XMRV and Occams Razor

Long time readers of ERV know that I do not believe a 'new' retrovirus, XMRV, is the causative agent of any human disease. It does not make sense as a real human pathogen, unless you disregard field basics (or make up new 'rules').

New cell transformation/cancer rules.

New transmission/epidemiology rules.

New immunology rules.

New virology rules.

New genetics/population biology/evolution rules.

Over and over and over with XMRV, Ive typed in frustration "THIS DOES NOT MAKE SENSE!!!!!"

Occams Razor, though not always useful in science, does provide us with a basic logical guide: Either XMRV is a real human pathogen that requires the upheaval of >decades of basic biology... or XMRV is a contaminant.

Its probably the latter, but that is just pure logic. Not science. Reality and logic do not always go hand-in-hand, especially with biology. Biology is a messy, silly process sometimes (see: evolution), and XMRV might indeed be a messy, silly virus that behaves bizarrely in humans. I can scream "THIS DOES NOT MAKE SENSE!!!!!" all I want, and it might not, but it still might be true.

While I was gone, a series of papers were published in Retrovirology that gave experimental support for the logical conclusion that XMRV is a contaminant:

Contamination of clinical specimens with MLV-encoding nucleic acids: implications for XMRV and other candidate human retroviruses (review)
Disease-associated XMRV sequences are consistent with laboratory contamination (blagged here)
An endogenous murine leukemia viral genome contaminant in a commercial RT-PCR Kit is amplified using standard primers for XMRV
Contamination of human DNA samples with mouse DNA can lead to false detection of XMRV-like sequences
Mouse DNA contamination in human tissue tested for XMRV

What does this research mean?

Its a collection of putative explanations for confusing results:

Why you can 'find' XMRV in some labs, and not others, in different parts of the world. Different people use different reagents. Its as variable as what kind of pop people like to drink. My labmate likes Superscript III for reverse transcription. I prefer a different brand for the same damn process. The lab next door might use something different. A lab in Germany might use something different. One of the papers showed how a reagent used in the 'XMRV positive' papers is absolutely contaminated by mouse DNA, thus can absolutely give false positive results via mouse ERVs. Other reagents popular in other labs were not contaminated ('XMRV negative'). Does this mean that all of the 'XMRV positive' papers that used the contaminated reagent were also contaminated and should be retracted? No. It means that this is a potential explanation for the discrepancies in XMRV positive/negative labs, and the pos labs must investigate this before they publish again.

Contamination can happen at collection. PCR reagents are not the only way you can get false positives. Another paper showed that their own samples, the actual samples, had mouse DNA in them. The second they were collected, they were contaminated, and they showed exactly what kind of test you need to perform to make sure that future samples are not likewise contaminated. Does this mean that pos papers were also contaminated prior to any other manipulation? No. It means that post labs must perform this test in the future to demonstrate that they dont have contaminated samples at the starting gate.

Cell lines are contaminated. I do HIV-1 research. When I want to do experiments in the lab, I can use primary cells, or cell lines. Primary cells would be like CD4+ T-cells purified from blood donated at a blood bank. Yes, they are very biologically relevant (they ARE the cells HIV-1 infects), but there are some problems. Primary cells are not all the same (from the same person, or from different people), they only live a couple of weeks, you can only get so many from a patient, meh. Another option is using cell lines. 'Immortal' (continually dividing, not literally immortal) pretty much identical cells. Not exactly biologically relevant (like my U87 cell line is actually glial cells forced to express CD4 and CCR5 or CXCR4-- HIV aint normally infecting glial cells), but good for consistency and quantity. Another problem? Cell lines can have a stage in their development in mice. This can lead to contamination with mouse viruses, like what happened to TZM-bls. Turns out MLV-like-viruses are in lots and lots of cell lines, and could be a huge potential source of false positives. This might also explain why the WPIs archaic-yet-proclaimed-better culture 'test' (I really cant emphasize enough how outdated viral culture is in 2010) is 'better'-- the longer you culture cells, the more you increase the contaminant false positive signal.

PCR protocol can effect detection of contamination. Okay, lets say you have a sample. It might be contaminated with mouse DNA. You work in a lab that might be contaminated with mouse viruses. Youve got PCR reagents you bought from the store. It might be contaminated with mouse DNA. Is there any way you can detect XMRV, and ignore all the contamination? Maybe. You use primers that are specific for what youre looking for, and will ignore stuff that kinda looks like what you are looking for, if you can. You must (at least) look for mouse IAPs as a marker for contamination (not mitochondrial DNA). You do not do what the 'XMRV positive' papers did and use nested PCR, which will amplify even the tiniest bits of anything that kinda looks like what you are looking for. Does this mean that the 'XMRV positive' papers amplified non-specific crap? No. It means they need to be extraordinarily careful in their primer design, and should not just use a primer set/PCR method that someone else used just because someone else used it. You must also extensively study the amplified sequence to ensure that the sequence is real (see below).

The stuff being sequenced is almost certainly crap. You cannot just sequence your 'new' virus and declare 'Mission Accomplished'. You cannot make value claims like 'there was sequence evolution' or 'there is sequence diversity' based off of eye-ball 'measurements' of your sequence. You need to collect your data and properly analyze it-- extensively. Or someone else will do it. And your 'eye-ball' might not match reality. This particular criticism I do lay at the feet of the 'XMRV positive' paper authors who made claims without the science to back them up. Particularly frustrating were these sequences very obvious homology to endogenous retroviruses, yet absolutely no serious investigation into contamination. XMRV bizarrely did not behave as a quasispecies, unlike every other retrovirus on the planet, yet no one appeared to be interested in investigating this astounding 'fact'. Shocking.

What are the long-term effects of this research? Does it 'prove' XMRV isnt real?


Its a bullshit filter.

Experimental evidence for Occams Razor-- XMRV is a contaminant, not a real human pathogen.

Because this is now published information, authors of 'XMRV pos' papers can no longer harumph 'ITS NOT CONTAMINATION!' to the press while doing no investigation into contamination. It is now part of the literature that reagents, mouse DNA, and viral RNA/DNA can confound results, and that sequence analysis supports this. You cannot publish an XMRV positive paper without now investigating and directly addressing all of this anymore (unless you publish in PNAS, I suppose...).

Contamination cannot be side-stepped or ignored anymore.



More like this

If Ive said it once, Ive said it a thousand times here on ERV-- Scientists are wrong all the time. ALLLLLLLL the time. Its what we *do*. We make a hypothesis, design an experiment, collect data, and refine the hypothesis, because the original hypothesis was wrong. Over and over and over and…
Judy Mikovits is playing whack-a-mole in her responses to the legion of anti-XMRV papers published to date: She bitches about one thing, or another thing, as the papers are published, but she misses the big picture. Its not that there are moles randomly popping up. The problem is youre entire 3…
Sorry, Hitch. Religion doesnt poison *everything*. Thanks to the bumbling efforts of 'respected' scientists like Judy Mikoivts, we now know that its not religion, but mouse DNA, that poisons everything. Relatively new readers of ERV are well aware of the fact that all efforts to connect XMRV to…
The simplest, simplest definition of evilution: Change over time. Your children are different from you. You are different from your parents. Your parents are different from their parents. And so on and so on and so on. But if Person A is older than you and has different DNA, that doesnt…

Great write up.

Peoblem is that patients have already made up their minds. Any cognitive dissonance that will occur because of future results will be reduced by assuming researchers are even more evil or the conspiracy is even vaster than they had first assumed....

Lovely clear explanation, thanks.

Shouldn't labs be using positive and negative controls in these experiments to pick up contamination problems? So if they find XMRV in a control which shouldn't have anything expect reagents in it at all, there's something completely screwy going on?

Apologies if this is an inappropriate or stupid question; when I was last in a lab there was no such thing as sequencing. So I can't fully imagine the procedures. But I remember one problem with the lab doing "Dr" Wakefield's measles tests was that its controls were either non-existent or screwy. (Yes, I know Wakefield's stuff was different, so I don't know how comparable the problems are.)

And, if I'm right, shouldn't the peer review process identify inadequate controls as a problem before publication?

@SC The problem is that there is no "positive" or "negative" control. There are positive and negative controls. What you use as a positive control or a negative control is based on the question you are asking. Even then there are additional controls to ensure everything is behaving correctly. IMO, these are the types of controls that ERV is referring to above. It is often easier to determine which controls were necessary after the fact, but scientists cannot go back in time using this new information to predict the appropriate controls. I am not defending the XMRV studies, simply pointing out that determining controls to use is not a black and white approach.

If you are trying to publish completely different or unexpected, you need to use more rigorous and numerous controls to better reinforce your conclusions. This is often where peer review breaks down (Im looking at you PNAS), and why Cell, Science, and Nature have many retractions. They go for the new awesome results that mean awesome new ways of looking at the universe..at least until it turns out that someone fucked up the protocol.

Sadly, many of the control studies ERV highlights should have been done in the original studies. However, scientists are human and if you get a result you really like, a scientist can (but should be trained not to) be less rigorous in order to diminish the chance that result turns out to be not true.


So far, outside of the Science study, Mikovits has two times tested a single pedigreed negative control together with positive samples in blinded fashion. Ruscetti did this one time. Mikovits found a false positive twice (one time in 1 out of 6 samples of the same control and one time 1/4) while Ruscetti also found a false postive in his testing of two negative control samples for serology. The one control sample was beforehand pedigreed as negative by both Mikovits and Ruscetti by PCR and culture (Mikovits) and serology (Ruscetti) themselves. Multiple times. And also by other labs (CDC and BSRI) by the way.

This perfect false positivity 'score' should worry patients, but it does not: they agree the well pedigreed negative could very well be a hard to find positive healthy person and this unprobable assumption keeps the faith running.

It really is another example of Occam's Razor. What is more probable? That the healthy control is in fact very weakly XMRV positive, that coincidentally this fact didn't turn up during testing the samples (say) 20 times by multiple assays beforehand, and that it only suddenly popped up as (correctly) positive when you started testing it in blinded fashion, using the (self proclaimed) weakest detecting method(s)? Or that Mikovits and Ruscetti are repeatedly reporting false positives on a healthy negative control?

RRM-- Yup. Check out the comments on Trines article.

SC-- Lorax pretty much got it :) I would lay more of the blame on the positive studies, though. Going back and reading them, even something as simple as explaining/showing their H2O controls are unnecessarily vague/confusing.


"They go for the new awesome results that mean awesome new ways of looking at the universe..at least until it turns out that someone fucked up the protocol.

Sadly, many of the control studies ERV highlights should have been done in the original studies."

ERV's list seems to be what are called heuristics in some disciplines. That gets translated as rule-of-thumb but they are more like a gauntlet composed of wet blankets.

In forensic antiquity authentication (for instance) they are used to counteract the "diatom dilemma" where the researcher (and everybody else) wants so badly for it to be real they 'forget' to look for regional markers when somebody shows up at the museum's back door holding a purported holy grail.

What is extremely disturbing to me about this is there have already been federal reimbursements for 1/5 of the cost of the WPI/VIP/XMRV test and a push for SSDI 100% blue book CFS disability on the grounds of an extant test (WPI/VIP/XMRV).

With that much political inertia and that much money on the line is it possible to turn back the clock if this is established to be contamination by a preponderance of evidence and an ever increasing demonstration of irreproducible results?

By Prometheus (not verified) on 03 Jan 2011 #permalink

The great thing about PCR is that it's so well-known and well-established that any lab can do it. The horrible thing about PCR is that it's so well-known and well-established that any lab can do it.

Probably the best write up of the recent Retrovirology papers I've seen. I think that the way they were over-hyped (especially the Wellfare trust one) has led to some down-playing their significance in response.

I've never really understood the lack of concern the WPI has shown about the possibility of contamination. Last year, I thought that they could have been doing testing we were not aware of... but nothing more solid has been published since.

I think it will be like the cold fusion people. It started out with some results that looked kind of funny, some labs with better equipment and better technique looked at it and found absolutely nothing.

The true believers are still at it, with their crude equipment that can't measure what they say they are measuring.

ERV, did you see this paper, which mysteriously came out 2 days later in a very obscure brand new journal?


First, their positive rate increases the more the sample is manipulated. This is alarming considering the Retrovirology series of papers.

Second, I question their use of ant-Env responses as their second most reliable form of detection. I am confused why the WPI does not look for antibodies to capsid in their patients. In the case of HIV infection the presence of anti-p24 antibodies comes well before those to Env. On a western blot p24 is often more intense. This can even be seen in Montagnier's initial identification of HIV paper when they took plasma from acutely infected patents, p24 is the only band detected. An anti-capsid response would be more indicative of ongoing replication rather than an Env response.

I have used numerous p24 (HIV) and p27 (SIV) antibodies and have never seen a cross reactive band on a blot. I cannot say the same about Env antibodies. For example the potent broadly neutralizing 4E10 antibody will also stain a biorad protein ladder. I believe 2F5 gives a similar result (could be wrong). Env proteins are quite large, contain odd protein sequences and are heavily modified, these could allow for ânon-specificâ binding of antibodies in patients. Capsid is much cleaner.

Last, a comment about the Towers Retrovirology paper. I think the deep sequencing data is the most damning. At the Cold Spring Harbor Retrovirus meeting this summer the Gag leader deletion kept coming up as proof that this virus was truly unique and came form an equally unique source. The Towers group detected it in common lab strain mouse genomic DNA. It does not help that some of the Envs identified also were incapable of infecting human cells or from other viruses :).

I canât want for this Summerâs Cold Spring Harbor Retrovirus Meeting!


The paper you are referring to, the addendum to the Science paper, is pretty old news. It was published ahead of print in July, I believe. I found the following to be pretty damning, not only in light of the Towers paper, but also in light of the Alter/Lo paper, after which Mikovits suddenly also found 'more diversity in sequences than originally reported'.

"The genetic variation between XMRV isolates currently identified is only 0.03%, despite the fact that the viral sequences were obtained from isolates from two vastly different diseases in patients from geographically distinct areas. This variation is smaller than the variation observed between HTLV-1 isolates. As in the case with HTLV, the lack of diversity implies that XMRV recently descended from a common ancestor."

Thanks Lorax, RPM, ERV for your kind explanations of the issues on controls. I know real laboratories are nothing like the cartoon version of flashing lights and "oh my gosh!" moments as the litmus paper turns blue.

I think I understand Lorax's comments about scientists not being able to go back in time to choose better controls with hindsight, but is this entirely correct? Can't an experiment be re-run with more/better controls? Obviously that takes time and money, both precious quantities.

The Occam's Razor comments are interesting. The related idea is also germane: extraordinary claims require extraordinary evidence. I wonder if the fact that the researchers were looking for something specific made them lose sight of how extraordinary their claims were? They got the result they wanted, so they believed it without a high (enough) level of skepticism. As ERV's earlier comments show, to those in the know, aspects of those claims made them implausible, suggesting that the high jump bar for the evidence level should have been upped much earlier, but perhaps they weren't as knowledgeable as our brainful host.

Of course, an exploratory experiment does not necessarily have full controls or enough data points for statistical signficance (as appropriate), but that's OK provided it's not going into the outside world.

Perhaps it's good for the general public to see the self-correcting mechanisms of science swing into action, but my (outsider's) view is that it would have been better overall for the retrovirus community to do its laundry in its own semi-private space rather than hang out its washing in public. But in these days of shock headlines, funding pressures and blogs, that's probably inevitable. I fear the babbling "huh, you scientists got that wrong, how do you know you're right now?" backlash from the ignorami.

As with cold fusion though, I think cases like these make fascinating science magazine articles (and teaching examples), bringing it home that even very good researchers get it wrong sometimes.

Drifting off-topic, but one of my most instructive experiences years ago was when we did some computer modelling (nothing to do with biology) and we were convinced the (potentially worrying) results were massively wrong. We checked and re-checked the model, and used independent alternative calculations as verification in a way that we would never have bothered with if the results had been in our expected range. The model was right, our pre-conceptions were wrong.

One of the main problems with the original papers is that they don't tell/show us their controls. The Mikovits paper esp. is just a short report, so the controls are talked about very little - they don't have space for pictures of clearly negative controls etc if they have them (although they did have space for a crappy electronmicrograph image - they need a better EM technician, frankly!). The Lo/Alter paper doesn't show controls well either (and shows lots of fuzzy bands that would worry me :S ). It's difficult to criticise when you can't tell what controls were used.

When I test for Epstein-Barr virus, I don't actually use DNA extracted from the virus - I use a sequence from the viral genome that doesn't amplify human DNA (BLAST checked and tested on real, EBV- and EBV+ human-derived DNA samples [ie extract DNA of whatever kind from a blood/cell sample, see what you get]) that is artificially synthesised, transformed into a plasmid, amplified and sequenced to be sure I really do only have a positive control for what I should be testing for.

Then, I screen everything when I run a qPCR (and something that looks uncontaminated on a gel can be contaminated at qPCR sensitivities) - RNA and DNA-free water, the water controls from each batch of DNA extraction, and DNA that I would expect to be EBV- (eg fetal kidney). And a dilution series of my positive control to give me an idea of viral load.

BUT - I have the benefit of having watched this XMRV fiasco unfold. It really hammers home that you should do science right, or not at all. It makes you hyper aware of how careful you need to be before you can say with confidence that a result is statistically significant.

"As with cold fusion though, I think cases like these make fascinating science magazine articles (and teaching examples), bringing it home that even very good researchers get it wrong sometimes."

Bingo....and thank you.


The XMRV CFS WPI drama is picture perfect example of Irving Langmuir's pathological science.

If my brain had been working when Mikovits started the weird attacks on scientists who wanted to critically examine her work, who couldn't reproduce it or who just wanted enough information about her protocols to try and reproduce it, I would have understood her behavior in terms of pulling ad hoc hypotheses.

While I think Fleischmann and Pons were rather unfairly branded, Mikovits is going to have a lot to answer for. Henry Miller seems to have had Judy's number five years ago:


By Prometheus (not verified) on 04 Jan 2011 #permalink

"I think I understand Lorax's comments about scientists not being able to go back in time to choose better controls with hindsight, but is this entirely correct? Can't an experiment be re-run with more/better controls? Obviously that takes time and money, both precious quantities."

Sorry to be vague above. I was speaking more esoterically than specifically. You are absolutely correct that experiments should be repeated and if you found something new or interesting you include the appropriate controls. A scientist (a good one anyway) should be thinking about the trivial possibilities to explain the really cool result and take steps to rule them out, thereby strengthening the argument that said result is cool.

However, there may be some new information that arises after you publish that suggests a new different explanation to your results. I do not think a scientist should be faulted with not have this new information. For example, to use one of ERV's examples, you may use a specific Taq polymerase in your work that gives a cool result. Importantly you find that your positive and appropriate negative controls work. However, 3 months later you can not repeat these results although the controls still work. It turns out someone ordered a different Taq polymerase. In your follow up studies you find that the original Taq was more sensitive and could allow some amplification of trace contaminants that the new Taq doesnt. (Actually that example sucks because the negative control should have caught the problem in the first place.)

Interesting. A very clear, well-written write up ERV. So, in your opinion, do you believe it would be possible to inoculate macaques or other primates with XMRV from one of these contaminated cell lines you mention and make them sick? If so, what would that illness look like?

Levi, people are doing those experiments right now in macaques. Problem is that you only get a short blip of XMRV viremia and then nothing...

I really don't believe patients have made up their minds. That's a generalisation. Most are waiting for further research. Also, I think it is wrong to suggest that the WPI has not taken contamination seriously. They have tested every stage of their study for it, used a non mouse lab, and always allowed others to retest their samples, such as the CDC. They are also having them retested with the IAP assay. Thought there are some issue with this.

RRM, Mikovits actually said they were finding P-MLVs back at last years Cold Spring Harbor meeting. Way before Lo et al. Also, fully sequenced clones from two patients display a 2% sequence variation, whereas gag and pol show considerable.

Charl, the controls for the Science paper were reported. Only not in Lombardi et al, but in the response from comments.

"The healthy control population,
which was similar in age and gender to the
patients, was composed of healthy people who
visited doctorsâ offices in the western United
States between 2006 and 2008. The greatmajority,
although not all, of the patients analyzed were
matched in geographic location with controls."

Published studies on Macaques show that the virus disappears from the blood and is then found in the organs.

Ah, Jel, I think you misunderstand what I mean - not who are the controls they used for the cases (that pretty well doesn't matter) but the technical(?) controls - when they extracted DNA from blood samples, did they also "extract" from a sample of (supposedly, because these things do get contaminated!) DNA-free water? If your extraction controls light up on a gel/register on qPCR, that could mean you mixed up tubes, or that you have a general contamination problem with the lab, or a reagent is contaminated. Another technical control would be eg. running LNCaP cells that have not had other cells co-cultured with them through all the test - FISH, anti-body testing, whatever - as the co-cultured cell lines; running all your PCR reagents (the water you use to make up a final reaction volume, any reference dye use use, the PCR mastermix, the Taq polymerase, whatever) with your primers, to be sure you aren't spotting a contaminant present at low levels in the reagents... Those are the controls I meant :)

The latest "news": not only patients, but now also the WPI themselves have fallen victim to the Galileo gambit:

"[Research groups who have been working on XMRV] understand that novel scientific discoveries, which threaten current dogma, will continue to be challenged until the evidence can no longer be denied. For instance, there are still those few who question the fact that HIV is the cause of AIDS. It took Nobel Prize winner, Dr. Barry Marshall, 17 years and three trials in which he infected and then cured himself of H-Pylori associated ulcers, before the medical world would accept the fact that the bacterium causes the disease. Today we are engaged in a new battle to prove that human gamma retroviral infections, such as XMRV, are underlying pathogens in neuro-immune diseases and untold cancers."


Sorry to tell you that I remain unconvinced. Mikovits has told so many things and, in hindsight, believers can mold quotes to whatever conclusion they want. Do you have a reputable source which reports (at the time of the meeting) that she did actually find PMLV's before the reports of Lo/Alter?

The "2%" figure is from an "online rebuttal document" that is being distributed among patients (which kinda proves my point don't you think, although of course I never meant to say that each and every patient in the whole wide world had made up their collective minds). Can you state a source for that assertion? I certainly don't hope that some wise guy on the forums was comparing an XMRV sequence with an MLV sequence from the PNAS authors.

ERV, thanks for continuing to apply your excellent brand of intellectual gamma rays to the XMRV mirage. Iâd like to add an additional, though not very erudite, application of Occamist reduction. Gender imbalance in diagnosis of CFS is widely noted (although poorly researched) in reports and is much repeated as fact â the female/male ratio of those afflicted with CFS is frequently stated to be 2:1. If XMRV (or any other virus) were the causation of CFS, it would be unique (?) as an infectious disease agent in having an exceptional preference for one gender over another, yet no pro XMRV causation hypothesis researchers have addressed this issue. In fact Mikovits and others have used the fact that the very diagnostic criteria which underlies the gender imbalance outcome, has not been used in comparative studies of their work, as a basis for criticising the adequacy of those comparative studies. So is XMRV twice as likely to cause disabling disease in women as it is in men ? In Occamist terms the uniqueness of the position is a complexity, which in the absence of any explanation for the uniqueness, militates against the answer being Yes. However we can not logically go from the answer Yes to directly dismissing XMRV as causative of CFS, because there remains the question of possible bias in the application of diagnostic criteria (for instance http://www.ncbi.nlm.nih.gov/pubmed/20189151 ). Yet if the criteria that Mikovits et al are insistent is essential to the identification of XMRV infected individuals is either a) so inadequate that within its correct application it produces a large false gender bias or b) is being so inappropriately applied by diagnosticians that it produces a large false gender bias, then how can such a diagnostic criteria support accurate identification of people with the âXMRV diseaseâ ? What we have is a classic logic loop, in Occamist terms an unsupportable complexity, which leads us to the inevitable deduction, XMRV has no identifiable relationship to CFS disease causation.

@ RRM âProblem is that patients have already made up their minds.â Give us a break, cognitively challenged ( http://www.springerlink.com/content/n57473152r062027/ ) if not yet cognitively dissonanced, we may be, but not all of us âpatientsâ are so enraptured with Mikovitsism that we canât exercise disbelief.

@ Prometheus âWhat is extremely disturbing to me about this is there have already been federal reimbursements for 1/5 of the cost of the WPI/VIP/XMRV testâ

Itâs worse than that. If you look at the WPI Form 990 submissions (guidestar http://www.guidestar.org/ ) for 2007 and 2008, youâll see that WPI received a $ million of Federal funding amounting to 45% of WPI income. Even at the level of building costs for the Centre of Molecular Medicine which the WPI makes much of as its research facility was only very partially paid for by the Whittemore family â total costs $77 million plus â Whittemores promised up to $5 million, only $1million of that had been called off according the most recent University of Nevada documents. The sole owner of the Lab licenced to sell private testing under the WPI protocol is Harvey Whittemore, WPI (Harveyâs wife and one of his senior employees are on the board!) says itâs ambition is to acquire the Lab, so maybe Harvey will donate it â still itâs hardly the most disinterested way of proceeding even if the intentions are good.

Well, if the results can be shown to be purely due to contamination, it is a short cut around the other problem, which is the definition of CFS used in various studies.

(Before anyone explodes, I believe that unexplained chronic fatigue exists as a problem and that calling that problem CFS is possibly reasonable. Some people clearly suffer from long term chronic and disturbing fatigue that has yet to be explained by detectable illness, sleep problems, malnutrition, etc, etc. These patients thus have a problem that can reasonably be referred to as CFS. Why they have it, whether it could be a variant of a known disease, or how to treat it remain open questions, but if someone claims and behaves as if they have disruptive chronic fatigue, well then, in my medical opinion, they do have a problem that is manifested by a reported sensation of chronic fatigue, and that is a health problem. To say otherwise would be silly. Whether or not use of terminology that may imply a common underlying pathophysiology is justified at this time is debatable - one could simply say that the patients suffer from self-reported severe, disruptive chronic fatigue, which explains the problem without resort to the term "syndrome". On the other hand CFS is brief and states the issue. I don't use "self-reported" as an insult here, but as clarification - there are also many cases of patients who attempt to deny obvious fatigue.)

Even if we assumed that all reported XMRV findings were correct, we'd have the problem that some labs reported a high incidence of XMRV in CFS patients, and others reported the opposite - but that the groups were in different countries and using slightly different definitions of CFS. This could have meant, if the findings had stood up, that XMRV can be associated with a CFS syndrome, but that there are also cases of XMRV-negative CFS.

Naturally, if the XMRV results are clearly shown to be the result of contamination, then the dilemma of explaining the conflicting results goes away.

The problems of explaining cases of puzzling chronic fatigue, and of whether or not a unifying terminology which implies a unifying pathophysiology, remain, but perhaps at least some confusing findings can be dispensed with.

Again, to anyone who experiences severe, unexplained chronic fatigue, by stating that I am not sure that all people who have this issue have the same underlying pathophysiology, I am NOT denying that you are experiencing a real, significant, and disruptive health problem.

Contamination cannot be side-stepped or ignored anymore.

Straw man. It never was.

XMRV is a contaminant, not a real human pathogen.

Brave words at this stage of the proceedings. Maybe even foolish.

I wonder if you will be so certain in a year?

WotWot-- It has been a year. A year + 3 months. And no one can support Mikovits findings (if they dont flatly refute them), including Mikovits.

Mikovits has had time, however, to speak side-by-side with Andrew Wakefield at an anti-vax conference, peddle XMRV 'testing' and XMRB 'drugs' and XMRV 'supplements'...

My position is not 'brave'. It is 'obvious in the face of a preponderance of evidence.'

From today's issue of Science:

Some had hoped that a project in which several U.S. labs are testing for XMRV in the same samples would clear up the picture. But so far this effort has been inconclusive. Four CFS patients' blood initially tested positive for XMRV at WPI and the U.S. Centers for Disease Control and Prevention but not at an NCI lab. When all three labs tested new samples from the same patients, none found XMRVâfor reasons that aren't yet clear, says Coffin. The group now plans to test blood from several dozen CFS patients and controls.


Andrew-- Ive already talked about that. And, to repeat what one of these papers said: There is not enough genetic diversity in the isolated sequences to assume these magic-eye viruses are real, replicating, human pathogens. The limited (read: non-existent) sequence variation points to contamination of reactions. Its not a real infectious agent, so 'how it started infecting humans' is a nonsense question.


I've referred to that investigationg from the Blood Working Group a couple of times. The telling circumstance that isn't mentioned in that snippet you quote, is that for the first time of testing WPI collected the blood, while for the second time of testing an independent phlebotomist collected the blood (and also, the second time samples were sent to the labs blinded).

I think it's reasonable to assume that XMRV contamination not only can occur in different ways, but also probably did (and does) occur in the WPI lab in different ways. It would make the most sense of the nonsensical results thus far.

It is 'obvious in the face of a preponderance of evidence.'


We shall see...

Your prediction (as I understand it) is that XMRV will turn out not to be pathogenic, is just a contamination issue, and careers will end because of it.

Mine is that you (and a few other naysayers) will be eating some serious humble pie over XMRV. Though I agree that careers are likely to be ended, whatever the technical result.

How about we meet here again in Jan 2012, and see how things are working out? Can't do fairer than that.

"The telling circumstance that isn't mentioned in that snippet you quote, is that for the first time of testing WPI collected the blood, while for the second time of testing an independent phlebotomist collected the blood (and also, the second time samples were sent to the labs blinded)."

I don't remember seeing that reported either. The Science paper, the Alter paper, and the new WPI UK study all collected CFS samples in a different way to control samples prior to blinding. It would be a bit of a coincidence if the CFS samples were always the contaminated ones (certainly the UK samples were collected by an independent phlebotomist) but it's certainly quite possible. What a disappointing end to this that would be.

One other political point: McClure's now come out claiming her positive prostate samples were contaminated. To me, this makes her opposition to XMRV in CFS more credible. When she was so confident that XMRV was not related to CFS, but was related to PC, I didn't take her entirely seriously. What a funny ending it would be to discover that XMRV was just a mirage all along.

WotWot -

Can't do fairer than that.

Actually, you could.

You could address the specific points that are made here -

1) Lack of consistent serum positive status for any group of patients with CSF by any definition, for XMRV.

2) Where positive results are obtained by PCR, they are not consistently replicable, and show characteristics that are suggestive of contaminant rather than pathogen (level of sequence variability).

3) Although not independently relevant, unprofessional, self-serving behavior by Mikovits which, in this context, adds to the impression of potential over-statement.

I have not "made up my mind" 100% on this issue, partly because I don't have specific detailed knowledge of all the original literature and rely mainly on reviews, but I would have to say that, at this point, there very are serious problems with the case for XMRV as a causative agent for CSF, to the extent that the logical default position is that it was a false report, until strongly documented otherwise. Those issues need to be addressed by anyone who makes the positive claim.

Meaningless, non-specific bombast such as yours serves no useful purpose.

(I will also note a few things on my own, for completeness - the difficulty of coming up with a rigorous, universally accepted definition of CSF and excluding all pertinent negatives complicates things, and I haven't seen any mention of studies of immune response, of quantification of ostensible viral load and its relationship with symptoms, or of any other supporting evidence beyond claims beyond PCR results as being either "positive" or "negative" - I could be wrong here and stand eager to be corrected if I am but that's my current understanding. For full completeness, putting this research aside, endogenous retrovirus mRNA and proteins can be expressed and indeed seem to have a major role in the evolution and development of the placenta, so an undetected role of erv expression in human health is not certainly possible.)

Jan 2010 prediction:

Furthermore, MLV is notorious for contaminating cell lines. ...We already know that 1:25 PCR results are going to be false positive, yet more patient samples were culture positive than PCR positive, which can only make sense if your friggen cell line is contaminated with Plain Jane MLV.

Jan 2011 status:

While I was gone, a series of papers were published in Retrovirology that gave experimental support for the logical conclusion that XMRV is a contaminant:

Contamination of clinical specimens with MLV-encoding nucleic acids: implications for XMRV and other candidate human retroviruses (review)
Disease-associated XMRV sequences are consistent with laboratory contamination (blagged here)
An endogenous murine leukemia viral genome contaminant in a commercial RT-PCR Kit is amplified using standard primers for XMRV
Contamination of human DNA samples with mouse DNA can lead to false detection of XMRV-like sequences
Mouse DNA contamination in human tissue tested for XMRV

You're already behind the 8-ball, WotWot. Though I wouldn't predict loss of careers.

Perhaps interesting:

Retrovirologist Dusty Miller is trying to duplicate the WPI findings using known XMRV positives (according to WPI/VipDx) and the IAP assay to check possible positive results for mouse contamination. He has stated that he has been warned about using the IAP assay by Mikovits and Ruscetti. They have apparently told Miller that false positives are possible, as they believe IAP sequences might be transferred by retroviruses like XMRV.

It looks like circular reasoning to me: Mikovits checks for contamination in her XMRV+ samples, finds that her samples have been contaminated according to the new sensitive IAP assay, but because her original results are true beyond question, it must be that finding IAP sequences in humans samples is actually proof that these people are infected with XMRV.

It may turn out that this IAP assay will become the most reliable XMRV test... ;-)

@In Vitro Infidelium (#23): I don't think the gender differential in CFS is inconsistent with a link to an infectious agent. Most autoimmune diseases show a greater female to male ratio, ALS being a notable exception. Nevertheless, identical twin studies in autoimmune diseases show only ~25-30% concordance. That highlights there is both a genetic and environmental component. The bulk of HIV infections in the beginning were in men and hemophiliacs. We now can easily explain that data. Given we don't know if and what the infectious agent is in CFS or it's mode of transmission, latency, ect., I would not rule out an infectious cause for CFS giving rise to the gender discrepancy.

With that said, I would rule out XMRV as playing a causative role in CFS (or likely anything else) at this juncture.

@ cynical1
Agreed â âlink to infectious agentâ should not exclude the possibility of a gender differential as a real characteristic of disease. But âlinkâ is not what is being claimed in the case of XMRV â there the argument is of direct infection as a cause of disease. A useful comparison is Rheumatic fever where an infectious agent (Streptococcus) prompts an autoimmune reaction; as it happens there is no recorded gender bias in succeptability to Rheumatic fever but that doesnât obviate autoimmune reactions to other infections being gender preferenced. Treatment of Rheumatic fever as a chronic condition may commence with anti streptococcus medication, but no-one would consider anti bacterials as a cure for Rheumatic fever which is a complex condition with multiple potential disabling outcomes needing multiple interventions. Unlike Rheumatic fever, CFS is not a discrete condition and multiple causes are likely amongst the diagnosed population, and some of those causes may well include autoimmune disease. One of the many dubious aspects of the XMRV/CFS fiasco is the movable line of CFS diagnosis where the pro WPI/Mikovits lobby argue for a progressive reduction in who qualifies for âproper diagnosis and therefore a potential to yield a positive XMRV test, so as to preserve the magical 70% XMRV positive figure. CFS is an inclusive categorisation based entirely on the exclusion of known disease, that 7 out of 10 people with CFS diagnosis would have a single causational infection is highly improbable, simply because there is no underlying commonality of categorised disease. At this stage it remains most likely that CFS is a categorisation of multiple chronic diseases which happen to share a broadly common symptomology. Diagnostician bias is almost inevitable in this circumstance and the over representation of women and the under representation of manual workers in the (largely unresearched) patient profiles is an issue that at least needs to be acknowledged.

Actually, you said "If XMRV (or any other virus) were the causation of CFS, it would be unique (?) as an infectious disease agent in having an exceptional preference for one gender over another, yet no pro XMRV causation hypothesis researchers have addressed this issue." Look up the stats on HIV in the US. There are many more men with HIV (and AIDS) than women. Since no one claimed anything about the modes of transmission of XMRV, in Occamist terms, the gender differential is not unique based on our knowledge of retroviruses.

Thanks all for the intriguing debate over the WPI's paper on XMRV and the assessments you give. I have ME/CFS/Fibro, (whatever you want to name it), and try to follow the latest advances in diagnosis and treatment. I appreciate the criticism of those who are experts in retrovirology and lab diagnosis. I have certainly considered the lab test currently available for patients to detect XMRV, but will await further information. I will be closely watching the science unfold. I am curious as to what you think about Mikovits' latest response below?
She writes:
"Most significantly, the recent Retrovirology publications failed to address the most important pieces of scientific evidence of human infection in the previous XMRV studies, including the fact that XMRV positive patients produce human antibodies to gamma retroviruses, XMRV integrates into human tissues, and infectious virus has been cultured from the blood of hundreds of patients with a diagnosis of Chronic Fatigue Syndrome and M.E. Humans do not make antibody responses to mouse DNA sequences from contaminated lab experiments. The Retrovirology studies only point out that XMRV research cannot be done in a mouse laboratory without extreme caution and should not rely solely on PCR methods."

Although not stated specifically, she also has claimed that three independent labs have supported her findings.
I'd be grateful for your comments on her statement.

Also, the NIH and the Dept. of Veterans Affairs have supported a grant that resulted in research last year from the Department of Pathology, University of Utah, Center for AIDS Research, Laboratory of Biochemical Pharmacology, Department of Pediatrics, Emory University School of Medicine and Veterans Affairs Medical Center, showing a retroviral that is effective in use against XMRV. I have also read another research paper that thoroughly discusses which combos of antivirals are the most effective against XMRV and states recommended combinations that are successful. What are your thoughts?

By CharlieJane (not verified) on 16 Jan 2011 #permalink

CharlieJane: ERV has talked about the antibody 'problem' before. Firstly, your immune system can mount an antibody response to things it has never seen due to the vast, vast diversity of antibodies your body produces (ERV or another immunologist who knows the numbers, if you could step in with them/the link, that would be great). And you can have antibodies to things that aren't pathogens - which is why you have allergic reactions, of why people get autoimmune diseases.

Secondly, those XMRV-specific antibodies? They aren't XMRV-specific. I can't find the right paper *head-desk*, but they found that XMRV antibodies (which are polyclonal) react to other mouse viruses, and other things besides. Antibody testing is a really rubbish way of picking up the presence or absence of XMRV with any confidence.

Charl: Thank you for your response. Have you read the transcript from the FDA December 14, 2010? "Blood Products Advisory Committee Meeting Transcript
What are your thoughts?

By CharlieJane (not verified) on 20 Jan 2011 #permalink

My son and I are HHV-6A, EBV, CMV and XMRV Positive. 6-10% of the folks that have posted on this board will test positive for XMRV. 6-10% of the folks reading this will test positive for XMRV. Around 70-87% of True CFS cases are linked with the viruses I mentioned above. Plus, a high percentage of these patients have opportunistic infections as well as viral reactivation. We are dealing with a stealth virus that infects the Brain and Spinal Cord. A potent retrovirus that activates with Androgen and Cortisol receptors. Opportunistic infections. It is easy to connect the dots from here. Also, now 2 starins of XMRV have been found. Stanford and Columbia Universities have teamed up to study the "Lyme Link" with XMRV. In one study, 100% of lyme patients tested positive for XMRV. You can believe anything you want. However, when the 6-10% of you fall ill; I bet you get tested.

By Julia Rachel (not verified) on 19 Feb 2011 #permalink

"I really cant emphasize enough how outdated viral culture is in 2010"

Does that mean that Koch's Postulate is broken and we should find new adapted "rules"?

By Tony Mach (not verified) on 26 Dec 2011 #permalink