Association between fatigue, peripheral serotonin, and L-carnitine in hypothyroidism and in chronic fatigue syndrome

Background Fatigue of unknown origin is a hallmark symptom in chronic fatigue syndrome (CFS) and is also found in 20% of hypothyroidism patients despite appropriate levothyroxine treatment. Here, we suggest that in these disorders, peripheral serotonin levels are low, and elevating them to normal range with L-carnitine is accompanied with reduced fatigue. Methods We conducted a retrospective analysis of follow-up clinical data (CFS N=12; hypothyroidism with fatigue N=40) where serum serotonin and fatigue levels were compared before vs. after 7 weeks of oral L-carnitine supplementation. Results After L-carnitine, serotonin increased (8-fold in CFS, Sig. = 0.002, 6-fold in hypothyroidism, Sig. < 0.001) whereas fatigue decreased (2-fold in both CFS and hypothyroidism, Sig. = 0.002 for CFS, Sig. < 0.001 for hypothyroidism). There was a negative correlation between serotonin level and fatigue (for CFS, rho = -0.49 before and -0.67 after L-carnitine; for hypothyroidism, rho = -0.24 before and -0.83 after L-carnitine). Conclusions These findings suggest a new link between low peripheral serotonin, L-carnitine, and fatigue.


Introduction
Fatigue is a common chronic cross-diagnostic symptom (1)(2)(3)(4).Recently, some emerging evidence has suggested a possible association between low peripheral serotonin and fatigue.On the one hand, low peripheral serotonin has been found during post-infectious fatigue following COVID-19 and other viruses (5) and in fibromyalgia patients (6,7).On the other hand, peripheral serotonin, once believed to mainly regulate intestinal mobility, is now understood to play key roles in regulating peripheral energy metabolism (8)(9)(10)(11)(12).Here, we examine if peripheral serotonin levels might be associated with fatigue in two other poorly understood conditions, chronic fatigue syndrome (CFS) and hypothyroidism with persistent fatigue.
In CFS, fatigue of unknown origin is a hallmark diagnostic feature (13).Its current prevalence in the U.S is 1.3% (14).Little is known about potential causes of CFS, though several recent studies point to activation of inflammatory cascades and issues with mitochondrial energy production (15)(16)(17)(18)(19)(20).Yet, currently there is no established laboratory test that could be used to diagnose or monitor disease progression or severity.Metabolomic, cytokine, mitochondrial, and natural killer cell meta-analyses have revealed an abnormal but mixed pattern, likely originating from both heterogeneity across patients and study methods (21)(22)(23)(24).Moreover, treatment is difficult and often leads to only partial alleviation of symptoms (13).
L-carnitine supplementation has been suggested to reduce fatigue in both hypothyroidism (26) and in CFS (30,31).Indicating a putative link between carnitine and serotonin, studies in mice have shown that carnitine supplementation increases serotonin concentration in the brain (32,33).Given that serotonin synthesis pathways are similar between the brain and periphery and across mammalian species, we here examine in humans the following hypotheses in both CFS and hypothyroidism with fatigue: (1) serum serotonin (S-5-HT) levels are low, (2) L-carnitine supplementation is associated with increased S-5-HT levels and reduced fatigue, and (3) there exists a correlation between peripheral S-5-HT and fatigue levels.

Study design and participants
The present study is a retrospective analysis of follow-up clinical data.The design is non-randomized, and each patient served as their own control (i.e., each patient was tested before and after L-carnitine supplementation).Ethical approval was obtained from the IRB committees at Faculty of Medicine, University of Helsinki, Helsinki, Finland, and at Feinberg School of Medicine, Northwestern University, Chicago, Illinois, USA.The sample consists of consecutive patients seen by the senior author over a 15-month period for CFS (5/2/2016 -8/23/2017) and an 11-month period for hypothyroidism (11/1/2016 -10/9/2017).Patients had been previously diagnosed with autoimmune hypothyroidism (initial N = 312) or CFS (initial N = 16).In addition, to be included in the sample, it was required that each patient spontaneously reported fatigue and chose to receive oral L-carnitine supplementation for 7 weeks, resulting in N = 52 for hypothyroidism (17% of patients) and N = 16 for CFS (all patients).Our L-carnitine dose was 300 mg twice a day, which falls within the typical range of 250 -3000 mg total dose per day across and within studies [e.g., (34)].The duration of 7 weeks was chosen based on a prior human study in CFS patients showing that the greatest improvement against fatigue took place between 4 and 8 weeks (30) and rodent studies showing effects on brain serotonin levels already after 21 -25 days of carnitine supplementation (32, 33).Patients that had received selective serotonin reuptake inhibitors (SSRIs) within the past 6 months were excluded, since SSRIs may alter S-5-HT levels (35), resulting in exclusion of 7 hypothyroidism and 2 CFS patients.Patients that did not return signed consent forms (N = 5 for hypothyroidism, N = 1 for CFS) were excluded.Finally, 1 CFS patient was removed during data analysis due to receiving a diagnosis of autoimmune thyroiditis after the study time window.All patients that satisfied these criteria were included, resulting in a final sample of 40 hypothyroidism (age mean 46 years, range 20 -79 years, 4 males) and 12 CFS (mean age 43 years, range 23 -68 years, 3 males) patients.Table 1 shows the detailed inclusion, exclusion, and removal criteria.In the sample with autoimmune hypothyroidism, the diagnosis had been made by endocrinologists at the Helsinki University Central Hospital, where the diagnostic criteria included elevated TPO antibodies, elevated TSH, reduced FT4, and ultrasound findings characteristic of autoimmune thyroiditis.All patients were on stable levothyroxine treatment and were, according to clinical and laboratory results, currently euthyreotic (Table 2).In CFS, the diagnosis had been made by internal medicine specialists at the Helsinki of University Central Hospital, where the diagnostic criteria followed the 2003 Canadian Consensus Criteria (36).All CFS patients had been found to be euthyreotic during the initial diagnosis as well as shortly before Lcarnitine (Table 2, samples taken 1 -4 months before starting Lcarnitine administration).

Outcomes and other collected variables
S-5-HT levels and behavioral fatigue levels had each been recorded twice, once before and once after the L-carnitine challenge.Specifically, on the 1 st physician visit (Week 0), the first fatigue score was recorded.The 1 st laboratory S-5-HT test was one week later (Week 1) followed by starting the L-carnitine regimen.The 2 nd S-5-HT laboratory test was six weeks later (Week 7), and the 2 nd physician visit with the second fatigue score one week thereafter (Week 8).On each physician visit, the fatigue score had been measured before discussing laboratory test results.Details of each are described below.

Fatigue score measurement
During the clinical visits, the treating physician recorded the fatigue level using Question 1 adapted from the Bristol rheumatoid arthritis fatigue numerical rating scale (BRAF-NRS) (37).Specifically, the patients were asked "How fatigued have you been on average during the past week on a scale of 0 -10, zero meaning as badly fatigued as you can imagine and 10 meaning not fatigued at all" (translated from Finnish).During data analysis, to be consistent with the convention where larger values reflect increasing symptom severity, the raw score axis was flipped (i.e., the raw values were subtracted from 10), resulting in final fatigue scores where 0 = no fatigue and 10 = maximal fatigue.Note that L-carnitine supplementation, started on Week 1 after the 1 st S-5-HT laboratory visit, continued until the 2 nd physician visit on Week 8 when the second fatigue score value was recorded.

S-5-HT assay
The S-5-HT (as well as all other laboratory) samples were collected at the Helsinki University Hospital Laboratory, where the S-5-HT protocol includes dietary restrictions for two days prior to the sample and collecting the samples between 8 -10 am.The dietary instructions were given in writing and read aloud to the patients by the treating physician ("For 2 days before the laboratory visit, do not consume foods containing serotonin or tryptophan, including avocado, banana, plum, peanuts, hickory nuts, pineapple, eggplant, tomato, kiwi, grapefruit, and honeydew", translated from Finnish).There was no fasting requirement.The laboratory personnel verbally confirmed that the patient had been aware of the dietary restrictions before each blood draw.S-5-HT was measured with the HPLC-ECD chromatographic method (38) using a fully clotted sample to accurately take into account the serotonin in circulating platelets.Specifically, blood was drawn into a pre-cooled serum test tube wrapped in a cold gel bag, mixed, and allowed to clot in cold at least 60 minutes.Then, the sample was centrifuged in cold, after which the serum was extracted and frozen.Next, the serum sample was shipped frozen overnight to a DAkkS accredited laboratory (L.A.D.R. GmbH, Labor Dr. Kramen & Kollegen, Geesthact, Germany) where the S-5-HT assays were conducted every 5 days and the samples were kept frozen until analysis.In this laboratory, intra-assay variation of the method has been reported at 4.5% (CV%) and the inter-assay variation at 4.7% (CV%).The detection limit was 15.9 nmol/l (2.8 µg/l) and the assay was linear between 45 and 8041 nmol/l (8 and 1417 µg/l).The 1 st and the 2 nd samples were analyzed independently, as soon as possible after each blood draw, using the same method described above.

Other data
In addition, the following clinical routine values had been measured once before the L-carnitine regimen: thyroid function (TSH, FT4), Hb, HbA1C, CRP, homocysteine, BMI, pulse rate, and oxygen saturation.All but BMI were within normal range (Table 2), which excludes frequent conditions that may independently cause fatigue (insufficient levothyroxine dose, anemia, diabetes with poor glycemic control, infection).All patients were non-smokers (39).Other diagnoses and medications were documented.

Statistical analysis
Some of the S-5-HT and fatigue scores were not normally distributed (Shapiro-Wilk test).Therefore, changes in fatigue and S-5-HT levels were examined with Related-Samples Wilcoxon Signed Rank Test within-subjects (before vs. after L-carnitine challenge).Correlations (simple and partial) between fatigue and S-5-HT levels were assessed using Spearman rho correlation coefficients.In addition, we conducted two post-hoc partial correlation analyses to examine possible confounding effects of age or TSH level.Finally, we computed subgroup control analyses based on the absence or presence of co-occurring diagnoses (Related-Samples Wilcoxon Signed Rank Test within-subjects before vs. after L-carnitine challenge, Spearman rho simple correlations between fatigue and S-5-HT levels).All significance (Sig.)values were 2-tailed.All statistics were computed using SPSS (version 28; IBM, Chicago, IL, USA).
The CFS sample (N = 12) included to following co-occurring diagnoses (N = 6) made at least 1 year before starting L-carnitine supplementation in the study: asthma (N =1), allergic rhinitis (N = 1), celiac sprue (N = 1), type 1 diabetes (N = 1), prostate hypertrophy (N = 1), and gastro-esophageal reflux disease (N = 1).Again, there were no new diagnoses within 1 year or during the L-carnitine supplementation, and medications were unchanged for at least 4 months before and during the study.

Control analyses for age or TSH level as confounding factors
In addition to the above planned statistical tests, we conducted two post hoc partial correlation analyses to examine possible confounding effects of age or TSH level as confounding factors.
Withage as a covariate, the Spearman rho values remained practically unchanged (differed by less than 0.02) regardless of if we excluded or included the covariate, separately for all 4 conditions [Hypothyroidism/CFS x Before/After L-carnitine].With the TSH level (available only before L-carnitine, since these were clinical data) as a covariate, in CFS the results again remained similar (rho value differences < 0.07).However, in hypothyroidism before L-carnitine, there was a correlation (rho = 0.376, Sig.= 0.017) between fatigue scores and TSH, which led to underestimation of the simple correlation between S-5-HT and fatigue scores (without TSH as a covariate, rho = -0.24Sig.= 0.131 as reported above, with TSH as a covariate rho = -0.364Sig.= 0.023).After L-carnitine, this confounding effect of TSH disappeared (rho value difference < 0.01).

Control analyses for co-occurring diagnoses
Finally, to examine if the absence vs. presence of any co-occurring diagnoses influenced the results, in another post hoc analysis we split each group into two subgroups: those that had no other diagnosis, and those that did have one or more additional diagnoses (for hypothyroidism, N = 19 and N = 21; for CFS, N = 6 and N = 6, respectively).Independently in each of the 4 subgroups, the changes in both S-5-HT and fatigue level from before to after Lcarnitine remained significant (Sig.< 0.001 for each hypothyroidism subgroup; Sig.= from 0.026 to 0.028 for each CFS subgroup).The correlations between S-5-HT and fatigue were also similar.Specifically, for hypothyroidism, all correlations that were significant or non-significant remained so in the subgroup analyses.For CFS, the subgroup without co-occurring diagnoses showed the same pattern of non-significant correlation before and significant correlation after L-carnitine; in contrast, the subgroup with co-occurring diagnoses flipped into significant correlation before and non-significant correlation after L-carnitine, most likely reflecting that at N = 6 it becomes difficult to obtain reliable correlation estimates.Overall, the subgroup analyses indicated that co-occurring diagnoses did not influence the main results or conclusions.

Discussion
The serum serotonin levels were low before L-carnitine, Lcarnitine supplementation was associated with a robust increase in peripheral serotonin and reduction of fatigue, and there was a statistically significant correlation between serum serotonin and fatigue levels in both patient groups.Thus, the results supported our hypotheses.This finding is mechanistically interesting because there are no published direct links between carnitine and peripheral serotonin synthesis.Further, to our knowledge, within-subjects correlations between S-5-HT and fatigue levels have not been previously reported in humans in any disorder.The fact that the findings were similar across CFS and hypothyroidism points to the possibility that low peripheral serotonin may be a common theme across some disorders that include fatigue.The control analyses suggested that age or co-occurring diagnoses did not bias the results.These findings and interpretations are indirectly supported by a recent report on fatigue in the long COVID-19 syndrome where lower peripheral serotonin was observed in patients than in healthy controls (5).To arrive at putative mechanisms, below we discuss 5-HT and carnitine, along with their possible links to fatigue.
Peripheral 5-HT is produced and secreted by the enterochromaffin cells in the gastrointestinal tract where it regulates intestinal mobility.Much of the peripheral 5-HT is picked up from the intestine by, and circulating in, platelets.In addition to intestinal mobility, it has become evident that peripheral serotonin participates in systemic energy homeostasis (10, 12).S-5-HT is produced in a two-step enzyme reaction from the essential amino acid tryptophan.The first reaction is catalyzed by tryptophan hydroxylase (TPH-1), which requires Ltryptophan, tetrahydrobiopterin (BH4), and O 2 as co-substrates, and has ferrous ion (Fe 2+ ) as an activator.The second enzyme, L-aromatic amino acid decarboxylase (AADC a.k.a.AAAD), changes the intermediate product to 5-HT.This step has 5-hydroxy-Ltryptophan as substrate, pyridoxal 5'-phosphate as an optional activator, and is influenced by Al 3+ , Zn 2+ , Ca 2+ , and Mg 2+ levels.Peripheral serotonin in hypothyroidism and CFS has previously received little attention, and S-5-HT measurements have been used mainly in the carcinoid tumors/syndrome.Current serotonin medications are aimed at increasing central 5-HT neurotransmission in the brain, and there are no approved indications for manipulating peripheral 5-HT, though peripheral 5-HT synthesis is being considered as a potential target for developing pharmacological agents in some disorders (8,9).Moreover, the peripheral and brain pools of 5-HT are synthesized separately and isolated from each other by the blood-brain-barrier; only the precursor 5-HTP passes the blood-brain barrier.
We are not aware of prior studies reporting low S-5-HT in hypothyroidism or CFS.A recent study suggested that plasma levels of 5-MT, a compound related to serotonin metabolism, may be decreased in CFS patients, but this has been interpreted as a confounder caused by serotonin-selective reuptake inhibitor (SSRI) use in these patients (40).The present results support the idea that peripheral serotonin metabolism may be altered in hypothyroidism and CFS.Theoretically, a low S-5-HT value would be expected to result from lack of its precursor tryptophan or disturbance in the two-step synthesis described above.Further, while low tryptophan dietary intake is rare, CFS and autoimmune hypothyroidism may be associated with an increment in indeloamine-dioxygenase (IDO) expression, which would promote the tryptophan-kynurenine pathway that decreases serotonin synthesis.Specifically, when IDO activity increases, more tryptophan is diverted away from serotonin synthesis and toward kynurenine.Interestingly, the aryl hydrocarbon receptor (AhR) of the tryptophan-kynurenine pathway serves as a critically important mediator between tryptophan metabolism, immune reactions, and tissue homeostasis (41)(42)(43), and several kynurenine metabolites have been associated with fatigue (44,45).
The principal role of carnitine is to transport long-chain fatty acids into the mitochondria for energy production via betaoxidation (46), but it also participates in glucose metabolism, detoxification of acyl moieties (47), and in the preservation of intracellular Coenzyme A (CoA) homeostasis (48).In some openlabel trials, carnitine supplementation has been suggested to reduce CFS symptoms (30, 31).However, studies examining carnitine levels in CFS patients have been inconclusive, showing either normal (49, 50) or reduced levels (30, 40, 51, 52) for at least some acylcarnitine forms (53).
While carnitine supplementation could have benefits, some recent studies suggest that chronic oral L-carnitine administration may be associated with an increased risk for atherosclerosis via increased trimethylamine (TMA) and/or trimethylamine N-Oxide (TMAO) levels (54, 55).If true, this would not favor long-term Lcarnitine supplementation, or may require the development of additional strategies to reduce TMA/TMAO levels (56).Yet, at this point, it is unclear if L-carnitine promotes atherosclerosis (57) and the relationship seems to be context dependent (58).
Mitochondrial dysfunction has been suggested to contribute to CFS (15)(16)(17)(18) and symptom severity seems to correlate with the degree The mechanisms that link carnitine intake, S-5-HT, and fatigue, in both CFS and hypothyroidism, are yet unknown.However, in the mouse brain, L-carnitine supplementation has been shown to increase brain 5-HT concentration (32, 33).Perhaps most tellingly, carnitine and 5-HT share the same cellular target organ, since carnitine participates in energy production at the mitochondrial level, and 5-HT receptors have a role in mitochondrial function in muscle tissue (11).As a putative bridge between carnitine and S-5-HT, increased energy production allowed by L-carnitine supplementation might promote peripheral 5-HT synthesis.This idea is indirectly supported by a study in fibromyalgia patients where supplementation with another key mitochondrial agent, CoQ 10 , increased low peripheral serotonin values (7).
Limitations of the study include a relatively small nonrandomized retrospective clinical sample.As typical for retrospective chart reviews, the outcome variables were limited to those that were systemically recorded on the patient charts during routine clinical outpatient visits, resulting in that some tests (e.g., corticotropin-releasing hormone stimulation test) that may have been helpful will need to be conducted in future prospective trials.Further, since this was an investigation focused a priori on S-5-HT levels, it remains unknown if the effects were specific to serotonin synthesis, or if other pathways were also affected.Moreover, it is not known if S-5-HT mediated the effects of L-carnitine on fatigue, as it is possible that L-carnitine could have effects unrelated to S-5-HT that could reduce fatigue and in parallel promote increased synthesis of peripheral 5-HT.Finally, for recording fatigue levels, the present study adopted BRAF-NRS Question 1 (37), which consists of a single item and is therefore well suited for clinical routine.Future prospective studies should use more extensive fatigue questionnaires, such as the Fatigue Assessment Scale (FAS) (64) or the 11-point fatigue numeric rating scale (NRS) (65,66).We also note that, while NRS, BRAF-NRS, and FAS have been validated in some disorders, currently there is no single fatigue behavioral instrument that has been validated in both hypothyroidism and CFS.
Future directions include prospective randomized controlled trials where the present findings are replicated with more detailed questionnaires and other tests for fatigue, depression, sleep, and stress.It would also be pertinent to add laboratory assays that allow putting the results in the larger context of cellular senescence, which seems to be disturbed in several disorders where fatigue has been associated with mitochondrial dysfunction.Specifically, recent research has suggested a bidirectional communication between metabolic dysfunction (loss of metabolic homeostasis) and cellular senescence (67,68).This relationship between metabolism and senescence has been considered for many chronic phenotypes, including for fatigue (69)(70)(71)(72)(73)(74)(75).Mitochondrial dysfunction has been implicated in the altered metabolic flexibility in such conditions (76,77).
In conclusion, the present results suggest new and unexpected links between fatigue, peripheral 5-HT, and L-carnitine in CFS and hypothyroidism, and offer testable hypotheses for future mechanistic and clinical studies.Peripheral serotonin and its putative link with L-carnitine appear to emerge as new candidates for understanding and treating fatigue across several disorders.

1 S- 5 -
FIGURE 1 S-5-HT Levels.Levels in chronic fatigue syndrome (A) and hypothyroidism with fatigue (B) before and after L-carnitine.Blue and red dots represent the individual observations, whereas the horizontal black lines show the mean ± SD values.

2 Fatigue
FIGURE 2 Fatigue Scores (0 = no fatigue, 10 = worst imaginable fatigue).Values in chronic fatigue syndrome (A) and hypothyroidism with fatigue (B) before and after L-carnitine.Blue and red dots indicate the individual observations, whereas the horizontal black lines show the mean ± SD values.

3
FIGURE 3Correlation Analyses.The panels show the data points and correlations between S-5-HT and fatigue level, separately for chronic fatigue syndrome (CFS, A) and hypothyroidism with fatigue (B).The Spearman rho values and linear trendlines are shown separately for before vs. after L-carnitine.

TABLE 2 Subject
Demographics and Clinical/Laboratory Test Results (mean ± SD).