In the April 2006 issue of the Journal of NeuroVirology, Poltermann et al (2006) report a lack of association between human cytomegalovirus (HCMV) and glioma. This subject is highly controversial, and the debate over the role of HCMV in glioma pathogenesis is ongoing. Given the fatal nature of these tumors, and the fact that very little is known about their etiology, extreme technical proficiency and scientific scrutiny must be used when examining potential causes. Several technical issues described in the article should be addressed to allow more accurate evaluation of the findings.

With regards to the immunohistochemistry protocol: First, the authors report using brain tumor sections 8 μm thick. Such thickness is typically used for surface staining and is known to be too thick for antigen-antibody binding. For optimal staining, a thickness of 6 μm is suggested (Cobbs et al, 2002). This allows complete deparaffinization, which is critical for proper enzyme digestion and antigen retrieval. Second, the authors fail to describe any postfixation of tissue sections with formalin prior to digestion, as well as any pepsin digestion before antigen retrieval. Formalin treatment (which stabilizes the antigen) is critical for epitope conditioning, and pepsin digestion is critical for HCMV detection because pepsin cleaves specific peptide residues that interfere with antigen-antibody binding. Third, the authors describe antigen retrieval using Citra buffer. This approach results in false-negative staining that could be overcome by combining both pepsin digestion and antigen retrieval (Polak et al, 2003) at 45◦ C to 50◦ C for 2.5 h; false-negative staining may result