Articles on Montoya and Mark Davis's Cytokine work

Montoya, Davis et al published their paper on cytokine signatures for ME in July.

Cytokine signature associated with disease severity in chronic fatigue syndrome patientsPNAS 2017 114 (34) E7150-E7158; published ahead of print July 31, 2017, doi:10.1073/pnas.1710519114
http://www.pnas.org/content/114/34/E7150.full

Now there has been a letter/article in response in PNAS that criticizes their methodology. Unfortunately its not public access
Cytokine signature in chronic fatigue syndrome
Megan E. Roerink, Matthew Buckland, Andrew R. Lloyd, and Jos W. M. van der Meer,
Extract:

And then a reply from Mark Davis

Reply to Roerink et al: Methods for recruitment, serum separation, and storage were the same for patients and controls

• Jose G. Montoya, Holden T. Maecker, Yael Rosenberg-Hasson and Mark M. Davis

Extract

 

Full text of the letter is at http://sci-hub.io/10.1073/pnas.1714011114 . Basically, they claim that they screwed up in the past and later decided their results weren't significant after all. I think this might be the same group who published a "nothing to see here" biomedical paper recently after sitting on it for years:

Full text of the author response is at http://sci-hub.io/10.1073/pnas.1714692114 They respond with a lot of the details of their methodology, basically establishing that they know what they're doing, unlike other groups . And I think this bit addresses concerns regarding different batch processing variability:

And they also refute the "nothing to see here", which is pretty relevant. If results were down to processing differences, it would be unlikely that random differences would consistently favor abnormally elevated results for ME patients:

 

I do love it when the serious pros ride into town, and show the amateurs how it is done.

 

The reply is reassuring.

 

Briefly: Roerink et al argued that methodological issues with blood plasma processing could account for the finding of higher cytokine TGF-beta in patients. Montoya et al responded that they looked at blood serum, not plasma, as their paper had spelt out.

Whatever the motivations behind the letter, I do like such methodological geek-sparring, which I think is good for keeping science honest. The tone of the initial letter was polite and I was like that it suggested specific ways to check that the TGF finding wasn’t down to contamination from platelets (that also produce TGF-beta).

Montoya and colleagues don’t respond to this specific point, instead emphasising that while the critical letter highlights issues with plasma, the Montoya paper used serum as they spelt out throughout the paper. This is very important, I think, not to mention a blunder by the critical authors. It’s all down to clotting and the difference between serum and plasma:


Differences between Serum and Plasma
Serum is the fluid you see leaking out of the wounds, it’s what’s left over from blood after it clots. Plasma is slightly different: it’s the fluid you get left with when the red and white blood cells, and platelets are separated away by centrifuging (which is why the first letter gets so excited about centrifugal details). You have to add anti-clotting agents to get plasma.

Key difference here is that clotting takes place to generate serum, which takes platelets out of the equation as they form the clots. Whether or not that removes all possibility of platelet-derived TGF-beta contamination being a factor, I don’t know. But it does mean that the letter from Roerink and colleagues is largely barking up the wrong tree.

 

There may be a problem with serum too:
(the second plasmin step is not relevant)

Nat Med. 1995 Sep;1(9):932-7.
Release and activation of platelet latent TGF-beta in blood clots during dissolution with plasmin.
Grainger DJ1, Wakefield L, Bethell HW, Farndale RW, Metcalfe JC.
Author information

Abstract
Transforming growth factor beta 1 (TGF-beta 1) is a platelet-derived cytokine involved in both normal wound healing and scarring. We show that human platelets contain two pools of latent TGF-beta 1, which constitute more than 95% of the total TGF-beta assayed in whole platelets. During clotting, one pool, the large latent TGF-beta complex consisting of latent TGF-beta binding protein (LTBP), the latency-associated peptide (LAP) and the 25-kD mature TGF-beta 1 dimer is released into the serum. A second pool, which contains LAP but not LTBP, is retained in the clot, but can be released by RGD peptide. When the clot is dissolved by plasmin this bound TGF-beta 1 is gradually activated and released. If similar mechanisms operate in vivo, the clot will act as a slow-release capsule of TGF-beta 1 activity during wound healing.
-------------------


But if the Montoya samples were all collected the same way and run on the same plates they should be OK.

Like Simon, I think it is good to have this sort of critique going on. You can never get enough of methodological cross-checking. Problems arise so easily.