Plasma proteomic profile reveals persistent immune activation in post-acute sequelae of SARS-CoV-2 infection

The COVID-19 pandemic, caused by the SARS-CoV-2 virus, has had profound implications for global health, with a substantial proportion of individuals experiencing persistent symptoms long after recovery from the acute infection. The Post-Acute Sequelae of SARS-CoV-2 infection (PASC), commonly referred to as Long COVID, represent a complex, multisystem syndrome that can persist for months or even years (1). Individuals with PASC report a diverse range of symptoms, including fatigue, cognitive dysfunction, dyspnea, and musculoskeletal pain, which can significantly impair quality of life and lead to long-term disability (2).

Despite a growing clinical recognition, the pathophysiology of PASC remains incompletely understood. The lack of validated molecular markers has hampered the development of both specific diagnostic tools and effective therapeutic interventions. However, accumulating evidence from independent studies suggests the presence of inflammatory marker proteins in plasma of individuals with PASC at various time points following SARS-CoV-2 infection (3–8).

Among the cytokines most frequently reported as upregulated in individuals with PASC are IL-1β, IL-6, and TNF-α (9). While their roles in the “cytokine storm” characteristic of severe acute COVID-19 are well established, there is no consensus that these cytokines are the primary drivers of chronic inflammation in PASC (10).

In a large UK cohort of hospitalized COVID-19 patients (the PHOSP-COVID study), plasma protein profiling at 5 months after discharge revealed that 13 proteins were significantly elevated in post-COVID individuals in the “very severe” recovery cluster compared with the “mild” cluster. These included IL-6, TGFA, CD83, SCGB3A2, CLEC4D, LAMP3, PLAUR, LGALS9, TFF2, EPO, FLT3LG, AGRN and FST (11).

In another longitudinal study of individuals with PASC, followed from 33 to 379 days after a mild COVID-19 infection, approximately 60% exhibited ongoing inflammation, while 40% did not. The inflammatory subgroup was characterized by persistent activation of IL-12/interferon-γ signaling and an NF-κB-driven inflammatory response, possibly initiated by TNF-α and resulting in elevated IL-6 expression (12).

The stratification of PASC into inflammatory and non-inflammatory subtypes has also been proposed by Woodruff and others (13), who identified 12 biomarkers associated with the inflammatory subtype, including IL-6, IL-8, and NF-κB.

Further efforts have been made to identify specific biomarkers in subgroups of PASC (14, 15). Given the heterogeneity of symptoms, several studies have attempted to associate plasma protein profiles with distinct post-COVID phenotypes, including cardiovascular, cognitive, gastrointestinal, fatigue, and anxiety/depression manifestations (14). Notably, children with PASC have shown aberrant plasma protein profiles characterized by elevated expression of pro-inflammatory and pro-angiogenic chemokines such as CXCL11, CXCL1, CXCL5, CXCL6, CXCL8, TNFSF11, OSM, and STAMBP1a (16).

It has been suggested that the chronic inflammation in PASC may be driven by viral persistence, with some studies reporting elevated levels of SARS-CoV-2 spike protein in the plasma of individuals with PASC up to one year or more after acute infection (17–19). However, this association remains uncertain and is still under debate.

Despite the growing number of studies on protein profile in PASC using mass spectrometry and proximity extension assay technologies, there remains a lack of research with extended follow-up, particularly in individuals who were not hospitalized during the acute phase of infection. Considering that certain clinical symptoms improve significantly over time, while others persist, it is important to conduct proteomic analyses in patients with very long-lasting symptoms. Moreover, the selection of a homogeneous study population is crucial to minimize confounding results by mixing individuals who experienced severe acute COVID-19 with those who had mild, non-hospitalized disease.

In this study, we used the Olink Explore 384 platform to profile plasma proteins in a cohort of 92 individuals with not previous hospitalized PASC and 73 matched controls who had experienced SARS-CoV-2 infection during the same time period but had fully recovered. In addition, we measured plasma levels of the SARS-CoV-2 spike protein. The primary aim was to determine whether individuals with PASC exhibit persistent inflammation and proteomic alterations detectable up to three years after the initial infection, and whether these alterations were associated with post-COVID symptoms. A secondary aim was to assess whether plasma spike protein levels were elevated in PASC patients and to explore their possible correlation with symptom profiles.

In this study, we demonstrate the presence of chronic inflammation in 92 PASC subjects at a mean follow-up of 34 months after a mild SARS-CoV-2 infection. The homogeneous cohort, together with the very long follow-up time, brings additional important information on the persistent proteomic inflammatory change associated to the post-COVID phenotype.

In a previous study on transcriptome analysis in PBMCs from a cohort of PASC individuals—also included in this study (26)—we reported activation of the JAK/STAT signaling pathway. In the current study, we identified elevated protein levels of OSM, IL-6, CSF3, CSF1, IL-12B, and IL-2, all known activators of the JAK/STAT pathway, consistent with our earlier findings.

Among the 26 DEPs, OSM was the most highly overexpressed (padj = 0.0019) and may represent the primary driver of JAK/STAT pathway overactivation in this cohort of patients. Beyond its role in immune activation and chronic inflammation, OSM is known to contribute to tissue remodeling and fibrosis, promoting fibroblast activation, extracellular matrix deposition, and fibrogenic responses across multiple organ systems (29, 30). Interestingly, elevated OSM levels have been reported both in acute severe SARS-CoV-2 infection and in PASC (31, 32). Russell et al. (31) identified OSM as significantly upregulated in lung tissue from patients who died of acute COVID-19, while a systematic review highlighted OSM as one of the consistently elevated biomarkers in PASC (32). Moreover, a recent study in children with PASC found OSM among the most prominently increased proteins (16). Together, these observations further support a potential role for OSM as a contributing driver of persistent inflammation in PASC.

We also identified additional overexpressed members of the IL-6 cytokine family, most notably IL-6, which plays a central role in both acute SARS-CoV-2 infection and PASC. In severe acute infection, IL-6 contributes to the cytokine storm together with TNF-α and IL-1β. Elevated IL-6 has been described as a strong biomarker of disease severity, correlating with ICU admission, need for ventilatory support, and increased mortality risk (33). Therapeutic agents targeting IL-6 or its receptor—such as Tocilizumab—are widely used to blunt the cytokine storm. Emerging evidence also suggests that IL-6 blockade may dampen long-term immune activation by reducing inflammatory gene expression in innate immune progenitor cells, potentially lowering the risk of developing PASC, though these findings await clinical validation (34). Recent data indicate that early IL-6 elevation, within four days of hospitalization, doubles the risk of developing PASC (35). Moreover, multiple cohort studies have identified persistently elevated IL-6 levels—often alongside IL-1β and TNF-α—for several months after the acute phase of infection, including in individuals who experienced only mild or moderate initial illness (36). Furthermore, IL-6 dysregulation has been especially implicated in neuropsychiatric manifestations of PASC, including fatigue, sleep disturbances, and depression (37). The sustained upregulation of both IL-6 and OSM observed in our cohort up to three years post-infection suggest that the inflammatory response triggered by SARS-CoV-2 can remain unresolved for years, supporting the concept of persistent immune activation as a key driver of long-term PASC symptoms. Our results are consistent with the broader literature and provide further evidence that IL-6–mediated pathways may represent a central mechanism in the pathophysiology of chronic post-COVID conditions, highlighting potential targets for therapeutic intervention.

In addition to OSM, we identified IL-1RA (padj = 0.0019) as one of the two top DEPs in our cohort. By competitively inhibiting the binding of IL-1α and IL-1β to the IL-1 receptor, IL-1RA functions as an endogenous anti-inflammatory mediator. Previous reports have shown that IL-1RA overexpression can mitigate the severity of acute COVID-19 infection (38). The fact that IL-1RA remains upregulated such a long time after infection may reflect a sustained attempt to counteract ongoing IL-1β-driven inflammation.

The third top protein in our random forest analysis was Angiopoietin-like protein 2 (ANGPTL2). This protein has been reported in preclinical studies to contribute to endothelial dysfunction and chronic inflammatory signaling, promoting vascular inflammation, tissue remodeling, and fibrosis, suggesting a potential role in persistent endothelial perturbation in PASC (39, 40).

Among the upregulated enrichment pathways, we found activation of both adaptive immune responses (Allograft Rejection and IL-2/STAT5 Signaling pathways) and innate immune responses (Inflammatory Response, TNF-α Signaling via NF-κB, and IL-6/JAK/STAT3 Signaling). While chronic activation of innate immunity may lead to a pro-inflammatory state—characterized by elevated cytokine levels and persistent activation of monocytes and other myeloid cells—the activation of adaptive immune responses may contribute to T cell dysregulation and autoimmune-like features, including autoantibody production.

We did not observe any correlation between plasma DEPs and post-COVID symptom severity or other clinical parameters, including SARS-CoV-2 variant (Omicron vs. earlier variants). The reasons why clinical symptoms did not align with measurable systemic inflammation in our cohort may be several. Self-reported symptoms such as fatigue, brain fog, or pain can fluctuate over time, and the assessment tools used may not have sufficient sensitivity or may not be fully standardized for this purpose. Single plasma measurements may also represent a limiting factor. Moreover, circulating proteins may not fully reflect tissue-specific damage or organ-specific inflammation, and other mechanisms, such as autonomic nervous system dysregulation, may contribute to persistent PASC symptoms.

Similarly, we did not identify distinct inflammatory subgroups within our cohort of patients. While these findings suggest that symptom patterns may not correspond to distinct inflammatory proteomic signatures, limitations in sample size, statistical power, symptom assessment tools, and natural variation of symptoms over time may have influenced our ability to detect such subgroups. Our observations are consistent with those reported by Talla et al. (12), who also reported no biologically defined inflammatory subgroups in PASC, but do not align with the results reported by Liew et al. (14). Notably, the only significant association identified in our study was between body mass index (BMI) and DEPs, also consistent with observations by Talla et al. (12). Elevated BMI is known to be associated with chronic low-grade systemic inflammation (41), but interestingly we found no association between BMI and post-COVID symptom burden. This suggests that BMI may act as a confounding factor in studies linking PASC to systemic inflammatory signatures, influencing proteomic signatures independently of clinical symptom severity.

Interestingly, the altered plasma protein profile in our PASC cohort partially overlapped with findings from the PHOSP-COVID study (11). That study identified 13 differentially expressed proteins, six of which were also found in our cohort: IL-6, TGFA, CD83, SCGB3A2, CLEC4D, and LGALS9. While the PHOSP-COVID cohort comprised previously hospitalized individuals assessed at 5 months and 1 year post-discharge, our cohort included non-hospitalized PASC patients evaluated up to 3 years post-infection. Together, these findings suggest that similar biological mechanisms may underlie persistent post-COVID symptoms regardless of acute disease severity, and that these mechanisms are likely initiated early and may persist for years.

Persistent viral reservoirs have been proposed as drivers of sustained inflammation in PASC (17–19). However, in our study spike protein levels resulted similar in both patients and controls and did not correlate with DEPs levels or symptoms. This finding does not provide support for the hypothesis that persistent spike protein is a major driver of, or strongly associated with, post-COVID symptoms, but the possibility that viral antigens persist in tissues or at levels below the detection threshold of plasma assays cannot be excluded. Given similar exposure histories to SARS-CoV-2 infection and vaccination in both groups, comparable spike protein levels were expected. The absence of elevated spike levels in PASC does not suggest the presence of ongoing viral replication in plasma, which aligns with our previous findings where we were unable to detect SARS-CoV-2 RNA in peripheral blood mononuclear cells (PBMCs) of individuals with PASC (26). Our results are consistent with a recent study that also found no significant increase in plasma spike protein levels in PASC patients compared to recovered controls, and reported no association between spike protein persistence and symptom severity or functional impairment (42). Taken together, these observations suggest that the inflammatory process in PASC may become self-sustaining and may not require ongoing viral replication in the circulation, though tissue reservoirs cannot be excluded.

A major strength of our study is the long follow-up time of post-COVID symptoms extended over 39 months (mean) in patients who were affected by the pre-omicron variant, accompanied by a detailed clinical phenotyping and comprehensive inflammatory proteomic profiling. Additional strengths include the relatively large number of study subjects and the homogeneity of the patient and control groups regarding ethnicity, socioeconomic back-ground, severity and time of acute infection.

A potential limitation of our study is the focus on a predefined inflammatory protein panel which may have limited the detection of alterations in other biologically relevant pathways, such as metabolic, neurovascular, or mitochondrial processes. Additionally, the cross-sectional design limits causal inference between persistent inflammation and PASC symptoms, and future longitudinal or repeated-measure studies will be needed to strengthen causal conclusions. Finally, our key differentially expressed proteins were not validated using orthogonal methods or independent cohorts in the current study, which should be considered when interpreting their potential biological relevance. Nevertheless, our findings are broadly consistent with previous reports, including the PHOSP-COVID study, with which we share several DEPs, supporting the reproducibility of these proteomic changes.

In summary, our extended proteomic analysis shows that key pro-inflammatory cytokines characterizing the acute phase of SARS-CoV-2 infection remain dysregulated in individuals with PASC, even long after the initial infection. This persistent immune perturbation suggests a failure of proper immune resolution that may not require detectable viral persistence in plasma, although viral antigen persistence in tissues cannot be excluded.

The authors thank the statistician Katja Gabrysch at the Uppsala Clinical Research Center for her assistance with the statistical analysis.

The author(s) declared that financial support was received for this work and/or its publication. The study was funded by grants from the Healthcare Board, Region of Uppsala, Sweden (FoU grant to SF), with contributions from the Open Medicine Foundation (to JB), Uppsala University and University Hospital (to ND), and Hjärnfonden 2022-0042 (to ND).

The datasets presented in this study can be found in online repositories. The names of the repository/repositories and accession number(s) can be found in the article/. 1

The studies involving humans were approved by Swedish Ethical Review Authority (2021-06852-0160) and conducted in accordance with the Helsinki declaration. The studies were conducted in accordance with the local legislation and institutional requirements. The participants provided their written informed consent to participate in this study.

SF: Conceptualization, Data curation, Formal analysis, Funding acquisition, Investigation, Methodology, Project administration, Resources, Software, Supervision, Validation, Visualization, Writing – original draft, Writing – review & editing. JK: Data curation, Investigation, Methodology, Software, Visualization, Writing – original draft, Writing – review & editing. JS: Investigation, Writing – original draft. JB: Investigation, Supervision, Writing – review & editing. ND: Supervision, Writing – review & editing, Data curation, Funding acquisition.

The author(s) declared that this work was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

The author(s) declared that generative AI was not used in the creation of this manuscript.

Any alternative text (alt text) provided alongside figures in this article has been generated by Frontiers with the support of artificial intelligence and reasonable efforts have been made to ensure accuracy, including review by the authors wherever possible. If you identify any issues, please contact us.

All claims expressed in this article are solely those of the authors and do not necessarily represent those of their affiliated organizations, or those of the publisher, the editors and the reviewers. Any product that may be evaluated in this article, or claim that may be made by its manufacturer, is not guaranteed or endorsed by the publisher.

The Supplementary Material for this article can be found online at: https://www.frontiersin.org/articles/10.3389/fimmu.2026.1775044/full#supplementary-material↗