Some small imaging studies have shown brain damage after hospitalized people had recovered from severe COVID-19. Can milder bouts injure the brain, too? Yes, say scientists led by Gwenaëlle Douaud and Stephen Smith at the University of Oxford, U.K. In Nature on March 7, they reported that four months after a mild COVID infection, adults ages 51 to 81 had slightly thinner gray matter and more signs of tissue damage in their olfactory areas than they did before infection, even after accounting for age-related brain changes that showed up in uninfected controls. This is the first longitudinal study on the brain effects of COVID that compares MRI scans before and after infection.

  • With U.K. Biobank data, scientists compared brains pre- and post-COVID.
  • After mild infection, abnormalities appeared in limbic system structures connected to sense of smell.
  • Whole brain size shrank slightly, and cognitive decline accelerated.
  • A separate study from Wuhan found worsening decline one year after severe COVID.

Executive function also declined more rapidly after illness. How these brain structure and cognition changes could influence dementia risk remains an important question.

“Having two time points to tease out subtle brain changes is the beauty of this study,” Markus Glatzel, University Medical Center Hamburg-Eppendorf, Germany, noted. Jonas Hosp, University Hospital Freiburg, Germany, agreed. “This impressive, high-quality study analyzed a unique imaging cohort in an unbiased way to show that systemic infectious diseases, such as COVID, may indeed alter brain structure,” he said.

Studies on brain health after recovery from COVID-19 infection have uncovered transient cerebrovascular damage, brain hypometabolism, worsening of pre-existing neurological problems, and impaired cognition (Jan 2021 news; Apr 2021 conference news; Hosp et al., 2021). However, most of this data is from MRI or PET studies of small cohorts, and none had pre-COVID data from which to draw.

To address this shortcoming, first author Douaud turned to the U.K. Biobank, a massive effort to collect comprehensive health and genetic information on half a million people aged 40 to 69. The biobank includes an imaging substudy of 100,000 participants (Oct 2016 news). Since 2016, it has scanned the brains of almost 43,000 volunteers older than 45.

Douaud identified biobank participants aged 51 to 81 who had had COVID-19 between two imaging sessions. They tested positive an average of 4.5 months before their second scan. The scientists matched each person to an uninfected control by age, sex, ethnicity, and time between pre- and post-COVID scans, which averaged three years. The scientists compared rates of change among infected people with rates of change in matched controls to account for age-related alterations in the brain. They collected structural, diffusion-weighted, and resting-state functional MRI scans on 384 controls and 401 COVID cases. Of the latter, only 15 had been in the hospital; the rest recovered at home from a mild to moderate course.

The scientists used an algorithm to identify changes in more than 2,000 image-based phenotypes (IDPs). These could be shapes, sizes, or metabolic activities of regions of interest, for example, the volume of the putamen or the diffusion index of the anterior cingulate cortex. In calculating rates of change due to the infection, the researchers gauged the effects of 6,300 non-imaging variables at baseline. Variables included lifestyle factors, family health history, and COVID-19 risk factors, such as obesity, blood pressure, smoking status, and diabetes. None significantly influenced longitudinal brain changes, indicating that COVID was indeed to blame.

Then how did COVID alter the brain? Sixty-five of these 2000+ IDPs  changed after illness. Of the five with the most robust correlation, whole brain volume shrank slightly more in COVID cases compared to controls, cerebrospinal fluid volume increased, and the lateral ventricles widened. The other two phenotypes were more localized. Tissue microstructure was compromised in the fronto-occipital fasciculus, which connects with areas of white matter altered after COVID. The orbitofrontal cortex, anterior cingulate cortex, hippocampus, parahippocampal gyrus, and amygdala, saw a rise in their diffusion indices, a proxy for tissue damage. These structures comprise the piriform network that functionally connects to the piriform cortex, which interprets smell.

Blunted or complete loss of smell is a telltale sign of early COVID infection. To see what other brain change might explain this, the scientists focused on 297 image-derived phenotypes related to smell. They found that infection had altered 68, most notably the orbitofrontal cortex had thinned, and diffusion indices rose in five regions of the piriform and olfactory tubercle functional networks. The researchers were unable to measure changes directly in the piriform cortex or olfactory bulb because those tiny areas near the sinuses tend to be distorted in MRI images.

To analyze brain differences another way, the scientists tracked changes in whole-brain cortical surface thickness and mean water diffusivity before and after COVID, then compared the changes to those in controls. Again, many of the same areas identified by IDP analysis were altered by infection (see image below). Overall, the most consistent abnormalities were within the orbitofrontal cortex and the parahippocampal gyrus.

“This well-done study by first-class authors points to a distinct anatomy within the brain that is selectively vulnerable to COVID,” remarked Marcus Raichle, Washington University, St. Louis. The imaging abnormalities held even after excluding the 15 hospitalized participants, who had more severe atrophy and tissue damage than people with milder infections.

Covid Change. Compared to controls, gray matter thinned after COVID in areas related to sensory processing (orbitofrontal cortex and insula), memory (parahippocampal gyrus and temporal pole), executive function (anterior cingulate cortex), and language processing (supramarginal gyrus, left). Mean diffusivity, a measure of tissue damage, rose in the orbitofrontal cortex, insula, anterior cingulate cortex, and amygdala (right). In general, atrophy and tissue damage were worse in the left hemisphere. [Courtesy of Douaud et al., Nature, 2022.]

How severe was the damage? In COVID cases, volumes fell and diffusivities rose an average of 0.2 to 2 percent more than they did in controls. The authors deemed this moderate. The hippocampus, for example, shrinks about 0.2-0.3 percent per year in older adults (Fraser et al., 2021).

Did this affect cognition? Before and after infection, the participants took these six tests: Trail Making to assess executive function; Symbol Digit, Pairs Matching, and total number of digits recalled correctly to measure memory; the fluid intelligence test to estimate reasoning; and speed of matching cards to measure reaction time. Only the time needed to complete the Trail Making tasks A and B differed after COVID. Participants took 8 and 12 percent longer to complete task A, which is numeric, and task B, which is alphanumeric, respectively (see image below).

Douaud thinks this tardiness might reflect poor executive function and reaction time rather than memory, since there was no difference in the three memory tasks. Intriguingly, the scientists linked slower completion of trails B to greater atrophy of the crus II of the cerebellum. Though this major hindbrain structure primarily controls balance and motor function, scientists now believe the crus II helps with executive function.

Sluggish Cognition. Compared to controls (blue), people who recovered from COVID (orange) took up to 35 percent longer (y axis) to connect numbers and letters in the Trail Making A (left) and B (right) tasks. The case/control difference grew larger with age. [Courtesy of Douaud et al., Nature, 2022.]

And ... Dementia?
The dreaded question, whether these COVID-related changes increase the risk for dementia or Alzheimer’s disease, is still open. That said, “It is a remarkably important finding that, even in mild COVID patients, the brain changed in regions we know are related to cognition and brain disorders, such as AD,” Ole Isacson, Harvard Medical School, Boston, told Alzforum. Raichle agreed. “The minute you involve the hippocampus and cortex, you bring in the default mode network that is implicated in AD,” he said.

Douaud and colleagues did not address links to dementia, but other researchers have looked cross-sectionally at changes six to 12 months after recovering from severe COVID and found hypometabolism in some of the same regions Douaud identified (Kas et al., 2021; Blazhenets et al., 2021).

New hints regarding dementia come from a longitudinal study published March 8 in JAMA Neurology (Liu et al., 2022). Scientists tracked 1,438 adults more than 60 years old who were hospitalized with COVID during the initial wave in Wuhan, China. Of those, 260 had been deemed to have severe COVID, while the others had moderate disease. The survivors and 438 uninfected spouses took the Informant Questionnaire on Cognitive Decline in the Elderly, and the Telephone Interview of Cognitive Status-40, six and 12 months after they had left the hospital. After one year, 12 percent of survivors developed dementia or mild cognitive impairment compared to 6 percent of controls. People who survived severe illness were 19 times likelier to experience progressive cognitive decline over the year than were controls. Those who had moderate COVID were 1.7 times more likely to decline during the first six months of recovery, but then their cognitive function stabilized.

While the data seem troubling, Glatzel sees no reason for panic. “Overall, COVID has pretty mild effects on the brain—no massive encephalitis, no dramatic influx of dementia cases after COVID infection. In the few cases I've seen, neuropathology returns to baseline a few months after infection, at least as judged using standard techniques,” he said. Indeed, previous studies have detected elevated plasma markers of brain injury, such as neurofilament light and glial fibrillary acidic protein, as well as lower Aβ and higher total tau and phosphotau-181, after severe COVID. All normalized within six months of hospital discharge (Sep 2021 conference news).

Intriguingly, a recent preprint reported diffuse Aβ deposits in the cortices of 10 people under age 60 who died from severe COVID-19 (Rhodes et al., 2022). Unlike amyloid plaques of AD, these did not bind thioflavin-S, and the scientists found similar deposits in five people younger than 66 who had died of acute respiratory distress, unrelated to COVID. Costantino Iadecola at Weill Cornell thinks these deposits are not the amyloid plaques seen in AD. “These Aβ deposits may turn into true fibrillar plaques, or they could be a transient response to hypoxia and cardiac arrest that go along with dying,” he told Alzforum. Colin Masters, University of Melbourne, Australia, agreed. “Very metabolically active brain areas, where APP processing is highest, are particularly susceptible to hypoxic damage and dementia,” Masters noted.

Many questions remain about brain health after COVID. “The U.K. Biobank study sets the stage for continued understanding of the vulnerability of select brain tissues and the permanence of these suspected neurodegenerative changes,” Mony de Leon and Anna Nordvig of Weill Cornell Medical College, New York, wrote (full comment below). Will the greater atrophy reverse, stop, or persist long-term? Douaud hopes to find out by analyzing biobank participant scans again in one to two years.—Chelsea Weidman Burke


  1. This study adds to our understanding that COVID-19 exposure is related to brain changes detected on MRI. By comparing subjects pre- and post-exposure in a longitudinal study design with controls, the authors effectively make the point that the brain is impacted by the disease. This contribution is important as it sets the stage for continued understanding of the biological mechanisms for the impact on brain, the vulnerability of select brain tissues, and the permanence of these suspected neurodegenerative changes.

    We expect this study will pave the way for many additional ones. For example, because this study was not designed to demonstrate the presence of a causal agent in brain tissue or CSF, the proximal cause of the nonspecific MRI brain changes remains unknown. Is this due to viral invasion or peripheral immunological signaling resulting in pathways that include CNS inflammatory, autoimmune, and hypoxic changes in the tissues and vessels?

    Second, the finding of MRI signals for the anatomy related to the olfactory system is important, but the specificity of this observation needs to be explored. Not all COVID-19 affected subjects show olfactory deficits (unexamined in the present study) and other diseases and exposures are known to affect this anatomy.

    Third, this two-point longitudinal study, while provocatively suggesting neural degeneration, begs extension to later time points to examine the permanence of the damage.

  2. The paper is very interesting, since having such a large imaging biobank of 42,000 individuals is really a big data gold mine. Applying this strategy of gathering large biobanks is expensive and time-consuming, especially since at the time of collecting you don’t expect something like a pandemic to happen. In this case the data is well-suited to observing the effects of COVID-19 on the population.

    The change in brain volume detected was (as mentioned in the paper) modest, but still significantly more than the 0.2–0.3 percent of expected brain volume loss per year. Relating structural changes in the anatomy to functional changes (cognitive tests) can be difficult, and further evidence would be needed to draw the connection between the two. That regions of taste and smell appear affected supports a COVID connection, and anosmia/ageusia being a prevalent symptoms.

    It is surprising to see that so few of the patients were hospitalized, and that only one of the 401 patients was mechanically ventilated. It would be interesting to know how the brain is affected after critical illness with COVID-19. If such brain atrophy is already seen in mild cases, one could suspect even more atrophy in those more critically ill. The lack of data on non-COVID hospitalized patients is a limitation, as mentioned by the authors. This would have shed more light on how specific the findings are for COVID.

    It would be good to see the baseline data for the cognitive scores and brain volumes before COVID-19. This could have been informative as it would give an impression of the extent of mild cognitive impairment and early dementia in the cohort before some contracted COVID-19, and in this way could show the effect of pre-existent frailty on the risk of COVID-19-based brain atrophy.

  3. The elderly are especially vulnerable to the coronavirus SARS-CoV-2, the cause of COVID-19, which is characterized in some patients by a post-infective syndrome termed long COVID (Vimercati et al., 2021; Søraas et al., 2021). Many symptoms of long COVID are likely central nervous system (CNS) based, driven by neuroinflammation, blood-brain-barrier dysfunction, and oxidative stress (Pinna et al., 2020). The sickest of COVID-19 patients show a decline in global intellectual performance greater than that of the average stroke patient (Hampshire et al., 2021). Over 50 percent of patients with long COVID who are not hospitalized rank their symptoms as moderate to severe seven months after recovery from COVID-19 (Davis et al., 2021). The most frequent symptoms are fatigue, post-exertional malaise, and cognitive dysfunction, including brain fog, poor attention, impaired executive function (difficulty thinking and problem solving), and short- and long-term memory loss.

    To determine the impact of SARS-CoV-2 infection in milder cases on brain pathology, Douaoud et al. assessed brain changes in 785 U.K. Biobank participants aged 51–81. Of these, 401 tested positive for infection with SARS-CoV-2 between their two scans, with 141 days on average separating their diagnosis and second scan. In controls, 384 days separated their scans. Significant longitudinal effects of infection with SARS-CoV-2 included a greater reduction in gray-matter thickness and tissue contrast in the orbitofrontal cortex and parahippocampal gyrus, greater changes in markers of tissue damage in regions functionally connected to the primary olfactory cortex, suggesting a degenerative spread of the disease via olfactory pathways, a greater reduction in global brain size, and a worsening of executive function (requiring more time to complete trail A and trail B of the Trail Making Test). These MRI and executive function effects remained after removal of data from 15 hospitalized cases.

    However, there were no memory impairments detected. The lack of memory impairments might be due to the fact that mild to moderate cases were included in this study and because there were subtle baseline differences; while no single cognitive score was different at baseline between controls and future cases, two cognitive principal components were different, suggesting slightly lower cognitive abilities for the future cases when compared with the controls. The lack of memory impairments might also be due to the fact that risk factors for long-term cognitive decline in long COVID, such as apolipoprotein E4 (Manzo et al., 2021; Miners et al., 2020), sex (males at greater risk for infection) (Gebhard et al., 2020), and pre-existing Type 2 diabetes (associated with excess adiposity) are all risk factors for AD as well and were not included in the analyses.

    As part of follow-up studies, gastrointestinal distress, the presence of anti-type I Interferon (IFN) antibodies (Su et al., 2022; Bastard, 2022), and vaccination against measles, mumps, and rubella (Ashford et al., 2021; Gold et al., 2020), all implicated in modulating the response to COVID-19 exposure, would be good to consider. This might reveal general patterns across neurological diseases. For example, increasing evidence supports a role for alterations in the gut microbiome and the gut-liver-brain axis in healthy aging (Wilmanski et al., 2021), Parkinson’s disease, and Alzheimer’s disease (Dodiya et al., 2019; Sampson et al., 2016; Kundu et al., 2022; Kundu et al., 2021). Such studies are timely and warranted to assess whether the brain pathology and cognitive injury seen following moderate COVID-19 are transient or persist long-term.


    . Association between Long COVID and Overweight/Obesity. J Clin Med. 2021 Sep 14;10(18) PubMed.

    . Persisting symptoms three to eight months after non-hospitalized COVID-19, a prospective cohort study. PLoS One. 2021;16(8):e0256142. Epub 2021 Aug 26 PubMed.

    . Neurological manifestations and COVID-19: Experiences from a tertiary care center at the Frontline. J Neurol Sci. 2020 Aug 15;415:116969. Epub 2020 Jun 3 PubMed.

    . Cognitive deficits in people who have recovered from COVID-19. EClinica lMedicine, 2021 eClinical Medicine

    . Characterizing long COVID in an international cohort: 7 months of symptoms and their impact. EClinicalMedicine. 2021 Aug;38:101019. Epub 2021 Jul 15 PubMed.

    . Could COVID-19 anosmia and olfactory dysfunction trigger an increased risk of future dementia in patients with ApoE4?. Med Hypotheses. 2021 Feb;147:110479. Epub 2021 Jan 5 PubMed.

    . Cognitive impact of COVID-19: looking beyond the short term. Alzheimers Res Ther. 2020 Dec 30;12(1):170. PubMed.

    . Impact of sex and gender on COVID-19 outcomes in Europe. Biol Sex Differ. 2020 May 25;11(1):29. PubMed.

    . Multiple early factors anticipate post-acute COVID-19 sequelae. Cell. 2022 Mar 3;185(5):881-895.e20. Epub 2022 Jan 25 PubMed.

    . Why do people die from COVID-19?. Science. 2022 Feb 25;375(6583):829-830. Epub 2022 Feb 24 PubMed.

    . MMR Vaccination: A Potential Strategy to Reduce Severity and Mortality of COVID-19 Illness. Am J Med. 2021 Feb;134(2):153-155. Epub 2020 Oct 23 PubMed.

    . Analysis of Measles-Mumps-Rubella (MMR) Titers of Recovered COVID-19 Patients. mBio. 2020 Nov 20;11(6) PubMed.

    . Gut microbiome pattern reflects healthy ageing and predicts survival in humans. Nat Metab. 2021 Feb;3(2):274-286. Epub 2021 Feb 18 PubMed.

    . Sex-specific effects of microbiome perturbations on cerebral Aβ amyloidosis and microglia phenotypes. J Exp Med. 2019 Jul 1;216(7):1542-1560. Epub 2019 May 16 PubMed.

    . Gut Microbiota Regulate Motor Deficits and Neuroinflammation in a Model of Parkinson's Disease. Cell. 2016 Dec 1;167(6):1469-1480.e12. PubMed.

    . Fecal Implants From App NL-G-F and App NL-G-F/E4 Donor Mice Sufficient to Induce Behavioral Phenotypes in Germ-Free Mice. Front Behav Neurosci. 2022;16:791128. Epub 2022 Feb 8 PubMed.

    . Integrated analysis of behavioral, epigenetic, and gut microbiome analyses in AppNL-G-F, AppNL-F, and wild type mice. Sci Rep. 2021 Feb 25;11(1):4678. PubMed.

  4. On the back of the recent data strengthening the link between Epstein-Barr virus and multiple sclerosis, the authors present data linking mild COVID symptoms to reduced cognition and cortical thinning in regions associated with olfaction over a period of 140 days. Strikingly, visual inspection of the data suggests that older people showed greater loss of cognition and cortical thinning than younger people, independent of the severity of their COVID symptoms.

    Longer-term follow-up of these subjects will enable us to evaluate whether the COVID infection changes patients’ symptoms towards Alzheimer’s or Parkinson’s disease, perhaps within a few years. The enrollment of these cohorts within the U.K. Biobank also potentially enables the identification biomarkers and additional risk factors such as APOE status that could influence whether the cortical thinning and cognitive changes are temporary, permanent, or progressive. Perhaps the extent of hyposmia will be a telling measure.

    From the perspective of discovering and developing therapeutics, rodents rely heavily on their sense of smell and provide an excellent translational model for the interactions between inflammation and cell-type-specific vulnerability related to COVID infections and potentially dementia.

  5. The detection of diffuse Aβ deposits in the cortices of 10 people under age 60 who died from severe COVID-19 (Rhodes et al., 2022) is intriguing. At first pass, however, I would guess this is the consequence of altered Aβ drainage or microglia uptake. There is an increasing body of evidence that microglia function involves the clearance of aggregation-prone peptides like Aβ. Since viral infections are known to modulate the physiological function of microglia, the detection of diffuse Aβ deposits might simply be the consequence of reduced Aβ clearance.

    In the case of long-term respiratory distress, various factors from treatment may further compromise the function of microglia, including anesthetics and benzodiazepines. The later may cause an overactivation of microglia in dendritic spine pruning, as we have shown recently, and may be the underlying cause of cognitive changes (Shi et al., 2022).

    Diffuse Aβ may not affect cognition and thus contribute to persistent cognitive impairment in COVID survivors, but I believe that its extracellular accumulation during long-term intensive care treatment may lead to the formation of fibrillary Aβ seeds and a higher risk of developing AD 25 years later.


    . Β-Amyloid Deposits in Young COVID Patients. January 14, 2022 The Lancet Preprint

  6. In response to Charlotte Teunissen, Lisa Vermunt, and Patrick Smeele's comment that "It would be good to see the baseline data for the cognitive scores and brain volumes before COVID-19":

    We thank you for your comments. We have provided these baseline data. For imaging, it's in Supplementary Table 2; for non-imaging (including cognition), it's in Supplementary Table 4.

Make a Comment

To make a comment you must login or register.


News Citations

  1. How Does COVID-19 Affect the Brain?
  2. COVID-19 Worsens Neurological Problems, Delirium
  3. 5,000 Down, 95,000 to Go: U.K. Biobank Releases First Brain-Imaging Data
  4. Aβ, Tau, and Other AD Markers Altered in COVID

Paper Citations

  1. . Cognitive impairment and altered cerebral glucose metabolism in the subacute stage of COVID-19. Brain. 2021 May 7;144(4):1263-1276. PubMed.
  2. . Longitudinal trajectories of hippocampal volume in middle to older age community dwelling individuals. Neurobiol Aging. 2021 Jan;97:97-105. Epub 2020 Oct 21 PubMed.
  3. . The cerebral network of COVID-19-related encephalopathy: a longitudinal voxel-based 18F-FDG-PET study. Eur J Nucl Med Mol Imaging. 2021 Jul;48(8):2543-2557. Epub 2021 Jan 15 PubMed.
  4. . Slow but Evident Recovery from Neocortical Dysfunction and Cognitive Impairment in a Series of Chronic COVID-19 Patients. J Nucl Med. 2021 Jul 1;62(7):910-915. Epub 2021 Mar 31 PubMed.
  5. . One-Year Trajectory of Cognitive Changes in Older Survivors of COVID-19 in Wuhan, China: A Longitudinal Cohort Study. JAMA Neurol. 2022 May 1;79(5):509-517. PubMed.
  6. . Β-Amyloid Deposits in Young COVID Patients. January 14, 2022 The Lancet Preprint

Further Reading


  1. . Adapting the UK Biobank Brain Imaging Protocol and Analysis Pipeline for the C-MORE Multi-Organ Study of COVID-19 Survivors. Front Neurol. 2021;12:753284. Epub 2021 Oct 29 PubMed.
  2. . Long-term microstructure and cerebral blood flow changes in patients recovered from COVID-19 without neurological manifestations. J Clin Invest. 2021 Apr 15;131(8) PubMed.

Primary Papers

  1. . SARS-CoV-2 is associated with changes in brain structure in UK Biobank. Nature. 2022 Apr;604(7907):697-707. Epub 2022 Mar 7 PubMed.