Technical Back and Forth between a critique of the first study and Dr. Mikovits
This is real technical stuff but perhaps someone has the experience to understand it.
http://f1000biology.com/article/id/1166366/dissent
Dr. Mikovits does agree that they should have discussed unpublished results in the media.
Matsuo Shuda and Patrick Moore
The discovery of the cause of chronic fatigue syndrome would be an extraordinary finding. Rather than providing extraordinary proof, this manuscript has flaws that leave the reader unsure of knowing precisely what was measured.
To detect the xenotropic murine leukemia-like virus (XMRV), the authors used nested-PCR on non-randomized and non-blinded samples, a recipe for uncontrolled PCR contamination. This technique re-amplifies previously cycled products and is inherently prone to intermittent false positivity that has occurred in our lab and many others (e.g. {1} and {2} on which I am the author). This is a concern in light of post-publication claims that XMRV detection rates among chronic fatigue syndrome (CFS) patients have climbed from 67% to 95%, and XMRV tests are now being sold and advertised on the internet at
http://www.redlabsusa.com . Southern blotting, which would allay this suspicion, was not done. Other results in the study also lack support. Flow cytometry and immunostaining with murine leukemia virus (MLV) antibodies were used to directly detect viral proteins in patient cells (see Figure 2A of the paper). The CFS peripheral blood cells have robust monotonic staining rather than the bimodal peaks that are expected from a mixture of infected and uninfected populations of peripheral blood cells. It is not certain whether this level of viremia for an exogenous retrovirus is medically possible. It may, perhaps, be possible but it seems improbable and is a pattern more consistent with a cross-reactive endogenous retroviral antigen. To confirm this finding, CFS peripheral blood cells (without negative controls in Figure 2B) were immunoblotted using cross-reactive spleen focus-forming virus (SFFV) and MLV antibodies. XMRV gp70 and p30 proteins are found at higher levels in 2 out of 5 CFS peripheral blood samples (1150 and 1221) than in the positive control -- HCD-57 cells directly infected with SFFV -- a very remarkable result. Repetition with negative control samples (see Figure 2C of the paper) has the higher molecular weight bands cut from the photograph, thus we cannot interpret potential positivity for p30 gag precursor proteins among the control samples (see CFS samples 1199 and 1220 in Figure 2B). Finally, the positive control HCD-57 cell lane in Figure 2C lane 8 has a completely different banding pattern from the very same control in Figure 2B lane 7 for the p30 gag protein. The elementary issue of whether the authors are measuring XMRV has to be clarified. The fundamental basis for the CFS case and control samples is also not defined at an appropriate level. The samples (supplementary online material) were "selected for this study from patients fulfilling the 1994 CDC Fukuda Criteria for Chronic Fatigue Syndrome (S1) and the 2003 Canadian Consensus Criteria for Chronic Fatigue Syndrome/myalgic encephalomyelitis (CFS/ME) and presenting with severe disability". These are two separate definitions, the latter published in the "Journal of Chronic Fatigue Syndrome" (which is no longer in print). It is unclear how the samples were selected from these two criteria. No references or cut-offs are given for tests used to clinically define the CFS patients as cases so we are unable to interpret the essential basis for the study. In addition, no description is given to indicate that controls were tested in the same manner as CFS patients; in fact, there is no description for negative control samples at all. For a disease whose diagnosis is controversial, a clear statement of where and how the cases and controls were selected is a critical first step.
Competing interests: None declared
Evaluated 18 Nov 2009
Author Response: Judy Mikovits, Whittemore Peterson Institute, Reno, United States
This dissent first discusses "the cause of CFS". We did not imply that XMRV caused CFS. We specifically state that our observation "raises several important questions". Is XMRV infection a causal factor in the pathogenesis of CFS or a passenger virus in the immunosuppressed CFS patient? The work presents a testable hypothesis that XMRV has a role in CFS pathogenesis. The key task for the scientific community is to define the scope of human disease associated with XMRV infection. We were highly concerned and vigilant about taking precautions to avoid PCR contamination. This was minimized using laboratory controls such as dedicated first- and second-round PCR areas in different buildings and consistent treatment of instruments and work areas with DNA ZAP and UV rays.
Although the original screen was performed using nested PCR, PCR positivity on a subset of samples was confirmed using single-round PCR in a different facility (Fig1A). Although not in the manuscript, PCR status was verified on identical samples not processed at WPI in a XMRV-free lab at NCI. Moreover, in normal samples analyzed in parallel, amplified sequences were found in only 3.7%, making it unlikely that 67% positivity reflects "uncontrolled PCR contamination". We should not have discussed unpublished data, particularly considering the lay media. To discuss more details about our more recent studies would repeat that error, but we were able to culture XMRV virus from plasma and detect antibodies in some samples which were PCR-negative (e.g. see #1118 in Fig1, 2A, 2D). These examples compelled us to further study PCR-negative patient samples. Flow cytometry data in the study were not meant to determine in vivo levels of XMRV protein expression or the level of clinical viremia. The goal was to determine if the XMRV DNA sequences detected represented the presence of infectious viral particles. The studies of XMRV protein expression in peripheral blood mononuclear cells (PBMC) were performed after PBMC were activated in culture with phytohemagglutinin and interleukin-2 for 7-14 days to allow the virus to spread through the culture. It should have been made clearer.
Although at earlier times these samples did have bimodal peaks, data shown were when most cells were virus positive. Not included in the paper was the ability of azidothymidine to block XRMV spread in vitro. The methodology for the immunoblots was also questioned. We used monoclonal antibodies which recognized the envelope of all xenotropic and polytropic but not ectopic murine leukemia viruses and reacted with XMRV env proteins of expected sizes. Concerning the relatively weak signal of the positive control, HCD-57/SFFV, test sample lanes contained 150-200ug of protein, only 30ug was loaded in the HCD/SFFV lane to prevent the positive control signal from overwhelming adjacent lanes. This study was the initial finding of infectious XMRV virions in human blood and a second association of XMRV and human disease.
Our observation of actively replicating virus, as well as antibodies directed against XMRV (which were not criticized) in the patient population examined, strengthened the paper and the hypothesis that this recently discovered virus is a human pathogen.
Clinically, CFS is a heterogeneous syndrome with diagnosis made on the basis of the exclusion of other diseases. Thus, the basis for diagnosis varies greatly and we expect there to be XMRV-positive and -negative CFS patients. Additional large-scale clinical studies using control groups are essential to determine whether XMRV is the cause of CFS. Also, rigorous independent validation is crucial. To facilitate this, the NCI, Cleveland Clinic and WPI are making virus reagents available through the NIH AIDS repository to any academic investigator by contacting the investigators involved.
Response added 7 Jan 2010
