This is day three of a five day series (thank goodness for the bank holiday weekend in the UK…) decoding the three most important studies from the past three years for five conditions. Yesterday: ME/CFS, today: Long COVID.
A brief note before we start. Long COVID has generated a larger volume of research in a shorter time than almost any condition in modern medical history. Choosing three studies from this field requires difficult decisions about what “most important” means. The three below were selected for their methodological rigour, the novelty of their findings, the mechanistic insights they provide, their direct relevance to patients navigating this condition and ultimately, my opinion. Several honourable mentions appear at the end.
Study 1: The Immune Profiling Study That Found a Biological Fingerprint of Long COVID
Klein J, Wood J, Jaycox J, et al. Distinguishing features of long COVID identified through immune profiling. Nature. 2023;623(7985):139–148.
Published in September 2023 in Nature, this study enrolled 275 individuals with or without Long COVID in a cross-sectional design. The research was led by Jon Klein and colleagues at Yale School of Medicine, the Icahn School of Medicine at Mount Sinai, with multiple affiliated institutions. Senior authors include Dr Akiko Iwasaki, one of the leading immunologists in the Long COVID field, together with Dr David Putrino, whose wearable research we have decoded separately.
The study used multidimensional immune phenotyping alongside unbiased machine learning to identify biological features that distinguished Long COVID participants from those who had recovered fully. Rather than testing a specific hypothesis about what might be wrong, the researchers mapped the full immune landscape and let statistical models identify what was actually different.
The findings were substantial. Long COVID participants showed marked differences in circulating myeloid and lymphoid cell populations compared to matched controls who had recovered from COVID-19. Myeloid cells, which include monocytes and macrophages, were present in altered proportions and showed evidence of abnormal activation. Lymphoid populations, including T cell subsets, were similarly disrupted in ways that differed from what is seen in acute infection or normal recovery.
Participants with Long COVID showed evidence of exaggerated humoral immune responses directed against SARS-CoV-2, meaning they were still mounting high antibody responses to the virus months after infection. This is consistent with the viral persistence hypothesis: if residual viral antigen or viral fragments remain in the body, the immune system continues producing antibodies against them. Higher antibody responses directed against Epstein-Barr virus were also observed among Long COVID participants compared to controls. EBV reactivation during immune dysregulation is a recognised mechanism in other post-viral conditions.
The most striking individual finding was the role of cortisol. When the machine learning model integrated all the immune phenotyping data to identify what best distinguished Long COVID participants from controls, decreased cortisol levels were the single most significant individual predictor. Cortisol is the body’s primary stress hormone, involved in regulating inflammation, energy metabolism, sleep-wake cycles, immune response, blood pressure, together with a range of other physiological processes. Its depletion in Long COVID participants raises questions about hypothalamic-pituitary-adrenal axis dysfunction, a finding consistent with the fatigue, sleep disruption, along with the immune dysregulation that characterise the condition.
This study does not prove causality. It cannot tell us whether the immune abnormalities cause Long COVID symptoms, whether they are a consequence of something else, or whether they are a necessary feature of the condition in all patients. The cross-sectional design provides a snapshot rather than a longitudinal picture. Replication in independent cohorts is needed. What it does provide is the most comprehensive biological characterisation of Long COVID immune features published to date, using methods that are both rigorous, crucially, hypothesis-agnostic.
Study 2: The Study That Proved the Virus Persists in Body Tissues After Mild Infection
Zuo W, He D, Liang C, et al. The persistence of SARS-CoV-2 in tissues and its association with long COVID symptoms: a cross-sectional cohort study in China. Lancet Infectious Diseases. 2024;24(8):845–855.
The viral persistence hypothesis for Long COVID has been discussed since early in the pandemic: the idea that residual SARS-CoV-2, either intact virus or viral fragments, remains in body tissues after apparent recovery, continuing to drive immune activation, inflammation, along with symptoms. This study, published in the Lancet Infectious Diseases in 2024, provides the most direct and comprehensive evidence for that hypothesis to date.
Researchers collected 317 tissue samples from 225 patients who had recovered from mild COVID-19, meaning patients who had not been hospitalised. Samples were collected at one, two, and four months post-infection. The tissues sampled were diverse: liver, kidney, stomach, intestine, brain, blood vessel, lung, breast, skin, and thyroid. Blood samples were also collected at each timepoint.
Viral RNA was detected in 30% of solid tissue samples at one month, 27% at two months, 11% at four months. The persistence was not confined to one or two tissue types. Viral RNA was distributed across all ten types of solid tissue tested. Subgenomic RNA, a marker of active viral replication rather than residual non-replicating fragments, was detected in 43% of the solid tissue samples that tested positive for viral RNA, suggesting that in a meaningful proportion of cases the virus was not simply leaving traces but retaining the capacity for replication.
The statistical link between viral persistence and Long COVID symptoms was strong. Patients with detectable viral RNA in their tissues were 5.17 times more likely to have Long COVID symptoms (odds ratio 5.17, 95% CI 2.64–10.13, p<0.0001). Higher viral copy numbers were also associated with greater symptom burden.
The importance of this study extends beyond confirming what had been suspected. It provides a mechanistic target. If residual viral persistence drives Long COVID in a subset of patients, antiviral therapy that clears the reservoir becomes a rational treatment strategy. The RECOVER-VITAL trial testing extended course Paxlovid in Long COVID patients was designed precisely around this hypothesis. This study also helps explain why some Long COVID patients improve spontaneously over time: viral clearance may occur gradually in some individuals, allowing immune dysregulation to resolve.
Limitations are worth noting carefully. This was conducted in a single country with specific viral variants circulating at the time of infection. The patients had mild acute disease, which is appropriate for studying the most common Long COVID phenotype but may not reflect what happens after severe acute infection. The study could not establish definitively whether the viral RNA detected was causing symptoms, correlated with something else causing symptoms, or present coincidentally. The mechanism by which persistent viral RNA drives the specific symptom profile of Long COVID remains incompletely characterised.
Study 3: The Muscle Biopsy Study That Made Post-Exertional Malaise Impossible to Dismiss
Appelman B, Charlton BT, Goulding RP, et al. Muscle abnormalities worsen after post-exertional malaise in long COVID. Nature Communications. 2024;15:17.
My personal favourite… We have decoded this study in detail in a previous piece, so the description here will be more concise. Published in January 2024 in Nature Communications, this study took skeletal muscle biopsies from 25 Long COVID patients with post-exertional malaise before exercise and one day after exercise triggered a crash, with the same protocol run in 21 healthy controls.
Before a single exercise bout, Long COVID muscle was already different from healthy muscle. Mitochondrial enzyme activity was reduced. The proportion of fast-twitch fibres that fatigue quickly was higher. These differences were present at baseline, without any exercise having occurred.
One day after exercise induced a crash, mitochondrial respiration dropped significantly in Long COVID patients while remaining stable in controls. Immune cells, specifically CD4+ T cells and mast cells, infiltrated the muscle tissue in ways not seen in controls. A key energy-generating enzyme called succinate dehydrogenase showed a significant interaction between group and time, falling in Long COVID patients after PEM while holding steady or rising in controls. Blood supply to the muscle was identical in both groups, ruling out the explanation that the damage was caused by reduced oxygen delivery, which is the hallmark of deconditioning.
The damage was visible under a microscope. It was measurable. It was specific to the crash window. The study provides the most direct evidence that post-exertional malaise involves objective cellular injury rather than subjective symptom amplification, reshaping how both researchers and clinicians think about exercise recommendations for this population.
A full decoded piece on this study including the subsequent debate in the published literature is available on Dysautonomia Decoded.
Honourable Mentions
The shared autonomic phenotype study by Novak and colleagues at Brigham and Women’s Hospital and Harvard Medical School (PLOS ONE, 2026) found that Long COVID and ME/CFS patients were essentially indistinguishable on detailed autonomic testing, with more than 90% of both groups showing reduced cerebral blood flow on standing. A full decoded piece is available on Dysautonomia Decoded.
The Walitt and Nath NIH deep phenotyping study (Nature Communications, 2024), which we decoded in the ME/CFS series yesterday, enrolled both post-infectious ME/CFS and Long COVID participants, documenting the autonomic, immune, along with sex-specific biological abnormalities across both groups simultaneously.
The microvascular retinal study by Wallraven and colleagues at the Technical University of Munich (BMC Medicine, 2026) found that post-COVID patients showed impaired retinal vessel dilation correlated directly with symptom severity, with ME/CFS patients showing nearly identical patterns. A full decoded piece is available on Dysautonomia Decoded.
The Bigger Picture
Three years of research have moved Long COVID from a disputed clinical entity to a condition with a growing biological evidence base. The immune system carries a measurable fingerprint in blood. The virus persists in body tissues in a substantial proportion of patients, linked statistically to symptoms with an odds ratio over five. Post-exertional malaise involves cellular damage visible under a microscope. These are not the findings of a functional or psychosomatic condition. They are specific, measurable, biologically coherent discoveries in peer-reviewed journals of the highest standing.
The research has not yet produced treatments that consistently work for most patients. That remains the most urgent gap. The mechanistic foundation being built by studies like these is what treatment development depends on.
The views and opinions expressed in Dysautonomia Decoded are my own and do not represent those of my employer or any affiliated organisation.
References
Klein J, Wood J, Jaycox J, et al. Distinguishing features of long COVID identified through immune profiling. Nature. 2023;623(7985):139–148. https://doi.org/10.1038/s41586-023-06651-y
Zuo W, He D, Liang C, et al. The persistence of SARS-CoV-2 in tissues and its association with long COVID symptoms: a cross-sectional cohort study in China. Lancet Infectious Diseases. 2024;24(8):845–855. https://doi.org/10.1016/S1473-3099(24)00171-3
Appelman B, Charlton BT, Goulding RP, et al. Muscle abnormalities worsen after post-exertional malaise in long COVID. Nature Communications. 2024;15:17. https://doi.org/10.1038/s41467-023-44432-3
Novak P, Systrom DM, et al. Shared autonomic phenotype of long COVID and myalgic encephalomyelitis/chronic fatigue syndrome. PLOS ONE. 2026;21(1):e0341278. https://doi.org/10.1371/journal.pone.0341278
Walitt B, Nath A, et al. Deep phenotyping of post-infectious myalgic encephalomyelitis/chronic fatigue syndrome. Nature Communications. 2024;15(1):907. https://doi.org/10.1038/s41467-024-45107-3
Wallraven T, Günthner R, et al. Microvascular remodeling and endothelial dysfunction across the post-COVID-19 spectrum. BMC Medicine. 2026;24(1):456. https://doi.org/10.1186/s12916-026-05144-9
This is amazing and so validating to read… thank you!
very interesting. Have any studies been done in the long term physiological difference for those who were exposed to and infected with covid 19 at different times of the outbreaks.