All metabolites with a q < 0.15 are included. N/A stands for Not Assigned. HMDB stands for Human Metabolome Database.
Figure 1. Box plot distribution of logged values for metabolites scored as being statistically different between controls (red) and patients (blue) at p < 0.05 and q < 0.15 by the Wilcoxon test. HMDB identity and test values can be found in Table 2. Y-axis scale is log10 transformed data.
"For example, cAMP and IMP are compounds known to be involved in many aspects of human body function, such as purine metabolism, chemical energy storage in muscles, and intra-cellular signal transduction. It is therefore extremely difficult to pinpoint a singular pathway linked to ME/CFS status or symptoms based on such compounds or, on the contrary, using compounds of which there is little to no knowledge, such as 2’-O-methylcytidine or gamma-glutamylthreonine. The latter molecule, however, is mentioned as a potential compound of interest among many other biomarkers to determine liver toxicity of a given agent in a patent . Results from our previous work  had focused attention on liver injury biomarkers.
Another metabolite of major interest is alpha-ketoglutarate because it is part of the “Energy” super-pathway and the TCA cycle sub-pathway. Indeed, the Krebs cycle is a pathway that consistently surfaces in ME/CFS metabolomics analysis across platforms and populations. Because fatigue is a major debilitating symptom of this disease, it has long been speculated that the energy metabolism of patients is dysfunctional. Several studies directly point to abnormal energy metabolism due to flawed TCA and urea cycles or directly upstream with putative impairment in pyruvate dehydrogenase [18,20]. A pilot study using a patented nutraceutical treatment hypothesized to boost the activity of this enzyme, and consequently the Krebs cycle, describes substantial improvements to the health and condition of treated patients . Nevertheless, alpha-ketoglutarate is involved in numerous metabolic pathways such as carnitine metabolism, lysine metabolism and branched-chain amino acids, to name a few, so that a focus on a single pathway as the foundation of the disabling symptoms of ME/CFS is presently unjustified.
The “Cofactors and Vitamins” category encompasses metabolites with disparate properties, as exemplified by heme and gamma-CEHC. Higher levels of heme, part of the “Hemoglobin and Porphyrin metabolism”, and lower levels of gamma-CEHC, part of the “Tocopherol metabolism”, were measured in ME/CFS patients compared to controls in our cohort (Figure 1). Heme is a vital component of many metalloproteins, the most well-known being hemoglobin, and is synthesized in the liver and the bone marrow. As the Metabolon® sample preparation is methanol-based, protein precipitation is expected even though protein-bound heme could still be released depending on the level of heme coordination. Because we used plasma, which is a cell-free matrix, it is anticipated that there is a greater contribution from “free heme” to the measurement of heme abundance unless substantial hemolysis occurred. High concentration of free heme in plasma is a biomarker for sickle cell disease severity, in which increased levels of inflammatory biomarkers such as lactate dehydrogenase, bilirubin, high reticulocytes count, and lipids are detected . "
This study includes a ~~replication~~ [EDIT: striek that. on closer reading this is more of a statistical comparison than a replication] of Naviaux and it ~~replicates~~ a lot, but, frustratingly, not the adenosine finding!
"The next comparison was between the complete Metabolon® dataset, and the female-gender dataset from Naviaux et al. . Out of the 154 metabolites with common HMDB identities, only two, adenosine and flavin adenine dinucleotide (FAD), behave statistically differently between the two studies (Table 4). This difference could be due to the collection method of the plasma, which was done in EDTA for our samples while lithium-heparin tubes were used by Naviaux et al. ."
overall they find the previous studies mostly ~~replicate~~ compare well
"The final combinations included Germain et al.  and Armstrong et al. , Germain et al.  and Naviaux et al. , and Armstrong et al.  and Naviaux et al. , and the results were very similar, with 87%, 98% and 91% of metabolites, respectively, not behaving statistically differently between studies (Table 4)."
But note that the statistically significant findings of the previous studies may be in the remaining part that did not replicate, as with adenosine.
A lot of studies get published in this part of the forum but this one is more important than many. It is a major paper featuring some very good science, a huge array of metabolites under consideration, and some interesting findings.
It is focused on something called redox - I think that's a term we're going to start seeing a lot more! (so if someone could explain it that would be great!)
Biological energy is frequently stored and released by means of redox reactions.
Photosynthesis involves the reduction of carbon dioxide into sugars and the oxidation of water into molecular oxygen. The reverse reaction, respiration, oxidizes sugars to produce carbon dioxide and water. As intermediate steps, the reduced carbon compounds are used to reduce nicotinamide adenine dinucleotide (NAD+) to NADH, which then contributes to the creation of a proton gradient, which drives the synthesis of adenosine triphosphate (ATP) and is maintained by the reduction of oxygen. In animal cells, mitochondria perform similar functions.
Free radical reactions are redox reactions that occur as a part of homeostasis and killing microorganisms, where an electron detaches from a molecule and then reattaches almost instantaneously. Free radicals are a part of redox molecules and can become harmful to the human body if they do not reattach to the redox molecule or an antioxidant. Unsatisfied free radicals can spur the mutation of cells they encounter and are, thus, causes of cancer.
Blood samples were collected in NYC by ME/CFS expert physician Dr. Susan Levine in EDTA
tubes and shipped overnight by FedEx in a Styrofoam box to Cornell University-Ithaca Biotechnology
Building, where plasma was separated from cells by centrifugation at 500 g for 30 min, before being
stored at 80 C until further analysis.
This extended processing time probably means any metabolites with a short half life will be difficult to measure. I keep hoping that one group will detect increased mast cell mediators but they have a very short half life. Blood needs to be immediately processed and frozen to detect those. The unknown is the unfrozen handling time at Metabolon so maybe they can't be measured anyway.
Would be nice to see some hard evidence for statements I've heard such as "up to 50% of ME/CFS patients have MCAS"
A snippet to feed @mariovitali's confirmation bias for his liver theory.
"Finally, gamma-glutamyl-threonine, a dipeptide part of the “Peptides” class, was detected at significantly higher levels in the blood of patients compared to controls (Figure 1i and Table 2). Apart from the fact that this compound is an intermediate breakdown product of protein degradation very little is known about its physiological effect in blood. It is, however, cited in a patent as a potential biomarker for liver toxicity determination ."
Next snippet (I'm writing these as I go through the paper):
They are able to compare patients metabolites to patterns in known diseases. They find a few, many of which have something to do with redox.
The best statistical significance comes from something called Carnitine Palmitoyl Transferase Deficiency II
Carnitine palmitoyltransferase II (CPT II) deficiency is a condition that prevents the body from using certain fats for energy, particularly during periods without food (fasting).
There are three kinds, two of which are pretty much fatal . The third kind has some intriguing aspects....
The myopathic form is the least severe type of CPT II deficiency. This form is characterized by recurrent episodes of muscle pain (myalgia) and weakness and is associated with the breakdown of muscle tissue (rhabdomyolysis). The destruction of muscle tissue releases a protein called myoglobin, which is processed by the kidneys and released in the urine (myoglobinuria). Myoglobin causes the urine to be red or brown. This protein can also damage the kidneys, in some cases leading to life-threatening kidney failure. Episodes of myalgia and rhabdomyolysis may be triggered by exercise, stress, exposure to extreme temperatures, infections, or fasting. The first episode usually occurs during childhood or adolescence. Most people with the myopathic form ofCPT II deficiencyhave no signs or symptoms of the disorder between episodes.
They looked at 832 metabolites. Of course some look significantly different between groups simply by chance when you consider so many. They applied a multiple comparison correction to the whole dataset ... and it left them with nothing significant. (those corrections are brutal in terms of turning findings insignificant). So they then applied that multiple comparison correction to the various metabolite groups ("super pathways"). After doing that they had 14 metabolites to look at.
It's not ideal but they make it very clear so we know what we're looking at.
"Undoubtedly, the statistical power of our analysis was weakened by the combination of a 51-subject population and an 832-metabolites array, explaining the limited number of metabolites we establish as significantly different in Table 2 and Figure 1, and only after super-pathway subgrouping."
What lends some hope is the similarities between the metabolites that suggests they are not random
If we consider that metabolites showing up could be the result of methodology differences, it could be that both Adenosine, IMP and cAMP may be low.
All of them have related metabolism.
ATP produces cAMP, that produces AMP that produces Adenosine and IMP that both produces Inosine.
I bet Inosine is not showing as a low metabolite because of some methodology issue .
The key problems could be either a lack of intracellular ATP to produce cAMP, either an adenylyl cyclase inhibition
Some figures to show how these métabolites are interconnected:
Figure 1 The Inosine metabolism:
"Figure 2: Model of compartmentalized cAMP signaling in neurons. (1) Electrical activity and calcium entry activate soluble adenylyl cyclase (sAC), causing an increase in intracellular cyclic adenosine monophosphate (cAMP) and activation of tetrameric protein kinase A (PKA) via dissociation of the regulatory (R) and catalytic subunits (C). (2) PKA regulates its own activity by phosphorylating and activating phosphodiesterase 4 (PDE4) in a negative feedback loop controlling cAMP. A-kinase anchoring protein scaffolds (such as AKAP6) confer specificity to cAMP signals by facilitating PKA-mediated phosphorylation of downstream effectors such as extracellular signal-regulated kinases (ERKs) and nuclear factor of activated T-cells (NFAT) transcription factors that promote survival and growth signaling. (3) Compartmentalized cAMP-synthesis (mediated by sAC) can also specify pro-survival signaling in the nucleus by directly activating cAMP response element-binding protein (CREB)-mediated transcription. (4) Similarly, mitochondrial-associated sAC is well positioned to respond to metabolic changes in the cell due to its sensitivity to bicarbonate (HCO3) that is converted from CO2 and H2O generated during respiration."
"Pathways of adenosine metabolism. Adenosine is formed either by hydrolysis of AMP or by hydrolysis of SAH, which arises from the action of methyltransferases. Adenosine can be metabolized by ADA or ADK into inosine and AMP, respectively. Note the AMP ͞ adenosine futile cycle, which is catalyzed by ADK and 5 Ј -nucleotidase. "
SAM ϭ S -adenosylmethionine;
SAH ϭ S -adenosylhomocysteine;
X ϭ methyl-group acceptor;
X-CH 3 ϭ methylated
You may be right on Inosine. Here's a chart of raw data for Inosine for this study. What you notice is many patients have the exact same low level. Where levels were below the threshold of detection, they "impute" them. Lots of controls and patients were below the threshold. It's not clear why Metabolon include this metabolite in their matrix if they mostly can't measure it!
Because of this its hard to know the true levels of Inosine
However adenosine does not show this pattern and patient have higher levels. Patients are higher even if you exclude p1, who is an outlier.
Inosine is an endogenous purine nucleoside that is produced by catabolism of adenosine. Adenosine has a short half-life (approximately 10 s) and is rapidly deaminated to inosine, a stable metabolite with a half-life of approximately 15 h.
So cAMP is meant to be statistically different between controls and patients. However when you plot out the data and give it the eye test you can see there really is not a lot of difference. In my humble non scientific opinion I would prefer a much larger sample size to say for sure that cAMP is statistically different. Hopefully this is where Maureen Hansons NIH grant comes in.
@Murph Were you able to find Figure S1 and Figure S2?
Sorting through the raw data quickly in Excel I thought the data for tauroursodeoxycholate looked significant.
Then I found this in the paper
The volcano plot tool combines fold changes and non-parametric testing for an alternative exploration of the data in order to bring in some biological significance to statistical analysis. A total of 7 metabolites stood out when using a fold change threshold of two and raw p < 0.05, while assuming an unequal group variance. Out of the 7, two overlap with the statistically significant metabolites described in the above section, namely heme and IMP.
All the other metabolites identified in the volcano plot had higher abundances in patients compared to controls (Figure S1). These included tauroursodeoxycholate (TUDCA), reported both as a cytoprotective agent and a chemical chaperone; 3-hydroxybutyrylcarnitine 1 and 3-hydroxybutyrate(BHBA), both involved in ketosis, a metabolic process associated with energy and glucose; piperine, an alkaloid found in herbs and spices; and histamine, a compound known to be involved in many aspects of the human body, including local immune responses, acting as a central neurotransmitter and a vasodilator to name a few.
Dug a bit deeper into histamine. Looked at the raw data with Excel. Seems to be that there is a correlation between histamine and perfluorooctanesulfonic acid (PFOS) I removed controls with histamine >200K and removed patients with histamine <200K. This is what I get. Thoughts?
Perfluorooctanesulfonic acid (conjugate baseperfluorooctanesulfonate) (PFOS) is an anthropogenic fluorosurfactant and global pollutant. PFOS was the key ingredient in Scotchgard, a fabric protector made by 3M, and numerous stain repellents. It was added to Annex B of the Stockholm Convention on Persistent Organic Pollutants in May 2009. PFOS can be synthesized in industrial production or result from the degradation of precursors. PFOS levels that have been detected in wildlife are considered high enough to affect health parameters, and recently higher serum levels of PFOS were found to be associated with increased risk of chronic kidney disease in the general US population. "This association was independent of confounders such as age, sex, race/ethnicity, body mass index, diabetes, hypertension, and serum cholesterol level."[