Neil Renwick1, Thomas Schulz2 and Jaap Goudsmit1
Kaposi's sarcoma (KS) is a vascular tumor that was brought to the attention of the medical community over a century ago in a fascinating case series in which purple-coloured nodular skin lesions were observed on five elderly men, with widespread cutaneous and visceral involvement reported in one patient at autopsy (Kaposi, 1872 as cited in Ober, 1988). This 'Classic' variant of KS is rare and the majority of cases are found in elderly Mediterranean men and Jewish people born in Eastern Europe (Rothman, 1962). Large-scale epidemiological studies have highlighted three other variants of KS; 'African' or 'Endemic' KS in young black adults and children in equatorial Africa (Davies and Lothe, 1962; Oettle, 1962), 'Iatrogenic' or 'Posttransplant' KS in patients who have previously received immunosuppressive therapy or in those who are organ transplant recipients (Klepp et al., 1978; Harwood et al., 1979; Penn, 1979) and 'AIDS-related' or 'Epidemic' KS, first noted in immunocompromised homosexual men from New York and California (Friedman-Kien et al., 1981). All four forms of KS have a predilection for males and the male:female ratio ranges from 2.3:1 in Iatrogenic KS to 106:1 in Epidemic KS and the clinical course of epidemic/AIDS-KS is more aggressive (Friedman-Kien and Saltzman, 1990). The strikingly unusual pattern of KS distribution begs the question as to what features, if any, are common to the four epidemiological variants of KS.
A major advance in our comprehension of KS epidemiology came from studying the distribution of 13616 KS cases among 90990 persons with AIDS reported to the Centers for Disease Control, Atlanta, until March 31, 1989. Detection of KS in 21% of males who acquired HIV-1 through homosexual or bisexual contacts in comparison with <7% in all other HIV transmission groups such as heterosexuals of Caribbean, African and other origin, intravenous drug users, transfusion recipients and persons with haemophilia suggested that epidemic KS might be a sexually transmitted infection (Beral et al., 1990). Similar conclusions for the existence of a separate KS agent as the cause of Kaposi's sarcoma in the context of HIV-1 infection were drawn from a Canadian cohort study (Schechter et al., 1991).
Discovery and classification of KSHV in the rhadinovirus lineage of the gammaherpesvirinae
Given that there may be a separate KS agent, it was reasoned that the genomic difference between proposed infected (KS) and uninfected (skin) tissues from an individual patient with AIDS-KS should be that of the infectious agent. Two herpesvirus-like DNA sequences, named KS330Bam and KS631Bam, were discovered using the PCR-based technique, Representational Difference Analysis, that preferentially identifies these differences. It was possible to amplify the KS330233 sequence by PCR from DNA extracted from 25/27 (93%) AIDS-KS tissues compared with 6/39 (15%) lymph nodes and lymphomas from AIDS patients without KS (X2=38.2, p<10-6), suggesting that this Kaposi's sarcoma-associated herpesvirus (KSHV) might be the proposed KS agent (Chang et al., 1994). The associations between KSHV (also known as Human Herpesvirus 8 (HHV8) for taxonomic purposes) and Kaposi's sarcoma among other diseases are discussed in more detail below.
Preliminary phylogenetic analysis of a 20.7-kb fragment from an AIDS-KS genomic library placed KSHV in the gamma2 (Rhadinovirus) lineage of the gammaherpesvirinae along with Herpesvirus Saimiri (HVS). The closest human relative is the gamma1 (Lymphocryptovirus) herpesvirus, Epstein-Barr Virus (EBV) (Moore et al., 1996a). This finding is intriguing for the relationship that gammaherpesviruses have with lymphoproliferative disorders and cancer; HVS can induce malignant lymphomas in owl monkeys and marmosets (Aotus and Saguinus sp.) (Melendez et al., 1969b) and EBV infection is linked to a fatal B-cell proliferation in young males with X-linked lymphoproliferative disease, posttransplant lymphomas, immunoblastic lymphomas in AIDS patients, Burkitt's Lymphoma, Nasopharyngeal Carcinoma and Hodgkin's Disease (reviewed in Rickinson and Kieff, 1996).
Rhadinoviruses are found in many species and several examples have been identified in non-human primates; HVS in Saimiri sciureus (Melendez et al., 1968, 1969a), ateline herpesviruses (HVA) in Ateles geoffroyi (Melendez et al., 1972), rhesus rhadinovirus (RRV) in macaques (Desrosiers et al., 1997) and retroperitoneal fibromatosis herpesvirus in both Macaca mulatta (RFHVMm) and Macaca nemestrina (RFHVMn) (Rose et al., 1997). Two distinct rhadinoviral sequences from African Green Monkeys have also been identified and these viruses are termed Chlorocebus rhadinovirus (ChRV) 1 and 2 (Greensill et al., 2000). Phylogenetic analysis of the DNA polymerase gene sequences from these primates show that there may be two distinct rhadinoviral lineages in Old World primates comprising respectively KSHV/RFHV/ChRV1 and RRV/ChRV2 (Fig.1).
DNA Neighbor-joining tree for a 2850-bp fragment from the DNA polymerase. Sequences were aligned with ClustalX (Thompson J.D. et al, Nucleic Acids Res. 1997 25 (24):4876-82), gapstripped, and analyzed using the PHYLIP Neighbor program based on F84 model distances (PHYLIP version 3.5c; J. Felsenstein and the University of Washington)
Of these animal rhadinoviruses only the New World representatives (HVS, ateline herpesvirus) and RRV grow well in tissue culture and can therefore be used in animal models. HVS causes lymphomas in owl monkeys and marmosets (Aotus and Saguinus sp.) (Melendez et al., 1969b) and despite the close relationship to KSHV, RRV only causes benign lymphoproliferations in experimentally infected macaques (Wong et al., 1999). RFHV has been detected in macaques suffering from retroperitoneal fibromatosis (RF) (Rose et al., 1997). Currently this virus can not be grown in tissue culture and attempts to induce RF experimentally by inoculations with RFHV-containing macaque samples have not yet been successful (Bosch et al.,1999).
The nucleotide sequence of KSHV provides insight into KS pathogenesis
The genome of KSHV has been mapped with cosmid and phage genomic libraries from a Primary Effusion Lymphoma (PEL) cell line (BC-1) and a KS biopsy specimen (Russo et al., 1996; Neipel et al., 1997b,c). The BC-1 KSHV genome and, due to an overall genomic difference of only 0.4%, the KS biopsy genome have a 140.5-kb-long unique region (LUR) which is flanked by multiple 801bp terminal repeat sequences (Fig.2). Within the LUR, 81 potential Open Reading Frames (ORFs) with greater than 100 amino acids have been identified and several additional spliced genes have since been added to this list. The numbering of KSHV ORFs is based on positional homologies with HVS due to substantial collinearity between these genomes whereas ORFs without positional homologues are numbered consecutively with a K prefix. The overall G+C content in the LUR is 53.5% and 84.5% in the terminal repeat sequence. The LUR consists of both conserved herpesviral genes that are involved in herpesvirus replication and structure and non-conserved genes that may provide insights into the pathogenetic mechanisms of this virus (Russo et al., 1996). This review will focus more specifically on viral proteins (designated with the prefix v(-)) such as interleukin 6 (vIL-6/ORFK2), macrophage inflammatory proteins (vMIP-II/ORFK4 and vMIP I/ORFK6), (vBck/ORFK4.1), (vBcl-2/ ORF 16), interferon regulatory factors 1-4 (vIRF1/ORFK9, vIRF2/ORFK10, vIRF3/ORFK10.1, vIRF4/ORFK4), FLICE-inhibiting protein (v-FLIP/ ORF71), cyclin (v-cyclin/ORF72), interleukin8- like/ G protein-coupled receptor (v-IL8R/ vGPCR/ORF74) and a multiple membrane spanning protein (LAMP/ORFK15). The presence in the viral genome of open reading frames with significant homology to mammalian genes involved in cellular growth control suggests that Amolecular mimicry of cell cycle regulatory and signaling proteins is a prominent feature of this virus'(Russo et al., 1996). The function and expression of these viral genes are discussed in more detail below.
Figure 2. The KSHV genome is 140.5kb long and is flanked by multiple terminal repeat sequences which are depicted as hatched squares. Three classes of transcript are recognised (Sarid et al., 1998); class I transcripts are constitutively expressed in PEL cells and these latent transcripts are coloured black, class II transcripts are expressed at low levels during latency but can be induced chemically and are shaded grey and class III transcripts are only present following chemical induction and are represented in white. It should be noted that the nut-1 transcript is not translated. Blocks of structural genes that are conserved between most gammaherpesviruses are labeled I-V. This diagram shows "non-conserved" genes that are discussed in the text and the map is not to scale.
Tracing the origin of KSHV
Comparison of protein sequences predicted from the genomes of Varicella-Zoster virus (VZV) and EBV suggests a common mammalian herpesviral ancestor (Davison and Taylor, 1987). Phylogenetic studies of alphaherpesvirinae genes and proteins using maximum parsimony and distance methods with evaluation by bootstrap analysis have shown that the molecular evolution of mammalian alphaherpesvirinae in the majority of cases parallels the evolution of their mammalian hosts (McGeogh and Cook, 1994). Based on the assumption of virus and host co-speciation over the same evolutionary timescale, similar phylogenetic analyses on common mammalian herpesviral ancestral genes such as uracil-DNA glycosylase imply that the differentiation of alpha, beta and gammaherpesvirinae occurred between 180 and 220 million years ago and that major sublineages in the herpesvirinae were generated before the mammalian radiation of 60-80 million years ago with subsequent coevolution of virus and host (McGeogh et al., 1995).
Analysis of the KSHV genome suggests that 42 genes are common to a set of genes that the mammalian herpesviral ancestor may have possessed. Eighteen genes are shared by the gammaherpesvirinae whereas 9 genes belong to the rhadinovirus lineage and 17 genes are specific to KSHV sublineage. There is also some preliminary phylogenetic evidence that the two Old World _2 herpesviruses KSHV and RRV separated from the New World HVS sublineage at the same time as the Old and New World primate separation approximately 35 million years ago and that KSHV and RRV sublineages separated with their hosts about 25 million years ago (McGeoch and Davison, 1999).
Variation in ORFK1 sequences can be used to identify KSHV subtypes
Precedents for sequence variation in specific regions of gammaherpesvirus genomes are found in the work on EBV nuclear antigens, EBNA-2A, -2B, -3A, -3B and -3C (Dambaugh et al., 1984; Sixbey et al., 1989; Sample et al., 1990), EBV Latent membrane proteins LMP-1, -2A and -2B (Sample et al., 1989; Miller et al., 1994; Busson et al., 1995) and HVS Saimiri Transformation Protein (STP) A- B- and C- alleles (Medveczky et al., 1984).
The most variable region to date in the KSHV genome appears to be ORFK1. The full length ORFK1 protein is predicted to encode a transmembrane protein (aa 1-289) consisting of the following regions; an extracellular domain (aa1-228) with an amino terminal signal peptide and an immunoglobulin receptor-like domain with two variable regions (VR1 and VR2), a transmembrane domain (aa 229-261) and a C-terminal cytoplasmic domain (aa 262-289) which contains an immunoglobulin-receptor tyrosine-based activation motif (ITAM). (Russo et al., 1996; Lagunoff and Ganem, 1997; Neipel et al., 1997b,c; Lee et al., 1998a; Nicholas et al., 1998; Cook et al., 1999; Zong et al., 1999).
Phylogenetic analyses using ORFK1 sequences obtained from different geographic regions, including 50 Classic and AIDS-KS and 10 Primary Effusion Lymphoma (PEL)/Body Cavity Based Lymphoma (BCBL) DNA samples, enabled the definition of four major subtypes (A, B, C and D) with bootstrap values greater than 70% and 83% for linear and radial unrooted dendrograms respectively (Nicholas et al., 1998; Zong et al., 1999). Prototype A, B and C DNA sequences have been previously defined on the basis of lesser degrees of intersubtype (1-1.5%) nucleotide variation in ORFs26/75 (Zong et al., 1997). In contrast, the ORFK1 sequences of these prototypes differed by 5.8-14.6% at the nucleotide level and 14-29% at the amino acid level and this sequence appears to be more reliable for examining strain variability. Interestingly almost 85% of the nucleotide differences led to amino acid substitutions and the majority of these changes were concentrated in VR1 (aa 54-92) and VR2 (aa 199-227) for all subtypes with more extensive involvement of the amino terminus in subtype B. Minimal differences were seen in the putative transmembrane domain for all three subtypes whereas the amino acid sequence of the cytoplasmic tail for subtype A and C differed by 12/38 (32%) amino acids from subtype B. Subtype variants were defined in this study on amino acid differences greater than 5% or on the presence or absence of specific diagnostic deletions in the ORFK1 sequence. There is evidence that particular subtypes and variants are associated with particular geographic regions; A1, A4 and C3 variants were frequently seen in USA AIDS-KS cases whereas B subtypes predominated in African KS cases or people of African heritage and C subtypes were mostly found in Classic, Posttransplant and AIDS-KS cases from the Middle East and Asia. A rare D subtype was identified in patients of Pacific Island origin. This study states that if cospeciation of virus and host is assumed then the geographic distribution of KSHV may be the result of isolation and founder effects associated with the migration of human populations out of Africa over the past 35000-60000 years (Zong et al., 1999).
Similar results were obtained by another study that determined complete (n=23) or near complete (n=25) ORFK1 sequences from 58 tumor and peripheral blood samples from patients with Classic KS, Posttransplant KS, AIDS-KS, lymphoproliferative disorders and asymptomatic KSHV infection (Cook et al., 1999). Pair-wise comparisons of ORFK1 amino acid sequences from these samples varied from 0.4 to 44%. Phylogenetic analysis of 52 ORFK1 DNA and protein sequences (aa28-243) using distance, parsimony and maximum likelihood methods also distinguished A, B and C subtypes with the lowest bootstrap value of 87%. No samples from the proposed subtype D region were analysed in this study. Most of the amino acid differences were seen between amino acids 53-95 and 149-240, which overlaps with the aforementioned hypervariable regions. Less variability is seen in the transmembrane and cytoplasmic domains, and the tyrosine residues of the ITAM motifs (Y-SL, YTQP) were invariant. This study narrows the definition of subtype variant however subtype B predominated in Africa while subtypes A and C were found more frequently in Europe. No correlation was noted between subtype and more aggressive KSHV-related disease or geographic regions (Cook et al., 1999). Both studies have noted a high degree of non-synonymous amino acid changes in the K1 gene although whether this is confined to the hypervariable regions or the entire gene requires further evaluation. In either case this finding suggests that ORFK1 is under a selective pressure in comparison to the rest of the KSHV genome (Cook et al., 1999; Zong et al., 1999). Further evidence of strain variability has been provided and it has been noted that strains do not always correspond to the geographic origin of the sample (Meng et al., 1999; Fouchard et al., 2000). The functions of ORFK1 and the significance of the variation are discussed below.
Some KSHV strains may have undergone recombination with a related gammaherpesvirus.
Two highly divergent versions of the Aright@ end of the KSHV genome between ORF75 and the terminal repeat regions have been noted (Glenn et al., 1999; Poole et al., 1999). Unlike ORFK1 these two variants do not appear to be under evolutionary pressure and may have arisen as a result of recombination with a related gammaherpesvirus (Glenn et al., Poole et al., 1999). Furthermore, additional recombination events involving KSHV strains have been noted throughout the viral genome (Poole et al., 1999). It should also be noted that, as is the case of the EBNA-1 protein of EBV, length variations of the LANA/ORF73 internal repeat region exist in different viral strains but geographic links of this length polymorphism have not yet been established (Rainbow et al., 1997; Gao et al., 1999).
Following the discovery of KSHV (Chang et al., 1994), seroepidemiologic and nucleic acid-based assays were designed to study the geographic distribution of KSHV.
Serological and nucleic-acid based assays for KSHV detection
Serological assays The demonstration of KSHV sequences in all forms of KS implies that antibodies to KSHV should be detectable in sera at the time of KS diagnosis and thereby serve as positive reference material (Olsen and Moore, 1998). Currently negative reference sera are chosen from a population that is at low risk for developing KS. Sensitivity and specificity data for assays can be calculated from these references and it is acknowledged that a low proportion of the negative group may have positive test results.
Latent antigens Antibodies to a high molecular weight (224-234kDa) latent nuclear antigen (LNA or LANA) can be detected by Western Blot (WB) or immunofluorescence assays (IFA) on B cell lines that were established from patients with BCBL/PEL with latent KSHV infection (Gao et al., 1996a,b; Kedes et al., 1996; Simpson et al., 1996). LANA has been shown to be encoded by ORF73 following expression of this gene in bacterial and mammalian expression systems and subsequent demonstration of the 222-234kDa doublet by WB of the nuclear extract (Kedes et al., 1997b; Kellam et al., 1997; Rainbow et al., 1997).
Using AIDS-KS and Classic KS sera as positive reference material and blood donor sera from low prevalence countries such as the United States and the United Kingdom as a negative reference, both LANA WB and IFA formats have high sensitivity and specificity; KSHV antibodies are detected in 71-100% of AIDS-KS and Classic KS sera but only in 0-4% of blood donors (Gao et al., 1996a,b; Kedes et al., 1996; Simpson et al., 1996).
Lytic antigens Lytic antigens have been identified to increase the sensitivity of KSHV immunoassays in order to estimate more accurately the prevalence of KSHV within populations and to avoid the problem that people may seroconvert to different antigens at different stages of KSHV infection (Goudsmit et al., 2000). Antibodies to lytic KSHV antigens can be detected by IFA on PEL cell lines that have been pre-treated with phorbol esters or sodium butyrate (Lenette et al., 1996; Miller et al., 1996,1997; Smith et al., 1997). Lytic antigens can be detected by WB or radioimmunoprecipitation of chemically induced BCBL-1 cell lines and also by whole virus enzyme-linked immunoassay as KSHV virions can be produced from the same cells by adding 12-o-tetradecanoylphorbol-13-acetate (Miller et al., 1996, 1997; Smith et al., 1997; Chandran et al., 1998a; Chatlynne et al., 1998). Recombinant structural antigens such as a capsid-related protein (ORF65) and membrane glycoprotein gp35-37 (ORFK8.1) are also useful for seroepidemiological purposes (Simpson et al., 1996; Chandran 1998a,b; Raab et al., 1998; Lang et al., 1999). Less discriminating antigens are not discussed here.
The reported sensitivity and specificity values for assays detecting antibodies to lytic antigens vary in different studies. Generally, however, lytic IFAs have a sensitivity close to 100% (Lennette et al., 1996; Smith et al., 1997; Chandran et al., 1998a,b; Rabkin et al., 1998) Different rates of antibody detection in US blood donors reported with this assay suggest that this specificity may vary in different laboratories (Lennette et al., 1996; Smith et al., 1997; Chandran et al., 1998a,b). Recombinant ORF65 proteins and gp35-37/K8.1 give sensitivities in the range of 75-85%, with some variability if different expression vectors or purification methods are used (Simpson et al., 1996; Chandran et al., 1998a,b; Raab et al., 1998; Lang et al., 1999). The specificity of recombinant ORF65 and K8.1 based assays is generally higher than 80% and may approach 90% depending on the assay conditions used (Simpson et al., 1996; Chandran et al., 1998a,b; Raab et al., 1998; Lang et al., 1999)
A Western Blot assay using recombinant ORF73, ORF65 and ORFK8.1A and B proteins detects 89% of HIV-infected individuals with KS and 7% of HIV-uninfected individuals without KS (Zhu et al., 1999). Although these assays are suitable for seroepidemiology, they are not optimised for diagnostic use. One of the challenges of this field is to produce a standardised assay with optimal sensitivity and specificity using a combination of antigens.
PCR based studies The earliest study to detect KSHV by nested PCR in peripheral blood showed that 52% of patients with AIDS-KS were positive (Whitby et al., 1995). Similarly KSHV DNA was detected in 14-20% of semen samples from patients with AIDS-KS (Gupta et al., 1996; Howard et al., 1997). As all KS biopsies are infected with KSHV (Olsen and Moore, 1998), it seems that nested PCR is not sensitive enough for more accurate determination of KSHV prevalence. In addition, a study on KSHV sequences in biopsies and cultured spindle cells of Epidemic, Iatrogenic and Classic KS suggests that the rate of KSHV detection by PCR may also be dependent on the extent of clinical disease (Aluigi et al., 1996). Viral DNA sequences in PBMCs from patients with Classic KS were detected in 17/23 (74%) cases of disseminated KS in comparison with 7/17 (41%) indolent KS cases suggesting again that the extent of clinical disease is important (Brambilla et al., 1996). Similar results were found in PBMCs of patients with 13/16 (81%) of active but only 2/10 (20%) of remitted AIDS-KS (Poggi et al., 1997). Semiquantitative analysis of Southern blots showed higher levels of KSHV DNA in those patients with multicentric and visceral involvement than in localized disease (Mendez et al., 1998). Despite these caveats, PCR studies have been invaluable in our understanding of the geographic and population distributions of KSHV, association of KSHV with disease, transmission of KSHV and the natural history of KSHV infection and are discussed alongside the seroepidemiology results in this review.
Geographic distribution of KSHV
Seroprevalence of KSHV Although there are a variety of KSHV serological tests, it is possible to demonstrate that certain geographic regions have comparatively high or low KSHV seroprevalences (reviewed in Schulz, 1999).
In Northern Europe, KSHV appears to be rare among the general population. KSHV antibodies to both LANA and ORF65 are detected in less than 3% of United Kingdom blood donors and Danish railway workers (Simpson et al., 1996; Melbye et al., 1998). Similarly KSHV antibodies to LANA and ORF65 are found in 7% of Swedish blood donors and by LANA alone in 3% of French myeloma/lymphoma patients (Marcelin et al., 1997; Enbom et al., manuscript in preparation). KSHV antibodies to ORF65 were only present in 1.3% of 534 Swedish women (Tedeschi et al., 1999). KSHV seroprevalences rise dramatically in countries in Southern Europe. Twenty percent of control subjects in a case-control study for Classic KS in Greece had antibodies to LANA and/or ORF65 (Simpson et al., 1996). KSHV seroprevalence is 4% in Milan using latent (BCP-1) IFA and LANA WB (Gao et al., 1996b).In other regions of Italy up to 35% of blood donors are KSHV seropositive by LANA IFA and/or ORF65 EIA/WB and a marked regional variation has been noted (Calabro et al., 1998; Whitby et al., 1998). Five percent of Swiss blood donors had antibodies to ORF65 ELISA (Regamey et al., 1998a).
In North America, antibodies to ORF65 and/or LANA are detected in less than 5% of blood donors (Gao et al., 1996a,b; Kedes et al., 1996; Lennette et al., 1996; Simpson et al., 1996). This range increases from 0-14% using a panel of aliquoted sera that were tested to study interlaboratory and interassay variation of seven immunofluorescence assays and ELISAs (Rabkin et al., 1998). Anti-LANA antibodies were virtually undetectable (<1%) in other studies on US blood donors (Kedes et al., 1997b; Chatlynne et al., 1998; Chandran 1998b). Use of a lytic IFA to detect antibodies against an undefined antigen increases the seroprevalence of KSHV among US blood donors to 8-29% (Lennette et al., 1996; Chandran et al., 1998b; Rabkin et al., 1998). ORFK8.1 WB and whole virus ELISA detect similar proportions (8-11%) of HIV negative men and blood donors (Chandran et al., 1998b; Chatlynne et al., 1998). In Central and South America, 7.4% of blood donors were seropositive to LANA IFA in Sao Paolo, Brazil (Caterino-de-Araujo et al., 1999). Only 9/250 (3.6%) blood donors were seropositive with the Whole Virus ELISA or lytic and latent IFA in Jamaica (Manns et al., 1998). In Honduras, 12% of HIV-uninfected men were positive by lytic IFA (Sosa et al., 1998).
In Africa antibodies to LANA and ORF65 have been reported in 6-53% of tested samples such as HIV negative patients, antenatal mothers, hospitalized patients and unknown sources and regional variation is again noted (Lennette et al., 1996; Gao et al., 1996a,b; Simpson et al., 1996; Ariyoshi et al., 1998, Bestetti et al., 1998; Mayama et al., 1998; Olsen et al., 1998; Wilkinson et al., 1998). In one study, KSHV seroprevalence ranges from 32% in Zimbabwe to 100% in the Ivory Coast using a lytic IFA (Lennette et al., 1996).
For the remainder of the world, use of the LANA IFA detected no KSHV antibodies to serum samples of unknown origin in the Dominican Republic, Guatemala and Haiti but 10-29% of the same sera were considered positive with a lytic IFA (Lennette et al., 1996). Thirteen percent of HIV negative women in Haiti had antibodies to LANA (Goedert et al., 1997). In Saudi Arabia 6/44 (7%) control sera were ORF65 EIA positive (Qunibi et al., 1998).
PCR studies have also shown that there is geographic variation in KSHV seroprevalence but lack of sensitivity makes it difficult to determine seroprevalence rates. Nonetheless KSHV DNA was not found in PBMCs in blood donors from the United Kingdom, United States, Spain or France (Collandre et al., 1995; Whitby et al., 1995; Heredia et al., 1996; Humphrey et al., 1996; Marchioli et al., 1996) whereas it was found infrequently (8-11%) in some (Bigoni et al., 1996; Viviano et al., 1997) but not other studies using Italian people as control subjects (Luppi et al., 1996). KSHV DNA has also been detected in 10% of HIV negative antenatal mothers from The Gambia (Ariyoshi et al., 1998) and in 8% of febrile Zambian children (Kasolo et al., 1997).
AIDS-related Kaposi's sarcoma
KSHV seroprevalence in risk groups for HIV-1 transmission The hypothesis that AIDS-KS may be caused by a KS agent implies that the distribution of the agent should parallel the distribution of Kaposi's sarcoma among HIV-1 transmission groups such that KSHV seroprevalence is higher among homosexual/men than all other HIV-1 transmission groups (Beral et al., 1990). Prevalences are considered first in terms of serological results and second in terms of PCR findings.
In the United States, antibodies to LANA have consistently been found to be more prevalent (22-35%) in HIV-infected homosexual men in comparison to other HIV transmission groups such as patients with haemophilia, transfusion recipients and injecting drug users (Gao et al., 1996b: Kedes et al., 1996; Lennette et al., 1996). The gradient of KSHV seroprevalence was again seen in HIV transmission groups using a lytic IFA that detected antibodies in 100% of homosexual men compared to 23% of injecting drug users and 21% of women (Lennette et al., 1996).
Similar distribution patterns of KSHV have been found in HIV transmission groups in Europe. In Denmark, The Netherlands and the United Kingdom, KSHV antibodies to ORF65 and LANA are detected more frequently in homosexual men/bisexual men (30-39%) as opposed to <7% of those with haemophilia or injecting drug users (Simpson et al., 1996; Melbye et al., 1998; Renwick et al., 1998). The pattern of distribution holds true even in Italy where inspite of a higher seroprevalence in the general population, 62% of homosexual men have antibodies to LANA or ORF65 compared to 11% in intravenous drug users and 17% in heterosexuals (Calabro et al., 1998). This pattern is seen consistently in studies from Italy (Rezza et al., 1998).
KSHV is only rarely detected in homosexual men unless they have KS when using nested PCR to detect KSHV DNA in PBMCs. In the United Kingdom, 11/143 (7.6%) HIV-infected homosexual men without KS were KSHV positive versus 0/134 (0%) blood donors and 0/26 (0%) oncology patients (Whitby et al., 1995). Comparable results were reported on KSHV DNA in PBMCs in 3/23 (13%) homosexual men with AIDS but without KS and 0/19 (0%) AIDS patients with haemophilia in the United States (Moore et al., 1996b) and in 1/4 (25%) HIV-uninfected homosexual men and 0/20 (0%) blood donors in France (Lebbe et al., 1997). Similar results were reported from France and Switzerland (Dupon et al., 1997; Quinlivan et al., 1997). The detection of KSHV by PCR is not useful for determining KSHV prevalence however the presence of this virus in other cells and body fluids is discussed below.
Association of KSHV with AIDS-KS That KSHV is the causative agent of KS, among other diseases, is usefully considered in terms of criteria that describe a cause and effect relationship. Disease causation can be viewed in terms of the following features; (1) The prevalence of the disease should be higher in those exposed than not exposed to the proposed cause and conversely, exposure to this cause should be more common in those with the disease than those without the disease. In the same vein, the incidence of the disease should be higher in persons who are exposed than not exposed as shown in prospective studies. As the strength of association, as measured by an odds ratio or relative risk, between a causative agent and its disease is based on prevalence or incidence data in exposed and unexposed groups, the implication is that the stronger the association, the more likely that the relationship is causal (2) These associations should be observed at different places and times under varying circumstances (3) Exposure to the cause should precede the appearance of disease (4) In some instances, there should be a dose-response relationship between the severity of exposure and severity of outcome which in turn means that prevention or elimination of the risk factor or modification of the host response should produce identical trends in disease expression (5) The specificity of the relationship in that a single suspected cause produces a single effect provides weak supportive evidence for causation (6) Although not always necessary, additional support for a causal relationship comes from biological plausibility and the relationship should be consistent with the natural history of the disease (7) Experimental evidence should support the association (Hill, 1965; Evans et al., 1978).
Studies that detect antibodies to LANA, ORF65 or vp40 using IFA, WB or EIA have all demonstrated high prevalences of KSHV antibodies in patients with AIDS-KS (range: 51.6-87.3%) compared with HIV-infected controls without KS (range:12.9-43.7%). As a result significant odds ratios (range: 3.7-18.7) have been found for the presence of KSHV antibodies between these cases and controls (Gao et al., 1996a,b; Kedes et al., 1996; Miller et al., 1996; Simpson et al., 1996; Renwick et al., 1998).
The detection of KSHV by PCR in PBMCs of AIDS-KS patients (range: 34.7-90.9%) and HIV-infected controls (range 0-18.7%) has similarly demonstrated a strong association between KSHV infection and KS with odds ratios varying from 2.3-440.0 (Whitby et al., 1995; Moore et al., 1996b; Lefrere et al., 1996; Marchioli et al., 1996; Humphrey et al., 1996). This association remains strong even in a region of high KSHV seroprevalence in the general population such as The Gambia (Ariyoshi et al., 1998). A strong association was not detected in the minority of studies (Collandre et al., 1995; Decker et al., 1996). The important temporal association between risk factor and disease was also demonstrated by PCR prior to the use of more sensitive serological assays (Whitby et al., 1995).
Based on the Hill criteria for disease causation, results from prospective cohort studies in differing locations have provided the most convincing arguments to date that KSHV is the causative agent of KS within HIV-1 infected individuals. Not only are the associations between KSHV and KS consistently strong but also KSHV antibodies are usually detected before the appearance of clinical KS lesions. This temporal relationship has been quantified in the San Francisco Men's Health Study, Amsterdam Cohort Studies and a cohort of Danish homosexual men among others (Gao et al., 1996a; Martin et al., 1998; Melbye et al., 1998; Renwick et al., 1998; O'Brien et al., 1999; Rezza et al., 1999). Fifty percent of HIV and KSHV coinfected individuals developed AIDS-KS within 5-10 years and a similar result was seen in Italy and the United States where approximately one third of HIV-KSHV coinfected men developed KS within 10 years (Martin et al., 1998; O'Brien et al., 1999; Rezza et al., 1999). Other data that are necessary to fulfill the Hill criteria are provided by the Amsterdam Cohort Studies on HIV-1 infection. The incidence of KSHV infection is highest among HIV-infected homosexual men (6.2/100 person-years) compared with HIV-uninfected homosexual men (2.6/100 person-years) and injecting drug users (0.7/100 person-years). As KS cases were observed exclusively in HIV-infected homosexual men, KSHV incidence data parallels the distribution of KS in these cohorts. Statistical models have also proved invaluable in evaluating the risk that KSHV infection confers on developing KS; a time-dependent Cox proportional hazards model has shown that KSHV antibodies (HR=3.27: 95%CI: 1.86-5.77) and declining CD4 counts (HR=0.59: 95%CI 0.51-0.68) are independent predictors of Kaposi's sarcoma in the context of HIV-1 infection. The timing of KSHVand HIV infections may be important as KSHV seroconversions that follow HIV-1 seroconversion appear to increase the risk for developing KS (HR=5.17: 95%CI 2.88-9.27) (Renwick et al., 1998).
KSHV seroprevalence in risk populations and association with Classic KS
KSHV seroprevalences range from 94.4-100% among persons with Classic KS and from 3.7-19.1% for matched or blood donor controls and odds ratios, when calculable, range from 130-257 (Gao et al., 1996b; Simpson et al., 1996; Calabro et al., 1998). Incidence rates for Classic KS have been estimated in Denmark, Italy, Greece, Sweden, the United Kingdom and the United States (Geddes et al., 1994, 1995; Franceschi and Geddes, 1995).For the most part, these incidence rates of Classic KS are paralleled by KSHV seroprevalences in blood donors from these countries, as determined by immunoassays performed by one research group to minimise interobserver and interassay variation. This is exemplified in that the 8-fold difference in annual incidence rates of Classic KS for British men over 60 years (0.42/100,000: 1971-80) and Italian men over 50 years of age (3.44/100,000: 1976-84) is mirrored by a similar difference in KSHV seroprevalence in Italian (24%) and British (3%) blood donors (Simpson et al., 1996; Calabro et al., 1998; Melbye et al., 1998; Rabkin et al., 1998; Enbom et al., 2000; reviewed in Schulz, 1999).
For countries such as Italy which have comparatively high incidence rates of classic KS in comparison to the rest of the world, these rates are higher in Southern Italy, Sicily and Sardinia than in central or Northern Italy (Geddes et al., 1994, 1995; Francheschi and Geddes, 1995). The seroprevalence of antibodies to LANA in blood donors are higher in Southern Italy (24.6%) than Northern Italy (7.3%), and the highest seroprevalence (35.0%) is seen in Sicily (Whitby et al., 1998). Regional variation has again been demonstrated with antibodies to ORF65 and LANA in Italian blood donors with the highest seroprevalence (32%) in Sardinia (Calabro et al., 1998). Twenty five of one hundred (25%) sera obtained from pregnant women in Sardinia were KSHV seropositive by lytic IFA (Serraino et al., 1998) and high titers of anti-lytic and anti-latent antigens were found in KS patients in comparison to 50 HIV seronegative patients without KS (Cattani et al., 1999). Prospective cohorts are available for studying the relationship between KSHV and Classic KS.
KSHV seroprevalence in risk populations and association with Posttransplant KS
KS is also associated with receiving an organ transplant (Harwood et al., 1979; Penn, 1979). Case-control studies from different geographic regions have again demonstrated a strong association between KSHV antibodies and Posttransplant KS. In Italy, 10/11 (91%) of transplantation recipients with KS were seropositive to antibodies to LANA and ORF65 prior to transplantation compared to 2/17 (11.7%) organ recipients who served as controls (OR=75; 95%CI: 4.7-3500) (Parravicini et al., 1997b). In Saudi Arabia, where KS is the commonest tumor to follow organ transplantation, 13/14 (92.9%) renal transplant recipients with KS were positive to p40 and sVCA immunoblot assays compared to 5/18 (27.7%) (p<0.001) renal transplant recipients without KS (Qunibi et al., 1998). In France, 17/25 (68%) of transplant recipients with antibodies to KSHV LANA or ORF65 pre- or post-transplantation developed KS compared to 1/33 (3%) KSHV seronegative transplant recipients (p<0.00001) (Farge et al., 1999). Independent risk factors for KS in this transplantation population included origin in Africa or the Middle East, use of antilymphocyte sera for induction and KSHV antibodies (OR=28.4; 95%CI: 4.9-279) (Farge et al., 1999). In a separate study from France 16/166 (9.6%) transplant recipients were KSHV seropositive with LANA IFA (Frances et al., 1999). Twelve of these sixteen (75%) of the KSHV seropositive patients survived past the first year and three patients developed Posttransplant KS wheres no such disease occured in the 150 KSHV seronegative patients. In Switzerland, KSHV seroprevalence in 220 recipients of renal transplants increased from 14/220 (6.4%) at the time of transplantation to 39/220 (17.7%) after one year of follow-up using ORF65 ELISA (Regamey et al., 1998b). Two of the twenty five KSHV seroconverters in this study developed Posttransplant KS.
KSHV seroprevalence in risk populations and association with African KS
Data on HIV negative KS patients in Africa are scarce. One patient was positive by LANA IFA and Western Blot in Uganda (Gao et al., 1996b) and 28/28 (100%) serum samples from patients with Endemic/African KS were positive with both lytic and latent IFAs (Lennette et al., 1996).
The existence of a cofactor(s)
Despite high prevalences (13.8-24.1%) of KSHV in blood donors from Italy, population-based incidence rates of Classic KS are low at 0.7-3/100000. The predilection of KS for males does not seem to be explained by higher KSHV seroprevalences in men which at best are three fold higher than women in some but not all studies (Calabro et al., 1998; Whitby et al., 1998; Angeloni et al., 1998). Similarly Endemic KS is more common in Central Africa and East Africa than in the rest of the continent, however KSHV seroprevalences are similar in West Africa and South Africa (Ariyoshi et al., 1998; Wilkinson et al., 1998) than in East Africa (Lennette et al., 1996; Simpson et al., 1996; Mayama et al., 1998). In a study of 16 KS patients from West Africa where there are comparable amount of HIV-1 and HIV-2, 14 patients had HIV-1, 1 patient had HIV-2 and 1 patient was coinfected by HIV-1 and HIV-2 (Ariyoshi et al., 1998). This study suggests that HIV-1 is a more effective co-factor than HIV-2 even when adjusted for CD4 count and that the role of HIV infection is not solely immunosuppression. Another unexplained aspect of KSHV epidemiology is that KS is as common as an AIDS-defining illness in HIV-infected drug users, patients with haemophilia and heterosexually transmitted women in KSHV endemic countries and countries where KSHV infection is rare suggesting that mode of transmission or the timing of infection plays a critical role (Casabona et al., 1991). It is also possible that a more virulent strain of KSHV or co-factor is an important determinant of disease expression.
Transmission of KSHV
Although the epidemiology of AIDS-KS suggests that KSHV is transmitted preferentially but not exclusively among HIV transmission groups, more formal evidence has been acquired using seroepidemiological results from several cross-sectional and prospective cohort studies on homosexual men.
In a cohort of Danish homosexual males, the presence of antibodies to KSHV LANA and ORF65 was independently associated by multivariate analysis with the number of receptive anal intercourses (OR=2.83; p=0.03) and sex with men from the United States (OR=2.27; p<0.05). Multivariate analysis of KSHV seroconversion over the follow up period to 1996 showed that KSHV seroconversion was independently associated with visits to homosexual communities in the United States (RR=2.1; p=0.03) and also HIV positive status (RR=2.0, p=0.03). A decline in KSHV incidence in the early 1980s was attributed to changes in lifestyle (Melbye et al., 1998).
Among participants and controls from the San Francisco Men's Health Study, no anti-LANA antibodies were detected among exclusively heterosexual men whereas KSHV seroprevalence was 12.5% among men who reported mostly homosexual activity and 39.6% among exclusively homosexual men (Martin et al., 1998). KSHV seropositivity was also strongly associated with base-line HIV infection (p<0.001). In addition the prevalence of KSHV increased linearly with the number of male intercourse partners in the preceding two years, the relative prevalences of KSHV were significantly raised among persons with a self-reported history of sexually transmitted disease and the highest prevalence of KSHV was found among men with more than five years of regular homosexual intercourse (Martin et al., 1998). In the Amsterdam Cohort Studies (1984-1996) risk factors were examined for those who were KSHV positive at enrolment and also for those who seroconverted during the course of the study. The independent risk factors that were associated with KSHV seroconversion include orogenital insertive orogenital sex (OR 5.95; 95%CI: 2.88-12.29) or orogenital receptive sex (OR 4.29; 2.11-8.71) with more than five partners in the past six months, older age (OR 2.89; 1.13-7.34 when older than 45 years) and preceding HIV infection (OR: 2.47; 1.53-3.99) (Dukers et al., 2000).
In the Sydney HIV cohort, HIV-1 infected homosexual men with KSHV antibodies determined by ORF65 WB and LANA were more likely to report more casual sexual partners and insertive oroanal contact with casual partners although these findings were not statistically significant (Grulich et al., 1999). Other recent evidence is provided from the San Francisco Young Mens Health Study in which 39/79 (48.1%) young homosexual men were seropositive and this was linked to the number of male sex partners (Blackbourn et al., 1999). Sera from 2718 patients in a London sexually transmitted disease clinic were tested with latent IFA and 198/2718 (7.3%) were positive with independent risk factors being homo- and bisexuality, birth in Africa, a history of syphilis, HSV-2 and HIV-infection (Smith et al., 1999). There was also no evidence in this study for sexual transmission among heterosexuals. KSHV antibodies were again noted to be more common in HIV-positive homosexual men (Verbeek et al., 1998).
There is considerable disagreement between studies on the precise mode of KSHV transmission between homosexual men however the mode of transmission must be either exclusively or a combination of orogenital, oro-anal, oro-oral or ano-genital insertive or receptive sex. The importance of the oral cavity as a reservoir of KSHV infection is seen in the following studies; KSHV has been detected in saliva by PCR or culture in HIV-1 infected persons with or without KS and in one HIV-negative individual with KS (Whitby et al., 1995; Vieira et al., 1997; Koelle et al., 1997). KSHV DNA sequences were found in 25/76 (32.9%) DNA samples extracted from saliva of HIV-infected patients but in 0/39 (0%) HIV-negative patients (Boldogh et al., 1996). KSHV is detected by nested PCR in 14/32 (43.8%), 11/24 (45.8%) and 2/24 (0.08%) of throat swab, saliva and urine samples respectively for KS patients but in almost no controls (Cattani et al., 1999). Ten of fourteen (71.4%) of HIV-infected archived oral biopsy samples were KSHV PCR positive and 20 control samples were negative (Di Alberti et al., 1997a). Other studies have shown KSHV by PCR in saliva and nasal secretions of HIV-1 infected individuals and KSHV seropositive persons but not in control subjects (Blackbourn et al., 1998; Lucht et al., 1998). Interestingly the number of HHV-8 copies per microgram of tissue or body fluid from KS patients was highest in PBMCs followed by saliva and semen and undetectable in faeces (La Duca et al., 1998). No KSHV sequences were detected by PCR on faeces from HIV-1 infected persons (Whitby et al., 1995).
The varying proportions of KSHV in semen samples and prostate glands is contentious and may be subject to geographic variation for example Italy (reviewed in Blackbourn et al., 1997b). Since this review these findings have remained contentious. Most reports show that KSHV DNA is absent in semen of HIV-1 infected men without KS but found in 25-33% of individuals with AIDS-KS (Diamond et al., 1997; Huang et al., 1997). Another study has detected KSHV in a large proportion of HIV-1 infected men (Bobroski et al., 1998). There was no evidence for KSHV infection in healthy Danish semen donors or in prostatic tissue from HIV-negative adults (Rubin et al., 1998; Kelsen et al., 1999) however KSHV was found in prostatic tissue by PCR in 5 patients with past or current KS (Diamond et al., 1998). In samples from Central Africa, almost half (53%) of the HIV-positive African patients with KS had detectable KSHV in their semen and in 13% of HIV-negative donors by nested PCR of ORF26 (Belec et al., 1998). These differences are probably due to variations in the assay or reflect geographic and population differences; a recent multicenter study has shown that it is easy to contaminate the HHV8 PCR and in those that were not contaminated, KSHV DNA was detected in less than 8% of all HIV-infected and uninfected donors (Pellett et al., 1999).
Only 13/387 (3.4%) women from the San Francisco bay area had KSHV antibodies to LANA IFA and 12 of the KSHV seropositive were also HIV seropositive (Kedes et al., 1997a). In London, 13/169 (18.3%) women attending a sexually transmitted disease clinic were KSHV seropositive, and were more likely to be African than born elsewhere (Whitby et al., 1999). KSHV can be detected in cervicovaginal secretions in HIV-infected women (Calabro et al., 1999;Whitby et al., 1999). It is not clear why so few HIV-1 infected women get Kaposi's sarcoma.
Transmission before puberty appears to be rare in the United States (Blauvelt et al., 1997) but does occur in countries where KSHV is more widespread; 6/40 (15%) Sardinian children were found to have antibodies to KSHV LANA and/or ORF65 (Calabro et al., 1998). Age distribution of KSHV antibodies using the same tests as above showed that the KSHV seroprevalence for adults was reached well before puberty in Ugandan children but also rare before the age of 2 (Mayama et al., 1998). Correlation with hepatitis B infection suggests that KSHV is transmitted horizontally in conditions of close contact and crowding (Mayama et al., 1998). Evidence for intrafamilial clustering has been seen in Italy (Angeloni et al., 1998). In South Africa, KSHV infection was common among children below the age of puberty but increased above puberty (Wilkinson et al., 1998). 4/53 (8%) of Zambian children admitted to hospital with a first febrile episode had detectable KSHV by PCR in their PBMCs (Kasolo et al., 1997)
In South Africa, seventeen (16%) of 107 healthy mothers (60 black, 24 white, 5 Asian and 18 mixed race) and 9 (8%) of their 112 children were KSHV seropositive by LANA IFA (Bourboulia et al., 1998). Eight of the nine (88%) seropositive children had a seropositive mother and this study concluded that children less than 10 were probably the result of mother-child transmission, although whether this occurred pre-, peri or post-partum is not yet known. A subsequent study showed that the probability of mother-to-child transmission increased with increasing maternal antibody titer (Sitas et al., 1999). In Zambia, 183 (48.4%) of 378 pregnant women were KSHV seropositive and 5 children with KS all had KSHV seropositive mothers (He et al., 1998). A study of 215 Ugandan children showed that antibodies to ORF65 and LANA were independently associated with Hepatitis B infection and adult seroprevalence rates were reached before puberty suggesting a horizontal route of transmission (Mayama et al., 1998). Similar results were found to latent and lytic antigens in Cameroon where beginning at 4 years of age there was a steady increase in seroprevalence from 27.5% to 39% in the 12-14-year age groups and 48% above 15 years suggesting that KSHV infection occurs during childhood in Africa (Gessain et al., 1999). In Egypt, the seroprevalence of antilytic antibodies exceeded 50% in children older than 6 years and the prevalence stabilised at ten years (Andreoni et al., 1999). In children less than 2 years, KSHV infection is however rare, even in endemic countries, arguing against transmission through breast milk (Goedert et al., 1997; Mayama et al., 1998; Gessain et al., 1999; Lyall et al 1999).
The presence of KSHV antibodies prior to transplantation in 10/11 (91%) of transplant recipients in the Italian study and in all 5 Posttransplant KS patients for whom pretransplant sera were available in the French study suggests that KS is mainly due to reactivation of KSHV in regions of KSHV endemicity (Parravicini et al., 1997b; Farge et al., 1999). A similar argument for KSHV reactivation was seen in France (Frances et al., 1999). In Switzerland, a non-endemic country, the appearance of IgM antibodies to ORF65 in 8/10 (80%) renal transplant recipients within three months of transplantation suggests that KSHV is transmitted through renal allografts or blood transfusion (Regamey et al., 1998b). There is some evidence that KSHV reactivates quickly after transplantation and that increased HHV8 DNA levels in PBLs is associated with KS development (Mendez et al., 1999). KS remission seems to coincide with reduction or cessation of immunosuppression (Moosa et al., 1998).
The detection of KSHV by reverse-transcription PCR from KSHV negative CD19 cells that were inoculated with the filtered supernatant of phorbol ester and interleukin-6 stimulated CD19 cells from a KSHV seropositive North American blood donor implies that KSHV may be transmitted parenterally through blood transfusion (Blackbourn et al., 1997a). The frequency of this occurrence is questioned by finding no KSHV DNA by KS330233 PCR in 19 multiply-transfused recipients (Lefrere et al., 1997). KSHV antibodies as detected by LANA IFA were found in 1/74 (1.3%) Romanian children who were HIV-1 positive as a result of parenteral transmission (Marcelin et al., 1998). A weak association between antibodies to ORF65 but not to LANA and a history of intravenous injections were found among African children younger than 12 years (Mayama et al., 1998).
KSHV and its association with diseases other than KS
KSHV was detected by PCR in 3 B-cell immunoblastic lymphomas from AIDS patients (Chang et al., 1994). Subsequent investigation for KSHV by PCR or Southern hybridisation of a panel of 193 DNA samples that were extracted from lymphomas from patients with (n=41) and without AIDS (n=151) showed that 8 positive samples all belonged to the body cavity based lymphomas (Cesarman et al., 1995a). This finding was confirmed and the tumor was termed primary effusion lymphoma (Cesarman et al., 1996; Nador et al., 1996). Cell lines such as BC-1, -2 and -3 and BCP-1 that have been established from these lymphomas have been invaluable for seroepidemiological and pathogenetic studies on latent and lytic KSHV growth (Cesarman et al., 1995b; Arvanitakis et al., 1996; Boshoff et al., 1998; Renne et al., 1996).
Due to an epidemiologic association between Multicentric Castleman's Disease (MCD), an atypical lymphoproliferative disorder, and KS in AIDS patients, KSHV was investigated and detected in 14/14 (100%) cases of HIV-associated MCD and 7/17 (41.1%) HIV-negative MCD (Soulier et al., 1995). Constant detection of KSHV in these diseases is reasonable evidence of causation as they are too rare to be considered in terms of the Hill criteria. Pathogenesis of these diseases is not discussed in this review.
The KSHV literature is filled with reports on the association or lack of association between KSHV and an extensive list of dermatologic, haematologic, neurologic and oncologic disease, infectious disease syndromes and systemic disorders. The most controversial reports to date are those concerning sarcoidosis and multiple myeloma (Di Alberti et al., 1997b; Rettig et al., 1997). As this article is confined to Kaposi's sarcoma these disease associations are not discussed.
The pathology of KS
The pathology of KS is similar for all four epidemiological variants. At a macroscopic level, KS lesions tend to progress through patch (macular), plaque and nodular stages. KS lesions are usually first observed on the skin and the histological events here described begin in the upper dermis. Patch lesions are characterised by perivascular infiltrates of lymphocytes and plasma cells and irregular dilated lymphatic-like spaces lined by endothelial cells. The plaque stage is notable for more extensive infiltration of blood vessels through the dermis and the presence of short fascicles of spindle-shaped cells. In the nodular stage the dermis is replaced by sheets of spindle cells with ovoid nuclei, which are seldom mitotic, and vascular spaces that give the lesion the appearance of a honeycomb. More aggressive lesions are characterised by more mitotic events and cellular atypia (Calonje and Wilson-Jones, 1997). Although spindle cells appear to be the proliferative component of the KS lesion, there is considerable debate as to their origin and whether their proliferation represents a hyperplastic polyclonal process rather than a monoclonal malignancy.
These doubts emerge from the following observations and studies. At a clinical level, examples of the natural history of KS include reports of lesions that have regressed following the administration of thalidomide for a concomitant disorder or antiretroviral therapy and this behaviour seems uncharacteristic for a malignant process (Soler et al., 1996; Blum et al., 1997; Lebbe et al., 1998; Wit et al., 1998). Cells that are present or derived from KS lesions express a host of endothelial, fibroblastic and smooth muscle surface antigens and argues that the lesions contain a heterogeneous rather than a homogeneous cell population (Regezi et al., 1993; Kaaya et al., 1995; Lebbe et al., 1997). Sometimes it is possible to establish immortalized cell lines from KS tissue, of which KS Y-1 can induce malignant tumors in immunodeficient mice (Siegal et al., 1990; Lunardi-Iskandar et al., 1995; Albini et al., 1997). Disagreement surrounds the issue of clonality of proliferating KS cells and therefore the classification of KS as a malignancy. As each somatic cell in women contains one maternal and one paternal X chromosome, clonality, as assessed by the methylation patterns of the X-linked androgen-receptor gene present in DNA from KS biopsies from women, show that KS can arise from different cells in the same patient and that evidence for polyclonality can be seen in nodular KS tissue although some late-stage KS tumors are monoclonal (Delabesse et al., 1997; Rabkin et al., 1997; Gill et al., 1998).
It is interesting to search for a disease with which to compare KS and its pathogenesis. It has been suggested that the extensive EBV replication that is observed in hairy leukoplakia is similar to KSHV replication in some KS tissues (Hendrier, 1999). Other researchers counter that lytic replication is seen only in the minority of KS lesions so that the analogy may be consistent with the polyclonal expansion of B cells by EBV in organ transplantation (Schulz and Moore, 1999). The electron microscopic and immunohistochemical bases for these analogies are discussed below. .
Immunohistochemistry of KS lesions; presence, location and expression of KSHV encoded genes in Kaposi's sarcoma
KSHV DNA sequences have been demonstrated by in situ hybridisation and/or amplification in spindle cells and endothelial cells that line the vascular spaces of patch, plaque and nodular KS lesions and are not present in normal endothelial cells (Boshoff et al., 1995; Li et al., 1996; Foreman et al., 1997; Kennedy et al., 1998). These viral sequences have also been detected in the following cells; B cells, CD8+ T cells, macrophages, monocytes, glandular epithelium from the prostate gland, non-neoplastic lymph nodes and normal gastrointestinal mucosa from HIV-infected individuals and circulating spindle cells (reviewed in Schulz, 1998; Flamand et al., 1996; Blasig et al., 1997; Sirianni et al., 1997a,b; Kliche et al., 1998; Trovato et al., 1999).
The detection of circular KSHV genomes in non-stimulated BCBL cell lines is considered as evidence of persistent or latent KSHV infection (Decker et al., 1996). More formal studies on KSHV transcription have been performed on BC-1 cell lines and KSHV transcripts can be classified as either Class I, II or III depending on their production under standard growth conditions and inducibility with tetradecanoylphorbol acetate (TPA) or latent, immediate early, early and late transcripts based on the time of their production pre- and post-treatment with sodium butyrate (Sarid et al., 1998; Sun et al., 1999). Interestingly for the pathogenesis of Posttransplant KS, the lytic cycle in KSHV can be activated by hydrocortisone (Hudnall et al., 1999). Class I transcripts are presumed to be latent due to their constitutive expression under standard growth conditions and lack of chemical inducibility and include, to date, v-FLIP (ORFK13/ ORF71), v-cyclin (ORF72) and LANA (ORF73) whereas Class II transcripts such as nuclear transcript 1 (nut-1/T1.1), vIL6 (ORFK2), vMIP-II (ORFK4), vMIP-I (ORFK6), vIRF (ORFK9) and T0.7 (ORFK12) are present constitutively and inducible with TPA and Class III transcripts such as ORF65, ORFK1 and vIL8R (ORF74) are only present following induction with TPA (Sarid et al., 1998). A second large study on KSHV viral transcription agrees with the designation of the latent transcripts, with the exception of v-cyclin (ORF72) which was not formally investigated (Sun et al., 1999). Time course studies of transcript production show that BC-1 cells post induction with sodium butyrate produce immediate early transcripts from Rta (ORF50), vIL-6 (ORFK2) and vMIP-II (ORFK4), early transcripts from vMIP-I (ORFK6), T0.7 (ORFK12), vBcl-2 (ORF16) and v-IL8R (ORF74) and late transcripts from vp19 (ORF65) (Sun et al., 1999). Several studies on the expression of individual genes have provided additional evidence to support the designation of the above transcripts as either lytic or latent (reviewed in Schulz, 1998; Davis et al., 1997; Horenstein et al., 1997; Reed et al., 1998; Sturzl et al., 1998, 1999).
Circular episomes which are indicative of latent persistent herpesviral infection were detected in KS biopsies (Decker et al., 1996). Of the class I/latent/constitutive genes, as defined in the PEL cell lines (Sarid et al., 1998; Sun et al., 1999), ORF73/LANA and ORF72/v-cyclin are known to be expressed in KS spindle cells, either by in situ hybridisation, or immunohistochemistry, or both (Staskus et al., 1997; Sturzl et al., 1997, Davis et al., 1997; Rainbow et al., 1997, Dupin et al., 1999). These findings indicate latent persistence of KSHV in the majority of KS spindle cells. However, a proportion of KS spindle cells has been shown to express lytic cycle (immediate-early, early and late) genes (Blasig et al., 1997; Sturzl et al., 1998 ; Sun et al., 1999). In addition KSHV virions can also be detected by light and transmission electron microscopy in some KS lesions (Orenstein et al., 1997). The extent of lytic and latent gene expression in KS tissue and the tissue specific expression of genes will also help determine an analogy for the pathology of KS. In early KS lesions, KSHV is present in <10% of cells forming the ectatic vessels whereas it is present in >90% of spindle cells in nodular KS and colocalizes with VEGFR3 (Dupin et al., 1999).
Necessary conditions for KS development
It is possible to study the actions of KSHV on microvascular endothelial cells in culture (Flore et al., 1998; Moses et al., 1999). Purified KSHV particles cause long term proliferation and survival of these microvascular endothelial cells and telomerase activity and anchorage independent growth is also observed (Flore et al., 1998). Interestingly in this study, KSHV was only present in a subset of cells and the survival of uninfected cells seems to be due to a paracrine mechanism, which is a recurrent theme in KS pathogenesis as discussed below. Predominantly latent infection, as determined by the expression of ORFK12 and LANA/ORF73, was noted in an in vitro culture system using these cells and lytic infection, as determined by ORF59 and ORFK8.1 expression, was observed with a small percentage of cells (Moses et al., 1999). KSHV infection of endothelial cells produces spindle-shape that resemble those in the KS lesion and show transformation characteristics such as loss of contact inhibition and acquisition of anchorage independent growth (Moses et al., 1999).
It is also possible to culture and study spindle cells from KS lesions and the following experimental studies implicate the growth-dependence of KS-derived cells on cytokines and angiogenesis in the pathogenesis of this vascular tumor. It is possible to culture cells from KS lesions that produce factors which support their own growth and that of others such as fibroblasts and endothelial cells and injection of these cultured cells into nude mice produces vascular tumors composed of mouse cells that are similar to KS (Salahuddin et al., 1988). These cells have been shown to grow in the presence of cytokines such as IL-1_, IL-6, Interferon-_, Oncostatin M and Tumor Necrosis Factor _ (Miles et al., 1990, 1992; Nair et al., 1992). In turn KS-derived cells in culture produce a variety of cytokines namely; basic Fibroblast Growth Factor (bFGF), IL-1_, IL-6, granulocyte-macrophage colony-stimulating factor (GM-CSF) and vascular endothelial growth factor (VEGF) (Ensoli et al., 1989; Miles et al., 1990; Bailer et al., 1995; Cornali et al., 1996). Reactivation and persistence of KSHV in B cells and monocytes is maintained by Th1 cytokines such as IFN-_ produced in KS (Monini et al., 1999). In addition, the synergistic effects of HIV-1 Tat and bFGF promote angiogenesis in nude mice and cell lines from AIDS-KS lesions express high levels of both VEGF and VEGF receptors (Ensoli et al., 1994; Masood et al., 1997).
Although the remainder of this review focuses on the role of KSHV in KS pathogenesis, it is important to emphasize the considerable evidence that implicates host-derived cytokines and angiogenic factors in this process. Future reviews on KS tumorigenesis are likely to pursue this theme of a complex interplay between KSHV-, co-factor- and host encoded- factors.
These proteins will be discussed using the classification proposed by Sarid et al. and genes that encode viral structural and assembly proteins are not discussed.
All 3 of these mRNAs are expressed in BCBL-1 cells prior to lytic induction and are also detectable in >70% of KS spindle cells in primary KS tumors (Dittmer et al., 1998).
The dual ORF assignment represents a historic disagreement over whether this gene was unique to KSHV or conserved among herpesviruses (Russo et al., 1996; Neipel et al., 1997b,c; Thome et al., 1997). ORFK13/ORF71 belongs to a group of genes which encode a FLICE protease inhibitory protein (v-FLIP) and cells that express v-FLIP are protected against apoptosis induced by the CD95 death receptor (Thome et al., 1997). Viral FLIP inhibits caspase activation and thereby prevents cells from Fas-mediated apoptosis and inoculation of v-FLIP transduced murine B lymphoma cells into immunocompetent mice led to aggressive tumor formation with a high rate of survival and growth (Djerbi et al., 1999). In this study, v-FLIP was considered a tumor progression factor. This protein may have a dual role in anti-apoptosis and tumor progression.
The association of cyclins with cyclin-dependent kinases (cdks) leads to the formation of an active holoenzyme that can phosphorylate many target molecules that are involved in the regulation of the cell replicative cycle (reviewed in Cannell and Mittnacht, 1999). KSHV v-cyclin stimulates cdk6 to phosphorylate the retinoblastoma protein and H1 histones (Chang et al., 1996; Li et al., 1997; Godden-Kent et al., 1997; Swanton et al., 1997). The v-cyclin/cdk6 holoenzyme is not inhibited by CDK inhibitors such as p16Ink4a, p21Cip1 and p27Kip1 and the holoenzyme stimulates the progression from G1 to S in quiescent fibroblasts (Swanton et al., 1997). In addition the holoenzyme can phosphorylate the p27Kip1 inhibitor which subsequently degrades and can also bypass p27Kip1-imposed G1 arrest (Ellis et al., 1999; Mann et al., 1999). V-cyclin appears to trigger apoptosis after the S phase in the presence of cdk6 but this phenomenon can be blocked by the activity of vBcl-2 (Ojala et al., 1999). V-cyclin can also activate transcription of human cyclin gene A in quiescent cells (Duro et al., 1999). All these studies implicate KSHV v-cyclin in cellular dysregulation and this protein may play a role in the proliferative component of the KS lesion.
Antibodies against LANA have also been used to demonstrate extensive expression of this protein in KS spindle cells (Rainbow et al., 1997; Katano et al., 1999; Kellam et al., 1999). LANA colocalizes with the viral episome in interphase nuclei and along mitotic chromosomes suggesting that LANA may be involved in episome maintenance during cell division (Ballestas et al., 1999). LANA has also been shown to be associated with nuclear heterochromatin (Szekely et al., 1998) and to bind to RING3, a member of the homeotic gene family fsh, an interaction that appears to mediate its phosphorylation (Platt et al., 1999).
The importance of cytokines such as IL-6 in KS pathogenesis has already been discussed. KSHV ORFK2 encodes an IL-6 molecule with a low degree of amino acid homology (24.8%) to human IL-6 (Moore et al., 1996a; Neipel et al., 1997a; Nicholas et al., 1997). This viral cytokine is expressed by PEL cells and CD20+ B cells in lymphoid tissue from KSHV infected patients but not KS tissue and suggests that vIL-6 may contribute to haematopoietic rather than endothelial cell proliferation (Moore et al., 1996c). This study mooted the idea of tissue specific expression of KSHV genes in KSHV-related pathology.
Viral IL-6 is functional and can maintain the proliferation of IL-6 dependent mouse and human cell lines (Moore et al., 1996c; Nicholas et al., 1997; Burger et al., 1998). This cytokine activates the JAK/STAT signaling pathway via interaction with the gp130 subunit independent of the IL-6R alpha chain (Molden et al., 1997; Burger et al., 1998; Wan et al., 1999). Viral IL-6 is a multifunctional cytokine that promotes haematopoiesis, plasmacytosis and angiogenesis (Aoki et al., 1999). Variation in vIL-6 expression between KSHV associated diseases such as PEL, MCD and KS has been noted (Cannon et al., 1999). These results partly agree with another study that had found vIL-6 transcripts in 6/6 KSHV positive, HIV seronegative patients with MCD but in no patients with KS (Parravicini et al., 1997a). The average levels of vIL-6 RNA in infected cells from PEL and MCD were an order of magnitude higher than those from KS, arguing again for varying expression of genes in KSHV-related pathologies (Staskus et al., 1999)
vMIP-I, vMIP-II and vBck (ORFs K4, K6 and K4.1)
ORFs K4 and K6 encode chemokines that contain a C-C dicysteine dimer (Moore et al., 1996a; Russo et al., 1996; Neipel et al., 1997a; Nicholas et al., 1997). ORFK6 encodes vMIP-I and this chemokine blocks entry of CCR5 dependent primary HIV-1 strains by interacting with the CCR5 receptor (Moore et al., 1996c). vMIP-I also seems to be an agonist of the CCR8 co-receptor (Dairaghi et al., 1999; Endres et al., 1999). ORFK4 encodes vMIP-II and this chemokine blocks HIV-1 infection of CD4+ cells that express CCR3 (Boshoff et al., 1997). There is evidence that vMIP-II blocks HIV-1 entry through CCR3, CCR5, CX3CR1 and CXCR4 receptors and that vMIP-II can block HIV-1 entry into microglia in culture (Chen et al., 1988; Kledal et al., 1997; Hibbitts et al., 1999). These viral chemokines have other functions; vMIP-II is chemoattractive for eosinophils and Th2 cells and both vMIPs I and II are highly angiogenic in the chorioallantoic assay (Boshoff et al., 1997; Sozzani et al., 1998). Whether vMIP-II has agonist or antagonist activity on the CCR8 co-receptor remains to be settled (Sozzani et al., 1998; Dairaghi et al., 1999). ORFK4.1 encodes a viral chemokine, vBck, which is related to human MIP-1 beta and the macrophage chemoattractant protein although no functional data is yet available (Neipel et al., 1997b,c).
T1.1/ nut-1 RNA (ORFK7)
KSHV infected cells produce a 1.1 kb transcript termed T1.1/nut-1RNA that accumulates in the cell nucleus and is not translated into protein (Sun et al., 1996; Zhong et al., 1996). T1.1/nut-1 RNA is transcribed by RNA polymerase II, has a poly-A tail but does not have a trimethylguanosine cap (Sun et al., 1996; Zhong and Ganem, 1997). These studies suggest that this RNA is associated with ribonucleoprotein complexes and may be involved in the control of either viral or cellular transcript splicing.
Viral Bcl-2 shares a low degree of amino acid homology (15-20%) with human cellular homologs (Russo et al., 1996; Neipel et al., 1997b,c; Cheng et al., 1997; Sarid et al., 1997). Functional studies suggest that vBcl-2 inhibits apoptosis in yeast or Sindbis virus-infected cells although these studies disagree on the ability of vBcl-2 to form a dimer protein with human Bcl-2 (Cheng et al., 1997; Sarid et al., 1997). Overexpression of vBcl-2 delayed the process of apoptosis in 293 cells and an anti-apoptotic role in pathogenesis is postulated (Friborg et al., 1998).
ORF50 and ORFK8
The protein that is encoded by ORF50 is similar to the EBV lytic switch BRLF1 protein which belongs to the herpesvirus family of R transactivators (Rta) and it activates lytic KSHV replication (Russo et al., 1996; Lukac et al., 1998; Sun et al., 1998). The ORF50 transcript is classified as Class II/III or as an immediate early gene and also encodes ORFK8 (Sarid et al., 1998; Zhu et al., 1999; Seaman et al., 1999). The C-terminal region of ORF50 strongly activates transcription when targeted to DNA (Lukac et al., 1999). In this study ORF50 was considered a molecular switch as expression of an ORF50 protein that lacks the C-terminal region suppresses viral replication that should be induced by stimuli such as phorbol ester and sodium butyrate and also prevents spontaneous reactivation of KSHV from latency. Three percent of spindle-shaped cells in KS biopsies express Rta (Sun et al., 1999).
The major ORF50 transcript also encodes K8 and this protein resembles the ZEBRA/Zta/EB1 transcriptional activator protein of EBV (Gruffat et al., 1999; Seaman et al., 1999; Zhu et al., 1999). This protein is also expressed in 3% of spindle-shaped cells in KS biopsies as determined by in situ hybridisation (Sun et al., 1999).
ORFK8.1 is immunogenic, encodes the glycoprotein gp35-37 and splicing of the transcript yields ORFK8.1A and B proteins (Raab et al., 1998; Chandran et al., 1998a,b; Lang et al., 1999; Zhu et al., 1999). Interestingly the antigens that are detected by lytic IFA are predominantly ORFK8.1 (Li et al., 1999). This study has shown by immunofluorescence and immunoelectron microscopy that ORFK8.1 appears to be acquired by the KSHV virion during budding and antibodies to this protein were detected in 18/20 (90%) patients with KS patients and in 0/10 (0%) patients without KS. Cytotoxic T-cell responses have been detected against ORFK8.1 and also ORFK1 and K12 (Osman et al., 1999).
vIRF-1 (ORFK9) and vIRF-2 (unassigned ORF)
Viral IRF-1 and-2 are encoded by ORFK9 and an unassigned ORF upstream from ORFK11 respectively (Moore et al., 1996a; Burysek et al., 1999). Viral IRF-1 inhibits the interferon-beta signaling pathway in HeLa and 293 cells, can transform NIH3T3 cells and induce tumors when injected into nude mice (Gao et al., 1997). Viral IRF-1 again was noted to inhibit IFN-mediated signal transduction and cause morphological changes in NIH3T3 cells consistent with cellular transformation and similar inhibitory results to interferon stimuli were found in HUVEC cells (Li et al., 1998; Zimring et al., 1998). The inhibitory effects of vIRF-1 seem to be mediated by binding with cellular IRFs and p300 (Burysek et al., 1999a). Viral IRF-1 may also activate transcription (Roan et al., 1999). Viral IRF-2 also can bind specific cellular transcription factors (Burysek et al., 1999b). These studies provide evidence for viral regulation of cellular genes and there are two further vIRFs (ORFK10.1 and ORFK11) to be studied.
T0.7 encodes a small protein termed Kaposin whose RNA is expressed in abundance in unstimulated BCBL-1 cells and is detectable in the majority of KS tumors of differing epidemiological variants (Zhong et al., 1996; Staskus et al., 1997). Expression of this protein in Rat-3 cells produces focal transformation and subcutaneous injection into nude mice produced highly vascular and undifferentiated sarcomas (Muralidhar et al., 1998). A recent report shows that the genomic region around ORFK12 region yields several transcripts and proteins for which the functions remain to be determined (Sadler et al., 1999).
There are two variants of the ORFK15 membrane protein with several src homology 2 (SH2)-like motifs and potential tumor necrosis factor receptor-associated factor (TRAF) binding sites in its cytoplasmic tail. This protein may combine features of LMP1 and LMP2A which are involved in cellular transformation and maintenance of viral latency in B cells respectively (Glenn et al., 1999; Poole et al., 1999). This protein localizes on the surface and intracellular membranes of 293 cells and interacts with TRAFs 1, 2 and 3. The detailed functions of this protein are yet to be defined (Glenn et al., 1999).
The high degree of amino acid variation in ORFK1 has been discussed above and suggests that this transmembrane protein is under a selective pressure. Many clues to the importance and function of ORFK1 come from the transforming proteins of other gammaherpesviruses; ORFK1 is the positional analogue of saimiri transforming protein (STP) in HVS and the latent membrane protein 1 (LMP-1) of EBV (reviewed in Hayward, 1999). A role in cellular transformation is surmised following the demonstration that rodent fibroblasts that express ORK1 show evidence of transformation and that ORFK1 can replace the C-strain of STP in the HVS genome and induce lymphoma in the common marmoset (Callithrix jacchus) (Lee et al., 1998b). Mutational analysis of a chimaeric construct of human CD8_ polypeptide and the carboxy terminus of ORFK1 has demonstrated that the ITAM motif is functional and is required for signal transduction (Lee et al., 1998b). K1 multimerizes in the membrane and appears to transduce a signal in the absence of an exogenous cross-linking ligand (Lagunoff et al., 1999).
Viral GPCR can signal through the phosphoinositide-inositol triphosphate-protein kinase C pathway in the absence of an IL-8 agonist and also stimulates rat fibroblasts to proliferate (Arvanitakis et al., 1997). Expression of the G protein-coupled receptor in a focus-formation assay led to focal transformation of NIH3T3 cells and subsequent injection of neomycin-selected cells into nude mice caused tumors that consisted of spindle-shaped cells (Bais et al., 1998) In this study, media from KSHV-GPCR-expressing NIH3T3 cells but not controls stimulated growth of HUVECs and induced microtubule formation in the microtubule formation and co-culture Matrigel assays and this process could be inhibited by anti-VEGF antibody. In addition, KSHV-GPCR activates two protein kinases, JNK/SAPK and p38MAPK, which are activated by inflammatory cytokines that are known to induce angiogenesis and stimulate VEGF production in KS spindle cells (Samaniego et al., 1994; Cornali et al., 1996; Bais et al., 1998).Constitutive signaling from this receptor can be augmented by interleukin-8 and growth-related protein-_ and inhibited by human interferon-_-inducible protein, vMIP-II, stromal cell-derived factor 1-_ and GPCR-specific kinases (Geras-Raaka et al., 1998a,b,c; Gershengorn et al., 1998; Rosenkilde et al., 1999). Interestingly, the related human receptor, CXCR2, can transform NIH3T3 cells if the DRY sequence at positions 138-140 is changed to the VRY sequence that is seen is KSHV GPCR (Burger et al., 1999). The amino terminus of GPCR is necessary for high affinity chemokine binding but not for constitutive activity (Ho et al., 1999). However v-GCR appears to be only expressed in a subpopulation of KS spindle cells, which presumably undergo lytic replication (Kirshner et al., 1999). It is encoded on a bicistronic mRNA downstream of ORFK14 and protein expression may require an unusual mechanism such as translational reinitiation, internal ribosome entry or leaky ribosomal scanning (Kirshner et al., 1999).
Several serological and nucleic acid based assays have been designed to detect KSHV and despite the limitations of imperfect reference standards and assay variability, it is possible to discern geographic regions and populations with comparatively high and low KSHV seroprevalences. Although KSHV DNA and/or antibodies are detected in the majority of patients and strongly associated with all epidemiological variants of KS, it is clear that most KSHV-infected individuals do not develop KS. A co-factor appears to be required for this vascular tumor to be expressed clinically; HIV-1 is a necessary condition for AIDS-KS development and the contributory roles of inflammatory cytokines, angiogenic factors, HIV-induced immunosuppression and HIV-1 tat have been discussed. Immunosuppressive therapy plays a role in the development of Posttransplant KS, however the role of immunosuppression, immunoactivation and/or co-factors remain to be established for Classic and Endemic KS. Studies on KS pathogenesis need to consider both the host and the co-factor in addition to KSHV.
Seroepidemiological studies have provided useful data on KSHV transmission. There is general agreement that KSHV is sexually transmitted among homosexual men although the precise mode of transmission is still hotly contested. Regardless of this disagreement, the public health message to practise safe sex is clear. In endemic countries, transmission in childhood through close contact appears to be important. It is important to elucidate all modes of KSHV transmission as part of the drive to eliminate a preventable disease.
KSHV is present in all KS tumors. Functional studies of individual KSHV-encoded proteins provide endless possibilities for the way that KSHV may cause KS; vIL6 is involved in cell proliferation, vMIP-I and -II are angiogenic, vBcl-2 could be anti-apoptotic, the interferon regulatory factors may downregulate the host response to interferon. Kaposin, ORFK1 and vIRF-1 may be transforming proteins and ORF74/GPCR has both angiogenic and transforming properties. Cellular defense mechanisms such as cell cycle shutdown and apoptosis may be prevented by vBcl-2, v-cyclin and vIL-6.
It is interesting that very few of these genes are expressed in the majority of KS spindle cells. Latent genes such as v-FLIP (anti-apoptosis), kaposin (cell transformation), LANA (episomal maintenance) and v-cyclin (cell proliferation) are expressed in most KS spindle cells indicating latent persistence of the virus. Expression of early/late lytic genes such as vGPCR in a few spindle cells may however contribute to pathogenesis e.g. by increasing the release of cellular growth factors such as VEGF. This may be explained that by the time KS is apparent, the destructive lytic cycle is no longer needed.
KSHV has been associated with a number of other diseases and these associations need to be explored. Little doubt remains around the relationship with Multicentric Castleman's Disease and Primary Effusion Lymphoma and these disease phenotypes may be dependent on tissue specific expression of KSHV genes. The availability of a cell culture system an/or an animal model will extend the studies that can be done on KS pathogenesis and also the evaluation of antiviral agents and pathogenesis-based therapies. The ultimate reward for studying a viral-induced tumor is that one day this disease should be eradicated.
I would like to thank Dr. Julie Greensill, University of Liverpool, and Margreet Bakker, University of Amsterdam, for their invaluable assistance