Rebuttal to Berenson and Vescio

REPRODUCIBILITY is a requirement in establishing whether a scientific finding is valid. Not only must results be reproducible, but they must be routinely reproducible in multiple laboratories. The field of KSHV research in its short history has a number of examples of startling reports of disease associations, including sarcoidosis, posttransplant skin tumors, angiosarcoma, and T-cell lymphomas that have not been reproducible.1 The dispute over whether KSHV plays a role in multiple myeloma involves the reproducibility of the initial findings based on Rettig et al.2 

Aside from Berenson’s group, three other groups have reported results that could be consistent with the hypothesis that KSHV is causally associated with multiple myeloma.3-5 Brousset et al3 have been cited as a group that was able to confirm the association in their French series. Although Berenson has asserted that KSHV/HHV8 cannot be detected by PCR on fresh bone marrow biopsies, nevertheless, Brousset et al3 were able to detect KSHV/HHV8 KS330₂₃₃ from 18 of 20 acetone-fixed and paraffin-embedded bone marrow biopsies. Agbaliki et al5 were also able to detect KSHV from 5 of 10 paraffin-embedded bone marrows from patients with MM. In contrast, many other groups have not been able to detect KSHV/HHV8 even in fresh bone marrow cores or aspirates.6-12 

The debate then shifts to the type of cell that is permissive for KSHV/HHV8 infection in the bone marrow. Apparently, the virus is only readily detectable after long-term culture in a subpopulation of cells of “unknown phenotype.”13 Several laboratories have tried culturing stromal cells, some using the protocol of Rettig et al, to amplify the virus-carrying cell type and have failed to detect KSHV/HHV8.6,8,11,14-16 Contrary to Berenson et al’s rebuttal, the sensitivity of the PCR assays used in some of these studies was quantified and was found to be extremely sensitive, in the case of KS330₂₃₃, detecting a few ORF 26 copies of 10⁵ cells.6,7,14 Berenson et al mention that the presence of viral DNA was “confirmed” using primers to various KSHV ORFs, referring to their unpublished or in press reports. Furthermore, the original basis for suggesting a connection between KSHV/HHV8 and multiple myeloma was the possibility that the virus-encoded vIL-6 protein acts as a paracrine factor for myeloma cells.2 However, even assuming that KSHV/HHV8 is present, no KSHV vIL-6 protein is detectable by immunohistochemistry in any multiple myeloma lesions,6 eliminating the theoretical underpinning for this hypothesis.

Berenson et al suggest that KSHV in situ hybridization studies confirm their PCR-based observations.2,17 These studies purport to show a majority of cultured cells infected with KSHV. Furthermore, the virus must be present in multiple myeloma tissues at greater than homeopathic amounts to drive tumor cell replication through a paracrine mechanism. When only one virus copy is present per cell (the level of infection for mixed tumor and nontumor KS lesions), virus is readily detectable by Southern hybridization and does not require PCR amplification for detection.18 If the strong in situ hybridization signal is truly reflective of the degree of infection by KSHV, then any laboratory should be able to detect virus by PCR or even Southern hybridization. A λ phage library could be easily made from the degree of infection reported by Berenson in in situ hybridization studies. Have larger portions of the viral genome been pulled out from such libraries?

With respect to serologic testing, only Gao et al4 have found any significant positivity for KSHV/HHV8-specific antibodies in multiple myeloma patients. Careful examination of this report shows that the immunofluorescent assay (IFA) detecting LANA (ORF73) antigen is positive in only 3 of 27 (11%) of multiple myeloma patients, whereas this group found 14 of 27 (52%) positive by LANA immunoblotting. Examination of the “LANA” immunoblot bands in Fig 1 of this report is not convincing for true seropositivity. Additionally, they also found the sera of control cancer patients to have a prevalence of KSHV/HHV8 antibodies higher than expected for the general population. In contrast, many other groups familiar with these seroassays have not been able to show an association between KSHV/HHV8 and multiple myeloma.6-9,12,16,19-21 Berenson et al make a point that there are basepair changes in the sequence of ORF 65 in multiple myeloma patients that would result in a negative or insensitive ORF 65.2 Western seroassay. It should be stressed that the indirect IFA frequently used in KSHV serologic testing detects ORF73 not ORF65, so basepair changes in ORF65 should not affect the results using IFA.

In summary, neither viral DNA nor antibodies can be reproducibly found in multiple myeloma patients. There now exists a large enough body of data from multiple laboratories that do not support the findings of Rettig et al,6-12,14-16,19-24 and the known epidemiology of KSHV does not make sense with the known epidemiology of multiple myeloma.25 KSHV research is young, and several very competent laboratories have been misled by technical artifacts that are part of the natural growing pains for any new field. Although we would be delighted if Berenson et al’s results implicating KSHV in the pathogenesis of multiple myeloma were correct, however, this does not appear to be the case.