A blood test performed at the time of Sars-CoV-2 infection promises to predict Long Covid syndrome, identifying patients most at risk of ailments one year after infection. This is the news that emerges from a study published in ‘eBioMedicine – The Lancet’, conducted by researchers from University College London (Ucl) on small numbers. If the results are confirmed by broader investigations, the authors envisage the possibility of predicting post Covid-19 syndrome by offering a test already at the time of the diagnosis of positivity to the pandemic coronavirus. The future test would measure the levels of a series of proteins, the plasma concentrations of which were found to be particularly high in those infected with persistent symptoms 12 months after infection.
Scientists analyzed plasma samples from 54 healthcare workers with Covid confirmed by molecular swab or antibody tests, taken every week for 6 weeks in spring 2020, comparing them to samples collected in the same period from 102 healthcare workers who had not been infected with SARS. CoV-2. Using targeted mass spectrometry techniques, the authors investigated how Covid-19 affected plasma protein levels over 6 weeks. They thus detected abnormal, very high concentrations for 12 of the 91 proteins evaluated, highlighting that the degree of abnormality in protein concentrations was associated with the severity of symptoms. The team also noted that, at the time of diagnosis of Sars-CoV-2 positivity, abnormal levels of 20 proteins were predictive of disturbances that persisted one year after infection. Most of these ‘spy’ proteins were linked to anticoagulant and anti-inflammatory mechanisms.
The researchers then turned to artificial intelligence for help, training a machine learning algorithm that learned to examine participants’ protein profiles and was able to distinguish all 11 operators who reported at least one persistent symptom 12 months after infection. . Another machine learning tool was used to estimate the probability that the test would be wrong, indicating a possible error rate of 6%.
“Our study shows that even a mild or asymptomatic Covid infection alters the blood plasma protein profile,” explains Gaby Captur, lead author of the work that was funded by the National Institute for Health and Care Research, Great Ormond Street Hospital Biomedical Research Center (Nihr Gosh Brc), the British Heart Foundation and the Barts charity.
“Our Long Covid prediction tool must be validated in an independent and larger group of patients – specifies the scientist – However, according to our approach a test that predicts the risk of Long Covid at the time of initial infection” by Sars -CoV-2 “could be implemented quickly and cost-effectively. Our method of analysis is in fact readily available in hospitals and is high-throughput in the sense that it can analyze thousands of samples in one afternoon.”
“If we can identify the people who could develop Long Covid – comments senior author Wendy Heywood – this will pave the way for the experimentation of treatments such as antivirals administered in the early stages of infection, to understand if they can reduce the risk of Long Covid. “.