minute was discarded and saliva collected over the next 5 min was used for the studies. Saliva was aliquoted and frozen at +70 °C until the time of assay.
2.2 Laboratory tests
Cytokines (IL-1β, interferon-γ, IL-10, IL-6, TNF-α and TNF-α Receptor II, [TNFRII]) were detected using the enzyme-linked immunosorbent assay (ELISA, Endogen brand, Perbio Science, Helsingborg, Sweden). RNA in saliva (0.5 ml) was extracted using the Boom method  and HIV RNA was measured using the Roche Amplicor HIV-1 Monitor test (Nutley, NJ). Plasma HIV RNA was detected in 0.2 ml plasma using the NASBA/Nuclisens kit (Bio Merieux Organon Teknika, Netherlands).
2.3 Statistical analysis
Chi-square analysis was performed to assess significance of the frequency of detection of cytokines in saliva and plasma. The Wilcoxon rank-sum test was used to determine if seropositivity affected detection of cytokines. The Spearman rank correlation was used to determine the association of HIV RNA in saliva with plasma HIV RNA and cytokines. If nothing was detected in samples (cytokines or RNA), the detection limit was assigned to that sample for statistical analysis. The lower limit of detection for cytokines (in pg ml+1) was IL-1β= 1, IFN-γ= 4, IL-6 = 2, IL-10 = 5, TNF-α= 1 and TNFRII = 2; for HIV RNA (in RNA copies ml+1) saliva = 100; plasma = 4000.
3.1 Cytokines in saliva
Levels of cytokines in saliva and plasma were determined for a total of 88 women consisting of four groups; HIV-seronegative individuals (n= 18); HIV-seropositive individuals with absolute CD4 counts of <200 cells mm+3 (n= 10); seropositives with CD4 counts of 200–500 cells mm+3 (n= 30); and seropositives with >500 CD4 cells ml+1 (n= 30). Of the seropositive patients, 30 were on no anti-retroviral therapy (mean ± SD plasma HIV RNA = 24,844 ± 41,412), five were on monotherapy (55,940 ± 98,331), 24 were receiving combination therapy (25,878 ± 68,437), and 11 were on highly active anti-retroviral therapy (HAART; 8694 ± 14,936). There were no significant differences in plasma or saliva virus loads between the treatment groups (p > 0.05, Kruskal Wallis test). The ages of the women ranged from 26 to 58 and were also not significantly different between the four groups. IL-1β, IL-6, TNF-α and TNFRII in saliva were assessed since they have been associated with inflammation and/or HIV disease severity and TNFRII has been reported to be more stable in secretions than TNF-α[26–30]. Interferon-γ and IL-10 were assessed because they are associated with type-1 and type-2 immune responses, respectively.
Of all assayed cytokines, IL-1β was most frequently detected in saliva samples (79/88 samples >1 pg ml+1, Fig. 1). Interferon-γ was detected in 60/88 (>4 pg ml+1) samples while IL-10 was detected in 54/88 samples (>5 pg ml+1). The other three cytokines were detected in saliva at relatively low frequencies; IL-6 was detected in 15 samples (>2 pg ml+1), TNFRII was detected in 2 (>2 pg ml+1), and TNF-α in 6 (>1 pg ml+1). Interestingly, while TNFRII was rarely detected in saliva, it was the most frequently detected cytokine in plasma samples (64/88). The other cytokines were detected in plasma at the following frequencies IL-1β= 46/88; IL-10 = 25/88; interferon-γ= 2/88; IL-6 = 4/88; and TNF-α= 7/88. Based on these frequencies, IL-1β, interferon-γ, IL-10 and IL-6 were all detected significantly more frequently in saliva than plasma while TNFRII was detected more frequently in plasma than in saliva (p < 0.05, Chi-square analysis, Table 1). Detection of IL-1β or IL-10 in saliva from individuals was not significantly associated with detection of those two cytokines in plasma from the same individuals (p= 0.33 and 0.5, respectively, Chi-square analysis). For the other four cytokines, an association of saliva and plasma detection could not be assessed since cytokines in either one or both of the compartments were too infrequently detected. Overall, the concentrations of IL-1β, interferon-γ and IL-10 were significantly higher in saliva than in plasma while conversely TNFRII was significantly higher in plasma than in saliva (all p < 0.001, Mann–Whitney U test).
Cytokine levels in saliva and plasma. Cytokines in saliva (S) and plasma (P) were assayed for HIV-seronegative individuals (seroneg., n= 18) or HIV-seropositive individuals with CD4 counts of <200 cell mm+3 (n= 10), 200–500 cells mm+3 (n= 30) or >500 cells mm+3 (n= 30). The lower limit of detection for the cytokines (in pg ml+1) was IL-1β= 1, IFN-γ= 4, IL-6 = 2, IL-10 = 5, TNF-α= 1 and TNFRII = 2. Each box encloses the second and third quartile of the observations with the median value of the variable displayed as a line. The lines extending from the top and bottom of each box mark the minimum and maximum values.
↵TNRII levels were higher in subjects with CD4 <200 vs. other seropositive groups.
For the cytokines that were detected in more than 30% of the saliva samples, cytokine level association with HIV seropositivity was assessed (Wilcoxon rank-sum test). Interferon-γ levels were significantly higher in the saliva of seropositives than in HIV-seronegative donors (p= 0.02). Conversely, IL-10 was significantly higher in the saliva and plasma of seronegatives than in seropositives (p= 0.0004 and 0.0003, respectively). IL-1β in saliva or plasma was not significantly different between the two groups (p= 0.08 and 0.9, respectively). TNFRII levels in plasma were also higher in seropositives than in seronegatives (p= 0.000006) as has been previously reported . When the ratio of interferon-γ to IL-10 in saliva was calculated, there was a significant difference between seronegatives and seropositives (p= 0.001) with the two having ratios of 0.7 and 2.3, respectively.
Associations between the three most frequently detected saliva cytokines were also calculated. IL-1β levels were significantly associated with interferon-γ levels and IL-10 levels while interferon-γ and IL-10 levels were also significantly associated (p < 0.001, Spearman rank correlation).
When the CD4 count of HIV-seropositives was analyzed for association with cytokine level, TNFRII in plasma was significantly different with the highest levels in the group with the lowest CD4 count (p < 0.0001). IL-10 in saliva was marginally associated with CD4 cell count (p= 0.047). In contrast, IL-1β in plasma or serum, IL-10 in plasma and interferon-γ in saliva were not significantly associated with CD4 cell count (p > 0.5). The ratio of interferon-γ to IL-10 in saliva was also not significantly different among the three seropositive CD4 groups.
3.2 HIV levels in saliva and plasma
HIV RNA was detected (>100 RNA copies ml+1) in 20/70 saliva samples from seropositive individuals while it was detected in 32 of the plasma samples (p= 0.01, Chi-square, Fig. 2). There was a significant correlation between the levels of virus in saliva and plasma (p= 0.001, Spearman rank correlation, Fig. 2(b)). However, there were no significant associations between virus load in the saliva or plasma with any of the saliva or plasma cytokine levels, ratio of interferon-γ to IL-10 or CD4 cell count. Additionally, neither the virus load in saliva or the cytokine levels in saliva were significantly associated with several indicators of oral inflammation including gingival bleeding (number of teeth with bleeding, range 0–15, mean ± SD = 1.7 ± 2.9), number of lesions in the mouth (range 0–3, mean ± SD = 0.4 ± 0.5) or loss of attachment (pockets of 4 or more mm of depth, range 0–126, mean ± SD = 53 ± 34).
HIV RNA levels in saliva and plasma. (a) HIV RNA levels (RNA copies ml+1) were determined in HIV-seropositive individuals with CD4 counts of <200 cell mm+3 (n= 10), 200–500 cells mm+3 (n= 30) or >500 cells mm+3 (n= 30). Each box encloses the second and third quartile of the observations with the median value of the variable displayed as a line. The lines extending from the top and bottom of each box mark the minimum and maximum values. The graph includes data from all 70 patients. (b) Graph of HIV RNA copies in plasma versus HIV RNA copies in saliva. The regression line is shown.
The aim of this study was to determine if saliva HIV RNA levels were associated with either inflammatory cytokines or plasma HIV RNA levels. While a previous study showed a significant correlation between saliva and plasma samples that were positive for detection of HIV RNA , this is the first study to show that plasma HIV RNA levels were significantly correlated with saliva HIV RNA levels. It should be noted that some of the subjects were treated with HAART. However, no significant correlations were noted between cytokine levels in saliva and saliva HIV RNA levels. This was somewhat unexpected since Shugars et al.  showed that 5 out of 67 HIV-positive subjects had oral hyper-excretion of HIV RNA which was defined as a fourfold or higher viral load in saliva than in plasma. The five hyper-excretors had elevated HIV-associated periodontal disease and gingival inflammation suggesting that increased inflammatory cytokines may have also been present orally, although that was not assessed. In the current study, gingival bleeding, mouth ulceration and gingival pockets were not associated with saliva HIV RNA levels. In the current study however, five subjects with possible discordance between saliva and plasma HIV RNA levels were noted with undetectable plasma HIV RNA, but with between 211 and 1741 HIV RNA copies per milliliter of saliva. Nevertheless, notably increased oral levels of any of the six cytokines or oral inflammation markers were not observed for those five subjects (data not shown). The lack of an ability to observe significant correlations between cytokines and virus levels may have been due to a confounding effect of anti-retrovirus treatment of some patients. Thus, inflammation may not be able to increase oral virus production in the presence of HAART. However, there was no difference in plasma virus levels between HAART and untreated patients although “effective” treatment and time of treatment were not considered. This lack of a difference may have been due to the fact that the group with high CD4 numbers had relatively low plasma virus levels (see Section 3) and was treated with HAART at a lower frequency.
A number of published studies that have analyzed salivary cytokines have reported values standardized for salivary protein levels (e.g. ). Protein standardization is performed since saliva protein content can as much as double due to dehydration while flow rate decreases accordingly  and age also has effects on saliva flow and protein concentration . In contrast, most studies of HIV in body fluids express virus levels as RNA copies ml+1[8,11]. We chose to express all cytokine and virus levels adjusted to volume so that we could compare virus levels between fluids and with previous studies and for the additional reason that if the saliva and plasma cytokines were both adjusted based on protein content, then the very high levels of albumin in plasma would have made the plasma cytokines appear to be very low compared to saliva. Therefore, a limitation of this study is the inability to effectively compare the results to other studies that have data normalized cytokines to protein levels. Another possible limitation of this study is that factors in saliva that could potentially interfere with measurement of cytokines were not assessed.
An important implication of the finding that the level of HIV RNA in saliva is associated with plasma HIV levels but not local cytokine levels is that when blood HIV levels are highest, the levels of infectious HIV in saliva would be highest, although infectivity was not assessed in this study. Therefore, this study would suggest an increase in transmission risk associated with high plasma HIV levels. However, many studies show little or no transmission of HIV due to exposure to saliva containing HIV [7,9,33]. This may imply that anti-viral factors in saliva inactivate HIV and reduce transmission risk as has been suggested .
Another goal of this study was to assess the effect of HIV-seropositivity and CD4 level on oral cytokines. Interferon-γ was significantly higher, while IL-10 was significantly lower in the saliva of seropositives than seronegatives. A previous study  showed that saliva from HIV-infected subjects with oral candidiasis had higher interferon-γ levels than either HIV-infected individuals or healthy controls. However, there was no significant difference in levels of saliva interferon-γ between HIV-infected subjects with no oral infections and healthy controls. This suggests that it is possible that the significant difference between seropositive and seronegative saliva interferon-γ levels seen in our study could have been due to oral infections in the seropositive group. Leigh et al.  found that healthy seronegative individuals had both interferon-γ and IL-10 detected in saliva while HIV-infected persons had a more TH2-type oral cytokine mixture due mostly to decreased interferon-γ. These results do not correspond with the results in the current study. The difference between the studies may have been due to differing numbers of saliva donors with candidiasis in the HIV-positive group since it was reported that those patients had a tendency to have higher IL-10 levels. Further, different frequencies of smokers in the seronegative or seropositive groups in either study could contribute to differences in the finding of the two studies since smoking was found to influence cytokine levels .
This study also assessed the relationship between saliva and plasma cytokine levels. Interestingly, we found that IL-1β, interferon-γ, IL-10 and IL-6 were all detected significantly more frequently and found at significantly higher levels in saliva than plasma while TNFRII was detected more frequently and at higher levels in plasma than saliva. It is possible that the higher concentrations of IL-1β, interferon-γ and IL-10 in saliva could reflect local production of these cytokines. The higher levels of IL-1β, interferon-γ, IL-10 in saliva may be due to local stimulation or inflammation and possibly contribute to inflammation. However, we observed no correlation between salivary cytokines and markers of inflammation. There was no significant association between any of the cytokine levels in saliva and plasma. This lack of correlation may also imply that the salivary cytokines measured in this study are produced locally rather than derived from blood. However, part of the reason for this lack of correlation is likely to be that only IL-1β and IL-10 were detected at high enough frequencies in both fluids to determine if their detection was related.
In summary, this study shows that the cytokines interferon-γ and IL-10 are differentially expressed in HIV-seropositives versus-seronegatives and that the level of HIV RNA in saliva is associated with plasma HIV levels but not local cytokine levels. Thus, these findings provide important information on the risk of oral transmission of HIV as well as the immune status of the oral cavity during HIV infection.
This work was supported by National Institutes of Health grant P01HD40539. Thanks to Yvonne Albert for excellent technical work and Larry Thomas for helpful discussions. Some of the data in this manuscript were collected by the Women's Interagency HIV Study (WIHS) Collaborative Study Group with centers (Principal Investigators) at New York City/Bronx Consortium (Kathryn Anastos); Brooklyn, NY (Howard Minkoff); Washington DC Metropolitan Consortium (Mary Young); The Connie Wofsy Study Consortium of Northern California (Ruth Greenblatt); Los Angeles County/Southern California Consortium (Alexandra Levine); Chicago Consortium (Mardge Cohen); Data Coordinating Center (Alvaro Mu~noz). The WIHS is funded by the National Institute of Allergy and Infectious Diseases, with supplemental funding from the National Cancer Institute, the National Institute of Child Health & Human Development, The National Institute on Drug Abuse, and the National Institute of Craniofacial and Dental Research. U01-AI-35004, U01-AI-31834, U01-AI-34994, U01-AI-34989, U01-HD-32632, U01-AI-34993, U01-AI-42590, M01-RR00079 and M01-RR00083.
(2003) Genital tract and plasma human immunodeficiency virus viral load throughout the menstrual cycle in women who are infected with ovulatory human immunodeficiency virus. Am. J. Obstet. Gynecol. 188, 122–128.
(1998) Th1/Th2 cytokine expression in saliva of HIV-positive and HIV-negative individuals: a pilot study in HIV-positive individuals with oropharyngeal candidiasis. J. Acquir. Immune Defic. Syndr. Hum. Retrovirol. 19, 373–380.
(1998) The effect of acute infectious illnesses on plasma human immunodeficiency virus (HIV) type 1 load and the expression of serologic markers of immune activation among HIV-infected adults. J. Infect. Dis. 178, 1642–1648.