Nikhil Prasad Fact checked by:Thailand Medical News Team Sep 30, 2026 48 minutes ago
Long COVID may not be a single biological disorder but a collection of different molecular disturbances, with particular symptoms carrying their own distinctive protein fingerprints, according to new research examining nearly 2,900 proteins in people previously infected with SARS-CoV-2.
Researchers identified 235 proteins associated with individual long COVID symptoms, revealing distinctive
molecular patterns behind fatigue, breathlessness, memory problems, hair loss, and other persistent effects.
The findings provide unusually detailed evidence that symptoms including fatigue, shortness of breath, memory difficulties, hair loss, anxiety, chest pain, and loss of smell and taste may be associated with different biological processes.
Nearly 2,900 Proteins Put Under the Microscope
The study involved 495 SARS-CoV-2 survivors participating in the Gutenberg COVID-19 Study. Their average age was 53.2 years.
Researchers initially measured 2,945 plasma proteins using high-throughput proteomic technology. After excluding 46 proteins because of detection limitations or technical variability, 2,899 proteins were included in the analysis.
Participants had reported 61 possible post-COVID symptoms. Because reliable statistical modeling requires sufficient numbers of people with each condition, symptoms occurring in fewer than 10 participants were excluded, leaving 21 for detailed investigation.
Machine-learning analysis ultimately identified 235 unique proteins associated with 15 post-COVID symptoms.
Fatigue was the most prevalent symptom analyzed, affecting 58 participants, or 11.7%.
IL1A Linked to Five Different Symptoms
Among the most striking discoveries involved interleukin-1 alpha (IL1A), a protein involved in inflammatory signaling.
IL1A was associated with shortness of breath, hair loss, depression, anxiety, and memory problems, making it the protein appearing across the greatest number of different symptom signatures.
This is potentially important because IL1A is a pro-inflammatory mediator involved in regulating inflammatory responses. Previous research has suggested that persistent inflammation could contribute to some post-COVID abnormalities, although the new findings do not prove that IL1A causes long COVID symptoms.
Another notable protein was paraoxonase 3 (PON3). It was associated with fatigue, slow movements, depression, mood swings, and loss of interest or pleasure.
Paraoxonase proteins normally participate in processes that protect against oxidative damage. The researchers therefore suggest that altered PON3 could provide another avenue for investigating the biological mechanisms associated with persistent symptoms.
Breathlessness Has a Particularly Complex Protein Signature
Not all symptoms had similarly sized molecular fingerprints. Shortness of breath had the largest signature, involving 51 proteins, suggesting that persistent breathlessness after COVID-19 may involve a particularly complex combination of biological changes.
By contrast, the molecular signatur
e for anxiety contained only three proteins: IL1A, ANXA1, and SWAP70.
Concentration difficulties were associated with five proteins: PRKAG3, KHK, CHP1, CDC25A, and ANXA1.
The appearance of ANXA1 in both anxiety and concentration problems also illustrates how individual proteins can cross conventional symptom categories.
Smell, Taste, and Memory Reveal More Molecular Clues
Loss or alteration of smell and taste showed particularly strong molecular similarities. The two symptoms shared 11 proteins, representing the greatest protein overlap observed between symptoms.
Among proteins associated with olfactory dysfunction, the researchers highlighted TLR4, ANXA1, SFTPA1, RNF5, PTGES2, and NUDT15 as candidates deserving further investigation. Collectively, these proteins are involved in processes including inflammation, immunity, and metabolism.
Memory problems produced another intriguing molecular profile. Proteins associated with memory impairment included MAPT, IL1A, NTF3, SHBG, AIF1, and RTN4R. These candidates point toward potential involvement of neuroinflammation, synaptic function, and other neurological processes.
However, these associations remain exploratory. They do not establish that any particular protein directly produces memory impairment or another long COVID symptom.
Six Different Symptom Groups Identified
This
Thailand Medical News report highlights another important finding: similarities between protein profiles allowed the researchers to separate symptoms into six molecular clusters.
The first contained fatigue, shortness of breath, loss of interest or pleasure, and slow movements.
A second connected anxiety with palpitations, while a third brought together hair loss, memory problems, mood swings, and concentration difficulties.
The fourth cluster contained smell and taste disturbances, while sleep disturbances and depression formed the fifth. Interestingly, chest pain formed its own sixth cluster.
Researchers identified 59 molecular pathways associated with these clusters. The fatigue and breathlessness cluster showed enrichment of pathways involving metabolic processes, including fatty-acid metabolism, as well as cellular signaling such as ATM signaling.
The cluster containing hair loss and cognitive and mood symptoms involved immune-inflammatory responses and cellular signaling. Chest pain, meanwhile, was uniquely associated with positive regulation of exocytosis, a cellular process through which substances are released from cells.
Protein Findings Reflected Objective Clinical Tests
One particularly valuable part of the study was the comparison between molecular signatures and objective clinical measurements.
Higher protein scores associated with memory and concentration problems corresponded with poorer performance on the Montreal Cognitive Assessment, a widely used cognitive screening test.
Protein patterns associated with smell and taste disturbances were similarly related to performance on an objective 12-item smell-identification test.
The molecular signature associated with mood swings was related to scores on the PHQ-9, which measures depressive symptoms, while the sleep-disturbance protein signature was associated with the Jenkins Sleep Scale.
Interestingly, the protein signature associated with shortness of breath did not show clinically relevant relationships with conventional measurements of lung function, including vital capacity and forced expiratory volume in one second.
Symptom-Specific Models Perform Better
The researchers also discovered that examining individual symptoms produced more useful molecular information than treating post-COVID illness as one broad condition.
A generalized post-COVID model produced an AUC of 0.62, whereas the initial symptom-specific model for chest pain reached 0.84.
More stringent nested cross-validation generated more conservative values ranging from 0.52 to 0.74. Hair loss achieved an AUC of 0.74, while chest pain reached 0.71.
Overall, protein-based models performed better than models relying only on age and sex in 11 of the 15 symptom comparisons, or 73.3%.
Institutions
The researchers were from the University Medical Center of Johannes Gutenberg University Mainz; German Center for Cardiovascular Research; Institute of Medical Biostatistics, Epidemiology and Informatics; Research Center for Immunotherapy; Institute of Immunology; Institute for Quantitative and Computational Biosciences; German Cancer Consortium; Centre for Healthy Ageing; Rhine-Main Neuroscience Network; Center for Thrombosis and Hemostasis; and Institute of Molecular Biology, Germany.
Conclusions
The findings provide further evidence that long COVID may represent a collection of biologically distinct but overlapping conditions rather than one uniform disease process. Different symptoms were associated with different protein combinations and biological pathways involving inflammation, metabolism, immune activity, cellular signaling, and neurological processes. However, because the research was cross-sectional and exploratory, it cannot establish whether these proteins cause the symptoms or are consequences of them. Larger longitudinal studies will be necessary before these molecular signatures can be used for diagnosis, patient classification, or targeted treatment.
The study findings were published in the peer reviewed journal: Molecular & Cellular Proteomics.
https://www.mcponline.org/article/S1535-9476(26)00166-0/fulltext
Read Also:
https://www.thailandmedical.news/articles/long-covid
Share
Tweet
Share
