We used PCR and serologic testing to study the prevalence of XMRV in 162 well-characterized North American patients with prostate cancer. We surveyed cancers with intermediate to advanced stages, and utilized multiple PCR assays, including tests with conserved primers to allow the detection of diverse XMRV and MLV-related sequences. Nearly half the specimens were also tested in triplicate to maximize the detection sensitivity of low-copy sequences. We used a broadly reactive WB assay to test all patients for antibodies. Our findings document a very low prevalence (1.9%) of XMRV sequences in prostate tissue DNA and absence of antibody positivity in all specimens. Combined these data do not support an association of XMRV or related MLV with prostate cancer.
The predominantly negative PCR results were observed despite the use of 1 ug of DNA which is 4–10× higher than the input DNA used in previous studies that reported XMRV detection 
. Likewise, the observed low prevalence of XMRV may not be explained by a decreased PCR assay sensitivity since we screened specimens with the same assay used originally by Urisman et al. 
. In the three specimens with detectable sequences, we noted that the copy number was very low as nested PCR and replicate testing was often needed for detection. Importantly, in all three specimens we were unable to detect mouse mtDNA despite the use of a highly sensitive assay, which suggests that the source of XMRV is unlikely the result of contamination with mouse DNA. These results are important since mouse DNA contamination was reported to be the source of XMRV in a recent study of prostate cancer tissues 
. The authors of this study used a PCR assay for the mouse intracisternal A particle (IAP) which they reported to be more sensitive than a mouse-specific D-loop mtDNA-based assay 
. Although DNA specimens were not available for testing in the IAP test, we have found that the IAP assay is equally sensitive to our COX2 real-time mtDNA assay (data not shown), thus further excluding mouse DNA contamination as a source for our positive PCR results. In addition, the DNA specimens were prepared at FCCC where only human biological specimens, and not cell lines, are handled, reducing further the possibility of murine contamination.
Sequence analysis of the PCR-positive specimens was highly informative because it confirmed that all three specimens were XMRV-related. Also, the finding of a viral strain in three prostate cancer patients that is distinct from the XMRV seen in previous studies is significant and demonstrates a broader viral diversity 
. This would be an expected result consistent with virus evolution during spread and persistence. The absence of antibodies and plasma viremia in these three patients is noteworthy because it may reflect sequestered or latent infections. Loss of antibody during a latent infection, while atypical of most retroviral infections, has been described previously for natural infection of macaques with simian type D retroviruses (SRV) 
. SRV in macaques is associated with outbreaks of severe immune deficiency in primate colonies and latent SRV infection in these antibody negative animals is confirmed with greater sensitivity using PCR analysis 
. Our results are also consistent with those seen recently in macaques experimentally infected with XMRV in which tissues at necropsy are PCR-positive but viremia and detection of provirus in PBMCs disappear quickly, followed by loss of antibody detection 
. Although one study reported the detection of XMRV neutralizing antibody in 11 patients with prostate cancer, these data were not confirmed by more sensitive methods such as WB, or by PCR testing in all cases, and thus may represent nonspecific reactivity 
. Longitudinal studies may better define host responses to XMRV infection.
Our study population contained 15 persons with the mutant QQ RNase L allele. However, none were XMRV-positive, which contrasts with the original findings by Urismann et al. 
. The XMRV-infected persons in our study had either the homozygous wild-type (RR) or heterozygous (RQ) RNase L R462Q alleles. Our results are consistent with other US and European studies which also identified higher frequencies of the homozygous QQ mutant allele in XMRV-negative persons 
. Combined, these results suggest that XMRV infection is not associated with this allelic form of RNase L.
Although our data do not support an association of prostate cancer with XMRV, it is important to understand whether XMRV has any causal role in prostate cancer when it is infrequently detected. In general, gammaretroviruses like XMRV induce malignant transformation by insertional mutagenesis, so that malignant cells in a tumor are all clonally infected. This mechanism of carcinogenesis has been found in animals as well as in children exposed to MLV-derived vectors in gene-therapy trials 
. Therefore, the low-frequency of XMRV-infected cells found in all three patients are inconsistent with a direct role of XMRV in the prostate carcinogenesis in these patients. Previously, a human prostate cancer cell line, 22Rv1, has been shown to express high levels of XMRV, a finding that supports a role of XMRV in prostate cancer tumorigenesis 
. However, it was reported recently that the XMRV expressed by 22Rv1 may not be of human origin, but most likely arose via recombination of two overlapping XMRV-like genomes during passage of the prostate tumor in inbred mice (T. Paprotka et al.
Conference on Retroviruses and Opportunistic Infections). Furthermore, others have shown that XMRV integration sites cloned from prostate tissues of two of nine patients may have been the result of contamination with DNA from experimentally infected DU145 cells 
, while other XMRV integration sites are not near tumor suppressor genes or proto-oncogenes 
. Combined, these results raise questions about the role of XRMV in prostate cancer. Nonetheless, more work is needed to better understand the prevalence of XMRV and MLVs in humans and their role in human disease.