Findings

One hundred and forty-two clinical samples belonging to 49 patients with epidemic KS histologically diagnosed in the Pathology Department at the Institute of Tropical Medicine Pedro Kourí between 2004-2007 were included. The research was approved by local and national ethics committees; all subjects provided their written informed consent. Clinical, immunological and epidemiological data from each patient are depicted in Table 1.

Table 1 Demographical, epidemiological and clinical characteristic of the study population
Full size table

Different types of samples (41 saliva, 48 tissues, 26 PBMC, and 27 plasma) obtained from each individual were tested at the same time. PBMC and plasma were obtained by separation of 20 mL of citrated whole blood using a Ficoll separation gradient (SIGMA, UK). Paraffin was removed from tissues by xylene treatment according to published protocols [1]. DNA extraction was performed using the QIAamp® DNA Mini Kit (QIAGEN, Germany) and the genomic DNA (gDNA) concentration was determined using spectrophotometer (GeneQuant II, Pharmacia Biotech, EUA) and adjusted to 100 ng, with the exception of plasma where 10 uL were directly used since it was not possible to quantify the gDNA. In order to obtain the standard DNA for absolute quantification, gDNA was extracted from the BCBL-1 cell line and the DNA copy numbers was determined by spectrophotometer as well. Then, the OD concentration was converted to DNA copy number following methods published elsewhere [2]. Once the gDNA copy number from BCBL-1 cell line was calculated, it was adjusted to 1 million copies and a standard curve with serial ten-fold dilutions (from 106 to 10 copies) assayed in triplicates was prepared. Once the run finished, the standard curve was automatically generated by the LightCycler software version 3.3, using the Second Derivative Maximum (SDMM) method. In additon, the detection of Human β globin gene was used as internal control in each clinical sample (with exception of plasma) for measuring of the exact amount of input DNA [3].

RT-PCR primers and conditions were described previously by Watzinger et al [4] with minor modifications adapted for the LightCycler 1.5 [4]. Samples were considered negative if the Ct value exceeded cycle 40, or if the copy number was below 10 copies. All patients or samples with more than 10 copies were considered positive, thus infected with HHV-8. For statistical analysis SPSS 11,5 (Inc. SPSS, Chicago, IL, the USA) and Statgraphic were used. Tests for comparison of proportions between the averages were performed using the Chi-square test and the Pearson coefficient of correlation for 95% confidentiality.

To our knowledge, this is the first report, where HHV-8 viral load has been simultaneously determined in four different fluids and cells (affected tissue, saliva, plasma and PBMC) from the same AIDS-KS patients. Forty-eight (98%) of the 49 AIDS-KS patients had detectable levels of HHV-8 DNA at least in one of the samples studied, (67% of patients had more than 2 samples positives), with the virus being more frequently detected in KS lesions (44 tissues, 91.6%) followed by saliva in 78% and PBMC in 69.2%. The detection probability and viral load being significantly lower in plasma (37%). There was a positive correlation between the detection of HHV-8 in tissue and the detection in saliva or PBMC (p < 0.01) that was not observed for plasma (Table 2). Although PCR inhibitions could be a possible explanation for the 4 negative results from KS lesions, several other factors that may have limited KSHV detection, like lesion sampling, since the lesion sample was not always the same for molecular detection and diagnosis. In addition, a possible histological inaccuracy could not be excluded because some patients were diagnosed as early stage of KS [5]. However, the percentage of KSHV DNA detected in the present study is similar to previous results published by Mendez and colleagues [6] and higher than the 87% of KSHV DNA detection reported by Kennedy and co-authors [7].

Table 2 KSHV load levels in different fluid and cells from each KS patient.
Full size table

All fresh frozen tissues, PBMC and saliva samples, were confirmed to have the same amount of the input gDNA in100 ng, by detecting similar crossing point (Cp) amplification signal of the Human β globin gene among them (average Cp: 24.08, range: 22-25); in contrast, paraffin-embedded tissues showed a gDNA amplification signal after cicle 29 (range:27-32), probably due to degradation [8]. This result was consistent with the detection of HHV-8 load significantly higher in fresh frozen tissue than in paraffin-embedded tissue (p < 0.003) (Table 2).

Despite the difference detected between fresh frozen and paraffin-embedded tissue, when copy numbers from different samples (tissue-plasma, tissue-saliva, tissue-PBMC, plasma-saliva, PMBC-saliva, and PBMC-plasma) within a given individual were compared, a significantly higher HHV-8 viral load in KS lesions compared to all the other samples (p < 0.05) was detected (Figure 1). One of the reasons that may explain why the higher HHV-8 load is detected in tissue, specifically in the case of HIV infection, is that the HIV tat protein promotes the replication of HHV-8-stimulated cell proliferation and inhibits apoptosis of spindle cells infected with HHV-8 [9, 10]. Recent reports have described a higher HHV-8 load of AIDS-KS lesions compared with endemic KS, although the differences were not statistically significant.

Figure 1
figure 1

Comparison of the logarithmic KSHV load detected in different samples of epidemic KS patient using the RT-PCR assay of a conserved region of ORF-26.

Full size image

HHV-8 load in saliva and PBMC was significantly lower than in tissue (p < 0.05) and significantly higher than in plasma (p < 0.05; Figure 1), however there were no statistic differences between the viral load in saliva and PBMC (p > 0.05). Of note, saliva represented the second most frequent source for HHV-8 detection (Figure 1). In the light of previous reports [11–13] and consistent with the present results transmission by saliva may contribute to the spread of HHV-8 infection among the HIV seropositive population besides sexual intercourse [14]. It still remains controversial why, if saliva is the main source of virus, HHV-8 infection shows a sexual pattern of transmission. Nevertheless, saliva is the only mucosal fluid in which infectious HHV-8 has been identified, although factors associated with HHV-8 salivary shedding remain unclear [15]. Moreover, future research will need to assess the possible impact of saliva in HHV-8 transmission among Cuban HIV seronegative individuals.

Most authors agree that the viral genome present in PBMC of AIDS-KS patients occurs in a latent state, but the role of HHV-8 persistence in PBMC with regard to the pathogenesis of KS remains unclear [16]. The lower copy number of HHV-8 detected in PBMC in comparison to KS tissue, could be due to its latent state within this compartment, in contrast to viral expression present in lesional spindle cells [17].

Harrington and colleagues demonstrated that the presence of HHV-8 in plasma is intermittent [18]. For other viral infections, plasma viral load has been used as the best marker to estimate disease progression [19–21]. However, this does not seem to be true for KS and HHV-8, as has been proposed by Polstra and colleagues [22]. Hundred percent of Cuban AIDS-KS patients, in whom HHV-8 was detected in plasma also tested positive in other clinical samples (saliva, KS tissue, PBMC). Thus, viremia could occur in those patients with active viral replication, alternating with latency periods where the virus is not detected in plasma, as has been reported for other members of the Herpesviridae family [23].

There was a significant Pearson's correlation when each patient's tissue HHV-8 levels were compared with saliva and PBMC (p < 0.01); however, this correlation did not exist for plasma samples (Figure 2). We did not find association between the HHV-8 viral load in all the samples analyzed with: CD4 cell counts, HIV viral load, type of KS (cutaneous, mucocutaneous and systemic), the histological stage (macular, patch and tumoral) nor the number of years elapsed between the KS diagnosis and death (p > 0.05).

Figure 2
figure 2

Graphic representation of KSHV loads in different samples per each AIDS-KS patient.

Full size image

In our study, Caucasian AIDS-KS patients were more likely to be positive for HHV-8 infection than black or mulatto KS patients (p = 0.006). There is no previous report regarding the association of HHV-8 levels with skin color. However, it has been described several times that AIDS-KS is more frequent in individuals with white skin [24]. Some authors have proposed a genetic predisposition to HHV-8 infection in individuals with certain HLA types (A*6801 and DRB1*04) that tend to have increased viral excretion in saliva e.g. in African women [25]. MSM were more likely to have three-times higher HHV-8 genome copies in KS lesions when compared with tissues from heterosexuals individuals (OR 3; 95% CI 1.1 to 12.5). This seems to be a very important finding that confirms what has been reported by other authors regarding sexual risk behaviors that are more prevalent in homosexual intercourse [26]. With the present study we confirm that HHV-8 produces a systemic infection in different body compartments and that the highest HHV-8 levels were produced in lesional KS tissue. Furthermore, saliva has been recognized as an important reservoir for HHV-8 transmission.

One of the main challenges for researchers is to identify the exact mode of virus transmission following sexual intercourse, as the presence of HHV-8 in semen, anal secretions or vaginal fluid has not been consistently demonstrated [27–30].