Seeing through the fog: a neuroendocrine explanation for post-COVID cognitive deficits

Standfirst

Gonadotropin-releasing hormone is implicated in cognitive functions, and its loss is a factor in pathological brain ageing. There are similarities between these processes and the neurological and cognitive deficits observed in patients with long COVID. Here, we explore the hypothesis that neuroanatomical and transcriptomic alterations associated with long-COVID could stem from this neuroendocrine perturbation.

Several years after the start of the pandemic, high numbers of papers on COVID-19 continue to be published. Setting aside the scientific compulsion to understand the pathophysiology of a rather singular disease, we also have an increasing awareness of the economic, social and individual costs of long COVID, also known as post-COVID syndrome or post-acute sequelae of COVID. Long COVID has all the makings of a second, delayed but far more insidious, pandemic. Long COVID is associated with a range of symptoms, depending on the physiological functions or organ systems affected: respiratory, cardiovascular, immunological, gastrointestinal, endocrine and neurological¹. This broad range of symptoms is in keeping with the broad cellular and tissue tropism of SARS-CoV-2, which is made possible by the growing list of host proteins that enable it to infect its target cells. Indeed, cognitive dysfunctions, such as memory or attention deficits and ‘brain fog’, are among the most common and debilitating symptoms reported by patients with long COVID. In the past year, several studies have shown that SARS-CoV-2 neuroinvasion leads to persistent neuroanatomical, transcriptomic and functional changes reminiscent of accelerated or pathological brain ageing. In addition, these papers suggest that the infection or loss of certain hypothalamic cell types might be the linchpin linking reproductive and metabolic perturbations with impaired cognitive function.

The long-term endocrine consequences of COVID-19 are now well documented, with almost all endocrine organs and tissues found to be susceptible to SARS-CoV-2. For instance, study after study has revealed an increased incidence of new-onset type 1 diabetes mellitus and type 2 diabetes mellitus (for a comprehensive review of endocrine perturbations attributable to COVID-19, see²). However, in many cases, the hormonal profiles of the patients are reminiscent not of an infection of the peripheral organ but of a central impairment or infection of the relevant hypothalamic–pituitary axis. This seems to be the case in a subpopulation of male patients who contracted COVID-19 and who, regardless of the severity of the initial infection, display persistent or delayed new-onset hypotestosteronemia without the compensatory increase in gonadotropin levels that would be indicative of a functional hypothalamic–pituitary–gonadal axis³. Interestingly, in postmortem analysis of brain samples, we observed at least two invasion routes for the virus; an olfactory route and a haematological route through the hypothalamic median eminence. In all the samples, this neuroinvasion seems to lead to the damage and loss of an irreplaceable cell type that is in contact with both the nasal epithelium and the pituitary portal blood circulation underlying the median eminence: neurons producing the master reproductive hormone, gonadotropin-releasing hormone (GnRH). In association with these dead and dying neurons, the expression of GnRH was also greatly affected in all the postmortem patient brains examined. This phenomenon likely explains the hypogonadotropic hypogonadism observed in living patients. Additionally, we also observed the widespread infection of tanycytes in the median eminence³; these cells regulate GnRH release into the pituitary portal circulation and replace the blood–brain barrier in this region. Tanycytes also actively transport several metabolic hormones to hypothalamic regions that regulate the central response to these peripheral signals. Indeed, in our cohort of patients with long COVID, patients who had persistent hypothalamic dysfunction also experienced perturbed body weight homeostasis.

Interestingly, the loss of GnRH is also implicated in the progressive cognitive decline associated with Down syndrome and seen in THY::TAU22 mice, an animal model of Alzheimer disease⁴; both these conditions are also characterized by metabolic dysfunction. Hypophysiotropic GnRH neurons are also known to project to brain areas involved in cognition, which express the GnRH receptor. Intriguingly, replacing physiological levels and patterns of GnRH led to improved cognitive performance in people with Down syndrome and in mouse models of both Down syndrome and Alzheimer disease. This approach also improved functional brain connectivity in patients and normalized the electrophysiological and transcriptomic characteristics (including the expression of myelin-related genes) of the hippocampus of these mice⁴. Together, these studies suggest that the SARS-CoV-2-mediated loss of GnRH neurons, or merely a reduction or dysregulation of GnRH expression, might result in reduced hippocampal connectivity (and thus volume), altered myelination and gene expression reminiscent of neurodegenerative conditions such as Alzheimer disease, and cognitive deficits. These projected changes correspond almost exactly to the findings of three of the latest long COVID studies.

A Spanish study of 84 patients 1 year after the COVID-19 infection and analysis of postmortem tissue samples from seven patients who died during an acute COVID-19 infection revealed a reduction in the volume of multiple hippocampal areas and associated brain regions implicated in cognition. Other grey and white matter abnormalities were also seen, as well as several cellular and molecular changes, including axonal degeneration, reactive astrogliosis and loss of myelin-related proteins⁵. Similar changes have been observed in other studies. For instance, a UK Biobank prospective study of 401 individuals who contracted COVID-19 and 384 controls, at a shorter post-infectious interval of around 5 months on average, revealed a reduction in overall brain size as well as grey matter thickness in limbic areas that correlated with longitudinal cognitive changes⁶. Similar structural changes could underlie functional deficits found to be equivalent to 10 years of cognitive ageing observed even 2 years after the acute infection in another UK-based cohort study⁷. Interestingly, the cellular and gene expression profiles of the frontal cortex and hippocampus of individuals who died during a SARS-CoV-2 infection were remarkably similar to those observed in the postmortem brains of patients with Alzheimer disease⁸. These findings reflect the molecular signatures reminiscent of ageing observed in the postmortem brain samples of patients with severe COVID-19 in a previous study⁹. While a wide variety of mechanisms and aetiologies have been postulated to explain these structural and functional consequences of SARS-CoV-2 infection, the possibility that these changes, in brain regions known to be GnRH-sensitive, might be the repercussion of dying or dysfunctional GnRH neurons cannot be ruled out.

Finally, the importance of the first postnatal activation of GnRH neurons in infancy, an event known as minipuberty, in the establishment of both cognition and other brain functions is also becoming increasingly evident ¹⁰. In this context, we observed in human fetal tissue and cell lines derived from embryonic GnRH neurons that these neurons express several viral entry factors and seem to be susceptible to SARS-CoV-2 infection from the embryonic period ³. GnRH neurons originate in the nose during embryonic development before migrating to their final destinations in the brain and continue to maintain a connection with the nasal epithelium. They thus represent a vulnerable cell population both from the point of view of maternal–fetal transmission of the virus and postnatal or paediatric infections, which could result in serious neurodevelopmental consequences. Studies of long COVID in children and adolescents have been understandably sparse, and endocrine investigations have taken into account the lockdown period as a whole rather than comparing infected and non-infected children. However, birth cohorts or paediatric cohorts of children from the pandemic period known to have been infected could be monitored for the appearance of such consequences, which could run the gamut from cognitive deficits to metabolic dysregulation to alterations of puberty and fertility, given what we now know about the role of this small population of neurons.

The coming years will reveal to what extent long COVID mimics the cellular, molecular and cognitive changes observed in pathological brain development or ageing, including Alzheimer disease, or in the breakdown of the central control of bodily homeostasis. Further research could also reveal whether therapeutically replacing physiological levels of a reproductive hormone, GnRH, during particular temporal windows could stop or slow these changes, thus limiting the wave of non-infectious disorders that could appear.