The results of this exploratory questionnaire-based study of the effect of treatment with an SSRI in PCS show that an SSRI contributes significantly to the reduction of PCS symptoms. 63.4% (n = 95) of patients reported a reasonably good to strong decrease in symptoms and an improvement in functioning. This increased quality of life can contribute to social participation. Four patients had developed clinical depression or anxiety disorder during PCS. Treatment with an SSRI also eliminated these disorders. People who had felt despondent due to PCS also felt less gloomy after an SSRI, but all attributed this to the reduction in their PCS symptoms.
An explanation for the skewed gender distribution in this study—also found in other research in PCS—may be that many genes for the immune system lie on the X chromosome so that men are more likely to have severe Covid-19 infections, but women are more likely to have more severe PCS symptoms that last longer7,49.
The high level of education in this study compared to other studies1 can probably be explained by the recruitment of patients through LinkedIn. The overrepresentation of female patients from healthcare and education could be caused by the predominance of female workers in these professions on the front lines of the pandemic. Moreover, they returned to work quickly after their Covid-19 infection which is a risk factor for PCS1. Many other risk factors for PCS in this study are consistent with Davis’s review article1, such as: asthma, allergies, connective tissue diseases, Epstein-Barr virus etc. A notable addition to the known risk factors is factor V Leiden thrombophilia. In this study, there were two patients with this coagulation disorder, while the prevalence in the general population is only 1:500050. Patients with this coagulation disorder have an 80-fold increased risk of thrombosis.
All patients in our group (n = 95) were chronically fatigued during PCS while in a German study11 only 19/42 patients of the PCS study population reported chronic fatigue such as in ME/CFS. An explanation for this difference could be that fatigue in PCS increases with time. While the duration of PCS in the German study was only six months, in our study it averaged 15 months. Also, in this German study, of the other PCS patients without chronic fatigue (23/42), only 15 had a neurological or cognitive impairment. In our study, 100% had a neurological or cognitive impairment. However, neurocognitive symptoms begin only a month to a few months after Covid-19 infection and worsen over time51.
Table of Contents
Strength and weakness of the study
No validated questionnaires are yet available for PCS. However, with a new disease, it is important to learn about all symptoms, so we used a questionnaire that included open-ended questions. Thus, we discovered that symptoms can shift over time. Dyspnoea and decrease in smell seem to decrease over time, while fatigue and brainfog seem to increase. Through the open-ended questions we also discovered new symptoms, such as dissociatieve symptoms and not being able to chew properly.
Using three different instruments to determine treatment effectiveness is a further strength of this study. There is strong evidence for the reliability of these measures. The Bell score is a widely used instrument in research on (chronic) fatigue, although not validated. Furthermore, the rating of the Open question outcomes was found to be reliable. Importantly, the three effect measures correlated strongly with each other, supporting the reliability of the individual measures.
The main weakness of this study is that it is not a randomized controlled trial (RCT). We had no control group. Therefore, a placebo effect cannot be ruled out. However, it is known that 85% of patients who have symptoms two months after Covid-19 infection still have them after one year1. ME/CFS and dysautonomia are usually lifelong1. So without treatment with an SSRI, many PCS patients may suffer from these conditions in permanently.
And there is more evidence that a placebo effect may not fully explain the positive results. A placebo effect usually occurs shortly after the start of an intervention and diminishes again after a few weeks, unless a positive expectation is given again52. However, 72 patients (n = 95) still had no response in the first weeks, but instead suffered side effects. The 24 patients who used the SSRI for more than six months reported that the effect was maintained, while they were not asked by us to do so, as we had no treatment relationship with the patients. A cohort study without a control group also does not exclude natural recovery. However, because the participants had been seriously ill for 1.5 years and deteriorated over time, it seems highly unlikely that, if they had received an SSRI and recovered after a few weeks, this would have been due to natural recovery.
Working with self-reporting is always vulnerable to biases. However, self-report often expresses the experience of patients better. In psychiatry not all physiological parameters can be measured. We also asked the patients to rate their complaints on a scale of 1 to 10. By inviting them to compare the severity of their complaints, we introduced more structure into the self-report.
Finally, low-cost SSRIs are covered by insurance and there is no cost to the patient, which argues against a placebo effect53. However, the possibility of a placebo effect can only be completely ruled out by an RCT.
The Bell score before Covid-19/PCS and the Bell score during PCS before starting an SSRI were completed retrospectively by all patients. This may have led to some bias. The complaint of PEM was not part of the 8 complaints in the Score list; patients had to list and score this on their own under “other complaints”. This may have led to underreporting of this important symptom. LinkedIn gave an overrepresentation of patients with a higher level of education, but perhaps also of a group of initially healthy people who were fully employed. It is precisely in this group that the impact of PCS as well as the outcomes of an SSRI may be well observed.
Potential mechanisms of action of SSRIs in PCS
Dysregulation of the tryptophan system
Normally the catabolic kynurenine pathway (KP) degrades 95% of the essential amino acid tryptophan to produce the vital energy cofactor NAD+. The rest of the tryptophan serves as a precursor for serotonin and melatonin. (See Fig. 6). In addition to NAD+, the KP generates different metabolites: kynurenine, kynurenic acid, 3OH-kynurenine, quinolinic acid, anthranilic acid and 3OH-anthranilic acid. The KP is stimulated by inflammation and in PCS it is found overactive26,27,28,30,32.
The overactive KP absorbs more than 95% of the tryptophan in PCS. Research shows that the amount of tryptophan in blood is decreased in PCS patients30,31,32. Tryptophan deficiency is probably already present during the infection with Covid-19, because Covid-19 attaches with its spike proteins to the Angiotensin Converting Enzyme (ACE2) receptors in the intestines, while tryptophan also uses the same receptor to be absorbed from the intestines54. There is a significant relationship between the level of metabolites in blood and the severity of cognitive impairment in PCS28, (p < 0.001)30,31. The KP has normally potentially neurotoxic as well as neuroprotective aspects. An overactive KP during a severe infection however, is toxic for neurons, especially for serotonin neurons. Not only does the KP hijack tryptophan away from the serotonin pathway28,30,31, the metabolite kynurenine and oxidative stress also lowers the level of tetrahydrobiopterin (BH4)1,55,56, an important coenzyme of the serotonin pathway. Moreover, BH4 is an important coenzyme of the dopamine pathway—and to the norepinephrine pathway20,55—thus the overactive KP damages these two neurotransmitter systems too. When there is no longer enough tryptophan, this can lead to serotonin depletion16.
SSRI’s make the serotonin in the neurons more available through inhibition of reuptake and can partly compensate for the deficiency of tryptophan57. Moreover, SSRIs lower oxidative stress58. That could be an explanation for our finding that PCS patients often have fewer complaints due to SSRIs. But that may come to an end when all the serotonin from the neurons is in the synapses. Nevertheless, after half a year or more (in this research 24 patients), many people still felt good when using an SSRI. So we can (hypothetically) conclude that an SSRI probably indirectly contributes to—when there is a lack of serotonin in the neurons—that these neurons start to pull harder on the available tryptophan. So SSRIs could slow down the overactive KP. There is no theoretical argument against this hypothesis16.
But there are more avenues for future research advancing this line of clinical research based on tryptophan metabolism. IDO2 expression can be halted by an AHR antagonist28, such as the dietary supplement resveratrol59 or the experimental anticancer drug IK-17560. Furthermore, kynurenine is a glutaminergic receptor antagonists and quinolinic acid even blocks this receptor28. N-acetylcysteine (NAC) not only produces the antioxidant glutathione, but is also a glutaminergic receptor agonist61. The poisonous quinolinic acid and kynurenine acid are nicotinic receptor antagonists. Nicotine is a nicotinic receptor agonist. To stick nicotine patches helps PCS patients. This may be not only because nicotine is a nicotinic receptor agonist and therefore an opponent of these poisonous metabolites, but nicotine is a strong acetylcholine (ACh) agonist as well62. ACh is the most important neurotransmitter in muscles. Furthermore, folic acid (vit. B11)63 can promote the conversion of B2 back into B455,63 in favour of the serotonin and dopamine pathway.
Disrupted HPA axis
After a stressful event, the hypothalamus secretes corticotropin-releasing hormone (CRH), which causes the pituitary gland to secrete adrenocorticotropic hormone (ACTH), which in turn causes glucocorticoids (GCs) to be released by the adrenal cortex.
In PCS cortisol (glucocorticoids) levels were on average halved19. That is an indication that the HPA axis is disrupted14,19. In comparison, in ME/CFS, the HPA axis is less severely disrupted12. GCs released by the adrenal glands, act on almost all types of immune cells and perform evident immunosuppressive and anti-inflammatory functions14,15,16,17,18. SSRIs affect the HPA axis16,64. For example by increasing GC receptor density in the hypothalamus and hippocampus (the memory control centre). Further on, acute administration of 5-HTP receptor ligands increased the plasma levels of ACTH and cortisol in both animals and humans15,16. But the effect is not one-dimensional. For example: in patients with major depression with high cortisol levels, an SSRI lowers the cortisol level but triggers a higher Cortisol Awaking Response (CAR)14,15,16,17,19. In ME/CFS, an SSRI works moderately in only one-third of patients12. SSRIs seem to work better in PCS than in ME/CFS. This may be an indication that SSRIs are (partly) effective in PCS by influencing this hormone axis. Another indication of this hypothesis is our report of the patient who was given hydrocortisone because PCS caused her only functioning adrenal gland to fail, but who recovered with an SSRI, after which she was able to taper off the hydrocortisone. In this patient, the SSRI apparently restored not only the HPA axis, but also the disrupted hypothalamic-pituitary-thyroid axis15,16. The question is therefore whether—apart from the HPA axis—more hormonal axes starting from the hypothalamus are disturbed by Covid-19/PCS. Thus, reported changes in the menstrual cycle after Covid-1965 could also result from a disrupted hypothalamic-pituitary–gonadal axis.
Disrupted brain stem
The brain stem, the oldest part of our brain, is responsible for basic functions such as body temperature, sleep–wake rhythm, heart rate, breathing, blood pressure, digestion, eye movements, urination, hearing, tasting, chewing, swallowing, and feeling movement and gravity. Neurotransmitters that are especially important there include serotonin, norepinephrine and dopamine10. The serotonergic neurons start in the raphe nuclei in the pons and may exert their influence there10.
The Covid-19 virus enters the brain stem cells easily attaching to the many ACE2 receptors66. Hypometabolic areas in the pons were found in a study of three PCS patients. Typically, SSRIs worsen sleep quality67, though, not always68. In this study, 29 patients reported in the open-ended question that their sleep improved on the SSRI. This could be explained by the influence of the SSRI on the brainstem. But this may not be the only reason that patients sleep better. We saw in Fig. 6 that the sleep hormone melatonin originates from serotonin. When the serotonin neurons withdraw more from the available tryptophan, more melatonin can be produced in addition to more serotonin. The decrease in palpitations, shortness of breath, gastrointestinal complaints, better temperature regulation (one patient) and the ability to chew better (one patient) also supports the idea that SSRIs may (partially) restore the neurotransmitter systems disrupted in the brainstem.
Disrupted autonomic nervous system (ANS) balance
Dysautonomia, especially POTS, is often a symptom of PCS. The clinical picture with a Bell score of 20 or less (almost half of the patients in this study) is striking, as shown by the answers to the open questions: overstimulated patients with palpitations lying limp and exhausted on the couch. In this state, the brainstem shows heightened arousal and seems to be in the fight-or-flight response along with the sympathetic nervous system14,69. This is a primitive survival mechanism in the face of danger, in which the sympathetic ANS dominates. The ‘fight-or-flight’ response activates the HPA axis so that extra cortisol is released and glucose is released into the muscles for action69. Instead, measured cortisol levels in PCS patients average only 50% of normal19 and the muscles are weak rather than ready to contract. The muscle weakness is not only because no glucose is released in the muscles. It is also because PCS is associated with autoantibodies against the G protein-coupled adrenergic receptor and muscarinic acetylcholine receptor70. Acetylcholine is the main neurotransmitter that sets muscles in motion.
The vagus nerve of the (parasympathetic) ANS also originates from the pons in the brainstem. But this nerve—just like muscles—uses acetylcholine as a neurotransmitter, so an SSRI cannot intervene there. One speaks of a paralyzed vagus nerve, probably because of the autoantibodies against the muscarinic acetylcholine receptor1. The sympathetic ANS predominates over the parasympathetic ANS in PCS, while G protein-coupled autoantibodies against the adrenergic receptor are also present. Perhaps the (nor)epinephrine pathway has fewer problems with the overactive KP, because it does not use tryptophan as a precursor, but phenylalanine. SSRIs often seem to reduce POTS and palpitations. This is unrelated to an inhibition of serotonin reuptake, as serotonin does not affect the ANS. The inhibition of norepinephrine reuptake may reduce POTS, but could theoretically lead to palpitations. The reduction in palpitations must be found in another mechanism, such as the influence of SSRIs on the anterior cingulate cortex (ACC) in the brain61.
Brainfog and sensory overload responded best to treatment with an SSRI. The raphe nuclei (pons) in the brain stem is the location of the origin of the serotoninergic system. From there, axons are sent throughout the CNS10. So SSRIs can intervene throughout the whole brain.
Dissociative symptoms also disappeared. In sensory overload and dissociation, there is sensory overload due to lack of filtering. The primary unimodal sensory brain regions do not cooperate well with the associative sensory brain regions71,72. It is known that an SSRI can sometimes help with this71,73.
Many PCS patients struggle with forgetfulness1. In the hippocampus, the control centre of memory, serotonergic neurons are dominant74 SSRIs also stimulate the production of serotonin cells in the hippocampus74. Possibly partly because of this, the patients’ forgetfulness decreased.
Sigma1 receptor agonist
The SSRIs fluvoxamine and fluoxetine have been shown to have extra anti-inflammatory effects during Covid-19 infection by inhibiting sphingomyelinase acid (ASM)33. Furthermore, an SSRI reduces the pro-inflammatory cytokines Interleukin 2 (IL 2) and IL 17 in the CNS. In this case, the SSRI must be a sigma1 receptor agonist. This opioid receptor is inter alia involved in reducing virus replication and inhibiting reactivation of herpes viruses such as Epstein-Barr Virus (EBV) and subsequent endoplasmic reticulum (ER) stress and inflammation35. We recommended only SSRIs who are sigma1 receptor agonists34,35. One patient was first given sertraline by the GP, with no response. After switching to citalopram, she did respond reasonable good. In five other patients, GPs prescribed the sigma1 receptor antagonists sertraline (n = 4) or paroxetine (n = 1). These five patients reported good (n = 2), reasonable good (n = 2) or moderate (n = 1) improvement. There was no evidence in favour of or against a difference in effect measured by the open question depending on whether patients received sigma1 agonists or antagonists (BF = 0.43). If we include the patient who changed from sertraline (no effect) to citalopram (reasonable good effect), no significant difference remains (BF = 0.39) This is only anecdotal evidence that the mechanism via the sigma1 receptor plays a role in the action of SSRI in PCS. However, the group using a sigma1 antagonist in this study is too small for a proper statistical analysis.
Positive influence of SSRIs on the circulatory system
Many Covid-19 and PCS patients have microclots indicative of coagulation problems. Microclots impede oxygen and nutrients flow to organs and tissues8. Platelets are involved in clotting. Platelets transport serotonin, because serotonine has a function in clotting. With serotonin deficiency, platelets become less functional. Because SSRIs inhibit the reuptake of serotonin in platelets, they prolong clotting time and could theoretically dissolve microclots75,76. The two patients with factor V Leiden thrombophilia responded well and moderately, respectively. This could mean that the anticoagulant effect of SSRIs—assuming that microclots played a role in these patients—might contribute to their response. If it were confirmed that PCS is more common in factor V Leiden thrombophilia, this coagulation disorder should be added to the list of risk factors.
But an SSRI can benefit blood circulation in other ways, too. They may show an anti-inflammatory effect on endothelial cells. SSRIs reduce the expression of cytokine-induced endothelial adhesion. This makes it difficult for circulating adhesion molecules, such as monocytes, to adhere77. This mechanism may partially explain their cardioprotective effects76.
Subgroup with fibromyalgia
Muscle pain and weakness decreased the least in the total group (n = 95). We also see this in the patients with fibromyalgia (n = 5) who reported little or no improvement after using an SSRI.
The group of non-responders (n = 9) were more severely ill with Covid-19 infection than those who did report good response to SSRIs. It is possible that the cascade of severe inflammation caused by Covid-19 in the CNS78 released much histamine. Inhibitory histamine receptors lower serotonin in the CNS, preventing an SSRI from effectively releasing extracellular serotonin40. By adding Histamine1 and 2 antagonists to treatment, an SSRI could in principle still become effective40. However, the lack of response to an SSRI may also have been caused by a serious factor: namely, neuropathology similar to Alzheimer’s disease is found in Covid-19 infections. In the CNS β-amyloid aggregations, plaque formations, tauopathy and cell death have been described79. In these conditions, an SSRI cannot possibly be effective anymore.
When should the SSRI be phased out?
Our hypothesis is that for many PCS symptoms, neurotransmitter systems are not damaged but dysregulated and most likely suffering from a tryptophan deficiency due to an overactive KP. But after a chronic course of PCS lasting two years, SSRIs cannot be expected to ”reset” these systems in a short time as long as the kynurenine pathway is still overactive. By comparison, the treatment duration for initial depression is six months to one year and for recurrent depression usually at least 3 years80. In toxic drug-induced depersonalization disorder, sometimes 6 years of treatment is advised (first author’s clinical experience)73. Drug-induced depersonalization disorder is phenomenologically similar to PCS, excluding fatigue and muscle pain. Poor stimulus selection, sensory overload, derealization and brainfog are similar in the two conditions. The preliminary recommendation is to continue treatment with an SSRI for at least 1.5—2 years. More research is needed to support our hypothesis regarding the resetting of neurotransmitter systems by SSRIs. Another possibility could be that SSRIs only suppress symptoms because of the tryptophan deficiency. Contrary to this is the experience of two patients (the first one with a very low Bell score of 20 at the start of SSRI treatment) who discontinued the SSRI after 8 months and a year, respectively, because they felt completely healthy for 2 and 7 months, respectively. They continue to do well (seven and four months after discontinuation). However, most patients mention that, despite the good response to the SSRI, they must be cautious of not exceeding their limits. The PEM had become much less, but with too many activities in a row, a relapse—albeit shorter and less severe than before—could follow.