Cytokine responses to exercise and activity in patients with chronic fatigue syndrome: Case control

[RogerBlack]

FOIA to determine if they actually had any of the 2009 samples remaining at the time of the latter analysis would be interesting. It would seem to me that the sane thing to do if you do at that point when you're suspecting they're dodgy, would be to reanalyse.

 

[Valentijn]

FOIA to determine if they actually had any of the 2009 samples remaining at the time of the latter analysis would be interesting.

Might as well get the anonymized individual results as well, while they're at it. They probably won't provide means, on the usual basis of being too stupid to calculate them in less than 18 hours, so I suppose we'd have to calculate them ourselves.

 

[Hutan]

The write up is quite bizarre. If you establish that your methodology is inconsistent you bin you results and either try again or forget it.

So the lab work was done in 2009 and 2011. And yet the study is only being published in 2017?

Maybe the study was indeed filed in the bin, but was resurrected now for a reason? Just maybe someone decided that there needed to be a UK study that throws doubt on the idea that there are cytokine differences in ME/CFS - in time for any NICE review?

 

[Valentijn]

This was also published as an abstract in June 2015, with no mention of a lab error:

Case control study of circulating cytokines in chronic fatigue syndrome

P.D. White, L. Clark, J.M. Thomas, M. Murphy, M. Buckland
Queen Mary University of London, Academic Psychological Medicine,
London, UK

Background: A hypothesis suggests that chronic fatigue syndrome is associated with elevated circulating concentrations of cytokines, supported by the role of infections triggering the illness and the similarity with acute sickness behaviour, which is associated with elevated circulating cytokines. A preliminary study supported the roles of transforming growth factor beta and tumour necrosis factor alpha.

Method: We measured serum protein levels and mRNA of 11 cytokines at baseline, after commuting across London, and before and after exercise. After exercise, samples were taken immediately, 3 h and 2 days later. We studied 24 patients with CFS and 21 age- and sex-matched healthy but sedentary controls, following a preliminary study.

Results: There were significant differences between patients and controls in TGFβ protein levels both at baseline and all other measurement points. No other cytokine was abnormally distributed. Neither commuting nor exercise had a significant effect on TGFβ or any other cytokine concentrations. Therewere no significant differences in mRNA between groups at any time point.

Conclusion: This study replicates the results in both our preliminary study and a systematic review that circulating TGFβ was at a higher concentration at all times in patients with CFS compared to controls. In this study, mRNA for TGFβ was no different between patients and controls. This finding cannot be related to post-exertional exacerbation of symptoms, but needs further exploration. Studying cytokines in other body tissues, such as cerebrospinal fluid may provide different results.

doi:10.1016/j.jpsychores.2015.03.139

 

[Valentijn]

It was published an additional time in 2012 as an abstract for the
"15th Annual Meeting of the European Association for Consultation-Liaison Psychiatry and Psychosomatics (EACLPP) & 29th European Conference on Psychosomatic Research (ECPR)", again without any mention of a lab failure:

22 - Cytokine responses to exercise and activity in patients with chronic fatigue syndrome: Proof of principle study

Clark L. 1, Buckland M. 2, Vleck V. 3, White P. 1

1 Barts and The London School of Medicine and Dentistry,
2 Barts and The London NHS Trust, England,
3 Department of Human Kinetics, Technical University of Lisbon, Portugal

Objective: The aim of this study was to test whether the characteristic feature of post-exertional malaise in CFS/ME is associatedwith elevated cytokine plasma concentrations TGFbeta, TNF-alfa, IL-6 and IL-8.

Method: This proof of principle study compares the cytokine responses of 24 CFS/ME cases and 21 healthy sedentary controls, to the physical activity involved in 1) commuting from home to hospital and 2)completing a sub-maximal aerobic exercise test.

Results: Cases were significantly more likely to report increased fatigue at rest, and after commuting and aerobic exercise. Cases spent significantly less time exercising (p = 0.02). Plasma levels of TNF-alfa and IL-6 were most often unmeasurable in both cases and controls and so statistical analysis was not appropriate. Plasma levels of TGFbeta were significantly elevated in cases at all times compared to controls, including at rest in the morning prior to commuting. Plasma levels of IL- 8 were not elevated in cases compared to controls except at 2 days after aerobic exercise. Changes in TGF-beta and IL- 8 seen after commuting and after aerobic exercise were not significantly different between groups. Increased post-commuting fatigue, pain and malaise were not associated with an increased plasma concentration of TGF-beta in cases compared to controls. Depressed and anxious mood, physical deconditioning and sleep disturbance did not confound the differences in TGF-beta and IL-8 concentrations between cases and controls.

Conclusions: We found elevated concentrations of TGF-beta in cases of CFS, which were not confounded by depressed or anxious mood, physical deconditioning or sleep disturbance, but this elevation was not associated with post-exertional fatigue. We did not replicate previous findings of elevated TNF-alpha or IL-6.

Keywords: chronic fatigue syndrome, cytokines, post-exertional malaise, aerobic exercise, TGF-?

 

[Cheshire]

Thanks @Valentijn for these abtracts.
I am more and more inclined to think that the timing of publication for this paper is not due to chance.

There are no dates for submission or peer review, but this message:
"This article has been accepted for publication and undergone full peer review but has not been
through the copyediting, typesetting, pagination and proofreading process which may lead to
differences between this version and the Version of Record."

Is it frequent or unusual to have such a message?

Edit: all the other same accepted papers bear the same message, so nothing unusual...

 

[user9876]

Thanks @Valentijn for these abtracts.
I am more and more inclined to think that the timing of publication for this paper is not due to chance.

There are no dates for submission or peer review, but this message:
"This article has been accepted for publication and undergone full peer review but has not been
through the copyediting, typesetting, pagination and proofreading process which may lead to
differences between this version and the Version of Record."

Is it frequent or unusual to have such a message?

Clicking on the full publication history of the main paper gives

Publication History

• Accepted manuscript online:5 August 2017
• Manuscript Accepted:5 July 2017
• Manuscript Revised:27 June 2017
• Manuscript Received:21 December 2016

 

[Jonathan Edwards]

This was also published as an abstract in June 2015, with no mention of a lab error:

I thought that might be the case. The mention of a result that was then claimed be an artefact suggested that it previously been made public and they wanted to 'put the record straight'.

 

[Valentijn]

I still haven't found a protocol. It probably exists, since it's mentioned, so it could probably get produced with an FOIA requist. But one thing to note is that their previous cytokine study from 2004 found a similar difference in TGF-beta between patients and controls.

2004 TGF-beta (pg/mL):

2017 TGF-beta (ng/mL):

2004 TGF-beta median values were approximately 18.1 times as high in patients versus controls:

However, we found elevated concentrations of plasma transforming growth factor beta (TGF-β), even before exercise, in subjects with CFS (median (IQR) of 904 (182-1072) pg/ml) versus controls (median (IQR) of 50 (45-68) pg/ml) (P < .001).

2017 TGF-beta median values were only approximately 2.3 times as high in patients versus controls. The supposedly faulty Batch 1 values were about 3 times as high as the later batches.

So it seems odd that they claim operator error for one batch, but not later batches processed by the same person, nor their earlier study which found even huger differences.

It's also interesting that their 2004 patients were a lot more disabled on the SF36-PF, with a median score of 40 compared to a median score of 60 in 2017.

Something I strongly dislike is that all controls from 2004 were "chosen" from hospital staff, and included an unstated number of employees and employee friends/family in 2017. Choosing controls who know about the study and possibly have an agenda or want to help a researcher who has an agenda can skew results quite a bit - for example, pushing harder on a CPET, or doing something extra which might help increase TGF-beta.

So how normal were the controls TGF-beta values? In the 2017 study they used serum, which seems to have a normal range of 3,465-13,889 pg/mL according to several lab sites, including Mayo. The controls had a median of 18,900, averaged over all time points, which would seem to be pretty elevated. (Side note: several of the 2017 graphs have units of measurement mislabeled, probably due to shortening long numbers to fit better.)

 

[Valentijn]

The more I look at it, the less sense Table 7 makes to me. The combined 11 patients and 19 controls tested by Operator 2 are also listed as Batch 2 and Batch 3. Are they saying that only Batch 1 is Operator 1, and Batch 2 and Batch 3 are Operator 2? It seems like a very confusing way to present the data:

 

[Snow Leopard]

The more I look at it, the less sense Table 7 makes to me. The combined 11 patients and 19 controls tested by Operator 2 are also listed as Batch 2 and Batch 3. Are they saying that only Batch 1 is Operator 1, and Batch 2 and Batch 3 are Operator 2? It seems like a very confusing way to present the data

Yes, it seems "operator 2" = Batch 2 + Batch 3.

So the statistical power is basically halved (11 patients vs 24 patients) due to the loss of "operator 1".

If they provided the actual distribution of data / mean and SD, then we could see whether a larger sample size would have had a significant result.


There are some sketchy methods to estimate it, but well...

Having said that, given the data they've provided, they'd probably need a sample size of over 100 for p<0.05

 

[wastwater]

I can get onto the back of this following my route foxc1 then Bmp4 then tgf-b

 

[Dolphin]

It's interesting that they claim they noticed a bimodal pattern of TGF levels, leads to speculation as to "true" vs "false" CFS cases, given they used the non-specific CDC "empirical criteria" (that found 10x more patients in CDC population based studies, compared to the original CDC "Fukuda" criteria) and the relatively mild mean scores on the SF-36, Fatigue scale compared to other studies. Perhaps there are some non-Fukuda (and non-CCC) cases that are skewing the results?

No, the empirical criteria are Reeves et al 2005. They used Reeves et al 2003 which are basically the Fukuda (1994) criteria.

 

[Snow Leopard]

No, the empirical criteria are Reeves et al 2005. They used Reeves et al 2003 which are basically the Fukuda (1994) criteria.

My mistake.

 

[Dolphin]

Cytokines, such as TGF-â, could thus affect the perception of pain and fatigue through both direct and indirect effects.

I reckon there's a good chance that Peter White suggested this sentence. He is big into this idea of interoception that we are perceiving that we are more fatigued and in more pain than we really are. Lots of things they could talk about but they bring this up even though their study wasn't looking at the issue.

Full paragraph:

This study does not rule out a role for local release of cytokines in the central nervous system (CNS). There is a two way relationship between the immune system and the CNS [3, 4]. Cytokines can cross the blood brain barrier, particularly when the blood brain barrier is modified by factors such as intercurrent infections or distress [38]. Cytokines are locally released by glial cells in the CNS, and TGF-â can cause central motor fatigue in animals [5, 39]. Cytokines, such as TGF-â, could thus affect the perception of pain and fatigue through both direct and indirect effects. There is some recent evidence to support neuro-inflammation in CFS/ME [40]. TGF- â appears to have complex effects in the brain, but may promote a pro-inflammatory state and disturbance of blood brain barrier function [41].