Unraveling Links between Chronic Inflammation and Long COVID: Workshop Report

The effects of an acute COVID-19 infection may remain in some people for weeks, months, or longer. The term “long COVID” was coined in a tweet by Elisa Perego in May 2020 (1) to describe these long-term effects. According to the Centers for Disease Control and Prevention, long COVID is “signs, symptoms, and conditions that continue or develop after acute COVID-19 infection” (https://www.covid.gov/be-informed/longcovid/definitions). The National Academies of Sciences, Engineering, and Medicine’s definition adds persistence of symptoms after onset: “signs, symptoms, and conditions present four weeks or more after the initial phase of infection; may be multisystemic; and may present with a relapsing–remitting pattern and progression or worsening over time, with the possibility of severe and life-threatening events even months or years after infection” (https://www.nationalacademies.org/our-work/examining-the-working-definition-for-long-covid). Recently, it has been suggested that because long COVID’s systemic and multiorgan effects follow a viral infection, it should be named postacute sequelae of SARS-CoV-2 (PASC).

The estimated incidence of long COVID is 1% among COVID-19 survivors, and more than 1 million Americans live with it today (2). Symptoms commonly include lingering loss of smell and taste, hearing loss, extreme fatigue, and “brain fog.” Persistent cardiovascular and respiratory problems, muscle weakness, and neurologic issues have also been documented. Some studies suggest that long COVID is due to prolonged inflammation after SARS-CoV-2 infection; however, the underlying mechanisms remain unclear. In September 2023, the trans–National Institutes of Health (NIH) Chronic Inflammation Working Group brought together experts worldwide for a 3-d virtual workshop to discuss current research about long COVID (https://events.cancer.gov/nih/inflammation-long-covid). Definitions and classifications of long COVID, potential mechanisms for its multiorgan effects, the role of inflammation and immune system dysregulation, comorbidities that might contribute to it, and potential treatment strategies were discussed.

Long COVID consists of many potentially overlapping entities with different biological causes, sets of risk factors, and outcomes. The lack of biomarkers for diagnostic testing, the wide variety of multiorgan and associated symptoms involved, the variation in the SARS-CoV-2 virus, its manifestations and management strategies over time, and the duration and persistence of symptoms have complicated the development of a standard definition, diagnosis, and treatment of long COVID. Josh Fessel summarized the ongoing research programs and activities related to COVID-19 at the NIH and unmet clinical needs. The NIH has a significant research commitment for long COVID, with 29 active projects and total funding of $735 million as of August 2023. The Researching COVID to Enhance Recovery (RECOVER) program is the largest trans-NIH initiative to comprehend predictors, mechanisms, diagnostics, treatments, and prevention strategies for long COVID (https://recovercovid.org/). It includes observational and mechanistic studies and clinical trials by multiple teams nationwide. Fessel emphasized that long COVID research faces an unprecedented challenge compared with acute COVID-19 due to its diverse clinical manifestations and a prolonged interval between exposure and onset of the condition with no animal models and specific testing. The virus’s evolving nature and changes in prevalent strains, advances in acute COVID-19 treatments, and the diverse medical histories of affected individuals further complicate the understanding and treatment of long COVID.

Efforts to comprehend the origins of long COVID include studying the direct viral damage to many tissues with downstream effects, viral sanctuary sites, and dysregulated inflammatory responses. Chronic inflammation is particularly interesting because it may explain the variable nature of long COVID’s manifestations. It offers a framework to unify hypotheses and explore treatment options (e.g., association of many inflammatory genes with the cellular response to COVID-19) (https://www.genome.jp/pathway/hsa05171). An increasing number of clinical trials on long COVID, including patient-led research collaborative (3) and acute COVID-19 studies, continue to add to our understanding of long COVID. As such, there are reasons for optimism in addressing the challenges associated with long COVID (https://erictopol.substack.com/p/a-break-from-covid-waves-and-a-breakthrough).

There are significant challenges in developing a unified classification for long COVID. Amy Proal described the viral reservoir theory that the SARS-CoV-2 virus replicates and maintains its RNA in various tissues (4) long after infection, rendering blood-based clinical tests for diagnosing long COVID challenging (5, 6). The virus may persist in the brain stem, spinal cord, tongue, olfactory epithelium, and tonsillar tissue (7, 8), and such viral reservoirs may downregulate the immune response and reactivate latent pathogens, leading to immune dysregulation, microbiome dysbiosis, and epithelial barrier permeability (4).

Melissa Haendel described the effort to use real-world data in the National COVID Cohort Collaborative dataset (https://covid.cd2h.org/about/). This is difficult due to differences in record quality in the electronic health records (EHRs) and inconsistent testing for long COVID (9). A machine learning algorithm based on patient characterization and data clustering was developed to identify potential patients with long COVID, using features such as dyspnea, fatigue, vaccination status, and number of outpatient visits (10). Different subtypes of long COVID identified by clustering features include neurologic, metabolic, and cardiopulmonary subcategories, suggesting that long COVID is a heterogeneous disease with immunological relationships. This approach can be used in any EHR and can help in identifying patients for interventions in clinical trials (11).

Stratification of patients with long COVID using pathophysiologic and immune features may help to reveal potential biological mechanisms contributing to the pathogenesis of long COVID. Upinder Singh reviewed work on defining the pathophysiology of long COVID subtypes using the RECOVER data. Almost 26,000 people, demographically representative of the U.S. population, have been enrolled (as of September 2023) in RECOVER’s prospective 4-y study. Singh described RECOVER’s long COVID cohort of 13,754 adult participants with >6 mo of follow-up (12), identifying 37 symptoms, with postexertional malaise, brain fog, and fatigue being the most prominent. Further work includes human subject retention, refining symptoms over long follow-up, integrating laboratory and imaging data, focused studies to define pathophysiology and mechanisms, and integrating results from other ongoing studies.

Evaluating the durability of immune memory and serum proteomic changes among a cohort of individuals infected at the start of the outbreak was presented by Julie McElrath. These individuals included outpatients with mild to moderate infections, first responders, and others with occupational and/or other known exposures associated with long COVID. In this cohort, 25% exhibited faster IgM and IgA Ab responses but similar memory B cell and T cell responses compared with individuals who had recovered from the infection. Serum proteomics revealed five participant clusters, highlighting the presence of inflammatory and noninflammatory subsets in patients with long COVID. The inflammatory signatures persisted in the long COVID subset, correlating with higher clinical activity scores and the potential for tailored treatment based on inflammatory profiles, particularly in individuals with higher body mass indices (BMIs) and older ages. Future studies will follow the longitudinal cohorts to elucidate the immune memory and immune signatures of long COVID and to understand the effects of the variants and responses upon potential treatments and eventual recovery.

Although SARS-CoV-2 is one of the most studied viruses, more research is needed to better understand the long-term impact and the chronicity of systemic effects following an initial SARS-CoV-2 infection. Speakers emphasized that potential mechanisms in the development of long COVID may include viral persistence; viral protein–induced immune response, cytokine production, and inflammation; cell death and inflammasome activation; adaptive immunity exhaustion; autoimmunity; altered virus-specific T and B cell differentiation; elevated (or abnormal) levels of the coagulation factors (13); and altered metabolic, genetic, and epigenetic factors. Dysregulation of vagus nerve and brainstem signaling has also been implicated in long COVID, affecting autonomic functions and behavioral responses (14).

Saurabh Mehandru described patients reporting a range of gastrointestinal (GI) symptoms, including loss of appetite; abdominal pain; nausea; and signs such as weight loss, altered bowel motility patterns, and new or exacerbated irritable bowel syndrome (15). Intestinal viral reservoirs are possibly associated with attenuated inflammation and lessened “sterilizing protection” during the acute phase of infection as SARS-CoV-2 may persist in the GI tract well beyond the resolution of clinical disease (16), as suggested by prolonged fecal shedding of SARS-CoV-2 in patients with long COVID (17). Although Mehandru’s group could not culture live virus, innate and adaptive immune cell abnormalities persist in the intestinal mucosa of COVID-recovered individuals (18). This led to the conclusion that delayed resolution of inflammation, autoimmunity, and viral persistence may all contribute to the pathogenesis of GI symptoms observed in long COVID (19).

Long COVID encompasses various conditions, each exhibiting various clinical manifestations across multiple body systems, as highlighted by Joseph Bellanti. The condition has affected millions of people worldwide, potentially leading to lifelong disabilities (20). Bellanti reiterated numerous mechanisms contributing to the pathogenesis of long COVID, including the persistence of the virus (4) or the spike protein (21), reactivation of latent viruses, viral superantigen activation of the immune system, multitissue damage, autoimmunity, and disruption of the gut microbiota (22). Variations in elements of the innate immune system among individuals could serve as a critical factor in the diverse progression of the disease (20). Genetic susceptibility to SARS-CoV-2 and epigenetic reprogramming of hematopoietic stem and progenitor cells may also play a role in developing long COVID (23).

Steven Josefowitz described how SARS-CoV-2 triggers epigenetic regulation, resulting in lasting cellular memory of inflammation in hematopoietic stem cells and innate immune cells (24). In survivors of severe COVID-19, persistent epigenetic alterations in circulating monocytes have been observed, leading to heightened inflammatory responses lasting from months to 1 y. Blood progenitors after severe SARS-CoV-2 infection displayed skewed myelopoiesis, fostering increased production of inflammatory monocytes and cytokines. IL-6 plays a partial role in this process, and IL-6R administration at acute infection can reduce inflammation in long COVID. These data have potential implications for understanding and managing the inflammatory state observed in patients with long COVID. Indeed, in the discussion, the presenters noted the link between IL-6–dependent epigenetic programs observed in blood progenitors and the persistent increases in IL-6, TNF, and IFN-γ signatures observed in a subtype of inflammatory long COVID (25). Future studies should address whether the persistence of these cytokines is linked to inflammatory hematopoiesis, contributing to chronic inflammation and clinical features of long COVID.

Eleni Gavriilaki presented clinical evidence from the early phase of the pandemic suggesting that severe SARS-CoV-2 infection could lead to endothelial damage and thrombocytic complications contributing to multiorgan damage (26, 27). Excessive complement activation is a key driver of COVID-19 immunothrombosis and severity (28), and the SARS-CoV-2 spike proteins can directly activate the alternative complement pathway (29). Gavriilaki’s team analyzed genetic and clinical data from patients hospitalized with COVID-19 infection and found variants associated with complementopathies and thrombotic microangiopathy in patients with severe COVID-19. A rigorous algorithm developed by the team could predict intensive care unit hospitalization and sex differences in disease susceptibility (30). A novel artificial neural network using five easily accessible laboratory indices to predict early intensive care unit hospitalization in patients with COVID-19 has been developed (31). Such artificial intelligence–based models and algorithms could be used to predict high-risk patients who would benefit from prophylactic or preemptive treatments. Longitudinal multiomics studies of patients with COVID-19 further reveal associations of long COVID with specific autoantibodies, viremia, and type 2 diabetes (32).

Richard Becker discussed how SARS-CoV-2 infection triggers a distinct acute and multiorgan inflammatory response. The severity of acute COVID-19 symptoms is now thought to be directly correlated with inflammation markers (33, 34), because inflammation is common in multiorgan systems during viral infection (33, 35). The COVID-19 Multiomics Blood Atlas Consortium reported that immune signatures in early COVID-19 infections were organ- and tissue-specific with the hallmarks of inflammatory cells and inflammasome networks (36), corroborated by COVID-19 tissue atlases (37). In addition to inflammation, immune dysfunction, microbial dysbiosis, autoimmune and immune priming, clotting and endothelial dysfunction, and neurologic signaling dysfunction occur in long COVID (3). In a longitudinal study of patients diagnosed with COVID-19, inflammatory markers and autoimmune Abs persisted for at least 3 mo, followed by a change in plasma biomarkers (32), underscoring the need for long-term follow-up to understand the pathobiology of long COVID. Moreover, these persistent inflammatory signatures were influenced by the viral load, the autoantibodies formed, and preexisting conditions (32). All these observations suggest that early tissue injury contributes to long COVID conditions, especially in some patients in whom proinflammatory mediators persist and participate in long COVID phenotypes.

Post-COVID-19 neurologic sequelae include cognitive, mood, and sleep disorders; dysautonomia; diverse pain syndromes; exertional intolerance; and premature fatigue. Pathogenic mechanisms for neurologic symptoms include blood–brain barrier (BBB) and blood–nerve barrier disruption, perivascular inflammation, microglial activation, and neuronal damage, with possible links to mast cells and spike protein persistence.

Raimund Helbok emphasized that proving the causality for neurologic disorders after SARS-CoV-2 infection remains challenging because virus persistence has been proved only in small numbers of autopsy cases, and certain neurologic manifestations such as headache, memory deficits, and difficulty in concentrating are observed under various conditions and are nonspecific (38). Efforts to understand the prevalence, underlying mechanisms, and outcomes of these COVID-19–related neurologic manifestations (39) are proceeding to establish global consensus and harmonize data elements (40). Changes in brain structure have also been associated with SARS-CoV-2–induced neuroinflammation (41), and cognitive assessments revealing residual cortical dysfunction and cognitive impairment in patients have been observed (42). Long-term observations are needed to understand the reversibility of these changes and identify potential long-lasting effects of SARS-CoV-2 infection on the peripheral nervous system and CNS.

Avindra Nath noted that persistent viral infection, immune dysregulation, or a combination of both likely causes the pathophysiology of neurologic disorders after COVID-19. One possible mechanism could be that anti-idiotypic Abs formed in response to Abs to the SARS-CoV-2 spike protein activate endothelial cells, which then disrupt tight junctions, leading to fibrinogen leakage across the BBB, resulting in neuroinflammation and neurodegeneration. Breakdown of the BBB is evidenced by the presence of perivascular fibrinogen found in the brain in autopsy studies of patients with COVID-19 (43). Thus, complement-dependent Ab-mediated damage to endothelial cells results in platelet aggregation, thrombus formation in the blood vessels, and leakage of serum proteins into the brain parenchyma. This leads to the infiltration of macrophages, the release of cytokines, astrogliosis, and microglial cell activation and neuron damage via neuronophagia (44). Questions that need addressing include whether similar microvascular pathology and neuroinflammation occur in the brains of patients with long COVID, what role is played by the persistence of viral Ags, and whether immunotherapy can modify the course of the illness.

Anne Louise Oaklander discussed the mechanisms, diagnosis, and management of post–COVID-19 peripheral neuropathies. Small-fiber polyneuropathy appears to be the most common because small fibers preferentially bind SARS-CoV-2 and interact with immune cells (45). Studies show that symptoms of long COVID peripheral neuropathies mostly follow mild COVID-19 illnesses (46) and reflect persistent nerve inflammation rather than infection. Because evidence most often implicates dysimmunity in patients without classic neuropathy (initially idiopathic small-fiber polyneuropathy) (47), trials of immunotherapies proved effective for other neuropathies, such as corticosteroids and pooled Igs, are increasingly being considered for patients with long COVID with confirmed neuropathies.

Theoharis Theoharides presented information about the SARS-CoV-2 spike protein, neuroinflammation, and brain fog. Patients with long COVID show symptoms that fail to modulate inflammatory pathways (48). Neurovascular inflammation, activation of microglia, and neuronal damage may result from the SARS-CoV-2 spike protein stimulating the brain perivascular mast cells (49) to secrete proinflammatory molecules that activate microglia (50), potentiating neuropsychiatric symptoms, fatigue, and brain fog after COVID-19. Emerging evidence also suggests that the flavonoid luteolin and its analog (eriodictyol) significantly inhibit the stimulatory effects of the spike protein on the activation of mast cells and microglia and reduce the brain fog effects of COVID-19 (49, 51).

James Morrissey summarized studies on clotting issues seen in patients with COVID-19. Early in the COVID-19 pandemic, what appeared to be hyperinflammation and abnormal coagulation profiles resulted in a prothrombotic state. Many patients had strikingly high d-dimer plasma levels (52). Normally, fibrinolysis is the source of plasma d-dimers, but in patients with COVID-19, paradoxically, there is evidence of diminished fibrinolysis. Similarly, mechanisms connecting acutely elevated d-dimers to long COVID are undetermined. COVID-19–associated coagulopathy (CAC) is a life-threatening complication of SARS-CoV-2 infection, and its underlying mechanisms are only partially understood, although the fibrinolytic and complement systems appear to be key players (53). Under normal conditions, the pro- and antithrombotic systems are tightly regulated. Infection with SARS-CoV-2 appears to dysregulate this delicate balance, leading to a prothrombotic state in some patients (54). The causes of the prothrombotic state in patients with long COVID seem to be different from the classical disseminated intravascular coagulation with significantly elevated plasma fibrinogen levels and d-dimers in patients even 6–12 mo after COVID-19 hospitalization. Some evidence supports the involvement of complex interactions between the innate immune response, coagulation, fibrinolytic pathways, and the vascular endothelium, resulting in a procoagulant condition. Overall, hyperinflammation appears to be associated with CAC through a variety of mechanisms involving the alveolar epithelium, endothelial dysfunction, complement activation, monocytes/macrophages, cytokine storm, NETosis, platelets, and prothrombotic autoantibodies contributing to microclot formation and possible cognitive deficits. Thus, acute CAC may set the stage for long COVID.

James Harker discussed the role of the immune system and inflammation in lungs after COVID-19 and its link to ongoing respiratory symptoms, the immunoproteomic landscape of the post–COVID-19 lung, and the long-term outlook for patients with long COVID. There are lingering respiratory effects of wide diversity and severity in patients with long COVID, including breathlessness, cough, interstitial lung disease, and fibrosis, with more pronounced symptoms observed in hospitalized patients (55). Among these patients, objective assessment of lung anatomy and pulmonary function, such as computed tomographic imaging, forced expiratory volume, total expiratory volume, forced vital capacity, and gas transfer tests exhibit significant functional loss in lungs 3–6 mo after infection (56–61). Systemic (blood) and local (bronchoalveolar lavage) immunophenotyping of patients in an observational study also revealed persistent increases in both innate and adaptive immune cells 3–12 mo after infection (57). Persistent changes were observed in the airway immune microenvironment, and a strong correlation between airway proteome alterations, particularly in immune regulation, elevated cell death, and tissue repair, with ongoing cell damage and repair, was seen in patients with long COVID (62, 63). Still, more studies are needed to unravel the role of molecular and cellular alterations in the resolution or long-term effects of COVID-19.

James Stone presented information about the direct involvement of cardiac pathology in long COVID. Multiple studies have identified different pathologies, particularly in patients who had not been vaccinated and/or had not received targeted antiviral therapies. During the acute phase, pathological changes in the heart included different forms of acute myocyte injury, such as right ventricular strain injury or acute ischemic injury, with heart tissues exhibiting histological characteristics of myocyte necrosis that was confirmed by myocyte bound complement activation product, diffuse macrophage infiltration, multifocal lymphocytic myocarditis, focal pericarditis, and endocardial or small vessel thrombi (64). The myocardium (as well as blood) was positive for the virus in 60–100% of patients who died of severe COVID-19 infection, with some degree of inflammation in the heart muscle, either full-fledged myocarditis or lower levels of inflammation not reaching the level of myocarditis (34, 65–68). Additionally, myocardial inflammation appears to strongly correlate with the disease duration in a virus-dependent fashion in the extended acute phase (68). These pathological changes are evident in both acute and postacute phases, albeit at lower frequencies in the latter. Additional pathologic changes specifically relevant to the postacute phase include myocardial replacement fibrosis and small vessel arteriosclerosis. Although the level of virus in the heart is relatively low in the postacute stages, viral persistence in the myocardium may play a role in some patients with long COVID (69). When assessing cardiac viral persistence in long COVID, the contributions of both reinfections, even if asymptomatic, and vaccine protein and/or mRNA persistence (70) should be considered. Because confounding factors may exist, the pathologic and molecular changes discussed should be validated in large studies.

Edward Conway’s presentation focused on the complement system, a host defense mechanism against foreign substances that may become harmful if overly activated. SARS-CoV-2 triggers complement activation through multiple pathways, potentially damaging tissue (53). Complement hyperactivation is implicated in long COVID, as evidenced by increased activity in patients (71); however, the underlying molecular and cellular mechanisms need further elaboration. Complement proteins and fragments play roles in cellular processes and T cell functions, with implications in long COVID, potentially involving tissue-resident memory T cells (TRMs) and their cytotoxic actions (72), possibly contributing to syndromes such as brain fog and chronic respiratory disease (73). In an imbalanced environment, reactivation of TRMs may lead to excessive inflammation and local organ damage, suggesting potential therapeutic targets modulating the complosome pathway and TRM responses.

The impact of long COVID on aging was highlighted by Shabnam Salimi. Chronic inflammation in the elderly (inflammaging) is thought to be a precursor to various chronic diseases and reduced overall system function. COVID-19 expedites age-associated inflammation and influences hallmarks of aging such as inflammation control, autophagy, and genomic stability (74, 75). Aging research observations are relevant to symptoms/syndromes seen in long COVID because persistent inflammation seen in long COVID can mimic the senescence-associated secretory phenotype, exacerbating frailty and multimorbidity.

Divaker Choubey discussed the parallels between autoimmune diseases and long COVID, underlining common risk factors such as genetics, sex, and aging. Both conditions involve autoantibodies, a type I IFN signature, and chronic systemic inflammation associated with senescent cell accumulation. Aging and sex are significant risk factors for COVID-19 and long COVID and are linked to increased ACE-2 viral receptor expression. These common pathways lead to altered cellular communication, mitochondrial dysfunction, epigenetic changes, and telomere attrition. The “priming” of the immune system and a higher load of aged cells in certain individuals with autoimmune disease may increase the risk of severe long COVID symptoms. Targeting senescent cell elimination could be a potential therapeutic approach to enhance the health of patients with long COVID.

Charisse Madlock-Brown discussed her analysis of selection criteria and symptom representation used for long COVID as part of the RECOVER initiative that uses COVID-19–related data from the National COVID Cohort Collaborative. Her study evaluated these factors’ impact and modeled the selection criteria’s effects on research outcomes. Identifying long COVID cases has posed challenges due to potential biases and varying data collection methods, influencing symptom representation. EHRs are limited by data availability and biases tied to patient access and healthcare use that affect the generalizability of such long COVID studies.

Charis Bridger Staatz presented the analysis of EHR from two British birth cohorts, the National Child Development Study and the 1970 British Cohort Study, to understand the relationship between COVID-19 severity and BMI. Evidence shows that BMI is related to long COVID (76). A meta-analysis of 208 studies found a linear dose–response relationship between BMI and hospital admission and deaths caused by COVID-19 (77). Staatz explained that obesity is considered a state of low-grade inflammation, and a higher BMI over the life span may increase the risk of disease later in life, including COVID-19 infection, illness severity, and long COVID (78).

Dhruv Khullar discussed disparities in long COVID and possible risk factors such as viral exposure, socioeconomic status, and access to quality medical care that might have caused a higher prevalence of new or lingering symptoms after acute COVID-19 among Black and Hispanic patients in New York City as compared with White patients. He presented data analyzed from the EHRs of >60,000 patients in New York City health care systems with information on race/ethnicity, baseline comorbidities, severity of initial illness, and long-term outcomes. Black and Hispanic patients had a higher range of potential long COVID symptoms and conditions, including diabetes, chest pain, headache, and joint pain. Hospitalized patients had higher rates of incident symptoms, mainly respiratory, circulatory, and nervous system disorders; joint pain; and fatigue. Recent data from other parts of the United States are consistent with these findings. For example, data from the U.S. Census Bureau suggest that Black and Hispanic individuals had the highest rates of lingering symptoms after acute COVID-19 among racial/ethnic groups (79).

Linda Geng discussed viral persistence in long COVID and therapeutic trials involving antiviral(s), mAbs, and other pharmacologic agents, presenting data from the Selective Trial of Paxlovid for PASC (STOP-PASC), a randomized, double-blind, placebo-controlled study. This trial aims to assess the efficacy and safety of nirmatrelvir/ritonavir in treating long COVID symptoms and explore the potential utility of both physiological and digital biomarkers (through wearables) of long COVID. Challenges in the clinical care of long COVID include a heterogeneous spectrum of long COVID manifestations, functional debilitation in many patients, its unclear pathophysiology, and the lack of biomarkers and established therapies (80). In the Stanford Long COVID Clinic cohort, patients presented 12 symptoms, on average, at the initial visit, reflecting the complexity of the condition in clinical evaluation (81).

Amy Arnsten presented clinical studies using guanfacine and n-acetyl cysteine (NAC) to treat cognitive deficits in long COVID. Her studies suggest that brain fog, a well-documented effect consisting of several debilitating cognitive deficits, is likely due to the impaired dorsolateral prefrontal cortex (dlPFC) function. This brain region is essential for higher cognitive tasks such as abstract reasoning, working memory, memory recall, and executive functions and is highly vulnerable to inflammatory insults. dlPFC has unusual molecular requirements, including reliance on N-methyl-d-aspartate and nicotinic α7 receptor neurotransmission, which can be disrupted by kynurenic acid (KYNA) inflammatory signaling (82), making the region highly vulnerable to inflammatory insults. High levels of KYNA markedly reduce the neuronal firing of dlPFC neurons needed for working memory. Glutamate carboxypeptidase II inflammatory signaling further reduces firing by dysregulating potassium channel opening. Studies reveal that NAC and α-2A adrenoceptor agonists such as guanfacine can restore dlPFC function in aged macaques with naturally occurring KYNA- and glutamate carboxypeptidase II–mediated inflammation. These effects are thought to be through anti-inflammatory actions, including inhibition of KYNA production and restoration of potassium channel regulation in dlPFC (83). Arnsten’s studies indicate that NAC and guanfacine reduce symptoms of brain fog in patients with long COVID (84).

Colin Berry summarized the CISCO-21 study, a randomized clinical trial examining the feasibility, safety, and effectiveness of resistance-based exercise for mitigating long-term physical and psychological effects in patients with COVID-19 in various stages of recovery. The clinical trial started enrolling participants in June 2021 and tests personalized, simple movements to enhance muscle strength and aerobic capacity (85).

COVID-19 vaccines are highly effective in preventing severe disease and may lessen the severity of long COVID (86). Therefore, widespread vaccination is vital. Developing specific diagnostic tests, relying on clinical assessments, laboratory tests, and patient history for its diagnosis, is needed because the absence of standardized diagnostic criteria and fragmented data from health care settings hinders understanding and management of long COVID. The absence of specific classification criteria and biomarkers makes chronic inflammation from long COVID challenging, and exploring inflammatory markers during the acute phase of infection may offer insights into long COVID. Similarities with other postinfection syndromes suggest the potential existence of SARS-CoV-2 reservoirs, emphasizing the need for dedicated autopsy repositories for research. Studying viral RNA persistence mechanisms can shed light on long COVID’s biological mechanisms. Further work is needed to understand the potential of SARS-CoV-2 latency and its implications for immunomodulatory therapies. Molecular-level understanding and precise markers for subtypes and risk factors are crucial. Ongoing research is investigating the involvement and impact of CAC and the potential roles of RBCs in dysregulated fibrinolysis in long COVID. Differentiating COVID-19–related lung fibrosis from other comorbidities and preexisting lung conditions that heighten the risk of long COVID warrants further research. Chronic infections, including prior or concurrent COVID-19 infections, may influence cardiomyopathy. Quantifying the relationship between comorbidities and clinical outcomes is essential for future health system planning.

Future studies on long COVID must consider how cases are being defined (e.g., Are these people hospitalized with acute COVID-19? Did they need to have a positive COVID-19 test result in medical records, or was a positive home test used?) and the heterogeneity of populations being studied (e.g., race, age, sex, geographic area, severity of initial illness, baseline health status, access to timely care, timeline of follow-up, multiple clinical presentations).

Long COVID is heterogeneous, with unclear pathophysiology and functional debilitation in many patients. It is likely a syndrome of syndromes caused by a combination of biological mechanisms such as chronic inflammation, persistent viral infection, immune dysregulation, microbiota dysbiosis, blood clotting and endothelial abnormalities, and dysfunctional neurologic signaling. The clinical diagnosis of long COVID is challenging because of the various symptoms observed and the lack of diagnostic biomarkers. Although more research is needed, this meeting underscored the importance of considering the role of chronic inflammation in long COVID while developing biomarkers for diagnosing and understanding the causes of long COVID, designing clinical trials including nonpharmacological interventions (such as exercise and diet), and conducting well-thought-out epidemiologic and basic biologic studies.

The views expressed are the authors’ own and do not necessarily represent the views of the National Institutes of Health or the U.S. government.

The authors have no financial conflicts of interest.

We sincerely thank all speakers for their thoughtful contributions to the workshop. List of the speakers: Josh Fessel, M.D., Ph.D. (National Center for Advancing Translational Sciences, Bethesda, MD); Amy Proal, Ph.D. (PolyBio Research Foundation, Kenmore, WA); Melissa Haendel, Ph.D. (University of Colorado Anschutz Medical Campus, Denver, CO); Saurabh Mehandru, M.D. (Icahn School of Medicine at Mount Sinai, New York, NY); Eleni Gavriilaki, M.D., Ph.D. (Aristotle University of Thessaloniki, Greece); Upinder Singh, M.D. (Stanford University, Stanford, CA); Julie McElrath, M.D., Ph.D. (Fred Hutchinson Cancer Center, Seattle, WA); Joseph A. Bellanti, M.D. (Georgetown University, Washington, DC); Steven Z. Josefowicz, Ph.D. (Cornell University, New York, NY); Richard C. Becker, M.D. (University of Cincinnati, Cincinnati, OH); Prof. Raimund Helbok, M.D., Ph.D. (Department of Neurology, Johannes Kepler University Linz, Linz, Austria); Avindra Nath, M.D. (National Institute of Neurological Disorders and Stroke, Bethesda, MD); Anne Louise Oaklander, M.D., Ph.D. (Massachusetts General Hospital, Boston, MA); Prof. Theoharis Theoharides, M.Sc., Ph.D., M.D. (Nova Southeastern University, Fort Lauderdale, FL); James Morrissey, Ph.D. (University of Michigan, Ann Arbor, MI); James Harker, Ph.D. (Imperial College London, London, UK); James Stone, M.D., Ph.D. (Massachusetts General Hospital, Boston, MA); Colin Berry, M.D., Ph.D. (University of Glasgow, Glasgow, UK); Edward Conway, M.D., Ph.D. (University of British Columbia, Vancouver, BC, Canada); Shabnam Salimi, M.D. (University of Maryland, Baltimore, MD); Divaker Choubey, Ph.D. (University of Cincinnati, Cincinnati, OH); Charisse Madlock-Brown, Ph.D. (University of Iowa, Iowa City, IA); Charis Bridger Staatz, Ph.D. (University College London, London, UK); Dhruv Khullar, M.D. (Cornell University, New York, NY); Linda C. Geng, M.D., Ph.D. (Stanford University, Stanford, CA); Amy Arnsten, Ph.D. (Yale University, New Haven, CT).