A patient-centric modeling framework captures recovery from SARS-CoV-2 infection, 2023, Ruffieux et al

[Andy]

Abstract

The biology driving individual patient responses to severe acute respiratory syndrome coronavirus 2 infection remains ill understood. Here, we developed a patient-centric framework leveraging detailed longitudinal phenotyping data and covering a year after disease onset, from 215 infected individuals with differing disease severities.

Our analyses revealed distinct ‘systemic recovery’ profiles, with specific progression and resolution of the inflammatory, immune cell, metabolic and clinical responses. In particular, we found a strong inter-patient and intra-patient temporal covariation of innate immune cell numbers, kynurenine metabolites and lipid metabolites, which highlighted candidate immunologic and metabolic pathways influencing the restoration of homeostasis, the risk of death and that of long COVID. Based on these data, we identified a composite signature predictive of systemic recovery, using a joint model on cellular and molecular parameters measured soon after disease onset. New predictions can be generated using the online tool http://shiny.mrc-bsu.cam.ac.uk/apps/covid-19-systemic-recovery-prediction-app, designed to test our findings prospectively.

Open access, https://www.nature.com/articles/s41590-022-01380-2

 

[Andy]

Immunologic prediction of long COVID

In addition to the acute phase of SARS-CoV-2 infection, a significant percentage of patients experience a prolonged illness with varying symptomatology. Longitudinal SARS-CoV-2 patient-centric immunologic, inflammatory and metabolic data collection has allowed the generation of a composite signature to predict recovery.
....
Symptoms of PASC comprise cough, fatigue, brain fog, myalgias and arthralgias, chest pain and many others, including disruption of smell and taste2. Many of these somatic complaints are relatively nonspecific and have been noted with other viral infections, including Epstein–Barr virus (EBV)4. The anosmia (loss of smell) and aguesia (loss of taste) may be fairly COVID-19 specific, with SARS-CoV-2 infecting the support cells for the olfactory nerves5. However, much of what is experienced during long-term post-SARS-CoV-2 infection overlaps with chronic fatigue syndrome/myalgic encephalitis, and it has been hypothesized to result from EBV reactivation6. Similarly, multiple sclerosis has been linked with EBV infection7. Nonetheless, the biology and immunology of PASC remain unclear, with commonly suggested hypotheses ranging from effects of pathogen remnants to autoimmunity to microbiome dysbiosis to direct tissue damage4. Multiple reports have explored risk definition for PASC development8, but the study by Ruffieux et al. explores patient-centric clinical, inflammatory, immunologic and metabolic parameters longitudinally to predict outcomes3.

https://www.nature.com/articles/s41590-022-01396-8

 

[SNT Gatchaman]

The methodology/analysis is complex but some summary quotes from the discussion —

Our joint modeling framework of the longitudinal immunologic, metabolic and clinical data on a SARS-CoV-2 cohort indicated protracted temporal covariation patterns that highlighted acute-phase inflammation as a common denominator interlinked with incomplete clinical, immunologic and metabolic recovery up to a year after disease onset, and suggested that coordinated dynamics of the innate immune system, kynurenine and host lipid metabolism were likely underpinning restoration of overall homeostasis.

These profiles constituted distinct patient ‘systemic recovery’ categories, which characterized disease course, risk of death and of long COVID beyond peak clinical severity.

Finally, an early prognosis of recovery for a new patient can be obtained from our pilot predictive model [link], whose excellent performance in our cohort warrants independent validation to evaluate clinical actionability.

Our data suggested that a limited number of pathophysiologic processes impact many of the parameters appearing together in early composite signatures predictive of poor prognosis. Of note, both the inflammatory immunopathology of severe acute SARS-CoV-2 infection [...] and hemophagocytic lymphohistiocytosis, a cytokine storm syndrome, are characterized by elevated triglycerides—likely linked to protracted hyper-inflammation. NK cells play a central role in anti-viral immunity through the secretion of pro-inflammatory cytokines and cytotoxic activity, and NK cell dysfunction is also a key criterion of hemophagocytic lymphohistiocytosis.

Pro-inflammatory cytokines activate the kynurenine pathway through induction of indoleamine 2,3-dioxygenase (IDO-1), as observed in our cohort. Kynurenine-pathway activation has been implied in linking inflammation and central nervous system alterations by favoring the degradation of tryptophan toward 3‐hydroxykynurenine and quinolinic acid, both of which appeared in the first predictive signature, and by reducing serotonin production, which appeared in the second signature.

 

[SNT Gatchaman]

For reference, the condition they refer to, HLH, is a cytokine release ("storm") syndrome and can be primary/familial or secondary, eg with infection, autoimmune/inflammatory conditions and malignancy, esp. in children. I note that Jo and SnowLeopard have previously commented that there isn't good evidence for cytokine storm itself in COVID but am assuming the authors here are more interested in shared mechanisms, esp relating to NK cells.

Some recent references —

Traffic jam within lymphocytes: A clinician's perspective (2023)
Cytokine Storm Syndrome (2023)
Malignancy-associated haemophagocytic lymphohistiocytosis (2022, paywall)
Recent advances in the treatment of hemophagocytic lymphohistiocytosis and macrophage activation syndrome (2022, paywall)
A Review of Current and Emerging Therapeutic Options for Hemophagocytic Lymphohistiocytosis (2022, paywall)

 

[Sasha]

This seems a pretty amazing approach...

 

[Hutan]

Here we adopted a wider perspective to understand organismal recovery, based on the same parameters as before9 (immune cell subsets, serum cytokines and C-reactive protein (CRP) levels) but collected over an extended follow-up period of 12 months, as well as newly established data (polar metabolites, glycoproteins and lipoproteins), and patient questionnaires addressing long-term symptoms of disease.

We recruited 215 SARS-CoV-2 PCR-positive patients (hereafter COVID-19 patient, CovP)

Briefly, inpatient classes (C to E) were sampled at enrollment, approximately weekly up to 4 weeks, and then every 2 weeks up to 12 weeks. Discharged CovPs were asked to provide a follow-up sample 4–8 weeks after enrollment. Outpatient classes (A and B) were sampled at enrollment and subsequently after approximately 2 and 4 weeks. Participant recall beyond the original study period9occurred at approximately 3, 6 and 12 months following recruitment. Additionally, 14 CovPs were newly included, after discharge from hospital.

CRP levels, five serum cytokines (interferon (IFN)-γ, interleukin (IL)-10, IL-1β, IL-6, tumor necrosis factor (TNF)), 36 polar metabolites, 103 glycoproteins and lipoproteins, and 33 immune cell subsets were quantified from the blood samples collected at the above time points.

Follow-up questionnaires10 assessing long-term symptoms were obtained from symptomatic CovPs (classes B to E) between 3 and 11 months after symptom onset (up to three questionnaires per CovP, median interval between two consecutive questionnaires for a same CovP: 5 months, IQR 1.5 month). The cohort also comprised 45 uninfected healthcare workers (hereafter healthy control, HC), defined by negative PCR test and serology, for whom blood samples were obtained at enrollment only.

Group 3 (persistent inflammation) showed widespread and long-lasting cellular alterations. Most notably, plasmablast and non-naive HLA-DR+CD38+ CD8+T cell numbers were still increased, whereas the numbers of plasmacytoid dendritic cells, CD4+ follicular helper T-like cells, Vγ9 Vδ2lo γδ T cells, MAIT cells, naive B cells and CD4+ regulatory T (Treg) cells were still decreased in the fourth time window (weeks 13–27; Fig. 3a). Similarly, there were persistent metabolic alterations (weeks 13–27) in group 3 compared to HCs, such as increased kynurenine, quinolinic acid and glutamic acid, and decreased levels of tryptophan, indole-3-acetic acid and serotonin (Fig. 3b), as well as increased levels of VLAB, GlycA and GlycB (Fig. 3c). Using group-level longitudinal mixed models with time modeled as a continuous variable, we next assessed baseline group effects (that is, group differences at symptom onset) and group × time interaction effects. Nearly half of the parameters included in this study had significant baseline and/or interaction effect(s) (for instance, NK cells, CD8+ TEM cells, myeloid dendritic cells, CD8+ TN cells, GlycA and GlycB displayed both baseline and interaction effects, CD4+ follicular helper T-like cells, Vγ9 Vδ2lo γδ T cells, MAIT cells, kynurenine, quinolinic acid and tryptophan had significant baseline effects only, and HLA-DR+CD38+ CD8+ T cells had significant interaction effects only; Fig. 3a–c). These parameters, when measured early after infection, might therefore contain information regarding an individual’s ability to recover. In all, these results indicated that recovery groups 1–3 had dissimilar parameter recovery rates, with group-3 CovPs experiencing persisting biologic disruptions up to 6 months after symptom onset.

 

[Hutan]

On the Long Covid analysis:

We next asked how biologic recovery profiles related to long COVID, using questionnaires on long-term symptoms (covering respiratory, neurological, gastrointestinal and other physical sequelae)10. These questionnaires were collected from 65 CovPs (54% female, median age 51 years old, IQR 29 years old) between months 2 and 11 after symptom onset. A comparison of the reported symptoms indicated that CovPs in group 3 reported more neurological symptoms (fatigue, muscle weakness, pain, difficulty eating, drinking, swallowing) compared to group 1 (Fig. 4a). Of note, in group 3, fatality and mechanical ventilation (a source of non-infection-related sequelae) added unavoidable limitations.

A latent factor analysis, which aimed to characterize the joint manifestation of the reported symptoms, identified two latent factors, the first of which (LFA1) appeared to be driven by the neurological symptoms. In particular, new neurology in limbs, fatigue and muscle weakness were attributed the largest loadings for LFA1 (Fig. 4a). As it accounted for most of the modeled variability, LFA1 also served as a natural patient-level latent proxy for long-term clinical manifestations, and was associated with the recovery groups (Fig. 4b).

As such, LFA1 was significantly higher for CovPs from group 3 (poorer score), and lower for CovPs from group 1 (Fig. 4b). However, there was substantial variability within each recovery group. In particular, two patients, CV0165 in group 1 and CV0201 in group 2, had high composite scores, although their cellular, inflammatory and molecular trajectories had essentially returned to normal levels by week 7 after symptom onset

This suggested that systemic, subjectively perceived sequelae persisted in these individuals despite absent, or rapidly resolving, inflammation and cellular/molecular disruption.

 

[Hutan]

I struggled to get much out of this paper that is relevant to Long Covid. Perhaps the sample that was followed for months wasn't big enough to have sufficient Long covid participants. It looks as though people with persistent fatigue are most likely to have initial features consistent with Groups 2 and 3, rather than Group1. But, that doesn't narrow things down a lot.

I suspect that there is data in this study that could tell us more about those with persistent fatigue. Unfortunately the symptoms they investigated aren't particularly relevant to LC-ME/CFS.