Thursday, June 16, 2011

DUSTY MILLER REDUX

Second Email Exchange Between 
Dusty Miller and CFS Central

CFS Central:  You wrote: "Regarding the data in Lombardi et al., I was initially convinced by their extensive analysis, in particular, their ability to grow virus from patient materials. Indeed, we entered the CFS/XMRV field largely on the basis of these data. However, published and unpublished data now indicate that all of these methods were flawed.”
 
As far as I'm aware, no published or unpublished data indicates that the methods of Lombardi et al. were flawed. To what studies are you referring? 

Dr. Dusty Miller:  In the recent Science Express paper by Knox, ..., and Levy, the authors used three techniques to detect XMRV and related retroviruses in humans with CFS, and found none.  These techniques included nested PCR, an assay for infectious virus, and assays for neutralizing antibodies and other inhibitory factors in blood.  Importantly, many of the CFS subjects they studied came from Dr. Peterson's practice in Incline Village, and represent some of the same patients that Lombardi et al. studied.  Therefore, this is a close replication of the Lombardi et al. study.  The Knox study is in addition to many other negative studies cited in this paper (refs. 7-12).  

Lastly, the XMRV viral sequences deposited by Lombardi et al. in GenBank all are very closely related the VP62 XMRV sequence (see the supplemental material in Knox et al.), and the VP62 plasmid from Robert Silverman was apparently present in the labs of Lombardi et al., strongly indicating contamination of the Lombardi et al. PCR assays by VP62 plasmid DNA.  Putting all of this together leads me to conclude that the methods of Lombardi et al. must have been flawed.

CFS Central:  You also wrote: "Regarding the constant accusation that no one has carefully replicated Lombardi's methods, this is not true as far as most scientists are concerned. Initial reports attempting to replicate the study did have flaws, but many later studies are convincing.”

The problem with these later studies as I see it is that not one of the so-called replication studies was a bona fide replication. The patient cohorts and/or methods used were not identical. I learned in 9th grade science that being identical in cohort and methods are necessary for a true replication—otherwise you’re introducing variables that may account for the different findings. In your view, could the different methods/cohorts account for the differences in the findings? Why or why not?
 
Miller:  Please see my response above.  It is difficult to perform an exact replication study, which would involve going to the Mikovits lab and watching whoever did the previous study repeat it with the same patient samples.  Besides which, you may have also learned in 9th grade science that scientific results must be generalizable, that is, competent scientists must be able to repeat the experiments under somewhat different conditions and obtain similar results.  If the claimed result can only be obtained by performing the experiment in one spot in the Mikovits lab, perhaps while singing a particular song, then the results are not generalizable and should be looked on with suspicion. 

19 comments:

  1. Dr. Miller sets up a straw man argument when he suggests that it would be impractical and kind of silly to have to go to the identical spot in the same lab in order to faithfully replicate an experiment.

    No one is suggesting such ridiculous measures. He is simply avoiding honestly answering Mindy's question about replication.

    ReplyDelete
  2. I really think Dusty Miller and his nonsense have been given too much attention already. Let's see his study. I see no need for further comment on an unpublished study.

    Patricia Carter

    ReplyDelete
  3. That's probably the problem he can't do a replication study because no XMRV positives are willing. And judging by his actions it was a wise move on their part. I doubt he'll get his hands on any Peterson patients now either (with the exception of a couple maybe). I'm sure they've figured out by now how they were used. Problem is no one knows if they were actually patients from the Lombardi study. Especially now that we know the authors of the Knox study did not have patient information, or results of who the XMRV postives or negatives were. For all they know it could have been random samples from the CAA biobank.

    ReplyDelete
  4. Dr. Miller is again being disingenuous.

    For example, he says " Importantly, many of the CFS subjects they studied came from Dr. Peterson's practice in Incline Village, and represent some of the same patients that Lombardi et al. studied."

    Well O.K., that may well be true, but how many of those patients actually tested XMRV +ve in the Lombardi et. al. study?

    And what about the differences in assays which resulted in Knox being unable to find virus in anyone and hence produce yet another 0/0 study.

    In her excellent blog http://treatingxmrv.blogspot.com/2011/06/motivation.html, Dr. Jamie Deckoff Jones suggests a whole host of reasons why Knox and Peterson would not have been motivated to make an honest effort to replicate Lombardi et al. Clearly, it doesn't suit Dr. Miller's negatively biased position (nor his allegiances) to address those kinds of issues.

    Glibly superficial answers doesn't fool anyone and, as ever, the devil is to be found in the all-important details.

    ReplyDelete
  5. Yeah I dunno about that. The Levy/Knox paper might have been a good effort. But they didn't even use the same cell line type as Lombardi. And there were other non-minor differences in their PCR protocol. It wasn't a shoddy study by a long shot, just as the Singh paper wasn't–but it didn't duplicate the methods of the Lombardi paper exactly. Using different methods, even if they are "supposed" to work in theory, introduced new variables. Nobody is going to be able to get to the bottom of this until someone goes back to the basics and tries the exact methods of the original paper. Its really that simple.

    ReplyDelete
  6. What an obvious dodge...and a really nasty thinly veiled swipe at Mindy as well.

    Color me unimpressed and unconvinced.

    *sigh* and I had such hope for the Hutch too...

    ReplyDelete
  7. Dusty, you have lost it. The word similar, does not mean replication. No study has attempted to replicate Lombardi or Lo et al.

    The fact is, it is very easy to replicate their assays, but no one can be bother to use the same variables as those labs. If you cannot control your annealing temperatures, buffer, the primers, the PCR type, etc. then you should all leave the field of virology. After all this is what you train to do and have equipment to do for you. Any idiot can set their professional oven in top restaurant kitchen to the same temperature each time they make soufflé. Perhaps you should observe Heston Blumenthal making ice cream:
    http://www.youtube.com/watch?v=D6CLoRuvGcY

    You would never get away with using unvalidated HIV assay to declare people negative and you won't get away with the same thing with HGRVs.

    Here are some of the variable they have not replicated. Making ridiculous comments about singing a song is only trying to insult the intelligence of other scientists.

    Enhancing agents
    Matric analyses
    Mg2+ concentration
    Annealing temperature
    Primer design
    Cycle number
    Pipetting and thermal cyclervariables
    Touchdown PCR
    etc.

    You know full well what was changed in Knox et al.

    ReplyDelete
  8. Very interesting, "If the claimed result can only be obtained by performing the experiment in one spot in the Mikovits lab, perhaps while singing a particular song, then the results are not generalizable and should be looked on with suspicion." Apparently Silverman, the Cleveland Clinic, the NCI and the Ruscettis played no role.

    I would like to point out that the singular event hypothesized in Paprotka et al. means that work is not even replicable in principle. We also see that failure to detect virus can be supporting evidence in the early passages, or it may be conveniently ignored when a second piece of the puzzle fails to materialize. Detecting virus can mean detecting a single gene sequence, or this can be taken as evidence the virus is not present. Very, very interesting.

    ReplyDelete
  9. We now see that having a positive control resembling the result invalidates a study via contamination. We have previously seen that hundreds of negative controls which remain negative are irrelevant. Real scientists must be able to work entirely without controls.

    ReplyDelete
  10. If this new paper was about to be published you would think his university would have had a press release in advance. Funny I can't find one.

    ReplyDelete
  11. re. my comment about the Hep B vaccine... So, because CFS has occurred in clusters often, it makes this cluster somehow less interesting? not sure why I got jumped on for that one.

    ReplyDelete
  12. All I know is when samples are handled the same way the patients in the lombardi science paper were 67% positive and controls 4% positive.
    In the Lo et. al. study by the NIH and FDA there was 87% positive and controls 7% using Coffin's requested detection methods for contamination.
    Until someone shows that those studies were contaminated and how it happened the science of HGRV's should not be left behind.
    If the WPI and Alter discover contamination they will act as scientists and admit so.
    I feel a scientific endeavor of this importance with so many contradictory results with many of the positive results ignored or having funding denied means research need to continue into HGRV's in humans.

    ReplyDelete
  13. Mindy, I have a suggestion. (sorry, you're doing all the hard work researching and blogging and I'm doing nothing but come up with a suggestion) :- discussions of detection and replication methods is always going to get technical and muddled. I think the most productive line of questioning is the differentiation of controls and patients in Lombardi; now repeated (blinded) by Cornell Uni. This is much more clear-cut because it is independent of the technical issues.

    I would actually have liked to hear his comments on that; because I though he was somewhat in the middle (or used to be).

    PeterW

    ReplyDelete
  14. @Peter: The technical details are not muddled at all for anyone who understands them. The negatives studies don't refute the findings. Not sure why you highlight the Hanson study is blinded, as Lombardi et al. was also blinded.

    ReplyDelete
  15. Andy said...
    re. my comment about the Hep B vaccine... So, because CFS has occurred in clusters often, it makes this cluster somehow less interesting? not sure why I got jumped on for that one.


    Andy, you've been around M.E. circles for a while now. Why does M. Bloomer's hepatitis vaccine outbreak merit a "whoa" when we're speaking of an epidemic? Since the epidemiology of this illness is, as I wrote before, dismal and undermined by rather sinister agendas, why aren't you saying to M. Bloomer and all of us, wow, something like this happens and nothing is ever said or done.

    What I was objecting to was my sense that your tone suggested that all that was necessary was to tell the right people and they'd get on the job and investigate it thoroughly. From the point of view of M.E. history, this has NEVER happened. Think about that. How much opportunity for intellectual, scientific knowledge has been lost, let alone the suffering it represents.

    You inhabit a middle class world where you expect responsiveness to your medical problems, where you don't have to lie to emergency room personnel about a disabling condition because of fearing their dismissal of your pressing difficulties. We do not.

    Your blind spot is shared by people who are less well intentioned than you are, but in the end, it amounts to the same thing, a denial of the history and severity of this illness, both in outbreaks and other modes of transmission.

    (there's an excellent thread on m.e. outbreaks running on PR: http://forums.phoenixrising.me/showthread.php?12275-ME-outbreaks-in-the-Swiss-Army-in-the-1930s

    akrasia

    ReplyDelete
  16. http://www.plosone.org/article/info%3Adoi%2F10.1371%2Fjournal.pone.0020874?utm_source=feedburner&utm_medium=feed&utm_campaign=Feed%3A+plosone%2FVirology+%28PLoS+ONE+Alerts%3A+Virology%29

    While this study is not investigating ME/CFS it does suggest that XMRV has a proclivity to immune dysfunctional cell lines in the laboratory. How can blood samples of "controls" show only 3.7% infection while that of ME/CFS samples show 67%? Because the contaminant can infect the immune-dysfunctional cell culture. Can antibodies be produced in the cell culture? If so, that might explain why many of the ME/CFS samples had antibodies to XMRV.

    ReplyDelete
  17. Sfanos et al. used unproven assays to detect XMRV. They didn't find XMRV though. The sequences they say are in the GenBank are also not there.

    You cannot get an immune reaction from a contaminant.

    ReplyDelete
  18. @Ian Hodgson PhD - Even if XMRV selectively infects cell lines that are disabled in their immune response, it can't explain the discrepancy in positives between patients and controls as you claim. Thats because all samples are blinded and tested the same way. Patients and controls (and water controls) are run through the same cell lines for culture using the same method. So if the cell lines are contaminated, then you'd expect all of the samples to show positive not just the patients.

    ReplyDelete
  19. Rather than participate in a flame war about his own work, I would like to tell Dusty why I think the proposed recombination hypothesis was a strategic blunder of epic proportions, XMRV completely aside. I hope this will not disappoint other readers of this blog, and will benefit some scientists whose work might be harmed.

    You might just remember some heated public debate in Cambridge during the 1970s over the potential for recombination origin of human pathogens in research laboratories. Please choose whatever probabilities you like from the current debate over the origin of XMRV and insert these in those transcripts. How do you like the effect?

    There is a second probability I have not heard mentioned, the probability of actually observing such a recombination event. If this is not wildly improbable, there must be many more recombinations never observed. Had Sfanos et al. not had a putative origin for XMRV to blame they might have assumed recombination responsible for the contamination of those cell lines which changed during their study.

    Please consider that card-carrying researchers will not be the only ones finding fragments of retroviruses that fit together 'like two pieces of a puzzle'.

    Your professional time is likely more valuable to you than my time is to me. I don't think I'm wasting your time with the above argument. You may want to have the numbers at your fingertips when you talk to university attorneys.

    ReplyDelete

Comments are welcome and moderated for appropriate content.