Please sign the petition: http://www.change.org/petitions/an-open-letter-to-the-editors-of-science-2
To the editors of Science:
I was greatly saddened to see your May 31, 2011, “Editorial Expression of Concern” (EEC) regarding Lombardi et al (Oct 2009).
As rightly stated by Judy Mikovits (1), this action on your part is “premature” to say the least. In fact, your EEC all but admits this outright: “Science eagerly awaits the outcome of these further studies and will take appropriate action when their results are known.” I dare say you've already taken action (albeit of a purely inappropriate sort) in the form of this editorial and your request for retraction. Unfortunately, these actions will do little but cast a shadow of intimidation over the future of XMRV/HGRV research, a future that you simultaneously concede is still playing out. Was it your intention to foster intimidation and short-circuit the scientific process?
In order to partially justify your concern over Lombardi et al, you state: “Since then, at least 10 studies conducted by other investigators and published elsewhere have reported a failure to detect XMRV in independent populations of CFS patients.” This line of reasoning is fundamentally flawed in at least two ways:
Firstly, it selectively ignores any and all evidence supporting Lombardi et al. Much of this evidence has been concisely presented by Judy Mikovits in her response (1), so I will not repeat it in detail. Nonetheless, the palpable bias of this statement is very disconcerting.
Secondly, this argument belies a woefully unscientific reliance on counting studies instead of analyzing methods. None of these negative studies, including Knox et al, represents a genuine replication attempt. This has been made clear by Annette Whittemore’s detailing of many (though not all) methodological differences in response to your EEC (2). It should not need to be reiterated that no number of non-replication studies can ever substitute for genuine replication, especially when employing techniques as intricate as those in question. How is it that the world’s premier scientific publication can be persuaded by quantity over quality, by tally over technique, by rhetoric over replication?
This latter issue is problematic for other reasons as well. It creates a self-fulfilling outcome dictated largely if not entirely by money and influence. Quantity is trumpeted as “consensus,” which in turn creates perceptual bias in favor of this faux “consensus.” It is well-known that since the publication of Lombardi et al, its authors have been denied for funding numerous times (six-plus times by the NIH) and have been blocked from publishing further evidence in support of their original paper. The truism that “negative studies are rarely published” has been turned upon its head with respect to XMRV research, where those who have demonstrated an ability to find the virus are starved for resources and shunned at every turn, while those who demonstrate abhorrence toward genuine replication are lavished with funds and granted journal space without so much as the peer-review equivalent of a gentle pat down.
You further support your concern by appealing to the conclusions of Knox et al and Paprotka et al.
With respect to Knox et al, it is made clear in the response from the Lombardi et al authors (1, 2, 3) that no attempt was made to faithfully replicate their methods. PCR is a complex process dependent on many variables, many if not all of which were modified by the authors of Knox et al. Their assays failed to detect XMRV in clinically positive samples. As such, Knox et al is merely another "failure to detect" using novel, clinically-unvalidated methods. I would normally have thought such banal attempts rather below the stature of Science. Certainly no amount of publication prominence can transform absence of evidence into evidence of absence, as your EEC implies.
The issue of clinical assay validation is a very critical point worth exploring further. None of the authors of negative studies to date have proven the ability of their assays to detect virus in human clinical samples. In the cases of Knox et al and Shin et al, the authors clearly demonstrated the insufficiency of their assays against clinical positives. In all of these negative studies, only the analytical sensitivity and specificity of their assays was shown, and this was assumed to be sufficient. Needless to say this would not be sufficient for any assays intended for clinical diagnosis. As the issue here is to determine disease association with a retrovirus with a very low copy number, it is vital to prove that one's assay can detect viral nucleic acid sequences in their natural matrix and naturally occurring structures (circular DNA, pre-integration complexes, and integrated DNA), which are very different chemically and physically from plasmid spiked into water, placental DNA, or even blood.
With respect to Paprotka et al, you claim that “laboratory contamination with XMRV produced by a cell line (22Rv1) derived from these early xenograft experiments is the most likely explanation for detection of the virus in patient samples.” This claim embraces a profound, unsupported leap from possible origination event to ubiquitous presence in laboratories that have never used any known contaminated materials. It is also strange that this presumed (not proven) contamination seems to have a strong and persistent affinity for patient samples over control samples.
Furthermore, the conclusion of Paprotka et al that this supposed recombination is the "most likely" origination event is largely speculative and powered primarily by the exclusion of very plausible alternative explanations. For example, their failure to find XMRV in progenitor prostate cancer xenografts does not conclusively rule out the possibility that one (or more) of these xenografts was already infected (especially with the data showing prostate cancer association). Given the difficulty of many researchers in finding XMRV via PCR, as well as the fact that propagation of these cell lines requires use of materials (e.g. testosterone) that would induce replication, it is quite plausible that Paprotka et al were only able to detect XMRV in the resultant cell line despite its earlier introduction via one of the xenografts. This explanation could be deemed speculative, but unfortunately it is no less supported by their evidence than their own conclusions.
Lastly, I want to point out the unscientific nature of retraction itself. Barring cases of clearly demonstrated fraud or manipulation (charges that no one could plausibly level at Lombardi et al), all data is important to the scientific process, both true and false. Science is the process of data accumulation; it is a record of both mistakes and successes. It is not a retrospective whitewash of everything except what is deemed correct at present. For one thing, some "mistakes" are vindicated in time. For another, genuine mistakes are still instructive to other researchers. If, as many seem to believe, Lombardi et al is the result of pervasive and devious (to the point of mimicking a real human infection in multiple ways!) contamination, is there truly no value in retaining the one study that has most thoroughly revealed the extent and nature of this issue and the potential experimental errors that exacerbate it?
As the gatekeepers of the world’s preeminent scientific publication, I believe you have a moral obligation to steward the honest pursuit of truth, no matter how frail and beleaguered, through the seething proclivities of human weakness. Here you have not only failed in this task, you have sacrificed honest inquiry at the altar of expediency. I am left wondering: what happened to science, Science?
In light of these concerns, I believe it behooves the editors Science to publicly account for the following questions:
1. Was Science made aware of the gross conflicts of interest between Konstance Knox and the WPI prior to its publication of Knox et al? If so, how did it deem these conflicts of interest acceptable and why were they not included as part of the publication? If not, will the publication of Knox et al be reconsidered with respect to this information? I sincerely hope that Science hasn't become a willful vehicle for vendetta.
2. Who were the peer reviewers of both Paprotka et al and Knox et al? Have any of the reviewers displayed prior public bias against the original Lombardi et al findings?
3. Will you grant the authors of Lombardi et al time and space to augment the findings of their original paper with new data?
4. Will you consider retracting your “Editorial Expression of Concern” in light of the numerous and varied criticisms leveled by myself and others?
I sincerely hope you will provide answers to these very pertinent questions in order to restore faith in your publication following the rather brazen and unwarranted issuance of this editorial.