Stratification of hospitalized COVID-19 patients into clinical severity progression groups by immuno-phenotyping and machine learning, Mueller, 2022

Stratification of hospitalized COVID-19 patients into clinical severity progression groups by immuno-phenotyping and machine learning
Yvonne M. Mueller, Thijs J. Schrama, Rik Ruijten, Marco W. J. Schreurs, Dwin G. B. Grashof, Harmen J. G. van de Werken, Giovanna Jona Lasinio, Daniel Álvarez-Sierra, Caoimhe H. Kiernan, Melisa D. Castro Eiro, Marjan van Meurs, Inge Brouwers-Haspels, Manzhi Zhao, Ling Li, Harm de Wit, Christos A. Ouzounis, Merel E. P. Wilmsen, Tessa M. Alofs, Danique A. Laport, Tamara van Wees, Geoffrey Kraker, Maria C. Jaimes, Sebastiaan Van Bockstael, Manuel Hernández-González, Casper Rokx, Bart J. A. Rijnders, Ricardo Pujol-Borrell & Peter D. Katsikis

Abstract
Quantitative or qualitative differences in immunity may drive clinical severity in COVID-19. Although longitudinal studies to record the course of immunological changes are ample, they do not necessarily predict clinical progression at the time of hospital admission.

Here we show, by a machine learning approach using serum pro-inflammatory, anti-inflammatory and anti-viral cytokine and anti-SARS-CoV-2 antibody measurements as input data, that COVID-19 patients cluster into three distinct immune phenotype groups. These immune-types, determined by unsupervised hierarchical clustering that is agnostic to severity, predict clinical course.

The identified immune-types do not associate with disease duration at hospital admittance, but rather reflect variations in the nature and kinetics of individual patient’s immune response. Thus, our work provides an immune-type based scheme to stratify COVID-19 patients at hospital admittance into high and low risk clinical categories with distinct cytokine and antibody profiles that may guide personalized therapy.

Link
Full Text

 

As I read this (with a very much non-expert eye and a little tongue-in-cheek), it suggests to me that the 3 immuno-types might possibly represent.

Balanced response immunotype (BRI): "All good, for you this infection is actually mild"
Excessive inflammation immunotype (EXI): "Your ECMO machine is right this way Sir/Madam"
Low antibody immunotype (LAI): Later no evidence of seroconversion, but long COVID symptoms. "Are you sure you had an infection Monsieur/Madame or is it just a belief?"

(Obviously these were hospitalised patients in the study).

 

Sorry to be a bit slack here, have only read this thread. Wondering if this sort of phenotyping is infection-specific. i.e. if you tested a person who was not infected with covid 19, or if they were infected with something else entirely, would they always be categorised into the same phenotype?

 

No it's a Machine Learning classifier - specific to C19 in this instance. You could apply the same technique to other infections/diseases, which may or may not be successful. In this instance, with loadsa data and some guidance as inputs, the "black box" validated 3 groupings, which the researchers have chosen to label as above. The immunology is complex so I'm only trying to see it from a high level.

I think this sort of thing is going to be helpful in understanding C19/long COVID. I also think that similar techniques can be applied to ME, as it seems useful when there are multiple and variable small derangements, rather than a single clearly abnormal bio-marker (given the current state of the art).