XMRV CFS UK study #II

[Gerwyn]

Looking at the information Suzy has supplied here, together with the supporting online material for the Lombardi et al Science paper (I found the pdf!), I would say on face value that it would appear all patients fulfilled both, especially as no discussion of stratification between both criteria sets is given in the paper.

However, I believe it does need clarifying by the WPI. I guess it needs to be established whether ALL patients in the sample met the Canadian criteria (by definition those sort of patients SHOULD meet Fukuda symptoms criteria anyway).

I also wanted to say that I also think there are various fundamental problems with the testing for XMRV methodology. Technical mircrobiological testing methodology is not my area of knowledge, hence my not pressing that issue. But I understand it's vital importance.

Nevertheless, the issue of patient cohort is also crucial, not just in regard to XMRV, but for future 'CFS' research. Tom Kindlon kindly put up a paper on co-cure today showing that the Canadian criteria is starting to be engaged with in research projects. Leonard Jason led a good validation project of Canadian criteria in 2005. I have been involved in endeavouring to ensure that PACE trial authors do not get to ignore the Canadian guidelines (without being called on frequently and publicly it at least) since the trial was in planning stages and since. So the issue of patient cohorts is a hot one.

I wonder - might it be useful to consider starting a new thread on establishing Canadian Cohorts in 'CFS' research?

yes please

 

[natasa778]

Thanks a lot Gerwyn. Btw have you seen this one http://www.jstor.org/pss/30112876

Not sure how many of you guys suffer chronic GI problems? which of course could be caused by other things, but this would be a relatively easy study to carry out for xmrv... It would certainly fit what we seen in autism to a "T" and would fit the theory of xmrv residing in epithelial (gut lining) and lymphatic tissue

any thoughts?

 

[Dx Revision Watch]

UK XMRV University College London PhD Project: A role for XMRV in human disease

In December, I submitted a request under FOI for information around this PhD Project to University College London. I received a response on 16 December.

Since the Project is apparently being funded by the MRC, I will need to address my information request direct to the MRC as little information was held by UCL.

UCL's response, first, followed by my request for information, followed by Project details.


16 December 2009

Further to your request dated 7 December 2009 regarding a PhD project, "A role for XMRV in human disease". Please find below our response:

1. UCL does not hold this information.

2. "A role for XMRV in human disease." There is no separate scientific title.

4. UCL does not hold this information.

5. As described on findaPhD website.

6-11 UCL does not hold this information.

12. March 2011.

13. September 2014.

14. MRC UK.

15. MRC UK.

----------------

7 December 2009

Subject: Request for information under FOIA: PhD Project: A role for XMRV in human disease

To: Records Manager, FOI, and Data Protection Officer, The Records Office, University College London, Gower Street, London WC1E 6BT
foi@ucl.ac.uk

Re: Request for information under FOIA in respect of PhD Project:

http://www.findaphd.com/search/showproject.asp?projectid=18971

Division of Infection & Immunity, University College London

Project Supervisors: Prof G Towers; Dr P Kellam

Project title: A role for XMRV in human disease

Laboratory supervisor: Prof Greg Towers

Clinical supervisor: Prof Deenan Pillay

[...]

I request the following information under the Act:


1] Any Identification or Reference code assigned to Project

2] Project's Public Title;
Project's Scientific Title

3] Study hypothesis/rationale (where applicable)

4] Ethics approval and any reference numbers attached to this approval

5] Study design

6] Countries of recruitment; Centres of recruitment
Through what means will prospective participants be recruited?


The Project Application notice states:

"Xenotropic murine retrovirus (XMRV) has recently been associated with chronic fatigue syndrome as well as prostate carcinoma in humans (1-3)."

"2. We will use these assays to measure XMRV load in chronic fatigue patient samples as well as, well but XMRV infected control samples, with a view to establishing whether viral load relates to disease, episodes of illness and/or severity."


7] For what Diseases/conditions/study domains are patient samples to be collected?

Control group is given as "well but XMRV infected control samples".

Through what means will control samples be assembled?


8] Participants - inclusion criteria

and please also state criteria by which patients are to be defined, eg CDC Fukuda 1994; Canadian Consensus Criteria 2003.

Will patients undergo functional assessments?
Who will be responsible for carrying out functional assessments?
How are patients to be assessed?

9] Participants - exclusion criteria

10] Target number of participants

11] Patient information material: please provide copies of any patient information material

12] Anticipated start date

13] Date by which Project is anticipated to complete

14] Sources of funding

15] Sponsor details

--------------------

PhD Project: A role for XMRV in human disease

A role for XMRV in human disease

Laboratory supervisor: Prof Greg Towers

Clinical supervisor: Prof Deenan Pillay

Xenotropic murine retrovirus (XMRV) has recently been associated with chronic fatigue syndrome as well as prostate carcinoma in humans (1-3). XMRV is a murine endogenous virus found in the genome of mice and until recently has been thought to be absent from the human population. It is now becoming clear that XMRV has transmitted to humans by a process of zoonosis, presumably from mice, and appears to be associated with a variety of diseases not previously associated with viral infection.

1. We will establish quantitative PCR assays and serology assays including enzyme linked immunosorbant assays (ELISA) to detect and quantify XMRV. Importantly, assays used to detect related murine leukaemia viruses in the lab are expected to be suitable.

2. We will use these assays to measure XMRV load in chronic fatigue patient samples as well as, well but XMRV infected control samples, with a view to establishing whether viral load relates to disease, episodes of illness and/or severity.

3. The receptor for XMRV has been identified. We will seek human polymorphism in the xenotropic receptor and assess which human cells express it. We will also establish which cells in vivo in blood express the receptor and which cells are infected with XMRV by quantitative PCR on sorted subsets of B and T cells from XMRV infected individuals.

This project proposes to address some of the most important questions surrounding the recently described XMRV infection of humans and to seek a therapeutic strategy for XMRV treatment. We expect it to be a competitive project and the experiments performed are likely to be influenced by ongoing studies published as we go. We expect that the candidate will be fully trained in modern techniques of molecular virology during the course of this project.


TO APPLY Send THREE COPIES of your CV (including full contact details of two academic referees) a personal statement and an indication of your top two preferences, on a separate page, from the list of projects below to:
Isabel Lubeiro, Division of Infection & Immunity, Windeyer Building, 46 Cleveland Street, London W1T 4JF.
CLOSING DATE: 23 NOVEMBER 2009​

 

[Angela Kennedy]

yes please

Ok - should I start in 'Action Alerts! and Advocacy'?

I'll dig out evidence I have on the issue and post up links if that's ok?

 

[Dx Revision Watch]

@ Cort, again,

If you were referring to the Kerr, Bishop, Gow et al study, I think it is important to note generally, and for those who support the CFS Research Foundation (CFSRF) with donations, that the CFSRF were co-funders of this study.

In December, the CFSRF published a statement in its newsletter on the WPI XMRV study findings. I recorded this on my own site, in this posting, below. Note that Anne Faulkner refers to another Kerr, Bishop study (highlighted in blue) which the CFSRF will be funding (or part funding).


http://meagenda.wordpress.com/2009/12/12/cfs-research-foundation-cfsme-xmrv-is-there-a-connection/

CFS Research Foundation: CFS/ME XMRV Is there a connection?

December 12, 2009

Prefaced with Notes by ME agenda:

The Chief Executive of the MRC, Sir Lezek Borysiewicz, is to step down in 2010 ( Source: ME Association News page ).

Dr Tim Harrison PhD, DSc, FRCPath. is a Trustee of the CFS Research Foundation and member of the CFSRF Research Committee.

Professor Stephen T. Holgate FMedSci, MRC Clinical Professor of Immunopharmacology, University of Southampton is a member of the CFS Research Foundations Research Committee.

Professor Holgate chairs the MRCs CFS/ME Expert Panel.

Dr Jonathon Kerr is a member of the MRCs CFS/ME Expert Panel.

Dr Paul Kellam BSc PhD, Department of Infection, University College London is also a member of the CFS Research Foundations Research Committee and one of the project supervisors for the UCL PhD Project: Project title:

A role for XMRV in human disease Division of Infection & Immunity, University College London: Project Supervisors: Prof G Towers; Dr P Kellam

http://www.findaphd.com/search/showproject.asp?projectid=18971

http://www.cfsrf.com/index.html

CFSRF Newsletter

CFS/ME XMRV Is there a connection?

It is likely that you will have heard or read about the interesting work being carried out by Dr Judy Mikovits and her team at the Whitmore Peterson Institute in Reno, Nevada to see if the retrovirus XMRV (exenotrophic murine leukaemia virus-related virus) might be associated with CFS/ME.

This research has been given tremendous coverage by the media throughout the world and while anyone suffering from CFS/ME must feel a degree of excitement we must caution restraint. A good deal more work needs to be done before too many claims can be made as to the relevance of this virus in CFS/ME.

Recently the retrovirus XMRV was found in the tumour tissue of a subset of prostate cancer patients. Both XMRV positive cancer and CFS/ME have been linked to alterations in a certain antiviral enzyme. The team in Nevada decided to carry out a study to see if this retrovirus might be associated with CFS/ME.

When the team analysed blood taken from 101 CFS/ME patients 68 (67%) tested positive to XMRV genes compared with only 8 (3.7%)out of 218 healthy controls. They stated that their results are consistent with the hypothesis that CFS/ME patients mount a specific immune response to XMRV. The have discovered a highly significant association between XMRV and CFS/ME.

The research associating XMRV with CFS/ME leaves many questions to be answered. First, it will be necessary for the study to be repeated. Over the years there have been claims for other retroviruses in other illnesses which have come to nought so it is essential that this research is found to have been concluded correctly and for the conclusions reached to be confirmed in independent studies around the world.

We have to ask the question is XMRV a cause or factor in the pathogenesis of CFS/ME oe a passenger virus in the immunosuppressed CFS patient population. Several other viruses have been linked to CFS/ME, for instance the Epstein-Barr virus, enteroviruses or herpes viruses, so we must ask what is their relationship to XMRV and the presence or absence off theses viruses.

Another question must be to ask if the virus XMRV causes CFS/ME or is it just more common in people with the illness.

In the USA the National Institutes of Health (NIH) have taken this research very seriously. They have called meetings of different departments to discuss the implications of these findings, and they and various groups throughout the world are currently setting out to determine whether this association can be confined for CFS/ME patients in Europe and other countries. They have also made a grant of $2 million to take the research further.

Dr Jonathan Kerr and Dr Judy Mikovits have been awarded $2 million from the NIH to study the disease mechanisms in CFS/ME. $1 million has been awarded to the research team in Nevada, the other $1 million has been awarded to Dr Jonathan Kerr at St Georges University of London, the scientist well known to all the supporters of the CFS Research Foundation who carried out the research which discovered 88 genes which were abnormal in CFS/ME patients but remained normal in healthy people. Dr Kerr will study CFS/ME patients to identify important genes which are turned on and off, proteins in the immune system (cytokines) and mutations in the DNA. Some of these American patients have developed Mantle Cell Lymphoma (MCL) after many years of having CFS/ME; these patients will also be included.

The CFS Research Foundation tackles some of the questions.

In spite of the large grant which Dr Jonathon Kerr has received from the NIH the Research Committee has decided that it is imperative that we know if UK and USA patients are infected with XMRV. So the Foundation is to fund a study to establish whether there is a relationship between XMRV and CFS/ME by testing samples from the UK and the USA. Dr Jonathan Kerr and Dr Kate Bishop, who is working at the national Institute for Medical Research in London, are planning to examine patients with CFS/ME and match comparison groups. They will test for the virus itself as well as for the immune responses to this virus. It is of course, vitally important to confirm or refute the finding recently published in the USA.

The Gene Work Continues

While this work is causing such excitement the work of gene expression continues. Of the 88 genes which are abnormal in the CFS/ME group but normal in the control group, Dr Kerr found that these genes could be divided into 7 subtypes. What was so interesting was that theses subtypes were associated with distinct differences in their clinical patterns and severity. Each of these subtypes had a different list of genes which were abnormal.

In a further study Dr Kerr tackled a problem which always causes great concern to CFS/ME sufferers and their families and friends. For years there has been dissension among doctors and scientists as to whether CFS/ME patients were suffering from endogenous depression. Many sufferers felt that this was holding up scientific research. Dr Kerr tested the genes of people with endogenous depression and compared them with the genes of 29 healthy blood donors. Gene levels in the endogenous depressed patients were similar to those in normal controls, but, importantly they are different from the CFS/ME patients.

Dr Kerr and his team are currently extending the previous findings by including a larger number of well-defined patients. These investigations are being conducted on a blinded basis in order to ensure that there has not been any potential bias on the technical aspects of the study. The samples have recently been collected by Dr Tim Harrison, a Reader in Molecular Virology at University College London Medical School, who visited St Georges Hospital to prepare the blinding. The samples were placed in tubes, each one coded, and then frozen. Dr Harrison will keep the code, and no one else will know it until the time set for unblinding.

You will see that this team will have made sure that their findings are accurate. This contrasts with some previous attempts carried out by other groups on a purely empirical treatment methods that have no firm scientific basis. The research being conducted at the present time by Dr Kerr and his team may well result in not only a reliable diagnostic test but also the initial steps for appropriate therapy based on firm scientific data.

The Future

The outlook for CFS/ME research has never been brighter. Increasingly, doctors and scientists are believing that this is an organic disease which need organic research. The paper from Nevada suggesting that the retrovirus XMRV might be associated with CFS/ME has caused great interest and scientists throughout the world have been attempting to repeat this study. Whether or not it is confirmed we already know that virus infection is important in CFS/ME.

We have some encouraging news from the Medical Research Council (MRC). For years people with CFS/ME, their relations, friends and some research scientists have been frustrated by the MRCs concentration on psychiatrists when conducting research into this illness. This has now changed. The Chief Executive of the MRC, Sir Lezek Borysiewicz is anxious that CFS/ME research should go ahead in a wide field. This must be the best possible news.

The Foundation is seeking new studies of a high standard. We shall have to re-double our efforts to produce these studies and we hope we can receive some part funding from the MRC. We see the possibility of our research expanding and producing even more radical results as it has in the past. The speed at which we can go forward is up to all of us. We can now look to the future with even greater hope.

Anne Faulkner, Honorary Director​

 

[oerganix]

"The Gene Work Continues

While this work is causing such excitement the work of gene expression continues. Of the 88 genes which are abnormal in the CFS/ME group but normal in the control group, associated with distinct differences in their clinical patterns and severityDr Kerr found that these genes could be divided into 7 subtypes. What was so interesting was that theses subtypes were . Each of these subtypes had a different list of genes which were abnormal."

Is there a link where one might find what Dr Kerr's 7 subtypes are?

thanks

 

[Esther12]

"The Gene Work Continues

While this work is causing such excitement the work of gene expression continues. Of the 88 genes which are abnormal in the CFS/ME group but normal in the control group, associated with distinct differences in their clinical patterns and severityDr Kerr found that these genes could be divided into 7 subtypes. What was so interesting was that theses subtypes were . Each of these subtypes had a different list of genes which were abnormal."

Is there a link where one might find what Dr Kerr's 7 subtypes are?

thanks

A lot of the details can be found here (plus links to the papers):

http://forums.aboutmecfs.org/showth...ene-Expression-study-published&highlight=kerr

From the little I understood, it sounded less compelling than I hoped. It's useful for showing a clear distinction from depression - but it seems like most of the psych lobby had realised that anyway.

 

[Advocate]

I wonder - might it be useful to consider starting a new thread on establishing Canadian Cohorts in 'CFS' research?

Hi Angela,

That sounds like a great idea. If you do start a new thread, please give a link to it on this old thread so that we don't miss the new one. Old threads that I've subscribed to always come into my e-mail box, but unless I check the website every day I sometimes miss good new ones.

 

[Angela Kennedy]

Hi Angela,

That sounds like a great idea. If you do start a new thread, please give a link to it on this old thread so that we don't miss the new one. Old threads that I've subscribed to always come into my e-mail box, but unless I check the website every day I sometimes miss good new ones.

Thanks Advocate, I will do that. I'll start the new thread anon!

 

[Gerwyn]

Thanks Advocate, I will do that. I'll start the new thread anon!

Kerr knows about the issues re non virologists drawing blood! and if that cohort is correct it would form the basis of a definitive XMRV study if they choose

 

[Gerwyn]

Thanks a lot Gerwyn. Btw have you seen this one http://www.jstor.org/pss/30112876

Not sure how many of you guys suffer chronic GI problems? which of course could be caused by other things, but this would be a relatively easy study to carry out for xmrv... It would certainly fit what we seen in autism to a "T" and would fit the theory of xmrv residing in epithelial (gut lining) and lymphatic tissue

any thoughts?

I know that endogenous retroviral expression is associated with ulcerative colitis now IBS type problems are common in ME---I wonder?? we know that endogenous gamma like sequences are involved HIV is causative not a bad place to look---A very interesting idea!

 

[_Kim_]

Hi Angela,

That sounds like a great idea. If you do start a new thread, please give a link to it on this old thread so that we don't miss the new one. Old threads that I've subscribed to always come into my e-mail box, but unless I check the website every day I sometimes miss good new ones.

Angela started it here:

Establishing research cohorts of 'Canadian' criteria patients: an urgent need

 

[bullybeef]

I know that endogenous retroviral expression is associated with ulcerative colitis now IBS type problems are common in ME---I wonder?? we know that endogenous gamma like sequences are involved HIV is causative not a bad place to look---A very interesting idea!

That is fascinating Gerwyn. I was diagnosed with colitis prior to ME!! The consultant later said the type of 'colitis' I had was common!! Another consultant then suggested I see a shrink. Mmmm..I'm sure I have heard that suggestion elsewhere on here!

 

[kurt]

Control Testing?

"Also, if WPI had to take up to four blood samples to find XMRV in their CFS group, did they also look that hard for XMRV in the controls? Did they also take four blood samples from each control?"

Apologies if I've missed any responses to this, but this is an incredibly important question. Does anyone know if WPI did the same multiple testing on the controls ?
Sorry if there's already been a response, I'm unable to spend much time on the Forum at present.

Yes, that is an incredibly important question, I posted that originally and believe nobody has responded. There is no evidence that controls were tested that extensively in the Science article, and I have not found any statement by Mikovitz since that suggests they were (maybe I missed it?). But this might just be an omission in the article, they might have tested controls that thoroughly and just not mentioned it since they also did not talk about multiple samples with CFS patients... But still, Mikovitz has now several times made a point of saying that multiple tests must be run on multiple samples to find XMRV with PCR testing, in order to communicate that this is a very difficult bug to find. That is one way of dismissing the failed validation studies in the UK, and the commercial tests for XMRV that are/were available. But maybe she is revealing something unintentionally here. There is no question WPI was searching for and trying to build a case for XMRV in CFS, and worked hard to find it in the blood of CFS patients. Did they actually collect four samples at different times from each control patient, and then test each of those samples four times? Somehow that seems like something that would have been reported.

What would this mean? Well, if they took four samples and tested from each CFS sample four times, but just tested controls once, that would mean the CFS patients had 16x the testing 'resolution' that the controls did. So effectively, you have to multiply the control positives by 16 to compare them to the CFS positives statistically. That means that rather than 3.7% positives in the controls there were the statistical equivalent of approximately 3.7 X 16 = 60% positives, which is so close to the 67% positive rate in the CFS samples that there would be functionally no difference between the groups and XMRV is actually ubiquitous. That would raise some questions about the prostate cancer studies of XMRV.

We really need to see some blinded testing so things like this can not happen. Or maybe WPI can clear up how they worked the statistical comparison if there was any differences in testing processes between the groups. No matter how much you believe your own theory, you just can not assume that one group needs more testing than the other group (controls), if that happened, the entire PCR part of the Science paper's conclusions would be invalid.

WPI needs to address this question, after all, they are the ones who have elaborated how difficult they worked to find XMRV in the CFS samples. Did they work as hard with the control samples? My guess is that they did, but they still need to make this clear, they have added new information about the CFS samples and that raises this question.

 

[Hope123]

What would this mean? Well, if they took four samples and tested from each CFS sample four times, but just tested controls once, that would mean the CFS patients had 16x the testing 'resolution' that the controls did. So effectively, you have to multiply the control positives by 16 to compare them to the CFS positives statistically. That means that rather than 3.7% positives in the controls there were the statistical equivalent of approximately 3.7 X 16 = 60% positives, which is so close to the 67% positive rate in the CFS samples that there would be functionally no difference between the groups and XMRV is actually ubiquitous. That would raise some questions about the prostate cancer studies of XMRV.

We really need to see some blinded testing so things like this can not happen. Or maybe WPI can clear up how they worked the statistical comparison if there was any differences in testing processes between the groups. No matter how much you believe your own theory, you just can not assume that one group needs more testing than the other group (controls), if that happened, the entire PCR part of the Science paper's conclusions would be invalid.

WPI needs to address this question, after all, they are the ones who have elaborated how difficult they worked to find XMRV in the CFS samples. Did they work as hard with the control samples? My guess is that they did, but they still need to make this clear, they have added new information about the CFS samples and that raises this question.

Kurt, my reading of it was that the 67% is the number of positives subject-wise (out of 101 CFS subjects) rather than by the blood tubes drawn. There are some statistical adjustments to be made when mutliple tests are run as statistically, the more tests run, the more spuriously positive results you get but I don't think they needed to do this since they based it on subject rather than tubes of blood. I do agree though that they needed to be as rigourous with the controls as they do with the cases. The stuff out from the CROI conference about XMRV being hard to find in blood reinforces the need to be uniform in their techniques.

 

[kurt]

Kurt, my reading of it was that the 67% is the number of positives subject-wise (out of 101 CFS subjects) rather than by the blood tubes drawn. There are some statistical adjustments to be made when mutliple tests are run as statistically, the more tests run, the more spuriously positive results you get but I don't think they needed to do this since they based it on subject rather than tubes of blood. I do agree though that they needed to be as rigourous with the controls as they do with the cases. The stuff out from the CROI conference about XMRV being hard to find in blood reinforces the need to be uniform in their techniques.

Yes, if the control testing was not as rigorous, they would need to make the statistical adjustment I worked out above. That would require a change to their results and invalidate their conclusions. However, if controls were run blind in the same exact batches as CFS cases, and multiple draws from each control were used, then you are correct and no adjustments would be needed as the basis for measure would be by the person and not by the tubes run. The only reason I brought this up is that controls are often from a convenient sample, which usually would include only one draw, and in this case there were many more controls than CFS cases, suggesting they were a convenient sample and may have indeed been handled differently.

Another way to adjust for this (if controls were handled single-pass) would be to take only the first measure from each CFS case, which would reduce the positives by 16X, so the real positive rate for CFS samples would be 67%/16 = 4.2%. Not much higher than the controls.

As for making an adjustment without equal numbers of tubes from controls, you would have to know the rate of false negatives in the controls and what causes false results. To find that out they would have to take a group of controls and test them also 16 times just like the CFS cases. Without knowing the false result risk there is no way to make a fair adjustment and it is imperative that controls be run blind alongside the CFS samples. IF that was not done, they have to make an adjustment like I have described. Would be nice to know if that was done.

 

[Mark]

Well, if they took four samples and tested from each CFS sample four times, but just tested controls once, that would mean the CFS patients had 16x the testing 'resolution' that the controls did. So effectively, you have to multiply the control positives by 16 to compare them to the CFS positives statistically. That means that rather than 3.7% positives in the controls there were the statistical equivalent of approximately 3.7 X 16 = 60% positives, which is so close to the 67% positive rate in the CFS samples that there would be functionally no difference between the groups and XMRV is actually ubiquitous. That would raise some questions about the prostate cancer studies of XMRV.

Indeed, if the testing applied to the CFS patients were different to, and 16 times more powerful than, that applied to the controls, then this would explain the WPI results, leaving only the mystery of why other researchers round the world are unable to detect any trace of this near-ubiquitous retrovirus.

But do you not think that either one of the WPI team, Dr Coffin, the Science review panel, or somebody at some point would have noticed this flaw?

Is it not frankly inconceivable that the WPI results were based on research in which they looked 16 times as hard for XMRV in CFS patients as they did for controls? Can we really imagine that the people involved in this study could fail to notice this point?

Alternatively, perhaps we should imagine that the WPI did realise that their whole approach was fundamentally dishonest, but they managed to swindle a bunch of scientists into failing to notice that their healthy controls were tested in a completely different way to the CFS patients?

Right from the outset, the WPI reaction to questions like this has had a hint of irritation about it, and from the word go they have reacted such questions by pointing out that their study was reviewed and published by Science. I think they've been quite restrained, actually, in dealing with the more insulting suggestions. If it were me, sarcasm mode would have kicked in at some point, and I would have asked in reply: "Do you really think your theory is so sophisticated that neither we nor the Science editors spotted that error? For some reason, you seem to think we're idiots!".

It's true that there has been an information deficit in certain areas, and perhaps we the patient community have done less than we could have to ask these sorts of questions of the WPI directly. All that detail does need teasing out, if only to head off the obvious potential flaws in the study that so many have alluded to in the absence of a clear description of why they are not possible. I suspect the WPI were hoping that the fact of publication in Science, and all the work they had to do to achieve that, meant that they were now exempt from impertinent questions about whether they had made some utterly basic and childish error. I suspect they were hoping everyone would focus on their findings and their implications, rather than delving into details of their methodology that Science magazine had already pored over for the best part of a year.

But frankly, I do still find these sorts of theories about the WPI study to be offensive and insulting to the WPI researchers. To include a methodological flaw of that magnitude would require either the utmost incompetence or the most profound dishonesty - and we would rightly tear them to shreds if it turned out the whole thing was no more than a giant swindle.

No offence intended in the above to you personally Kurt, your perspective is very valuable on this forum, and your sober notes of caution are extremely helpful for a hothead like me! But I still can't understand how you consider possibilities like this one to be realistic? Wouldn't the WPI have to be either phenomenally stupid or incredibly evil to produce such twisted science? I mean, the scenario you suggest is way beyond any flaw we've ever found in Simon Wessely's work, and with over 500 papers to his name in about 20 years, he's been churning out the studies at the rate of one a fortnight for the last 2 decades!

 

[Hope123]

Mark, many journals, and I would certainly expect Science to have a panel or ability to assemble a panel, have a statistics editorails staff so I believe that it would be been run by this group. Also, the reviewers picked for the paper would usually have some background to pick up on this. Studies and thus papers are not perfect but they should be adequate enough to pass some degree of muster to be published. I've a little experience on both sides - reviewing and submitting papers.

Kurt, the stats are more complicated if I remember right than what you're writing. The method I have a little familiarity with is Bonferroni correction but there are other methods. In any case, we don't have the data to figure this out. If anyone's so inclined:

http://en.wikipedia.org/wiki/Multiple_testing

 

[Hope123]

You know, I had to read Kerr's paper multiple times to understand what was going on because of the different cohorts and my fuzzy mind.

What stands out to me overall is the poor immune response of the BLT group and both CFS groups (SGUL and Glasgow), not just the low number positives but the poor affinity of their antibodies. I'm not sure if makes a difference that the neutralization test was initiated on the BLT group but then expanded to SGUL and Glasgow group. It seems like the neutralization assay overall is still pretty rough and based on comparison of neutralization curves between control neutralization antibodies and patient serum. To be expected I guess since these are early days.

I wonder why Kerr and company would pick a potentially immunosuppressed group to base the assays on -- BLT consists of sick people attending clinics and pregnant women who would be somewhat immunosuppressed. This might be why you get 3/226 positive in BLT and 22/157 positive in SGUL controls.

Maybe the message overall is antibody testing and PCR are both not great methods used only by themselves to identify XMRV in folks with CFS. (although the WPI supposedly has different results)

And what's deal with the positives in the SGUL controls being collected at one blood donation site? Kerr hints at a local outbreak, then doubts it. Hopefully, they're able to track down who these folks are and what, if any, special circumstances they have - do they work in a mouse lab? (As an undergrad, the group I worked with had a ritual of going down to donate blood and have juice and cookies together -- a twice monthly bonding event.)

 

[raul]

What would this mean? Well, if they took four samples and tested from each CFS sample four times, but just tested controls once, that would mean the CFS patients had 16x the testing 'resolution' that the controls did. So effectively, you have to multiply the control positives by 16 to compare them to the CFS positives statistically. That means that rather than 3.7% positives in the controls there were the statistical equivalent of approximately 3.7 X 16 = 60% positives, which is so close to the 67% positive rate in the CFS samples that there would be functionally no difference between the groups and XMRV is actually ubiquitous. That would raise some questions about the prostate cancer studies of XMRV.

We really need to see some blinded testing so things like this can not happen.

I completely agree with the importance of treating patient and control samples equally and with the advantages of blinded testing. In general this factor would be less than 16 given that samples from the same person would test positive more than once and that such repeated tests would not be independent of each other.

Say 8% of control volunteers have the virus and the test detects the virus 50% of the time. Then the first time I test I get 4% positives. This leaves 4% undetected infected volunteers. The second time I test half of these, that is 2% overall, turn out positive. If we do this 16 times we end up with F=4%+2%+1%+0.5%+ ....= something very close to 8% but less than 8%.
The final 8% may be further reduced if tests for the same person are correlated, as having a first false negative increases the chances of a second one.

So if the test is very insensitive the dominant factor is the number of repetitions. If the test is reasonably sensitive, say in the order 20-30%, the dominant factor is going to be what the real infection percentage is, so that with 16 repetitions you get very close to the real number.

BTW, I think it would be interesting to have a category for healthy people in the test polls to get a feel for sensitivity and specificity when results become available.