www.thelancet.com/infection Published online February 10, 2025 https://doi.org/10.1016/S1473-3099(24)00769-2 1 Personal View Lancet Infect Dis 2025 Published Online February 10, 2025 https://doi.org/10.1016/ S1473-3099(24)00769-2 PolyBio Research Foundation, Medford, MA, USA (A D Proal PhD, M B VanElzakker PhD) ; Department of Infectious Diseases and Unit of Post- COVID Huddinge, Karolinska University Hospital, Stockholm, Sweden (S Aleman MD) ; Department of Medicine Huddinge (S Aleman) and Department of Women’s and Children’s Health (Prof P Brodin MD) , Karolinska Institutet, Stockholm, Sweden; HIV entry and Laboratory of Mucosal Immunity, Institut Cochin, Paris, France (Prof M Bomsel PhD) ; Université Paris Cité, CNRS, INSERM, Institut Cochin, Paris, France (Prof M Bomsel) ; Department of Immunology and Inflammation (Prof P Brodin) and Medical Research Council Laboratory of Medical Sciences (Prof P Brodin) , Imperial College London, London, UK; Center for Infectious Medicine, Department of Medicine Huddinge, Karolinska Institutet, Huddinge, Sweden (M Buggert PhD) ; Department of Pathology and Laboratory Medicine (Prof S Cherry PhD) , Department of Microbiology (M Locci PhD) , Department of Systems Pharmacology and Translational Therapeutics (Prof E J Wherry), and Institute for Immunology and Immune Health (M Locci PhD, M M Painter PhD, Prof E J Wherry PhD) , Perelman School of Medicine, University of Pennsylvania, Philadelphia, PA, USA; Emerging Pathogens Section, Critical Care Medicine Department, Clinical Center, National Institutes of Health, Bethesda, MD, USA (D S Chertow MD) ; Laboratory of Virology, National Institute of Allergy and Infectious Diseases, National Institutes of Health, Hamilton, MT, USA Targeting the SARS-CoV-2 reservoir in long COVID Amy D Proal, Soo Aleman, Morgane Bomsel, Petter Brodin, Marcus Buggert, Sara Cherry, Daniel S Chertow, Helen E Davies, Christopher L Dupont, Steven G Deeks, E Wes Ely, Alessio Fasano, Marcelo Freire, Linda N Geng, Diane E Griffin, Timothy J Henrich, Stephen M Hewitt, Akiko Iwasaki, Harlan M Krumholz, Michela Locci, Vincent C Marconi, Saurabh Mehandru, Michaela Muller-Trutwin, Mark M Painter, Etheresia Pretorius, David A Price, David Putrino, Yu Qian, Nadia R Roan, Dominique Salmon, Gene S Tan, Michael B VanElzakker, E John Wherry, Johan Van Weyenbergh, Lael M Yonker, Michael J Peluso There are no approved treatments for post-COVID-19 condition (also known as long COVID), a debilitating disease state following SARS-CoV-2 infection that is estimated to affect tens of millions of people. A growing body of evidence shows that SARS-CoV-2 can persist for months or years following COVID-19 in a subset of individuals, with this reservoir potentially driving long-COVID symptoms or sequelae. There is, therefore, an urgent need for clinical trials targeting persistent SARS-CoV-2, and several trials of antivirals or monoclonal antibodies for long COVID are underway. However, because mechanisms of SARS-CoV-2 persistence are not yet fully understood, such studies require important considerations related to the mechanism of action of candidate therapeutics, participant selection, duration of treatment, standardisation of reservoir-associated biomarkers and measurables, optimal outcome assessments, and potential combination approaches. In addition, patient subgroups might respond to some interventions or combinations of interventions, making post-hoc analyses crucial. Here, we outline these and other key considerations, with the goal of informing the design, implementation, and interpretation of trials in this rapidly growing field. Our recommendations are informed by knowledge gained from trials targeting the HIV reservoir, hepatitis C, and other RNA viruses, as well as precision oncology, which share many of the same hurdles facing long- COVID trials. Introduction The COVID-19 pandemic has made it increasingly clear that infectious pathogens can drive not just acute illness, but also debilitating chronic disease. The US Centers for Disease Control and Prevention estimates that approxi- mately 6% of Americans infected with SARS-CoV-2 develop chronic symptoms or sequelae, referred to as post-COVID-19 condition (also known as long COVID). 1–3 Similar incidence has been noted in settings across the globe. 4 Although multiple biological factors are being evaluated as contributors to long COVID, 3,5 a growing body of research centres around the persistence of SARS-CoV-2 as a driver of disease in at least some individuals. 6–13 This SARS-CoV-2 reservoir might drive inflammation, hinder virus-directed immune responses, or disturb the function of infected cells, contributing to long COVID and other complications. If persistence of SARS-CoV-2 causes chronic disease, then the virus is an obvious target for therapeutic studies. Early trials of antivirals and monoclonal antibodies in long COVID are underway. However, because of uncertainty regarding the mechanisms and measures of SARS-CoV-2 persistence, several considerations should guide the design and interpretation of these and future trials. Here, we draw from RNA virus biology, efforts to target the HIV reservoir, and clinical oncology to outline major considerations for the design and implementation of trials targeting SARS-CoV-2 persistence in long COVID. The SARS-CoV-2 reservoir Although the initial expectation in the infectious disease community was that SARS-CoV-2 infection would be transient in immunocompetent individuals, multiple studies now support the existence of a SARS-CoV-2 reservoir in at least a subset of people with long COVID (appendix p 1). These studies have identified single- stranded or double-stranded SARS-CoV-2 RNA or proteins well beyond the acute phase of infection in tissues such as the gut 7,8,14–18 or in host cells such as platelets or megakaryocytes. 19 Immune responses suggestive of ongoing stimulation by viral antigens have also been documented in individuals with long COVID. 20–22 Multiple teams have identified SARS-CoV-2 proteins in plasma up to 14 months after initial infection, 10,11,23–27 which are believed to be translated from viral RNA in tissue reservoir sites and leak into the circulation. Although SARS-CoV-2 persistence has been documented even in people without long COVID, evidence linking SARS-CoV-2 persistence to long COVID is growing. In a large study from the National Institutes of Health RECOVER programme, participants who reported long-COVID symptoms affecting heart and lung, brain, and musculoskeletal systems were approximately twice as likely to have detection of SARS-CoV-2 proteins circulating in their blood during the post-acute phase. 11 The totality of the evidence suggests that some people with long COVID harbour a tissue-based reservoir of SARS-CoV-2, with variable detection in the circulation using present technologies, which might drive inflammation or immune dysregulation, and ultimately result in long COVID. 3 Establishment and maintenance of the reservoir The establishment of the reservoir begins during acute SARS-CoV-2 infection. Both host and viral dynamics likely play a role. A recent analysis of post-acute antigen persistence in plasma showed that those with more 2 www.thelancet.com/infection Published online February 10, 2025 https://doi.org/10.1016/S1473-3099(24)00769-2 Personal View (D S Chertow) ; Department of Respiratory Medicine, University Hospital Llandough, Cardiff, UK (H E Davies MD) ; University School of Medicine, University Hospital of Wales, Cardiff, UK (H E Davies) ; Division of Genomic Medicine, Environment & Sustainability (Prof C L Dupont PhD) , Department of Informatics (Y Qian PhD) and Department of Infectious Diseases (G S Tan PhD, M Freire PhD) , J Craig Venter Institute, University of California San Diego, La Jolla, CA, USA; Division of HIV, Infectious Diseases, and Global Medicine (Prof S G Deeks MD, M J Peluso MD) and Division of Experimental Medicine (Prof T J Henrich MD) , University of California, San Francisco, CA, USA; The Critical Illness, Brain Dysfunction, Survivorship Center at Vanderbilt University Medical Center, Nashville, TN, USA (Prof E W Ely MD) ; Veteran’s Affairs Tennessee Valley severe initial infection were more likely to have subsequent antigen detection. 10 Others have shown that the duration of viral shedding or maximum viral load during the acute phase correlates with subsequent long COVID. 28,29 These observations suggest that individuals with a higher early viral burden might have a greater viral inoculum that persists at primary infection sites such as the lungs or gut, or that seeds distant tissue sites. One autopsy study identified single-stranded and subgenomic SARS-CoV-2 RNA or protein in dozens of tissues including brain, nerve, and ocular tissue up to 230 days after COVID-19 onset. 30 The seeding of distant reservoir sites is reminiscent of post-Ebola syndrome, where studies have identified viral reservoirs in immune- privileged sites such as testes, eyes, and brain long after acute infection. 31 A related possibility is that host immune status or failure of the immune system to adequately clear the virus drives the establishment of a reservoir. For example, some studies have connected long COVID to decreased neutralising antibody function. 32 Immune dysfunction facilitating the seeding of a reservoir could partly explain why some patients with long COVID report mild acute infections but develop symptoms gradually over time. Such hypotheses are compatible with studies showing that efforts to reduce viral burden or enhance the immune response (eg, antiviral treatment 33–37 and vaccination 38,39) appear in some cases to protect against the development of long COVID. Once a reservoir is established, there are several mechanisms by which it could be maintained (figure 1). SARS-CoV-2 RNA could persist without producing new extracellular virions, potentially via mutation or immune escape, as observed with other single-stranded RNA viruses. 40 RNA could persist and be translated, provoking an immune response even if it is not fully replicating. The identification of viral proteins in plasma during the post-acute phase 10,11,23 suggests that translation is occurring, as does higher expression of markers of activation and exhaustion on SARS-CoV-2-specific CD8 + T cells among some individuals with long COVID in comparison to those who fully recovered. 41 Another possibility is that the virus persists and replicates, at least periodically or at low amounts, and is capable of infecting new target cells. This scenario is supported by studies that have identified double-stranded, antisense, or subgenomic RNA (transcripts synthesised as products of replication) in tissues, immune cells, or whole blood from people with long COVID. 7–9,19 However, these transcripts could represent active or previous replication of the virus, as it is possible the viral RNA persists for some time in protected intracellular microenvironments. Periodic viral replication is one potential explanation for the fluctuating symptoms reported by many patients with long COVID. Well designed studies targeting each mechanism, together with careful monitoring of the effects on reservoir biomarkers, will ultimately be needed to reveal the primary drivers of reservoir persistence (table 1). Therapeutic approaches to target the reservoir Here, we briefly review therapeutic strategies that could target the reservoir in clinical trials meant to address persistent SARS-CoV-2 as a driver of long COVID (table 2). Targeting the virus Antivirals are straightforward candidates for reservoir- targeting trials. Antivirals do not directly destroy viruses but rather disrupt specific components of their replication cycle, preventing production of additional virions. They are typically small molecules that target enzymes such as RNA-dependent RNA polymerase or viral protease. Antivirals such as nirmatrelvir–ritonavir, ensitrelvir, and remdesivir are being tested in several long-COVID trials (NCT05595369, NCT05668091, 43 NCT05823896, NCT06161688, NCT05911906, and NCT06511063). Although one trial (NCT05576662) showed no symptomatic benefit of a 15-day treatment course of nirmatrelvir–ritonavir compared with ritonavir placebo among 155 people with long COVID, 44 others are pursuing Figure 1: Potential forms of SARS-CoV-2 reservoir persistence Each form of persistence likely requires targeting by different therapeutics, or combinations of therapeutics, for the most effective treatment. SARS-CoV-2 Infected cell SARS-CoV-2 single-stranded RNA remains, without necessarily being translated into viral proteins or new virions SARS-CoV-2 RNA might be capable of driving translation of viral proteins without necessarily producing new virions SARS-CoV-2 RNA drives production of new virions periodically, and perhaps at low levels, which can potentially infect new cells for long-term reservoir maintenance Coreceptors Cytoplasm www.thelancet.com/infection Published online February 10, 2025 https://doi.org/10.1016/S1473-3099(24)00769-2 3 Personal View Geriatric Research Education Clinical Center, Nashville, TN, USA (Prof E W Ely) ; Department of Pediatrics (Prof A Fasano MD, L M Yonker MD) , Mucosal Immunology and Biology Research Center (Prof A Fasano, L M Yonker) , and Division of Neurotherapeutics (M B VanElzakker) , Massachusetts General Hospital, Boston, MA, USA; Harvard Medical School, Boston, MA, USA (Prof A Fasano, L M Yonker) ; J Craig Venter Institute, Department of Medicine, Stanford University School of Medicine, Stanford, CA, USA (L N Geng MD) ; W Harry Feinstone Department of Molecular Microbiology and Immunology, Johns Hopkins Bloomberg School of Public Health, Baltimore, MD, USA (Prof D E Griffin PhD) ; Laboratory of Pathology, Center for Cancer Research, National Cancer Institute, National Institutes of Health, Bethesda, MD, USA (S M Hewitt MD) ; Department of Immunobiology (Prof A Iwasaki PhD) , and Center for Infection and Immunity (Prof A Iwasaki, Prof H Krumholz MD) , Yale University School of Medicine, New Haven, CT, USA; Howard Hughes Medical Institute, Chevy Chase, MD, USA (Prof A Iwasaki) ; Center for Outcomes Research and Evaluation, Yale New Haven Hospital, New Haven, CT, USA (Prof H Krumholz MD) ; Section of Cardiovascular Medicine, Department of Internal Medicine, Yale School of Medicine, New Haven, CT, USA (Prof H Krumholz) ; Department of Health Policy and Management, Yale School of Public Health, New Haven, CT, USA (Prof H Krumholz) ; Emory University School of Medicine and Rollins School of Public Health, Atlanta, GA, USA (Prof V C Marconi MD) ; Atlanta Veterans Affairs Medical Center, Decatur, GA, USA (Prof V C Marconi) ; Precision Immunology Institute (Prof S Mehandru MD) and Department of Rehabilitation and Human Performance (Prof D Putrino PhD) , Icahn School of Medicine at Mount Sinai, New York, NY, USA; Henry D Janowitz Division of Gastroenterology, Department of Medicine, Icahn School of the approach in larger studies that include more targeted phenotypes, along with longer treatment duration. Monoclonal antibodies (mAbs) are also promising therapeutic candidates, as these agents can have two activities: directly neutralising replicating virus, thereby serving as an antiviral to halt the replication cycle, and promoting clearance of viral protein. If effector function is maintained, they may also promote clearance of infected cells expressing viral proteins at the cell surface via antibody-dependent cellular cytotoxicity. Although no agents of this class are currently approved or authorised for treatment of COVID-19, several are available for prophylaxis, and new agents are under development. There are also approaches under development to directly inactivate and degrade viral RNA, including antisense oligonucleotides and CRISPR-based RNA targeting. 45–48 These therapies could target viral RNA even if no replication is occurring. However, research is needed to determine feasibility. In particular, under- standing viral reservoir locations (eg, the gut) is important since such treatments might need to be directed to the tissue where RNA persists. Enhancing the host immune system The presence of a SARS-CoV-2 reservoir in long COVID suggests that the virus might escape immune-mediated clearance, perhaps due to dysfunctional immune responses, or reduced activity of local cytotoxic T-cell and natural killer (NK) cell responses in tissues, particularly within immune-privileged sites. Murine norovirus— another RNA virus—can persist in an immune-privileged enteric niche via induction of a CD8 + T-cell differentiation state that fails to detect and clear viral reservoirs. 49 Cellular immune dysfunction and general immune dysregulation have been observed in long COVID, 41,50,51 and in some cases connected to SARS-CoV-2 persistence. 25,52 Some, 41,49 but not all, 25 studies to date have observed subtle changes in T-cell responses that warrant further investigation to determine if SARS-CoV-2-specific T cells are exhausted or otherwise dysfunctional in patients with long COVID. If T-cell exhaustion or NK cell dysfunction, or both, are connected to SARS-CoV-2 persistence, enhancing these innate and adaptive immune responses might facilitate better control of persistent infection. Relevant therapies include cytokines such as interleukin superagonists, which are being studied in HIV to restore and boost T-cell and NK cell responses 53 or cytokines with antiviral potential (eg, interferon and IL-15). PD-1 and PD-L1 checkpoint blockade therapies could be considered. Whether therapeutic COVID-19 vaccination is beneficial in inducing immune responses that clear the viral reservoir could also be investigated. Thus far, although some people with long COVID have reported symptom benefit from vaccination, 38,39 others have reported worsening of symptoms. 54 It should be noted that the current SARS-CoV-2 vaccines such as mRNA and viral- vectored vaccines generate neutralising antibody Mechanisms Advantages Considerations Direct acting Direct acting antivirals Disrupt viral replication Several approved; oral options available; antiviral agents with different synergistic mechanisms of action might more successfully target reservoirs Tissue penetration to sites of persistence requires further study; extended courses might be needed if reservoir persists within long-lived host cell types; drug–drug interactions can complicate administration Antisense oligonucleotides and other CRISPR-based approaches Inactivate or destroy viral ssRNA Could potentially inactivate viral RNA whether or not translating or replicating, and even if it has escaped immune clearance Oligonucleotides and protein-based therapeutics often need to be specifically directed to the tissue where RNA persists and delivery should consider cellular internalisation strategies Acquired and innate immunotherapies Monoclonal antibodies Bind extracellular proteins and virions Can be updated for different variant targets; maintain longer-term therapeutic levels Unlikely to work if virus is not extracellular or protein not expressed on surface of infected cell; might or might not have access to target tissue; might or might not have retained effector function to facilitate clearance of infected cells; evolving variants might become resistant Cytokine-based therapies (eg, interleukin superagonists) Activate T and natural killer cell immunity or innate defenses Could still work even if persistent virus has escaped immune-mediated clearance Stimulating the immune response might cause patients to temporarily feel worse before feeling better PD-1 and PD-L1 checkpoint blockade therapies Restore function of exhausted T cells Supports T cells to target intracellular reservoir Further investigation needed to elucidate which patients with long COVID show T cell exhaustion; potentially complex risk profile Therapeutic vaccines (including next generation vaccines) Increase immune response and antibody response towards persistent antigen Can be updated for different variants Development of next-generation vaccines that include viral epitope targets beyond spike, and that elicit antibody-dependent cell-mediated responses or T-cell mediated responses might better direct clearance of persistently infected cells Table 1: Potential reservoir trial therapeutics and mechanisms of action 4 www.thelancet.com/infection Published online February 10, 2025 https://doi.org/10.1016/S1473-3099(24)00769-2 Personal View Medicine at Mount Sinai, New York, NY, USA (Prof S Mehandru) ; Institut Pasteur, Université Paris-Cité, HIV, Inflammation and Persistence Unit, Paris, France (Prof M Muller-Trutwin PhD) ; Department of Physiological Sciences, Faculty of Science, Stellenbosch University, Stellenbosch, South Africa (Prof E Pretorius PhD) ; Department of Biochemistry and Systems Biology, Institute of Systems, Molecular and Integrative Biology, Faculty of Health and Life Sciences, University of Liverpool, Liverpool, UK (Prof E Pretorius) ; Division of Infection and Immunity (Prof D A Price MD) responses and T-cell immunity only to the spike protein. The development of next-generation vaccines that include expanded viral protein targets or elicit antibody- dependent cell-mediated responses might better direct clearance of persistently infected cells. This effort mirrors those to develop therapeutic vaccines in other chronic viral infections such as HIV. Key considerations for clinical trials targeting the reservoir The optimal design of reservoir-targeting trials requires consideration of multiple factors (figure 2). Measures of persistence Because tissue-based studies have provided compelling evidence of a SARS-CoV-2 reservoir, use of tissue biopsy is an optimal approach for assessing the effect of trials targeting the reservoir. Some tissues (eg, gut and lymph nodes) can be safely sampled in living patients (figure 3). After processing and preservation of the tissue, investigators can assess the presence of RNA using hybridisation, sequencing, or amplification approaches (table 3). Detection of double-stranded RNA 55 can be used to infer the presence of ongoing or previous replicating virus. Ideally, these assessments would occur before and after intervention. However, tissue collection is invasive, expensive, and technically challenging. A small tissue specimen might miss neighbouring cells harbouring SARS-CoV-2. The reservoir might also persist in tissues such as the brain or myocardium, which are likely inaccessible given the risk of sampling such organs. Hence, the field would benefit tremendously from a persistence biomarker akin to the plasma HIV RNA level NCT number Therapeutic Location Recruitment target Status Treatment of established long COVID PREVAIL-LC NCT06161688 Ensitrelvir fumaric acid (S-217622) vs placebo University of California, San Francisco, CA, USA 40 participants Active RECOVER-VITAL NCT05965726 Nirmatrelvir–ritonavir for 25 days or 15 days vs placebo ritonavir Duke University, Durham. NC, USA 900 participants Active imPROving Quality of LIFe in the long COVID patient NCT05823896 Nirmatrelvir–ritonavir for 15 days vs placebo ritonavir Karolinska Institutet, Solna, Sweden 400 participants Active PaxLC NCT05668091 Nirmatrelvir–ritonavir for 15 days vs placebo ritonavir Yale University, New Haven, CT, USA 100 participants Completed STOP-PASC NCT05576662 Nirmatrelvir–ritonavir for 15 days vs placebo ritonavir Stanford University, Palo Alto, CA, USA 168 participants Completed outSMART-LC NCT05877508 AER002 administered once intravenously vs placebo University of California, San Francisco, CA, USA 30 participants Active AT1001 for the treatment of long COVID NCT05747534 Larazotide (AT1001) vs placebo in children and young adults Massachusetts General Hospital, Boston, MA, USA 48 participants Active Antiviral clinical trial for long COVID-19 NCT06511063 Truvada or selzentry vs placebo Icahn School of Medicine at Mount Sinai, New York, NY, USA 90 participants Active ERASE-LC NCT05911906 Remdesivir University of Plymouth, Devon, UK 72 participants Active ESSOR NCT05999435 LAU-7b vs placebo Multiple hospitals, Canada 272 participants Active Study to evaluate the efficacy and safety of ampligen in patients with post-COVID conditions NCT05592418 Rintatolimod vs placebo Hunter-Hopkins Center, Charlotte, NC, USA 80 participants Completed RSLV-132 in participants with long COVID NCT04944121 RSLV-132 vs placebo Resolve Therapeutics, Alabama and Florida, USA 112 participants Completed Prevention of long COVID ACTIV-2 NCT04518410 Amubarvimab and romlusevimab vs placebo for acute COVID-19 Sites in the USA, Argentina, Brazil, Mexico, Philippines, and South Africa 847 participants Completed SOLIDARITY NCT04978259 Remdesivir during acute COVID-19 hospital stay up to 10 days in addition to standard care Clinical Urology and Epidemiology Working Group, Helsinki, Finland 202 participants Completed PANORAMIC ISRCTN30448031, NIHR135366 Molnupiravir vs placebo during acute COVID-19 University of Oxford, Oxfordshire, UK 783 participants Completed SCORPIO-SR* jRCT2031210350 Ensitrelvir during acute COVID-19 vs placebo 92 institutions in Japan, Viet Nam, and South Korea 1821 participants, 2694 target Completed COVID-OUT* NCT04510194 Metformin vs fluvoxamine vs ivermectin vs metformin plus fluvoxamine vs metformin plus ivermectin vs placebo University of Minnesota, Minneapolis, MN, USA 1323 participants Completed *The trial was not designed to target the reservoir, but secondary analysis found that metformin had an antiviral effect. Table 2: Clinical trials designed to target or prevent the SARS-CoV-2 reservoir www.thelancet.com/infection Published online February 10, 2025 https://doi.org/10.1016/S1473-3099(24)00769-2 5 Personal View and Systems Immunity Research Institute (Prof D A Price) , Cardiff University School of Medicine, University Hospital of Wales, Cardiff, UK; Gladstone Institutes (Prof N R Roan PhD) and Department of Urology (Prof N R Roan) , University of California, San Francisco, CA, USA; Department of Infectious Diseases, Institut Fournier, Paris, France, (D Salmon MD) ; Direction of International Relations Assistance Publique Hôpitaux de Paris, Paris, France (D Salmon) ; Laboratory of Clinical and Epidemiological Virology, Rega Institute for Medical Research, Department of Microbiology, Immunology and Transplantation, KU Leuven, Leuven, Belgium (J Van Weyenbergh PhD) Correspondence to: Dr Amy D Proal, PolyBio Research Foundation, Medford, MA 02155, USA aproal@polybio.org or in HIV—an objective measurement that uses an easily accessible body fluid (eg, blood), is sensitive and specific, has a rapid turnaround time, is scalable, and for which a change (eg, reduction) is clearly associated with a clinical outcome. 56 The development and validation of multiple assays for SARS-CoV-2 persistence is among the most urgent efforts in this field. But although multiple studies have identified SARS-CoV-2 RNA or proteins in blood obtained from patients with long COVID, 9,10,25 detection of these antigens is not nearly as reliable as gold standard biomarkers, such as plasma HIV RNA, for several reasons. The degree to which antigens from tissue sites reach the circulation is unknown, and participants harbouring tissue reservoirs could fail to have antigen detected in blood. Using current assays, detection in blood varies within an individual over time. 10,23 The cause of such fluctuations, although unknown, might reflect differences in translational activity in tissue, leading to periods when proteins are released into the circulation at different rates. It is also possible that only tiny amounts of protein leak into circulation from reservoir sites; these might be below the detection limit of many assays and consequently missed. For this reason, ultrasensitive assays (eg, Simoa) are used in some studies. However, even these assays might require additional iterations to accurately identify protein below the detection limit. 10,11,25 SARS-CoV-2 RNA and protein in the blood of individuals with long COVID might also persist inside host cells including monocytes, megakaryocytes, or platelets. 19 High-throughput assays capable of measuring cell- associated viral RNA, such as digital transcriptomics 9 in whole blood, might be needed to account for this possibility. Given these gaps, it is imperative to accelerate efforts to further develop and optimise assays for SARS-CoV-2 RNA and protein in accessible matrices, such as whole blood, plasma, stool, and saliva. These efforts should include defining assay sensitivity, specificity, positive and negative predictive value, reproducibility, determinants of within-person and between-person variation, and the likelihood of antigen detection in body fluids if the reservoir is localised to tissue. The detection limitations of the assay used should also be considered. For example, blood samples from people with long COVID might have high amounts of endogenous anti-SARS-CoV-2 IgG that could form immune complexes, impairing antigen or specific antibody detection with some assays. Protocols that remove these endogenous IgGs or dissociate immune complexes might be required before measuring antigens via immunoassay. Ideally, assays would distinguish inert from replicating virus, would quantify viral transcripts, and would document if mutations occur over time. Development of new assays, such as CRISPR- based approaches for viral RNA detection, 57 or changes in the immune system 20 should also be prioritised. These are key considerations because reliable persistence measurables in blood and saliva would not only increase our ability to interpret trials, but also allow trials to be more inclusive, potentially enabling sample collection at home for bedbound patients with severe disease. Eligibility determination Because SARS-CoV-2 can drive many forms of physiological dysfunction, a portion of patients meeting symptom-based long-COVID diagnostic criteria could potentially be affected by underlying issues other than SARS-CoV-2 persistence. Thus, to successfully trial therapeutics targeting the SARS-CoV-2 reservoir in long COVID, teams should recruit or analyse those patients who have persistent virus. The recruitment of such patients is not straightforward for the reasons outlined above. In general, there are two approaches to testing drugs targeting the SARS-CoV-2 reservoir. Ideally, investigators would recruit only participants confirmed to harbour a SARS-CoV-2 reservoir and assess changes in reservoir measures before and after the intervention. For example, one trial of larazotide, a synthetic peptide regulator of gut permeability, for long COVID requires confirmation of spike antigenemia for eligibility (NCT05747534). Due to variability in plasma protein detection (10–60%), and questions about assay sensitivity, anywhere from Figure 2: Main considerations for clinical trials targeting the reservoir mAbs=monoclonal antibodies. Eligibility determination Recruitment based on, or post hoc analyses stratifying by, viral persistence is recommended Safety Adverse effects of some drugs may differ between acute and long COVID, and some agents have considerable drug–drug interactions Blinding and the placebo effect Due to potential placebo effect, inclusion of a control group is essential for interpretation of most studies Duration and timing of treatment Longer treatment courses (months rather than weeks) might be needed to eradicate a reservoir Tissue penetration of the drug Reservoir-targeting therapeutics must achieve adequate levels in locations of persistence On-study reinfections Investigators should have a plan to handle on-study reinfections, including emphasising the importance of testing Measures of persistence Tissue biopsy is ideal, but more accessible biospecimens (eg, blood, etc) should be collected for certain persistence assays Combination trials Combination trials of drugs (eg, antivirals and mAbs) might be required to most effectively target reservoirs Evolving SARS-CoV-2 variants Broadly neutralising SARS-CoV-2 antibodies might be needed to target mixed reservoirs or those established after infection with modern variants Prevention of long COVID Interventions of acute COVID-19 treatment should include post-acute viral persistence as a prespecified outcome 6 www.thelancet.com/infection Published online February 10, 2025 https://doi.org/10.1016/S1473-3099(24)00769-2 Personal View Assist Prof Michael J Peluso, Division of HIV, Infectious Diseases, and Global Medicine, University of California, San Francisco, CA 94110, USA michael.peluso@ucsf.edu See Online for appendix 80 to 480 people might need to be screened to recruit the sample size of 48 participants with spike antigenemia. This effort might be warranted because the afore- mentioned trial of nirmatrelvir–ritonavir 44 —which did not reach its primary endpoint—did not have available any validated or fit-for-purpose measure to inform recruitment. 43 An alternative, and more common, approach is to recruit based on meeting an established case definition for long COVID, and to store biospecimens for post-hoc analyses focused on the reservoir. Orthogonal assays— for example, using a combination of protein-based and RNA-based measures in blood—can be used to assess evidence of virus persistence at baseline, even if viral persistence is not a key endpoint of the study. In such cases, post-hoc analyses stratifying by the presence or absence of viral persistence can be prespecified to improve the validity of clinically significant findings. This approach also allows for reanalysis of outcomes should new persistence assays become available. Depending on the proportion of people with long COVID confirmed to have a viral reservoir, one might not observe a benefit in the treatment group and might only see an effect in the subgroup with persistence. Investigators might consider calculating the sample size needed to detect an effect assuming different proportions (eg, 10%, 25%, and 50%) of participants with SARS-CoV-2 persistence in the study cohort. This calculation would inform the design and contextualise the results. Stratification by biological sex could also reveal subgroup benefit. For example, a phase 2 study of RSLV-132, a human RNase fused to IgG1 Fc that can degrade extracellular RNA, was recently completed. 47 Although the primary endpoint of fatigue improvement at day 71 was not met, earlier timepoints revealed statistically significant improvement in fatigue. Importantly, women showed greater responses to RSLV-132 than men, potentially because TLR7, which detects viral RNA, is expressed at higher concentrations in women. 58 Combination trials To date, most long-COVID trials have tested single agents. If SARS-CoV-2 persistence is driven by active replication, synergistic use of agents with different mechanisms of action, including targeting viral proteins at different phases of the viral life cycle (eg, viral proteases, RNA-dependent RNA polymerase, or other proteins), might be necessary. This possibility has been suggested in acute COVID-19; although monotherapy is the current standard, combining pyrimidine biosynthesis inhibitors with antiviral nucleoside analogues synergistically inhibits SARS-CoV-2 infection. 59 Alternatively, combining direct-acting antivirals with mAbs could simultaneously halt intracellular replication, enhance clearance of infected cells, and neutralise extracellular virus and protein. This approach needs to be tested directly. Another consideration relates to the fact that the relationships—that is, the connections and directionality— between virus persistence and multiple other mechanisms Figure 3: SARS-CoV-2 clinical trials persistence metrics 1 2 3 6 7 5 4 Biopsy SARS-CoV-2 RNA and protein in the numbered tissue reservoir sites can be measured in samples collected via biopsy. Blood and other fluids SARS-CoV-2 proteins in blood or other body fluids can be measured via ultrasensitive single-molecule arrays. Protein measured via these assays is likely translated by virus in tissue reservoir sites, and leaks into the circulation via exosome transport where it can be measured. However, even these assays may require further iteration to improve their detection abilities. Host cells SARS-CoV-2 RNA and protein inside host cells, including cells of the immune system (eg, monocytes, platelets, or megakaryocytes), can be measured via methods such as digital transcriptomics or RNAScope probes. Analysis could require collection of whole blood in which host cells are not depleted via centrifugation as with plasma. Biopsy sites 1. Olfactory epithelium, tonsils, adenoids, and salivary gland 2. Lungs 3. Lymph nodes 4. Gut 5. Endometrium 6. Muscle 7. Bone marrow After appropriate tissue processing and preservation, the presence of RNA can be accessed via hybridisation, sequencing, or amplification approaches. Assessment of viral replication, for example via methods to detect double-stranded RNA, can be used to infer if active virus is present. www.thelancet.com/infection Published online February 10, 2025 https://doi.org/10.1016/S1473-3099(24)00769-2 7 Personal View proposed to be important in long COVID have yet to be confirmed. Perhaps a dysfunctional immune system, driven by the reactivation of Epstein–Barr virus, 60 fibrin formation, 61 or microbiome perturbations, allows SARS-CoV-2 to persist while simultaneously causing long COVID. In such a case, addressing the primary insult, or addressing multiple mechanisms simultaneously (eg, a SARS-CoV-2 antiviral and a herpesvirus antiviral) might be better than targeting the reservoir alone. Combination trials to simultaneously address virus persistence and dysregulated immunity might also be important. There are multiple considerations for such studies. The SARS-CoV-2 reservoir might persist in immune-protected tissue microenvironments that are Purpose of assay Considerations Nucleic acid Quantitative PCR (real time or droplet digital) Quantitate the number of specific viral RNA target sequences in blood, cell-free fluids, or bulk tissue lysates Nucleic acid sequence mutations might affect primer binding and fragmentation of genetic material might lead to assay failure; presence of RNA does not equate with active replication Target-specific probe-based assays Examine specific viral RNA targets using probe- based methods with or without concomitant analysis of human gene transcription Nucleic acid sequence mutations might affect primer binding and fragmentation of genetic material might lead to assay failure; presence of RNA does not equate with active replication; can also multiplex assays to interrogate hundreds of viral and host targets in a single reaction and set range of input RNA concentrations that can be used at one time In situ hybridisation Examine the presence of the target in the native tissue with high sensitivity Contingent on tissue integrity and preservation; can target multiple RNA regions simultaneously within a specific gene or subgenomic region and provides spatial information, but background fluorescence or non-specific binding or trapping of probe might complicate interpretation Digital spatial profiling Examine specific viral RNA and human transcript targets using either probe-based or less biased RNA sequencing methods to provide spatial information Contingent on tissue integrity and preservation, can target multiple RNA regions simultaneously within a specific gene or subgenomic region and provides spatial information, but background fluorescence or non-specific binding or trapping of probe might complicate interpretation; some assays allow single-cell transcript resolution; low throughput and data analysis is challenging CRISPR-Cas target recognition Direct detection of viral RNA that can allow quant