Cortisol levels in ME/CFS

[Hutan]

Low cortisol or otherwise abnormal cortisol comes up frequently in the ME/CFS literature. Many researchers and clinician have been convinced that ME/CFS is the result of managing stress poorly and that it's all about a HPA axis (Hypothalamus, Pituitary and Adrenal Gland axis) problem. They have therefore repeatedly looked for abnormalities in cortisol levels.

However, if you look carefully at the studies that are supposed to support this contention, they tell a different story.

An early study claiming to show low cortisol in ME/CFS had an extremely biased selection. Mean levels found in studies have been variable - a tendency to lower but still normal levels in ME/CFS is easily explainable by the lower activity levels and relatively undemanding lives people with ME/CFS tend to lead. We know that people who routinely undertake physically demanding exercise have higher levels of cortisol - few people do that once they develop ME/CFS.

There's a similar story with peak morning cortisol. If people don't have to get up at 6.30 am, rush around getting the family organised, getting out the door and navigating rush hour traffic, then they don't have the same need for that peak morning cortisol as a person with ME/CFS who might be waking at 9 am and then doing a bit of online work from the couch.

We have also seen a post-Covid study where steroids taken for symptoms following severe Covid-19 infections probably suppressed cortisol levels.

So, I thought I'd do a bit of a casual review, a survey of some of the relevant studies, with links to the forum discussion about them. Edit - if you want to post about a study or make some comments about what I have posted , go right ahead.

 

[Hutan]

Early studies
Melatonin Levels in Women with Fibromyalgia and Chronic Fatigue Syndrome, 1999, Korszun et al
This was a tiny study, but it found no evidence of abnormal levels of cortisol (or melatonin) in a small sample of women with CFS.


Small adrenal glands in chronic fatigue syndrome: a preliminary computer tomography study, 1999, Scott et al
This one is really incredible. As I said on that thread:
"So, they screened people in a fatigue clinic for adrenal gland functioning (using the ACTH test). They don't say how many people they screened, but they do say that they only found 8 people with faulty adrenal gland functioning.

Comparing the size of the adrenal glands of these 8 subjects with 55 heathy controls, they found that the selected CFS subjects had smaller adrenal glands.

Of course, this tells us nothing about people with CFS with normal ACTH test results. Possibly these 8 people don't even have CFS - perhaps they have some other medical issue. Given there is no good evidence for people with ME/CFS having, on average, abnormal ACTH or cortisol levels, I don't think this study contributes to knowledge about ME/CFS.

And yet, it has recently been cited as evidence for an HPA axis-based theory of ME/CFS etiology."

 

[Hutan]

Early studies
Salivary cortisol output before and after cognitive behavioural therapy for chronic fatigue syndrome, 2008, Roberts
That was a study I had not looked at until now. I'm constantly amazed at how bad these studies are. Wesseley and Chalder were co-authors.

The hypothesis was 'we know that people with CFS have low cortisol, we know that CBT fixes CFS, so we should see an increase in cortisol levels after CBT'.

Actually, the levels of cortisol in the participants were normal, before and after CBT - a fact that was not acknowledged in the paper.

Actually, even with the 'mark your own homework' approach of therapists assessing the improvement of people after the CBT intervention, a significant proportion of people were assessed as unchanged or worse after CBT.

And actually, the 'responders' to CBT had a smaller increase in cortisol levels than the non-responders.

This paper is therefore evidence of normal cortisol values in people with (ME)/CFS and evidence of changes in cortisol levels within the normal range being irrelevant to ME/CFS severity.

 

[Hutan]

Review of early studies
The neuroendocrinology of chronic fatigue syndrome, 2003, Cleare
Cleare, who was a coauthor of the Roberts paper above, had earlier done this review of the evidence relating to HPA axis dysfunction in CFS.

Cleare notes:

Summarizing these results, about half of the studies found some evidence for lowered cortisol levels at some point in the day.

Most studied has been the hypothalamic-pituitary-adrenal (HPA) axis; although the quality of many studies is poor, the overall balance of evidence points to reduced cortisol output in at least some patients, with some evidence that this is linked to symptom production or persistence.

He must have been disappointed with the outcome of the 2008 Roberts study (mentioned in the post above), which showed that changing the cortisol output did not change symptoms.

There's a paywall on the 2003 review, but I expect, based on what I have seen of later studies that most of the included studies did not find levels of cortisol outside the normal range for the majority of CFS participants.

We might have to come back to this review, after looking at each of the studies.

 

[Hutan]

Cortisol in hair
Hair and salivary cortisol in a cohort of women with chronic fatigue syndrome, 2018, Roerink et al
This more recent one is from Knoop's team
People with CFS and the healthy controls didn't have statistically different cortisol levels in saliva throughout the day or in hair compared to controls. However, that did not stop the authors from noting in the Highlights 'Hypocortisolism has been described in CFS', which is sort of true, people have said it, but their study did not support that idea.

But, but, they know there must be something to find...

In addition, the interaction between the HPA axis and the immune system, both expected to play an important role in CFS, is unclear.

All they found was a difference in the cortisol awakening response

People with CFS did have a statistically lower 'cortisol awakening response' (4.2 nmol/L vs 6.3 mol/L for the controls).

If you look at the standard deviations given with each average, you can see the very high variation.
(i.e. 4.2 + or - 5.4; 6.1 + or -6.3)
There is a huge overlap in values - some people with CFS had higher values, some healthy controls had lower values. So, even though statistically significant, it is not a very impressive difference.

As I've noted, differences in awakening cortisol are to be expected, based on different lifestyle demands on the ME/CFS and control cohorts including normal wakening times. It has also been found that people in pain have a lower awakening cortisol.

I'm going to copy the chart and the conclusion the authors drew from it here, just because it illustrates how delusional people can be when their ideas about an illness become dogma:

"This study confirms the altered dynamics of the HPA axis in a group of CFS patients, and for the first time shows that this might also be present for long-term cortisol measures." (i.e. in hair)

Here the authors are summing up how that result in Figure 2 above is so important. HCC is the hair cortisol.

In addition, there was a trend to lower HCC in patients (1). Especially the latter has important implications for future studies. When lower HCC can be confirmed in other studies, it might be used as a more stable and reliable tool for the evaluation of the HPA axis in CFS (2). Earlier studies found that urinary cortisol is able to predict response to CBT , and that salivary cortisol normalizes with adequate treatment (3) . However, in the light of the previously mentioned limitations of salivary cortisol, HCC might be an interesting alternative to predict the effect of treatment.

1. the hair cortisol of the people with CFS was not significantly different to that of healthy controls - see Figure 2. What's the point of reporting p values if you are going to ignore them?
2. the anakinra trial failed to show a faulty HPA axis [the anakinra trial was part of this study]
3. Salivary cortisol was normal without treatment in this study

I find myself amazed again on reading the things these authors wrote about their null result. The very slight difference in hair cortisol means between the CFS and healthy controls can be explained by different lifestyles. I would not be surprised if some of the healthy controls were hyper-healthy - we know that people who exercise a lot tend to have a higher level of cortisol. But, look at the
overlap in the chart! Hair cortisol level has no discriminative power.

 

[Hutan]

Japanese study of yoga
The longitudinal effects of seated isometric yoga on blood biomarkers, autonomic functions, & psychological parameters of [PwCFS], 2019, Oka et al

This study measured cortisol levels in blood before and after 2 months of seated yoga (classes and practiced at home, for an average of 5.8 sessions per week). 15 participants, meeting multiple ME/CFS criteria.

Cortisol (μg/dL) reference range 3–20

Before 11.5 ± 8.3
After 10.4 ± 6.3
P value 0.552

So, the participants had very normal cortisol before and after the intervention. There was no significant change in cortisol after the yoga.

I don't think the Japanese researchers had got the BPS memo about needing to find low levels of cortisol in people with ME/CFS, and that increasing cortisol levels is the aim of exercise interventions. The Japanese researchers noted approvingly that cortisol dropped when measured at the beginning of a yoga session and then at the end of a yoga session.

 

[Hutan]

Characterization of Cortisol Dysregulation in Fibromyalgia and Chronic Fatigue Syndromes: A State-Space Approach, 2020, Pednekar et al
This one is a thoroughly awful paper, it's hard to say what they found. There does not look to be any differences in cortisol between CFS and healthy control cohorts. That does not stop them suggesting that more studies on cortisol will produce the answers to CFS.

Characterizing CFS and FMS based on the cortisol alteration will help us to develop novel methods for treating these disorders.

 

[Hutan]

An Etiological Model for Myalgic Encephalomyelitis/Chronic Fatigue Syndrome, 2011, Jason et al
This paper isn't a study, but I just wanted to flag it to show that even people that we might think understand ME/CFS quite well can have some whacky ideas built on this false idea that cortisol is central to the pathology of ME/CFS.

Once this system is charged or kindled, it can sustain a high level of arousal with little or no external stimulus and eventually this could lead to hypocortisolism. Seizure activity may spread to adjacent structures of the limbic-hypothalamic-pituitary axis in the brain, which might be responsible for the varied symptoms that occur among patients with ME/CFS. In addition, kindling may also be responsible for high levels of oxidative stress, which has been found in patients with ME/CFS.

Also noting @Snow Leopard's point about day time napping possibly confounding morning cortisol levels

Some people latch on the finding of low salivary cortisol in mornings, but those people don't understand that there are confounding factors - patients may think they're being careful and complying with the study by actually getting up at a regular time in the morning and then testing (as opposed to sleeping in to their usual waking time as they usually do). This alone can cause the effect.

Napping during the day can also alter the daily cortisol response, without having any underlying pathology.

Edited to add:
Determinants of cortisol awakening responses to naps and nighttime sleep
which makes it clear that the morning peak is an expression of the Cortisol Awakening Response, which is a response to waking up, rather than the time of day. People who nap will have a different daily architecture of cortisol production. Loss of sleep during the night also affects cortisol patterns.

 

[Hutan]

Hydrocortisone studies

Low-dose hydrocortisone in chronic fatigue syndrome: a randomised crossover trial, 1999, Cleare, Wessely et al

Mean urinary free cortisol at baseline was 105 (84–127; reference range for laboratory 30–250 nmol per 24 h). The mean value after hydrocortisone was 141 (100–181) nmol per 24 h and was 97 (76–119) nmol per 24 h after placebo (95% CI for difference 14 to 74, p=0·006).

32 participants. The participants' urinary free cortisol levels were normal throughout, although did increase with hydrocortisone.


Low-Dose Hydrocortisone for Treatment of Chronic Fatigue Syndrome - A Randomized Controlled Trial, 1998, McKenzie et al
70 people.

On the HPA axis hypothesis: (the cortisol levels in this sample were entirely normal - both morning cortisol and after ACTH. The values are given but no comment is made about their normality)

Fourth, the data bear importantly on the basic hypothesis under which this study was undertaken, namely that CFS symptoms are perpetuated through suboptimal activity of the HPA axis.15That the basal and stimulated-cortisol levels did not correlate with illness severity in this study (data not shown), nor were they predictive of clinical improvement or response to treatment, argue against the hypothesis.

Positive responses to hydrocortisone treatment were limited to one questionnaire among quite a number, were small and were in line with the positive response that healthy people also experience.

 

[Sasha]

Early studies
Salivary cortisol output before and after cognitive behavioural therapy for chronic fatigue syndrome, 2008, Roberts

[...]
This paper is therefore evidence of normal cortisol values in people with (ME)/CFS and evidence of changes in cortisol levels within the normal range being irrelevant to ME/CFS severity.

It never fails to amaze me how many research studies by BPS proponents, when conventionally interpreted, show the exact opposite of what the author claim, PACE being a classic.

 

[Creekside]

Good review. I've seen so many people excited about some study, ignoring the many other studies that didn't support the findings.

 

[rvallee]

Somehow even though much the same has happened in Long Covid, including garbage quality and odd misinterpretation, the idea is just as much alive there as well. Once a conclusion has been accepted without evidence, no amount of evidence will change trust in the conclusion.

It seems to work just like compromising someone with a small bribe. At this point you can squeeze them further by threatening to rat them out on the small bribe. Then you can move to make them do things they wouldn't do, and threaten them further.

They have a lot of theories, but not a lot of evidence. However they love their theories far more than they love evidence.
- The Aristocracts Evidence-based medicine

 

[Hutan]

Long covid studies
Distinguishing features of Long COVID identified through immune profiling, 2022, Klein, Iwasaki et al
This Iwasaki paper has been very influential in promoting the idea of low cortisol in people with Long Covid.
A large study

Analysis of circulating immune mediators and various hormones also revealed pronounced differences, with levels of cortisol being uniformly lower among participants with Long COVID relative to matched control groups. Integration of immune phenotyping data into unbiased machine learning models identified significant distinguishing features critical in accurate classification of Long COVID, with decreased levels of cortisol being the most significant individual predictor.

The cortisol difference is pretty large though and they looked for an association with time of sample collection and found none. I think you'd expect more overlap if this were due to sleep patterns or morning activity levels. Could be nothing, but I wouldn't be so quick to dismiss cortisol, especially with 101 patients.

Noting @cassava's excellent post on the Iwasaki thread - some references to papers to include.
Also my post in that thread about a paper that monitored the cortisol levels of Polish rowers undergoing training. Cortisol levels change, within weeks, with baseline activity levels.
And also my posting about the possibility that steroid medications during the acute illness and subsequently have supressed endogenous cortisol production

I think the problem is that "Long covid" is such a mix of conditions. Finding the "distinguishing features of heterogeneous conditions - twitchy lungs and lingering coughs; ME/CFS; heart issues; anosmia; and more - isn't really possible.

So I think what has happened instead is that at least two things are favouring the lower average cortisol. One is the issue I've already posted about - people with post-covid ME/CFS have, on average, a lower activity than the controls, so their bodies have adjusted the level of cortisol down a bit. The other issue I think might be at work here is that some of the people with lingering symptoms are taking various corticosteroids which is suppressing their endogenous cortisol. There might also be a residual effect from steroid treatments during the acute covid-19 infection for people who had severe acute illnesses - these people are probably more likely to still have symptoms.

 

[Snow Leopard]

Edited to add:
Determinants of cortisol awakening responses to naps and nighttime sleep
which makes it clear that the morning peak is an expression of the Cortisol Awakening Response, which is a response to waking up, rather than the time of day. People who nap will have a different daily architecture of cortisol production. Loss of sleep during the night also affects cortisol patterns.

The key point is that cortisol is a feed-forward metabolic hormone. Salivary cortisol in particular adjusts to your daily sleep-wake cycle and morning activity patterns. So waking up at a different time from normal will mean a different level when awakening. Similarly, someone retired who doesn't do much in the morning will have a different (flatter) pattern to someone who gets up and has an active work life in the morning.

The claim about time of collection is misleading as they are not measuring the key factor - time between usual wake time and time of collection. Time from midnight is not the variable we wish to control for! And in fact if time from midnight is the same between groups, but the usual wake time of patients is different from controls, then this indicates a confounding factor!

 

[Hutan]

A German study of post-covid syndrome with a good sized sample of nearly 1000 people with PCS found no differences in (presumably serum) cortisol or ACTH levels. 11.6% of people were assessed as meeting ME/CFS criteria (I'm not sure which one)
Persistent symptoms and clinical findings in adults with post-acute sequelae of COVID-19/post-COVID-19 syndrome in the second year....[], 2025, Peter

 

[SNT Gatchaman]

Stellate Ganglion Block reduces symptoms of SARS-CoV-2-induced ME/CFS: A prospective cohort pilot study (2025, Fatigue: Biomedicine, Health & Behavior)

This did show low morning cortisol (unchanged in short term follow-up after a series of stellate ganglion blockades, for LC-ME/CFS).

[Saliva samples] were obtained within 30 minutes of awakening, before brushing teeth or eating, and at least ten minutes after rinsing with water. Subjects agreed to refrain from alcohol consumption the day before collection. Samples were refrigerated, then brought to the clinic on ice. Cartridges were centrifuged 2 minutes at 1000 x g at 4 ⁰C. Filtrate was transferred to cryovials in 100 µl aliquots and stored at -80 ⁰C until use. The Invitrogen cortisol ELISA kit was used as per protocol.

Morning salivary cortisol levels for all subjects were well below the normal threshold of 10 ng/ mL at baseline (3.9 ng/ml +/− 1.4), and no significant changes were detected after treatment at either short – or long-term follow-up (4.4 ng/ml +/− 1.7 and 3.9 ng/ml +/−1.2)

 

[Snow Leopard]

Stellate Ganglion Block reduces symptoms of SARS-CoV-2-induced ME/CFS: A prospective cohort pilot study (2025, Fatigue: Biomedicine, Health & Behavior)

This did show low morning cortisol (unchanged in short term follow-up after a series of stellate ganglion blockades, for LC-ME/CFS).

Morning cortisol is confounded when it is collected at a different amount of time after (usual) wake time for patients versus controls.

Patients often change their wake times during a study whereas healthy controls generally don't. Especially one that involves sleep monitoring!

 

[Hutan]

This did show low morning cortisol (unchanged in short term follow-up after a series of stellate ganglion blockades, for LC-ME/CFS).

The researchers screened 40 consecutive patients and only chose 10 for the study. It isn't all that clear why someone made the cut for the study. There's a lot of potential for bias there, and the researchers very easily could have included a requirement for the baseline to show low morning cortisol. The study is small and lacks credibility.

 

[AliceLily]

I did the 24 hour urine cortisol test in my first year of very severe ME. I remember there being very specific instructions which I followed. They didn't find anything wrong. My ME symptoms were at their worst too.

 

[SNT Gatchaman]

Immunological and clinical markers of post-acute sequelae of COVID-19: Insights from mild and severe cases six months post-infection (2025, Preprint: BioRxiv) —

no significant difference in cortisol levels was found between patients with and without PASC