• human herpesvirus 8
• human immunodeficiency virus 1
• seroprevalence
• polymerase chain reaction
• immunohistochemistry

Recent studies have assessed the seroprevalence of human herpesvirus 8 (HHV-8) in patients infected with human immunodeficiency virus 1 (HIV-1) and healthy blood donors in Brazil.^1–3 We decided to assess the prevalence of HHV-8 infection by another strategy, based upon the polymerase chain reaction (PCR) and immunodetection of HHV-8 positive cells in a series of tissue samples from patients living in the São Paulo state in Brazil. For that purpose, we used three different primers corresponding to three different regions of the viral genome, using a PCR protocol described previously.⁴ In parallel, we used primer sets specific to the open reading frame K1 (ORF-K1) sequence to classify the viral strains.⁵ Lastly, we applied immunohistochemistry with an anti-latent nuclear antigen 1 (LNA-1) (LN53; ABI, Columbia, Maryland, USA) antibody on routinely processed paraffin wax embedded sections of lymphoid tissues.⁴

The patients (n = 82) were selected randomly at the department of pathology of the State University of Campinas (UNICAMP), São Paulo, Brazil. The samples comprised reactive lymph nodes removed from adult patients (HIV negative), treated for cancer of the gastrointestinal tract, male and female urogenital tract, and head and neck. Ages ranged from 22 to 87 years. DNA was found to be intact in only 25 patients. Three of these 25 patients had HHV-8 DNA sequences in the lymph nodes, as assessed using three different sets of primers—ORF-26, ORF-72, and ORF-75—after one round of PCR, as described previously.⁴ In addition, according to the description of subtypes by Zong et al,⁵ the analysis of the ORF-K1 sequences showed that in one patient HHV-8 belonged to group A (subtype A) and in the other two the virus belonged to group B (subtype B). Unexpectedly, immunohistochemistry was negative in all 82 patients, including the three PCR positive ones. Overall, our results, although based on a small series, are in keeping with those published by Caterino de Araujo et al.¹ A recent interesting serological study performed in 781 Brazilian Amerindians from different tribes revealed a high prevalence of HHV-8 in this population.³ In that study,³ the viral strains were of different subtypes, but the authors described a new variant (subtype E), which is hyperendemic in Brazilian Amerindians. Therefore, as suggested by our study and that of Caterino de Araujo,¹ HHV-8 subtype B seems to prevail in the Saõ Paulo state. We were surprised by the lack of LNA-1 positive cells in tissue sections from the three PCR positive patients. This could result from the scarcity of HHV-8 infected cells or a restricted expression of latent genes, such as LNA-1, in reservoir cells. This phenomenon has been seen in Epstein-Barr virus (EBV) positive bystander lymphoid cells with the lack of expression of most latent genes (EBV encoded nuclear antigen 2 negative, latent membrane protein 1 negative).

We cannot completely exclude technical problems related to antigen alteration by tissue processing, but samples from patients with Kaposi's sarcoma from the same institution, processed with the same protocol, were strongly positive for LNA-1. HHV-8 serology was not available for these patients, but our results imply that the real incidence of HHV-8 infection is underestimated. However, the correlation of PCR detection with serology should be undertaken in a larger series of patients to determine the seroprevalence of the virus, at least in São Paulo state. In the report by Biggar et al there was a discrepancy between the serology results and PCR on mononuclear cells from peripheral blood of seropositive patients.³ This last technique appears to be less sensitive in detecting HHV-8 positive patients but remains mandatory to perform a strain subclassification.