Comment on Detection of Mycotoxins in Patients with Chronic Fatigue Syndrome Toxins 2013, 5, 605-617

The paper by Brewer et al. (2013) has a key methodologic flaw [1]. The control group selected was inappropriate, resulting in an invalid comparison and findings.

The essence of a case-control study is to compare a case group having a disease with a group from the same general source population that did not develop the disease, but had the same opportunity to develop the disease and be included in the case group. When the case and control groups are compared, differences in exposure may suggest possible causes of the disease, or factors associated with causes.

In [1], diagnosis of chronic fatigue syndrome (CFS) was apparently the sole criterion of selection for the cases, which seems appropriate [1]. After inclusion, over 90% of cases were found to have biomarkers of exposure to specific fungal toxins of interest, which were suspected of involvement in causing the disease. After inclusion, most also reported a history of exposure to water damaged buildings (WDB), where these toxin exposures are presumed to have occurred. The reported WDB exposure, in over 90% of the cases, was not related to their original selection as a case group. The controls, on the other hand, were defined as “[h]ealthy control patients with no known toxic mold exposures in water-damaged buildings.” Thus controls were free of CFS and also without reported history of exposure to WDB environments, the presumed source of the toxin exposures.

An appropriate control group would have consisted of individuals without diagnosed CFS, chosen as much as possible from a population who might have ended up in the case group if they had developed CFS. To exclude from the controls those without opportunity for the exposure of interest is completely inappropriate. This control selection strategy, aside from making the results invalid, suggests the authors may not have understood the essential purpose and requirements of a case-control comparison. Normally, a case-control study of the disease and exposures of interest in this study would be conducted by comparing a group of people with CFS diagnosed by specific criteria, and a group without diagnosed CFS. There would be no consideration, in the selection of either cases or controls, of what exposures the subjects thought they had been exposed to. That would involve a very subjective and imprecise way to select subjects, might have little to do with actual exposures, and most importantly, would likely introduce bias into the analysis.

It is not evident that other types of control groups would be preferable. For instance, controls who had CFS but were not knowingly exposed to WDB would give you limited useful information. The reported exposures would have no demonstrable association with disease since all the subjects would have the disease, but the results would show, among people with diagnosed CFS, whether thinking you had prior WDB exposure was associated with specific mycotoxin exposures. Alternatively, investigating whether reporting prior WDB exposure was associated with higher biomarkers of fungal mycotoxins, but in groups selected without respect to disease and not biased by this association, would be an interesting but different study.

It is important to point out that the problems with the study are related not to the selection of cases, but only to the selection of controls. Proper selection of cases but inappropriate selection of controls can make a case-control comparison invalid. I would hope that in their response, Dr. Brewer et al. deal clearly and directly with the issue of the control group selection, and provide their explicit opinion on the issue of whether the stated use in the study of both non-CFD status and non-WDB history to select controls was correct. (Apparently the only epidemiologist involved in the original paper, Dr. Madison, has died, so she cannot respond, and the remaining authors may not fully understand the criticisms or be able to respond to this question.) Also, despite the statement in the original comment by Dr. Osterman (2016) that the case-control comparison was “rigged,” that is not an issue that can be or needs to be resolved [2]. The important issue is the invalid control selection, regardless of whether due to intention or error.

While a claim may be made that the article by Brewer et al. (2013) was only a reported case series and not intended to be an epidemiologic case-control study, this is not a credible claim [1]. The researchers studied a diseased group, and the “results were compared to healthy control subjects previously reported by the same testing laboratory.” The comparison group was defined as “[h]ealthy control patients with no known toxic mold exposures in water-damaged buildings.” Their urine specimens “were used to develop reference data for the control group used in this study.” Mycotoxins “in the urine of patients and controls were statistically analyzed to determine if a difference existed between the two groups.” So even if the authors, including the epidemiologist, somehow did not realize their study would be read as an epidemiologic case-control comparison, this will be the universal interpretation of readers, and this is how the paper should be evaluated.

I think it would be unfortunate if Brewer et al. (2013) were cited as documenting a relationship between CFS and a body burden of mycotoxins [1]. This relationship may or may not exist, but this paper has not shown evidence to support it. I would advise the journal that in the future, review of any submitted manuscript about toxins that involves an epidemiologic study should include careful epidemiologic review.

 

Source: Mendell MJ. Comment on Detection of Mycotoxins in Patients with Chronic Fatigue Syndrome Toxins 2013, 5, 605-617. Toxins (Basel). 2016 Nov 7;8(11). pii: E324. https://www.ncbi.nlm.nih.gov/pmc/articles/PMC5127121/ (Full article)

 

Reply to Comment on Detection of Mycotoxins in Patients with Chronic Fatigue Syndrome Toxins 2013, 5, 605-617 by John W. Osterman, M.D.

This paper [1] was an observational case study. It was not intended to be, nor have we ever indicated that it was, an epidemiologic study [2]. One of the authors (Dr. Brewer) is an infectious disease specialist, who treats a number of patients with chronic fatigue syndrome (CFS). Dr. Brewer’s primary responsibility is to properly diagnose and treat these patients and ensure their wellbeing. In 2012, Dr. Brewer began to test patients for the presence of mycotoxins using the RealTime Lab’s mycotoxin panel. As he saw and treated more and more chronic fatigue patients, he began to see an association between the presence of mycotoxins and the symptoms of CFS. As this association became more apparent, Dr. Brewer discussed these findings with other experts in the field of mycotoxins. It was decided that these observations had potentially important clinical implications and the group decided to proceed with publication of this collection of clinical cases. The patients reported in our study were included based on their diagnosis (CFS) and not their exposure history.

These observations did lead to a hypothesis that perhaps the patients had internal fungal growth leading to both the symptoms of CFS and the presence of the mycotoxins produced by the fungi. Subsequently, this resulted in a treatment regimen for fungal colonization/infection in the sinuses, the results of which improved both the patient’s health and reduced the concentration of mycotoxins.

Never did the authors state or imply that mycotoxins caused CFS and never did we undertake a controlled study to look at CFS in a mycotoxin positive and a mycotoxin negative population. The major finding was the association between mycotoxins and CFS. In the paper (discussion section) several ideas were addressed (e.g., mitochondrial toxicity) as to possible pathophysiologic mechanisms.

The reference to the negative controls of another study, where the individuals were not exposed to a water damaged and potentially mold infested environment, was only meant to point out that the entire general population does not harbor elevated levels of mycotoxins, and/or the molds that produce them (despite low levels of exposure in the environment and potential mycotoxin-exposure in foods).

Much work would be and is needed to link mycotoxins and or mold as the causative agent of CFS and the authors understand that this would necessitate a clinical study with the appropriate mycotoxin negative controls. While this may be a future project, the focus now is on patient treatment and presentation of case histories such as the ones in this paper.

In summary, this was a clinical observation, not an epidemiological study. The findings are provocative and may have important implications for these types of illnesses. The results will hopefully stimulate and promote further investigation by our group and others.

 

Source: Brewer J, Thrasher JD, Hooper D. Reply to Comment on Detection of Mycotoxins in Patients with Chronic Fatigue Syndrome Toxins 2013, 5, 605-617 by John W. Osterman, M.D. Toxins (Basel). 2016 Nov 7;8(11). pii: E323. https://www.ncbi.nlm.nih.gov/pmc/articles/PMC5127120/ (Full article)

 

Comment on Detection of Mycotoxins in Patients with Chronic Fatigue Syndrome. Toxins 2013, 5, 605-617

Abstract:

The paper by Brewer et al. entitled “Detection of Mycotoxins in Patients with Chronic Fatigue Syndrome. Toxins 2013, 5, 605–617” is so methodologically flawed that it should never have been published in the scientific literature [1].

In this paper, the authors measure the presence of mycotoxins in the urine of 112 patients suffering from chronic fatigue syndrome (CFS). These finding are then compared to urine samples from 55 healthy control subjects “… with no history of exposure to WDB (water damaged buildings) or moldy environment…” (sic). Not surprisingly, there were more people from the CFS group with mold exposure than in the comparison group. These results are not surprising because, BY DEFINITION, the control group had no history of exposure to mold. By purposely choosing a control group with no history of mold exposure, the authors have statistically rigged their results in such a way that only a positive relationship will be found when compared to the CFS group.

Using the same approach, the authors could test urine from their CFS patients for the presence of caffeine metabolites and compare the results to urine from a group not exposed to caffeinated beverages; they would find more caffeine metabolites in the CFS group for the same methodological reasons, the control group having been purposely selected to be not exposed. The same would be true for nicotine metabolites in the CFS patients’ urine using urine from non-smokers as a comparison group or comparing urinary animal protein metabolites from the CFS group to animal protein metabolites in urine from vegetarians. The results from these studies would show a positive but erroneous association between CFS and caffeine, nicotine and animal protein. The same is true for the relationship that Brewer et al. purportedly found in this study of CFS and mold. The findings from this study are misleading and meaningless.

This study is an example of extreme selection bias and is akin to showing that men are shorter than women by comparing the height of an average group of men to that of women on the national basketball team!

Given the mountain of “junk” science on the Internet, I feel that a credible on-line scientific journal must ensure rigorous methodological standards for the papers it publishes. Such was not the case for this paper.

 

Source: Osterman JW. Comment on Detection of Mycotoxins in Patients with Chronic Fatigue Syndrome. Toxins 2013, 5, 605-617. Toxins (Basel). 2016 Nov 7;8(11). pii: E322. https://www.ncbi.nlm.nih.gov/pmc/articles/PMC5127119/ (Full article)

 

‘PACE-Gate’: When clinical trial evidence meets open data access

Abstract:

Science is not always plain sailing and sometimes the voyage is across an angry sea. A recent clinical trial of treatments for chronic fatigue syndrome (the PACE trial) has whipped up a storm of controversy. Patients claim the lead authors overstated the effectiveness of cognitive behavioural therapy and graded exercise therapy by lowering the thresholds they used to determine improvement. In this extraordinary case, patients discovered that the treatments tested had much lower efficacy after an information tribunal ordered the release of data from the PACE trial to a patient who had requested access using a freedom of information request.

© The Author(s) 2016.

 

Source: Geraghty KJ. ‘PACE-Gate’: When clinical trial evidence meets open data access. J Health Psychol. 2016 Nov 1. pii: 1359105316675213. [Epub ahead of print] https://www.ncbi.nlm.nih.gov/pubmed/27807258