Circulating miRNAs Expression in Myalgic Encephalomyelitis/Chronic Fatigue Syndrome 2023 Soffritti et al

[John Mac]

Abstract:
Myalgic encephalomyelitis/chronic fatigue syndrome (ME/CFS) is a complex multifactorial disease that causes increasing morbidity worldwide, and many individuals with ME/CFS symptoms remain undiagnosed due to the lack of diagnostic biomarkers.

Its etiology is still unknown, but increasing evidence supports a role of herpesviruses (including HHV-6A and HHV-6B) as potential triggers. Interestingly, the infection by these viruses has been reported to impact the expression of microRNAs (miRNAs), short non-coding RNA sequences which have been suggested to be epigenetic factors modulating ME/CFS pathogenic mechanisms.

Notably, the presence of circulating miRNAs in plasma has raised the possibility to use them as valuable biomarkers for distinguishing ME/CFS patients from healthy controls.

Thus, this study aimed at determining the role of eight miRNAs, which were selected for their previous association with ME/CFS, as potential circulating biomarkers of the disease.

Their presence was quantitatively evaluated in plasma from 40 ME/CFS patients and 20 healthy controls by specific Taqman assays, and the results showed that six out of the eight of the selected miRNAs were differently expressed in patients compared to controls; more specifically, five miRNAs were significantly upregulated (miR-127-3p, miR-142-5p, miR-143-3p, miR-150-5p, and miR-448), and one was downmodulated (miR-140-5p).

MiRNA levels directly correlated with disease severity, whereas no significant correlations were observed with the plasma levels of seven pro-inflammatory cytokines or with the presence/load of HHV-6A/6B genome, as judged by specific PCR amplification.

The results may open the way for further validation of miRNAs as new potential biomarkers in ME/CFS and increase the knowledge of the complex pathways involved in the ME/CFS development.

ETA: Academic Editor: Vincent C. Lombardi

pdf
https://mdpi-res.com › ijms › ijms-24-10582

 

[Andy]

Online version, https://www.mdpi.com/1422-0067/24/13/10582

Fukuda criteria.

 

[Hutan]

Research from Italy and Latvia

They note recent research suggesting specific microRNAs might be different in ME/CFS:

More recently, growing interest has been given to microRNAs (miRNAs) as potential biomarkers in ME/CFS; however, so far, no miRNA has been validated for clinical diagnosis [23,24,25]. MiRNAs are short sequences (18–23 nucleotides) of non-coding RNA with essential roles in regulating gene expression at the post-transcriptional level, which is suggested to have a role in many pathological pathways. Specifically, circulating miR-124, miR-448, and miR-551b have been found to be differentially expressed in patients with rheumatoid arthritis (RA), systemic lupus erythematosus (SLE), Sjögren’s syndrome (SS), and ulcerative colitis, with respect to healthy controls, and thus have been suggested as biomarkers for autoimmune diseases [26]. High-throughput miRNome sequencing of plasma from ME/CFS subjects evidenced differential expression of miR-127-3p, miR-142-5p, and miR-143-3p compared to non-CFS controls [24]. In addition, the upregulation of miR-140-5p and miR-150-5p expression has been reported both in plasma and peripheral blood mononuclear cells (PBMCs) of ME/CFS subjects compared to healthy controls or associated with ME/CFS response to post-exertional malaise induction [23,25,27], and an influence of the nutritional status and gender of patients has been observed [27]. The main findings regarding the mentioned miRNAs and their association and role in ME/CFS disease are summarized in Table 1.

They therefore set out to examine specific miRNA.

Based on these observations, the aim of the present study was to determine the potential role of autoimmunity-associated miRNAs as biomarkers of ME/CFS. To this purpose, circulating miR-124-3p, miR-127-3p, miR-140-5p, miR-142-5p, miR-143-3p, miR-150-5p, miR-448, and miR-551b-3p were analyzed in plasma from 40 ME/CFS patients and 20 healthy controls (CTRs). Correlations were searched between miRNAs’ expression and disease severity, plasma pro-inflammatory cytokines, and HHV-6 infection/reactivation.

This therefore is a replication study, and a step along from the exploratory analyses we have seen with thousands of candidate molecules.

 

[Hutan]

40 ME/CFS - clinically diagnosed from an out patient clinic
20 healthy controls

Pro-inflammatory cytokines
IFN-y - not different
IL-17A - a lot lower in ME/CFS p=0.0003
Il-2 - a lot lower in ME/CFS p=0.0001
IL21 - a lot lower in ME/CFS p=0.001
IL23 - not different, but the numbers look weird. Controls are 1.6 +-235; ME/CFS are 407+-115 (means +-SE)
IL6 - a lot lower in ME/CFS p=0.0270
TNF-a - a lot lower in ME/CFS p=0.005

The results look a bit weird to me; I think it's possible that some processing issue resulted in the much lower results in ME/CFS. Certainly, there's nothing there to suggest that ME/CFS is an inflammatory disease.


HHV6A/B presence

a total of 9/40 patients (22.5%) resulted in being positive for the presence of HHV-6A/6B
In the control group, only 2/20 subjects (10%) resulted in being positive for viral presence

This is not evidence for HHV6A/B being part of ME/CFS pathology. The results are similar to another recent study looking at the presence of HHV6 genetic material in the blood of people with ME/CFS.

miRNA Plasma Levels in ME/CFS Patients
Of the 8 miRNA, there are some that look a bit different.
mir-124 - not different

mir-127 - not different
authors continued to look at it on the basis of a difference between controls and mild ME/CFS, but there really doesn't look to be a difference

mir-140 - lower in ME/CFS p=0.007
in the text earlier, this was reported as higher in ME/CFS than controls after exercise

mir-142 - higher in ME/CFS p=0.02
Previously reported as increased in ME/CFS in the Brenu 2014 study (see next post) - potentially a replication

mir-143 - not different
authors continued to look at it on the basis of a difference between controls and mild ME/CFS, but there really doesn't look to be a difference

mir-150 - higher in ME/CFS p=0.03
in the text earlier, another study was reported as finding this higher in ME/CFS than controls after exercise - potentially a replication

mir- 448 - higher in ME/CFS p=0.0001
This was one of the proposed autoimmunity markers. It had the clearest difference versus controls

mir-551 - not different

 

[Hutan]

High-Throughput Sequencing of Plasma MicroRNA in CFS/ME, 2014, Brenu, Staines, Marshall-Gradisnik et al
I've made a thread for that 2014 paper that found that miR-127-3p, miR-142-5p, and miR-143-3p were increased relative to controls. It's a small rather unreliable study, but it's interesting to see the replication of the increased mir-142 result.

Among the miRNAs found to be upregulated in ME/CFS patients, miR-142-5p has been reported as being overexpressed in most diseases linked to immunological disorders [46]

Andy previously posted this paper: microRNA-142–mediated repression of phosphodiesterase 3B critically regulates peripheral immune tolerance, 2019, Lord et al relevant to the possible mir-142 up-regulation.

 

[Sly Saint]

article on paper by MERUK
Can microRNAs be used to diagnose ME/CFS?
https://www.meresearch.org.uk/can-micrornas-be-used-to-diagnose-me-cfs/

 

[Hutan]

article on paper by MERUK

That's a good analysis. Here's the Discussion:

A quick glance at the summary table from a recent systematic review analysing potential ME/CFS biomarkers will tell you that an enormous number of microRNAs have been analysed, with minimal overlap between studies. Despite the limitations of this study, it is useful that the researchers chose to study microRNAs that they had noted in previous literature as this gives the opportunity to compare findings.

The three microRNAs increased in the patient group (miR-142, mir-150 and mir-448) were also found to be upregulated in previous data. However, they also mention that there are mixed results between studies regarding the expression of miR-140 and the presence of pro-inflammatory cytokines.

With the help of ME Research UK funding, researchers such as Dr Bhupesh Prusty and Dr Francisco Westermeier have also touched upon the relationship between ME/CFS and microRNAs. However, this is a vast and complex field, and we need larger studies with robust methodology to build upon and validate previous findings. At present it seems that quite a few more steps are needed before microRNAs could be considered a frontrunner in the search for biomarkers.

I don't feel quite so negative towards microRNAs as potential biomarkers though. I'd like to see more work done on them in studies with carefully selected patients (i.e. not using a Fukuda criteria). The technology is there, so some more work on this seems to me to be a worthwhile thing to do.

The analysis did not comment on the possible issues with the pro-inflammatory cytokines, other than to note that the authors mention that the use of NSAIDS might have affected the findings.

 

[ME/CFS Skeptic]

The largest difference was found for miR-448 which had not been tested in ME/CFS before but it was found to be increased in various auto-immune diseases (the article mentions RA, SLE, SS, UC). Might be something to look out for in other studies.

 

[SNT Gatchaman]

The largest difference was found for miR-448 which had not been tested in ME/CFS before but it was found to be increased in various auto-immune diseases (the article mentions RA, SLE, SS, UC)

The precursor miR-448 gene is on the X chromosome. It's upregulated by interferon. It suppresses IDO1 in CD8+ T cells (at least in the colorectal cancer tumour microenvironment, from the abstract below).

miR-448 upregulation -> IDO1 downregulation -> decreased CD8+ T cell apoptosis -> more cytotoxic T cells.

I.e. miR-448 (via ↓ IDO1) tips toward less cancer and potentially more autoimmunity.

---
miR-448 targets IDO1 and regulates CD8+ T cell response in human colon cancer (2019)
Lou, Qiong; Liu, Ruixian; Yang, Xiangling; Li, Weiqian; Huang, Lanlan; Wei, Lili; Tan, Huiliu; Xiang, Nanlin; Chan, Kawo; Chen, Junxiong; Liu, Huanliang

Indoleamine 2,3-dioxygenase 1 (IDO1) is a critical regulator of T cell function, contributing to immune tolerance. Upregulation of IDO1 has been found in many cancer types; however, the regulatory mechanisms and clinical significance of IDO1 in colon cancer are still unclear.

Here, we investigated the role of dysregulated microRNA (miRNA) targeting IDO1 in the colon cancer microenvironment. We elucidated IDO1 function by performing cell-based assays and establishing transplanted tumor models in BALB/c mice and BALB/c nude mice. We evaluated IDO1 protein expression by immunohistochemistry (IHC) in a tissue microarray (TMA) and analyzed IDO1 mRNA expression with The Cancer Genome Atlas (TCGA). We screened miRNAs targeting IDO1 by using a dual luciferase reporter assay. We tested the function of microRNA-448 (miR-448) by using western blotting (WB) and fluorescence-activated cell sorting (FACS).

We demonstrated that stable IDO1 overexpression enhanced xenograft tumor growth in BALB/c mice but not in BALB/c nude mice. We also revealed the involvement of posttranscriptional regulation of IDO1 in colon cancer by observing IDO1 protein levels and mRNA levels. Furthermore, ectopic expression of miRNA mimics suggested that miR-448 could significantly downregulate IDO1 protein expression. Notably, we proved that miR-448 suppressed the apoptosis of CD8+ T cells by suppressing IDO1 enzyme function.

Our findings indicated that IDO1 suppressed the CD8+ T cell response in colon cancer. miR-448, as a tumor-suppressive miRNA, enhanced the CD8+ T cell response by inhibiting IDO1 expression. The results provide a theoretical basis for the development of new immunotherapy for the treatment of colon cancer.

Link | PDF (Journal for ImmunoTherapy of Cancer)

---
Some other papers that may relate, which I haven't read yet are —

Interactions of IDO and the Kynurenine Pathway with Cell Transduction Systems and Metabolism at the Inflammation–Cancer Interface (2023, Cancers)

Metabolic reprogramming by miRNAs in the tumor microenvironment: Focused on immunometabolism (2022, Frontiers in Oncology)

miR-aculous new avenues for cancer immunotherapy (2022, Frontiers in Immunology)

The Regulatory Effects of MicroRNAs on Tumor Immunity (2022, BioMed Research International)

The Role of Indoleamine 2, 3-Dioxygenase 1 in Regulating Tumor Microenvironment (2022, Cancers)

Role of microRNAs in remodeling the tumor microenvironment Review (2020, International Journal of Oncology)

Suppression of TLR4 by miR-448 is involved in Diabetic development via regulating Macrophage polarization (2019, Journal of Pharmacy and Pharmacology)

 

[Hutan]

There's this paper (not open access):
miR-448 regulates potassium voltage-gated channel subfamily A member 4 (KCNA4) in ischemia and heart failure
Results: The expression of KCNA4 is diminished in ischemia and human heart failure tissues with ventricular tachycardia. Previously, we have shown that miR-448 is upregulated in ischemia and inhibition can prevent arrhythmic risk after myocardial infarction. The 3'-untranslated region of KCNA4 has a conserved miR-448 binding site. miR-448 bound to this site directly and reduced KCNA4 expression and the transient outward potassium current. Inhibition of miR-448 restored KCNA4.

Conclusion: These findings showed a link between Kv1.4 downregulation and miR-448-mediated upregulation in ischemia, suggesting a new mechanism for the antiarrhythmic effect of miR-448 inhibition.


It's an interesting link between ischemia and miR-448. That observation that its gene is on the X chromosome is also interesting. I'd really like to see some more replication of the miR studies.

 

[forestglip]

Overall, increased levels of plasmatic miRNAs in the ME/CFS group were more evident by stratifying patients according to severity, showing a positive correlation between abundance of miRNAs and more severe symptoms. This was evident for miR-448, which was overexpressed in all symptoms’ subgroups, as compared to healthy controls, with increasing values correlating with ME/CFS severity, (p < 0.0001, p = 0.001, and p = 0.0015, for Subgroups 1, 2, and 3, respectively).

Seems strange to say it's correlated with severity based on p values, which will be influenced by sample size as well as effect size. And the sample sizes vary:

Severe (1): 5 cases
Moderate (2): 22 cases
Mild (3): 13 cases



Sure, severe having the lowest p value while also having the lowest sample size might mean something. But between mild and moderate, the graph shows that mild is higher than moderate, in the same direction as severe. It has a higher p value because it has far fewer cases.

 

[Utsikt]

Seems strange to say it's correlated with severity based on p values, which will be influenced by sample size as well as effect size. And the sample sizes vary:

Severe (1): 5 cases
Moderate (2): 22 cases
Mild (3): 13 cases

View attachment 25766

Sure, severe having the lowest p value while also having the lowest sample size might mean something. But between mild and moderate, the graph shows that mild is higher than moderate, in the same direction as severe. It has a higher p value because it has far fewer cases.

I wonder what the actual p-value for group 2 is. It could be anything between 0.00050 and 0.00149.

And I agree that using p-values that way is very weird.

 

[jnmaciuch]

It has a higher p value because it has far fewer cases.

P-value is going to take into consideration both mean difference and variance. Meaning that if there is a lot of variability, it's actually much harder to find a statistically significant association with a low sample size. You'd need to substantially increase power by increasing sample size to see a statistically significant association with a small mean difference and large variance.

So a very small p-value in a small sample size indicates that the difference was big enough and the variation was minimal enough that the correlation between variable X and Y are very very likely not due to chance.

P-value does mean correlation regardless of effect size--it really just means that whatever difference there is, regardless of how small or how variable, is likely to be associated with your other variable in a way that is not simply due to chance.

 

[jnmaciuch]

I wonder what the actual p-value for group 2 is. It could be anything between 0.00050 and 0.00149.

And I agree that using p-values that way is very weird.

What do you mean the actual p-value for group 2? It would be p=0.001, no?

 

[Utsikt]

What do you mean the actual p-value for group 2? It would be p=0.001, no?

No, that’s a rounded value because they use more decimals for the other two groups.

 

[forestglip]

P-value is going to take into consideration both mean difference and variance. Meaning that if there is a lot of variability, it's actually much harder to find a statistically significant association with a low sample size. You'd need to substantially increase power by increasing sample size to see a statistically significant association with a small mean difference and large variance.

So a very small p-value in a small sample size indicates that the difference was big enough and the variation was minimal enough that the correlation between variable X and Y are not due to chance.

P-value does mean correlation regardless of effect size--it really just means that whatever difference there is, regardless of how small or how variable, is likely to be associated with your other variable in a way that is not simply due to chance.

I'm not sure if you're agreeing with me. But the variance in the mild and moderate in the chart looks nearly identical. And the median is higher. The only reason I can see that it would have a higher p value is due to its smaller sample size.

They can say severity correlates to how sure they are that the results are not due to chance, but that's not really the same as correlating with miRNA levels. You could have 1000 people in group C with slightly higher levels than group A and 2 people in group B with very high levels. That wouldn't suggest to me that there's a trend of higher levels going from A > B > C based on p values that came out that way based on how many people they recruited.

 

[jnmaciuch]

No, that’s a rounded value because they use more decimals for the other two groups.

Ah I suspect it's 0.0010--many programs will just drop the extra 0 automatically.

 

[Jonathan Edwards]

Seems strange to say it's correlated with severity based on p values, which will be influenced by sample size as well as effect size.

I don't think they were suggesting that the trend in p values represented a correlation. But you are right to say that this is a totally unconvincing presentation of a 'correlation with severity'. There is no correlation with severity because the means produce a v-shaped profile: up down up.

Looking at that my immediate reaction is that it is pretty good evidence for the observation either being spurious or epiphenomenal. It raises more questions than it answers.

 

[Utsikt]

Ah I suspect it's 0.0010--many programs will just drop the extra 0 automatically.

If that’s the case it’s a flaw that should be corrected. The last 0 isn’t extra, it’s just as significant as any other digit would have been.

 

[forestglip]

I don't think they were suggesting that the trend in p values represented a correlation.

That's what it looks like to me:

This was evident for miR-448, which was overexpressed in all symptoms’ subgroups, as compared to healthy controls, with increasing values correlating with ME/CFS severity, (p < 0.0001, p = 0.001, and p = 0.0015, for Subgroups 1, 2, and 3, respectively).

The only values they cite are p values for "increasing values". It can't apply to actual values because of the "V" shape you mention.