In the important study ‘Serotonin reduction in post-acute sequelae of viral infection' by Wong et al. (16) they investigated PCS in four human cohorts, in animal models of viral infection and in organoid cultures. First, they conducted a study among 1,540 PCS patients who presented to a post-COVID center with severe complaints. They identified eight clusters of patients based on clinical symptoms, varying from mainly physical problems, such as loss of strength in muscles, to mainly neurocognitive complaints such as memory disorders. The researchers performed targeted plasma metabolomics on 58 representative PCS patients 3–22 months after infection and found serum serotonin reduction compared with 30 healthy controls.

For this important finding they present three causes: a) diminished intestinal absorption of the serotonin precursor tryptophan. Because of downregulation of genes of the angiotensin converting enzyme (ACE2) these receptors are strongly decreased. Furthermore, not only tryptophan, but also the COVID-19 virus with its spike proteins attaches to these receptors (20, 21). As a consequence, during the COVID-19 infection, tryptophan has to compete with the virus over a reduced number of ACE2 receptors; b) micro-clots of thrombocytes. Thrombocytes contain serotonin. The micro-clots reduce the number of thrombocytes and thus the availability of serotonin; and c) enhanced monoamine oxidase (MAO) that promotes the breakdown of serotonin.

In a study by Sadlier et al. (17), a cohort of 20 hospitalized PCS patients were compared to 20 healthy controls, 4–6 months and 6–9 months after infection. Levels of multiple metabolites with immunomodulatory properties were elevated like quinolinate, a toxic KP metabolite. There were reduced serotonin levels and among other things the serum glutamate (a neurotransmitter) level was increased.

Su et al. (18) performed a longitudinal multi-omic analysis in COVID-19 patients (n = 209). This cohort was followed immediately after the COVID-19 infection and had less severe symptoms. They measured autoantibodies, specific COVID-19 RNAemia, metabolic profiles, global plasma proteomic and peripheral blood mononuclear cells (PBMCs) in blood draws. They found no reduced serum serotonin levels compared with 457 healthy controls. What stands out is that the patients reporting neurological symptoms exhibited elevated proteins associated with the negative regulation of the circadian sleep/wake cycle. The hormone melatonin is responsible for this and is produced in the brain (in the pineal gland) from serotonin.

Wong et al. conclude that PCS patients with serious complaints have a greater chance of permanently retaining reduced serotonin levels than PCS patients with mild complaints. They checked this with a cohort of Peluso et al. (22) and found that serum serotonin levels did indeed negatively correlate with the severity of the complaints.

However, in the retrospective study by Mathé et al. (19) no reduced serum serotonin levels were found using the Liquid Chromatography—Mass Spectrometry (lc-ms/ms) technology in a cohort of 34 PCS patients at least 6 months after infection and with serious complaints, which they compared with 14 healthy controls.

Although the study by Wong and colleagues is the most comprehensive of all the studies with interesting and important results, I agree with the conclusion of Mathé and colleagues that serum serotonin is not a reliable biomarker in PCS and should not be used in routine diagnostic assessment, based on two arguments.

The first reason is that serotonin cannot cross the blood-brain barrier (14). It appears that only some peripheral serotonin reaches the brain via the cranial nerve, the vagus nerve (16). This nerve normally uses Acetylcholine (Ach) as neurotransmitter (9). So, peripheral serum serotonin level is not directly related to the serotonin level in the brain. Based on animal models, Wong et al. assume that serotonin in the brain is not reduced in PCS. In vivo, however, it is technically very difficult to measure serotonin in the brain. With all possible techniques [microdialysis, functional magnetic resonance imaging (fMRI), fast-scan cyclic voltammetry (FSCV), genetically encoded serotonin indicators (GESIs) and positron emission tomography (PET)] either the spatiotemporal resolution is too poor or the technique is too invasive or/and too expensive (23). Wong et al. conclude that with reduced serotonin in the peripheral serum in PCS, less serotonin can move up the vagus nerve to the hippocampus, the control center of memory, possible causing the memory disorders in PCS. In our article in which we describe a study into the treatment of 95 PCS patients with selective serotonin reuptake inhibitors (SSRIs; 16), we give however another explanation. We hypothesize that serotonin reduction also occurs in the brainstem and the brain. After all the pons in the brainstem is the origin of the serotoninergic system and from there, axons are sent throughout the central nervous system (CNS; 6, 7). The afferent vagus nerve also arises from the pons (6, 7) and not from the hippocampus, which Wong and colleagues assume (16). The brainstem is full of ACE2 receptors, to which not only tryptophan but also the COVID-19 virus can attach (20). Hypometabolic areas are found in the pons in PCS (24, 25).

Recent research from Besteher et al. (26) confirms this argument. They found with fMRI scans from PCS patients suffering from neuropsychiatric symptoms (n = 30) significantly larger gray matter volumes (GMV) than in healthy controls (n = 20). For example in the prefrontal cortex (PFC)—which is involved in a range of higher order cognitive functions and in the hippocampus, control center of memory (27). In these brain areas the neurotransmitter serotonin predominates (27, 28). The authors state the enlargement of the GMV could be a sign of recovery through neurogenesis or compensation (26). Another potential explanation is cerebral swelling caused by immune reactions (26). Given that the neuropsychiatric symptoms persist, it seems likely that the enlargement of the GMV is caused by pathology. Moreover, it provides a plausible explanation for the positive effect of SSRIs on neurocognitive disorders in PCS when there are serotonin balance problems in those brain regions (13).

Furthermore, Su et al. (18) found that melatonin, which is produced in the brain from serotonin, was reduced. This is an additional support—contrary to the conclusion of Wong at al. —that cerebral serotonin may be reduced.

The second reason to reject serotonin as a biomarker, is the variability in the degree of serum serotonin reduction between the cohorts in the different studies (16–19). The causes of this variability can probably be found in the many different variables between the studies. Such as: the time passed between infection and measurement: ranging from 0 to 22 months, the severity of the PCS complaints, their exact quantification (especially for subjective complaints such as neurocognitive complaints) and to which of the eight subgroups the patients belonged in a special cohort. I believe that the methodology used and therefore the results in these studies vary too much to conclude that serotonin is a reliable biomarker in PCS research.

Unlike serotonin, tryptophan can cross the blood-brain barrier (9, 14) and may therefore be a better biomarker option (13, 15). In the case of a comparative study however, the above variables should preferably be more comparable.

Beside the three causes for the serotonin reduction given by Wong and colleagues, there may be a fourth cause: the KP, a pathway to create the energy factor nicotinamide adenine dinucleotide (NAD+), which interacts extensively with the immune system, seems strongly activated in COVID-19 and PCS (15, 29–31). This results in the formation of many toxic kynurenine metabolites (15, 29–31). This process demands a lot of tryptophan (14; see Figure 1) and because tryptophan is an important precursor of serotonin, a deficiency of tryptophan can also cause a deficiency of serotonin (9).

In the Wong et al. study, the kynurenine metabolites decline as PCS lasted longer. Therefore, the researchers conclude that an activated KP may not be a major cause of serotonin reduction. However, in a study by Guo et al. (30) PCS patients show persistently elevated levels of INDO-2, an enzyme which stimulates the production of kynurenine (Figure 1). In the study by Cron (15) the PCS patients had elevated levels of kynurenine metabolites (such as quinoline), while tryptophan was depleted. Additionally, Cysique et al. found a significant relationship between the level of toxic kynurenine metabolites in blood and the severity of cognitive impairment in PCS (29). These authors conclude that the severity of neurocognitive symptoms seems to be directly related to the degree of overactivity of the KP. The more active the KP, the less tryptophan is left for the production of serotonin.

Figure 1 illustrates that serotonin deficiency can lead to a melatonin deficiency too. The hormone melatonin regulates the circadian sleep/wake cycle (17, 32). Many PCS patients have sleep problems (13, 33).

Too much kynurenine due to a runaway positive feedback loop of the KP, blocks tetrahydrobiopterin (BH4), a coenzyme for the production of the neurotransmitter dopamine, which in turn ensures the production of the neurotransmitter (nor)epinephrine (9). Norepinephrine from the sympathetic autonomic nervous system increases the frequency and force of muscle contractions (34) why PCS patients with muscle complaints have more symptom reduction with an SNRI (selective serotonin and norepinephrine reuptake inhibitor) compared with an SSRI (13).

If we look at the toxic KP metabolites, we see that both kynurenine acid and quinolinic acid are glutaminergic receptor antagonists. This causes glutamate (a neurotransmitter) accumulation (35) which leads to various problems, such as reduced concentration and palpitations (35), complaints that PCS patients often suffer from (13, 33). That is why we recommend in our article (13) research into N-acetylcysteine as a drug to restore the glutaminergic balance in PCS (35).